Sexually dimorphic effects of SARM1 deletion on cardiac NAD⁺ metabolism and function

Abstract

Nicotinamide adenine dinucleotide (NAD⁺) decline is repeatedly observed in heart disease and its risk factors. Although strategies promoting NAD⁺ synthesis to elevate NAD⁺ levels improve cardiac function, whether inhibition of NAD⁺ consumption can be therapeutic is less investigated. In this study, we examined the role of sterile-α and TIR motif containing 1 (SARM1) NAD⁺ hydrolase in mouse hearts, using global SARM1-knockout mice (KO). Cardiac function was assessed by echocardiography in male and female KO mice and wild-type (WT) controls. Hearts were collected for biochemical, histological, and molecular analyses. We found that the cardiac NAD⁺ pool was elevated in female KO mice, but only trended to increase in male KO mice. SARM1 deletion induced changes to a greater number of NAD⁺ metabolism transcripts in male mice than in female mice. Body weights, cardiac systolic and diastolic function, and geometry showed no changes in both male and female KO mice compared with WT counterparts. Male KO mice showed a small, but significant, elevation in cardiac collagen levels compared with WT counterparts, but no difference in collagen levels was detected in female mice. The increased collagen levels were associated with greater number of altered profibrotic and senescence-associated inflammatory genes in male KO mice, but not in female KO mice.

NEW & NOTEWORTHY We examined the effects of SARM1 deletion on NAD⁺ pool, transcripts of NAD⁺ metabolism, and fibrotic pathway for the first time in mouse hearts. We observed the sexually dimorphic effects of SARM1 deletion. How these sex-dependent effects influence the outcomes of SARM1 deficiency in male and female mice in responses to cardiac stresses warrant further investigation. The elevation of cardiac NAD⁺ pool by SARM1 deletion provides evidence that targeting SARM1 may reverse disease-related NAD⁺ decline.

INTRODUCTION

Heart disease is the leading cause of deaths globally. It has several underlying risk factors such as physical inactivity, aging, diabetes, obesity, and hypertension (1). These stresses promote cardiac dysfunction and hypertrophy, driven by signaling such as metabolic derangements and profibrotic cascade. Decline in nicotinamide adenine dinucleotide (NAD⁺), an essential electron carrier in cellular metabolism, has been implicated in heart disease (2, 3). However, the role of cellular mechanisms of NAD⁺ decline in cardiac dysfunction needs further investigation.

NAD⁺ is an indispensable redox cofactor that carries electrons generated from glycolysis and Kreb’s cycle into the mitochondrial electron transport chain (4). The reduced form NADH donates electrons in oxidoreductase reactions and regenerates the oxidized form NAD⁺. The importance of NAD⁺ in cellular energetics and function supports its crucial role in cardiac function. Therefore, NAD⁺ redox imbalance and depletion contribute to the progression of cardiac dysfunction induced by genetic, hypertensive, and metabolic stresses in mice and humans (5–9). Although elevation of cardiac NAD⁺ levels by activating the NAD⁺ synthesis pathway mitigates cardiac dysfunction (6–8, 10), inhibition of NAD⁺ hydrolysis to maintain cellular NAD⁺ homeostasis can also be a therapeutic approach. Cellular NAD⁺ is degraded to nicotinamide (NAM) by enzyme classes such as PARPs and sirtuins, which also serve important cellular functions such as DNA repair and protein deacetylation (11), and by dedicated NAD⁺ hydrolases such as sterile-α and TIR motif containing 1 (SARM1), CD38, and bone marrow stromal cell antigen 1 (Bst1). Since the latter mainly serves to hydrolyze NAD⁺, they could be potential therapeutic targets against NAD⁺ decline in diseases. In this study, we explored how the deletion of one of the NAD⁺ hydrolases, SARM1, affects NAD⁺ homeostasis and cardiac function (11).

SARM1 is a mitochondria-associated, intracellular NAD⁺ hydrolase known for its role in axonal injury and degeneration (12). SARM1 is activated by low NAD⁺ levels (13, 14), and its activation leads to further NAD⁺ hydrolysis, mitochondrial dysfunction, and energetic catastrophe (15). Therefore, SARM1 deficiency is protective against axonal degeneration and in models of neuronal injury (16, 17). Inhibition of SARM1 activity could be an attractive therapeutic target for diseases involving NAD⁺ decline. However, the role of SARM1 in cardiac function has not yet been explored.

