Psychosocial experiences are associated with human brain mitochondrial biology
Edited by Moses Chao, New York University Langone Medical Center, New York, NY; received October 11, 2023; accepted April 30, 2024
Significance
Psychosocial experiences predict health trajectories, but the underlying mechanism remains unclear. We report that positive psychosocial experiences are linked to greater abundance of the mitochondrial energy transformation machinery, whereas negative experiences are linked to lower abundance. Overall, psychosocial experiences accounted for 18 to 25% of the variance in protein abundance for complex I, the largest and most upstream mitochondrial oxidative phosphorylation (OxPhos) enzyme. At single-cell resolution, positive psychosocial experiences were particularly related to glial cell mitochondrial phenotypes. As a result, opposite associations between glial cells and neurons were naturally masked in bulk transcriptomic analyses. Our results suggest that mitochondrial recalibrations in specific brain cell types may represent a potential psychobiological pathway linking psychosocial experiences to human brain health.
Abstract
Psychosocial experiences affect brain health and aging trajectories, but the molecular pathways underlying these associations remain unclear. Normal brain function relies on energy transformation by mitochondria oxidative phosphorylation (OxPhos). Two main lines of evidence position mitochondria both as targets and drivers of psychosocial experiences. On the one hand, chronic stress exposure and mood states may alter multiple aspects of mitochondrial biology; on the other hand, functional variations in mitochondrial OxPhos capacity may alter social behavior, stress reactivity, and mood. But are psychosocial exposures and subjective experiences linked to mitochondrial biology in the human brain? By combining longitudinal antemortem assessments of psychosocial factors with postmortem brain (dorsolateral prefrontal cortex) proteomics in older adults, we find that higher well-being is linked to greater abundance of the mitochondrial OxPhos machinery, whereas higher negative mood is linked to lower OxPhos protein content. Combined, positive and negative psychosocial factors explained 18 to 25% of the variance in the abundance of OxPhos complex I, the primary biochemical entry point that energizes brain mitochondria. Moreover, interrogating mitochondrial psychobiological associations in specific neuronal and nonneuronal brain cells with single-nucleus RNA sequencing (RNA-seq) revealed strong cell-type-specific associations for positive psychosocial experiences and mitochondria in glia but opposite associations in neurons. As a result, these “mind-mitochondria” associations were masked in bulk RNA-seq, highlighting the likely underestimation of true psychobiological effect sizes in bulk brain tissues. Thus, self-reported psychosocial experiences are linked to human brain mitochondrial phenotypes.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Psychosocial experiences predict health outcomes and longevity through unresolved mechanisms that likely involve the brain (1). Positive psychosocial factors such as having a larger social network or greater purpose in life are associated with reduced risk of mortality and higher cognitive functioning at an older age (e.g., refs. 2–6). And negative psychosocial factors including social isolation, loneliness, and depression are associated with increased risk for cognitive impairments (e.g., refs. 7–9). However, the subcellular and molecular pathways that either produce these states or that transduce external psychosocial exposures into lifespan-altering cellular processes within the brain and body remain unclear. Two lines of research implicate mitochondria in the stress-disease cascade.
On the one hand, mitochondria are targets of stress and represent a potential pathway for the biological embedding of psychosocial exposures (10–14). The brain sustains its function by relying on energy (i.e., Adenosine triphosphate (ATP)) produced through oxidative phosphorylation (OxPhos) within mitochondria (15–19). Stressful experiences and chronic psychosocial stressors are believed to trigger neuroendocrine, molecular, and functional recalibrations that affect macroscopic brain properties [leading, for example, to hippocampal atrophy (20) and cortical thinning (21)], as well as subcellular and organellar structures, including mitochondria, that support numerous neuronal and glial functions (11, 22). In preclinical studies, acute and chronic psychological stressors affect mitochondrial content and functions across multiple tissues (e.g., refs. 23–25; for a review see ref. 10). In bulk tissue analyses of rodent brains, chronic stress affects the expression of mitochondrial genes involved in energy transformation (26, 27) as well as the abundance of proteins involved in mitochondrial energy metabolism, oxidative stress, and mitochondrial import and transport (27–29), likely signaling increased energy demand. In humans, chronic psychological stress and psychopathology are associated with molecular and functional alterations in specific aspects of mitochondrial biology measured in blood cells, including whole blood mitochondrial DNA copy number (mtDNAcn) (30, 31), enzymatic activities (32, 33), and cellular respiration (34–38), suggesting that psychosocial experiences affect mitochondrial biology across multiple tissues. These data implicate energetic and mitochondrial recalibrations as features of cellular responses to mental stress and possibly other psychosocial exposures.
On the other hand, mitochondrial biology is also a driver of behavior and affective experiences. In preclinical rodent models, mitochondrial DNA defects alter physiological stress reactivity (39–41) and influence mood-like behaviors (40, 42). Alteration in mitochondrial OxPhos in defined brain regions (nucleus accumbens, prefrontal cortex) influences social behavior as well as anxiety-like and depression-like behaviors [(27, 43–49), reviewed in ref. 50]. In syngenic animals (with the same genome), we also found that differences in mitochondrial energy production capacity in brain tissue accounted for 20 to 50% of animal-to-animal differences on anxiety-like behaviors and social avoidance (49). In human studies, psychopathology, autism, and/or neurodegeneration are also related to alterations in brain mitochondrial OxPhos components and mtDNA (e.g., refs. 27 and 51 see ref. 52 for a review). Thus, while chronic stress can affect brain mitochondrial biology, mitochondrial biology may also directly influence mood and behavior.
Existing human studies documenting a connection between daily psychosocial experiences and mitochondria have mostly been conducted in peripheral immune cells (32, 34–37). While these and other studies identified mitochondrial biology as a potential psychobiological pathway, findings in leukocytes are limited by high interindividual variation in cell type abundances, rapid cell turnover of mitotic cell (sub)populations, and dynamic immune processes that influence mitochondrial biology but say little about the “health” or systemic properties of mitochondria themselves (53). In comparison, the brain contains neurons and glial cells that exist in relatively stable proportions (compared to immune cells) across the lifespan. Moreover, mitochondrial metabolism specifically directs cellular activities including neuronal excitability and presynaptic neurotransmitter release (54), neurogenesis (55), as well as glial cell properties including astrocyte proliferation (56), microglia activation (57, 58), and oligodendrocyte differentiation (59). Glial cells may in fact contribute to brain sensing of peripheral metabolic and hormonal signals and be preferentially sensitive to systemic energy metabolism (60). For reasons that may relate to the preferential oxidation of glucose and unique metabolic requirements for OxPhos complex I in brain tissue, in human genetic diseases and animal models, the brain is particularly vulnerable to OxPhos complex I defects (61, 62). Neuronal developmental rates and cellular aging trajectories also are under mitochondrial regulation (63, 64), opening the possibility that mitochondrial alterations could influence brain aging and the risk of neurodegenerative diseases (65). This point is supported by well-documented alterations of mitochondria biology, particularly OxPhos and mtDNA, in brains of individuals with Alzheimer’s disease (AD) (66–69).
Given the central role of the brain in the processing of psychosocial exposures as well as in the production of affective experiences, it represents the most relevant tissue to investigate mitochondria psychobiological associations. In this study, we examine how prospectively collected positive and negative psychosocial factors relate to postmortem protein and transcript abundance of known mitochondrial pathways in hundreds of human brains, in bulk brain tissue, and in single cells (Fig. 1A).
Fig. 1.
Results
General Approach.
We used brain multiomics and psychosocial data from the ROS and MAP cohort studies (70, 71) (study design and cohort characteristics are shown in Fig. 1 A and B, SI Appendix, Fig. S1, and Table 1). This dataset includes longitudinal prospectively acquired psychosocial data, complemented by untargeted quantitative proteomics offering a broad coverage of mitochondrial proteins from the postmortem brain dorsolateral prefrontal cortex (DLPFC), a brain area involved in executive functions and emotional regulation (72), and sensitive to psychological stress exposure (27, 73–75).
Table 1.
| ROSMAP | ||
|---|---|---|
| TMT | Pseudobulk RNAseq | |
| Total n | 400 | 424 |
| Study, MAP | 252 (63%) | 210 (50%) |
| Study, ROS | 148 (37%) | 214 (50%) |
| Female, n (percent) | 282 (70%) | 288 (68%) |
| No cognitive impairment (NCI), n (percent) | 168 (42%) | 143 (35%) |
| Mild cognitive impairment (MCI), n (percent) | 101(25%) | 110 (27%) |
| Alzheimer’s dementia (AD), n (percent) | 123 (31%) | 157 (38%) |
| Age (years) at death, mean (SD) | 89.2 (6.5) | 89.2 (6.8) |
| Amyloid, mean (SD) | 4.7 (4.7) | 4.5 (4.5) |
| Tangles, mean (SD) | 4.8 (5.0) | 6.0 (7.2) |
| Postmortem interval in hours, mean (SD) | 7.8 (0.5) | 7.7 (5.1) |
| Positive score, n | 90 | 32 |
| Negative score, n | 172 | 99 |
We approached mitochondria as multifaceted organelles that are best studied as multiple functional domains (32, 76, 77), rather than by a single indicator. To capture interpretable biological signatures of mitochondrial health, we aggregated >1,100 mitochondrial genes as MitoPathways (78) (for study coverage, see Dataset S2). Because each mitochondrial pathway reflects the average of several functionally related genes, MitoPathways are both more biologically interpretable and also more statistically robust than single-gene or single-protein approaches. To probe the multifaceted mitochondria in an unbiased fashion, we initially assessed the seven broad MitoPathways scores from the MitoCarta 3.0 database (78), which include all facets of mitochondrial biology, subsequently refining our approach to more specific based on the most robust results.
