Humoral Immunity to Stress Proteins and Periodontal Disease J Periodontol • October 1999 1185 Humoral Immunity to Stress Proteins and Periodontal Disease Dennis E. Lopatin,* Charles E. Shelburne,† Neal Van Poperin,* Charles J. Kowalski,* and Robert A. Bagramian‡ Background: There is evidence that microbial heat shock (stress) proteins (Hsp) are immunodominant antigens of many microorgan- isms. Immunity to these proteins has been shown in non-oral infections to contribute to protection. This study was undertaken to assess the relationship(s) between immunity to human and microbial heat shock proteins, periodontal disease status, and colonization by periodontal disease-associated microorganisms. Methods: Subgingival plaque and blood samples obtained from 198 patients during an earlier clinical study were examined for the presence of specific periodontal disease-associated microorganisms and anti- bodies to selected human and microbial heat shock proteins (Hsp70, Hsp90, DnaK, and GroEL). Particle concentration immunofluorescence assay (PCFIA) was used to detect anti-Hsp antibodies and slot immunoblot assay (SIB) was used to detect subgingival plaque species. Regression models were used to examine the contribution of age, gender, gingival index, probing depth, attachment loss, calculus index, plaque index, and microbial colonization to the anti-Hsp anti- body concentrations. Results: Our studies demonstrated that, when evaluated by ANOVA, patients with higher anti-Hsp (Hsp90, DnaK, and GroEL) antibody con- centrations tended to have significantly (P ≤0.05) healthier periodontal tissues. This was most obvious when the relationship between mean probing depths and antibody concentrations were studied. For Hsp90 antibodies, 2 variables (probing depth and P. gingivalis concentration) were found to have significant contributions (R = 0.293, P <0.0002). The equation derived from the regression model was y = 12558 − 2070*PD + 1842*PG. This confirmed the inverse relationship with probing depth and the positive relationship with colonization by P. gingivalis. Attempts to model the other stress protein antibodies were not successful. Conclusions: We believe that the present observations reflect the presence of protective anti-Hsp antibodies, rather than simply the pres- ence of the microorganism in the gingival sulcus. The clinical signifi- cance of these observations lies in the potential of identifying patients who are at risk for developing periodontal disease based on their in- ability to mount an immune response to specific Hsp or Hsp epitopes, as well as the development of vaccines based on Hsp epitopes. J Periodontol 1999;70:1185-1193. KEY WORDS Periodontal diseases/microbiology; protein, microbial heat-shock; protein, stress; antibody formation; immunity; humoral. A ssociations between host response and severity or progression of periodontal disease have classically been stud- ied in the context of immunity directed against specific micro- organisms.1 This has been true even though dental plaque repre- sents a highly complex array of microorganisms, each having both unique and common antigens that contribute to its overall antigenicity and immunogenicity. Studying the potential role of host immunity in the context of such a complex anti- gen mixture is difficult and one is never certain whether differences in host response represent changes in magnitude of the response to a specific antigen or in qualitative differences in the antigenic array. Recently, there has been a greater focus among investigators upon specific purified bacterial antigens such as lipopolysaccharides, lipote- ichoic acids, fimbrial proteins, and other cell membrane-associated antigens.2-6 Heat shock proteins (Hsp) have been extensively evaluated in numerous medically significant dis- eases such as candidiasis, Lyme disease, Chlamydia, and tubercu- losis. We questioned whether this family of evolutionarily-conserved proteins had a significant role in periodontal disease.7 These pro- teins, defined in part by the mol- e c u l a r s i z e o f t h e i r p r o t o t y p e * Department of Biologic and Materials Sciences, School of Dentistry, University of Michigan, Ann Arbor, MI. † 3M, BioMaterials Technology Center, St. Paul, MN. ‡ Department of Periodontics, Prevention, and Geriatrics, University of Michigan, Ann Arbor, MI. members, have significant roles in normal cellular function as molecular chaperones8,9 and defense against environmental stresses.10 They have been shown to have an important role in inflammatory mechanisms,11 autoimmune diseases,12-14 and atherosclerosis.15 In addition, these proteins have been shown to be the immunodominant antigens of many microorganisms including periodontal disease- associated microorganisms such as Porphyromonas gingivalis16-19 and Actinobacillus actinomycetem- comitans.20-21 Because these proteins have been so highly conserved in evolution we felt that they might be used to assess “basal” or “generic” immunity