May 6, 1989 - anesthetized mice. Vaginal washes were obtained by flush- ing the vaginal vault with phosphate-buffered saline as described by Ramsey et al.
Vol. 57, No. 8
INFECTION AND IMMUNITY, Aug. 1989. p. 2441-2446
0019-9567/89/082441-06$02.00/0 Copyright © 1989. American Society for Microbiology
Humoral Immune Response to Chlamydial Genital Infection of Mice with the Agent of Mouse Pneumonitis KYLE H. RAMSEY,'* WILBERT J. NEWHALL V,2 AND ROGER G. RANK1 Department of Microbiology and Immunology, University of Arkansas for Medical Sciences, Little Rock, Arkansas 72205,l and Departmnent of Medi ine Indianta University Sc/hool of Medicine, Indianapolis, Indiana 462232 Received 4 April 1989/Accepted 6 May 1989
The objective of this study was to characterize the humoral immune response to chlamydial genital infection of mice with the mouse pneumonitis agent (MoPn). With an enzyme-linked immunoabsorbent assay, immunoglobulin G antibodies to MoPn were first detected in plasma by day 14. Peak plasma antibody concentrations were reached by day 49, and this reponse did not decline significantly throughout the 300-day monitoring period. Immunoglobulin A against MoPn could first be detected in pooled vaginal washes by day 21 after infection and had reached peak concentrations by day 28, but anti-MoPn immunoglobulin G was not consistently present in secretions. The antibody response in secretions had declined slightly by day 300. Immunoblot analysis revealed that the early phase of the plasma antibody response to MoPn as a result of genital infection was against lipopolysaccharide, the major outer membrane protein, and a 62-kilodalton (kDa) protein. In secretions, early-phase immunoglobulin A antibodies were directed to the major outer membrane protein and lipopolysaccharide. Late reactions to 15-, 22-, and 83-kDa proteins in plasma were noted. Late reactions to the 62-kDa protein in secretions were also noted. The cause of these late responses remains unexplained. When mice were challenged intravaginally with MoPn at 50-day intervals after the primary infection, it was found that mice inoculated on day 100 or after were susceptible to reinfection. Susceptibility could not be related to a decline in the antibody concentration in plasma or secretions or in the antibody response to specific components of MoPn as measured by immunoblot analysis.
A murine model for chlamydial genital infection with the mouse pneumonitis (MoPn), a Clalamndia trachomnatis biovar, has been previously described (2). It was found that both humoral and cell-mediated immune (CMI) responses are elicited as a result of this infection (1). Furthermore, the development of chronic infections in athymic nude mice established the importance of T cells in resolving the infection (22). However, it could not be determined whether the requirement for T cells in resolution of the infection reflected a necessity for functional CMI mechanisms or for T-cell help in the antibody response or both. Recently, we have reported that mice are capable of resolving MoPn genital infections in the absence of an antibody response and that resistance to reinfection is not significantly altered by the absence of antibody (16). Nonetheless, these findings did not rule out antibody as a contributing factor in protection from MoPn genital infection but merely emphasized that CMI alone could resolve a genital infection and provide a protective response. Additionally, studies of guinea pigs infected genitally with the agent of guinea pig inclusion conjunctivitis (GPIC) have clearly indicated an important role for humoral immunity in resolution of primary infection and resistance to reinfection (17, 19, 20, 24). Thus, the goal of this study was to characterize the kinetics of the antibody response in serum and genital secretions over an extended period of time and to define the antigen-specific antibody response at both sites. We also wanted to determine whether the level of antibody or the presence of antibody to certain chlamydial outer membrane components could be associated with either the resolution of a primary infection or resistance to reinfection.
