Sex differences in metabolic pathways are regulated by Pfkfb3 and Pdk4 expression in rodent muscle

Skeletal muscles display sexually dimorphic features. Biochemically, glycolysis and fatty acid β-oxidation occur preferentially in the muscles of males and females, respectively. However, the mechanisms of the selective utilization of these fuels remains elusive. Here, we obtain transcriptomes from quadriceps type IIB fibers of untreated, gonadectomized, and sex steroid-treated mice of both sexes. Analyses of the transcriptomes unveil two genes, Pfkfb3 (phosphofructokinase-2) and Pdk4 (pyruvate dehydrogenase kinase 4), that may function as switches between the two sexually dimorphic metabolic pathways. Interestingly, Pfkfb3 and Pdk4 show male-enriched and estradiol-enhanced expression, respectively. Moreover, the contribution of these genes to sexually dimorphic metabolism is demonstrated by knockdown studies with cultured type IIB muscle fibers. Considering that skeletal muscles as a whole are the largest energy-consuming organs, our results provide insights into energy metabolism in the two sexes, during the estrus cycle in women, and under pathological conditions involving skeletal muscles.

A nimal species have developed a variety of sex differences in their structures and functions. In most animals, this sexual dimorphism is most obvious in the reproductive system. Additionally, however, many animals exhibit sexually dimorphic appearances involving body size, exterior body parts, and feather color. Even many mammalian species demonstrate visually appealing features that differ between males and females. Skeletal muscle is a representative organ whose structures and activities differ between the two sexes 1 .
Skeletal muscle is composed of multiple types of fibers that differ in terms of their morphological, biochemical, and physiological properties. In rodents, these fibers are largely divided into four types (types I, IIA, IIB, and IIX) based on which the myosin heavy chain (MYH) gene is expressed [2][3][4] . As for the functional features of the fibers, type I demonstrates the slowest contraction, while types IIA, IIX, and IIB exhibit successively faster contraction 5,6 . Morphologically, the type IIB fiber is the largest and the type I fiber is the smallest. Biochemically, type IIB fiber has the highest glycolytic activity, while type I fiber demonstrates the highest oxidative capacity. These differences in energy metabolism and the differential ATPase activities of myosin heavy chains have been studied in relation to functional differences among the fiber types 1,7 . In addition to these representative fibers, other fiber types that express multiple types of MYH or minor types of MYH have been detected 8,9 .
The sexually dimorphic features of skeletal muscles have been investigated extensively. The total mass of muscle fibers and their individual sizes, fiber type composition, and skeletal muscle energy metabolism, contractile strength, and fatigability were found to be different between the two sexes 1,10,11 . In addition, the preferred fuels for energy metabolism have been shown to be glucose and fatty acids in the muscles of males and females, respectively 12,13 . Although this difference has been thought to play a fundamental role in the sexually dimorphic functions of skeletal muscle, the mechanisms that induce this feature remain unknown. In addition to these morphological, physiological, and biochemical studies, recent deep-sequence studies revealed sexually dimorphic gene expression in skeletal muscles [14][15][16][17] .
Sex steroids have been investigated as the primary mechanism underlying sex differences 1,18,19 . In fact, some male-specific characteristics of skeletal muscles, such as heavier muscle weight and larger fiber size, were shown to be the result of testosterone 20,21 . Testosterone acts by binding to the androgen receptor (AR/NR3C4), thereby regulating target gene expression. Several genes whose functions are closely related to the male-biased anabolic activity of muscles were shown to be targets of AR [22][23][24] .
Many studies have advanced our understanding of the functional characteristics of skeletal muscles. Unfortunately, however, much of this research was performed with whole muscle rather than with specific fiber types. Since skeletal muscle consists of multiple fiber types, investigation of each type is thought to be essential for the overall comprehension of the functional properties of skeletal muscle. Therefore, in the present study, we aimed to investigate sexual dimorphisms of skeletal muscle at the level of particular fiber types. Our analyses of transcriptome datasets revealed that two key genes underlie sexually dimorphic metabolism. The male-predominant glycolytic activity was found to be due to male-enriched expression of Pfkfb3 (phosphofructokinase-2), while female-predominant fatty acid β-oxidation was found to be due to E2 (estradiol)-enhanced expression of Pdk4 (pyruvate dehydrogenase kinase 4).