in a variety of disease processes and would not necessar- ily be specifically linked to a single microorganism. In the present study, we have employed a panel of purified stress proteins and measured serum IgG antibody levels to them using particle concentration fluorescence immunoassay. These data were then compared to clinical measurements of periodontal disease and colonization by microorganisms com- monly associated with periodontal disease. The spec- imens used in this study were derived from an Amish and non-Amish population examined in an earlier study.22,23 MATERIALS AND METHODS Study Population This study focused on the Amish and non-Amish liv- ing in rural southeastern Michigan. The clinical data and archived specimens collected from a total of 198 Amish (n = 145) and non-Amish (n = 53) specimens were used. This represented all of the subjects recruited during the first year of an earlier study.22,23 The results of that study indicated that the Amish did not present as a unique population that could be dis- tinguished from the non-Amish. This study was per- f o r m e d u s i n g c l i n i c a l f i n d i n g s a n d s p e c i m e n s archived from that study. Participation rate for these subjects was approximately 77%. Ages ranged from 18 to 79 years; 49% (97) males and 51% (101) females. Since the non-significant differences found in clinical measurements between the Amish and non-Amish were attributable to oral hygiene practice, later analyses treated all subjects as a single popula- tion. Subsequent regression modeling confirmed this assumption. The oral disease in this population is consistent with age-associated chronic adult peri- odontitis with no evidence of early-onset periodontal diseases. Clinical Measurements One dentist conducted examinations of all existing teeth with the exception of third molars using the Silness and Löe plaque index,24 Löe and Silness gin- gival index,25 and modified Ramfjord’s periodontal disease index for probing depth and attachment loss measurements.26 Loss of periodontal attachment and pocket depth were measured to the nearest millime- ter using a periodontal probe. Measurements were taken from 6 sites of each tooth: disto-buccal, mid- buccal, mesio-buccal, disto-lingual, mid-lingual, and mesio-lingual. Blood Collection Blood samples were collected in 10 ml vacuum tubes without anticoagulant. After clotting, the tubes were centrifuged and the serum frozen at −80°C. Plaque Specimens Plaque samples were collected from the 6 most peri- odontally diseased sites based on probing depth and/or inflammation. These 6 samples were then pooled. This approach allowed us to assess the flora representative of the most diseased sites in each sub- ject. Supragingival plaque about the most diseased sites was removed with a sterile curet and discarded. A second sterile curet was introduced into the sulcus or periodontal pocket and extended as far apically as possible. The root surface was then sampled with a curet and the adherent plaque on the scaler tip trans- ferred to a screw-capped cryotube containing 0.5 ml phosphate buffered saline (PBS; 0.05 M sodium phos- phate, 0.15 M NaCl, pH 7.4) with 0.5% formaldehyde, EDTA (2 mM), PMSF (1.0 mM), pepstatin A (0.1 mM) and leupeptin (0.5 mg/l). Sample vials were stored at −20° until assayed. Slot Immunoblot Assay (SIB) The SIB was performed as previously described.27 The samples were ultrasonicated§ (3 to 5 seconds; 60% power) to disrupt aggregates of plaque particles. Nitrocellulose sheets‖ were soaked for 15 minutes in TBS (0.05 M NaCl, 10.0 mM Tris-HCl, 10.0 mM Tris- base, pH 7.4) and inserted into the slot blot manifold.¶ Standards (pure cultures) or undiluted plaque sam- ples were applied (10 µl) to each well of the slot blot manifold which was then evacuated with gentle appli- cation of vacuum. The nitrocellulose membrane was then removed from the manifold for further process- ing. All subsequent incubations were performed on a rocking table# at room temperature. Microbial detection and quantitation was performed after first immersing the nitrocellulose membrane in TBS containing 0.5% non-fat dried milk** for 30 to 60 minutes to block unoccupied binding sites on the nitrocellulose membrane. The appropriate antibacter- ial antibody, diluted in TBS-Tween 20 (TBS-T; 0.05% Tween 20) containing 0.5% milk proteins** was Humoral Immunity to Stress Proteins Volume 70 • Number 10 1186 § Kontes Instruments. ‖ BA-85, Schleicher & Schuell, Keene, NH. ¶ Minifold II, Scheicher & Schuell. # Hoefer Scientific Co., San Francisco, CA. ** BLOTTO, Kirkegaard & Perry Laboratories, Gaithersburg, MD. J Periodontol • October 1999 Lopatin, Shelburne, Van Poperin, Kowalski, Bagramian 1187 applied and allowed to incubate for 1 hour. Following three 5-minute washes in TBS-T, goat anti-rabbit IgG conjugated to alkaline phosphatase†† diluted in TBS- T containing 0.5% milk proteins was applied and incubated for 1 hour. After all washes, the BCIP-NBT