Age and sex shape plasma lipid associations to skeletal muscle mitochondrial respiration and H₂O₂ emission

Abstract

Aging changes the lipidome and mitochondrial function in a sex-dependent manner, yet their associations remain poorly understood. Twenty-four younger (7M/17F) and forty-three older (21M/22F) adults underwent blood draws and skeletal muscle biopsies for this cross-sectional investigation. Plasma lipidomic profiling was performed via liquid chromatography-tandem mass spectrometry, while peak mitochondrial O₂ utilization (OXPHOS) and hydrogen peroxide (H₂O₂) emission were assessed using high-resolution respirometry. Plasma lipidomic analysis annotated 535 lipid species across 28 different lipid classes. Lipid-age associations were identified in four lipid classes for both sexes with twelve lipid classes demonstrating sex-specific associations, including triglycerides (TG), carnitines (CAR), and fatty acids (FA). For lipid-OXPHOS interactions, the primary lipid class and species associated with higher OXPHOS exclusively in males were ceramides (CER) and dimethyl cholesterol esters (dimethyl-CE), while TG were the primary lipid species associated with impaired OXPHOS in females. For lipid-H₂O₂ interactions_, the primary lipid class and species associated with higher H₂O₂ were methyl desmosteryl esters (methyl-DE), methyl cholesterol esters (methyl-CE), FA and TG in males whereas females exhibited 10 (CE, SM, LPC, dihexosylceramides (Hex2Cer), LPE, PI, HexCer, LPI, CAR, and hexosyl-N-acetylneuraminyl-ceramides (Hex2NeuAcCer)) lipid classes associated exclusively with H₂O₂ emission. These findings establish novel age- and sex-specific relationships between age-related changes in plasma lipids and skeletal muscle mitochondrial function, revealing distinct lipid signatures for respiration and H₂O₂ emission in males and females.

Introduction

Aging is the primary risk factor for the development of cardiovascular disease (CVD), which continues to remain the leading cause of mortality worldwide [1]. This increase in CVD risk with aging is driven, in part, by progressive changes to mitochondrial-specific biological processes (i.e., fission and fusion, DNA mutations, cellular senescence and oxidative stress) resulting in organelle dysfunction [2, 3]. Mitochondria are dynamic organelles responsible for energy production from cellular oxidative metabolism, orchestration of cell signaling pathways and regulation of cell homeostasis [4–7]. Consequently, mitochondrial dysfunction underlies several age- and metabolic-related diseases and independently predicts both cardiovascular-related and all-cause mortality [8–10]. Therefore, given the central role of mitochondrial function in health and longevity [6, 11], understanding the mechanisms contributing to impaired mitochondrial function with aging is clinically relevant for reducing future cardiovascular and cardiometabolic disease risk.

Age-related alterations to major lipid species and lipid pathways in addition to lipid accumulation, peroxidation, and impaired fatty acid utilization are associated with mitochondrial dysfunction [12, 13]. These lipids are key biological molecules that influence numerous cellular processes including cell membrane stability and signaling, energy storage and metabolism [14–17]. Current evidence indicates that specific lipid classes and species are critical for maintaining mitochondrial function, supporting proper assembly of protein complexes in the electron transport chain (ETC) and influencing responsiveness to mitochondrial damage [18–20]. In human skeletal muscle, phosphatidylcholines (PC), phosphatidylethanolamines (PE), and phosphatidylinositols (PI) levels typically increase with aging and are inversely related to mitochondrial respiratory function [21–23]. In contrast, plasma levels of lysophosphatidylcholines (LPCs) decrease with aging and are directly related to impaired in vivo mitochondrial oxidative capacity [24, 25]. Other plasma lipid species including ceramides (CERs), sphingomyelins (SMs), and triglycerides (TG), increase with aging and are related to mitochondrial respiration and the generation of mitochondrial-derived reactive oxygen species (mtROS), which can negatively influence mitochondrial function [26–30]. Therefore, understanding the influence of these lipid-driven alterations on skeletal muscle mitochondrial bioenergetics holds promise for identifying novel strategies to attenuate mitochondrial aging. However, the impact of age on the relationship between primary plasma lipid class levels and skeletal muscle mitochondrial function, measured ex vivo in humans, has not been investigated.

In addition to age, biological sex influences differences in plasma lipid species and skeletal muscle mitochondrial metabolism [31, 32]. Specifically, mitochondrial substrate utilization, antioxidant capacity, respiratory function, and mitochondrial content and biogenesis are affected by age and sex [33, 34]. While both sexes experience mitochondrial dysfunction with aging, the physiological impact may be more pronounced in females, potentially due to age-related declines in estrogen signaling and altered lipid metabolism [35–37]. Evidence from several lipidomic studies demonstrates that females typically have higher plasma concentrations of fatty acids, SMs, and PCs whereas males commonly exhibit higher plasma TGs and lysophospholipids with aging [32, 38, 39]. Nonetheless, the relationship between the aging lipidome, mitochondrial respiration and mtROS production measured ex vivo in males and females has yet to be elucidated. Therefore, understanding how plasma lipid-muscle mitochondrial dynamics are influenced by both age and sex is crucial to expand our current understanding of cardiometabolic health and CVD risk with aging.

The purpose of this study was to determine the association between plasma lipid classes and skeletal muscle mitochondrial function in males and females and if these relationships are influenced by age. We hypothesized that the concentration of lipid species within a specific plasma lipid class would be associated with skeletal muscle mitochondrial respiration (OXPHOS) and H₂O₂ emission in males and females and impacted by aging. Furthermore, we hypothesized that differences in lipid species concentration across age, OXPHOS and H₂O₂ emission ranges within a lipid class would operate in a sex-specific manner. As an exploratory aim, for any lipid classes showing significant associations with OXPHOS or H₂O₂ emission, we examined whether specific lipid species within those classes would be associated with the respective outcome.

