A synthetic 10‐kD heat shock protein (hsp10) from Mycobacterium tuberculosis modulates adjuvant arthritis

The heat shock protein, hsp10, is an abundant protein in Mycobacterium tuberculosis (Mtb), its nucleotide sequence encoding a protein of 99 amino acids with a molecular mass of 10±7kD. This sequence is phylogenetically conserved, being represented by the GroES homologue of Escherichia coli. Hsp 10 and GroES are members of the chaperonin 10 family of molecular chaperones, and GroES is necessary for the optimal activity of GroEL, a member of the chaperonin 60 family and the E coli homologue of mycobacterial hsp65. Since hsp65 has been implicated in both experimental and human rheumatoid arthritis, we aimed to assess the immunomodulatory effects of its co‐chaperonin, hsp10, in experimental arthritis. Our results show that an aqueous solution of a mycobacterial hsp10 delayed the onset and severity of adjuvant‐induced arthritis in rodents when administered after disease induction but before joint involvement occurred. This biological activity was specific for the hsp10 of Mtb, since neither GroES nor the rat homologue was effective. Using synthetic hsp10 fragments, the activity was localized to the N‐terminal region of the molecule. Assessment of circulating antibody levels to mycobacterial hsp10 and hsp65 indicated that all arthritic rats had increased litres to both hsp10 and hsp65: hsp10‐treated rats showed further elevation of this humoral response not only to hsp10 but also to hsp65 when compared with the untreated arthritic control. This is the first report of the immunomodulatory activity of mycobacterial hsp10 in experimental arthritis, and exhibits a potential role for this co‐chaperonin in pathophysiological situations.


INTRODUCTION
Hsp 10-kD (hsplO) is an abundant protein in Mycobacterium tuberculosis (Mtb), its nucleotide sequence encoding a protein of 99 amino acids with a molecular mass of 10.7 kD [l 11. Widely conserved among different organisms, sequence homologies have been defined among Mycobacteria [12] and enteric bacteria [13]. More recently, a rat hsplO was identified [14], and the human homologue has been cloned [I 51.
Functionally, hsplO is a chaperonin and co-operates with hsp65 to achieve optimal protein folding and assembly [16]. Since hsp65 is also highly immunogenic, several investigators have addressed the question of immunoreactivity elicited by the ,O-kD hsps. Immunization of guinea pigs with whole Mtb sonicates followed by challenge with hsplO induced DTH [17], and the 10-kD antigen was later shown to be an important Heat shock (stress) proteins (hsps) are ubiquitous throughout nature, essential to life and are induced by a number of stressors, including inflammatory mediators; they are classified into families according to molecular weight [l]. Since stress protein genes are highly conserved across many genera, the ability of immune hosts to recognize their protein products (self) as antigens raises important questions about the regulation of the immune response, autoimmunity occurring due to failure of these discriminatory mechanisms [2]. In particular, a possible link between hsps and rheumatoid arthritis (RA) [3-71 and hsps and experimental arthritis [8-101 has been reviewed by many authors. Mtb, towards which Mtb-specific T cell lines proliferate [18]. Thus, it appears that hspl0 shows immunodominance. elicits T and B cell immunc responses, and may have immunomodulatory activity. Previously, using immunohistology, we have demonstrated enhanced expression of hsplO and hsp65 in human arthritic synovial membrane [27]. Since, in vitro, hspl0 functions co-operatively with hsph5 [16], we studied the immunogenicity of a chemically synthesized hspl0 protein from Myco. tuberculosis and its elrect on adjuvant-induced arthritis (AA) in rats.

Animals
Wistar rats were obtained from Charles River (Margate, UK) and Lewis rats from Harlan Olac (Bicester, UK). The rats were fed a standard diet and tap water. Rats weighed z15Og at the start of the experiment. Young female rats were chosen because they are more susceptible to arthritis induction.

