Estrogen Regulates Transcription Factors STAT-1 and NF-κB to Promote Inducible Nitric Oxide Synthase and Inflammatory Responses¹

Inflammation, mediated by the production of inflammatory cytokines, chemokines, and other inflammatory molecules, is an essential body response that functions to eliminate extrinsic assaults. However, exaggerated or prolonged inflammation can lead to severe debilitating or even fatal inflammatory diseases. Thus, a better understanding of the generation and regulation of inflammatory molecules is critical. Current studies from our laboratory and others have revealed that in vivo estrogen exposure promotes inflammatory responses that include enhanced secretion of Th1-related cytokines (IFN-γ, IL-12, IL-1β) and inflammatory chemokines (MCP-1 and MCP-5) and induction of enzymes involved in proinflammation (inducible NO synthase (iNOS)⁴ and cyclooxygenase-2; Refs. 1, 2, 3, 4, 5, 6, 7). However, the molecular mechanism of estrogen-mediated proinflammatory responses remains unclear.

iNOS is the primary isoform present in immune cells that is capable of producing high levels of NO (8, 9). NO is a unique signaling messenger, which plays an important role in the control of a number of immune defenses against microbes, viruses, and parasites (9, 10, 11). However, inappropriate or excessive NO production is detrimental and has been implicated in the pathogenesis of many chronic inflammatory diseases including asthma, multiple sclerosis, hypercholesterolemia, atherosclerosis, arthritis, and systemic lupus erythematosus (12, 13, 14). Considering the versatile role of NO in the immune system and the fact that its aberrant production can be detrimental, it is important to understand the mechanism of estrogen-mediated promotion of iNOS and NO in splenocytes.

Multiple transcription factors have been shown to regulate iNOS gene expression in response to different stimulations. Activation of the NF-κB pathway and/or the IFN-γ-JAK-STAT-1 pathway has been shown to be essential for iNOS induction in various cell types (15, 16, 17, 18). Inhibition of STAT-1 or NF-κB signaling decreased LPS- and/or IFN-γ-induced iNOS gene expression (17, 19, 20, 21). Further, a role of IFN regulatory factor (IRF)-1 and IFN Consensus Sequence Binding Protein (ICSBP or IRF-8) in iNOS gene expression is suggested by the observations that IFN-γ-stimulated iNOS was reduced in both IRF-1^−/− and ICSBP^−/− mouse macrophages (22, 23, 24). Nevertheless, other cell types from IRF-1^−/− mice, such as chondrocytes and pancreatic islet cells, have exhibited normal iNOS induction (24, 25). This suggests that the transcription factors required for induction of iNOS may vary among cell types. In addition, studies have shown that transcription factors such as C/EBPβ, CREB, and AP-1 are also involved in iNOS regulation (17). Generally, maximal induction of iNOS requires synergistic activation of multiple transcription factors (16, 26, 27).

The focal point of these studies was to reveal the molecular mechanism of estrogen-mediated promotion of iNOS, with particular emphasis on the critical role of the key transcription factors STAT-1 and NF-κB on iNOS gene expression in splenocytes. We found that estrogen enhances STAT-1 DNA-binding activity without increasing the expression of phosphorylated and total STAT-1. Further, our data revealed that estrogen induces serine protease-mediated proteolysis of STAT-1 and NF-κB, which may alter and enhance the activity of these transcription factors. Moreover, we demonstrate that modified STAT-1 and NF-κB proteins in splenocytes from estrogen-treated mice have differential roles in regulating different inflammatory molecules such as iNOS, IFN-γ and MCP-1. Together, our studies add new knowledge with regard to the mechanism of estrogen-mediated regulation of inflammation.

