ORIGINAL ARTICLE

Genetic variants of CC chemokine genes in experimentalautoimmune encephalomyelitis, multiple sclerosis andrheumatoid arthritis

J Ockinger1, P Stridh1, AD Beyeen1, F Lundmark2, M Seddighzadeh3, A Oturai4, PS Sørensen4,AR Lorentzen5,6, EG Celius5, V Leppa7,8, K Koivisto9, PJ Tienari10,11, L Alfredsson12, L Padyukov3,J Hillert2, I Kockum1, M Jagodic1,13 and T Olsson1,13

1Neuroimmunology Unit, Department of Clinical Neuroscience, Center of Molecular Medicine, Karolinska Institutet, Stockholm,Sweden; 2MS Research Group, Department of Clinical Neuroscience, Center of Molecular Medicine, Karolinska Institutet, Stockholm,Sweden; 3Rheumatology Unit, Department of Medicine, Center of Molecular Medicine, Karolinska Institutet, Stockholm, Sweden;4Department of Neurology, Danish Multiple Sclerosis Research Center, Copenhagen University Hospital, Copenhagen, Denmark;5Department of Neurology, Oslo University Hospital, Ulleval, Norway; 6Institute of Immunology, Oslo University Hospital,Rikshospitalet, Norway; 7Unit of Public Health Genomics, National Institute for Health and Welfare, Helsinki, Finland; 8Institute forMolecular Medicine and Helsinki Biomedical Graduate School, University of Helsinki, Helsinki, Finland; 9Central Hospital of Seinajoki,Seinajoki, Finland; 10Molecular Neurology Program, Biomedicum Helsinki, University of Helsinki, Helsinki, Finland; 11Department ofNeurology, Helsinki University Central Hospital, Helsinki, Finland and 12Institute of Environmental Medicine, Karolinska Institutet,Stockholm, Sweden

Multiple sclerosis (MS) is a complex disorder of the central nervous system, causing inflammation, demyelination and axonaldamage. A limited number of genetic risk factors for MS have been identified, but the etiology of the disease remains largelyunknown. For the identification of genes regulating neuroinflammation we used a rat model of MS, myelin oligodendrocyteglycoprotein (MOG)-induced experimental autoimmune encephalomyelitis (EAE), and carried out a linkage analysis in anadvanced intercross line (AIL). We thereby redefine the Eae18b locus to a 0.88 Mb region, including a cluster of chemokinegenes. Further, we show differential expression of Ccl2, Ccl11 and Ccl11 during EAE in rat strains with opposite susceptibilityto EAE, regulated by genotype in Eae18b. The human homologous genes were tested for association to MS in 3841 cases and4046 controls from four Nordic countries. A haplotype in CCL2 and rs3136682 in CCL1 show a protective association to MS,whereas a haplotype in CCL13 is disease predisposing. In the HLA-DRB1*15 positive subgroup, we also identified anassociation to a risk haplotype in CCL2, suggesting an influence from the human leukocyte antigen (HLA) locus. We furtheridentified association to rheumatoid arthritis in CCL2, CCL8 and CCL13, indicating common regulatory mechanisms forcomplex diseases.Genes and Immunity (2010) 11, 142–154; doi:10.1038/gene.2009.82; published online 29 October 2009

Keywords: multiple sclerosis; experimental autoimmune encephalomyelitis; rheumatoid arthritis; chemokine; AIL;meta analysis

Introduction

Multiple sclerosis (MS) is a chronic inflammatory diseaseof the central nervous system, causing demyelinationand axonal damage. The etiology is largely unknown,but a number of genetic variants contributing tosusceptibility are identified, including HLA-DRB1,1–3

IL7R4,5 and IL2RA.5 Complex disorders involving chronic

inflammation and dysregulation of the immune systemmay share common genetic risk factors.6 Examples ofgenes associated to several inflammatory diseasesinclude IRF5, which is associated to systemic lupuserythematosus,7 rheumatoid arthritis (RA),8 inflamma-tory bowel disease9 and MS10, and CLEC16A, associatedto MS11 and type I diabetes.12–14 Identification ofmediators of inflammation and other pathogenic me-chanisms is crucial for the understanding of the disease,and experimental models have successfully been appliedin positional cloning and functional studies of genes thatregulate complex human diseases, including CTLA-415

and CBLB.16

To identify genetic regions that influence severity andsusceptibility to neuroinflammation, we have used a ratmodel of MS, myelin oligodendrocyte glycoprotein

Received 2 July 2009; revised 3 September 2009; accepted 15September 2009; published online 29 October 2009

Correspondence: Dr J Ockinger, Neuroimmunology Unit, Depart-ment of Clinical Neuroscience, Karolinska Institutet, KarolinskaUniversity Hospital, CMM, L8:04, Stockholm 171 76, Sweden,E-mail: johan.ockinger@ki.se13These authors contributed equally to this study.

Genes and Immunity (2010) 11, 142–154& 2010 Macmillan Publishers Limited All rights reserved 1466-4879/10 $32.00

www.nature.com/gene

(MOG)-induced experimental autoimmune encephalo-myelitis (EAE), and carried out a linkage analysis in anadvanced intercross line (AIL). The aim was to identifycandidate genes in a previously identified locus, Eae18b,on rat chromosome 10.17–19 This genetic region is also acandidate region for different rat models of arthritis20–22

and the homologous region on human chromosome17 showed suggestive linkage in a meta analysis oflinkage studies in MS.23 Eae18b contains a number ofinteresting candidate genes for MS and other inflamma-tory disorders, including a cluster of chemokine genes.Chemokines are small proteins with chemotactic andimmunoregulatory properties, which are important forguiding leukocyte trafficking, in health and duringinflammation.24,25 There is evidence for a role of specificchemokines in regulation of disease and cell migration inboth MS26 and EAE.27 The chemokine genes have alsobeen included in linkage and association studies,28–31

but despite substantial evidence for linkage and associa-tion to this chemokine cluster, data lack conclusivereplication.32

In this study, we refine the Eae18b locus to a 0.88 Mbregion containing the chemokine gene cluster and showthat the expression of chemokines Ccl2, Ccl11 and Ccl1 isregulated by the locus. We further present evidence forassociation of CCL2, CCL1 and CCL13 in a large NordicMS material, including a possible modulatory effect fromthe human leukocyte antigen (HLA)-DR1*15 locus.

Materials and methods

Advanced intercross lineThe AIL33 used in this study originated from the EAE-susceptible DA and EAE-resistant PVG.av1 rat strainsthat share the RT1.AV1 MHC-haplotype, thus allowingthe identification of non-MHC genes regulating disease.The AIL was produced as previously described.34 InbredDA rats were originally obtained from the Zentralinstitutfur Versuchstierzucht (Hannover, Germany) andPVG.av1 rats from Harlan UK (Blackthorn, UK). Animalswere bred and housed at the Karolinska UniversityHospital (Stockholm, Sweden), in polystyrene cagescontaining aspen wood shavings with free access towater and standard rodent chow and a 12 h light–darkcycle. The animals were routinely monitored for patho-gens according to a health-monitoring program for ratsat the National Veterinary Institute, in Uppsala, Sweden.The local ethical committee in northern Stockholmapproved the experiments.

EAE induction and phenotypic evaluationRecombinant MOG (amino acids 1–125 from theN-terminus) was expressed in Escherichia coli andpurified to homogeneity by chelate chromatography aspreviously described.35 The purified protein, dissolved in6 M urea, was dialyzed against PBS to obtain a semi-precipitated preparation that was stored at �20 1C. Ratsbetween 8 and 16 weeks of age were anesthetized withisofluorane (Forene, Abbott Laboratories, Abbot Park, IL,USA) and immunized with a single subcutaneousinjection in the dorsal base of the of the tail with 200mlof inoculum containing rMOG (amino acids 1–125) (20mgper rat) in saline emulsified with IFA (Sigma Aldrich,St Louis, MO, USA), ratio 1:1. The rats were weighed

and monitored daily for clinical signs of EAE from days 7to 10 until the end of the experiment, that is 31–38 dayspost immunization. The clinical score was graded asfollows: (0) no clinical signs of EAE; (1) tail weakness ortail paralysis; (2) hind leg paraparesis or hemiparesis; (3)hind leg paralysis or hemiparalysis; (4) tetraplegia ormoribund; and (5) death. The following clinical para-meters were assessed for each animal: EAE incidence(clinical signs for 41 day), onset of EAE (the first daythat clinical signs were observed), maximum EAE score(the highest clinical score observed during EAE),cumulative EAE score (the sum of daily clinical scores),duration of EAE (the number of days with EAE) andweight loss ([weight at day 8 p.i.�minimum weightduring the experiment]/weight at day 8 p.i.).

