Introduction

Coronary heart disease is relatively rare in premenopausal women compared to postmenopausal women or men. After menopause, the sex protection in women decreases, and cardiovascular risk parameters are adversely impacted [1]. Pathogenic consequences of menopause and sex differences in disease prevalence are attributed to differences in estrogen levels supplied to cardiovascular tissues. Estradiol (17β-estradiol, E2) atheroprotection has been documented in many animal models of atherosclerosis [2–5], including murine models with targeted mutations (reviewed in [6]).

In humans, observational epidemiological studies strongly support an atheroprotective role for endogenous E2 [7], largely attributed to reductions in low-density lipoprotein (LDL) and increases in high-density lipoprotein (HDL) cholesterol [8, 9], antioxidant actions [10, 11], and a direct biological effect on cardiovascular tissue [12–14]. In contrast, clinical trials have failed to confirm the benefits of combined or unopposed estrogen reported by earlier observational studies [15–17]. The hypothesis that hormone therapy is an effective preventive measure for coronary heart disease has been challenged by these trials, perhaps in part due to the inclusion of postmenopausal women with subclinical coronary artery plaque. The weight of the evidence indicates that older women and those with subclinical or overt coronary heart disease should not take hormone therapy. However, timing of initiation of hormone therapy in relation to menopause onset and age might reduce coronary risk over time, presumably by affording atheroprotection prior to the onset of established vascular disease [18]. In addition, factors including the method of administration, dose, and duration of use of exogenous hormone may also be relevant. Nonetheless, the overall clinical utility of hormone therapy for atheroprevention remains controversial and complex.

Estrogen’s long-term effects are generally ascribed to transcriptional modulation of target genes through estrogen receptors (ERs) [19]. Two ERs, ERα (ESR1) and ERβ (ESR2), have been characterized [20, 21]. They are encoded on two separate genes (ESR1 and ESR2), are distinct functionally and structurally, and have overlapping though not identical tissue expression patterns [21–23]. Functional estrogen receptors are expressed in vascular endothelial cells [24], smooth muscle cells [25, 26], and macrophages [27]. Normal coronary arteries of premenopausal women demonstrate normal ER expression, whereas in atherosclerotic vessels of postmenopausal women, ERs are downregulated [28]. Thus, changes in expression of ERs accompany atherogenesis, though the direction and magnitude of the association needs clarification.

In humans and rodents, both ERs may mediate protection against vascular injury [25, 29]. However, the individual contribution of sex steroid hormones and their cognate receptors to the atherosclerosis and its progression remains poorly understood. In LDL-receptor (LDL-R) and apolipoprotein (Apo) E-deficient mice, ERα is a major mediator of E2’s atheroprotective effects on advanced atherosclerotic lesions [30–33]. However, studies to distinguish the atheroprotective effects of E2 from those of ERα in the development of early atherosclerosis and examine the mechanism of E2 protection in fatty-steak formation are limited. In comparing the effects of E2 treatment on spontaneous development of atherosclerotic lesions in ApoE−/− mice (with or without the ERα), the major atheroprotective effect of E2 was substantially abrogated in mice lacking both ApoE and ERα [33]. However, some of the protective effects of E2 appeared to be ERα independent.

Thus, there is continued need to define the individual contribution(s) of estrogen and estrogen receptors to atherosclerosis, the stage(s) of atherosclerosis when estrogen is atheroprotective, and the stage-specific mechanism(s) involved in vasculoprotection. The goal of this study was to directly examine the action of the ERα in E2-mediated atheroprotection, using a murine model of early-stage atherosclerosis (fatty streaks) deficient for the ERα, in order to directly examine the action of the ERα in early E2-mediated atheroprotection, identify the individual contribution(s) of estrogen and/or the ERα to early atherogenesis, and address mechanisms of E2 protection in early vascular disease. The study design permitted us to evaluate three potential individual and/or interrelated mechanisms for atheroprotection in early-onset disease: [1] 17 β-estradiol (E2), [2] the estrogen receptor-alpha (ERα), and [3] serum lipids.

Materials and Methods

All experiments were performed in compliance with the National Institutes of Health (NIH) Guidelines and in accordance with protocols approved by the Animal Care and Use Committee of the University of California, Davis.