Methods

Study design

To understand the association between lipid species and mitochondrial function with aging, we performed a semi-targeted lipidomic profiling of plasma coupled with high resolution respirometry of skeletal muscle mitochondria (Fig. 1A). In total, 67 adults (28 M/39F, mean age 52 ± 20 y) including 24 younger (7 M/17F, mean age 27 ± 7 y) and 43 older adults (21 M/22F, mean age 67 ± 7 y) were recruited to participate in this cross-sectional investigation. Participants were excluded if they had a body mass index (BMI) > 35 kg/m², uncontrolled hypertension (e.g., systolic blood pressure > 160 mmHg or diastolic blood pressure > 100 mmHg), were a smoker, or had a history of or currently had cardiovascular, metabolic, or liver disease or cancer. All older female participants were postmenopausal (> 1 y since last menses). Written informed consent was obtained from all participants prior to participation. All experimental protocols and procedures were approved by the Institutional Review Board at the University of Utah (IRB #111,321) and the Salt Lake City Veterans Affairs medical Center (VAMC). For all study protocols, participants reported to Utah Vascular Research Laboratory (UVRL) located in the Salt Lake City VAMC Geriatric, Research, Education and Clinical Center after an overnight fast (8–12 h) from food, caffeine, alcohol and exercise as previously described [40].

Fig. 1.

A) Study design depicting the participants, methodology and analysis of plasma lipidomics by liquid chromatography tandem mass spectrometry and skeletal muscle mitochondrial O₂ utilization by high resolution respirometry, and B) pie chart depicting the number of annotated lipids species and their respective lipid classes. Figure created in BioRender

Anthropometrics, resting heart rate and blood pressure, body composition and physical activity

Upon arrival at the laboratory, anthropometric measurements of height and weight were used to calculate body mass index (BMI). Following 5 min of uninterrupted rest in a quiet thermoneutral room, resting heart rate was assessed by a 3-lead electrocardiogram (AcqKnowledge, BIOPAC systems) and resting brachial blood pressure (BP) was measured using an automated sphygmomanometer (Tango M2, SunTech Medical). All resting measures were made in triplicate and averaged for all analyses. Dual x-ray absorptiometry (DEXA) was used to assess body composition including total body fat percentage, fat mass and lean muscle mass. All participants wore an accelerometer for a minimum of fourteen continuous days to assess physical activity (PA) status (i.e., average steps/day) (GT1M; Actigraph, Pensacola, FL, USA). Cardiometabolic health scores were derived based on Life’s Essential 8 for the following parameters: BMI, brachial BP, non-high-density lipoprotein cholesterol, blood glucose and PA [41].

Blood chemistry

Venous blood samples were obtained from the antecubital vein by a trained member of the research team. Standard blood chemistry (i.e., lipid panel) was analyzed by the Veterans Affairs blood laboratory for total cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL), triglycerides and blood glucose. Additional plasma samples were aliquoted and stored at −80 °C for future analysis.

Lipid extraction

Liquid chromatography-mass spectrometry (LC–MS) grade methanol was obtained from Honeywell Burdick & Jackson (Morristown, NJ), and 1-butanol from Fisher Scientific (Waltham, MA). Internal standards were acquired from Avanti Polar Lipids (Birmingham, AL), including C16 Ceramide-d7 (d18:1-d7/16:0; 860,676), C18 Ceramide-d7 (860,677), C24 Ceramide-d7 (860,678), C24:1 Ceramide-d7 (860,679), C17 Glucosyl(β) Ceramide (860,569), C18:1 Dihydroceramide (86,062), Sphingosine-1-Phosphate-d7 (860,659), Sphinganine-1-Phosphate-d7 (860,694), Sphingosine-d7 (860,657), Oleic Acid-d9 (861,809), and the SPLASH Lipidomix standard (330,707). Palmitic acid-d31 (16,497) and d3-Palmitoylcarnitine (C16; 26,569) were obtained from Cayman Chemical (Ann Arbor, MI).

Lipids were extracted using a modified version of the single-phase Alshehry method, employing a methanol/1-butanol mixture as described previously [42, 43]. Briefly, 20 µL of plasma was aliquoted into a 96-well plate. Samples were randomized, and 200 µL of extraction solvent containing internal standards was added to each well. Plates were sonicated at 4 °C for 10 min then centrifuged at 1,100 × g for 10 min at 4 °C. Supernatants were transferred to a new 96-well plate for LC–MS/MS analysis. Pooled quality control (QC) and process blank samples were prepared alongside the experimental samples.

LC–MS/MS analysis

Lipid extracts were analyzed using a Sciex 7500 + QTRAP mass spectrometer coupled to a Waters ACQUITY UPLC system. Chromatographic separation was performed on a Waters ACQUITY UPLC CSH C18 column (2.1 × 100 mm, 1.7 µm) with a VanGuard pre-column (2.1 × 5 mm, 1.7 µm). The column was maintained at 60 °C, with a flow rate of 0.4 mL/min and an injection volume of 1 µL. Mobile phase A was acetonitrile/water (60:40, v/v) with 10 mM ammonium formate and 0.1% formic acid; mobile phase B was isopropanol/acetonitrile/water (90:9:1, v/v/v) with the same additives. The gradient was as follows: 0–2.4 min, 15–30% B; 2.4–3.0 min, 30–48% B; 3.0–13.2 min, 48–82% B; 13.2–13.8 min, 82–99% B; 13.8–16.7 min, 99% B; 16.7–16.8 min, 99–15% B, followed by a 4 min re-equilibration at 15% B.