Protocol for induction of AA
Heat-killed human strains C, DT and PN of M-yco. tuberculosis (Central Veterinary Laboratory, Weybridge, UK) were finely ground in a pestle and mortar and suspended in light paraffin oil to a filial concentration of 10mgiml. Rats were inoculated intradermally at the base of the tail with a total of 100 pl of the suspension [31]. The day of arthritis induction was dcsignatcd as day 0 and disease severity evaluated using the following standard clinical scoring system ( [31], adapted from the work of Currey & Ziff [32]): 0, no inflammation; 1, slight redness and swelling of the foot; 2, swelling of the foot such that the tendons are no longcr visible; 3, swelling extending to the ankle joint; 4, gross inflammation and deformity of the ankle joint; the tail was scored 0 or 1 according to the absence or presence of cutaneous nodules; ears were scored 0 or I according to the absence or presence of subcutaneous haemorrhages. Scores were summed, giving a potential maximum of 18 for each animal. Animals were killed aftcr 28 days (earlier if 0 1996 Blackwell Science Ltd. Clinicnl and Espcrimental Immunology, adverse reactions ensued or protcction was minimal) and exsanguinaled. Sera were stored at -20°C.

Treatment with hsplO after A A induction
On day 0, AA was induced in Lewis and Wistar rats. On days 4, 5 and 6 groups of five to six animals were injected with hsplO or related fragments, each receiving 50 pg of protein/peptide in 100 pl PBS intradermally at the base of the tail. A control group received PBS alone on days 4, 5 and 6, while the absolute control was not treated and received Mtb only on day 0. To assess the specificity of the hsplO response similar experiments were carried out on Lewis rats using rhsp65, E. coli GroES, or a chemically synthesized rat hspl0 in place of the mycobacterial hspl0.
Prolonged treatment with synthctic hsplO On day 0, AA was induced in Lewis rats. The rats were treated with a dose of 50 kg of hsplO in 100 pl PBS on days 0,2,4,6,8, 10, 12 and 14. Control groups received 1 0 0~1 PBS only (PBS control) or no injection (absolute control).
ELISA for untibodies to hsp65 and hsplO Serum samples were thawed immediately before use, and circulating antibodies to recombinant Myco. leprae hsp65 (rhsp65 antibodies) and to synthetic mycobacterial hsplO (hspl0 antibodies) were measured by conventional ELISA as previously described [33]. Briefly, microtitre plates were coated with rhsp65 or synthetic hsplO in carbonate buffer pH 9.6. Plates were blocked with 1% bovine serum albumin (BSA) to minimize non-specific reactions and test sera diluted accordingly: 1:25-1:200 for hsp65 antibodies and 1:400-1:51 200 for hspl0 antibodies. Peroxidase-conjugated rabbit anti-rat immunoglobulin (Dako Ltd, High Wycombe, UK) was added and the plates were developed by the addition of orthophenylene diamine supplemented with hydrogen peroxide. Absorbance was monitored at a wavelength of 492 nm.
All sera were assayed in duplicate and a reference serum was included in duplicate on each plate. Since readings can vary greatly between assays, the use of a reference serum (OD492 control) allows a more accurate comparison. The reference serum was from a pool of eight naive rats, and OD492 readings (& s.e.m.) varied as follows: 0.027 (iO.003): it w7as also stored in aliquots with the test sera at -20°C. lnterassay and intra-assay variabilities were always < 10%.
Results were calculated as the ratio (OD492 sample/OD492 control) using serum dilutions of 150 for rhsp65 antibodies and 1:12 800 for hspl0 antibodies. The dilutions were chosen so that the levels lay on the linear portion of the curve.

Statistical analysis
Statistical comparisons for clinical scores and antibody levels were made using the Mann-Whitney U-test with Minitab data1 analysis software (Minitab Inc, Coventry, UK). Lewis rats (Fig. lb) experienced a more modest, yet still significant protective effect ( P < 0-01 versus Abs control and P < 0.02 versus PBS control on days 9-14; P < 0.02 on days 17 and 18 and P < 0.05 on days 7 and 19 versus Abs control).

Treatment with hsp10 in PBS after A A induction
This experiment was repeated, with analagous results, in six separate experiments using different batches of hspl0, but, due to restrictions in the amount of synthetic compound available in each batch, a full dose response was not ascertained. Therefore, the results presented in Figs la,b do not necessarily represent optimal modulation. In Lewis rats, the pattern of response was very reproducible: on day 10, there was = 80% reduction in clinical scores, but this was not sustained.
By day 12, only a 60% reduction in severity was observed, and, by days 18, 19 and 20, animals showed only a 15-20% benefit (e.g. in Fig. lb, on day 20, mean clinical scores for hsplO treatment were 12-2 while the untreated Abs control group showed a mean score of 16.3; this equates to a 25% improvement).
All further investigations on the characterization of this protective response were carried out using Lewis rats. The reasons for this choice are discussed below.