MATERIALS AND METHODS Animals. Female BALB/c mice were obtained either from SASCO, St. Louis, Mo., or from Harlan Sprague-Dawley, Indianapolis, Ind. Animals were housed in cages covered with fiberglass filter bonnets and were supplied with food and water ad libitum in an environmentally controlled room at 24°C with a cycle of 12 h of light and 12 h of darkness. Infection of mice. Mice were inoculated intravaginally with 0.03 ml of a suspension of McCoy cell-grown MoPn in sucrose phosphate glutamate buffer (pH 7.2) while under sodium pentobarbitol anesthesia (16). Mice were 7 to 8 weeks of age at the time of infection. Assessment of infection. Infection was assessed by isolation of MoPn from cervical-vaginal swabs in McCoy cells and subsequent quantitation of inclusion-forming units (IFUs) by using techniques described previously (16). Quantitation of the antibody response. Plasma was obtained by puncture of the retro-orbital venous plexus of anesthetized mice. Vaginal washes were obtained by flushing the vaginal vault with phosphate-buffered saline as described by Ramsey et al. (16). Antibody responses were quantitated in plasma and vaginal washes by using an enzyme-linked immunoabsorbent assay (7, 16) that used peroxidase-conjugated goat anti-mouse immunoglobulin G (IgG) or IgA (Fc specific; Organon Teknika, Malvern, Pa.) for the detection of antibody in plasma and secretions, respectively. SDS-PAGE and electrophoretic transfer. Chlamydial proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (10). Some gels were stained with Coomassie brilliant blue, destained, and then photographed. Molecular weights of chlamydial proteins were estimated by comparison with a plot of migration distance versus the log of the molecular weight of known standards. For immunoblotting purposes, approximately 325 pg of HeLa cell-
agent of
*
Corresponding author. 2441
2442
INFECT. IMMUN.
RAMSEY ET AL.
grown, gradient-purified, MoPn elementary bodies were solubilized and then resolved on a slab gel (140 by 1.5 by 130 mm) consisting of a 20-mm 3% acrylamide stacking gel and a 110-mm 12.5% acrylamide separating gel. Resolved MoPn proteins were electrophoretically transferred to a nitrocellulose membrane (NCM) by using a modification of the procedure described by Towbin et al. (26). The transfer was conducted in a Tris-glycine tank buffer (25 mM Tris-192 mM glycine) containing 20% (vol/vol) methanol. Electrophoresis was at 60 V for 4 h at 4°C. Sarkosyl extraction. Outer membranes from purified MoPn elementary bodies were derived by extraction with Nlauroyl sarcosine (Sarkosyl; Sigma Chemical Co., St. Louis, Mo.) (3). Immunoblotting. After electrophoretic transfer, the NCM was incubated for 1 h at room temperature in Tris buffer (10 mM Tris-145 mM NaCl, pH 7.3) containing 10% horse serum (TB-10) to block excess protein-binding sites. The NCM was then sliced vertically into identical 5-cm strips and frozen in test tubes at -70°C until needed. Plasma samples (diluted 1:100 [vol/vol] in TB-10) and vaginal washes (diluted 1:25 [vol/vol] in TB-10) were incubated on NCM strips for 2 h at room temperature with constant gentle agitation. The TB-10 and unbound antibody was then aspirated from the tubes, and the NCM strips were washed three times in 155 mM NaCl (3 min per wash). To detect bound antibody, NCM strips were then incubated with peroxidase-conjugated goat anti-mouse IgG (plasma) or IgA (vaginal washes) (Fc specific; Organon Teknika) diluted 1:1,000 (vol/vol) and 1:500 (vol/vol), respectively, for 1 h. This was followed by repetition of the wash step described above and visualization of the bound conjugated antibody by using a 4-chloro-1-naphthol developing solution (12). Experimental design. In the first experiment, mice were inoculated with a suspension of MoPn containing 9.8 x 106 IFU, and cervical vaginal swabs were taken at 5-day intervals thereafter. When the experiment was repeated, the animals were inoculated with 2.7 x 106 IFU, and to confirm infection, genital swabs were collected 5 days postinoculation. Animals not becoming infected by 5 days postinoculation were excluded from the experiment. In both experiments, parallel groups of five mice each were infected simultaneously. One group was reserved for monitoring of both the primary infection and the primary antibody response to the infection by collection of genital swabs, plasma, and vaginal washes at the specified intervals. The remaining groups were challenged intravaginally at 50-day intervals post-primary inoculation using either one inoculation (experiment 1) or inoculation on 2 consecutive days (experiment 2). Genital swabs were collected at 3-day intervals postchallenge. Plasma and vaginal washes were collected at 7-day intervals postchallenge. RESULTS Infection kinetics. When mice were inoculated intravaginally with MoPn, they acquired an infection that peaked 5 to 10 days postinoculation (Fig. 1). The infection declined thereafter, with subsequent isolation attempts yielding low numbers of IFUs recovered from three of five animals at day 15. All animals became isolation negative by day 20. Repeated attempts to isolate MoPn from genital swabs after 20 days postinoculation were unsuccessful. Humoral immune response kinetics. After primary genital infection with MoPn, IgG antibody responses were first detected in plasma by day 14 postinoculation and peaked by
U)
z 105.