Genotyping in ratsGenotyping of animals was carried out on DNAextracted from ear or tail tip tissues according to astandard protocol.36 In total, 794 rats were genotyped(428 females and 366 males) corresponding to 772clinically monitored rats and 22 additional rats, forwhich phenotypic data could not be obtained because ofdeath after anesthesia. The incidence of EAE was 29%,affecting 152 females and 72 males, as previouslydescribed.34 The region analyzed in the G10 AIL includesthe Eae18b region previously linked to MOG-EAE.19 The7.93 Mb region delimited by markers D10Arb27 andD10Rat27 was genotyped with 10 microsatellite markers(Proligo, Paris, France). Primer sequences are found onwww.ensembl.org or http://rgd.mcw.edu. PCR amplifi-cation was carried out as previously described37 with[g-33P] ATP end-labeled forward primers. The PCRproducts were size fractioned on 6% polyacrylamidegels and visualized by autoradiography. Two indepen-dent observers evaluated all genotypes manually.

Linkage analysisLinkage analysis was carried out in R/qtl 1.08,38 usingthe Haley–Knott regression model, with sex and breed-ing couple as covariates. The confidence interval wasdefined using the markers outside a 1.5 reduction inLOD score approximately equivalent to a 95% confidenceinterval.39 In order to better define the position of theQTL within the mapped region, the probability for thelocation of a QTL was calculated using a bootstrapapproach. Simulated pedigrees were sampled withreplacement from the observed AIL individuals to createa new dataset with the same number of samples (794),which was mapped using a single-QTL model in R/qtl.The maximum LOD and the location of that maximumwere recorded and the re-sampling was repeated 1000times to obtain an estimate of the probability of a QTLbeing present within the confidence interval of the QTL.

Expression analysisThe thirteenth generation of the AIL was produced from25 breeding couples, and offspring was genotyped asdescribed above, with microsatellite markers D10G97and D10Rat98, flanking the identified QTL. In all, 51animals were homozygous for DA (19 rats) or PVG.av1(32 rats) at both markers. The background was notgenotyped. Animals (12 weeks old) were immunizedwith MOG, as describe above, and on day 7 postimmunization, animals were killed using CO2 and

CC chemokines in EAE, MS and RAJ Ockinger et al

143

Genes and Immunity

inguinal lymph nodes were dissected out and snap-frozen in microtubes and stored at �70 1C until use. Forexpression analysis in the parental DA and PVG.av1strains five animals of each strain were immunized, asdescribed above, and the spinal cord and inguinal lymphnodes were dissected out when the DA animals showedEAE symptoms (X1) day 10–14, post immunization. Apaired PVG.av1 animal was selected for each DA rat, andwas dissected the same day as the DA rat, as PVG.av1rats do not show EAE symptoms at this dose. Tissueswere disrupted using Lysing Matrix D tubes (MPBiomedicals, Irvine, CA, USA) on a FastPrep homogeni-zer (MP Biomedicals) and mRNA was extracted usingRNeasy mini columns (Qiagen, Hilden, Germany),including on column DNA-digestion. Reverse transcrip-tion was carried out using random hexamer primers(Gibco, Carlsbad, CA, USA) and Superscript ReverseTranscriptase (Invitrogen, Carlsbad, CA, USA). Real-timePCR was performed using a BioRad iQ5 iCyclerDetection System (BioRad Laboratories, Hercules, CA,USA) with a three-step PCR protocol (95 1C for 10 min.followed by 40 or 50 cycles of 95 1C for 15 s, 60 1C for 30 sand 72 1C for 30 s), using SYBR green as fluorophore.(BioRad Laboratories). Relative expression was calcu-lated as the ratio between the target and Gapdh. Serial 10-fold dilutions from a pool of undiluted samples withinthe study were used as standard. Primers sequences forthe quantitative real time PCR are found in Supplemen-tary Table 3.

Re-sequencingRe-sequencing templates were amplified from genomicDNA using a standard three-step PCR protocol withprimers purchased from Proligo. Amplicons wereextracted using PCR CleanUp (Qiagen) and used ina sequencing reaction including BigDye terminator v3.1(Applied Biosystems, Carlsbad, CA, USA). Fragmentswere separated and recorded on an ABI 3100 (AppliedBiosystems) and sequences were analyzed with VectorNTI software (InforMax, Carlsbad, CA, USA).

Association studiesCase–control material for MS. The first Swedish MS case–control study (Sweden 1) consisted of 1023 subjects livingin Sweden, originating from Sweden or other Nordiccountries. The patients fulfilled the McDonald criteria fordefinite MS40 and were recruited by neurologists at theKarolinska University Hospital. The control groupsconsisted of 1215 blood donors. These subjects lived inthe Stockholm area and originated from Sweden or otherNordic countries. Details of the collection, as well asadditional data regarding the cohort, have previouslybeen reported.41

To follow up our original association to MS, associatedsingle-nucleotide polymorphisms (SNPs) were investi-gated in an additional 2546 patients and 3943 controlsfrom Sweden, Norway, Denmark and Finland. Sweden 2included 952 cases and 665 controls, Norway included548 cases and 554 controls, Denmark included 512 casesand 553 controls, and Finland 806 cases and 1059controls. All patients from the Nordic countries werediagnosed with MS according to the McDonald criteria40

except in Finland, where the Poser’s criteria wereapplied.42 Controls from Sweden 1 and Norway arehealthy blood donors and controls from Denmark

are healthy blood donors and hospital staff. Controlsfrom Finland are population based, whereas controlsfrom Sweden 2 are matched to the MS patients withregards to age, sex and area of residence. Oral and/orwritten informed consent was given from all individualsinvolved in the study.

Case–control material for RA. A total of 2360 RA patients(wherein 1373 patients were positive for anti-citrulli-nated peptide antibodies (ACPA)) fulfilling the 1987revised criteria of the American College of Rheumatol-ogy43 and 958 population-based controls from thepreviously described EIRA cohort44,45 were included inthe study. Detection of antibodies directed towardscitrullinated peptides was performed using the Immuno-scan-RA Mark2 ELISA test (Euro-Diagnostica, Malmo,Sweden). A level 425 U ml�1 was regarded as beingpositive according to instructions in the kit and asconfirmed by validation at the Clinical Immunologylaboratory at Uppsala University Hospital, Uppsala,Sweden.

Genotyping of human materialGenotyping in all case–control cohorts, except Finland,was performed using Taqman SNP Genotyping Assaysfrom Applied Biosystems. PCR amplifications wereperformed on GeneAmp PCR System 9700 (AppliedBiosystems) thermal cyclers, end-point readings on a7900HT Real-Time PCR System (Applied Biosystems)and allele calling was carried out in SDS v1.2 software(Applied Biosystems). About 2–5 ng DNA was used ineach PCR reaction. Genotyping in the Finnish cohort wasperformed using the Sequenom MassARRAy iPLEXGold platform (Sequenom, San Diego, CA, USA), with15 ng of genomic DNA/24-SNP multiplex. Genotypeswere automatically called in Sequenom Typer 4.0(Sequenom) and manually verified.

HLA typing. Human leukocyte antigen typing wasconducted with sequence-specific primers using OlerupSSP HLA-A low-resolution kit and Olerup SSP DR low-resolution kit (Olerup SSP, Saltsjobaden, Sweden). Forthe Finnish cohort, the rs3135388 SNP was genotypedusing the Sequenom iPlex platform, for the determina-tion of the HLA-DR1*15 status.

AssociationAssociations including different models of allelic effectswere calculated using logistic regression in the SNPassocpackage46 in R.47 Allele frequencies and frequencies ofhaplotypes (including all SNPs in each gene) wereestimated in Haploview 4.148 and haplotype frequencieswere confirmed in UNPHASED 3.1.49 No haplotypesspanning over several genes was investigated, as themarkers were selected with a focus on the genes, not torepresent the genetic polymorphisms in the wholeregion. Meta analyses were carried out using theMantel–Haenszel method with fixed effects in the Rpackage rmeta (http://cran.r-project.org).

CCL2 ELISA in cerebrospinal fluid (CSF)Cell free CSF from 38 MS patients with relapsing-remitting disease included in the Sweden 1 cohort and41 patients diagnosed for other neurological diseases atthe Neurological clinic at Karolinska University Hospital,

CC chemokines in EAE, MS and RAJ Ockinger et al

144

Genes and Immunity

Solna was used for the analysis of protein levels ofchemokines. The control group consisted of 41 subjectswith other neurological diseases (mean age 43.3 years,range 21–83 years; 20 females and 21 males; all withoutIgG oligoclonal bands in CSF) with the followingdiagnoses: unspecified sensory disturbance (n¼ 9), un-specified headache (n¼ 3), migraine (n¼ 4), non inflam-matory peripheral neuropathy (n¼ 4), vertigo (n¼ 3),paresthesia (n¼ 3), unspecific white matter changes(n¼ 3), psychological symptoms (n¼ 2), neuralgia(n¼ 1), carpal tunnel syndrome (n¼ 1), cervicalgia(n¼ 1), amyotrophic lateral sclerosis (n¼ 1), dissociativesyndrome (n¼ 1), myalgia (n¼ 1), hypesthesia (n¼ 1),diplopia (n¼ 1), n. pudendus damage (n¼ 1) andsyncope (n¼ 1). All CSF samples were centrifuged andcells were separated immediately after sampling and thecell-free CSF was stored at �70 1C until use. Proteinlevels of CCL2, CCL7, CCL11, CCL8, CCL13 and CCL1 inCSF were measured using Duoset ELISA kits (R&DSystems, Minneapolis, MN, USA) according to manu-facturer’s instructions on Microlon 96 well plates(Greiner Bio-One GmbH, Frickenhausen, Germany).Only CCL2 was found in detectable levels. Values wereanalyzed using Mann–Whitney U test in GraphPadPrism version 5 (GraphPad, San Diego, CA, USA).

