Gastroenterology 2014;147:680–689

BASICAND

TRANSLATIONALAT

Variants in Nicotinamide Adenine Dinucleotide Phosphate OxidaseComplex Components Determine Susceptibility to Very Early OnsetInflammatory Bowel DiseaseSandeep S. Dhillon,1,2 Ramzi Fattouh,1 Abdul Elkadri,1,2,3 Wei Xu,4 Ryan Murchie,1

Thomas Walters,1,3 Conghui Guo,1 David Mack,5 Hien Q. Huynh,6 Shairaz Baksh,6

Mark S. Silverberg,2,7 Anne M. Griffiths,1,3 Scott B. Snapper,8 John H. Brumell,1,2,9 andAleixo M. Muise1,2,3

1SickKids Inflammatory Bowel Disease Center and Cell Biology Program, Research Institute, The Hospital for Sick Children,Toronto, Ontario, Canada; 2Institute of Medical Science, University of Toronto, Toronto, Ontario, Canada; 3Division ofGastroenterology, Hepatology, and Nutrition, Department of Pediatrics, University of Toronto, The Hospital for Sick Children,Toronto, Ontario, Canada; 4Princess Margaret Hospital and Dalla Lana School of Public Health, University of Toronto, Toronto,Ontario, Canada; 5Division of Pediatric Gastroenterology, Children’s Hospital of Eastern Ontario, Ottawa, Ontario, Canada;6 Division of Pediatric Gastroenterology, Stollery Children’s Hospital, Edmonton, Alberta, Canada; 7Mount Sinai HospitalInflammatory Bowel Disease Group, University of Toronto Group, Zane Cohen Center for Digestive Diseases, Toronto, Ontario,Canada; 8Division of Pediatric Gastroenterology, Hepatology, and Nutrition, Department of Medicine, Children’s HospitalBoston; Division of Gastroenterology and Hepatology, Brigham & Women’s Hospital, Department of Medicine, Harvard MedicalSchool, Boston, Massachusetts; and 9Department of Molecular Genetics, University of Toronto, Toronto, Ontario, Canada

Abbreviations used in this paper: CGD, chronic granulomatous disease;CYBA, cytochrome b light chain and encodes p22phox protein; CYBB,cytochrome b-245 or NADPH oxidase 2 and encodes Nox2 or gp91phox;GFP, green fluorescent protein; IBD, inflammatory bowel disease; MSMD,Mendelian susceptibility to mycobacterial disease; NADPH, nicotinamideadenine dinucleotide phosphate; NBT, nitroblue tetrazolium; NCF1,neutrophil cytosol factor 1 and encodes p47 phox protein; NCF2, neutrophilcytosol factor 2 and encodes p67 phox protein; NCF4, neutrophil cytosolfactor 4 and encodes p40 phox protein; PID, primary immunodeficiency;RAC1 and 2, Ras-related C3 botulinum toxin substrate 1 and 2; ROS,reactive oxygen species; SNP, single nucleotide polymorphism; VEOIBD,very early onset inflammatory bowel disease; VEOUC, very early onsetulcerative colitis.

© 2014 by the AGA Institute0016-5085/$36.00

http://dx.doi.org/10.1053/j.gastro.2014.06.005

BACKGROUND & AIMS: The colitis observed in patients withvery early onset inflammatory bowel disease (VEOIBD;defined as onset of disease at younger than 6 years of age)often resembles that of chronic granulomatous disease (CGD)in extent and features of colonic inflammation observed byendoscopy and histology. CGD is a severe immunodeficiencycaused by defects in the genes that encode components of thenicotinamide adenine dinucleotide phosphate (NADPH) oxi-dase complex. We investigated whether variants in genes thatencode NADPH oxidase components affect susceptibility toVEOIBD using independent approaches. METHODS: We per-formed targeted exome sequencing of genes that encodecomponents of NADPH oxidases (cytochrome b light chain andencodes p22phox protein; cytochrome b-245 or NADPH oxi-dase 2, and encodes Nox2 or gp91phox; neutrophil cytosolfactor 1 and encodes p47 phox protein; neutrophil cytosolfactor 2 and encodes p67 phox protein; neutrophil cytosolfactor 4 and encodes p40 phox protein; and Ras-related C3botulinum toxin substrate 1 and 2) in 122 patients withVEOIBD diagnosed at The Hospital for Sick Children, Univer-sity of Toronto, from 1994 through 2012. Gene variants werevalidated in an independent International Early Onset Pedi-atric IBD Cohort Study cohort of patients with VEOIBD. In asecond approach, we examined Tag single nucleotide poly-morphisms in a subset of patients with VEOIBD in which theNOX2 NADPH oxidase genes sequence had been previouslyanalyzed. We then looked for single nucleotide poly-morphisms associated with the disease in an independentInternational Early Onset Pediatric IBD Cohort Study cohort ofpatients. We analyzed the functional effects of variants asso-ciated with VEOIBD. RESULTS: Targeted exome sequencingand Tag single nucleotide polymorphism genotyping identified11 variants associated with VEOIBD; the majority of patientswere heterozygous for these variants. Expression of thesevariants in cells either reduced oxidative burst or altered in-teractions among proteins in the NADPH oxidase complex.Variants in the noncoding regulatory and splicing elements

resulted in reduced levels of proteins, or expression of alteredforms of the proteins, in blood cells from VEOIBD patients.CONCLUSIONS: We found that VEOIBD patients carry het-erozygous functional hypomorphic variants in components ofthe NOX2 NADPH oxidase complex. These do not cause overtimmunodeficiency, but instead determine susceptibility toVEOIBD. Specific approaches might be developed to treat in-dividual patients based on their genetic variant.

Keywords: VEOIBD; CGD; Genetics; Phagocytes.

hronic granulomatous disease (CGD) is a severe im-

Cmunodeficiency caused by genetic defects in compo-nents of the NOX2 nicotinamide adenine dinucleotidephosphate (NADPH) oxidase complex (Figure 1). The muta-tions identified in the NADPH oxidase gene in CGD patientsmostly result in the complete loss of phagocytes’ ability tomount the sufficient respiratory burst required to kill in-vading pathogens, leading to a severe immunodeficiencywith

Figure 1. Functional vari-ants in the NADPH oxidasegenes in VEOIBD andCGD.(A) Direct-exon sequencingof VEOIBD yielded a num-ber of novel and rare vari-ants with changes infunction and resulting al-tered NADPH oxidasecomplex activity. A da-maging G364R variant ofgp91phox was found nearthe FAD-binding domain. Inaddition, a R90H variant inthe PI(3,4)P2-binding do-main of p47phox, which waspreviously shown to resultin reduced ROS produc-tion, was found in a subsetof patients. p67phox wasshown to have many func-tional coding variants,mostnotably in protein–proteinbinding; these were R38Q,H389Q, N419I, andG501R,decreasing binding toRac2, Vav1, p40phox, andp47phox, respectively. R308Q, a rare variant in the PB1domain of p40phox, wasshown to reduce binding top67phox. Lastly, 2 non-coding variants were foundto reduce protein expres-sion of Rac2 and p22phox.(B) Representation ofNADPH oxidase genesvariants found in VEOIBD(purple star) and mutationsin CGD (red circle).

