Estrogen and the Vascular Endothelium: The Unanswered Questions

Abstract

Premenopausal women have a lower incidence of cardiovascular disease (CVD) compared with their age-matched male counterparts; however, this discrepancy is abolished following the transition to menopause or during low estrogen states. This, combined with a large amount of basic and preclinical data indicating that estrogen is vasculoprotective, supports the concept that hormone therapy could improve cardiovascular health. However, clinical outcomes in individuals undergoing estrogen treatment have been highly variable, challenging the current paradigm regarding the role of estrogen in the fight against heart disease. Increased risk for CVD correlates with long-term oral contraceptive use, hormone replacement therapy in older, postmenopausal cisgender females, and gender affirmation treatment for transgender females. Vascular endothelial dysfunction serves as a nidus for the development of many cardiovascular diseases and is highly predictive of future CVD risk. Despite preclinical studies indicating that estrogen promotes a quiescent, functional endothelium, it still remains unclear why these observations do not translate to improved CVD outcomes. The goal of this review is to explore our current understanding of the effect of estrogen on the vasculature, with a focus on endothelial health. Following a discussion regarding the influence of estrogen on large and small artery function, critical knowledge gaps are identified. Finally, novel mechanisms and hypotheses are presented that may explain the lack of cardiovascular benefit in unique patient populations.

Cardiovascular disease (CVD) remains the leading cause of death in men and women. While premenopausal women have a lower incidence of CVD compared with age-matched males, this difference is abolished following menopause (1, 2). The decline of plasma estrogen during both natural and early menopause is associated with an increase in cardiovascular disease risk (3). Supplementation with estrogen has been shown to improve vascular function in some studies (4, 5), suggesting that the hormone elicits cardioprotective benefits. However, clinical cardiovascular outcomes from estrogen-containing hormone therapy have been variable. Data have shown that long-term estrogen supplementation for contraception (oral contraceptives) can increase oxidative stress, promote endothelial dysfunction (6), and increase future incidence of hypertension and CVD (7). Older women on hormone replacement therapy (HRT) also have an increased risk for CVD (8, 9). Observational and survey studies have shown that compared with cisgender (cis) females and transgender (trans) males, trans females and biological males undergoing estrogen supplementation have higher rates of myocardial infarction and mortality due to CVD (10‐13).

Vascular endothelial dysfunction, or the loss of nitric oxide (NO) bioavailability due to increased oxidative stress, precedes the development of CVD and is a strong predictor of future cardiovascular pathology. Preclinical studies largely indicate that estrogen has a protective effect within endothelium by increasing production of NO, vascular endothelial growth factor, and other mediators that augment endothelial migration and proliferation (14, 15). However, it still remains unclear why these estrogen-induced, anti-atherosclerotic effects within the vasculature do not translate to improved CVD outcomes. The goal of this review is to explore our current understanding of the influence of estrogen on the human vascular endothelium and highlight novel mechanisms and hypotheses that may explain the apparent disconnect between estrogen supplementation and cardiovascular health. Discussion is focused on biological sex differences in vascular function associated with estrogen use, a topic important for understanding CVD risk in both cis and trans females.

Effect of Estrogen on Human Macrovascular Function

Endothelial dysfunction is an early indicator of CVD and quantification of brachial artery dilation in response to increased flow rates (flow-mediated dilation; FMD) is the most commonly used, noninvasive methodology for assessing endothelial health. Decreased FMD correlates with impairment of coronary artery function and strongly predicts future adverse cardiac events (16, 17). The meta-analysis by Green et al examined 20 studies (374 total comparisons) and concluded that the FMD response is primarily mediated by vasoprotective NO (18). The prognostic value and ease of the technique has allowed investigators to use FMD to study effects of age, estrogen, and menopause on large artery function and to correlate these effects to either the prevention or promotion of CVD (16, 17).

The HUNT3 Fitness Study assessed brachial artery FMD in 4739 healthy adults (20-89 years of age). A marked reduction in FMD in healthy men was noted in their 30s compared with healthy women in their 40s. FMD was consistently higher in women compared to men (25% higher in their 20s, 27% higher in 30s, 34% in 40s, 23% in 50s, 13.2% in 60s) until 70 years of age, after which no significant differences were observed (19). As for younger cohorts, Hopkins et al assessed FMD in youth (6-18 years of age) and found that the female youth had greater FMD compared to the male youth, with the greatest differences at 17 to 18 years of age. Only in the female group was a small increase in FMD observed after age 15, suggesting a critical window at the end of female puberty (higher estrogen [E2] levels) when sex differences in FMD become most apparent (20). These studies support the notion that FMD, a predominantly endothelial- and NO-dependent phenomenon, is influenced by biological sex and estrogen.

