Ceramide Biomarkers Predictive of Cardiovascular Disease Risk Increase in Healthy Older Adults After Bed Rest

Abstract

Acute bed rest places older adults at risk for health complications by disrupting homeostasis in many organ systems, including the cardiovascular system. Circulating ceramides are emerging biomarkers predictive of cardiovascular and metabolic health and have recently been shown to be sensitive indices of cardiovascular (CV) risk. Therefore, the purpose of this study was to characterize the time course of changes in circulating ceramides in healthy younger and older adults after 5 days of bed rest and to determine whether short-term bed rest alters CV-related circulating ceramides. We hypothesized that circulating ceramides predictive of poor cardiometabolic outcomes would increase following 5 days of bed rest. Thirty-five healthy younger and older men and women (young: n = 13, old: n = 22) underwent 5 days of controlled bed rest. Fasting blood samples collected daily during the course of bed rest were used to measure circulating ceramides, lipoproteins, adiponectin, and fibroblast growth factor 21 (FGF21) levels. The primary findings were that circulating ceramides decreased while ceramide ratios and the cardiac event risk test 1 score were increased primarily in older adults, and these findings were independent of changes in circulating lipoprotein levels. Additionally, we found that changes in circulating adiponectin, FGF21 and the 6-minute walk test (6MW) inversely correlated with CV-related circulating ceramides after bed rest. The results of this study highlight the sensitivity of circulating ceramides to detect potential CV dysfunction that may occur with acute physical disuse in aging.

Sphingolipids, Disuse, Blood lipids, Inactivity, CV risk factors

Physical disuse has whole body ramifications, taxing organs responsible for glucose homeostasis (liver, adipose, and muscle), and conferring damaging effects to the CV system (1). With aging, the risk for CV complications are elevated and further accelerated in older adults who are inactive (2). Whole body disuse in the form of bed rest causes CV dysfunction, and accelerates aging and markers of frailty (3). Interestingly, circulating lipoprotein markers, commonly used to diagnose cardiometabolic risk, are not altered with 16–35 days of bed rest (4,5), and are poor biomarkers for early detection of cardiac dysfunction (6,7). Thus, novel, sensitive diagnostic tools that can detect early deterioration in heart health are needed.

Ceramides are a subset of bioactive sphingolipids that have been linked to cardiometabolic diseases and have similar physiologic functions to cholesterol (8). Circulating ceramides are well known to be associated with insulin resistance, type 2 diabetes mellitus, prediabetes, and obesity (9–11). Moreover, circulating ceramides are linked with an increased risk of CV-related death (6,12). Even though it is known that chronic physical inactivity induces CV dysfunction, the relationship between circulating ceramides and CV health as a result of short-term period of physical disuse, typical to that of the duration of hospitalization (eg, 5–7 days) (13,14) is limited.

Common circulating ceramide species linked to poor metabolic and CV outcomes are elevated levels of ceramide C16:0 (ceramide with palmitic acid), C18:0 (ceramide with stearic acid), C20:0 (ceramide with arachidic acid), and C24:1 (ceramide with lignoceric acid:1 double bond) (8). We and others have observed that inhibiting serine palmitoyltransferase (SPT, the rate-limiting enzyme in ceramide synthesis) pharmacologically with myriocin or genetically via haploinsufficiency for the SPTLC2 subunit is insulin sensitizing (15). Depleting ceramides through knocking out downstream ceramide synthesis enzymes (CerS1, Degs1) also rescues insulin resistance (16,17), steatohepatitis (18), and depleting ceramide levels in the heart reduces cardiomyocyte apoptosis in rodent models (19). Ceramides are involved in cardiovascular disease (CVD) development through atherosclerosis, impaired vasodilation, and cardiomyocyte apoptosis (20–22). Importantly, calculating ceramide ratios of C16:0, C18:0, or C24:1 to the more abundant circulating ceramide species, C24:0, are sensitive biomarkers with greater prediction of CV-related death than low-density lipoprotein cholesterol (LDL) (23). These observations have recently led to using circulating ceramide species and ratios to develop the cardiac event risk test 1 (CERT1) score that predicts CV complications and death, whereas lipoproteins did not (6). Neither ceramide ratios nor CERT1 scores have been determined prospectively through short-term physical disuse (as modeled by bed rest) in younger and older adults and in those with no prior known CVD or metabolic dysfunction.

