Research ArticleTemporal Identification of Dysregulated Genes andPathways in Clear Cell Renal Cell Carcinoma Based onSystematic Tracking of Disrupted Modules

Shao-Mei Wang1 Ze-Qiang Sun2 Hong-Yun Li2 Jin Wang2 and Qing-Yong Liu2

1Center for Kidney Disease Jinan Central Hospital Affiliated to Shandong University Jinan Shandong 250013 China2Department of Urinary Surgery Qianfoshan Hospital Affiliated to Shandong University 16766 Jingshi Road JinanShandong 250014 China

Correspondence should be addressed to Qing-Yong Liu qingyongliu ayeahnet

Received 18 June 2015 Revised 31 July 2015 Accepted 11 August 2015

Academic Editor Konstantin Blyuss

Copyright copy 2015 Shao-Mei Wang et alThis is an open access article distributed under the Creative CommonsAttribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Objective The objective of this work is to identify dysregulated genes and pathways of ccRCC temporally according to systematictracking of the dysregulated modules of reweighted Protein-Protein Interaction (PPI) networks Methods Firstly normal andccRCC PPI network were inferred and reweighted based on Pearson correlation coefficient (PCC) Then we identified alteredmodules using maximumweight bipartite matching and ranked them in nonincreasing order Finally gene compositions of alteredmodules were analyzed and pathways enrichment analyses of genes in altered modules were carried out based on ExpressionAnalysis Systematic Explored (EASE) test Results We obtained 136 576 693 and 531 disrupted modules of ccRCC stages I II IIIand IV respectively Gene composition analyses of altered modules revealed that there were 56 common genes (such as MAPK1CCNA2 and GSTM3) existing in the four stages Besides pathway enrichment analysis identified 5 common pathways (glutathionemetabolism cell cycle alanine aspartate and glutamate metabolism arginine and proline metabolism and metabolism ofxenobiotics by cytochrome P450) across stages I II III and IV Conclusions We successfully identified dysregulated genes andpathways of ccRCC in different stages and these might be potential biological markers and processes for treatment and etiologymechanism in ccRCC

1 Introduction

Clear cell renal cell carcinoma (ccRCC) is the most commontype of kidney cancer and accounts for approximately 60to 70 of all renal tumors [1] Patients with ccRCC comprisea heterogeneous group of patients with variable pathologicstage and grade used to stratify patients and infer prognosis[2] However providing patients with reliable informationabout anticipated treatment response is challenging due tothe molecular heterogeneity of ccRCC [3] Delineating thepathogenesis of ccRCC by investigating the gene and epige-netic changes and their effects on key molecules and theirrespective biologic pathways is of crucial importance for theimprovement of current diagnostics prognostics and drugdevelopment [4] For example studies suggest that ccRCC isclosely associated with tumor suppressor von-Hippel Lindau

(VHL) gene mutations that lead to stabilization of hypoxiainducible factors (HIF-1120572 and HIF-2120572 also known as HIF1Aand EPAS1) in both sporadic and familial forms [5 6]

With the advances of high-throughput experimentaltechnologies large amounts of Protein-Protein Interaction(PPI) data are uncovered which make it possible to studyproteins on a systematic level [7 8] In addition a PPInetwork can be modeled as an undirected graph wherevertices represent proteins and edges represent interactionsbetween proteins to prioritize disease associated genes orpathways and to understand the modus operandi of diseasemechanisms [9 10] But it has been noticed that PPI data areoften associated with high false positive and false negativerates due to the limitations of the associated experimentaltechniques and the dynamic nature of protein interactionmaps which may have a negative impact on the performance

Hindawi Publishing CorporationComputational and Mathematical Methods in MedicineVolume 2015 Article ID 313740 11 pageshttpdxdoiorg1011552015313740

2 Computational and Mathematical Methods in Medicine

of complex discovery algorithms [11] Many computationalapproaches have been proposed to assess the reliabilityof protein interactions data An iterative scoring methodproposed by Liu et al [12] was selected to evaluate thereliability and predict new interactions and it has been shownto perform better than other methods However studyingmultiple diseases simultaneously makes it challenging todiscern clearly the intricate underlying mechanisms

In addition it is important to effectively integrate omicsdata into such an analysis for example Chu and Chen[13] combined PPI and gene expression data to construct acancer perturbed PPI network in cervical carcinoma to studygain- and loss-of-function genes as potential drug targetsMagger et al [14] combined PPI and gene expression datato construct tissue-specific PPI networks for 60 tissues andused them to prioritize disease genes Beyond straightforwardscoring genes in the gene regulatory network it is crucial tostudy the behavior of modules across specific conditions ina controlled manner to understand the modus operandi ofdisease mechanisms and to implicate novel genes [15] sincesome of the important genes may not be identifiable throughtheir own behavior but their changes are quantifiable whenconsidered in conjunctionwith other genes (eg asmodules)What is required therefore is systematic tracking genepathways and module behavior across specific conditions ina controlled manner

Therefore in this paper we performed a temporal (stagesI II III and IV of ccRCC) analysis between normal controlsand ccRCC patients to identify disrupted genes and pathwaysby systematically tracking the altered modules of reweightedPPI network To achieve this we firstly inferred normaland ccRCC cases of different stages PPI networks based onPearson correlation coefficient (PCC) next clique-mergingalgorithmwas performed to exploremodules in PPI networkand we compared these modules to identify altered modulesthen gene composition of thesemoduleswas analyzed finallypathways enrichment analysis of genes in altered moduleswas carried out based on Expression Analysis SystematicExplored (EASE) test

2 Materials and Methods

21 Inferring Normal and ccRCC PPI Network

211 PPI Network Construction We utilized a dataset ofhuman PPI network the Search Tool for the Retrievalof Interacting GenesProteins (STRING) which comprised16730 genes and 1048576 interactions [16] For STRING self-loops and proteins without expression value were removedThe remaining largest connected component with score ofmore than 08 was kept as the selected PPI network whichconsisted of 8590 genes and 53975 interactions

212 Gene Expression Dataset and Dataset Preprocessing Amicroarray expression profile E-GEOD-53757 from ArrayExpress database was selected for ccRCC related analysis E-GEOD-53757which existed onAffymetrixGeneChipHumanGenome U133 Plus 20 Platform was divided into 4 groupsaccording to tumor stage (stages I II III and IV)There were

24 19 14 and 15 ccRCC patients at stages I II III and IVrespectively the number of normal controls in each stage wasequaled to its patientsrsquo number

The expression profile was preprocessed by standardmethods consisting of ldquormardquo [17] ldquoquantilesrdquo [18] ldquomasrdquo[19] and ldquomedianpolishrdquo [17] To be specific ldquormardquo methodwas carried out for background correction to eliminateinfluences of nonspecific hybridization [17] The quantilenormalization algorithm was a specific case of the trans-formation 119909

1015840

119894= 119865minus1

(119866(119909119894)) where we estimated 119866 by the

empirical distribution of each array and119865 using the empiricaldistribution of the averaged sample quantiles [18] Perfectmatch (PM)mismatch (MM) correction was conducted byldquomasrdquo method [19] Summarization of the probe data wasconducted by ldquomedianpolishrdquo [17] A multichip linear modelwas fit to data from each probe set In particular for a probeset 119896 with 119894 = 1 119868

119896probes and data from 119895 = 1 119869

arrays we fitted the following model log2(PM119896119894119895) = 120572119896

119894+ 120573119896

119895+

120576119896

119894119895 where 120572

119894was a probe effect and 120573

119895was the log

2expression

valueNext the data were screened by feature filter method of

gene filter package and the number of genes with multipleprobes was 20102 At last we obtained the gene expressionvalue for each gene including 20102 genes from 144 samples(72 normal controls and 72 ccRCC patients)

213 Reweighting Gene Interactions by PCC Gene interac-tions in network based on ccRCC patients of different stages(stages I II III and IV) and their normal controls werereweighted by PCC which evaluated the probability of twocoexpressed gene pairs PCC is a measure of the correlationbetween two variables giving a value between minus1 and +1inclusively [20] The PCC of a pair of genes (119909 and 119910)which encoded the corresponding paired proteins (119906 and V)interacting in the PPI network was defined as

PCC (119909 119910)

=

1

119904 minus 1

119904

sum

119894=1

(

119892 (119909 119894) minus 119892 (119909)

120590 (119909)

) sdot (

119892 (119910 119894) minus 119892 (119910)

120590 (119910)

)

(1)

where 119904 was the number of samples of the gene expressiondata 119892(119909 119894) or 119892(119910 119894) was the expression level of gene 119909 or119910 in the sample 119894 under a specific condition 119892(119909) or 119892(119910)

represented themean expression level of gene119909 or119910 and119892(119909)

or 119892(119910) represented the standard deviation of expression levelof gene 119909 (or 119910)

The PCC of a pair of proteins (119906 and V) was defined asthe same as the PCC of their corresponding paired genes (119909and 119910) which was PCC(119906 V) = PCC(119909 119910) If PCC(119906 V) has apositive value there is a positive linear correlation between119906 and V In addition we defined PCC of each gene-geneinteraction as weight value of the interaction

22 Identifying Modules from the PPI Network In this papermodule-identification algorithm is based on clique-merging[21 22] and is similar to the method proposed by Liu et al[12] It consisted of three steps in the first step it found

Computational and Mathematical Methods in Medicine 3

all the maximal cliques from the weighted PPI networkMaximal cliques were evaluated by a fast depth-first searchwith pruning-based algorithm proposed by Tomita et al[23] It utilized a depth-first search strategy to enumerate allmaximal cliques and effectively pruned nonmaximal cliquesduring the enumeration process

In the second step we assigned a score to each clique thescore of a clique119862was defined as its weighted density 119889

119882(119862)

119889119882

(119862) =

sum119906isin119862Visin119862119908 (119906 V)|119862| sdot (|119862| minus 1)

(2)

where119908(119906 V)was theweight of the interaction between 119906 andV We ranked these cliques in nonincreasing order of theirweighted densities 119862

1 1198622 119862

119896

Finally we went through this ordered list repeatedlymerging highly overlapping cliques to build modules Forevery clique 119862

119894 we repeatedly looked for a clique 119862

119895(119895 gt 119894)

such that the overlap |119862119894cap 119862119895||119862119895| ge 119905 where 119905 = 05

was a predefined threshold for overlapping [15] If such 119862119895

was found we calculated the weighted interconnectivity 119868119908

between 119862119894and 119862

119895as follows

119868119908

(119862119894 119862119895)

