Normal/disease-paired cDNA library subtraction for molecular marker development

MOLECULAR BIOTECHNOLOGY Volume 27, 2004

cDNA Library Subtraction 119RESEARCH

119

Molecular Biotechnology © 2004 Humana Press Inc. All rights of any nature whatsoever reserved. 1073–6085/2004/27:2/119–125/$25.00

Abstract

Normal/Disease-Paired cDNA Library Subtractionfor Molecular Marker Development

Ning Wu,* Yandong Li, and Kanyand Matand

*Author to whom all correspondence and reprint requests should be addressed: Dr. Ning Wu, Center for Biotechnological Research andEducation, Research Building, Room, 109, Langston University, Highway 33, Langston, OK 73050. E-mail: [email protected]

This study consists of a novel strategy for the identification of potential cancer exclusively expressedgenes, which might lead to the development of valuable diagnostic molecular markers. Normal- and cancer-paired tissues from the same patient were collected and subjected to the construction of primary complemen-tary deoxyribonucleic acid (cDNA) libraries. The cancer cDNA library was generated by subtracting normalcDNAs from the primary cancer library. The remaining clones in the subtracted cancer library were se-quenced and Basic Local Alignment Search Tool (BLAST)-analyzed against current nonredundant andest_human databases in the GenBank. The clones that had no matches with any known gene sequencesexcept the human genomic sequences were identified as ideal candidates for potential molecular markerdevelopment. The candidate marker sequence and associated primer sequences were identified on the targetclone sequence of the region with the least, or no, matched homologs. The specificity of the markers wasmeasured by a polymerase chain reaction test experiment with the DNA templates from different humannormal tissues, genomic DNA, and additional patients’ tissues. Results from bioinformatical and experimen-tal approaches used suggest that the methodology has a high potential to identify cancer exclusive tran-scripts. Thus, it might result in the development of diagnostic markers.

Index Entries: Molecular marker; cancer; cDNA; subtraction; polymerase chain reaction; expressionsequence tag.

1. IntroductionIdentification of genes exclusively or abun-

dantly expressed in cancer may yield novel tumormolecular diagnostic markers (1). The suppres-sion subtraction hybridization technology hasbeen applied in the analysis of gene differentialexpression in carcinogenesis for potential markergene discovery (2). However, the potential formolecular marker diagnostic value of severalidentified target genes has been hampered by theirlow-level transcription in normal tissues. To over-come this limitation, the expression of targetgenes has to be tightly suppressed in normal non-neoplastic tissues, although it remains active intheir respective cancer cells (3). This article pre-sents a novel strategy with the potential of en-hancing cancer-specific gene discovery by usingthe method of paired complementary deoxyribo-

nucleic acid (cDNA) library subtraction. By sub-tracting all normal cDNAs from the cancer cDNAlibrary, only genes that were exclusively expressedin the cancer cells were identified. The DNA prim-ers of potential diagnostic genetic markers werethen developed and tested by polymerase chainreaction (PCR) (4).

2. Materials and Methods2.1. Human Total Ribonucleic AcidExtraction and mRNA Purification

Normal and cancer tissue samples were col-lected from a single patient to minimize geneticdifferences between individuals. Human renal cellcarcinoma and adjacent normal kidney tissueswere collected separately from a 61-yr-old Asianfemale. The tissues were first flash-frozen in liq-uid nitrogen, then ground to fine powder, and ho-

04/JW675/Wu 119-126 5/18/04, 11:40 AM119


120 Wu, Li, and Matand

mogenized using a prechilled mortar and pestleset and a homogenizer. Total RNA was isolatedusing a modified standard procedure (5), and mes-senger RNA (mRNA) was purified using magneticparticle technology primed by biotinylated oligo dTprimer and purified by Streptavidin magneticbeads (New England Biolabs, Beverly, MA) (6).The quality and concentration of both total RNAand mRNA were determined by spectrophotom-eter at 260- and 280-nm wavelengths.