Data were acquired in both positive and negative ion modes using scheduled multiple reaction monitoring (MRM). Source conditions included: ion source gas 1 at 45 psi, gas 2 at 60 psi, curtain gas at 45 psi, collision gas at 9 psi, and source temperature at 400 °C. The spray voltage was set to 5200 V in positive mode and 4500 V in negative mode. Data acquisition was performed using Sciex Analyst software; peak integration and quantification were conducted using MultiQuant. Lipids were quantified relative to class-specific internal standards and reported in pmol/mL plasma. This method was adapted from Huynh et al. [44].

Skeletal muscle biopsy

Muscle samples were collected from the vastus lateralis (VL) under sterile conditions as previously described [45]. Muscle biopsies were performed in the fasted condition following local anesthesia (2% xylocaine) using a Bergstrom needle with suction ∼15 cm proximal to the knee at a depth of ∼3.5 cm below the muscle fascia [46]. Immediately following the biopsy, ∼20 mg of muscle was stored in ice-cold biopsy preservation fluid (BIOPS; 2.77 mM CaK2EGTA, 7.23 mM K2EGTA, 6.56 mM MgCl2, 0.5 mM DTT, 50 mM K-MES, 20 mM imidazole, 20 mM taurine, 5.77 mM Na2ATP, and 15 mM phosphocreatine, pH 7.1 at 4 ◦C) for the subsequent respiratory analyses.

Skeletal muscle mitochondrial respiration and H₂O₂ emission

Muscle samples were prepared and permeabilized as described previously [45, 47, 48]. In brief, adipose, and connective tissues were removed from the muscle fibers in 4 °C BIOPS by careful use of fine-tip forceps under a microscope (SZX10; Olympus, Center Valley, PA, USA). The muscle fibers were then permeabilized with saponin (50 ug/mL) in BIOPS for 30 min. Following the saponin treatment, the muscle fibers were rinsed twice for 10 min in mitochondrial respiration medium (MIR05; 2.77 mM CaK2EGTA, 7.23 mM K2EGTA, 6.56 mM MgCl2, 0.5 mM DTT, 20 mM imidazole, 5.77 mM ATP, 15 mM phosphocreatine, 50 mM K-MES, and 20 mM taurine, pH 7.0).

High-resolution respirometry (O2k, OROBOROS Instrument, Innsbruck, Austria) was used to assess mitochondrial O₂ utilization. Permeabilized muscle fibers were placed in an O2k-chamber containing 2 ml of the MIR05 solution that was continuously stirred and maintained at 37 °C. Respiration rate was measured using a substrate-uncoupler-inhibitor-titration (SUIT) protocol.

The SUIT protocol utilized the following substrates (final concentration in respiration chamber): glutamate (10 mM) and malate (2 mM) for Complex I activity (State 2 respiration), ADP (5 mM) for complex I oxidative phosphorylation (State 3—complex I), succinate (10 mM) to activate Complex II to measure combined State 3—Complex I and II supported OXPHOS, rotenone (0.5 uM) to inhibit complex I (State 3—Complex II), and oligomycin (5 uM) to inhibit complex V (State 4).

In a separate experimental protocol designed to assess mitochondrial H₂O₂ emission, Amplex UltraRed (10 µM), horseradish peroxidase (1 U/ml), and superoxide dismutase (5 U/ml) were added to the chamber prior to substrate addition. Pyruvate (5 mM), malate (2 mM), and succinate (10 mM) were then added to capture peak H₂O₂ emission, which was recorded using a Fluorescence-Sensor Green of the LED2-Module (O2k-Fluo, OROBOROS Instrument, Innsbruck, Austria). Before each experiment, H₂O₂ standard curves were performed to convert the fluorescence signal to a H₂O₂ emission value [48]. The Oroboros DatLab system was used for all analyses and results were normalized to the dry muscle tissue weight, which was measured after the protocol ended.

Statistical analysis

For an initial overview, we performed differential abundance testing of plasma lipid species between age groups and sexes using MetaboAnalystR version 6.0 [49], with significance thresholds of raw p < 0.05 and false discovery rate p < 0.01. To test our primary hypotheses, we used linear mixed-effects models to examine associations between plasma lipid concentrations and age, OXPHOS, and H₂O₂ emission. Plasma lipid concentrations were natural log-transformed to address right-skewed distributions. Log-transformed lipid concentrations, age, OXPHOS, and H₂O₂ emission were z-scored (standardized) to enable interpretation of main effects in the presence of interactions and to facilitate comparison of effect sizes across predictors. We fit three linear mixed-effects models, with full interactions terms and lower-order terms, with plasma lipid concentration as the dependent variable: (1) Lipid-Age: lipid class × age × sex; (2) Lipid-OXPHOS: lipid class × age × sex × OXPHOS; and (3) Lipid-H₂O₂ Emission: lipid class × age × sex × H₂O₂ emission. All models included random intercepts for lipid species and random slopes for age, allowing baseline concentrations and age trajectories to vary across species. This approach simultaneously evaluated all lipid species, treated age as continuous, and leveraged species-specific variation to increase statistical power and detect higher-order interactions. Although each participant contributed measurements for multiple lipid species across lipid classes, we modeled random intercepts by species rather than by participant because species-level variation which accounted for significantly more variance in lipid concentration than participant-level variation. For each model, we tested whether the association between plasma lipid concentration and the model’s primary predictor (age, OXPHOS, or H₂O₂ emission) differed across lipid classes and between sexes, using omnibus contrasts for significant three-way interactions (p < 0.05). We tested whether the marginal slopes (standardized β coefficients) of the primary predictor for each model were significantly different from zero (p < 0.05) for each lipid class and sex combination, without correcting for multiple comparisons, as no cross-class comparisons were made. We conducted pairwise comparisons to test whether the β coefficients differed between males and females within each lipid class, with Benjamini–Hochberg false-discovery rate correction (q < 0.05) across all lipid classes. For our exploratory aim, we calculated Pearson correlations between plasma lipid concentrations and the model’s primary predictor for individual lipid species within lipid classes that showed significant associations (p < 0.05) in the linear mixed-effects model. These exploratory correlations were not adjusted for multiple comparisons. Results from the linear mixed-effects models are reported as standardized β coefficients, 95% confidence intervals (CI), and p-values. We used a two-way analysis of variance (ANOVA) to compare participant characteristics and skeletal muscle mitochondrial function across age group and sex, with significance indicated by p < 0.05. Statistical analyses were performed using Stata (Stata 17; StataCorp LLC, College Station, TX, USA) or GraphPad prism (10.6.1).

