genome-wide pathway analysis identifies genetic pathways

ORIGINAL ARTICLE

Genome-Wide Pathway Analysis IdentifiesGenetic Pathways Associated with Psoriasis
Adria Aterido1, Antonio Julia1, Carlos Ferrandiz2, Lluıs Puig3, Eduardo Fonseca4,Emilia Fernandez-Lopez5, Esteban Dauden6, Jose Luıs Sanchez-Carazo7, Jose Luıs Lopez-Estebaranz8,David Moreno-Ramırez9, Francisco Vanaclocha10, Enrique Herrera11, Pablo de la Cueva12,Nick Dand13, Nuria Palau1, Arnald Alonso1, Marıa Lopez-Lasanta1, Raul Tortosa1,Andres Garcıa-Montero14, Laia Codo15, Josep Lluıs Gelpı15, Jaume Bertranpetit16, Devin Absher17,Francesca Capon13, Richard M. Myers17, Jonathan N. Barker13,18 and Sara Marsal1
Psoriasis is a chronic inflammatory disease with a complex genetic architecture. To date, the psoriasis herita-bility is only partially explained. However, there is increasing evidence that the missing heritability in psoriasiscould be explained by multiple genetic variants of low effect size from common genetic pathways. Theobjective of this study was to identify new genetic variation associated with psoriasis risk at the pathway level.We genotyped 598,258 single nucleotide polymorphisms in a discovery cohort of 2,281 case-control individualsfrom Spain. We performed a genome-wide pathway analysis using 1,053 reference biological pathways. A totalof 14 genetic pathways (PFDR � 2.55 � 10e2) were found to be significantly associated with psoriasis risk. Usingan independent validation cohort of 7,353 individuals from the UK, a total of 6 genetic pathways were signif-icantly replicated (PFDR � 3.46 � 10e2). We found genetic pathways that had not been previously associated withpsoriasis risk such as retinol metabolism (Pcombined ¼ 1.84 � 10e4), the transport of inorganic ions and aminoacids (Pcombined ¼ 1.57 � 10e7), and post-translational protein modification (Pcombined ¼ 1.57 � 10e7). In the latterpathway, MGAT5 showed a strong network centrality, and its association with psoriasis risk was further vali-dated in an additional case-control cohort of 3,429 individuals (P < 0.05). These findings provide insights intothe biological mechanisms associated with psoriasis susceptibility.

Journal of Investigative Dermatology (2016) 136, 593e602; doi:10.1016/j.jid.2015.11.026

1Rheumatology Research Group, Vall d’Hebron Research Institute,Barcelona, Spain; 2Dermatology Department, Hospital UniversitariGermans Trias i Pujol, Badalona, Barcelona, Spain; 3DermatologyDepartment, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain;4Dermatology Department, Complejo Hospitalario Universitario de ACoruna, A Coruna, Spain; 5Dermatology Department, HospitalUniversitario de Salamanca, Salamanca, Spain; 6Dermatology Department,Hospital Universitario La Princesa, Madrid, Spain; 7DermatologyDepartment, Hospital General Universitario de Valencia, Valencia, Spain;8Dermatology Department, Hospital Universitario Fundacion Alcorcon,Madrid, Spain; 9Dermatology Department, Hospital Universitario VirgenMacarena, Sevilla, Spain; 10Dermatology Department, HospitalUniversitario 12 de Octubre, Madrid, Spain; 11Dermatology Department,Hospital Universitario Virgen de la Victoria, Malaga, Spain; 12DermatologyDepartment, Hospital Universitario Infanta Leonor, Madrid, Spain;13Division of Genetics and Molecular Medicine, King’s College LondonSchool of Medicine, Guy’s Hospital, London, UK; 14Spanish National DNABank, Universidad de Salamanca, Salamanca, Spain; 15Life Sciences,Barcelona Supercomputing Centre, Barcelona, Spain; 16Spanish NationalGenotyping Centre (CeGen), Universitat Pompeu Fabra, Barcelona, Spain;17HudsonAlpha Institute for Biotechnology, Huntsville, Alabama, USA; and18St John’s Institute of Dermatology, King’s College London, London, UK

Correspondence: Antonio Julia, Rheumatology Research Group, Valld’Hebron Research Institute, Barcelona, 08035, Spain. E-mail: [email protected] or Sara Marsal, Rheumatology Research Group, Vall d’HebronResearch Institute, Barcelona, 08035, Spain. E-mail: [email protected]

Abbreviations: BC, betweenness centrality; DC, degree centrality; FDR, falsediscovery rate; GWAS, genome-wide association studies; SNP, singlenucleotide polymorphism

Received 13 July 2015; revised 27October 2015; accepted 12November 2015;accepted manuscript published online 29 December 2015; corrected proofpublished online 23 January 2016

ª 2015 The Authors. Published by Elsevier, Inc. on behalf of the Society for Inv

INTRODUCTIONPsoriasis is a common chronic inflammatory disease of theskin that affects approximately 2% of the worldwide popu-lation (Nestle et al., 2009). In psoriasis, immune cells infil-trate the skin leading to an increased proliferation ofkeratinocytes (Ferenczi et al., 2000; Gudjonsson and Elder,2007). It is a genetically complex disease with a complexmode of inheritance (Vyse and Todd, 1996). HLA class I geneHLA-C*0602 haplotype association explains the largest partof the known heritability of psoriasis (Nair et al., 2006;Strange et al., 2010).

Genome-wide association studies (GWAS) have beensuccessful in the characterization of the genetic architectureof many complex human diseases (Manolio, 2010). To date,more than 15 GWAS have been performed using large pso-riasis cohorts from Caucasian and Asian populations andhave collectively identified more than 50 susceptibility locifor psoriasis (Bowes et al., 2015; Tsoi et al., 2015b; Yin et al.,2015; Zuo et al., 2015). Despite progress in characterizingpsoriasis genetic etiology, loci outside the HLA region onlyexplain less than 25% of the estimated psoriasis heritability(Tsoi et al., 2012; Yin et al., 2014).

Recent research has shown that the missing heritability ofcomplex human diseases can be explained by commongenetic variants, rare variants or a combination of genetic,epigenetic, and environmental interactions (Gibson, 2012).From these, common genetic variants could explain more

estigative Dermatology. www.jidonline.org 593

http://dx.doi.org/10.1016/j.jid.2015.11.026

http://crossmark.crossref.org/dialog/?doi=10.1016/j.jid.2015.11.026&domain=pdf

mailto:[email protected]



