Epitope Specificity of Anti–Heat Shock Protein 65/60 Serum Antibodies in Atherosclerosis
Abstract Levels of specific antibodies (Ab) against mycobacterial and human heat shock protein (hsp) 65/60 are increased in the sera of patients with atherosclerotic lesions and have been demonstrated to be capable of mediating endothelial cytotoxicity. To clarify the antigen epitopes recognized by these serum Abs, Ab binding to hsp65 deletion mutants (Dms), as well as to overlapping 15-mer and 8-mer hsp65 peptides, was assessed. Western blotting of hsp65 Dms indicated the presence of at least one epitope between amino acid (aa) residues 171and 276, recognized by both high-titer sera and affinity-purified anti-hsp65/60 Ab. Fluorescence immunoassays using 53 15-mer peptides and Pin ELISA using 526 7-mer peptides demonstrated three distinct, conserved sequences with high affinity to high-titer sera and purified anti-hsp65/60 Ab. Two N-terminal sequences, aa 97-109 and aa 179-187, and one C-terminal sequence, aa 504-512, were identified. These three epitopes recognized by anti-hsp65/60 Ab may serve as autoantigens in certain circumstances in vivo. This phenomenon could contribute to the initiation of atherosclerosis by an autoimmune reaction.
- Received March 18, 1996.
- Accepted July 2, 1996.
There has been much recent interest in the role of immune mechanisms in the development of atherosclerosis.1 2 Both cellular and humoral immune responses against hsp60, a phylogenetically conserved protein, have been studied extensively.3 The high interspecies sequence homology is reflected in how often immune reactions to hsp60 elicit an autoimmune reaction. hsp60, as a target of humoral and cellular immune reactions, has been reported to play a role in various human diseases,4 from rheumatoid arthritis and scleroderma to cystic fibrosis and atherosclerosis.5
We have previously demonstrated an increase of titers of anti-hsp65/60 Abs in the sera of individuals with carotid atherosclerotic lesions (Bruneck Study6 ). These Abs were proven to bind both mycobacterial hsp65 and its human counterpart, hsp60, with high affinity, indicating recognition of certain well-conserved epitopes on these molecules. As a consequence of cross-reactivity, they recognize hsp60-expressing cultured endothelial cells and macrophages treated with heat shock, as well as endothelial and intima cells in atherosclerotic lesions, where cells are subject to a variety of stressors resulting in cell damages.7 Recent investigations point to a cytotoxic effect of anti-hsp65/60 Abs on interaction with heat-shocked (42°C for 30 minutes) endothelial cells8 or macrophages.8A
The aim of the present work was to identify the epitope(s) on the hsp molecule recognized by anti-hsp65/60 Ab in sera of atherosclerotic patients as a basis for future characterization of autoreactivity and cytotoxicity toward human cells. Herein we provide evidence that these Abs bind specifically to three different hsp65 epitopes, which may be important for understanding the immune reaction to hsp in vivo.
Serum Samples and Ab Determination
Human sera were derived from the Bruneck Study, an epidemiological study of cardiovascular risk factors in clinically healthy individuals.9 Sera with high or low anti–mycobacterial hsp65 titers were obtained from probands, aged >65 years, with sonographically proven atherosclerotic lesions in their carotid arteries. Details of carotid sonography and hsp Ab determination by ELISA have been described at length.6 8 All high-titer sera selected had anti-hsp65 Ab titers of more than 1:1280 in the presence of carotid atherosclerosis, whereas Ab titers of low-titer sera did not exceed 1:40.
Affinity Chromatography of Anti-hsp65/60 Ab
Purification of serum anti-hsp65/60 Ab was performed following an established method described elsewhere.8 Briefly, immunoglobulin of pooled high-titer serum was precipitated by a standard (NH4)2SO4 procedure and incubated in a chromatography column with 2 mL agarose gel beads (Affi-Gel Kit, Biorad), coupled with 3 mg recombinant hsp65 (Stressgene Biotech Corp). Specific Abs were recovered by 20 mmol/L HCl elution, pooled and equilibrated with PBS, pH 7.2. Anti-hsp65/60 Ab titers of purified immunoglobulins were similar to original high-titer serum (1:1280), whereas unbound immunoglobulin had no measurable hsp Ab titer (<1:10).