We used a similar approach to integrate psychosocial experiences, defined here as the psychosocial exposures and psychological states in an individual’s outer and inner world. This includes i) positive factors: social network size, late-life social activity, purpose in life, time horizon, and well-being; and ii) negative factors: adverse childhood experiences, negative life events, social isolation, depressive symptoms, negative mood, and perceived stress. From these constructs, we built a positive and a negative score within which longitudinal variables were averaged across the follow-up period (average duration: 4.3 y ± 2.2 (SD), range 1 to 10 y; see SI Appendix, Fig. S2 for score composition and intercorrelations), providing more stable estimates of each person’s psychosocial and mental states prior to death, which can then be related to mitochondrial biology.
Psychosocial Experiences Are Associated with OxPhos Protein Abundance.
Mitochondrial proteomics and psychosocial data were available for 400 participants (70% women, 98% White Caucasian). At the time of death, 43% of individuals had normal cognition (NCI), 26% had mild cognitive impairment (MCI), and 31% had a diagnosis of Alzheimer’s disease (AD) as defined in refs. 79–81. To limit the dimensionality and complexity of our analyses, we use the seven primary MitoCarta 3.0 MitoPathways on the basis of previous literature and mechanistic upstream drivers of OxPhos function: “OxPhos,” “protein import, sorting, and homeostasis,” “mitochondrial dynamics and surveillance,” “signaling,” “metabolism,” “small molecule transport,” and “mitochondrial central dogma” (78). Correlation analyses of the protein-based MitoPathway scores with either positive or negative psychosocial scores showed that OxPhos consistently exhibited the strongest psychobiological associations (Fig. 1C).
Positive psychosocial factors were positively associated with “protein import, sorting, and homeostasis” (rho = 0.27; 95%CI = 0.07 to 0.45), “mitochondrial dynamics and surveillance” (rho = 0.23; 95%CI = 0.03 to 0.42), and “OxPhos” protein abundance (rho = 0.33; 95%CI = 0.13 to 0.50). On the other hand, negative psychosocial score was negatively associated only with “OxPhos” (rho = −0.27; 95%CI = −0.40 to −0.14), indicating that individuals reporting more negative experiences have lower abundance of mitochondrial OxPhos proteins in the DLPFC.
In stratified analyses, the associations with OxPhos were found across both men (positive: rho = 0.45; 95%CI = 0.10 to 0.70, negative: rho = −0.17; 95%CI = −0.41 to −0.17) and women (positive: rho = 0.28; 95%CI = 0.03 to 0.50, negative: rho = −0.30; 95%CI = −0.45 to −0.13). Similar psychosocial–OxPhos associations were observed both among individuals with some cognitive impairment (positive: rho = 0.47; 95%CI = 0.02 to 0.76, negative: rho = −0.30; 95%CI = −0.58 to 0.04) or without cognitive impairment (positive: rho = 0.34; 95%CI = 0.05 to 0.57, negative: rho = −0.39; 95%CI = −0.54 to −0.21) (SI Appendix, Fig. S2 and Dataset S3), supporting the generalizability of these findings linking positive and negative psychosocial factors to more and less DLPFC OxPhos protein content, respectively.
Psychobiological OxPhos Associations in BLSA and with Tandem Mass Tag (TMT) Proteomics.
In a much smaller study cohort with multiple brain areas, the Baltimore Longitudinal Study of Aging (BLSA) (n = 47, 36% female, 58% NCI, 42% AD; see SI Appendix, Fig. S3 and Dataset S2 for cohort characteristics), there were no significant associations between psychosocial factors and mitochondrial pathways in the middle frontal gyrus. However, in the precuneus area of the same participants, positive psychosocial experiences were positively associated with “mitochondrial dynamics and surveillance” protein abundance (rho = 0.47; 95%CI = 0.14 to 0.71; n = 15), while the reverse was found for negative psychosocial experiences (rho = −0.34; 95%CI = −0.56 to −0.07; n = 26) (SI Appendix, Fig. S4). Thus, while not a straightforward validation of our findings, the directionality of findings aligns with the one in the DLPFC found in ROSMAP. Together with other work showing mitochondrial variations across mouse brain areas (49), this calls for future studies assessing mitochondrial psychobiological associations across different brain regions.
In ROSMAP, we found no significant relationship between psychosocial scores and DLPFC mtDNAcn, mtDNA density (mtDNAcn/mitochondrial content score), or a general mitochondrial housekeeping score (Fig. 1D). Although people who reported more positive and negative experiences tended to have greater (rho = 0.20; 95%CI = −0.01 to 0.39) and lower (rho = −0.33; 95%CI = −0.33 to −0.05) total estimated mitochondrial mass, respectively (Fig. 1D), correcting OxPhos abundance for mitochondrial mass showed that the psychosocial experiences and OxPhos associations were not fully explained by differences in total mitochondrial content (Fig. 1E). These results thus largely rule out that psychosocial experiences would be only associated with unspecific changes in mitochondrial mass, for example driven by a general increase or decrease in energy demand. Instead, our results, taking into account both OxPhos proteins and mitochondrial content, suggest that psychosocial experiences are linked to differences in OxPhos proteins on a per-mitochondrion basis (i.e., a qualitative change in mitochondrial phenotype).
To explore this possibility with the highest possible degree of specificity, we examined more specific MitoPathways quantifying specific components of the OxPhos system, including complexes I through V. The strongest association with psychosocial exposures was for nDNA-encoded subunits belonging to complex I (positive: rho = 0.38; 95%CI = 0.19 to 0.55, negative: rho = −0.36; 95%CI = −0.48 to −0.23) (Fig. 1E and SI Appendix, Figs. S5–S7). OxPhos complex I is the major entry point for electrons into the electron transport chain (ETC), and is the OxPhos component whose abundance and impaired activity in the prefrontal cortex have been the most frequently associated with neurodegeneration and neurodevelopmental disorders (51, 52).
We then sought to confirm these data in the selected reaction monitoring proteomics (SRM, n = 1,208; see Dataset S1 for cohort characteristics) ROSMAP dataset, which is substantially larger but contains only a small subset of targeted mitochondrial proteins (16 OxPhos proteins total, vs. 128 for the TMT dataset). Computing scores as with the TMT data, the associations between psychosocial factors and complex I nDNA-encoded subunits showed the same direction, but the effects were only significant for negative psychosocial factors (rho = −0.24; 95%CI = −0.13 to −0.03) (SI Appendix, Fig. S8). This suggests two things: first, that our findings partially generalize in a larger, heterogenous sample; and second, in keeping with the multifaceted nature of mitochondria and multiprotein nature of OxPhos complexes (49), that the mitochondrial indices built from comprehensive sets of proteins covering >90% of OxPhos proteins (in the TMT DLPFC dataset, Fig. 1) might be required to sensitively detect mitochondrial psychobiological associations.
Domain-Specific OxPhos Psychobiological Associations.
Next, we dissected the separate contributions of each instrument that make up the positive and negative psychosocial scores to see which contributed the most to the psychobiological associations with OxPhos protein abundance. For the positive psychosocial score, the significant measures contributing the most to the OxPhos association were well-being (rho = 0.31; 95%CI = 0.11 to 0.48) and late-life social activity (rho = 0.20; 95%CI = 0.06 to 0.32). Well-being was assessed according to six major themes (purpose in life, personal growth, positive relations with others, self-acceptance, autonomy environmental mastery) (82, 83). Late-life social activity captures how often participants engaged in activities involving social interaction (84). For the negative score, negative mood (rho = −0.24; 95%CI = −0.39 to −0.08) and negative life events (rho = −0.19; 95%CI = −0.34 to −0.03) had the strongest effect sizes (SI Appendix, Fig. S9, stratified analysis are shown in SI Appendix, Fig. S10 and Dataset S3). Thus, both individual experiences (well-being and mood) and objectifiable factors (social activity and life events) relate to DLPFC brain mitochondrial biology.
To quantify the relative and combined contributions of psychosocial experiences on mitochondrial proteins, we performed multiple linear regressions controlling for relevant covariates (holding constant the effects of these variables) including cognitive status, as AD has been associated with altered brain mitochondria biology (66–69). Positive and negative psychosocial factors explained 11% (P = 0.0015) and 12% (P < 0.0001) of the variance in nDNA-encoded subunits of complex I abundance, respectively (SI Appendix, Fig. S11). Since individuals with more positive experiences may also report less negative experiences, we created a combined score taking into account their shared variance. Combined, the psychosocial factors assessed in this study explained 18% of the variance in nDNA-encoded complex I subunit abundance (r2 = 0.18, P < 0.0001) (Fig. 2). After adjusting for cell type abundance using TMT-derived indices of excitatory and inhibitory neurons, astrocytes, oligodendrocytes, pericytes, and endothelial cells, the proportion of explained variance was 16% (r2 = 0.16, P = 0.0004, SI Appendix, Fig. S11D), indicating that even after taking into account the proportion of neurons and glial cells in the brain of each participant, complex I abundance remains significantly related to psychosocial experiences. In stratified analysis where we included only participants with NCI, the model fit increased to 25% (from 18%) of variance in DLPFC complex I abundance attributable to psychosocial factors (r2 = 0.25, P = 0.001, SI Appendix, Fig. S11D). Thus, these results suggest a potential cross-talk between psychosocial experiences and brain mitochondrial OxPhos capacity.
Fig. 2.
Mitochondrial Psychobiological Associations Are Cell Type Specific.
To examine whether the proteomic findings above (reflecting the OxPhos machinery) were also observed at the level of gene expression (largely reflecting the active cellular programs rather than the machinery), we repeated the same analytical approach with bulk RNA sequencing (RNA-seq) data. We analyzed three different brain areas: DLPFC (N = 1,092), posterior cingulate cortex (PCC) (N = 661), and anterior caudate (AC) (N = 731) from the ROSMAP cohort (see Dataset S1 for cohort characteristics). Unlike with proteomics, there were no consistent associations between positive nor negative experiences and RNA transcript-based mitochondrial pathways (SI Appendix, Figs. S12 and S13). This suggested that either the mitochondrial psychobiological associations observed in the tissue proteome are limited to posttranscriptional factors (e.g., protein synthesis, stability, or other factors), or that the bulk transcriptome data are too noisy and do not provide the level of sensitivity or specificity required to detect true associations between psychosocial experiences and mitochondrial features.
