New humanized HLA–DR4–transgenic mice that mimic the sex bias of rheumatoid arthritis

ARTHRITIS & RHEUMATISMVol. 56, No. 1, January 2007, pp 69–78DOI 10.1002/art.22213© 2007, American College of Rheumatology

New Humanized HLA–DR4–Transgenic Mice That Mimic theSex Bias of Rheumatoid Arthritis

Veena Taneja,1 Marshall Behrens,1 Ashutosh Mangalam,1 Marie M. Griffiths,2

Harvinder S. Luthra,1 and Chella S. David1

Objective. To generate a mouse model that canmimic human rheumatoid arthritis (RA). A major dif-ference between RA in humans and collagen-inducedarthritis (CIA) in mice is the lack of sex bias andautoantibodies in the animal model. We usedDRB1*0401-transgenic mice to understand the role ofDR4 in susceptibility and sex bias in RA.

Methods. A transgenic mouse was generated thatlacked all endogenous mouse class II genes (AEo) andexpressed the RA susceptibility allele HLA–DRB1*0401.These transgenic mice were tested for incidence, sever-ity, and sex distribution of CIA.

Results. DRB1*0401.AEo mice developed CIA pre-dominantly in females and produced rheumatoid fac-tors, similar to the features of human RA. Anotherfeature similar to human RA is the expression of class IImolecules on antigen-presenting cells as well as T cells.Activated and sorted CD4� T cells can present DR4-restricted type II collagen (CII)–derived peptide invitro, but cannot process the antigen. This suggests arole for these cells in epitope presentation locally injoints, which affects disease severity. After challengewith CII, female mice had higher cellularity and in-creased T cell proliferation and produced higher levelsof proinflammatory cytokines than did the male mice.

Conclusion. DR4.AEo mice expressed HLA simi-lar to humans and displayed increased arthritis suscep-tibility in females, thus mimicking RA in humans. This

model may be valuable for studying sex differencesobserved in humans and for understanding why auto-immunity is increased in women. These mice may alsobe useful for developing future therapeutic strategies.

Rheumatoid arthritis (RA) is an autoimmunedisease that results from a complex interplay of bothgenetic and environmental factors. The major geneticcontribution comes from genes located in the HLA classII region. In most populations, DRB1*0401, 0404, and0101 are associated with RA. Gregersen and coworkers(1) suggested a “shared epitope” hypothesis to explainthe association of various DR alleles with arthritis.According to the hypothesis, DRB1 alleles sharing amotif at positions 67, 70, 71, and 74 are implicated in thepathogenesis of RA by their ability to present similarantigens, leading to the development of arthritis.DRB1*0402 was found to be associated with non-susceptibility to RA; however, both DRB1*0401 andDRB1*0402 can present the DR4-restricted immuno-dominant peptide of type II collagen (CII) (2,3).

Another hypothesis suggested that “shared epi-topes” could be involved in T cell receptor (TCR) selectionbecause peptides from DRB1 alleles are presented byHLA–DQ in the thymus (4). This is supported by thefinding that naturally processed peptides presented by classII molecules are derived from endogenous class II mole-cules (5). However, it is difficult to experimentally assessthe role of a single allele and its contribution to theshared epitope in order to confirm this hypothesis inhumans.

RA is a chronic inflammatory disease character-ized by synovial inflammation and erosion of bone andcartilage, which lead to the destruction of joints. Patientswith RA have CII-reactive T cells and antibodies, whichsuggests that CII is a candidate autoantigen in diseasepathogenesis (6). Collagen-induced arthritis (CIA) hasbeen used for the last 2 decades as an experimentalmodel with which to study RA. Our previous studies

Supported by NIH grant AR-30752. The HLA class II–transgenic mice were produced with support from NIH grant AI-14764. Dr. Taneja is recipient of an Arthritis Investigator award fromthe Arthritis Foundation.

1Veena Taneja, PhD, Marshall Behrens, BS, Ashutosh Man-galam, PhD, Harvinder S. Luthra, MD, Chella S. David, PhD: MayoClinic, Rochester, Minnesota; 2Marie M. Griffiths, PhD: Departmentof Veterans Affairs Medical Center, and University of Utah, Salt LakeCity, Utah.

Address correspondence and reprint requests to VeenaTaneja, PhD, Department of Immunology, Mayo Clinic, 200 FirstStreet SW, Rochester, MN 55905. E-mail: [email protected].

Submitted for publication March 12, 2006; accepted in revisedform August 16, 2006.

69

have shown that A�o-transgenic mice expressing HLA–DQA1*0301/DQB1*0302 are highly susceptible to CIA(7) and can present multiple human CII peptides (8).Previous studies using DR4/H-2 mice showed that DR4could modulate arthritis, thus suggesting an importantrole of this haplotype in RA (3,9,10).

In the present study, the DRB1*0401.A�o-transgenic mice, which lack functional A� and E� genesbut which express A� and E� genes, did not developCIA. A potential problem with these mice is the expres-sion of endogenous H-2E chain. To overcome thisproblem, we mated DRB1*0401.A�o mice withMHCII�/� mice, which lack all of the endogenous classII molecules (11), to generate DR4.AEo mice that lackall 4 classic murine chains (A�, A�, E�, and E�). Theonly class II molecules expressed in these transgenicmice are DR4 molecules (E�DRB1*0401). Unlike en-dogenous class II molecules, HLA–DR molecules intransgenic mice are expressed on a subset of CD3� Tcells that are also known to be present in humans. In vivostudies after induction of CIA showed an increasedsusceptibility to disease in female DR4 mice. A sex biasin terms of disease was not observed in the DQ8.AEo

mice. The observations in DR4.AEo-transgenic mice aresimilar to those in humans with RA and may provide amodel with which to study sex bias for arthritis inhumans.