Wild-type male C57BL/6 mice, 3–4 wk old, were purchased from Charles River Laboratories and housed in the animal facility at the Center for Molecular Medicine and Infectious Diseases. As is the long-standing standard practice in our laboratory, ∼1 wk after acclimatization, mice were orchiectomized and surgically implanted with Silastic capsules containing 17β-estradiol (estrogen; Sigma-Aldrich) or empty (placebo control) Silastic implants (2, 6, 28, 29, 30, 31). These estrogen implants slowly release estrogen over many weeks, and serum estrogen concentrations (222.67 ± 41.9 pg/ml at 7 wk) achieved by this method of implantation have been previously published (6). Mice were fed a commercial pellet diet devoid of estrogenic hormones (7013 National Institutes of Health-31 Modified 6% Mouse/Rat Sterilizable Diet; Harlan-Teklad). The Institutional Animal Use and Care Committee at the Virginia Polytechnic Institute and State University approved all animal procedures.

Mice were euthanized 6–8 wk after implantation. Splenocytes were isolated and cultured using procedures that have been described in detail before (5, 6, 28, 29, 31). The splenocytes were adjusted to 5 × 10⁶ cells/ml before plating. RPMI 1640 (phenol red free) supplemented with 10% charcoal-stripped FBS (Atlanta Biologicals), 2 mM l-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 1% nonessential amino acids (Mediatech) was used for splenocyte culture. For Con A and IFN-γ stimulation, equal volumes of 10 μg/ml Con A (Sigma-Aldrich) or 10 ng/ml rIFN-γ (eBioscience) in medium were added to seeded cells (final concentrations at 5 μg/ml and 5 ng/ml, respectively). For nonstimulated cells (media controls), equal volumes of complete media were added to seeded cells. For JAK2 kinase and NF-κB inhibitor studies, the splenocytes were pretreated with vehicle (DMSO or ethanol), 20 μM AG490 (a selective JAK2 tyrosine kinase inhibitor; Sigma-Aldrich), or 10 μM A77 1726 (a selective NF-κB inhibitor; Axxora) for 1 h before stimulation with Con A in the presence of the inhibitor or vehicle. For the serine protease inhibitor study, 50 μM AEBSF (Calbiochem) was added to plated cells with Con A at the same time.

The murine macrophage cell line, RAW264.7 (American Type Culture Collection), was cultured in RPMI 1640 supplemented with 10% charcoal-stripped FBS (Atlanta Biologicals), 2 mM l-glutamine, 100 IU/ml penicillin, and 100 μg/ml streptomycin (Mediatech).

EMSAs were performed as described in detail in our previous report (6). Briefly, 5′-biotin-labeled DNA-binding probes for STAT-1 (IFN-γ-activated sequence (GAS) probe, forward: GATCGTGATTTCCCCGAAATGACG and reverse complementary: CGTCATTTCGGGGAAATCACGATC) and NF-κB (κB DNA probe, forward: GATCGAGGGGACTTTCCCTAGC and reverse complementary: GCTAGGGAAAGTCCCCTCGATC) were synthesized by Integrated DNA Technologies and annealed. For the binding reaction, 5 μg of nuclear extract proteins (2–3 μl) were incubated with 20 fmol of biotin-labeled dsDNA probe at room temperature for 25 min in 20 μl of 1× binding buffer (12 mM HEPES (pH 7.5), 80 mM NaCl, 5 mM DTT, 5 mM MgCl₂, 0.5 mM EDTA, 1 μg of poly(deoxyinosinate-deoxycytidyate), and 5% glycerol). The binding complexes were separated on 5% Tris-borate-EDTA polyacrylamide gels (Bio-Rad), transferred to a positively charged nylon membrane (Pierce), and then detected using a Chemiluminescent Nucleic Acid Detection Module (Pierce). The images were captured and the signal intensities were analyzed using a Kodak Image Station 440.

