Self-Reported Sleep Efficiency and Duration are Associated with Bioenergetic Function in Peripheral Blood Mononuclear Cells (PBMCs) of Adults

Abstract

Poor sleep may impair systemic mitochondrial bioenergetics, but this relationship has not been examined in humans. This study examined associations of self-reported sleep with peripheral blood mononuclear cell (PBMC) bioenergetics in adults. Forty-three participants completed the Pittsburgh Sleep Quality Index from which sleep indices were calculated. PBMCs were analyzed for bioenergetics using extracellular flux analysis. Sleep efficiency was positively correlated with maximal respiration and spare capacity. Lower sleep efficiency and longer sleep duration were associated with lower Bioenergetic Health Index in age-, sex-, and body mass index-adjusted models. Findings indicate that sleep is related to systemic bioenergetic function in humans.

1. Introduction

Sleep plays critical roles in aging, health, and disease, but these functions are poorly understood at the cellular level. Disrupted and insufficient sleep are associated with numerous chronic and aging-related conditions including cardiovascular disease [1], type 2 diabetes [2], and Alzheimer’s disease [3], each of which involves marked mitochondrial dysfunction [4–6]. Measurement of bioenergetic processes, in which mitochondria convert nutrients into cellular energy, provides an innovative opportunity to examine downstream markers of mitochondrial dysfunction. As such, bioenergetics may represent a key cellular process through which disturbed sleep contributes to diseases of aging.

Bioenergetic dysfunction is a hallmark of aging and pathology and has been considered a “canary in the coal mine” for accelerated aging and chronic disease development [7,8]. Recent methodological advances now permit minimally invasive assessment of cellular bioenergetics in platelets and peripheral blood mononuclear cells (PBMCs), which reflect systemic bioenergetic capacity including skeletal muscle, cardiac muscle, and brain energetics [9,10]. PBMC bioenergetics are compromised in patients with Parkinson’s disease [11] and chronic fatigue syndrome [12] and are related to indices of aging including gait speed, grip strength, and brain volume, and cognitive function [13–15]. Given the rapidly growing understanding of the role of bioenergetic impairment in aging and disease, identifying potentially modifiable determinants of such dysfunction, such as sleep, is vital.

Emerging evidence suggests that disrupted, short, and long sleep may compromise bioenergetic function. In Drosophila, short-term sleep deprivation impaired oxidative phosphorylation and electron transport activity, increased reactive oxygen species production, and reduced activity of antioxidant enzymes [16]. Mice exposed to 72 hours of sleep deprivation exhibited decreased activity of mitochondrial electron chain complexes I-III [17], which are directly involved in cellular ATP generation. Disrupted and insufficient sleep in humans has been linked with oxidative damage, which harms mitochondria and compromises bioenergetic capacity [18]. Specifically, low sleep efficiency and short sleep duration have been related to indices of oxidative DNA damage [19], while short sleep duration was associated with systemic oxidative stress in a large sample of adults [20]. Although little research has examined the relationship of long sleep with bioenergetic function, long sleep duration is consistently linked with adverse health outcomes [21] and with elevated levels of circulating inflammatory cytokines [22], which are associated with reduced bioenergetic function in PBMCs [14]. Taken together, this research suggests that poor sleep efficiency and both short and long sleep duration may be related to impaired bioenergetics.

To our knowledge, there are currently no studies examining the relationship between sleep and bioenergetic function in humans, which would provide critical insight into a novel mitochondrial link between sleep, aging, and disease. The purpose of this study, one of the largest to date assessing bioenergetic function in humans [23], was to examine associations between indices of self-reported sleep (i.e., satisfaction, alertness, timing, efficiency, and duration) [24] and systemic bioenergetic function, measured in PBMCs of community-dwelling adults. We hypothesized that that lower sleep efficiency would be associated with lower bioenergetic function, and that sleep duration would demonstrate an inverted-U-shaped relationship with bioenergetic function, with both short and long sleep associated with lower bioenergetic function. Relationships of other components of sleep (satisfaction, daytime alertness, timing) with indices of bioenergetic function were explored.