A
4 z 1
7 io3 -J
10
10
0
z
0
5
15
10
20
25
DAYS FIG. 1. Course of MoPn genital infection. Each point represents the total number of IFUs isolated from an individual animal at 5-day intervals after inoculation.
day 49 (Fig. 2). This response remained constant throughout the monitoring period. When IgA levels were measured locally in the genital tract, it was found that antibodies directed against MoPn could first be detected in vaginal secretions 21 days after infection (Fig. 2). This response peaked soon thereafter on day 28 and remained near peak levels throughout most of the monitoring period, declining only slightly by day 120. Attempts to detect IgG in pooled vaginal washes proved that this class of antibody was present inconsistently or at low levels throughout the monitoring period (data not shown). Analysis of the outer membrane protein structure of MoPn. The protein profile of MoPn elementary bodies was very similar to those of three human serovars of C. trachomatis (Fig. 3A). The major outer membrane protein (MOMP) had an estimated molecular mass of 44 kilodaltons (kDa). MoPn was found to lack a protein having a mass of ca. 60 kDa but had a relatively high amount of a ca. 62-kDa protein. Other predominant proteins include 72-, 46-, 29-, and 15-kDa proteins. The latter two proteins are analogous to the putative adhesins described by others (5, 27), on the basis of their reactivity with a monoclonal antibody specific for those proteins (W. J. Newhall V, unpublished observation). The outer membrane protein profile was obtained after extraction of elementary bodies with Sarkosyl and analysis 10
1031
0
50
100
150
200
T
T
250
300
DAYS FIG. 2. Antibody titers in plasma and vaginal secretions. Shown IgG titers (-) and IgA titers in pooled vaginal washes of five animals (0).
are mean plasma
VOL. 57, 1989
IMMUNE RESPONSE TO CHLAMYDIAL INFECTION OF MICE
2443
B
A M
0
L2
0
A
2
N
3 MP
*
--
_
68K-
45K-
r
-_-60K -
_
_
72 '6 OK 46K MOMP
]
_
_
_72K - 62K
--46K -MOMP
c22K 10
-MOMP
0 z
0
D10
U)
:3 1O0 0 z 3
6
9
12
DAYS -15 K
-LPS 50 100 150 200
0
FIG. 6. Course of MoPn genital infection resulting from a challenge inoculation 150 days after the primary infection. Each point represents the total number of IFUs isolated from an individual animal on each sample date. Also shown is a plot of the mean of the primary infection from Fig. 1 (0).
challenge
similar to that of animals in the first experiment in length and intensity.
group was
FIG. 5. Immunoblot analysis of IgA response in vaginal secretions. Samples were taken from a pool of five animals. Postchallenge sample was obtained after a challenge infection on day 300.
DISCUSSION vaginal challenge. Additionally, IgA antibodies specific for the 15-kDa protein were detected in vaginal secretions after challenge infection, once again paralleling the plasma IgG response.
