Estrogen replacement regimen and brain infusion of lipopolysaccharide differentially alter steroid receptor expression in the uterus and hypothalamus
- First Online:
- Cite this article as:
- Marriott, L.K., McGann-Gramling, K.R., Hauss-Wegrzyniak, B. et al. Endocr (2007) 32: 317. doi:10.1007/s12020-008-9038-1
The regimen of estrogen replacement can alter the consequences of estrogen therapy and stressors. To determine the long-term effects and interaction of these systems on the brain and periphery, adult female rats were infused with lipopolysaccharide (LPS) into the fourth ventricle of the brain for 4 weeks, and ovariectomized rats were administered either constant or pulsed regimens of estrogen replacement (17β-estradiol) until sacrifice at 8 weeks. Constant, but not pulsed, estrogen replacement reduced ERα and increased HSP90, HSP70, and PRB uterine protein levels. Both estrogen regimens increased ERβ, HSP27, and PRA uterine proteins. Both regimens reduced hypothalamic levels of ERα, but not ERβ, HSP, or PR. No changes were observed in the hippocampus. Long-term brain infusion of LPS activated microglia and reduced body weight, but did not alter corticosterone or nitrotyrosine levels. LPS infusion into intact rats suppressed uterine weight, increased ERα and decreased HSP90 in the uterus. LPS did not alter uterine weight in ovariectomized rats treated with constant or pulsed estrogen. Together, these data suggest the timing of estrogen replacement and neuroinflammatory stressors can profoundly affect uterine and hypothalamic steroid receptor expression and may be important parameters to consider in the post-menopausal intervention with estrogen.
KeywordsEstrogen receptor (ER; ERα (ESR1), ERβ (ESR2))Heat shock protein (HSP; HSP90 (HSPAA1), HSP70 (HSPA1B), HSP27 (HSPB2))Progesterone receptor (PGR; PRA, PRB)Lipopolysaccharide (LPS)Estrous cycleFluctuating regimen
Hormone replacement is used to treat undesirable symptoms associated with menopause and has been explored as a possible preventative therapy for diseases such as Alzheimer’s disease, stroke and cardiovascular disease (for review ). Results from observational studies have conflicted with randomized controlled trials in all three disease states (for review ), which has raised caution concerning the use of estrogens and progestins. Clinical trials, meta-analyses, and animal studies have attempted to reconcile these findings to determine the factors influencing the outcome of hormone therapy in post-menopausal women.
The timing of estrogen initiation and replacement regimen appear to be two factors that contribute significantly to the efficacy of hormone therapy . Initiating hormone therapy closer to menopause has been explored as a means of altering the risk of cardiovascular disease . Studies in rats support the importance of estrogen timing as constant estradiol replacement enhanced working memory when initiated immediately after ovariectomy but not after prolonged hormone deprivation . Thus, the timing of constant hormone therapy initiation may be important for prevention, but does not appear to be effective if disease processes have already begun (for review ).
The regimen of estrogen replacement also appears to play a role and studies have begun to address whether pulsed administration of estrogens may be more efficacious [6–8]. Quality of life in women is improved by pulsed, but not constant estrogen therapy . Moreover, pulsed estrogen replacement reversed cognitive impairments in primates and ovariectomized rats, whereas constant estrogen had no effect [10, 11]. The differential response produced by constant and pulsed estrogen regimens may be mediated, in part, by the ability of pulsed estrogen to better activate rapid signaling pathways important for neuroprotection and cognition .
Estrogen receptors (ER) mediate many of these rapid signaling pathways and ERα plays an important role in protecting the brain, cardiovascular and immune systems from inflammatory insults [13–15]. Brain inflammation is a component of several diseases present in the post-menopausal population and hormones may alter disease progression through their effects on inflammation (for review ). For example, ERs are expressed on inflammatory cells, such as microglia and monocytes, and estrogen can block inflammatory signaling and gene expression induced by inflammatory stressors in the brain, such as lipopolysaccharide (LPS; for review ). In turn, inflammatory processes may alter the efficacy of hormone therapy in post-menopausal women . Thus, the regimen of constant estrogen replacement or its combined presence with inflammation may underlie some of the differences observed between experimental animal models and recent clinical trials.