2. Material & methods

A community sample was recruited via ads placed in a local newspaper, flyers placed in local establishments, and a website that provides information about studies available in the area. Potential participants who expressed interest (N = 111) were screened for eligibility and provided with information about the study via a phone call. Individuals were excluded from participating if they were employed as a night shift worker, were pregnant or nursing, or taking daily nonsteroidal anti-inflammatory medication. A total of 71 eligible individuals expressed continued interested in participating and were included as part of a larger project that has been described previously [25], while 43 participants (79.1% female) between the ages of 48 and 70 (M = 61.63, SD = 5.99) completed the study procedures included as part of this cross-sectional study. Cellular energetics were evaluated only among the subset of participants described here after a laboratory technician was hired and properly trained after the larger study had been initiated. There were no significant differences between present study participants and those from the larger project on any of the included demographic or sleep-related variables included in the present study (all p values > .12). Participants were not asked to fast prior to their appointment but were instructed to avoid consuming caffeine and certain high-fat foods (e.g., butter, bacon, mayonnaise, deep fried foods), and to avoid strenuous exercise during the morning of their study visit. A list of prohibited caffeinated substances and foods was provided to participants before their appointment. All visits were scheduled for 8am to control for potential diurnal variation. Written informed consent was obtained from all study participants prior to a blood draw and the completion of self-report questionnaires. The Pennsylvania State University Institutional Review Board approved the study procedures.

2.1. Measures

2.1.1. Sleep

Participants completed the Pittsburgh Sleep Quality Index (PSQI), a widely utilized 19-item questionnaire that assesses habitual sleep quality over the previous month [26]. In the present study, specific items from the PSQI were used to quantify key indices of sleep (i.e., satisfaction, alertness, timing, efficiency, and duration). Sleep satisfaction was assessed by a single item of overall sleep quality on a scale ranging from 0 (very good) to 3 (very bad). Alertness was operationalized as the degree to which participants had trouble staying awake while driving, eating meals, or engaging in social activities on a scale ranging from 0 (not during the past month) to 3 (three or more times per week). The midpoint between reported bedtime and wake time was used as an indicator of sleep timing. Sleep efficiency was calculated by dividing the reported sleep duration by the reported time spent in bed (based on reported time spent trying to fall asleep). Sleep duration was calculated as the number of reported hours per night typically spent sleeping during the past month.

2.1.2. Mitochondrial oxygen consumption

PBMCs were isolated using Histopaque gradient centrifugation. Briefly, 10mL of heparinized whole blood was added to a 50 ml falcon tube followed by RPMI (with phenol red; Sigma-Aldrich; St. Louis, MO) with Pen/Strep media up to 35 mL. The mixture was then added to a separate 50 ml falcon tube containing a total of 15ml Histopaque®-1077 (Sigma-Aldrich; St. Louis, MO), which was then spun at 400 × g for 30 minutes at room temperature with no acceleration/brake. Next, the PBMC layer was collected and placed into a 15 mL tube that was then filled with RPMI (phenol red free). The 15 mL tube was spun for 10 minutes at 1500 RPM (acceleration and brake = 9) and the cells were re-suspended in 4 mL of washing media.

Mitochondrial Stress Test Base media with glucose was prepared by adding 500μl of 100mM Sodium Pyruvate (1mM), 225μl of D-Glucose (4.5mM) and 500μl of 200mM Glutamine (2mM) to 50 mL of Seahorse XF Base Media (Seahorse Bioscience, Billerica, MA). The media was filtered using a 2μm filter and warmed to 37°C in a water bath. Using the contents of the Seahorse XFe24 Flux Assay Kit (Seahorse Bioscience, Billerica, MA), the utility plate was filled with 1 mL of Seahorse XF calibrant (Seahorse Bioscience, Billerica, MA). The Hydro Booster was placed on top of the Utility Plate and the Sensor Cartridge was lowered through the openings on the Hydro Booster plate, into the Utility Plate, submerging the sensors in XF calibrant. The cartridge was then incubated in a non-CO2 incubator at 37°C overnight.