Results

High-resolution linkage analysis in ratsA region on rat chromosome 10 has previously beenlinked to the susceptibility and severity of EAE, incrosses using the susceptible DA and the resistantPVG.av1 or ACI rat strains.18,19 In order to further define

the region regulating EAE severity, the tenth generationof a rat AIL,50 originating from the MHC-identical DAand PVG.av1 strains, was used for linkage analysis. Atotal of 10 microsatellite markers (D10Arb27-D10Rat27)in the previously defined Eae18b were genotyped in 794AIL rats. Linkage analysis redefined the QTL to a0.88 Mb region, linked primarily to severity of EAE,(Figure 1 and Table 1), with the peak marker D10Kini1located at 70.27 Mb. The affected animals with DAgenotype at D10Kini1 show longer disease durationcompared with animals with heterozygous or PVGgenotype (DA/DA: 17.5 days, DA/PVG: 12.0 andPVG/PVG: 8.0, P¼ 0.0012). Additional phenotypes re-presenting severity of EAE (cumulative and maximumEAE score) are also linked to the same peak marker(D10Kini1), and have similar confidence intervals,whereas weight loss during disease is linked to abroader, overlapping region (Table 1). Phenotypesrepresenting susceptibility to EAE (incidence and dayof onset of EAE) as well as anti-MOG IgG subtype titers(total IgG, IgG1 and IgG2b) in sera were also investi-gated, but neither susceptibility phenotypes nor anti-MOG IgG titers showed any significant linkage to theregion.

Further, we applied a bootstrap approach to supportthe linkage and to better position the most likely locationof the QTL. The frequencies of the maximum LOD scorewithin the region (Figure 1), obtained from 1000resamples, indicated that the most likely location forthe main effect QTL is located at the marker D10Kini1, amarker that maps between Ccl7 and Ccl11. This refinedQTL described here is located within the previouslydefined QTL,19 but the confidence interval is reducedfrom 3 to 0.88 Mb. In total, 10 genes (known and

2

4

6

8

10

12 Duration of EAE

LOD

Sco

re

0.4

0.1

0.2

0.3

Freq. of counts

95% C.I. (0.88 Mb)

D10A

rb27O

T24.18

D10R

at29

D10R

at123

D10R

at98D

10Rat155

D10K

ini1

D10R

at58

OT

28.48

D10R

at27

Figure 1 Linkage analysis in the tenth generation of an advanced intercross line (AIL) defines Eae18b, regulating severity of myelinoligodendrocyte glycoprotein-induced experimental autoimmune encephalomyelitis (MOG-EAE). Linkage analysis carried out on 794 rats,genotyped with 10 microsatellite markers, with duration of EAE (number of days with score X1) as phenotype. The scale on the x axis isbased on the physical position of the markers (Ensembl v52). The horizontal light gray bar represents the confidence interval for the identifiedQTL, defined by the microsatellite markers outside a reduction of 1.5 LOD score. Dark gray vertical bars represent the frequency of maximumLOD score at a specific location, based on 1000 re-samplings from the original cohort (see Materials and methods for details).

CC chemokines in EAE, MS and RAJ Ockinger et al

145

Genes and Immunity

predicted) are located within the confidence interval forduration of EAE (Table 2), including a cluster of genescoding for chemokines Ccl2, 7, 11, 12 and 1. Ccl2 has beenstudied in relation to induction and relapses ofEAE,27,51,52 but all chemokines in the cluster are involvedin chemotaxis and immunostimulation. We thereforeinvestigated further all chemokine genes in the clusterfor genetic regulation of neuroinflammation.

Genotype at Eae18b regulates expression of Ccl2, Ccl1and Ccl11A pattern of expression in both lymph nodes and spinalcord at the onset of disease that correlates with diseasesusceptibility indicated dysregulation of chemokinepathways during disease (Supplementary Figure 1).The EAE resistance conferred by PVG genotype may beexplained in part by the upregulation of Ccl11 and Ccl12in the lymph node and downregulation of Ccl2 and Ccl1in the spinal cord. To investigate if this chemokinedysregulation is genetically regulated and contributes todisease, we looked at genomic sequence and expressionlevels between susceptible and resistant strains. Exons ofthe chemokine genes were re-sequenced, but no geneticdifferences between the DA and PVG.av1 strains weredetected (data not shown). However, when re-sequen-cing up to 2000 bp upstream of the transcription startsites, multiple polymorphisms were detected, indicatingregulation of transcription as a likely disease modifier. Inaddition, the genetic influence from Eae18b on expressionof chemokine genes was investigated in the thirteenthgeneration of the rat AIL. In total, 51 animals from 25breeding couples were stratified according to the

genotype at Eae18b (markers D10Got97 and D10Rat98)into two groups, DA or PVG homozygous. The inter-crossed animals represent a highly genetically diversebackground and the stratified groups will thereby onlybe genetically identical within, and in close proximity ofthe Eae18b locus. Expression of Ccl2, Ccl1 and Ccl11mRNA in lymph nodes seven days after immunizationwas regulated by the genotype in the region (Figure 2).Susceptible DA alleles predispose for higher expressionof Ccl2 and Ccl1, whereas resistant PVG alleles predis-pose for higher expression of Ccl11 (Figure 2). The lack ofexonic polymorphisms between the tested rat strains incombination with the expression profiles in the stratifiedAIL animals and parental strains indicates geneticallydetermined cis-regulated expression differences as themain mechanism, whereby Eae18b regulates EAE sus-ceptibility. It furthermore puts focus on Ccl2, Ccl1 andCcl11 and their human homologues for the regulation ofneuroinflammation.

Association to CCL2, CCL13 and CCL1 in MS. Havingestablished genetic linkage with the chemokine genecluster with severity of experimental neuroinflammationand expression of Ccl2, Ccl1 and Ccl11, we set out toinvestigate the effects of genetic variances in thehomologous human chemokine locus including CCL2,CCL7, CCL11, CCL8, CCL13 and CCL1 on chromosome 17(29.6–29.7 Mb) and their impact on MS susceptibility. Aninitial scan in 1023 Swedish MS patients and 1215controls (Sweden 1) using 18 SNPs revealed associationto rs159313 in CCL13, using a dominant model(P¼ 0.019) (OR:1.24 [1.04:1.48]), and suggestive associa-

Table 1 Eae18b on rat chromosome 10 is linked to the severity of EAE

Phenotype Peak marker LOD Position (Mb) Confidence interval (position, Mb)

Weight loss D10Rat58 11.99 71.07 D10Rat123 (69.68) D10Rat27 (77.09)Cumulative EAE score D10Kini1 10.54 70.27 D10Rat123 (69.68) D10Rat58 (71.07)Maximum EAE score D10Kini1 9.23 70.27 D10Rat98 (70.19) OT28.48 (71.97)Duration of EAE D10Kini1 11.13 70.27 D10Rat98 (70.19) D10Rat58 (71.07)

Abbreviation: EAE, experimental autoimmune encephalomyelitis.Linkage analysis was carried out in 794 rats, and identified linkage to cumulative and maximum EAE score, as well as duration of EAE andweight loss during disease, all representing severity of EAE. The confidence intervals are defined by the marker flanking a 1.5 reduction inLOD. Positions of genetic markers retrieved from Ensembl v 52.

Table 2 Eae18b includes a cluster of chemokine genes

Location (bp) RGD symbol Description Strand

70256263 Ccl2 Chemokine (C–C motif) ligand 2 +70267281 Ccl7 Chemokine (C–C motif) ligand 7 +70279161 Ccl11 Chemokine (C–C motif) ligand 11 +70321041 Ccl12 Chemokine (C–C motif) ligand 12 +70380671 Ccl1 Chemokine (C–C motif) ligand 1 �70628526 Tmem132e Transmembrane protein 132E +70825997 OTTMUSG00000000913 Projected MGI (automatic) +70983767 Znf830 Zinc finger protein 830 +71000287 Lig3 Ligase III, DNA, ATP-dependent +71025035 Rffl E3 ubiquitin-protein ligase rififylin �

Abbreviations: ATP, adenosine triphosphate; DNA, deoxyribose nucleic acid; EAE, experimental autoimmune encephalomyelitis.List of genes included in the confidence interval for linkage to duration of EAE. Location (start position) and description of genes wereretrieved from Ensembl v52 (www.ensembl.org), and gene symbols from Rat Genome Database (http://rgd.mcw.edu).

CC chemokines in EAE, MS and RAJ Ockinger et al

146

Genes and Immunity

0.0

0.5

1.0

1.5

2.0

0.0

0.5

1.0

1.5

2.0

0

1

2

3

4

5

0.0

0.5

1.0

1.5

2.0

0.0

0.5

1.0

1.5

2.0

Ccl2 Ccl7 Ccl11

p=0.019

Ccl1Ccl12

p=0.0001

p=0.015

Ccl2

vs. G

apdh

Ccl7

vs.G

apdh

DA/DA PVG/PVG DA/DA PVG/PVG

Ccl11

vs.

