|Home | About | Journals | Submit | Contact Us | Français|
Pharmacologically increased estrogen levels have been shown to lower hepatic and plasma proprotein convertase subtilisin/kexin type 9 (PCSK9) levels in animals and humans. We hypothesized that physiological changes in estrogen levels influence circulating PCSK9, thereby contributing to the known wide inter-individual variation in its plasma levels, as well as to the established increase in LDL cholesterol (LDL-C) with normal aging. Circulating PCSK9, estradiol, and other metabolic factors were determined in fasting samples from 206 female and 189 male healthy volunteers (age 20–85 years), The mean levels of PCSK9 were 10% higher in females than in males (P < 0.05). PCSK9 levels were 22% higher in postmenopausal than in premenopausal (P < 0.001) females. Within the group of premenopausal females, circulating PCSK9 correlated inversely to estrogen levels, and PCSK9 was higher (305 ng/ml) in the follicular phase than in the ovulatory (234 ng/ml) or the luteal (252 ng/ml) phases (P < 0.05). Changes in endogenous estrogen levels during the menstrual cycle likely contribute to the broad inter-individual variation in PCSK9 and LDL-C in normal females. PCSK9 levels increase in females after menopause but not in men during this phase in life. This likely contributes to why LDL-C in women increases in this period.
Elevated plasma LDL cholesterol (LDL-C) is a major risk factor for cardiovascular disease (1). The liver has a key role in the regulation of LDL-C, controlling its synthesis and catabolism, the latter through modulating the number of LDL receptors (LDLRs) (2). LDLRs are regulated by transcriptional and posttranscriptional mechanisms. In addition to transcriptional control of LDLRs via the sterol-regulatory element binding protein (SREBP)-2 pathway (3), the importance of posttranscriptional regulation by proprotein convertase subtilisin/kexin type 9 (PCSK9) has been established lately (4). Circulating PCSK9, presumably secreted from the liver, binds to LDLRs and promotes their lysosomal degradation (4–6). Plasma levels of PCSK9 generally correlate with those of LDL-C (7, 8). Also, PCSK9 is regulated by SREBP-2 (3), and there are gain- or loss-of-function genetic variants of PCSK9 that show altered affinity to the LDLR (8–12). Furthermore, animal and human studies have shown that PCSK9 is also controlled by hormones such as glucagon (13), estrogen (14), growth hormone (14, 15), insulin (13, 16), and thyroid hormone (17).
LDL-C levels are known to increase by 30–50% during normal aging, mainly due to a reduced plasma clearance rate of LDL particles (18, 19). This age-related increase in LDL-C is more pronounced in women, particularly following menopause (20). Also, PCSK9 increases with age, but only in women (8). The mechanism for these gender differences is unclear. We previously reported that pharmacologic doses of estrogen strongly reduce hepatic PCSK9 mRNA and protein levels in the rat (13). Further, we found that the induction of endogenous estrogens to supraphysiological levels in conjunction with in vitro fertilization reduces plasma PCSK9 and LDL-C levels (14). However, from post hoc analysis of postmenopausal women receiving estrogen substitution therapy, it has been reasoned that loss of endogenous estrogen may not be the cause for increased PCSK9 with age (8).
In the present work, we aimed to evaluate whether the physiological changes of estrogen levels that occur during the menstrual cycle or with normal aging are linked to changes of circulating PCSK9 and LDL-C. We also explored to determine whether serum PCSK9 levels in a cohort of normal women and men were linked to variation in plasma lipids, glucose or insulin levels, or markers of cholesterol and bile acid turnover.
From a previously described cohort of 435 healthy Swedish volunteers (21), samples from 395 subjects (206 females and 189 males) were available for analysis. Their baseline characteristics are shown in Table 1. Subjects had been excluded if they had a previous history of coronary heart disease, type 2 diabetes, or dyslipidemia, or were under any medication including postmenopausal hormonal therapy or contraceptive drugs. Other lifestyle parameters, such as smoking and alcohol consumption, were also considered. Less than 10% of this cohort were smokers and less than 2% had an excessive intake of alcohol; none of these factors could be seen to influence other measurements.
Serum from samples drawn after an overnight fast was stored at −80°C (21).
All subjects gave informed written consent to participate in the study, which was approved by the Ethics Committee of Karolinska Institutet. In females, menopausal status and menstrual phases were evaluated from questionnaires.
