Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Immunol. Author manuscript; available in PMC 2013 February 15.
Published in final edited form as:
PMCID: PMC3303601

Plac8-dependent and iNOS-dependent mechanisms clear Chlamydia muridarum infections from the genital tract1


Chlamydia trachomatis urogenital serovars replicate predominately in genital tract epithelium. This tissue tropism poses a unique challenge for host defense and vaccine development. Studies utilizing the Chlamydia muridarum mouse model have shown that CD4 T cells are critical for clearing genital tract infections. In vitro studies have shown that CD4 T cells terminate infection by up regulating epithelial iNOS transcription and nitric oxide production. However, this mechanism is not critical as iNOS-deficient mice clear infections normally. We recently showed that a subset of Chlamydia-specific CD4 T cell clones could terminate replication in epithelial cells using an iNOS-independent mechanism requiring T cell degranulation. We advance that work using microarrays to compare iNOS-dependent and iNOS-independent CD4 T cell clones. Plac8 was differentially expressed by clones having the iNOS-independent mechanism. Plac8-deficient mice had delayed clearance of infection, and Plac8-deficient mice treated with the iNOS-inhibitor N-monomethyl-L-arginine were largely unable to resolve genital tract infections over 8 weeks. These results demonstrate that there are two independent and redundant T cell mechanisms for clearing C. muridarum genital tract infections; one dependent on iNOS, the other dependent on Plac8. While T cells subsets are routinely defined by cytokine profiles, there may be important subdivisions by effector function, in this case CD4Plac8.


Chlamydia trachomatis serovars D-K are sexually transmitted bacterial pathogens that cause urethritis, endometritis, salpingitis, epididymitis, and pelvic inflammatory disease. In women, infections ascend into the upper reproductive tract causing scarring responsible for ectopic pregnancies and infertility. Aggressive public health measures based on antibiotic treatment of source cases and their sexual partners are partially counterproductive due to deleterious effects on herd immunity (13). There is a consensus among researchers and public health experts that a Chlamydia vaccine is likely necessary to reduce the incidence of Chlamydia trachomatis infections, and the medical resources committed to treating them. A critical component of rational vaccine development is defining the immune parameters that mediate/correlate with protective immunity.

For viral vaccines, neutralizing antibodies are the immune parameter that correlates with protective immunity against primary infections. Existing data from the C. muridarum mouse model suggests that immunity to urogenital serovars of Chlamydia is complex. Passive transfer of antibodies from immune mice has little effect on primary infections of the genital tract (4), and B cell-deficient mice clear primary genital tract infections similar to wild type controls (5). Clearance of primary genital tract infections is dependent on T cell immunity, and T cell-immunity is sufficient to clear the genital tract (6). MHC class I antigen presentation is dispensable for protective host immunity while MHC class II is absolutely required (7). Recent data in experimental mouse models argues for a supportive role for antibodies in vaccine-generated immunity in the genital tract (8, 9). A large body of research has identified CD4 T cells of the Th1 subset as the critical parameter for protective host immunity (6). The protective Th1 subset generated by vaccination appears to be multifunctional CD4 T cells that secrete IFN-γ & TNF-α, with or without IL-2 production (10).

In vitro studies using a Chlamydia-specific CD4 T cell clone capable of terminating C. muridarum replication in epithelial cells identified a probable mechanism for sterilizing CD4 T cell-mediated immunity that was dependent on iNOS, IFN-γ, and T cell-epithelial cell contact via LFA-1 - ICAM-1 (1114). However, follow up studies showed iNOS-knockout mice were not compromised in clearance of C. muridarum genital tract infections (15, 16), and that IFN-γ-knockout mice cleared 99.9% of C. muridarum from the genital tract with near normal kinetics (17, 18). Furthermore, mice deficient in the known T cell killing mechanisms perforin & FasL-Fas were also able to clear C. muridarum genital tract infections with normal kinetics (19). Additional knockout mice and antibody depletion studies have shown that mice deficient in TNF-α receptors (20), IL-12 (18, 20), or IL-1β (21) have varying degrees of compromised C. muridarum genital tract infection clearance. However, cytokine deficiencies have pleiotropic effects on host innate and adaptive immunity, and do not identify specific effector mechanisms responsible for terminating Chlamydia replication in reproductive tract epithelium.

Our lab recently identified an iNOS-independent mechanism for terminating C. muridarum replication in epithelial cells (22). Using a panel of Chlamydia-specific CD4 T cell clones we showed that a subset of clones could terminate C. muridarum replication in the presence of the iNOS inhibitor N-monomethyl-L-arginine (MLA), while a separate clone was completely inhibited by MLA as previously reported in the literature. The iNOS-independent CD4 T cell clones were also relatively independent of IFN-γ, and could be partially blocked using phenyl arsenic oxide (PAO), an inhibitor of T cell degranulation. Potent blocking of the iNOS-independent CD4 clones required both an inhibitor of degranulation and an iNOS inhibitor, suggesting two independent mechanisms capable of terminating C. muridarum replication in infected epithelial cells. All the Chlamydia-specific CD4 T cell clones were Th1, made IFN-γ when activated, and were capable of recognizing infected epithelial cells (23).

To further investigate the findings above, we compared two iNOS-independent CD4 T cell clones to two iNOS-dependent CD4 T cell clones using gene expression micro arrays. Micro array analysis identified a candidate second mechanism for terminating C. muridarum replication in epithelial cells that was then investigated in vivo utilizing an existing knockout mouse. Interesting results of those studies are reported here.

Materials and Methods


4–5 week old female C57BL/6J and B6.129S6-Plac8 (009598) mice were purchased from The Jackson Laboratories (Bar Harbor, MA). All mice were housed in Indiana University Purdue University-Indianapolis (IUPUI) specific-pathogen-free facilities (SPF). The IUPUI Institutional Animal Care and Utilization Committee approved all experimental protocols.

Cells, T cell clones and bacteria

C57epi.1 epithelial cells and Chlamydia-specific CD4 T cell clones uvmo-1, uvmo-2, uvmo-4, and spl4-10, were derived from C57BL/6 mice and cultured as previously described (23). Mycoplasma-free Chlamydia muridarum (Nigg), previously known as C. trachomatis strain mouse pneumonitis (MoPn) (Nigg) was grown in McCoy cells as previously described (23).