MATERIALS AND METHODS

Transgenic mice. The generation of HLA–DRB1*0401(DR4)–transgenic mice has been described previously (12).DRB1*0401.A�o mice were mated with MHCII�/� (AEo) mice(11) to generate DR4.AEo mice. In a similar manner, wegenerated DQ8.AEo mice. DR4.A�o and DQ8.A�o mice (3,7)were used as controls. Mice of both sexes between the ages of8 and 12 weeks were used in this study and were bred andmaintained in the pathogen-free Immunogenetics Mouse Col-ony at the Mayo Clinic in accordance with the InstitutionalAnimal Care and Use Committee (IACUC). All experimentsincluded littermate controls and were performed with theapproval of the IACUC.

Flow cytometry. The expression of DR�, H-2E, andTCR V�-chains on peripheral blood lymphocytes from trans-genic mice were analyzed by flow cytometry using the followingmonoclonal antibodies: L227 (anti-DR), IVD12 (anti-DQ),14-4-4s (anti-E�), and GK1.5 (anti–mouse CD4) (AmericanType Culture Collection, Manassas, VA), and B20.6 (anti-V�2), KT4-10 (anti-V�4), MR9-4 (anti-V�5.1.2), MR9-8(anti-V�5.1), 44-22-1 (anti-V�6), TR310 (anti-V�7), KJ-16(anti-V�8.1.2), F23.2 (anti-V�8.2), MR10-2 (anti-V�9), KT11(anti-V�11), 14.2 (anti-V�14), and KJ23a (anti-V�17) (BDPharMingen, San Diego, CA). Conjugated antibodies againstCD3, CD4, CD8, Mac-1, and B220 (BD PharMingen) werealso used. For analysis of DR expression on activated cells,primed spleen cells were challenged in vitro with CII or

staphylococcal enterotoxin B (SEB). All cell surface markerswere analyzed in cells pooled from 2 mice per strain, andexperiments were performed 2–3 times.

Induction and evaluation of CIA. Pure native chick CIIwas obtained by multiple-step purification as described previ-ously (13). CIA was induced in transgenic animals by immu-nization with CII (100 �g of CII emulsified in Freud’s com-plete adjuvant [CFA]) according to the standard protocoldescribed elsewhere (3). Mice were studied in 2 groups toensure reproducibility of results. The onset and progression ofCIA was monitored from week 3 through week 12 postimmu-nization. The severity of arthritis in each paw was graded on ascale of 0–3, as described previously (14). The mean arthritisscore was determined in arthritic animals only.

Histopathologic analysis. Mice were killed 10–12weeks after immunization, and the paws were removed, decal-cified, and fixed. Sections were stained with hematoxylin andeosin and examined for infiltrations and erosions. Spleen,thymus, liver, salivary gland, and kidney were removed fromnaive DR4 mice, then frozen and sectioned. Sections werestained with anti-DR antibody (L227) that had been conju-gated with fluorescein isothiocyanate or phycoerythrin, andsome sections were stained with CD4 that had been conjugatedwith phycoerythrin. Slides were also stained with 4�,6-diamidino-2-phenylindole.

T cell proliferation assay. Mice were immunized intra-dermally at the base of the tail and 1 hind footpad with 200 �gof CII emulsified 1:1 in CFA (Difco, Detroit, MI). Ten dayspostimmunization, draining lymph nodes/spleen were removedand cultured in vitro. Lymph node cells (1 � 106) werecultured in HEPES buffered RPMI 1640 containing 5% heat-inactivated horse serum and antibiotics (streptomycin andpenicillin) in 96-well flat-bottomed tissue culture plates. Cellswere challenged by the addition of RPMI 1640 (100 �l;negative control), concanavalin A (20 �g/ml, positive control),or native collagen (50 �g/ml).

For inhibition experiments, culture supernatant con-taining 25 �g/ml of monoclonal antibody GK1.5 (anti-CD4),anti-DR (L227), or Lyt2 (anti-CD8) was added to the cells andchallenged in vitro with 50 �g/ml of CII. The cells wereincubated for 48 hours at 37°C. During the last 18 hours, thecells were pulsed with 3H-thymidine (1 �Ci/well). At the end ofthe assay, the cells were harvested using a plate harvester, andthe incorporated radioactivity was determined using an auto-mated counter (Microbeta; PerkinElmer Wallac, Gaithers-burg, MD). Results are expressed as a stimulation index (SI).

Transgenic and transgene-negative littermates weretested for T cell responses to human CII–derived peptide254-273, which was synthesized and purified at the Mayo ClinicPeptide Facility. The mice were primed with 200 �g of peptideand then challenged in vitro with 100 �g/ml of the peptide.

For presentation of CII and its derived human CIIpeptide 254–273, CD4� cells were sorted with a fluorescence-activated cell sorter (FACS) from lymph node cells isolatedfrom primed mice. CD4� cells were cultured in vitro in thepresence or absence of antigen-presenting cells (APCs). Up to5 � 105 CD4� sorted cells were used for culturing alone withCII (50 �g/ml). In the other cultures, 5 � 105 irradiated spleencells were used as APCs. DR restriction was studied by addinganti-DR antibody (25 �g/ml) to the culture. Cells were cul-tured in RPMI 1640 medium as described above, and cellproliferation was determined by 3H-thymidine incorporation.