Aside from puberty, the transition to menopause is another period in which changes in FMD are apparent. Moreau et al reported a progressive reduction in brachial artery FMD in peri- (between 46 and 54 years old) and postmenopausal women (>52 years old) compared with premenopausal women. FMD in late postmenopausal women was almost 50% lower than that of premenopausal women, with the most significant reduction seen during the menopausal transition (approximately 35% decrease in FMD). These reductions remained significant after adjusting for metabolic risk factors, sex hormone concentrations, prior hormonal contraceptives/therapy, and vasomotor symptoms (21). Deciphering whether the reduction in FMD is due to loss of estrogen vs the normal aging process has proven challenging. However, vascular function studies among women with low endogenous plasma estrogen levels have added clarity. Inhibiting gonadotropin releasing hormone and thus lowering endogenous estrogen production by the ovaries, leads to a decrease in FMD in pre- and postmenopausal women, an effect reversed by estrogen supplementation (22). Kalantaridou et al reported a decrease in FMD in young women with premature ovarian failure compared with age- and BMI-matched healthy women. Prescription of oral estrogen/progesterone cyclical therapy for 6 months exhibited an improvement in FMD to values comparable to the control group (4). Similarly, physically active women with functional amenorrhea due to a suppressed hypothalamus-pituitary-axis exhibited a reduction in FMD (23), which was restored once the natural menstrual cycle returned or supplemental estrogen was administered (5). These observations extend into other large artery beds as well. In carotid arteries, Iwamoto et al demonstrated vascular function was dependent on estrogen levels as premenopausal women had higher FMD during the luteal phase (high plasma estrogen) of their menstrual cycle, which was found to be reduced during the menopause transition (24).

Estrogen-induced improvement in large artery function has also been observed in trans females undergoing gender-affirming therapy (25). In fact, brachial artery FMD in trans females taking ethinyl estradiol or conjugated equine estrogen (CEE) for at least 5 months, was reported to be similar to that of age-matched cis females (25). Notably, low serum testosterone levels in trans females, which serves as an index for adequate estrogen therapy, independently predicted FMD (25). In middle-aged cis males, endogenous estrogen as opposed to testosterone levels are positively associated with FMD, independent of age or lipid levels (26). Accordingly, in a placebo-controlled, double-blinded randomized clinical trial, a reduction in plasma estrogen through inhibition of aromatase in young healthy men led to an almost 43% decrease in FMD (27). Although it remains unclear whether the enhanced function in the vasculature associated with estrogen supplementation is due to estrogen alone vs a decrease in testosterone, the evidence that estrogen is responsible for improvement in large artery function is strong.

Effect of Estrogen on Human Microvascular Function

Most knowledge regarding the influence of estrogen on the endothelium has emerged from studying large conduit arteries; however, evidence continues to emerge that the health of the microvasculature, networks of small resistance arterioles that feed essentially every organ in the human body, may be better at predicting future CVD (28). Unlike conduit arteries where endothelial dysfunction manifests as a decrease in FMD, arterioles have the unique ability to maintain dilation during disease and compensate for the loss of NO by switching to the use of hydrogen peroxide (H₂O₂) to elicit smooth muscle relaxation (29). Although this transition in vasoactive mediator allows microvessels to dilate in response to increased flow (demand) despite disease, the formation of H₂O₂ creates a proinflammatory and prothrombotic vascular environment as opposed to the anti-inflammatory, anti-atherosclerotic milieu generated by NO.

Using the noninvasive approach of nailbed videocapillaroscopy, Clapauch et al reported a decrease in microvascular dilation in postmenopausal women, which improved following a brief, 1-hour exposure to 17β-estradiol (30, 31). Campisi et al measured myocardial blood flow using positron emission tomography (PET) during cold pressor testing in postmenopausal women and found that long-term estrogen therapy improved mean blood flow (marker of microvascular function) only in those without prior coronary risk factors (32). It is important to note that estrogen only improved vascular function in healthy postmenopausal women.

Our recent work interrogated the effect of age and estrogen in a sex-specific manner on the human microvasculature. Using isolated human resistance arterioles, we revealed that females maintain NO-mediated flow-induced dilation (FID) throughout their lifespan (30-78 years); however, exposure to 100 nM of E2 for 16 to 20 hours induced a switch from NO to H₂O₂-mediated dilation regardless of age/menopause status. In arterioles from females with coronary artery disease (CAD), a trend toward reduced FID was seen. This was in contrast to what was observed in microvessels from males. Following treatment with E2, vessels from cis-males males had a significant reduction in the vasodilatory capacity to flow regardless of disease status. In addition to impaired FID, the arterioles also did not fully dilate in response to the endothelium-independent dilator papaverine, suggesting dysfunction of the vascular smooth muscle (33). Elucidating the mechanisms that contribute to both endothelial-dependent and endothelial-independent dysfunction from E2 is necessary to reduce microvascular damage and potential future CVD risk in cis and trans females who are exposed to estrogen.

Lack of Translation to Improved Outcomes

Despite the vast amount of data showing a beneficial effect of estrogen on large artery FMD in both sexes, the question of whether estrogen supplementation decreases future cardiovascular risk remains debatable. Potential mechanisms and hypotheses that may explain the discordance between the vasculoprotective effects observed in preclinical/clinical studies and cardiovascular outcomes are discussed here.

Estrogen

Natural vs synthetic estrogen

Plasma estrogen is found in one of 3 forms: 17β-estradiol (E2), estrone, and estriol (concentrations detailed in Table 1). E2, the most abundant and potent form, is primarily produced by the ovaries in premenopausal women. Estrone, a less potent form of estrogen, replaces E2 as the most abundant circulating female hormone in post- vs premenopausal women (34), and levels of both decrease 10-fold with menopause. During menopause, aromatase found within adipose, skin, brain, and bone tissue is responsible for generating the majority of circulating E2 and estrone (61). Estriol is the primary pregnancy-related estrogen and is maintained at low levels in nonpregnant women (35). In males, the testes produce approximately 20% of circulating estrogens, with the remaining amount formed via aromatase conversion of testosterone to estrogen. Plasma concentrations of estrone are higher compared with E2 in males, with levels similar to postmenopausal females (34).

Table 1.