Therefore, the purposes of the present study were to characterize the circulating ceramide response more than 5 days of bed rest in both healthy younger and older adults and, secondly, to determine whether 5 days of bed rest alters CV-related circulating ceramides. We hypothesized that ceramide biomarkers predictive of poor cardiometabolic health would increase in both young and older adults following 5 days of bed rest.

Methods

Subject Characteristics

Thirty-five healthy men and women (young: N = 13, old: N = 22) were included in this study. Detailed subject characteristics are described in Table 1. Most of the samples were obtained from two previously published 5-day bed rest studies (24,25). Briefly, all subjects were recruited within the Salt Lake City area and screened for pre-existing medical conditions. All subjects read and signed the informed consent and the study was reviewed and approved by the University of Utah Institutional Review Board and conformed to the Declaration of Helsinki and Title 45, U.S. code of Federal Regulations, Part 46, “Protection of Human Subjects.”

Table 1.

Subject Characteristics.

	Young		Old		2 × 2 ANOVA		(p)
	Pre	Post	Pre	Post	Age	Time	Interaction
N (M/F)	13 (7/6)	13 (7/6)	22 (11/9)	22 (11/9)
Age (years)	23.4 ± 3.2	23.4 ± 3.2	67.8 ± 5.5	67.8 ± 5.5	<0.01	>0.99	>0.99
Body composition (kg)
Weight	67.2 ± 13.6	65.8 ± 13.3	74.6 ± 10.3	73.3 ± 10.5	0.55	0.42	0.89
Whole lean mass	46.2 ± 10.0	46.0 ± 9.6	48.0 ± 8.5	46.4 ± 8.4	0.74	<0.01	<0.01
Leg lean mass	15.6 ± 3.5	15.6 ± 3.4	15.1 ± 2.9	14.4 ± 2.8	0.52	<0.01	<0.01
Fat mass	17.3 ± 5.5	17.5 ± 5.6	24.1 ± 6.3	24.3 ± 6.3	<0.01	0.12	0.99
Percent body fat	0.26 ± 0.1	0.26 ± 0.1	0.32 ± 0.1	0.33 ± 0.1	0.01	<0.01	0.89
BMI (kg/m2)	22.1 ± 3.4	21.7 ± 3.2	24.9 ± 2.7	24.5 ± 2.9	0.96	0.17	0.01
Lipoproteins and cholesterol (mg/dL)
HDL	31.0 ± 10.3	32.7 ± 12.2	34.3 ± 12.7	31.7 ± 10.6	0.66	0.66	0.10
LDL	86.7 ± 13.3	91.9 ± 5.5	90.8 ± 8.3	91.4 ± 5.9	0.61	0.06	0.17
LDL/HDL	3.0 ± 0.7	3.6 ± 3.0	3.1 ± 1.5	3.3 ± 1.5	0.79	0.19	0.55
T-cholesterol	117.7 ± 20.2	124.6 ± 14.9	125.2 ± 18.1	123.0 ± 13.5	0.59	0.32	0.07
Triglycerides	91.6 ± 44.8	119.7 ± 84.8	93.4 ± 21.2	109.8 ± 71.0	0.87	0.16	0.69
Circulating hormones
Adiponectin (µg/mL)	11.9 ± 5.9	10.5 ± 5.0	17.3 ± 8.4	15.3 ± 6.8	0.04	<0.01	0.59
FGF21 (pg/mL)	102.4 ± 65.4	46.7 ± 31.6	222.7 ± 138.5	143.6 ± 107.1	<0.01	<0.01	0.52
Insulin sensitivity
HOMA-IR	1.2 ± 0.3	1.4 ± 0.3	1.3 ± 0.8	1.4 ± 0.9	0.71	0.10	0.43
CLIX-IR	8.4 ± 3.4	5.5 ± 2.3	5.1 ± 2.1	3.5 ± 1.9	<0.01	<0.01	0.05
Matsuda Index	10.5 ± 2.6	7.2 ± 1.5	9.3 ± 5.4	7.7 ± 5.6	0.84	<0.01	0.21
Fasting insulin (uU/mL)	5.3 ± 1.0	6.3 ± 1.2	5.8 ± 3.3	5.9 ± 3.8	0.98	0.14	0.22
Fasting glucose (mg/dL)	88.2 ± 7.8	90.7 ± 4.1	94.9 ± 6.5	96.8 ± 11.6	0.02	0.09	0.84

Note: ANOVA = analysis of variance; BMI = body mass index; HDL = high-density lipoprotein; LDL = low-density lipoprotein; FGF21 = fibroblast growth factor 21; HOMA-IR = homeostatic model assessment-insulin resistance; CLIX-IR = Clamp-Like Index-Insulin resistance. Data are mean ± SD. Bold p values indicate statistical significance (p < .05).

