Acute effects of a ketone monoester, whey protein, or their coingestion on mTOR trafficking and protein-protein colocalization in human skeletal muscle

Abstract

We recently demonstrated that acute oral ketone monoester intake induces a stimulation of postprandial myofibrillar protein synthesis rates comparable to that elicited following the ingestion of 10 g whey protein or their coingestion. The present investigation aimed to determine the acute effects of ingesting a ketone monoester, whey protein, or their coingestion on mechanistic target of rapamycin (mTOR)-related protein-protein colocalization and intracellular trafficking in human skeletal muscle. In a randomized, double-blind, parallel group design, 36 healthy recreationally active young males (age: 24.2 ± 4.1 yr) ingested either: 1) 0.36 g·kg⁻¹ bodyweight of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate (KET), 2) 10 g whey protein (PRO), or 3) the combination of both (KET + PRO). Muscle biopsies were obtained in the overnight postabsorptive state (basal conditions), and at 120 and 300 min in the postprandial period for immunofluorescence assessment of protein translocation and colocalization of mTOR-related signaling molecules. All treatments resulted in a significant (Interaction: P < 0.0001) decrease in tuberous sclerosis complex 2 (TSC2)-Ras homolog enriched in brain (Rheb) colocalization at 120 min versus basal; however, the decrease was sustained at 300 min versus basal (P < 0.0001) only in KET + PRO. PRO and KET + PRO increased (Interaction: P < 0.0001) mTOR-Rheb colocalization at 120 min versus basal; however, KET + PRO resulted in a sustained increase in mTOR-Rheb colocalization at 300 min that was greater than KET and PRO. Treatment intake increased mTOR-wheat germ agglutinin (WGA) colocalization at 120 and 300 min (Time: P = 0.0031), suggesting translocation toward the fiber periphery. These findings demonstrate that ketone monoester intake can influence the spatial mechanisms involved in the regulation of mTORC1 in human skeletal muscle.

NEW & NOTEWORTHY We explored the effects of a ketone monoester (KET), whey protein (PRO), or their coingestion (KET + PRO) on mTOR-related protein-protein colocalization and intracellular trafficking in human muscle. All treatments decreased TSC2-Rheb colocalization at 120 minutes; however, KET + PRO sustained the decrease at 300 min. Only PRO and KET + PRO increased mTOR-Rheb colocalization; however, the increase at 300 min was greater in KET + PRO. Treatment intake increased mTOR-WGA colocalization, suggesting translocation to the fiber periphery. Ketone bodies influence the spatial regulation of mTOR.

INTRODUCTION

The mechanistic target of rapamycin complex 1 (mTORC1) is a central regulator of muscle protein synthesis (MPS) in response to a variety of anabolic stimuli including nutrient availability, growth factors, and mechanical stimulation (1). Regulation of the mTORC1 pathway involves protein-protein interactions and intracellular protein translocation (2–5) that elicit changes in downstream kinase activity (4, 6). Namely, the subcellular dissociation of tuberous sclerosis complex 2 (TSC2; negative regulator of mTOR) from Ras homolog enriched in brain (Rheb; positive regulator of mTOR) and reciprocal binding of Rheb with mTOR’s catalytic domain to promote its activation are pivotal interactions regulating mTORC1 activity. In addition, the trafficking of mTORC1 towards the muscle cell periphery, now understood as the site of protein synthesis (1, 7), is coordinated by the lysosome (1). In recent years, the analysis of mTORC1 translocation and protein complex colocalization has become increasingly common in human studies attempting to unravel the molecular mechanisms underpinning the mTORC1-related stimulation of MPS rates, in part because traditional Western blot readouts of pathway activation can be dissociated from acute changes in MPS rates (8) and protein kinase activity (9). In an initial study, Song and colleagues (6) reported that mTOR and the lysosomal protein LAMP2 were highly colocalized in skeletal muscle at rest; however, resistance exercise resulted in rapid translocation of mTOR-LAMP2 towards the cell periphery. Simultaneously, resistance exercise promoted TSC2-Rheb dissociation, and increased mTOR-Rheb colocalization post-exercise. Since this initial investigation, a number of studies have reported changes in mTORC1 cellular distribution and/or protein-protein colocalization in human skeletal muscle in response to nutrient stimuli, exercise, or nutrient × exercise interactions (for review, see Ref. 1). These spatial mTORC1 regulatory events are now thought to contribute to increased mTORC1 activity (10) and the stimulation of MPS rates in response to anabolic stimuli.