Summary of estrogen formulations, plasma concentration, and effects on the vasculature

Estrogen Types	Plasma Concentration	Effects on Vasculature
17β-Estradiol (E2)	Premenopause: 40–500 pg/mL (34) *Peak levels of 200–500 pg/mL during the pre-ovulatory luteinizing hormone surge; the end of the follicular phase Postmenopause: <20 pg/mL (34) Males: 10–40 pg/mL (34) Pregnancy: 5 ng/mL (35) Trans females: 100–500 pg/mL. Endocrine Society recommendation: <200 pg/mL (36)	Improved brachial artery FMD (37‐40) Increased NO production (41‐43) Increased ROS production (high dose (43); aged/senescent vessels (44)) Improved in vivo microvascular function (healthy postmenopausal females (30‐32)) Switch to H2O2-mediated dilation in arterioles from healthy females regardless of age (32) Impaired microvascular endothelial and smooth muscle function in biological males (33) Protection from senescence (45‐47) Upregulation of sphingolipid signaling (48‐50)
Estrone	Premenopause: 30–200 pg/mL (51) Postmenopause: 28–35 pg/mL (52) Males: <60 pg/mL (34) Pregnancy: 3–5 ng/mL (35)	Increased NO production (similar to E2) (53)
Estriol	Pregnancy: 1–3 ng/mL (35) Nonpregnant male and female: <0.2 ng/mL (35)	Improved FMD in older females (54) Decreased smooth muscle inflammation (55)
Conjugated equine estrogen (CEE)	Variable based on patient needs	Increased monocyte adhesion to endothelium (56) Lower NO production compared to natural estrogens (53) Improved FMD in healthy, early menopausal women (39, 40) and in trans females (25)
Ethinyl estradiol		No increase in NO production (57) Influence on FMD in females depends on the progesterone used for contraception (reviewed in (58)) Improved FMD in trans females (25)
Transdermal E2		Improved FMD and carotid artery compliance in postmenopausal women when coadministered with antioxidants (59, 60)

Abbreviations: FMD, flow-mediated dilation; NO, nitric oxide; ROS, reactive oxygen species.

Various forms of synthetic estrogen are used for oral contraception, hormone replacement, and gender affirmation. The most commonly used estrogen in oral contraceptives is ethinyl estradiol, typically prescribed in combination with progestins (62). Ethinyl estradiol is reported to be 500 times more potent than E2 and irreversibly binds estrogen receptors, which increases the duration of estrogenic actions at lower doses (63). For postmenopausal HRT, typically oral/transdermal E2 are prescribed in combination with progesterone. Oral conjugated equine estrogen (CEE) was prescribed more commonly prior to the Women's Health Initiative (64). CEE consists of E2 as well as many ring B unsaturated estrogens, which may account for its 2- to 6-fold higher affinity for estrogen receptors compared to E2 alone (65). Common formulations of E2 that are used for gender-affirming therapy in trans females include oral/transdermal E2 and injectable estradiol valerate and cypionate. Plasma levels of E2 in trans females are similar to those seen among premenopausal cis females (36).

It is apparent that the various estrogen formulations can elicit drastically different biological effects. It was recently reported that equilin, one of the estrogens within CEE, is not as effective at increasing endothelial nitric oxide synthase (eNOS) mRNA expression and activity compared to E2 or estrone (53). Similarly, ethinyl estradiol, the synthetic estrogen most commonly used for oral contraception, does not increase NO production in human endothelial cells (57). Much of our understanding regarding the vasculoprotective effects of estrogen stems from studies that utilize natural E2 as opposed to synthetic forms of the hormone. Studies focused on elucidating the effects of synthetic estrogens on the endothelium are warranted and may provide clarity as to the lack of clear benefit for those taking synthetic formulations of estrogen. Table 1 provides a summary of the known effects of natural and synthetic estrogens on the vasculature.

Estrogen metabolites and vascular health

Estrogens (E2 and estrone) are metabolized by various CYP-450 isoforms into 14 known metabolites (66), some of which remain bioactive. CYP1A1 and CYP1B1 are expressed within the vascular endothelium (67) and can both metabolize estradiol and estrone into 2-hydroxyestradiol/estrone and its methylated form, 2-methoxyestradiol/estrone, both of which are associated with vascular benefits (68). For instance, Barchiesi et al used human aortic smooth muscle cells to show that 2-methoxyestradiol inhibited proliferation and induced cell cycle arrest. The study also concluded that this estrogen metabolite is capable of preventing neointima formation in a rodent model of aortic injury (68). Whereas CYP1A1 favors 2-hydroxylation, CYP1B1 preferentially metabolizes estrogen to 4-hydroxyestradiol/estrone and 4-methoxyestradiol/estrone, compounds that unlike the anti-mitogenic actions of 2-methoxyestradiol, are associated with promoting abnormal cell growth and estradiol-induced cancers (69). While these studies shed light on how estrogen metabolites influence smooth muscle cells, there is a lack of information regarding these compounds and endothelial health. The fact that CYP1B1 activity is influenced by many factors, including but not limited to age, diet, smoking, lifestyle, and genetics (70), begs the question whether alteration in estrogen metabolites formed, from more beneficial to more ominous compounds, can damage the endothelium and may, at least in part, explain the lack of translation between preclinical studies showing improved CVD outcomes with hormone therapy and failed randomized clinical trials.