September 2014 VEOIBD and the NOX2 NADPH Oxidase Genes 681

BASICAN

DTR

ANSLAT

IONA

LAT

recurrent infections.1,2 Interestingly, up to 40% of CGD pa-tients develop intestinal inflammation that is very similar tothe colitis observed in inflammatory bowel disease (IBD)patients.3–5 Very early onset inflammatory bowel disease(VEOIBD) is defined as disease onset before 6 years of age6,7

and has increased in incidence dramatically during the pastdecade.8 It primarily affects the colon,1,2 and the diseaseextent and location and endoscopic and histologic

appearance of the colonic inflammation often resembles thecolonic disease observed in CGD.

Recent large-scale genetic studies have identified >160genetic loci that affect risk of developing IBD.9–11 However,the causal genes in the loci, the mechanisms by which thesevariants contribute to IBD pathogenesis, and the influenceon specific IBD phenotypes, are mostly unclear. Commonsingle nucleotide polymorphisms (SNPs) in the NADPH

682 Dhillon et al Gastroenterology Vol. 147, No. 3

BASICAND

TRANSLATIONALAT

oxidase complex are associated with adult-onset IBD,including neutrophil cytosol factor 4 and encodes p40 phox

protein (NCF4)12–14 and Ras-related C3 botulinum toxinsubstrate 1 and 2 (RAC1 and 2).12,15 Additionally, in a recentstudy,12 a novel variant of p67phox (encoded by neutrophilcytosol factor 2 and encodes p67 phox protein [NCF2]) thatreduced binding to RAC2 was uniquely associated withVEOIBD.12 Therefore, we investigated the role of theseNADPH oxidase genes in VEOIBD through Tag SNP geno-typing and targeted exome sequencing and identified func-tional variants that reduced oxidative burst, altered NADPHoxidase complex protein–protein interaction, or decreasedgene expression. Overall, our study revealed that functionalvariants in the NOX2 NADPH oxidase complex genes that donot cause an overt immunodeficiency are associated withVEOIBD susceptibility.

MethodsAll results are presented according to the Strengthening the

Reporting of Genetic Associations guidelines.16 Two comple-mentary approaches were carried out to define the geneticassociation of the NADPH oxidase genes (cytochrome b lightchain and encodes p22phox protein [CYBA], cytochrome b-245or NADPH oxidase 2 and encodes Nox2 or gp91phox [CYBB],NCF1, NCF2, NCF4, RAC1, RAC2) with VEOIBD (illustrated inFigure 2). The first approach used targeted exome sequencingof the NADPH oxidase gene in a cohort of VEOIBD patients andvalidated in a replication cohort. The second approach was toanalyze a subset of VEOIBD patients previously genotyped forTag SNPs covering the NADPH oxidase genes and to replicateassociated variants in an independent VEOIBD cohort.

Study DesignTargeted Exome Sequencing. One hundred and

twenty-two VEOIBD patients were sequenced using targeted

enrichment of the exon-coding for the NADPH oxidase genes(CYBA, CYBB, NCF1, NCF2, NCF4, RAC1, RAC2) using AgilentSureSelect target enrichment and sequenced on the IlluminaHiSeq 2000/2500 with exon primer and sequencing protocolsdesigned by the Beckman Coulter Genomics (http://www.beckmangenomics.com/) as described previously17 (Figure 2).Sanger sequencing and Taqman sequencing were used to verifyall genetic variants and SNPs identified in the NADPH oxidasegenes at The Centre for Applied Genomic (TCAG; http://www.tcag.ca). A validation cohort of 480 healthy controls and thereplication cohort consisting of 164 VEOIBD patients and 980healthy control individuals genotyped using Taqman at theCentre for Applied Genomics, The Hospital for Sick Children, asdescribed previously.12,15

Tagged Single Nucleotide PolymorphismsGenotyping. In order to examine intronic variants notcovered in the targeted exome sequencing approach (Figure 2),a VEOIBD subset of patients identified in a previous NADPHoxidase gene Tag SNP study were examined for associationwith VEOIBD. Briefly, 72 Tag SNPs in the NADPH oxidase genes(4 in CYBA, 6 in CYBB, 14 in NCF2, 20 in NCF4, 10 in RAC1, and18 in RAC2) were selected from the International HapMapProject (www.hapmap.org) phase II, data release 23a (with aminor allele frequency >1%) as part of an Illumina GoldengateCustom Chip, as described previously.12,15 From that study, 159VEOIBD patients were identified and compared with 913healthy controls. SNPs associated with VEOIBD were replicatedin a cohort consisting of 633 subjects (153 VEOIBD patientsand 480 healthy controls) using Taqman at the Centre forApplied Genomics, The Hospital for Sick Children.

Analysis. Single nucleotide and insertion/deletion vari-ants identified by either targeted exome sequencing or Tag SNPgenotyping were validated by Sanger sequencing. Function andminor allelic frequency were searched for using National Heart,Lung and Blood Institute Exome Sequencing Project ExomeVariant Server (http://evs.gs.washington.edu/EVS/),18 NationalCenter for Biotechnology Information dbSNP (http://www.

Figure 2. Flow chart ofgenetic studies.

September 2014 VEOIBD and the NOX2 NADPH Oxidase Genes 683

BASICAN

DTR

ANSLAT

IONA

LAT

ncbi.nlm.nih.gov/projects/SNP/),19 National Institute of Envi-ronmental Health Sciences FuncPred (http://snpinfo.niehs.nih.gov/snpinfo/snpfunc.htm),20 Polyphen2 (http://genetics.bwh.harvard.edu/pph2/),21 SIFT (http://sift.jcvi.org/),22 FastSNP(http://fastsnp.ibms.sinica.edu.tw/),23 Human Splicing Finder(http://www.umd.be/HSF/),24 and pfSNP (http://pfs.nus.edu.sg/).25

Setting. Patients included in the study were recruitedfrom the Inflammatory Bowel Disease Centre at The Hospitalfor Sick Children, University of Toronto, and were diagnosedwith VEOIBD between the years 1994 and 2012, with aconfirmed diagnosis before 6 years of age (definition based onour recent modification4 to the Paris classification26). All pa-tients were found to have typical IBD without any clinical,histologic, or biochemical evidence of immunodeficiency.

