Elevated plasma ceramide levels in post-menopausal women: a cross-sectional study

Plasma ceramide levels in women are positively correlated with age

An initial analysis of the data revealed the existence of a statistically detectable interaction between sex and ceramide levels, which prompted us to focus our attention on age-dependent changes occurring in either sex. The scatter plot reported in Fig 1A illustrates the total ceramide levels in plasma of individual women aged 20-78 years. Pearson’s analysis of the data revealed a statistically significant positive correlation between ceramide levels and age (r = 0.378; p = .0004). Because the largest accrual in plasma ceramides occurred between the age of 40 and 50 years, which is coincident with menopause, in a secondary analysis we grouped the data according to the subjects’ menopausal status. We found a statistically detectable difference between pre-menopausal women (20-54 years) and post-menopausal women (47-78 years) (Fig 1B). In particular, the levels of long-chain ceramide (d18:1/18:0) (p = .0035, unpaired Student’s t-test), very long-chain ceramides (d18:1/24:0) (p = .0012) and (d18:1/24:1) (p < .0001), and dihydroceramide (d18:0/24:1) (p = .0340) were higher in post-menopausal relative to pre-menopausal women (Fig 1B). No differences were found in the levels of ceramide (d18:1/16:0) (p = .3526) and dihydroceramide (d18:0/24:0) (p = .3633). In contrast with these findings in women, men showed no significant age-dependent increases in plasma ceramides (r = 0.143; p = .208) (Fig 2A). Male subjects in the age groups 19-54 and 55-80 years displayed comparable levels of circulating ceramide (d18:1/18:0) (p = .7112), (d18:1/24:0) (p = .7895), (d18:1/24:1) (p = .0847) and dihydroceramide (d18:0/24:1) (p = .9014). However, dihydroceramide (d18:0/24:0) was significantly lower in men >55 years, compared to younger men (p = .0003) (Fig 2B).

Figure 1. Scatter plot of plasma ceramide concentrations in women aged 20 to 78 years. (A) Total ceramide levels in 84 female subjects included in the study. Pearson’s correlation is considered statistically significant at p < .05. (B) Average levels of individual ceramide species in pre-menopausal women (20-54 years, n = 44, open bars) and post-menopausal women (47-78 years, n = 40, closed bars). Results are expressed as mean $\pm$ SEM. *p < .05, **p < .01, ***p < .001; unpaired Student’s t-test.

Figure 2. Scatter plot of plasma ceramide concentrations in men aged 19 to 80 years. (A) Total ceramide levels in 80 male subjects included in the study. Pearson’s correlation is considered statistically significant at p < .05. (B) Average levels of individual ceramide species in men aged 19-54 years (n = 48, open bars) and 55-80 years (n = 32, closed bars). Results are expressed as mean $\pm$ SEM. *p < .05, **p < .01, ***p < .001; unpaired Student’s t-test.

In an additional analysis, we compared total ceramide levels in plasma of pre- and post-menopausal women with those measured in men of the same age group (Fig 3). The results show that pre-menopausal women had significantly lower levels of circulating ceramides (p < .05, 2-way ANOVA followed by Bonferroni post-hoc test) relative to men of similar age (Fig 3A). The difference disappeared after menopause (p > .05) (Fig 3A).

Figure 3. Plasma ceramide and estradiol concentrations in men and women. (A) Plasma ceramide levels in, left, pre-menopausal women (20-54 years, n = 44) and age-matched men (19-54 years, n = 48) and, right, post-menopausal women (47-78 years, n = 40) and age-matched men (55-80 years, n = 32). (B) Plasma estradiol levels in, left, pre-menopausal women (20-54 years, n = 44) and age-matched men (19-54 years, n = 48) and, right, post-menopausal women (47-78 years, n = 40) and age-matched men (55-80 years, n = 32). *p < .05, **p < .01, ***p < .001; 2-way ANOVA followed by Bonferroni post-hoc test (women 20-54 years versus men 19-54 years). # p < .05, ## p < .01, ### p < .001; 2-way ANOVA followed by Bonferroni post-hoc test (women 20-54 years versus women 47-78 years).

