INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The link of a set of related HLA-DR alleles, DRB1*0101 (DR1), DRB1*0401 (DR4/Dw4), DRB1*0404 (DR4/Dw14), DRB1*0405 (DR4/Dw15), and DRB1*1402 (DR6/Dw16), which share an amino acid sequence EQ(K/R)RAA and also have an increased susceptibility to develop rheumatoid arthritis (RA) has been well established (11, 16). The mechanism by which the susceptibility motif predisposes to RA is at present unknown. However, it has been suggested that an environmental factor interacting with a genetic predisposition may be one mechanism involved in the pathogenesis of this debilitating disease (8). Recent studies have demonstrated cross-reactivity between a sequence QKRAA, which is present in the DnaJ heat shock protein of Escherichia coli and the RA susceptibility sequence (1). We have identified an amino acid sequence similarity between the RA susceptibility motif and the sequence ESRRAL found in the hemolysin of Proteus mirabilis (17). Furthermore, elevated levels of antibodies to EQRRAA (13) and ESRRAL (17) have been found in RA patients. Apart from reactivity with antibody, the RA associated motif has also been shown to play a role in peptide binding (7), and we have identified a structurally related peptide found in P. mirabilis urease LEIEKDFTTYGEE and collagen type XI, DELSPETSPYVKE, which satisfy the criteria for selectively binding the RA associated motif (19). Our studies were undertaken to determine the extent of cross-reactivity between the RA associated motif KDLLEQKRAAVDTYC and the non-RA-associated motif KDILEDERAAVDTYC compared to the bacterial sequence LGSISESRRALQDSQR and to determine whether the two structurally related peptides present in P. mirabilis and type XI collagen bind selectively to the RA-associated motif.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Peptide synthesis. The peptides were assembled by using an automated Milligen Biosearch model 9050 Pepsynthesiser on NovaSyn TG flow resin with a loading value of 0.41 mmol · g/liter, functionalized with a Rink amide linker. Acylation cycles with (four equivalents) Fmoc (9-fluorenylmethoxycarbonyl) amino acid, preactivated with 2(1-H-benzo.triazo-1-yl)-1,1,3-tetramethyl uronium tetrafluoroborate (TBTU) and di-isopropyl ethylamine (DIPEA) in a molar ratio (1:1:1.5 by volume) were carried out in DMF for 30 to 40 min. Fmoc deprotection was achieved with 20% piperidine in DMF. N-terminal acetylation was carried out by using pentafluorophenyl acetate (four equivalents) for 30 min. Reactions were monitored by UV detection of the column effluent at 365 nm and color tests for free amino groups (9). Complete peptidyl resins were washed successively with DMF, methanol, and diethyl ether before being dried in vacuo. The peptidyl resin was treated with a solution (10 ml) of trifluoroacetic acid (TFA)-H₂O-triisopropylsilane (94:5:1 by volume) for 90 min at room temperature. The resin was then filtered through a sintered funnel and washed with TFA, and the combined filtrate was dried by rotary evaporation. Residual TFA was azeotroped with diethyl ether. The products were lyophilized from water and purified by preparative high-performance liquid chromatography with a Vydac 218TP54 column on a Waters 990 photodiode array system. Solvent A was 0.1% TFA, and solvent B was 10% solvent A plus 90% acetonitrile. All peptides used had a purity of >90%, and their masses were determined by using a Bioanalysis MALDI-TOF instrument with a-cyano-4-hydroxycinnamic acid as the matrix.

Peptide antisera. Synthetic peptides of HLA-DRB1*0401, HLA-DRB1*0402, and P. mirabilis hemolysin sequence were conjugated to the carrier protein keyhole limpet hemocyanin (KLH) (ICN Biomedical Ltd.) by using m-maleimidobenzoyl-N-hydroxysuccinimide ester (6). The conjugate was purified by gel filtration with PD10 columns (Sigma Chemical Ltd.). New Zealand White rabbits received three subcutaneous injections at 2-week intervals. A 250-µl aliquot of a 1-mg/ml solution (synthetic peptide conjugated to KLH) was added to 250 µl of Specol adjuvant (Central Veterinary Institute). Peptide antiserum reactivity was continuously monitored by measuring the immune response to various concentrations of both the target peptide and KLH alone. The rabbits were bled 14 days after the last immunization, and the resultant sera were stored at 20°C.