Western blotting was used to analyze the expression of proteins in whole-cell, nuclear, and cytoplasmic extracts as we described previously (5, 6). For whole-cell extracts, cell pellets were washed with ice-cold PBS and then lysed with CelLyticM Cell Lysis Reagent (Sigma-Aldrich). Nuclear and cytoplasmic extracts were prepared using NE-PER nuclear and cytoplasmic extraction reagents (Pierce). The blot images were captured and analyzed using a Kodak Image Station 440. Abs against NF-κB p65 (c-20 and F-6), p50 (NLS), IRF-1, and total STAT-1 were purchased from Santa Cruz Biotechnology. Abs against phosphorylated STAT-1 (Tyr⁷⁰¹ and Ser²⁷⁰) were purchased from Cell Signaling. The Ab against β-actin (loading control) was obtained from Abcam.

Griess assays were used to measure nitrite, an indicator of NO production in the culture medium as described in detail before (5). The levels of IFN-γ and MCP-1 in culture supernatants were determined by ELISAs as described in our previous studies (2, 31, 32).

A nucleofector device and mouse macrophage nucleofector kit (Amaxa) were used to transfect siRNA oligonucleotides to mouse splenocytes. All of the siRNA oligonucleotides were purchased from Dharmacon. Briefly, 1.5 × 10⁷ freshly isolated splenocytes were pelleted, resuspended in 100 μl of mouse macrophage nucleofector solution, and mixed with 3.5 μg of siRNA oligonucleotidess. The sample was transferred to a cuvet and then transfected using the optimal nucleofector program m001. Twenty-four hours after transfection, the splenocytes were stimulated with Con A for 24 or 48 h. The cells were collected for Western blot analysis. Supernatants were used for measurement of NO, IFN-γ, and MCP-1.

All values are means ± SEM. To assess statistical significance between placebo- and estrogen-treated mice, two-tailed unpaired t tests were performed using GraphPad InStat version 3.0a for Macintosh (GraphPad Software). For evaluation of the effect of a specific inhibitor or siRNA, the levels of nitrite, IFN-γ, and MCP-1 in specific inhibitor-treated or siRNA-transfected cells were shown as a percentage of the level of the corresponding control (regarded as 100%). Paired t tests were performed to assess the statistical significance of the effect of inhibitors and specific siRNA.

In our previous report, we have shown that in vivo estrogen treatment of mice markedly augmented the ability of Con A-activated splenocytes to release higher levels of IFN-γ and NO (2, 5). Further, by using IFN-γ^−/− knockout mice, we have demonstrated that estrogen-mediated up-regulation of iNOS and NO is IFN-γ dependent (5). In this report, we investigated whether induction of iNOS and NO by Con A is mediated through IFN-γ activation of the JAK-STAT-1 signaling pathway. As we previously reported (5), NO production in estrogen-treated splenocytes was significantly higher than that of splenocytes from placebo-treated mice (11.03 ± 2.99 μM vs 5.39 ± 1.77 μM; p < 0.001) following stimulation with Con A for 48 h (Fig. 1,A). Inhibition of JAK-STAT-1 signaling with a specific JAK2 tyrosine kinase inhibitor, AG490, resulted in ∼45% and 60% reduction in NO production in splenocytes from placebo- and estrogen-treated mice, respectively (Fig. 1 B).

To further determine the role of STAT-1 in Con A-stimulated iNOS in splenocytes, we used STAT-1 siRNA to specifically inhibit STAT-1 gene expression. The STAT-1 siRNA has no effect on the expression of other STATs such as STAT-3 and STAT-4 (supplemental Fig. 1).⁵ As indicated, inhibition of STAT-1 protein expression markedly decreases Con A-stimulated iNOS protein expression (Fig. 1,C), and NO production (Fig. 1 D) in cells from both placebo- and estrogen-treated mice. Taken together, our data clearly indicate that the IFN-γ-JAK-STAT-1 signaling pathway plays an important role in Con A-stimulated iNOS expression and NO production in mouse splenocytes.