substrate solution‡‡ was applied and color develop- ment allowed to proceed to its maximum. Absorbed polyclonal antisera prepared in rabbits specific for P. gingivalis, A. actinomycetemcomitans, T. denticola, and B. forsythus were used in the SIB assay. The specificity of these antisera has been previously described.28,29 Particle Concentration Fluorescence Immunoassay (PCFIA) PCFIA was performed as described by Shelburne et al.30 Stress proteins used in these studies were lim- ited to those that were commercially available since periodontal pathogen-specific stress proteins were either not available or were available in amounts insuf- ficient for screening purposes. The initial step in the performance of PCFIA was the preparation of antigen- coupled latex particles. Briefly, 100 µg of the specific stress protein (human Hsp70, human Hsp90, E. coli DnaK, and E. coli GroEL) was diluted to 10 ml in 0.1M sodium phosphate (pH 7. 0) and combined with 10 mg of carbodiimide§§ and 1ml of carboxyl latex particles‖‖ (0.8 µM, 5% w/v,) and incubated for 1 to 2 hours at 37°C with rotation. The coated particles were washed 3 times in 0.1M sodium phosphate and resuspended in 20 ml of PBS + 0.2% sterile milk proteins. Twenty µl of the coated particles were dispensed into each well of 96 well filtration plates.‖‖ Sera were diluted 1:100 in PBS + 0.2% sterile milk proteins, added to wells (20 µl/ well) in triplicate, and incubated 30 minutes at room temperature. The particles were then washed and 20 µl of FITC labeled anti-human IgG (various vendors) added. After an additional 15-minute incubation the particles were washed 3 times and the fluorescence of the labeled anti-human IgG captured by the solid phase was measured by epifluorimetry and expressed in rela- tive fluorescence units (RFU). Controls included pooled human serum to control for inter-plate variability and FITC-labeled anti-human IgG in the absence of 1:100 dilution of patient serum to control for non-specific binding. Preliminary studies indi- cated that a 1:100 serum dilution was optimal, resulting in detection of all sample fluorescence at the same gain setting (gain = 25) in the linear range of the machine. The relative amounts of fluorescence in each well was used as a semi- quantitative representation of the amount of anti- body present in the sample. Statistical Analysis The relationships between specific indices of periodontal disease, antibody level, and colo- nization of subgingival dental plaque were assessed by ANOVA. Fisher’s method for multiple comparisons was used for pair-wise comparison of the groups. The contribution of these variables to stress protein anti- body concentration, as well as those of age, gender, and Amish/non-Amish status, was assessed in multi- ple and stepwise regression models.¶¶ Results with P ≤0.05 were considered significant. RESULTS Clinical Measurements A complete description of the clinical status of this population is available in earlier reports.22,23 An overview of the clinical status of the subjects is pro- vided in Table 1. All analyses described below used full mouth means for all clinical measurement (prob- ing depth, attachment loss, and gingival index). Colonization and Clinical Measurements The relationship between colonization and the clinical measurements was determined by ANOVA. As shown in Table 2, when probing depth was examined, incr easing pr obing depth was associated with increased colonization by all species tested. In all cases but A. actinomycetemcomitans, these increases were significant. We observed no statistically signifi- cant relationships when gingival index or attachment loss were evaluated. Anti-Stress Protein Serum Antibody Concentrations and Clinical Measurements Subjects were stratified into 3 groups (0, 1, and 2) based on their mean whole mouth gingival index (0 = 0 to <0.5; 1 = 0.5 to <1.5; 2 = >1.5). The anti-stress pro- tein antibody concentrations in these 3 groups were then compared by ANOVA. As shown in Figure 1B, a consistent trend is noted. Gingival index strata of 0 and 1 were associated with higher antibody concentrations, while scores greater than 1 (significant amounts of gin- †† Bio-Rad Laboratories, Hercules, CA. ‡‡ Kirkegaard & Perry Laboratories. §§ Sigma Chemical Co., St. Louis, MO. ‖‖ Epicon, IDEXX, Westbrook, ME. ¶¶ Stat-View, SAS Institute, Cary, NC. Table 1. Clinical Description of Subject Population Clinical Measurement Mean ± S.D. Range of Means by Subject (minimum-maximum) Plaque index 0.87 ± 0.59 0.26-1.93 Gingival index 0.83 ± 0.63 0.17-2.32 Probing depth (mm) 2.65 ± 0.55 1.34-4.38 Attachment loss (mm) 1.24 ± 0.81 0.23-4.66 givitis) were associated with lower levels of antibody. In the case of anti-Hsp90 and anti-GroEL anti- bodies, the concentrations were significantly (P <0.05) lower in the stratum 2 category compared to the 0 category. For the other stress proteins, while trends were apparent, the differences were not within the 95% confidence level. A