Interferon-ε Protects the Female Reproductive Tract from Viral and Bacterial Infection

See allHide authors and affiliations

Science  01 Mar 2013:
Vol. 339, Issue 6123, pp. 1088-1092
DOI: 10.1126/science.1233321

A Role for IFN-ɛ

Type I interferons (IFNs) are critical cytokines involved in host defense against pathogens, particularly viruses. IFN-ɛ is an IFN-like gene encoded within the type I IFN locus in mice and humans whose function has not been characterized. Fung et al. (p. 1088) created mice with a genetic deletion in Ifn-ɛ and found that, like other type I IFNs, IFN-ɛ signals through the IFN-α receptors 1 and 2. However, unlike these other cytokines, which are primarily expressed by immune cells and are induced upon immune cell triggering, IFN-ɛ was expressed exclusively by epithelial cells of the female reproductive tract in both mice and humans and its expression was hormonally regulated. IFN-ɛ–deficient mice were more susceptible to infection with herpes simplex virus 2 and Chlamydia muridarum, two common sexually transmitted pathogens.


The innate immune system senses pathogens through pattern-recognition receptors (PRRs) that signal to induce effector cytokines, such as type I interferons (IFNs). We characterized IFN-ε as a type I IFN because it signaled via the Ifnar1 and Ifnar2 receptors to induce IFN-regulated genes. In contrast to other type I IFNs, IFN-ε was not induced by known PRR pathways; instead, IFN-ε was constitutively expressed by epithelial cells of the female reproductive tract (FRT) and was hormonally regulated. Ifn-ε–deficient mice had increased susceptibility to infection of the FRT by the common sexually transmitted infections (STIs) herpes simplex virus 2 and Chlamydia muridarum. Thus, IFN-ε is a potent antipathogen and immunoregulatory cytokine that may be important in combating STIs that represent a major global health and socioeconomic burden.

Type I interferons (IFNs) are crucial in host defense because of their antipathogen actions and ability to activate effector cells of the innate and adaptive immune responses (1, 2). The type I IFN locus contains genes encoding 13 IFN-α subtypes, IFN-β, and IFN-ω (3) whose promoters contain acute response elements [such as interferon regulatory factors (IRFs) and NF-κB in IFN-β], which ensure rapid induction of these genes by pattern-recognition receptor (PRR) pathways (4, 5). This locus also contains a gene, which we previously designated IFN-ε, but whose function has remained uncharacterized.

Interferon-ε shares only 30% amino acid homology to a consensus IFN-α sequence and to IFN-β. Therefore, we first demonstrated that IFN-ε was a type I IFN by showing that it transduced signals via the Ifnar1 and Ifnar2 receptors (see the supplementary materials and methods) (6). Incubation of recombinant Ifn-ε with bone marrow–derived macrophages (BMDMs) from wild-type (WT) mice induced IFN-regulated genes (IRGs) such as Irf-7 and 2′5′oas (which encodes oligoadenylate synthetase) (Fig. 1, A and B), whereas these IRGs were not induced in BMDMs from Ifnar1- or Ifnar2-deficient mice. Accordingly, Ifn-ε should be classified as a type I IFN.

Fig. 1

Interferon-ε signals through the type I IFN receptor but is not induced by TLR ligands nor regulated by IRFs. (A and B) BMDMs from WT, Ifnar1/, and Ifnar2/ C57BL/6 mice were stimulated with recombinant mouse Ifn-α1, Ifn-β, or Ifn-ε (0.1 μg/ml) for 3 hours. (A) Irf-7 and (B) 2′5′oas expression was measured by quantitative real-time fluorescence polymerase chain reaction (qRT-PCR). Data are expressed as mean ± SEM (error bars) of at least three independent experiments. (C) BMDMs from C57BL/6 WT mice were treated with a range of TLR ligands or transfected with Poly (I:C) and Poly (dA:dT) for 3 hours at 37°C. Ifn-β, Il-6, and Ifn-ε were measured by qRT-PCR. Data are expressed as mean ± SEM of at least three independent experiments. (D) Luciferase reporter plasmids containing Ifn-α, Ifn-β, p125, or Ifn-ε were cotransfected with empty vector or IRF-3, IRF-7, or IRF-5 expression vectors into HEK293 cells. Data are expressed as mean ± SEM. All values are means of at least three independent experiments. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 (unpaired Student's t test).