Given the critical role of STAT-1 in regulating the expression of iNOS, we next determined whether in vivo estrogen treatment up-regulated iNOS expression in splenocytes by increasing the expression and/or activity of STAT-1. The Western blot data from analysis of whole-cell extracts indicated that there was constitutive phosphorylation of STAT-1 at the Ser⁷²⁷ site (p-STAT-1-Ser), but not at the Tyr⁷⁰¹ site (p-STAT-1-Tyr) in freshly isolated splenocytes (Fig. 2,A). Following Con A stimulation, there was a substantial increase of p-STAT-1-Tyr in splenocytes stimulated with Con A for 3 h from both placebo- and estrogen-treated mice. The level of p-STAT-1-Tyr was decreased with increased stimulation time at 6 and 24 h of Con A stimulation. Overall, the Western data with whole cell extracts revealed that the expression level of phosphorylated STAT-1 and total STAT-1 in splenocytes from estrogen-treated mice is comparable with that from placebo-treated mice (Fig. 2 A).

Once activated, phosphorylated STAT-1 dimerizes, translocates into the nucleus, binds to specific promoters, and induces the expression of target genes. We therefore determined whether estrogen treatment increases the nuclear expression of phosphorylated STAT-1. The data indicated that the expression of phosphorylated STAT-1 is not increased in the nuclei of splenocytes from estrogen-treated mice when compared with that from placebo controls. Indeed, the densitometry analysis of Western data indicated that the expression level of phosphorylated STAT-1 was decreased in splenocytes from estrogen-treated mice at 6 and 24 h of Con A stimulation (Fig. 2,B). In cells from estrogen-treated mice, STAT-1 was modified and truncated STAT-1 forms were frequently observed (Fig. 2,B). This posttranslational modification of STAT-1 seems to be a dynamic process given that in some samples, full-length and truncated STAT-1 forms coexisted. Fig. 2 B depicts two representative STAT-1 expression patterns in cells from estrogen-treated mice. Also, the modification was only observed for nuclear STAT-1, not cytoplasmic STAT-1 (supplemental Fig. 2).

We then determined whether this posttranslational modification of STAT-1 in the nuclei would affect the DNA-binding activity of STAT-1. Experiments including EMSAs with GAS DNA and κB DNA competitors, supershift assays, and in vitro DNA-binding assays revealed that Con A activated NF-κB proteins bind to the GAS DNA oligonucleotides specifically. We could not separate formed p65/p50 GAS DNA-binding complexes from STAT-1 GAS DNA-binding complexes in cells from placebo-treated mice (supplemental Fig. 3). Therefore, to exclusively detect the STAT-1-binding signal intensity, we included 1 pmol of unlabeled κB DNA oligonucleotides in the binding reaction to compete out the binding of NF-κB protein to GAS DNA probes. As shown in Fig. 2,C, there was no STAT-1 DNA binding in freshly isolated cells (t₀). Con A stimulation induced STAT-1 DNA binding complexes in nuclei of splenocytes from both placebo- and estrogen-treated mice. Although the nuclear level of phosphorylated STAT-1 was not increased in the samples from estrogen-treated mice after 3 h of Con A stimulation, the STAT-1 binding signal intensity was significantly stronger when compared with that from placebo-treated mice (Fig. 2, C and D). Because the STAT-1 in cells from estrogen-treated mice was truncated (Fig. 2,B), the STAT-1 DNA-binding complex in cells from estrogen-treated mice was slightly smaller than that from placebo-treated mice (Fig. 2,C). By 24 h of Con A stimulation, the binding activity of STAT-1 completely disappeared (Fig. 2,C), corresponding with the decrease of phosphorylated STAT-1 in the nuclei of splenocytes (Fig. 2 B). Overall, our data show that although in vivo estrogen treatment does not increase the expression level of phosphorylated and total STAT-1, it does induce posttranslational modification of nuclear STAT-1 and enhances the DNA-binding activity of STAT-1.