Generation and Use of Mice

Mice heterozygous for the ERα gene (ERα+/−), obtained from Dr. Dennis Lubahn (University of Missouri) and mated to yield a breeding colony and progeny lacking the ERα gene (ERα−/−), and wild-type littermate controls having intact ERα (ERα+/+) were used. The correct genotype was confirmed by polymerase chain reaction (PCR) amplification of genomic DNA. Tail DNA from progeny of heterozygous matings was genotyped as previously described [34] using DNeasy spin columns (Qiagen kit) per the manufacturer’s instructions. PCR amplification was used to distinguish homozygous mutants from heterozygotes and wild-type animals. Primers used to amplify products specific for the wild-type gene and disrupted ERα gene were synthesized as custom primers (Life Technologies). Primer sequences for presence of the targeted ERα gene in homozygous mutants (a 649-bp fragment only) and animals with the normal wild-type gene (a 239-bp fragment only) were as follows:

  • ERα−/− (KO) forward 5′ TGAATGAACTGCAGGACGAG 3′ and reverse 5′ AATATCACGGGTAGCCAACG 3′

  • ERα+/+ (WT) forward 5′ CTACGGCCAGTCGGGCAT 3′ and reverse 5′ AGACCTGTAGAAGGCGGGAG 3′

Mice were housed in a temperature-, 12 h light-, and humidity-controlled environment in a dedicated pathogen-free barrier facility at the University of California, Davis. ERα gene deletion mice have a mixed genetic background from strains 129/J and C57BL/6 J, an atherosclerosis-susceptible strain. They are infertile but otherwise phenotypically normal [34]. They do not produce a fully functional ERα and have no demonstrable full-length wild-type ERα protein [34]. However, the disrupting neo-sequence used to generate the knockout results in a variant ERα protein (not the full-length ERα wild-type protein) contained an internal deletion of a portion of the N-terminal region of the receptor. This alternative ERα protein has residual low-level, high-affinity, estradiol-binding activity [35, 36] but, importantly, lacks the specificity to be from either ERα or ERβ (Liu and Lubahn, unpublished data).

Atherogenesis Model and Experimental Design

Experimental animals were weaned at 4 weeks. To manipulate estrogen status, age-matched ERα−/− and ERα+/+ female mice were castrated at 6–7 weeks of age by bilateral ovariectomy (OVX) under anesthesia (2.5% Avertin). OVX mice were then implanted with subcutaneous hormone pellets (Innovative Research, Toledo, OH, USA) designed for slow release of 6 μg/day of exogenous E2 as 17β-estradiol (exogenous E2-6; n = 19 and n = 18 for ERα−/− and ERα+/+ mice, respectively), or 8 μg/day of exogenous E2 as 17β-estradiol (exogenous E2-8; n = 21 and n = 20 for ERα−/− and ERα+/+ mice, respectively) for a total of up to 4 months. The 6-μg/day E2 dose was chosen based on analogous models indicating that this dosage attains blood levels at peak estrus necessary to lower plasma lipids and plaque size [30–32]. The 8-μg/day E2 dose was chosen to assess the dose–response of exogenous E2 and the impact of higher E2 dose and levels on atheroprotection. OVX ERα−/− (n = 22) and ERα+/+ (n = 20) mice implanted with placebo vehicle pellets served as E2-deficient controls. In addition, ERα−/− (n = 18) and ERα+/+ (n = 19) female mice with ovaries left intact served as an additional study group for evaluating the comparative vascular effect of endogenous E2. Intact ERα-deficient females have generally been excluded from prior studies due to plasma E2 levels that are higher than wild-type mice as a consequence of their estrogen insensitivity [37]. However, we chose to include these mice to evaluate the effect of higher endogenous estrogen in our model.

To induce atherosclerotic lesions, mice were fed a high-fat, high-cholesterol, atherogenic diet [38, 39], herein referred to as the atherogenic diet (Purina; 15% wt/wt butter fat, 1.25% wt/wt cholesterol, and 0.5% cholic acid) for 4 months starting 1 week after surgical recovery from OVX. The diet was formulated to be phytoestrogen-free. Aortas (five to eight from each genotype) were sampled for lesion formation at baseline (age 6–7 weeks, immediately before cholesterol feeding), and after 2 and 4 months on the atherogenic diet.