Knowing that different cell types contain vastly different mitochondrial phenotypes (49, 53, 85), we therefore quantified mitochondrial phenotypes in specific brain cell types using single-nucleus RNA-seq of ROSMAP DLPFC samples (n = 424, 68% women; 35% NCI, 27% MCI 38% AD, cohort characteristics are shown in Table 1). Of the 87 distinct cell types and subtypes previously identified (86), we first examined mitochondrial psychobiological associations with transcript levels in cell types where the same MitoPathways as above were sufficiently well represented (>50% of participants and >50% coverage of mitochondrial genes). Based on these criteria, we included the following six major cell types: excitatory and inhibitory neurons, microglia, astrocytes, oligodendrocytes, and oligodendrocyte progenitor cells (OPCs) (Fig. 3A).
Fig. 3.
At this level of cellular resolution, we observed remarkable divergence in the magnitude and direction of the correlations between positive and negative experiences and MitoPathways DLPFC gene expression (Fig. 3B and Dataset S4). Averaging all MitoPathways together to decrease the false discovery rate, we found that individuals reporting more positive experiences had higher mitochondrial gene expression in both oligodendrocytes (rhoavg = 0.19) and OPCs (rhoavg = 0.18). In contrast, positive experiences were negatively correlated with neuronal MitoPathways, both in excitatory (rhoavg = −0.07) and inhibitory neurons (rhoavg = −0.16).
These divergent neuron vs. glial cell patterns were particularly apparent when examining the well-being scale alone. From the lowest to highest associations, we find neurons rhoavg = 0.02 (range = −0.10 to 0.21), astrocytes rhoavg = 0.23 (−0.03 to 0.34), microglia rhoavg = 0.24 (0.06 to 0.43), oligodendrocytes (0.08 to 0.41), and OPCs rhoavg = 0.28 (−0.22 to 0.45). Fig. 3C presents the contrasts in standardized effect sizes between positive and negative experiences and all MitoPathways, for each major cell type. For this analysis, the null hypothesis is that both positive and negative exposures are not associated with MitoPathway expression (i.e., r = 0.00) and that positive and negative scores or scales yield similar random distributions (i.e., effect size between positive and negative experiences is null (Cohen’s d = 0). In contrast, glial cells exhibited large contrasting effect sizes, indicating that psychosocial experiences are consistently associated in the same direction with multiple DLPFC single-cell MitoPathways, particularly among oligodendrocytes (Cohen’s d = 1.95) and OPCs (d = 2.25), but not in neurons (d < 1.13) (Fig. 3D).
We then expanded these analyses of mitochondrial psychobiological associations at the highest possible level of resolution across 23 cell subtypes having sufficient coverage (>50% of participants, >50% coverage of mitochondrial genes). Although underpowered for these analyses, as for our results among the six major cell types, these data also highlighted how individuals who report higher positive experiences tended to have greater MitoPathway gene expression in glial cells, but not in neurons (Fig. 3 E–G and Dataset S4). Comparing nuclear vs. mtDNA-encoded gene associations also suggested that mtDNA-encoded OxPhos subunits showed a positive association with psychosocial experiences across cell types (positive score: rhoavg = 0.23), while this was not the case for nDNA-encoded OxPhos subunits (positive score: rhoavg = –0.03) (Fig. 3E), possibly reflecting differential regulation of nuclear and mitochondrial genomes.
Together, these single cell-level data provide two main insights. First, they suggest that mitochondrial psychobiological associations in the mitochondrial proteome may represent an underestimate of the true effect sizes. Second, they suggest that psychosocial factors have differential effects on mitochondrial biology in specific brain cells, opening unresolved questions around the role of glial mitochondrial biology in affective experiences and behaviors.
Discussion
In this study, we used longitudinal psychosocial data and postmortem DLPFC proteomics in a sample of older adults with and without cognitive impairments to evaluate the association between psychosocial experiences and brain mitochondrial biology. Individuals reporting more positive experiences, such as greater well-being, had greater brain tissue OxPhos complex I protein abundance, while the opposite effect was found for negative psychosocial experiences. Considering their independent contributions (people feeling more positive may report fewer negative experiences), we find that ~18 to 25% of the variance in complex I abundance between individuals was attributable to self-reported psychosocial experiences. This fraction of explained variance is substantial given the measurement errors inherent to self-reported questionnaires as well as to postmortem brain omic measurements. Moreover, extending these results using single nucleus transcriptomics results showed that the connection between psychosocial factors and mitochondrial biology may be either predominantly, or perhaps restricted to, nonneuronal glial cells. This suggests that the proportion of explainable variance in complex I proteins noted above is likely an underestimate, since it is derived from bulk brain tissue where multiple cell types are aggregated. Combined, our direct measurements of brain proteins (the active agents supporting electron transport chain function) together with our single-cell resolution transcriptomics data (showing large cell-type-specific associations) provide evidence linking psychosocial experiences and mitochondrial biology in the aging human brain.
Our findings linking psychosocial experiences to brain mitochondria agree with a vast body of animal literature documenting an effect of chronic stress exposure on various aspects of mitochondrial OxPhos. In animals, chronic stress leads to reduced protein expression specifically of complex I (87), complex I enzymatic activity (25, 88–91), and complex I–driven respiration (24). Thus, in line with our human brain findings, chronic stress may preferentially affect brain mitochondrial complex I biology for reasons that remain unclear.
Although animal studies document a causal effect of stress exposure on mitochondrial biology, our findings could reflect the reverse association. DLPFC mitochondrial bioenergetics could contribute to shaping affective experiences, such as one’s state of well-being reported on questionnaires. This interpretation is supported by a growing body of work in rodents and humans indicating that mitochondrial OxPhos defects can alter behavior (e.g., refs. 24, 45, and 48). Social dominance, depressive, and anxiety-related behaviors have been associated with variation in brain mitochondrial OxPhos capacity (43–46, 49). In the human brain, postmortem studies show reduced complex I activity in relation to depression (see ref. 52 for a review), and individuals with mitochondrial DNA defects leading to reduced OxPhos capacity exhibit a higher incidence of depressive symptoms compared to the general population (92). Using noninvasive proton magnetic resonance spectroscopy imaging in humans, a study also found that trait anxiety is associated with downregulated nucleus accumbens taurine levels (93), a metabolite essential for complex I activity (94, 95). Finally, complex I occupancy quantified by positron emission tomography (i.e., a proxy for protein abundance) was lower in the anterior circulate cortex of adults with autism spectrum disorder than controls, and lower abundance was correlated with more severe social communication deficits (51), thus suggesting that variation in mitochondrial energy production capacity in specific brain regions may influence social behaviors, and possibly other psychosocial experiences.
The brain, and in particular the DLPFC area, is directly relevant to the embedding of psychosocial experiences as it is involved in executive functions and emotional regulation. Sustained cognitive demand directly affects brain DLPFC metabolism characterized by glutamate accumulation (96), which can have neurotoxic effects when chronically occurring such as when being exposed to a chronically stressful condition. Adversity and chronic stress (73), which overlap with the negative psychosocial constructs assessed in our study, lead to both structural and functional DLPFC changes such as reduction of grey matter volume (74) and/or impairment of brain functional connectivity involved in attention shifts (75). Our findings open the exciting possibility that alterations of mitochondrial OxPhos capacity could mediate the effects of social factors on brain energetics and therefore affect downstream brain and cognitive functions. Future studies should determine whether psychosocial factors shape cognitive trajectories through alterations of brain mitochondrial OxPhos capacity or through other related mitochondrial functions or behaviors (76).
In relation to mitochondria in other tissues, our findings align and contrast with the mixed literature in immune cells. Consistent with the current brain study, we previously found that positive mood predicted higher OxPhos enzymatic activities in PBMCs on subsequent days (12 to 48 h), but mitochondrial activities did not predict mood on subsequent days, suggesting mood-to-mitochondria directionality for these effects (97). Interestingly, the observed effect size in PBMCs was r2 = 0.13 to 0.16, comparable to r2 = 0.18 for the current results in the brain. Other studies have also found mitochondrial alterations to be a potential mechanism for the biological embedding of early adversity (98, 99). In contrast, a study found that childhood maltreatment was associated with higher mitochondrial respiration in live PBMCs (37), possibly as a result of increased proinflammatory activity, since activated cells producing high cytokine levels must consume more energy. Similarly, another study found that women experiencing greater stress quantified by allostatic load gave birth to children with greater mitochondrial content at age 3 to 6 y (100), possibly reflecting the activation of energy-costly proinflammatory programs among immune cells (101, 102). Some of the divergences between these results in immune cells and our brain data could reflect the different cell types used, and/or more generally the nature of immune vs. brain cells.
An important caveat of our study is the unresolvable directionality of effects. We can conceive of four biologically plausible scenarios linking psychosocial and mitochondrial outcomes. First, it is plausible that psychosocial experiences affect brain activity and therefore directly shape mitochondrial biology. Second, as discussed above, one’s mitochondrial biology could affect behavior and perception of self-reported psychosocial experiences on questionnaires. Third, a bidirectional relationship may exist, such that chronic stress exposure directly affects an individual’s mitochondrial biology and subsequently affects their perception of social events. And fourth, other factors such as environmental toxicant could affect both mitochondria and psychosocial experiences through independent mechanisms. However, the emerging picture in the literature is that all those pathways are interactive, and thus, our results may reflect the outcome of those complex interactions.