On the day of each participant visit, a Seahorse XFe24 Cell Culture Microplate (Seahorse Bioscience, Billerica, MA) was pretreated by adding 50μl of Geltrex (Thermo-Fisher Scientific, Grand Island, NY) to each well and incubated at 37°C for between 45–120 minutes. Approximately 500,000 cells/mL were aliquoted from re-suspended cells and pipetted into three wells for triplicate testing. Each corresponding media was then added to wells containing re-suspended cells for a final volume of 500μl/well. The plate was then centrifuged at 600 × g (acceleration and deacceleration = 5) for 10 minutes followed by incubation at 37°C for between 45–60 minutes. Oxygen consumption rates were measured using standard manufacturer protocol (Seahorse Bioscience, Billerica, MA). Briefly, oxygen consumption rates were measured three times approximately eight minutes apart under each of four conditions: basal and after injections of oligomycin (4μM), carbonyl cyanide 4-phenylhydrzone (FCCP; 6μM), and Antimycin A/Rotenone (2μM). The resulting measurements allow for the calculation of the following indices of bioenergetic function: basal respiration, ATP-linked respiration, proton leak, maximal respiration, spare capacity (i.e., the difference between basal and maximal respiration), and non-mitochondrial respiration. A composite Bioenergetic Health Index (BHI) was calculated as BHI = log(ATP-linked × spare capacity)/(proton leak × non-mitochondrial), where higher scores indicate greater bioenergetic health [27].

2.2.3. Demographics

Participants self-reported their age and sex. Height and weight were measured by study staff to calculate body mass index (BMI). These variables were used as covariates in analyses.

2.3. Statistical analyses

Participants provided complete data for all study variables. Descriptive statistics were performed to determine the normality of distributions. Bivariate correlations between study variables were calculated. Separate regression models estimated associations of sleep efficiency and sleep duration with BHI and were adjusted for age, sex, and BMI. The model estimating BHI from sleep duration was first performed with a linear sleep duration term followed by the addition of a quadratic term (sleep duration-squared). All statistical analyses were performed using SPSS version 26 [28].

3. Results

Descriptive statistics for study variables are provided in Table 1. Proton leak was the only variable with a non-normal distribution and was log₁₀ transformed to improve distribution normality. Bivariate Pearson correlations are provided in Table 2. Lower sleep efficiency was associated with lower maximal respiration (r = .36, p = .02) and spare capacity (r = .38, p = .01), as well as lower BHI-assessed bioenergetic health (r = .57, p < .01). Longer sleep duration was associated with lower bioenergetic health as measured by the BHI (r = − .46, p < .01). To aid interpretation of results, a scatterplot depicting bivariate Pearson correlations of sleep efficiency and sleep duration with the BHI are illustrated in Figure 1. Later sleep timing was associated with higher basal respiration (r = .35, p = .02), ATP-linked respiration (r = .36, p = .02), maximal respiration (r = .33, p = .03), spare capacity (r = .31, p = .04), and non-mitochondrial respiration (r = .34, p = .03); however, sleep timing was not associated with the BHI (r = .04, p = .81).

Table 1.

Descriptive statistics

Variable	Mean (SD) or Number (%)
Sleep satisfaction	0.98 (0.52)
Daytime alertness	0.28 (0.63)
Sleep timing, hours	2.84 (1.11)
Sleep efficiency, %	85.80 (9.09)
Sleep duration, hours	8.21 (1.32)
Basal respiration1	59.89 (21.52
ATP-linked respiration1	54.65 (20.09)
Proton leak1	6.02 (8.86)
Maximal respiration1	247.86 (121.15)
Spare capacity1	187.32 (106.64)
Non-mitochondrial respiration1	29.25 (20.82)
Bioenergetic Health Index	4.77 (1.13)
Age, years	61.63 (5.99)
Sex
Female	34 (79.1)
Male	9 (20.9)
Body mass index	27.14 (5.48)

Note.