Cytokine ELISAs

5×104 T cell clones were activated in 96 well tissue culture plates by immobilized anti-CD3 monoclonal antibody 145-2c11 (BD Biosciences, San Jose, CA), 0.5 μg/ml in PBS overnight at 4°C (washed once), in RPMI media containing 1 ηg/ml recombinant murine IL-7 (R&D Systems, Minneapolis, MN). Relative levels of interferon-gamma (IFN-γ), tumor necrosis factor-alpha (TNF-α), IL-2, and IL-10 in culture supernatants were determined by ELISA using capture and biotinylated monoclonal antibody pairs with recombinant murine standards according to the manufacturer's protocols. IFN-γ ELISA: XMG1.2; IL-2 ELISA 5H4/1A12 (Pierce-Endogen; Rockford, IL). TNF-α ELISA: TN3-19.12/C1150-14; IL-10 ELISA: JES5-2A5/SXC-1 (BD Biosciences). Detection was accomplished with Streptavidin-HRP (BD Biosciences) and TMB substrate (Sigma Chemical Co., St. Louis, MO).

Gene expression micro array analysis

Chlamydia-specific CD4 T cell clones uvmo-2, uvmo-3, uvmo-4, and spl4-10 were purified by ficoll-hypaque (histopaque 1083; Sigma Chemical Co.) at the end of their culture cycle and then grown for 3 days in their usual culture media including growth factors, without antigen stimulation. On day three, total RNA was isolated from each T cell clone using a protocol that included an RNAse-free DNAse I treatment step (RNAeasy; Qiagen, Valencia, CA). With assistance from The Indiana University Center for Medical Genomics, gene expression patterns were analyzed using the Affymetrix Mouse ST 1.0 Array that analyzes 28,853 murine genes. Samples were labeled using the standard Affymetrix protocol for the WT Target Labeling and Control Reagents kit according to the Affymetrix user manual: GeneChip® Whole Transcript (WT) Sense Target Labeling Assay GeneChip. Individual labeled samples were hybridized to the Mouse Gene 1.0 ST GeneChips® for 17 hours then washed, stained and scanned with the standard protocol using Affymetrix GCOS (GeneChip® Operating System). GCOS was used to generate data (CEL files). Arrays were visually scanned for abnormalities or defects. CEL files were imported into Partek Genomics Suite (Partek, Inc., St. Louis, Mo). RMA signals were generated for the core probe sets using the RMA background correction, quantile normalization and summarization by Median Polish. Summarized signals for each probe set were log2 transformed. These log transformed signals were used for Principal Components Analysis, hierarchical clustering and signal histograms to determine if there were any outlier arrays. Untransformed RMA signals were used for fold change calculations. Data was analyzed using a 1-way Anova (analysis of variance) using log2 transformed signals for all four CD4 T cell clones, and contrasts were made comparing iNOS-dependent uvmo-4 and spl14-10 individually to the combined expression of the iNOS-independent T cell clones uvmo-2 and uvmo-3. Fold changes were calculated using the untransformed RMA signals. Genes up or down regulated 5-fold with p values <0.001 for either uvmo-4 or spl4-10 compared to uvmo-2/uvmo-3 were considered in the final analysis (supplemental table 1). The microarray data presented here is available in the Gene Expression Omnibus database ( under accession number GSE32128.

RT-PCR analysis

Chlamydia-specific CD4 T cell clones uvmo-2, uvmo-3, uvmo-4, and spl4-10 were purified by ficoll-hypaque at the end of their culture cycle and then grown for 3 days in their usual culture media including growth factors, without antigen stimulation. On day three, total RNA was isolated from each T cell clone using a protocol that included an RNAse-free DNAse I treatment step. Specific mRNA gene reverse transcription and applification was performed using AMV reverse transcriptase/Tfl DNA polymerase in a one step system (AccessQuick RT-PCR; Promega, Madison, WI). Amplification conditions were (1) 48°C for 45 min; (2) 95°C 2 min; (3) 95°C 30 sec; (4) 56°C 20 sec; (5) 72°C 30 sec; (6) goto step three 37 times; (7) 72°C 7 min; (8) hold 4°C, using an MJ Research J200 PCR machine. All gene analyses were done on the same set of RNA samples. PCR primers, amount of total RNA in amplification, and expected product sizes shown in table 1. RT-PCR reactions were run on 2% agarose gels with ethidium bromide; images inverted for presentation. The “no RT” controls generated no RT-PCR products (data not shown). PCR primers used are shown in table 1.

Table 1
RT-PCR primers

Epithelial cell infections/Chlamydia replication/titration

C57epi.1 cell monolayers in 48 well plates were untreated or treated with IFN-γ (10 ηg/ml) for 10 h prior to infection. Wells were infected with 3 IFU per cell. After addition of C. muridarum the plates were spun at 300 × g for 30 minutes. 4 h post infection the inoculums were removed and CD4 T cell clones were added in RPMI 1640 T cell medium. 32 h post infection, wells were scraped, harvested with an equal volume of sucrose-phosphate-glutamate acid buffer (SPG buffer) and stored at −80° C until C. muridarum titers determined on McCoy cell monolayers using an anti-Chlamydia LPS antibody and FITC-labeled goat anti-mouse IgG (Rockland Immunochemicals, Gilbertsville, PA) as previously described (24).

Genital tract infections

One week prior to infection mice were treated with 2.5 μg of medroxyprogesterone delivered subcutaneously (depo-Provera, Pfizer Pharmaceuticals, New York, NY). Lightly anesthetized mice were infected vaginally with 5×104 inclusion forming units (IFU) of C. muridarum biovar Nigg in 10 μl of SPG buffer. Mice were swabbed M-W-F to monitor the clearance of C. muridarum from the genital tract as previously described (25).

Pathology Scoring

A simple scoring system was utilized to assess macroscopic genital tract pathology. Hydrosalpinx was scored as 0, 1, or 2 reflecting no, unilateral, or bilateral hydrosalpinx respectively. Mice have a bi-fed uterus (2 uterine horns) that was similarly scored 0, 1, or 2 for hydro-uterus. The maximum pathology score for an individual mouse was 4.