Aortic Atherosclerotic Lesion Analysis

To determine extent of characteristics and extent of atherosclerotic lesions in aortas and determine any regional differences in atherosclerosis susceptibility, we adapted and optimized previously published and reproducible methods for atheroma quantification [40], as previously described [41]. Oil Red O-stained (Sigma) cryosections (10 μm thick) defined and characterized neutral lipid in the vessel wall, permitting histological assessment and quantitative evaluation of lesions in three artery segment locations (proximal segment = proximal aortic arch; mid segment = suprarenal aorta; distal segment = infrarenal aorta proximal to the iliac bifurcation). Fatty streaks (herein, early lesions) were analyzed for area, location, and quantity by direct imaging (Olympus BX-40 Microscope, Olympus DP11 Camera with 64 MB smart media chip). Quantitative total lesion area and mean lesion size (μm2 ± SEM) for each early lesion was determined morphometrically (Image Pro Plus, version 4.1). Qualitative morphological assessment of lesion complexity used, as parameters, extent of subintimal lipid deposition and ORO staining and absence or presence of foam cells. Wall thickness, disruption of vessel wall architecture, and distortion of nuclear proliferation were not assessed as they were not part of the early-lesion pathology in the study’s early-lesion model. For each mouse, a total of 36 cryosections (discarding every other section) were obtained from the proximal, mid, and distal aortic segments for extensive sampling of a total of >2,100 μm per aorta.

Measurement of Uterine Weight

The entire uterus was collected at baseline and 2 and 4 months to assess in vivo exposure to estradiol. Serosal fat was removed, the uterus divided from the utero-tubal junction bilaterally, transected at the junction of the uterine body and cervix, and uterine wet weight recorded relative to body weight.

Quantification of Plasma Estradiol

Plasma levels of free 17β-estradiol were determined in study mice at baseline and 2 and 4 months on the atherogenic diet. Assays were performed in duplicate by enzyme immunoassay (Cayman Chemical, Ann Arbor, MI, USA) on nonpooled plasma samples.

Quantification of Plasma Lipids

Plasma lipid levels were measured in study mice at baseline and 2 and 4 months on the atherogenic diet and analyzed as follows: triglyceride (TG) with Triglyceride (INT) Reagent (Sigma); high-density lipoprotein cholesterol (HDL-C) with a colorimetric reagent kit (Wako) after precipitating the ApoB-containing particles; and total cholesterol (TC) using a GM7 Analox Analyzer (Analox). All assays were performed in duplicate on nonpooled plasma samples.

Statistical Analysis

Lesion area and lesion number were quantified by calculating the total lesion area and total number of lesions, respectively, in each of the 36 aortic sections for each mouse aorta. Comparisons between study groups were made using Student’s t test for independent samples (two-tailed), and analysis of variance, using MS-Office 2000 Excel software for PC and Sigma Stat version 2.03 (SPSS). Genotype (ERα −/− versus ERα +/+) and E2 status (OVX, Intact, E2-6, and E2-8) were the grouping variables. Correlations between parameters were analyzed using simple linear regression. Probability values at α < 0.05 identified all statistically significant comparisons.

Results

Plasma Estradiol

We determined plasma estradiol levels in study mice. We successfully achieved the desired hormonal status in all of the mice (Fig. 1a). OVX reduced circulating plasma E2 to very low levels in ERα−/− and ERα+/+ mice. E2 levels in intact ERα−/− mice were more than twice those of intact ERα+/+ mice (p < 0.05), consistent with the previously reported ER-dependent functional estrogen insensitivity of ERα−/− mice [42]. Plasma E2 in exogenous E2-replaced ERα−/− and ERα+/+ mice were at or above peak estrus levels, previously reported necessary for physiologic atheromatous plaque reduction [30–32]. With E2 replacement, female mice have been previously reported to develop carcinoma of the cervix and urinary tract and reproductive tract pathology [43]. None of our study mice developed cervical carcinoma with E2. However, three mice developed endometritis, hydronephrosis, and bladder enlargement unassociated with infection. These mice were excluded from subsequent analysis.

Fig. 1
figure 1

a Plasma estradiol levels in ERα−/− and ERα+/+ female mice. Levels of plasma 17β-estradiol (pg/ml; mean ± SEM) in ERα−/− and ERα+/+ female study mice are shown for ovariectomized (OVX, n = 14–15), intact (n = 6–10), and ovariectomized replaced with estradiol pellets designed to deliver 6 μg/day (E 2 -6; n = 10–12) and 8 μg/day (E 2 -8; n = 8–13). Statistically significant (p < 0.05) and non-significant (ns) comparisons are shown for each genotype. There were no significant differences for treatment group comparisons between genotypes. b Uterine weights in ERα−/− and ERα+/+ female mice. Uterine weights (g; mean ± SEM) in ERα−/− and ERα+/+ female study mice are shown for ovariectomized (OVX, n = 4–6), intact (n = 5–7), and ovariectomized replaced with estradiol pellets designed to deliver 6 μg/day (E 2 -6; n = 13–14) and 8 μg/day (E 2 -8; n = 6–14). Statistically significant (p < 0.01) and non-significant (ns) comparisons are shown for each genotype. In addition, there were significant differences (p < 0.01) for uterine weight comparisons between ERα+/+ E2-6 and E2-8 and all ERα−/− treatment groups

