Sestrin 3 regulation in type 2 diabetic patients and its influence on metabolism and differentiation in skeletal muscle

Abstract

In mammals, the sestrin family is composed of three stress-responsive genes. Ablation of sestrin in Drosophila attenuates longevity, which is accompanied by increased S6K phosphorylation and decreased AMPK phosphorylation. Nevertheless, the metabolic role of sestrins in mammals is not comprehensively understood. We characterized the expression of individual sestrin family members and determined their role in vastus lateralis muscle biopsies from participants with normal glucose tolerance (NGT) or type 2 diabetes (T2D). Expression of sestrin 1 or sestrin 2 mRNA was unaltered between the NGT and T2D participants. Conversely, sestrin 3 mRNA was increased in T2D patients and correlated with fasting plasma glucose, 2-h postprandial plasma glucose and HbA_1c. A trend for increased sestrin 3 protein was observed in T2D patients. In human primary myotubes, sestrin 3 mRNA increased during differentiation, and this response was unaltered in T2D-derived myotubes. Long-term treatment of myotubes with insulin or AICAR decreased sestrin 3 mRNA. Exposure of myotubes to the reactive oxygen species H₂O₂ increased mRNA expression of sestrin 1 and 2, whereas sestrin 3 was unaltered. siRNA-mediated silencing of sestrin 3 in myotubes was without effect on insulin-stimulated glucose incorporation into glycogen or AICAR-stimulated palmitate oxidation. These results provide evidence against sestrin 3 in the direct control of glucose or lipid metabolism in human skeletal muscle. However, siRNA-mediated sestrin 3 gene silencing in myotubes increased myostatin expression. Collectively, our results indicate sestrin 3 is upregulated in T2D and could influence skeletal muscle differentiation without altering glucose and lipid metabolism.

type 2 diabetes (T2D) is a chronic, progressive disease that arises from defects in glucose and lipid metabolism. Insulin resistance is a key feature of T2D and can be manifested at the molecular level through impairments in insulin signaling (14) and defects in energy-sensing pathways (23). AMP-activated protein kinase (AMPK) and mammalian target of rapamycin (mTOR) are cellular energy sensors that respond in a reciprocal manner to environmental cues, such as nutrient supply or cellular stress, to control metabolism and growth in insulin-sensitive tissues (12, 26). Actions of AMPK and mTOR are altered in insulin target tissues from animal models of T2D (17). Thus, the identification and characterization of molecular integrators that titer AMPK and mTOR signaling in response to environmental cues may provide insight into the prevention of insulin resistance in T2D.

Sestrins have been identified as intermediates that integrate signal transduction through AMPK and mTOR (15). Consequently, sestrins have emerged as candidates involved in the development of insulin resistance due to their ability to mediate energy-sensing pathways (6). Originally, sestrins were recognized for their importance in detoxifying reactive oxygen species (ROS) through regeneration of peroxiredoxins (8). Three sestrin family members have been identified in humans and mice. Sestrin 1 (or p53-activated gene number 26; PA26) was identified in a human osteosarcoma cell line by using a tetracycline-regulated approach to activate p53 (4, 25). Sestrin 2 (or hypoxia-inducible gene number 95; HI95) was detected in a hypoxia screen microarray by using a human glioma cell line (9). Structural analysis of sestrin 1 led to the discovery of sestrin 3 (22). Regulation of sestrin 1 and sestrin 2 occurs mainly through p53-depedendent pathways (4, 13). Sestrin 3 activity is influenced by the transcriptional regulator FoxO1, a conserved downstream phosphorylation target of Akt (10, 21). Sestrin 1 and sestrin 2 inhibit mTOR complex 1 (mTORC1) by increasing AMPK phosphorylation (6). In vitro expression of sestrin 3 in mouse embryonic fibroblasts reduces mTORC1 activity, as evident by a reduction in S6K phosphorylation (21). Since altered AMPK and mTOR signaling is implicated in the development of insulin resistance, sestrins may play a role in the pathogenesis of T2D.