Degranulation experiments

5×104 CD4 T cell clone cells were incubated in wells with media, phorbol myristate acetate (PMA; Sigma) (10 ηg/ml) & calcium ionophore A23187 (500 ng/ml; Sigma), PMA & A23187 + immobilized anti-CD3 (0.5 μg/ml; 50 μl PBS overnight 4° C) (145-2C11; BD Biosciences), or media containing 0.1% tritonX-100 for 4 h at 37° C. Culture supernatants were collected and levels of granzyme B were determined using a murine granzyme B ELISA (mouse granzyme B duoset; R&D systems, Minneapolis, MN).

%Degranulation=measured value in experimental well- spontaneous release(media)×100tritonX-100value(maximal release)spontaneous release

Statistical Analysis

Summary figures for each experimental investigation are presented means and standard deviations (SD) or as `pooled' means and standard error of the mean (SEM). Figure legends indicate the number of independent experiments pooled to generate each figure. Student's two-tailed t-tests were used to assess significance of experimental data. Homogeneity of variances was assessed using a folded F-test. Statistical analysis of Chlamydia shedding was performed using repeated measures analysis of variance to test for differences between groups over time. All data were verified to meet analytic assumptions. Nonparametric tests were also performed and showed similar results. Analyses were performed using SAS 9.2 (SAS Institute, Cary, NC).


Characteristics of Chlamydia-specific CD4 T cell clones used in gene expression micro array

We previously characterized a panel of ten Chlamydia-specific CD4 T cell clones including uvmo-2, uvmo-3, uvmo-4, and spl4-10 (22, 23). The uvmo-2 and uvmo-3 T cell clones are potent terminators of C. muridarum replication, with or without IFN-γ pretreatment of epithelial monolayers, and make significant amounts of IFN-γ when activated by infected epithelial cells (23). They have two potent mechanisms to terminate replication; one dependent on iNOS, the other dependent on T cell degranulation. Significant reversal of the ability of uvmo-2 and uvmo-3 to terminate infection required simultaneous use of two inhibitors; an iNOS-inhibitor (MLA) and an inhibitor of T cell degranulation (PAO) (22). Availability of the degranulation-dependent mechanism allowed uvmo-2 & uvmo-3 to terminate C. muridarum replication in the presence of iNOS inhibitors (iNOS-independent). The uvmo-4 and spl4-10 CD4 T cell clones make modest amounts of IFN-γ when activated by infected epithelial cells, and they require IFN-γ pretreatment of the epithelial monolayer to control C. muridarum replication (23). The ability of Spl4-10 to terminate C. muridarum replication in epithelial cells is iNOS-dependent, and was completely reversed by iNOS inhibition with MLA (22). The mechanism utilized by uvmo-4 to terminate C. muridarum replication in epithelial cells was not previously determined. Because uvmo-4 required IFN-γ pretreatment of epithelial cells to control C. muridarum replication, we hypothesized that it was dependent on iNOS similar to spl4-10. C57epi.1 epithelial monolayers pretreated 10 h with 10 ηg/ml IFN-γ in the absence or presence of 1 mM MLA were infected with C. muridarum for 4 h, then co-cultured without and with uvmo-4 at a T cell:epithelial cell ratio of ~0.75:1. Thirty-two hours post infection the wells were harvested by addition of SPG buffer plus scraping. Replication of C. muridarum under each condition was quantified by plating recovered IFU on McCoy monolayers (Figure 1). The ability of uvmo-4 to block >90% of C. muridarum recovery required IFN-γ pretreatment of the epithelial monolayer. iNOS inhibition with MLA restored C. muridarum recovery from untreated epithelial cells, and restored C. muridarum recovery to >50% of control values in IFN-γ pretreated epithelial cells. Termination of C. muridarum replication by uvmo-4 was dependent on iNOS production of nitric oxide. Identification of uvmo-4 as an iNOS-dependent CD4 T cell clone gave us the desired CD4 T cell panel for gene expression micro array analyses; i.e. two iNOS-dependent CD4 T cell clones (uvmo-4 and spl4-10) and two iNOS-independent CD4 T cells clones (uvmo-2 and uvmo-3).

Figure 1
Uvmo-4 is an iNOS-dependent CD4 T cell clone. C57epi.1 epithelial cells untreated (panel A)or pretreated for 10 h with IFN-β (panel B), in the absence or presence of iNOS inhibitor MLA, were infected with C. muridarum at 3 IFU per cell. Inocula ...

Uvmo-2, -3, -4, and spl4-10 would be classified Th1 based on IFN-γ production, but there are significant differences in their cytokine profiles (Figure 2). When activated via the T cell receptor complex uvmo-2, uvmo-3, and uvmo-4 produce TNF-α (50–100 ηg/ml) and IL-2 (3–10 ηg/ml) in addition to IFN-γ; spl4-10 produces significant amounts of IL-10 (14 ηg/ml), lesser amounts of IFN-γ & TNF-γ, and a negligible amount of IL-2.

Figure 2
Multifunctional Chlamydia-specific CD4 T cell clones with differing cytokine profiles. 5×104 T cells were cultured, without activation or activated by immobilized anti-CD3 monoclonal antibody, in RPMI media containing 1 ηg/ml IL-7. 20 ...

To assess the role of cytokines in the anti-Chlamydia effector mechanism of these four CD4 T cell clones, we activated each T cell clone with UV-inactivated C. muridarum pulsed irradiated splenocytes in media containing IL-2 (25 units/ml) and IL-7 (2 ηg/ml), harvesting the conditioned media 48 h later. Control RPMI media and fresh cell-free conditioned media from each clone (40% vol/vol in RPMI media) were exchanged for the epithelial media covering monolayers of C57epi.1 epithelial cells infected 4 h earlier with 3 IFU of C. muridarum per cell. 32 h post infection the wells were harvested by adding SPG buffer and scraping. Replication of C. muridarum under each condition was quantified using McCoy monolayers (Figure 3). The final IFN-γ concentration under each experimental condition was comparable for all four clones, as was the relative inhibition of C. muridarum replication. Fresh T cell conditioned media was 2–3 logs more potent than recombinant murine IFN-γ at 10 ηg/ml, which causes only a 2–3-fold reduction in recovered IFU/well in C57epi.1 cells (22). The cytokine-mediated inhibition of replication was almost completely reversed in the presence of MLA, arguing that the mechanism of replication inhibition by cytokines in the absence of T cells was potent induction of epithelial iNOS and nitric oxide production.

Figure 3
The role of T cell cytokines and iNOS in controlling C. muridarum replication in epithelial cells. Fresh T cell clone conditioned culture supernatant (40% vol/vol), without and with 1mM MLA, was exchanged for the infecting inocula 4 h post-infection of ...