ELISA. Antibody responses were measured by peptide enzyme-linked immunosorbent assay (ELISA). Briefly, the assay was carried out as follows: flat-bottomed, rigid polystyrene microtiter plates (Dynatech) were coated with 100 µl of synthetic peptide (5.0 µg peptide per ml) and KLH (10 µg/ml), overnight at 4°C. After adsorption and being washed with phosphate-buffered saline (PBS; pH 7.4) containing 0.1% (vol/vol) Tween 20 (Sigma), the plates were saturated with 1% (wt/vol) bovine serum albumin (Sigma)-PBS-Tween 20 and incubated for 1 h at 37°C, followed by further washing with PBS-Tween 20. Different dilutions of peptide rabbit antisera in PBS-Tween were then added, and the plates were incubated for 90 min at 37°C, followed by washing with PBS-Tween. Peroxidase-conjugated goat anti-rabbit class-specific immunoglobulin G (Dako, Ltd.) diluted 1/500 in PBS-Tween was added, and the plates incubated for 90 min at 37°C. After another washing the substrate solution, 0.5 mg of 2,2'-azinobis(3-ethylbenzthiazoline-sulfonic acid) (ABTS) (Sigma) per ml in citrate-phosphate buffer (pH 4.1) containing 0.98 mM H₂O₂ (Sigma), was added to each well. Development of the plates took place at room temperature in the dark for 20 min. The reaction was stopped with 2 mg of sodium fluoride solution (Sigma) per ml, and the optical density (OD) was measured at a wavelength of 630 nm with a micro-ELISA plate reader (MR 600; Dynatech). For inhibition studies, the inhibitors KDLLEQKRAAVDTYC, LGSISESRRALQDSQR, and KDILEDERAAVDTYC (100 µg/ml) were incubated overnight at 37°C with the three peptide antisera before ELISAs were carried out. All ELISAs were carried out in triplicate, and the mean OD value (± the standard error [SE]) was calculated for each sample.

Mouse fibroblast (Dap.3) cells. Mouse fibroblast Dap.3 cells transfected with HLA-DRB1*0401, HLA-DRB1*0402, and untransfected cells, together with L243 (anti-DR) in supernatant form, were kindly provided by R. Lechler of the Department of Immunology, Hammersmith Hospital, London, England. Both sets of transfected and untransfected cells were maintained as described previously (2). However, transfected cells were also grown in the presence of G418 (200 µl/ml) (Gibco). The formation of an adherent monolayer of cells indicates healthy growth.

Peptide antiserum assay. An aliquot of 10⁶/ml per fluorescence-activated cell sorter tube (Falcon) of cells was suspended in PBA (PBS plus 0.1% [wt/vol] plus 0.1% [wt/vol] azide), the tubes centrifuged at 4°C at 1,500 rpm (Omnifuge 2.0RS; Heraeus Sepatech) for 5 min, and the supernatant was discarded. Mouse cells (transfected or untransfected) were incubated on ice for 1 h with L243, with a second antibody only, as well as with different dilutions of rabbit peptide antisera. After a washing in PBA, aliquots of swine anti-rabbit fluorescein isothiocyanate-labelled total immunoglobulin (Dako) diluted in PBA (1/24) were added to the assay tubes and incubated for 1 h on ice. The cells were washed and resuspended in PBA. The percentage of cells which fluoresced above an arbitrarily set level by the use of the negative control (second antibody only) were detected with a FACScan by using the software provided (Becton Dickinson) with the permission of N. Staines of the Infection and Immunity Section, Division of Life Sciences, King's College, London, England.

Peptide binding assay. Mouse transfectants of DRB1*0401 and DRB1*0402 and untransfected cells were incubated on ice for 90 min, with concentrations of test or control peptide as described above that ranged between 0.25 and 1 mg/ml. After a washing in PBA, the mean cell fluorescence values, defined by the use of the negative control peptide, were detected with a FACScan.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Dilution studies with peptide antisera. The results obtained in this study demonstrate cross-reactivity between KDLLEQKRAAVDTYC and LGSISESRRALQDSQR (Fig. 1A and B) but not with the Dw10 motif KDILEDERAAVDTYC when tested by ELISA (Fig. 1C).

View larger version (41K): [in this window] [in a new window]   FIG. 1.   Anti-peptide antiserum dilution response curves with ELISA. The antisera were raised against the following KLH conjugates of peptide: CKDLLEQKRAAVDTYC, CLGSISESRRALQDSQR, and CKDILEDERAAVDTYC. The binding of antisera and preimmune serum was determined by using the uncoupled peptides LGSISESRRALQDSQR (A), KDLLEQKRAAVDTYC (B), and KDILEDERAAVDTYC (C) adsorbed onto the ELISA plate. Also shown is the inhibition of binding by the indicated peptide.