= radic

sum119906isin(119862119894minus119862119895)

sumVisin119862119895 119908 (119906 V)10038161003816100381610038161003816119862119894minus 119862119895

10038161003816100381610038161003816sdot

10038161003816100381610038161003816119862119895

10038161003816100381610038161003816

sdot

sum119906isin(119862119895minus119862119894)

sumVisin119862119894 119908 (119906 V)10038161003816100381610038161003816119862119895minus 119862119894

10038161003816100381610038161003816sdot1003816100381610038161003816119862119894

1003816100381610038161003816

(3)

If 119868119908(119862119894 119862119895) ge 119905 then 119862

119895was merged into 119862

119894forming a

module else 119862119895was discarded

We captured the effect of differences in interactionweights between normal and ccRCC cases through theweighted density-based ranking of cliques Weighted densityassigned higher rank to larger and stronger cliquesThereforewe expected cliques with lost proteins or weakened interac-tions to go down the rankings resulting in altered modulegeneration thereby capturing changes in modules betweennormal and ccRCC cases

23 Comparing Modules between Normal and ccRCC Con-ditions The approach to compare modules between normaland ccRCC conditions is similar to the method proposed bySrihari and Ragan [15] In detail 119867

119873and 119867

119879represented

the PPI network of normal controls and ccRCC patientsidentifying the sets of modules 119878 = 119878

1 1198782 119878

119898 and

119879 = 1198791 1198792 119879

119899 respectively For each 119878

119894isin 119878 module

correlation density 119889119888(119878119894) was defined as

119889119888(119878119894) =

sum119909119910isin119878119894

PCC ((119909 119910) 119872)

1003816100381610038161003816119878119894

1003816100381610038161003816sdot (

1003816100381610038161003816119878119894

1003816100381610038161003816minus 1)

(4)

Correlation densities of ccRCC modules (119889119888(119879119894)) were

calculated similarlyDisrupted or altered module pairs were evaluated by

modeling the set Υ(119878 119879) as maximum weight bipartitematching [24] Firstly we build a similarity graph 119872 = (119881

119872

119864119872

) where 119881119872

= 119878 cup 119879 and 119864119872

= cup(119878119894 119879119895) 119869(119878

119894 119879119895) ge

119905119869 Δ119862(119878119894 119879119895) ge 120575 whereby 119869(119878

119894 119879119895) = |119878

119894cap 119879119895||119878119894cup 119879119895| was

the Jaccard similarity andΔ119862(119878119894 119879119895) = |119889

119888(119878119894)minus119889119888(119879119894)|was the

differential correlation density between 119878119894and 119879

119895 and 119905

119869and

120575 were thresholds with 23 and 005 [15] 119869(119878119894 119879119895) weighted

every edge (119878119894 119879119895) We next identified the disrupted module

pairs Υ(119878 119879) by detecting the maximum weight matchingin 119872 and we ranked them in nonincreasing order of theirdifferential density Δ

119862 At last we inferred genes involved

in ccRCC as Γ = 119892119892 isin 119878119894

cup 119879119895 (119878119894 119879119895) isin Υ(119878 119879)

and ranked in nonincreasing order of Δ119862(119878119894 119879119895) To identify

altered modules we matched normal and ccRCCmodules bysetting high 119905

119869 which ensured that the module pairs either

had the same gene composition or had lost or gained only afew genes

24 Functional Enrichment Analysis To further investigatethe biological functional pathways of genes in altered mod-ules fromnormal controls and ccRCC Kyoto Encyclopedia ofGenes and Genomes (KEGG) pathway enrichment analysiswas performed by Database for Annotation Visualizationand Integrated Discovery (DAVID) [25] KEGG pathwayswith 119875 value lt 0001 were selected based on EASE testimplemented in DAVID EASE analysis of the regulatedgenes indicated molecular functions and biological processesunique to each category [26] The EASE score was used todetect the significant categories In both of the functionaland pathway enrichment analyses the threshold of theminimum number of genes for the corresponding term gt 2was considered significant for a category

119901 =

(119886+119887

119886) (119888+119889

119888)

(119899

119886+119888 )

(5)

where 119899was the number of background genes 1198861015840 was the genenumber of one gene set in the gene lists 1198861015840+119887was the numberof genes in the gene list including at least one gene set 1198861015840 + 119888

was the gene number of one gene list in the background genes1198861015840 was replaced with 119886 = 119886

1015840minus 1

3 Results

31 Analyzing Disruptions in ccRCC PPI Network Weobtained 20102 genes of normal and ccRCC cases afterpreprocessing and then investigated intersections betweenthese genesrsquo interactions and STRING PPI network andidentified PPI networks of normal and ccRCC cases Thenormal 119867

119873and ccRCC 119867

119879PPI networks of different stages

(stages I II III and IV) displayed equal numbers of nodes(8050) and interactions (49151) Although their interactionscores (weights) were different from each other as shownin Figure 1 there was no statistical significance betweennormal and ccRCC cases in different stages in whole levelbased on Kolmogorov-Smirnov test (119875 gt 005) Howeverthe score distribution between the ccRCC networks andnormal networks was different especially for stages III andIV in the score distribution 0sim03 (Figures 1(c) and 1(d))Examining these interactions more carefully distributionsamong different stages were also different and changes ofccRCC networks and normal networks were more and moreobvious from stage I to stage IV

4 Computational and Mathematical Methods in Medicine

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(a)

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(b)

NormalccRCC

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

(c)

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(d)

Figure 1 Score-wise distributions of interactions normal versus ccRCC cases (a) represents stage I of ccRCC (b) represents stage II (c)represents stage III and (d) represents stage IV

32 Analyzing Disruptions in ccRCC Modules Clique-merging algorithm was selected to identify disrupted oraltered modules from normal and ccRCC PPI network inthis paper In detail we performed a comparative analysisbetween normal 119873 and ccRCC 119879 modules to understanddisruptions at the module level Maximal cliques of normaland ccRCC PPI network were obtained based on fast depth-first algorithm and maximal cliques with the threshold ofnodes gt 5 were selected for module analysis Overall wenoticed that the total number of modules (1895) as well asaverage module sizes (20235) was almost the same acrossthe two conditions and four stages Table 1 showed overallchanged rules of weighted interaction density betweennormal modules and ccRCC modules we could find that

maximal and average weighted density of normal case wassmaller than that of ccRCC for each stage in detail theaverage weighted density of stages III (0075) and IV (0089)was a little higher than that of stages I (0068) and II (0046)while in the overall level the difference of module densityscores had no statistical significance between normal andccRCC cases in different stages with 119875 gt 005 Further therelationship between modules weighted density distributionand numbers of modules was illustrated in Figure 2 Themodule numbers were different when the interaction densityranged from 005 to 025 especially for stages II III andIV of ccRCC These differences might be the reasons ofweighted density changes of ccRCC from different stages(Table 1)

Computational and Mathematical Methods in Medicine 5

Table 1 Correlations of normal and ccRCC modules of different stages

Module set Stage I Stage II Stage III Stage IVNormal ccRCC Normal ccRCC Normal ccRCC Normal ccRCC

PCC correlationMaximal 0315 0324 0254 0278 0324 0339 0294 0326Average 0068 0083 0046 0074 0075 0087 0089 0084Minimum minus0076 minus0072 minus0073 minus0073 minus0092 minus0074 minus0078 minus0057

0

100

200

300

400

500

600

700

Mod

ules

minus005 005 01 015 02 025 03 0350Weighted interaction density of modules

NormalccRCC

(a)

0 005 01 015 02 025 03minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(b)

0

100

200

300

400

500

600

700

Mod

ules

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

NormalccRCC

(c)

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(d)

Figure 2 Weighted interaction density distribution of modules in normal and ccRCC cases (a) represents stage I of ccRCC (b) representsstage II (c) represents stage III and (d) represents stage IV

6 Computational and Mathematical Methods in Medicine

Table 2 Overall conditions of changed module correlation densityof ccRCC stages

ccRCC stages Changed module correlation densityMaximum Average Minimum

I 0255 0015 minus0195II 0254 0028 minus0192III 0253 0012 minus0246IV 0155 minus0006 minus0240

Stage IStage II

Stage IIIStage IV

0

200

400

600

800

1000

Mod

ule p

airs

minus005minus01minus015minus02 005 01 015 02 025 030Changed module correlation density

Figure 3 Module correlation density distributions of stage I stageII stage III and stage IV

Next we obtained disrupted module pairs (ccRCC mod-ule and its relative normal module) based on modeling theset Υ(119878 119879) as maximum weight bipartite matching and thencalculated their PCC difference values (also called changedmodule correlation density value) With the thresholds 119905

119869=

23 and 120575 = 005 the overall conditions of changed modulecorrelation density of stages I II III and IV in ccRCC hadno significant difference (119875 gt 005 Table 2) An overalldecrease in maximum correlations of ccRCC modules withdeepened stage was observed besides minimum correlationdensity of stage III was the smallest among the four stages Inaddition changed module correlation density distributionswere shown in Figure 3 and the number of modules wasdifferent in the same density interval of four stages especiallyin the distribution interval of minus005sim020 For stage IVmodule distributions firstly increased and then decreasedwith density increase the maximum was reached at sectionof 0sim005

33 In-Depth Analyses of Disrupted Modules When restrict-ing random inspection correction of modules under condi-tion of 119875 lt 001 we obtained 136 576 693 and 531 disruptedmodules of stages I II III and IV respectively Meanwhile atotal of 1026 genes were obtained of these disrupted modulesin detail 317 genes of stage I 450 genes of stage II 658 genes

Table 3 Common genes of disruptedmodules based on four ccRCCstages

Number Genes1 MAPK12 CDC63 CDKN1A4 SF3B65 CPSF36 SRSF67 SRSF18 U2AF19 SRSF410 CCNB111 ESPL112 NCAPH13 KIF1114 BUB1B15 CDC2016 CCNA217 CCNB218 MAD2L119 CENPF20 ALDH4A121 NCBP122 MGST123 GSTZ124 GSTM225 GSTM526 GSTM327 GSTA328 SNRPD329 CDKN1B30 NDUFAB131 RNPS132 ALB33 LPA34 GOT135 GLUD136 FTCD37 GLUD238 ALDH3A239 ALDH9A140 ALDH1B141 ALDH7A142 NAGS43 GATM44 ASNS45 ACY346 ASPA47 GOT248 ASS149 GAD250 ALDH251 MGST252 MGST353 GSTO154 GSTM455 RPA356 UPF3B

of stage III and 690 genes of stage IVTherefore 56 commongenes existing in four stages were explored (Table 3) such asMAPK1 CCNA2 and GSTM3