Results
Sexual dimorphism in muscle fiber size. We attempted to confirm sexual dimorphism in muscle fiber size (cross-sectional area (CSA)) in five different skeletal muscles (quadriceps, tibialis anterior, gastrocnemius, triceps, and soleus) in 8-week-old male and female mice. In all skeletal muscles examined, fiber CSAs were larger in males than in females ( Supplementary Fig. 1a). Mammalian skeletal muscles consist of four types classified according to the type of myosin heavy chain expressed: type I, IIA, IIB, and IIX fibers can be distinguished by the predominant expression of MYH7 (encoding MYH1), MYH2 (encoding MYH2A), MYH4 (encoding MYH2B), and MYH1 (encoding MYH2X), respectively 4,25,26 . Antibodies against MYH2A and MYH2B were used to identify type IIA and IIB fibers, respectively (Supplementary Fig. 1b). In the skeletal muscles examined, the CSAs of type IIB fibers (a major fast fiber type in fast-twitch skeletal muscles in rodents) were larger in males than in females. We focused on this fiber type in the following studies to further investigate the sexual dimorphism in CSA.
We next addressed the age at which the sex difference emerged. The CSAs of quadriceps type IIB fibers were examined postnatally at 2, 3, 4, and 8 weeks. A slight difference between sexes appeared at 4 weeks, and the difference became obvious at 8 weeks ( Supplementary Fig. 1c). Amd (S-adenosylmethionine decarboxylase) and Smox (spermine oxidase), both of which are required for polyamine synthesis, were reported to exhibit male-enriched and androgen-induced expression 14,15,17,27 . The expressions of these genes were examined in quadriceps type IIB fibers. Similar to the CSAs, slight male-enriched expressions were observed at 4 weeks, and were clearly apparent at 8 weeks ( Supplementary Fig. 1d).
Effect of sex steroids on muscle fiber size. To examine the effect of sex steroids on CSAs, we prepared muscles from sham-operated males and females, castrated males (Cas), ovariectomized females (Ovx), mice treated with DHT (dihydrotestosterone) after gonadectomy (Cas+DHT and Ovx+DHT), and mice treated with E2 after gonadectomy (Cas+E2 and Ovx+E2), as shown in Supplementary Fig. 2. Of note, as the control female, we used diestrus mice that had undergone sham operation.
After the skeletal muscles of the mice were stained for MYH2B (Fig. 1a), the CSAs of the MYH2B-positive type IIB fibers were measured ( Fig. 1b and Supplementary Fig. 1e). Immunofluorescence analysis suggested that type IIB fibers were larger in males than in females, and that DHT treatment enlarged CSAs regardless of sex. Statistical analyses indicated that the CSAs of the quadriceps, tibialis anterior, triceps, and soleus were larger in males than in females (Fig. 1c). With a few exceptions, DHT increased the CSAs of the muscles above regardless of sex, whereas E2 did not. Interestingly, the quadriceps fibers enlarged by DHT remained larger in males than in females.
Sexually dimorphic gene expression in quadriceps type IIB fibers. Since the CSAs of quadriceps muscle type IIB fibers exhibited clear sexual dimorphism and DHT dependency, we decided to obtain transcriptomes of these fibers. In all 10 experimental mouse groups (sham-operated males and females, males and females transplanted with a DHT-containing or empty pellet after gonadectomy, and males and females injected with E2-containing or corn oil after gonadectomy), single fibers were prepared from an area of the quadriceps where~95% fibers are type IIB (Supplementary Fig. 3a). Thereafter, the fiber types were determined by RT-PCR ( Supplementary Fig. 3b).