Antibodies to rhsp65 and hsplU
To assess the humoral immune response in hspl0-protected Lewis rats, we measured the antibody lcvcl to rhsp65 and hspl0. Figure 2a shows the antibody titre to hspl0. The hspl0-treated group showed a significantly elevated antibody level (median 31.2) compared with the Abs control (median 7.22, P < 041) and the PBS control (rncdian 11.67, P < 0.02) groups. Similarly (Fig. 2b), the rhsp65 antibody level in the hspl0-treated group was significantly elevated (median 4.3) in comparison with a PBS control (median 2.2, P < 0.05) and approached significance towards the absolute control (median 1.97, P < 0.08). th Days assessed to localize the biological activity of hspl0; groups of Lewis rats received 50 pg peptide in 100 pl PBS on days 4,5 and 6, as described. As previously stated, optimal activity was not determined. In Table 1, two separate experiments (a and b) are reported: only fragment 1-25 showed any significant protective effect. There was a significant delay of the onset at day I1 ( P < 0.02 versus the arthritic control), but later (day 20) a tendency to reach high clinical scores. Figure 3 compares the kinetics of the disease course using peptide fragment 1-25 and the whole synthetic mycobacterial hspl0. By day 19, hsplO treatment produced a 21% reduction in clinical score, while fragment 1-25 had ceased to be effective. Fragment 1-58 was similar in effect, showing an early, although statistically nonsignificant, effect on disease severity, but later clinical scores were similar to the control groups. Fragments 25-99, 51-99 and 75 99 had no effect on disease kinetics when administered on days 4, 5 and 6 after AA induction.

Specijicity of the mycobacterial hsplO modulation
To assess specificity and to control for any artefacts in synthesis, an irrelevant peptide, p24, was synthesized. This peptide corresponds to the 104 amino acids of the C-terminus of HIV virus [29] and is therefore of equivalent length to hsplO but structurally unrelated. As Table 2a shows, p24 was ineffective when tested under the same experimental conditions. Neither rhsp65 (Table 2b) nor the rat hsplO (Table 2c) could mimic the biological activity of mycobacterial hspl0. GroES, the E. coli homologue of hsplO (Table 2d), showed a slight but statistically insignificant activity which was lost immediately following onset of disease.   Groups were of six rats. Assessment of disease severity was made by clinical scoring. Levels of significance compared with the absolute control were * P < 0.01; **P < 0.02. Levels of significance compared with the PBS control were t P < 0.01 (Mann-Whitney (/-test).

Prolonged therapeutic treatment with hsplO in PBS
Because early treatment with hsplO appeared to delay the onset and decrease the severity of AA, we determined whether prolonged treatment could achieve and maintain increased protection against arthritis. As the arrows in Fig. 4 indicate, rats were treated with hsplO in PBS on days 0, 2, 4, 6, 8, 10, 12 and 14 after AA induction. The profile of disease for the hspl0-treated groups was almost identical to the previous experiments carried out with a shorter treatment. There was a clear delay of onset ( P < 0.01 versus PBS control and Abs control from day 6 to day 13), but again the disease progressed, although less severely, maintaining significantly lower clinical scores compared with the Abs control ( P < 0.02 on days 13, 14 and 16).