70 TANEJA ET AL

Figure 1. Characterization of DR4.AEo mice. Expression of DR4 insections of A, spleen, B, thymus, C, kidney, D, liver, and E, salivary glandfrom DRB1*0401.AEo mice was studied by immunostaining with fluores-cein isothiocyanate (FITC)–conjugated anti-DR antibody and with 4�,6-diamidino-2-phenylindole (DAPI). F–I, Thymus sections showing stainingwith phycoerythrin-conjugated CD4 (F), FITC-conjugated DR (G), asuperimposition of CD4- and DR-stained sections, showing some CD4�cells that are also DR� (H), and lack of staining with isotype controlantibody (I). These sections were also stained with DAPI. (Originalmagnification � 10 in A–E; � 40 in F–I.) J, Dot plots of gated splenicCD3� cells showing similar percentages of CD4� and CD8� cells inDR4.A�o-transgenic and DR4.AEo-transgenic mice. Results are from 1of 3 experiments performed with pooled cells from 2 mice per strain perexperiment. Values are percentages.

Figure 2. Higher expression of DR4 in DR4.AEo-transgenic micethan in DR4.A�o-transgenic mice. The percentage expression ofDR4 on CD4� cells from male and female AEo-transgenic mice isincreased in activated cells. A, Expression of DR in naive DR4.A�o

(gray line) and DR4.AEo (black line) mice in thymus and spleencells and in gated splenic B220� and CD3� cells. Values arepercentages. Dotted line represents the isotype control. B, Expres-sion of DR on gated CD3�,CD4� and CD3�,CD8� cells from naivemice and on activated T cells from mice primed with type II collagen(CII) or staphylococcal enterotoxin B (SEB). Values are percent-ages in cells from male (black line) and female (gray line) mice andin the isotype control (dotted line). The difference in meanfluorescence intensity between naive and activated CD4� cellswas significant (P � 0.05). Results are representative of 2–3experiments performed with cells pooled from 2 mice per strainper experiment for all experiments.

NEW HUMANIZED HLA–DR4–TRANSGENIC MOUSE MODEL OF RA 71

An SI �2 was considered to be a positive response. Theexperiment was performed twice with cells pooled from 2 miceper experiment.

Measurement of cytokines. The cytokines interferon-�(IFN�), interleukin-10 (IL-10), IL-18, transforming growthfactor � (TGF�), and tumor necrosis factor � (TNF�) weremeasured by capture enzyme-linked immunosorbent assay(ELISA), using kits obtained from BD PharMingen and fol-lowing the manufacturer’s instructions.

Measurement of anti-CII antibodies. Levels of anti-chick and anti-mouse CII IgG antibodies were measured insera obtained 35 days after CII immunization. A standardELISA was used for the analyses, and data are reported as theoptical density.

Measurement of rheumatoid factors (RFs). IgG andIgM RFs were measured by ELISA in mouse sera obtained 35days after priming, as previously described (15). Briefly,ELISA plates were coated overnight at 4°C with rabbit IgG.After washing, sera (1:40 dilution) were added, and the plateswere incubated for 45 minutes at room temperature. Plateswere then washed 5 times with phosphate buffered salinecontaining 0.05% Tween 20. Subsequently, wells were incu-bated for 1 hour with horseradish peroxidase–conjugatedrabbit anti-mouse IgG (Fc specific) or rabbit anti-mouse IgM(�-chain specific) (both from Pierce, Rockford, IL). Afterwashing, 3,3�,5,5�-tetramethylbenzidine substrate (Sigma, St.Louis, MO) was added, and the absorbance spectrum wasdetermined with an automated spectrophotometer (Bio-Rad,Hercules, CA). Sera from MRL/lpr and B10 mice were used aspositive and negative controls, respectively.

Measurement of antibodies to cyclic citrullinated pep-tide (anti-CCP). Serum levels of anti-CCP antibodies weremeasured by ELISA using a kit obtained from Axis-Shield(Kimbolton, UK).

Statistical analysis. The difference in the incidence ofarthritis between groups was analyzed using Fisher’s exact test.Antibody levels, onset of arthritis, mean arthritis scores in ar-thritic mice, and numbers of CD4� cells were compared usingthe nonparametric Mann-Whitney 2-tailed U test. The mean flowintensity of DR expression was compared by Kolmogorov-Smirnov test. All other significance assessments were calcu-lated using Student’s t-test or t-test with unequal variance.

RESULTS

Expression of class II molecules on T cells inAEo-transgenic mice. The DR4.AEo-transgenic micedeveloped normally and had no gross phenotypic abnor-malities. Immunostaining of various organs showed ex-pression of DR4 in the thymus and spleen, whereas noexpression in the kidney or liver and rare patchy expres-sion in salivary gland was observed (Figures 1A–E).

Immunostaining of thymus sections with antibod-ies to CD4 and DR showed some DR�,CD4� cells(Figures 1F–I). Subsets gated on splenic CD3� cellsshowed a similar number of CD3�,CD4� andCD3�,CD8� cells in both transgenic mouse strains (Fig-ure 1J). Expression of DR varied for transgenic mice, with30–65% of cells staining positive for the transgene by flowcytometry in cells from the spleen and thymus (Figure 2A).