Five hsp65 Dms (Fig 1⇓) were prepared as detailed previously.10 All mutants were fusion proteins with an N-terminal β gal residue and were generated by transforming E. coli M1070 with EcoRI-Sal I DNA fragments of the plasmid pRIB 1011, which has been subcloned into plasmid pEX2. The product of pEX2 plasmid–transformed E. coli M1070 served as a control for β gal alone. Samples, including Dms (1.5 μg each), recombinant hsp65 (0.5 μg), and β gal (pEX2, 1.5 μg), were diluted 1:10 (vol/vol) in sample buffer and separated on a 10% polyacrylamide gel under reducing conditions. Proteins were electrophoretically blotted onto nitrocellulose membranes (BA85, Schleicher & Schuell, Inc). After blocking with 2% milk powder/PBS (Merck) for 1 hour, filters were probed with anti-hsp65/60 Ab (50 μg/mL) or rabbit anti–β gal Ab (East-Acres Biologicals, 5 μg/mL) for 3 hours. Reactions were visualized by an HRP–anti-human Ig conjugate (catalog No. P212, Dakopatts) or an HRP–anti-rabbit Ig conjugate (catalog No. P217, Dakopatts), respectively, and by subsequent addition of 4-chloro-1-naphthol/hydrogen peroxide (Sigma).
A total of 53 biotinylated mycobacterial hsp65 peptides, each 15 aa long with an overlap of 5 aa, spanning the entire sequence of hsp65, were tested for recognition by anti-hsp65/60 Abs. Peptides were generated by standard solid-phase methods on a peptide synthesizer using FMOC aa pentafluorophenyl esters (Cambridge Research Biochemicals Ltd; see Reference 1111 ). Human anti-hsp65/60 Ab (100 μL, diluted 1:1000 vol/vol in PBS) or high-titer serum (100 μL, diluted 1:500 vol/vol in PBS) were coated onto each well of an ELISA plate (Petra Plastic, catalog No. 11041) at 4°C overnight. After washing with 0.05% Tween 20/PBS and blocking with 1 mol/L glycine (Merck) for 1 hour, 1 μg per well of each biotinylated peptide in 100 μL PBS was added. Reaction was detected by addition of europium-labeled streptavidin (Wallac, catalog No. 1244-360, diluted 1:800) and enhancement solution (Wallac, catalog No. 1244-104) and measured with a fluorometer (Wallac).
To rule out any specific or unspecific binding of certain peptides to plastic, blocking agent, or other assay reagents, we used a split-well technique to assess the reaction with or without the coating serum Ab for each single hsp65 peptide. The reaction in the absence of a peptide served as a negative control.
To encompass the entire mycobacterial hsp65 sequence, 526 7-mer peptides with an overlap of 6 aa were synthesized onto polyethylene pins using an epitope scanning kit (Cambridge Research Biochemicals Ltd, Reference 1212 ). Ab binding to the hsp65 synthetic peptide, immobilized on polyethylene pins, was measured by using a modified ELISA. The pins were blocked with 200 μL ELISA buffer (PBS containing 1% BSA, 1% ovalbumin, and 0.1% Tween 20) for 1 hour at room temperature. After washing with PBS-Tween 20, pins were incubated with 100 μL of anti-hsp65/60 Ab (1 μg per well in 100 μL ELISA buffer supplemented with 0.1% azide) or unbound Ig fraction as a control, for another hour at 4°C. After addition of an HRP–goat anti-human Ig conjugate (Jackson ImmunoResearch Labs, Inc; catalog No. 109-035-064), diluted 1:2000 in ELISA buffer, for 1 hour, pins were washed and the reaction was visualized by its substrate ABTS. Absorbance was read at 410 nm on a MicroELISA Autoreader (Dynatech Labs, Inc).
To assess specificity, each test was performed in duplicate, comparing peptide binding of anti-hsp65/60 Ab in the presence or absence of blocking by 50-fold excess recombinant hsp65 (50 μg per well). Results were considered specific when blocking exceeded a Δ optical density >0.05, and blocking of most of the specific epitopes exceeded 50% by far. Assay controls, without addition of anti-hsp65/60 Ab, were completely negative on all 526 pins. Pins were used repeatedly after thorough cleaning by sonication in disruption buffer (1% SDS, 0.1% 2-mercaptoethanol, and 0.1 mol/L sodium hydrogen phosphate).
To analyze the epitope specificity of anti-hsp65/60 Ab, we selected the sera of 10 patients with both high-titer anti-hsp65/60 serum Abs (Ab titer >1:1280) and severe carotid atherosclerotic lesions. These representative sera were either used directly or subjected to affinity purification over hsp65 columns. Thus, hsp65 epitope specificity was determined for each single serum, for the serum pool, and for the specific hsp65/60 Abs purified from these pooled sera.