Presently, the metabolic role of sestrins in humans is unknown. Genetic ablation of Drosophila sestrin (dSesn) leads to age-associated pathologies associated with metabolic disease, including triglyceride accumulation, increased ROS, mitochondrial dysfunction, skeletal muscle degeneration, and cardiac malfunction (15). Whether individual sestrin isoforms differentially control skeletal muscle metabolism in humans is unclear. Due to the potential importance of sestrins in muscle growth and metabolic homeostasis (15), we focused our studies on human skeletal muscle and tested the hypothesis that sestrins play a role in glucose and lipid metabolism. We measured the expression of the three sestrin family members in skeletal muscle biopsies from participants with normal glucose tolerance (NGT) or T2D. Given the role of sestrins in detoxifying ROS, we determined the effect of the oxidizing agent H₂O₂ on sestrin mRNA expression in cultured human myotubes. Finally, we determined the effects of sestrin 3 siRNA-mediated gene silencing on glucose and lipid metabolism and skeletal differentiation.

MATERIALS AND METHODS

Study participants and skeletal muscle biopsies.

Participants were recruited and were classified as either NGT or T2D based on fasting glucose, 2-h postprandial glucose, and HbA_1c. Participants were matched for age, BMI, and waist circumference (Table 1). Vastus lateralis muscle biopsies were obtained as described (11). The nature, purpose, and possible risks of the study were explained to all participants before informed consent was obtained. The investigation was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Karolinska Institutet.

Table 1.

Clinical characteristics of the NGT and T2D participants

	NGT	T2D	P Value
n	12	10
Sex, M/F	5/7	6/4
Age, yr	58 ± 2	61 ± 1	NS
BMI, kg/m2	30.0 ± 0.6	29.2 ± 1.0	NS
Waist circumference, cm	98.4 ± 2.3	101.5 ± 2.6	NS
Fasting glucose, mM	5.6 ± 0.2	8.8 ± 0.9	<0.05
2-h Postprandial glucose, mM	7.1 ± 0.4	16.4 ± 1.2	<0.05
HbA1c, %	4.8 ± 0.1	6.0 ± 0.3	<0.05

Results are means ± SE for normal glucose tolerant (NGT) vs. type 2 diabetic (T2D).

Materials.

Reagents for cell culture including DMEM-F12 + GlutaMAX1 (31331), DMEM + GlutaMAX 1 g/l glucose (21885), fetal bovine serum (10270), penicillin-streptomycin (15140), and fungizone (15290) were from Invitrogen (Carlsbad, CA). siRNA against human sestrin 3 (L-018289-01) was from Dharmacon (Lafayette, CO). TaqMan assays for human sestrin 3 (Hs00376220_m1), human sestrin 2 (Hs00900115_m1), human sestrin 1 (Hs00902787_m1), human myostatin (Hs00976237_m1), and β₂-microglobulin (4310886E), β-actin (4310881E), and 18S (4319413E) were from Applied Biosystems (Foster City, CA). An antibody against sestrin 3 was from Abcam (Cambridge, MA). Antibodies against pAMPK-Thr¹⁷², pS6K-Thr³⁸⁹, pFoxO1-Ser²⁵⁶, and GAPDH were from Cell Signaling Technology (Beverly, MA). An antibody against p53 was from Santa Cruz Biotechnology (Santa Cruz, CA).

RNA extraction, cDNA synthesis, and quantitative real-time PCR.

Total RNA was extracted using the RNeasy kit from Qiagen (Hilden, Germany). Cells were lysed, and RNA was precipitated using 70% ethanol. RNA was eluted in RNase-free water using spin columns. To prevent DNA contamination, an on-column DNase digestion was performed. RNA content and quality were assessed using a NanoDrop spectrophotometer from Thermo Fischer Scientific (Waltham, MA). cDNA was generated by using the High Capacity Reverse Transcription Kit (Applied Biosystems). PCR reactions contained cDNA and TaqMan Gene Expression Master Mix/TaqMan Fast Advanced Master Mix (Applied Biosystems), TaqMan primers, and RNase-free water. Multiple housekeeping genes were tested (β₂-microglobulin, 18S, and β-actin) and β₂-microglobulin was the most stable. PCR reactions were performed and monitored on a StepOnePlus qRT-PCR machine (Applied Biosystems). Gene expression data were compared using the ΔΔC_T method.

Western blot analysis.

Protein lysates derived from muscle biopsies or cultured myotubes were mixed with Laemmli buffer. Proteins were resolved using SDS-PAGE following transfer of protein to PVDF membranes. Phosphorylated and total protein was visualized using antibodies followed by enhanced chemiluminescence.

Human skeletal muscle cultures and siRNA-mediated gene silencing.

