Genome-wide identification of Chlamydia trachomatis antigens associated with tubal factor infertility.

Twenty-five to 35% of patients presenting for infertility evaluation have tubal disease ( 1 - 4 ). Chlamydia trachomatis is the primary sexually transmitted infection responsible for tubal factor infertility (TFI) ( 5 - 7 ) with C. trachomatis antibodies in approximately 70% of patients ( 8 ). C. trachomatis infected cells produce inflammatory cytokines ( 9 , 10 ) which may contribute to upper genital tract inflammatory damage ( 11 - 13 ). Lunefeld, et al found that among patients undergoing in vitro fertilization, those with C. trachomatis antibodies had decreased pregnancy rates ( 14 ). C. trachomatis infection is often asymptomatic so patient history cannot dictate the presence of tubal disease ( 15 , 16 ). Searching for biomarkers to predict chlamydial infection-associated tubal infertility is under intensive investigation. Elevated titers of anti- C. trachomatis antibodies are associated with TFI, but detection of overall antibody levels lacks the sensitivity and specificity required for differential diagnosis ( 17 ). Measuring anti- C. trachomatis antibodies at the single antigen level may offer increased sensitivity and specificity for predicting TFI. Elevated anti-chlamydial heat shock protein 60 (HSP60, CT110) antibodies are associated with TFI ( 18 - 27 ). Anti-HSP60 antibodies are associated with decreased pregnancy rates in patients with an ectopic pregnancy history ( 17 ). When HSP60 antibodies are in follicular fluid, there are decreased implantation rates ( 28 , 29 ). Some have postulated that chlamydial HSP60 incites a strong inflammatory response that may cross-react with the highly conserved human HSP60 ( 25 , 30 , 31 ). HSP60 may induce T-cell responses that contribute to the tubal damage ( 32 , 33 ). Regardless of how HSP60 or anti-HSP60 antibodies can mechanistically contribute to tubal disease, the specificity of anti-HSP60 antibody as a predictor for TFI has demonstrated that it may be used to differentially diagnose TFI. We have developed a C. trachomatis whole-genome scale protein array that can profile antigen specificities of anti- Chlamydia trachomatis antibodies ( 34 ). We hypothesized that C. trachomatis -infected women with TFI or controls (IFC) may develop antibodies recognizing unique sets of antigens.

Sera from TFI or IFC were titrated using HeLa cells infected with either C. trachomatis or C. pneumoniae organisms as antigens in an IFA. TFI Patients developed high titers of antibodies to C. trachomatis (p<0.001) but not C. pneumoniae (p=0.269) ( Table 1 ). When the patients were categorized based on levels of anti-chlamydial antibodies, most TFI patients developed high titers of anti- C. trachomatis antibodies (61.3%) while most IFC patients displayed lower titers (82.6%; p<0.001). It is difficult to use the quantitative difference in overall anti- C. trachomatis antibodies to diagnose TFI. To identify antigens that are recognized by TFI patients, Anti- C. trachomatis antibodies in 24 TFI and 11 IFC patients were mapped at the genome-wide scale since these patients displayed an overall anti- C. trachomatis antibody titer above 1:1000. As shown in Fig.1 , these 35 sera recognized C. trachomatis antigens distributed across the genome (panel A) with 265 antigens recognized by at least one antiserum and 643 antigens not detected by any sera (B). Many C. trachomatis antigens are recognized by both groups of patients, but there are antigens preferentially recognized by either group as highlighted with color-coded brackets (C). Thirty antigens were significantly recognized by TFI based on either mean OD values (Student’s t -test) or recognition frequency (Fisher’s exact test). Reactivity was confirmed using absorption against either HeLa alone or C. trachomatis -infected HeLa lysates as described previously ( 36 ). To identify antigens that can be used to predict TFI in infertility clinics, we identified antigens that were uniquely recognized by TFI patients. The 30 antigens preferentially recognized by TFI patients (as described above) were reacted with sera from all 54 patients (including 31 TFI and 23 IFC) regardless of their overall