Full size image

Uterine weights were measured to functionally assess in vivo E2 action in our system (Fig. 1b). In ERα+/+ mice, the very small, atrophic uteri of E2-deficient OVX female mice confirmed the absence of estrogenic stimulation. ERα+/+ mice with endogenous E2 had significantly (p < 0.01) larger uteri than OVX females. Uterine weights of OVX ERα+/+ female mice replaced with exogenous E2-6 and E2-8 were over 8-fold larger than OVX ERα+/+ females (p < 0.05) and 4-fold larger than intact ERα+/+ mice (p < 0.05). As expected, in ERα−/− mice, uteri were consistently small and did not significantly enlarge with E2 treatment. Although E2 has reported effects on body weight in humans and several animal species [44, 45], there were no significant differences in body weight between ERα−/− and ERα+/+ mice in our different treatment groups and experimental time points (data not shown).

Plasma Lipids

Figure 2 summarizes fasting plasma lipids for total cholesterol, triglycerides, and high-density lipoprotein cholesterol for OVX, intact, and E2-replaced ERα−/− and ERα+/+ female mice while on the atherogenic diet. TC was lower at baseline in ERα+/+ mice than in ERα−/− mice (73 ± 11 mg/dl versus 104 ± 5 mg/dl, respectively, p = ns). Compared to baseline, after 4 months on the atherogenic diet, fasting TC increased significantly (p < 0.05) in OVX, endogenous E2, and exogenous E2-6 ERα−/− and ERα+/+ mice (32%, 28%, and 30% versus 55%, 58%, and 55% for each genotype, respectively; p = ns for treatment group comparisons between genotypes). However, the atherogenic diet-associated increase in TC in ERα−/− and ERα+/+ mice was not affected by estrogen treatments in any of the study groups, except for a significant (p < 0.05) reduction in TC with exogenous E2-8 only observed in both ERα−/− and ERα+/+ mice.

Fig. 2
figure 2

Plasma lipid levels in ERα−/− and ERα+/+ female mice. Fasting lipid levels (mg/dl; mean ± SEM) for total cholesterol (TC), triglyceride (TG), and high-density lipoprotein (HDL) were determined in non-pooled plasma of ERα−/− (n = 50 samples for each lipid) and ERα+/+ (n = 49 samples for each lipid) female mice on the atherogenic diet for animals ovariectomized (OVX), intact, and ovariectomized replaced with estradiol pellets designed to deliver 6 μg/day (E 2 -6) and 8 μg/day (E 2 -8). *p < 0.05 versus lipid baseline values (not shown, see text) prior to cholesterol feeding for each genotype; **p < 0.05 versus TC in other treatment groups for a given genotype

Full size image

In addition, TC levels on the atherogenic diet in OVX and intact ERα+/+ mice did not correlate with lesion area and correlated poorly with lesion area in exogenous E2-6 ERα+/+ mice (r = 0.56; p = ns). However, in ERα+/+ mice, the reduction in TC with exogenous E2-8 correlated strongly with reduction in fatty streak size (r = 0.83; p < 0.05). In ERα−/− mice, aortic lesion area did not correlate with TC in any of the treatment groups. Thus, with the sole exception of E2-8 in ERα+/+ mice, the dramatic reduction in lesion size with E2 (endogenous or exogenous) in ERα−/− and ERα+/+ mice was not a consequence of an estrogen-mediated reduction in TC levels.

There were no significant differences in baseline fasting plasma TG levels between ERα−/− and ERα+/+ female mice (75 ± 6 and 81 ± 13 mg/dl, respectively). While on the atherogenic diet, fasting levels of plasma TG increased significantly (p < 0.05) in all E2 mouse groups, compared to baseline or OVX (Fig. 2). However, increases in TG did not correlated with lesion formation in any of the treatment groups or genotypes.

Fasting levels of HDL-C were similar at baseline in ERα−/− and ERα+/+ mice (13 ± 2 and 17 ± 4 mg/dl, respectively). Compared to baseline, HDL levels tended to be significantly greater in hormone replaced ERα−/− mice than ERα+/+ mice. However, there were no significant changes in HDL-C levels between treatment groups in ERα+/+ mice on the atherogenic diet (Fig. 2). Furthermore, there was no correlation between lesion area and HDL-C levels in either genotype.