While our study design is strengthened by integrated indices from prospectively acquired psychosocial data, as well as postmortem proteomics and single nucleus gene expression data in a large sample of individuals, it also presents several technical limitations. Proteomics lacks sensitivity to detect proteins present at very low abundance (including some mitochondrial proteins) and highly hydrophobic mitochondrial proteins embedded within lipid membranes (103). Because the enzymatic activities of proteins are posttranslationally regulated by several factors, protein abundance also may not directly translate into mitochondrial functions and behaviors (76). Future studies should complement static, omic-based measures of molecular mitochondrial features with direct measures of mitochondrial activities and functions. Postmortem gene expression may also only poorly reflect gene expression in the living human brain, calling for studies of living samples where possible (104), although achieving the scale required to address psychobiological hypotheses may be challenging. Other limitations include the lack of psychosocial measures proximal to death in some participants with AD (not able to provide questionnaire-based data), which we expect to contribute noise to our analyses, likely resulting in underestimated psychobiological associations. Finally, the use of self-reported measures of psychosocial experience is not without limitations (e.g., high variability, desirability bias), but the prospective nature of the assessments in this study likely prevented recall bias.
In summary, the present study links individual psychosocial experiences (antemortem) with human brain mitochondrial biology (postmortem). Our results document a robust psychobiological association (18 to 25% of shared variance) between psychosocial experiences and brain DLPFC mitochondrial OxPhos complex I. Moreover, single-nucleus studies of psychosocial–mitochondrial associations suggest that glial and neuronal cell subtypes may either respond to or contribute to human psychosocial experiences in opposite directions. Cell type specificity therefore represents an important concept to integrate into our emerging understanding of the processes linking psychosocial factors, brain energetics, and mitochondria. While the directionality of these effects and the underlying mechanisms remain to be examined in detail, our results are consistent with the notion that recalibrations in mitochondrial energetics may transduce the effects of psychosocial exposures into molecular and biological changes that shape brain health and aging.
Materials and Methods
Study Participants.
We used data from ROSMAP (N = 400) (70, 71, 105) and BLSA (N = 47) (106–109). These studies have measured postmortem DLPFC protein abundance using untargeted proteomics. Clinical characteristics of the study participants included in the analysis are shown in Table 1 and Dataset S1.
ROSMAP.
We repurposed postmortem brain proteomics and RNAseq data from two cohort studies of adults, the Rush Memory and Ageing Project (MAP), and the Religious Orders Study (ROS) (70, 110, 111). ROS participants are catholic nuns, priests, and brothers, from about 40 groups across the United States. MAP study participants were enrolled primarily from about 40 retirement communities throughout northeastern Illinois, with additional participants enrolled through home visits. Participants were free of dementia at study enrollment and agreed to annual evaluations and brain donation at death. The ROS and MAP studies were approved by the Institutional Review Board of Rush University Medical Center. All participants signed informed consent, Anatomical Gift Act, and a repository consent to share data and biospecimens.
Alzheimer’s dementia (AD) clinical diagnosis proximate to death was based on a review of selected annual clinical data after death by the study neurologist who did not have access to neuropathologic data. Cognitive status was defined as no cognitive impairment (NCI), MCI, or Alzheimer’s dementia (AD) from the final clinical diagnosis of dementia, as reported previously (79–81). Postmortem AD pathology was measured as described previously (79, 112), and AD classification was defined based on the National Institutes of Ageing-Reagan criteria (113). The NIA-Reagan criteria for the pathologic diagnosis of AD (high or intermediate likelihood of AD) (79) and final clinical diagnosis of cognitive status (AD and no other cause of cognitive impairment) were used to define pathologic AD as described previously (114). MCI refers to participants with cognitive impairment but who did not meet criteria for dementia.
BLSA.
The National Institute on Aging’s BLSA was used as an additional study cohort. TMT proteomic data from DLPFC brain tissue were available for 13 cognitively healthy individuals and 34 AD cases (36% women). The BLSA study was approved by the Institutional Review Board and the National Institute on Aging. Human research at the NIH and the BLSA participants provided written informed consent (106).
Postmortem neuropathological evaluation was conducted at the Johns Hopkins Alzheimer’s Disease Research Center with the Uniform Data Set. Amyloid plaque distribution was assessed according to the CERAD criteria and neurofibrillary tangle pathology was assessed with the Braak staging. Participants were classified as cognitively healthy if they were evaluated as cognitively healthy within 9 mo of their death and presented low CERAD (0.13 ± 0.35) and Braak (2.26 ± 0.94) measures of amyloid and tau neuropathology. Participants were categorized as AD cases when classified as demented at the last clinical research assessment and when the brains showed high CERAD (2.9 ± 0.31) and Braak (5.4 ± 0.82) scores (consistent with moderate to severe neuropathological burden). Asymptomatic AD cases were participants evaluated as cognitively normal proximate to death, with brains presenting high CERAD (2.1 ± 0.52) and moderate Braak (3.6 ± 0.99).
Psychosocial Variables.
ROSMAP.
Psychosocial variables were either collected once or annually. After participants were diagnosed with dementia, psychosocial assessments were no longer included in the yearly assessments. Thus, the psychosocial assessments for individuals with MCI and AD may be less proximal to death and therefore less likely to be related to postmortem biological parameters. Positive psychosocial exposures measured annually included social network size (115), late-life social activity (84), purpose in life (2), well-being (82, 83), and time horizon. Time horizon was measured using 6 items (“1. Many opportunities await me in the future.”, “2. I expect that I will set new goals in the future.”, “3. I have the sense that time is running out.”, “4. I can do anything I want in the future.”, “5. There are limited possibilities in my future.”, “6. As I get older, I experience time as more limited.”) for which participants were asked to rate how each item applies to themselves using a 7-point Likert rating scale [Strongly agree (1), Agree (2), slightly agree (3), Neither agree nor disagree (4), Slightly disagree (5), Disagree (6), Strongly disagree (7)]. Items (1,2 and 4) that are positively worded are flipped so that higher ratings on all items indicate a longer perceived time horizon. The total score is the mean of the item ratings, with a higher score indicating a longer perceived time horizon. Negative psychosocial exposures measured annually included negative life events (116), perceived social isolation (8), depressive symptoms (117), negative mood (118), and perceived stress (119). Adverse childhood experiences (120) were measured once. Procedure to request data and detailed information on the questionnaires can be found on the Rush Alzheimer’s Disease Center (RADC) Research Resource Sharing Hub website: https://www.radc.rush.edu/.
ROSMAP psychosocial variables summary scores.
Each questionnaire score was transformed into a z-score (x-mean/SD) before being transformed into a t-score with an average of 100 and a SD of 10 ((x*10)+100). Before computing the summary scores, the longitudinal variables were averaged across the follow-up. An index of positive exposures and an index of negative exposures were computed by averaging the t-scores for the positive and negative questionnaire scores respectively. The positive and negative scores included at least one assessment for each questionnaire. To create a global score that includes overall psychosocial experiences, positive variables were reversed and used to compute a combined negative score (social network size variable was excluded).
BLSA.
For the BLSA cohort, summary scores of positive and negative psychosocial exposures were derived from the Activities and Attitudes Questionnaire (121). Detailed descriptions of the items included and calculations are provided in SI Appendix, Doc 1. Depressive symptoms were assessed using the Center for Epidemiologic Studies Depression Scale (CESD) (122). Additionally, depressive and anxiety symptoms were also assessed using the Cornell Medical Index (123). For all the variables, the average of the values across the follow-up were used.
Brain Mitochondrial Indices.
ROSMAP proteomics.
TMT isobaric labeling mass spectrometry.
Postmortem DLPFC (Brodmann area 9) proteomics data were measured with TMT isobaric labeling mass spectrometry (71) and were available for 400 (70% women) individuals including 168 cognitively normal individuals, 101 MCI, 123 AD, and 8 had other causes of dementia. A total of 12,691 unique proteins were detected; 7,901 were kept after quality control.
Briefly, protein digestion was performed after tissue homogenization as previously described in ref. 124. An equal amount of protein from each sample was aliquoted and digested in parallel to be used as the global pooled internal standard (GIS) in the TMT batches. Before TMT labeling, sample randomization was performed by covariates (age, sex, PMI, diagnosis, etc.) into 50 batches (8 cases per batch). The TMT 10-plex kit (ThermoFisher 90406) was used to label the samples (N = 400) and the pooled GIS standards (N = 100) as previously described (124, 125). In each batch, the 8 middle TMT channels were used to label individual samples and the TMT channels 126 and 131 were used to label GIS standards. High-pH fractionation was performed on an Agilent 1100 HPLC system as previously described (71, 126). Peptide eluents separation was done on a self-packed C18 (1.9 μm, Maisch, Germany) fused silica column [25 cm × 75 μM internal diameter (ID); New Objective, Woburn, MA] by a Dionex UltiMate 3000 RSLCnano liquid chromatography system (ThermoFisher Scientific) and monitored using an Orbitrap Fusion mass spectrometer (ThermoFisher Scientific). Proteome Discoverer software (version 2.3, ThermoFisher Scientific) was used to analyze the RAW files. The protein abundances were log2 transformed and regressed for batch, study, age at death, and postmortem interval, prior to downstream analyses. The data are available through the synapse.org AMP-AD Knowledge Portal (www.synapse.org; SynapseID: syn17015098) (127).
SRM quantitative proteomics.
A small set of targeted proteins (N = 119) were measured using SRM quantitative proteomics as described previously (128) in postmortem DLPFC brain tissue of 381 cognitively normal individuals, 286 MCI, 519 AD, and 22 with other causes of dementia (68% women). The data are available through the synapse.org AMP-AD Knowledge Portal (www.synapse.org; SynapseID: syn10468856) (129).
ROSMAP RNA-seq.