Satellite cells were isolated from vastus lateralis or rectus abdominus muscle biopsies. Human muscle cells were also purchased from Lonza Group (Switzerland). Satellite cell-derived myoblasts and Lonza muscle cells were grown to confluence in DMEM-F12 + GlutaMAX1 supplemented with 20% fetal bovine serum, penicillin-streptomycin, and fungizone under 7.5% CO₂. Myotube differentiation was initiated by replacing growth medium with DMEM + GlutaMAX 1 g/l glucose supplemented with 2% fetal bovine serum, penicillin-streptomycin, and fungizone. Experiments in satellite cell-derived myotubes or Lonza muscle cells were performed 7 days or 4 days after differentiation, respectively. For long-term studies of myotubes, differentiation of myoblasts to myotubes began with an 8-day treatment without or with insulin (400 pM), AICAR (200 μM), or glucose (25 mM). For siRNA-mediated sestrin 3 gene silencing, Lonza skeletal muscle cells were cultured in a six-well plate at 70% confluence. On day 1, cells were incubated with 80 pmol of scramble siRNA or siRNA against Sestrin 3 complexed with transfection reagent Lipofectamine 2000 for 6 h. After transfection, cells were placed in differentiation medium. On day 3, the transfection performed on day 0 was repeated. Cells were used for experiments on day 4.

Insulin-stimulated glucose incorporation into glycogen.

Human primary myotubes were serum starved for 4 h. Myotubes were then incubated in the absence or presence of 120 nM insulin for 30 min and thereafter incubated with d-[U-¹⁴C]glucose for 90 min. Cells were lysed in 0.03% SDS, and cold glycogen was added. The mixture was boiled at 99°C for 1 h, and glycogen was precipitated overnight at −20°C by the addition of 100% ethanol. The radioactivity in the collected pellet was measured using a WinSpectral 1414 liquid scintillation counter. Data were normalized to protein content.

AICAR-stimulated palmitate oxidation.

Human primary myotubes were serum starved for 2 h and then incubated in the absence or presence of 2 mM AICAR for 6 h in medium containing [9,10-³H(N)]palmitic acid. The myotube supernatant was incubated with charcoal to remove the remaining nonoxidized palmitate. In the remaining supernatant, the radioactivity of ³H₂O was measured using the WinSpectral 1414 liquid scintillation counter. Data were normalized to protein content.

Statistical analysis.

Data are presented as means ± SE. Differences were analyzed using Student's t-test or two-way ANOVA. Data were considered significantly different when P < 0.05. Statistical comparisons were performed using SPSS.

RESULTS

Skeletal muscle sestrin 3 mRNA expression is increased in T2D and correlated with clinical markers of glucose homeostasis.

mRNA expression of sestrin 1, sestrin 2, and sestrin 3 was determined in skeletal muscle biopsies from the age- and BMI-matched NGT and T2D participants. Expression of sestrin 1 and sestrin 2 was unaltered between the NGT and T2D participants, whereas expression of sestrin 3 was increased 41% in T2D patients (P < 0.05; Fig. 1). A trend for increased sestrin 3 protein abundance was noted in skeletal muscle from T2D patients (Fig. 1, D and E), suggesting changes in mRNA-protein dynamics in T2D. We next examined the relationship between clinical measures of glucose homeostasis and sestrin 3 expression. Sestrin 3 mRNA expression correlated with fasting blood glucose, 2-h postprandial blood glucose and HbA_1c (Fig. 2A). Skeletal muscle sestrin 1 mRNA expression correlated with sestrin 2 mRNA expression (Fig. 2B). Conversely, sestrin 3 mRNA expression did not correlate with either sestrin 1 or sestrin 2 mRNA expression (Fig. 2B).

Fig. 1.

Sestrin expression in skeletal muscle from normal glucose tolerant (NGT) and type 2 diabetic (T2D) participants. mRNA expression of sestrin 1 (A), sestrin 2 (B), and sestrin 3 (C) was determined in skeletal muscle biopsies from NGT (n = 12, open bar) or T2D (n = 10, filled bar) participants. mRNA was normalized to β₂-microglobulin. D: sestrin 3 protein abundance was measured in skeletal muscle biopsies from the NGT (n = 10) and T2D (n = 11) participants. E: quantifcation of sestrin 3 protein abundance obtained under D. *P < 0.05 vs. NGT.