In this study, we used a global knockout mouse model of SARM1 (KO) to investigate its effect on NAD⁺ metabolism, inflammation, cardiac remodeling, structure, and function. Since sex differences often affect the risk factors for cardiovascular diseases, we factored in sex as a biological variable in our experimental design (18, 19). All the procedures and measurements were performed in both male and female mice. SARM1 deletion had no significant effects on body weight and cardiac function and geometry in both male and female mice. However, sex-specific differences were observed in the effects of SARM1 deletion on cardiac NAD⁺ metabolism and fibrosis.

METHODS

Animal Care and Experiments

All animal care and procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the Oklahoma Medical Research Foundation (Protocol No. 22-38) and were performed in accordance with IACUC regulations. C57BL/6 and SARM1-KO mice with the same background (Stock No. 018069) were obtained from The Jackson Laboratory, and 3- to 4-mo-old male and female mice of both genotypes were used for the experiments. For each experiment, n = 6 per each genotype per sex provided a sufficient sample size based on power analysis: 50% difference in NAD⁺ pool between animal groups, power at 0.8 with a standard deviation of 1.4. Randomization of animals into experimental groups was not needed since there were no treatments/interventions involved. The study included only healthy, unstressed mice of both sexes and genotypes, and there were no other exclusion criteria.

Echocardiography

Systolic function and cardiac geometry were assessed by echocardiography through parasternal long-axis view. M-mode images were recorded using VEVO 2100 system (VisualSonics) on lightly anesthetized mice when the heart rate was within 500–600 beats/min. Parameters such as fractional shortening (FS), left ventricular internal dimension at diastole (LVID;d) or systole (LVID;s), interventricular septal thickness in diastole (IVS;d), and left ventricular posterior wall thickness in diastole (LVPW;d) were measured, blindly analyzed, and calculated using the software package in the VEVO 2100 system. Parameter of diastolic function, the E′/A′ ratio was measured by tissue-Doppler imaging of mitral annulus. All measurements were made and used to calculate the above parameters by averaging three cardiac cycles.

Tissue Harvest and Processing

On the day of tissue harvest, mice were fasted for 6 h and anesthetized using isoflurane. Rib cages were cut open to expose the hearts, and blood was collected by cardiac puncture. Heart weight (HW) and tibia length (TL) were measured to calculate the heart weight-to-tibia length ratio. Cardiac tissue was collected and snap-frozen in liquid nitrogen. Frozen tissues were pulverized using a Tissuelyzer II (Qiagen) for downstream biochemical assays.

NAD⁺ Assay

Pulverized cardiac tissue (10–15 mg) was used to measure NAD⁺ and NADH levels using commercially available kit (E2ND-100; BioAssay Systems), complying with the manufacturer’s instructions. NAD⁺ pool was calculated as the sum of the levels of NAD⁺ and NADH, together expressed at NAD(H) levels (Fig. 1B).

Figure 1.

SARM1 deficiency increases cardiac NAD⁺ levels and induces sex-based changes in NAD⁺ consumption transcripts. Quantitative PCR analysis was used to confirm deletion of Sarm1. A: Sarm1 transcript levels. NAD⁺ and NADH levels were quantified by a colorimetric assay. B: NAD⁺ pool was measured in mouse hearts. Quantitative PCR analyses were used to assess transcript expression of NAD⁺ consumption enzymes in the hearts of male and female wild-type (WT) and SARM1-KO mice. Cd38 (C), Bst1 (D), Parp1 (E), Sirt1 (F), Sirt5 (G), Sirt6 (H), and Sirt7 (I) transcript levels were measured. J: heatmap summarizing NAD⁺ consumption gene expression. Expression data are represented as fold-change with respect to WT mice of same sex. *P < 0.05, **P < 0.01 vs. WT; n = 6 per group. Blue charts, male mice data; red charts, female mice data. On the heatmap, gene names with underlined text signify P < 0.05 vs. WT. KO, knockout; mRNA, messenger RNA; NAD⁺, nicotinamide adenine dinucleotide; NADH, reduced nicotinamide adenine dinucleotide; PCR, polymerase chain reaction; SARM1, sterile-α and TIR motif containing 1.