To broadly determine anti-hsp65/60 Ab specificity and thus narrow down the immunodominant parts on the hsp65 molecule, we investigated anti-hsp65/60 Ab binding to a panel of hsp65 Dms. Dms were separated on an SDS–polyacrylamide gel under reducing conditions, blotted, and probed with human high-titer serum or anti-hsp65/60 Ab.
Since all the Dm preparations contained various amounts of GroEL, the hsp60 homologue of E. coli, simple measurement of protein content was not applicable for adjustment of administration of equal amounts of Dms for electrophoresis. However, since hsp65 Dms were fused with β gal, we were able to standardize the amount of Dm per lane by measuring β gal content, using probes for the blotted Dms with an anti–β gal Ab, as demonstrated in Fig 2B⇓. All lanes except the recombinant hsp65 showed similar signals when probed with anti–β gal Ab, thus indicating similar amounts of Dms (Fig 2B⇓). Since each molecule of Dm obviously is fused to a single β gal molecule, the concentrations of the fused hsp65 fragments were thus also comparable.
All three Dms covering the C-terminal part of hsp65, including pRIB 1463 (aa 404-540, Fig 2A⇑, lane 1), pRIB 1444 (aa 303-540, lane 2), and pRIB 1451 (aa 276-540, lane 3), were negative when probed with high-titer serum or anti-hsp65/60 Ab. However, mutant pRIB 1404 (aa 2-540), which encompassed the entire hsp65 molecule, clearly revealed positive reaction, indicating the presence of at least one epitope within the stretch aa −2-276 recognized by the specific Abs (lane 5). In addition, Dm pRIB 1426, comprising aa residues 171-540 was also recognized by high-titer serum and the anti-hsp65/60 Ab (lane 4). No signal except reactivity to 60-kD E. coli–GroEL, which was present in all Dm preparations, could be detected when β gal (pEX2) alone was probed by anti-hsp65/60 Ab (lane 7). In addition, a strong signal occurred when recombinant hsp65, as a positive control, was used (lane 6). These data led us to conclude that at least one epitope is present within aa 171-276 and is recognized by specific anti-hsp65/60 Ab. But it cannot be excluded that the region aa 276-540 also comprises further epitopes, which perhaps were hidden due to the fusion with β gal.
For a more detailed characterization of the epitope pattern recognized by anti-hsp65/60 Ab, we used a fluoroimmunoassay with a series of 15-mer hsp65 peptides overlapping by 5 aa. Ab binding to 53 peptides spanning the entire sequence of the hsp65 molecule and 10 nonsense peptides was analyzed and compared (Fig 3⇓).
A specific reaction to three distinct hsp65 peptides was found when probing with anti-hsp65/60 Ab. Pooled high-titer sera or individual sera showed the same binding pattern. The most N-terminally located sequence recognized comprised aa 91-105 and revealed high-affinity binding to the serum Abs. The next peptide sequence recognized by both anti-hsp65/60 Ab and high-titer serum was another N-terminal one, covering aa 171-185, which surprisingly is identical to the T-cell epitope associated with adjuvant arthritis in rats. The detection of these two N-terminal epitopes confirmed the data obtained from the binding studies to hsp65 Dms. In addition, a specific reaction with a third, C-terminally located peptide could be observed. This C-terminal peptide at aa 501-515 repeatedly showed the strongest binding to the anti-hsp65/60 Ab and antisera. The remaining 50 hsp65 peptides and control peptides showed no reactivity with the Abs.
As a control, serum Abs deprived of anti-hsp65/60 Ab did not recognize the three peptides described above, nor did they bind to any of the other peptides used.
To prove the specificity of the test, the anti-hsp65/60 Ab–coated wells were preincubated with a 10-fold concentration of recombinant hsp65 before the specific peptide was added. The binding of each of the three specific peptides could be blocked by the presence of whole recombinant hsp65. Preincubation with a similar concentration of BSA or ovalbumin had no effect on peptide recognition by the anti-hsp65/60 Ab (data not shown).
To confirm the results obtained in the FIA and to further narrow down the hsp65 epitope sequences recognized by anti-hsp65/60 Abs, we used a Pin ELISA, recording the reactivity of anti-hsp65/60 Abs to a total of 526 8-mer, 7-mer–overlapping hsp65 peptides synthesized on plastic pins. Due to this highly overlapping synthesizing mode, with only a single different aa between one pin and the next, we were able to define the anti-hsp65/60 Ab binding to the exact sequence involved. In this way, each of the aa sequences demonstrated with the FIA was also recognized and more closely defined by the Pin ELISA (Figs 4⇓ and 5⇓). For the N-terminal sequence aa 91-105 determined by FIA, we could define the second half of the peptide, aa 97-105, together with the four adjacent aa (-VAAG) in C-direction as the exact sequence reactive with hsp65/60 Ab (sequence I: aa 97-109, Fig 5⇓). Concerning the second FIA-derived sequence aa 171-185, once again the C-terminal half of the peptide beginning from aa 179 plus two adjacent aa (-EL-) proved to be the exact sequence recognized by anti-hsp65/60 Ab (sequence II, Fig 5⇓). In a similar fashion, we could confirm Ab binding to the C-terminally located hsp65 sequence by Pin ELISA. Positive reaction was observed just in the middle of the FIA epitope (aa 501-515, Fig 5⇓), thus narrowing Ab specificity to the sequence aa 504-512 (sequence III).