Ketone bodies [i.e., β-hydroxybutyrate (β-OHB), acetoacetate, and acetone] are lipid-derived molecules whose endogenous production is amplified in the liver during periods of carbohydrate restriction. In addition to serving as metabolic fuel substrates, ketone bodies are metabolites now recognized to play important roles as signaling molecules capable of altering gene expression and cellular function (11). Orally ingested exogenous ketone delivery is a recent and practical approach used to increase circulating ketone bodies, avoiding the need for carbohydrate restriction to amplify endogenous ketogenesis. A growing area of renewed interest is the protein anabolic and/or anticatabolic actions of ketone bodies in skeletal muscle. In a seminal study, infusion of DL-β-OHB to ∼2 mM was reported to stimulate MPS rates in humans (12). More recently, an exogenous ketone monoester coingested with a protein-carbohydrate supplement increased the phosphorylation status of select proteins within the mTORC1 pathway in human skeletal muscle subsequent to exercise (13). In the same study (13), the authors reported that ketone bodies potentiated the increase in mTORC1 activation and protein synthesis in leucine-stimulated myotubes. In support of these findings, we recently demonstrated that acute ingestion of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate (KET), whey protein (PRO), or their coingestion (KET + PRO) in the overnight postabsorptive state elicited a similar stimulation of myofibrillar protein synthesis (MyoPS) rates in young males (14). Therefore, the purpose of the present study was to determine the effects KET, PRO, and KET + PRO on mTORC1-related protein-protein colocalization and intracellular protein trafficking in human skeletal muscle. We hypothesized that KET, PRO, and KET + PRO would augment mTOR-related protein-protein colocalization and translocation versus basal conditions (i.e., overnight postabsorptive conditions).

MATERIALS AND METHODS

Participants and Ethical Approval

Thirty-six recreationally active healthy young (age: 24.2 ± 4.1 yr; body weight: 72.1±8.4 kg; body fat: 20.9 ± 5.8%; BMI: 23.4 ± 2 kg/m²; means ± SD) males were recruited for this study. The participants were part of a larger project conducted within our laboratory (14). Participants gave written informed consent to participate in a protocol approved by the Faculty of Medicine Institutional Review Board at McGill University on human experimentation and in accordance with the Declaration of Helsinki revised in 2013.

Experimental Design

The details of this randomized, double-blind, parallel group study (registered at clinicaltrials.gov as NCT04565444) have been described previously (14). Briefly, participants reported to the laboratory at ∼0730 h in the ∼10 h overnight postabsorptive state for a single test visit. Following the collection of a baseline muscle biopsy (t = 0 min) from the vastus lateralis, one of three nutritional treatments were administered: 0.36 g·kg⁻¹ bodyweight of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate (KET), 10 g whey protein concentrate (PRO), or the combination of both (KET + PRO). Treatments were matched for volume and were prepared in opaque plastic bottles. Additional muscle biopsies were then obtained at t = 120 and 300 min in the postprandial period. Muscle biopsy samples (∼30 mg) were freed from any visible blood, adipose, and connective tissue, immediately mounted in Optimal Cutting Temperature Compound (Tissue Plus, Fisher HealthCare), and frozen in isopentane cooled by liquid nitrogen before immediate storage at −80°C for subsequent immunofluorescence analysis of mTORC1-related colocalization and translocation.