anti- C. trachomatis antibody titers. As listed in Figure 2 , the HSP60 (CT110) reacted with 22 of the 31 TFI and 4 of 23 IFC sera with a specificity and sensitivity of 82.6% and 71.0% respectively in predicting TFI. To further increase specificity, a 5-fold dilution (final dilution of 1:4000) was used to reduce the false-positive rate ( Figure 2 ). At this dilution, 10 of the 30 antigens, including CT110, CT322, CT376, CT381, CT414, CT443, CT681, CT795, CT798 & CT813, failed to react with any sera from the IFC group. Thus, these 10 antigens were uniquely recognized by TFI patients with a detection specificity of 100% (panel A). Dilution of the samples decreased detection sensitivity. HSP60 (CT110) only reacted with 11 out of 31 TFI sera, dropping the detection sensitivity to 35.5% along with three immunodominant antigens (CT795, CT798, CT813). CT443 reacted with 18 of the 31 TFI sera, maintaining a sensitivity of 58.1%. When the reactivity of the 10 antigens was analyzed at each individual antiserum level (Panel B), we found that the 10 antigens all together reacted with 21 independent sera of 31 total TFI sera, with a sensitivity of 67.7%. More importantly, this sensitivity can be maintained using fewer antigens. Combining HSP60 with CT376, CT381 & CT798 (total 4 antigens) or CT443 with CT381 (only 2 antigens) maintains sensitivity of 67.7%. As for other immunodominant antigens with a detection specificity of <100%, their detection sensitivity can be 80% even after serum dilution. Due to their ability to generating false positive results by reacting with IFC samples, it is clinically undesirable to use these antigens for screening for TFI.

C. trachomatis organisms cause pathologies in the fallopian tubes, leading to complications such as ectopic pregnancy and infertility. Since infertility can be caused by many different factors, distinguishing tubal infertility from other causes is required for guiding treatment plans. There is an ongoing debate on whether C. trachomatis -specific serology can aid in differential diagnosis of TFI. The goal of the current study is to test whether we can identify C. trachomatis antigens that can improve specificity and sensitivity in detecting TFI. The finding that anti- C. trachomatis but not anti- C. pneumoniae antibodies are highly associated with TFI is consistent with a well-established concept in the literature ( 42 - 44 ). Efforts have been made to develop individual C. trachomatis antigen-based detection methods. Previous reports demonstrate that anti-HSP60 antibodies are detected in 70-80% of TFI patients ( 21 , 22 , 45 , 46 ). Our genome-wide search for additional markers of TFI not only confirmed these findings but also revealed new information for further increasing specificity and sensitivity in detecting TFI. 30 antigens were preferentially recognized by TFI patients. At 1:800 dilution, HSP60 reacted with 22 of the 31 TFI sera (71% sensitivity) and 4 of the 23 IFC sera (82.6% specificity), which is consistent with previous findings and suggests our ELISA is comparable to other assays. When the sera were diluted to 1:4000 in order to further increase specificity, 10 of the 30 antigens achieved 100% specificity. Although the sensitivity decreased, careful examination revealed that a combination of 2 antigens [CT381 and CT443 (Outer membrane complex B, OmcB)], or 4 antigens [CT110 (HSP60), CT376, CT381 & CT798] detected TFI with a specificity and sensitivity of 100% and 67.7% respectively ( Fig. 2 ). We conclude that these combinations of antigens improved the C. trachomatis serology approach for diagnosing tubal infertility over using HSP60 alone (35.5% sensitivity), which represents a clinically significant improvement. Hysterosalpingogram (HSG) has a detection specificity and sensitivity of 83% and 65% respectively for detecting tubal pathology ( 47 ). C. trachomatis antigen-based serology diagnosis has numerous advantages over HSG besides improved detection, including sparing patients from the discomfort, radiation, and potential for infectious sequellae. This conclusion is consistent with previous reports that elevated chlamydial antibody levels are comparable to HSG ( 48 ) in diagnosing TFI and that HSG does