Study Design

Five days of bed rest (Monday–Friday) took place at the University of Utah Center for Clinical and Translational Science as described in our prior studies (24,25). Briefly, all procedures were performed after an overnight fast. On the first and final day of best rest dual-energy x-ray absorptiometry (DEXA) was performed and an oral glucose tolerance test was conducted on the fourth day of bed rest to determine whole and leg lean mass and fat mass and glucose tolerance, respectively. Additionally, a 6MW was conducted before and at the end of bed rest to determine the relationship of lipid species to physical performance. 6MW is commonly used in clinical practice and correlates well to aerobic capacity (VO₂) (26).On each morning of bed rest a fasting blood sample was drawn and serum and plasma samples were collected and used for analysis.

Dietary and Physical Activity Control Prior to Bed Rest

Prior to bed rest, individuals were given instruction to record food intake more than 5 days (including one weekend day). Caloric intake was calculated using the Food Processor Nutrition Analysis software (Salem). To measure physical activity, individuals were provided with a step activity monitor (Orthocare Innovations, Mountain Lake Terrace).

Dietary Control During Bed Rest

The night prior to entering the Clinical Research Unit, subjects were given a meal consisting of a caloric and macronutrient density based on the individuals’ body weight. During bed rest, total caloric intake was predetermined by the research dietician using the Harris–Benedict equation adjustment for no physical activity (15% protein, 55% carbohydrate, and 30% fat). Individuals were also provided water ad libitum throughout the 5-day bed rest period.

Insulin Sensitivity and Cardiovascular Risk Calculations

Homeostatic model assessment-insulin resistance (HOMA-IR), Clamp-Like Index-Insulin Resistance (CLIX-IR), and Matsuda Index were used to determine insulin sensitivity, and the CERT1 was used to determine cardiovascular (CV) risk. The following calculations were used to determine each, HOMA-IR: (fasting plasma insulin × fasting plasma glucose)/22.5 (27), CLIX-IR: serum creatinine (×0.85 if male)/(mean AUC_glucose × mean AUC_C-peptide) × 6,600 (AUC = area under the curve) (28), Matsuda Index: (10,000/square root of [fasting glucose × fasting insulin] × [mean glucose × mean insulin during oral glucose tolerance test]) (29). CERT1: ceramide C16:0, C18:0, C24:1, and the ratios C16:0/C24:0, C18:0/C24:0, and C24:1/C24:0 were compared to all study subjects. If the variable belonged to the third quartile, the individual received +1 point, and if to the fourth quartile, +2 points. The score ranges from 0 to 12, with 12 being the highest risk (6).

Circulating HDL, LDL, Total Cholesterol, Adiponectin, and Fibroblast Growth Factor 21

High-density lipoprotein (HDL), LDL, and total cholesterol were determined with colorimetric assays (Abcam, #ab65390) for pre- and post-bed rest plasma and serum samples. The LDL fraction also contained very low-density lipoprotein (VLDL), thus the LDL fraction described herein represents a composite of both lipoproteins. Circulating adiponectin and fibroblast growth factor 21 (FGF21) were determined utilizing an enzyme-linked immunosorbent assay (ELISA) following manufacturer protocols. Inter- and intra-assay coefficient of variation for adiponectin was 5.6% and 5.7%, respectively. Inter- and intra-assay coefficient of variation for FGF21 was 2.8% and 3.9%, respectively. (EMD Millipore, adiponectin: EZHADP-61K, FGF21: EZRMFGF21-26K).