Results

Participant characteristics and skeletal muscle mitochondrial respiration rates of the participant population are presented in Tables 1 and 2. In brief, older females had lower fat mass and higher cholesterol, HDL cholesterol, and were more physically active compared to older males. Older females also had higher resting heart rate, brachial blood pressure, cholesterol, triglycerides and a lower cardiometabolic health score compared to younger females (all, Table 1, p < 0.05). No differences were observed between younger and older males (Table 1, p > 0.05). Compared to older males, older females had lower state 2 complex I, state 3 complex I, state 3 complex I + II and state 4 respiration rates (all, Table 2, p < 0.05). No age-related differences were observed for mitochondrial respiration in males and no age- or sex-related differences were observed for H₂O₂ emission (all, Table 2, p > 0.05).

Table 1.

Participant characteristics

	Young						Old
	Male (n=7)			Female (n=17)			Male (n=21)			Female (n=22)
Age, y	25	±	7	27	±	6	69	±	6b	64	±	5a,c
Resting HR, bpm	67	±	14	73	±	10	65	±	10	66	±	8a
BMI, kg/m2	25	±	4	24	±	4	28	±	4	27	±	5
Body fat, %*	27	±	5	36	±	5b	32	±	6	39	±	7c
Fat mass, kg*	22.1	±	0.8	25.0	±	0.7	30.5	±	1.0	28.5	±	1.1c
Lean mass, kg*	54.9	±	0.7	40.9	±	0.5b	58.6	±	0.7	40.9	±	0.9
Systolic BP, mmHg	119	±	12	113	±	7	122	±	14	127	±	16a
Diastolic BP, mmHg	74	±	10	71	±	5	75	±	8	78	±	9a
Pulse pressure, mmHg	45	±	6	41	±	7	47	±	13	49	±	10a
Mean arterial pressure, mmHg	89	±	11	85	±	5	90	±	8	95	±	11a
Cholesterol, mg/dL	206	±	58	170	±	29b	177	±	39	205	±	37a,c
LDL, mg/dL	150	±	56	105	±	32b	114	±	38b	127	±	33
HDL, mg/dL	49	±	7	57	±	14	50	±	11	65	±	18c
Triglycerides, mg/dL	107	±	54	72	±	46	90	±	39	103	±	38a
Glucose, mg/dL	89	±	5	86	±	7	92	±	7	90	±	6
Cardiometabolic health score	80	±	15	90	±	12	73	±	11	76	±	14a
Physical activity, steps/day#	5,599	±	2,684	6,236	±	2,145	4,361	±	2,252	6,738	±	3,919c
Comorbid Conditions
Obesity, n (%)	2 (26)			2 (12)			7 (33)			5 (23)
Prediabetes, n (%)	0 (0)			2 (12)			4 (19)			2 (9)
Dyslipidemia, n (%)	3 (43)			4 (24)			4 (19)			11 (50)

*n=16 young females, n=18 old males, n=19 old females, ^#n= 16 young females. BMI, body mass index; BP, blood pressure; HDL, high-density lipoprotein; HR, heart rate; LDL, low-density lipoprotein. Data are mean±SD. p<0.05 vs ^ayoung females, ^byoung males and ^cold males

Table 2.

Skeletal muscle mitochondrial respiration and H₂O₂ emission

	Young						Old
	Male (n=7)			Female (n=16)			Male (n=18)			Female (n=20)
State 2 Complex I respiration, pmol/s/mg	4.5	±	1.6	3.8	±	1.8	4.7	±	2.0	2.6	±	1.4a,c
State 3 Complex I respiration, pmol/s/mg	11.2	±	3.5	9.7	±	5.8	11.0	±	4.4	7.9	±	4.4c
State 3 Complex I+II respiration, pmol/s/mg	68.9	±	12.2	66.1	±	25.1	81.2	±	22.6	58.2	±	17.7c
State 3 Complex II respiration, pmol/s/mg	55.5	±	8.4	47.5	±	18.1	51.9	±	18.1	42.6	±	12.0
State 4 respiration, pmol/s/mg	15.1	±	5.3	12.0	±	4.3	17.3	±	9.2	12.3	±	5.5c
H2O2 emission at State 3, pmol/s/mg	0.84	±	0.28	0.70	±	0.56	0.78	±	0.52	0.90	±	1.0

*n=12 young females. H₂O₂, hydrogen peroxide. Data are mean±SD. p<0.05 vs ^ayoung females, ^byoung males and ^cold males

Plasma lipidome across age, sex, and lipid class

Plasma lipidomic analysis annotated 535 lipid species across 28 different lipid classes (Fig. 1B). As a preliminary overview, differential abundance testing indicated that 214 lipid species (209 higher, 5 lower) differed for older vs. younger adults (Fig. 2 A) and that 6 lipid species (5 higher, 1 lower) differed for females vs. males (Fig. 2B). The rigorous linear mixed-effects models detected significant (all, p < 0.05) three-way interactions between lipid class, sex, and each primary predictor (age, OXPHOS, or H₂O₂ emission).