Estrogen Signaling in the Endothelium

There are 3 major isoforms of estrogen receptors (ERs): the ER-α, ER-β, and G protein–coupled estrogen receptor (GPER). All 3 ERs are widely expressed in the vasculature (71) and through their activation contribute to the acute (nongenomic) and chronic (genomic) effects of E2 (summarized in Fig. 1).

Figure 1.

Estrogen receptor signaling in the vascular endothelium. Membrane-bound ER-α co-localizes with calveolin-1 and signals through phosphatidylinositol 3-kinase (PI3K) and protein kinase B (Akt) as well as mitogen activated protein kinase (MAPK) to phosphorylate eNOS, triggering an increase in NO. ER-α signaling also increases intracellular calcium through phospholipase-C/inositol 14,5-triphosphate pathway which then subsequently binds calmodulin and activates eNOS (72). Membrane-bound ER-α can also activate eNOS in response to flow. GPER transactivates epithelial growth factor receptor (EGFR), and signals via ERK1/2/PI3K/Akt to activate eNOS (73). GPER upregulates calcium through the inositol 14,5-triphosphate and ryanodine receptors on the endoplasmic reticulum as well as transient receptor potential channel on the plasma membrane (74) which triggers calcium/calmodulin signaling to promote eNOS phosphorylation (75). GPER signaling also increases expression of transcription factor Kruppel-like-factor 2 (KLF-2) (74), which subsequently increases eNOS expression (76) and glycolytic pathways. Estrogen increases activity of sphingosine kinase, S1PR1, and S1P transporters, which increases NO production (49, 77). Nuclear estrogen receptors upregulate antisenescent proteins (45), increase expression of eNOS, and enhance neutral sphingomyelinase (50) /sphingolipid signaling. Image created using Biorender.com and published with permission.

Nuclear receptor signaling

Expression of nuclear receptors ER-α and ER-β is dynamic and can be modified depending on level of estrogen exposure. Chronically, high estrogen levels seem to upregulate ER-α and downregulate ER-β. Data from preclinical models show an increase in the ER-β:ER-α ratio during low estrogen states (age/menopause) in female rats (78). Using immunofluorescence techniques, Gavin et al identified that ER-α in endothelial cells collected from premenopausal women is highest during the late follicular phase when plasma E2 levels are elevated. This is in contrast to a 33% lower expression in endothelial cells from postmenopausal women. Although the endothelial cells originated from the antecubital vein as opposed to arteries, this further supports that ER-α positively correlates with plasma estrogen levels (79).

Evidence suggests that receptor expression not only changes with E2 levels, but also with disease status. For instance, Cruz and colleagues isolated small arteries (350 μm) from postmenopausal women diagnosed with CAD and using immunofluorescence showed that these arteries have a higher ER-β:ER-α ratio compared with arteries from age-matched women without CAD (80). Interestingly, immunostaining of coronary arteries from male cadavers (mean age 67 years) showed that the amount of ER-α was approximately half of the amount of ER-β identified in the intima. ER-β expression positively correlated with coronary calcification and atherosclerosis as measured via microradiography and histology, respectively (81). Important questions remain as to whether an elevated ER-β:ER-α ratio promotes vascular endothelial damage during disease vs if the shift toward increased ER-β is compensatory, or both.

The 2 nuclear estrogen receptors appear to share some commonality regarding function but there is a wide range of evidence suggesting the receptors can act in opposition. For instance, ER-α signaling promotes proliferation of breast cancer cells which is in sharp contrast to the antiproliferative effects observed with stimulation of ER-β (82). Similarly, in prostate cancer, ER-β exerts largely tumor-suppressive effects (83) while ER-α promotes tumor growth (84). Evidence that the nuclear receptors are the yin and yang of estrogen signaling is further solidified in the study by Bhavani and colleagues that used human recombinant ER-α and ER-β transfected into HepG2 cells. They elegantly showed that increasing the ER-β:ER-α ratio by 2:1 or 10:1 leads to an inhibition of ER-α transcriptional activity by 33% and 66%, respectively. Of note, this effect was observed even in the presence of higher concentrations of E2 (100 nM) (85), reinforcing the strong negative influence of ER-β. It is plausible that differences in the ratio of these 2 receptors can help determine the response to estrogen; however, there is a paucity of studies that delve into how this ratio ultimately affects endothelial health.

G protein–coupled estrogen receptor

In addition to the traditional role of the ligand-activated nuclear receptors, estrogen is also capable of binding to transmembrane receptors to mediate rapid signaling events. G protein–coupled estrogen receptor 1 (GPER) is a seven transmembrane-domain G protein–coupled receptor (GPCR) that shares numerous ligands with ER-α and ER-β, including E2 and the selective estrogen receptors (SERMs) such tamoxifen (72, 86). In the human microvasculature, endothelial expression of GPER is lower in males and postmenopausal females compared to premenopausal females (33). Endothelial-specific expression of GPER across sexes and with age remains unknown in conduit arteries and preclinical models.

The binding affinity of E2 to GPER is considerably lower (K_d 3-6 nM (72)) compared with that of the nuclear receptors (Kd 0.1-1 nM (87)), questioning the significance of GPER in vivo (E2 plasma concentrations in picomolar ranges). However, preclinical studies using GPER-knockout mice reported decreased NO production, increased atherosclerosis, and overall cholesterol levels with loss of GPER expression. Ovariectomized mice treated with the GPER agonist G1 also had lower atherosclerosis development (73) and hypertension (88). Of note, studies have shown that G1 can also bind certain variants of ER-α, thus effects are likely not specific to GPER activation (89). Moreover, while these studies establish the importance of GPER in maintaining beneficial cardiovascular health, they do not probe whether these are estrogen-dependent effects.