¹= oxygen consumption rate measured in pmol/minute.

Table 2.

Pearson correlations between study variables (N = 43)

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Sleep satisfaction	--
2. Daytime alertness	.12	--
3. Sleep timing	.07	− .12	--
4. Sleep efficiency	− .18	− .07	.31*	--
5. Sleep duration	− .05	− .38	.01	− .48*	--
6. Basal respiration	.19	− .06	.35*	.14	− .07	--
7. ATP-linked respiration	.04	.03	.36*	.24	− .14	.85*	--
8. Proton leak	.28	− .03	.05	− .23	.07	.31*	− .05	--
9. Max respiration	− .11	− .01	.33*	.36*	− .14	.75*	.79*	.08	--
10. Spare capacity	− .17	− .01	.31*	.38*	− .16	.66*	.72*	.02	.91*	--
11. Non-mitochondrial respiration	.08	− .08	.34*	.01	.16	.34*	.58*	− .41*	.45*	.44*	--
12. Bioenergetic Health Index	− .29	.01	.04	.57*	− .46*	.16	.26	− .40*	.36*	.38*	− .08	--
13. Age	.12	− .09	.08	− .19	.37*	− .13	.01	− .01	.05	.07	.20	− .23	--
14. Sex	− .02	− .14	− .06	− .21	− .07	.02	− .14	.11	− .15	− .16	− .10	− .18	− .20	--
15. Body mass index	− .08	.01	.30	− .09	.23	− .19	− .15	.05	− .17	− .16	− .02	− .17	.29	− .18

Note. Sex coded as 0 = male and 1 = female.

^*p < 0.05.

Figure 1.

Scatterplots illustrating linear associations of (A) sleep efficiency and (B) sleep duration with bioenergetic health among community-dwelling adults (N = 43).

Linear regression analyses demonstrated that lower sleep efficiency and longer sleep duration were associated with lower bioenergetic health (β = .52, p < .01 and β = − .43, p < .01, respectively) above and beyond participant age, sex, and BMI (Table 3). The quadratic sleep duration term was not associated with the BHI (β = − .10, p = .92) when added to the sleep duration model, suggesting that the association between sleep duration and bioenergetic function was linear rather than inverted-U-shaped as hypothesized.

Table 3.

Regression models predicting the Bioenergetic Health Index (N = 43)

	Sleep Efficiency (linear)				Sleep Duration (linear)				Sleep Duration (quadratic)
Predictor	β	B (95% CI)	SE	p	β	B (95% CI)	SE	p	β	B (95% CI)	SE	p
Constant		1.78 (− 3.88, 7.44)	2.79	.53		10.53 (6.50, 14.55)	1.99	< .01		10.18 (2.50, 17.86)	3.79	.01
Age	− .12	− .02 (− .08, .03)	.03	.40	− .09	− .02 (− .08, .04)	.03	.56	− .09	− .02 (− .08, .04)	.03	.57
Sex	− .12	− .32 (− 1.10, .46)	.39	.41	− .24	− .67 (− 1.46, .12)	.39	.09	− .24	− .67 (− 1.47, .14)	.40	.10
Body mass index	− .11	− .02 (− .08, .04)	.03	.44	− .09	− .02 (− .08, .04)	.03	.55	− .09	− .02 (− .08, .04)	.03	.56
Sleep efficiency	.52	.07 (.03, .10)	.02	< .01
Sleep duration					− .43	− .37 (− .63, − .11)	.13	<.01	− .33	− .29 (− 1.84, 1.27)	.77	.71
Sleep duration2									− .10	− .01 (− .10, .09)	.05	.92

Note. Sex coded as 0 = male and 1 = female.

4. Discussion

This study, one of the largest using blood-based bioenergetics to date [23], found that lower self-reported sleep efficiency and longer sleep duration were associated with lower systemic bioenergetic function in PBMCs of community-dwelling adults. In exploratory analyses, later sleep timing was related to elevated indices of bioenergetic function including higher basal respiration, ATP-linked respiration, maximal respiration, spare capacity, and non-mitochondrial respiration. These results suggest that multiple components of sleep are related to bioenergetic indices of mitochondrial function and may represent a novel cellular pathway linking sleep with aging, health, and disease.