not add to the medical knowledge on whether C. trachomatis infection contribute to tubal pathology ( 49 ). It is unknown whether these antigens themselves or antigen-specific immune responses contribute to the inflammatory pathologies in the fallopian tubes. The protein CT443, or OmcB, displayed the highest rate of reactivity with TFI patient sera. OmcB is a highly conserved immunodominant antigen, but the precise location of OmcB in the organisms and its role during infection is poorly understood ( 34 , 50 - 52 ). Despite the overwhelming evidence of C. trachomatis infection association with TFI, not every patient is infected with C. trachomatis or developed immune responses to C. trachomatis . Interventions such as early antibiotic therapy may cause a negative or low titers in patients, but it is unlikely that tubal pathology would be attributed to the C. trachomatis infection in these patients. Tubal pathology in TFI patients without positive C. trachomatis titers might be caused by other sources of infection such as Neisseria gonorrhoeae ( 53 , 54 ) and Mycoplasma genitalium ( 55 , 56 ). Thus, to further increase the sensitivity in diagnosing TFI, other infection causes should also be taken into account. The current study has presented a promising approach for globally identifying C. trachomatis -specific serological markers for diagnosis of TFI, Although statistically, the identified antigens or their combinations seem to provide definitive diagnosis of TFI, large sample sizes of infertile women from more diverse sources must be analyzed with a genome-wide scale proteome assay to maximize the potential of the C. trachomatis antigen-specific serology-based diagnosis of TFI. Efforts are underway to expand our studies to China where we hope to enroll several thousand patients.

31 TFI and 23 IFC patients were enrolled at the University of Texas Health Science Center at San Antonio following Institutional Review Board approval. All women were at least 21 years old and underwent diagnostic laparoscopy with chromotubation as part of their infertility evaluation. Diagnosis of tubal infertility was defined as fallopian pathology consistent with hydrosalpinx, fimbrial phimosis, or peri-tubal adhesions. Exclusion criteria included prior tubal ligation, surgical finding of endometriosis, or a history of pelvic infection or inflammation other than pelvic inflammatory disease such as appendicitis. IFC patients had normal pelvic findings and tubal patency at laparoscopy. After the blood draw, serum samples were stored at −20°C until analyzed. As previously described, HeLa cells (American Type Culture Collection, Manassas VA 20108) were cultured in DMEM (GIBCO PRL, Rockville, MD) with 10% fetal calf serum (FCS; GIBCO BRL) at 37°C with 5% Carbon dioxide (CO 2 ) ( 34 - 36 ). C. trachomatis serovar D or Chlamydia pneumoniae AR39 organisms were grown, purified and titrated as previously described ( 36 - 38 ). For immunofluorescence assay, chlamydial organisms were used to infect HeLa cells grown on glass coverslips in 24-well plates. The sub-confluent HeLa cells were treated with DMEM containing 30 μg/ml of DEAE-Dextran (Sigma, St. Louis, MO) for 10 minutes at 37°C. After removal of DEAE-Dextan solution, chlamydial organisms were added to the wells for 2 hours at 37°C. The infected cells were continuously cultured in DMEM with 10% FCS and 2μg/ml of cycloheximide (Sigma, St. Louis, MO). Anti-chlamydial organism antibodies in human sera were titrated using an Immunofluorescence assay (IFA) as previously described ( 34 , 36 , 39 , 40 ). Briefly, HeLa cells grown on coverslips were infected with C. trachomatis or C. pneumoniae organisms, fixed 48h post-infection for C. trachomatis and 72h for C. pneumoniae with 2% paraformaldehyde, and permeabilized with 2% saponin at room temperature for 1 hour. After blocking, human sera were added to the Chlamydia-infected cell samples. The primary Ab binding was visualized with a goat anti-human IgG conjugated with Cy3 (red; Jackson ImmunoResearch Laboratories, West Grove, PA), and DNA was labeled with Hoechst dye (blue; Sigma-Aldrich). The highest dilution of a serum that still gave a positive reactivity was defined as the titer of the given