The DR molecule is expressed at higher levels inAEo-transgenic mice than in A�o-transgenic mice. Todetermine the expression of HLA transgene in mice ascompared with humans, we studied the expression ofDR4 on spleen or lymph node B220� and CD3� cells.As expected, B220� cells stained positive for DR4(Figure 2A). FACS analysis of CD3�, CD4�, andCD8� cells demonstrated expression of DR on CD3�cells as well as CD3� subsets in these transgenic mice(Figure 2B). When activated with CII and SEB, theexpression of DR4 on CD3�,CD4� cells was increased.Activated antigen-specific CD4� cells consistentlyshowed significantly higher expression of DR, as deter-mined by mean flow intensity, in female mice than inmale mice, although the difference was not statisticallysignificant. However DR expression was significantlyhigher in activated CD4� cells compared with naiveCD4� cells (P � 0.05). The number of cells expressingDR was similar in CD8� naive and SEB-activated cells.

Table 1. Incidence of CIA in DRB1*0401-transgenic mice*

Mouse strain

CIA incidence, no.positive/no. tested (%)

CIA severity,mean � SD

CIA onset,mean � SD weeks

Anti-CII antibodies,mean � SD

Females Males Total Chick CII Mouse CII

DR4.AEo 12/23 2/13† 14/36 (39) 3.5 � 1.2‡ 5.2 � 1.4§ 0.6 � 0.1 0.4 � 0.2DR4.A�o 1/5 1/8 2/13 (15) 1.5 � 0.5 6.5 � 0.7 ND NDDQ8.AEo 6/9 8/11 14/20 (70) 6 � 1.2 5.3 � 1.3 0.7 � 0.5 0.7 � 0.0DQ8.A�o 5/8 4/7 9/15 (60) 5.1 � 1.8 5.6 � 1.1 0.8 � 0.0 0.7 � 0.0AEo 0/5 0/3 0/8 (0)A�o 0/3 0/3 0/6 (0)

* Values represent all mice from 2 experiments. No significant differences were observed in the DQ8.A�o and DQ8.AEo mouse strains. CIA �collagen-induced arthritis; anti-CII � anti–type II collagen; ND � not done.† P � 0.03 versus female DR4.AEo mice.‡ P � 0.06 versus DR4.A�o mice.§ P � 0.04 versus DR4.A�o mice.

72 TANEJA ET AL

To test whether the DR4 transgene was func-tional in the selection of various TCRs in transgenicmice, we studied the V� T cell repertoire profile. AllDR4 mice showed partial deletion of V�5.1 and V�4(data not shown) as compared with B10 mice (3). Anincreased number of V�6� T cells were selected byDR4-transgenic mice as compared with B10 mice. Acomparison of the V� profile in A�o-transgenic mice andAEo-transgenic mice suggested that there was a similarprofile, except for an increased frequency of V�8, whichwas not statistically significant. Findings in the DQ8.AEo

mice were similar to those in the DQ8.A�o describedpreviously (7).

Presence of DR4 and disease susceptibility infemale mice. Our previous study showed that only 15%of DR4.A�o mice develop CIA (unpublished observa-tions). In the present study, we compared the incidence,severity, and sex distribution of arthritis in DR4.AEo

and DR4.A�o mice (Table 1). In DR4.AEo mice immu-nized with chick CII, 39% (14 of 36) developed arthritis.However, in DR4.AEo mice, CIA was observed predom-inantly in females (52% versus 15% of males; P � 0.03)

Figure 3. Sex bias in susceptibility to collagen-induced arthritis in DR4-transgenic mice, with predomi-nantly female mice developing disease. A, Incidence and onset of arthritis in male and female DR4.A�o

(AboM and AboF) and DR4.AEo (AEoM and AEoF) mice, showing increased susceptibility and earlieronset in female DR4.AEo mice. Arthritis was scored as described elsewhere (14). B, Hematoxylin andeosin staining of paw sections from an arthritic DR4.AEo mouse (top), showing infiltration of cells in thesynovium and erosive arthritis in the joint of a digit, and from a normal mouse (bottom), without arthritis.C, Rheumatoid factor (RF) levels in sera collected 35 days after immunization. Group 1 represents typeII collagen (CII)–primed DR4.AEo mice (n � 12), group 2 represents CII-primed DQ8.AEo mice (n �15), group 3 represents naive DR4 mice (n � 6), group 4 represents Freund’s complete adjuvant(CFA)–injected DR4 mice (n � 6), and group 5 represents AEo mice (n � 10). Both DR4.AEo andDQ8.AEo mice produced IgG and IgM RFs. DR4.AEo mice immunized with CFA alone and naive micewere used as controls. Bars show the mean for each strain. Upper and lower horizontal lines indicate themeans in MRL-lpr (positive control) and CII-primed B10 (negative control) mice, respectively. Data pointsbelow the mean for the negative controls indicate mice that did not develop arthritis. OD � optical density.

NEW HUMANIZED HLA–DR4–TRANSGENIC MOUSE MODEL OF RA 73

(Figure 3A). There was no significant difference in theseverity of arthritis in female versus male DR4.AEo

mice, although arthritis onset occurred much earlier inthe females (P � 0.03). A sex bias in the susceptibility toarthritis was not observed in DQ8-restricted arthritis(Table 1). None of the transgene-negative littermates(A�o and AEo) developed arthritis.