http://www.jidonline.org

Tab

le1.Pathwaysassociated

withpsoriasis

risk

andvalidated

inthereplicationstage

Pathway

Datab

ase

Gen

esSN

PsD

1PD

FDRD

PDE

FDRDE

SNPsR

1PR

FDRR

PRE

FDRRE

PC

Inflam

matory

response

2Biocarta

29

628

<9.99�

10e8

5.25�

10e5

1.35�

10e2

4.71�

10e2

606

<3.33�

10e7

5.77�

10e7

1.53�

10e2

1.58�

10e2

1.06�

10e12

Naturalkiller

Tcell2

Biocarta

29

638

<9.99�

10e8

5.25�

10e5

1.17�

10e2

4.71�

10e2

603

<3.33�

10e7

5.77�

10e7

1.59�

10e2

1.59�

10e2

1.06�

10e12

DNArepair2

Reactome

112

2,050

1.33�

10e4

9.36�

10e3

ee

1,962

<3.33�

10e7

5.77�

10e7

ee

1.10�

10e9

Aminoacid

tran

sportacross

theplasm

amem

brane

Reactome

31

1,025

2.00�

10e4

1.24�

10e2

ee

993

4.00�

10e5

5.63�

10e5

ee

1.57�

10e7

Post-translational

protein

modification

Reactome

188

5,965

2.00�

10e4

1.24�

10e2

ee

5,725

4.00�

10e5

5.63�

10e5

ee

1.57�

10e7

Tran

sportto

theGolgian

d

subsequen

tmodification

Reactome

33

1,557

3.33�

10e4

1.95�

10e2

ee

1,516

3.20�

10e3

4.38�

10e3

ee

1.57�

10e5

AsparagineN-linked

glycosylation

Reactome

81

2,760

4.00�

10e4

2.11�

10e2

ee

2,639

8.67�

10e3

1.13�

10e2

ee

4.72�

10e5

Tran

sportofinorgan

icions

andam

inoacids

Reactome

94

4,010

4.00�

10e4

2.11�

10e2

ee

3,872

2.00�

10e5

3.25�

10e5

ee

1.57�

10e7

Retinolmetab

olism

KEG

G64

1,512

5.33�

10e4

2.55�

10e2

ee

1,406

2.79�

10e2

3.46�

10e2

ee

1.84�

10e4

Abbreviations:C,co

mbined

;D,disco

very

cohort;E,

exclusionIL12Bgene;

FDR,falsedisco

very

rate;KEG

G,Kyo

toEn

cycloped

iaofGen

esan

dGen

omes;P,

empirical

set-based

P-value;

R,replicationco

hort.

1Number

ofsinglenucleo

tidepolymorphismsmap

pingto

aparticu

larpathway.

2Increasedpermutationsto

refinetheP-value(n

¼10,000,000).

A Aterido et al.Genome-Wide Pathway Analysis in Psoriasis

594

than 60% of the heritability of the most prevalent autoim-mune diseases (Golan et al., 2014). Importantly, most of thesecommon genetic variants are characterized by having loweffect sizes (Park et al., 2010).

Although GWAS based on single markers have successfullyidentified disease-susceptibility variants, this strategy is notadequate to identify genetic variants with low effect sizes thatare genuinely associated with disease risk (Du et al., 2012). Insingle-marker GWAS, a large number of genetic variants aretested for association with a complex trait. To avoid falsepositive results, a stringent genome-wide significantthreshold must be used (Johnson et al., 2010). This conser-vative threshold, however, does not allow the identificationof modest effect risk loci, unless extremely large samplessizes of cases and controls are used (Wang et al., 2010).Importantly, single-marker GWAS consider only the individ-ual effect of each single nucleotide polymorphism and ignorethe joint effect of multiple causal genetic variants as well asthe biological context where disease genes operate (Zhanget al., 2010).

Functionally related genes have been shown to collectivelycontribute to disease susceptibility, including those loci thatdo not reach individually the genome-wide significantthreshold (Zhong et al., 2010). Recently, new methods thatare able to analyze genetic associations at the pathway levelhave been developed (Gui et al., 2011). Pathway-based ap-proaches are robust statistical methodologies that integrategenetic and biological knowledge to test whether sets offunctionally related genes are jointly associated with acomplex trait (Ramanan et al., 2012). Therefore, pathway-based methods increase the statistical power of the associa-tion analysis by reducing the number of association tests thatmust be performed and allow a functional interpretation ofthe results (Wu et al., 2010).

Pathway-based analyses have been recently performed tostudy the genetic basis of cancer subtypes using either selectedcandidate pathways, but also at a genome-wide scale (Chenet al., 2014; Koster et al., 2014). Although the genome-widepathway analysis can have a high computational cost, thisapproach is able to identify novel genetic pathways associatedwith disease risk. The identification of new pathways associ-ated with disease risk could increase the probability to developnew therapeutic strategies in complex diseases such as psori-asis. To date, however, the genome-wide pathway analysisapproach has not been performed in psoriasis.

To gain a better understanding of the genetic risk basisof psoriasis, we performed a genome-wide pathwayanalysis on a large multicenter cohort of patients withpsoriasis. In this study, we analyzed the association of1,053 reference biological pathways using 1,263 patientswith psoriasis and 1,558 controls from Spain. Using anindependent cohort of 2,178 cases and 5,175 controlsfrom the UK, we then performed a validation study of thesignificantly associated pathways in the discovery cohort.With this approach, we identified genetic pathways thathad not been previously associated with psoriasis risksuch as retinol metabolism, transport of inorganic ions andamino acids, and post-translational protein modification.These results provide important insights into the geneticetiology of psoriasis.

Journal of Investigative Dermatology (2016), Volume 136

Figure 1. Gene overlap of genetic pathways associated with psoriasis risk. (a) Heat map representing the percentage of genes that are shared between each

pathway pair. (b) Venn diagram of the overlapping pathways representing the transport of inorganic ions and amino acids process as well as the number of genes

shared between them. (c) Venn diagram of the overlapping pathways representing the post-translational protein modification process as well as the number of

genes shared between them.


RESULTSIdentification of genetic pathways associated withpsoriasis risk

In the discovery stage, the genome-wide pathway analysisidentified a total of 26 genetic pathways significantly asso-ciated with psoriasis risk after multiple test correction (PFDR <0.05, Supplementary Table S1 online). The complete resultsof the genome-wide pathway analysis performed in the dis-covery study are shown in Supplementary Table S2 online.

From the 26 significantly associated pathways, we foundthat 14 pathways included IL12B gene. After HLA-C*0602,IL12B is one of the strongest known genetic risk factors forpsoriasis. To confirm that the observed pathway associationswere the result of the joint effect of multiple genes and notthe result of a single risk locus strongly associated with thedisease, we removed IL12B from these genetic pathways andtested again for association. After extracting IL12B, two ge-netic pathways—“Inflammatory response” and “Natural killerT cell”—remained significantly associated with psoriasis risk(PFDR < 0.05). Consequently, only these two pathways fromthe group containing IL12B gene were selected for replica-tion. Together with the other 12 pathways, a total of 14different genetic pathways were finally tested for validation inthe UK population. Using this independent case-controlcohort we significantly validated the association of 9genetic pathways with psoriasis risk (PFDR < 0.05, Table 1).