Variables associated with plasma ceramides

Next, we used multivariable linear regression models to test the association between age and ceramides and adjust for potential covariates for which data had been collected (Table 1). These factors included hypertension (32/164 subjects, 16 women), tobacco smoking (23/164 subjects, 17 women), use of anti-hypercholesterol (12/164 subjects, 3 women) or contraceptive agents (6/164 subjects, 6 women), obesity (0/164 subjects) and diabetes (0/164 subjects). We did not take into account the number of cigarettes smoked as a variable of multivariable linear regression analysis. Plasma lipid levels were not collected for these samples. The adjusted linear regression analysis confirmed that ceramide (d18:1/24:0) (β (SE) = 5.67 (2.38); p = .0198) and ceramide (d18:1/24:1) (β (SE) = 2.88 (0.61); p < .0001) were positively associated with age in women and also, unexpectedly, revealed an opposite, albeit weaker, trend with ceramide (d18:1/16:0) (β (SE) = -.08 (.04); p = .0285) (Table 2A). Interestingly, the analysis also pointed to a significant association, found only in women, between hypertension and ceramide (d18:1/16:0) (β (SE) = 4.78 (1.42); p = .0012) and (d18:1/18:0) (β (SE) = 2.48 (.97); p = .0126) (Table 2A). Of note, 15 out of 16 women affected by hypertension were in the post-menopausal group. In men (Table 2B), the analysis unmasked a statistically detectable association between age and ceramide (d18:1/24:1) (β (SE) = 1.86 (.77); p = .0179). Finally, we found no associations, in either sex, between ceramide levels and any other variable, including smoking, contraceptive use or anti-hypercholesterol agents, obesity and diabetes (Table 2A-B).

Plasma ceramide levels are negatively correlated with estradiol in women, but not in men

Next, we investigated a possible association between plasma levels of estradiol, which are known to fall significantly at menopause, and ceramides. Estradiol was measured with a competitive binding immunoassay. As expected, plasma estradiol was higher in pre-menopausal women (<55 years) compared to men in the same age group (p < .001, 2-way ANOVA followed by Bonferroni post-hoc test) (Fig 3B). After menopause, estradiol levels sharply decreased in both sexes (Fig 3B). Fig 4 illustrates the results of Pearson’s analyses of ceramide levels in female subjects of all ages. The results show a statistically significant negative correlation between estradiol and ceramide (d18:1/24:1) (r = -0.294; p = .007), a non-significant negative trend between estradiol and ceramide (d18:1/24:0) (r = -0.202; p = .066) and no correlations between estradiol and other ceramide species. By contrast, in men, no correlation was observed between estradiol and any ceramide species, including ceramide (d18:1/24:1) (r = -0.034; p = .763) (Fig 5), which was found to be correlated with aging (β (SE) = 1.86 (.77); p = .0179) (Table 2B).

Figure 4. Pearson’s correlation analysis between estradiol and levels of various ceramide species in plasma from 84 women aged 20 to 78 years. (A) Ceramide (d18:1/24:1); (B) Ceramide (d18:1/24:0); (C) Ceramide (d18:1/16:0); (D) Ceramide (d18:1/18:0). Correlation is considered statistically significant at p < .05.

Figure 5. Pearson’s correlation analysis between estradiol and levels of various ceramide species in plasma from 80 men aged 19 to 80 years. (A) Ceramide (d18:1/24:1); (B) Ceramide (d18:1/24:0); (C) Ceramide (d18:1/16:0); (D) Ceramide (d18:1/18:0). Correlation is considered statistically significant at p < .05.

Estradiol suppresses ceramide accumulation in vitro

Sphingolipid-derived mediators such as ceramide, sphingosine, and sphingosine-1-phosphate regulate steroidogenesis [21], but it is still unknown whether estrogen hormones influence sphingolipid metabolism. To gain insight into the causality of the negative correlation observed between estradiol and ceramide in women, we asked whether the sex hormone might regulate ceramide mobilization (i.e. formation and/or degradation) in human MCF7 breast cancer cells, a cell line that expresses the estrogen receptor α (ERα) and β (ERβ) [22]. MCF7 cells were treated with estradiol (10 nM) for 24 h and ceramides quantified in lipid extracts by LC-MS/MS. Results indicate that the exposure to estradiol causes a substantial reduction in ceramide (d18:1/16:0) (p = .02, unpaired Student’s t-test), (d18:1/24:0) (p = .0002) and (d18:1/24:1) (p = .0006) (Table 3), thereby suggesting that estradiol causes a downregulation in the biosynthesis or/and an upregulation in the degradation of these substances.

Ethics statement

Investigation has been conducted in accordance with the ethical standards and according to the Declaration of Helsinki and according to national and international guidelines and has been approved by the authors' institutional review board.