The RNAs recovered from fibers positive for MYH2B (encoded by Myh4) were pooled and subjected to mRNA sequencing. Transcriptome datasets with sufficient quality for the following analyses were obtained from all experimental groups above ( Supplementary Fig. 3c). Genes whose CPM (counts per million mapped reads) values were >10.0 in either the sham-operated males or females were extracted as all expressed genes (6978 genes) and utilized for the following analyses. Of these genes, 68 and 60 demonstrated more than 2.0-fold enrichment in males and females, respectively (Fig. 2a, Table 1). As described in detail below, two key genes for energy metabolism, Pdk4 (pyruvate dehydrogenase kinase 4) and Pfkfb3 (phosphofructokinase-2), were among the male-enriched genes.
Expression profiles of the male-and female-enriched genes in the 10 experimental groups were analyzed by hierarchical clustering (Fig. 2b, c). According to the clustering profile of the male-enriched genes, male, Cas+DHT, and Ovx+DHT mice were classified into one subgroup. Nearly half of the maleenriched genes exhibited DHT-dependent expression. A similar effect of DHT was observed in the ovariectomized females. Likewise, clustering of the female-enriched genes indicated that the same experimental groups were likely to form a subgroup. Comparison of female, Cas+E2, and Ovx+E2 mice suggested that a certain number of the female-enriched genes were activated by E2 in both sexes. Even though both DHT and E2 affected gene expression, it is likely that the effects of DHT are more evident than those of E2.
We unexpectedly found that a group of male-enriched genes was activated in castrated males (Cas+P) following empty pellet implantation, although this phenomenon was not observed in ovariectomized females (Ovx+P). Since the reason for this unexpected gene activation by control treatment was unknown, we carefully examined the subsequent results.
Principal component analysis of whole transcriptome data was used to classify the mouse groups. Male, Cas+DHT, and Ovx +DHT mice comprised a separate subgroup (SG1 in Fig. 2d). The classification of female, Cas+E2, and Ovx+E2 mice was unclear. The experimental group, Cas+P, was classified apart from the others, perhaps due to the unexpected gene activation by the empty pellet implantation described above. It was reported that testosterone levels were different among mouse strains, and that of C57BL/6 was significantly lower than those of CD-1, CH3, and FVB 28 . Studies using mice with higher testosterone levels might provide us with more pronounced sexually dimorphic gene expression.
Functions related to male-and female-enriched genes. Gene ontology analyses were conducted on the sex-biased genes. The polyamine biosynthetic process was identified as a potential biological process related to the male-enriched genes (Supplementary Table 1). Among the polyamine synthetic genes, Odc1 (ornithine decarboxylase 1), Amd1/2, and Smox showed higher expressions in males than in females and were induced by DHT Cas+DHT/Ovx+DHT a) Fig. 1 Effects of sex steroids on skeletal muscle fiber size. a Quadriceps muscle samples were prepared from eight experimental groups: sham-operated male and female mice (CTR), gonadectomized mice (Cas for males and Ovx for females), DHT-treated male mice (Cas+DHT), and female mice (Ovx+DHT) after gonadectomy, and E2-treated male mice (Cas+E2) and female mice (Ovx+E2) after gonadectomy ( Supplementary Fig. 2). Type IIA (red) and type IIB (green) fibers were detected with MYH2A and MYH2B antibodies, respectively. Blue: DAPI. Scale bar = 50 μm. b The CSAs of the type IIB fibers were measured using approximately 4800 to 6000 fibers per specimen. The distributions of CSA sizes (horizontal axis) and frequencies (vertical axis) are shown. Male and female data are indicated by blue and orange bars, respectively. The same studies were performed using the tibialis anterior, gastrocnemius, triceps, and soleus muscles ( Supplementary Fig. 1e). A representative result from three biologically independent samples is shown. c The CSA size distribution was compared among the eight experimental groups above. The data were analyzed as described in the "Methods" section. The box and whisker plots with the same letter are not significantly different from each other (p < 0.01). Table 2a). The expressions of these genes were not upregulated by empty pellet implantation. These polyamine synthetic genes were previously shown to be male enriched and androgen inducible 14,15,17,27 . As for their function in skeletal muscles, they were reported to suppress muscle atrophy and promote hypertrophy 29,30 , possibly by regulating cellular proliferation and viability 31,32 as well as protein synthesis 33 and autophagy 34,35 . Together, it is likely that the sexual differences in muscle sizes primarily depend on the amounts of androgens in males and females.