Lesion Area, Number, and Distribution

ERα−/− and ERα+/+ mice developed aortic lesions typical of fatty streaks [38]. The fatty-streak lesion area (averaged throughout the period on the atherogenic diet) for all treatment groups is summarized in Fig. 3. Endogenous and exogenous E2 were both very effective in inhibiting early lesion formation. In both ERα−/− and ERα+/+ female mice, endogenous E2 in mice with intact ovarian function and exogenous E2 in estrogen-replaced OVX mice significantly reduced lesion area compared with E2-deficient OVX mice. Compared to OVX, lesions in mice with endogenous E2 were 64% and 55% smaller in ERα−/− and ERα+/+ mice, respectively (p < 0.001). In addition, following exogenous E2 replacement, lesions were more than 90% smaller (p < 0.001) in both ERα−/− and ERα+/+ mice compared to OVX mice. No significant difference in lesions was observed between the two doses of exogenous estradiol (E2-6 and E2-8) in ERα−/− or ERα+/+ mice. In addition, the rate at which lesions developed over time appeared to decline with increasing estrogenic action in both ERα−/− and ERα+/+ mice but did not differ between the two genotypes (data not shown).

Fig. 3
figure 3

Early lesion formation in ERα+/+ and ERα−/− female mice. Lesion area (μm2 ± SEM) of early atherosclerotic fatty streaks in the aortas of ERα−/− (n = 80) and ERα+/+ female mice (n = 77) was measured at baseline and 2 and 4 months of cholesterol feeding (n = 5–7 mice each per genotype, treatment group, and time point). Lesion area is shown for mice ovariectomized (OVX), intact, and ovariectomized replaced with estradiol pellets designed to deliver 6 μg/day (E 2 -6) and 8 μg/day (E 2 -8). Data are mean ± SEM lesion area (in μm2). *p < 0.001 compared to the indicated treatment groups for a given genotype

Full size image

For each of the E2 treatment groups, representative histological lesions of female ERα−/− and ERα+/+ mice after 4 months on the atherogenic diet are shown (Fig. 4). Early atherosclerotic lesions were present in the intimal layer of aortas of both ERα−/− and ERα+/+ mice as early as 2 months on the atherogenic diet, with more extensive lesions developing over the 4 months of the study. Lesions were distinguished by lipid infiltration into the vessel wall and the presence of fatty streaks and did not have characteristics of advanced lesions, confirming our model of early atherosclerosis. Compared to endogenous and exogenous E2, OVX ERα−/− and ERα+/+ mice had early lesions that were more extensive and lipid rich and with larger deposits of intra- and extra-cellular lipid. Lesions in ERα−/− and ERα+/+ endogenous E2 mice had less lipid deposits than OVX mice but more fatty streaking of the sub-endothelium than exogenous E2 mice. Lesion formation was essentially absent in mice replaced with exogenous E2.

Fig. 4
figure 4

Representative cross sections of lesions from the aortas of ERα−/− And ERα+/+ mice. Histochemistry of examples of representative cross sections (10 μm) of lesions from the proximal aorta of cholesterol fed (4 months) ERα−/− (right panel) and ERα+/+ (left panel) female mice for the following experimental groups: ovariectomized (OVX), ovaries intact, and ovariectomized replaced with estradiol pellets designed to deliver 6 μg/day (E 2 -6) and 8 μg/day (E 2 -8). The vessels have been stained for lipid with Oil-red-O and counterstained with Gill’s Hematoxylin. A normal vessel prior to the initiation of cholesterol feeding, baseline, is shown for comparison. The representative images (×20 magnification) demonstrate red-staining subintimal lipid deposits and fatty streaks. Cross sections are oriented with the lumen to the right and the adventitial surface to the left of each image. Size markers = 20 μm

Full size image

We also characterized early lesions by quantity and segment location in the aorta. The segmental distribution of early lesions in aortas of ERα−/− (n = 51, 48, 54, and 54 total segments for OVX, intact, E2-6, and E2-8, respectively) and ERα+/+ mice (n = 51, 51, 51, and 54 total segments for OVX, intact, E2-6, and E2-8, respectively) is shown in Fig. 5. All mice showed predilection for early lesion formation in the proximal aortic segment, with 83–100% of all early lesions localized to this region of the aorta. The mid aorta contained 2–18% of early lesions, whereas <3% of early lesions were in the distal aorta. Differences in distribution of lesions were significant (p < 0.01) between segments (proximal versus mid versus distal aorta), and compared to endogenous or exogenous E2, there was significantly greater (~4-fold) proximal distribution of lesions in OVX mice. However, lesion distribution was not significantly affected by ERα status.