We next determined whether activation of IFN-γ-JAK-STAT-1 signaling alone is sufficient to induce iNOS and NO in splenocytes. As a control, we included the murine macrophage cell line, RAW264.7. Unlike RAW264.7 cells, there was no NO production in IFN-γ-stimulated splenocytes from either placebo- or estrogen-treated mice (Fig. 3,A). Further, Western blot analysis indicated that there is also no iNOS protein expression in whole-cell extracts from IFN-γ-stimulated splenocytes (Fig. 3,B). The activation of IFN-γ-STAT-1 signaling in IFN-γ-stimulated splenocytes from either placebo- or estrogen-treated mice was confirmed by the expression of IRF-1 (Fig. 3,B), phosphorylated STAT-1 (Fig. 3,C), and STAT-1 DNA-binding activity (Fig. 3,D), which paralleled those from Con A-stimulated cells. Because we have shown that estrogen inhibited the expression of IRF-1 (28) and enhanced STAT-1 DNA-binding activity (Fig. 2,C), as expected, we observed an inhibition of IRF-1 expression and increased STAT-1 DNA-binding activity in IFN-γ-stimulated cells from estrogen-treated mice compared with that from placebo-treated mice (Fig. 3, B and D). The above data clearly indicated that although it is critical, activation of the IFN-γ-STAT-1 signaling pathway alone is not sufficient to induce the expression of iNOS in splenocytes. Thus, the IFN-γ-JAK-STAT-1 signaling pathway is likely to act synergistically with other signaling pathway(s) to induce iNOS and NO in Con A-activated mouse splenocytes.

In addition to STAT-1, the transcription factor NF-κB is also critical for iNOS induction (17). A previous report from our laboratory has shown that estrogen enhances NF-κB signaling and that activated NF-κB proteins bind to the promoter of the iNOS gene (6). As expected, inhibition of NF-κB signaling with a selective NF-κB inhibitor, A77 1726, resulted in a ∼40% reduction of NO production in splenocytes from both placebo and estrogen-treated mice (Fig. 4,A). Further, inhibition of NF-κB p65 expression with p65 siRNA significantly decreased the production of Con A-stimulated NO in splenocytes from both placebo- and estrogen-treated mice (Fig. 4,B). There is also decreased induction of NO in p50 siRNA-transfected splenocytes from estrogen-treated mice (Fig. 4,B). Western blot analysis further confirmed that the expression of iNOS protein was inhibited in p65 siRNA-transfected splenocytes (Fig. 4 C). These data indicated that the NF-κB signaling pathway plays an important role in the regulation of Con A-stimulated iNOS-NO in mouse splenocytes.

In addition to iNOS, we have shown that estrogen up-regulated other inflammatory molecules such as IFN-γ and MCP-1 (2, 6, 31, 32). Here, we found that inhibition of either NF-κB or STAT-1 decreased Con A-stimulated IFN-γ in splenocytes from placebo-treated mice. In cells from estrogen-treated mice, inhibition of NF-κBp65, but not STAT-1, significantly decreased Con A-induced IFN-γ (Fig. 5,A). These data suggested that NF-κB, not STAT-1, plays an important role in regulation of Con A-stimulated IFN-γ in splenocytes from estrogen-treated mice. Impressively, although NF-κB signaling has also been shown to play an important role in regulating the expression of MCP-1 (33, 34), inhibition of NF-κBp65 did not decrease Con A-induced MCP-1 in splenocytes from estrogen-treated mice. Rather, there is increased MCP-1 production in NF-κB p65 siRNA-transfected cells from estrogen-treated mice (Fig. 5,B). Further, inhibition of STAT-1 also increased Con A-stimulated MCP-1 production in splenocytes from estrogen-treated mice (Fig. 5 B). The effect of STAT-1 and NF-κB on MCP-1 could not be fully determined in placebo-treated mice because the Con A-stimulated MCP-1 level was very low, and hence MCP-1 could not be measured in all placebo samples. These data imply differential roles of STAT-1 and NF-κB in the regulation of MCP-1 in splenocytes from estrogen-treated mice. Overall, our data showed that STAT-1 and NF-κB signaling in splenocytes from estrogen-treated mice have differential regulatory roles in regulating three inflammatory molecules, NO, IFN-γ, and MCP-1.