In fact, recent studies have questioned whether E2 directly activates GPER. Tutzauer et al reported that vasorelaxation of mice caudal arteries was seen in response to G1, an agonist of GPER, even when the receptor was knocked out. They further showed lack of GPER response to E2 in yeast and HEK293 cells expressing recombinant human GPER (90). Another group used MCF7 cells and showed that, unlike ER-β or ER-α, transfection of GPER did not influence estrogen-mediated rapid signaling (91). On the contrary, prior studies using COS7 kidney cells transfected with GPER have suggested that E2 does bind GPER (72). While the reason for these contradictory findings remains unclear, most of these studies utilized transfected GPER in a variety of cell models. Thus, whether changes in estrogen levels throughout the menstrual cycle, following menopause, or estrogen supplementation for contraception/hormone therapy influence GPER-mediated vascular health in vivo is a critical knowledge gap.

Estrogen and nitric oxide

Although historically thought to be localized in the cytoplasm, roughly 5% to 10% of ER-α and ER-β are found within the plasma membrane (91). Using mice models that specifically allow for nuclear ER-α activation while preventing membrane-bound ER-α signaling, Guivarc’h et al highlighted that nuclear ER-α primarily mediates transcriptional changes that confer cardiovascular protection; on the contrary, the membrane-bound receptors primarily contribute to acute NO production in response to estrogen (92). Membrane-bound ER-α which when bound to E2, co-localizes with calveolin-1 and signals through phosphatidylinositol 3-kinase (PI3K) and protein kinase B (Akt) as well as mitogen activated protein kinase (MAPK) to phosphorylate eNOS, triggering an increase in NO. ER-α signaling also increases intracellular calcium through phospholipase-C/inositol 14,5-triphosphate pathway which then subsequently binds calmodulin and activates eNOS (72) (Fig. 1).

Studies from ER-α knockout mice suggest that NO production in response to estrogen, and vasculoprotection in general, is primarily mediated by ER-α. Darblade et al showed that in ovariectomized mice, estrogen treatment increased NO production within the aorta of wild-type and ER-β knockout mice as opposed to ER-α knockout mice (93). This is further supported by Lu et al, who showed that a lack of ER-α expression in HUVECs prevents estrogen-induced activation of AKT and ERK pathways, known activators of NO production (94). GPER can also increase NO levels (41), by transactivating epithelial growth factor receptor which activates the ERK1/2/PI3K/Akt/eNOS pathway (73). GPER upregulates calcium through the inositol 14,5-triphosphate and ryanodine receptors on the endoplasmic reticulum as well as transient receptor potential channel on the plasma membrane (74) which triggers calcium/calmodulin signaling to promote eNOS phosphorylation (75). GPER signaling also increases expression of transcription factor Kruppel-like-factor 2 (KLF-2) (74), which subsequently increases eNOS expression (76). However, as mentioned, its importance in vivo remains questionable.

Using a porcine epicardial coronary ring model, Traupe et al showed that the ER-α agonist 4,4′,4″-(4-propyl-[¹H]pyrazole-13,5-triyl)tris-phenol (PPT) was responsible for eliciting both rapid (5 minutes) and sustained (60 minutes) NO-dependent dilation whereas administration of an ER-β agonist (2,3-bis[4-hydroxyphenyl]-propionitrile; DPN), resulted in a delayed and sustained dilation. The dilatory effect due to activation of ER-β was unaffected in the presence of the nitric oxide synthase inhibitor N-nitro-L-arginine methylester (L-NAME) but dilation was inhibited by charybdotoxin and apamin. This suggests that ER-β activation results primarily in an endothelium-hyperpolarizing mechanism of dilation rather than NO production (42). Further, vasoconstriction in response to the thromboxane A2 receptor agonist U46619 and overall oxidative stress production was reduced in the presence of PPT as opposed to DPN, implying that ER-α may be primarily responsible for E2's NO-producing effects. Of note, these observed effects were similar in both the male and female vasculature in pigs (42). This reiterates the notion that changes in the ER-β:ER-α ratio with age, disease, and estrogen levels may play an important role in regulating endothelial health and production of NO. Moreover, prior data show lower ER-α activation by CEE compared to endogenous estrogens (53). Numerous B-ring unsaturated estrogens within CEE also have a 2-fold higher affinity for ER-β relative to ER-α (95). Thus, postmenopausal women on oral CEE might not only have lower expression of ER-α, but also lower activation of the receptors present. This provides further evidence that E2's beneficial NO-producing effects may not translate with CEE specifically.

ER-α may promote NO production in resistance arteries independent of estrogen exposure, a concept that has recently emerged from work by Favre et al. By generating male mice with a mutation in membrane-bound ER-α that prevents anchorage to the membrane and caveolin-1, the investigators reported reduced FID in isolated mesenteric arteries. Activation of eNOS in these mice also remained unaltered by flow compared to WT mice. The remaining FID in the mutated mice was significantly abolished in the presence of PEG-catalase and restored with Mito-Tempol, suggesting that FID relied on production of mitochondrial reactive oxygen species (ROS). Of note, mice that selectively lacked nuclear ER-α but had intact membrane ER-α showed no reduction in FID or eNOS activation (96). This data suggests that membrane-bound ER-α can enhance NO bioavailability and reduce oxidative stress independent of ligand availability. Knowing that the ER-α:ER-β ratio decreases with age, menopause, and CAD, and that there is a reduction in overall estrogen level with both age and menopause, this begs the question whether the changes in ligand-independent signaling reduces NO production in older/diseased individuals and postmenopausal women.