Gene expression micro arrays comparing iNOS-independent T cell clones to iNOS-dependent T cell clones

Our previous study showed that the most potent CD4 T cell clone terminators of C. muridarum replication, uvmo-2 and uvmo-3, utilized induction of epithelial iNOS, and a degranulation-dependent effector mechanism that was not functional in spl4-10. We also showed that uvmo-2 and uvmo-3 recognized and lysed infected epithelial cells late in replication cycle; too late to catch C. muridarum in the non-infectious reticulate body stage (eclipse phase). Late recognition of infected epithelial cells by uvmo-2 & uvmo-3 argued against their termination mechanism being a function of earlier presentation of their nominal antigens, i.e. not early recognition. Because the uvmo-2 & uvmo-3 could terminate replication in the presence of iNOS inhibitor MLA, we hypothesized that they had an unidentified T cell effector mechanism. To gain insight into this unidentified degranulation-dependent effector mechanism we performed gene expression micro arrays comparing iNOS-independent (uvmo-2 & uvmo-3) to iNOS-dependent (uvmo-4 & spl4-10) T cell clones. We recognized that micro array analyses of four T cell clones would not permit statistically significant conclusions about T cell mechanisms, but postulated that the resulting data would suggest testable hypotheses. T cell clones at the end of their culture cycle were ficoll-hypaqued to remove cellular debris, washed, then cultured for 72 h in media with usual growth cytokines. Total RNA was harvested and used for Affymetrix mouse ST1.0 gene expression micro arrays; four independent RNA isolations for four independent micro array experiments were done to minimize background noise. Figure 4 shows principal component analysis from the micro arrays comparing the four T cell clones. Interestingly the two iNOS-independent CD4 T cell clones derived from different mice (uvmo-2 & 3) clustered tightly together and away from the iNOS-dependent clones (uvmo-4 & spl4-10), indicating that they had a similar expression pattern for the genes that accounted for the greatest variability in the data set. In supplemental table 1 we list the 193 genes that differed + or – 5-fold comparing the individual expression patterns of spl4-10 and uvmo-4 to the combined expression pattern of uvmo-2 & uvmo-3. The genes that were the most intriguing to us and the primary focus of this paper were those potentially useful as biomarkers or possibly involved in an effector mechanism (Table 2). Uvmo-2 & uvmo-3, the iNOS-independent terminators of C. muridarum replication, had high mRNA levels of Casd1 (CAS1 domain containing 1) and Plac8 (placenta-specific 8), and low levels of perforin (Prf1). The iNOS-dependent T cell clones uvmo-4 & spl4-10 had very low Casd1 and Plac8 levels and high levels of perforin. RT-PCR analysis confirmed the micro array data for Casd1, Plac8, and perforin in iNOS-independent (uvmo-2,-3) and iNOS-dependent (uvmo-4 & spl4-10) CD4 T cell clones (Figure 5).

Figure 4
Principal Component Analysis of the gene expression micro array data for the four CD4 T cell clones. Red = uvmo-2; blue = uvmo-3; light purple = uvmo-1; dark purple = spl4-10.
Figure 5
RT-PCR confirmation of micro array data. mRNA levels for Casd1, Plac8, perforin (Prf1), and β actin were analyzed using an independent total RNA isolation for each CD4 T cell clone. RT-PCR products were run on a 2% ethidium bromide agarose gel; ...
Table 2
Biomarker and Mechanism Genes

Plac8 knockout mice are compromised in their ability to clear C. muridarum genital tract infections

High level expression of Plac8 by the iNOS-independent CD4 T cell clones was particularly interesting because Plac8 knockout mice were previously shown to have a defect in clearing Klebsiella pneumoniae peritoneal infections; a phenotype ascribed to a killing defect in Plac8-deficient neutrophils (26). In the original Plac8 knockout mouse study, investigation of tissue distribution showed expression of Plac8 in bulk T cell populations by western blotting. To investigate whether Plac8 was important in host defense against C. muridarum genital tract infections we challenged Plac8 knockout mice, comparing them to C57BL/6J control mice.

C57BL/6J and Plac8 knockout mice were treated with medroxyprogesterone to synchronize estrous, then infected vaginally with 5×104 IFU of C. muridarum. The iNOS inhibitor MLA was added to the drinking water on day 52 and continued thereafter. Mice were swabbed thrice weekly through 80 days post-infection then assessed for genital tract pathology. Recovered IFUs from genital swabs were titered on McCoy monolayers (Figure 6). Plac8 mice had a clear defect in clearance of C. muridarum from the genital tract. C57BL/6J control mice cleared the infection by day 26 while Plac8 knockout mice cleared at day 47. Clearance kinetics were identical through the first three weeks of infection before diverging during clearance of the final ~1000 IFU/swab from the genital tract. Addition of iNOS inhibitor MLA to the drinking water post-clearance on day 52 led to transient recovery of viable C. muridarum from Plac8 knockout mice on days 56 & 59, but not from C57BL/6J mice at any time point. With MLA treatment all the Plac8 knockout mice shed viable Chlamydia on at least one day; two mice on a single day (either day 56 or day 59), and four mice on both days. These results mirror previously published experiments done by Ramsey et al with iNOS knockout mice where cyclophosphamide treatment (lymphocyte/neutrophil depletion) post-clearance led to transient recovery of viable C. muridarum from iNOS knockout mice but not C57BL/6J control mice (27). Our data combined with those of Ramsey et al argue that sterilizing immunity in the genital tract is dependent on two separate mechanisms; one mechanism dependent on iNOS, the other mechanism dependent on Plac8. Clearance of transient C. muridarum shedding in iNOS and Plac8 knockout mice in post-clearance experiments with cyclophosphamide and MLA respectively argues for a third mechanism, possibly the antibody-dependent mechanism described by Morrison et al (4, 28).

Figure 6
Plac8 is required for sterilizing immunity. Wild type C57BL/6J and Plac8 female knockout mice in experimental groups of 6 mice each were treated with medroxyprogesterone, then infected vaginally one week later with 5×104 IFU of C. muridarum. On ...