Patients did not have evidence of chronic granulomatousdisease (CGD). CGD is caused by mutations in any 1 of 5 genes,including X-linked CGD caused by mutation in the geneencoding p91-phox (CYBB; 300481); autosomal recessive cy-tochrome b-negative CGD (233690), caused by mutations in theCYBA gene (608508); autosomal recessive cytochromeb–positive CGD type I (233700), caused by mutation in theNCF1 gene (608512); autosomal recessive cytochromeb–positive CGD II (608515), caused by mutation in the NCF2gene (608515); and autosomal recessive cytochromeb–positive CGD type III (613960), caused by mutation in theNCF4 gene (601488) (OMIM: http://www.omim.org/entry/306400). VEOIBD patients did not have CGD as clinically, bio-chemically, or genetically defined, with no previously describeddisease causing CGD mutation for either X-lined or 2 CGDmutations for autosomal recessive disease, no atypical orchronic infection, and no diagnostically deficient neutrophilburst as measured by nitroblue tetrazolium (NBT) or dihy-drorhodamine. Therefore, these VEOIBD patients do not haveCGD and would not require prophylactic antibiotics or otherCGD specific treatments.

Participants. This was a cohort study that examined thegenetics of VEOIBD patients. All VEOIBD subjects wererecruited from The Hospital for Sick Children, Toronto, Canada,and had a confirmed diagnosis of IBD before 6 years of age,based on our recent modification4 to the Paris classification,26

and were therefore classified as VEOIBD.4 Patients with aknown immunodeficiency, including CGD, were excluded fromthe study. Phenotypic information and DNA samples were ob-tained from study subjects with approval of the InstitutionalReview Ethics Board for IBD genetic studies at The Hospital forSick Children in Toronto. Replication cohorts had Ethics Boardapproval for genetic and phenotypic studies at the individualinstitutions. Written informed consent was obtained from allparticipants. The healthy controls were obtained from theCentre for Applied Genomics (Ontario Population GenomicsPlatform; plates used: 1–15; a complete description of thiscontrol population can be found at http://www.tcag.ca/cyto_population_control_DNA.html).

To conduct systematic quality control on the raw geno-typing data, we examined SNPs genotyped for the initial cohortonly as reported previously.12,15 No SNP deviated significantlyfrom Hardy-Weinberg Equilibrium in the controls (P < .001).The pediatric and healthy control cohorts along with strictquality-control measures were carried out as described previ-ously for this cohort.12,27

VariablesConstructs and Antibodies. Myc-tagged human

p67phox complementary DNA was cloned into the pCDNA3vector (Invitrogen, Carlsbad, CA), as described previously12;green fluorescent protein (GFP)–tagged human p40phox com-plementary DNA was obtained from the laboratory of Dr JohnBrumell and GFP-tagged human p47phox complementary DNAwas obtained from OriGene (pCMV6-AC-GFP, RG201233). Mu-tations in N419I and G501R of p67phox and in R308Q of p40phox

were generated by site-directed mutagenesis using theQuikChange II Site-Directed Mutagenesis Kit according tomanufacturer’s instructions. All construct sequences wereverified using Sanger sequencing. For Western blot analysis,primary antibodies used were mouse monoclonal anti-Myc(Millipore, Billerica, MA), mouse monoclonal anti-GFP (Medi-mabs, Mont Royal, Canada), and polyclonal p47phox (R360,rabbit; from the laboratory of Dr John Brumell). Secondaryantibodies used were goat anti-mouse horseradish peroxidase.Enhanced chemoluminescence (Pierce, Rockford, IL) was usedto visualize protein bands.

Transient Transfection. HEK293T cells were main-tained in Dulbecco’s modified Eagle medium containing 10%heat-inactivated fetal bovine serum and antibiotic-antimycotic(Sigma, St Louis, MO) at 37�C in a 5% CO2 humidified incu-bator. Approximately 1.5 � 106 cells were plated per 100-mmtissue culture dish and grown overnight to achieve 60%–70%confluence. The following day, cells were cotransfected withplasmids containing p67phox, p47phox, or p40phox comple-mentary DNA using the calcium phosphate transient trans-fection method according to manufacturer’s instructions. Sixhours post transfection, the media was replaced with normaltissue culture media. Twenty-four hours post transfection, cellswere lysed using standard protein lysis.

Coimmunoprecipitation StudiesCells lysates were incubated with 2 mg anti-Myc (Millipore)

or anti-GFP (Medimabs) antibody for 1 hour at 4�C. Forty mi-croliters of Protein G-agarose (50% mixture) (Sigma) wasadded to this antibody-lysate solution and incubated for anadditional hour at 4�C. The beads were isolated by centrifu-gation and washed 4 times with lysis buffer. For Western blotanalysis, beads were boiled at 95�C for 5 minutes in 5� sodiumdodecyl sulfate sample buffer, centrifuged, and then loaded intothe gel. All densitometry measurements to indicate proteinexpression were measured using ImageJ software. Electronicimages of Western blots were analyzed; band intensity levels ofbinding partners, as measured by ImageJ, were normalized tothat of the immunoprecipitated protein.

Nitroblue Tetrazolium. Venous blood was collected inpotassium-EDTA tubes where polymorphonuclear cells were iso-lated and purified using PolymorphPrep (CosmoBio USA, Carls-bad, CA), according to the manufacturer’s instructions. The NBTcolorimetric assay was performed by aliquoting 1.5 � 106 poly-morphonuclear cells per well of a 24-well plate in 500 mL Hank’sBalanced Salt Solution. Cells were allowed to adhere at 37�C for30 minutes. Media were replaced with medium with each stim-ulant: phorbol myristate acetate A (P8139, 1 mM; Sigma), ZymosanA (Z4250, 20 mg/mL; Sigma), and diphenyleneiodonium chloride(D2926, 10 mM; Sigma). Three hundred microliters of NBT solu-tion (Sigma; N5514, 10-mg tablet suspended in Hank’s Balanced

Tab

le1.Rep

licated

SNPsIden

tified

byTa

rgeted

Exo

meSeq

uenc

ing

Gen

e(protein)

SNPna

me

Res

ultin

gfunc

tiona

lcha

nge

MAF,

%Disea

setype

Gen

etic

mod

el

Disco

very

coho

rtRep

licationco

hort

Com

bined

analys

is

Pva

lue

OR

(95%

CI)

Pva

lue

OR

(95%

CI)

PValue

OR

(95%

CI)

RAC1

(RAC1)

rs35

7618

91

Intron

icStron

gGATA

bindingsite

23

6.0

VEOIBD

Dom

inan

t1.11

�10

�2

1.9(1.2–3.2)

8.7�

10�1

0.95

(0.5–1.8)

7.4�

10�2

1.4(0.9–2.1)

Red

uces

ROS

VEOUC

Dom

inan

t2.13

�10

�3

2.7(1.4–5.2)

7.8�

10�1

1.1(0.5–2.5)

1.7�

10�2

1.8(1.1–2.8)

NCF2

(p67

phox )

rs35

0125

21

Cod

ing

N41

9IRed

uces

bindingto

NCF4

<1

VEOUC

Dom

inan

t9.43

�10

�3

5.3(1.3–21

.2)

3.2�

10�1

2.8(0.3–23

)6.89

�10

�3

4.3(1.4–13

)

CYBA

(p22

phox )

rs72

5507

04

Promoter

SP1site

loss

23

Red

uces

p22

phoxex

press

ion

<1

VEOCD

Dom

inan

t2.05

�10

�3

5.3(1.6–17

.3)

4.11

�10

�3

7.8(1.5–40

)4.54

�10

�5

5.8(2.2–15

)

CI,co

nfiden

ceinterval;MAF,

minor

allele

freq

uenc

y;OR,od

dsratio

;VEOCD,ve

ryea

rlyon

setCrohn

’sdisea

se.