similar pattern is seen when examining the relationship between mean probing depth and anti-stress protein antibody con- centration (Fig. 1A). The highest antibody concentrations were found in subjects in the 2 mm and 3 mm strata. Subjects who have mean probing depths greater than 3 mm (3 mm stratum) have significantly (P <0.05) reduced antibody concentrations for all stress proteins, except Hsp70. Figure 1C shows the same relationships for attachment loss. Again, the same trend is obvious and is significant (P <0.05) for Hsp90 and DnaK. We then examined the relationship between the presence of anti-stress protein antibodies and colo- nization of the subgingival plaque by specific peri- odontal disease-associated microorganisms (Fig. 2). In this figure, we show the relationship between the antibody concentrations and 3 levels of colonization as measured by the slot immunoblot assay. These levels (low, moderate, and high) correspond to <0.2 × 105, 0.2 to 3 × 105, and >3 × 105 microorganisms27 in the undiluted plaque sample assessed in the assay. As shown in Figure 2A, there is an association between colonization and anti-Hsp90 concentration. In all cases, there is a significantly (P <0.05) lower anti- Hsp90 concentration associated with the lowest level of colonization regardless of the bacterial species examined. In all cases, with the exception of A. actino- mycetemcomitans, the highest antibody concentration is associated with the highest level of colonization. There does not appear to be any significant associ- ation in the relationship between the anti-Hsp70 con- centrations and colonization. There is an apparent association with B. forsythus colonization; however, it is not statistically significant. This is also the case when the associations between anti-GroEL antibodies and colonization are examined (Fig. 2C). In the cases of A. actinomycetemcomitans and B. forsythus while the antibody concentrations appear to be lower at the highest level of colonization, they are not statistically significant. Finally, significant (P <0.05) relationships between the anti-DnaK antibody concentrations and coloniza- tion by P. gingivalis and T. denticola are evident in Figure 2D. Though not significant, A. actinomycetem- comitans reveals a trend with anti-DnaK antibodies similar to that observed for the other three antibodies. In all cases, the antibody concentration at the highest level of colonization is lower than the preceding con- centration. Similarly, B. forsythus reveals this pattern with anti-DnaK and anti-GroEL antibodies. Regression Modeling of Anti-Stress Protein Antibodies In order to begin to understand the variables that con- tribute to the anti-stress protein antibody concentra- tions, we used step-wise and multiple regression models which examined the contribution of age, gen- der, Amish/non-Amish status, gingival index, probing depth, attachment loss, calculus index, plaque index, and microbial colonization. In the case of the Hsp90 antibodies, 2 variables were found to have significant contributions to the antibody concentration having a R = 0.293 and P <0.0002. These variables were prob- ing depth (PD) and P. gingivalis concentration (PG). The equation derived from the regression model was y = 12558 − 2070*PD + 1842*PG. This confirmed the inverse relationship with probing depth and the posi- tive relationship with colonization by P. gingivalis. Attempts to model the other stress protein antibodies were not successful. DISCUSSION There has been a considerable focus on stress proteins in current infectious disease research due to the obser- vations that these protein families are immunodomi- nant antigens of many human pathogens.7 As a result we felt that it was timely to assess the association between immunity to common stress proteins and one of the most common microbial-mediated diseases in humans, periodontal disease. We selected 4 common, commercially available stress proteins, 2 human and Humoral Immunity to Stress Proteins Volume 70 • Number 10 1188 Table 2. Colonization Versus Mean Probing Depth Mean SIB Score (Mean ± Sem) Microorganism 2 mm* 3 mm* 4 mm* P† n = 75 n = 87 n = 36 P. gingivalis 0.66 ± 0.06 0.89 ± 0.05‡ 1.00 ± 0.21 <0.0004 A. actinomycetemcomitans 0.86 ± 0.04 0.95 ± 0.04 1.08 ± 0.19 NS B. forsythus 0.63 ± 0.05 0.90 ± 0.05‡ 0.89 ± 0.17 <0.0005 T. denticola 0.69 ± 0.05 0.94 ± 0.05‡ 1.15 ± 0.26§ <0.0005 * Stratification categories: 2 mm, mean probing depth <2.5 mm; 3 mm, mean probing depth ≥2.5 and <3.5; 4 mm, mean probing depth ≥3.5 mm. † ANOVA. ‡ Significant difference (P <0.005) between 2 mm and 3 mm categories. § Significant difference (P <0.01) between 2 mm and 4 mm categories. J Periodontol • October 1999 Lopatin, Shelburne, Van Poperin, Kowalski, Bagramian 1189 2 microbial in origin. Human Hsp70 and Hsp90 were used because they are amongst the most common of the stress protein families in eukaryotic systems.10 Since there is evidence in studies of Candida albicans that conser