Autophagy in Myf5+ progenitors regulates energy and glucose homeostasis through control of brown fat and skeletal muscle development

Autophagy in Myf5+ progenitors regulates energy and glucose homeostasis through control of brown fat and skeletal muscle development Atg7 deletion in Myf5+ progenitors blocks autophagy in brown adipose tissue and muscle, affecting their differentiation and function. Knockout mice have higher body temperatures and glucose intolerance, underscoring the importance of autophagy in these processes.


INTRODUCTION
The metabolic syndrome is a major health issue affecting B25% of the US population [1]. While disturbances in energy balance contribute to the metabolic syndrome, the mechanisms leading to energy imbalance are unclear. Adipose tissues and skeletal muscle (SKM) have pivotal roles in regulating energy and glucose homeostasis [2]. Excess energy is stored as lipid in white adipose tissue (WAT), whereas brown adipose tissue (BAT) expends energy by generating heat [3]. SKM [2] and BAT [4] maintain glucose homeostasis via glucose uptake in response to insulin, and intriguingly, both tissues originate from myogenic factor 5-positive (Myf5þ) progenitors [3]. It is thus conceivable that factors affecting Myf5þ progenitors will dysregulate energy balance through effects on BAT and SKM differentiation.
Macroautophagy (MA) entails formation of LC3-II-positive autophagosomes that sequester and target cytoplasmic cargo for lysosomal degradation [5]. In addition to quality control, MA regulates lipid metabolism by degrading lipid droplets (LD) via lipophagy [6]. Overnutrition and aging decrease MA in liver [6] and hypothalamic neurons [7], respectively, suggesting that metabolic defects in these conditions occur, in part, from reduced MA. MA also controls energy balance by regulating WAT differentiation [8]. Loss of a key MA gene, Atg7, in aP2þ adipocytes decreases WAT differentiation [8], and remarkably, Atg7 À / À WAT acquires BATlike features [8]. As MA modulates WAT development, we asked whether MA in Myf5þ progenitors controls BAT development. Here we show that mice lacking Atg7 in Myf5þ progenitors (Knock out, KO) show loss of MA in Myf5-derived tissues, BAT and SKM. Loss of Atg7 disrupts BAT differentiation, and surprisingly, promotes 'Beige' (brown adipocyte-like) cell [9] development in inguinal (ing) WAT that contributes to increased energy expenditure and raised body temperature. KO mice show reduced SKM differentiation and mass and are glucose intolerant, thus revealing a key role for MA in Myf5þ progenitors in regulating energy and glucose homeostasis through effects on BAT and SKM development.

RESULTS AND DISCUSSION Loss of Atg7 in Myf5þ cells disrupts MA in BAT/SKM
To determine the effect of loss of MA during BAT development, we knocked out Atg7 in Myf5þ progenitors by crossing Atg7 Flox/Flox [10] with Myf5-Cre mice [11]. KO mice displayed absence of ATG7, decreased pre-autophagosome-associated ATG5-ATG12 levels, LC3-I accumulation and loss of autophagosome-bound LC3-II in BAT and SKM (EDL, extensor digitorum longus; Fig 1A) without modifying those in epididymal (e) WAT or heart (Fig 1B,C). Atg7 deletion in BAT and SKM was verified by qPCR analyses for diminished Atg7 expression (Fig 1D), while those in eWAT ( Fig 1D) or heart (supplementary Fig S1A online) remained unaffected. Atg5 expression was comparable in tissues from control (Con) and KO mice ( Fig 1D). Moreover, ATG5-ATG12 and LC3-II levels remained equivalent in spleen, liver, lung, kidney, mediobasal hypothalamus (MBH) and perinephric fat from Con and KO mice ( Fig 1E). As small subsets of progenitors in ingWAT and eWAT express myf5 [12], we failed to detect Atg7 deletion in WAT from KO mice (Fig 1B). In fact, compensatory increases in ATG7 levels were detected in eWAT from KO mice (supplementary Fig S1B online), although increases in ATG7 did not enhance MA flux (not shown). Despite increased Atg7 expression in ingWAT (supplementary Fig S1C online), ATG7 levels remained comparable in ingWAT from Con and KO mice (supplementary Fig S1B online).

Autophagy in Myf5þ progenitors
N. Martinez-Lopez et al scientific report online) maintained lower body weights, although high-fat diet (HFD)-fed male Con and KO mice acquired comparable weights ( Fig 1G). Decreased body weights in KO mice were largely from reduced lean mass as determined by quantitative NMR (qNMR; Fig 1H), although analyses of organ weights revealed decreased BAT ( Fig 1I) and SKM weights (Figs 1J,K), and a trend towards smaller eWAT pads ( Fig 1L). HFD-fed KO mice also displayed reduced lean mass when compared with Con (supplementary Fig S1E online). Furthermore, RD-( Fig 1M) or HFD-fed mice (supplementary Fig S1E online) did not redistribute fat between their visceral and subcutaneous depots.