We previously reported that infusion of LPS into the female rat brain simulates a low-grade neuroinflammatory response which can alter the effects of estrogen replacement measured after 6 days . Brain infusion of LPS produced an unexpected increase in uterine weight in rats treated with constant, but not pulsed, estrogen that was concomitant with a 90% reduction in ERα, consistent with reports that ERα knockout mice may be more sensitive to inflammatory insults . Reductions in ERα combined with HPA axis-induced elevations in corticosterone or progesterone are thought to play a role in stress-induced increases in uterine weight [19, 20]. Thus, we hypothesized that LPS stimulates the HPA axis to produce elevated progesterone and corticosterone, which can enhance uterine growth if uterine ERα levels are suppressed. To test whether activation of the HPA axis is necessary to increase uterine growth, the current study infused LPS into the brain for several weeks to maintain a neuroinflammatory reaction (i.e. activated microglia) without the involvement of the HPA axis, which returns to baseline in response to chronic stressors. Thus, despite reduced levels of uterine ERα, we hypothesized that brain inflammation would be insufficient to increase uterine weight if the HPA axis was no longer activated. Moreover, we hypothesized that if constant estrogen can alter ERα levels, longer durations of replacement would be necessary to reduce ERα levels in brain regions important for endocrine homeostasis or learning and memory, such as the hypothalamus or hippocampus, respectively. The purpose of the current study was to compare the long-term effects (8 weeks) of constant versus pulsed estrogen replacement on steroid-related protein levels and determine whether inflammatory stressors in the brain alters the physiological consequences of these regimens.
Downregulation of ERα in the uterus and hypothalamus by constant estrogen
Biochemical markers after 8 weeks of treatment
Circulating estradiol (pg/ml)
Circulating progesterone (ng/ml)
Estrogen binding sites in the hippocampus (fmol/mg protein)
9.0 ± 4.0a
26.9 ± 5.9a
30.1 ± 5.2
20.8 ± 6.6a
15.6 ± 4.3a
27.2 ± 4.9
18.3 ± 5.3a
7.9 ± 3.0b
20.7 ± 5.5
7.5 ± 3.0a
6.2 ± 1.2b
29.8 ± 4.7
12.4 ± 7.0a
6.6 ± 1.5b
29.3 ± 5.5
11.9 ± 2.4a
4.6 ± 0.9b
27.5 ± 5.2
27.9 ± 6.1b
5.7 ± 0.7b
19.4 ± 5.2
11.0 ± 1.8a
5.2 ± 1.0b
31.1 ± 4.2
27.0 ± 5.5b
8.9 ± 2.9b
33.7 ± 5.5
31.5 ± 6.5b
8.7 ± 1.5b
31.3 ± 4.5
Induction of uterine ERβ by both estrogen regimens
OVX reduced ERβ protein in the uterus 62% from intact levels (F(2,83) = 10.20, P = 0.0001, Fig. 1c) which was restored by both estrogen treatments (2.7× and 2.9× of OVX controls for pulsed and constant estrogen, respectively; F(4,83) = 5.299, P < 0.001). ERβ levels were insensitive to LPS in the uterus (P = 0.80) and were unaltered in the hypothalamus (P = 0.68) or hippocampus (P = 0.99) with any treatment (Fig. 1d).
Regulation of uterine heat shock proteins (HSP90, HSP70, and HSP27)
Uterine HSP70 levels were not altered by OVX (P = 0.75) or pulsed estrogen (P = 0.34), but increased 27% following 8 weeks of constant estrogen replacement (1.3× of OVX controls; F(4,85) = 3.359, P < 0.02; Fig. 2b). LPS treatment did not alter HSP70 levels overall (P = 0.29) or within intact rats (P = 0.07). Likewise, there was no change in HSP70 protein levels in the hypothalamus by LPS (P = 0.99) or estrogen treatment (P = 0.51, data not shown).
HSP27 levels did not differ between intact and OVX rats (P = 0.83), but increased significantly following constant or pulsed estrogen replacement (3.7× and 3.5× of OVX controls, respectively; F(4,38) = 9.95, P < 0.0001; Fig. 2c). HSP27 strongly correlated with uterine ERβ levels (r(36) = 0.66, P < 0.0001), consistent with recent reports that HSP27 is an ERβ-associated protein involved in its signaling and localization in vitro . Uterine HSP27 levels were not altered by LPS (P = 0.96) and HSP27 levels in the hypothalamus were undetectable using either a rabbit polyclonal or mouse monoclonal antibody (385877 and EMD-35, respectively; Calbiochem, San Diego, CA, data not shown).