Serum PCSK9 was determined with a quantitative sandwich enzyme immunoassay ELISA (catalog number Circulex CY-8079; CycLex Co., Ltd., Japan), according to manufacturer’s instructions. Concentrations of PCSK9 levels are given in nanograms per milliliter.
Serum estradiol [(E2) 1,3,5(10)-estratriene-3,17β-diol; 17β-estradiol] and follicle-stimulating hormone (FSH) were determined by ELISAs according to the manufacturer’s instructions (β-estradiol, catalog number DE2693; FSH, catalog number DE1288; Demeditec Diagnostics GmbH, Germany). E2 concentration is expressed in picograms per milliliter, FSH in milliunits per milliliter. ELISA plates were analyzed with a Tecan plate reader (Infinite® 200 pro series).
Serum insulin was assayed by a solid phase two site enzyme immunoassay (Ultra-sensitive human insulin ELISA kits; catalog number 10-1132-01; Marcodia, Sweden) according to manufacturer’s protocol. Concentrations of insulin are expressed in milliunits per liter (1 mU/l = 6.0 pmol/l). The homeostasis model assessment-estimated insulin resistance (HOMA-IR) was calculated according to the formula [insulin (mU/l) × fasting plasma glucose (mmol/l)]/22.5.
Total and HDL cholesterol (HDL-C), TGs, and glucose were determined using routine clinical chemistry techniques. LDL-C was calculated according to Friedewald, Levy, and Fredrickson (22). Serum levels of 7α-hydroxy-4-cholestene-3-one (C4), a marker of bile acid production, and unesterified lathosterol, a marker of cholesterol synthesis, were determined as described (21).
Data are presented as means ± SD. The significance of differences within multiple groups was compared by one-way ANOVA followed by Tukey’s post hoc comparisons test. In Table 1, the significance between two groups (males vs. females and premenopausal vs. postmenopausal) were tested by unpaired t-test. Frequency distributions in Fig. 1 were assessed by chi-square analysis. Correlations in Table 2 were analyzed by nonparametric Spearman correlation test. GraphPad Prism version 5 (GraphPad Software, SanDiego, CA) was used.
Fasting levels of PCSK9 ranged from 113 to 831 ng/ml, with a mean level of 290 ng/ml. The distribution of PCSK9 levels was skewed, but was normalized after logarithmic transformation (Fig. 1A). In line with previous results (7, 8), there was a clear difference in PCSK9 levels between males (mean, 276 ng/ml) and females (304 ng/ml; P < 0.05). This gender difference was explained by the fact that females >50 years of age had higher levels (mean, 330 ng/ml) than those <50 years of age (276 ng/ml; P < 0.05), whereas PCKS9 levels were similar in the corresponding groups of male subjects (Fig. 1B–G). Thus, with advancing age, both PCSK9 and LDL-C levels increased in females, whereas only LDL-C increased in males (Fig. 2).
To further characterize the influence of endogenous estrogen on PCSK9 in the females, we divided them according to their reported menopausal status. There were markedly lower E2 levels in postmenopausal as compared with premenopausal women, and a corresponding increase in their FSH levels (Table 1). There was also a negative correlation between LDL-C and E2 in females (Rs = −0.22; P < 0.01). We next compared the levels of PCSK9, LDL-C, and total cholesterol between the two groups, and could establish that they were all significantly higher in postmenopausal women (Table 1). Our results also demonstrated positive correlations between PCSK9 and LDL-C within all (Rs = 0.43; P < 0.0001), premenopausal (Rs = 0.42; P < 0.0001), and postmenopausal (Rs = 0.24; P < 0.05) females, and in all males (Rs = 0.26; P < 0.01) (Fig. 3A–D).
To further explore the relation between PCSK9 and E2, we analyzed their relationship within the female cohort. There was an inverse relationship in premenopausal women (Rs = −0.27; P < 0.01), but not in postmenopausal women (Rs = −0.12; NS; Table 2).
The fact that PCSK9 was related to E2, but not to age, in the premenopausal group prompted us to explore its relation to the menstrual cycle. Based on their reported menses, these subjects were further divided into three groups (23): follicular phase (days 1–11), ovulation phase (days 12–18), and luteal phase (days 19–32). As expected, the E2 level was highest at ovulation (Fig. 4A). PCSK9 levels were 235 ng/ml in this phase, and higher in the luteal (253 ng/ml) and follicular (305 ng/ml) phases (Fig. 4B). LDL-C and total cholesterol levels showed the same trend (Fig. 4C, D), whereas there were no differences in HDL-C or TG levels (Fig. 4E, F).