We next determined whether IFN-ε was induced by PRR pathways. Primary BMDMs, murine embryonic fibroblasts, and the murine macrophage cell line RAW264.7 were treated with synthetic ligands of Toll-like receptors (TLRs) 2, 3, 4, 7/8, and 9; cytosolic DNA sensors or AIM2 inflammasomes potently induced known PRR response genes such as Ifn-β and/or interleukin-6 (Il-6) (79). By contrast, there was no significant change in the expression of Ifn-ε upon stimulation with these activators (Fig. 1C and fig. S1, A and B). Because all PRRs induce type I IFN expression through the activation of the IRF family of transcription factors (5), we then examined whether IRFs could directly regulate the Ifn-ε promoter. IRF-3, IRF-7, and IRF-5 induced promoter activity of Ifn-β, Ifn-α, and p125 (5) luciferase reporters in human embryonic kidney (HEK) 293 cells (Fig. 1D). By contrast, we did not observe any alteration of Ifn-ε promoter activity (Fig. 1D). Semliki Forest virus infection of RAW264.7 cells stimulated the expression of the positive control antiviral response gene 2′5′oas, but not Ifn-ε expression (fig. S1C). Furthermore, Ifn-ε expression was not altered during in vivo infection with herpes simplex virus 2 (HSV-2) or Chlamydia muridarum (see below), nor by stimulation of human endometrial cell lines with PRR ligands (fig. S1D). This lack of regulation of Ifn-ε gene expression by conventional PRR pathways is consistent with the lack of response elements for these pathways [IRFs, NF-κB, STAT (signal transducers and activators of transcription), ISRE (interferon-stimulated response element)] in the Ifn-ε proximal promoter compared with other type I IFN genes (fig. S1E).

Because Ifn-ε was not regulated by PRR pathways, we examined its constitutive expression. The expression of Ifn-α and -β was undetectable in all organs (Fig. 2A). Similarly, the expression of Ifn-ε was not detectable at significant levels in any organ except the uterus, cervix, vagina, and ovary (Fig. 2A). Immunohistochemistry results show that Ifn-ε was expressed in the luminal and glandular epithelial cells of the endometrium (Fig. 2B). In support of these data, the uterine expression levels of Ifn-ε did not differ in NOD (nonobese diabetic)/SCID (severe combined immunodeficient)/IL-2rγ−/− mice [which are deficient in mature T, B, and natural killer (NK) cells] relative to WT mice, indicating that the aforementioned cells do not express detectable levels, nor do they regulate this cytokine (fig. S1F). This finding differs from the expression of conventional type I IFNs, which are usually expressed in hemopoietic cells.

Fig. 2

Interferon-ε is expressed in the FRT in both mice and humans. (A) Mouse organs were harvested and Ifn-ε expression was measured by qRT-PCR, normalized to 18S RNA and presented relative to Ifn-ε expression in kidney. Data are expressed as the mean ± SEM (error bars) of at least three individual mice. (B) Representative images showing Ifn-ε localization in uterine tissue (at estrous stage) of WT and Ifn-ε/ C57BL/6 mice by immunohistochemistry. Scale bars, 50 μm. These images are representative of at least five individual mice. (C and D) Ifn-ε expression was measured by qRT-PCR in mouse uterus at different stages of (C) the estrous cycle and (D) pregnancy. Data are expressed as mean ± SEM of at least three separate experiments. (E) Ifn-ε expression was determined by qRT-PCR in ovariectomized (OVX) mice and OVX mice treated with estrogen (E OVX). Data are expressed as mean ± SEM of at least six individual mice and are representative of at least two separate experiments. (F) A cDNA panel of human tissues was examined for IFN-ε expression by qRT-PCR, and the results were expressed relative to IFN-ε expression in the kidney. (G) Epithelial cells were isolated from endometrial samples of postmenopausal women or those at different stages of the menstrual cycle, and IFN-ε expression was measured by qRT-PCR. Values are presented relative to IFN-ε expression in the human endometrial cell line ECC-1. Data are expressed as mean ± SEM of six individual patient samples. *P < 0.05; **P < 0.01; ***P < 0.001 [(A to F) unpaired Student's t test,(G) Mann-Whitney U test].