To further investigate the role of Plac8 in clearance of genital tract infections, C57BL/6J and Plac8 knockout mice were pretreated with medroxyprogesterone on day T-7, continuous MLA was added to the drinking water at day T-1, and then the mice were infected vaginally on day 0 with 5×104 IFU of C. muridarum. Mice were swabbed thrice weekly through day 56 post-infection; recovered IFU were titered on McCoy monolayers (Figure 7). Plac8 knockout mice treated with MLA were markedly compromised in their ability to clear C. muridarum compared to C57BL/6J mice under the same experimental conditions. Clearance kinetics diverged after day 14, with all C57BL/6J mice clearing the infection by day 40 compared to only 1 of 8 Plac8 knockout mice clearing the infection by the day 56. In addition to the absolute difference in final clearance, there is a difference in the slope of the clearance curve with the Plac8 clearance curve being nearly flat from day 21 through the end of the experiment. Clearance of C. muridarum from one Plac8 knockout mouse on the last day of the experiment raises the possibility that Plac8 knockout mice treated with MLA would clear genital tract infections over a longer time course. Two Plac8 knockout mice infected with continuous MLA drinking water experienced an inflammatory response that sealed the vaginal vault, making it impossible to do sequential monitoring of Chlamydia shedding. This outcome is unique in the authors' experience with the C. muridarum mouse model. These mice were killed at day 28 due to increasing abdominal girth. Both mice appeared to be well by activity level but were judged to be at risk for acute decompensation. Gross reproductive tract pathology in these mice is shown in Figure 8. Both Plac8 knockout mice had gross distortion of the genital tract anatomy with pan dilatation, hydrouterus or hemouterus, and severe hydrosalpinx. The lumenal fluid and spleen were harvested from the second mouse and C. muridarum quantified on McCoy monolayers. Significant IFUs were recovered from the hydrosalpinx fluid (25,000 IFU/ml), and from the spleen (8,400 IFU) suggesting dissemination and extra-genital tract replication when both the iNOS-dependent and the Plac8-dependent mechanisms are compromised.

Figure 7
Inhibition of iNOS at initiation of infection compromises clearance of C. muridarum from the genital tract of Plac8 knockout mice. Wild type C57BL/6J (11 mice) and Plac8 knockout (10 mice) female mice were treated with medroxyprogesterone, then infected ...
Figure 8
Severe pathology in Plac8 knockout mice treated with MLA at time of vaginal infection with C. muridarum. The two Plac8 knockout mice that could not be monitored in the experiment shown in figure 7 were killed on day 28 post-infection and compared to an ...

Assessment of macroscopic genital tract pathology for the experiments shown in figures 6 and and77 showed no difference between C57BL/6J mice and Plac8 knockout mice when MLA was administered on post-infection day 52 (Delayed MLA), but a significant difference when MLA was administered concurrently with infection (Initial MLA) (see Figure 9).

Figure 9
Mice in delayed-MLA (figure 6) and initial-MLA (figure 7) infection experiments were scored for macroscopic pathology at the end of those experiments. Statistical comparisons were made between the C57BL/6J and Plac8 knockout mice for each experiment. ...

CD4 T cell clones that do not express Plac8 have no apparent defect in degranulation

Our previous study showed that the iNOS-independent mechanism utilized by uvmo-2 & uvmo-3 was dependent on T cell degranulation. In neutrophils, Plac8 localizes to the granules. It is unclear from existing data whether Plac8 is in granules or regulates trafficking of granules (see discussion). With respect to degranulation-dependent mechanism utilized by uvmo-2 & uvmo-3, the two simplest explanations for Plac8 function would be either directly as an antimicrobial effector molecule analogous to human granulysin, or indirectly as a regulator of events in degranulation, i.e. participation in delivery of an unidentified effector molecule. To test whether T cells that do not express Plac8 are deficient in degranulation, we compared the degranulation capacity of our C57BL/6 derived Plac8+ CD4 T cell clones uvmo-2 & uvmo-3 to our Plac8 CD4 T cell clones spl4-10 & uvmo-4. The T cell clones were degranulated with PMA/A23187 and PMA/A23187 plus anti-CD3. Degranulation was measured as % granzyme B release as measured by ELISA (Figure 10). The four CD4 T cell clones had a wide range of intracellular granzyme B levels (26–390 pg/ml). There was no obvious defect in degranulation in any of the CD4 T cell clones, or any correlation of degranulation efficiency with presence or absence of Plac8 expression. Plac8 does not appear to have a role in T cell degranulation.

Figure 10
Plac8 expression is not required for degranulation. The CD4 T cell clones were degranulated with either PMA & A23187 (gray bars) or PMA & A23187 + immobilized anti-CD3 (black bars). There was no correlation between Plac8 expression (uvmo-2,-3) ...


Data presented in this study establish that there are two redundant mechanisms for clearance of C. muridarum from the genital tract. The first mechanism is dependent on T cell & IFN-γ–mediated up regulation of epithelial iNOS transcription and nitric oxide production as previously described in the literature by Igietseme et al (29). The second mechanism is dependent on Plac8, and likely on T cell degranulation based on our in vitro studies (22). The existence of an iNOS-independent mechanism capable of clearing C. muridarum from the genital tract explains longstanding confusion about the role of iNOS in protective immunity. Mice deficient in iNOS clear C. muridarum because they still have the Plac8-dependent mechanism; conversely, Plac8-deficient mice clear infection because they still have the iNOS-dependent mechanism. In this study we showed that dual deficient mice, genetically deficient in Plac8 and pharmacologically deficient in iNOS activity (nitric oxide), were severely compromised in their ability to resolve a C. muridarum genital tract infection over 8 weeks. In addition, we showed that Plac8 contributes to sterilizing immunity as live C. muridarum could be recovered from Plac8 knockout mice that had cleared genital tract infections by administration of an iNOS inhibitor (MLA). That result combined with a previous study showing that cyclophosphamide treatment led to recovery of viable C. muridarum from iNOS knockout mice that had cleared a genital tract infection argues that sterilizing immunity requires both iNOS-dependent and Plac8-dependent mechanisms.