Induction of uterine PRs
Brain infusion of LPS suppressed uterine weight in intact rats
Brain infusion of LPS altered circulating estradiol and progesterone levels
Consistent with the low uterine weights and diestrus vaginal cytology, OVX rats had low estradiol levels (12.4 ± 2.7 pg/ml) that increased 2.4× by constant, but not pulsed, estrogen replacement (29.6 ± 3.6 and 17.9 ± 3.2 pg/ml, respectively; F(4,85) = 3.75, P < 0.008; Table 1). Estradiol levels were 61% lower in pulsed estrogen-treated rats infused with LPS (11.0 ± 1.8 pg/ml) compared to CSF (27.9 ± 6.1 pg/ml, F(4,94) = 2.606, P < 0.05), possibly due to LPS effects on the cytochrome p450 system . In contrast, LPS had no effect on estradiol levels in rats receiving constant estrogen (P = 0.63) or in ovary-intact rats (P = 0.17). Circulating estradiol levels were assayed twice in triplicate with similar results and estradiol levels correlated with increases in uterine weight (r(93) = 0.22; P < 0.03) as well as the reductions in ERα protein levels in the uterus (r(74) = −0.33; P < 0.004) and hypothalamus (r(48) = −0.35; P < 0.02).
Circulating progesterone levels were significantly reduced from intact levels by 8 weeks of OVX (21.0 ± 3.7 and 6.3 ± 0.9 ng/ml, respectively; F(2,96) = 21.8, P < 0.0001; Table 1) and were not restored by estrogen replacement (P = 0.12) or LPS infusion (P = 0.44) in OVX rats. Intact rats also showed no difference in progesterone levels when infused with LPS (P = 0.14). Circulating progesterone and corticosterone levels correlated (r(96) = 0.42, P = 1.99 × 10−5), consistent with the finding that progesterone is produced by the adrenal gland as a biosynthetic precursor to corticosterone . Moreover, removal of ovarian sources of progesterone by OVX resulted in a stronger correlation with a steeper slope between the two variables (r(76) = 0.77, P = 3 × 10−16) than observed in intact rats (r(17) = 0.51, P < 0.03).
Protein levels assessed by immunoblot in rats treated with pulsed or constant estrogen for 8 weeks
We expected to observe differences in hippocampal ER levels, consistent with the regimen-dependent effects on spatial working memory ; however, no changes in levels of either ER subtype were observed by immunoblot or radioligand binding in the current study, consistent with results seen after 6 days of treatment . In the hypothalamus, several studies reported conflicting effects of estrogen replacement on ER levels [33–37], possibly due to nuclei sampled, the timing of intervention or regimen of estrogen replacement. We observed no change in either ER subtype after 6 days ; however, both estrogen regimens reduced ERα, but not ERβ, protein by 35% following 8 weeks of treatment. The hypothalamus plays a major role in endocrine homeostasis and these actions are mediated predominantly by ERα . Thus, systems that utilize ERα-dependent induction of hormonal signals, such as the HPG and HPA axes [38, 39], may be particularly sensitive to long-term estrogen replacement.
The Pgr gene is regulated in an ER-dependent manner  and increases in the uterus in response to estrogen administration . PRA and PRB expression are controlled by two different promoters that are both estrogen-regulated  and consistent with this notion, uterine PRA increased following both regimens whereas only constant estrogen was sufficient to increase PRB. PRB expression can be stimulated via ERα, but not ERβ ; although, it is unclear why PRB levels were higher in constant estrogen-treated rats who had reduced levels of uterine ERα. PRA can repress PRB , which functions as a more potent activator of transcription of PR target genes . Thus, it is possible that differences in promoter usage by PRA and PRB may play a role in the ability of estrogen regimens to differentially alter these isoforms.
Heat shock proteins play an essential role in cellular trafficking, protein folding and protecting cells from stress (for review ). Uterine HSP70 and HSP90 levels increased in the current study by constant, but not pulsed, estrogen replacement. These HSPs and their co-chaperones can alter ERα stability and signaling [46–48] as well as play a role in ER ligand binding . In contrast to HSP70 and HSP90, both estrogen regimens increased uterine HSP27 levels by 350%. HSP27 strongly correlated to increases in uterine ERβ in the current study, consistent with the finding that HSP27 can associate with ERβ to act as a co-repressor of estrogen signaling .
Many steroid receptors can directly modulate inflammatory processes [16, 51, 52] and it has been suggested that the regimen of constant estrogen or its combined presence with inflammation may influence the efficacy of estrogen replacement in post-menopausal women. We infused LPS into the brain to test the effect of low-grade inflammation on ER, PR, and HSP expression during estrogen replacement. While all rats showed LPS-induced increases in activated microglia, we were surprised to find that only intact rats showed LPS effects on ERα and HSP90 expression. Fasting has been shown to alter ERα expression , but no correlation was observed between body weight and any of the steroid proteins measured in the current study (data not shown). ERα plays a role in modulating inflammatory processes [13–15, 54] and, in response to the inflammatory cytokine, TNFα, can be recruited to the Tnf promoter with HSP90 as part of a transcriptional complex .