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

2 Computational and Mathematical Methods in Medicine

of complex discovery algorithms [11] Many computationalapproaches have been proposed to assess the reliabilityof protein interactions data An iterative scoring methodproposed by Liu et al [12] was selected to evaluate thereliability and predict new interactions and it has been shownto perform better than other methods However studyingmultiple diseases simultaneously makes it challenging todiscern clearly the intricate underlying mechanisms

In addition it is important to effectively integrate omicsdata into such an analysis for example Chu and Chen[13] combined PPI and gene expression data to construct acancer perturbed PPI network in cervical carcinoma to studygain- and loss-of-function genes as potential drug targetsMagger et al [14] combined PPI and gene expression datato construct tissue-specific PPI networks for 60 tissues andused them to prioritize disease genes Beyond straightforwardscoring genes in the gene regulatory network it is crucial tostudy the behavior of modules across specific conditions ina controlled manner to understand the modus operandi ofdisease mechanisms and to implicate novel genes [15] sincesome of the important genes may not be identifiable throughtheir own behavior but their changes are quantifiable whenconsidered in conjunctionwith other genes (eg asmodules)What is required therefore is systematic tracking genepathways and module behavior across specific conditions ina controlled manner

Therefore in this paper we performed a temporal (stagesI II III and IV of ccRCC) analysis between normal controlsand ccRCC patients to identify disrupted genes and pathwaysby systematically tracking the altered modules of reweightedPPI network To achieve this we firstly inferred normaland ccRCC cases of different stages PPI networks based onPearson correlation coefficient (PCC) next clique-mergingalgorithmwas performed to exploremodules in PPI networkand we compared these modules to identify altered modulesthen gene composition of thesemoduleswas analyzed finallypathways enrichment analysis of genes in altered moduleswas carried out based on Expression Analysis SystematicExplored (EASE) test

2 Materials and Methods

21 Inferring Normal and ccRCC PPI Network

211 PPI Network Construction We utilized a dataset ofhuman PPI network the Search Tool for the Retrievalof Interacting GenesProteins (STRING) which comprised16730 genes and 1048576 interactions [16] For STRING self-loops and proteins without expression value were removedThe remaining largest connected component with score ofmore than 08 was kept as the selected PPI network whichconsisted of 8590 genes and 53975 interactions

212 Gene Expression Dataset and Dataset Preprocessing Amicroarray expression profile E-GEOD-53757 from ArrayExpress database was selected for ccRCC related analysis E-GEOD-53757which existed onAffymetrixGeneChipHumanGenome U133 Plus 20 Platform was divided into 4 groupsaccording to tumor stage (stages I II III and IV)There were

24 19 14 and 15 ccRCC patients at stages I II III and IVrespectively the number of normal controls in each stage wasequaled to its patientsrsquo number

The expression profile was preprocessed by standardmethods consisting of ldquormardquo [17] ldquoquantilesrdquo [18] ldquomasrdquo[19] and ldquomedianpolishrdquo [17] To be specific ldquormardquo methodwas carried out for background correction to eliminateinfluences of nonspecific hybridization [17] The quantilenormalization algorithm was a specific case of the trans-formation 119909

1015840

119894= 119865minus1

(119866(119909119894)) where we estimated 119866 by the

empirical distribution of each array and119865 using the empiricaldistribution of the averaged sample quantiles [18] Perfectmatch (PM)mismatch (MM) correction was conducted byldquomasrdquo method [19] Summarization of the probe data wasconducted by ldquomedianpolishrdquo [17] A multichip linear modelwas fit to data from each probe set In particular for a probeset 119896 with 119894 = 1 119868

119896probes and data from 119895 = 1 119869

arrays we fitted the following model log2(PM119896119894119895) = 120572119896

119894+ 120573119896

119895+

120576119896

119894119895 where 120572

119894was a probe effect and 120573

119895was the log

2expression

valueNext the data were screened by feature filter method of

gene filter package and the number of genes with multipleprobes was 20102 At last we obtained the gene expressionvalue for each gene including 20102 genes from 144 samples(72 normal controls and 72 ccRCC patients)

213 Reweighting Gene Interactions by PCC Gene interac-tions in network based on ccRCC patients of different stages(stages I II III and IV) and their normal controls werereweighted by PCC which evaluated the probability of twocoexpressed gene pairs PCC is a measure of the correlationbetween two variables giving a value between minus1 and +1inclusively [20] The PCC of a pair of genes (119909 and 119910)which encoded the corresponding paired proteins (119906 and V)interacting in the PPI network was defined as

PCC (119909 119910)

=

1

119904 minus 1

119904

sum

119894=1

(

119892 (119909 119894) minus 119892 (119909)

120590 (119909)

) sdot (

119892 (119910 119894) minus 119892 (119910)

120590 (119910)

)

(1)

where 119904 was the number of samples of the gene expressiondata 119892(119909 119894) or 119892(119910 119894) was the expression level of gene 119909 or119910 in the sample 119894 under a specific condition 119892(119909) or 119892(119910)

represented themean expression level of gene119909 or119910 and119892(119909)

or 119892(119910) represented the standard deviation of expression levelof gene 119909 (or 119910)

The PCC of a pair of proteins (119906 and V) was defined asthe same as the PCC of their corresponding paired genes (119909and 119910) which was PCC(119906 V) = PCC(119909 119910) If PCC(119906 V) has apositive value there is a positive linear correlation between119906 and V In addition we defined PCC of each gene-geneinteraction as weight value of the interaction

22 Identifying Modules from the PPI Network In this papermodule-identification algorithm is based on clique-merging[21 22] and is similar to the method proposed by Liu et al[12] It consisted of three steps in the first step it found

Computational and Mathematical Methods in Medicine 3

all the maximal cliques from the weighted PPI networkMaximal cliques were evaluated by a fast depth-first searchwith pruning-based algorithm proposed by Tomita et al[23] It utilized a depth-first search strategy to enumerate allmaximal cliques and effectively pruned nonmaximal cliquesduring the enumeration process

In the second step we assigned a score to each clique thescore of a clique119862was defined as its weighted density 119889

119882(119862)

119889119882

(119862) =

sum119906isin119862Visin119862119908 (119906 V)|119862| sdot (|119862| minus 1)

(2)

where119908(119906 V)was theweight of the interaction between 119906 andV We ranked these cliques in nonincreasing order of theirweighted densities 119862

1 1198622 119862

119896

Finally we went through this ordered list repeatedlymerging highly overlapping cliques to build modules Forevery clique 119862

119894 we repeatedly looked for a clique 119862

119895(119895 gt 119894)

such that the overlap |119862119894cap 119862119895||119862119895| ge 119905 where 119905 = 05

was a predefined threshold for overlapping [15] If such 119862119895

was found we calculated the weighted interconnectivity 119868119908

between 119862119894and 119862

119895as follows

119868119908

(119862119894 119862119895)

= radic

sum119906isin(119862119894minus119862119895)

sumVisin119862119895 119908 (119906 V)10038161003816100381610038161003816119862119894minus 119862119895

10038161003816100381610038161003816sdot

10038161003816100381610038161003816119862119895

10038161003816100381610038161003816

sdot

sum119906isin(119862119895minus119862119894)

sumVisin119862119894 119908 (119906 V)10038161003816100381610038161003816119862119895minus 119862119894

10038161003816100381610038161003816sdot1003816100381610038161003816119862119894

1003816100381610038161003816

(3)

If 119868119908(119862119894 119862119895) ge 119905 then 119862

119895was merged into 119862

119894forming a

module else 119862119895was discarded

We captured the effect of differences in interactionweights between normal and ccRCC cases through theweighted density-based ranking of cliques Weighted densityassigned higher rank to larger and stronger cliquesThereforewe expected cliques with lost proteins or weakened interac-tions to go down the rankings resulting in altered modulegeneration thereby capturing changes in modules betweennormal and ccRCC cases

23 Comparing Modules between Normal and ccRCC Con-ditions The approach to compare modules between normaland ccRCC conditions is similar to the method proposed bySrihari and Ragan [15] In detail 119867

119873and 119867

119879represented

the PPI network of normal controls and ccRCC patientsidentifying the sets of modules 119878 = 119878

1 1198782 119878

119898 and

119879 = 1198791 1198792 119879

119899 respectively For each 119878

119894isin 119878 module

correlation density 119889119888(119878119894) was defined as

119889119888(119878119894) =

sum119909119910isin119878119894

PCC ((119909 119910) 119872)

1003816100381610038161003816119878119894

1003816100381610038161003816sdot (

1003816100381610038161003816119878119894

1003816100381610038161003816minus 1)

(4)

Correlation densities of ccRCC modules (119889119888(119879119894)) were

calculated similarlyDisrupted or altered module pairs were evaluated by

modeling the set Υ(119878 119879) as maximum weight bipartitematching [24] Firstly we build a similarity graph 119872 = (119881

119872

119864119872

) where 119881119872

= 119878 cup 119879 and 119864119872

= cup(119878119894 119879119895) 119869(119878

119894 119879119895) ge

119905119869 Δ119862(119878119894 119879119895) ge 120575 whereby 119869(119878

119894 119879119895) = |119878

119894cap 119879119895||119878119894cup 119879119895| was

the Jaccard similarity andΔ119862(119878119894 119879119895) = |119889

119888(119878119894)minus119889119888(119879119894)|was the

differential correlation density between 119878119894and 119879

119895 and 119905

119869and

120575 were thresholds with 23 and 005 [15] 119869(119878119894 119879119895) weighted

every edge (119878119894 119879119895) We next identified the disrupted module

pairs Υ(119878 119879) by detecting the maximum weight matchingin 119872 and we ranked them in nonincreasing order of theirdifferential density Δ