treatment in both sexes (Supplementary
As for the female-enriched genes, the terms "collagen fibril organization," "wound healing," and "skeletal system development" were found by the gene ontology analysis. Since collagen genes were commonly included in these processes, we examined their expressions in the datasets. The expressions of many collagen genes were higher in females than in males (Supplementary Table 2b) but were not clearly affected by ovariectomy or E2 treatment.
Differential regulation of Pdk4 and Pfkfb3 genes by sex steroids. Because diestrus mice with low E2 serum concentrations were used as the control females, it was assumed that a certain population of genes potentially activated by E2 would not be included in the female-enriched genes. Therefore, the E2 induction ratios for all of the expressed genes (6978) were calculated in males and females (Cas+E2/Cas+oil and Ovx+E2/Ovx+oil) and plotted in Fig. 3a. The accumulation of genes in the upper right and lower left quadrants suggested that many genes were activated and suppressed by E2 regardless of sex. However, this pattern was not observed when only the female-enriched genes were analyzed (Fig. 3b). Moreover, many genes exhibiting relatively high activation or suppression by E2 were excluded from the female-enriched genes. Female-enriched Col genes were not activated intensively by E2, even in female mice. As a consequence, this analysis identified E2-induced genes that were not members of the female-enriched gene population. Interestingly, two key genes regulating energy metabolism, Pdk4 and Pcx (pyruvate carboxylase), were included (Fig. 3a).
Likewise, the DHT induction ratios for all of the expressed genes and the male-enriched genes were calculated (Fig. 3c, d). Expectedly, Smox and Amd1/2 were localized in the upper right quadrant, indicating that they were activated by DHT regardless of sex. Although Pdk4 and Pfkfb3 were male-enriched genes, their localizations differed from those of Smox and Amd1/2, suggesting that their expressions were relatively unaffected by DHT in both sexes.
Possible contribution of PFKFB3 to male-predominant glycolysis. PFKFB3 plays a crucial role in glycolytic regulation by producing fructose-2,6-bisphosphate, which robustly activates PFKM (the type of phosphofructokinase-1 found in muscles) 36,37 (Fig. 4a). Therefore, the male-enriched expression of Pfkfb3 suggested that glycolytic activity in quadriceps type IIB fibers would be higher in males than in females. To examine this, we prepared muscle fibers from a particular region (white quadriceps) 38 of the quadriceps ( Supplementary Fig. 3a), and the extracellular acidification rate (ECAR), an index of glycolytic activity, was determined. As expected, the ECAR was approximately two-fold higher for muscle fibers from males than from females (Fig. 4b).
We assumed that male-biased glycolytic activity could be caused by higher expressions of glycolytic genes in males as well as by the upregulated expression of Pfkfb3. However, neither the transcriptome data nor qRT-PCR showed male-enriched expression of any glycolytic genes (Fig. 4c). Moreover, their expressions were not largely affected by gonadectomy or sex steroid treatments. In contrast, both transcriptomic analysis and qRT-PCR (Fig. 4d, e) showed that Pfkfb3 expression was decreased significantly by castration but not by ovariectomy. Consistent with the results shown in Fig. 3a and b, DHT treatment failed to reverse the decreased expression caused by castration. Empty pellet implantation did not affect the expression of Pfkfb3.   Expectedly, the male-enriched expression of PFKFB3 was detected at the protein level ( Fig. 4f, Supplementary Fig. 5a). The results above strongly suggested that male-predominant glycolysis can be achieved by the male-enriched expression of Pfkfb3 alone. To verify this, we performed a knockdown study of the gene. We first examined if the gene could be suppressed by a general procedure for siRNA knockdown in cultured muscle fibers. These fibers were transfected with siRNAs for Pfkfb3 or control siRNA for 6, 12, or 24 h, then the mRNA was quantified by qRT-PCR. The siRNA treatment suppressed Pfkfb3 expression to approximately 25% at 6 h and then to 40-50% at 12 and 24 h after the treatment (Fig. 5a). Likewise, PFKFB3 was decreased at  Of note, the activity of the siPfkfb3-treated male-derived fibers decreased to approximately the same level as the siControltreated female-derived fibers. Taken together, these results strongly suggest that the male-predominant glycolytic activity of quadriceps type IIB fibers can be established largely by the male-enriched expression of Pfkfb3.