684 Dhillon et al Gastroenterology Vol. 147, No. 3

BASICAND

TRANSLATIONALAT

Salt Solution at 1 mg/mL) was added and the total mixture wasincubated at 37�C in 5% CO2 humidified incubator for 30 minutes.Cells were washed and fixed in 2.5% paraformaldehyde fixativeand then lysed and solubilized in 240 uL 2M KOH and 280 uLdimethyl sulfoxide. Absorbance was then measured at 570 nm.

Data Sources Measurement. Phenotypic informationand DNA samples were obtained from study subjects withapproval of the Institutional Review Ethics Board for IBDgenetic studies at The Hospital for Sick Children in Toronto.Replication cohorts had Ethics Board approval for genetic andphenotypic studies at the individual institutions. Writteninformed consent was obtained from all participants. Theaffected subjects were recruited from International Early OnsetPediatric IBD Cohort Study sites and healthy controls wereobtained from the Centre for Applied Genomics.

Statistical MethodsAssociation analyses of the discovery and replication co-

horts were used to test associations of the NADPH oxidase SNPsand variants with VEOIBD, very early onset Crohn’s disease,and very early onset ulcerative colitis (VEOUC) vs healthycontrols. Logistic regression analysis was applied for an addi-tive model and Pearson c2 tests were applied for dominant andrecessive models. This analysis was done using the Goldenhelix(SVS 7.6.4) program. The combined cohort was analyzed usingthe same protocol with pooled population data from bothcohorts. For densitometry measurement statistics, the Bonfer-roni post-test was used with n ¼ 3, with each full experimentrepresenting 3 triplicate experiments.

ResultsTargeted Exome Sequencing

To capture rare and novel variants, we carried out tar-geted exome sequencing of the NADPH oxidase genes (CYBA,CYBB, NCF1, NCF2, NCF4, RAC1, RAC2; Figure 1) in 122VEOIBD patients without any clinical or laboratory evidenceof CGD (as described in Methods and reported according tothe Strengthening the Reporting of Genetic Associationsguidelines16). All identified NADPH oxidase mutations wereheterozygous and validated using both Sanger sequencingand Taqman and compared with 480 healthy control in-dividuals and replicated in an independent replication cohort.

We first examined 3 SNPs that were associated withVEOIBD in the targeted exome sequencing discovery cohortand in the combined analysis (Table 1 and Figure 3). Thisincluded an intronic RAC1 SNP (rs35761891) conferring astronger GATA-1 transcription-factor binding site that wasassociated with both VEOIBD and VEOUC. An exonic NCF2SNP (rs35012521) resulting in p67phox N419I variant wasassociated with VEOUC (Pcombined ¼ 6.89 � 10�3; oddsratio ¼ 4.3; 95% confidence interval: 1.4–13). And a pro-moter SNP in CYBA (rs72550704) causing a loss of anSP1 transcription factor binding site23 implicated in up-regulation of gene expression28 was associated with veryearly onset Crohn’s disease (Pcombined ¼ 4.54 � 10�5; oddsratio ¼ 5.8; 95% confidence interval: 2.2–15).

InourVEOIBDdiscovery targetedexomesequencing cohort,we also identified 8 variants (5 SNPs and 3 novel variants)predicted to result in NADPH oxidase complex dysfunction that

Figure 3. Forest plot ofNADPH oxidase SNPassociated with VEOIBD.Forest plot of the oddsratios of associated vari-ants with VEOIBD pheno-types. For each variant,the effect allele (EA), minorallele frequency (MAF), Pvalue, and odds ratio withthe 95% confidence inter-val is shown.

September 2014 VEOIBD and the NOX2 NADPH Oxidase Genes 685

BASICAN

DTR

ANSLAT

IONA

LAT

were not observed in the 1560 genotyped healthy controls(Table 2 and Figure 3). This included a coding R90H p47phox

variant (encoded by NCF1; rs3447) that was previously shownby Olsson et al in rheumatoid arthritis patients to significantlydecrease reactive oxygen species (ROS) production.29 In addi-tion, an NCF2 SNP (rs17849502) resulting in a p67phox H389Qvariant previously associated with systemic lupus erythema-tosus,30 was found in 14 VEOIBD patients. This p67phox H389Qvariant was previously shown to reduce ROS production andwas suggested to reduce p67phox binding to VAV1, an importantprocess for FcyR-dependent ROS production.30

Tagged Single Nucleotide PolymorphismGenotyping

In order to examine intronic SNPs not identified in thetargeted exome sequencing approach, a subanalysis exam-ining a subset of VEOIBD previously genotyped for Tagged

Table 2.SNPs and Novel Variants Identified by Targeted Exom

Gene (protein)Amino acid change,

SNP name Protein domain

CYBB (gp91phox) G364R,rs141756032

Ferric reductase

NCF1 (p47phox) R90H,rs13447

PX domain

NCF2 (p67phox) R38Q,rs147415774

TPR1 domain

NCF2 (p67phox) H389Q,rs17849502

PB1 domain

NCF2 (p67phox) G501R,Novel

Between PPRand SH3 2 domains

NCF2 (p67phox) P454S,rs55761650

Exon 15

NCF4 (p40phox) R308Q,rs141160114

PB1 domain

RAC2 (RAC2) Noncodingchromosome

22, 37627176, (C/T)

Intron 5

hnRNP, heterogeneous nuclear ribonucleoproteins; MAF, minoraPopulation frequencies refer to data collected by National HearVariant Server18 (http://evs.gs.washington.edu/EVS/), limited to

NADPH oxidase SNPs12 was carried out. We found novelassociations with 2 RAC2 SNPs (rs1476002 and rs1476002),an NCF4 SNP (rs18813313), and an RAC1 SNP(rs10951982) (Table 3 and Figure 3). In the combinedanalysis, all SNPs remained associated with VEOIBD andrs1476002 had the strongest association with VEOUC(Precessive ¼ 7.77 � 10�7; odds ratio ¼ 7.0; 95% confidenceinterval: 2.9–17), suggesting the importance of the RAC2 inVEOUC susceptibility.