While eNOS activation is associated with beneficial anti-inflammatory, antithrombotic, and anti-atherosclerotic effects, constitutive NO production through inducible nitric oxide synthase (iNOS) can lead to pathological levels of NO accumulation, peroxynitrite formation, and oxidative damage. Using HUVECs, Cho et al showed that while 1nM of E2 (48 hours) increased eNOS mRNA levels, this effect was less pronounced in cells treated with 1mM E2. The latter group exhibited a greater increase in iNOS expression, an observation that was associated with increased cell permeability compared with the 1nM E2 treatment. Notably, while the eNOS regulation was disrupted in the absence of calcium, iNOS modulation was calcium independent (43). Wong and colleagues similarly showed iNOS upregulation within the smooth muscle cells with higher doses of E2 (97). Whether vascular estrogen concentrations reach the higher dosage levels within cis and trans females undergoing estrogen therapy or taking oral contraception remains unclear. However, it is possible that noncyclic and constant estrogen exposure may increase iNOS-dependent NO production.

Estrogen and prostaglandins

In addition to NO, endothelial prostaglandin production through cyclooxygenase enzymes plays a critical role in mediating vascular tone and health. While conversion of prostaglandins to prostacyclin promotes an antithrombotic and vasodilatory response, conversion to thromboxane elicits constriction and thrombosis (98). As such, imbalances in thromboxane:prostaglandin levels contribute to various cardiovascular diseases (98). Li et al showed that compared to aortas from rats with intact ovaries, those from ovariectomized rats had a decrease in vasopressin-induced thromboxane A2 release due to lower endothelial and smooth muscle mRNA expression of cyclooxygenase-2 and thromboxane synthase enzymes. Estrogen supplementation in the ovariectomized rats prevented this effect (99), suggesting that estrogen potentiates prostanoid pathways associated with vasoconstriction and thrombosis. These findings are in agreement with other preclinical studies reporting increased vasoconstrictor response in females (100, 101). Clinically, studies show an increase in urinary thromboxane levels in postmenopausal women taking oral HRT (102), and a decrease in prostacyclin levels in women taking long-term estrogen-containing oral contraceptives (103), which may partly explain the association with thrombosis and hypertension in females taking contraception/menopausal hormone therapy (104, 105). Together, these studies question whether chronic estrogen therapy may shift prostaglandin ratios toward detrimental thromboxane-increasing pathways within the endothelium, and questions remain as to how these effects vary between cyclic vs noncyclic estrogen exposure as well as in trans females. It should be noted that numerous studies also report that estrogen increases vasodilatory and protective prostacyclin levels (106‐108); however, these effects are largely seen within the cerebral vasculature. Additionally, ER-α (42) and GPER (109) signaling can reduce vasoconstrictive prostaglandin production, an effect not associated with ER-β (42). This highlights that variations in receptor levels with age, menopause, and CAD all likely influence the dynamic regulation of beneficial vs detrimental prostaglandin production in response to estrogen (107).

Aging and Resistance to Estrogen Signaling

An intriguing explanation for the mixed results in postmenopausal HRT trials, referred to as the “timing-hypothesis,” proposes that the longer amount of time that has passed in a low estrogen state increases the likelihood of estrogen-induced damage once it re-enters the circulation. The “timing-hypothesis” is supported by the fact that acute estrogen-induced vasodilation is reduced in uterine arteries from post- compared to premenopausal women (110). Acute and chronic oral E2 therapy leads to a greater FMD increase in women less than 5 years since menopause compared with those who have experienced menopause for over 5 years (111), providing further evidence that timing is key. In line with this finding, randomized clinical trials have found no difference in FMD following CEE-containing hormone therapy in older healthy women and females diagnosed with CAD (average age 65) (37, 38), compared to early menopausal women (<64 years) who displayed an improvement in FMD with hormone therapy (39, 40).

More recently, follow-up of the Early vs Late Intervention Trial with Estradiol (ELITE) showed that estrogen levels following HRT were negatively associated with carotid intimal-media thickness (CIMT) in early menopausal women (<6 years since menopause) but positively associated with CIMT in late menopausal women (>10 years since menopause). Women in both groups had similar plasma estrogen levels following hormone therapy (112). Interestingly, the rate of progression of CIMT was also lower in early menopausal women on hormone therapy compared to those on placebo, but no difference was seen in late menopausal women. On the contrary, plaque burden within the coronary arteries and total stenosis was similar between women treated with hormone therapy and those receiving placebo, regardless of time since menopause (113). The investigators concluded that estrogen prevents arterial lesion formation or slows progression of atherosclerosis but does not reverse established lesions. Furthermore, it is plausible that estrogen depletion likely leads to accelerated vascular damage and such irreversible arterial aging and/or arteriolosclerosis limits the beneficial effects of estrogen on the vasculature in older women. Potential mechanisms as to how estrogen may damage the vasculature are described here.