Fig. 2.

A) Volcano plot indicating the lipid species that differed (up and down regulated) with aging and B) volcano plot indicating the lipid species that differed (up and down regulated) between females and males, each using a threshold of p<0.05 and an adjusted p-value false discovery rate of p <0.01

Lipid-age associations across lipid class and sex

The results of the linear mixed effects for lipid-age are presented in Fig. 3. The slopes of the relationship between plasma lipid concentration and age were significant (all, p < 0.05) in both males and females for 4 lipid classes including LPE, PI, CER, and LPC.In females only, the slopes of the relationship between plasma lipid concentration and age were significant (all p < 0.05) for an additional 10 lipid classes including TG, PG, DG, PS, SM, PE, Hex2NeuAcCer, FA, CAR, PC, HexCer and CE. In males only, the slopes of the relationship between plasma lipid concentration and age were significant (p < 0.05) for Cholic Acid.The slopes of the relationship between plasma lipid concentration and age were significantly different (all p < 0.05) between males and females for 15 lipid classes including Cholic Acid,PS, PG, TG, DG, PE, SM, PI, CE, LPE, PC, HexCer, CER, CAR and LPC.

Fig. 3.

Forest plots for the association between lipid class and age stratified by sex. Symbols indicate the lipid-age coefficient, representing the change in lipid species concentration for a given change in age within a lipid class in males (blue) and females (red), where bold symbols indicate significant coefficients. Asterisks indicate significantly different lipid-age coefficients between males and females within a lipid class. CAR, carnitines; CE, cholesterol esters; CER, ceramides; DE, double bond equivalents; DG, diglycerides; Dimethyl-CE, dimethyl cholesterol esters; FA, fatty acids; HexCer, hexosylceramides; Hex2Cer, dihexosylceramides; Hex3Cer, trihexosylceramides; Hex2NeuAcCer, hexosyl-N-acetylneuraminyl-ceramides; LPC, lysophosphatidylcholines; LPE, lysophosphatidylethanolamines; LPI, lysophosphatidylinositol; Methyl-CE, methyl cholesterol esters; Methyl-DE, methyl desmosteryl esters; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SM, sphingomyelin; SPBP, sphingoid based phosphates; ST, sterols and TG, triglycerides

Lipid-OXPHOS associations across lipid class and sex

The results of the linear mixed effects for lipid-OXPHOS are presented in Fig. 4. The slopes of the relationship between plasma lipid concentration and OXPHOS were significant (all p < 0.05) in both males and females for 6 lipid classes including PS, DG, LPC, PI, TG, and PC. In females only, the slopes of the relationship between plasma lipid concentration and OXPHOS were significant (all p < 0.05) for an additional 2 lipid classes, FA and PE. In males only, the slopes of the relationship between plasma lipid concentration and OXPHOS were significant (all p < 0.05) for an additional 7 lipid classes including CER, Dimethyl-CE, SM, LPE, CE, and Cholic Acid. The slopes of the relationship between plasma lipid concentration and OXPHOS were significantly different (all p < 0.05) between males and females for 7 lipid classes including PS, DG, CER, TG, SM, PI, and Cholic Acid.

Fig. 4.

Forest plots for the association between lipid class and OXPHOS stratified by sex. Symbols indicate the lipid-OXPHOS coefficient, representing the change in lipid species concentration for a given change in OXPHOS within a lipid class in males (blue) and females (red), where bold symbols indicate significant coefficients. Asterisks indicate significantly different lipid-OXPHOS coefficients between males and females within a lipid class. CAR, carnitines; CE, cholesterol esters; CER, ceramides; DE, double bond equivalents; DG, diglycerides; Dimethyl-CE, dimethyl cholesterol esters; FA, fatty acids; HexCer, hexosylceramides; Hex2Cer, dihexosylceramides; Hex3Cer, trihexosylceramides; Hex2NeuAcCer, hexosyl-N-acetylneuraminyl-ceramides; LPC, lysophosphatidylcholines; LPE, lysophosphatidylethanolamines; LPI, lysophosphatidylinositol; Methyl-CE, methyl cholesterol esters; Methyl-DE, methyl desmosteryl esters; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SM, sphingomyelin; SPBP, sphingoid based phosphates; ST, sterols and TG, triglycerides

Lipid-H₂O₂ emission associations across lipid class and sex

The results of the linear mixed effects for lipid-H₂O₂ emission are presented in Fig. 5. The slopes of the relationship between plasma lipid concentration and H₂O₂ emission were significant (all p < 0.05) in both males and females for 5 lipid classes including FA, TG, PE, PC, and PS. In females only, the slopes of the relationship between plasma lipid concentration and H₂O₂ emission were significant (all p < 0.05) for an additional 10 lipid classes including CE, SM, LPC, Hex2Cer, LPE, PI, HexCer, LPI, CAR, and Hex2NeuAcCer. In males only, the slopes of the relationship between plasma lipid concentration and H₂O₂ emission were significant (all p < 0.05) for an additional 5 lipid classes including Methyl-DE, Methyl-CE, PG, Dimethyl-CE, and CER. The slopes of the relationship between plasma lipid concentration and H₂O₂ emission were significantly different (all p < 0.05) between males and females for 10 lipid classes including Methyl-DE, FA, HexCer, CE, CAR, PC, CER, PE, TG, and PS.