In agreement with previous reports (7, 8), PCSK9 correlated positively with total cholesterol and LDL-C, BMI, glucose, insulin, HOMA index, and TG levels, whereas it did not correlate with HDL-C (Table 2).
Of some interest was the fact that insulin correlated more strongly than glucose levels with PCSK9 in both males and females, and that no correlation was seen within PCSK9 and glucose levels in premenopausal group (Table 2). Calculated HOMA index was also positively correlated with PCSK9 in both genders (Table 2). Further, plasma PCSK9 showed a positive correlation with the marker of cholesterol synthesis, lathosterol/cholesterol, in males (Rs = 0.32; P < 0.001). There were no correlations between plasma PCSK9 and bile acid synthesis, measured as C4c (not shown).
The fasting level of PCSK9 in healthy subjects has a marked inter-individual variation, and it is influenced by age and gender (7, 8, 11). We here demonstrate that, in menstruating women, the estrous cycle strongly relates to PCSK9 levels. These changes in estrogen, as well as the loss of endogenous estrogen after menopause, are likely to influence the levels of LDL-C in normal women. The concept of estrogen as a modulator of PCSK9 is in agreement with previous reports by Persson and colleagues, who showed that estrogen can reduce circulating LDL-C at pharmacological or supraphysiological levels by downregulation of hepatic and plasma PCSK9 in both animals and humans (13, 14). It was previously reported that LDL-C follows an opposite trend to estrogen during the menstrual cycle (24). However, post hoc analysis of subjects on estrogen treatment in a population study did not suggest any obvious effects of exogenous estrogens on PCSK9 levels (8). If this finding is related to the fact that endogenous and exogenous estrogens may have different effects on lipid and lipoprotein metabolism, as shown previously in oophorectomized women (25), is unclear.
While the present study presents strong support for the view that increasing levels of PCSK9 following gradual loss of endogenous estrogen at least partly explain the marked increase of LDL-C that occurs with normal aging in women, additional mechanisms are probably also involved. Furthermore, it is obvious that while the increase in LDL-C with age in males was rather similar to that seen in females, there was no evidence of an increase in PCSK9 in men. Other hormone deficiencies, such as those of growth hormone (15) and thyroid hormone (26), could contribute to the age-dependent reduction of LDLR function and ensuing increase in circulating LDL-C. In particular, the potential role of testosterone (in both genders) still remains to be studied. The possibility that estrogen may exert some of its effects on LDLRs via modulation of growth hormone secretion should also be considered (14, 27).
It has been shown that circulating PCSK9 levels relate to insulin and glucose levels (7, 8, 28). This was confirmed in the present study. Although glucose was not related to circulating PCSK9 in premenopausal females, the calculated HOMA index indicated an influence of PCSK9 on insulin sensitivity. Brouwers et al. (29) reported that PCSK9 levels are higher in patients with familial hyperlipidemia, and that they correlated with markers of cholesterol biosynthesis. In the present work, we found a positive correlation between PCSK9 and lathosterol/cholesterol levels in healthy males, but not in females. This probably reflects their coordinated regulation through the SREBP-2 pathway (30–32). In contrast to the finding of a coregulation of bile acid synthesis and PCSK9 in mice reported by Parker et al (33), we did not observe any relationships between these variables in healthy humans. This may well be the consequence of yet another marked species difference of cholesterol metabolism, but needs further study.
In conclusion, PCSK9 levels increase in females after menopause. Variation of endogenous estrogen levels during the menstrual cycle likely contributes to the inter-individual variation in PCSK9 and LDL-C in normal females, and may partly explain the age-related increase in LDL-C in this gender. Further studies on the role of hormonal regulation of circulating PCSK9 in the control of LDL-C in humans should be of great interest.
The authors thank Mrs. Ingela Arvidsson and Mrs. Lisbet Benthin for their expert technical assistance.
This study was supported by the Swedish Research Council, the Stockholm City Council (ALF), the Swedish Heart-Lung Foundation, the Fondation Leducq, the NovoNordisk Foundation, the Knut and Alice Wallenberg Foundation, and the Cardiovascular Program, Karolinska Institute/Stockholm City Council. The authors have nothing to disclose in relation to this work.