Our finding of lower sleep efficiency associated with lower BHI, maximal respiration, and spare capacity is consistent with research in animal models that has linked poor sleep efficiency with lower bioenergetic function and with greater oxidative stress and inflammation. Experimental sleep disruption in Drosophila substantially diminished bioenergetic efficiency, including maximal respiration and spare capacity, in addition to increasing oxidative stress and reducing antioxidant activity [16]. Basal respiration and proton leak were not affected, suggesting that maximal respiration and spare capacity may be the bioenergetic parameters that are most sensitive to reduced sleep efficiency, which is consistent with our results. The association of sleep efficiency with maximal respiration was similar in magnitude to the previously observed association between maximal respiration and fatigue-related vitality (inverse) [29], but smaller than for the inflammatory cytokine interleukin-6 (IL-6; inverse), grip strength, and a composite measure of physical function (positive) [14]. In humans, lower actigraphy-assessed sleep efficiency was associated with reduced urinary 8-hydroxydeoxyguanosine, an indicator of elevated oxidative DNA damage [18], and with elevated circulating levels IL-6 [30], which is associated with reduced maximal respiration and spare capacity in PBMCs [14]. These data are consistent with our findings, but the observational nature of such results and the complex interrelationships between inflammation, oxidative stress, and bioenergetic function [31] limits speculation into causal pathways. Low levels of circulating melatonin may also play a role in the link between poor sleep efficiency and lower bioenergetic function. In addition to its role as a darkness-signaling and sleep-promoting hormone, melatonin accumulates in large quantities in mitochondria where it scavenges reactive oxygen species and upregulates antioxidant enzyme activity to maintain respiratory complex activities and the electron transport chain, thereby preserving bioenergetic function [32,33]. Experimental sleep studies that manipulate sleep efficiency and evaluate its impact on inflammation, oxidative stress, melatonin, and bioenergetics are needed to disentangle these putative mechanisms.

Our hypothesis regarding sleep duration and bioenergetic function was not supported, as we observed a negative linear association between sleep duration and bioenergetic function rather than an inverted-U-shaped relationship. A large literature has documented that both short (often defined as < 6 hours of sleep) and long sleep duration (i.e., >9 hours per night) are associated with many indices of morbidity [21] and mortality [34], but we found only evidence for an association of longer sleep duration with lower bioenergetic function. Our results may be explained by compelling evidence for a bidirectional relationship between sleep duration and oxidative stress: In Drosophila, experimentally lengthening sleep increased resistance to oxidative stress and enhanced antioxidant activity, but reducing neuronal oxidative stress resulted in shorter sleep duration [35]. Consistent with the free radical flux theory of sleep [36], the authors hypothesized that oxidative stress serves as a sleep-initiating signal, with sleep functioning to lower the oxidative burden. Given that the BHI is a sensitive measure of oxidative stress [37], the negative linear association between bioenergetic health and sleep duration in our study may reflect a scenario in which higher BHI scores indicate insufficient levels of oxidative stress to trigger adequate sleep duration, while lower BHI scores reflect sufficiently high levels of oxidative stress to initiate adequate sleep. A related theory posits that ATP released during neurotransmission stimulates the production of sleep-promoting inflammatory cytokines IL1β and TNFα [38], providing an inflammatory mechanism by which longer sleep duration may reflect increased brain energetic activity, which is correlated with peripheral bioenergetic function [10]. Experimental human research will be critical to investigate putative bidirectional relationships between sleep, oxidative stress, inflammation, and mitochondrial bioenergetics.