We found that Ifn-ε expression varied approximately 30-fold at different stages of the estrous cycle, with the lowest levels during diestrus and the highest at estrus (Fig. 2C). During pregnancy, uterine Ifn-ε expression was dramatically reduced at day 1.5 postcoitus (pc) and was lowest at day 4.5 pc, coincident with the time of embryo implantation (Fig. 2D). Ifn-ε expression was also reduced in pseudo-pregnant mice 4.5 days pc after mating with vasectomized males (Fig. 2D), which suggests that maternal hormones, not the embryo or its products, were required for the reduction in Ifn-ε. In addition, there was a slight increase in expression of Ifn-ε (1.8- to 1.9-fold) 8 hours pc, though expression had returned to normal levels by 16 hours pc, showing that neither seminal fluid nor sperm directly suppress Ifn-ε expression (fig. S1G). Because changes in expression occur after mating with vasectomized or intact males, expression fluctuations are likely to be secondary to physiological and hormonal changes, which are known to be comparable at day 4.5 pc whether or not conception occurs. Together, these data are consistent with Ifn-ε expression being hormonally regulated. To evaluate this finding, we ovariectomized female mice and administered ovarian sex steroid hormones. Estrogen administration induced Ifn-ε expression more than sixfold (Fig. 2E). This hormonal regulation was not observed for expression of other conventional type I IFNs (10).

Expression analysis of a panel of tissues confirmed the lack of basal expression of IFN-ε in all organs in women, with the exception of the endometrium (Fig. 2F). To determine whether human IFN-ε was also regulated in different hormonal states, we tested epithelial cells isolated from uterine endometrium from six women in secretory or proliferative stages of the menstrual cycle or after menopause. IFN-ε expression was highest in the proliferative phase when estrogen levels were high and was approximately 10-fold lower in the secretory phase when estrogen levels were low and progesterone was high. IFN-ε levels were virtually undetectable in samples from postmenopausal women (Fig. 2G) (11). Consistent with the epithelial cell origin of this cytokine, several endometrial cancer–derived cell lines were found to express IFN-ε (fig. S1H).

Next, we generated Ifn-ε/ mice to characterize the pathophysiological functions of this gene (fig. S2, A to E, and table S1). No differences were detected in male and female fertility (fig. S3A), or in the reproductive organs from male and female mice (fig. S3B) and immune organs characterized by immunophenotyping (fig. S3, C to H).

The basal levels of 2′5′oas, Irf-7, and Isg15 were significantly reduced in uteri from Ifn-ε−/− mice, similar to the very low levels observed in Ifnar1/ mice (Fig. 3, A to C), indicating that Ifn-ε did signal in vivo. IRG levels in other organs were the same between WT and Ifnε−/− mice (fig. S3I). Furthermore, this difference in IRG levels resulting from constitutive Ifn-ε expression was similar in magnitude to the induction of these IRGs in WT mice that were given Ifn-α, -β, or -ε intravaginally (fig. S4) and to the degree of altered expression observed after Chlamydia or HSV-2 infection (see below). These data demonstrate that expression of IFN-ε in the female reproductive tract (FRT) is required for maintaining basal levels of IRGs, which play an important role in innate immunity.