serum sample. Serum samples were serially diluted and the appropriate dilutions were repeated multiple times based on the results obtained from prior dilutions in order to obtain a more accurate titer for each serum. Images were acquired with an Olympus AX70 fluorescence microscope equipped with multiple filter sets (Olympus, Melville, NY) as previously described ( 36 , 40 ). Glutathione S -transferase (GST) fusion protein enzyme-linked immunosorbent assay (ELISA) for detecting human antibody recognition of chlamydial proteins was carried out as previously described ( 36 ). Bacterial lysates containing individual chlamydial GST fusion proteins were added to 96 well microplates pre-coated with glutathione (Pierce, Rockford, IL) at a 1:10 dilution in PBS with a total volume of 200 μl/well. Lysates containing GST alone, as negative, and GST-chlamydial protease-like activity factor (CPAF), as positive controls, were also included on each plate. The plates were incubated overnight at 4°C to allow GST fusion proteins to bind to the plate-immobilized glutathione then blocked with 2.5% milk in PBS and washing with PBST (PBS with 0.05% Tween 20; Sigma-Aldrich). The human sera was pre-absorbed with a bacterial lysates containing GST at 4°C overnight, then incubated with Glutathione beads (bioWorld, Dublin, OH) for 1 hour at room temperature to reduce background caused by non-specific human antibodies. The human antibody reactivity was detected with a goat anti-human-IgG, IgA & IgM conjugated with horse-radish peroxidase (HRP; Jackson ImmunoResearch Laboratories) plus the substrate 2,2′-azino-bi(2-ethylbenzothiazoline-6-sulforic acid) diammonium salt (ABTS; Sigma). The optical density (OD) was measured at 405nm using a microplate reader (Molecular Devices Corporation, Sunnyvale, CA). To confirm the antibody binding specificity, all sera were further absorbed with lysates made from either HeLa cells alone or C. trachomatis serovar D-infected HeLa cells prior to reacting with the fusion protein-coated plates. The absorption was carried out as following: HeLa cells with or without chlamydial infection were lysed via sonication at 2 × 10 7 cells per ml of PBS containing a cocktail of protease inhibitors. The pre-diluted serum samples were incubated with cell lysates overnight at 4°C prior to reacting with the plate-immobilized chlamydial fusion proteins. The antibody binding that remained positive after HeLa-alone lysate absorption but significantly reduced by Chlamydia-HeLa lysate absorption was considered true positive. Data were analyzed using SPSS v. 15.0 software (IBM, Chicago, IL) as previously described ( 36 , 39 ). Briefly, titer values were log-transformed to produce a normal distribution and analyses were performed on transformed values. Student’s t -Test was utilized to assess anti- C. trachomatis and anti- C. pneumoniae antibodies to evaluate overall mean differences between the 2 groups of patients. Because the a priori hypothesis was that the TFI group would have higher titers than the IFC group, a one-tailed analysis was used for the C. trachomatis data, but a two-tailed analysis was performed on the C. pneumonia because there was no a priori hypothesis. Because the antibody titers had large variations within a given group, the serum titers were evaluated by ranges of 1:10,000 (High). Chi-Squared and Fisher’s Exact Test were employed to compare overall antibodies to C. trachomatis and antibodies to C. pneumoniae . ELISA results were analyzed using Student’s t -Test and Fisher’s Exact Test as appropriate. For the genome-wide ELISA, both Student’s t -Test (for comparing quantitative OD value data) and Fisher’s Exact Test (for comparing the number of sera positively reacted with a given antigen) were preformed. Using both methods allows us to identify C. trachomatis antigens that are both clinically and statistically significant. When Student’s t -Test was utilized, the OD values after subtracting background from the same plate were used. When Fisher’s Exact Test was utilized, a response was determined positive when the OD value was equal to or greater than 2 standard deviations above the mean calculated from the same 96 well plate as described previously ( 39 , 41 ).