RNA-seq was performed from DLPFC (N = 1,102), PCC (N = 661) AC (N = 731) tissues as previously described (130, 131). Briefly, samples were extracted using Qiagen’s miRNeasy mini kit (cat. no. 217004) and the RNase-free DNase Set (cat. no. 79254). Quantification by Nanodrop and quality evaluation was performed by Agilent Bioanalyzer. The Illumina HiSeq with 101-bp paired-end reads was used for sequencing. To pass quality control, the mean coverage for samples was set at 95 million reads (median 90 million reads). Data included in the analysis met quality control criteria. A normalized expression level was computed for each gene by subtracting the mean expression across all samples and dividing by the SD. Before downstream analysis, the expression levels were regressed for batch, library size, percentage of coding bases, percentage of aligned reads, percentage of ribosomal bases, percentage of UTR base, median 5 prime to 3 prime base, median CV coverage, study (ROS or MAP), and postmortem interval (PMI). The data are available through the synapse.org AMP-AD Knowledge Portal (www.synapse.org; SynapseID: syn3388564) (132).
ROSMAP single-nucleus pseudo-bulk RNAseq data.
Single-nucleus RNAseq was performed from frozen DLPFC specimens (N = 424) as described previously (86, 133). Gray matter was extracted and dissociated into nuclei suspension. The Chromium Single Cell 3’ Reagent Kits version 3 (10x Genomics) were used to construct the single-nucleus RNA-seq libraries following the manufacturer’s protocol. The sample single-nucleus RNA-seq libraries were then sequenced using HiSeqX and NovaSeq sequencers (Illumina). CellRanger (v6.0.0; 10x Genomics) with the “GRCh38-2020-A” transcriptome and the “include-introns” option were used to process FASTQ files. CellBender software was used for Cell calling and ambient RNA removal.
Cell clustering was performed based on existing cell type annotations (134) as described previously (133). A stepwise clustering approach was used to first identify 7 major cell types which were then subdivided into 92 cell subtypes. Pseudo-bulk UMI count matrices were constructed for each cell type and cell subtype by extracting and aggregating UMI counts of the cell (sub)type of interest for each participant, and normalizing them by sequencing depth. Pseudo-bulk UMI counts normalization was done by using the trimmed mean of M-values (TMM) method of edgeR, and log2 of counts per million174 mapped reads (CPM) were calculated using the voom function of limma (version 3.44.3). Low expression genes (log2CPM<2.0) were filtered out. Expression levels were quantile normalized after batch effects correction using ComBat. We selected cell types detected for >50% of participants and which had coverage for >50% of mitochondrial genes to compute the MitoPathways scores. Values were regressed for batch, PMI, age at death, study, sex, and cognitive status prior analysis. The raw and pseudo-bulk data are available through the AD Knowledge Portal (https://www.synapse.org/#!Synapse:syn31512863) (135).
ROSMAP WGS mtDNAcn.
WGS libraries were prepared using the KAPA Hyper Library Preparation Kit in accordance with the manufacturer’s instructions. Briefly, 650 ng of DNA from DLPFC tissue was sheared using a Covaris LE220 sonicator (adaptive focused acoustics). After bead-based size selection, selected DNA fragments were end-repaired, adenylated, and ligated to Illumina sequencing adapters. To evaluate final libraries, fluorescent-based assays were used including qPCR with the Universal KAPA Library Quantification Kit and Fragment Analyzer (Advanced Analytics) or BioAnalyzer (Agilent 2100). Then, libraries were sequenced on an Illumina HiSeq X sequencer (v2.5 chemistry) using 2 x 150bp cycles. R/Bioconductor (packages GenomicAlignments and GenomicRanges) was used to calculate the median sequence coverages of the autosomal chromosomes and of the mitochondrial genome. Ambiguous regions were excluded using the intra-contig ambiguity mask from the BSgenome package. The mtDNAcn was calculated as (covmt/covnuc) × 2. Before downstream analysis, mtDNAcn was z-standardized and logarithmized as described previously (66). The data are available through the synapse.org AMP-AD Knowledge Portal (www.synapse.org; SynapseID: syn25618990) (136).
BLSA Proteomics.
TMT proteomics was measured on postmortem samples from 47 participants from two brain regions (middle frontal gyrus and precuneus). Detailed description of the method is provided in SI Appendix, Doc 2. Values were regressed for age at death, PMI, sex, and AD status before analysis.
Mitochondrial Pathways’ Indices Calculation.
Each datapoint (protein or transcript level) was transformed into a z-score ((x-mean)/SD) before being transformed into a t-score with an average of 100 and a SD of 10 ((x*10)+100. Then for each score, t-scores (protein or transcript levels) were averaged to build a score representing the average abundance for each mitochondrial pathway (MitoPathways) category. The MitoPathways were defined using MitoCarta 3.0. mitochondrial gene annotations (78) (for study coverage see Dataset S2). An estimated mitochondrial content index was calculated using the average protein expression of all mitochondrial genes referenced in MitoCarta 3.0. OxPhos enrichment scores were computed by calculating the ratio of each OxPhos category divided by the mitochondrial content score. Mitochondrial DNA density was computed using the ratio of mtDNA copies per cell (mtDNAcn derived from WGS) relative to mitochondrial content per cell (mito content index) as described previously (137). To ensure our findings were not confounded by variation of cellular content, we investigated the relationship between a set of housekeeping proteins (derived from ref. 138) and psychosocial scores and found no evidence of an association (Fig. 1D). Finally, since both OxPhos and mitochondrial mass were associated with psychosocial scores (Fig. 1D), the ratio of OxPhos protein levels to mitochondrial content score was computed to obtain a score representing specific mitochondrial OxPhos enrichment. To further increase the specificity of our analyses, we computed separate scores for subunits encoded by nuclear DNA (nDNA) and mtDNA. To add proportion of different cell types as covariate in our regression model, TMT-derived cell type estimates were calculated using average protein expression of the following genes: All neurons (RBFOX3), Astrocytes (AQP4, S100B, SLC1A3, ALDH1L1), Endothelial cells (CLDN5), GABAergic inhibitory neurons (GAD1, GAD2, SST, RELN), Glutamatergic neurons (SLC17A6, SLC17A7, SLC17A8), Microglia (AIF1, TMEM119), Oligodendrocytes (MOBP, MOG, MAG, PLP1), Pericytes (PDGFRB, RGS5). As reported in the results, including cell type proportions as covariates had a moderate influence on the magnitude of the effect sizes for the association between psychosocial experiences variance in nDNA-encoded complex I subunit abundance, but did not alter the direction or statistical significance of the results.
Statistical Analyses.
Protein and transcript levels were regressed for technical variables (see above) and postmortal interval (PMI) prior to analysis. The associations between psychosocial scores and mitochondrial biology indices were assessed using Spearman’s rho correlation. Multivariate linear regressions were used to measure the relationship between psychosocial scores and respiratory complex I nDNA-encoded protein abundance scores adjusted for mitochondrial content while controlling for the effect of sex and cognitive status. We used principal component analysis to visualize major cell type clustering after excluding genes with low expression across the entire dataset. Statistical analyses were performed with Prism 9 (GraphPad, CA) and RStudio version 1.4, and R version 4.0.4. Statistical significance was set at P < 0.05.
Data, Materials, and Software Availability
TMT isobaric labeling mass spectrometry (ROSMAP) data have been deposited in www.synapse.org (syn17015098) (127). Anonymized [TMT isobaric labeling mass spectrometry (BLSA)] data have been deposited in www.synapse.org (syn39213792) (139).
Acknowledgments
We acknowledge participants of the ROS and MAP and BLSA studies. Support for this work was provided by NIH grants NHI RF1AG057473 (D.A.B. and P.L.D.J.), U01AG061356 (D.A.B. and P.L.D.J.), P30AG16101 (D.A.B.), R01AG15819 (D.A.B.), R01AG070438 (P.L.D.J.), R01AG17917 (D.A.B.), U01AG46152 (D.A.B. and P.L.D.J.), U01AG61356 (D.A.B. and P.L.D.J.), R01AG036836 (P.L.D.J.), U01AG061357 (N.T.S.), R01AG061800 (N.T.S.), RF1AG062181 (N.T.S.), RF1AG076821 (M.P.), GM119793 (M.P.) and MH122706 (M.P.), the Intramural Program of the National Institute on Aging (M.T.), a NARSAD young investigator award (C.T.), the Nathaniel Wharton Fund and The Baszucki Brain Research Fund (M.P.).
Author contributions
C.T. and M.P. designed research; V.M., H.-U.K., M.F., A.L., V.A.P., C.H., D.M.D., N.T.S., A.P.W., T.S.W., Y.W., M.T., L.F., D.A.B., and P.L.D.J. contributed new reagents/analytic tools; C.T. and A.S.M. analyzed data; and C.T., A.S.M., C.S., M.F., L.F., and M.P. wrote the paper.
Competing interests
The authors declare no competing interest.
Supporting Information
Appendix 01 (PDF)
- Download
- 2.31 MB
Dataset S01 (XLSX)
- Download
- 10.86 KB
Dataset S02 (XLSX)
- Download
- 298.67 KB
Dataset S03 (XLSX)
- Download
- 88.47 KB
Dataset S04 (XLSX)
- Download
- 193.56 KB
References
1
B. S. McEwen, Physiology and neurobiology of stress and adaptation: Central role of the brain. Physiol. Rev. 87, 873–904 (2007).
2
P. A. Boyle, L. L. Barnes, A. S. Buchman, D. A. Bennett, Purpose in life is associated with mortality among community-dwelling older persons. Psychosom. Med. 71, 574 (2009).
3
P. A. Boyle, A. S. Buchman, L. L. Barnes, D. A. Bennett, Effect of a purpose in life on risk of incident Alzheimer disease and mild cognitive impairment in community-dwelling older persons. Arch. Gen. Psychiatry 67, 304–310 (2010).