Fig. 3

Ifn-ε/ mice are more susceptible to HSV-2 vaginal infection. (A) Isg15, (B) Irf-7, and (C) 2′5′oas expression between WT and Ifn-ε/ C57BL/6 mice was determined by qRT-PCR. The values represent means ± SEM (error bars) of four individual mice. (D to F and H) Mice pretreated with medroxyprogesterone acetate (Depo-Ralovera, Pfizer) at day –5 were infected with HSV-2 (D, E, H) at a level of 2400 or (F) 24 pfu per mouse on day 0. (D) Representative images demonstrating overt genital lesions, redness, and swelling in HSV-2–infected Ifn-ε/ mice at day 7 pi; these qualities are absent in C57BL/6 WT mice. (E)Clinical scores of WT and Ifn-ε/ C57BL/6 mice during the 7-day course of infection. Data are means ± SEM of five individual mice and are representative of at least three separate experiments. (F and G) HSV-2 titers (pfu) from vaginal tissue of WT and Ifn-ε/ C57BL/6 mice infected with (F) 2400 and (G) 24 pfu, respectively, at day 3 pi were determined by titration of clarified vaginal tissue samples on Vero cell monolayers by plaque assay. Data are expressed as mean ± SEM of five individual mice. (H) HSV-2 titers from homogenates of vaginal tissue, spinal cord, and brain stem of infected WT and Ifn-ε/ C57BL/6 mice at day 7 pi were determined as in (F) and (G). Data are expressed as mean ± SEM of five individual mice. *P < 0.05; **P < 0.01 (unpaired Student's t test).

To determine whether Ifn-ε is important in protecting the FRT from viral infection, we examined the effect of genital HSV-2 infection in Ifn-ε−/− mice. After a sublethal dose of a clinical isolate of HSV-2 strain 186 (12), Ifn-ε−/− mice had significantly more severe clinical scores of disease [day 6 and 7 postinfection (pi)] with severe epidermal lesions evident compared with WT mice (Fig. 3, D and E). These effects were observed at virus doses of 24 and 2400 plaque-forming units (pfu) per mouse (Fig. 3, F and G) and were consistent with elevated viral titers in infected vaginal tissues of Ifn-ε−/− mice at day 3 pi, compared with WT animals. At the low dose of 24 pfu, Ifn-ε was protective, as virus was only detectable in the null mice and not in WT animals. In addition, Ifn-ε−/− mice had significantly higher viral titers in the spinal cord and brain stem 7 days pi, consistent with either increased replication or retrograde transport of virus (Fig. 3H). Notably, there was no significant change in the expression of Ifn-ε in the first 3 days after viral infection, consistent with our in vitro data showing that this gene is not pathogen induced (fig. S5A). The susceptibility of Ifn-ε−/− was less than that of Ifnar1−/− mice, which cannot respond to Ifn-α, -β, or -ε (fig. S5B). However, because Ifn-β and IRGs were not induced less in Ifn-ε−/− mice, the protective effects of Ifn-ε in this model of a prevalent sexually transmitted infection (STI) were independent of other type I IFNs (fig. S5, C to F).

We next investigated the role of Ifn-ε in a murine model of FRT infection by Chlamydia, the most prevalent bacterial STI (13, 14). After a sublethal, intravaginal infection of WT and Ifn-ε−/− mice with C. muridarum (15), Ifn-ε−/− mice displayed more severe clinical signs of disease from 7 until 30 days pi (Fig. 4A). More bacteria were detected in vaginal swabs of Ifn-ε−/− mice throughout the course of infection (Fig. 4B). C. muridarum recovery from vaginal lavage 3 days pi in WT mice had not increased from day 1 inoculum levels, but there was a 40-fold increase in the levels of bacteria in Ifn-ε−/− mice (Fig. 4C). We also observed significantly increased levels of Chlamydia at 30 days pi, indicative of increased chlamydial growth in the upper FRT (uterine horns) of Ifn-ε−/− mice compared with very low levels in WT mice (Fig. 4D). This finding in particular indicates that Ifn-ε−/− mice are substantially more susceptible to (and less able to clear) an ascending infection in the FRT than WT mice. Because NK cells have a protective role against this infection (16), we measured their levels at 3 days pi. Notably, both the percentage and total numbers of these cells were decreased in the uteri of Ifn-ε−/− mice (fig. S6, A and B). Importantly, there were no changes in Ifn-ε RNA expression at the early or late stages of the infection (fig. S6C), consistent with our in vitro data showing that Ifn-ε is not regulated by PRR pathways. Furthermore, production of Ifn-β and IRGs was higher than the levels in WT mice (fig. S7, A to D), indicating that the protective effects of Ifn-ε were not solely due to priming for the production of other type I IFNs. To demonstrate that Ifn-ε could directly mediate protection against infection, we observed a dose-dependent reduction in bacteria (Fig. 4E), demonstrating that reconstitution of (progesterone) lowered Ifn-ε levels protected against this bacterial infection.