Data from our in vitro studies and others in vivo studies strongly suggest that clearance of Chlamydia from the genital tract is not dependent on physical killing of infected epithelial cells. We have previously shown that three Chlamydia-specific CD4 T cell clones that were potent terminators of C. muridarum replication in epithelial cells recognized infected cells late in the replication cycle (22). Those clones did not lyse 18 h infected epithelial cells (optimal recognition time point) in 4 h killing assays, but eventually lysed the epithelial monolayer after the reticulate body (RB) to elementary body (EB) transition was complete. On that basis we concluded that cytolysis was not likely critical for these clones to terminate infection, and that termination of infection required a direct antimicrobial attack on EB. The iNOS-independent CD4 T cell-mediated termination of infection was inhibited by phenyl arsenine oxide (PAO), an inhibitor of T cell degranulation, suggesting delivery of an anti-Chlamydia effector molecule. That data combined with in vivo studies showing that perforin knockout mice, FasL knockout mice, and perforin-Fas dual knockout mice clear infections with normal kinetics argues in favor of clearance mechanisms that are independent of cytolysis.

In the current study we have additional data that support a non-cytolytic mechanism for T cell-mediated clearance of genital tract infections. CD4 T cell clones that are potent terminators of Chlamydia replication in epithelial cells turn out to have low levels of perforin, while less potent Chlamydia-specific CD4 T cell clones have high levels of perforin. On that basis it does not appear that high perforin expression correlates with terminating infection in epithelial cells. The small amounts of perforin present in perforin low CD4 T cell clones must be sufficient to deliver an anti-Chlamydia effector molecule to the epithelial cytosol, or there may be a perforin independent pathway. Our previous report showed that PAO and concanamycin A prevented potent CD4 T cell clone-mediated termination of Chlamydia replication in epithelial cells. Concanamycin A is an H+/ATPase inhibitor that causes perforin degradation by raising the pH of T cell granules. We equivocated about interpreting the concanamycin A (CMA) results because CMA was also a potent inhibitor of C. muridarum replication necessary for T cell recognition of infected cells. It is possible that concanamycin A elevation of pH in T cell granules may also cause degradation of molecules other than perforin, including the anti-Chlamydia effector molecule.

The biological function of Plac8 is unclear from existing data. Plac8 is a small 112 aa cysteine-rich protein originally described as being enhanced in the reproductive tract (30). Plac8 is expressed by macrophages, neutrophils, B cells, T cells and epithelial cells (lung & small bowel specifically), and has a highly conserved human homolog (26). Some studies have linked Plac8 to hematopoetic cell survival (3133), though that biology did not appear to be relevant for neutrophils from wild type and Plac8 knockout mice (26). Plac8 knockout mice were more susceptible to Klebsiella pneumoniae peritonitis, and their neutrophils were compromised in their ability to kill ingested bacteria in vitro. In neutrophil fractionation experiments Plac8 localized to granules and was not found in the cytosol. SDS-PAGE eletrophoretic mobility suggested that Plac8 was not processed (26). Plac8 does not have a conventional signal sequence. There is a potential signal peptide cleavage site between amino acids 17 and 18, but only 7% of known mammalian cleavage sites share these amino acid sequence characteristics (SignalP; Our data and the existing Klebsiella data are most easily explained by one of two scenarios. Plac8 is either an antimicrobial effector molecule analogous to human granulysin, or it is a critical component of the pathway that delivers or activates an anti-Chlamydia effector molecule in the cytosol of infected epithelial cells. In this report we showed that Plac8 does not appear to have a role in T cell degranulation. It is attractive to speculate that reactive sulfhydryl groups on cysteine-rich Plac8 react with cysteine-rich OmcA & OmcB in the Chlamydia EB outer membrane leading to neutralization. Our data show that Plac8 is not ubiquitously expressed by CD4 T cells. Limited by an n=4, it appears that CD4 T cells that express Plac8, CD4Plac8, utilize both iNOS-dependent and degranulation-dependent mechanisms for terminating C. muridarum replication in epithelial cells, while Plac8 CD4 T cells appear to be dependent on nitric oxide. It is plausible that dual effector Chlamydia-specific CD4Plac8 T cells are the optimal protective subset in the genital tract.

The susceptibility of Plac8 knockout mice to C. muridarum genital tract infections is likely due to a defect in T cell immunity because nude mice sufficient in neutrophils, Plac8, and iNOS are incapable of clearing primary C. muridarum primary genital tract infections (34), time to clearance in wild type mice is not affected by neutrophil depletion (35, 36), and clearance is entirely dependent on MHC class II- restricted T lymphocytes (7). The details of how Plac8 is critical for cellular immunity to Chlamydia remain to be determined. It is interesting to note that both iNOS- and Plac8-dependent mechanisms for terminating Chlamydia replication in epithelial cells are not likely to be cytolytic mechanisms that expose intracellular bacteria to extracellular effector mechanisms such as neutralizing antibodies or neutrophils. Rather both mechanisms appear to kill Chlamydia within intact epithelial cells. In this “toxic bag” model, Chlamydia is trapped and neutralized within the infected cells by a combination of nitric oxide and Plac8 or a Plac8-dependent effector. This type of mechanism is intellectually satisfying as the majority of CD4 T cells appear to recognize infected epithelial cells after the noninfectious RB to infectious EB transition has occurred (23). Because Plac8 is also expressed by epithelial cells that are susceptible to infection, we speculate that it exists in an inactive state in those cells. It is possible that CD4 T cells not only deliver active Plac8 to infected cell cytosol, but that active Plac8 or other injected T cell molecule triggers activation of epithelial Plac8 in an amplified final pathway.

In this study CD4 T cell conditioned media was a very potent inhibitor of C. muridarum replication in murine epithelial cells through an iNOS-dependent mechanism; greater than two logs more potent than recombinant IFN-γ in our previous studies and those of others. This finding suggests that in vitro studies using recombinant IFN-γ (22, 3739) may have significantly underestimated the contribution of cytokines to controlling C. muridarum replication in murine cells due either to lesser biologic activity of recombinant IFN-γ or more likely, contributions of additional cytokines such as TNF-γ to a cytokine-mediated anti-Chlamydia epithelial response. The iNOS promoter once activated by IFN-γ can be further up regulated by TNF-α (40). TNF-α has been shown to be an important cytokine during genital tract clearance (20) and in protective immunity induced by Chlamydia T cell vaccines (10, 41). Scurlock et al in a recent paper focused on possible IL-17 contributions to Chlamydia immunity showed that IFN-γ and TNF-γ in genital tract secretions of C. muridarum infected mice peaked on day 4 and were essentially undetectable by day 14 (42). Doxycycline experiments suggest that adaptive immunity takes hold about day 10 post-infection (43). We propose a working model in which early innate immunity in a genital tract awash with cytokines including IFN-γ & TNF-α clears 2–3 logs of C. muridarum during the first 10 days of infection, and that clearance of the last 3–4 logs of C. muridarum by the adaptive cellular immune response can be accomplished by either cytokine-dependent (iNOS) or Plac8-dependent mechanisms; sterilizing cellular immunity requires both iNOS- & Plac8-dependent mechanisms.