119862 At last we inferred genes involved

in ccRCC as Γ = 119892119892 isin 119878119894

cup 119879119895 (119878119894 119879119895) isin Υ(119878 119879)

and ranked in nonincreasing order of Δ119862(119878119894 119879119895) To identify

altered modules we matched normal and ccRCCmodules bysetting high 119905

119869 which ensured that the module pairs either

had the same gene composition or had lost or gained only afew genes

24 Functional Enrichment Analysis To further investigatethe biological functional pathways of genes in altered mod-ules fromnormal controls and ccRCC Kyoto Encyclopedia ofGenes and Genomes (KEGG) pathway enrichment analysiswas performed by Database for Annotation Visualizationand Integrated Discovery (DAVID) [25] KEGG pathwayswith 119875 value lt 0001 were selected based on EASE testimplemented in DAVID EASE analysis of the regulatedgenes indicated molecular functions and biological processesunique to each category [26] The EASE score was used todetect the significant categories In both of the functionaland pathway enrichment analyses the threshold of theminimum number of genes for the corresponding term gt 2was considered significant for a category

119901 =

(119886+119887

119886) (119888+119889

119888)

(119899

119886+119888 )

(5)

where 119899was the number of background genes 1198861015840 was the genenumber of one gene set in the gene lists 1198861015840+119887was the numberof genes in the gene list including at least one gene set 1198861015840 + 119888

was the gene number of one gene list in the background genes1198861015840 was replaced with 119886 = 119886

1015840minus 1

3 Results

31 Analyzing Disruptions in ccRCC PPI Network Weobtained 20102 genes of normal and ccRCC cases afterpreprocessing and then investigated intersections betweenthese genesrsquo interactions and STRING PPI network andidentified PPI networks of normal and ccRCC cases Thenormal 119867

119873and ccRCC 119867

119879PPI networks of different stages

(stages I II III and IV) displayed equal numbers of nodes(8050) and interactions (49151) Although their interactionscores (weights) were different from each other as shownin Figure 1 there was no statistical significance betweennormal and ccRCC cases in different stages in whole levelbased on Kolmogorov-Smirnov test (119875 gt 005) Howeverthe score distribution between the ccRCC networks andnormal networks was different especially for stages III andIV in the score distribution 0sim03 (Figures 1(c) and 1(d))Examining these interactions more carefully distributionsamong different stages were also different and changes ofccRCC networks and normal networks were more and moreobvious from stage I to stage IV

4 Computational and Mathematical Methods in Medicine

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(a)

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(b)

NormalccRCC

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

(c)

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(d)

Figure 1 Score-wise distributions of interactions normal versus ccRCC cases (a) represents stage I of ccRCC (b) represents stage II (c)represents stage III and (d) represents stage IV

32 Analyzing Disruptions in ccRCC Modules Clique-merging algorithm was selected to identify disrupted oraltered modules from normal and ccRCC PPI network inthis paper In detail we performed a comparative analysisbetween normal 119873 and ccRCC 119879 modules to understanddisruptions at the module level Maximal cliques of normaland ccRCC PPI network were obtained based on fast depth-first algorithm and maximal cliques with the threshold ofnodes gt 5 were selected for module analysis Overall wenoticed that the total number of modules (1895) as well asaverage module sizes (20235) was almost the same acrossthe two conditions and four stages Table 1 showed overallchanged rules of weighted interaction density betweennormal modules and ccRCC modules we could find that

maximal and average weighted density of normal case wassmaller than that of ccRCC for each stage in detail theaverage weighted density of stages III (0075) and IV (0089)was a little higher than that of stages I (0068) and II (0046)while in the overall level the difference of module densityscores had no statistical significance between normal andccRCC cases in different stages with 119875 gt 005 Further therelationship between modules weighted density distributionand numbers of modules was illustrated in Figure 2 Themodule numbers were different when the interaction densityranged from 005 to 025 especially for stages II III andIV of ccRCC These differences might be the reasons ofweighted density changes of ccRCC from different stages(Table 1)

Computational and Mathematical Methods in Medicine 5

Table 1 Correlations of normal and ccRCC modules of different stages

Module set Stage I Stage II Stage III Stage IVNormal ccRCC Normal ccRCC Normal ccRCC Normal ccRCC

PCC correlationMaximal 0315 0324 0254 0278 0324 0339 0294 0326Average 0068 0083 0046 0074 0075 0087 0089 0084Minimum minus0076 minus0072 minus0073 minus0073 minus0092 minus0074 minus0078 minus0057

0

100

200

300

400

500

600

700

Mod

ules

minus005 005 01 015 02 025 03 0350Weighted interaction density of modules

NormalccRCC

(a)

0 005 01 015 02 025 03minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(b)

0

100

200

300

400

500

600

700

Mod

ules

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

NormalccRCC

(c)

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(d)

Figure 2 Weighted interaction density distribution of modules in normal and ccRCC cases (a) represents stage I of ccRCC (b) representsstage II (c) represents stage III and (d) represents stage IV

6 Computational and Mathematical Methods in Medicine

Table 2 Overall conditions of changed module correlation densityof ccRCC stages

ccRCC stages Changed module correlation densityMaximum Average Minimum

I 0255 0015 minus0195II 0254 0028 minus0192III 0253 0012 minus0246IV 0155 minus0006 minus0240

Stage IStage II

Stage IIIStage IV

0

200

400

600

800

1000

Mod

ule p

airs

minus005minus01minus015minus02 005 01 015 02 025 030Changed module correlation density

Figure 3 Module correlation density distributions of stage I stageII stage III and stage IV

Next we obtained disrupted module pairs (ccRCC mod-ule and its relative normal module) based on modeling theset Υ(119878 119879) as maximum weight bipartite matching and thencalculated their PCC difference values (also called changedmodule correlation density value) With the thresholds 119905

119869=

23 and 120575 = 005 the overall conditions of changed modulecorrelation density of stages I II III and IV in ccRCC hadno significant difference (119875 gt 005 Table 2) An overalldecrease in maximum correlations of ccRCC modules withdeepened stage was observed besides minimum correlationdensity of stage III was the smallest among the four stages Inaddition changed module correlation density distributionswere shown in Figure 3 and the number of modules wasdifferent in the same density interval of four stages especiallyin the distribution interval of minus005sim020 For stage IVmodule distributions firstly increased and then decreasedwith density increase the maximum was reached at sectionof 0sim005

33 In-Depth Analyses of Disrupted Modules When restrict-ing random inspection correction of modules under condi-tion of 119875 lt 001 we obtained 136 576 693 and 531 disruptedmodules of stages I II III and IV respectively Meanwhile atotal of 1026 genes were obtained of these disrupted modulesin detail 317 genes of stage I 450 genes of stage II 658 genes

Table 3 Common genes of disruptedmodules based on four ccRCCstages

Number Genes1 MAPK12 CDC63 CDKN1A4 SF3B65 CPSF36 SRSF67 SRSF18 U2AF19 SRSF410 CCNB111 ESPL112 NCAPH13 KIF1114 BUB1B15 CDC2016 CCNA217 CCNB218 MAD2L119 CENPF20 ALDH4A121 NCBP122 MGST123 GSTZ124 GSTM225 GSTM526 GSTM327 GSTA328 SNRPD329 CDKN1B30 NDUFAB131 RNPS132 ALB33 LPA34 GOT135 GLUD136 FTCD37 GLUD238 ALDH3A239 ALDH9A140 ALDH1B141 ALDH7A142 NAGS43 GATM44 ASNS45 ACY346 ASPA47 GOT248 ASS149 GAD250 ALDH251 MGST252 MGST353 GSTO154 GSTM455 RPA356 UPF3B

of stage III and 690 genes of stage IVTherefore 56 commongenes existing in four stages were explored (Table 3) such asMAPK1 CCNA2 and GSTM3

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Computational and Mathematical Methods in Medicine 3

all the maximal cliques from the weighted PPI networkMaximal cliques were evaluated by a fast depth-first searchwith pruning-based algorithm proposed by Tomita et al[23] It utilized a depth-first search strategy to enumerate allmaximal cliques and effectively pruned nonmaximal cliquesduring the enumeration process

In the second step we assigned a score to each clique thescore of a clique119862was defined as its weighted density 119889

119882(119862)

119889119882

(119862) =

sum119906isin119862Visin119862119908 (119906 V)|119862| sdot (|119862| minus 1)

(2)

where119908(119906 V)was theweight of the interaction between 119906 andV We ranked these cliques in nonincreasing order of theirweighted densities 119862

1 1198622 119862

119896

Finally we went through this ordered list repeatedlymerging highly overlapping cliques to build modules Forevery clique 119862

119894 we repeatedly looked for a clique 119862

119895(119895 gt 119894)

such that the overlap |119862119894cap 119862119895||119862119895| ge 119905 where 119905 = 05

was a predefined threshold for overlapping [15] If such 119862119895

was found we calculated the weighted interconnectivity 119868119908

between 119862119894and 119862

119895as follows

119868119908

(119862119894 119862119895)

= radic

sum119906isin(119862119894minus119862119895)

sumVisin119862119895 119908 (119906 V)10038161003816100381610038161003816119862119894minus 119862119895

10038161003816100381610038161003816sdot

10038161003816100381610038161003816119862119895

10038161003816100381610038161003816

sdot

sum119906isin(119862119895minus119862119894)

sumVisin119862119894 119908 (119906 V)10038161003816100381610038161003816119862119895minus 119862119894

10038161003816100381610038161003816sdot1003816100381610038161003816119862119894

1003816100381610038161003816

(3)

If 119868119908(119862119894 119862119895) ge 119905 then 119862

119895was merged into 119862

119894forming a

module else 119862119895was discarded

We captured the effect of differences in interactionweights between normal and ccRCC cases through theweighted density-based ranking of cliques Weighted densityassigned higher rank to larger and stronger cliquesThereforewe expected cliques with lost proteins or weakened interac-tions to go down the rankings resulting in altered modulegeneration thereby capturing changes in modules betweennormal and ccRCC cases

23 Comparing Modules between Normal and ccRCC Con-ditions The approach to compare modules between normaland ccRCC conditions is similar to the method proposed bySrihari and Ragan [15] In detail 119867

119873and 119867

119879represented

the PPI network of normal controls and ccRCC patientsidentifying the sets of modules 119878 = 119878

1 1198782 119878

119898 and

119879 = 1198791 1198792 119879

119899 respectively For each 119878

119894isin 119878 module

correlation density 119889119888(119878119894) was defined as

119889119888(119878119894) =

sum119909119910isin119878119894

PCC ((119909 119910) 119872)