Gap

dh

DA/DA PVG/PVG

Ccl12

vs.Gap

dh

Ccl1

vs. G

apdh

DA/DAPVG/PVGDA/DA PVG/PVG

Figure 2 Genetically heterogeneous animals stratified on the genotype at Eae18b show differential expression of chemokines. mRNAexpression in lymph nodes collected day 7 post immunization from the thirteenth generation advanced intercross line (AIL) rats, with DAgenotype (n¼ 19) and PVG genotype (n¼ 32) at the Eae18b locus. The relative expression of chemokine mRNA is calculated using GapdhmRNA expression as reference. Mean values with standard error are presented, and statistical significance was calculated using Mann–Whitney non-parametrical test. P-values 40.05 are not shown.

0.001

0.01

0.1

1

0.05

rs28

5765

6 a

rs28

5765

7 b

rs45

86 c

rs13

900

d

rs30

9123

7 b

rs30

9132

2 e

rs47

9589

6 a

rs37

4450

8 f

rs18

6018

4 b

rs47

9589

8 b

rs31

3803

5 a

rs31

3803

6 b

rs31

3803

8 f

rs15

9313

b

rs20

7207

0 d

rs31

3668

2 d

rs31

3803

2 b

rs15

9279

a

CCL2 CCL7 CCL11 CCL8 CCL13 CCL1

Sweden ISweden IIDenmarkNorwayFinlandRA (ACPA-pos)

Figure 3 Single-nucleotide polymorphisms (SNPs) in the chemokine gene cluster on chromosome 17 are associated to multiple sclerosis(MS) and rheumatoid arthiritis (RA). Uncorrected P-values for associations in case–control cohorts for MS from Sweden (open and blackcircles), Denmark (open triangles), Norway (black circles) and Finland (gray diamonds) and a case–control cohort including anti-citrullinatedpeptide antibodies (ACPA) positive RA patients from Sweden (open boxes). The original screen was performed in Sweden I, and onlyselected SNPs were genotyped in all populations. Linkage disequilibrium for all included SNPs was calculated from Sweden I; color intensityindicates D0. The relative position and predicted function for each SNP is indicated by superscript letters: aupstream (50), bintronic, cexonicsynonymous, d30UTR, edownstream (30) and fexonic non-synonymous.

CC chemokines in EAE, MS and RAJ Ockinger et al

147

Genes and Immunity

tion to SNPs in CCL2, CCL8 and CCL1 (Figure 3,Supplementary Table 1a). We then proceeded to analyzethe haplotypes within the investigated genes andidentified association to haplotypes within CCL2 andCCL13 (Figure 4 and Table 3). As the chemokine genesare small and a relatively low number of SNPs weregenotyped, the analyzed haplotypes for each gene

included all genotyped SNPs within the gene. ForCCL2, a haplotype including three or four SNPs givevery similar results, as LD between rs4586 and rs13900 ishigh (r2: 0.69).

Meta-analysis of association to chemokine genes in five NordicMS cohorts. Selected SNPs in CCL2, 8, 13 and 1 (based

0.32 0.50 0.7 1.26 2.00

Sweden 1 (1023/1215)

Sweden 2 (952/665)

Norway (548/554)

Denmark (512/553)

Finland (806/1059)

Summary (3841/4046)

0.63 0.71 0.79 0.89 1.00 1.12 1.26 1.41

CCL1 RS3136682 C/TCCL2 GCC CCL13 TC

0.89 1.00 1.12 1.26 1.41

OR=0.86 (0.76:0.97)p=0.0134

OR=1.07 (1.01:1.14)p=0.033

OR =0.71 (0.56:0.91)p=0.0036

n.a

Odds Ratio

Figure 4 Genetic variants of chemokine genes are associated with multiple sclerosis (MS). Meta analysis of selected haplotypes in CCL2(rs2857656, rs2857657, rs4586) and CCL13 (rs159313, rs2072070), and rs3136682 in CCL1, including MS cohorts from four Nordic countries. Thesize of the gray squares are inversely correlated to the standard error of the mean (s.e.) in each population, lines represent the odds ratio (OR)and confidence interval. Black diamonds indicate the OR and confidence interval for the combined analysis. n.a.: not analyzed.

Table 3 Genetic variants of chemokine genes are associated with MS

(a)Country CCL2 GCC CCL13 TC CCL1 rs3136682 C/T

rs2857656-rs2857657-rs4586 rs159313-rs2072070

Cases Controls OR (CI) P-value Cases Controls OR (CI) P-value Cases Controls OR (CI) P-value

Sweden 1 0.077 0.095 0.80 (0.64:0.99) 0.0403 0.549 0.501 1.21 (1.08:1.37) 0.0016 0.031 0.043 0.72 (0.45:1.15) 0.1625Sweden 2 0.084 0.098 0.84 (0.66:1.07) 0.1618 0.549 0.527 1.09 (0.95:1.26) 0.2153 0.043 0.047 0.90 (0.56:1.45) 0.6585Norway 0.101 0.129 0.76 (0.58:0.99) 0.041 0.528 0.541 0.95 (0.80:1.13) 0.5621 0.034 0.033 1.04 (0.53:2.05) 0.9045Denmark 0.133 0.124 1.09 (0.84:1.41) 0.5211 0.544 0.539 1.02 (0.86:1.21) 0.3661 0.028 0.055 0.50 (0.26:0.97) 0.03335Finland NA NA NA NA 0.482 0.481 1.00 (0.88:1.14) 0.9554 0.025 0.044 0.55 (0.33:0.94) 0.01786

(b)

Country CCL13 CC

rs159313-rs2072070

Cases Controls OR (CI) P-value

Sweden 1 0.004 0.031 0.11 (0.05:0.24) 2.57� 10�11

Sweden 2 0.002 0.006 0.34 (0.10:1.20) 0.0798Norway 0.003 0.001 2.50 (0.30:20.60) 0.355Denmark 0.036 0.068 0.51 (0.34:0.77) 0.001Finland 0.003 0.004 0.83 (0.28:2.46) 0.7366

Abbreviation: CI, confidence interval; MS, multiple sclerosis; NA, not analyzed; OR, odds ratio.(a) Frequencies of haplotypes in CCL2 and CCL13 and allele frequencies in CCL1, odds ratio (OR) with confidence intervals (CI) and P-valuesfor each cohort included in the meta-analysis. (b) A rare haplotype in CCL13 is associated with reduced for risk of MS. Haplotype frequencies,odds ratios (OR) with confidence intervals (CI) and P-values for each cohort of a haplotype in CCL13 not included in the meta-analysisbecause of significant heterogeneity between sample cohorts.

CC chemokines in EAE, MS and RAJ Ockinger et al

148

Genes and Immunity

on significant and suggestive associations in the Swedishmaterial) were further tested in additional case–controlgroups from Sweden (Sweden 2) (952 cases, 655controls), Denmark (512 cases, 553 controls), Norway(548 cases, 554 controls) and Finland (806 cases, 1059controls). A number of associated SNPs and haplotypeswere identified in all Nordic cohorts (Figure 3, Supple-mentary Table 1b–e). Meta-analysis was carried out on allsingle SNPs and haplotypes, thus showing significant orsuggestive association. Haplotypes in CCL2 (GCC;rs2857656, rs2857657, rs4586) and 13 (TC; rs159313,rs2072070) as well as a single SNP (rs3136682) in CCL1were significant in meta-analysis (Figure 4, Table 3)(uncorrected P-values). The presented meta-analysespassed the test for heterogeneity,53 and were analyzedusing the Mantel–Haenszel method with fixed effects.One additional haplotype including a rare variant of theCCL13 gene (CC; rs159313, rs2072070) showed significantheterogeneity in allele frequency between the studygroups, and was therefore excluded from the meta-analysis. However, significant association was seen inSweden 1 and Denmark, indicating a possible impor-tance of this rare haplotype (Table 3b).

Investigation of the influence from HLA type, sex and othercovariates. The influence of the previously known riskgenes for MS, HLA-DRB1*152,3 and HLA-A*0254 on theassociations of chemokine genes were tested in all HLA-typed individuals in Sweden 1 (985 cases, 1141 controls).In this analysis of the HLA-DRB1*15 positive subgroup(615 cases, 330 controls), markers in CCL2 (rs2857656,P¼ 0.046, OR¼ 1.24 and rs13900, P¼ 0.031, OR¼ 1.27),CCL11 (rs1860184, P¼ 0.033, OR¼ 1.26) and CCL13(rs159313, P¼ 0.0055, OR¼ 1.52) showed association toMS (Supplementary Table 2a), also apparent when usinga multiplicative model55,56 for HLA-DRB1*15 and che-mokine gene effects (data not shown). In contrast, noSNPs were associated to MS in the HLA-DRB1*15negative subgroup (408 cases, 885 controls). However,no significant interaction between carrying a HLA-DRB1*15 allele and chemokine genotype was identifiedin a regression model, probably due to lack of sufficient

power. Analyses using stratification for the protectiveHLA-A*02 allele or sex, respectively, were carried out butas no significant effects on the association was detected(data not shown) these variables were excluded fromfurther analysis. Moreover, when Multiple SclerosisSeverity Score,57 known for 906 of the Swedish patients,was used as outcome variable in a linear regressionmodel, no association was shown between severity of MSand genotype in the chemokine genes. As HLA-DRB1*15was identified as a possible modifier of the association inSweden 1, the association was also evaluated in theDRB1*15-positive subgroups of Sweden 2 and cohortsfrom Norway, Denmark and Finland (SupplementaryTables 2b–e) and included in a subsequent meta-analysis(Figure 5, Table 4). This analysis identified a new variantof the CCL2 haplotype (GCT; rs2857656, rs2857657,rs4586) conferring susceptibility and confirmed theassociation to the CCL13 haplotype (TC; rs159313,rs2072070), (Figure 5). Association to the CC haplotypein CCL13 or rs3136682 in CCL1 could not be tested in theDRB1*15-positive subgroups, because of low minor-allele frequency.