Fig. 2.

Correlation between sestrin expression and clinical parameters. A: expression of sestrin 3 was correlated with HbA_1c, basal glucose (mM), and 2-h postprandial glucose (mM). B: correlation between sestrin 1 and sestrin 2 mRNA. Correlation between sestrin 3 and sestrin 1 or sestrin 2 mRNA. mRNA expression data were normalized to β₂-microglobulin. NGT and T2D participants are indicated as filled and open circles, respectively.

Sestrin 3 expression in primary human myotubes.

Primary human skeletal muscle cultures were used to investigate the direct role of sestrin 3 in myotube differentiation. Satellite cell-derived myoblasts from skeletal muscle biopsies obtained from the NGT and T2D participants were propagated, passaged, and differentiated into myotubes. Sestrin 3 mRNA expression was increased in myotubes differentiated from primary human myoblasts. This response was similar between NGT and T2D (Fig. 3A). Long-term treatment of differentiated human myotubes with antidiabetic agents alters insulin action (2). Thus, we investigated whether insulin, AICAR, or high-glucose exposure for 8 days altered sestrin 3 expression in cultured myoblasts during differentiation to myotubes. Long-term insulin or AICAR treatment decreased sestrin 3 mRNA (Fig. 3B). Conversely, sestrin 3 mRNA was unaltered by high-glucose treatment.

Fig. 3.

Sestrin 3 expression during myotube differentiation. A: sestrin 3 mRNA was examined in primary skeletal muscle cells derived from NGT or T2D participants. Primary skeletal muscle cells were studied as undifferentiated myoblasts or differentiated myotubes. *P < 0.05 for myotubes vs. corresponding myoblasts. B: myotubes were left untreated (basal) or incubated for 8 days with 400 pM insulin, 200 μM AICAR, or 25 mM glucose. mRNA expression data were normalized to β₂-microglobulin; n = 6. *P < 0.05 vs. basal.

Effect of the ROS H₂O₂ on sestrin mRNA expression.

To probe the antioxidant function of sestrin isoforms in primary human muscle cells, myotubes were treated with the oxidizing agent H₂O₂ for 2 or 6 h, and sestrin isoform mRNA expression was determined. Sestrin 1 and sestrin 2 mRNA expression was increased, whereas sestrin 3 mRNA expression was unchanged after the 6-h H₂O₂ treatment (Fig. 4A). H₂O₂ treatment increased p53 protein expression and decreased S6K-Thr³⁸⁹ phosphorylation, whereas FoxO1-Ser²⁵⁶ phosphorylation was unaltered (Fig. 4B). GAPDH was analyzed to confirm equal sample loading (Fig. 4B).

Fig. 4.

Effect of the reactive oxygen species (ROS) H₂O₂ on sestrin mRNA expression. Differentiated myotubes were incubated in the absence (time 0) or presence of 1 mM H₂O₂ for 2 or 6 h. A: mRNA expression of sestrin 1, sestrin 2, and sestrin 3 was determined and normalized to β₂-microglobulin. B: protein abundance of p53, phosphorylated S6K-Thr³⁸⁹, FoxO1-Ser²⁵⁶, and GAPDH were visualized by Western blot analysis; n = 3, *P < 0.05 vs. time 0.

Effect of siRNA-mediated sestrin 3 gene silencing on substrate metabolism and signaling.

The role of sestrin 3 in glucose and lipid metabolism was examined following sestrin 3 silencing in primary human myotubes. siRNA-mediated gene silencing reduced sestrin 3 mRNA expression by 58 ± 3.1% (P < 0.05; Fig. 5A) and protein abundance by 45 ± 4.6% (P < 0.05; Fig. 5B) compared with human myotubes transfected with a scramble siRNA. Insulin increased glucose incorporation into glycogen in the differentiated myotubes. siRNA-mediated silencing of sestrin 3 did not alter glycogen synthesis under basal or insulin-stimulated conditions (Fig. 5C). AICAR increased palmitate oxidation in differentiated myotubes. siRNA-mediated gene silencing of sestrin 3 did not alter basal or AICAR-stimulated palmitate oxidation (Fig. 5D). We also examined AMPK and mTOR signaling pathways in cultured myotubes after silencing of sestrin 3. Phosphorylation of AMPK-Thr¹⁷² and S6K-Thr³⁸⁹ was determined as a measure of AMPK and mTOR signaling, respectively. In addition, the link between sestrin 3 and its upstream regulator FoxO1 was evaluated through phosphorylation of pFoxO1-Ser²⁵⁶. Sestrin 3 knockdown did not alter pAMPK-Thr¹⁷², pS6K-Thr³⁸⁹, or pFoxO1-Ser²⁵⁶ phosphorylation in cultured myotubes incubated in the absence or presence of AICAR (Fig. 5E). To further determine a role for sestrin 3 in the regulation of cell growth, myostatin mRNA expression was determined following sestrin 3 silencing to investigate a role for sestrin isoforms in muscle growth. siRNA-mediated sestrin 3 gene silencing increased myostatin expression in differentiated myotubes 1.5-fold (Fig. 5F).