DISCUSSION
AA is recognized as a T cell-mediated disease, since early therapeutic intervention with T cell antisera (e.g. anti-CD4) abrogates disease induction [32,34]; similar immunomodulation of human RA has been attempted using T cell MoAbs, with variable success [35]. The finding that prior injection of mycobacterial hsp65 (chaperonin 60) protected Lewis rats against AA induction [36] fuelled speculation of a cure for arthritis. However, Hogervorst et al. [37] demonstrated that administration of hsp65 following disease induction, exacerbated arthritis. The mechanism is clearly complex. Since co-operativity between chaperonins has been demonstrated [16], we examined the effect of mycobacterial hsplO (chaperonin 10) on the early response to induction of adjuvant disease.
In this study, we have described a new biological effect of a chemically synthesized chaperonin 10 protein, the mycobacterial 10-kD heat shock protein, hspl0. Although hsplO appeared to have a more pronounced effect on AA in Wistar rats (Fig. la), we chose to examine AA in Lewis rats, as much of the published work with regard to hsp protection uses this model [8-lo]. Moreover, the Lewis rat may be more indicative of the clinical situation, since differences in the response of Lewis rats may be attributable to the known impairment in their hypothalamo-pituitary-adrenal axis [38], a condition also noted in patients with RA [39].
Modulation of AA was specific to the mycobacterial form, since GroES (the E. coli homologue) and the rat homologue did not elicit protection in our system. Inactivity of the rat (autologous) hsplO is interesting, as rat hsplO shows almost 100% homology with human EPF [26], an immunomodulatory molecule [23-251. Patients with RA often show a transient remission during pregnancy [40], perhaps akin to the effect seen here with hspl0.
As our results show, the specificity of the system is exquisite, since there is no observed response to self hsplO (rat) or GroES (E. coli) while the mycobacterial protein promotes a suppressive response. This accords well with a recent report by Anderton et a!. [41] using chaperonin 60 proteins, which showed that, although mycobacterial hsp6S can protect Lewis rats against AA induction, the corresponding self (rat) protein is ineffective. Moreover, they identified a peptide sequence in the N-terminus (residues 256-270) of hsp65, and immunization with this peptide was able to confer protection against AA in Lewis rats: a homologous epitope occurred in rat chaperonin 60. Using T cell lines derived from rats immunized with this chaperonin 60 peptide 256-270 (self or mycobacterial) they inferred that recognition of cross-reactive epitopes on self and bacterial hsp60, having low M H C affinity, might be used to maintain a protective autoreactive memory T cell population, thereby eliminating the requirement for mycobacterial involvement. In our study, only mycobacterial fragments were tested.
An alternative hypothesis, derived following studies using a murine arthritis model, suggests that autoimmune disease arises through stimulation of immunologically ignorant autoreactive T cells [42]. In this model, although arthritis is induced in the absence of Mtb by a non-antigenic paraffin oil (pristane), arthritic mice display T and B cell immunity to both mycobacterial and mammalian chaperonin 60, and mice can be protected against arthritis induction by inoculation with hsp65. Examination of the mechanisms involved demonstrated that arthritic mice showed T cell reactivity to multiple epitopes on hsp65 while protected mice had a limited repertoire. Therefore it is possible that, in our experiments, this 'epitope spreading' was limited by early treatment with hspl0.
From the humoral aspect we have demonstrated that in Lewis rats with AA, antibody production to hsplO is significantly elevated, underlining the importance of hsplO as a relevant antigen in this experimental disease. Furthermore, rats receiving three consecutive injections of hsplO in an aqueous solvent (PBS) after adjuvant injection (Freund's complete adjuvant (FCA)) mounted, in comparison with the arthritic controls, a significantly elevated response to hsp 10 itself and, interestingly, also to hsp65. Cross-reactivity between chaperonin antibodies is not surprising since the genes are highly conserved; however, the degree of cross-reactivity is highly dependent on the nature and sitc of antigen presentation [43]. Sequence similarities within the E. coli chaperonins GroEL (hsp65 homologue) and GroES (hspl0 homologue) are documented [44] and, using the Daresbury computer database, we showed that hsplO shares 69% homology with the 65-kD (My'co. Zeprae) antigen [33]. Thus, it is not inconceivable that our results might be explained by a sequence or by a stoichiometric cross-reaction. Moreover, as increased hsp65specific antibody levels are associated with resistance to AA induction in Wistar rats [45], it is attractive to hypothesize that the 10-kD antigen is acting as a modulator of the immune response towards related heat shock (chaperonin) proteins.
Recently, a 12-kD protein of Mycn. tuberculosis has been ascribed a protective role in a model of experimental encephalomyelitis [46], and Ferrero vt ul. [47] have evidence that a protein belonging to the GroES class is a protective antigen in mucosal infection in mice. Our results provide further support for the potential use of chaperonin 10 proteins in the therapy of autoimmune diseases and infection.