Study subjects

We recruited 164 Italian subjects (84 women) from 19 to 80 years of age (Table 1). A primary criterion for subject selection was the absence of major medical illnesses and, particularly, conditions that had been previously linked to ceramide alterations. A subdivision of the subjects by age and menopause status is reported in Table S1. There were no significant differences between male and female subjects with regard to age, years of education, ethnic background, cognitive status, as assessed by the Mini-Mental State Examination (MMSE), obesity and diabetes, hypertension and use of anti-hypercholesterol drugs. By contrast, there was difference in smoking status (20.23% women versus 7.5% men). Moreover, 6 of 44 pre-menopause women used contraceptives at time of enrollment. None of the post-menopause women was under hormone replacement therapy (HRT).

Exclusion criteria were: (i) suspicion of cognitive impairment or dementia based on MMSE [46] (score ≤ 26, consistent with normative data collected in the Italian population) and confirmed by a detailed neuropsychological evaluation using the mental deterioration battery [47] and clinical criteria for Alzheimer’s dementia [48] or mild cognitive impairment [49]; (ii) subjective complaints of memory difficulties or other cognitive deficits, regardless of whether or not these interfered with daily life, as assessed by clinical evaluation; (iii) vision and hearing loss that could potentially influence testing results, as assessed by clinical evaluation; (iv) major medical illnesses (i.e., unstable diabetes; obesity; obstructive pulmonary disease or asthma; hematological and oncological disorders, as assessed by clinical evaluation and pertinent clinical testing; pernicious anemia; significant gastrointestinal, renal, hepatic, endocrine, or cardiovascular system diseases, as assessed by clinical evaluation and pertinent clinical testing; recently treated hypothyroidism, as assessed by clinical evaluation; (v) current or reported psychiatric disease, as assessed by the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision (DSM-IV-TR SCID) [50] or neurological disease, as assessed by clinical evaluation; and (vi) known or suspected history of alcoholism or drug addiction. Finally, because ceramides have been previously involved in the pathogenesis of neurodegenerative disorders [15,51,52], we excluded subjects who showed brain abnormalities or vascular lesions as determined by using a recently published semi-automated method [53]. The menopausal status was prospectively assessed during clinical interviews. Women were defined as post-menopausal when showing 12 consecutive months of amenorrhea that was not due to other obvious pathological or physiological causes.

Blood collection and analyses were approved by the Santa Lucia Foundation Ethics Committee. A written consent form was signed by all participants after they received a full explanation of the study procedures. Informed consent has been obtained.

Variables examined in relation to ceramide levels

All variables were assessed for each patient using the same methods. Demographic variables considered in linear regression analysis included age and sex. Medical history covariates included hypertension, use of anti-hypercholesterol agents and use of contraceptives. Current and former smoking status was ascertained by an oral interview.

Chemicals

Solvents and chemicals were purchased from Sigma Aldrich (Milan, Italy). Ceramide standards were from Avanti Polar Lipids (Alabaster, AL, USA).

Blood collection

Blood was drawn by venipuncture in the morning after an overnight fast, and collected into 10-ml tubes containing spray-coated EDTA (EDTA Vacutainer, BD Biosciences, San Diego, CA, USA). Plasma was obtained by blood centrifugation at 400 × g at 4 °C for 15 min. The plasma divided into aliquots was stored at −80 °C until analyses.

Lipid extraction

Lipids were extracted using a modified Bligh and Dyer method [54]. Briefly, plasma samples (50 µL) or cell pellets were transferred to glass vials and liquid-liquid extraction was carried out using 2 mL of a methanol/chloroform mixture (2:1 vol/vol) containing the odd-chain saturated ceramide (d18:1/17:0) as an internal standard. After mixing for 30 s, lipids were extracted with chloroform (0.5 mL), and extracts were washed with liquid chromatography-grade water (0.5 mL), mixing after each addition. The samples were centrifuged for 15 min at 3500 x g at room temperature. After centrifugation, the organic phases were collected and transferred to a new set of glass vials. To increase overall recovery, the aqueous fractions were extracted again with chloroform (1 mL). The two organic phases were pooled, dried under a stream of N₂ and residues were dissolved in methanol/chloroform (9:1 vol/vol, 0.07 mL). After mixing (30 s) and centrifugation (10 min, 5000 x g, room temperature) the samples were transferred to glass vials for analyses.