1003816100381610038161003816119878119894

1003816100381610038161003816sdot (

1003816100381610038161003816119878119894

1003816100381610038161003816minus 1)

(4)

Correlation densities of ccRCC modules (119889119888(119879119894)) were

calculated similarlyDisrupted or altered module pairs were evaluated by

modeling the set Υ(119878 119879) as maximum weight bipartitematching [24] Firstly we build a similarity graph 119872 = (119881

119872

119864119872

) where 119881119872

= 119878 cup 119879 and 119864119872

= cup(119878119894 119879119895) 119869(119878

119894 119879119895) ge

119905119869 Δ119862(119878119894 119879119895) ge 120575 whereby 119869(119878

119894 119879119895) = |119878

119894cap 119879119895||119878119894cup 119879119895| was

the Jaccard similarity andΔ119862(119878119894 119879119895) = |119889

119888(119878119894)minus119889119888(119879119894)|was the

differential correlation density between 119878119894and 119879

119895 and 119905

119869and

120575 were thresholds with 23 and 005 [15] 119869(119878119894 119879119895) weighted

every edge (119878119894 119879119895) We next identified the disrupted module

pairs Υ(119878 119879) by detecting the maximum weight matchingin 119872 and we ranked them in nonincreasing order of theirdifferential density Δ

119862 At last we inferred genes involved

in ccRCC as Γ = 119892119892 isin 119878119894

cup 119879119895 (119878119894 119879119895) isin Υ(119878 119879)

and ranked in nonincreasing order of Δ119862(119878119894 119879119895) To identify

altered modules we matched normal and ccRCCmodules bysetting high 119905

119869 which ensured that the module pairs either

had the same gene composition or had lost or gained only afew genes

24 Functional Enrichment Analysis To further investigatethe biological functional pathways of genes in altered mod-ules fromnormal controls and ccRCC Kyoto Encyclopedia ofGenes and Genomes (KEGG) pathway enrichment analysiswas performed by Database for Annotation Visualizationand Integrated Discovery (DAVID) [25] KEGG pathwayswith 119875 value lt 0001 were selected based on EASE testimplemented in DAVID EASE analysis of the regulatedgenes indicated molecular functions and biological processesunique to each category [26] The EASE score was used todetect the significant categories In both of the functionaland pathway enrichment analyses the threshold of theminimum number of genes for the corresponding term gt 2was considered significant for a category

119901 =

(119886+119887

119886) (119888+119889

119888)

(119899

119886+119888 )

(5)

where 119899was the number of background genes 1198861015840 was the genenumber of one gene set in the gene lists 1198861015840+119887was the numberof genes in the gene list including at least one gene set 1198861015840 + 119888

was the gene number of one gene list in the background genes1198861015840 was replaced with 119886 = 119886

1015840minus 1

3 Results

31 Analyzing Disruptions in ccRCC PPI Network Weobtained 20102 genes of normal and ccRCC cases afterpreprocessing and then investigated intersections betweenthese genesrsquo interactions and STRING PPI network andidentified PPI networks of normal and ccRCC cases Thenormal 119867

119873and ccRCC 119867

119879PPI networks of different stages

(stages I II III and IV) displayed equal numbers of nodes(8050) and interactions (49151) Although their interactionscores (weights) were different from each other as shownin Figure 1 there was no statistical significance betweennormal and ccRCC cases in different stages in whole levelbased on Kolmogorov-Smirnov test (119875 gt 005) Howeverthe score distribution between the ccRCC networks andnormal networks was different especially for stages III andIV in the score distribution 0sim03 (Figures 1(c) and 1(d))Examining these interactions more carefully distributionsamong different stages were also different and changes ofccRCC networks and normal networks were more and moreobvious from stage I to stage IV

4 Computational and Mathematical Methods in Medicine

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(a)

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(b)

NormalccRCC

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

(c)

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(d)

Figure 1 Score-wise distributions of interactions normal versus ccRCC cases (a) represents stage I of ccRCC (b) represents stage II (c)represents stage III and (d) represents stage IV

32 Analyzing Disruptions in ccRCC Modules Clique-merging algorithm was selected to identify disrupted oraltered modules from normal and ccRCC PPI network inthis paper In detail we performed a comparative analysisbetween normal 119873 and ccRCC 119879 modules to understanddisruptions at the module level Maximal cliques of normaland ccRCC PPI network were obtained based on fast depth-first algorithm and maximal cliques with the threshold ofnodes gt 5 were selected for module analysis Overall wenoticed that the total number of modules (1895) as well asaverage module sizes (20235) was almost the same acrossthe two conditions and four stages Table 1 showed overallchanged rules of weighted interaction density betweennormal modules and ccRCC modules we could find that

maximal and average weighted density of normal case wassmaller than that of ccRCC for each stage in detail theaverage weighted density of stages III (0075) and IV (0089)was a little higher than that of stages I (0068) and II (0046)while in the overall level the difference of module densityscores had no statistical significance between normal andccRCC cases in different stages with 119875 gt 005 Further therelationship between modules weighted density distributionand numbers of modules was illustrated in Figure 2 Themodule numbers were different when the interaction densityranged from 005 to 025 especially for stages II III andIV of ccRCC These differences might be the reasons ofweighted density changes of ccRCC from different stages(Table 1)

Computational and Mathematical Methods in Medicine 5

Table 1 Correlations of normal and ccRCC modules of different stages

Module set Stage I Stage II Stage III Stage IVNormal ccRCC Normal ccRCC Normal ccRCC Normal ccRCC

PCC correlationMaximal 0315 0324 0254 0278 0324 0339 0294 0326Average 0068 0083 0046 0074 0075 0087 0089 0084Minimum minus0076 minus0072 minus0073 minus0073 minus0092 minus0074 minus0078 minus0057

0

100

200

300

400

500

600

700

Mod

ules

minus005 005 01 015 02 025 03 0350Weighted interaction density of modules

NormalccRCC

(a)

0 005 01 015 02 025 03minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(b)

0

100

200

300

400

500

600

700

Mod

ules

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

NormalccRCC

(c)

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(d)

Figure 2 Weighted interaction density distribution of modules in normal and ccRCC cases (a) represents stage I of ccRCC (b) representsstage II (c) represents stage III and (d) represents stage IV

6 Computational and Mathematical Methods in Medicine

Table 2 Overall conditions of changed module correlation densityof ccRCC stages

ccRCC stages Changed module correlation densityMaximum Average Minimum

I 0255 0015 minus0195II 0254 0028 minus0192III 0253 0012 minus0246IV 0155 minus0006 minus0240

Stage IStage II

Stage IIIStage IV

0

200

400

600

800

1000

Mod

ule p

airs

minus005minus01minus015minus02 005 01 015 02 025 030Changed module correlation density

Figure 3 Module correlation density distributions of stage I stageII stage III and stage IV

Next we obtained disrupted module pairs (ccRCC mod-ule and its relative normal module) based on modeling theset Υ(119878 119879) as maximum weight bipartite matching and thencalculated their PCC difference values (also called changedmodule correlation density value) With the thresholds 119905

119869=

23 and 120575 = 005 the overall conditions of changed modulecorrelation density of stages I II III and IV in ccRCC hadno significant difference (119875 gt 005 Table 2) An overalldecrease in maximum correlations of ccRCC modules withdeepened stage was observed besides minimum correlationdensity of stage III was the smallest among the four stages Inaddition changed module correlation density distributionswere shown in Figure 3 and the number of modules wasdifferent in the same density interval of four stages especiallyin the distribution interval of minus005sim020 For stage IVmodule distributions firstly increased and then decreasedwith density increase the maximum was reached at sectionof 0sim005

33 In-Depth Analyses of Disrupted Modules When restrict-ing random inspection correction of modules under condi-tion of 119875 lt 001 we obtained 136 576 693 and 531 disruptedmodules of stages I II III and IV respectively Meanwhile atotal of 1026 genes were obtained of these disrupted modulesin detail 317 genes of stage I 450 genes of stage II 658 genes

Table 3 Common genes of disruptedmodules based on four ccRCCstages

Number Genes1 MAPK12 CDC63 CDKN1A4 SF3B65 CPSF36 SRSF67 SRSF18 U2AF19 SRSF410 CCNB111 ESPL112 NCAPH13 KIF1114 BUB1B15 CDC2016 CCNA217 CCNB218 MAD2L119 CENPF20 ALDH4A121 NCBP122 MGST123 GSTZ124 GSTM225 GSTM526 GSTM327 GSTA328 SNRPD329 CDKN1B30 NDUFAB131 RNPS132 ALB33 LPA34 GOT135 GLUD136 FTCD37 GLUD238 ALDH3A239 ALDH9A140 ALDH1B141 ALDH7A142 NAGS43 GATM44 ASNS45 ACY346 ASPA47 GOT248 ASS149 GAD250 ALDH251 MGST252 MGST353 GSTO154 GSTM455 RPA356 UPF3B

of stage III and 690 genes of stage IVTherefore 56 commongenes existing in four stages were explored (Table 3) such asMAPK1 CCNA2 and GSTM3

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

4 Computational and Mathematical Methods in Medicine

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(a)

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(b)

NormalccRCC

0

2000

4000

6000

8000

10000

12000

Inte

ract

ions

02 03 04 05 06 07 08 09 101Interaction scores

(c)

0

2000

4000

6000

8000

10000

12000In

tera

ctio

ns

02 03 04 05 06 07 08 09 101Interaction scores

NormalccRCC

(d)

Figure 1 Score-wise distributions of interactions normal versus ccRCC cases (a) represents stage I of ccRCC (b) represents stage II (c)represents stage III and (d) represents stage IV

32 Analyzing Disruptions in ccRCC Modules Clique-merging algorithm was selected to identify disrupted oraltered modules from normal and ccRCC PPI network inthis paper In detail we performed a comparative analysisbetween normal 119873 and ccRCC 119879 modules to understanddisruptions at the module level Maximal cliques of normaland ccRCC PPI network were obtained based on fast depth-first algorithm and maximal cliques with the threshold ofnodes gt 5 were selected for module analysis Overall wenoticed that the total number of modules (1895) as well asaverage module sizes (20235) was almost the same acrossthe two conditions and four stages Table 1 showed overallchanged rules of weighted interaction density betweennormal modules and ccRCC modules we could find that

maximal and average weighted density of normal case wassmaller than that of ccRCC for each stage in detail theaverage weighted density of stages III (0075) and IV (0089)was a little higher than that of stages I (0068) and II (0046)while in the overall level the difference of module densityscores had no statistical significance between normal andccRCC cases in different stages with 119875 gt 005 Further therelationship between modules weighted density distributionand numbers of modules was illustrated in Figure 2 Themodule numbers were different when the interaction densityranged from 005 to 025 especially for stages II III andIV of ccRCC These differences might be the reasons ofweighted density changes of ccRCC from different stages(Table 1)