The possible contribution of PDK4 to female-predominant fatty acid metabolism. Female skeletal muscles use fatty acid β-oxidation rather than glycolysis for energy production 12,39 . Therefore, we examined the expression of genes involved in fatty acid β-oxidation in the transcriptome datasets and using qRT-PCR (Fig. 6a, Supplementary Table 2c). Although many of these genes showed a tendency to be activated by E2, none demonstrated female-enriched or E2-enhanced expression over twice the baseline level. By contrast, our transcriptome data revealed E2-enhanced expression of Pdk4, which was further confirmed by qRT-PCR (Fig. 3, Fig. 6b, c). In addition to the level of the mRNA, PDK4 was increased in the E2-treated female muscle fibers at the level of protein (Fig. 6d, Supplementary Fig. 5c).
As for the function of PDK4, studies so far have established that the enzyme promotes the utilization of fatty acids for energy by suppressing the pyruvate dehydrogenase complex through phosphorylation 40 (Supplementary Fig. 4). Expectedly, the phosphorylation level of PDH (P-PDH) increased in the E2-treated female muscle (Fig. 6d, Supplementary Fig. 5c). Therefore, we hypothesized that female-predominant fatty acid β-oxidation is attributable to E2-induced Pdk4. To investigate this, we determined the fatty acid-dependent oxygen consumption rate (OCR) using cultured muscle fibers from male mice and diestrus female mice treated with or without E2 for 24 h. As shown in Fig. 6e, the fatty acid-dependent OCR was similar in male mice and diestrus female mice. As expected, the OCR was enhanced two-fold by E2 treatment.
Finally, we investigated the effect of Pdk4 knockdown. When muscle fibers were treated with siPdk4, the amount of Pdk4 mRNA was decreased to 40% by 6-h treatment, and to 20% by 9and 12-h treatments (Fig. 6f). This is consistent with the level of PDK4 protein decreased by the siPdk4 treatment (Fig. 6g,  Supplementary Fig. 5d). Next, we examined whether the phosphorylation level of PDH is decreased by the siPdk4 treatment. As expected, it was found that the phosphorylation levels of PDH in the E2-treated female and male were significantly decreased (Fig. 6h, Supplementary Fig. 5e). Finally, muscle fibers treated with siRNA for 9-h were subjected to the aforementioned OCR assay. Pdk4 knockdown resulted in a decrease of fatty acid-dependent OCR in all samples. Interestingly, the enhancement of fatty acid dependency by E2 treatment was nullified by Pdk4 knockdown (Fig. 6i). Taken together, these results strongly suggest that female-predominant fatty acid utilization is attributable to E2-induced Pdk4 gene expression.

Discussion
Many transcriptome datasets have been obtained from skeletal muscles to evaluate the effects of exercise, metabolic diseases, aging, etc. [41][42][43] . Some of them uncovered sexually dimorphic gene expression while others demonstrated the effects of sex steroids [14][15][16][17] . In addition to these transcriptome analyses, many studies have characterized the sexually dimorphic structures and functions of skeletal muscles 1 . One fundamental difference regards energy metabolism: male skeletal muscles preferentially utilize glycolysis, while female muscles tend to rely on mitochondrial fatty acid β-oxidation 7 . It was also shown, regardless of sex, that the main type of metabolism in some skeletal muscle fibers is anaerobic glycolysis, while in others it is aerobic  44 . Therefore, the sexual dimorphism in the energy metabolism of skeletal muscle is due in part to the preponderance of glycolytic fibers in males and of oxidative fibers in females. However, it is also possible that sexually dimorphic metabolism is the result of differential metabolic activities intrinsic to male and female fibers. To investigate this issue, we focused on type IIB fibers, which are the most abundant type of fiber in fast-twitch muscles in rodents 45,46 .