Fig. 5.

Effect of siRNA-mediated sestrin 3 gene silencing on skeletal muscle metabolism. Human myotubes were transfected with scramble siRNA (Scr siRNA) or siRNA against human sestrin 3 (Sesn3 siRNA). A: sestrin 3 mRNA expression normalized to β₂-microglobulin. *P < 0.05 vs. Scr siRNA transfected cells. B: representative immunoblot showing sestrin 3 protein abundance. C: basal (open bar) and insulin-stimulated (120 nM; filled bar) glycogen synthesis. *P < 0.05 vs. basal. D: Basal (open bar) and AICAR-stimulated (2 mM; filled bar) palmitate oxidation. *P < 0.05 vs. basal. E: representative immunoblots showing protein abundance of phosphorylated AMPK-Thr¹⁷², S6K-Thr³⁸⁹, FoxO1-Ser²⁵⁶, and GAPDH in myotubes incubated for 6 h in the absence (Basal) or presence of AICAR (2 mM). F: myostatin mRNA was measured via real-time qPCR; n = 3. *P < 0.05 vs. Scr siRNA transfected.

DISCUSSION

Sestrins are stress-responsive genes that balance AMPK and mTOR signaling. A link between sestrin expression and metabolic homeostasis is evident from gene ablation of dSesn in Drosophila (15). dSesn provides a feedback loop to prevent excessive TORC1 activation and ROS accumulation and prevention against age-related metabolic disorders of lipid metabolism and skeletal muscle degeneration (15). Mammalian sestrins also play an important role in metabolic control. In mice fed a high-fat diet, sestrin 2 protein is increased in liver and skeletal muscle, whereas sestrin 3 abundance is increased only in skeletal muscle (16). In contrast to Drosophila, where dSesn ablation has a protective effect on metabolism, sestrin 2 ablation in mice exacerbates obesity-induced mTORC1-S6K activation, glucose intolerance, insulin resistance, and hepatosteatosis (16). Interestingly, peripheral and hepatic metabolism is unaltered in lean sestrin 2 knockout mice, indicating that sestrin 1 or 3 may play a compensatory role. Indeed, double knockout of sestrin 2 and sestrin 3 provokes hepatic mTORC1-S6K activation and peripheral insulin resistance in lean mice fed a chow diet (16), highlighting an important role of sestrin 3 in metabolism. Collectively, these experimental studies in genetically modified model organisms support a role for sestrins in metabolic homeostasis, but clinical evidence is lacking.

Here, we report that the sestrin 3 transcript is elevated in skeletal muscle from T2D patients vs. BMI-matched NGT participants, which is mirrored by a trend toward increased protein abundance. Sestrin 1 and 2 expression is unaltered. This contrasts with findings of elevated sestrin 2 mRNA and protein abundance in liver and skeletal muscle from obese high-fat diet-fed rodents (16). Skeletal muscle sestrin 3 mRNA expression in T2D patients correlated with fasting blood glucose, 2-h postprandial blood glucose and HbA_1c levels, implying a relationship with the degree of insulin sensitivity. Since a clear elevation in protein levels among T2D patients was not observed, the mechanism by which T2D impacts sestrin 3 mRNA translation to protein remains uncertain. Increased plasma glucose is a clinical hallmark of T2D. However, long-term treatment of primary human myotubes with high glucose did not alter sestrin 3 mRNA expression, suggesting that factors other than hyperglycemia account for the increase skeletal muscle sestrin 3 expression in T2D patients. Treatment of human myotubes with insulin or the AMPK activator AICAR decreased sestrin 3 mRNA. Given that FoxO1 regulates sestrin 3 mRNA expression through binding of the promoter region of sestrin 3 (10), the decrease in sestrin 3 mRNA after long-term insulin or AICAR treatment may be related to exclusion of phosphorylated FoxO1 from the nucleus, since both insulin-stimulated Akt- and stress-induced AMPK-mediated phosphorylation of FoxO1 lead to nuclear export of FoxO1 (24). Collectively, our findings provide further evidence of the link between sestrin 3 expression and signal transduction through Akt and AMPK pathways. Nevertheless, gene silencing of sestrin 3 in primary human myotubes did not directly affect glucose uptake or lipid oxidation. Thus, sestrin 1 and 2 may compensate for sestrin 3 depletion. Alternatively, residual sestrin 3 may be sufficient for glucose and lipid metabolism.