ved Hsp90 epitopes, shared between humans and Candida, were responsible for protection against systemic candidiasis, we felt that evaluating the association between anti-human Hsp antibodies and oral disease might be informative.31 We chose GroEL and DnaK (Hsp65 and Hsp70) since they rep- resented the most highly conserved stress proteins of microbial origin.10 The clinical specimens were derived from a patient population that was from an earlier study of an Amish community.22,23 Based on an evaluation of the oral health of this population and comparisons with non- Amish living in the same geographical location, we felt that, with the exception of a lack of routine oral hygiene, there were no significant differences between the Amish and the non-Amish in this area. While this is a randomly-selected population with few members pre- senting with advanced periodontitis, there is a clear relationship between disease severity (probing depth) and level of colonization by putative periodontal pathogens. Our studies also demonstrated a clear relationship between the serum concentration of 3 of the 4 specific anti-stress protein antibodies and clinical measures of periodontal disease. The serum antibodies (IgG) were highest when the subjects have the best oral health. This was most obvious when the mean probing depths are examined. It was interesting that throughout these analyses, while the antibodies reactive with human Hsp70 tended to follow the same trends, they showed the least statistically significant relationships. However, its microbial homologue, DnaK, demonstrated consid- Figure 1. Relationship between clinical measurements and serum anti-stress protein antibody concentration. Antibody concentration for each stress protein is expressed in relative fluorescence units (RFU) as determined by particle concentration fluorescence immunoassay (PCFIA). Error bars indicate standard error of the mean for the antibody concentration in each clinical category. Statistical significance of antibody titers (P ≤0.05): *significant differences between lowest and highest clinical stratum; †significant differences between middle and highest clinical stratum. A. Mean gingival index; subjects were stratified into 3 groups (0, 1, and 2; white, gray, and black bars, respectively) based on their mean whole mouth gingival index (0 = 0 to <0.5; 1 = 0.5 to <1.5; 2 = >1.5). Sample sizes for categories 0, 1, and 2 were 87, 66, and 45, respectively. B. Mean probing depth; subjects were stratified into 3 groups (2 mm, 3 mm, and 4 mm; white, gray, black bars, respectively) based on their mean whole mouth probing depths (2 mm, <2.5 mm; 3 mm, ≥2.5 and <3.5; 4 mm, ≥3.5 mm). Sample sizes for categories 2 mm, 3 mm, and 3 mm were 68, 72, and 58, respectively. C. Mean attachment loss; subjects were stratified into 3 groups (2 mm, 3 mm and 4 mm; white, gray, black bars, respectively) based on their mean whole mouth attachment losses (2 mm, <2.5 mm; 3 mm, ≥2.5 and <3.5 mm, ≥3.5 mm). Sample sizes for categories 2 mm, 3 mm and 4 mm were 86, 79, and 33, respectively. Humoral Immunity to Stress Proteins Volume 70 • Number 10 1190 erable disease-associated differences. This would sug- gest that even with the considerable homology within the Hsp70 family, the epitopes that are being recog- nized immunologically are probably those which are not evolutionarily conserved. In contrast, a very signif- icant inverse relationship between disease and human Hsp90 antibody concentration probably indicates that highly conserved epitopes are being recognized. While we did not possess purified microbial Hsp90 protein at the time of this study, investigations in progress in our laboratory demonstrate a high degree of cross-reactivity between human Hsp90 and the Hsp90 homologues of P. gingivalis and A. actinomycetemcomitans based on antibody reactivity (unpublished data). Similarly, serum IgG antibody concentrations reactive with GroEL followed similar trends. Mean gingival indices and attachment losses also followed these trends. Our observations are somewhat similar of a report by Schett et al.32 They examined anti-Hsp65 antibod- ies in sera (IgG) and found that the antibody concen- Figure 2. Relationship between serum anti-stress protein antibody concentration and colonization. Antibody concentration for each stress protein is expressed in relative fluorescence units (RFU) as determined by particle concentration fluorescence immunoassay (PCFIA). Error bars indicate standard error of the mean for the antibody concentration in each colonization category. Colonization by selected microorganisms is expressed as low (SIB score = 0), moderate (SIB score = 1) and high (SIB score = 2 + 3) (white, gray, and black bars, respectively). Statistical significance of antibody titers (P ≤0.05): *for significant differences between low and moderate colonization; †for significant differences between low and high colonization; and ‡for significant differences between moderate and