Characterization of the genetic pathways associated withpsoriasis risk

To discard the presence of redundant pathways, we evalu-ated the level of gene overlap between all associated path-ways. From the nine validated genetic pathways, we foundthat the “Amino acid transport across the plasma membrane”and “Transport of inorganic ions and amino acids” pathways,

as well as the “Asparagine N-linked glycosylation,” “Trans-port to the Golgi and subsequent modification,” and “Post-translational protein modification” pathways had a highdegree of overlap between them (>95% of shared genes,Figure 1a). Consequently, and to avoid redundancy, only thepathway showing the highest level of significance wasselected to represent each biological process. The “Transportof inorganic ions and amino acids” (Pcombined ¼ 1.57 � 10e7,Figure 1b) and “Post-translational protein modification”(Pcombined ¼ 1.57 � 10e7, Figure 1c) pathways were there-fore selected from each overlapping pathway group. The“Inflammatory response” (Pcombined ¼ 1.06 � 10e12), “Nat-ural killer T cell” (Pcombined ¼ 1.06 � 10e12), “DNA repair”(Pcombined ¼ 1.10 � 10e9), and “Retinol metabolism”(Pcombined ¼ 1.84 � 10e4) pathways did not show a signifi-cant degree of overlap and were therefore considered asindependent biological processes.

Within the final group of six genetic pathways associatedwith disease risk and representing independent biologicalprocesses, we analyzed the association between each partic-ular gene and psoriasis risk (Table 2). We found 37 small-effectgenes that were nominally associated with psoriasis risk bothin the discovery and replication cohorts (P � 1.29 � 10e2,Table 3). The complete list of genetic associations obtainedfrom each genetic pathway is shown in SupplementaryTable S3 online. The linkage disequilibrium pattern betweenthe SNPs mapping to each genetic pathway associated withpsoriasis risk is shown in Supplementary Figure S1 online.

Functional-based networks associated with psoriasis risk

To understand the relevance of each particular gene within thegenetic pathway associated with psoriasis risk, we used bio-logical knowledge to build the associated functional-basednetwork (Figure 2). Using known or predicted functional

www.jidonline.org 595


Table 2. Association results of the top five genes involved in each pathway associated with psoriasis risk

Pathway1 Database SNPD COORD A1 A2 ORD PD GeneD SNPR COORD A1 A2 ORR PR GeneR

Inflammatory response Biocarta rs20541 5:131995964 A G 0.72 4.18 � 10e5 IL13,IL4 rs2965012 1:218786549 A C 0.83 7.56 � 10e4 TGFB2

rs11739623 5:131864152 A G 1.21 1.79 � 10e3 IL5 rs2243123 3:159709651 G A 1.14 1.06 � 10e3 IL12A

rs2799083 1:218581617 G A 1.22 2.82 � 10e3 TGFB2 rs25890 5:131437562 G A 0.88 1.09 � 10e3 CSF2

rs2366408 3:159696099 A C 1.19 3.35 � 10e3 IL12A rs20541 5:131995964 A G 0.86 2.41 � 10e3 IL13,IL4

rs2069837 7:22768027 G A 1.33 3.93 � 10e3 IL6 rs4963517 12:6947800 A G 0.90 2.94 � 10e3 CD4

Natural killer T cell Biocarta rs20541 5:131995964 A G 0.72 4.18 � 10e5 IL4 rs4297265 1:67852335 G A 0.83 4.01 � 10e7 IL12RB2

rs11739623 5:131864152 A G 1.21 1.79 � 10e3 IL5 rs749873 2:136817088 G A 0.84 2.61 � 10e5 CXCR4

rs2799083 1:218581617 G A 1.22 2.82 � 10e3 TGFB2 rs2965012 1:218786549 A C 0.83 7.56 � 10e4 TGFB2

rs2114808 2:137249556 G A 0.81 3.09 � 10e3 CXCR4 rs2243123 3:159709651 G A 1.14 1.06 � 10e3 IL12A

rs2366408 3:159696099 A C 1.19 3.35 � 10e3 IL12A rs25890 5:131437562 G A 0.88 1.09 � 10e3 CSF2

Retinol metabolism KEGG rs2173201 4:100250970 A C 0.77 5.82 � 10e5 ADH1C,ADH1B rs7188923 16:81336356 A G 0.89 1.81 � 10e3 BCMO1

rs4148295 4:70475866 C A 1.23 3.41 � 10e4 UGT2A1 rs10882144 10:94852448 A G 0.87 2.55 � 10e3 CYP26A1

rs17614939 4:70360229 G A 0.78 5.21 � 10e4 UGT2B4 rs4319546 12:57346828 A G 0.89 4.96 � 10e3 RDH16

rs2279345 19:41515702 A G 0.84 2.44 � 10e3 CYP2B6 rs4405788 2:72235688 A G 0.90 5.48 � 10e3 CYP26B1

rs17864686 2:234591339 A G 1.25 3.37 � 10e3 UGT1A8 rs11670760 19:41336795 G A 1.12 5.73 � 10e3 CYP2A6

DNA repair Reactome rs240956 6:111616051 A C 1.46 3.16 � 10e6 REV3L rs458017 6:111696091 G A 1.65 1.40 � 10e13 REV3L

rs20541 5:131995964 A G 0.72 4.18 � 10e5 RAD50 rs2240116 9:35094373 A G 1.36 5.22 � 10e4 FANCG

rs2213178 8:48816716 A G 1.29 6.11 � 10e5 PRKDC rs7099120 10:131015367 A G 1.15 9.51 � 10e4 MGMT

rs2985689 14:50098031 C A 1.28 1.66 � 10e3 POLE2 rs3783819 14:61316264 A G 0.89 1.13 � 10e3 MNAT1

rs1887181 10:131594850 G A 1.46 1.86 � 10e3 MGMT rs11693731 2:58887650 A G 0.89 1.13 � 10e3 FANCL

Post-translational protein

modification

Reactome rs1007108 1:26104973 A G 1.43 2.74 � 10e6 MAN1C1 rs9886302 7:70751484 A G 0.81 7.29 � 10e6 WBSCR17

rs10865331 2:62551472 A G 1.25 7.88 � 10e5 B3GNT2 rs7220464 17:7210836 A C 0.85 2.09 � 10e5 EIF5A

rs3791312 2:135183045 G A 0.71 8.04 � 10e5 MGAT5 rs4528932 3:118941441 A G 1.17 5.28 � 10e5 B4GALT4

rs1495086 8:15378013 A G 0.78 1.02 � 10e4 TUSC3 rs7780461 7:151641016 A G 1.24 8.26 � 10e5 GALNTL5

rs977905 3:5882683 G A 1.24 1.69 � 10e4 EDEM1 rs12262718 10:17343706 A G 1.31 8.68 � 10e5 ST8SIA6

Transport of inorganic ions

and amino acids

Reactome rs12661704 6:111560890 A G 1.60 2.34 � 10e7 SLC16A10 rs12661704 6:111560890 A G 1.42 2.38 � 10e9 SLC16A10

rs10205402 2:40710953 A G 0.77 8.78 � 10e6 SLC8A1 rs2385844 2:220839453 G A 0.85 9.23 � 10e6 SLC4A3

rs532237 20:48467560 G C 1.31 1.09 � 10e4 SLC9A8 rs6012750 20:48430680 A G 0.86 6.81 � 10e5 SLC9A8

rs538385 13:30229665 G A 0.82 6.91 � 10e4 SLC7A1 rs1874361 1:205908186 A C 1.15 1.63 � 10e4 SLC26A9

rs17050441 4:139402774 G A 1.27 1.14 � 10e3 SLC7A11 rs11668878 19:47268373 A C 1.27 1.65 � 10e4 SLC1A5

Abbreviations: A1, minor allele; A2, major allele; COORD, SNP coordinates in build GRCh37/hg19; D, discovery cohort; KEGG, Kyoto Encyclopedia of Genes and Genomes; OR, odds ratio; P, P-value;R, replication cohort.1The detailed description of the “Inflammatory response” and “Natural killer T cell” pathways corresponds to the association results after excluding the IL12B gene from the genome-wide pathway analysis.