Our data indicated that estrogen induced a posttranslational modification of nuclear STAT-1, which resulted in truncated STAT-1. Proteolytic processing has been shown to play an important role in regulation of STAT protein-mediated signaling (35). Here, we found that selective inhibition of the activity of serine proteases with AEBSF completely abolished the truncated STAT-1 isoforms in cells from estrogen-treated mice (Fig. 6,A). Moreover, we found that STAT-1 binding activity is obviously decreased in AEBSF-treated splenocytes from estrogen-treated mice, and there was a concomitant reappearance of full-length STAT-1 (Fig. 6). However, AEBSF treatment has no effect on the expression and binding activity of STAT-1 in cells from placebo-treated mice.

Additionally, we have recently shown that in vivo estrogen treatment increases NF-κB transcriptional activity even though it markedly inhibits nuclear expression of NF-κB p65, c-Rel, and RelB proteins (6). In addition, estrogen also modifies p50 protein in nuclei as indicated by truncated p50 protein (6). Here, we found that the truncated p50 protein had disappeared and been replaced with a p50 protein of normal size in AEBSF-treated cells from estrogen-treated mice (Fig. 7,A). Surprisingly, we found that NF-κBp65 was also evident in the nuclei of AEBSF-treated cells from estrogen-treated mice (Fig. 7,A). Moreover, p65/p50 κB DNA-binding complexes, which were inhibited by in vivo estrogen treatment, were evident in AEBSF-treated cells from estrogen-treated mice (Fig. 7,B). The small binding complexes (referred to as c and d; Ref. 6), which were observed only in cells from estrogen-treated mice, had disappeared and were no longer observed in AEBSF-treated cells. By using a p65 Ab (F-6), which recognizes the N-terminal sequence of p65, a C-terminal truncated p65 isoform (p65s, ∼35 kDa) in the nuclei of splenocytes from estrogen-treated mice was evident (Fig. 7,C). Inhibition of serine proteases with AEBSF resulted in noticeable disappearance of p65s, which was replaced with full-length p65 (Fig. 7 C).

Further, we investigated the effect of inhibition of serine protease on Con A-stimulated inflammatory molecules in splenocytes. As indicated, with the inhibition of serine protease, we observed a significant decrease of Con A-stimulated NO and iNOS protein expression in cells from estrogen-treated but not placebo-treated mice (Fig. 8, A and B). Inhibition of serine protease significantly decreased IFN-γ production in cells from both placebo- and estrogen-treated mice (Fig. 8,C). In contrast to iNOS and IFN-γ, with the inhibition of serine protease, Con A-stimulated MCP-1 in splenocytes from estrogen-treated mice was markedly increased (Fig. 8 D). Taken together, our data revealed that in vivo estrogen treatment induced serine protease-mediated proteolysis of nuclear STAT-1 and NF-κB, which contributed to increased production of NO and IFN-γ, but not MCP-1, in Con A-activated splenocytes.