Sestrins protect against ROS-induced oxidative stress and cell damage via the tumor suppressor p53 (5), and this may be of physiological relevance for the regulation of insulin sensitivity (20). Given earlier evidence that overexpression of any sestrin family member is sufficient to decrease ROS in a Ras-induced ROS cell model (13), sestrin inhibition would be predicted to prevent ROS-induced insulin resistance. Our study reveals that H₂O₂ treatment of primary human myotubes increased mRNA of sestrin 1 and 2 without altering sestrin 3. These findings suggest that sestrin 1 and 2, rather than sestrin 3, are the main stress-inducible proteins involved in the suppression of ROS production and protection from oxidative stress. Understanding the role of sestrin 3 in ROS production is complicated further by two alternatively spliced forms of sestrin 3 that encode two different protein products (7, 10). Nevertheless, siRNA-mediated sestrin 3 gene silencing lowers ROS levels to that of cells treated with control siRNA (21). Collectively, these findings indicate that sestrin isoforms have specific roles in maintaining ROS status. However, the clinical implication remains unclear, since the absolute capacity for mitochondrial ROS production in skeletal muscle from either obese insulin-resistant or T2D patients is unaltered (1).

Studies from Drosophila and mouse models support an indirect connection between sestrins and muscle growth (15). Here, we report that sestrin 3 siRNA-mediated gene silencing increased myostatin expression, further implyng a role in muscle growth. Myostatin, a member of the TGFβ family, is a secreted protein produced mainly in skeletal muscle that acts as a negative regulator of muscle growth (18). Myostatin-null mice exhibit muscle hypertrophy (18) as well as partial suppression of fat accumulation and improvements in glucose metabolism in genetic animal models of obesity and diabetes (19). Conversely, in T2D patients and streptozotocin-induced diabetic rodents, myostatin mRNA expression is elevated (3). Our evidence that sestrin 3 gene silencing increased myostatin expression in cultured myotubes is consistent with increased skeletal muscle myostatin expression in T2D patients (3) and muscle degeneration arising from dSesn ablation (15). These findings imply that sestrin 3 may play a role in muscle growth by modulating myostatin expression. Aberrant growth signals may negatively impact skeletal muscle glucose metabolism and lead to the development of T2D.

In conclusion, skeletal muscle expression of sestrin 3 mRNA is elevated in T2D patients and correlated with clinical markers of markers of glucose homeostasis. Sestrin 3 knockdown in cultured human myotubes does not impact insulin sensitivity or lipid oxidation. However, sestrin 3 knockdown increases myostatin expression, implying a role in muscle growth. Molecular pathway analysis of muscle hypertrophy and factors involved in ROS production may further unravel the specific role of sestrin 3 in skeletal muscle physiology.

GRANTS

The Strategic Diabetes Program at Karolinska Institutet, European Research Council Ideas Program (ICEBERG, ERC-2008-AdG23285), Swedish Research Council, Swedish Diabetes Association, Strategic Research Foundation, Knut and Alice Wallenberg Foundation, and the Stockholm County Council supported this research.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: E.B.M.N., M.E.O., and J.R.Z. conception and design of research; E.B.M.N. and M.E.O. performed experiments; E.B.M.N., M.E.O., and J.R.Z. analyzed data; E.B.M.N., M.E.O., and J.R.Z. interpreted results of experiments; E.B.M.N. prepared figures; E.B.M.N. and J.R.Z. drafted manuscript; E.B.M.N., M.E.O., and J.R.Z. edited and revised manuscript; E.B.M.N., M.E.O., and J.R.Z. approved final version of manuscript.