Endothelial senescence

Endothelial senescence, defined as a decrease in cell proliferation and stable cell cycle arrest, is a hallmark of chronological aging and is strongly linked to CVD. Senotherapy, or the removal of senescent cells specifically in the vasculature, has therefore emerged as a promising therapy in the fight against heart disease. Studies suggest that the degree of vascular senescence may have a significant role in determining how a blood vessel will respond to exogenous estrogen. For example, using isolated carotid arteries from ovariectomized non-senescent (SAMR) and senescence-accelerated (SAMP) mouse models, the vasoconstrictive response to phenylephrine was reduced in vessels from the SAMR mice regardless of early- vs late-onset of estrogen supplementation, implying that estrogen maintains its dilatory effect regardless of how much time was spent in an estrogen-depleted state. This was in contrast to the augmented phenylephrine-induced constriction observed in the SAMP mice treated with late-onset estrogen (45 days since ovariectomy) as opposed to early-onset (day of ovariectomy). Of note, an increase in thromboxane A2 and a decrease in NO production was associated with the enhanced constriction (114). In another study, ovariectomized SAMR mice showed reduced NO production in response to acetylcholine receptor activation, which was restored with E2 supplementation. This is in contrast to SAMP mice where E2 supplementation did not restore NO production (78). Similarly, unlike in arteries from young male mice (6 months), mesenteric arteries from wild-type aged mice (24 months) did not dilate in response to 10⁻⁹ to 10⁻⁷ E2 administration (115). These studies elegantly highlight the potential role of aging and senescence in the vascular response to estrogen; however, questions remain as to how E2 signaling is mechanistically altered in senescent vs non-senescent endothelial cells.

Estrogen can also protect against vascular senescence, an observation that has been credited to influencing many processes, including but not limited to, decreasing autophagy (114), increasing telomerase (116), and increasing anti-senescent sirtuin 1 (SIRT1) (45) expression. In cultured human endothelial cells, estrogen has been shown to prevent senescence induced by oxidized low-density lipoprotein (45) and high cell passage (46). On the contrary, a recent study reported that trans females have an increase in circulating levels of plasminogen activator inhibitor-1 (marker of senescence) compared to cis males (117); however, whether this is associated with endothelial senescence remains unknown. The study was also underpowered to assess differences between those on vs not on hormone therapy. While the link between estrogen and cellular senescence exists, there are many knowledge gaps regarding the timing and dosage of estrogen that prevents or potentially promotes this age-associated process.

Oxidative stress

The oxidative stress theory of aging hypothesizes that accumulation of oxidative damage within cells is largely responsible for age-related functional loss. Although the mechanisms of oxidative stress–induced aging remain elusive, low estrogen states such as menopause are associated with both an increase in ROS and age-related vascular changes (eg, increased arterial stiffness). Menopause has been linked to deficiencies in both the eNOS substrate L-arginine (118) and tetrahydrobiopterin (BH4; an essential cofactor for eNOS (119)), deficiencies that both favor uncoupling of eNOS and production of oxidative stress. BH4, which decreases with age, also suppresses NADPH oxidase (NOX) activity to suppress superoxide production (120). A reduced amount of BH4 was found in aortas from ovariectomized rats whereas estrogen supplementation restored the cofactor (121). The positive relationship between estrogen and BH4 was also highlighted by a study showing that estrogen upregulates expression of the BH4-producing enzyme GTP cyclohydrolase I in response to hyperglycemia in bovine aortic endothelial cells (122).

BH4 administration improves endothelium-dependent dilation (brachial FMD) after a high-fat meal in postmenopausal women (123). Furthermore, carotid artery compliance and brachial artery FMD was shown to improve 3 hours after BH4 administration and 2 days after transdermal estradiol in estrogen-deficient postmenopausal women. Notably, the combination of both did not lead to any additional improvements, and this effect was not observed in premenopausal women (59). In a prior study, the same group reported similar levels of improvement in carotid artery compliance following vitamin C administration (60), which promotes conversion of BH3 to BH4 (124). Exposure to estrogen in aged vessels may therefore uncouple eNOS, leading to production of ROS as opposed to the vasculoprotective compound NO. Restoring BH4 therapeutically might be an attractive strategy to improve CVD outcomes in those being treated with exogenous estrogen therapy.

Another source of vascular ROS production is NOX, an enzyme whose expression and activity is positively associated with male sex, estrogen depletion, and aging (125, 126). Although the significance of GPER remains questionable, Meyer at al reported a reduction in oxidative stress–related cardiovascular pathologies in aged male GPER-knockdown mice. More specifically, they showed that GPER signaling is associated with increased superoxide production through NOX1 in aorta of aged mice; thus, knockdown of GPER improved aortic endothelium-dependent vasodilation (44). The authors went on to propose GPER inhibitors as a novel drug class for decreasing NOX-dependent oxidative stress, as both pharmacological and genetic reduction of GPER reduced NOX-1 expression (44). Whether these results reflect ligand-independent signaling effects of GPER remains unclear and explorations of these pathways in older females and following menopause are also needed.

Endothelial sphingolipid balance

Evidence continues to emerge that sphingolipids, a group of biologically active signaling lipids, have a significant impact on vascular function. Accumulation of ceramide can cause vascular endothelial dysfunction (127), but it can be metabolized to sphingosine-1-phosphate (S1P) which stimulates NO production through S1P-receptor 1 (S1PR1) (48). Estrogen administration to cultured endothelial cells has been shown to result in a rapid and transient increase in sphingosine kinase activity and S1P production. This effect, along with activation of eNOS, was abolished in cells treated with siRNA targeted to sphingosine kinase or S1PR1, highlighting the critical role of S1P formation in mediating estrogen's ability to increase NO (49). Estrogen also increases activation of ATP-binding cassette transporters C1, G2, and S1P transporter spinster homolog 2 (SPNS2) (77), which favors extracellular export of S1P allowing it to activate its receptor in an autocrine manner.