2.2. Construction of Normal/Cancer-PairedcDNA Libraries

Two primary cDNA libraries, normal kidneyand renal cell carcinoma, were constructed usingthe modified directional cloning method (7). Onemicrogram of mRNA was reverse-transcribed us-ing the M-MLV reverse transcriptase (Promega,Madison, WI) with twice the amount of OligodT18–NotI primer for the first-strand cDNA syn-thesis in 20 µL at 37°C for 2 h. The second-strandcDNA was synthesized in 160 µL at 16°C for 2 hin the presence of 0.15 mM β-NAD, 3.75 mMdithiothreitol, and enzymes of ribonuclease H,DNA polymerase I, and Escherichia coli DNAligase. After ligation to EcoRI adapters, the cDNAmolecules were loaded onto a self-designed gelfiltration column for size selection. The cDNAfragments, which were larger than 500 bp, werecombined together and digested with NotI. Thenthey were cloned directionally at EcoRI andNotI sites into the phagemid vector—pT7T3-Pac(Pharmacia, Piscataway, NJ) with M13 F/R and T7/T3 primers at both ends of the multicloning sites.The qualities of both normal and cancer cDNA li-braries were determined by plating assay (count-ing for colony-forming unit [cfu]) and wholelibrary and random-picked colony-restriction di-gestion using EcoRI and NotI (checking the insertDNA digestion pattern and calculating the recom-binant efficiency).

2.3. Subtraction of Normal cDNAs FromCancer cDNA Library

The cDNA clones of a normal kidney librarywere subtracted from the cancer cDNA library togenerate a subtracted cancer library using a modi-

fied normalization/subtraction method based onthe calculation of reassociation kinetics (C0tnumber) (8). Both primary cDNA libraries wereamplified, and the double-stranded cDNAs werepurified using Plasmid Midi Kit (Qiagen, Valencia,CA) and then reelectroporated into E. coli TOP10F'bacteria (Invitrogen, Carlsbad, CA) for single-stranded circular cDNA generation by superinfec-tion of bacteria with the M13KO7 help phage.The single-stranded cDNAs were purified byhydroxyapatite (HAP) chromatography at 60°Cusing the buffer containing 0.12M sodium phos-phate, pH 6.8, 10 mM ethylenediamine–tetra-acetic acid (EDTA), and 1% sodium dodecylsulfate (SDS). The single-stranded cDNAs fromthe cancer library were defined as “tracers,” andthose from the normal library were used as tem-plates for PCR to generate partial second-strandcDNAs, which were complementary to the cDNAinserts and were defined as the subtraction “driv-ers.” The PCR reaction included the Taq PCRCore Kit (Qiagen, Valencia, CA), approx 1–2 ngof circular single-stranded cDNAs, 1 µM of eachof T7/T3 primers, and 0.2 mM of deoxy–nucle-otide-triphosphates. The driver DNAs were puri-fied by ethanol precipitation. The subtractionhybridization was performed in 20 µL vol in thepresence of 50% formamide, 1% SDS, 0.01MTris-HCl, pH 7.5, 0.01M EDTA, 0.5M NaCl, 50ng of tracers (single-stranded cDNAs from thecancer library), 2.5 µg of drivers (PCR productsof normal cDNAs), and the blocking oligonucle-otides, which included 40 µg of 5' blocker (5'-TAATACGACTCACTATAGGGAATTTGGCCCTCGAGG CCAAGAATTCGGCACGAGG-3'),40 µg of 3' blocker (5'-AAAAAAAAAAAAAAAAAAAAGCGGCCGCAAGCTTATTCCCTTTAGTGAG GGTTAAT-3') and 10 µg of tailblocker (5'-(A)50-3'). The reaction was set at 30°Cfor 17 h with the C0t number of about 75. Aftersubtraction–hybridization, the hybridized par-tially double-stranded cDNAs were separatedfrom the nonhybridized single-stranded cDNAsby HAP chromatography using similar buffer in-dicated earlier. The single-stranded cDNAs wereconverted into partial double-stranded cDNAs bylimited extension using primer 5'-GACTGGT

04/JW675/Wu 119-126 5/18/04, 11:40 AM120


cDNA Library Subtraction 121

GAGTACTCAACC AAGTC-3' and T7 Sequenasev 2.0 (Amersham, Piscataway, NJ) at 37°C for 30min. The converted cDNAs were then electroporatedinto E. coli TOP10 bacteria (Invitrogen, Carlsbad,CA) for amplification. The amplified subtractedcancer library was then proportionally diluted andplated for colony picking and sequencing process.