Lipid Isolation

Lipids were extracted from 20 μL of serum mixed in 200 μL of ice-cold PBS. Each sample was spiked with internal standards to quantify levels of individual lipids. These standards were obtained from Avanti Polar Lipids (Alabaster, AL) and included C17 ceramide (d18:1/17:0) N-heptadecanoyl-D-erythro sphingosine (#860517), C17 sphingomyelin (d18:1/17:0) N-heptadecanoyl-D erythrosphingosylphosphorylcholine (#860585), C18:1 dihydroceramide (d18:0/18:1(9Z)) (#860624), and lactosyl (ß) C24 ceramide N-(tetracosanoyl)-1-ß-lactosyl-sphing-4-ene (#110762). After 15 minutes of incubation in chilled methanol with a brief vortex every 5 minutes, samples were pelleted at 15,000g for 5 minutes at 4°C. The supernatant was removed and mixed with 30 μL of 1 M KOH (in methanol). Following overnight incubation at 50°C in a sand bath, the samples were concentrated in a speedvac at room temperature to 200–300 μL of the remaining sample. Next, 25 μL of glacial acetic acid, 300 μL of ddH₂O and 500 μL of liquid chromatography (LC) grade methyl tertiary-butyl ether were added, vortexed and pelleted at 15,000g for 2 minutes at 4°C. The last step was repeated without the addition of ddH₂O and then the samples were completely dehydrated in a speedvac at RT for 4 hours and stored at −20°C until LC mass spectrometry (MS) analysis.

Liquid Chromatography with Tandem Mass Spectrometry

Before analysis, lipid extracts were resuspended in 200 μL of HPLC grade methanol, pelleted and the clear supernatant was transferred to glass vials. Lipid extracts were separated on an Acquity UPLC CSH C18 1.7 μm 2.1 mm × 50 mm (Waters, Milford, MA) column maintained at 60°C connected to an Agilent HiP 1290 Sampler and Agilent 1290 Infinity pump, equipped with an Agilent1290 Flex Cube and an Agilent 6490 triple quadrupole mass spectrometer (Agilent Technologies, Inc., Santa Clara, CA). Sphingolipids were detected using dynamic multiple reaction monitoring (MRM) in positive ion mode. Source gas temperature was set to 210°C, with a gas (N₂) flow of 11 L/min and a nebulizer pressure of 30 psi. Sheath gas temperature was 400°C, sheath gas (N₂) flow was 12 L/min, capillary voltage was 4,000 V, nozzle voltage as 500 V, high pressure RF was 190 V and low-pressure RF was 120 V. The injection volume was 2 μL and the samples were analyzed in a randomized order with the pooled quality control sample injection eight times throughout the sample queue. Mobile phase A consists of ACN:H₂O (60:40 v/v) in 10 mM ammonium formate and 0.1% formic acid, and mobile phase B consists of IPA:ACN:H₂O (90:9:1 v/v) in 10 mM ammonium formate and 0.1% formic acid. The chromatography gradient started at 15% mobile phase B, increased to 30% B more than 1 minute, increased to 60% B from 1 to 2 minutes, increased to 80% B from 2 to 10 minutes, and increased to 99% B from 10 to 10.2 minutes, where it was held until 14 minutes. Post-time was 5 minutes and the flow rate was 0.35 mL/min throughout. Collision energies and cell accelerator voltages were optimized using sphingolipid standards with dynamic MRM transitions as [M+H]⁺ → [m/z = 284.3] for dihydroceramides, [M-H₂O+H]⁺ → [m/z = 264.2] for ceramides and [M+H]⁺ → [m/z = 184.4] for all sphingomyelins. Sphingolipids without available standards were identified based on high-resolution LC-MS, quasi-molecular ion and characteristic product ions. Their retention times were either taken from high-resolution LC-MS data or inferred from the available sphingolipid standards. The results from LC-MS experiments were collected using an Agilent Mass Hunter Work station and analyzed using the software package Agilent Mass Hunter Quant B.07.00. Sphingolipids were quantitated based on peak area ratios to the standards added to the samples at extraction. Serum subfraction lipid samples were processed as one batch each.

Statistical Analysis

Subject characteristics were analyzed with a two-way ANOVA, determining mean differences due to age, bed rest (time), and the interaction of age by time (GraphPad Prism 8). Independent sample t-tests were used to determine differences at baseline between younger and older adults. Mean changes across time are displayed with profile plots, showing the change from baseline for each Day 1–5, with a line connecting the means separately for young and old subjects. To model these data, we fit a mixed effects linear regression model, which accounted for the within subject repeated measures across time, a between subjects group main effect (young vs old), and group by time interactions using indicator variables for time so that the model passed through all of the means exactly. Postfit Wald tests were used to compare specific means between groups or compare within subjects change between different days while using variance estimates from the full model (Stata version 15.1). Pearson’s correlations were utilized to determine relationships between ceramides and physiological and physical function measurements.