CCL2 protein levels are upregulated in MS,and are determined by the CCL2 genotypeTo evaluate if the detected SNPs or haplotypes influencethe protein levels of the chemokines, all chemokinesincluded in the association studies were analyzed usingELISA. Only CCL2 was found in detectable amountsusing the selected method. mRNA levels of the chemo-kine genes were also investigated using real-timequantitative PCR, but were undetectable in most samples(data not shown). CSF samples from 38 patientsdiagnosed with MS were compared with 41 controlsamples from individuals diagnosed with other (non-inflammatory) neurological diseases. CCL2 levels in theCSF were higher in MS patients (0.35 ng ml�1), comparedwith other neurological diseases controls (0.24 ng ml�1),P-value¼ 0.0048 (Figure 6a). Furthermore, MS patientscarrying the C allele of rs4586 in the CCL2 gene hadhigher protein levels (0.45 ng ml�1) compared with theTT homozygous patients (0.31 ng ml�1) (P¼ 0.034)

0.79 1.00 1.26 1.58 2.00 2.51

Sweden 1 (615/330)

Sweden 2 (368/193)

Norway (315/164)

Denmark (206/31)

Summary (1649/978)

CCL2 GCT CCL13 TC

Odds Ratio0.56 0.71 0.89 1.12 1.41

Finland (245/260) n.a

OR: 1.22 (1.07,1.39)p=0.0031

OR: 1.14 (1.02,1.28)p=0.026

Figure 5 Haplotypes in CCL2 and CCL13 are associated to multiple sclerosis (MS) in the HLA-DR1*15 positive subgroups. Meta analysis ofhaplotypes in CCL2 (rs2857656, rs2857657, rs4586) and CCL13 (rs159313, rs2072070) in the HLA-DR1*15 positive subgroups from four Nordicpopulations. The size of the gray squares are inversely correlated to the standard error of the mean (s.e.) in each population, lines representthe odds ratio (OR) confidence interval. Black diamonds indicate the OR and confidence interval for the combined analysis. n.a.: notanalyzed.

CC chemokines in EAE, MS and RAJ Ockinger et al

149

Genes and Immunity

(Figure 6b). The SNP rs4586 is included in the CCL2haplotypes, and showed suggestive association in theinitial scan (Sweden 1), wherein the T allele is associatedwith MS.

Association between chemokine gene polymorphisms and RAA number of chemokines, including CCL2, have beensuggested to regulate the inflammatory processes in thejoints of RA patients.58 Production of CCL2 is upregu-lated in synovial fluid and sera in RA patients,59 and inaddition, a polymorphism in the proximity of CCL2showed suggestive association to RA in a small Spanishstudy.60 CCL13, produced by the chondrocytes incartilage, is upregulated in the RA joint and enhancesthe proliferation of synovial cells.61,62 In addition, thechemokine cluster region has showed suggestive sig-nificance in a whole genome-association study of ACPApositive RA patients.63

Therefore, we choose to include patients and controlsfrom the Swedish EIRA cohort and genotyped 2360 RApatients (1373 ACPA positive) and 958 controls for anumber of SNPs based on the association to MS. SingleSNPs in CCL2 (rs4586, P¼ 0.015, OR¼ 1.34), CCL8(rs3138038, P¼ 0.0059, OR¼ 2.37) and CCL13 (rs159313,P¼ 0.049, OR¼ 1.12) are associated with the risk of RA inACPA positive set (Figure 3, Supplementary Table 1f).When analyzing the whole group of patients, includingACPA negative patients, only the SNPs in CCL2 and

CCL8 were significant. The haplotypes identified in MSwere not found to be associated to RA.

Discussion

Despite many genes reported to be associated with MS,replications are generally lacking for majority, owing tothe small effect of MS genes, heterogeneity, gene–geneand gene–environment interactions. Additional strate-gies are needed to prioritize and support some of thesenumerous gene associations. In this study we have useda comparative genetic approach, combining evidencefrom a rat model of MS and association in several case–control cohorts for MS, in order to confirm and dissectthe contribution of a chemokine gene cluster. The high-resolution linkage analysis in the tenth generation of anAIL confirmed the previously established linkage to thisregion, and further reduced the confidence interval to a0.88 Mb region. The accumulation of recombinations inthe AIL enabled a more precise location, compared withprevious studies. We also used a bootstrap approach fora better positioning of the main genetic effect in theregion. Both the bootstrap approach and the classiclinkage analysis identified the same genetic location(marker D10Kini1) as the most probable for the QTL.Only 10 genes are located in the identified confidenceinterval in rat, including five chemokine genes (Ccl2, 7,

Table 4 HLA-DR1*15 status influences the association to haplotypes in the chemokine cluster

Country CCL2 GCT CCL13 TC

rs2857656-rs2857657-rs4586 rs159313-rs2072070

Cases Controls OR (CI) P-value Cases Controls OR (CI) P-value

Sweden 1 0.463 0.405 1.26 (1.04:1.54) 0.016 0.536 0.474 1.28 (1.06:1.56) 0.01Sweden 2 0.485 0.410 1.33 (1.03:1.73) 0.026 0.589 0.542 1.21 (0.94:1.57) 0.14Norway 0.442 0.436 1.03 (0.77:1.38) 0.89 0.527 0.532 0.97 (0.73:1.29) 0.89Denmark 0.429 0.327 1.50 (0.83:2.78) 0.17 0.539 0.555 0.96 (0.54:1.69) 0.89Finland NA NA NA NA 0.492 0.484 1.03 (0.80:1.33) 0.85

Abbreviations: CI, confidence interval; OR, odds ratio.Haplotype frequencies, odds ratios (OR) with confidence intervals (CI) and P-values for association, for haplotypes in CCL2 and CCL13identified in the HLA-DRB1*15 positive subgroups of four Nordic cohorts.

CC/CT TTMS OND0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

CC

L2 c

onc.

(ng

/ml)

CC

L2 c

onc.

(ng

/ml)

p = 0.0048 p = 0.039

Figure 6 The CCL2 concentration in cerebrospinal fluid (CSF) is (a) elevated in multiple sclerosis (MS) patients compared to controls, and is(b) regulated by genotype at CCL2 (rs4586). CCL2 levels in CSF from MS patients (n¼ 38) and controls (n¼ 41) diagnosed with other non-inflammatory neurological diseases (OND) were measured with ELISA. Median values are indicated, and statistical significance wascalculated using Mann–Whitney non-parametrical test.

CC chemokines in EAE, MS and RAJ Ockinger et al

150

Genes and Immunity

11, 12 and 1). Ccl2 and other chemokines in this regionhave been implicated as regulators of neuroinflamma-tion52 and regulators of infiltration of immune cells intothe central nervous system during EAE.27 The hereindefined confidence intervals do, however, not includethe second cluster of chemokines (including Ccl5, 9, 6, 3and 4) located further upstream (71.61–71.76 Mb) on ratchromosome 10, also suggested as modulators ofimmunoreactions in EAE.64

Using a genetically heterogeneous population ofanimals from the thirteenth generation of an AIL, andstratifying for the genotype in this region, someimportant findings were obtained. In the lymph nodesduring the induction of encephalitogenic immuneresponse, Ccl2 and Ccl1 are upregulated in animals withsusceptible genotype (DA), whereas Ccl11 is upregulatedin animals with the resistant genotype (PVG) in theregion. As the chemokines are known to attract andactivate immune cells with different specificity, it’s likelythat the composition and activation state of the immunecells attracted to the lymph nodes or the spinal cord bydifferent chemokines is influenced by this differentialexpression. It has also been suggested that hetero- orhomodimerization of chemokines can alter the signalingcapability,65 and the ratio of specific heterodimers couldbe affected by the differential expression seen here.However, additional studies will be required to describethe relationship between individual polymorphisms andregulation of transcription or expression. We are atpresent developing a congenic strain including theEae18b locus, which will be the optimal tool forevaluating the genetic effect and molecular mechanismsof this locus in EAE.