2.4. Clone Sequencing, Data Analysis,and Molecular Marker Identification

All the clones in the subtracted cancer librarywere plated and subjected to DNA sequencingfrom both directions. DNA sequencing was per-formed using ABI PRISM® 3100 Genetic Ana-lyzer and BigDye™ Terminators Sequencing Kit(Applied Biosystems, Foster City, CA). The se-quences were submitted to GenBank for BasicLocal Alignment Search Tool (BLAST) analysisagainst nonredundant (nr) and est_human nucle-otide databases. According to the BLAST results,the potential molecular marker was determinedthrough the query sequence area, which had noneor the least homolog-matching responses. The pairof PCR primers for the potential molecular mark-ers were designed and synthesized for the follow-ing PCR test experiments.

2.5. Potential Molecular MarkerTest Experiments

A PCR experiment was designed to test thespecificity of selected potential molecular mark-ers. Additional paired normal and kidney cancertissues from other patients were collected. The to-tal RNAs were extracted, and the mRNAs werepurified using the magnetic particle technologyand reverse-transcribed to the first-strand cDNAs.Further, human genomic DNA and seven humannormal-tissue cDNA libraries, including brain,colon, kidney, liver, lung, stomach, and uteruswere also used in the test experiment.

3. Results and DiscussionBoth primary kidney normal and cancer cDNA

libraries contained 106 cfu and 91% recombinantefficiency. The subtracted cancer cDNA libraryhad no colony growing in the plating assay beforeamplification. The restriction analysis showed thatboth primary libraries contained similar DNA-

smear areas above 500 bp. However, several clear,significant DNA bands were observed only in thesubtracted library. This suggested that a few genesremained after the whole library had been sub-tracted. The plating assay for the amplified sub-tracted library generated about 150 colonies,which were subsequently sequenced. Of all se-quences that passed BLAST search, a total of 25individual sequences were identified (Table 1).Seventeen sequences had certain matches withknown gene homologues located in the nr database.The other eight sequences showed no matches withany known genes except human chromosomal se-quences. However, all those eight sequences hadmatched with NCI CGAP cDNA clones—gener-ated from different cancer cDNA libraries—in theest_human database, and five of them had the bitsscore higher than 450. This suggests that thosesequences may be silent under normal environ-ment and expressed exclusively under cancer con-ditions. Those five sequences were the idealcandidates for the development of molecularmarkers.

One clone named hkms60 (566 nucleotides),which was used to test this novel strategy, had anexact match of human chromosome 7 (GI: 12025623) genomic sequence (bits score: 1007, E value:0.0) in the region of 120350K to 120360K. Noknown genes had been identified in this chromo-somal region. About 55% of hkms60 sequence—from position 220 to 566 bp—showed severalhomolog matches among which the best match wasan expressed sequence tag (EST) from NCI_CGAPkidney cancer cDNA library (GI: 9509330; 308nucleotides matched with bits score 569, E value:e-160). However, there was only one homolog(GI: 12373517) with 58 nucleotides that matchedwith hkms60 (bits score: 54, E value: 8e-05) in thefirst 220-bp region. The least matched region fromthe 71–240-bp position (170 nucleotides) with theoptimal diagnostic PCR primers (5'-GTGTGCATGGCTCTATGTAA-3' and 5'-CTATGAAGGGGTAACACCAA-3') on both ends is the potentialmolecular marker.

PCR test reactions of seven normal tissuecDNAs generated no amplified PCR products onall normal tissues except the genomic control and

04/JW675/Wu 119-126 5/18/04, 11:40 AM121



Tab

le 1

BL

AS

T R

esul

ts o

f S

ubtr

acte

d cD

NA

Lib

rary

Clo

nes

Clo

neQ

uery

Sco

reId

enti

ties

No.