Uterine reproductive physiology and decidual growth are directly regulated by HSP90 and its co-chaperones , which can be altered by LPS [57, 58]. Stress can disrupt ovarian cyclicity (for review ) and inoculation of the uterus with inflammatory stressors, such as LPS or bacteria, can increase inflammatory cytokines, alter circulating hormone levels and prolong diestrus [25, 26]. Consistent with these findings, most intact LPS-infused rats were in diestrus after 8 weeks and had reduced uterine weight that correlated to activated microglia in the brain. While others have documented uterine effects within 2 weeks of inflammatory stressor initiation [25, 26], we report that these physiological changes can persist weeks after cessation of LPS infusion. These data highlight the long-term impact of inflammatory stress on reproductive function.
The peripheral effects of brain infusion of LPS could be mediated by the HPA axis, the sympathetic nervous system or LPS leakage from the brain . We found no evidence of inflammatory cytokines (IL1β, TNFα) in the blood or brain of male rats after brain infusion of LPS (G.L. Wenk, unpublished observations); however, sex steroids can alter blood–brain-barrier permeability . Downregulation of ERα combined with HPA axis-induced elevations in corticosterone or progesterone is hypothesized to mediate uterine growth [18–20]. Therefore, our use of chronic LPS, which attenuates the HPA response  allowed us to test whether the corticosterone and progesterone components of the LPS-induced inflammatory response were necessary for the increased uterine growth. Consistent with our hypothesis, the 85% decline in uterine ERα with constant estrogen was not sufficient to alter uterine weight in LPS-infused rats, as the HPA response was similar to CSF controls. It remains to be determined whether injections of corticosterone or progesterone would be sufficient to increase uterine weight in constant estrogen-treated rats; however, our work suggests that these factors may contribute significantly to the uterine response. Other factors that may play a role include ER chaperones and thyroid hormones, which can crosstalk with estrogens and are affected by stressors (for reviews, see [63, 64].
The timing of hormone therapy after menopause is being increasingly recognized as an important factor in the success of the therapy ; however, the data presented here suggest that the regimen of estrogen replacement also plays a role. Constant and pulsed estrogen regimens altered steroid-related proteins in a tissue- and time-dependent manner, even when administered at the time of OVX. Many of these proteins are also affected by inflammatory stressors, suggesting that stress and estrogen regimen may become important factors in the post-menopausal intervention with estrogen.
Materials and methods
Experimental groups and numbers
Hormone group ovariectomized (OVX)
Brain infusion of LPS
Verification of LPS effectiveness
CSF (mean ± SEM)
LPS (mean ± SEM)
Activation of thalamic microglia
90.2 ± 5.3 nmol/mg protein
109.1 ± 4.6 nmol/mg protein
F(1,97) = 7.155, P < 0.01
177.1 ± 5.4 g
147.8 ± 4.7 g
F(1,97) = 35.957, P < 0.001
564.4 ± 54.0 ng/ml
693.3 ± 67.3 ng/ml
n.s., P = 0.13
4.01 + 0.08 nM
3.84 + 0.08 nM
n.s., P = 0.14
Immediately after implantation of the cannula, rats were bilaterally ovariectomized (OVX) or left intact. OVX rats were assigned to either a constant or pulsed estrogen replacement regimen (see Fig. 5). Rats receiving the constant replacement regimen were implanted (s.c.) with 5 mm silastic capsules containing estrogen (25% 17β-estradiol and 75% cholesterol) or oil (100% cholesterol) at the time of surgery, as previously described [18, 66]. Rats receiving the pulsed estrogen replacement regimen received one injection every 4 days beginning the morning after surgery. All injections were administered between 10:00 a.m. and 12:00 p.m. and consisted of either estrogen (10 μg 17β-estradiol dissolved in 100 μl sesame oil; Sigma, St. Louis, MO) or oil vehicle (100 μl sesame oil). The injection paradigm was selected to mimic the proestrus peak seen in the normal rat estrous cycle and has previously been used to enhance learning and memory . Serial blood sampling from jugular catheters previously verified both estrogen regimens and determined that estradiol levels resulting from these injections returned to baseline within 24 h . Approximately the same amount of estradiol was delivered by both estrogen regimens, as determined by area under the curve calculations extrapolated from serial sampling (data not shown). Serum levels of estradiol, progesterone and corticosterone were quantified using 125I radioimmunoassay (RIA) kits (Diagnostic Systems Laboratories; Webster, TX), according to the manufacturer’s instructions.