Functional StudiesNext we determined if the identified coding variants had

abnormal NADPH oxidase functions. First, we examined theNCF2 variant (rs35012521) that resulted in a p67phox N419Iamino acid change in the PB1 domain. When transfectedinto HEK 293T cells, on coimmunoprecipitation (Figure 4A),the N419I variant located in the PB1 domain showed

e Sequencing

MAFaVEOIBD patients(n ¼ 122), n (%) Defect

0.005% (males) 1 (0.8) Predicted damaging21

5.5%30 12 (9.8) Reduces ROS production30

Novel, no data 1112 (9.0) Reduces binding toRac212

5.0% 14 (11.5) Reduces binding to Vav131

Novel, no data 1 (0.8) Predicted damaging21

decrease in p47phox

binding0.008% 1 (0.8) Exonic splicing enhancer

site lost23

0.0001% 1 (0.8) Predicted damaging21

decrease in p67phox bindingNovel, no data 1 (0.8) hnRNP A1 site gain24

splicing dysregulation

allele frequency.t, Lung and Blood Institute Exome Sequencing Project ExomeEuropean Americans.

Tab

le3.Rep

licated

NADPH

Oxidas

eGen

esSNPsAss

ociatedWith

VEOIBD

Iden

tified

byTa

gSNPGen

otyp

ing

Gen

e(protein)

SNPna

me

MAF,

%Ass

ociatio

nMod

el

Disco

very

coho

rtRep

licationco

hort

Com

bined

analys

is

Pva

lue

OR

(95%

CI)

Pva

lue

OR

(95%

CI)

Pva

lue

OR

(95%

CI)

RAC1(RAC1)

rs10

9519

8224

VEOIBD

Additive

1.21

�10

�2

0.67

(0.5–0.9)

4.14

�10

�2

0.73

(0.5–0.99

)2.12

�10

�3

0.71

(0.57–

0.89

)RAC2(RAC2)

rs14

7600

213

VEOIBD

Rec

essive

2.71

�10

�3

4.6(1.7–12

.6)

1.6�

10�1

2.5(0.67–

9.56

)2.16

�10

�4

3.9(1.8–8.6)

13VEOUC

Rec

essive

2.85

�10

�4

8.0(2.6–24

.5)

9.0�

10�3

5.7(1.31–

24.4)

7.77

�10

�7

7.0(2.9–17

)RAC2(RAC2)

rs73

9041

42VEOIBD

Additive

1.13

�10

�2

0.8(0.6–0.9)

2.06

�10

�1

0.8(0.6–1.1)

5.34

�10

�3

0.8(0.6–0.9)

NCF4

(p40

phox )

rs18

8311

37

VEOIBD

Rec

essive

7.22

�10

�3

7.9(1.8–35

.6)

NA

NA

7.65

�10

�3

6.0(1.3–27

)

CI,co

nfiden

ceinterval;MAF,

minor

allele

freq

uenc

y;NA,no

tap

plicab

le;OR,od

dsratio

.

686 Dhillon et al Gastroenterology Vol. 147, No. 3

BASICAND

TRANSLATIONALAT

reduced binding to p40phox (encoded by NCF4) comparedwith the wild-type p67phox, suggesting the importance ofp40phox mediated regulation of p67phox in the formation ofthe NADPH complex (Figure 1).

Similar to NCF2 variants, the R308Q p40phox (encoded byNCF4) variant (rs147415774) also located in the PB1domain responsible for p40phox�p67phox binding, showedsimilarly reduced binding to p67phox (encoded by NCF2),further supporting the importance of this interaction inproper NADPH complex formation (Figure 4A). Finally, theG501R p67phox (encoded by NCF2) novel variant waslocated in the SH3 domain, which is responsible for inter-action with p47phox, was found in 1 patient from the tar-geted exome sequencing discover cohort. We found thatthe G501R p67pho variant reduced binding betweenp47phox�p67phox (Figure 4B), demonstrating the overallimportance of protein–protein interactions in NADPH com-plex formation in VEOIBD (Figure 1).

In addition to protein-binding defects, we also identifiedvariants in the noncoding regulatory and splicing elements ofthe NADPH oxidase genes in a number of VEOIBD patients.The CYBA (encoding p22phox) variant (rs72550704) that wasfound to be associated with very early onset Crohn’s diseaseresulted in a loss of an SP1 binding site.23 This variant wasfound to reduce p22phox protein expression in Patient 1 (aheterozygous carry of the variant), who also had reducedROS production (Supplementary Table 1 and SupplementaryFigure 1A), although it was not directly shown that thereduction in p22phox protein expression resulted in subse-quent ROS reduction. In this particular case, both the patientand mother carried the functional variant and therefore othergenetic or environmental factors must also play a role insusceptibility to disease.

In addition, a novel variant in RAC2 (chromosome 22,37627176 [C/T]) that was predicted to alter splicing regu-lation of RAC2 through the addition of a heterogeneousnuclear ribonucleoprotein A1 site was also identified in apatient who developed steroid-dependent colitis (a hetero-zygous carry of the variant). We determined that this novelintronic variant in RAC2 led to the 5 new alternative splicingforms of the RAC2 messenger RNA (SupplementaryFigure 1B, C), resulting in loss of exons and truncationsthat were not seen in control samples lacking this variant.Overall, these altered splicing forms resulted in decreasedRAC2 protein expression, as seen in Western blot analysis(Supplementary Figure 1B).

Variants That Result in Reduced Reactive OxygenSpecies Production

To determine if the NADPH oxidase variants resulted infunctional defects, we retrospectively examined clinicalNADPH oxidase ROS production and identified only 5 of the122 targeted exome sequencing VEOIBD patients who hadprevious ROS measured by NBT for clinical purposes. The 5patients with identified NADPH oxidase variants all hadreduced ROS production, although within normal limits(Table 3). Four patients had heterozygous NADPH oxidasevariants and 1 patient (Patient 4) had compound

Figure 4. Functional studies of rare and private mutation in the NADPH oxidase genes associated with VEOIBD. (A) HEK293Tcells were cotransfected with wild-type and variant p67phox-myc and p40phox-GFP constructs (p67phox-WT, p67phox-N419I,p40phox-WT, and p40phox-R308Q) p40phox-GFP was immunoprecipitated from cell lysates using anti-GFP primary antibody.Levels of p67phox-myc bound to p40phox-GFP were determined by Western blot using anti-myc antibodies. Densitometry ofimmunoprecipitated p67phox-myc was normalized to total levels of immunoprecipitated p40phox-GFP. n ¼ 3 experimentalreplicates; ***P < .05 as calculated by the Bonferroni post-test. (B) HEK293T cells were cotransfected with wild-type andvariant p67phox-myc and p47phox-GFP constructs (p67phox-WT, p67phox-G501R and p40phox-WT) p67phox-myc was immu-noprecipitated from cell lysates using anti-myc primary antibody. Levels of p47phox-GFP bound to p67phox-myc were deter-mined by Western blot using anti-GFP. Densitometry of immunoprecipitated p47phox-GFP was normalized to total levels ofimmunoprecipitated p67phox-myc. n ¼ 3 experimental replicates; ***P < .05 as calculated by the Bonferroni post-test.