AAterid

oetal.

Gen

ome-W

idePath

way

Analysis

inPsoriasis

Journal

ofInvestigative

Derm

atology

(2016),Volume136

596

Table 3. Genes associated with psoriasis risk in thediscovery and replication stages for each validatedpathway

Pathway1 Database Gene2 PD PR

Inflammatoryresponse

Biocarta IL12A 3.35 � 10e3 1.06 � 10e3

IL12B 3.02 � 10e10 1.69 � 10e18

IL13 4.18 � 10e5 2.41 � 10e3

IL4 4.18 � 10e5 2.41 � 10e3

TGFB2 2.82 � 10e3 7.56 � 10e4

Natural killer T cell Biocarta CXCR4 3.09 � 10e3 2.61 � 10e5

IL12A 3.35 � 10e3 1.06 � 10e3

IL12B 3.02 � 10e10 1.69 � 10e18

IL4 4.18 � 10e5 2.41 � 10e3

IL4R 7.35 � 10e3 1.37 � 10e3

TGFB2 2.82 � 10e3 7.56 � 10e4

Retinol metabolism KEGG ADH1B 5.82 � 10e5 1.21 � 10e2

UGT2B4 5.21 � 10e4 6.13 � 10e3

RPE65 5.10 � 10e3 7.34 � 10e3

DNA repair Reactome FANCL 3.22 � 10e3 1.13 � 10e3

MGMT 1.86 � 10e3 9.51 � 10e4

RAD50 4.18 � 10e5 2.41 � 10e3

REV3L 3.16 � 10e6 1.40 � 10e13

RFC3 2.14 � 10e3 1.94 � 10e3

Transport of inorganicions and amino acids

Reactome SLC16A10 2.34 � 10e7 2.38 � 10e9

SLC1A4 2.04 � 10e3 7.44 � 10e3

SLC38A1 1.34 � 10e3 9.44 � 10e3

SLC43A2 1.29 � 10e2 1.15 � 10e2

SLC7A1 6.91 � 10e4 6.62 � 10e3

SLC7A11 1.14 � 10e3 6.22 � 10e3

SLC7A7 5.09 � 10e3 2.77 � 10e3

SLC8A1 8.75 � 10e6 1.30 � 10e3

SLC9A8 1.09 � 10e4 6.81 � 10e5

SLC9A9 4.37 � 10e3 3.22 � 10e3

Post-translationalprotein modification

Reactome ALG10 3.89 � 10e3 4.15 � 10e3

B3GNT2 7.88 � 10e5 1.01 � 10e3

EDEM1 1.69 � 10e4 5.81 � 10e4

EIF5A 3.42 � 10e4 2.09 � 10e5

FUT8 5.30 � 10e3 1.87 � 10e4

GALNT1 5.22 � 10e4 9.72 � 10e4

MAN1A1 2.82 � 10e4 1.04 � 10e2

MAN2A1 7.29 � 10e3 3.53 � 10e3

MGAT5 8.04 � 10e5 9.34 � 10e3

SEMA6D 2.06 � 10e3 1.46 � 10e3

ST8SIA6 8.17 � 10e3 8.68 � 10e5

TUSC3 1.02 � 10e4 6.07 � 10e3

Abbreviations: D, discovery cohort; KEGG, Kyoto Encyclopedia of Genesand Genomes; OR, odds ratio; P, P-value; R, replication cohort.1The detailed description of the “Inflammatory response” and “Naturalkiller T cell” pathways corresponds to the association results beforeexcluding the IL12B gene from the genome-wide pathway analysis.2Genes contained in the genetic pathways that were nominally associatedwith psoriasis risk in the discovery and replication stages.


associations between the pathway genes, functional-basednetworks are a powerful approach to represent and analyzethe topological structure of a biologic pathway.

To characterize the network properties of the resultingfunctional-based networks, we determined the betweenness

centrality (BC) and degree centrality (DC) statistics(Supplementary Table S4 online). These two measures areuseful to identify those network elements (genes in this case)that are likely to be more influential in the structure of thenetwork. BC and DC have been widely used to identify thegenes that are more likely to be essential for pathway func-tionality (Hahn and Kern, 2005; Joy et al., 2005;Vallabhajosyula et al., 2009). We found that SLC7A11 fromthe “Transport of inorganic ions and amino acids” pathwayandMGAT5 from the “Post-translational protein modification”pathway had markedly high BC values (BC � 0.1). From these,MGAT5 gene also showed a much stronger DC value thanSLC7A11 (DCMGAT5 ¼ 19, DCSLC7A11 ¼ 3).

Given the strong network centrality properties found forMGAT5 gene in the “Post-translational protein modification”pathway, we decided to further test the association of this keygene with psoriasis risk in an independent case-controlcohort. Using this additional replication cohort, we signifi-cantly validated the association of MGAT5 gene with psori-asis risk (P ¼ 1.3 � 10e2; odds ratio [95% confidenceinterval] ¼ 0.85 [0.74e0.96]).

Functional analysis of MGAT5 variation

MGAT5 encodes for a key enzyme in the N-glycosylationpathway, a post-translational process that is directly impli-cated in T-cell activation and differentiation (Demetriouet al., 2001). To assess the functional role of MGAT5 inpsoriasis pathogenesis, we evaluated the association betweengenetic variation at MGAT5 gene and the levels of T-cellsurface glycosylation. Flow cytometry analysis of in vitroactivated CD4þ and CD8þ T cells obtained from 27 patientswith psoriasis showed an increase in N-glycosylation levelsin patients carrying one or two copies of the protective allele(G) compared with homozygous individuals for the risk allele(A) (Figure 3). The increased glycosylation levels in in-dividuals carrying at least one copy of (G) allele wasobserved both in activated CD8þ and CD4þ T cells. In CD4þ

T lymphocytes, the glycosylation level was significantlyhigher in GG homozygotes compared with AA homozygotes(P ¼ 0.01, Figure 3).

DISCUSSIONGenome-wide association analyses have successfully identi-fied more than 50 loci associated with psoriasis susceptibility.To date, however, the genetic basis of psoriasis is still notcompletely understood. In this study, we have performed agenome-wide pathway analysis of psoriasis genetic risk. Us-ing a discovery cohort from Spain and an independent cohortfrom the UK, we have identified and validated the associationof six genetic pathways with psoriasis susceptibility. Impor-tantly, these validated pathways include biological processessuch as retinol metabolism, transport of inorganic ions andamino acids, and post-translational protein modification thathad not been previously associated with psoriasis risk at thegenetic level. In addition, analyzing the network properties ofthese validated pathways we have found that MGAT5 genehas a strong centrality in the post-translational proteinmodification pathway. Using an additional independentcase-control cohort from Spain, we have further replicatedthe association of MGAT5 with psoriasis risk. Taken together,



Figure 2. Functional-based network of each genetic pathway associated with psoriasis risk. (a) “Inflammatory response.” (b) “Natural killer T cell.” (c) “Retinol

metabolism.” (d) “DNA repair.” (e) “Transport of inorganic ions and amino acids.” (f) “Post-translational protein modification.” The color of each gene represents

the P-value of its association with psoriasis in the negative logarithmic scale, ranging from the lowest significance (green) to the strongest (red). The gene shape

represents the association with the disease in neither the discovery nor the replication study (square), only in either the discovery or the replication study (circle)

and the association found in both discovery and replication studies (rhombus). The edge width is proportional to the confidence of the functional association

between two genes. Disconnected genes are hidden.