IDL

engt

h (b

p)H

omol

og(b

its)

(%)

1hk

ms1

548

0nr

: H

omo

sapi

ens

beta

ine-

hom

ocys

tein

e S-

met

hylt

rans

fera

se m

RN

A40

898

est_

hum

an:

qn24

h12.

x1 N

CI_

CG

AP

_Kid

5 H

omo

sapi

ens

cDN

A c

lone

420

982

hkm

s25

343

nr:

Glu

tath

ione

S-t

rans

fera

se A

2 su

buni

t51

395

est_

hum

an:

oq62

e11.

s1 N

CI_

CG

AP

_Kid

6 H

omo

sapi

ens

cDN

A c

lone

561

943

hkm

s27

742

nr:

Hom

o sa

pien

s ri

boso

mal

pro

tein

L24

(R

PL

24)

mR

NA

414

93es

t_hu

man

: ti

56b0

8.x1

NC

I_C

GA

P_L

ym12

Hom

o sa

pien

s cD

NA

clo

ne42

093

4hk

ms2

940

7nr

: H

omo

sapi

ens

carb

onyl

red

ucta

se 1

(C

BR

1) m

RN

A43

290

est_

hum

an:

tw78

a01.

x1 N

CI_

CG

AP

_Ut3

Hom

o sa

pien

s cD

NA

clo

ne44

090

5hk

ms3

163

1nr

: H

omo

sapi

ens

cyto

chro

me

P45

0, s

ubfa

mil

y X

XV

IIB

(25-

hydr

oxyv

itam

in D

-1- α

-hyd

roxy

lase

), p

olyp

epti

de 1

(C

YP

27B

1) m

RN

A71

890

est_

hum

an:

wh6

9b08

.x1

NC

I_C

GA

P_K

id11

Hom

o sa

pien

s cD

NA

clo

ne71

890

6hk

ms1

021

4nr

: H

.sap

iens

mR

NA

for

DR

ES

9 pr

otei

n54

94es

t_hu

man

: tc

90e0

7.x1

NC

I_C

GA

P_C

LL

1 H

omo

sapi

ens

cDN

A c

lone

5494

7hk

ms4

283

8nr

: H

uman

ST

S W

I-12

968

547

97es

t_hu

man

: w

a03e

08.x

1 N

CI_

CG

AP

_Kid

11 H

omo

sapi

ens

cDN

A c

lone

557

978

hkm

s44

938

nr:

TH

IOS

UL

FA

TE

SU

LF

UR

TR

AN

SF

ER

AS

E (

HU

MA

N)

648

92es

t_hu

man

: xk

04e0

7.x1

NC

I_C

GA

P_C

019

Hom

o sa

pien

s cD

NA

clo

ne64

892

9hk

ms2

384

1nr

: H

omo

sapi

ens

glut

athi

one

pero

xida

se 3

(pl

asm

a) (

GP

X3)

mR

NA

7491

est_

hum

an:

yn67

b03.

s1 S

oare

s ad

ult b

rain

N2b

5HB

55Y

Hom

o sa

pien

s cD

NA

clo

ne90

9310

hkm

s47

814

nr:

Hom

o sa

pien

s I

fact

or (

com

plem

ent)

(im

mun

oflu

ores

cenc

e (I

F))

mR

NA

926

95es

t_hu

man

: xx

31a1

1.x1

NC

I_C

GA

P_U

t1 H

omo

sapi

ens

cDN

A c

lone

932

9611

hkm

s48

744

nr:

Hom

o sa

pien

s go

lgi a

utoa

ntig

en, g

olgi

n su

bfam

ily

a, 3

(G

OL

GA

3) m

RN

A91

696

est_

hum

an:

xn59

c11.

x1 S

oare

s_N

HC

eC_c

ervi

cal_

tum

or H

omo

sapi

ens

cDN

A92

496

12hk

ms6

323

2nr

: P

LA

SM

A S

ER

INE

PR

OT

EA

SE

(H

UM

AN

)38

596

est_

hum

an:

xm07

h12.

x1 N

CI_

CG

AP

_Ut4

Hom

o sa

pien

s cD

NA

clo

ne38

596

13hk

ms5

710

7nr

: H

uman

rib

osom

al p

rote

in S

14 g

ene,

com

plet

e cd

s20

299

est_

hum

an:

hh80

e02.