Computational and Mathematical Methods in Medicine 5

Table 1 Correlations of normal and ccRCC modules of different stages

Module set Stage I Stage II Stage III Stage IVNormal ccRCC Normal ccRCC Normal ccRCC Normal ccRCC

PCC correlationMaximal 0315 0324 0254 0278 0324 0339 0294 0326Average 0068 0083 0046 0074 0075 0087 0089 0084Minimum minus0076 minus0072 minus0073 minus0073 minus0092 minus0074 minus0078 minus0057

0

100

200

300

400

500

600

700

Mod

ules

minus005 005 01 015 02 025 03 0350Weighted interaction density of modules

NormalccRCC

(a)

0 005 01 015 02 025 03minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(b)

0

100

200

300

400

500

600

700

Mod

ules

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

NormalccRCC

(c)

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(d)

Figure 2 Weighted interaction density distribution of modules in normal and ccRCC cases (a) represents stage I of ccRCC (b) representsstage II (c) represents stage III and (d) represents stage IV

6 Computational and Mathematical Methods in Medicine

Table 2 Overall conditions of changed module correlation densityof ccRCC stages

ccRCC stages Changed module correlation densityMaximum Average Minimum

I 0255 0015 minus0195II 0254 0028 minus0192III 0253 0012 minus0246IV 0155 minus0006 minus0240

Stage IStage II

Stage IIIStage IV

0

200

400

600

800

1000

Mod

ule p

airs

minus005minus01minus015minus02 005 01 015 02 025 030Changed module correlation density

Figure 3 Module correlation density distributions of stage I stageII stage III and stage IV

Next we obtained disrupted module pairs (ccRCC mod-ule and its relative normal module) based on modeling theset Υ(119878 119879) as maximum weight bipartite matching and thencalculated their PCC difference values (also called changedmodule correlation density value) With the thresholds 119905

119869=

23 and 120575 = 005 the overall conditions of changed modulecorrelation density of stages I II III and IV in ccRCC hadno significant difference (119875 gt 005 Table 2) An overalldecrease in maximum correlations of ccRCC modules withdeepened stage was observed besides minimum correlationdensity of stage III was the smallest among the four stages Inaddition changed module correlation density distributionswere shown in Figure 3 and the number of modules wasdifferent in the same density interval of four stages especiallyin the distribution interval of minus005sim020 For stage IVmodule distributions firstly increased and then decreasedwith density increase the maximum was reached at sectionof 0sim005

33 In-Depth Analyses of Disrupted Modules When restrict-ing random inspection correction of modules under condi-tion of 119875 lt 001 we obtained 136 576 693 and 531 disruptedmodules of stages I II III and IV respectively Meanwhile atotal of 1026 genes were obtained of these disrupted modulesin detail 317 genes of stage I 450 genes of stage II 658 genes

Table 3 Common genes of disruptedmodules based on four ccRCCstages

Number Genes1 MAPK12 CDC63 CDKN1A4 SF3B65 CPSF36 SRSF67 SRSF18 U2AF19 SRSF410 CCNB111 ESPL112 NCAPH13 KIF1114 BUB1B15 CDC2016 CCNA217 CCNB218 MAD2L119 CENPF20 ALDH4A121 NCBP122 MGST123 GSTZ124 GSTM225 GSTM526 GSTM327 GSTA328 SNRPD329 CDKN1B30 NDUFAB131 RNPS132 ALB33 LPA34 GOT135 GLUD136 FTCD37 GLUD238 ALDH3A239 ALDH9A140 ALDH1B141 ALDH7A142 NAGS43 GATM44 ASNS45 ACY346 ASPA47 GOT248 ASS149 GAD250 ALDH251 MGST252 MGST353 GSTO154 GSTM455 RPA356 UPF3B

of stage III and 690 genes of stage IVTherefore 56 commongenes existing in four stages were explored (Table 3) such asMAPK1 CCNA2 and GSTM3

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Computational and Mathematical Methods in Medicine 5

Table 1 Correlations of normal and ccRCC modules of different stages

Module set Stage I Stage II Stage III Stage IVNormal ccRCC Normal ccRCC Normal ccRCC Normal ccRCC

PCC correlationMaximal 0315 0324 0254 0278 0324 0339 0294 0326Average 0068 0083 0046 0074 0075 0087 0089 0084Minimum minus0076 minus0072 minus0073 minus0073 minus0092 minus0074 minus0078 minus0057

0

100

200

300

400

500

600

700

Mod

ules

minus005 005 01 015 02 025 03 0350Weighted interaction density of modules

NormalccRCC

(a)

0 005 01 015 02 025 03minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(b)

0

100

200

300

400

500

600

700

Mod

ules

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

NormalccRCC

(c)

0 005 01 015 02 025 03 035minus005

Weighted interaction density of modules

0

100

200

300

400

500

600

700

800

900

Mod

ules

NormalccRCC

(d)

Figure 2 Weighted interaction density distribution of modules in normal and ccRCC cases (a) represents stage I of ccRCC (b) representsstage II (c) represents stage III and (d) represents stage IV

6 Computational and Mathematical Methods in Medicine

Table 2 Overall conditions of changed module correlation densityof ccRCC stages

ccRCC stages Changed module correlation densityMaximum Average Minimum

I 0255 0015 minus0195II 0254 0028 minus0192III 0253 0012 minus0246IV 0155 minus0006 minus0240

Stage IStage II

Stage IIIStage IV

0

200

400

600

800

1000

Mod

ule p

airs

minus005minus01minus015minus02 005 01 015 02 025 030Changed module correlation density

Figure 3 Module correlation density distributions of stage I stageII stage III and stage IV

Next we obtained disrupted module pairs (ccRCC mod-ule and its relative normal module) based on modeling theset Υ(119878 119879) as maximum weight bipartite matching and thencalculated their PCC difference values (also called changedmodule correlation density value) With the thresholds 119905

119869=

23 and 120575 = 005 the overall conditions of changed modulecorrelation density of stages I II III and IV in ccRCC hadno significant difference (119875 gt 005 Table 2) An overalldecrease in maximum correlations of ccRCC modules withdeepened stage was observed besides minimum correlationdensity of stage III was the smallest among the four stages Inaddition changed module correlation density distributionswere shown in Figure 3 and the number of modules wasdifferent in the same density interval of four stages especiallyin the distribution interval of minus005sim020 For stage IVmodule distributions firstly increased and then decreasedwith density increase the maximum was reached at sectionof 0sim005

33 In-Depth Analyses of Disrupted Modules When restrict-ing random inspection correction of modules under condi-tion of 119875 lt 001 we obtained 136 576 693 and 531 disruptedmodules of stages I II III and IV respectively Meanwhile atotal of 1026 genes were obtained of these disrupted modulesin detail 317 genes of stage I 450 genes of stage II 658 genes

Table 3 Common genes of disruptedmodules based on four ccRCCstages

Number Genes1 MAPK12 CDC63 CDKN1A4 SF3B65 CPSF36 SRSF67 SRSF18 U2AF19 SRSF410 CCNB111 ESPL112 NCAPH13 KIF1114 BUB1B15 CDC2016 CCNA217 CCNB218 MAD2L119 CENPF20 ALDH4A121 NCBP122 MGST123 GSTZ124 GSTM225 GSTM526 GSTM327 GSTA328 SNRPD329 CDKN1B30 NDUFAB131 RNPS132 ALB33 LPA34 GOT135 GLUD136 FTCD37 GLUD238 ALDH3A239 ALDH9A140 ALDH1B141 ALDH7A142 NAGS43 GATM44 ASNS45 ACY346 ASPA47 GOT248 ASS149 GAD250 ALDH251 MGST252 MGST353 GSTO154 GSTM455 RPA356 UPF3B

of stage III and 690 genes of stage IVTherefore 56 commongenes existing in four stages were explored (Table 3) such asMAPK1 CCNA2 and GSTM3

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

6 Computational and Mathematical Methods in Medicine

Table 2 Overall conditions of changed module correlation densityof ccRCC stages

ccRCC stages Changed module correlation densityMaximum Average Minimum

I 0255 0015 minus0195II 0254 0028 minus0192III 0253 0012 minus0246IV 0155 minus0006 minus0240

Stage IStage II

Stage IIIStage IV

0

200

400

600

800

1000

Mod

ule p

airs

minus005minus01minus015minus02 005 01 015 02 025 030Changed module correlation density

Figure 3 Module correlation density distributions of stage I stageII stage III and stage IV

Next we obtained disrupted module pairs (ccRCC mod-ule and its relative normal module) based on modeling theset Υ(119878 119879) as maximum weight bipartite matching and thencalculated their PCC difference values (also called changedmodule correlation density value) With the thresholds 119905

119869=

23 and 120575 = 005 the overall conditions of changed modulecorrelation density of stages I II III and IV in ccRCC hadno significant difference (119875 gt 005 Table 2) An overalldecrease in maximum correlations of ccRCC modules withdeepened stage was observed besides minimum correlationdensity of stage III was the smallest among the four stages Inaddition changed module correlation density distributionswere shown in Figure 3 and the number of modules wasdifferent in the same density interval of four stages especiallyin the distribution interval of minus005sim020 For stage IVmodule distributions firstly increased and then decreasedwith density increase the maximum was reached at sectionof 0sim005

33 In-Depth Analyses of Disrupted Modules When restrict-ing random inspection correction of modules under condi-tion of 119875 lt 001 we obtained 136 576 693 and 531 disruptedmodules of stages I II III and IV respectively Meanwhile atotal of 1026 genes were obtained of these disrupted modulesin detail 317 genes of stage I 450 genes of stage II 658 genes