Genetic polymorphisms in the homologous region onhuman chromosome 17 have been studied for involve-ment in MS susceptibility,29–32 as well as other auto-immune or chronic inflammatory diseases includingRA,60 myocardial infarction66,67 and allergy.68 A secondcluster including chemokine genes (CCL5, CCL16, CCL14,CCL15, CCL23, CCL18, CCL3 and CCL4) located furtherupstream has also been studied in relation to MS, but inthis study we have focused on the centromeric cluster, asthis was identified in the rat linkage study. Throughmeta-analysis of several MS cohorts from Nordiccountries we could collect a large number of MS patientsand controls from a relatively homogenous populationand thereby identify association to haplotypes in CCL2and CCL13, and a single marker in CCL1. The hereindefined haplotypes in CCL2 and CCL13 both include oneSNP (rs2857657 and rs159313, respectively), which wereincluded in the previously defined haplotypes.30 As theLD is high within these genes, we can hereby confirm theassociation of haplotypes in CCL2 and CCL13 to MS,originally suggested in a smaller study.30 In addition, weidentified association to rs3136682 in CCL1, wherein CTheterozygous state confers protection. It is interesting tonote that no individuals homozygous for the T allelewere found among the 3841 cases and 4046 controlsincluded in the study, suggesting that this genotypecould be fatal or that the SNP is relatively new.Associated haplotypes in both CCL2 and CCL13 werealso found in the meta-analysis of the HLA-DRB1*15positive subgroups but not in the HLA-DRB1*15 nega-tive subgroups. In the original Swedish material, anassociation to rs1860184 in CCL11 was also identified in

the HLA-DRB1*15 positive subgroup. It is interesting tonote that the original GCC haplotype in CCL2 was notassociated to MS in the HLA-DRB1*15 positive sub-group, but instead the GCT variant of the haplotype wasidentified. The GCT haplotype is associated to anincreased risk in the HLA-DRB1*15 positive subgroup,in contrast to the GCC haplotype that is protective inthe whole group. Taken together, these discrepanciesbetween the whole cohorts and HLA-DRB1*15 positiveand negative subgroups, respectively, indicate a modify-ing effect from the HLA locus, although no significantinteraction was detected.

Single-nucleotide polymorphisms in CCL2, CCL8 andCCL13 were also shown to be associated with ACPApositive RA. Even though the haplotypes identified inMS are not associated to RA, these findings point out theimportance of chemokine signaling in disease develop-ment and should encourage the investigation of thesegenes in other inflammatory diseases. However, not allinflammatory diseases share the same risk alleles, forexample PTPN22 is associated to RA, SLE, T1D andautoimmune thyroiditis,69,70 but not MS or Crohn’sdisease.71,72 Most likely this reflects differences in thedevelopment of different inflammatory diseases. Thediscrepancies in association to the chemokine genescould thus indicate that distinct sets of chemokinesorchestrate the infiltration of the target organs in MS andRA. Moreover, even when the same gene is indicated inseveral inflammatory diseases, individual polymorph-isms may have different or even opposite effect ondisease susceptibility, as seen for the ILR2A gene in MSand T1D.73 An analogy is seen here for CCL2 and CCL13,wherein single SNPs are associated to RA but haplotypesare associated to MS.

The associations observed in the individual cohorts aswell as in the meta-analyses provide solid evidence forthe association to MS, even though the effects contrib-uted from the genetic polymorphisms in the chemokinegene cluster are small. It could be speculated that thegenetic polymorphism in the human chemokine clusteraffects the severity of disease rather than susceptibility,as in rats. Here, the association to Multiple SclerosisSeverity Score was tested, but larger materials with evenbetter defined variables for disease severity would beadvantageous for these studies.

The association of SNPs in the chemokine gene clusterto both MS and RA prompted us to investigate thebiological effects of these polymorphisms. The primaryorgan in MS is the central nervous system andinvestigation of biological processes within this organwould reveal clues to the pathogenesis of MS. However,brain biopsies are rare and post mortem samplestypically involve late-stage disease or unusual circum-stances. In contrast, CSF is routinely sampled from MSpatients for diagnostic purposes, and may to some extentmirror the inflammatory events of the central nervoussystem. In this study we have established that CCL2levels are higher in the CSF from patients with arelapsing remitting MS disease course, compared withcontrols. This is in contrast to a number of previousstudies.74,75 This inconsistency between studies couldreflect differences in CCL2 production during differentstages of disease, differences between populations ordifferent composition of control groups in the studies, ashealthy controls are rarely used. In line with genetic

CC chemokines in EAE, MS and RAJ Ockinger et al

151

Genes and Immunity

association of CCL2 with MS, we could also establish anassociation between CCL2 genotype and CCL2 produc-tion; carrying one or two C alleles at rs4586 is associatedwith a higher CCL2 production, compared with a Thomozygous genotype. It is interesting to note that thesuggestive association to this SNP points out the Thomozygous genotype as a risk factor for MS, thusindicating that the relation between CCL2 levels in CSFand MS is not straightforward.

CCL2, CCL13 and CCL1, showing association with MS,have previously been investigated for involvement inhuman and experimental neuroinflammation, as well asother inflammatory diseases. For example, CCL2 canbind to a number of receptors, including CCR2, 5, 10,DARC and D6, and is rather well described in MS,although the genetic association has been ambiguousuntil now. Indeed, CCL2 is found in MS lesions26 andactivates macrophages/microglia in secondary progres-sive MS.76 CCL1 on the other hand, has only oneidentified receptor, CCR8, and its function in MS isunknown, but has been proposed as a regulator ofmacrophage activation in diabetes.77 The structure ofCCL13 was recently described78 and little is known aboutits function in MS, although the expression in peripheralT cells is downregulated upon therapeutic IVIG treat-ment in MS.79 Furthermore, CCL13 has been shown to beupregulated in the synovial fluid in RA patients, and is aregulator of proliferation of synovial cells.62,61 It is to benoted that CCL13 signals through receptors CCR2, 3, 5and D6 as do other chemokines in this cluster (CCL7, 8and 11). No direct homolog of the CCL13 gene is found inrats, but the gene shows a high sequence similarity to ratCcl11 (70%), whereas human CCL11 and rat Ccl11 share77% sequence similarity (www.ensembl.org). We there-fore hypothesize that CCL13 might be a human specificfunctional homolog of CCL11.

In conclusion, we have here demonstrated associationbetween polymorphisms in a chemokine gene clusterand inflammatory diseases in both rats and humans,and suggest genetically determined regulation ofexpression and protein levels as the main regulatorymechanism.

Conflict of interest

The authors declare no conflict of interest.

Acknowledgements

We thank Dr Mohsen Khademi and Maria Karlsson fortechnical assistance and Dr Boel Brynedal for criticalreview of the paper. We also thank all the patients andvolunteers who contributed to this study. All involvedhospital staff are thanked for their contribution, inparticular the neurologists Tuula Pirttila (University ofKuopio), Irina Elovaara (University of Tampere) andMauri Reunanen (University of Oulu) for their efforts inrecruiting Finnish MS patients, and Professor LarsKlareskog (Karolinska Institutet) for recruiting patientsto the Swedish RA cohort. This work was supported bygrants from the Swedish Research Council, EURATools(LSHG-CT-2005-019015), Neuropromise (LSHM-CT-2005-018637), the Swedish Association for Persons with

Neurological Disabilities, the Finnish Academy, theSigrid Juselius Foundation and Helsinki UniversityCentral Hospital.

References

1 Jersild C, Hansen G, Svejgaard A, Fog T, Thomsen M, DupontB. Histocompatibility determinants in multiple sclerosis,with special reference to clinical course. The Lancet 1973; 302:1221–1225.

2 Hillert J, Olerup O. Multiple sclerosis is associated with geneswithin or close to the HLA-DR-DQ subregion on a normalDR15,DQ6,Dw2 haplotype. Neurology 1993; 43: 163–168.

3 Lincoln MR, Montpetit A, Cader MZ, Saarela J, Dyment DA,Tiislar M et al. A predominant role for the HLA class II regionin the association of the MHC region with multiple sclerosis.Nat Genet 2005; 37: 1108–1112.

4 Lundmark F, Duvefelt K, Iacobaeus E, Kockum I, Wallstrom E,Khademi M et al. Variation in interleukin 7 receptor alphachain (IL7R) influences risk of multiple sclerosis. Nat Genet2007; 39: 1108–1113.

5 International Multiple Sclerosis Genetics Consortium, Hafler DA,Compston A, Sawcer S, Lander ES, Daly MJ et al. Risk allelesfor multiple sclerosis identified by a genomewide study.N Engl J Med 2007; 357: 851–862.

6 Anaya JM, Gomez L, Castiblanco J. Is there a common geneticbasis for autoimmune diseases? Clin Dev Immunoly 2006; 13:185–195.

7 Sigurdsson S, Nordmark G, Goring HH, Lindroos K, WimanAC, Sturfelt G et al. Polymorphisms in the tyrosine kinase 2and interferon regulatory factor 5 genes are associatedwith systemic lupus erythematosus. Am J Hum Genet 2005;76: 528–537.

8 Sigurdsson S, Padyukov L, Kurreeman FA, Liljedahl U,Wiman AC, Alfredsson L et al. Association of a haplotype inthe promoter region of the interferon regulatory factor 5gene with rheumatoid arthritis. Arthritis Rheum 2007; 56:2202–2210.