598

these findings contribute to a better understanding of thegenetic risk basis of psoriasis and provide important insightsinto the biological mechanisms associated with the diseasepathogenesis.

Retinol has been demonstrated to inhibit inflammatoryprocesses in dermatological diseases (Balato et al., 2013). Inparticular, retinol inhibits the regulatory activity of the nu-clear factor kappa B (NFKB) in the skin (Austenaa et al.,


2004). NFKB is an established transcriptional factor thatregulates multiple proinflammatory genes that are key inpsoriasis pathogenesis like tumor necrosis factor andinterleukin-17 (Goldminz et al., 2013). The NFKB signalingpathway has also been associated with the regulation of theproliferation of epidermal keratinocytes (Tsuruta, 2009).These findings are consistent with the elevated levels of NFKBthat have been found in lesional and non-lesional psoriatic

Figure 3. N-Glycosylation on activated T lymphocytes according to MGAT5

genotype. Boxplots of mean fluorescence intensity (MFI) of cell membrane

glycosylation of in vitro activated CD4þ (left) and CD8þ (right) T cells from

patients with psoriasis. Patients with one and two copies of the protective

(G) allele of MGAT5 SNP rs3791318 tend to have higher glycosylation levels,

thus increasing the threshold for T-cell receptor-mediated response as well as

lowering the threshold for cytotoxic T-lymphocyte-associated antigen-4-

mediated arrest of T-cell proliferation.


skin samples compared with non-psoriatic skin (Lizzul et al.,2005). Therefore, genetic variation in the retinol metabolismpathway could reduce the retinol production leading to aweakened NFKB signaling and, consequently, promotingboth inflammatory and proliferative hallmarks of psoriasis.

Psoriasis risk was also associated with the genetic pathwayimplicated in the transport of both inorganic ions and aminoacids. An increased transport of inorganic ions in CD4þ

helper T cells has been shown to contribute to autoimmuneand inflammatory diseases (Lang et al., 2014). In particular,the intracellular transport of calcium is crucial for controllingthe expression of proinflammatory genes in immune cells(Khananshvili, 2013; Vig and Kinet, 2009). Accordingly, thetransport of inorganic ions and amino acids pathway asso-ciated with psoriasis risk includes the SLC8A1 gene, whichmodulates the cytoplasmic calcium concentration (Clapham,2007). The transport of amino acids into T cells is essential tomaintain the increased production of proinflammatory cyto-kines in activated human T cells (Hayashi et al., 2013).Importantly, the expression of amino acid transporters hasbeen found to be differentially regulated in psoriatic inflam-matory processes (Jaeger et al., 2008). These results thereforesuggest that genetic variation in the transport of amino acidsand inorganic ions pathway could increase the risk todevelop psoriasis by modulating T-cell functionality.

The post-translational protein modification pathway isresponsible for the N-linked glycosylation of the asparagineresidues in the HLA molecules (Rudd et al., 2001). This post-translational modification pathway has been found to benecessary for the immune system tolerance to self-antigens(Ryan and Cobb, 2012). Previous studies have found that a

deficient or aberrant asparagine glycosylation can induceautoimmune diseases (Green et al., 2007). Also, post-translationally modified autoantigens have been associatedwith psoriasis (Iversen et al., 2011). In patients with psoriasis,the peptide glycosylation activity has been found to bemarkedly increased in comparison with healthy controls(Damasiewicz-Bodzek and Wielkoszynski, 2012). Further-more, specific post-translational modifications on glycopro-teins expressed on the surface of T lymphocytes have beenshown to target these cells to the inflamed skin (Fuhlbriggeet al., 1997). Therefore, genetic variation in the post-translational protein modification pathway could perturb theglycosylation processes that are crucial to maintain the im-mune system tolerance.

MGAT5 encodes for a key enzyme in the N-glycosylationpathway. This pathway has been directly implicated in T-cellactivation and autoimmunity (Demetriou et al., 2001). Recentresearch has found an association between MGAT5 glyco-sylation activity and multiple sclerosis etiology both inexperimental models and in humans (Grigorian andDemetriou, 2011; Mkhikian et al., 2011). In this study, wehave found that the MGAT5 is a key gene in the post-translational protein modification pathway associated withpsoriasis. Subsequently, we found that genetic variation atMGAT5 is associated with the level of glycosylation of in vitroactivated T cells. This result is consistent with previous find-ings showing that deficiency of MGAT5 glycosylation activityreduces the T-cell activation threshold and, consequently,promotes the triggering of autoimmune diseases (Demetriouet al., 2001). Further studies evaluating the implication ofthe T-cell surface glycosylation in clinically relevant outcomesin psoriasis such as skin severity are warranted.

The association of psoriasis risk with the inflammatoryresponse and the natural killer T-cell pathways involves morethan 10 immune-related genes, including IL12B. In a recentpathway analysis study using association results of a meta-analysis for psoriasis risk (Tsoi et al., 2015a), these twopathways were also found to be associated. These findings,however, were not validated using an independent cohort.Our study, therefore, provides strong confirmation of theimplication of these two genetic pathways in the risk ofpsoriasis. Also, the permutation-based approach used in ourstudy allowed to control for the potential bias associated withthe presence of strong linkage disequilibrium patterns withingenes. Our results indicate that the association of thesepathways is not only driven by IL12B gene, but it is the resultof the joint contribution of other small-effect genes in thesepathways. One of these genes is CXCR4, which encodes for achemokine receptor from the natural killer T-cell pathway(Colantonio et al., 2002). Although CXCR4 gene has not beenpreviously associated with psoriasis risk in single-markerGWAS, CXCR4 chemokine has been shown to reduce kera-tinocyte proliferation and, consequently, the expansion ofpsoriatic plaques by regulating the proliferative cytokinesignals that are activated in psoriatic lesions (Takekoshi et al.,2013). In addition, the inflammatory angiogenesis of psoriaticskin that leads to vascular remodeling has been recentlyshown to be modulated by CXCR4 chemokine (Zgraggenet al., 2014). Using the pathway analysis, we can thereforeidentify small-effect genes like CXCR4 that cannot be




600

detected by single-marker GWAS but that are biologicallyimplicated in key processes of the disease pathophysiology.