The onset of disease was earlier (mean � SD5.2 � 1.4 weeks in DR4.AEo and 6.5 � 0.7 weeks inDR4.A�o mice; P � 0.04) and the arthritis was moresevere (P � 0.06) in DR4.AEo mice as compared withDR4.A�o mice. The arthritis was more prominent in thedigits of the paws in DR4.AEo mice as compared withDQ8.AEo mice. Histopathologic analysis of sectionsstained with hematoxylin and eosin showed severe infil-tration and erosive arthritis of the digits of the paws inDR4 mice (Figure 3B).

Development of autoantibodies. All arthritic micedeveloped anti-CII antibodies in response to immuniz-ing and self collagen (Table 1), and all developed IgGand IgM RFs (Figure 3C). Levels of IgG-RF weresignificantly higher in DR4 mice than in either theirtransgene-negative littermates (P � 0.0001), CFA-immunized DR4-transgenic mice (P � 0.005), or naiveDR4-transgenic mice (P � 0.005). There was no differ-ence in IgG-RF and IgM-RF levels between DR4-transgenic and DQ8-transgenic mice. The levels of RFcorrelated with the severity of disease (data not shown).Very few mice produced RF levels as high as those in theMRL/lpr mice (positive controls). RF levels in B10 mice(negative controls) and in naive mice were similar tothose in mice immunized with CFA only. IgM-RF levelswere higher in DR4 mice compared with CFA-immunized and transgene-negative CFA-immunizedmice, although the difference did not reach statisticalsignificance (P � 0.08).

We also tested mice with and without CIA for thepresence of anti-CCP antibodies. Mice with CIA hadvery high levels of anti-CCP antibodies (mean � SD36 � 3 units/ml), whereas mice without CIA had verylow levels (8 � 4 units/ml) of these antibodies (P �0.005).

HLA-restricted in vitro response to CII in trans-genic mice. Cells isolated from CII-primed DR4.AEo

mice responded to CII when challenged in vitro (Figure4A). The response was CD4-mediated and was restrictedby the transgene, as indicated by the results of theinhibition assay. To determine if the response to DR4-restricted peptide in DR4.AEo mice is similar to that inDR4.A�o mice, we studied both transgenic mousestrains for in vitro responses to human CII–derivedpeptide 254–273. Both transgenic mouse strains

Figure 4. In vitro response to type II collagen (CII) and humanCII–derived peptide 254–273. Results suggest that CD4 cells fromDR4.AEo mice can present antigen in vitro. A, DR4.AEo micewere immunized with CII emulsified in Freund’s complete adjuvant.For the inhibition assay, spleen cells were cultured in the presence ofCII, either alone or with one of the following antibodies: GK1.5(anti-CD4), Lyt2 (anti-CD8), or L227 (anti-DR). In addition, CD4� Tcells from CII-primed mice were sorted by fluorescence-activatedcell sorting (FACS) and cultured in the presence and absence ofantigen-presenting cells (APCs) as described in Materials andMethods. SI � stimulation index; nt � not tested. B, Comparisonof the response of lymph node cells to human CII–derived peptide254–273 (HII 254–273) from peptide-primed DR4.A�o andDR4.AEo mice, showing a significantly higher response in thelatter group (P � 0.004). C, Cytokine profile (tumor necrosis factor� [TNF�], interleukin-18 [IL-18], IL-10, transforming growthfactor � [TGF�], and interferon-� [IFN�]) in DR4.AEo mice inresponse to in vitro challenge with CII and human CII–derivedpeptide 254–273. D, CD4� T cells from DR4.AEo mice primed withhuman CII–derived peptide 254–273 were sorted by FACSand cultured in the presence and absence APCs. The results in Aand D represent a comparison of collagen and its derived peptideusing sorted CD4� cells from DR4.AEo mice. Values are themean � SD of 2–3 experiments using pooled cells from 2–3 miceper experiment.

74 TANEJA ET AL

mounted a CD4-mediated and major histocompatibilitycomplex (MHC)–restricted response to the peptide,although a higher response was observed in DR4.AEo

mice (P � 0.004) (Figure 4B). DR4.AEo mice producedTh1 cytokines (IFN�, IL-18, and TNF�) as well asregulatory cytokines (IL-10 and TGF�) when challengedin vitro with CII or with peptide (Figure 4C).

Antigen presentation, but not processing, byCD4� T cells from DR4.AEo mice. We tested the abilityof CD4� T cells to process and present antigen in vitro.CD4� cells from DR4.AEo mice that had been primedwith CII and with human CII–derived peptide 254–274were sorted by flow cytometry after staining with specificantibody. The rest of the cells were designated as CD4�.In vitro assay was performed using spleen cells, sortedCD4� cells only, CD4� cells in the presence of irradi-ated APCs, and CD4� cells in the presence of irradiatedAPCs. Sorted CD4� cells mounted an in vitro response,with an SI of 3, when challenged with the peptide(Figure 4D), but not when challenged with CII (Figure4A). In the presence of irradiated APCs, sorted CD4�cells proliferated to an extent similar to that of wholespleen cells when challenged with peptide (SI of 5), butthe response to CII was much lower. The CD4� cellsresponded poorly to the peptide (data not shown).