Given the pivotal role of estrogen in immune modulation and in gender differences in immunity and autoimmunity, there is a need to better understand the mechanisms of estrogen regulation of inflammatory immune responses. Estrogen has been shown to exhibit both proinflammatory and anti-inflammatory immune responses in different studies. The divergent effects of estrogen on inflammatory responses could depend on whether estrogen treatment is in vivo or in vitro, length of estrogen exposure time, the timing of estrogen treatment (i.e., with regard to age of the animal and its maturity), the dose of estrogen, the species or tissue type examined, and stimuli used (3, 8, 36, 37). Contradictory to the finding that estrogen inhibits the inflammatory protein iNOS in most in vitro studies (19, 38, 39), in vivo studies indicate that estrogen increases the expression of iNOS in splenocytes (4, 5), peritoneal macrophages (3), rat uterus (40), and ovine coronary arteries (41). In contrast to the finding that in vitro estrogen treatment inhibited NF-κB signaling, which resulted in decreased inflammation in macrophages (19), we have shown that in vivo estrogen enhanced NF-κB transcription activity and production of inflammatory molecules (6). These reports suggest that a different molecular mechanism underlies the effect of short-term in vitro and relative long-term in vivo estrogen treatment on inflammatory responses. In this study, we delineated the molecular mechanisms underlying estrogen-mediated promotion of inflammatory responses in splenocytes. Primarily, we focused on estrogen regulation of STAT-1 and NF-κB, two key transcription factors that are involved in the generation of inflammatory molecules.

Although estrogen up-regulated the expression of IFN-γ in Con A-activated splenocytes, it inhibited the expression of IRF-1, a transcription factor induced by IFN-γ and known to play an important role in the regulation of IFN-γ-induced genes such as iNOS (28). In the present study, we demonstrated that IFN-γ-JAK-STAT-1 signaling is essential, but not sufficient, to induce iNOS in Con A-stimulated splenocytes. Activation of both IFN-γ-JAK-STAT-1 and NF-κB signaling is necessary for iNOS and NO production in Con A-activated mouse splenocytes. Inhibition of either IFN-γ-JAK-STAT-1 or NF-κB signaling significantly decreases the expression of iNOS in Con A-activated splenocytes (Figs. 1 and 4).

Increased IFN-γ production in splenocytes from estrogen-treated mice did not enhance the expression of phosphorylated and total STAT-1, a key mediator of IFN-γ signaling (Fig. 2, A and B). However, we found that nuclear STAT-1 in splenocytes from estrogen-treated mice is truncated and STAT-1 DNA binding activity was enhanced (Fig. 2). Thus far, truncated STAT isoforms generated by proteolytic processing have been identified for STAT-3, STAT-5, and STAT-6 (35, 42). Here, we are the first to report that in vivo estrogen treatment induces serine protease-mediated proteolysis of nuclear STAT-1, which resulted in truncated STAT-1 in splenocyte nuclei. Moreover, our data have suggested that this modification of nuclear STAT-1 attributes to increased STAT-1 DNA binding activity in splenocytes from estrogen-treated mice given that there are normal sized STAT-1 and decreased STAT-1 DNA-binding activity in AEBSF-treated cells (Fig. 6). Given that the expression of truncated STAT proteins has been potentially linked to the pathology of human diseases (42), it is important to further investigate whether estrogen induced truncation of STAT-1 is related to estrogen-mediated immune disorders.

In addition to STAT-1, the nuclear expression of NF-κBp65 and p50, which are blocked and truncated, respectively, in splenocytes from estrogen-treated mice, was also restored in AEBSF-treated cells (Fig. 7,A). We previously suspected that the absence of nuclear expression of NF-κB p65 in splenocytes from estrogen-treated mice could be caused by inhibition of nuclear translocation given that these proteins are still present in the cytoplasm (6), and estrogen has been shown to control p65 intracellular transportation by targeting microtubule-associated transport systems (19). However, the reappearance of nuclear expression of NF-κB p65 in AEBSF-treated cells suggests another possibility that nuclear p65 may be proteolyzed by serine protease. Additional experiments with a different NF-κB p65 Ab (F-6), which recognizes the N-terminal sequence of p65, confirmed the existence of truncated p65 isoforms (p65s) in nuclei of splenocytes from estrogen-treated mice (Fig. 7 C). Moreover, the nuclear expression levels of c-Rel, RelB, and IRF-1, which were inhibited by estrogen, were also increased in AEBSF-treated splenocytes from estrogen-treated mice and were comparable with those observed in cells from placebo-treated mice (supplemental Fig. 4). These data indicated that in vivo estrogen treatment could modify multiple transcription factors by inducing serine protease-mediated proteolysis.