KO mice display impaired BAT differentiation
To determine the effect of loss of MA in Myf5þ progenitors on BAT, we subjected BAT from Con and KO mice to qPCR analysis for BAT-and adipose-selective genes. KO BAT displayed decreased expression of BAT genes, ucp1, cidea, elovl3, prdm16 and zic1, and adipose genes, c/ebpa, c/ebpb, pparg and ap2 without modifying pgc1a, a transcriptional coactivator of BAT genes (Fig 2A). KO BAT also displayed B40% reduction in adrenergic b3 receptor (adb3) expression (Fig 2A), suggesting an attenuated ability to respond to catecholamines. In contrast to effects of loss of MA in aP2þ adipocytes, that is, acquisition of BAT-like features by eWAT and augmented BAT mass [8], loss of MA in Myf5þ progenitors suppressed BAT differentiation. Surprisingly, KO BAT displayed increased expression of additional UCP family members, ucp2 and ucp3 ( Fig 2B). As heat production is UCP1 dependent [13], the significance of increased ucp2/ucp3 expression remains unclear. Loss of ATG7 in Myf5þ progenitors did not modify eWAT differentiation indicated by comparable c/ebpa, pparg and ap2 expression in Con and KO mice ( Fig 2C). We verified that changes in mRNA expression in KO BAT correlated with protein levels. Indeed, KO BAT showed decreased levels of C/EPBa, C/EBPb, PPARg, fatty acid synthase (FAS), UCP1 and the mitochondrial marker cytochrome oxidase (COX) compared with Con BAT (Fig 2D). In contrast, C/EPBa, PPARg, perilipin (PLIN)1, PLIN3, FAS, stearoyl CoA desaturase 1 (SCD1), aP2 and GLUT4 levels remained intact in eWAT from KO mice ( Fig 2E), demonstrating selective impairment in BAT differentiation. Electron microscopic analyses of KO BAT verified decreased mitochondrial number and size with regions of mitochondrial destruction between areas of preserved mitochondria ( As Myf5þ progenitors give rise to BAT and SKM, we asked whether loss of MA in Myf5þ progenitors skewed the differentiation of these cells towards SKM. To test this, Con and KO BAT were analyzed for factors regulating muscle differentiation, that is, pax7 and pax3 that control the population of proliferative myogenic myf5þ cells, myf5, myod (myoblast determination protein) and myog (myogenin), which regulates conversion of myoblasts into myocytes [14]. Con and KO BAT had comparable pax7, myf5, myod and myog expression (Fig 2G), while pax3 remained undetectable (not shown). To analyze the fate of BAT derived from Atg7 À / À Myf5þ cells, Con and KO BAT were subjected to hematoxylin and eosin (H&E) staining, which revealed intense eosinophilic cytoplasm, increased LD and adipocyte size, and decreased LD number/cell in KO BAT indicating a departure from the typical features of BAT ( Fig 2H). Loss of MA in Myf5þ progenitors also impacted cold-induced BAT gene expression. BAT from cold-exposed (B4 1C for 75 min) KO mice failed to upregulate ucp1, cox4, cidea, elovl3 and adb3 genes to levels achieved by Con ( Fig 2J). KO mice were also deficient in their ability to reduce LD content in BAT indicating impaired lipid utilization (supplementary Fig S2E online).  Fig S3D online), as observed in adult KO mice ( Fig 2B). Furthermore, day 6 Atg7 À / À BAT showed altered mitochondrial morphology, that is, dilated intra-mitochondrial space and distorted mitochondrial cristae (supplementary Fig S3E  online), decreased b-oxidation rates (supplementary Fig S3F online), and increased LD content (supplementary Fig S3G online), suggesting that MA is required in the early steps of BAT development, that is, after the E16.5 stage.