9 Dideberg V, Kristjansdottir G, Milani L, Libioulle C, Sigurds-son S, Louis E et al. An insertion-deletion polymorphism inthe interferon regulatory factor 5 (IRF5) gene confers riskof inflammatory bowel diseases. Hum Mol Genet 2007; 16:3008–3016.

10 Kristjansdottir G, Sandling JK, Bonetti A, Roos IM, Milani L,Wang C et al. Interferon regulatory factor 5 (IRF5) genevariants are associated with multiple sclerosis in three distinctpopulations. J Med Genet 2008; 45: 362–369.

11 International Multiple Sclerosis Genetics Consortium. Theexpanding genetic overlap between multiple sclerosis andtype I diabetes. Genes Immun 2009; 10: 11–14.

12 Todd JA, Walker NM, Cooper JD, Smyth DJ, Downes K,Plagnol V et al. Robust associations of four new chromosomeregions from genome-wide analyses of type 1 diabetes. NatGenet 2007; 39: 857–864.

13 Hakonarson H, Grant SF, Bradfield JP, Marchand L, Kim CE,Glessner JT et al. A genome-wide association study identifiesKIAA0350 as a type 1 diabetes gene. Nature 2007; 448: 591–594.

14 Zoledziewska M, Costa G, Pitzalis M, Cocco E, Melis C, Moi Let al. Variation within the CLEC16A gene shows consistentdisease association with both multiple sclerosis and type 1diabetes in Sardinia. Genes Immun 2009; 10: 15–17.

15 Ueda H, Howson JM, Esposito L, Heward J, Snook H,Chamberlain G et al. Association of the T-cell regulatory geneCTLA4 with susceptibility to autoimmune disease. Nature2003; 423: 506–511.

16 Yokoi N, Komeda K, Wang HY, Yano H, Kitada K, Saitoh Yet al. Cblb is a major susceptibility gene for rat type 1 diabetesmellitus. Nat Genet 2002; 31: 391–394.

CC chemokines in EAE, MS and RAJ Ockinger et al

152

Genes and Immunity

17 Dahlman I, Wallstrom E, Weissert R, Storch M,Kornek B, Jacobsson L et al. Linkage analysis of myelinoligodendrocyte glycoprotein-induced experimental autoim-mune encephalomyelitis in the rat identifies a locus control-ling demyelination on chromosome 18. Hum Mol Genet 1999; 8:2183–2190.

18 Jagodic M, Kornek B, Weissert R, Lassmann H, Olsson T,Dahlman I. Congenic mapping confirms a locus on ratchromosome 10 conferring strong protection against myelinoligodendrocyte glycoprotein-induced experimental autoim-mune encephalomyelitis. Immunogenetics 2001; 53: 410–415.

19 Jagodic M, Becanovic K, Sheng JR, Wu X, Backdahl L,Lorentzen JC et al. An advanced intercross line resolvesEae18 into two narrow quantitative trait loci syntenic tomultiple sclerosis candidate loci. J Immunol 2004; 173:1366–1373.

20 Furuya T, Salstrom JL, McCall-Vining S, Cannon GW, Joe B,Remmers EF et al. Genetic dissection of a rat model forrheumatoid arthritis: significant gender influences on auto-somal modifier loci. Hum Mol Genet 2000; 9: 2241–2250.

21 Joe B, Cannon GW, Griffiths MM, Dobbins DE, Gulko PS,Wilder RL et al. Evaluation of quantitative trait loci regulatingseverity of mycobacterial adjuvant-induced arthritis in mono-congenic and polycongenic rats: identification of a newregulatory locus on rat chromosome 10 and evidence ofoverlap with rheumatoid arthritis susceptibility loci. ArthritisRheum 2002; 46: 1075–1085.

22 Joe B, Remmers EF, Dobbins DE, Salstrom JL, Furuya T,Dracheva S et al. Genetic dissection of collagen-inducedarthritis in chromosome 10 quantitative trait locus speedcongenic rats: evidence for more than one regulatory locusand sex influences. Immunogenetics 2000; 51: 930–944.

23 The Transatlantic Multiple Sclerosis Genetics Cooperative. Ameta-analysis of genomic screens in multiple sclerosis. MultScler 2001; 7: 3–11.

24 Griffiths-Johnson DA, Collins PD, Rossi AG, Jose PJ, WilliamsTJ. The chemokine, eotaxin, activates guinea-pig eosinophilsin vitro and causes their accumulation into the lung in vivo.Biochem Biophys Res Commun 1993; 197: 1167–1172.

25 Glabinski AR, Tani M, Tuohy VK, Tuthill RJ, Ransohoff RM.Central nervous system chemokine mRNA accumulationfollows initial leukocyte entry at the onset of acute murineexperimental autoimmune encephalomyelitis. Brain, BehavImmun 1995; 9: 315–330.

26 McManus C, Berman JW, Brett FM, Staunton H, Farrell M,Brosnan CF. MCP-1, MCP-2 and MCP-3 expression in multiplesclerosis lesions: an immunohistochemical and in situ hybri-dization study. J Neuroimmunol 1998; 86: 20–29.

27 dos Santos AC, Barsante MM, Arantes RM, Bernard CC,Teixeira MM, Carvalho-Tavares J. CCL2 and CCL5 mediateleukocyte adhesion in experimental autoimmune encephalo-myelitis—an intravital microscopy study. J Neuroimmunol2005; 162: 122–129.

28 Banisor I, Leist T, Kalman B. Involvement of beta-chemokinesin the development of inflammatory demyelination. J Neuroin-flammation 2005; 2: 7.

29 Vyshkina T, Kalman B. Haplotypes within genes of I2-chemokines in 17q11 are associated with multiple sclerosis: asecond phase study. Hum Genet 2005; V118: 67.

30 Vyshkina T, Shugart YY, Birnbaum G, Leist TP, Kalman B.Association of haplotypes in the [beta]-chemokine locus withmultiple sclerosis. Eur J Hum Genet 2004; 13: 240.

31 Bugeja MJ, Booth D, Bennetts B, Heard R, Rubio J, Stewart G.An investigation of polymorphisms in the 17q11.2-12 CCchemokine gene cluster for association with multiple sclerosisin Australians. BMC Med Genet 2006; 7: 64.

32 Vyshkina T, Sylvester A, Sadiq S, Bonilla E, Perl A, Kalman B.CCL genes in multiple sclerosis and systemic lupus erythe-matosus. J Neuroimmunol 2008; 200: 145–152.

33 Darvasi A. Experimental strategies for the genetic dissectionof complex traits in animal models. Nat Genet 1998; 18: 19.

34 Ockinger J, Serrano-Fernandez P, Moller S, Ibrahim SM,Olsson T, Jagodic M. Definition of a 1.06-Mb region linkedto neuroinflammation in humans, rats and mice. Genetics 2006;173: 1539–1545.

35 Amor S, Groome N, Linington C, Morris MM, Dornmair K,Gardinier MV et al. Identification of epitopes of myelinoligodendrocyte glycoprotein for the induction of experimen-tal allergic encephalomyelitis in SJL and Biozzi AB/H mice. JImmunol 1994; 153: 4349–4356.

36 Laird PW, Zijderveld A, Linders K, Rudnicki MA, Jaenisch R,Berns A. Simplified mammalian DNA isolation procedure.Nucleic Acids Res 1991; 19: 4293.

37 Jacob HJ, Brown DM, Bunker RK, Daly MJ, Dzau VJ,Goodman A et al. A genetic linkage map of the laboratoryrat, Rattus norvegicus. Nat Genet 1995; 9: 63.

38 Broman KW, Wu H, Sen S, Churchill GA. R/qtl: QTL mappingin experimental crosses. Bioinformatics (Oxford, England) 2003;19: 889–890.

39 Manichaikul A, Dupuis J, Sen S, Broman KW. Poor perfor-mance of bootstrap confidence intervals for the location of aquantitative trait locus. Genetics 2006; 174: 481–489.

40 McDonald WI, Compston A, Edan G, Goodkin D, HartungHP, Lublin FD et al. Recommended diagnostic criteria formultiple sclerosis: guidelines from the International Panelon the diagnosis of multiple sclerosis. Ann Neurol 2001; 50:121–127.

41 Roos IM, Kockum I, Hillert J. The interleukin 23 receptor genein multiple sclerosis: a case-control study. J Neuroimmunol2008; 194: 173–180.

42 Poser CM, Paty DW, Scheinberg L, McDonald WI, Davis FA,Ebers GC et al. New diagnostic criteria for multiplesclerosis: guidelines for research protocols. Ann Neurol 1983;13: 227–231.

43 Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF,Cooper NS et al. The American Rheumatism Association 1987revised criteria for the classification of rheumatoid arthritis.Arthritis Rheum 1988; 31: 315–324.

44 Klareskog L, Stolt P, Lundberg K, Kallberg H, Bengtsson C,Grunewald J et al. A new model for an etiology of rheumatoidarthritis: smoking may trigger HLA-DR (shared epitope)-restricted immune reactions to autoantigens modified bycitrullination. Arthritis Rheum 2006; 54: 38–46.