September 2014 VEOIBD and the NOX2 NADPH Oxidase Genes 687

BASICAN

DTR

ANSLAT

IONA

LAT

heterozygous NCF2 variants (encoding p67phox H389Q andN419I) and was diagnosed at 2 years of age with VEOIBD. Inthis patient, known NADPH oxidase–derived ROS-dependentfunctions,31 including colorimetric NBT and interleukin-8–and N-formyl-methionine-leucine-phenylalanine–inducedchemotaxis, were impaired (data not shown).

DiscussionOverall, through targeted exome sequencing, we identi-

fied heterozygous SNPs and novel variants in the NADPHoxidase complex genes in VEOIBD patients with eitherimpaired ROS production, reduced protein binding, orreduced gene expression. CGD is an X-linked or autosomalrecessive disease in which the vast majority of mutationsare found in NOX2 (Figure 1B), and almost all identifiedmutations are associated with markedly reduced, orabsence of, the protein expression, or rare missense variantsthat result in profound or complete disruption of enzymeactivity. In contrast, only one NOX2 variant was identified inVEOIBD patients, and the majority of those variants identi-fied in the NADPH oxidase genes occurred in binding

domains of the cytosolic components of the complex(p40phox�p67phox�p47phox; Figure 1A and B) that areuncommonly found in CGD (Figure 1B).

It has been previously observed that defective bindingbetween 2 NADPH oxidase complex proteins can alter otherprotein–protein interactions in the complex, as illustrated by astudy demonstrating that defective p67phox binding to RAC2also reduced binding between p67phox�NOX2.32 Therefore,we suggest that the NADPH oxidase variants identified inVEOIBD patients that reduce protein–protein interactionsmight also limit or delay complex formation, leading toreduced ROS production without the complete loss of functionobserved in CGD. Overall, these studies suggest that impairedNADPH oxidase assembly at the membrane can be a criticalstep in the development of colitis in these young patients.

Although these VEOIBD patients share similarities withthe colonic inflammation observed in CGD patients, theVEOIBD patients described here had heterozygous variantsin the autosomal genes of the NADPH oxidase complex and asmall number of patients studied had low-normal ROSproduction, with a minimum of 10% of normal ROS

688 Dhillon et al Gastroenterology Vol. 147, No. 3

BASICAND

TRANSLATIONALAT

production. Therefore, these VEOIBD patients did not haveCGD or suffer from chronic infection or multisystemic dis-ease associated with CGD. Female carriers of X-linked CGDwith pronounced X-linked inactivation (lyonization) whohave only 10% of normal ROS production do not developCGD.33 This also suggests that the levels of ROS productionobserved in VEOIBD patients with NADPH oxidase variantsare above the threshold of ROS production observed in CGDand, therefore, result in susceptibility to VEOIBD and notimmunodeficiency. However, the majority of VEOUC pa-tients with identified functional variants in the NADPH ox-idase pathway had left-sided disease, which is thepredominant location of the colitis observed in CGD pa-tients,7,8 as opposed to the pancolitis, which is morecommonly observed in VEOIBD34 (Supplementary Table 2).Therefore, these studies also suggest that hypomorphicNADPH oxidase functional variants increase the risk fordeveloping VEOIBD and the colonic inflammation is similarto that seen in the more severe immunodeficiency CGD.

Growing evidence suggests the importance of defectiveinnate immunity in IBD pathogenesis,35–37 including defec-tive neutrophil functions, such as reduced superoxide pro-duction,38 decreased recruitment,39 and impaired monocytefunctions.35 Primary immunodeficiencies (PIDs), includingCGD, are typically characterized by pediatric onset,35 and thismight explain the higher degree of association of functionalNADPH oxidase genes variants observed in our VEOIBDcohorts. The role of PID genes in VEOIBD is exemplified ininfants who develop colitis before 1 year of age (infantileVEOIBD), where mutations in IL10RA/B were found in pa-tientswith a distinct phenotype of severe colitis with perianaldisease and folliculitis17,40 and, more recently, mutations inTTC7A in infantile VEOIBD patients with apoptotic entero-colitis.41 Similar to the results presented here, variants inIL10RA have been found in higher prevalence in VEOUCwithout evidence of PID,17 suggesting that hypomorphicvariants in PID genes that do not result in overt immunode-ficiency can result in susceptibility to VEOIBD.

Interestingly, mirroring the binding defects describedhere in VEOIBD, changes in the assembly of the NADPHoxidase complex have been implicated in Mendelian sus-ceptibility to mycobacterial disease (MSMD), resulting inselectively reduced oxidative burst in macrophages.42

MSMD is a disease in which patients cannot mount resis-tance to mild strains of mycobacterial microbes and,although traditionally defective mutations in NOX2 (encodedby CYBB) cause CGD, in a rare MSMD case, 2 functionalcoding variants of NOX2 resulted in a decrease inmacrophage-specific NADPH oxidase complex formation.42

This further suggests the importance of NADPH oxidasegene variants in diseases less severe than CGD. In addition,p47phox (encoded by NCF1) has been genetically linked withrheumatoid arthritis through genetic association as well asanimal model studies.29,43 In addition, the H389Q codingvariant of p67phox (encoded by NCF2), also identified in thisstudy, was recently associated with systemic lupus erythe-matosus.30 These studies suggest that mutations in NOX2results in immunodeficiency (CGD or MSMD), and functionalvariants in other components of the NADPH oxidase

complex contribute to susceptibility to immune-relateddiseases, including a significant role in the development ofVEOIBD in a large cohort.

Supplementary MaterialNote: To access the supplementary material accompanyingthis article, visit the online version of Gastroenterology atwww.gastrojournal.org, and at http://dx.doi.org/10.1053/j.gastro.2014.06.005.

References

1. Marks DJ, Miyagi K, Rahman FZ, et al. Inflammatory

bowel disease in CGD reproduces the clinicopathologicalfeatures of Crohn’s disease. Am J Gastroenterol 2009;104:117–124.

2. Schappi MG, Smith VV, Goldblatt D, et al. Colitis inchronic granulomatous disease. Arch Dis Child 2001;84:147–151.

3. Werlin SL, Chusid MJ, Caya J, et al. Colitis in chronic gran-ulomatous disease. Gastroenterology 1982;82:328–331.

4. Muise AM, Snapper SB, Kugathasan S. The age of genediscovery in very early onset inflammatory bowel dis-ease. Gastroenterology 2012;143:285–288.

5. Levine A, Griffiths A, Markowitz J, et al. Pediatric modi-fication of the Montreal classification for inflammatorybowel disease: the Paris classification. Inflamm BowelDis 2011;17:1314–1321.

6. Benchimol EI, Guttmann A, Griffiths AM, et al. Increasingincidence of paediatric inflammatory bowel disease inOntario, Canada: evidence from health administrativedata. Gut 2009;58:1490–1497.