Immunofluorescence

Serial cross sections (7 µm) were stained for mTOR colocalization and digital images were captured using a Zeiss Axio Observer Z1 microscope (Göttingen, Germany) at ×40 0.75 numerical aperture (NA) magnification as previously described (6). For total protein quantification, 4 slide replicates for each participant were stained, imaged, and quantified on uncoated glass slides (VWR International, UK). Each slide contained duplicates from all three time points (t = 0, 120, and 300 min) for two participants from random treatment groups. All image-capturing parameters were kept constant between images, including exposure time, gain, image frame, and light intensity. For colocalization analysis, at least six images of randomly selected fields of view were captured per section with each image including ∼8 fibers on average. The average number of fibers analyzed per subject, per time-point was ∼122 (range: 37–202) for colocalization. For each subject, muscle samples were taken from three time-points (3 samples total), and average values were calculated from randomly selected fields of two serial sections for each time-point. Prior to image analysis, each image underwent deconvolution via Hi-Gaussian correction and despeckling. All image processing and quantification was carried out in Zeiss Zen lite (version 3.5, blue edition) and Pearson’s correlation coefficient was used to quantify correlations between TSC2-Rheb, mTOR-Rheb, mTOR-lysosome-associated membrane protein 2 (LAMP2), and mTOR-wheat germ agglutinin (WGA; sarcolemmal marker) to be consistent with previous human in vivo research (2, 3, 5, 6, 15). Theoretically, increases in colocalization between two time-points result in r values moving closer to 1.

Antibodies

The mouse monoclonal anti-mTOR (no. 05–1592) antibody was purchased from Millipore (Toronto, ON, Canada) and mouse monoclonal anti-TSC2 (no. AM1919b) was purchased from Abcepta (San Diego, CA). The corresponding conjugated secondary antibody to these antibodies was goat anti-mouse IgGγ1 Alexa 594 (no. A11005; Life Technologies Inc.). Antibodies targeting rabbit polyclonal-LAMP2 (no. AP1824d) were purchased from Abcepta (San Diego, CA) and Rheb (no. ab92313; Abcam) was purchased from Abcam (Toronto, ON, Canada). The corresponding antibody to these antibodies was goat anti-rabbit IgG (H&L) Alexa 488 secondary antibody (no. A11008; Life Technologies Inc.). Finally, wheat germ agglutinin (WGA-350, no. W11263, Fisher Scientific) was used to identify the cell periphery (i.e., sarcolemma) of muscle fibers. The protocol for the incubations for the primary and secondary antibodies as well as the dilutions have been described previously (6).

Statistical Analysis

Colocalization data (i.e., Pearson’s r) were analyzed via two-factor (treatment × time) repeated measures ANOVA with Bonferroni post hoc analysis following a main effect for time, treatment, or treatment × time interaction. Assumptions of the statistical models were assessed using Mauchly’s test, and the D’Agostino-Pearson omnibus normality test at a significance of P < 0.05. If a significant Mauchly’s test was determined, the Greenhouse-Geisser correction factor was used to adjust the degrees of freedom accordingly. For data that did not pass normality, values were transformed with the ln or square root of the value. The statistical analysis was performed on transformed data, but non-transformed data are presented in graphic form for clarity. GraphPad Prism for Windows (version 8.4.3; GraphPad Software; CA) was used for all statistical analysis. Statistical significance was set at P < 0.05. All data are presented as individual participant data along with means ± SD.

RESULTS

KET + PRO but Not KET or PRO Sustains the Decrease in TSC2 Colocalization with Rheb

An interaction effect was observed for TSC2-Rheb colocalization (Interaction: P < 0.0001). TSC2-Rheb colocalization decreased by ∼20 ± 5%, ∼16 ± 4%, and ∼17 ± 6% in KET, PRO, and KET + PRO at 120 min versus basal (all P < 0.0001), with no difference between treatments (all P > 0.05). TSC2-Rheb colocalization returned to values not different from basal at 300 min in KET and PRO, respectively, but remained decreased at 300 min by ∼19 ± 5% versus basal (P < 0.0001) in KET + PRO (Fig. 1, A and B).

Figure 1.

Tuberous sclerosis complex 2 (TSC2) and mechanistic target of rapamycin (mTOR) colocalization with Ras homolog enriched in brain (Rheb) in the basal state and in response to 0.36 g·kg⁻¹ bodyweight of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate (KET), 10 g whey protein concentrate (PRO), or the combination of both (KET + PRO) at 120 and 300 min in the postprandial period. Immunofluorescence quantification of mTOR/TSC2 (red) and Rheb (green) colocalization, displayed as a composite image (merge) (n = 12/condition). Yellow/orange regions represent TSC2 and Rheb (A) mTOR and Rheb colocalization (C). Each panel represents one subject from KET, PRO, and KET + PRO across the experimental time course. Group data are quantified and reported as TSC2-Rheb colocalization (B) and mTOR-Rheb colocalization (D). Scale bar represents 100-µm area. All data are presented as individual participant data along with means ± SD and were analyzed using two-way repeated measures ANOVA. Means with different letters are different within each treatment (P < 0.05). *Different from KET at same time point (P < 0.05). †Different from PRO at same time point (P < 0.05).