Table 3 Common genes of disruptedmodules based on four ccRCCstages

Number Genes1 MAPK12 CDC63 CDKN1A4 SF3B65 CPSF36 SRSF67 SRSF18 U2AF19 SRSF410 CCNB111 ESPL112 NCAPH13 KIF1114 BUB1B15 CDC2016 CCNA217 CCNB218 MAD2L119 CENPF20 ALDH4A121 NCBP122 MGST123 GSTZ124 GSTM225 GSTM526 GSTM327 GSTA328 SNRPD329 CDKN1B30 NDUFAB131 RNPS132 ALB33 LPA34 GOT135 GLUD136 FTCD37 GLUD238 ALDH3A239 ALDH9A140 ALDH1B141 ALDH7A142 NAGS43 GATM44 ASNS45 ACY346 ASPA47 GOT248 ASS149 GAD250 ALDH251 MGST252 MGST353 GSTO154 GSTM455 RPA356 UPF3B

of stage III and 690 genes of stage IVTherefore 56 commongenes existing in four stages were explored (Table 3) such asMAPK1 CCNA2 and GSTM3

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Computational and Mathematical Methods in Medicine 7

Cell cycleGlutathione metabolism

Drug metabolismMetabolism of xenobiotics by cytochrome P450

Arginine and proline metabolismECM-receptor interaction

DNA replicationSmall cell lung cancer

Oxidative phosphorylationRibosome

Parkinsonrsquos diseaseOocyte meiosis

Histidine metabolismAlzheimerrsquos disease

Dilated cardiomyopathy

Epithelial cell signaling in Helicobacter pylori infectionHypertrophic cardiomyopathy

Regulation of actin cytoskeletonArrhythmogenic right ventricular cardiomyopathy

Mismatch repairLinoleic acid metabolism

Neuroactive ligand-receptor interactionNucleotide excision repair

Huntingtonrsquos diseaseSpliceosome

RNA polymeraseProtein export

ProteasomeBeta-alanine metabolism

Cardiac muscle contractionFocal adhesion

Chemokine signaling pathwayProgesterone-mediated oocyte maturation

Retinol metabolismI II III IV

ExistingNot Existing

Vibrio cholerae infection

Figure 4 Distribution of pathways in stages I II III and IV Pathways were identified by KEGG with 119875 lt 0001 The light green squarerepresented the notion that one pathway did not exist in the stage while the dark one stood for the notion that the pathway existed in thestage

As we all know differentially expressed (DE) gene wasusually selected to screen significant genes between normalcontrols and disease patients thus we identified 2781 DEgenes between normal controls and ccRCC patients of fourstages based on Linear Models for Microarray Data packageTaking the intersection of common genes and DE genes intoconsideration we obtained 19 genes (ALB ASS1 GSTM3MAD2L1 ALDH1B1 ALDH4A1 MAPK1 GSTZ1 GATMFTCD CCNA2 CENPF GSTM4 ASNS CCNB1 NAGSACY3GSTA3 and ESPL1) which might play important rolesin the ccRCC development

Pathway enrichment analysis based on genes in disruptedmodules of different stages was performed and the resultswithin threshold 119875 value lt 0001 were shown in Figure 4there were 5 common pathways (glutathionemetabolism cellcycle alanine aspartate and glutamate metabolism arginineand proline metabolism and metabolism of xenobiotics bycytochrome P450) across stages I II III and IV

4 Discussions

The objective of this paper is to identify dysregulated genesand pathways in ccRCC from stage I to stage IV accord-ing to systematically tracking the dysregulated modules ofreweighted PPI networks We obtained reweighted normaland ccRCC PPI network based on PCC and then identifiedmodules in the PPI network By comparing normal andccRCC modules in each stage we obtained 136 576 693and 531 disrupted modules of stages I II III and IVrespectively Furthermore a total of 56 common genes (suchas MAPK1 and CCNA2) and 5 common pathways (egcell cycle glutathione metabolism and arginine and prolinemetabolism) of the four stages were explored based on genecomposition and pathway enrichment analysesThe commongenes and pathways from stage I to stage IV were significantfor ccRCC development if we control these signatures andthe biological progress in the early stage of the tumor theremight be positive effects on the therapy

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

8 Computational and Mathematical Methods in Medicine

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus042 108

052

minus023

08 021 067

minus008minus03

minus034

minus026

minus035 022 006

046011 107

022

minus023

(a)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 minus028

minus022

minus034minus056

015

minus016

minus031 036

07

minus048

minus065 minus037

minus034

minus032

minus062

minus026

minus001046

(b)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus018 021

109

minus031

057 029

minus029

minus003

minus013

minus084

011

minus041

minus018 minus01 018

032086

022

minus02

(c)

MAPK1

CEBPB

MAPK3

RELA

MAPK14NFKB1

RIPK2

minus046 074

minus037

029minus034

minus006

minus016

minus014

013

021

minus003

026minus016

minus063

023

minus021

014

04minus03

(d)

Figure 5 Swapping behavior in altered module (MAPK1 CEBPBMAPK3 RELAMAPK14NFKB1 and RIPK2) Nodes stood for genes andedges stood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genesin the module more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) representsstage III and (d) represents stage IV

MAPK1 (mitogen-activated protein kinase 1) whichencoded a member of theMAPK family acted as an integra-tion point for multiple biochemical signals and was involvedin a wide variety of cellular processes such as proliferationdifferentiation transcription regulation and development[27] Roberts and Der had reported that aberrant regulationof MAPK contributed to cancer and other human diseasessuch as ccRCC in particular theMAPK had been the subjectof intense research scrutiny leading to the development ofpharmacologic inhibitors for the treatment of cancer [28]Moreover MAPK participant biological processes were keysignaling pathways involved in the regulation of normal cellproliferation and differentiation For example an increasein the activation of MAPK signal transduction pathway wasobserved as the cancer progresses [29] MAPKextracellularsignal-related kinase pathway was activated in tumors andrepresented a potential target for therapy [30] ThereforeccRCC as a common tumor was related toMAPK closely

Furthermore we studied gene swapping behaviors insingle altered module of four stages taking MAPK familygenes related altered modules as an example As shown inFigure 5 we could discover that for a module (MAPK1CEBPB MAPK3 RELA MAPK14 NFKB1 and RIPK2) instages I II III and IV its gene compositions (nodes) werethe same but the interaction scores (edges) were differentThe interaction value betweenMAPK1 andMAPK3 was 052minus048 109 and minus003 in stages I II III and IV respectivelyand there was a weak correlation of the two genes in stageII It might explain differences of modules and existenceof dysregulated modules Swapping behavior in the alteredmodule (CCNA2MND1 CDC45 RFC4 CCNB1 and CDK4)was shown in Figure 6

CCNA2 cyclin A2 was expressed in dividing somaticcells and regulated cell cycle progression by interacting withcyclin-dependent kinase (CDK) kinases [31] Consistent withits role as a key cell cycle regulator expression of CCNA2

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Computational and Mathematical Methods in Medicine 9

minus0231

minus0411

0389

03903920206

0322

minus0242

09010494

02410425 0184

0603CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(a)

minus0213minus0128

0355

09010544

0931

0092

06930754 0178

0771 0666

0565CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0019

(b)

0223

minus0208

0317

07970380841

0484

0055

05631238

05090415 0763

1203CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

(c)

0268

0292

0431

017701380059

0072

06810293

0545 0225

0978CCNA2 CDK4

CDC45 RFC4

CCNB1MND1

minus0081

minus0004

(d)

Figure 6 Swapping behavior in altered module (CCNA2 CDK4 CDC45 RFC4 CCNB1 and MND1) Nodes stood for genes and edgesstood for the interactions of genesThe thickness of the edges represented the interaction scores or expression levels between two genes in themodule more thickness with higher value of expression scores (a) represents stage I of ccRCC (b) represents stage II (c) represents stageIII and (d) represents stage IV

was found to be elevated in a variety of tumors such asbreast cervical liver and kidney tumors [32] It was not clearwhether increased expression ofCCNA2was a cause or resultof tumorigenesisCCNA2-CDK contributed to tumorigenesisby the phosphorylation of oncoproteins or tumor suppressors[33] In our paper we had also proved that the correlationvalue betweenCCNA2 andCDK4 of the four stages was 06030565 1203 and 0978 in sequence (Figure 6)Wemight inferthat cell cycle played amedium role in correlations ofCCNA2and cancers thus cell cycle was discussed next

Cell cycle is the series of events that take place in a cellleading to its division and duplication and dysregulation ofthe cell cycle components may lead to tumor formation [34]It had been reported that alterations in activated proteins(cyclins and cyclin-dependent kinases etc) which led tofailure of cell cycle arrest may thus serve as markers of amore malignant phenotype and cell cycle-related genes aided

in discrimination of atypical adenomatous hyperplasia fromearly adenocarcinoma [35] Chen et al demonstrated thatcell cycle progression effects on NF-120581B activity representeda molecular basis underlying the aggressive tumor behavior[36] Besides cell cycle checkpoint inactivation allowedDNAreplication in aneuploid cells and may favor oncogenicgenomic [37] and a cell cycle regulator is potentially involvedin genesis of many tumor types such as ccRCC [38] Wecould conclude that cell cycle played a key role in the ccRCCprogress

Our results also showed that ccRCC had close rela-tionship with metabolism pathways such as glutathionemetabolism and arginine and proline metabolism Glu-tathione metabolism which played both protective andpathogenic roles in cancers was crucial in the removal anddetoxification of carcinogens [39] And the present reviewhighlighted the role of glutathione and related cytoprotective

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

10 Computational and Mathematical Methods in Medicine

effects in the susceptibility to carcinogenesis and in thesensitivity of tumors to the cytotoxic effects of anticanceragents [40] Recently Hao et al discovered that three sig-nificant pathways related to ccRCC namely arginine andprolinemetabolism aldosterone-regulated sodium reabsorp-tion and oxidative phosphorylation were observed [41]Arginineproline metabolism is a significant pathway inccRCC that had been discovered by Perroud et al previously[42] and the results were in accordance with our analysis

5 Conclusions

In conclusion we successfully identified dysregulated genes(such as MAPK1 and CCNA2) and pathways (such as cellcycle glutathione metabolism and arginine and prolinemetabolism) of ccRCC in different stages and these genes andpathwaysmight be potential biologicalmarkers and processesfor treatment and etiology mechanism in ccRCC

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank the editors and reviewers for the insightfulcomments and suggestions on this work They also thankJinan Evidence Based Medicine Science-Technology Centerfor the help of statistical and computational advice

References

[1] H An Y Zhu L Xu L Chen Z Lin and J Xu ldquoNotch1 predictsrecurrence and survival of patients with clear-cell renal cellcarcinoma after surgical resectionrdquo Urology vol 85 no 2 pp483e9ndash483e14 2015