No additional epitopes were detected on screening all 526 hsp65 peptides for their reactivity to the specific Abs. Binding to control peptides was extremely low, proving that high binding activity was not due to nonspecific reactions of the Ab. The same results were obtained with pooled or single high-titer sera. The high assay sensitivity allowed for an analysis of anti-hsp65/60 Abs within low-titer sera. Although specific Abs were quantitatively low, they recognized the same three sequences as those described for Abs derived from high-titer sera (data not shown).
In assessing the aa sequence homology between mycobacterial hsp65 and human hsp60 we found that sequence I and II exhibit a total of 58% and 55% identical or similar aa, respectively. As the most conserved one, sequence III contains 78% homologous areas and its N-terminal heptapeptide is completely identical with its human counterpart.
Abs to hsp65/60 are present in the circulation of atherosclerosis patients, as well as clinically healthy human subjects. We have shown that titers of these Abs correlated positively with the severity of atherosclerotic lesions detected by sonography in carotid arteries.6 In vitro, these human Abs showed cytotoxicity to endothelial cells, producing high levels of hsp60 in response to heat stress,8 indicating a possible autoimmune reactivity of these antigen-antibody interactions. In the present study, we provide the first evidence that these Abs react specifically with three short, linear epitopes present in hsp65/60, which may be involved as autoantigens in the pathogenesis of atherosclerosis. The question of possible further conformational epitopes was not yet addressed in this study.
It is known that 70% of children in Western countries have atherosclerotic lesions in their coronary arteries as early as 10 to 12 years of age13 14 and almost 100% of humans over 30 show lesions in the arterial wall despite being clinically healthy. The blood cholesterol concentration alone is not a plausible explanation, since most of these children and young adults in the Western countries have normal blood cholesterol levels. Alternatively, circulating Abs to hsp60, which are present in almost all humans, may contribute to the development of atherosclerosis.
In our studies, the three epitopes on hsp65/60 were recognized not only by Abs from patients but also by Abs from clinically healthy subjects, which is not surprising given that atherosclerotic lesions occur in the arterial walls of all our blood donors aged over 50 years. This suggests that a lower level of Ab to hsp65/60 might play a crucial role in the development of atherosclerosis and a high level in promoting atherosclerotic lesions to atheroma, a late stage of the lesion. Alternatively, susceptibility to stressors of the artery as a target organ may determine immunopathological damage.15 Arterial endothelial cells, for example, differ greatly from venous endothelial cells in the pattern of adhesion molecules and hsp expressed in response to certain kinds of stimulation.15A As a third factor determining a differential Ab response, the presence of as yet undefined conformational epitopes should be considered.
An important question is whether the three epitopes demonstrated in the present study directly mediate an autoimmune reaction possibly involved in the pathogenesis of atherosclerosis. Given the fact that increased hsp60 expression was seen in endothelial cells overlying atherosclerotic lesions and sites of major hemodynamic stress in general (eg, branching of large vessels; see References 7 and 167 16 ) and that in vitro surface expression of hsp in endothelial cells emerged and was recognized by these Abs to hsp60,8 we speculate that epitope(s) of hsp60 may be responsible for an antivascular autoimmune reaction.
Since mycobacterial hsp65 and E. coli GroEL display a high degree of sequence homology, it is likely that hsp65 possesses a very similar three-dimensional structure to that recently described for GroEL.17 As a consequence of this and because serum anti-hsp65/60 Abs also exhibit a high binding capacity to GroEL, the corresponding areas of Ab binding on the three-dimensional GroEL structure may be of interest. Interestingly, sequence I (aa 97-109) and sequence III (aa 504-512), located distantly from each other on the linear aa sequence, are closely related areas on the corresponding tertiary GroEL structure. The missing reactivity of anti-hsp65/60 Ab to sequence III in Dms lacking sequence I may corroborate the presence of a conformational epitope formed by these two structures. Both are part of the equatorial domain of GroEL and participate in two side-by-side α helices. Furthermore, both α helices are directed toward the outside of the 7-mer ring structure and do not involve any part of the central channel. In general, α-helical structures are considered to be good immunogens for Ab recognition and are also known to be capable of preserving their structure even in peptide form. In contrast, sequence II, resembling the arthritogenic T-cell epitope, is located within the small intermediate domain of GroEL.