KET + PRO Amplifies the Sustained Increase in mTOR Colocalization with Rheb versus KET and PRO

An interaction effect was observed for mTOR-Rheb colocalization (Interaction: P < 0.0001). mTOR-Rheb colocalization remained unchanged versus basal in KET. However, mTOR-Rheb colocalization increased by ∼25 ± 8% in PRO, and by ∼16 ± 8% in KET + PRO. The increase in mTOR-Rheb colocalization at 120 min was greater in PRO (P = 0.0069) and KET + PRO (P = 0.0018) versus KET. At 300 min, mTOR-Rheb colocalization remained increased by ∼5 ± 7% and by ∼14 ± 10% versus basal in PRO (P = 0.0378) and KET + PRO (P = 0.0007), respectively. At 300 min, mTOR-Rheb colocalization was greater in KET + PRO versus both KET (P = 0.0010) and PRO (P = 0.0008) (Fig. 1, C and D).

mTOR Colocalization with LAMP2 Decreases at 300 Min with KET + PRO, but Remains Highly Colocalized with KET and PRO

An interaction effect was observed for mTOR-LAMP2 colocalization (Interaction: P = 0.0412). mTOR-LAMP2 colocalization was unchanged versus basal at 120 and 300 min after the ingestion of both KET and PRO, but was decreased by ∼14 ± 11% at 300 min versus basal (P = 0.0033) after the ingestion of KET + PRO (Fig. 2, A and B).

Figure 2.

Lysosome-associated membrane 2 (LAMP2) and wheat germ agglutinin (WGA) colocalization with mechanistic target of rapamycin (mTOR) in the basal state and in response to 0.36 g·kg⁻¹ bodyweight of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate (KET), 10 g whey protein concentrate (PRO), or the combination of both (KET + PRO) at 120 and 300 minutes in the postprandial period. Immunofluorescence quantification of mTOR (red) and LAMP2/Rheb (green), and mTOR and WGA (blue) colocalization, displayed as a composite image (merge) (n = 12/condition). Yellow/orange regions represent mTOR and LAMP2 colocalization (A) and pink/magenta regions represent mTOR and WGA colocalization (C). Each panel represents one subject from KET, PRO, KET + PRO across the experimental time course. Group data are quantified and reported as mTOR-LAMP2 colocalization (B) and mTOR-WGA colocalization (D). Scale bar represents 100-µm area. All data are presented as individual participant data along with means ± SD and were analyzed using two-way repeated measures ANOVA. In B, means with different letters are different within each treatment (P < 0.05). In D, means with different letters are different (P < 0.05).

Treatment Ingestion Increases mTOR Colocalization with WGA with No Differences between Treatments

mTOR-WGA colocalization increased (Time: P = 0.0031) by ∼8 ± 20% in KET, ∼29 ± 22% in PRO, and ∼16 ± 39% in KET + PRO at 120 min versus basal (P = 0.0007), and by ∼19 ± 64% in KET, ∼14 ± 59% in PRO, and ∼19 ± 39% in KET + PRO at 300 min versus basal (P = 0.0154). No difference between treatments (Treatment: P = 0.1517) or treatment × time interaction (Interaction: P = 0.3767) was observed (Fig. 2, C and D). A detailed overview of the protein-protein colocalization and translocation data is provided in Supplemental Table S1.

DISCUSSION

We (14) and others (12) have demonstrated that ketone body administration stimulates muscle protein synthesis rates in humans (13). In the present study we evaluated the acute effects of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate, with (KET + PRO) and without (KET) dietary protein (10 g whey protein) coingestion on changes in mTORC1-related protein-protein colocalization and intracellular protein trafficking in human skeletal muscle at both 120 and 300 min in the postprandial period as compared to dietary protein intake (PRO). We report that KET can modulate the underlying spatial mechanisms of protein synthesis in human skeletal muscle, via decreasing TSC2-Rheb colocalization and increasing mTOR-WGA colocalization. KET + PRO further amplifies this effect by sustaining the decrease in TSC2-Rheb colocalization and enhancing mTOR-Rheb colocalization at 300 min versus KET and PRO.