Results

Bed Rest Effects on Body Composition, Circulating Factors, and Insulin Sensitivity

Older adults had higher fat mass, percent body fat, adiponectin, FGF21, fasting glucose, and lower insulin sensitivity (p < .05, Table 1) than younger adults. After bed rest, younger and older adults both increased body fat percentage (2%), decreased circulating levels of adiponectin (−9%) and FGF21 (−36%), and also decreased insulin sensitivity (−33%) (p < .05). Only the older adults lost body weight (−2%) and whole body and leg lean mass (−3%) (p < .05) after bed rest. There were no changes in lipoproteins (LDL, HDL, LDL/HDL, triglycerides) or cholesterol after bed rest.

Circulating Ceramides and Changes With Bed Rest in Younger and Older Adults

At baseline (pre), older adults had higher circulating ceramide C14:0, C16:0, C23:0, C24:0, C24:1, C26:0, C26:1, and total ceramides (p < .05) than younger adults (Table 2). Daily changes in circulating ceramides implicated in cardiometabolic risk detection (C16:0, C18:0, C20:0, C24:0, and C24:1) are displayed over the 5-day time course in (Figure 1). Ceramide C16:0 increased (14%) in younger adults only (p = .02) by Day 3 of bed rest, but returned to baseline levels on Day 4 (Figure 1A). Ceramide C18:0 increased in older adults by Day 3 (17%; p = .02) and Day 4 (25%, p = .002) compared with baseline, and was higher than younger adults (p = .02) on Day 4. However, ceramide C18:0 returned to baseline levels by Day 5 of bed rest (Figure 1B). Ceramide C20:0 increased by Day 4 of bed rest in older adults only (14%; p = .03), and returned to baseline by Day 5 (Figure 1C). Ceramide C24:0 decreased (−22%) at Day 5 compared with baseline (p < .001) in older adults and was lower than younger adults (p < .001). There was also an age-related difference in C24:0 at Day 4 (Figure 1D). Younger adults increased ceramide C24:1 (19%) at Day 4 compared with baseline (p < .001), and was different than older adults (p = .003), but this response returned back to baseline on Day 5 of bed rest (Figure 1E). Total ceramides decreased (−14%) in older adults only (p = .005) by Day 5 of bed rest (Figure 1F).

Table 2.

Means and Interquartile Ranges of Younger and Older Adults Circulating Ceramide Levels at Baseline (Pre-Bed Rest)

	Young	Old	p Value
Lipid (pmol/mL)
Cer d18:1, 14:0	0.8 (0.7–1.0)	1.1 (0.8–1.1)	0.01
Cer d18:1, 16:0	19.5 (16.3–23.8)	23.5 (18.4–28.0)	0.05
Cer d18:1, 16:1	2.6 (2.0–3.0)	2.9 (2.8–3.2)	0.22
Cer d18:1, 18:0	63.0 (46.2–59.0)	65.1 (44.0–75.9)	0.85
Cer d18:1, 20:0	52.8 (40.6–59.4)	61.7 (44.5–68.8)	0.24
Cer d18:1, 22:0	350.0 (244.1–402.7)	428.1 (306.0–500.6)	0.17
Cer d18:1, 22:1	10.1 (7.8–12.1)	11.8 (8.6–13.0)	0.26
Cer d18:1, 23:0	312.7 (248.1–385.0)	459.1 (342.2–589.6)	0.02
Cer d18:1, 24:0	1,106.8 (843.9–1,332.2)	1,556.6 (1,189.9–1,778.1)	0.02
Cer d18:1, 24:1	317.2 (251.3–377.3)	462.3 (353.1–538.7)	0.01
Cer d18:1, 26:0	15.9 (14.8–17.8)	22.0 (16.1–24.2)	0.01
Cer d18:1, 26:1	5.1 (4.1–5.7)	7.6 (5.9–8.8)	<0.01
Total ceramides	2,260.0 (1,709.5–2,638.1)	3,106.4 (2,449.9–3,568.7)	0.03

Notes: Data are mean (IQR). Bold p values indicate p < .05.

Figure 1.

Time course of percent change from pre-bed rest of circulating ceramide C16:0 (A), C18:0 (B), C20:0 (C), C24:0 (D), and C24:1 (E) throughout 5 days of bed rest in younger and older adults. #p < .05 versus pre-bed rest in young only. *p < .05 versus pre-bed rest in old only. †p < .05 indicates different between young versus old on respective day.