All but four of the 35 intersubunit contact sites17 are located distantly from the sequences we described as involved in the formation of linear B-cell epitopes. Sites critically involved in intersubunit binding are not found within sequence III but only at the C-terminal end of sequence I (Lys105 and Ala109). Notably, two intersubunit bounds (Thr181 and Leu183) are located exactly in the middle of sequence II (aa 179-187) and probably interfere with Ab binding to the hsp 7-mer. However, the existence of Abs against sequence II suggests their accessibility to the humoral immune system, where in contrast to T-cell recognition, no antigen processing occurs. Most probably, the degradation of the 7-mer ring structure during prokaryotic or eukaryotic cell death or physiological turnover of hsp molecules leads to increased accessibility of sequence II and immunogenic recognition of this cryptic epitope.
During the last few years, a number of major and minor T-cell epitopes have been identified,18 rendering hsp65 highly immunogenic for the cellular immune system. Our sequence II is identical with the major T-cell epitope thought to play a role in adjuvant arthritis in rats,19 and sequence I shows a great overlap with a T-cell epitope associated with the appearance of recurrent oral ulcers.20 Sharing of identical epitopes recognized by T and B cells is a well-known phenomenon in autoimmune disease.21 In addition to its role as humoral and cellular hsp65 epitope, sequence II shows homology with a peptide of toxic shock syndrome toxin-1.22
Ongoing studies of the possible existence of conformational B-cell epitopes on the hsp65 molecule may clarify anti-hsp65/60 Ab binding and correlate specific epitopes to arterial cell damage.
Selected Abbreviations and Acronyms
|ELISA||=||enzyme-linked immunosorbent assay|
|β gal||=||β galactosidase|
|hsp||=||heat shock protein|
We thank Dr L. Mizzen, Stressgen Biotechnologies Corp, Victoria, BC, Canada, for kindly providing recombinant Mycobacterium tuberculosis hsp65 and T. Öttl for preparation of photographs.
Seifert PS, Hansson GK. Complement receptors and regulatory proteins in human atherosclerotic lesions. Arteriosclerosis. 1989;9:802-810.
Cohen IR. Heat shock protein 60 and the regulation of autoimmunity. In: Stress Proteins in Biology and Medicine. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1990.
Xu Q, Dietrich H, Steiner HJ, Gown AM, Mikuz G, Kaufmann SHE, Wick G. Induction of atherosclerosis in normocholesterolemic rabbits by immunization with heat shock protein 65. Arterioscler Thromb. 1992;12:789-799.
Xu Q, Luef G, Weimann S, Gupta RS, Wolf H, Wick G. Staining of endothelial cells and macrophages in atherosclerotic lesions with human heat-shock protein reactive antisera. Arterioscler Thromb. 1993;13:1763-1769.
Schett G, Xu Q, Amberger A, Van der Zee R, Recheis H, Willeit J, Wick G. Autoantibodies against heat shock protein 65 mediate endothelial cytotoxicity. J Clin Invest. 1995;96:2569-2577.
Willeit J, Kiechl S. Prevalence and risk factors of asymptomatic extracranial carotid artery atherosclerosis: a population based study. Arterioscler Thromb. 1993;13:661-668.
Thole JE, Van Schooten WC, Keulen WJ, Hermans PW, Janson AA, De Vries RR, Kolk AH, Van Embden JDA. Use of recombinant antigens expressed in Escherichia coli K-12 to map B-cell and T-cell epitopes on the immunodominant 65-kilodalton protein of Mycobacterium bovis BCG. Infect Immun. 1988;56:1633-1640.
PDAY Study Group. Natural history of aortic and coronary atherosclerotic lesions in youth. Arterioscler Thromb. 1993;13:1291-1298.
Amberger A, Maczek C, Jürgens G, Michaelis D, Schett G, Trieb K, Eberl T, Jindal S, Xu Q, Wick G. Co-expression of ICAM-1, VCAM-1, ELAM-1 and Hsp60 in human arterial and venous endothelial cells in response to cytokines and oxidized low density lipoproteins. Cell Stress Chaperones. In press.
Anderton SM, Van der Zee R, Noordzij A, Van Eden W. Differential mycobacterial 65-kDa heat shock protein T cell epitope recognition after adjuvant arthritis-inducing or protective immunization protocols. J Immunol. 1994;152:3656-3664.