Rats were exposed to experimental conditions described above for 8 weeks, consistent with our previous work investigating these factors on spatial working memory . After 8 weeks (24 h after the last estrogen injection and between 12:00–5:00 p.m.), each rat was weighed, assessed for vaginal cytology, anesthetized with isoflurane and sacrificed by decapitation. Trunk blood was collected, serum isolated, and stored (−70°C) until analysis. Uterine weights were quantified and brain dissections were performed as previously described . Briefly, the entire hippocampus was taken bilaterally, the thalamus was isolated from between ∼−2.12 and −6.5 mm from Bregma and a hypothalamic sample was taken (∼3 mm width inferior to the anterior commissure between −2.12 and −0.8 mm from Bregma) which contained several hypothalamic nuclei including the medial preoptic and anterior hypothalamic areas, suprachiasmatic, periventricular, lateroanterior and paraventricular nuclei as well as the tuber cinerum. All results were analyzed by two-way ANOVA and Bonferroni post-tests (GraphPad Prism) unless described otherwise. Pearson’s product-moment correlation coefficient was used to determine the association between variables (GraphPad Prism).
Activated microglia were quantified in the thalamus by in vitro [3H]PK11195 filtration binding (1 nM; specific activity, 85.5 Ci/mmol; displaced by 20 μmol/l diazepam [18, 69]). The thalamus is sensitive to brain infusion of LPS in female rats [18, 66] and activation of thalamic microglia correlates to LPS-induced behavioral impairments on a Morris water maze task (L.K. Marriott, unpublished observations).
Estrogen receptor quantification
Hippocampal ERs were quantified by [2,4,6,7-3H(N)]-17β-estradiol radioligand binding (1 nM; specific activity, 95 Ci/mmol; displaced by 1 μM diethylstilbestrol [18, 70, 71]). The entire right hippocampus was assayed to ensure consistency between samples, as the density and subtype of the estrogen receptor varies between dorsal and ventral aspects of the hippocampus [72, 73]. ER levels were verified via immunoblot (see “Immunoblotting” below) in a subset of these rats using the contralateral hippocampus.
Hypothalamic, uterine and left hippocampal tissues were homogenized in immunoprecipitation buffer, sonicated, centrifuged and electrophoresed onto pre-cast 4–12% Bis–Tris NuPage gels (Invitrogen; Carlsbad, CA) as previously described . Equal amounts of protein (55 μg for hypothalamus; 30 μg for uterus; 10 μg for hippocampus) were loaded as determined by bicinchoninic acid assay (Pierce Biotechnology Inc., Rockford, IL). One animal per treatment group was included on each gel, and at least four gels were run per region. Protein was transferred to polyvinylidene diflouride membranes (Westran PVDF, 0.2 μM; Fisher) and blocked in 5% nonfat dry milk (NFDM) as described previously . Primary antibodies [ERα (AB15, 1:500; Neomarkers, Fremont, CA); ERβ (PA1310B, 1:1000; Affinity Bioreagents, Golden, CO); heat shock protein (HSP)90 and HSP70 (formally known as HSP90AA1 and HSPA1B, respectively; 1:2000, Becton-Dickinson, Franklin Lakes, NJ), HSP27 (formally known as HSPB2; 1:2000, 385877, Calbiochem, San Diego, CA); progesterone receptor (PR, formally known as PGR; AB13, 1:1000, Neomarkers); and actin (formally known as ACTB; 1:5000, Sigma)] were incubated (overnight at 4°C) in 5% NFDM/Tris-buffered saline containing 0.1% Tween-20 and 0.02% sodium azide to prevent bacterial growth. Appropriate secondary and tertiary antibodies conjugated to horseradish peroxidase were used , and bands were visualized and quantified as previously described . Recombinant human ERα (10 pg) or ERβ (100 pg) proteins (Affinity Bioreagents, Golden, CO) were loaded as positive controls. Blots were stripped (Reblot; Chemicon; Temecula, CA) and reprobed for each protein as described above.
Protein loading was controlled using actin; values did not differ between treatment groups in any region tested (data not shown). Band intensity was determined by subtracting out the background for each band and dividing by protein levels (i.e. actin). Blots were normalized to each other by dividing the maximal band intensity of all blots by the maximal band intensity of each individual blot. All treatment groups were equally represented on each gel and exposed to the same experimental conditions.
We would like to thank Drs. Damani N. Bryant, David S. Herman and Yannick Marchalant for their thoughtful critiques of the manuscript. Supported by the U.S. Public Health Service (AG030331, to GLW), the Alzheimer’s Association (IIRG-010-2654, to GLW), NIH (NS20311-23 to DMD) and Reproductive Biology Training Grant (T32-HD007133-27 to LKM).