The mTORC1 signaling cascade is a primary regulator of MPS rates in response to anabolic stimuli (1). The kinase activity of mTORC1 is regulated in part by protein-protein colocalization and intracellular localization dynamics (16). Amino acids represent key nutritional stimuli capable of activating mTORC1, with human studies demonstrating that increases in amino acid availability can result in recruitment of mTORC1 to the lysosomal surface (2), and/or result in its translocation to the muscle cell periphery (3, 5) to support its activation. In agreement with previous work demonstrating dissociation of TSC2 from the lysosome in response to nutrient-derived stimuli (17), our study revealed a decrease in TSC2-Rheb colocalization versus basal conditions at 120 min post-treatment ingestion in all treatments. However, whereas TSC2-Rheb colocalization returned to basal values at 300 min in both KET and PRO, the decrease in TSC2-Rheb colocalization persisted to 300 min in KET + PRO (Fig. 1, A and B). A reduction in TSC2-Rheb colocalization has previously been reported in human muscle in response to mixed-macronutrient meal administration at rest (5), and combined exercise-nutrient stimulation (2, 5, 6). TSC2 canonically acts as a negative regulator of Rheb (18, 19), preventing Rheb from becoming GTP-bound and activating mTORC1 (18). However, TSC2 dissociation from Rheb and the lysosome allows Rheb to activate mTORC1 (20). The protein kinase Akt can phosphorylate and deactivate TSC2, preventing TSC2-dependent inhibition of mTORC1 signaling (20). Although previous research has reported that β-OHB can activate the PI3K-Akt pathway in cardiomyocytes (21) and rodent skeletal muscle (22), we (14) and others (13) have reported no effect of ketone monoester intake on Akt^{(Ser 473)} phosphorylation in human muscle. Alternatively, TSC2 can be phosphorylated and deactivated via Akt-independent mechanisms involving Erk/MAPK signaling (1), a pathway also shown to be regulated via ketone bodies (23).

In a reciprocal manner, an increase in mTOR-Rheb colocalization versus basal values occurred in PRO and KET + PRO at both 120 and 300 min in the postprandial period; however, there was no change in mTOR-Rheb colocalization in KET. At 300 min, mTOR-Rheb colocalization was greater in KET + PRO versus both KET and PRO (Fig. 1, C and D). Our results align with previous research demonstrating an increase in mTOR-Rheb colocalization in human muscle in response to mixed-macronutrient meal administration at rest (5), and combined exercise-nutrient stimulation (2, 5, 6). Once the inhibitory effect of TSC2 on Rheb is alleviated, Rheb can become GTP-bound and activate mTORC1 (24). Overall, our results demonstrate that ketone monoester intake (KET) does not increase mTOR-Rheb colocalization but can further potentiate the protein/amino acid (KET + PRO) mediated increase in mTOR-Rheb colocalization in human skeletal muscle.

In contrast to in vitro and rodent studies suggesting the translocation of mTORC1 towards the lysosome, where its activation occurs (16, 25), is a key event in response to amino acid administration (25), we found no increase in mTOR-LAMP2 colocalization subsequent to treatment administration. However, this observation aligns with human research demonstrating that mTOR-LAMP2 colocalization remained unaltered following mixed meal ingestion and resistance exercise (6). Moreover, our findings are consistent with reports indicating marginal changes in mTOR-LAMP2 colocalization after egg white ingestion (2) and the absence of an increase in response to mixed meal ingestion both at rest, and during recovery from endurance exercise (5). However, a decrease in mTOR-LAMP2 colocalization occurred at 300 min in KET + PRO. One possible explanation could be variations in lysosomal content, indicated by changes in LAMP2 fluorescence intensity. Previous research has shown that ketone bodies can induce chaperone-mediated autophagy in vitro (26) and activate the autophagic-lysosomal pathway by stimulating AMPK and TFEB-mediated lysosomal biogenesis in cultured neurons (27). This may have led to an increase in the free-lysosomal pool (28), thereby affecting mTOR-LAMP2 association. Collectively, these results reinforce the proposition that the association of mTORC1 with the lysosome may be a requisite event for facilitating heightened rates of mRNA translation in human skeletal muscle; however, mTOR-LAMP2 colocalization is not always augmented in response to nutritional stimuli.