Fig. 5
figure 5

Lesion distribution in aortic segments of ERα−/− and ERα+/+ female mice. Lesion area (μm2; mean ± SEM) was determined in the proximal, mid, and distal aortic segments of female ERα−/− and ERα+/+ mice ovariectomized (OVX, n = 29), intact (n = 23), and ovariectomized replaced with estradiol pellets designed to deliver 6 μg/day (E 2 -6; n = 25) and 8 μg/day (E 2 -8; n = 24). See text for details. *p < 0.01 versus other treatment groups of the same genotype

Full size image

The number of lesions in aortas of exogenous (E2-6 and E2-8) ERα−/− mice was significantly lower (p < 0.05) than in aortas of OVX ERα−/− mice. In ERα−/− mice, the mean number of lesions in endogenous E2, E2-6, and E2-8 mice was 12 ± 5, 8 ± 2, and 10 ± 2, respectively, compared to a mean of 41 ± 7 lesions in the aortas of OVX mice (p < 0.05). In ERα+/+ mice, a similar pattern was observed with exogenous E2 mice having fewer lesion than OVX or intact mice.

Relationship of Plasma Estradiol and Lesions

In our system, early lesion formation was inversely correlated with plasma estradiol levels in both ERα−/− and ERα+/+ mice. This relationship is demonstrated in Fig. 6. A step-wise reduction in lesion size was observed with a rise in plasma estradiol, and a converse step-wise increase in lesion size was seen with a drop in plasma estradiol. OVX animals had a mean plasma estradiol of only 13 pg/ml (we assume that the residual measurable estradiol in OVX mice is due to adrenal sex steroid production and conversion to estradiol) and the greatest extent of lesions. Intact mice with endogenous E2, in which mean plasma estradiol levels were 58 pg/ml, demonstrated nearly 60% reduction in lesion size (p < 0.05). As expected, intact ERα−/− females had relatively greater endogenous E2 levels than intact ERα +/+ mice due to their relative insensitivity to estrogen. However, the differences in endogenous E2 levels between these mice were relatively small and not sufficient to result in significant differences in aortic lesions in our studies. However, with higher plasma estradiol levels (mean >247 pg/ml attained in exogenous E2 mice), lesion development was essentially eliminated compared to endogenous E2 and OVX mice (<0.05). The inverse relationship between mean serum estradiol and mean lesion area was unaffected by ERα status as it did not significantly differ between the two ERα genotypes.

Fig. 6
figure 6

Reduction in lesion area as a function of plasma estradiol. Lesion area (μm2 ± SEM) as a function of plasma 17β-estradiol (E2; pg/ml; mean) in ERα−/− and ERα+/+ female study mice. Estradiol values (mean, pg/ml) are indicated for mice ovariectomized, OVX (E2 = 13 pg/ml), intact (E2 = 58 pg/ml), and ovariectomized and replaced with E2 pellets designed to deliver 6 μg/day (E2-6 = 247 pg/ml) and 8 μg/day (E2-8 = 284 pg/ml)

Full size image

Discussion

The important new finding of our studies is that protection from development of early atherosclerotic lesions (fatty streaks) is dependent on estrogen but independent of the ERα. In addition, estrogen atheroprotection was estrogen-level dependent, with greater atheroprotection offered by exogenous estrogen compared to endogenous estrogen and loss of atheroprotection with ovariectomy.

Atheroprotection by E2 in atherogenic transgenic mouse models with targeted inactivation of the low-density lipoprotein receptor [32] and apolipoprotein E [30, 31] has been demonstrated. In these models, mice spontaneously develop advanced atherosclerosis. In a combined ApoE/ERα deletion model [33], ERα was reported to be a major mediator of E2 protection in advanced atherosclerotic lesions. However, whether the same occurs in early-stage lesions is unknown. In this paper, we demonstrate several novel findings: (1) E2 inhibits and delays the development of early lesions by ERα-independent mechanisms, as there was no significant interaction between ERα genotype and hormone treatment or hormone status. (2) There was an estrogen concentration-dependent response to atheroprotection. And (3) Endogenous and exogenous E2 are both highly, though not equally, effective in attenuating aortic lesion development and progression. In our model, inhibition of early lesions using exogenous E2 in ovariectomized animals was more atheroprotective than leaving the ovaries (and thus endogenous E2 levels) untouched. This was somewhat surprising since the dosage used was chosen to obtain blood E2 levels similar to those at peak estrous, though actual serum E2 levels attained were at times higher. However, we do not think that this is the reason that inhibition of early lesions using exogenous E2 therapy in ovariectomized animals was more effective at inhibiting early lesion formation than endogenous E2. A more likely possibility to consider is that cycling estrogen levels may attenuate the growth inhibitory effect of endogenous E2, in contrast to continuous therapy in our model.