In this study, we have also found a significant associationbetween the DNA repair genetic pathway and psoriasis risk.Together with the dysregulation of immune system processes,the epidermal hyperproliferation is another well-known bio-logical process implicated in the psoriasis pathophysiology(Wolf et al., 2012). The application of ultraviolet radiation inpsoriasis skin lesions to induce apoptosis in aberrantlyproliferating keratinocytes has proved to be a successfultreatment for the clearance of plaque psoriasis in approxi-mately 70% of patients (Weatherhead et al., 2011). The ul-traviolet radiation induces DNA damage that promotes thetranscription of the DNA repair pathway genes (Roos andKaina, 2006). Consequently, the enzymatic machinery ofthe pathway repairs the DNA damage and also triggers thecell death by activating the p53 apoptotic signaling (Lavinet al., 2005). Therefore, these results suggest that geneticvariation in the DNA repair pathway promotes an inefficientactivation of the p53 apoptotic signaling that leads to anincreased keratinocyte proliferation, as well as an inefficientresponse to ultraviolet therapy in patients with psoriasis.

Although the pathway-based analysis is a powerfulapproach to identify small-effect genetic variants associatedwith disease risk, this methodology is not exempt of limita-tions. Intergenic SNPs across the whole genome that mapphysically far away from genes were not included in thisstudy. These genetic variants could be known risk loci (e.g.,rs12188300 is associated with psoriasis risk and is locatedat >20Kb from IL12B gene) or may regulate the expression ofgenes through cis- and trans-expression quantitative trait locimechanisms (Gilad et al., 2008). Also, some SNPs might notbe functionally related to the closest genes. With theincreasing regulatory information derived from expressionquantitative trait loci and epigenomic data (Bernstein et al.,2010; Martens and Stunnenberg, 2013; Raney et al., 2011),intergenic SNPs could be integrated in the pathway-basedanalysis in the next few years.

The complex linkage disequilibrium structure of the HLAregion together with the strong association with the suscepti-bility to multiple common diseases has been shown togenerate false positive results in pathway-based methods(Wang et al., 2010). Following recent studies, in this study weremoved the SNPs mapping to this locus to perform the pre-sent pathway analysis (Chen et al., 2014). As a result, knownpathways associated with psoriasis risk that include genes fromthe HLA region, like the NFKB pathway, were not analyzed inthis study. Importantly, however, in this study we have foundand validated the association between genetic pathwaysrelated to IL12 signaling, an established genetic risk pathwayfor psoriasis and psoriasis risk. Also, within the associatedpathways there are known risk genes for psoriasis (e.g., REV3Land IL4 within the DNA repair and inflammatory responsepathways, respectively). Together, these results confirm theaccuracy of the present pathway-based approach to identifyrelevant genetic variation associated with psoriasis risk.

The present genome-wide pathway analysis has twoimportant strengths. First, we used PLINK software (Boston,MA) to identify genetic pathways associated with psoriasisrisk. This pathway analysis method uses genotype data in


contrast to the methodologies that are only based on asso-ciation statistics. An important limitation of these lattermethodologies is that they do not account for the linkagedisequilibrium between SNPs. This can result in highly biasedresults and a significant increase in false positive results(Wang et al., 2010). Instead, the pathway analysis approachthat we used, although can be computationally costly, effi-ciently overcomes these biases by maintaining the correctlinkage disequilibrium patterns between SNPs. Finally,compared with previous pathway-based studies in othercomplex diseases, we have performed a two-stage pathwayanalysis in two large cohorts from different populations.Using an independent population, we have validated geneticpathways associated with psoriasis risk.

In conclusion, using a genome-wide pathway analysisapproach we have identified to our knowledge previouslyunreported genetic pathways associated with psoriasis risk.These biological pathways include retinol metabolism,transport of inorganic ions and amino acids, and post-translational protein modification. The results of this studyrepresent an important contribution to the characterization ofthe genetic risk basis of psoriasis.

MATERIALS AND METHODSStudy population

A total of 1,263 patients with psoriasis and 1,558 controls were

recruited for the discovery stage (Supplementary Table S5 online).

An independent case-control cohort of 7,353 individuals from the

UK was used to validate the significantly associated pathways in the

discovery cohort. An independent cohort of 1,381 patients with

psoriasis and 2,048 controls from Spain was used to replicate the

association between MGAT5 gene and psoriasis risk (Supplementary

Materials, Supplementary Table S6 online).

All the procedures were followed in compliance with the prin-

ciples of the Declaration of Helsinki and all patients provided

written informed consent to participate in this study. The study and

the consent procedure were approved by the local Institutional Re-

view Board of each participating center.

DNA extraction and genome-wide genotyping

GWAS genotyping of the 2,821 individuals from the discovery

cohort was performed using Illumina Quad610 Beadchips (Illu-

mina, San Diego, CA) (Supplementary Materials). After the quality

control analysis, a final data set of 541,926 SNPs from 1,172 pa-

tients with psoriasis was available for the pathway-based analysis.

The genome-wide genotyping of the patients with psoriasis from

the validation stage was performed using the Illumina

Human660W-Quad (Illumina) and the healthy controls were gen-

otyped using the Illumina custom Human1.2M-Duo (Illumina) as

has been previously described (Strange et al., 2010). The final data

set used for the replication study included 515,703 SNPs from

2,178 patients with psoriasis. The genotyping of the MGAT5

replication cohort was performed using the Taqman real-time PCR

platform (Applied Biosystems, Foster City, CA) (Supplementary

Materials).

Pathway-based analysis

Gene set definition. Reference biological pathway annotation

databases BioCarta (www.biocarta.com), Kyoto Encyclopedia of

Genes and Genomes (Kanehisa and Goto, 2000), and Reactome

(Croft et al., 2014) were used to determine the global pathways

http://www.biocarta.com


(Supplementary Materials, Supplementary Tables S7 and S8 online).

The final gene set included in this study was composed of 215,948

SNPs mapping to 1,053 pathways.

Gene-set association analysis. The statistical association anal-

ysis was performed using the PLINK set-based test (Purcell et al.,

2007) (Supplementary Materials). To obtain the global statistical

significance of each validated pathway, we combined the empirical

P-values resulting from the discovery and replication stages using

Fisher’s method (Kugler et al., 2010). We tested the association of

1,053 pathways with psoriasis risk. The false discovery rate (FDR)

method (Hochberg and Benjamini, 1990) was used to account for

multiple testing.

Sensitivity analysis by removing the HLA and IL12B loci. In

pathway-based analysis, the presence of a single marker with very

strong effects can lead to false positive associations. In these cases,

the joint contribution of the pathway genes to disease risk is masked

and not adequately evaluated (Wang et al., 2010). Similar to pre-

vious studies, to avoid this type of spurious associations, we

removed all SNPs mapping to the HLA region (Megabases 25.6 to

33.3 in chromosome 6) (Chen et al., 2014). In the discovery stage,

we found genetic pathways in which the IL12B gene was signifi-

cantly associated with disease risk at a genome-wide scale. IL12B is

a well-known psoriasis risk gene that shows a large effect on disease

susceptibility and, like the HLA region, could generate false positive

results (Cargill et al., 2007; Nair et al., 2008; Zhu et al., 2013).