Circulating CV-Related Ceramide Markers After Bed Rest

After the first day of bed rest, C16:0/C24:0 (Figure 2A), C18:0/C24:0 (Figure 2B), and C20:0/C24:0 (Figure 2C) increased in older adults (p < .05), an effect that persisted throughout bed rest. Younger adults increased C16:0/C24:0 (Figure 2A) on Day 2 of bed rest and increased C24:1/C24:0 (Figure 2D) on Day 4 of bed rest (p < .05), yet these effects were transient as both returned to baseline levels on the following day. C24:1/C24:0 was elevated in older adults only (p < .05) on Days 3 and 5 of bed rest (Figure 2D), with older adults being higher than younger adults on Day 5 (p = .02). Older adults had higher C16:0/C24:0 than younger adults on Days 4 and 5, and higher C18:0/C24:0 and C20:0/C24:0 on Days 2, 3 and 4 of bed rest (p < .05, Figure 2A–C). The CERT1 score was increased in older adults on Days 3, 4, and 5 of bed rest with the score being higher in older versus younger adults on Day 4 (p = .04) (Figure 2E).

Figure 2.

Time course of percent change from pre-bed rest of circulating ceramide ratios C16:0/C24:0 (A), C18:0/C24:0 (B), C20:0/C24:0 (C), and C24:1/C24:0 (D) throughout 5 days of bed rest in younger and older adults. (E) CERT1 (cardiac event risk rest 1) scores in younger and older adults throughout 5 days of bed rest. #p < .05 vs pre-bed rest in young only. *p < .05 vs pre-bed rest in old only. †p < .05 indicates different between young vs old on respective day.

Correlations of Ceramides to Physiological and Physical Function Parameters

Percent change of C16:0/C24:0 and C20:0/C24:0 inversely correlated to changes in adiponectin after bed rest in older adults (C16:0/24:0 vs adiponectin, r = −.400, p < .05; C20:0/24:0 vs adiponectin, r = −.403, p < .05; Figure 3A and B). Additionally, percent change in C16:0/C24:0 and C18:0/C24:0 inversely correlated with the change in FGF21 after bed rest in older adults (C16:0/24:0 vs FGF21, r = −.475, p < .05; C18:0/24:0 vs FGF21, r = −.483, p < .05; Figure 3C and D). Ceramide species C18:0, C20:0, C24:1, and CERT1 score negatively correlated to insulin sensitivity at baseline but not after bed rest (Supplementary Table S1). Changes in HDL, LDL, or LDL/HDL ratio were not related to CERT1 score (Supplementary Table S2). 6MW did not change after bed rest in younger (pre: 649.8±17.5m − post: 634.4 ± 26.6m, p = .394) or older (pre: 650.3 ± 22.7 m − post: 644.2 ± 22.0 m, p = .547) adults. However, changes in C18:0/C24:0 (but not the other ratios or CERT1) inversely correlated to changes in 6MW distance in older adults (C18:0/C24:0 vs 6MW, r = −.552, p < .05; Figure 3E). Percent change in adiponectin correlated to the change in FGF21 after bed rest in all subjects (Figure 3F). Change in CERT1 score, ceramide ratios, adiponectin, FGF21, lipoproteins, or total cholesterol did not correlate with changes in insulin sensitivity.

Figure 3.

Relationship between changes in adiponectin (A and B), FGF21 (C and D), and 6MW (E) to changes in ceramide ratios in older adults. Changes in adiponectin to changes in FGF21 in younger and older adults (F) FGF21 = fibroblast growth factor-21; 6MW = 6-minute walk test.

Discussion

The primary finding in this study was that in as few as 5 days of bed rest, circulating ceramides ratios and CERT1 score, indicative of poor cardiometabolic health, were elevated in older adults who had no prior history of heart or metabolic disease. Interestingly, the bed rest-induced change in ceramide ratios inversely correlated with changes in circulating adiponectin, FGF21 and the 6MW in older adults bridging a relationship to a potential mechanism and links to CV performance. Secondly, bed rest reduced total circulating ceramides in older adults with ceramide C24:0 lower by 22%. Together, these data suggest that circulating ceramides that have previously been used to predict CV risk, increased in older adults within just a few days of bed rest, whereas traditional biomarkers LDL or HDL were not altered during this short duration.