Our findings also extend recent observations reported in ApoE KO mice in which E2 treatment failed to reduce aortic lesions in ApoE-deficient mice also lacking ERα [33]. We postulate that the differences between our findings and previous are due to two principal factors: (a) the difference in the animal model used and (b) the differences in estradiol effect in early- versus late-stage atherosclerosis. In the previous report, the experimental strain used was the heterozygous ERα deletion on the atherogenic ApoE null transgenic mouse background. In contrast, the current study used ERα-Neo null mice on a mixed 129/J and C57Bl/6J background. Such mixed background mice are relatively protected from atherosclerosis and thus, a cholesterol-containing diet was used to induce aortic fatty streaks that did not progress to more mature atheroma. We took advantage of this model to specifically study E2 and ERα effects on early-stage disease. Secondly, the current study examined the early stages of atherosclerosis, whereas the previous report examined more advanced lesions of atherosclerosis. The differential effects of E2 treatment on early-stage atherosclerosis as opposed to the more advanced stages and how ERα is associated with one but not the other are very important observations and distinctions with keen relevance to better understanding atherosclerotic cardiovascular disease. Sex steroid hormones are critical determinants of cardiovascular sex differences. Moreover, the effects of estrogens on the blood vessel wall depend upon the extent and complexity of atherosclerotic disease present at the time hormone therapy is initiated. As atherosclerosis evolves, the early vascular protective mechanisms of estrogen (e.g., increased NO and cyclooxygense 2 and decreased TNF-α, cell adhesion molecules, LDL oxidation/binding, platelet activation, and vascular smooth muscle cell proliferation) recede and are replaced by estrogen responses that may be deleterious (decrease in ER function/expression and vasodilation and increase in inflammatory activation and plaque instability). These events have recently been comprehensively reviewed by Turgeon et al. [46], Clarkson [47], and Manson [48]. Thus, it is becoming increasingly clear that the pleiotropic effects of hormone therapy on the vascular system and cells differ, depending not only on hormone status, but also importantly on the stage of atherosclerosis in the underlying blood vessel.

We also examined whether the atheroprotective effects of E2 on early atherosclerosis were lipid mediated. However, the extent of atheroprotection attributable to E2 in our study and previous studies [30–33] is clearly greater than can be accounted for by changes in lipid parameters alone. In LDL-R−/− mice [32], exogenous E2, but not endogenous E2, lowers cholesterol. In the ERα−/− and ERα+/+ mice, lipid levels at baseline and following an atherogenic diet were similar to those previously published for atherosclerosis-susceptible mouse strains [38] and concur with those reported in ApoE-deficient mice, where only high doses of exogenous E2 reduced plasma cholesterol [30]. However, the protective effects of endogenous and exogenous E2 against fatty-streak formation in our studies were not correlated with plasma levels of TC, TG, or HDL-C. Thus, in our model, the traditional factors that promote lesion formation, such as a high-fat diet and high plasma cholesterol, did not interact with ERα genotype as the changes in serum lipoproteins were independent of ERα expression and occurred in both ERα−/− and +/+ female mice.

Our studies had some limitations. The ERα-deficient mouse used (ERα-Neo KO) is not fully null [35, 49, 50]. In this mouse, the ERα gene was disrupted by the insertion of the neomycin gene into the first coding exon, resulting in deficiency of the full-length wild-type ERα protein [34]. This approach also results in the expression of two alternatively spliced transcripts of the disrupted ERα gene, E1 and E2 [35], and the expression of two variant ERα proteins (only one of them found in the aorta) [35, 49]. This alternative ERα protein has residual low-level, high-affinity, estradiol-binding activity [35, 36] but lacks the specificity to be from either ERα or ERβ (Liu and Lubahn, unpublished data). This variant also has some estrogen-dependent transcriptional activity [35, 49, 50]—ERα transcriptional activity is mediated through two activation functions, AF1 and AF2, whose activity is tightly regulated in a cell-specific manner through yet unknown processes. More recently, mice completely deficient of the ERα gene by excision of exon 2 (ERα-Δ2KO) [51] resulted in the complete abrogation of an estradiol effect [49]. Although our studies used the ERα Neo KO model, the results presented clearly indicate that any residual activity of the variant of ERα present in the aorta of this mouse model is not sufficient to protect against atherosclerosis in female mice, indicating that the expression of the full-length ERα is not required and that ERα transcriptional activity mediated through the AF transactivation function is dispensable in protection against fatty-streak formation in the vessel wall.