4
T. E. Seeman, T. M. Lusignolo, M. Albert, L. Berkman, Social relationships, social support, and patterns of cognitive aging in healthy, high-functioning older adults: MacArthur studies of successful aging. Health Psychol. 20, 243 (2001).
5
L. Fratiglioni, H.-X. Wang, K. Ericsson, M. Maytan, B. Winblad, Influence of social network on occurrence of dementia: A community-based longitudinal study. Lancet 355, 1315–1319 (2000).
6
T. E. Seeman, G. A. Kaplan, L. Knudsen, R. Cohen, J. Guralnik, Social network ties and mortality among tile elderly in the Alameda County Study. Am. J. Epidemiol. 126, 714–723 (1987).
7
J. T. Cacioppo, L. C. Hawkley, Perceived social isolation and cognition. Trends Cogn. Sci. 13, 447–454 (2009).
8
R. S. Wilson et al., Loneliness and risk of Alzheimer disease. Arch. Gen. Psychiatry 64, 234–240 (2007).
9
A. Terracciano et al., Personality and risk of Alzheimer’s disease: New data and meta-analysis. Alzheimer’s Dement. 10, 179–186 (2014).
10
M. Picard, B. S. McEwen, Psychological stress and mitochondria: A systematic review. Psychosom. Med. 80, 141–153 (2018).
11
M. Picard, B. S. McEwen, Psychological stress and mitochondria: A conceptual framework. Psychosom. Med. 80, 126 (2018).
12
M. Picard, C. Trumpff, Y. Burelle, Mitochondrial psychobiology: Foundations and applications. Curr. Opin. Behav. Sci. 28, 142–151 (2019).
13
T. E. Daniels, E. M. Olsen, A. R. Tyrka, Stress and psychiatric disorders: The role of mitochondria. Annu. Rev. Clin. Psychol. 16, 165–186 (2020).
14
J. Allen, H. J. Caruncho, L. E. Kalynchuk, Severe life stress, mitochondrial dysfunction, and depressive behavior: A pathophysiological and therapeutic perspective. Mitochondrion 56, 111–117 (2021).
15
Y. Chen, J. Zhang, How energy supports our brain to yield consciousness: Insights from neuroimaging based on the neuroenergetics hypothesis. Front. Syst. Neurosci. 15, 648860 (2021).
16
P. J. Magistretti, I. Allaman, A cellular perspective on brain energy metabolism and functional imaging. Neuron 86, 883–901 (2015).
17
D. E. Befroy, K. F. Petersen, D. L. Rothman, G. I. Shulman, Assessment of in vivo mitochondrial metabolism by magnetic resonance spectroscopy. Methods Enzymol. 457, 373–393 (2009).
18
Y. Ren et al., The impact of loneliness and social isolation on the development of cognitive decline and Alzheimer’s disease. Front. Neuroendocrinol. 69, 101061 (2023).
19
Y. Yu et al., A 3D atlas of functional human brain energetic connectome based on neuropil distribution. Cereb. Cortex 33, 3996–4012 (2023).
20
S. J. Lupien et al., Cortisol levels during human aging predict hippocampal atrophy and memory deficits. Nat. Neurosci. 1, 69–73 (1998).
21
A. N. Clausen et al., Combat exposure, posttraumatic stress disorder, and head injuries differentially relate to alterations in cortical thickness in military veterans. Neuropsychopharmacology 45, 491–498 (2020).
22
V. Rangaraju et al., Pleiotropic mitochondria: The influence of mitochondria on neuronal development and disease. J. Neurosci. 39, 8200–8208 (2019).
23
C. Batandier et al., Acute stress delays brain mitochondrial permeability transition pore opening. J. Neurochem. 131, 314–322 (2014).
24
Y. Kambe, A. Miyata, Potential involvement of the mitochondrial unfolded protein response in depressive-like symptoms in mice. Neurosci. Lett. 588, 166–171 (2015).
25
W. Liu, C. Zhou, Corticosterone reduces brain mitochondrial function and expression of mitofusin, BDNF in depression-like rodents regardless of exercise preconditioning. Psychoneuroendocrinology 37, 1057–1070 (2012).
26
R. G. Hunter et al., Stress and corticosteroids regulate rat hippocampal mitochondrial DNA gene expression via the glucocorticoid receptor. Proc. Natl. Acad. Sci. U.S.A. 113, 9099–9104 (2016).
27
M. Weger et al., Mitochondrial gene signature in the prefrontal cortex for differential susceptibility to chronic stress. Sci. Rep. 10, 18308 (2020).
28
S. Chakravarty et al., Chronic unpredictable stress (CUS)-induced anxiety and related mood disorders in a zebrafish model: Altered brain proteome profile implicates mitochondrial dysfunction. PLoS One 8, e63302 (2013).
29
M. D. Filiou et al., Proteomics and metabolomics analysis of a trait anxiety mouse model reveals divergent mitochondrial pathways. Biol. Psychiatry 70, 1074–1082 (2011).
30
A. R. Tyrka et al., Alterations of mitochondrial DNA copy number and telomere length with early adversity and psychopathology. Biol. Psychiatry 79, 78–86 (2016), https://doi.org/10.1016/j.biopsych.2014.12.025.
31
N. Cai et al., Molecular signatures of major depression. Curr. Biol. 25, 1146–1156 (2015).
32
M. Picard et al., A mitochondrial health index sensitive to mood and caregiving stress. Biol. Psychiatry 84, 9–17 (2018).
33
M. Zvěřová et al., Disturbances of mitochondrial parameters to distinguish patients with depressive episode of bipolar disorder and major depressive disorder. Neuropsychiatr. Dis. Treat. 15, 233–240 (2019).
34
C. Boeck et al., The association between cortisol, oxytocin and immune cell mitochondrial oxygen consumption in postpartum women with childhood maltreatment. Psychoneuroendocrinology 96, 69–77 (2018).
35
A. Karabatsiakis et al., Mitochondrial respiration in peripheral blood mononuclear cells correlates with depressive subsymptoms and severity of major depression. Transl. Psychiatry 4, e397 (2014).
36
C. Boeck et al., Targeting the association between telomere length and immuno-cellular bioenergetics in female patients with major depressive disorder. Sci. Rep. 8, 9419 (2018).
37
A. M. Gumpp et al., Childhood maltreatment is associated with changes in mitochondrial bioenergetics in maternal, but not in neonatal immune cells. Proc. Natl. Acad. Sci. U.S.A. 117, 24778–24784 (2020).
38
J. Hroudová, Z. Fišar, E. Kitzlerová, M. Zvěřová, J. Raboch, Mitochondrial respiration in blood platelets of depressive patients. Mitochondrion 13, 795–800 (2013).
39
M. Picard et al., Mitochondrial functions modulate neuroendocrine, metabolic, inflammatory, and transcriptional responses to acute psychological stress. Proc. Natl. Acad. Sci. U.S.A. 112, E6614–E6623 (2015).
40
U. Gimsa, E. Kanitz, W. Otten, S. M. Ibrahim, Behavior and stress reactivity in mouse strains with mitochondrial DNA variations. Ann. N. Y. Acad. Sci. 1153, 131–138 (2009).
41
T. L. Emmerzaal et al., Impaired mitochondrial complex I function as a candidate driver in the biological stress response and a concomitant stress-induced brain metabolic reprogramming in male mice. Transl. Psychiatry 10, 176 (2020).
42
T. Kasahara et al., Mice with neuron-specific accumulation of mitochondrial DNA mutations show mood disorder-like phenotypes. Mol. Psychiatry 11, 577–593 (2006).
43
F. Hollis et al., Mitochondrial function in the brain links anxiety with social subordination. Proc. Natl. Acad. Sci. U.S.A. 112, 15486–15491 (2015).
44
M. A. van der Kooij et al., Diazepam actions in the VTA enhance social dominance and mitochondrial function in the nucleus accumbens by activation of dopamine D1 receptors. Mol. Psychiatry 23, 569–578 (2018).
45
E. Gebara et al., Mitofusin-2 in the nucleus accumbens regulates anxiety and depression-like behaviors through mitochondrial and neuronal actions. Biol. Psychiatry 89, 1033–1044 (2021).
46
X. Xie et al., Depression-like behaviors are accompanied by disrupted mitochondrial energy metabolism in chronic corticosterone-induced mice. J. Steroid Biochem. Mol. Biol. 200, 105607 (2020).
47
F. Duclot, L. Sailer, P. Koutakis, Z. Wang, M. Kabbaj, Transcriptomic regulations underlying pair-bond formation and maintenance in the socially monogamous male and female prairie vole. Biol. Psychiatry 91, 141–151 (2022).
48
S. Ghosal et al., Mitofusin-2 in nucleus accumbens D2-MSNs regulates social dominance and neuronal function. Cell Rep. 42, 112776 (2023).
49
A. M. Rosenberg et al., Brain mitochondrial diversity and network organization predict anxiety-like behavior in male mice. Nat. Commun. 14, 4726 (2023).
50
D. H. Ülgen, S. R. Ruigrok, C. Sandi, Powering the social brain: Mitochondria in social behaviour. Curr. Opin. Neurobiol. 79, 102675 (2023).
51
Y. Kato et al., Lower availability of mitochondrial complex I in anterior cingulate cortex in autism: A positron emission tomography study. Am. J. Psychiatry 180, 277–284 (2023).
52
L. Holper, D. Ben-Shachar, J. Mann, Multivariate meta-analyses of mitochondrial complex I and IV in major depressive disorder, bipolar disorder, schizophrenia, Alzheimer disease, and Parkinson disease. Neuropsychopharmacology 44, 837–849 (2019).
53
S. Rausser et al., Mitochondrial phenotypes in purified human immune cell subtypes and cell mixtures. Elife 10, e70899 (2021).