7. Huang JS, Noack D, Rae J, et al. Chronic granulomatousdisease caused by a deficiency in p47(phox) mimickingCrohn’sdisease.ClinGastroenterolHepatol2004;2:690–695.

8. Matute JD, Arias AA, Wright NA, et al. A new geneticsubgroup of chronic granulomatous disease with auto-somal recessive mutations in p40 phox and selectivedefects in neutrophil NADPH oxidase activity. Blood2009;114:3309–3315.

9. Jostins L, Ripke S, Weersma RK, et al. Host-microbeinteractions have shaped the genetic architecture of in-flammatory bowel disease. Nature 2012;491:119–124.

10. Franke A, McGovern DP, Barrett JC, et al. Genome-wide meta-analysis increases to 71 the number ofconfirmed Crohn’s disease susceptibility loci. Nat Genet2010;42:1118–1125.

11. Anderson CA, Boucher G, Lees CW, et al. Meta-analysisidentifies 29 additional ulcerative colitis risk loci,increasing the number of confirmed associations to 47.Nat Genet 2011;43:246–252.

12. Muise AM, Xu W, Guo CH, et al. NADPH oxidase com-plex and IBD candidate gene studies: identification of arare variant in NCF2 that results in reduced binding toRAC2. Gut 2012;61:1028–1035.

13. Rioux JD, Xavier RJ, Taylor KD, et al. Genome-wide as-sociation study identifies new susceptibility loci forCrohn disease and implicates autophagy in diseasepathogenesis. Nat Genet 2007;39:596–604.

September 2014 VEOIBD and the NOX2 NADPH Oxidase Genes 689

BASICAN

DTR

ANSLAT

IONA

LAT

14. Roberts RL, Hollis-Moffatt JE, Gearry RB, et al. Confir-mation of association of IRGM and NCF4 with ilealCrohn’s disease in a population-based cohort. GenesImmun 2008;9:561–565.

15. Muise AM, Walters T, Xu W, et al. Single nucleotidepolymorphisms that increase expression of the guano-sine triphosphatase RAC1 are associated with ulcerativecolitis. Gastroenterology 2011;141:633–641.

16. Thoeni C, Amir A, Guo C, et al. A novel nonsense mutationin the EpCAM gene in a patient with congenital tuftingenteropathy. J Pediatr Gastroenterol Nutr 2014;58:18–21.

17. Moran CJ, Walters TD, Guo CH, et al. IL-10R poly-morphisms are associated with very-early-onset ulcera-tive colitis. Inflamm Bowel Dis 2013;19:115–123.

18. Exome Sequencing Project. NGESP. Exome VariantServer. Seattle, WA: University of Washington, 2013.

19. dbSNP Short Genetic Variants. Volume 2013. Bethesda,MD: National Center for Biotechnology Information, 2013.

20. SNP Function Prediction (FuncPred). Volume 2013. Tri-angle, NC: National Institute of Environmental HealthSciences, 2013.

21. Adzhubei IA, Schmidt S, Peshkin L, et al. A method andserver for predicting damaging missense mutations. NatMethods 2010;7:248–249.

22. Kumar P, Henikoff S, Ng PC. Predicting the effects ofcoding non-synonymous variants on protein functionusing the SIFT algorithm. Nat Protoc 2009;4:1073–1081.

23. Yuan HY, Chiou JJ, Tseng WH, et al. FASTSNP: an al-ways up-to-date and extendable service for SNP func-tion analysis and prioritization. Nucleic Acids Res 2006;34:W635–W641.

24. Desmet FO, Hamroun D, Lalande M, et al. HumanSplicing Finder: an online bioinformatics tool to predictsplicing signals. Nucleic Acids Res 2009;37:e67.

25. Wang J, Ronaghi M, Chong SS, et al. pfSNP: an inte-grated potentially functional SNP resource that facilitateshypotheses generation through knowledge syntheses.Hum Mutat 2011;32:19–24.

26. Levine A, Kugathasan S, Annese V, et al. Pediatric onsetCrohn’s colitis is characterized by genotype-dependentage-related susceptibility. Inflamm Bowel Dis 2007;13:1509–1515.

27. Tyler AD, Milgrom R, Stempak JM, et al. The NOD2insCpolymorphism is associatedwithworse outcome followingileal pouch-anal anastomosis for ulcerative colitis. Gut2013;62:1433–1439.

28. Li L, He S, Sun JM, et al. Gene regulation by Sp1 andSp3. Biochem Cell Biol 2004;82:460–471.

29. Olsson LM, Nerstedt A, Lindqvist A-K, et al. Copy num-ber variation of the gene NCF1 is associated with rheu-matoid arthritis. Antioxid Redox Signal 2012;16:71–78.

30. Jacob CO, Eisenstein M, Dinauer MC, et al. Lupus-associated causal mutation in neutrophil cytosolic factor2 (NCF2) brings unique insights to the structure andfunction of NADPH oxidase. Proc Natl Acad Sci U S A2012;109:E59–E67.

31. Hattori H, Subramanian KK, Sakai J, et al. Small-molecule screen identifies reactive oxygen species askey regulators of neutrophil chemotaxis. Proc Natl AcadSci U S A 2010;107:3456–3551.

32. Diebold BA, Bokoch GM. Molecular basis for Rac2regulation of phagocyte NADPH oxidase. Nat Immunol2001;2:211–215.

33. Lekstrom-Himes JA, Gallin JI. Immunodeficiency dis-eases caused by defects in phagocytes. N Engl J Med2000;343:1703–1714.

34. Griffiths AM. Specificities of inflammatory bowel diseasein childhood. Best Pract Res Clin Gastroenterol 2004;18:509–523.

35. Vinh DC, Behr MA. Crohn’s as an immune deficiency:from apparent paradox to evolving paradigm. Expert RevClin Immunol 2012;9:17–30.

36. Khor B, Gardet A, Xavier RJ. Genetics and pathogenesisof inflammatory bowel disease. Nature 2011;474:307–317.

37. Rakoff-Nahoum S, Bousvaros A. Innate and adaptiveimmune connections in inflammatory bowel diseases.Curr Opin Gastroenterol 2010;26:572–577.

38. Hayee B, Rahman FZ, Tempero J, et al. The neutrophilrespiratory burst and bacterial digestion in Crohn’s dis-ease. Dig Dis Sci 2011;56:1482–1488.

39. Marks DJ, Harbord MW, MacAllister R, et al. Defectiveacute inflammation in Crohn’s disease: a clinical inves-tigation. Lancet 2006;367:668–678.

40. Glocker EO, Kotlarz D, Boztug K, et al. Inflammatorybowel disease and mutations affecting the interleukin-10receptor. N Engl J Med 2009;361:2033–2045.

41. Avitzur Y, Guo C, Mastropaolo LA, et al. Mutations intetratricopeptide repeat domain 7A result in a severeform of very early onset inflammatory bowel disease.Gastroenterology 2014;146:1028–1039.