x1 N

CI_

CG

AP

_GU

1 H

omo

sapi

ens

cDN

A c

lone

202

99

04/JW675/Wu 119-126 5/18/04, 11:40 AM122



hkm

s11

553

nr:

Hom

o sa

pien

s ur

omod

ulin

(ur

omuc

oid,

Tam

m-H

orsf

all g

lyco

prot

ein)

mR

NA

468

87es

t_hu

man

: w

a04b

02.x

1 N

CI_

CG

AP

_Kid

11 H

omo

sapi

ens

cDN

A c

lone

470

8715

hkm

s21

246

nr:

Hom

o sa

pien

s se

rine

/thr

eoni

ne k

inas

e 11

(P

eutz

-Jeg

hers

syn

drom

e) m

RN

A37

394

est_

hum

an:

tp84

d06.

x1 N

CI_

CG

AP

_Ut3

Hom

o sa

pien

s cD

NA

clo

ne38

394

16hk

ms1

823

0nr

: H

uman

ST

S W

I-13

531

5090

est_

hum

an:

wz8

9d08

.x1

NC

I_C

GA

P_B

rn25

Hom

o sa

pien

s cD

NA

clo

ne50

9017

hkm

s26

551

nr:

Hom

o sa

pien

s 3-

hydr

oxy-

3-m

ethy

lglu

tary

l-C

oenz

yme

A r

educ

tase

(

HM

GC

R)

gene

, com

plet

e cd

s68

095

est_

hum

an:

xt75

b04.

x2 N

CI_

CG

AP

_Ut2

Hom

o sa

pien

s cD

NA

clo

ne68

095

18hk

ms3

512

40nr

: G

enom

ic s

eque

nce

from

Hum

an 1

7, c

ompl

ete

sequ

ence

803

97es

t_hu

man

: tm

76a0

3.x1

NC

I_C

GA

P_B

rn25

Hom

o sa

pien

s cD

NA

clo

ne80

397

19hk

ms4

678

8nr

: H

uman

chr

omos

ome

14 D

NA

seq

uenc

e45

497

est_

hum

an:

xt85

a06.

x1 N

CI_

CG

AP

_Ut1

Hom

o sa

pien

s cD

NA

clo

ne45

497

20hk

ms6

056

6nr

: H

omo

sapi

ens

chro

mos

ome

7 cl

one

RP

11-7

02D

16, c

ompl

ete

sequ

ence

1007

96es

t_hu

man

: hw

25a0

9.x1

NC

I_C

GA

P_K

id11

Hom

o sa

pien

s cD

NA

clo

ne56

997

21hk

ms2

049

5nr

: H

omo

sapi

ens

chro

mos

ome

5 cl

one

CT

d-20

35K

4, c

ompl

ete

sequ

ence

3810

0es

t_hu

man

: no

hit

22hk

ms3

964

1nr

: H

omo

sapi

ens

PA

C c

lone

RP

4-67

6L20

fro

m 7

q35-

q36,

com

plet

e se

quen

ce60

84es

t_hu

man

: n1

77b1

0.s1

NC

I_C

GA

P_B

r2 H

omo

sapi

ens

cDN

A c

lone

5285

23hk

ms4

186

2nr

: H

omo

sapi

ens

chro

mos

ome

6p21

.3, H

LA

Cla

ss I

reg

ion,

sec

tion

5/2

075

792

est_

hum

an:

xm19

b03.

x1 N

CI_

CG

AP

_Ut4

Hom

o sa

pien

s cD

NA

clo

ne84

8524

hkm

s19

629

nr:

Hom

o sa

pien

s cl

one

RP

11-1

22G

11, c

ompl

ete

sequ

ence

3892

eest

_hum

an:

w14

3c01

.x1

NC

I_C

GA

P_U

t1 H

omo

sapi

ens

cDN

A c

lone

755

9625

hkm

s62

295

nr:

Hom

o sa

pien

s cl

one

RP

11-4

70M

17, c

ompl

ete

sequ

ence

515

97es

t_hu

man

: tm

56f1

0.x1

NC

I_C

GA

P_K

id11

Hom

o sa

pien

s cD

NA

clo

ne52

397

Abb

r: n

r, a

ll n

onre

dund

ant

Ge n

Ban

k +

EM

BL

+ D

DB

J +

Pro

tein

Da t

a B

a nk

(PD

B)

sequ

enc e

s (b

ut n

o E

ST

, S

TS

, G

SS

, or

HT

GS

sequ

ence

s);e

st_h

uman

, the

non

redu

ndan

t dat

abas

e of

Gen

Ban

k +

EM

BL

+ D

DB

J E

ST

div

isio

ns li

mit

ed to

the

hum

an o

rgan

ism

.