Fig. 5.

Forest plots for the association between lipid class and H₂O₂ emission stratified by sex. Symbols indicate the lipid-H₂O₂ coefficient, representing the change in lipid species concentration for a given change in H₂O₂ within a lipid class in males (blue) and females (red), where bold symbols indicate significant coefficients. Asterisks indicate significantly different lipid- H₂O₂ coefficients between males and females within a lipid class. CAR, carnitines; CE, cholesterol esters; CER, ceramides; DE, double bond equivalents; DG, diglycerides; Dimethyl-CE, dimethyl cholesterol esters; FA, fatty acids; HexCer, hexosylceramides; Hex2Cer, dihexosylceramides; Hex3Cer, trihexosylceramides; Hex2NeuAcCer, hexosyl-N-acetylneuraminyl-ceramides; LPC, lysophosphatidylcholines; LPE, lysophosphatidylethanolamines; LPI, lysophosphatidylinositol; Methyl-CE, methyl cholesterol esters; Methyl-DE, methyl desmosteryl esters; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SM, sphingomyelin; SPBP, sphingoid based phosphates; ST, sterols; TG, triglycerides

Correlations at the individual lipid species level across age and sex

For our exploratory aim, we calculated Pearson correlations between plasma lipid concentrations and for individual lipid species within lipid classes with significant relationship in the model. Specifically, higher species abundance of CERs and LPCs were associated with greater OXPHOS in males as a whole cohort and when split by age, higher species abundance of CERs and LPCs remained associated with greater OXPHOS in older males (all, p < 0.05, Fig. 6A-B). In females, higher abundance of TG species was associated with lower OXPHOS as a whole cohort but when split by age higher species abundance of LPCs and PIs were associated with greater OXPHOS in older females (all, p < 0.05, Fig. 7A-B). Higher abundance of TGs, FAs, and PCs were associated with greater H₂O₂ emission in males as a whole cohort and in older males (all, p < 0.05, Fig. 8A-B) whereas higher amounts of HexCer species were associated with lower H₂O₂ emission in females (all, p < 0.05, Fig. 9A-B).

Fig. 6.

Heatmap correlation matrix indicating the lipid species associated with skeletal muscle OXPHOS in A) all males, B) older males and C) younger males. All values are p<0.05.CER, ceramides; LPC, lysophosphatidylcholines; Dimethyl-CE, dimethyl cholesterol esters; LPE, lysophosphatidylethanolamines; HexCer: hexosylceramides; TG, triglycerides; YN, younger

Fig. 7.

Heatmap correlation matrix indicating the lipid species associated with skeletal muscle OXPHOS in A) all females, B) older females and C) younger females. All values are p<0.05. LPC, lysophosphatidylcholines; TG, triglycerides; DG, diglycerides; PE, phosphatidylethanolamines; PI, phosphatidylinositols; YN, younger

Fig. 8.

Heatmap correlation matrix indicating the lipid species associated with skeletal muscle H₂O₂emission in A) all males and B) older males. All values are p<0.05. TG, triglycerides; FA, fatty acids; PC, phosphatidylcholines; PE, phosphatidylethanolamines; Methyl-DE, methyl desmosteryl esters; CER, ceramides

Fig. 9.

Heatmap correlation matrix indicating the lipid species associated with skeletal muscle H₂O₂emission in A) all females and B) younger females. All values are p<0.05.HexCer, hexosylceramide; TG, triglycerides; YN, younger

Discussion

This study provides novel insight regarding the association between age-related changes in the plasma lipidome and mitochondrial function in males and females. Specifically, the findings demonstrate 1) distinct lipid classes are associated with skeletal muscle mitochondrial OXPHOS in males (CER, dimethyl-CE, HexCer, SM, LPE, CE, and Cholic Acid) and females (FA and PE), 2) distinct lipid classes are associated with skeletal muscle mitochondrial H₂O₂ emission in males (methyl-DE, methyl-CE, PG, dimethyl-CE, and CER) and females (CE, SM, LPC, Hex2Cer, LPE, PI, HexCer, LPI, CAR and Hex2NeuAcCer) and 3) specific lipid species within a given lipid class are associated in a sex-specific manner with OXPHOS and H₂O₂ emission in males (OXPHOS: CER, dimethyl-CE, LPC and LPE and H₂O₂: TG, FA, methyl-DE) and females (OXPHOS: TG, DG, and PE and H₂O₂: HexCer). To our knowledge, this is the first study to comprehensively assess the interactive effects of age and biological sex on the association between plasma lipid classes and species and ex vivo skeletal muscle mitochondrial OXPHOS and H₂O₂ emission. Collectively, these findings provide evidence of how age and biological sex differentially shape the plasma lipidome and its connection with skeletal muscle mitochondrial function.