[2] F Vincenzo G Novara A Galfano et al ldquoThe lsquostage sizegrade and necrosisrsquo score ismore accurate than theUniversity ofCalifornia Los Angeles Integrated Staging System for predictingcancer-specific survival in patients with clear cell renal cellcarcinomardquo BJU International vol 103 no 2 pp 165ndash170 2009

[3] V R Dondeti B Wubbenhorst P Lal et al ldquoIntegrativegenomic analyses of sporadic clear cell renal cell carcinomadefine disease subtypes and potential new therapeutic targetsrdquoCancer Research vol 72 no 1 pp 112ndash121 2012

[4] A H Girgis V V Iakovlev B Beheshti et al ldquoMultilevel whole-genome analysis reveals candidate biomarkers in clear cell renalcell carcinomardquo Cancer Research vol 72 no 20 pp 5273ndash52842012

[5] L W Marston T A Rouault J Mitchell and M K CherukurildquoNitroxide therapy for the treatment of von HippelmdashLindaudisease (VHL) and renal clear cell carcinoma (RCC)rdquo GooglePatents 2014

[6] T Tanaka T Torigoe Y Hirohashi et al ldquoHypoxia-induciblefactor (HIF)-independent expression mechanism and novelfunction of HIF prolyl hydroxylase-3 in renal cell carcinomardquoJournal of Cancer Research and Clinical Oncology vol 140 no3 pp 503ndash513 2014

[7] F Jordan T-P Nguyen and W-C Liu ldquoStudying protein-protein interaction networks a systems view on diseasesrdquoBriefings in Functional Genomics vol 11 no 6 pp 497ndash5042012

[8] S Srihari and H W Leong ldquoTemporal dynamics of proteincomplexes in PPI networks a case study using yeast cell cycledynamicsrdquo BMC Bioinformatics vol 13 article S16 2012

[9] J Zhao S Zhang L-Y Wu and X-S Zhang ldquoEfficientmethods for identifying mutated driver pathways in cancerrdquoBioinformatics vol 28 no 22 pp 2940ndash2947 2012

[10] S Srihari C H Yong A Patil and L Wong ldquoMethods forprotein complex prediction and their contributions towardsunderstanding the organisation function and dynamics ofcomplexesrdquo FEBS Letters 2015

[11] C Wu J Zhu and X Zhang ldquoIntegrating gene expressionand protein-protein interaction network to prioritize cancer-associated genesrdquo BMC Bioinformatics vol 13 article 182 2012

[12] G Liu J Li and L Wong ldquoAssessing and predicting proteininteractions using both local and global network topologicalmetricsrdquo Genome Informatics vol 21 pp 138ndash149 2008

[13] L-H Chu and B-S Chen ldquoConstruction of a cancer-perturbedprotein-protein interaction network for discovery of apoptosisdrug targetsrdquo BMC Systems Biology vol 2 article 56 2008

[14] O Magger Y Y Waldman E Ruppin and R Sharan ldquoEnhanc-ing the prioritization of disease-causing genes through tissuespecific protein interaction networksrdquo PLoS ComputationalBiology vol 8 no 9 Article ID e1002690 2012

[15] S Srihari andMA Ragan ldquoSystematic tracking of dysregulatedmodules identifies novel genes in cancerrdquo Bioinformatics vol29 no 12 pp 1553ndash1561 2013

[16] D Szklarczyk A Franceschini M Kuhn et al ldquoThe STRINGdatabase in 2011 functional interaction networks of proteinsglobally integrated and scoredrdquo Nucleic Acids Research vol 39no 1 pp D561ndashD568 2011

[17] R A Irizarry BM Bolstad F Collin LMCope BHobbs andT P Speed ldquoSummaries of Affymetrix GeneChip probe leveldatardquo Nucleic Acids Research vol 31 no 4 article e15 2003

[18] B M Bolstad R A Irizarry M Astrand and T P Speed ldquoAcomparison of normalizationmethods for high density oligonu-cleotide array data based on variance and biasrdquo Bioinformaticsvol 19 no 2 pp 185ndash193 2003

[19] B Ben affy Built-in Processing Methods 2013[20] N Gerhard ldquoPearson correlation coefficientrdquo in Dictionary of

Pharmaceutical Medicine p 132 Springer 2009[21] G Liu L Wong and H N Chua ldquoComplex discovery from

weighted PPI networksrdquo Bioinformatics vol 25 no 15 pp 1891ndash1897 2009

[22] S Srihari andHW Leong ldquoA survey of computationalmethodsfor protein complex prediction from protein interaction net-worksrdquo Journal of Bioinformatics and Computational Biologyvol 11 no 2 Article ID 1230002 2013

[23] E Tomita A Tanaka and H Takahashi ldquoThe worst-case timecomplexity for generating all maximal cliques and computa-tional experimentsrdquoTheoretical Computer Science vol 363 no1 pp 28ndash42 2006

[24] H N Gabow ldquoAn efficient implementation of Edmondsrsquo algo-rithm for maximum matching on graphsrdquo Journal of the ACMvol 23 no 2 pp 221ndash234 1976

[25] D W Huang B T Sherman and R A Lempicki ldquoSystematicand integrative analysis of large gene lists using DAVID bioin-formatics resourcesrdquo Nature Protocols vol 4 no 1 pp 44ndash572008

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Computational and Mathematical Methods in Medicine 11

[26] G Ford Z Xu A Gates J Jiang and B D Ford ldquoExpressionAnalysis Systematic Explorer (EASE) analysis reveals differen-tial gene expression in permanent and transient focal stroke ratmodelsrdquo Brain Research vol 1071 no 1 pp 226ndash236 2006

[27] E Siljamaki L Raiko M Toriseva et al ldquoP38120575 mitogen-activated protein kinase regulates the expression of tight junc-tion protein ZO-1 in differentiating human epidermal ker-atinocytesrdquo Archives of Dermatological Research vol 306 no 2pp 131ndash141 2014

[28] P J Roberts and C J Der ldquoTargeting the Raf-MEK-ERKmitogen-activated protein kinase cascade for the treatment ofcancerrdquo Oncogene vol 26 no 22 pp 3291ndash3310 2007

[29] I M Subbiah G Varadhachary A M Tsimberidou et alldquoAbstract 4700 One size does not fit all Fingerprintingadvanced carcinoma of unknown primary through compre-hensive profiling identifies aberrant activation of the PI3K andMAPK signaling cascades in concert with impaired cell cyclearrestrdquo Cancer Research vol 74 no 19 article 4700 2014

[30] B H OrsquoNeil L W Goff J S W Kauh et al ldquoPhase II study ofthe mitogen-activated protein kinase 12 inhibitor selumetinibin patients with advanced hepatocellular carcinomardquo Journal ofClinical Oncology vol 29 no 17 pp 2350ndash2356 2011

[31] A Cribier B Descours A L C Valadao N Laguette and MBenkirane ldquoPhosphorylation of SAMHD1 by cyclin A2CDK1regulates its restriction activity toward HIV-1rdquo Cell Reports vol3 no 4 pp 1036ndash1043 2013

[32] C H Yam T K Fung and R Y C Poon ldquoCyclin A in cell cyclecontrol and cancerrdquoCellular andMolecular Life Sciences vol 59no 8 pp 1317ndash1326 2002

[33] S H Woo S-K Seo S An et al ldquoImplications of caspase-dependent proteolytic cleavage of cyclin A1 in DNA damage-induced cell deathrdquo Biochemical and Biophysical Research Com-munications vol 453 no 3 pp 438ndash442 2014

[34] V Hirt BartholomausMathematical Modelling of Cell Cycle andTelomere Dynamics University of Nottingham 2013

[35] Y F Ho S A Karsani W K Yong and S N Abd MalekldquoInduction of apoptosis and cell cycle blockade by helichrysetinin A549 human lung adenocarcinoma cellsrdquo Evidence-BasedComplementary and Alternative Medicine vol 2013 Article ID857257 10 pages 2013

[36] G Chen M S Bhojani A C Heaford et al ldquoPhosphorylatedFADD induces NF-120581B perturbs cell cycle and is associatedwith poor outcome in lung adenocarcinomasrdquo Proceedings ofthe National Academy of Sciences of the United States of Americavol 102 no 35 pp 12507ndash12512 2005

[37] J D Gordan P Lal V R Dondeti et al ldquoHIF-120572 effects on c-Myc distinguish two subtypes of sporadic VHL-deficient clearcell renal carcinomardquo Cancer Cell vol 14 no 6 pp 435ndash4462008

[38] Y Yamada H Hidaka N Seki et al ldquoTumor-suppressivemicroRNA-135a inhibits cancer cell proliferation by targetingthe c-MYC oncogene in renal cell carcinomardquo Cancer Sciencevol 104 no 3 pp 304ndash312 2013

[39] G K Balendiran R Dabur and D Fraser ldquoThe role ofglutathione in cancerrdquo Cell Biochemistry and Function vol 22no 6 pp 343ndash352 2004

[40] N Traverso R Ricciarelli M Nitti et al ldquoRole of glutathionein cancer progression and chemoresistancerdquoOxidativeMedicineand Cellular Longevity vol 2013 Article ID 972913 10 pages2013

[41] J-F Hao K-M Ren J-X Bai et al ldquoIdentification of poten-tial biomarkers for clear cell renal cell carcinoma based on

microRNA-mRNA pathway relationshipsrdquo Journal of CancerResearch andTherapeutics vol 10 pp C167ndashC172 2014

[42] B Perroud J Lee N Valkova et al ldquoPathway analysis of kidneycancer using proteomics and metabolic profilingrdquo MolecularCancer vol 5 article 64 2006

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom

Submit your manuscripts athttpwwwhindawicom

Stem CellsInternational

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MEDIATORSINFLAMMATION

of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Behavioural Neurology

EndocrinologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Disease Markers

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

BioMed Research International

OncologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Oxidative Medicine and Cellular Longevity

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

PPAR Research

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Immunology ResearchHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

ObesityJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Computational and Mathematical Methods in Medicine

OphthalmologyJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Diabetes ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Research and TreatmentAIDS

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Gastroenterology Research and Practice

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Parkinsonrsquos Disease

Evidence-Based Complementary and Alternative Medicine

Volume 2014Hindawi Publishing Corporationhttpwwwhindawicom