While it is established that acute bed rest deteriorates CV function [eg, reduced VO_2max, plasma volume (30,31)], the common cholesterol markers of CV risk (total cholesterol, LDL, HDL) display mixed results in bed rest trials. For instance, 8 days of bed rest in healthy young men increased LDL and decreased HDL from baseline (32) whereas 10 days of bed rest or greater does not alter LDL or total cholesterol, and only lowers HDL (4,33). Therefore, cholesterol markers may not be sensitive enough to detect CV dysfunction and risk during acute periods of physical disuse as modeled with bed rest. The CERT1 score was developed from a case–control clinical trial to address the need to predict coronary events and optimize secondary prevention strategies (6). This score, a composite of circulating levels of ceramide C16:0, C18:0, C24:1, and the ratios C16:0/C24:0, C18:0/C24:0, and C24:1/C24:0, was indeed predictive of CV death in coronary artery disease cases versus controls (6). Using this sensitive biomarker index, we observed an increase in CERT1 score and ceramide ratios (C16:0/C24:0, C18:0/C24:0, and C24:1/C24:0) primarily in older but not younger adults that occurred early as Day 2 of bed rest and persisted during the remaining 5 days of bed rest. This suggests that short-term periods of whole-body disuse in healthy older adults initiates early changes in CV health, independent of changes in lipoprotein CV risk markers (total cholesterol, LDL, and HDL). Therefore, this study provides additional evidence that the CERT1 score and ceramide ratios may be a more sensitive diagnostic tool to detect early changes in the CV system. This statement is further evidenced by the relationship of C18:0/C24:0 to physical function (6MW), as physical function and C18:0/C24:0 are both predictors of mortality in individuals with CVD (6,34).

In the current study, we observed a decrease in circulating adiponectin levels following 5 days of bed rest in both young and older adults, and these levels were inversely correlated to C16:0/C24:0 and C20:0/C24:0. Adiponectin is a highly abundant endocrine hormone released by adipocytes and can promote vascular homeostasis improving atherosclerosis and thus decreasing CV complications (35). In cases of metabolic dysfunction, such as obesity and insulin resistance, adiponectin levels are decreased (36). Alternately, an increase in adiponectin coincides with improved insulin sensitivity and weight loss in obese humans at risk for CVD (36,37). Interestingly, adiponectin protects cardiomyocytes from apoptosis by reducing ceramides (38), while also decreasing inflammation in endothelial cells and monocytes, alleviating atherosclerosis development (35). Thus, these reports provide further evidence of the relationship between adiponectin and ceramide ratios that was observed after bed rest in our data set. Together, our results support that short-term bed rest lowers adiponectin levels and are associated with changes in circulating ceramide ratios. The study of adiponectin with physical disuse may be of future interest regarding CV health.

Like adiponectin, FGF21 also decreased with bed rest and inversely correlated with ceramide ratios. FGF21 is recognized as a modulator of metabolism in health and disease. Although primarily produced in the liver, FGF21 can be expressed in several tissues such as skeletal muscle, adipose, pancreas, and brain (39). Interestingly, we previously observed that FGF21 administration in rodents increased adiponectin levels resulting in a decrease in ceramides (40), bridging a relationship between FGF21, adiponectin, and ceramides. Similarly, we observed a positive correlation between changes in circulating FGF21 to changes in adiponectin, suggesting that there could be a mechanistic link between the two in our clinical samples. Studies of exercise and protein supplementation have eluded that FGF21 is regulated by amino acids. For example, exercise depletes amino acid pools and subsequently increases circulating FGF21 (through GCN2 activation and subsequently ATF4 activation in the liver). Similarly, low protein diets result in increases in circulating FGF21, whereas high protein diets and amino acids decrease FGF21. Since prior studies support that bed rest increases plasma amino acids (41,42), it is tempting to speculate that circulating amino acids may be upstream of these hormone and ceramide changes following disuse. Future in vitro studies examining the relationship between amino acid levels, FGF21 signaling, and ceramide would be interesting. Regardless, our data support that FGF21 and adiponectin may be an additional biomarker to diagnose health complications of physical disuse, and also may be a potential mechanism inducing changes in circulating ceramides.