54
M. J. Devine, J. T. Kittler, Mitochondria at the neuronal presynapse in health and disease. Nat. Rev. Neurosci. 19, 63–80 (2018).
55
R. Iwata, P. Vanderhaeghen, Regulatory roles of mitochondria and metabolism in neurogenesis. Curr. Opin. Neurobiol. 69, 231–240 (2021).
56
C. Fiebig et al., Mitochondrial dysfunction in astrocytes impairs the generation of reactive astrocytes and enhances neuronal cell death in the cortex upon photothrombotic lesion. Front. Mol. Neurosci. 12, 40 (2019).
57
J. Ye et al., Electron transport chain inhibitors induce microglia activation through enhancing mitochondrial reactive oxygen species production. Exp. Cell Res. 340, 315–326 (2016).
58
S. B. Shaikh, L. F. Nicholson, Effects of chronic low dose rotenone treatment on human microglial cells. Mol. Neurodegener. 4, 1–13 (2009).
59
R. Schoenfeld et al., Oligodendroglial differentiation induces mitochondrial genes and inhibition of mitochondrial function represses oligodendroglial differentiation. Mitochondrion 10, 143–150 (2010).
60
S. Nampoothiri, R. Nogueiras, M. Schwaninger, V. Prevot, Glial cells as integrators of peripheral and central signals in the regulation of energy homeostasis. Nat. Metab. 4, 813–825 (2022).
61
A. Quintana, S. E. Kruse, R. P. Kapur, E. Sanz, R. D. Palmiter, Complex I deficiency due to loss of Ndufs4 in the brain results in progressive encephalopathy resembling Leigh syndrome. Proc. Natl. Acad. Sci. U.S.A. 107, 10996–11001 (2010).
62
A. Quintana et al., Fatal breathing dysfunction in a mouse model of Leigh syndrome. J. Clin. Invest. 122, 2359–2368 (2012).
63
R. Iwata, P. Casimir, P. Vanderhaeghen, Mitochondrial dynamics in postmitotic cells regulate neurogenesis. Science 369, 858–862 (2020).
64
G. Sturm et al., OxPhos defects cause hypermetabolism and reduce lifespan in cells and in patients with mitochondrial diseases. Commun. Biol. 6, 1–22 (2023).
65
R. H. Swerdlow, J. M. Burns, S. M. Khan, The Alzheimer’s disease mitochondrial cascade hypothesis: Progress and perspectives. Biochim. Biophys. Acta 1842, 1219–1231 (2014).
66
H.-U. Klein et al., Characterization of mitochondrial DNA quantity and quality in the human aged and Alzheimer’s disease brain. Mol. Neurodegener. 16, 1–17 (2021).
67
S. Mostafavi et al., A molecular network of the aging human brain provides insights into the pathology and cognitive decline of Alzheimer’s disease. Nat. Neurosci. 21, 811–819 (2018).
68
W. S. Liang et al., Alzheimer’s disease is associated with reduced expression of energy metabolism genes in posterior cingulate neurons. Proc. Natl. Acad. Sci. U.S.A. 105, 4441–4446 (2008).
69
R. A. Neff et al., Molecular subtyping of Alzheimer’s disease using RNA sequencing data reveals novel mechanisms and targets. Sci. Adv. 7, eabb5398 (2021).
70
D. A. Bennett et al., Religious orders study and rush memory and aging project. J. Alzheimers Dis. 64, S161–S189 (2018).
71
C. Robins et al., Genetic control of the human brain proteome. Am. J. Hum. Genet. 108, 400–410 (2021).
72
K. N. Ochsner, J. J. Gross, The cognitive control of emotion. Trends Cogn. Sci. 9, 242–249 (2005).
73
A. F. Arnsten, Stress weakens prefrontal networks: Molecular insults to higher cognition. Nat. Neurosci. 18, 1376 (2015).
74
E. B. Ansell, K. Rando, K. Tuit, J. Guarnaccia, R. Sinha, Cumulative adversity and smaller gray matter volume in medial prefrontal, anterior cingulate, and insula regions. Biol. Psychiatry 72, 57–64 (2012).
75
C. Liston, B. S. McEwen, B. Casey, Psychosocial stress reversibly disrupts prefrontal processing and attentional control. Proc. Natl. Acad. Sci. U.S.A. 106, 912–917 (2009).
76
A. Monzel, J. Enriques, M. Picard, Multifaceted mitochondria: Moving mitochondrial science beyond function and dysfunction. Nat. Metab. 5, 546–562 (2023).
77
B. K. Chacko, D. Zhi, V. M. Darley-Usmar, T. Mitchell, The Bioenergetic Health Index is a sensitive measure of oxidative stress in human monocytes. Redox Biol. 8, 43–50 (2016).
78
S. Rath et al., MitoCarta3. 0: An updated mitochondrial proteome now with sub-organelle localization and pathway annotations. Nucleic Acids Res. 49, D1541–D1547 (2021).
79
D. Bennett et al., Neuropathology of older persons without cognitive impairment from two community-based studies. Neurology 66, 1837–1844 (2006).
80
D. A. Bennett et al., Natural history of mild cognitive impairment in older persons. Neurology 59, 198–205 (2002).
81
D. A. Bennett et al., Decision rules guiding the clinical diagnosis of Alzheimer’s disease in two community-based cohort studies compared to standard practice in a clinic-based cohort study. Neuroepidemiology 27, 169–176 (2006).
82
C. D. Ryff, C. L. Keyes, The structure of psychological well-being revisited. J. Pers. Soc. Psychol. 69, 719 (1995).
83
R. S. Wilson et al., The influence of cognitive decline on well-being in old age. Psychol. Aging 28, 304 (2013).
84
A. S. Buchman et al., Association between late-life social activity and motor decline in older adults. Arch. Intern. Med. 169, 1139–1146 (2009).
85
C. Fecher et al., Cell-type-specific profiling of brain mitochondria reveals functional and molecular diversity. Nat. Neurosci. 22, 1731–1742 (2019).
86
G. S. Green et al., Cellular dynamics across aged human brains uncover a multicellular cascade leading to Alzheimer’s disease. bioRxiv [Preprint] (2023). https://doi.org/10.1101/2023.03.07.531493 (Accessed 3 April 2023).
87
I. Perić, V. Costina, A. Stanisavljević, P. Findeisen, D. Filipović, Proteomic characterization of hippocampus of chronically socially isolated rats treated with fluoxetine: Depression-like behaviour and fluoxetine mechanism of action. Neuropharmacology 135, 268–283 (2018).
88
G. T. Rezin et al., Inhibition of mitochondrial respiratory chain in brain of rats subjected to an experimental model of depression. Neurochem. Int. 53, 395–400 (2008).
89
J. L. Madrigal et al., Glutathione depletion, lipid peroxidation and mitochondrial dysfunction are induced by chronic stress in rat brain. Neuropsychopharmacology 24, 420–429 (2001).
90
P. Rinwa, A. Kumar, Piperine potentiates the protective effects of curcumin against chronic unpredictable stress-induced cognitive impairment and oxidative damage in mice. Brain Res. 1488, 38–50 (2012).
91
P. Rinwa, A. Kumar, Modulation of nitrergic signalling pathway by American ginseng attenuates chronic unpredictable stress-induced cognitive impairment, neuroinflammation, and biochemical alterations. Naunyn Schmiedebergs Arch. Pharmacol. 387, 129–141 (2014).
92
E. Morava et al., Depressive behaviour in children diagnosed with a mitochondrial disorder. Mitochondrion 10, 528–533 (2010).
93
A. Strasser, L. Xin, R. Gruetter, C. Sandi, Nucleus accumbens neurochemistry in human anxiety: A 7 T 1H-MRS study. Eur. Neuropsychopharmacol. 29, 365–375 (2019).
94
C. J. Jong, J. Azuma, S. Schaffer, Mechanism underlying the antioxidant activity of taurine: Prevention of mitochondrial oxidant production. Amino Acids 42, 2223–2232 (2012).
95
S. W. Schaffer, K. Shimada-Takaura, C. J. Jong, T. Ito, K. Takahashi, Impaired energy metabolism of the taurine-deficient heart. Amino Acids 48, 549–558 (2016).
96
A. Wiehler, F. Branzoli, I. Adanyeguh, F. Mochel, M. Pessiglione, A neuro-metabolic account of why daylong cognitive work alters the control of economic decisions. Curr. Biol. 32, 3564–3575.e5 (2022).
97
M. Picard et al., A mitochondrial health index sensitive to mood and caregiving stress. Biol. Psychiatry 84, 9–17 (2018).
98
C. Boeck et al., Inflammation in adult women with a history of child maltreatment: The involvement of mitochondrial alterations and oxidative stress. Mitochondrion 30, 197–207 (2016).
99
C. Boeck et al., The association between cortisol, oxytocin, and immune cell mitochondrial oxygen consumption in postpartum women with childhood maltreatment. Psychoneuroendocrinology 96, 69–77 (2018).
100
L. E. Gyllenhammer et al., Maternal inflammation during pregnancy and offspring brain development: The role of mitochondria. Biol. Psychiatry Cogn. Neurosci. Neuroimaging 7, 498–509 (2022).
101
N. Bobba-Alves, R.-P. Juster, M. Picard, The energetic cost of allostasis and allostatic load. Psychoneuroendocrinology 146, 105951 (2022).
102
M. Schwaiger et al., Altered stress-induced regulation of genes in monocytes in adults with a history of childhood adversity. Neuropsychopharmacology 41, 2530–2540 (2016).
103
K. Chandramouli, P. Y. Qian, Proteomics: Challenges, techniques and possibilities to overcome biological sample complexity. Hum. Genomics Proteomics 2009, 239204 (2009).