Because the inhibition of NF-κB p65 and STAT-1 with their corresponding siRNA decreased iNOS expression in cells from estrogen-treated mice, it is possible that truncated STAT-1 and p65s isoforms still play positive regulatory roles for iNOS gene expression. At this time, we are not clear how truncated STAT-1 has stronger DNA-binding activity than full-length STAT-1. In addition, with the complete deletion of the transactivation domain in the C terminus, how p65-short isoforms play positive regulatory roles in splenocytes from estrogen-treated mice is also not apparent. A recent study revealed that a truncated p65 isoform (p35 RelA) was induced in in vitro Leishmania-infected macrophages and subsequently formed p35RelA/p50 dimers had transcriptional activity (43). As mentioned in our previous report (6), there is an unknown, small p50 containing NF-κB binding complex d in Con A-activated splenocytes from estrogen-treated mice (Fig. 7,B). This binding complex d was not observed in AEBSF-treated splenocytes (Fig. 7 B). Whether this NF-κB binding complex d is formed by p65s (∼35 kDa) and p50 is not known and requires further investigation. In addition, we have shown that in vivo estrogen exposure induced the expression of a transcriptional coactivator, B cell lymphoma 3 (Bcl3), which binds to the iNOS promoter with NF-κBp50 in Con A-activated splenocytes from estrogen-treated mice (6). Bcl3 can bind to p50/p50 and p52/p52, which lack transactivation domains and confer the ability of these homodimers to induce NF-κB-responsive genes (44, 45). Therefore, it is possible that p65s exerts its positive regulatory activity by interaction with p50 and Bcl3. However, additional experiments are required to confirm this assumption.

In contrast to decreased iNOS and NO production, we observed increased MCP-1 induction in p65 siRNA and STAT-1 siRNA-transfected cells from estrogen-treated mice (Fig. 5). Consistent with this finding, although the induction of iNOS, NO, and IFN-γ was inhibited, MCP-1 induction was enhanced when serine protease was inhibited in cells from estrogen-treated mice (Fig. 8). These data suggested that truncated STAT-1 and p65 could also play negative regulatory roles for certain inflammatory molecules such as MCP-1. The complete understanding of the differential roles of modified STAT-1 and p65 in regulation of different inflammatory molecules in splenocytes from estrogen-treated mice requires further investigation. Considering that inflammatory gene expression involves multiple transcription factors and/or coactivators, it is possible that estrogen modifies the interactions between these transcription factors/coactivators, which alters the activity of transcriptional complexes and results in either increased or decreased inflammatory gene expression. It is also possible that estrogen treatment regulates transcription factors other than STAT-1 and NF-κB to enhance the induction of MCP-1. In concurrence with this, we found that estrogen treatment increases the expression and binding activity of C/EBPβ (supplemental Fig. 5), another important transcription factor that regulates MCP-1, iNOS, and many other inflammatory molecules (17, 46, 47, 48).

Our data suggest that serine protease plays an important role in regulating inflammatory responses by proteolyzing transcription factors, and that this is associated with estrogen exposure. Nevertheless, the underlying mechanism by which estrogen induces serine protease activity is unknown. In addition, because serine proteases play important roles in inflammatory responses, as well as in innate and adaptive immunity (49), it is worthwhile to further investigate which type of serine protease was induced by estrogen, and whether estrogen-induced serine protease is directly involved in regulation of inflammatory responses. Given the increasing importance of immune tissue-derived iNOS, IFN-γ, and MCP-1 in health and disease, studies on estrogen-induced regulation of these proinflammatory molecules might offer a better understanding of diseases and aid in devising new therapeutic interventions for estrogen-related immune disorders.

We thank Mr. Carmine Graniello for his technical support. We also thank Mr. Peter Jobst, Ms. Connie Kingrea, and other animal care staff members.

The authors have no financial conflict of interest.