Higher cellularity and IFN� production by fe-male mice than by male mice. A comparison of femaleand male DR4.AEo mice showed a higher number ofsplenic CD3� and CD4� cells in the females than in themales (P � 0.05) (Figure 5A). The high cellularity infemale mice was associated with higher spleen weight ascompared with that in male mice (P � 0.01). In vitro,lymph node cells isolated from mice primed with immu-nodominant human CII–derived peptide and CIImounted a stronger response and produced higher levelsof proinflammatory cytokines in cells from females thanin cells from males, with a significant increase in IFN�levels (P � 0.05) (Figure 5B). Characterization of acti-vated cells after immunization consistently showed ahigher number of activated CD4� cells that were alsopositive for CD25� and CD69� in female mice, al-though the difference did not reach statistical signifi-cance. Even though the total number of V�8 cells werenot different between female and male mice (results notshown), expansion of V�8� CD4 T cells was higher infemale than in male mice after immunization (P � 0.05).In addition, analysis of sera from primed mice showedthat female mice produced higher levels of anti-CIIantibodies in vivo than did male mice (Figure 5D), butthe difference did not reach statistical significance (P �0.06).

DISCUSSION

Mouse models of RA have often been critiquedas having inherent immunologic shortcomings: expres-

Figure 5. Higher cellularity and higher production of inflammatorycytokines and antibodies in female DR4.AEo mice than in maleDR4.AEo mice. Values are the mean � SD of 2–3 experiments usingpooled cells from 2–3 mice per experiment. � � P � 0.05 for femalesversus males. A, Numbers of spleen cells and cell subsets in male andfemale DR4.AEo mice. B, Lymph node cells isolated from mice primedwith human type II collagen (CII)–derived peptide 254–273 (HII254-273) and with CII, showing higher levels of proinflammatorycytokines (interleukin-18 [IL-18] and interferon-� [IFN-�]) in femalesthan in males. Cytokines were measured in culture supernatants byenzyme-linked immunosorbent assay (see Materials and Methods fordetails). C, Levels of anti-CII antibodies (Abs) in sera obtained 35 daysafter immunization from DR4.AEo mice, showing higher levels infemales than in males. Only data from mice that developed collagen-induced arthritis are shown. OD � optical density.

NEW HUMANIZED HLA–DR4–TRANSGENIC MOUSE MODEL OF RA 75

sion of endogenous mouse class II chains obfuscatesinterpretation of the data; mice do not express class IImolecules on T cells; and mice models fall short ingenerating endocrinologic conditions that reproduce thefemale-biased susceptibility to RA in humans. Themodel presented here is the first to share similaritieswith humans for the major differences observed inhuman RA and mouse models of arthritis. Mice used inthe present study lack all the classic murine endogenousclass II chains.

One major difference between human and mouseclass II molecules is that human class II molecules areexpressed on T cells, especially activated T cells, whereasmouse class II molecules are not (16,17). Human T cellclones can present antigen (18). Thus, expression of classII molecules on T cells could potentially play a role inhuman autoimmune diseases. During the onset of dis-ease, activated CD4 T cells infiltrate the target tissue,which results in inflammation. The class II� CD4 T cellscould present locally available peptides and activateother cells. DR4.AEo mice express class II molecules onT cells, a feature similar to that in humans. Although ourstudy does not directly implicate these T cell–expressingclass II molecules in disease pathogenesis, a probablescenario can be envisaged. We observed increased ex-pression of DR in activated T cells. CD8 cells showed ahigher number of DR� cells when activated by CII ascompared with SEB, although the difference was notstatistically significant. One possibility could be thatcompared with CII, SEB leads to vigorous expansion ofCD8 cells, resulting in a lower percentage of cells thatare positive for DR.

The other major difference between RA andmodels of CIA is that whereas there is a significant sexbias and production of autoantibodies such as RFs andanti-CCP in RA (19), neither of these occurs in CIA.Studies in humans have shown that anti-CCP antibodyproduction precedes the onset of RA and are associatedwith the presence of DR4 (20,21). The DR4-transgenicmice described herein were shown to produce RFs aswell as anti-CCP antibodies. Thus, we have generated anew humanized mouse model of RA that mimics thephenotype of the human disease and shows a sex bias.

Data from the present study of DRB1*0401-transgenic mice show that DR4 renders female micesusceptible to the development of CIA in the absence ofall of the endogenous class II molecules. DR4 occurs inlinkage disequilibrium with DQ8 in humans and isinherited en bloc as a haplotype. DQ8-transgenic micedid not develop sex-biased CIA, which lends support toour observations that the sex bias could be due to DR4.Previous studies using the CIA model showed a contri-

bution of DR4, but all the features of human RA werenot presented in those models (3,9,10). This could havebeen due to the presence of endogenous class II chainsthat made it difficult to interpret the observations.

As in humans with RA, the reason why femalemice are more susceptible to the development of CIAthan male mice is not yet clear. Hormones have longbeen predicted to be one of the major reasons for thisbias in autoimmunity in humans (22). Another impor-tant sex difference that can contribute to the immuneresponse is the higher number of absolute CD4� cells inwomen than in men (23). The data presented herereproduce the findings in humans. We found that thefemale mice responded with much stronger T cell re-sponses to the DR4-restricted peptide than did the malemice. One possible reason for this stronger responsecould be the high cellularity and increased number ofCD4� cells in the female mice. A stronger in vitroresponse to the immunodominant peptide, increasedamounts of inflammatory cytokines, IFN�, and autoan-tibodies in female mice may be other reasons for the sexbias and earlier onset of disease, since there was nodifference in the severity of arthritis between the twosexes. A recent study showed that deprivation of andro-gen in mice increases the cellularity of primary andperipheral lymphoid organs and T cell proliferation (24).This suggests that androgen might play a role in protect-ing male mice from the development of CIA. Estrogenhas been shown to influence the immune response byinfluencing B cell survival and modulating dendritic cellfunction (25,26). We observed a higher number of Bcells in female mice, even though the difference did notreach statistical significance.