Atg7 in Myf5þ cells in early BAT development
To determine whether acutely inhibiting MA impacts BAT differentiation in adult mice, we injected BAT of Atg7 Flox/Flox mice with Cre-expressing adenoviruses (Cre AdV) or an empty vector, and mice were killed after 5 days following an acute cold stress. Cre AdV injections decreased BAT Atg7 mRNA by B30% (supplementary Fig S3H online) possibly from reduced accessibility of viruses into the entire BAT pad. This acute reduction of Atg7 expression decreased ucp1 and elovl3 expression (supplementary Fig S3I online) without modifying ucp2 or ucp3 expression (supplementary Fig S3J online). Bodipy stains from cold-exposed Cre AdV-injected mice revealed increased LD content (supplementary Fig S3K online) as observed in KO mice (supplementary Fig S2E online), suggesting that in addition to its role in early BAT development, MA controls BAT differentiation and lipid metabolism during adulthood. It is thus likely that postdevelopmental changes in MA, such as with age [7], will alter BAT differentiation and lipid metabolism.

KO mice exhibit increased body temperature
To test the physiological outcome of impaired BAT differentiation, Con and KO mice were subjected to core body temperature analyses. Surprisingly, despite abnormalities in the molecular signature of BAT, KO mice maintained higher body temperature at basal conditions and during cold exposure (Fig 3A). To explore the mechanism for increased body temperature, we asked whether constitutive increases in energy expenditure raised body  Fig S4F online) indicating sustained fat oxidation during both cycles. Higher energy expenditure did not occur from increased locomotor activity (Fig 3H), in fact, HFD-fed KO mice displayed decreased dark cycle z axis movements compared with Con mice (supplementary Fig S4G online). Values are mean ± s.e. ***Po0.001.

Autophagy in Myf5þ progenitors
N. Martinez-Lopez et al scientific report As KO mice exhibited smaller eWAT pads and reduced RER, we asked whether constitutive increases in WAT lipolysis provided the lipid fuel to sustain higher energy expenditure rates in KO mice. Indeed, KO mice displayed smaller white adipocytes (Fig 3I) and B2.5-fold increase in adb3 expression (Fig 3J) in a compensatory response to maintain adrenergic signaling. Coldexposed KO mice also increased their adb3 expression in ingWAT by B30% (Fig 3K). Although Con and KO mice showed equivalent basal serum-free fatty acid (FFA) and glycerol levels (Fig 3L), KO mice exhibited modest increases in circulating FFA and glycerols in response to intraperitoneal (i.p.) isoproterenol (Fig 3L), and significantly elevated serum FFA following cold exposure (Fig 3M). These results allow us to speculate that WAT lipolysis-driven increases in FFA availability/oxidation probably contribute to raised body temperature in KO mice.

'Beige' cells/BAT increase energy expenditure in KO mice
To identify the tissues that oxidized WAT-derived FFA in KO mice, we asked whether defective MA in BAT triggered 'Beige' cell [9] development in WAT. Acute depletion of ATG7 in BAT (via Cre AdV) did not modify basal or cold-induced expression of 'Beige' genes, tmem26 or tbx1, in eWAT (Fig 4A,B). In contrast, ATG7 depletion led to B1.5-fold increase in basal tbx1 expression in ingWAT ( Fig 4C) and an approximately two-to threefold increase in tmem26 and tbx1 expression following cold exposure (Fig 4D). H&E stains of ingWAT from Cre AdV-injected mice confirmed presence of multi-loculated brown adipocyte-like cells (Fig 4E) that increased with cold exposure (Fig 4F; supplementary Fig S5A online).
To determine whether KO mice displayed 'Beige' cell development, we subjected WAT from Con and KO mice to qPCR analyses for 'Beige' genes [9]. As expected, we observed significantly increased expression of tmem26 and tbx1 in ingWAT ( Fig 4G) but not eWAT (not shown) from 4-month (mo)-old coldexposed KO mice. In fact, eWAT from 10-mo-old KO mice displayed decreased basal tmem26 and tbx1 expression (Fig 4H), while those of brown adipocyte genes, hspb7, fbxo31, eva1 and ebf3 [9], remained intact (Fig 4I). Given that small subsets of adipocyte progenitors in WAT express myf5 [12], we speculate that loss of Atg7 in a pool of eWAT-resident Myf5þ cells impacted 'Beige' cell development in eWAT, while 'Beige' cells in iWAT possibly originate from redundant lineages and thus remained intact. In consistency with 'Beige' cell development, ingWAT but not eWAT, from KO mice displayed approximately As SKM participates in thermogenesis [15], we asked whether FFA oxidation in SKM contributed to increased energy expenditure in KO mice. Despite comparable COX levels in various SKM groups from Con and KO mice (not shown), soleus from KO mice displayed higher COX levels (supplementary Fig S5B online). Soleus (Fig 4K), and not gastrocnemius (GA; supplementary Fig S5C online), from cold-exposed KO mice displayed increased cox4, nd1 (subunit of NADH dehydrogenase) and pgc1a expression, while cpt1b and cpt2 (fatty acid translocase) or ucp2 and ucp3 (supplementary Fig S5D online) remained comparable to Con. Increased b-oxidation in soleal explants verified their contribution to increased energy expenditure in KO mice (Fig 4L).
Equivalent b-oxidation rates in liver (supplementary Fig S5E  online) from Con and KO mice excluded its role in increasing energy expenditure. Surprisingly, BAT from KO mice displayed increased b-oxidation compared with Con mice (Fig 4M). Despite the apparent defect in utilizing intrinsic lipid stores (supplementary Fig S2E online), Atg7 À / À BAT from adult mice maintained higher b-oxidation rates, in all likelihood, from WATderived FFA (modeled in supplementary Fig S7 online). Indeed, in contrast to reduced b-oxidation in KO BAT from pups (supplementary Fig S3F online), Atg7 À / À BAT from adult mice displayed increased b-oxidation ( Fig 4M) in a likely compensatory mechanism to meet thermogenic requirements in adults.