Although our findings indicate that E2 protects against the development of atherosclerosis by an ERα- and lipid-independent mechanism, we cannot conclude that the early atheroprotective effects of E2 occurred entirely via non-receptor-mediated pathways. In C57BL/6J mice, E2 prevents the vascular injury response to carotid artery denudation independent of ERα [52], perhaps proceeding via the ERβ [53]. Although the response to vascular injury differs from that of the atherosclerotic response to high fat feeding, a mechanistic role for ERβ seems unlikely in our system as the ERβ does not appear to be involved in early stages of atherosclerosis development [6]. Estrogen receptor ERβ localized to the vascular intima is correlated with coronary calcification and thus more advanced atherosclerosis in pre- and postmenopausal women [54], suggesting a role for this ER subtype in more advanced vascular disease. Recently, however, heat shock protein (HSP) 27, an ERβ-associated protein, was demonstrated to have attenuated expression in coronary atherosclerosis and to modulate estrogen signaling in minimally diseased arteries [55]. Thus, in our system, vascular protective effects mediated by ERβ receptor, its isoforms [56], an as yet uncharacterized ER, HSPs, or other factors cannot be excluded and remain the work for future studies. Lastly, ERα- or ERβ-independent effects of E2 have been demonstrated for catechol estrogens (estrogens converted to catechol estrogens via oxidation) where inhibition of cell growth by E2 is independent from ERs [57].

Furthermore, in clinical studies, the common estrogen receptor-alpha (ER-alpha) IVS1-401 C/C genotype polymorphism identifies a group of women (approximately 20%) who have augmented response to hormone therapy. This has been demonstrated on levels of HDL cholesterol and intermediate cardiovascular risk markers, such as E-selectin [58, 59]. Although the impact of this polymorphism on ERα transcription and other estrogen-sensitive intermediate and clinical end points has not yet been established, the augmented effects of ER polymorphisms on hormone therapy regulation of lipids and other vascular mediators may be additional possible mechanisms by which ER polymorphisms might influence estrogen action in early atheroprotection. Thus, the implications of differing experimental results in various genetic models as extrapolations to human disease suggests that a polygenomic approach will be needed to fully understand whether exogenous estrogen treatments will be appropriate for all women and the timing of this treatment in relationship to atherosclerosis development.

In summary, this study addresses the atheroprotective role of estradiol that is uncoupled to that of the ERα and whether the mechanism of action of estradiol in atheroprotection is ERα dependent, or not, in early-stage lesions. Our findings demonstrate that, in early lesions of atherosclerosis, the mechanism of atheroprotection is estradiol-mediated and not dependent on the extent of hyperlipidemia, a reduction in serum lipids by estrogen, or on the presence or absence of the ERα. Our findings are important and provide new clues to the stage of vascular atherosclerosis at which estradiol and ERα protection occurs. Specifically, our observations establish that both endogenous and exogenous E2 retard the development of fatty-streak formation characteristic of early atherosclerosis in a manner that is closely tied to estrogen level, yet independent of a functional ERα. In contrast, and based on prior studies, the ERα is required for a major part, if not all, of the E2-mediated reduction in lesion size in plaques with more advanced lesion characteristics and is a major mediator in plaque maturation. Furthermore, the inhibitory effects of E2 on atherogenesis appear to be lost once atherosclerotic lesions are established [60].

In conclusion, our studies contribute to a better understanding of ERα-dependent and -independent mechanisms of E2 atheroprotection, linked to an emerging concept coupling estrogen action to the timing of its initiation in relationship to critical periods in the stages of atherogenesis. This knowledge is potentially very important, as one of the most critical areas of new focus in this field is understanding the molecular and phenotypic interface of estrogens with their receptors in the vascular wall vis-à-vis the stage(s) of atherosclerosis. A better understanding of this system may lead to a better understanding of how to optimize and when to target hormone therapies for cardiovascular disease protection in women.