Inhibition studies. Peptide antiserum raised against DRB1*0401 was inhibited by preincubation with 100 µg of LGSISESRRALQDSQR peptide per ml, as well as with KDLLEQKRAAVDTYC peptide. In a similar way, antisera raised against the P. mirabilis sequence was also inhibited by preincubation with 100 µg of KDLLEQKRAAVDTYC and LGSISESRRALQDSQR peptides per ml (Fig. 1A and B). The anti-CLGSISESRRALQDSQR antiserum prior to incubation with DRB1*0401 peptide had a mean antibody binding activity of 1.49 (±0.02) at a dilution of 1:200 and reacted at dilutions of up to 1:51,200. After incubation, the binding activity was reduced to 0.20 (±0.04) at a 1:200 dilution and reacted at dilutions up to 1:6,400. Similar results were obtained with the LGSISESRRALQDSQR peptide (Fig. 1A). The DRB1*0401 antiserum had a mean antibody binding activity to the DRB1*0401 peptide of 1.29 (±0.01) at 1:200 and bound at a dilution of up to 1:51,200. However, after incubation with the LGSISESRRALQDSQR peptide, the activity was reduced to 0.21 (±0.02) at 1:200, and the serum reacted at a dilution of up to 1:6400. Furthermore, similar results were obtained with the KDLLEQKRAAVDTYC peptide (Fig. 1B).

Flow cytometric analysis. Mouse fibroblast transfected cell lines expressing intact DRB1*0401 and DRB1*0402 molecules, together with the untransfected cells, were also used for cross-reactive studies. An arbitrarily set value of 10¹ for cell fluorescence was used as the cutoff point, as only 2% of the DRB1*0401 and DRB1*0402 transfected cells fluoresced positively with the negative control (second antibody only) (Fig. 2A1), since a positive control L243 was used (Fig. 2A2). At an antibody dilution of 1:160, 88% of DRB1*0401 transfected cells were positively bound by the CKDLLEQKRAAVDTYC (Fig. 2A3) and 79% were bound by the CLGSISESRRALQDSQR (Fig. 2A4) peptide antisera compared to 21% bound by the antiCKDILEDERAAVDTYC (Fig. 2A5) and 5% bound by the pooled preimmune rabbit (Fig. 2A6) sera. Both anti-CKDLLEQKRAAVDTYC and anti-CLGSISESRRALQDSQR antiserum samples bound to the transfected cells at dilutions of up to 1:10,240, whereas anti-CKDILEDERAAVDTYC stopped reacting at 1:2,560 and the pooled preimmune serum stopped reacting at 1:640 (Fig. 2B). Furthermore, increased binding to transfectants expressing DR4/Dw10 was demonstrated by using antiserum raised to DRB1*0402 peptide at dilutions of 1:40, 1:80, and 1:160 compared to anti-CKDLLEQKRAAVDTYC, anti-CLGSISESRRALQDSQR, and pooled preimmune rabbit sera. At an antibody dilution of 1:80, 22% of the DRB1*0402 transfected Dap.3 cells were positively bound by the anti-CKDILEDERAAVDTYC (Fig. 2C1), 7% by the anti-CKDLLEQKRAAVDTYC (Fig. 2C2), and 10% by the anti-CLGSISESRRALQDSQR (Fig. 2C3) peptide antisera and 9% were bound by the pooled preimmune rabbit sera (Fig. 2C4). All three peptide antisera and the preimmune serum stopped binding to the DR4/Dw10 cells at a dilution of 1:2,560 (Fig. 2D). However, there was no difference in binding between the individual peptide antiserum and the pooled preimmune rabbit serum to the untransfected mouse cells (Table 1).

View larger version (75K): [in this window] [in a new window]   FIG. 2.   (A) Immunofluorescence profiles for 104 transfected (DR4/Dw4) Dap.3 cells with fluorescence greater than 101. Fluorescence intensity (x axis) versus cell number (y axis). Quadrant plots show binding of negative control (secondary antibody only) (panel 1), L243 (anti-DR) (panel 2), anti-EQKRAA (panel 3), anti-ESRRAL (panel 4), anti-EDERAA (panel 5), and pooled preimmune serum (panel 6). The peptide and preimmune serum were used at a 1:160 dilution. The marker (M1) indicates the positive area of binding. (B) Dilution studies of antisera raised against CKDLLEQKRAAVDTYC, CLGSISESRRALQDSQR, and CKDILEDERAAVDTYC peptides and pooled preimmune rabbit serum binding to mouse fibroblast transfected cell line Dap.3 expressing HLA-DRB1*0401 (DR4/Dw4). The percentages of cells which fluoresce at levels greater than the arbitrarily set level of 101 are shown. (C) Immunofluorescence profiles for 104 transfected (DR4/Dw10) Dap.3 cells with a fluorescence greater than 101. The fluorescence intensity (x axis) is plotted versus cell number (y axis). Quadrant plots show binding of anti-EDERAA (panel 1), anti-EQKRAA (panel 2), anti-ESRRAL (panel 3), and pooled preimmune serum (panel 4). The peptide and preimmune sera were used at a 1:80 dilution. The marker (M1) indicates positive area of binding. (D) Dilution studies of antisera raised to CKDLLEQKRAAVDTYC, CLGSISESRRALQDSQR, and CKDILEDERAAVDTYC peptides and pooled preimmune serum binding to the mouse fibroblast transfected cell line Dap.3 expressing HLA-DRB1*0402 (DR4/Dw10). The percentages of cells which fluoresce at levels greater than the arbitrarily set level of 101 are shown.