To understand the sex differences in autoimmu-nity, the basic immune response needs to be studied.Since the innate immune response drives the adaptiveresponse, both the MHC and the cells of innate immuneresponses may be important in the development ofautoimmunity. While studies in humans are difficult toundertake and interpret, the model described here pro-vides a very good choice with which to study themechanism of sex bias in autoimmunity.

MHC molecules have been shown to be impor-tant in the selection of the T cell repertoire in thethymus (27). Our data show that DR4 can positively andnegatively select the TCR V� profile, thus suggestingthat it can present peptides in the thymus. A significantobservation in the present study was that CD3� cellsexpressed class II molecules, a feature in humans, butnot in mice. To determine if the HLA transgene ex-pressed on T cells of transgenic mice plays a role indisease pathogenesis, we tested the ability of CD4� T

76 TANEJA ET AL

cells to process and present antigen. Sorted CD4� cellsmounted an in vitro response when challenged with thepeptide, but not when challenged with CII, which sug-gests that they can present peptides but cannot processwhole CII. This could indicate that in humans, activatedCD4� T cells could present antigen locally, leading toexacerbation of disease.

Our data suggest that DR4 can present antigen ina MHC-restricted manner and mount an immune re-sponse in vivo. DR4.AEo mice mounted a strongerresponse than DR4.A�o mice to DR4-restrictedcollagen-derived peptide and produced the proinflam-matory cytokines IFN�, IL-18, and TNF� and theregulatory cytokines IL-10 and TGF� in response to invitro challenge with the immunizing CII and peptide.IL-10 has been associated with protection from arthritisin most of the experimental models studied (28). How-ever, it can also be produced by CD4� T cells inresponse to autoantigen presented by B cells (29). TGF�has been shown to promote bone destruction in animalmodels of CIA and in humans with RA, which is relatedto an increase in the expression of proinflammatorycytokines and metalloproteinases (30). IL-18 induced byTNF� and IFN� are important in the development ofCIA (31).

The new humanized mice provide a better modelfor studying arthritis since they mimic the major featuresof the disease in humans. In conclusion, we have shownthat the new transgenic mice are similar to humans intheir expression of class II molecules on CD4 T cells,that activated CD4� T cells can present peptides but donot process the antigen, and that the new AEo-transgenic mice develop autoimmunity in a sex-biasedmanner (a female to male ratio of 3:1) and produceautoantibodies similar to humans with RA.

ACKNOWLEDGMENTS

We thank Julie Hanson and Tad Trejo for maintainingthe transgenic mice and Michele Smart for screening thetransgenic mice.

AUTHOR CONTRIBUTIONS

Dr. Taneja had full access to all of the data in the study andtakes responsibility for the integrity of the data and the accuracy of thedata analysis.Study design. Drs. Taneja and David.Acquisition of data. Mr. Behrens and Dr. Mangalam.Analysis and interpretation of data. Drs. Taneja, Griffiths, Luthra, andDavid.Manuscript preparation. Drs. Taneja and David.Statistical analysis. Dr. Taneja.

REFERENCES

1. Gregersen PK, Silver J, Winchester RJ. The shared epitopehypothesis: an approach to understanding the molecular geneticsof susceptibility to rheumatoid arthritis. Arthritis Rheum 1987;30:1205–13.

2. Diab BY, Lambert NC, L’Faqihi FE, Loubet-Lescoulie P, dePreval C, Coppin H. Human collagen II peptide 256-271 prefer-entially binds to HLA-DR molecules associated with susceptibilityto rheumatoid arthritis. Immunogenetics 1999;49:36–44.

3. Pan S, Taneja V, Griffiths MM, Luthra HS, David CS. Comple-mentation between HLA-DR4 (DRB1*0401) and specific H2-Amolecule in transgenic mice leads to collagen-induced arthritis.Hum Immunol 1999;60:816–25.

4. Zanelli E, Krco CJ, Baisch JM, Cheng S, David CS. Immuneresponse of HLA-DQ8 transgenic mice to peptides from the thirdhypervariable region of HLA-DRB1 correlates with predispositionto rheumatoid arthritis. Proc Natl Acad Sci U S A 1996;93:1814–9.

5. Kirschmann DA, Duffin KL, Smith CE, Welply JK, Howard SC,Schwartz BD, et al. Naturally processed peptides from rheumatoidarthritis associated and non-associated HLA-DR alleles. J Immu-nol 1995;155:5655–62.

6. He X, Kang AH, Stuart JM. Accumulation of T cells reactive totype II collagen in synovial fluid of patients with rheumatoidarthritis. J Rheumatol 2000;27:589–93.

7. Nabozny GH, Baisch JM, Cheng S, Cosgrove D, Griffiths MM,Luthra HS, et al. HLA-DQ8 transgenic mice are highly susceptibleto collagen-induced arthritis: a novel model for human polyarthri-tis. J Exp Med 1996;183:27–37.

8. Krco CJ, Pawelski J, Harders J, McCormick D, Griffiths MM,Luthra HS, et al. Characterization of the antigenic structure ofhuman type II collagen. J Immunol 1996;156:2761–8.

9. Rosloniec EF, Brand DD, Myers LK, Esaki Y, Whittington KB,Zaller DM, et al. Induction of autoimmune arthritis in HLA-DR4(DRB1*0401) transgenic mice by immunization with human andbovine type II collagen. J Immunol 1998;160:2573–8.