Results of challenge infection. It has been previously reported that mice infected genitally with MoPn display solid immunity to reinfection for a period of at least 60 days after the primary infection (1). The data shown in Table 1 support these observations. In the first experiment, mice were resistant to a second infection at 50 and 100 days after the primary infection (approximately 30 and 80 days after resolution of the primary infection). However, intravaginal inoculation of MoPn 150 days after the primary infection resulted in infection of five of five mice. The secondary infection was abbreviated in length and intensity (Fig. 6) compared with the primary one. The results of the second experiment indicate that two of four mice challenged on day 100 and one of five mice challenged on day 150 became isolation positive. However, it should be noted that only low numbers of IFUs were recovered 3 days after challenge in each of the isolationpositive animals. Isolation attempts at 6 and 9 days postchallenge were negative for each of these animals. When animals were challenged 200 days after the primary infection, four of five were isolation positive. The second infection in this
In this study, we have characterized the humoral immune MoPn genital infections. By enzyme-linked immunosorbent assay, we found that IgG antibodies first appeared in plasma on day 14 and that IgA antibodies could first be detected in vaginal washes by day 21. Attempts to detect IgG in vaginal washes have proven that this class of antibody appears in secretions inconsistently and at low levels throughout the monitoring period (data not shown). Since antibody appeared in plasma and secretions 14 to 21 days after inoculation and since this is a time period correlating roughly to the time of resolution of the infection, one might suspect that antibody affects recovery from the infection. However, we have recently shown that mice are capable of resolving MoPn genital infections in the absence of an antibody response (16). Therefore, the appearance of antibody during this period probably indicates that antibody has at best a participatory role but is not essential for the resolution of infection in this model. A significant aspect of this study is the finding that mice become susceptible to reinfection. This would imply that some protective facet of the immune response wanes sufficiently to allow the establishment of a second infection. It has also been suggested that humans develop some protective immunity after a chlamydial genital infection (25) and that this protection may be short-lived (9). Similarly, in the response to
TABLE 1. Results of intravaginal challenge with MoPn No. of animals infected/no. challenged (challenge dose per mouse [IFUs]") on postinfection day:
Expt 50
1 2 "
0/5 (1.8 x 107) 0/5 (2.7 x 106 and 6.6 x 105)
100
150
0/5 (3.3 x 107)
5/5 (6.9 x 107)
2/4 (1.2 x 106 and 9.1 x 105)
1/5 (2.4 x 106 and 7.1 x 105)
For experiment 2, mice were inoculated on 2 consecutive days with the respective doses indicated.
bND, Not done.
200
250
3/4 (4.7 x 107) 4/5 (8.4 x 106 and 3.0 x 105)
5/5 (3.4 x 107)
NDb
VOL. 57, 1989
IMMUNE RESPONSE TO CHLAMYDIAL INFECTION OF MICE
marmoset model of chlamydial genital infection it was found that animals become reinfected after a primary infection but that the second infection lasts for a shorter time (8). Also similar to the mouse, guinea pigs are solidly immune to challenge infections shortly after recovery, but this immune response eventually subsides to allow a second, abbreviated infection (4, 21). The present data indicate that susceptibility to challenge infection did not correlate well with decreases in antibody titer in plasma or secretions, nor could it be associated with a decline in the antibody response to a specific MoPn antigen. Thus, one might suspect that CMI mechanisms are responsible for protection against a challenge infection in this model. Unfortunately, we have not yet determined whether an association exists between susceptibility to reinfection and a decrease in CMI. However, previous results indicated that mice are resistant to a second infection in the absence of antibody response (16). Some of our current results differ from those obtained in the guinea pig-GPIC model. In that model, both humoral and CMI responses are essential elements in resolution of the infection (18, 24) as well as in resistance to reinfection (19, 20, 23). Similar to the mouse-MoPn model, reports on the guinea pig-GPIC model of chlamydial genital infection describe antibody reactions to the 61-kDa protein (analogous to the 62-kDa protein of MoPn), MOMP, and LPS early in the response (4). Unlike the mouse system, genital infections of guinea pigs also elicit antibody responses to 84-, 72-, 47-, 33-, 27-, 19-, and 15-kDa proteins early in the response to the infection. Also unlike the mouse model, the response to each of these chlamydial components declines over time in the guinea pig. Surprisingly, in the murine model, the peak response was seen much later (91 days for the murine model compared with 28 days for the guinea pig model), and not only did the response to analogous components not decrease, it actually increased with time. Furthermore, among