104
L. E. Liharska et al., A study of gene expression in the living human brain. medRxiv [Preprint] (2023). https://doi.org/10.1101/2023.04.21.23288916 (Accessed 1 September 2023).
105
L. Yu et al., Cortical proteins associated with cognitive resilience in community-dwelling older persons. JAMA Psychiatry 77, 1172–1180 (2020).
106
L. Ferrucci, The Baltimore Longitudinal Study of Aging (BLSA): A 50-year-long journey and plans for the future. J. Gerontol. A Biol. Sci. Med. Sci. 63, 1416–1419 (2008).
107
V. Swarup et al., Identification of conserved proteomic networks in neurodegenerative dementia. Cell Rep. 31, 107807 (2020).
108
A. P. Wingo et al., Large-scale proteomic analysis of human brain identifies proteins associated with cognitive trajectory in advanced age. Nat. Commun. 10, 1–14 (2019).
109
T. G. Beach et al., Arizona study of aging and neurodegenerative disorders and brain and body donation program. Neuropathology 35, 354 (2015).
110
D. A. Bennett, J. A. Schneider, Z. Arvanitakis, R. S. Wilson, Overview and findings from the religious orders study. Curr. Alzheimer Res. 9, 628–645 (2012).
111
D. A. Bennett et al., Overview and findings from the rush memory and aging project. Curr. Alzheimer Res. 9, 646–663 (2012).
112
D. Bennett et al., Apolipoprotein E ε4 allele, AD pathology, and the clinical expression of Alzheimer’s disease. Neurology 60, 246–252 (2003).
113
B. T. Hyman, J. Q. Trojanowski, Editorial on consensus recommendations for the postmortem diagnosis of Alzheimer disease from the National Institute on Aging and the Reagan Institute Working Group on diagnostic criteria for the neuropathological assessment of Alzheimer disease. J. Neuropathol. Exp. Neurol. 56, 1095 (1997).
114
H.-U. Klein, M. Schäfer, D. A. Bennett, H. Schwender, P. L. De Jager, Bayesian integrative analysis of epigenomic and transcriptomic data identifies Alzheimer’s disease candidate genes and networks. PLoS Comput. Biol. 16, e1007771 (2020).
115
D. A. Bennett, J. A. Schneider, Y. Tang, S. E. Arnold, R. S. Wilson, The effect of social networks on the relation between Alzheimer’s disease pathology and level of cognitive function in old people: A longitudinal cohort study. Lancet Neurol. 5, 406–412 (2006).
116
R. S. Wilson et al., Negative social interactions and risk of mild cognitive impairment in old age. Neuropsychology 29, 561 (2015).
117
L. S. Radloff, The CES-D scale: A self-report depression scale for research in the general population. Appl. Psychol. Meas. 1, 385–401 (1977).
118
D. Watson, L. A. Clark, A. Tellegen, Development and validation of brief measures of positive and negative affect: The PANAS scales. J. Pers. Soc. Psychol. 54, 1063 (1988).
119
A. D. Turner, B. D. James, A. W. Capuano, N. T. Aggarwal, L. L. Barnes, Perceived stress and cognitive decline in different cognitive domains in a cohort of older African Americans. Am. J. Geriatr. Psychiatry 25, 25–34 (2017).
120
R. S. Wilson et al., Childhood adversity and psychosocial adjustment in old age. Am. J. Geriatr. Psychiatry 14, 307–315 (2006).
121
E. W. Burgess, R. S. Cavan, R. J. Havighurst, Your Activities and Attitudes (Science Research Associates, 1949).
122
W. W. Eaton, C. Smith, M. Ybarra, C. Muntaner, A. Tien, “Center for epidemiologic studies depression scale: Review and revision (CESD and CESD-R)” in The Use of Psychological Testing for Treatment Planning and Outcomes Assessment: Instruments for Adults, M. E. Maruish, Ed. (Lawrence Erlbaum Associates Publishers, 2004).
123
K. Brodman, A. J. Erdmann, I. Lorge, H. G. Wolff, T. H. Broadbent, The Cornell Medical Index: An adjunct to medical interview. J. Am. Med. Assoc. 140, 530–534 (1949).
124
L. Ping et al., Global quantitative analysis of the human brain proteome in Alzheimer’s and Parkinson’s disease. Sci. Data 5, 1–12 (2018).
125
E. C. Johnson et al., Deep proteomic network analysis of Alzheimer’s disease brain reveals alterations in RNA binding proteins and RNA splicing associated with disease. Mol. Neurodegener. 13, 1–22 (2018).
126
P. Mertins et al., Reproducible workflow for multiplexed deep-scale proteome and phosphoproteome analysis of tumor tissues by liquid chromatography–mass spectrometry. Nat. Protocols 13, 1632–1661 (2018).
127
E. C. B. Johnson et al., Data from “Large-scale proteomic analysis of Alzheimer’s disease brain and cerebrospinal fluid reveals early changes in energy metabolism associated with microglia and astrocyte activation”. Synapse. https://www.synapse.org/#!Synapse:syn17015098. Deposited 5 July 2022.
128
L. Yu et al., Targeted brain proteomics uncover multiple pathways to Alzheimer’s dementia. Ann. Neurol. 84, 78–88 (2018).
129
L. Yu et al., Data from “Targeted brain proteomics uncover multiple pathways to Alzheimer’s dementia”. Synapse. https://www.synapse.org/#!Synapse:syn10468856. Deposited 18 December 2019.
130
T. Raj et al., Integrative transcriptome analyses of the aging brain implicate altered splicing in Alzheimer’s disease susceptibility. Nat. Genet. 50, 1584–1592 (2018).
131
S. Oveisgharan et al., Estrogen receptor genes, cognitive decline, and Alzheimer disease. Neurology 100, e1474–e1487 (2023).
132
T. Raj et al., Data from: “Integrative transcriptome analyses of the aging brain implicate altered splicing in Alzheimer’s disease susceptibility.” Synapse. https://www.synapse.org/#!Synapse:syn3388564. Deposited 18 October 2021.
133
M. Fujita et al., Cell-subtype specific effects of genetic variation in the aging and Alzheimer cortex. bioRxiv [Preprint] (2022). https://doi.org/10.1101/2022.11.07.515446 (Accessed 8 December 2022).
134
A. Cain et al., Multicellular communities are perturbed in the aging human brain and Alzheimer’s disease. Nat. Neurosci. 26, 1267–1280 (2023).
135
M. Fujita et al., Data from “Cell-subtype specific effects of genetic variation in the aging and Alzheimer cortex.” Synapse. https://www.synapse.org/#!Synapse:syn31512863. Deposited 2 March 2023.
136
H-.U. Klein et al., Data from “Characterization of mitochondrial DNA quantity and quality in the human aged and Alzheimer’s disease brain”. Synapse. https://www.synapse.org/#!Synapse:syn25618990. Deposited 18 October 2021.
137
C. Trumpff et al., Mitochondrial respiratory chain protein co-regulation in the human brain. Heliyon 8, e09353 (2022).
138
L. Jiang et al., A quantitative proteome map of the human body. Cell 183, 269–283.e9 (2020).
139
D. M. Duong, C. Hurst, N. T. Seyfried, blsa_7batch_tmtmultiregion. Synapse. https://www.synapse.org/#!Synapse:syn39213792. Deposited 7 October 2023.
Information & Authors
Information
Published in
Cover image: Pictured is an artistic illustration of the interplay between AI and the natural world. A collection of articles in the Physics Meets Machine Learning Special Feature explores how advances in machine learning can help solve problems in the natural sciences and how ideas from physics and biology can help improve machine learning. Together, the articles highlight interconnected and emerging research in the topics of physics and machine learning, with applications including protein folding, material assembly, and neural networks. See the Introduction to the Special Feature, e2403580121. Image credit: Alex Eben Meyer for the Simons Foundation.
Classifications
Copyright
Copyright © 2024 the Author(s). Published by PNAS. This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
Data, Materials, and Software Availability
TMT isobaric labeling mass spectrometry (ROSMAP) data have been deposited in www.synapse.org (syn17015098) (127). Anonymized [TMT isobaric labeling mass spectrometry (BLSA)] data have been deposited in www.synapse.org (syn39213792) (139).
Submission history
Received: October 11, 2023
Accepted: April 30, 2024
Published online: June 18, 2024
Published in issue: July 2, 2024
Keywords
Acknowledgments
We acknowledge participants of the ROS and MAP and BLSA studies. Support for this work was provided by NIH grants NHI RF1AG057473 (D.A.B. and P.L.D.J.), U01AG061356 (D.A.B. and P.L.D.J.), P30AG16101 (D.A.B.), R01AG15819 (D.A.B.), R01AG070438 (P.L.D.J.), R01AG17917 (D.A.B.), U01AG46152 (D.A.B. and P.L.D.J.), U01AG61356 (D.A.B. and P.L.D.J.), R01AG036836 (P.L.D.J.), U01AG061357 (N.T.S.), R01AG061800 (N.T.S.), RF1AG062181 (N.T.S.), RF1AG076821 (M.P.), GM119793 (M.P.) and MH122706 (M.P.), the Intramural Program of the National Institute on Aging (M.T.), a NARSAD young investigator award (C.T.), the Nathaniel Wharton Fund and The Baszucki Brain Research Fund (M.P.).
Author Contributions
C.T. and M.P. designed research; V.M., H.-U.K., M.F., A.L., V.A.P., C.H., D.M.D., N.T.S., A.P.W., T.S.W., Y.W., M.T., L.F., D.A.B., and P.L.D.J. contributed new reagents/analytic tools; C.T. and A.S.M. analyzed data; and C.T., A.S.M., C.S., M.F., L.F., and M.P. wrote the paper.
Competing Interests
The authors declare no competing interest.
Notes
This article is a PNAS Direct Submission.
Authors
Metrics & Citations
Metrics
Citation statements
Altmetrics
Citations
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.