The use of sphingolipids as an aging biomarker is understudied (43), and our findings provide a needed contribution to the small pool of literature supporting that total circulating ceramides are elevated with aging (43,44). Higher baseline circulating ceramides in older adults are likely not to be induced by adiponectin or FGF21, since both of these negative regulators of ceramide expression were increased with age. However, the increase in adiponectin and FGF21 in older adults could be compensatory as a physiological response intended to sequester the rising ceramide levels. Rather, age-related changes in ceramides are likely driven by higher circulating free fatty acids that are commonly increased with age (45), though free fatty acids were not determined in this study and therefore is only speculative.

Higher circulating ceramide levels, particularly the C16:0, C18:0, C20:0, and C24:1 species, have been associated with T2D, obesity, and poor cardiorespiratory fitness (9,10,46). Despite robust insulin resistance following bed rest, we observed a transient increase only in C18:0 and C24:1, in which both increased by Days 3 or 4 of bed rest then returned to baseline by Day 5. Recently, C18:0 [the dominant species in skeletal muscle (47)] was reported to be unchanged in the muscle of young adults after 7 days bed rest (48). Therefore, the physiological relevance of the temporal increases in these circulating ceramides in response to bed rest are questionable. Due to the transient response of ceramide C18:0 observed in our study along with no change observed in muscle ceramide C18:0 in young adults in the citation above, it is unlikely that the observed circulating ceramide response reported herein directly arises from the skeletal muscle tissue.

Though total ceramides decreased with bed rest, this was largely driven by a decrease in ceramide C24:0. Ceramide C24:0 is the dominant ceramide species found in the circulation contributing to nearly 50% of the ceramide pool and presumably arising from the liver (47). Interestingly, ceramide C24:0 has been shown to be lower in patients with fatal myocardial infarctions and positively associated with incidence of CV disease (6,49). Since all ceramide ratios used to calculate CERT1 scores were divided into ceramide C24:0, the decrease in C24:0 is the primary factor responsible for increased CERT1 scores in older adults and therefore CV risk.

In summary, we found that 5 days of bed rest decreased circulating ceramides in older but not younger adults. The changes in circulating ceramides in older adults increased ceramide ratios and the CERT1 score which are predictive of CVD risk and death. The C18:0/C24:0 ceramide ratio, corresponded to lower physical performance (6MW), and adiponectin and FGF21 changes were associated with ceramide ratios indicating these hormones may be potential mechanisms contributing to bed rest-induced deteriorations in CV function. Exploring circulating ceramides as a tool to more accurately and sensitively diagnose or identify disease risk is of great interest. Recently, we used unbiased machine learning methods and identified sphingolipids as more sensitive biomarkers of coronary artery disease than conventional lipid markers cholesterol and triglycerides (50). Thus, ceramides and other sphingolipids along with phospholipid markers may be better biomarkers to identify CV risk. The use of such tools such for other disease states as well as during physical disuse are likely to be of interest in the future to researchers.

Funding

This work was supported by the National Institute on Aging Grants R01 AG-050781 (to M.J.D.), National Institutes of Health (DK112826 and DK108833 to W.L.H. and DK115824, DK116888, and DK116450 to S.A.S.), National Institutes of Health under Ruth L. Kirschstein National Research Service Award (NIH 1T32HL139451NIH 1T32HL139451 to Z.S.M.), the Juvenile Diabetes Research Foundation International (JDRF 3-SRA-2019-768-A-B to W.L.H. and 3-SRA-2019-768-A-B to S.A.S.), the American Diabetes Association (#1-18-ICTS-046 to S.A.S.), the American Heart Association (17GRNT33670881 to S.A.S.), the Margolis Foundation (to S.A.S.). Clinical inpatient and outpatient services were supported by an institutional center grant (CTSA) provided by the National Center for Advancing Translational Sciences (UL1-TR001067). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Acknowledgments

We would like to thank the CCTS nursing, dietary, and medical staff for their assistance with the blood sampling and patient care during the inpatient and outpatient visits. We are also grateful for the support from the Metabolomics Core at the Health Sciences Center of the University of Utah.

Author Contributions

M.J.D., S.A.S., and J.J.P. designed the research proposal; M.J.D., J.J.P., A.I.M., P.T.R., Z.S.M., G.J.S., A.M.P., W.L.H., and S.A.S. conducted the research; M.J.D., J.J.P., and G.J.S. analyzed the data; M.J.D. and J.J.P. wrote the article, M.J.D. had primary responsibility for final content. All authors read and approved final draft of the article.

Conflict of Interest

S.A.S. is a cofounder, consultant, and shareholder for Centaurus Therapeutics.