eight animals this appearance of new specificities was amazingly consistent. Additionally, when guinea pigs were given a second infection they developed more intensified reactions to each of the aforementioned components, whereas mice did not develop such an apparent anamnestic response in plasma, as measured either by immunoblotting or by enzyme-linked immunosorbent assay (data not shown). However, the reactions to the 62-kDa protein, MOMP, the 17-kDa protein, and LPS did intensify in secretions after challenge in the murine model. Several explanations could be offered to account for the differences observed between the two models. The most plausible explanation would take into account that the immune mechanisms responsible for resolution of the infection and resistance to reinfection may differ between the two species of chlamydiae or between the host species. Similar long-term studies on the antibody response in humans are not currently available. However, immunoblot analysis after the infection in mice indicates that the antibody reaction pattern is similar to that described for humans by Newhall and co-workers (15). After the infection in humans, antibodies may be produced to a 15-kDa protein, MOMP, a circa 60-kDa protein, and other proteins that migrate on sodium dodecyl sulfate-polyacrylamide gel electrophoresis in a pattern similar to those of MoPn. An interesting difference is the reaction with the comigrating 62-kDa proteins in mice. While antibody responses in humans are directed against both the 60- and 62-kDa proteins of human strains, our evidence suggests that MoPn-infected mice produce antibody primarily against the 62-kDa cys-
2445
teine-rich outer membrane protein of MoPn (analogous to the 60-kDa protein of human strains). The apparent lack of reactivity with the Sarkosyl-soluble 62-kDa protein (analogous to the 62-kDa protein of human strains) remains unex-
plained. It is interesting that as late as 91 days after the primary infection (corresponding to 70 to 75 days after resolution of the primary infection) immunoblot analysis revealed several new antibody specificities in the plasma and secretions of eight of eight animals tested. The prolonged antibody response in genital secretions was also unexpected. One explanation for this late response is that a population of suppressor cells may be present early in the response to this infection and that this population of cells is either reduced or their functional activity is decreased with time. The appear-
of suppressor cells in response to ocular chlamydial infections has been reported elsewhere (28), albeit after multiple inoculations. Another explanation is that the appearance of new antibody specificities late in the response and the prolonged antibody response in secretions are due to a latent infection. Hence, as the protective immune response wanes, the reshedding of MoPn after the apparent resolution of the infection may lead to a boost in the antibody response. Evidence for this type of infection has been reported to occur in vitro (11, 13). However, in preliminary studies with this model, attempts to isolate MoPn from tissue homogenates from the spleen, genital tract, bone joints, and other organs after resolution of the infection have been unsuccessful (R. G. Rank, unpublished data). More detailed studies are needed to be fully conclusive on the possibility of latent infections in this model. The present observations combined with data gathered from previous studies indicate that the humoral immune response is of questionable value in providing resistance to MoPn genital infections. Thus, in this model, CMI may play the predominant role in resolution of the infection and resistance to reinfection. The results, therefore reinforce the murine model as a tool to study the mechanisms of CMI against chlamydial genital infection.
ance
ACKNOWLEDGMENTS This study was supported by Public Health Service grants A126328 and Al23044 from the National Institutes of Health and by the University of Arkansas for Medical Sciences Foundation Fund. LITERATURE CITED 1. Barron, A. L., R. G. Rank, and E. B. Moses. 1984. Immune response in mice infected in the genital tract with mouse pneumonitis agent (Chianydia trachoinatis biovar). Infect. Immun. 44:82-85. 2. Barron, A. L., H. J. White, R. G. Rank, B. L. Soloff, and E. B. Moses. 1981. A new animal model for the study of Chlamydia trachomatis genital infections: infection of mice with the agent of mouse pneumonitis. J. Infect. Dis. 143:63-66. 3. Batteiger, B. E., W. J. Newhall V, and R. B. Jones. 1985. Differences in outer membrane proteins of the lymphogranuloma venereum and trachoma biovars of Chulaindia tracholnatis. Infect. tmmun. 50:488-494. 4. Batteiger, B. E., and R. G. Rank. 1987. Analysis of the humoral immune response in chlamydial genital infection in guinea pigs. Infect. Immun. 55:1767-1773. 5. Hackstadt, T. 1986. Identification and properties of chlamydial polypeptides that bind eucaryotic cell surface components. J. Bacteriol. 165:13-20. 6. Hatch, T. P., M. Miceli, and J. E. Sublett. 1986. Synthesis of disulfide-bonded outer membrane proteins during the development of cycle of Chlainydia psittaci and Chlamnydia trachotna-
2446
RAMSEY ET AL.