04/JW675/Wu 119-126 5/18/04, 11:40 AM123



Fig. 1. PCR reactions of seven normal tissue cDNA libraries. Lane 1: DNA ladder; lane 2: kidney cancer; lanes3–9: brain, colon, kidney, liver, lung, stomach, and uterus; lane 10: human genomic DNA.

Fig. 2. PCR treatments of four patients’ paired kidney normal and cancer cDNAs. (A) Test experiment: the PCRproducts are 170 bp. (B) Control reaction: b-actin 3' end fragment (720 bp). Lane 1: DNA ladder; lane 2: humangenomic DNA; lanes 3, 5, and 7: patients’ normal kidney cDNAs; lanes 4, 6, and 8: patients’ kidney cDNAs.

kidney cancer tissue (Fig. 1). Additional PCR testresults of different patients’ kidney normal andcancer cDNAs with the human genomic and β-actin control experiments are showed in Fig. 2.Only kidney cancer cDNAs that included genomiccontrol generated 170-bp amplified products asexpected. The results suggest that hkms60 existedin normal human genomic DNA but was tran-scribed only in cancerous cells. This study pro-vides strong evidence in support of our newlydeveloped normal/cancer paired cDNA library sub-traction strategy. It has been a reliable technique forthe development of potential molecular markers withhigh specificity and diagnostic utility.

AcknowledgmentsWe thank Dr. Jason Feng for technical advice

and Penny Xia for laboratory assistance.

References1. Wu, C. G., Habib, N. A., Mitry, R. R., Reitsma, P. H.,

van Deventer, S. J., and Chamuleau, R. A. (1997)Overexpression of hepatic prothymosin alpha, a novelmarker for human hepatocellular carcinoma. Br. J.Cancer 76, 1199–1204.

2. Ahmed, F. E. (2002) Molecular techniques for studyinggene expression in carcinogenesis. J. Environ. Sci. HealthPart C Environ. Carcinog. Ecotoxicol. Rev. 20, 77–116.

3. Lacroix, J., Becker, H. D., Woerner, S. M., Rittgen, W.,Drings, P., and von Knebel Doeberitz, M. (2001) Sen-sitive detection of rare cancer cells in sputum and pe-

04/JW675/Wu 119-126 5/18/04, 11:40 AM124



ripheral blood samples of patients with lung cancer bypreproGRP-specific RT-PCR. Int. J. Cancer 92, 1–8.

4. Raj, G. V., Moreno, J. G., and Gomella, L. G. (1998)Utilization of polymerase chain reaction technology inthe detection of solid tumors. Cancer 82, 1419–1442.

5. Puissant, C. and Houdebine, L.-M. (1990) An improve-ment of the single-step method of RNA isolation byacid guanidinium thiocyanate-phenol-chloroform ex-traction. Biotechniques 8, 148, 149.

6. Bach, H. J., Hartmann, A., Trevors, J. T., and Munch,J. C. (1999) Magnetic capture-hybridization method

for purification and probing of mRNA for neutral pro-tease of Bacillus cereus. J. Microbiol. Methods 37,187–192.

7. Soares, M. B. (1994) Construction of directionallycloned cDNA libraries in phagemid vectors. In Auto-mated DNA Sequencing and Analysis. Adams, M. D.,Fields, C., and Venter, J. C., eds., Academic Press,New York, pp. 110–114.

8. Bonaldo, M. F., Lennon, G., and Soares, M. B. (1996)Normalization and subtraction: two approaches to fa-cilitate gene discovery. Genome Res. 6, 791–806.

04/JW675/Wu 119-126 5/18/04, 11:40 AM125

Documents

Normal/disease-paired cDNA library subtraction for molecular marker development