high colonization. A) Anti-Hsp90 antibodies; B) Anti-Hsp70 antibodies; C) Anti-GroEL antibodies; D) Anti-DnaK antibodies. Sample sizes for low, medium, and high colonization categories by microorganism are as follows: P. gingivalis, 43, 44, 111 (L,M,H, respectively); A. actinomycetemcomitans, 51, 110, 37; B. forsythus, 91, 74, 33; and T. denticola, 83, 71, 44. J Periodontol • October 1999 Lopatin, Shelburne, Van Poperin, Kowalski, Bagramian 1191 trations were highest in periodontitis. This conflicts with our observations that in periodontitis the anti-Hsp anti- body levels are lowest. A possible explanation for this discrepancy is not immediately obvious; however, spe- cific differences between our studies exist. There is a considerable difference in sample size between the studies. In their study, a total of 23 subjects were divided into 3 groups. There are also differences between the antigens tested. We tested only E. coli GroEL and Schett et al.32 reported serum (IgG) data only for recombinant mycobacterial Hsp65. Finally, and perhaps the most important difference is that in our random population, we do not have the severe periodontitis that is observed in the Schett groups. And as a result, our most diseased group is considerably healthier than theirs and resembles their intermediate (gingivitis) disease. Our study also examined the association between colonization by specific microbial species and anti- stress protein antibodies. We found no significant associations between anti-GroEL or anti-Hsp70 con- centrations and colonization by the specific micro- organisms in our test panel. Again, we found this interesting since relationships between antibody con- centration and colonization were observed with anti- DnaK antibodies, an Hsp70 homologue, again indicating that, in this case, the non-conserved (anti- bacterial) epitopes are probably more important. However, we observed the most significant associa- tions between the anti-Hsp90 concentrations and microbial colonization. Since the antigen used to test for anti-Hsp90 antibodies is derived from human sources, the antibodies detected most likely represent those directed against conserved epitopes, rather than species-specific or anti-bacterial. In all cases, except for A. actinomycetemcomitans, there appeared to be a significant association between increased col- onization and anti-Hsp90 antibody concentration. Statistical modeling clearly demonstrated the inverse relationship with probing depth and the positive rela- tionship with colonization. The model very clearly defines the relationship between P. gingivalis and Hsp90. The most surprising observation is the inverse relationship between the antibody concentrations and oral disease. Most prior studies of humoral immunity to periodontal disease-associated microorganisms and oral health have shown a positive correlation.1 Thus, the present observations may actually be identifying the presence of protective antibodies, rather than sim- ply presence of the microorganism in the gingival sul- cus. If these are, in fact, protective antibodies, the next critical step will be to localize the antigens using these antibodies in order to help assess the role of the anti- gen in the virulence of the microorganism. It will be critical to determine if the stress proteins play a role in adherence and colonization, or if they interfere with normal mammalian cell function once the microorgan- ism has established its infection. Homologues of specific stress protein families have been demonstrated to be present in oral bacteria, including F. nucleatum, P. intermedia, P. nigrescens, P. melaninogenica, A. actinomycetemcomitans, and P. gingivalis. Hsp60/GroEL and Hsp70/DnaK homo- logues have been detected using cross-reactive anti- bodies against prototype members.16,19,20,33,34 Homologues of DnaK, GroEL, and groESL have been cloned and sequenced from P. gingivalis and A. actin- omycetemcomitans.17,18,20,33,35,36 In addition, the hsp90 gene homologue of A. actinomycetemcomitans was “accidentally” cloned by Winston et al.37 There have been numerous suggestions that the Hsp of pathogens may play a role in their viru- lence38,39 Studies by Matthews and Burnie31 demon- strate that immunity to Candida albicans Hsp90 is important for survival from systemic candidiasis. Other studies have shown that the normally non- pathogenic S. cerevisiae can be made pathogenic by causing an over-expression of the hsp90 gene.40 Mutations in an hsp gene (htrA) in Salmonella typhimurium result in reduced virulence. Finally, reports by Kirby et al. have shown that the potent bone-resorbing mediator of A. actinomycetemcomi- tans is homologous to GroEL.41 These findings sup- port the importance of understanding the role of stress proteins in microbial virulence and the function of the host response in blocking these effects. Our studies employing stress protein homologues reveal that there is considerable conservation of Hsp epitopes with