Accordingly, we removed this psoriasis susceptibility locus (from

158,741,791 to 158,757,481 base pairs in chromosome 5) from the

significant pathways and we repeated the analysis. We excluded 73

and 58 SNPs from the discovery and replication studies, respectively.

Characterization of the genetic pathways associated withpsoriasis risk

Genetic pathways involved in similar biological processes may share

genes. To identify pathways representing different and independent

biological processes, we computed the gene overlap between each

pair of genetic pathways associated with psoriasis risk

(Supplementary Materials).

The statistical significance of the association between pathway

genes and psoriasis risk was determined according to the most sig-

nificant SNP mapping to each particular gene.

Analysis of the functional-based networks associated withpsoriasis risk

The biological knowledge representing the functional association

between gene pairs was used to build the functional-based network

of each genetic pathway associated with psoriasis risk. To identify

those genes that are more likely to play a central role in the genetic

pathways associated with psoriasis risk, we analyzed the net-

work statistical properties of each functional-based network

(Supplementary Materials). Using the genes that were nominally

associated with psoriasis risk in both discovery and replication

stages, we identified the most influential gene according to the

highest values of these network statistics.

Functional analysis of MGAT5 variation

Following the methodology previously described (Chen et al., 2009),

we evaluated the association of MGAT5 psoriasis risk variant with

the level of cell surface glycosylation of in vitro activated CD4þ and

CD8þ T cells isolated from n ¼ 27 patients with psoriasis

(Supplementary Materials).

CONFLICT OF INTERESTThe authors state no conflict of interest.

ACKNOWLEDGMENTSThis study was funded by of the Spanish Ministry of Economy and Competi-tiveness, grant numbers: PSE-010000-2006-6 and IPT-010000-2010-36.

SUPPLEMENTARY MATERIAL

Supplementary material is linked to the online version of the paper at www.jidonline.org, and at http://dx.doi.org/10.1016/j.jid.2015.11.026.

REFERENCES

BIOCARTA Pathways. [http://www.biocarta.com].

Austenaa LM, Carlsen H, Ertesvag A, et al. Vitamin A status significantly altersnuclear factor-kappaB activity assessed by in vivo imaging. Faseb J2004;18:1255e7.

Balato A, Schiattarella M, Lembo S, et al. Interleukin-1 family members areenhanced in psoriasis and suppressed by vitamin D and retinoic acid. ArchDermatol Res 2013;305:255e62.

Bernstein BE, Stamatoyannopoulos JA, Costello JF, et al. The NIH RoadmapEpigenomics Mapping Consortium. Nat Biotechnol 2010;28:1045e8.

Bowes J, Budu-Aggrey A, Huffmeier U, et al. Dense genotyping of immune-related susceptibility loci reveals new insights into the genetics of psori-atic arthritis. Nat Commun 2015;6:6046.

Cargill M, Schrodi SJ, Chang M, et al. A large-scale genetic association studyconfirms IL12B and leads to the identification of IL23R as psoriasis-riskgenes. Am J Hum Genet 2007;80:273e90.

Clapham DE. Calcium signaling. Cell 2007;131:1047e58.

Colantonio L, Recalde H, Sinigaglia F, et al. Modulation of chemokine re-ceptor expression and chemotactic responsiveness during differentiation ofhuman naive T cells into Th1 or Th2 cells. Eur J Immunol 2002;32:1264e73.

Croft D, Mundo AF, Haw R, et al. The Reactome pathway knowledgebase.Nucleic Acids Res 2014;42:D472e7.

Chen D, Enroth S, Ivansson E, et al. Pathway analysis of cervical cancergenome-wide association study highlights the MHC region and path-ways involved in response to infection. Hum Mol Genet 2014;23:6047e60.

Chen HL, Li CF, Grigorian A, et al. T cell receptor signaling co-regulatesmultiple Golgi genes to enhance N-glycan branching. J Biol Chem2009;284:32454e61.

Damasiewicz-Bodzek A, Wielkoszynski T. Advanced protein glycation inpsoriasis. J Eur Acad Dermatol Venereol 2012;26:172e9.

Demetriou M, Granovsky M, Quaggin S, et al. Negative regulation of T-cellactivation and autoimmunity by Mgat5 N-glycosylation. Nature 2001;409:733e9.

Du Y, Xie J, Chang W, et al. Genome-wide association studies: inherentlimitations and future challenges. Front Med 2012;6:444e50.

Ferenczi K, Burack L, Pope M, et al. CD69, HLA-DR and the IL-2R identifypersistently activated T cells in psoriasis vulgaris lesional skin: blood andskin comparisons by flow cytometry. J Autoimmun 2000;14:63e78.

Fuhlbrigge RC, Kieffer JD, Armerding D, et al. Cutaneous lymphocyte antigenis a specialized form of PSGL-1 expressed on skin-homing T cells. Nature1997;389:978e81.

Gibson G. Rare and common variants: twenty arguments. Nat Rev Genet2012;13:135e45.

Gilad Y, Rifkin SA, Pritchard JK. Revealing the architecture of gene regulation:the promise of eQTL studies. Trends Genet 2008;24:408e15.

Golan D, Lander ES, Rosset S. Measuring missing heritability: inferring thecontribution of common variants. Proc Natl Acad Sci USA 2014;111:E5272e81.

Goldminz AM, Au SC, Kim N, et al. NF-kappaB: an essential transcriptionfactor in psoriasis. J Dermatol Sci 2013;69:89e94.

Green RS, Stone EL, Tenno M, et al. Mammalian N-glycan branching protectsagainst innate immune self-recognition and inflammation in autoimmunedisease pathogenesis. Immunity 2007;27:308e20.

Grigorian A, Demetriou M. Mgat5 deficiency in T cells and experimentalautoimmune encephalomyelitis. ISRN Neurol 2011:374314.




http://dx.doi.org/10.1016/j.jid.2015.11.026

http://www.biocarta.com

http://refhub.elsevier.com/S0022-202X(15)00242-0/sref2












































































602

Gudjonsson JE, Elder JT. Psoriasis: epidemiology. Clin Dermatol 2007;25:535e46.

Gui H, Li M, Sham PC, et al. Comparisons of seven algorithms for pathwayanalysis using the WTCCC Crohn’s Disease dataset. BMC Res Notes2011;4:386.

Hahn MW, Kern AD. Comparative genomics of centrality and essentiality inthree eukaryotic protein-interaction networks. Mol Biol Evol 2005;22:803e6.

Hayashi K, Jutabha P, Endou H, et al. LAT1 is a critical transporter of essentialamino acids for immune reactions in activated human T cells. J Immunol2013;191:4080e5.

Hochberg Y, Benjamini Y. More powerful procedures for multiple significancetesting. Stat Med 1990;9:811e8.

Iversen OJ, Lysvand H, Hagen L. The autoantigen Pso p27: a post-translationalmodification of SCCA molecules. Autoimmunity 2011;44:229e34.

Jaeger K, Paulsen F, Wohlrab J. Characterization of cationic amino acidtransporters (hCATs) 1 and 2 in human skin. Histochem Cell Biol 2008;129:321e9.

Johnson RC, Nelson GW, Troyer JL, et al. Accounting for multiple compari-sons in a genome-wide association study (GWAS). BMC Genomics2010;11:724.