Analysis of mRNA Expression

Total RNA was extracted from pulverized cardiac tissue using RNeasy Fibrous Tissue Mini Kit (Cat. No. 74704; Qiagen). RNA concentration and quality were measured on Nanodrop (Thermo Fisher Scientific), and 500 ng RNA was used to prepare cDNA using Maxima FirstStrand cDNA Synthesis kit (Thermo Fisher Scientific). Quantitative polymerase chain reaction was carried out using Bio-Rad CFX Real-Time PCR System and Taqman probes. The data were normalized to the expression of the reference gene, myoglobin (MYB), and expressed as fold-change over wild-type (WT) control. Seaborn data visualization library was used in Python to generate heatmaps for mRNA expression data.

Histology

During tissue harvest, 1- to 2-mm-thick heart tissues were cut cross-sectionally at the midsection and fixed in 4% paraformaldehyde for 24 h. The medium was changed to 70% ethanol, and fixed tissues were embedded in paraffin blocks and sectioned. Slide-mounted tissue sections were stained with Masson’s trichrome for collagen detection. The stained sections were scanned with an Aperio Scanscope AT2 (Leica) at ×40 magnification. Color images of stained sections were captured using ImageScope and were exported as image files. Quantitative image analysis of whole stained sections was performed by an analyst blind to the experimental groups, using Aperio Bright-field Image Analysis. Toolbox software and a color deconvolution algorithm were used to measure RGB color vectors of blue (positive trichrome stain) and total stained tissue area. Cardiac fibrosis was quantified using percentages of collagen (trichrome-blue stain component) and total stained tissue area (in mm²).

Statistical Analysis

All analyses were performed using GraphPad Prism 9.0. All the observations from each group were included in the analyses, and data are expressed as means ± SE. Differences in the means of the two groups were analyzed by Student’s t test, and a P < 0.05 was considered significant.

RESULTS

SARM1 Deficiency Increases Cardiac NAD⁺ Levels in Female Mice

We first confirmed the deletion of SARM1 through quantitative PCR analysis, and Sarm1 transcript was detected in male and female WT hearts, but was not detectable in male and female KO hearts (Fig. 1A). SARM1 deficiency significantly elevated cardiac NAD⁺ pool in female mice and showed an upward trend in male mice (Fig. 1B). To check whether SARM1 deletion causes compensatory changes in other NAD⁺ hydrolases, we measured the expression of Cd38 and Bst1. Cardiac Cd38 expression was slightly but significantly upregulated in female KO mice, but not in male mice (Fig. 1C). There was no significant difference in the expression of Bst1 in both male and female KO hearts, compared with respective WT hearts (Fig. 1D). Although SARM1 is known to be activated by stress-induced phosphorylation or NAD⁺ metabolites (13, 20), our findings indicate that baseline SARM1 NAD⁺ hydrolase activity contributes to the maintenance of cardiac NAD⁺ levels.

SARM1 Deficiency Leads to More Changes in Cardiac NAD⁺ Consumption and Synthesis Transcripts in Male Mice than in Female Mice

Apart from SARM1 and other NAD⁺ hydrolases, PARPs and sirtuins are major consumers of NAD⁺ inside the cell. We observed significant downregulation of Parp1, Sirt1, Sirt5, Sirt6, and Sirt7 transcripts in male KO mice compared with male WT mice (Fig. 1, E–I). However, no significant changes were observed in the expression of these genes in female KO mice (Fig. 1, E–I). SARM1 deficiency induced changes in a greater number of NAD⁺ consumption genes in male hearts than in female hearts (Fig. 1J).

In the salvage pathway of NAD⁺ biosynthesis, nicotinamide generated from the NAD⁺ hydrolysis of SARM1 is recycled to synthesize NAD⁺. Upon SARM1 deletion, we observed upregulation of nicotinamide phosphoribosyltransferase (Nampt), the rate-limiting enzyme of the salvage synthesis pathway, in male hearts but not in female hearts (Fig. 2A). However, the transcripts of other NAD⁺ synthesis enzymes such as nicotinamide nucleotide adenylyltransferase 2 (Nmnat2), Nmnat3, nicotinamide riboside kinase 2 (Nmrk2), and quinolinate phosphoribosyl transferase (Qprt) were significantly downregulated in male SARM1-KO hearts, but not in female hearts (Fig. 2, B–E). These findings indicate that sexual dimorphism of NAD⁺ metabolic gene expression exists in responses to SARM1 deficiency (Fig. 1J and Fig. 2F).

Figure 2.