45 Stolt P, Bengtsson C, Nordmark B, Lindblad S, Lundberg I,Klareskog L et al. Quantification of the influence of cigarettesmoking on rheumatoid arthritis: results from a populationbased case-control study, using incident cases. Ann Rheum Dis2003; 62: 835–841.

46 Gonzalez JR, Armengol L, Sole X, Guino E, Mercader JM,Estivill X et al. SNPassoc: an R package to perform wholegenome association studies. Bioinformatics (Oxford, England)2007; 23: 644–645.

47 R Development Core Team. R: A Language and Environment forStatistical Computing. R Foundation for Statistical Computing:Vienna, 2009.

48 Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis andvisualization of LD and haplotype maps. Bioinformatics(Oxford, England) 2005; 21: 263–265.

49 Dudbridge F. Likelihood-based association analysis for nucle-ar families and unrelated subjects with missing genotype data.Hum Hered 2008; 66: 87–98.

50 Darvasi A, Soller M. Advanced intercross lines, an experi-mental population for fine genetic mapping. Genetics 1995;141: 1199.

51 Kennedy KJ, Strieter RM, Kunkel SL, Lukacs NW, Karpus WJ.Acute and relapsing experimental autoimmune encephalo-myelitis are regulated by differential expression of the CCchemokines macrophage inflammatory protein-1alphaand monocyte chemotactic protein-1. J Neuroimmunol 1998;92: 98–108.

52 Jee Y, Yoon WK, Okura Y, Tanuma N, Matsumoto Y.Upregulation of monocyte chemotactic protein-1 and CC

CC chemokines in EAE, MS and RAJ Ockinger et al

153

Genes and Immunity

chemokine receptor 2 in the central nervous system is closelyassociated with relapse of autoimmune encephalomyelitis inLewis rats. J Neuroimmunol 2002; 128: 49–57.

53 Woolf B. On estimating the relation between blood group anddisease. Ann Hum Genet 1955; 19: 251–253.

54 Brynedal B, Duvefelt K, Jonasdottir G, Roos IM, Akesson E,Palmgren J et al. HLA-A confers an HLA-DRB1 independentinfluence on the risk of multiple sclerosis. PLoS ONE 2007; 2:e664.

55 Modin H, Olsson W, Hillert J, Masterman T. Modes of actionof HLA-DR susceptibility specificities in multiple sclerosis.Am J Hum Genet 2004; 74: 1321–1322.

56 Barcellos LF, Oksenberg JR, Begovich AB, Martin ER, SchmidtS, Vittinghoff E et al. HLA-DR2 dose effect on susceptibility tomultiple sclerosis and influence on disease course. Am J HumGenet 2003; 72: 710–716.

57 Roxburgh RH, Seaman SR, Masterman T, Hensiek AE, SawcerSJ, Vukusic S et al. Multiple Sclerosis Severity Score: usingdisability and disease duration to rate disease severity.Neurology 2005; 64: 1144–1151.

58 Haringman JJ, Smeets TJ, Reinders-Blankert P, Tak PP.Chemokine and chemokine receptor expression in pairedperipheral blood mononuclear cells and synovial tissue ofpatients with rheumatoid arthritis, osteoarthritis, and reactivearthritis. Ann Rheum Dis 2006; 65: 294–300.

59 Koch AE, Kunkel SL, Harlow LA, Johnson B, Evanoff HL,Haines GK et al. Enhanced production of monocyte chemoat-tractant protein-1 in rheumatoid arthritis. J Clin Invest 1992; 90:772–779.

60 Gonzalez-Escribano MF, Torres B, Aguilar F, Rodrıguez R,Garcıa A, Valenzuela A et al. MCP-1 promoter polymorphismin Spanish patients with rheumatoid arthritis. Hum Immunol2003; 64: 741–744.

61 Iwamoto T, Okamoto H, Kobayashi S, Ikari K, Toyama Y,Tomatsu T et al. A role of monocyte chemoattractant protein-4(MCP-4)/CCL13 from chondrocytes in rheumatoid arthritis.FEBS J 2007; 274: 4904–4912.

62 Iwamoto T, Okamoto H, Iikuni N, Takeuchi M, Toyama Y,Tomatsu T et al. Monocyte chemoattractant protein-4 (MCP-4)/CCL13 is highly expressed in cartilage from patients withrheumatoid arthritis. Rheumatology (Oxford, England) 2006; 45:421–424.

63 Plenge RM, Seielstad M, Padyukov L, Lee AT, Remmers EF,Ding B et al. TRAF1-C5 as a risk locus for rheumatoidarthritis—a genomewide study. N Engl J Med 2007; 357:1199–1209.

64 Miyagishi R, Kikuchi S, Takayama C, Inoue Y, Tashiro K.Identification of cell types producing RANTES, MIP-1alpha and MIP-1 beta in rat experimental autoimmuneencephalomyelitis by in situ hybridization. J Neuroimmunol1997; 77: 17–26.

65 Crown SE, Yu Y, Sweeney MD, Leary JA, Handel TM.Heterodimerization of CCR2 chemokines and regulationby glycosaminoglycan binding. J Biol Chem 2006; 281:25438–25446.

66 Alioglu E, Turk U, Cam S, Abbasaliyev A, Tengiz I, Ercan E.Polymorphisms of the methylenetetrahydrofolate reductase,vascular endothelial growth factor, endothelial nitric oxidesynthase, monocyte chemoattractant protein-1 and apolipo-

protein E genes are not associated with carotid intima-mediathickness. Can J Cardiol 2009; 25: e1–e5.

67 McDermott DH, Yang Q, Kathiresan S, Cupples LA, MassaroJM, Keaney Jr JF et al. CCL2 polymorphisms are associatedwith serum monocyte chemoattractant protein-1 levels andmyocardial infarction in the Framingham Heart Study.Circulation 2005; 112: 1113–1120.

68 Rigoli L, Caminiti L, Di Bella C, Procopio V, Cuppari C, Vita Det al. Investigation of the eotaxin gene �426C-T, �384A-Gand 67G-a single-nucleotide polymorphisms and atopicdermatitis in Italian children using family-based associationmethods. Clin Exp Dermatol 2008; 33: 316–321.

69 Hinks A, Barton A, John S, Bruce I, Hawkins C, Griffiths CEet al. Association between the PTPN22 gene and rheumatoidarthritis and juvenile idiopathic arthritis in a UK population:further support that PTPN22 is an autoimmunity gene.Arthritis Rheum 2005; 52: 1694–1699.

70 Criswell LA, Pfeiffer KA, Lum RF, Gonzales B, Novitzke J,Kern M et al. Analysis of families in the multiple autoimmunedisease genetics consortium (MADGC) collection: the PTPN22620W allele associates with multiple autoimmune phenotypes.Am J Hum Genet 2005; 76: 561–571.

71 Begovich AB, Caillier SJ, Alexander HC, Penko JM, Hauser SL,Barcellos LF et al. The R620W polymorphism of the proteintyrosine phosphatase PTPN22 is not associated with multiplesclerosis. Am J Hum Genet 2005; 76: 184–187.

72 De Jager PL, Sawcer S, Waliszewska A, Farwell L, Wild G,Cohen A et al. Evaluating the role of the 620W allele of proteintyrosine phosphatase PTPN22 in Crohn’s disease and multiplesclerosis. Eur J Hum Genet 2006; 14: 317–321.

73 Maier LM, Lowe CE, Cooper J, Downes K, Anderson DE,Severson C et al. IL2RA genetic heterogeneity in multiplesclerosis and type 1 diabetes susceptibility and solubleinterleukin-2 receptor production. PLoS Genet 2009; 5:e1000322.

74 Sorensen TL, Tani M, Jensen J, Pierce V, Lucchinetti C, FolcikVA et al. Expression of specific chemokines and chemokinereceptors in the central nervous system of multiple sclerosispatients. J Clin Invest 1999; 103: 807–815.

75 Sindern E, Niederkinkhaus YM, Henschel LM, Ossege T,Patzold J, Malin P. Differential release of beta-chemokines inserum and CSF of patients with relapsing-remitting multiplesclerosis. Acta Neurol Scand 2001; 104: 88–91.

76 Tanuma N, Sakuma H, Sasaki A, Matsumoto Y. Chemokineexpression by astrocytes plays a role in microglia/macro-phage activation and subsequent neurodegeneration insecondary progressive multiple sclerosis. Acta Neuropathologi-ca 2006; 112: 195–204.

77 Cantor J, Haskins K. Recruitment and activation of macro-phages by pathogenic CD4 T cells in type 1 diabetes: evidencefor involvement of CCR8 and CCL1. J Immunol 2007; 179:5760–5767.

78 Barinka C, Prahl A, Lubkowski J. Structure of humanmonocyte chemoattractant protein 4 (MCP-4/CCL13). ActaCrystallographica 2008; 64 (Part 3): 273–278.

79 Pigard N, Elovaara I, Kuusisto H, Paalavuo R, Dastidar P,Zimmermann K et al. Therapeutic activities of intravenousimmunoglobulins in multiple sclerosis involve modulation ofchemokine expression. J Neuroimmunol 2009; 209: 114–120.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

CC chemokines in EAE, MS and RAJ Ockinger et al

154

Genes and Immunity