Sleep timing was the only other sleep component related to indices of bioenergetic function, as later sleep midpoint was associated with global energetic upregulation (i.e., higher basal, ATP-linked, maximal, and non-mitochondrial respiration, and higher spare capacity). Higher ATP-linked respiration and spare capacity are considered indicators of healthy bioenergetic function, but elevated basal and non-mitochondrial respiration can reflect mitochondrial dysfunction [26], complicating interpretation of results. Later habitual sleep timing, approximated by later chronotype, has been associated with metabolic complications including type 2 diabetes, obesity, and impaired glucose metabolism [39,40], but was not associated with oxidative stress compared to early chronotype [41]. Among individuals with normal chronotypes, skeletal muscle mitochondrial respiration peaked in the late evening and reached its nadir in the early afternoon [42], representing a reduction in bioenergetic activity from midnight throughout the morning. Individuals reporting later sleep timing (i.e., later chronotypes) may exhibit delayed peak energetic function, resulting in higher bioenergetic activity at a given time point in the morning, possibly explaining the association of later sleep timing with elevated bioenergetic function in our sample. However, these results were exploratory and should be interpreted with caution. Although participants were all tested at 8am, this clock time represents variable circadian time based on individual differences in endogenous circadian rhythms (e.g., melatonin), highlighting the need for inclusion of circadian rhythm indices (e.g., dim light melatonin onset). Given that habitual sleep timing is primarily driven by circadian rather than homeostatic sleep factors, the combined results of our study suggest that both circadian rhythms and sleep may be related to human bioenergetic function.

The findings of this study and their interpretations should be considered in light of several limitations which can be addressed by future research. First, the cross-sectional study design makes inferences about causal relationships merely speculative. Although we hypothesize a pathway from sleep to bioenergetics, the opposite is certainly plausible. For instance, in a Drosophila model of mitochondrial encephalomyopathies, a disorder involving stark bioenergetic dysfunction, flies exhibit disrupted sleep and circadian rhythms [43]. To garner evidence about causal relationships, laboratory studies should test whether bioenergetics are sensitive to acute sleep restriction and deprivation, and prospective studies can demonstrate whether changes in habitual sleep quality, duration, and timing predict changes in bioenergetic function over time. Second, the use of a single self-reported sleep questionnaire is not an ideal measure of habitual sleep characteristics. Self-reported sleep diaries and actigraphy should be incorporated into future sleep and bioenergetics studies and are well-suited for longitudinal research in large samples. Third, although the sample size in the present study was larger than most blood-based bioenergetics studies to date [23], our analyses were not powered to examine moderation by age, sex, or race/ethnicity, which are related to both sleep [44,45] and bioenergetics [46–48]. Therefore, larger and more diverse samples should be recruited to probe these potential interactions. Fourth, many lifestyle factors (e.g., alcohol consumption, smoking, diet, exercise) may impact mitochondrial function in adults [49]. Future studies may benefit from including such variables. Finally, we did not screen for sleep disorders (e.g., obstructive sleep apnea [OSA]) or chronotype, which may influence bioenergetic function. In particular, the intermittent hypoxia characteristic of OSA [50] has been associated with brain bioenergetic dysfunction [51]. Indices of OSA and/or hypoxia should therefore be assessed in future studies of sleep and bioenergetics.

In summary, this study showed that lower sleep efficiency and longer sleep duration were associated with worse bioenergetic health, while later sleep timing was associated with elevated overall bioenergetic activity. These findings demonstrate that multiple components of sleep are related to mitochondrial function in humans, and—given the ubiquitous role of mitochondrial decline in aging and disease—highlight sleep as a potentially modifiable determinant of bioenergetic health.

Highlights.

• The cellular mechanisms linking sleep with aging, health and disease are poorly understood.

• Animal research suggests that poor sleep impairs mitochondrial bioenergetic function, which plays an important role in aging and disease etiology and progression.

• This is the first study to examine the link between sleep and systemic bioenergetic function in humans and is among the largest human studies of blood-based bioenergetics to date.

• Lower sleep efficiency and longer sleep duration were associated with lower bioenergetic health.

• Bioenergetics may represent a potential target to probe mechanistic links between sleep, aging, health, and disease processes.

Data Availability Statement

The data that supports the findings of this study are available upon reasonable request to the corresponding author.