SARM1 deficiency changes the mRNA expression of NAD⁺ synthesis enzymes in male mice, but not in female mice. Quantitative PCR analyses were used to assess transcript expression of NAD⁺ synthesis enzyme transcripts in the hearts of male and female wild-type (WT) and SARM1-KO mice. Relative mRNA levels of Nampt (A), Nmnat2 (B), Nmnat3 (C), Nmrk2 (D), and Qprt (E) were measured. F: heatmap summarizing NAD⁺ synthesis gene expression. Expression data were represented as fold-change with respect to WT mice of same sex. *P < 0.05, **P < 0.01 vs. WT; n = 6 per group. Blue charts, male mice data; red charts, female mice data. On the heatmap, gene names with underlined text signify P < 0.05 vs. WT. KO, knockout; mRNA, messenger RNA; NAD⁺, nicotinamide adenine dinucleotide; NADH, reduced nicotinamide adenine dinucleotide; PCR, polymerase chain reaction; SARM1, sterile-α and TIR motif containing 1.

SARM1 Deficiency Does Not Change Body Weight, Cardiac Function, and Geometry in Male and Female Mice

Young WT and KO mice had similar body weights in both male and female cohorts (Fig. 3A). We next measured cardiac function and geometry by echocardiography. Systolic function, represented by fractional shortening (FS), showed no differences between WT and KO mice in both male and female cohorts (Fig. 3B). Similarly, diastolic function, represented by early-to-late ratio of peak diastolic velocity (E′/A′ ratio) did not differ between WT and KO mice in both male and female cohorts (Fig. 3C). Left ventricular internal dimension (LVID;d) and posterior and septum wall thicknesses (IVS;d and LVPW;d, respectively) were similar in WT and KO hearts in both male and female mice (Fig. 3D, Table 1). Heart weight-to-tibia length ratio, a marker of cardiac hypertrophy, showed no change upon SARM1 deletion in both male and female cohorts (Table 1). These results indicate that SARM1 deletion has no effects on cardiac structure and function at the baseline.

Figure 3.

SARM1 deficiency does not change body weight, cardiac function, and geometry in male and female mice at baseline. Body weight and cardiac function by echocardiography were measured in male and female, young wild-type (WT), and SARM1-KO mice. Body weight (A), fractional shortening (FS; B), early-to-late ratio of peak diastolic velocity (E′/A′ ratio; C), and left ventricular internal dimension (end-diastolic; LVID;d; D) were measured. Blue charts, male mice data; red charts, female mice data; n = 6 per group. BW, body weight; KO, knockout; SARM1, sterile-α and TIR motif containing 1.

Table 1.

SARM1 deficiency does not change baseline cardiac geometry in male and female mice

	Male		Female
	WT	SARM1-KO	WT	SARM1-KO
IVS;d	0.801 ± 0.04	0.88 ± 0.1	0.84 ± 0.09	0.77 ± 0.05
LVPW;d	0.64 ± 0.04	0.68 ± 0.05	0.56 ± 0.05	0.66 ± 0.03
LVID;d	3.3 ± 0.09	3.4 ± 0.19	3.4 ± 0.1	3.3 ± 0.24
LVID;s	1.39 ± 0.1	1.6 ± 0.13	1.7 ± 0.05	1.6 ± 0.18
HW/TL	6.5 ± 0.11	6.95 ± 0.65	5.5 ± 0.17	5.25 ± 0.16

Values are means ± SE. d, diastolic; HW, heart weight; IVS, interventricular septal thickness; KO, knockout; LVID, left ventricular internal dimension; LVPW, left ventricular posterior wall thickness; SARM1, sterile-α and TIR motif containing 1; s, systolic; TL, tibia length; WT, wild type.

SARM1 Deficiency Changes Cardiac Profibrotic and Prosenescence Markers in Sex-Specific Manner

Apart from its role as an NAD⁺ hydrolase, SARM1, also known as Myd88-5, modulates inflammatory pathways downstream of Toll-like receptors (TLRs) (21, 22). In the heart, activation of this inflammatory signaling can lead to cardiac fibrosis and remodeling (23). Therefore, we examined the impact of the loss of SARM1 on cardiac fibrosis and inflammation. We detected a small but significant increase in collagen levels in male KO hearts, but not in female hearts, compared with WT counterparts (Fig. 4A). However, this change did not appear to be physiologically relevant as the cardiac function of KO male mice was normal (Fig. 3). Increased fibrosis in KO male hearts was associated with upregulation of Ctgf and downregulation of Tgfb2 and Fn1 (Fig. 4, B–D). Interestingly, Col1a1 was unchanged in KO male hearts, but upregulated in KO female hearts (Fig. 4E).