Fig. 4

Ifn-ε/ mice are more susceptible to Chlamydia muridarum vaginal infection. (A to D) Mice were pretreated with progesterone at day –7 and infected intravaginally with 5 × 104 inclusion-forming units (IFU) of C. muridarum. (A) Clinical scores were recorded daily for 30 days. Data are means ± SEM (error bars) of at least six individual mice. (B) Bacterial recovery from vaginal swabs of WT and Ifn-ε/ C57BL/6 mice at different time points, as determined by qRT-PCR for bacterial major outer membrane protein. Data are means ± SEM of at least six individual mice. (C) Bacterial recovery, as measured by qRT-PCR from vaginal lavage at days 1 and 3 pi. Data are means ± SEM of at least six individual mice. (D) Bacterial 16S RNA from the uterine horns of WT and Ifn-ε/ C57BL/6 mice at 30 days pi was examined by qRT-PCR. Data are means ± SEM of at least six individual mice. (E) WT C57BL/6 mice were pretreated with progesterone at day –7 and treated intravaginally with rIfn-ε (2 or 4 μg) 6 hours before C. muridarum infection. Bacterial recovery from the vaginal lavage at day 3 pi was measured by qRT-PCR. PBS, phosphate-buffered saline. Data are means ± SEM of at least six individual mice. *P < 0.05; **P < 0.01; ***P < 0.001 (unpaired Student's t test).

The distinct properties of IFN-ε, compared with other type I IFNs (table S2), make IFN-ε the only one that protects against Chlamydia, whereas the others exacerbate disease (1720). All type I IFNs protect against HSV-2 infection (21, 22), with IFN-ε likely contributing because its constitutive expression by epithelial cells offers immediate efficacy at the site of first contact of mucosal pathogens. Interestingly, the increased susceptibility to FRT infections of women on progestagen-containing contraception (23, 24) may be explained by the lowering of Ifn-ε levels (fig. S8A) during progestin pretreatment that is required for all FRT infection models (25, 26). The local effect of IFN-ε is supported by our observation that IFN-ε makes no difference in a systemic model (fig. S8, B to D). Consistent with the importance of IFN-ε in FRT immunity, IFN-ε is evolutionarily conserved in eutherian mammals, particularly in residues predicted to contact the two receptor components (fig. S9) (27). Because STIs are major global health and socioeconomic problems, the distinctive regulatory and protective properties of IFN-ε may facilitate the development of new strategies for preventing and treating STIs and, perhaps, other diseases.

Supplementary Materials

Materials and Methods

Figs. S1 to S9

Tables S1 and S2

References (2833)

References and Notes

  1. Acknowledgments: We thank A. Mansell, R. Ferrero, and L. Salamonsen for their contributions; N. Bourke and S. Forster for helpful discussions and reading of the manuscript; K. Fitzgerald for reagents; and C. Berry for assistance with viral plaque assays. The data presented in this paper are tabulated in the main text and in the supplementary materials. This work was supported by funding from Australian National Health and Medical Research Council (P.J.H., N.E.M., P.M.H., J.R., C.E.G., and B.P.), the Australian Research Council (P.J.H., N.E.M., J.R.), the NIH via grant R01 AI053108 (D.J.C.), and the Victorian Government's Operational Infrastructure Support Program. P.J.H., N.E.M., K.Y.F., H.C., S.A.S., and N.D.W. hold International Patent Application number PCT/AU2011/000715, "Use of interferon epsilon in methods of diagnosis and treatment."
View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article