Joy MP, Brock A, Ingber DE, et al. High-betweenness proteins in theyeast protein interaction network. J Biomed Biotechnol 2005;2005:96e103.

Kanehisa M, Goto S. KEGG: Kyoto Encyclopedia of Genes and Genomes.Nucleic Acids Res 2000;28:27e30.

Khananshvili D. The SLC8 gene family of sodium-calcium exchangers(NCX)—structure, function, and regulation in health and disease. Mol As-pects Med 2013;34:220e35.

Koster R, Mitra N, D’Andrea K, et al. Pathway-based analysis of GWAs dataidentifies association of sex determination genes with susceptibility totesticular germ cell tumors. Hum Mol Genet 2014;23:6061e8.

Kugler KG, Mueller LA, Graber A. MADAM: an open source meta-analysistoolbox for R and Bioconductor. Source Code Biol Med 2010;5:3.

Lang F, Stournaras C, Alesutan I. Regulation of transport across cell mem-branes by the serum- and glucocorticoid-inducible kinase SGK1. MolMembr Biol 2014;31:29e36.

Lavin MF, Birrell G, Chen P, et al. ATM signaling and genomic stability inresponse to DNA damage. Mutat Res 2005;569:123e32.

Lizzul PF, Aphale A, Malaviya R, et al. Differential expression of phosphor-ylated NF-kappaB/RelA in normal and psoriatic epidermis and down-regulation of NF-kappaB in response to treatment with etanercept. J InvestDermatol 2005;124:1275e83.

Manolio TA. Genomewide association studies and assessment of the risk ofdisease. N Engl J Med 2010;363:166e76.

Martens JH, Stunnenberg HG. BLUEPRINT: mapping human blood cell epi-genomes. Haematologica 2013;98:1487e9.

Mkhikian H, Grigorian A, Li CF, et al. Genetics and the environment convergeto dysregulate N-glycosylation in multiple sclerosis. Nat Commun 2011;2:334.

Nair RP, Ruether A, Stuart PE, et al. Polymorphisms of the IL12B and IL23Rgenes are associated with psoriasis. J Invest Dermatol 2008;128:1653e61.

Nair RP, Stuart PE, Nistor I, et al. Sequence and haplotype analysis supportsHLA-C as the psoriasis susceptibility 1 gene. Am J Hum Genet 2006;78:827e51.

Nestle FO, Kaplan DH, Barker J. Psoriasis. N Engl J Med 2009;361:496e509.

Park JH, Wacholder S, Gail MH, et al. Estimation of effect size distributionfrom genome-wide association studies and implications for future discov-eries. Nat Genet 2010;42:570e5.

Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genomeassociation and population-based linkage analyses. Am J Hum Genet2007;81:559e75.


Ramanan VK, Shen L, Moore JH, et al. Pathway analysis of genomic data:concepts, methods, and prospects for future development. Trends Genet2012;28:323e32.

Raney BJ, Cline MS, Rosenbloom KR, et al. ENCODE whole-genome data inthe UCSC genome browser (2011 update). Nucleic Acids Res 2011;39:D871e5.

Roos WP, Kaina B. DNA damage-induced cell death by apoptosis. Trends MolMed 2006;12:440e50.

Rudd PM, Elliott T, Cresswell P, et al. Glycosylation and the immune system.Science 2001;291:2370e6.

Ryan SO, Cobb BA. Roles for major histocompatibility complex glycosylationin immune function. Semin Immunopathol 2012;34:425e41.

Strange A, Capon F, Spencer CC, et al. A genome-wide association studyidentifies new psoriasis susceptibility loci and an interaction between HLA-C and ERAP1. Nat Genet 2010;42:985e90.

Takekoshi T, Wu X, Mitsui H, et al. CXCR4 negatively regulates keratinocyteproliferation in IL-23-mediated psoriasiform dermatitis. J Invest Dermatol2013;133:2530e7.

Tsoi LC, Elder JT, Abecasis GR. Graphical algorithm for integration of geneticand biological data: proof of principle using psoriasis as a model. Bioin-formatics 2015a;31:1243e9.

Tsoi LC, Spain SL, Ellinghaus E, et al. Enhanced meta-analysis and replicationstudies identify five new psoriasis susceptibility loci. Nat Commun2015b;6:7001.

Tsoi LC, Spain SL, Knight J, et al. Identification of 15 new psoriasis susceptibilityloci highlights the role of innate immunity. Nat Genet 2012;44:1341e8.

TsurutaD.NF-kappaB linkskeratinocytesand lymphocytes in thepathogenesis ofpsoriasis. Recent Pat Inflamm Allergy Drug Discov 2009;3:40e8.

Vallabhajosyula RR, Chakravarti D, Lutfeali S, et al. Identifying hubs in pro-tein interaction networks. PLoS One 2009;4:e5344.

Vig M, Kinet JP. Calcium signaling in immune cells. Nat Immunol 2009;10:21e7.

Vyse TJ, Todd JA. Genetic analysis of autoimmune disease. Cell 1996;85:311e8.

Wang K, Li M, Hakonarson H. Analysing biological pathways in genome-wide association studies. Nat Rev Genet 2010;11:843e54.

Weatherhead SC, Farr PM, Jamieson D, et al. Keratinocyte apoptosis inepidermal remodeling and clearance of psoriasis induced by UV radiation.J Invest Dermatol 2011;131:1916e26.

Wolf R, Orion E, Ruocco E, et al. Abnormal epidermal barrier in the patho-genesis of psoriasis. Clin Dermatol 2012;30:323e8.

Wu MC, Kraft P, Epstein MP, et al. Powerful SNP-set analysis for case-controlgenome-wide association studies. Am J Hum Genet 2010;86:929e42.

Yin X, Low HQ, Wang L, et al. Genome-wide meta-analysis identifies mul-tiple novel associations and ethnic heterogeneity of psoriasis susceptibility.Nat Commun 2015;6:6916.

Yin X, Wineinger NE, Cheng H, et al. Common variants explain a largefraction of the variability in the liability to psoriasis in a Han Chinesepopulation. BMC Genomics 2014;15:87.

Zgraggen S, Huggenberger R, Kerl K, et al. An important role of the SDF-1/CXCR4 axis in chronic skin inflammation. PLoS One 2014;9:e93665.

Zhang K, Cui S, Chang S, et al. i-GSEA4GWAS: a web server for identificationof pathways/gene sets associated with traits by applying an improved geneset enrichment analysis to genome-wide association study. Nucleic AcidsRes 2010;38:W90e5.

Zhong H, Yang X, Kaplan LM, et al. Integrating pathway analysis and geneticsof gene expression for genome-wide association studies. Am J Hum Genet2010;86:581e91.

Zhu KJ, Zhu CY, Shi G, et al. Meta-analysis of IL12B polymorphisms(rs3212227, rs6887695) with psoriasis and psoriatic arthritis. Rheumatol Int2013;33:1785e90.

Zuo X, Sun L, Yin X, et al. Whole-exome SNP array identifies 15 new sus-ceptibility loci for psoriasis. Nat Commun 2015;6:6793.






































































































































































genome-wide pathway analysis identifies genetic pathways

Documents