Identifying the Plac8-dependent mechanism for clearing Chlamydia from the reproductive tract resolves longstanding confusion about the role of iNOS in protective immunity to Chlamydia genital tract infections by identifying a redundant second clearance mechanism. Investigating the CD4Plac8 T cell subset will likely contribute to our understanding of protective immunity, and hopefully identify surrogate biomarkers for protective immunity that prove useful during rational vaccine development. Casd1 is an uncharacterized open reading frame predicted to be a transmembrane glycoprotein with an extracellular domain. We are investigating Casd1 as a possible T cell accessory molecule, and potentially a convenient biomarker for the CD4Plac8 T cell subset.

Supplementary Material


The authors have no conflicts of interest related to the content of this manuscript. Critical assistance with micro array data analysis was provided by Jeanette McClintick and Howard Edenberg in the Indiana University Center for Medical Genomics.


1This research was supported by NIH grant R01AI070514.


1. Brunham RC, Pourbohloul B, Mak S, White R, Rekart ML. The Unexpected Impact of a Chlamydia trachomatis Infection Control Program on Susceptibility to Reinfection. J Infect Dis. 2005;192:1836–1844. [PubMed]
2. Brunham RC, Rekart ML. The arrested immunity hypothesis and the epidemiology of chlamydia control. Sex Transm Dis. 2008;35:53–54. [PubMed]
3. Rekart ML, Brunham RC. Epidemiology of chlamydial infection: are we losing ground? Sex Transm Infect. 2008;84:87–91. [PubMed]
4. Morrison SG, Morrison RP. A predominant role for antibody in acquired immunity to chlamydial genital tract reinfection. J Immunol. 2005;175:7536–7542. [PMC free article] [PubMed]
5. Su H, Feilzer K, Caldwell HD, Morrison RP. Chlamydia trachomatis genital tract infection of antibody-deficient gene knockout mice. Infect Immun. 1997;65:1993–1999. [PMC free article] [PubMed]
6. Morrison RP, Caldwell HD. Immunity to murine chlamydial genital infection. Infect Immun. 2002;70:2741–2751. [PMC free article] [PubMed]
7. Morrison RP, Feilzer K, Tumas DB. Gene knockout mice establish a primary protective role for major histocompatibility complex class II-restricted responses in Chlamydia trachomatis genital tract infection. Infect Immun. 1995;63:4661–4668. [PMC free article] [PubMed]
8. Farris CM, Morrison SG, Morrison RP. CD4+ T cells and antibody are required for optimal major outer membrane protein vaccine-induced immunity to Chlamydia muridarum genital infection. Infect Immun. 2010;78:4374–4383. [PMC free article] [PubMed]
9. Li W, Murthy AK, Guentzel MN, Chambers JP, Forsthuber TG, Seshu J, Zhong G, Arulanandam BP. Immunization with a combination of integral chlamydial antigens and a defined secreted protein induces robust immunity against genital chlamydial challenge. Infect Immun. 2010;78:3942–3949. [PMC free article] [PubMed]
10. Yu H, Karunakaran KP, Kelly I, Shen C, Jiang X, Foster LJ, Brunham RC. Immunization with live and dead Chlamydia muridarum induces different levels of protective immunity in a murine genital tract model: correlation with MHC class II peptide presentation and multifunctional Th1 cells. J Immunol. 2011;186:3615–3621. [PMC free article] [PubMed]
11. Igietseme JU. The molecular mechanism of T-cell control of Chlamydia in mice: role of nitric oxide. Immunology. 1996;87:1–8. [PubMed]
12. Igietseme JU. Molecular mechanism of T-cell control of Chlamydia in mice: role of nitric oxide in vivo. Immunology. 1996;88:1–5. [PubMed]
13. Igietseme JU, Ananaba GA, Bolier J, Bowers S, Moore T, Belay T, Lyn D, Black CM. The intercellular adhesion molecule type-1 is required for rapid activation of T helper type 1 lymphocytes that control early acute phase of genital chlamydial infection in mice. Immunology. 1999;98:510–519. [PubMed]
14. Igietseme JU, Wyrick PB, Goyeau D, Rank RG. An in vitro model for immune control of chlamydial growth in polarized epithelial cells. Infect Immun. 1994;62:3528–3535. [PMC free article] [PubMed]
15. Igietseme JU, Perry LL, Ananaba GA, Uriri IM, Ojior OO, Kumar SN, Caldwell HD. Chlamydial infection in inducible nitric oxide synthase knockout mice. Infect Immun. 1998;66:1282–1286. [PMC free article] [PubMed]
16. Ramsey KH, Miranpuri GS, Poulsen CE, Marthakis NB, Braune LM, Byrne GI. Inducible nitric oxide synthase does not affect resolution of murine chlamydial genital tract infections or eradication of chlamydiae in primary murine cell culture. Infect Immun. 1998;66:835–838. [PMC free article] [PubMed]
17. Cotter TW, Ramsey KH, Miranpuri GS, Poulsen CE, Byrne GI. Dissemination of Chlamydia trachomatis chronic genital tract infection in gamma interferon gene knockout mice. Infect Immun. 1997;65:2145–2152. [PMC free article] [PubMed]
18. Perry LL, Feilzer K, Caldwell HD. Immunity to Chlamydia trachomatis is mediated by T helper 1 cells through IFN-gamma-dependent and -independent pathways. J Immunol. 1997;158:3344–3352. [PubMed]
19. Perry LL, Feilzer K, Hughes S, Caldwell HD. Clearance of Chlamydia trachomatis from the murine genital mucosa does not require perforin-mediated cytolysis or Fas-mediated apoptosis. Infect Immun. 1999;67:1379–1385. [PMC free article] [PubMed]
20. Perry LL, Su H, Feilzer K, Messer R, Hughes S, Whitmire W, Caldwell HD. Differential sensitivity of distinct Chlamydia trachomatis isolates to IFN-gamma-mediated inhibition. J Immunol. 1999;162:3541–3548. [PubMed]