Figure 4.

SARM1 deficiency induces sex-based changes in the cardiac expression of profibrotic and prosenescence genes in mice. A: percent collagen was quantified by Masson-trichrome staining of heart tissue sections. Quantitative PCR analyses were used to assess transcript expression of profibrotic and prosenescence genes in the hearts of male and female, wild-type (WT), and SARM1-KO mice. Relative mRNA levels of Ctgf (B), Tgfb2 (C), Fn1 (D), Col1a1 (E), Gdf15 (F), p21/Cdkn1a (G), Trp53 (H), and Cxcl1 (I). J: heatmap summarizing profibrotic and prosenescence gene expression. Expression levels were normalized to WT mice of the respective sex. *P < 0.05, **P < 0.01 vs. WT; n = 6 per group. Blue charts, male mice data; red charts, female mice data. On the heatmap, gene names with underlined text signify P < 0.05 vs. WT. KO, knockout; mRNA, messenger RNA; PCR, polymerase chain reaction; SARM1, sterile-α and TIR motif containing 1.

We further examined the effect of SARM1 deletion on the transcripts of senescence-associated, proinflammatory cytokines, chemokines, and growth factors. The expression of growth differentiation factor 15 (Gdf15), a growth factor involved in cardiac remodeling, was significantly downregulated in male KO mice, but upregulated in female KO mice, compared with respective WT controls (Fig. 4F). Although the senescence marker p21 was upregulated in male KO mice (Fig. 4G), other markers such as Trp53 and Cxcl1 were significantly downregulated in male mice (Fig. 4, H and I). These senescence-associated inflammatory markers did not change in female KO mice (Fig. 4, G and I). As observed in the NAD⁺ metabolism transcripts, male mice showed more pronounced changes in the expression of profibrotic and proinflammatory transcripts, compared with female mice (Fig. 4J).

DISCUSSION

Decline in NAD⁺ levels is observed in diseased hearts in both humans and preclinical models (9, 24). The decline can be due to impaired NAD⁺ synthesis and/or excessive consumption by sirtuins, PARPs, and NAD⁺ hydrolases. SARM1 is one of the NAD⁺ hydrolases, but its role in heart function has never been described. In this study, we examined the effects of SARM1 deficiency in mouse hearts using SARM1 total knockout mice. The key findings of this study are 1) SARM1 deletion more robustly elevates baseline NAD⁺ pool in female hearts than in male hearts; 2) SARM1 deletion has sexually dimorphic effects on cardiac transcripts related to NAD⁺ metabolism, fibrosis, and senescence-associated inflammation; and 3) SARM1 deletion does not affect the baseline cardiac function in both male and female mice.

Under physiological conditions with high cellular NAD⁺ levels, SARM1 remains in its inactive conformation stabilized by the binding of NAD⁺ to an allosteric site (25). The enzyme is turned on into its NAD⁺-hydrolyzing active conformation by metabolic mechanisms such as elevated NMN-to-NAD⁺ ratio or by activated JNK signaling (13, 20). However, we observed elevated cardiac NAD⁺ levels in female KO mice (male KO showed a trended increase), even in the absence of any cellular stress. These results suggest that low levels of SARM1 activity exist at baseline to regulate cardiac NAD⁺ turnover and homeostasis.

We observed changes in greater number of NAD⁺ metabolic transcripts in male KO hearts than in female KO hearts, whereas female KO hearts had a more robust increase in NAD⁺ levels. Sex-based differences in NAD⁺ levels in mouse brains have been shown attributable to different energy metabolism between the sexes (26, 27). Among the transcripts of NAD⁺ metabolism, only Cd38 levels were elevated in female KO hearts. However, the slight elevation of Cd38 NAD hydrolase did not lead to NAD⁺ decline in female hearts. This could be due to the fact that CD38 is an ectoenzyme that regulates extracellular NAD⁺ metabolism (28). In male KO hearts, we observed decreased expressions of Parp1, Sirt1, Sirt5, Sirt6, and Sirt7 and increased expression of Nampt, which could contribute to elevated NAD⁺ levels. Conversely, genes of NAD⁺ salvage, Preiss-Handler, and de novo synthesis pathways (Nmnat2, Nmnat3, Nmrk2, and Qprt) were suppressed in male KO hearts, which could prevent NAD⁺ elevation. Since NAD⁺ pool is constantly turned over by consumption and synthesis mechanisms; the transcript data support that SARM1 deficiency may impact NAD⁺ synthesis and consumption fluxes in male hearts, but not in female hearts. Therefore, a future study measuring in vivo NAD⁺ flux (29) may allow further understanding of the sexual dimorphism in SARM1-dependent changes of cardiac NAD⁺ homeostasis.