10. Taneja V, Taneja N, Behrens M, Pan S, Trejo T, Griffiths MM, etal. HLA-DRB1*0402 (DW10) transgene protects collagen-in-duced arthritis-susceptible H2Aq and DRB1*0401 (DW4) trans-genic mice from arthritis. J Immunol 2003;171:4431–8.

11. Madsen L, Labrecque N, Engberg J, Dierich A, Svejgaard A,Benoist C, et al. Mice lacking all the conventional MHC class IIgenes. Proc Natl Acad Sci U S A 1999;96:10338–43.

12. Pan S, Trejo T, Hansen J, Smart M, David CS. HLA-DR4(DRB1*0401) transgenic mice expressing an altered CD4-bindingsite: specificity and magnitude of DR4-restricted T cell response.J Immunol 1998;161:2925–9.

13. Griffiths MM, Eichwald EJ, Martin JH, Smith CB, DeWitt CW.Immunogenetic control of experimental type II collagen inducedarthritis. I. Susceptibility and resistance among inbred strains ofrats. Arthritis Rheum 1981;24:781–9.

14. Wooley PH, Luthra HS, Stuart JM, David CS. Type II collagen-induced arthritis in mice. I. Major histocompatibility complex (Iregion) linkage and antibody correlates. J Exp Med 1981;154:688–700.

15. Taneja V, Taneja N, Paisansinsup T, Behrens M, Griffiths MM,Luthra H, et al. CD4 and CD8 T cells in susceptibility/protectionto collagen-induced arthritis in HLA-DQ8-transgenic mice: impli-cations for rheumatoid arthritis. J Immunol 2002;168:5867–75.

16. Chen D, Ueda R, Harding F, Patil N, Mao Y, Kurahara C, et al.Characterization of HLA DR3/DQ2 transgenic mice: a potentialhumanized animal model for autoimmune disease studies. EurJ Immunol 2003;33:172–82.

17. Yoshihiro A, Tada T. Epitopes associated with the MHC restric-tion site of T cells. J Exp Med 1987;165:87–96.

18. Hewitt CR, Feldmann M. Human T cell clones present antigen.J Immunol 1989;143:762–9.

NEW HUMANIZED HLA–DR4–TRANSGENIC MOUSE MODEL OF RA 77

19. Van Gallen FA, van Aken J, Huizinga TW, Schreuder GM,Breedveld FC, Zanelli E, et al. Association between HLA class IIgenes and autoantibodies to cyclic citrullinated peptides (CCPs)influences the severity of rheumatoid arthritis. Arthritis Rheum2004;50:2113–21.

20. Ferucci ED, Majka DS, Parrish LA, Moroldo MB, Ryan M, PassoM, et al. Antibodies against cyclic citrullinated peptide are asso-ciated with HLA–DR4 in simplex and multiplex polyarticular-onset juvenile rheumatoid arthritis. Arthritis Rheum 2005;52:239–46.

21. Nielen MM, van Schaardenburg D, Reesink HW, van de Stadt RJ,van der Horst-Bruinsma IE, de Koning MH, et al. Specificautoantibodies precede the symptoms of rheumatoid arthritis: astudy of serial measurements in blood donors. Arthritis Rheum2004;50:380–6.

22. Nelson JL, Ostensen M. Pregnancy and rheumatoid arthritis.Rheum Dis Clin North Am 1997;23:195–212.

23. Amadori A, Zamarchi R, De Silvestro G, Forza G, Cavatton G,Danieli GA, et al. Genetic control of the CD4/CD8 T-cell ratio inhumans. Nat Med 1995;1:1279–83.

24. Roden AC, Moser MT, Tri SD, Mercader M, Kuntz SM, Dong H,et al. Augmentation of T cell levels and responses induced byandrogen deprivation. J Immunol 2004;173:6098–108.

25. Grimaldi CM, Cleary J, Dagtas AS, Moussai D, Diamond B.

Estrogen alters thresholds for B cell apoptosis and activation.J Clin Invest 2002;109:1625–33.

26. Paharkova-Vatchkova V, Maldonado R, Kovats S. Estrogen pref-erentially promotes the differentiation of CD11c� CD11b (inter-mediate) dendritic cells from bone marrow precursors. J Immunol2004;72:1426–36.

27. Zanelli E, Gonzalez-Gay MA, David CS. Could DRB1 be theprotective locus in rheumatoid arthritis? Immunol Today 1995;16:274–8.

28. Finnegan A, Kaplan CD, Cao Y, Eibel H, Glant TT, Zhang J.Collagen-induced arthritis is exacerbated in IL-10-deficient mice.Arthritis Res Ther 2003;5:R18–24.

29. Rock KL, Benacerraf B, Abbas AK. Antigen presentation byhapten-specific B lymphocytes. I. Role of surface immunoglobulinreceptors. J Exp Med 1984;160:1102–13.

30. Cheon H, Yu SJ, Yoo DH, Chae IJ, Song GG, Sohn J. In-creased expression of pro-inflammatory cytokines and metallo-proteinase-1 by TGF-�1 synovial fibroblasts from rheumatoidarthritis and normal individual. Clin Exp Immunol 2002;127:547–52.

31. Tanaka M, Harigari M, Kawaguchi Y, Ohta S, Sugiura T, TakagiK, et al. Mature form of interleukin 18 is expressed in rheumatoidarthritis synovial tissue and contributes to interferon-� productionby synovial T cells. J Rheumatol 2001;28:1779–87.

78 TANEJA ET AL