HLA-peptide binding. Fluorescent labelled peptides from type II as well as type XI collagens and P. mirabilis bound without appreciable differences to mouse fibroblast transfectant cell lines expressing HLA-DRB1*0401 and DRB1*0402 molecules compared to the control peptide. The mean cell fluorescence as determined by the FACScan did not differ when the two transfectants and individual peptide concentrations of the test and control peptides were used (Table 2).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, peptide antisera raised in rabbits against the RA susceptibility sequence EQKRAA reacted with P. mirabilis hemolysin ESRRAL peptide. In a reciprocal manner, antisera against the hemolysin sequence demonstrated greater binding affinity towards the RA susceptibility motif than the EDERAA peptide sequence of HLA-DRB1*0402 found in the HLA-Dw10, an allele not associated with RA. These observations are compatible with the recent report of cross-reactivity between the RA susceptibility motif QKRAA and the DnaJ protein of E. coli QKRAA (1). In addition, the structurally related peptides LEIEKDFTTYGEE of P. mirabilis urease and DELSPETSPYVKE of type XI collagen showed low reactivity with HLA-DRB1*0401 molecules. Furthermore, recent studies have shown that the type II collagen sequence VEIRAEGNRFTY bound with high affinity to the RA-associated motif but not to the nonassociated DRB1*0402 (7). However, in this study, the VEIRAEGNRFTY peptide showed low reactivity with HLA-DRB1*0401 molecules. These findings could be due to a difference in methodology and further studies with affinity-purified HLA-DRB1*0401 molecules may increase the sensitivity of the peptide binding assay. The results obtained provide some evidence for a possible role of E. coli and P. mirabilis in the pathogenesis of RA. It has been reported that RA patients with active disease have specific antibodies against P. mirabilis (3, 5, 12), a finding which correlates with both levels of C-reactive protein (3) and isolation rates of P. mirabilis (18). A decrease in anti-Proteus antibody levels and a decrease in a modified Stoke disease activity index was observed in RA patients treated with a high fluid and vegetarian diet (10). Furthermore, RA patients were reported to have elevated antibodies to the 63-kDa hemolysin protein of P. mirabilis and to a 16-mer synthetic peptide containing the ESRRAL sequence (17). The P. mirabilis peptide binding by RA sera has recently been confirmed by an independent group (4). In a related study, it was reported that Japanese patients with RA have increased antibodies against a 16-mer synthetic peptide of DRB1*0405 which also contains the EQRRAA sequence (13).

The ESRRAL motif was found in 5 of 77,573 sequences according to the Protein Information Resource database: P. mirabilis, Serratia marcescens, Vibrio cholerae, Rickettsia tsutsugamushi, and Brucella ovis. Two other organisms, Pseudomonas aeruginosa DQRRAA and E. coli EQKRAA have a sequence identity with EQRRAA, the RA susceptibility motif. However, we were unable to find any elevation in antibody levels against E. coli, Serratia sp., and Pseudomonas sp. in RA patients, although significant titers were present against Proteus sp. (15). Apart from such antibody responses against microorganisms in RA, T-cell reactions might also play a role in the development of the disease. The presentation of such microbial peptides as LEIEKDFTTYGEE of P. mirabilis by the RA-associated DR motif to CD4⁺ T cells could lead to the initiation of disease because of cross-reactivity with HLA-presented peptides of self-antigens of collagen type II and collagen type XI due to molecular mimicry. This CD4⁺ response to such a peptide needs to be measured in individuals with the appropriate HLA-DR alleles with or without RA.

The results presented in this study suggest that antibodies raised against P. mirabilis ESRRAL antigens during urinary tract infections could subsequently bind, albeit with lower affinity, to DR4-positive cells in tissues expressing the class II HLA antigens EQKRAA and EQRRAA, fix complement, and so initiate local inflammation that could lead to destruction of self-tissues by antibody-dependent cell cytotoxicity. It is important to note that E. coli is responsible for 80% of urinary tract infections in women, while P. mirabilis accounts for approximately 15 to 18% of cases. Interestingly, it is known that RA patients suffer an increased incidence of urinary tract infections (14). Furthermore, Proteus urease contains a sequence, IRRET, which cross-reacts with the sequence LRREI found in the ₂ chain of type XI collagen (17), a component of hyaline cartilage, and this cross-reaction could play a role in the development of erosions found in this disease.

The mechanism of RA disease pathogenesis is as yet unknown; however, the possible involvement of microorganisms in RA would appear to be gaining greater acceptance. The most likely candidates for this disease pathogenesis are E. coli and P. mirabilis.