In addition to NAD⁺ hydrolysis, SARM1 is also an adaptor molecule for Toll-like receptor (TLR) signaling and regulates their downstream proinflammatory cascade (21, 22, 30). In the heart, TLR activation is associated with increased inflammation, cardiac hypertrophy, and remodeling (23, 31). In our study, loss of SARM1 was associated with elevation of collagen deposition in male hearts, but not in females. Other profibrotic markers such as Ctgf, Tgfb2, and Fn1 were also changed in only male hearts. Conversely, mRNA expression of Col1a1 was elevated in female hearts, but not in males. However, these changes did not affect cardiac function and geometry. Senescence-associated inflammatory transcripts including cytokines, chemokines, and growth factors such as Gdf15, p21, Trp53, and Cxcl1 also showed significant differences between WT and KO hearts, only in male mice. Although the changes of these transcripts did not robustly affect fibrosis and cardiac function, these molecular changes may prime male hearts for altered susceptibility to insults leading to fibrotic and inflammatory remodeling. Interestingly, TLR signaling has been speculated to contribute to sex differences in cardiovascular diseases (32).

Elevation of NAD⁺ levels improves cardiac function in various models of heart disease (7, 8, 10). Despite the elevated NAD⁺ pool, cardiac function and geometry of KO mice did not change in both sexes. Because of the sexual dimorphism discussed in RESULTS, the effects of SARM1 deletion on cardiac stress-induced dysfunction could differ between sexes. In addition, it is important to note a few limitations of our study. The cohort of female mice in our study had not been synchronized according to their estrous cycles. Female gonadal hormones are known to regulate metabolism (33) and may alter NAD⁺ levels to regulate transcription. Future studies using synchronized estrous cycles (34) are needed to examine whether the phases of the estrous cycle affect SARM1-dependent changes in the expression of genes such as sirtuins, which are known to be estrogen regulated (35). In male mice, we have not tested whether SARM1 deficiency could exert its effects through changes in circulating levels of testosterone. Although no previous report has explored this relationship, it is known that NAD⁺ can regulate spermatogenesis as a cofactor of testosterone biosynthesis (36). The deficiency of another NAD⁺ hydrolase CD38 can induce testosterone and increase cardiac function (37). Although we did not observe changes in cardiac function in SARM1-KO mice, this evidence from literature (37) still suggests a possibility that SARM1 may affect gene expression via testosterone signaling.

Taken together, our results demonstrate the role of SARM1 in cardiac NAD⁺ metabolism and function at the baseline. Elevation of NAD⁺ levels by SARM1 deletion indicates its potential as a therapeutic target to prevent NAD⁺ decline. Our findings show the sexually dimorphic role of SARM1 in mouse hearts and reiterate the importance of including both sexes in experimental designs.

GRANTS

This work was supported, in part, by National Institutes of Health Grants 1P20GM139763 (to C.F.L.) and R00AG051735 (to Y.A.C.); American Heart Association Grant 17SDG33330003 (to C.F.L.), Presbyterian Health Foundation of Oklahoma recruitment and pilot grants (to C.F.L.); and a Tobacco Settlement Endowment Trust’s Oklahoma Center for Adult Stem Cell Research grant (to C.F.L.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.F.L. conceived and designed research; H.L.N., Y.A.C., C.M.L., and C.F.L. performed experiments; H.L.N., K.E.M., and C.F.L. analyzed data; H.L.N. and C.F.L. interpreted results of experiments; H.L.N. and K.E.M. prepared figures; H.L.N. drafted manuscript; H.L.N., K.E.M., Y.A.C., and C.F.L. edited and revised manuscript; H.L.N., K.E.M., Y.A.C., C.M.L., and C.F.L. approved final version of manuscript.