Smaller myofibers and glucose intolerance in KO mice
Despite increases in energy expenditure, KO mice remained hyperglycemic (Fig 5A), euinsulinemic (Fig 5B), and displayed defective glucose clearance (Fig 5C,D) and insulin insensitivity (Fig 5E). Since MA maintains SKM mass [16] and glucose homeostasis [17], we asked whether loss of MA in Myf5þ progenitors affected myofiber size and, in turn, glucose homeostasis. In consistency with reduced MA in EDL ( Fig 1B); soleus, TA and GA from KO mice also displayed defects in MA (Fig 5F). KO mice presented with reduced GA myofiber crosssectional area by B25% (Fig 5G), absent centralized myonuclei ( Fig 5G) and reduced expression of atrophy markers, MuRF-1 and Atrogin-1 (Fig 5H), indicating absence of SKM degeneration [16]. Reduced myofiber size probably resulted from defective SKM differentiation, indicated by decreased expression of differentiated SKM marker, creatine kinase muscle and raised levels of myod and myog (Fig 5I), while pax7, pax3 and myf5 remained intact. Embryonic loss of Myf5þ cells fails to suppress myogenesis, suggesting significant contributions to muscle development from Myf5-independent lineages [18]. Indeed, Myf5þ cells contribute to adult myonuclei by B50% [18], and consequently, loss of Atg7 in Myf5þ progenitors modestly affected myocyte size supporting the previously described contribution to SKM development from both Myf5þ and Myf5-independent lineages [18]. The percentage and/or selectivity of myocytes that show defective autophagy in each SKM group following loss of Atg7 in Myf5þ progenitors remains to be seen.
To identify the tissues contributing to glucose intolerance, Con and KO mice fed HFD for 2 weeks were subjected to i.p. insulin (1 U per kg of body weight per 30 min), and SKM and fat were analyzed for Akt phosphorylation (P-Akt). While BAT, EDL and GA from KO mice displayed decreased P-Akt (Fig 5J; supplementary Fig S6A online), soleus and ingWAT (Fig 5J; supplementary Fig S6A online) presented with increased P-Akt. Consequently, 2-deoxyglucose uptake assays revealed modest increases in glucose uptake by soleus (supplementary Fig S6B  online). Intriguingly, loss of Atg7 in Myf5þ progenitors decreased irs1 and irs2 expression in GA without modifying those in eWAT (Fig 5K,L) or affecting insulin receptor expression in SKM or eWAT (supplementary Fig S6C online). Changes in irs1 and irs2 expression probably impacted glucose clearance in KO mice, although our findings cannot distinguish whether decreased irs1/irs2 expression occurred from defective SKM differentiation or from loss of MA per se. It remains possible that persistently raised mitochondrial oxidation introduced oxidative changes in SKM or BAT, which disrupted insulin signaling.
Whether the overall phenotype of KO mice is an effect of deficient MA or due to loss of a possible MA-independent function of ATG7 remains to be elucidated. Furthermore, how ATG7 in Myf5þ progenitors controls differentiation of progenitors into adipocytes remains unknown. It is possible that MA promotes differentiation of progenitors via its ability to modulate cellular energetic needs or eliminate regulatory proteins and/or maintain quality control. Alternatively, the established role for insulin in driving adipogenesis, and the effect of loss of MA on insulin signaling might explain why loss of Atg7 impacts adipose differentiation. Aging associates with reduced ATG7 levels [7] and it is likely that MA failure in Myf5þ progenitors with age interferes with tissue differentiation, which contributes to metabolic defects and sarcopenia. Maintaining MA activity in Myf5þ progenitors might help prevent abnormalities in glucose metabolism and/or sarcopenia observed with age.