In rat models of menopause, the ceramide:S1P rheostat shifts toward decreased ceramide metabolism, resulting in higher levels of ceramide and lower levels of S1P (50). In humans, plasma ceramide has been shown to not only increase with age but also negatively correlates with plasma estrogen levels in women (128). This aligns with evidence that premenopausal women have higher levels of plasma S1P compared with postmenopausal women as well as men (77). These findings support the role of estrogen in modulating the sphingolipid rheostat and suggest that the loss of estrogen may tip the scale toward ceramide accumulation and increased risk of future CVD.

On the contrary, other studies indicate that estrogen can stimulate an increase in neutral sphingomyelinase (NSmase) mRNA (50), a flow-sensitive, ceramide-forming enzyme within caveolae of endothelial cells. An intriguing question is whether the cyclical nature of natural estrogen during the menstrual cycle allows for bursts of estrogen to stimulate production of ceramide, which gets converted to S1P, providing vascular benefit. On the contrary, continuous exposure to estrogen via hormone therapy or oral contraceptives may result in tipping the sphingolipid balance toward high levels of ceramide.

Conclusion

Estrogen supplementation continues to be a critical therapy for many women who experience ovarian insufficiency, menopausal symptoms, or for the purpose of contraception as well as gender affirmation. This review highlights the critical knowledge gaps (Table 2) regarding how estrogen influences vascular endothelial function and presents potential mechanisms (summarized in Fig. 2) to explain the discrepancies observed between preclinical studies and human in vivo studies. The amount of prescribed estrogen is likely to increase due to the aging female population entering menopause as well as projected increased use for contraception and gender affirmation. Further exploration of the pathways described here may lead to novel therapies that meet the needs of any gender while also supporting a strong and robust vascular system.

Table 2.

Critical knowledge gaps regarding the influence of estrogen on the vascular endothelium

Key Knowledge Gaps
Understanding of human micro- and macrovascular function in cis and trans females in response to changes in estrogen levels Does microvascular endothelial function vary throughout the menstrual cycle, during menopausal transition, and postmenopause in cis females? How do changes in estrogen vs testosterone levels influence large vessel and microvascular flow-mediated dilation in trans females? How does exogenous chronic estrogen therapy influence microvascular endothelial function in cis and trans females?
Signaling differences across estrogen receptors with age, menopause, and disease and influence on vascular health Is an elevated ER-β:ER-α ratio causative in promoting vascular damage during disease or is it compensatory following disease? Are there sex, age, and disease-based differences in estrogen-mediated gene transcription? Is ligand-independent signaling impaired in older/diseased individuals and postmenopausal women? Do changes in plasma estrogen levels influence GPER-mediated vascular health in vivo? What is the role of GPER with age and change in estrogen status (eg, menopause)?
Effect of natural vs synthetic estrogen formulations as well as cyclic vs chronic estrogen exposure on endothelial health The influence of various estrogen formulations as well as mono/phasic administration on endothelial health/function. Does this change with change with age, menopause, and/or disease? Are ER-α and ER-β levels regulated similarly in response to synthetic estrogens and chronic estrogen treatment? Does endothelial gene transcription, NO production, prostaglandin production, and sphingolipid balance vary based on estrogen formulation and mono/phasic administration? Does noncyclic estrogen therapy increase iNOS expression/activity in comparison to eNOS upregulation seen with cycling/low levels of estrogen?
Effect of senescence and aging on endothelial response to estrogen Is estrogen signaling impaired in senescent vs non-senescent endothelial cells? What is the timing and dose of estrogen that prevents endothelial senescence?

Figure 2.

Summary overview of how estrogen levels relate to cardiovascular disease (CVD) risk in cis and trans females and proposed mechanisms contributing to endothelial (dys)function. Image created using Biorender.com and published with permission.

Abbreviations

Akt

protein kinase B

BH4

tetrahydrobiopterin

CAD

coronary artery disease

CEE

conjugated equine estrogen

CIMT

carotid intimal-media thickness

CVD

cardiovascular disease

DPN

2,3-bis(4-hydroxyphenyl)-propionitrile

E2

estrogen (17β-estradiol)

eNOS

endothelial nitric oxide synthase

ER

estrogen receptor

FID

flow-induced dilation

FMD

flow-mediated dilation

GPER

G protein–coupled estrogen receptor

H₂O₂

hydrogen peroxide

HRT

hormone replacement therapy

HUVEC

human umbilical vein endothelial cell

iNOS

inducible nitric oxide synthase

KLF-2

Kruppel-like-factor 2

NO

nitric oxide

NOX

NADPH oxidase

PI3K

phosphatidylinositol 3-kinase

PPT

4,4′,4″-(4-propyl-[1H]pyrazole-13,5-triyl)tris-phenol

ROS

reactive oxygen species

SAMR

non-senescent mouse model

SAMP

senescence-accelerated mouse model

S1P

sphingosine-1-phosphate

S1PR1

sphingosine-1-phosphate receptor 1

Funding

NHLBI R01HL160752 (J.K.F.), NHLBI K08HL141562-04S1 (J.K.F.), NHLBI K08HL141562 (J.K.F.), and American Heart Association predoctoral fellowship grant #909315 (G.S.).

Disclosures

None.

Data Availability Statement

Data sharing is not applicable to this review article as no new datasets were generated or analyzed.