Lipids are essential compounds involved in many biological and physiological processes including cell membrane integrity, signal transduction, enzyme activation and metabolism [14–17]. Current evidence from The Genetics of Lipid-Lowering Drugs and Diet Network (GOLDN) study and others have identified biological sex as a key-determinant of age-related changes in lipid species [32, 39, 50]. Females typically exhibit higher plasma concentrations of FAs, SMs, and PCs whereas males commonly exhibit higher plasma concentrations of TGs and lysophospholipids with aging [32, 38, 39]. Our data extend these findings by identifying novel relationships between plasma lipid concentration and age within numerous lipid classes, including LPEs, PIs, CERs and LPCs in both males and females, TGs, DGs, PGs, PSs, PEs, CARs and CEs in females and cholic acid in males. Furthermore, our data show that the slopes of the relationships between plasma lipid concentration and age were significantly different between males and females for TGs, DGs, PGs, PSs, PEs, PCs, PIs, LPEs, LPCs, CERs, CARs, CEs and cholic acid highlighting novel sex-differences. Overall, these data underscore the importance of considering both age and biological sex in plasma-lipid based metabolism research.

Changes in lipid homeostasis and membrane composition modulate both skeletal muscle mitochondrial structure and function, making lipids key regulators of mitochondrial health [20]. Previous evidence demonstrates that low levels of plasma LPCs are associated with impaired in vivo mitochondrial oxidative capacity with aging using ³¹P-MRS [24, 25]. We extend these findings in demonstrating, for the first time, that the slope of the relationship between plasma lipid concentrations and OXPHOS were significant for several lipid classes including TGs, DGs, PSs, PIs, PCs and LPCs in both males and females; PEs, and FAs in females and CERs, SMs, LPEs, CEs and cholic acid in males. Biological sex influences the slopes of these relationships with sex differences in TGs, DGs, PSs, PIs, CERs, SMs, FAs and cholic acid.

Phospholipids are a major lipid class that changes during the aging process, either increasing or decreasing, depending on the lipid species. LPCs and PIs were two primary phospholipid classes associated with alterations in OXPHOS in both males and females. LPCs have been shown to exert positive and negative effects on mitochondrial function depending on specific conditions [51–53]. Indeed, one potential mechanism by which higher amounts of LPCs influence mitochondrial OXPHOS is through the synthesis of cardiolipin, since LPCs are key intermediates in cardiolipin biosynthesis thus contributing to enhanced mitochondrial function [20]. For both males and females, a given increase in LPCs was associated with a similar increase in OXPHOS. Therefore, it appears that higher abundance of LPCs support a mitochondrial protective effect with aging in both males and females. Strategies aimed to improve LPCs may be beneficial for maintaining or enhancing mitochondrial metabolism and OXPHOS in males and females with aging.

PIs serve as transient signaling molecules that have recently been shown to regulate different aspects of mitochondrial dynamics including fission, fusion, mitophagy, vesicle formation and trafficking and lipid transfer [54]. PIs are essential cellular transducers for controlling mitochondrial metabolism via membrane contact and bioenergetics by coordinating peroxisome-mitochondria communication [55]. Importantly, deletion of specific PIs such as PI(4,5)P₂ results in decreased expression of lipid transporter proteins, and β-oxidation genes, which in turn can contribute to structural and functional impairments in the mitochondria, resulting in an inability to maintain adequate metabolic homeostasis. Interestingly, the slope between PI and OXPHOS was positive in females, but negative for males. These data suggest that PI abundance may be clinically relevant for mitochondrial health and may provide novel insight into the impact of age and sex on mitochondrial function.

Maintenance of the intracellular redox environment is also a major role of the mitochondria that is required for multiple cellular actions including metabolism and ROS production [56, 57]. In terms of bioenergetics, in the presence of excess triglycerides, mitochondria increase β-oxidation and the breakdown of fatty acids for energy, thereby promoting excess electron leakage from the ETC to react with molecular oxygen, forming ROS [58, 59]. In the current study, TG was a primary class associated with greater skeletal muscle mitochondrial H₂O₂ emission. Excess TGs promote excess mitochondrial ROS production including greater H₂O₂ emission through several mechanisms including mitochondrial uncoupling of the ETC from ATP synthesis, increased mitochondrial fission (i.e., fragmentation) linked to energy deficits and apoptosis, reduced antioxidant capacity and increased inflammatory signaling [60, 61]. Furthermore, there is a coordinated relationship between TG mobilization and mitochondria particularly under stress conditions, where TGs are mobilized from lipid stores to generate new mitochondrial machinery, including the membrane lipid, cardiolipin [62, 63]. Recent evidence demonstrates that impaired immobilization of TGs, via the lack of the TG lipase gene, adipose triglyceride lipase (ATGL), impairs stress-induced mitochondrial recovery [64]. Our data indicate that a given increase in TGs is associated with similar increases in skeletal muscle H₂O₂ emission for both males and females. This is noteworthy as alterations in TG composition may serve as an important cellular modification underlying mitochondrial dysfunction and cellular stress with aging. These data support the notion that changes in TG metabolism and composition with aging may contribute, in part, to metabolic and mitochondrial dysfunction and support the notion that hypertriglyceridemia is a clinically relevant condition linked to mitochondrial and metabolic dysfunction. Taken together, the influence of age and biological sex are important factors to consider regarding the relation between plasma lipid concentrations and mitochondrial function. Future longitudinal research is needed to understand the underlying physiology by which certain lipid classes and distinct lipid species may modulate mitochondrial function in a sex-specific manner.