With nutrient provision, mTORC1-lysosome complexes can undergo translocation to peripheral cell regions, and impeding this translocation prevents mTORC1 activation (29). Thus, aside from emphasizing the significance of mTOR localization to the lysosomal surface, it is widely accepted that the translocation of mTORC1 to the cell periphery is a crucial factor supporting its activation. In the present study, we demonstrate a similar increase in mTOR-WGA colocalization at 120 and 300 min versus basal following KET, PRO, and KET + PRO. This finding aligns with prior research demonstrating an increase in mTOR-WGA colocalization in response to nutritional stimuli administered at rest (15, 30, 31), indicative of mTOR translocation to peripheral fiber regions (6), and in close association with capillaries and nutrients (6, 32). The comparable increase in mTOR-WGA colocalization between treatments suggests that dietary protein/amino acid intake may not be required for the translocation of the mTORC1-lysosome complex toward the fiber periphery, since KET did not contain protein (14). Therefore, while it is well established that protein/amino acids promote mTOR localization, our findings suggest that ketone bodies may represent metabolites capable of promoting translocation of mTOR to the cell periphery, further highlighting the role of nonprotein factors in augmenting the molecular regulation of MPS (5).

Limitations of this study include lack of a true negative control treatment (e.g., flavored water), the single-sex and age-category nature of the study design (i.e., young males), and lack of confirmation that these small changes in localization/translocation impact downstream kinase activity in human skeletal muscle. Additional research is required to address these limitations and further clarify the physiological relevance of mTOR trafficking and protein-protein colocalization in human skeletal muscle. For example, determining the threshold of change in protein colocalization necessary to increase kinase activity, stimulate MPS rates, and alter changes in muscle phenotype in response to exercise training are pertinent topics for future research.

In conclusion, we demonstrate that acute intake of the ketone monoester (R)-3-hydroxybutyl (R)-3-hydroxybutyrate (KET), 10 g whey protein (PRO), or their coingestion (KET + PRO) decrease TSC2-Rheb colocalization; however, this decrease is sustained in KET + PRO. PRO and KET + PRO (but not KET) increased mTOR-Rheb colocalization; however, KET + PRO resulted in a sustained increase in mTOR-Rheb colocalization at 300 min that was greater than KET and PRO. All treatments induced a comparable sustained increase in mTOR-WGA colocalization, suggesting a similar capacity to increase in mTOR translocation to the fiber periphery, a central process in this kinase’s activation. Collectively, these molecular events may contribute to the ketone body-mediated stimulation of protein synthesis in human skeletal muscle.

DATA AVAILABILITY

Data described in the manuscript will be made available on request pending application to and approval from the corresponding author.

SUPPLEMENTAL DATA

Supplemental Table S1: https://doi.org/10.6084/m9.figshare.25664646.

GRANTS

McGill Sylvan Adams Sport Science Institute Award and Natural Sciences and Engineering Research Council of Canada Discovery Grant were awarded to T.A.C.-V.

DISCLOSURES

S.A.S is an employee of Iovate Health Sciences, a manufacturer of sports nutrition products, but did not provide nor develop the investigational supplements. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

S.J.H. and T.A.C.-V. conceived and designed research; S.J.H., J.L., J.A.M., and T.A.C.-V. performed experiments; S.J.H., Z.C.-B., and S.A.S. analyzed data; S.J.H., S.A.S., N.H., B.J.G., and T.A.C.-V. interpreted results of experiments; S.J.H. prepared figures; S.J.H. drafted manuscript; S.J.H., J.L., Z.C.-B., S.A.S., N.H., B.J.G., J.A.M., and T.A.C.-V. edited and revised manuscript; S.J.H., J.L., Z.C.-B., S.A.S., N.H., B.J.G., J.A.M., and T.A.C.-V. approved final version of manuscript.