21. Prantner D, Darville T, Sikes JD, Andrews CW, Jr., Brade H, Rank RG, Nagarajan UM. Critical role for interleukin-1beta (IL-1beta) during Chlamydia muridarum genital infection and bacterial replication-independent secretion of IL-1beta in mouse macrophages. Infect Immun. 2009;77:5334–5346. [PMC free article] [PubMed]
22. Jayarapu K, Kerr M, Ofner S, Johnson RM. Chlamydia-specific CD4 T cell clones control Chlamydia muridarum replication in epithelial cells by nitric oxide-dependent and - independent mechanisms. J Immunol. 2010;185:6911–6920. [PMC free article] [PubMed]
23. Jayarapu K, Kerr MS, Katschke A, Johnson RM. Chlamydia muridarum-specific CD4 T-cell clones recognize infected reproductive tract epithelial cells in an interferon-dependent fashion. Infect Immun. 2009;77:4469–4479. [PMC free article] [PubMed]
24. Johnson RM. Murine oviduct epithelial cell cytokine responses to Chlamydia muridarum infection include interleukin-12-p70 secretion. Infection and immunity. 2004;72:3951–3960. [PMC free article] [PubMed]
25. Schachter J, editor. Chlamydiae (Psittacosis-lymphogranuloma venereum-trachoma group) American Society for Microbiology; Washington, D.C.: 1980.
26. Ledford JG, Kovarova M, Koller BH. Impaired host defense in mice lacking ONZIN. J Immunol. 2007;178:5132–5143. [PubMed]
27. Ramsey KH, Miranpuri GS, Sigar IM, Ouellette S, Byrne GI. Chlamydia trachomatis persistence in the female mouse genital tract: inducible nitric oxide synthase and infection outcome. Infect Immun. 2001;69:5131–5137. [PMC free article] [PubMed]
28. Morrison SG, Morrison RP. Resolution of secondary Chlamydia trachomatis genital tract infection in immune mice with depletion of both CD4+ and CD8+ T cells. Infect Immun. 2001;69:2643–2649. [PMC free article] [PubMed]
29. Igietseme JU, Uriri IM, Hawkins R, Rank RG. Integrin-mediated epithelial-T cell interaction enhances nitric oxide production and increased intracellular inhibition of Chlamydia. J Leukoc Biol. 1996;59:656–662. [PubMed]
30. Galaviz-Hernandez C, Stagg C, de Ridder G, Tanaka TS, Ko MS, Schlessinger D, Nagaraja R. Plac8 and Plac9, novel placental-enriched genes identified through microarray analysis. Gene. 2003;309:81–89. [PubMed]
31. Rogulski K, Li Y, Rothermund K, Pu L, Watkins S, Yi F, Prochownik EV. Onzin, a c-Myc-repressed target, promotes survival and transformation by modulating the Akt-Mdm2-p53 pathway. Oncogene. 2005;24:7524–7541. [PubMed]
32. Li Y, Rogulski K, Zhou Q, Sims PJ, Prochownik EV. The negative c-Myc target onzin affects proliferation and apoptosis via its obligate interaction with phospholipid scramblase 1. Mol Cell Biol. 2006;26:3401–3413. [PMC free article] [PubMed]
33. Wu SF, Huang Y, Hou JK, Yuan TT, Zhou CX, Zhang J, Chen GQ. The downregulation of onzin expression by PKCepsilon-ERK2 signaling and its potential role in AML cell differentiation. Leukemia. 2010;24:544–551. [PubMed]
34. Rank RG, Soderberg LS, Barron AL. Chronic chlamydial genital infection in congenitally athymic nude mice. Infect Immun. 1985;48:847–849. [PMC free article] [PubMed]
35. Barteneva N, Theodor I, Peterson EM, de la Maza LM. Role of neutrophils in controlling early stages of a Chlamydia trachomatis infection. Infect Immun. 1996;64:4830–4833. [PMC free article] [PubMed]
36. Lee HY, Schripsema JH, Sigar IM, Murray CM, Lacy SR, Ramsey KH. A link between neutrophils and chronic disease manifestations of Chlamydia muridarum urogenital infection of mice. FEMS Immunol Med Microbiol. 2010;59:108–116. [PubMed]
37. Nelson DE, Virok DP, Wood H, Roshick C, Johnson RM, Whitmire WM, Crane DD, Steele-Mortimer O, Kari L, McClarty G, Caldwell HD. Chlamydial IFN-{gamma} immune evasion is linked to host infection tropism. Proc Natl Acad Sci U S A. 2005;102:10658–10663. [PubMed]
38. Coers J, Bernstein-Hanley I, Grotsky D, Parvanova I, Howard JC, Taylor GA, Dietrich WF, Starnbach MN. Chlamydia muridarum evades growth restriction by the IFN-gamma-inducible host resistance factor Irgb10. J Immunol. 2008;180:6237–6245. [PubMed]
39. Al-Zeer MA, Al-Younes HM, Braun PR, Zerrahn J, Meyer TF. IFN-gamma-inducible Irga6 mediates host resistance against Chlamydia trachomatis via autophagy. PLoS One. 2009;4:e4588. [PMC free article] [PubMed]
40. MacMicking J, Xie QW, Nathan C. Nitric oxide and macrophage function. Annu Rev Immunol. 1997;15:323–350. [PubMed]
41. Yu H, Jiang X, Shen C, Karunakaran KP, Jiang J, Rosin NL, Brunham RC. Chlamydia muridarum T-cell antigens formulated with the adjuvant DDA/TDB induce immunity against infection that correlates with a high frequency of gamma interferon (IFN-gamma)/tumor necrosis factor alpha and IFN-gamma/interleukin-17 double-positive CD4+ T cells. Infect Immun. 2010;78:2272–2282. [PMC free article] [PubMed]
42. Scurlock AM, Frazer LC, Andrews CW, Jr., O'Connell CM, Foote IP, Bailey SL, Chandra-Kuntal K, Kolls JK, Darville T. Interleukin-17 contributes to generation of Th1 immunity and neutrophil recruitment during Chlamydia muridarum genital tract infection but is not required for macrophage influx or normal resolution of infection. Infect Immun. 79:1349–1362. [PMC free article] [PubMed]
43. Su H, Morrison R, Messer R, Whitmire W, Hughes S, Caldwell HD. The effect of doxycycline treatment on the development of protective immunity in a murine model of chlamydial genital infection. J Infect Dis. 1999;180:1252–1258. [PubMed]