periodontal disease-associated microor- ganisms. Clearly, since we were able to demonstrate associations using purified human Hsp, autoimmunity (as opposed to autoimmune disease) may serve an important function in host defense in periodontal dis- ease. This is not a novel observation and has been proposed by other investigators studying the role of host immunity to stress proteins.34 The relationship between anti-Hsp90 immunity, oral health, and colonization is reminiscent of the observa- tions by Matthews and Burnie31 in their studies of sys- temic candidiasis. In interpreting our findings, it should be clear that we are not using antigens derived from any of the periodontal disease-associated microorgan- isms in our panel. The serum antibodies and the associ- ations that we have described are likely to reflect the contributions of specific antigenic epitopes that have been conserved through evolution due to their impor- tance to the inherent function of the stress protein. We would expect that if we had employed stress proteins derived from our microbial panel, significantly different results would have been obtained. In that case, we would predict that the predominant serum antibody response would be against the species-specific epitopes and would probably reflect infection “history” rather than protective antibodies. Our laboratory is currently cloning the Hsp90 homologue of P. gingivalis and once complete we will be able to address issues related to homology between human and P. gingivalis Hsp90 and the epitopes that are involved in this apparent protec- tive response. Once sufficient quantities of purified or recombinant Hsps of periodontal disease-associated microorganisms are available we will be able to begin to address the importance of immunity to conserved versus microbe-specific epitopes. Since Matthews and Burnie31 demonstrated that protection could be attrib- uted to the recognition of a specific conserved epitope, it may be possible to establish a genetic basis for resis- tance to P. gingivalis (or other species) colonization based on the ability to recognize a specific epitope. Further elucidation of this model is dependent on evalu- ating the immunity to specific Hsp90 epitopes. ACKNOWLEDGMENTS The authors acknowledge the expert technical assis- tance provided by Mrs. Christine A. Binsfeld in the per- formance of the PCFIA analysis. Mr. Shelburne is a Research Specialist in the BioMaterials Technology Center of 3M. This study was supported by U.S. Public Service Grants DE 11117, DE08571, and DE06998. REFERENCES 1. Ebersole JL. Systemic humoral immune responses in periodontal disease. Crit Rev Oral Biol Med 1990;1: 283-331. 2. Boutsl EA, Koseki T, Nishihara T, Ishikawa I. Characterization of the immunodominant antigens of Porphyr omonas gingivalis 381 in high-responder patients. Oral Microbiol Immunol 1996;11:236-241. 3. Wilson M, Reddi K, Henderson B. Cytokine-inducing components of periodontopathogenic bacteria. J Periodont Res 1996;31:393-407. 4. Wilson M. Biological activities of lipopolysaccharides from oral bacteria and their relevance to the pathogenesis of chronic periodontitis. Sci Prog 1995;78:19-34. 5. Yoshimura F, Sugano T, Kawanami M, Kato H, Suzuki T. Detection of specific antibodies against fimbriae and membrane proteins from the oral anaerobe Bacteroides gingivalis in patients with periodontal diseases. Microbiol Immunol 1987;31:935-941. 6. Daly CG, Seymour GJ, Kieser JB. Bacterial endotoxin: a role in chronic inflammatory periodontal disease? J Oral Pathol 1980;9:1-15. 7. Shinnick TM. Heat shock proteins as antigens of bacterial and parasitic pathogens. Curr Top Microbiol Immunol 1991;167:145-160. 8. Craig EA, Gambill BD, Nelson RJ. Heat shock proteins: molecular chaperones of protein biogenesis. Microbiol Rev 1993;57:402-414. 9. Hendrick JP, Hartl FU. Molecular chaperone functions of heat-shock proteins. Ann Rev Biochem 1993;62: 349-384. 10. Lindquist S. The heat-shock response. Ann Rev Biochem 1986;55:1151-1191. 11. Polla BS. A role for heat shock proteins in inflammation? Immunol Today 1988;9:134-137. 12. Minota S, Cameron B, Welch WJ, Winfield JB. Autoantibodies to the constitutive 73-kD member of the hsp70 family of heat shock proteins in systemic lupus erythematosus. J Exp Med 1988;168:1475-1480. 13. Minota S, Koyasu S, Yahara I, Winfield J. Autoantibodies to the heat-shock protein hsp90 in systemic lupus erythematosus. J Clin Invest 1988;81:106-109. 14. Winfield J, Jarjour W, Minota S. Stress protein auto- antibodies and the expression of stress proteins on the surface of human gamma-delta cells and other cells of the immune system. Chem Immunol 1992;53:47-60. 15. Kleindienst R, Schett G, Amberger A, et al. Athero- sclerosis as an autoimmune condition. Isr J Med Sci 1995;31:596-599. 16. Lu B, McBride BC. Stress response of Porphyromonas gingivalis. Oral Microbiol Immunol 1994;9:166-173. 