42. Bustamante J, Arias AA, Vogt G, et al. Germline CYBBmutations that selectively affect macrophages in kin-dreds with X-linked predisposition to tuberculousmycobacterial disease. Nat Immunol 2011;12:213–221.

43. Olofsson P, Holmberg J, Tordsson J, et al. Positionalidentification of Ncf1 as a gene that regulates arthritisseverity in rats. Nat Genet 2003;33:25–32.

Author names in bold designate shared co-first authorship.

Received January 2, 2014. Accepted June 3, 2014.

Reprint requestsAddress requests for reprints to: Aleixo M. Muise, MD, PhD, The Hospital forSick Children, 555 University Avenue, Toronto, Ontario, Canada, M5G 1X8.e-mail: aleixo.muise@utoronto.ca; fax: 416-813-6531.

AcknowledgmentsThe authors thank the all patients and their families described here.The following authors were part of the International Early Onset Pediatrics IBD

Cohort Study (www.NEOPICS.org): Sandeep S. Dhillon, Ramzi Fattouh, AbdulElkadri, Ryan Murchie, Thomas Walters, Conghui Guo, David Mack, Hien Q.Huynh, Anne M. Griffiths, Scott B. Snapper, John H. Brumell, and Aleixo M.Muise.

Conflicts of interestThe authors disclose no conflicts.

FundingThis project was funded by operating grants from the Canadian Institutes of HealthResearch to AMM and JHB (MOP97756 and MOP119457). AMM and SS aresupported by Leona M. and Harry B. Helmsley Charitable Trust to study very earlyonset inflammatory bowel disease. JHB held an Investigator in Pathogenesis ofInfectious Disease Award from the Burroughs Wellcome Fund. Infrastructure forthe Brumell Laboratory was provided by a New Opportunities grant from theCanadian Foundation for Innovation and the Ontario Innovation Trust. SBS issupported by National Institutes of Health grants HL59561, DK034854, andAI50950 and the Wolpow Family Chair in IBD Treatment and Research.

Supplementary Figure 1. Functional variants conferring decreased NADPH oxidase gene expression. (A) Densitometry ofWestern blot of p22phox variant (encoded by CYBA; SNP rs72550704), which was associated with Crohn’s disease diagnosisat younger than the age of 6 years and resulted in lower p22phox expression. Both the patient and her mother were hetero-zygous for this variant. (B) Densitometry of Western blot of Rac2 for a novel variant gaining an intronic heterogeneous nuclearribonucleoprotein (hnRNP) A1 (noncoding chromosome 22, 37627176 [C/T]) disrupting splicing regulation, which decreasedoverall expression levels. (C) Splicing variants of the RAC2 gene for the novel variant found in a single patient conferring a gainof an intronic hnRNP A1 site, disrupting splicing regulation. To assess for RAC2 splicing changes, patient and healthy controlmessenger RNA (approximately 1 ug) were reverse transcribed (random decamers [5 uM], 1� RT buffer, deoxynucleosidetriphosphate solution [0.5 mM], RNase inhibitor [10 U], and reverse transcriptase [100 U]) by incubating the solution at 44�C for1 hour and then inactivated at 92�C for 10 minutes. Thereafter, complementary DNA was amplified by polymerase chainreaction (PCR) using primers specific to RAC2 and then visualized on a 1% agarose gel. PCR products were cloned intoplasmids using the CloneJET PCR Cloning Kit, as per manufacturer’s instructions, before sending to sequencing. (C) RAC2splice variants complementary DNA: variant 1: deleted parts of exon 3–7 (244–294, 1146–1439); variant 2: deleted part of exon7 (244–294, 1146–1439); variant 3: deleted parts of exons 3–7 (244–300, 1050–1439); variant 4: deleted part of Exon 7(445–809, 1275–1439); and variant 5: deleted parts of exons 5 and 6 (244–534, 1052–1439).

689.e1 Dhillon et al Gastroenterology Vol. 147, No. 3

Supplementary Table 1.NADPH Oxidase Variants Found in Patients With Low-Normal ROS Production

Patient no./NOBI SNP Gene (protein)Heterozygous variant/

functional defect Clinical features

1/14.7 and 19.1 on2 occasions

rs72550704 CYBA(p22phox)

Promoter variantreduces p22phox expression

Dx at age 3 years with pancolitis VEOIBD

2/34 rs13447 NCF1(p47phox)

R90HPX domainreduces ROS production in

transfected COS cells30

Dx at age 19 months with pancolitis VEOIBD

3/40 and 53 on2 occasions

rs147415774 NCF2(p67phox)

R38Qreduces binding to Rac212

Dx at age 9 months with pancolitis VEOIBD

4/53 rs17849502 NCF2(p67phox)

H389Qreduces binding to Vav131

Dx at age 2 years with pancolitis VEOIBDLoop ileostomy at age 6 years and end

ileostomy at age 10rs35012521 NCF2

(p67phox)N419Ireduces binding to NCF4

5/48 rs35761891 RAC1(RAC1)

GATA binding site lost Dx at age 6 years with pancolitis VEOIBD

Dx, diagnosis; NOBI, neutrophil oxidative burst index (NOBI was carried out at The Hospital for Sick Children diagnosticlaboratory. Normal value defined as 32 to 300); COS, CV-1 (simian) in Origin.

Supplementary Table 2.Association of Presence of Functional Variants Found in the Targeted Exome Sequencing ProjectWith Crohn’s Disease and Ulcerative Colitis Disease Locations

Paris classification categoryPhenotype

present or absentWithout functional

variant, nWith functional

variant, n OR (95% CI) P value

L1: isolated ileal disease Absent 52 25 1.30 (0.39–4.38) .75Present 8 5

L2: colonic disease Absent 15 12 0.49 (0.19–1.25) .14Present 46 18

L3: ileocolonic disease Absent 57 27 1.03 (0.25–4.30) .68Present 4 3

L4a: proximal to ligament of Treitz Absent 54 25 1.54 (0.45–5.34) .52Present 7 5

L4b: distal to ligament of Treitz Absent 52 25 1.39 (0.44–4.326) .76Present 9 6

E1: ulcerative proctitis Absent 65 28 NA .58Present 2 0

E2: left-sided UC Absent 57 16 4.28 (1.56–11.69) .0047Present 10 12

E3: extensive disease Absent 60 25 NA 1.00Present 7 3

E4: pancolitis Absent 19 15 0.34 (0.138–0.85) .033Present 48 13

CI, confidence interval; NA, not applicable; OR, odds ratio; UC, ulcerative colitis.All disease locations are based on the Paris Classification systems with L1–L4b designating Crohn’s disease locations andE1–E4 designating locations for ulcerative colitis. Association was tested based on absence compared with presence offunctional and rare variants listed in Tables 1 and 2 using a c2 association test.

September 2014 VEOIBD and the NOX2 NADPH Oxidase Genes 689.e2