tis. J. Bacteriol. 165:379-385. 7. Hough, A. J., Jr., and R. G. Rank. 1988. Induction of arthritis in C57B1/6 mice by chlamydial antigen: effect of prior immunization or infection. Am. J. Pathol. 130:163-172. 8. Johnson, A. P., M. F. Osborn, B. J. Thomas, C. M. Hetherington, and D. Taylor-Robinson. 1981. Immunity to reinfection of the genital tract of marmosets with Chlarnydia trachomnatis. Br. J. Exp. Pathol. 62:606-613. 9. Katz, B. P., B. E. Batteiger, and R. B. Jones. 1987. Effect of prior sexually transmitted disease on the isolation of Chlamydia trachomatis. Sex. Transm. Dis. 14:160-164. 10. Laemmli, U. K. 1970. Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature (London) 227:680-685. 11. Lee, C. K., and J. W. Moulder. 1981. Persistent infection of mouse fibroblasts (McCoy cells) with a trachoma strain of Chlamydia trachoinatis. Infect. lmmun. 32:822-829. 12. Moller, S. A., and C. A. K. Borrebaeck. 1985. A filter immunoplaque assay for the detection of antibody-secreting cells in vitro. J. Immunol. Methods 79:195-204. 13. Moulder, J. W., N. J. Levy, and L. P. Schulman. 1980. Persistent infection of mouse fibroblasts (L cells) with Chlainvdia psittaci: evidence for a cryptic chlamydial form. Infect. Immun.
30:874-883. 14. Newhall, W. J., V. 1987. Biosynthesis and disulfide cross-linking of outer membrane components during the growth cycle of Chlamydia trachomnatis. Infect. Immun. 55:162-168. 15. Newhall, W. J., V, B. Batteiger, and R. B. Jones. 1982. Analysis of the human serological response to proteins of Chlarnydia tracho,natis. Infect. Immun. 38:1181-1189. 16. Ramsey, K. H., L. S. F. Soderberg, and R. G. Rank. 1988. Resolution of chlamydial genital infection in B-cell-deficient mice and immunity to reinfection. Infect. Immun. 56:1320-1325. 17. Rank, R. G., and A. L. Barron. 1982. Prolonged genital infection by GPIC agent associated with immunosuppression following treatment with estradiol, p. 391-394. In P.-A. Mardh, K. K. Holmes, J. D. Oriel, P. Piot, and J. Schachter (ed.), Chlamydial
INFECT. IMMUN.
infections. Elsevier Biomedical Press, New York. 18. Rank, R. G., and A. L. Barron. 1983. Effect of antithymocyte serum on the course of chlamydial genital infection in female guinea pigs. Infect. Immun. 41:876-879. 19. Rank, R. G., and A. L. Barron. 1983. Humoral immune response in acquired immunity to chlamydial genital infection of female guinea pigs. Infect. Immun. 39:463-465. 20. Rank, R. G., and B. E. Batteiger. 1989. Protective role of serum antibody in immunity to chlamydial genital infection. Infect. Immun. 57:299-301. 21. Rank, R. G., B. E. Batteiger, and L. S. F. Soderberg. 1988. Susceptibility to reinfection after a primary chlamydial genital infection. Infect. Immun. 56:2243-2249. 22. Rank, R. G., L. S. F. Soderberg, and A. L. Barron. 1985. Chronic chlamydial genital infection in congenitally athymic nude mice. Infect. Immun. 48:847-849. 23. Rank, R. G., L. S. F. Soderberg, M. M. Sanders, and B. E. Batteiger. 1989. Role of cell-mediated immunity in the resolution of secondary chlamydial genital infection in guinea pigs infected with the agent of guinea pig inclusion conjunctivitis. Infect. Immun. 57:706-710. 24. Rank, R. G., H. J. White, and A. L. Barron. 1979. Humoral immunity in the resolution of genital infection in female guinea pigs infected with the agent of guinea pig inclusion conjunctivitis. Infect. Immun. 26:573-579. 25. Schachter, J., L. D. Cles, R. M. Ray, and F. E. Hesse. 1983. Is there immunity to chlamydial infections of the human genital tract'? Sex. Transm. Dis. 10:123-125. 26. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretic transfer of proteins from polysaccharide gels to nitrocellulose sheets: procedure and some applications. Proc. Natl. Acad. Sci. USA 76:4350-4354. 27. Wenman, W. M., and R. U. Meuser. 1986. Chlamydia trachoinatis elementary bodies possess proteins which bind to eucaryotic cell membranes. J. Bacteriol. 165:602-607. 28. Young, E., and H. R. Taylor. 1986. Immune mechanisms in chlamydial eye infection. Development of T suppressor cells. Invest. Ophthalmol. Vis. Sci. 27:615-619.