17. Maeda H, Miyamoto M, Hongyo H, Nagai A, Kurihara H, Murayama Y. Heat shock protein 60 (GroEL) from Porphyromonas gingivalis: molecular cloning and sequence analysis of its gene and purification of the recombinant protein. FEMS Microbiol Lett 1994;119: 129-135. 18. Maeda H, Miyamoto M, Hongyo H, Nagai A, Kurihara H, Mur yama Y. Heat shock protein 60 (GroEL) from Porphyromonas gingivalis: molecular cloning and sequence analysis of its gene and purification of the recombinant protein. FEMS Microbiol Lett 1994:15; 124:121-112. 19. Vayssier C, Mayrand D, Grenier D. Detection of stress proteins in Porphyromonas gingivalis and other oral bacteria by western immunoblotting analysis. FEMS Microbiol Lett 1994;121:303-307. 20. Nakano Y, Inai Y, Yamashita Y, et al. Molecular and immunological characterization of a 64-kDa protein of Actinobacillus actinomycetemcomitans. Oral Microbiol Immunol 1995;10:151-159. 21. Lokensgard I, Bakken V, Schenck K. Heat shock response in Actinobacillus actinomycetemcomitans. FEMS Immunol Med Microbiol 1994;8:321-328. 22. Bagramian RA, Farghaly MM, Lopatin D, Sowers M, Syed SA, Pomerville JL. A comparison of periodontal disease among rural Amish and non-Amish adults. J Clin Periodontol 1994;21:386-390. 23. Bagramian RA, Farghaly MM, Lopatin D, Sowers M, Syed, SA, Pomerville JL. Periodontal disease in an Amish population. J Clin Periodontol 1993;20:269-272. 24. Silness J, Löe H. Periodontal disease in pregnancy. (II) Correlation between oral hygiene and periodontal condition. Acta Odontol Scand 1964;22:121-135. 25. Löe H, Silness J. Periodontal disease in pregnancy. (I) Prevalence and severity. Acta Odontol Scand 1963;21: 533-551. 26. Ramfjord SP. The periodontal disease index (PDI). J Periodontol 1967;38(Suppl.):602-610. 27. Van Poperin N, Lopatin DE. Slot immunoblot assay for detection and quantitation of periodontal disease- associated microorganisms in dental plaque. J Clin Microbiol 1991;29:2554-2558. 28. Loesche WJ, Lopatin DE, Giordano J, Alcoforado G, Hujoel P. Comparison of the benzoyl-DL-arginine- naphthylamide (BANA) test, DNA probes, and immunological reagents for ability to detect anaerobic periodontal infections due to Porphyromonas gingivalis, Treponema denticola, and Bacteroides forsythus. J Clin Microbiol 1992;30:427-433. 29. Loesche WJ, Lopatin DE, Stoll J, Van Poperin N, Hujoel PP. Comparison of various detection methods for Humoral Immunity to Stress Proteins Volume 70 • Number 10 1192 J Periodontol • October 1999 Lopatin, Shelburne, Van Poperin, Kowalski, Bagramian 1193 periodontopathic bacteria: can culture be considered the primary reference standard? J Clin Microbiol 1992;30: 418-426. 30. Shelburne CE, Sandberg GP, Binsfeld CA, Wolff LF, Curry RA. Monoclonal antibodies to lipopolysaccharide of four oral bacteria associated with periodontal disease. J Periodont Res 1993;28:1-9. 31. Matthews R, Burnie J. The role of hsp90 in fungal infection. Immunol Today 1992;13:345-348. 32. Schett G, Metzler B, Kleindienst R, et al. Salivary anti- hsp65 antibodies as a diagnostic marker for gingivitis and a possible link to atherosclerosis. Int Arch Allergy Immun 1997;114:246-250. 33. Koga T, Kusuzaki T, Asakawa H, Senpuku H, Nishihara T, Noguchi T. The 64-kilodalton GroEL-like protein of Actinobacillus actinomycetemcomitans. J Periodontal Res 1993;28:475-477. 34. Ando T, Kato T, Ishihara K, Ogiuchi H, Okuda K. Heat shock proteins in the human periodontal disease process. Microbiol Immunol 1995;39:321-327. 35. Hotokezaka H, Ohara N, Hayashida H, et al. Trans- criptional analysis of the groESL operon from Porphy- romonas gingivalis. Oral Microbiol Immunol 1997;12: 236-239. 36. Yoshida A, Nakano Y, Yamashita Y, Yu H, Ohishi M, Koga T. Isolation and characterization of the dnaKJ operon from Actinobacillus actinomycetemcomitans. DNA Sequence 1997;8:93-98. 37. Winston JL, Toth SI, Roe BA, Dyer DW. Cloning and characterization of the Actinobacillus actinomyce- temcomitans gene encoding a heat-shock protein 90 homologue. Gene 1996;179:199-204. 38. Lathigra RB, Butcher PD, Garb, TR, Young DB. Heat shock proteins as virulence factors of pathogens. Curr Top Microbiol Immunol 1991;167:125-143. 39. Fer nandez RC, Logan SM, Lee, SH, Hof fman PS. Elevated levels of Legionella pneumophila stress protein Hsp60 early in infection of human monocytes and L929 cells correlate with virulence. Infect Immun 1996;64: 1968-1976. 40. Hodgetts S, Matthews R, Morrissey G, Mitsutake K, Piper P, Bur nie J. Over-expression of Sacchar omyces cerevisiae hsp90 enhances the virulence of this yeast in mice. FEMS Immunol Med Microbiol 1996;16:229-234. 41. Kirby AC, Meghji S, Nair SP, et al. The potent bone- resorbing mediator of Actinobacillus actinomyce- temcomitans is homologous to the molecular chaperone GroEL. J Clin Invest 1995;96:1185-1194. Send reprint requests to: Dr. Dennis E. Lopatin, Department of Biologic and Materials Sciences, School of Dentistry, University of Michigan, 1011 North University Ave., Ann Arbor, MI 48109-1078. Fax: 734/764-2425. Accepted for publication February 8, 1999.