Ceramide quantification

Ceramides were analyzed by LC-MS/MS using an Acquity® ultra-performance liquid chromatography (UPLC) system coupled to a Xevo triple quadrupole mass spectrometer interfaced with electrospray ionization (Waters, Milford, MA, USA), as previously described [54]. Lipids were separated on a Waters Acquity® BEH C18 column (2.1 × 50 mm, 1.7 μm particle size) at 60 °C and eluted at a flow rate of 0.4 mL/min. The mobile phase consisted of 0.1% formic acid in acetonitrile/water (20:80 vol/vol) as solvent A and 0.1% formic acid in acetonitrile/isopropanol (20:80 vol/vol) as solvent B. A gradient program was used: 0.0–1.0 min 30% B, 1.0–2.5 min 30 to 70% B, 2.5–4.0 min 70 to 80% B, 4.0–5.0 min 80% B, 5.0–6.5 min 80 to 90% B, and 6.6–7.5 min 100% B. The column was reconditioned to 30% B for 1.4 min. The injection volume was 3 µL. Detection was done in the positive electrospray ionization mode. Capillary voltage was 3.5 kV and cone voltage was 25 V. The source and desolvation temperatures were set at 120 °C and 600 °C respectively. Desolvation gas and cone gas (N₂) flow were 800 L/h and 20 L/h, respectively. Plasma and cell-derived ceramides were identified by comparison of their LC retention times and MS/MS fragmentation patterns with those of authentic standards (Avanti Polar Lipids). Extracted ion chromatograms were used to identify and quantify the following ceramides and dihydroceramides (d18:1/16:0) (m/z 520.3 > 264.3), (d18:1/18:0) (m/z = 548.3 > 264.3), (d18:1/24:0) (m/z = 632.3 > 264.3), (d18:1/24:1) (m/z = 630.3 > 264.3), (d18:0/24:0) (m/z = 652.5 > 634.5) and (d18:0/24:1) (m/z = 650.5 > 632.5). Data were acquired by the MassLynx software and quantified using the TargetLynx software (Waters, Milford, MA, USA).

Estradiol quantification

Plasma 17-β-estradiol (E2) levels were quantified using a competitive binding immunoassay kit (Human E2 ELISA kit, Invitrogen, Milan, Italy) following manufacturer’s instructions. Briefly, plasma samples, controls and standard curve samples (50 µL) were incubated with E2-horseradish peroxidase conjugate (50 µL) and anti-estradiol antibody (50 µL) in a 96-well plate for 2 h, at room temperature, on a shaker set at 700 ± 100 rpm. Washing was carried out by completely aspirating the liquid, filling the wells with diluted wash buffer (0.4 mL) provided in the kit and then aspirating again. After repeating this procedure 4 times, chromogen solution (200 µL) was added to each well; reactions were run for 15 min and stopped adding 50 µL of the stop solution provided in the kit. Absorbance was measured at 450 nm and estradiol concentrations were calculated by interpolation from the reference curve.

Cell cultures and treatment

The MCF7 human breast cancer cell line [22,55] was a kind gift of Dr. Gennaro Colella (Mario Negri Institute, Milan, Italy). Cells were cultured in phenol red-free Dulbecco’s Modified Eagle’s Medium (DMEM) (Gibco by Life Technologies, Carlsbad, CA, USA) supplemented with charcoal-stripped fetal bovine serum (10%) (Sigma Aldrich, Milan, Italy) to starve cells from steroid hormones (starvation medium), L-glutamine (2 mM), and penicillin/streptomycin (100 µg/mL), in a humidified atmosphere (5% CO₂, 37 °C). Cells were seeded in 6-well plates (3 x 10⁵ cells/well) and cultured for 24 h. Estradiol (Sigma Aldrich, Milan, Italy) was dissolved in dimethyl sulfoxide (DMSO) and diluted in phenol red-free DMEM to a final concentration of 10 nM (0.1% final DMSO concentration). After 24 h incubation, the media were removed, cells were washed with phosphate-buffered saline, scraped and centrifuged (800 x g, 4 °C, 10 min). Protein concentrations were measured using the bicinchoninic acid assay (Pierce, Rockford, IL, USA) and cell pellets were stored at -80 °C until analyses.

Statistical analyses

Results are expressed as mean ± SEM (standard error of the mean). Sex differences in baseline demographic and health-related characteristics were examined using Fisher’s exact test for categorical variables and unpaired Student's t-test for continuous variables. Data were analyzed by unpaired Student's t-test or 2-way ANOVA followed by Bonferroni post-hoc test. Pearson’s correlation coefficient assessed the pairwise correlation between estradiol and ceramide levels. Significant outliers were excluded using the Grubbs’ test. Multivariable linear regression method was used to assess the association between ceramides and patients’ characteristics. Differences between groups were considered statistically significant at values of p < .05. The GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA, USA) and SAS 9.4 (SAS Institute, Cary, NC, USA) were used for statistical analyses.