Through our exploratory analysis, we identified 9 different CER species that were associated with greater OXPHOS in males. This is notable as multiple lines of evidence demonstrate higher CER accumulation with aging is associated with cardiometabolic disease [65] and contributes to metabolic and mitochondrial dysregulation [66]. Yet, the extent to which distinct CER species impact mitochondrial function and energy metabolism remains limited [67]. CER species differing in acyl chain length execute distinct features linked to mitochondrial function and OXPHOS including suppressed mitochondrial ETC activity, increased ROS, reduced mitochondrial membrane potential, mitophagy and apoptosis [68]. However, the effect of biological sex on the association between specific CER species and its relationship with skeletal muscle mitochondrial OXPHOS is unknown. Thus, our data provide novel insight linking distinct plasma CER species with enhanced skeletal muscle OXPHOS with aging in males. Additional research is needed to understand the potential mitochondrial protective effects of higher CERs in males with aging and in cardiometabolic dysfunctional populations.

For females, we identified 26 distinct TG species that were associated with lower OXPHOS. This is notable as higher amounts of circulating TGs are linked to age-related mitochondrial dysfunction in females [69]. Mitochondria generate energy through OXPHOS, which is fueled, in part, by FAs from the breakdown of TGs. However, while mitochondria can adapt to high levels of TGs by increasing energy expenditure and reducing oxidative stress in the short-term, chronically elevated levels of triglycerides (i.e., hypertriglyceridemia) can result in a maladaptive response whereby excess lipid accumulation promotes cell and organ toxicity (i.e., lipotoxicity) and lipid-induced redox imbalance leading to mitochondrial dysfunction [70]. Additional key mechanisms contributing to lower mitochondrial OXPHOS from hypertriglyceridemia include decreased mitochondrial efficiency and content, impaired insulin signaling, and increased mitochondrial uncoupling [28, 71]. However, the exact mechanisms underlying the physiological role of how TG species may modulate mitochondrial function in a sex-specific manner in females remains limited.

Strengths and limitations

A primary strength of this study is the use of LC–MS/MS to comprehensively assess alterations in the plasma lipidome with aging in males and females. Moreover, the direct assessment of mitochondrial O₂ utilization and H₂O₂ emission in permeabilized skeletal muscle fibers using high resolution respirometry provides valuable insight into the functional consequences of systemic lipidomic changes and its effects on oxidative metabolism. This study is not without limitations. First, the current sample consists of a primarily homogeneous sample of Caucasian adults potentially limiting the generalizability of these findings. Additionally, sex hormones were not assessed in younger or older females limiting the interpretation for the role of estrogen dysregulation on lipid metabolism and its effect on mitochondrial function. Furthermore, these results are correlational and cannot be interpreted as cause and effect. Despite these limitations these data present a novel framework for future studies and longitudinal work evaluating mechanisms for how changes in plasma lipid classes and species may impact skeletal muscle mitochondrial respiratory capacity and H₂O₂ emission in an age- and sex-dependent manner. Future investigations using a larger and more diverse population with or without disease/comorbid conditions to examine the cross-sectional and temporal changes in lipid alterations and their effects on mitochondrial health are needed to extend the current findings.

Conclusion

These data provide evidence that alterations in the plasma lipidome with aging are associated with skeletal muscle mitochondrial OXPHOS and H₂O₂ emission in a sex-specific manner. These findings offer novel insight for individual plasma lipid classes and species as clinically relevant targets, to improve metabolic and mitochondrial health and lower CVD risk with aging in males and females. Further, these data highlight the importance of considering sex-specific differences in aging and in lipid and skeletal muscle metabolism research.

Abbreviations

CAR

Carnitines

CE

Cholesterol esters

CER

Ceramides

DE

Double bond equivalents

DG

Diglycerides

Dimethyl-CE

Dimethyl cholesterol esters

FA

Fatty acids

HexCer

Hexosylceramides

Hex2Cer

Dihexosylceramides

Hex3Cer

Trihexosylceramides

Hex2NeuAcCer

Hexosyl-N-acetylneuraminyl-ceramides

LPC

Lysophosphatidylcholines

LPE

Lysophosphatidylethanolamines

LPI

Lysophosphatidylinositols

Methyl-CE

Methyl cholesterol esters

Methyl-DE

Methyl desmosteryl esters 

PC

Phosphatidylcholines

PE

Phosphatidylethanolamines

PG

Phosphatidylglycerols

PI

Phosphatidylinositols

PS

Phosphatidylserines

SM

Sphingomyelins

SPBP

Sphingoid based phosphates

ST

Sterols

TG

Triglycerides

Author contributions

N.A.C., B.A.R and J.D.T. conceived and designed research; N.A.C., B.A.R., H.R., H.S., J.A.M. performed experiments; N.A.C., B.A.R., J.A.M., R.M.B. analyzed data; N.A.C., B.A.R., H.R., H.S., J.A.M., B.E.H., R.N.S., R.S.R., M.J.D., R.M.B., J.D.T. interpreted results of experiments; N.A.C. and R.M.B. prepared tables and figures; N.A.C drafted manuscript; N.A.C., B.A.R., R.M.B., J.D.T. edited and revised manuscript; N.A.C., B.A.R., H.R., H.S., J.A.M., B.E.H., R.N.S., R.S.R., M.J.D., R.M.B., J.D.T. approved final version of the manuscript.

Funding

This work was funded, in part, by the Department of Veterans Affairs Advanced Fellowship in Geriatrics (to N.A.C.), U.S. Veterans Affairs Clinical Science Research and Development Career Development Award IK2CX002114 (to R.M.B), Department of Veterans Affairs Merit Award I01RX003810 (to J.D.T) and the National Institutes of Health Grant HL142603 (to J.D.T).

Data availability

Data will be made available from the corresponding author upon reasonable request.

Declarations

Conflict of interests

No conflicts of interest, financial, or otherwise are declared by the authors.