We found that male-predominant glycolytic activity could not be accounted for simply by enhanced glycolytic gene expression in males. Interestingly, however, Pfkfb3 was one of the maleenriched genes identified in this study. PFKFB3 mediates the conversion of F-6-P (fructose-6-phosphate) to F-2,6-BP (fructose-2,6-bisphosphate), the latter of which acts as a potent allosteric activator of PFKM (a muscle type of PFK-1), one of the glycolytic rate-limiting enzymes 47 . During supramaximal exercise, activated glycolysis rapidly increases lactate concentrations, causing muscle fiber pH to become acidic, and simultaneously causes rapid reductions in oxygen and glucose concentrations 48 . These conditions are known to decrease glycolytic activity by suppressing the action of PFKM. Interestingly, however, suppression of the liver type of PFK-1 can be released by the robust action of F-2,6-BP produced by PFKFB3 [49][50][51] . Because the concentration of F-2,6-BP was shown to correlate with the expression level of Pfkfb3/PFKFB3 52,53 , we assumed that the male-enriched expression of Pfkfb3 would ensure male-predominant glycolytic activity. Indeed, when knockdown was used to decrease the level of Pfkfb3 gene expression in male-derived fibers to that observed in female-derived fibers, the glycolytic activity in the former decreased to levels similar to those in the latter. As described above, the male-predominant glycolytic activity of fast-twitch muscles is thought to be due to the larger number of type IIB fibers in male muscles. In addition, our study revealed for the first time that male type IIB fibers are intrinsically capable of driving glycolysis more robustly than female fibers through the malebiased expression of a single gene, Pfkfb3.
The results of this study suggest the potential importance of the sexually dimorphic expression of Pfkfb3. Studies so far have investigated the mechanism of Pfkfb3 gene regulation from the viewpoint of glycolysis promotion in cancer cells. These studies implicated HIF1α (hypoxia-inducible factor 1α) in gene regulation 53,54 . In addition, testosterone and E2 were shown to activate HIF1α gene expression in prostate 55,56 and breast cancer cells 57 , respectively, suggesting that sex steroids could induce Pfkfb3 gene expression through HIF1α induction. Meanwhile, our current study of the quadriceps muscle suggested that stillunidentified factors besides testosterone are responsible for the male-enriched expression of Pfkfb3. Regarding these factors, it is interesting to note that the metabolic activities of preimplantation embryos are higher in males than in females 58 , and the number of X chromosomes may contribute to sexually dimorphic metabolism 59 . These studies suggest that genes localized on the sex chromosomes might play a role in the sexually dimorphic metabolism seen in XX and XY muscle fibers.
Our present study of cultured type IIB fibers demonstrated again the well-known fact that females preferentially utilize fatty acids for mitochondrial oxidation. This preference has been suggested to be due to E2-induced expression of genes related to fatty acid β-oxidation, including Cpt1b (which encodes a rate-limiting enzyme for β-oxidation), carnitine palmitoyltransferase 60,61 , Hadhb (hydroxyacyl-CoA dehydrogenase), and Pdk4 (pyruvate dehydrogenase kinase 4) 62 . Their results using whole gastrocnemius muscle correlate well with our findings using type IIB fibers, in that the induction ratio of Pdk4 by E2 was more evident than those of other β-oxidation genes.
It has been established that PDK4 induces a metabolic shift from glycolysis to fatty acid β-oxidation through phosphorylation, thereby suppressing the pyruvate dehydrogenase complex 63 . Indeed, the level of Pdk4 gene expression has been shown to correlate with the activity of fatty acid β-oxidation in cultured cells 40 . Taken together, we inferred that the preferential use of fatty acid β-oxidation in females is caused primarily by Pdk4 gene expression induced by E2. Expectedly, E2 treatment enhanced fatty acid-dependent mitochondrial oxygen consumption in female-derived muscle fibers, and this enhancement was canceled by the knockdown of Pdk4. Although we cannot exclude the possibility that female-predominant fatty acid β-oxidation is attributable to E2-activated β-oxidation genes such as Cpt1b, our knockdown studies of cultured muscle fibers demonstrated that E2-induced Pdk4 gene expression was a more likely cause.
We also observed E2-activated expression of the Pcx gene, whose product mediates an anaplerotic reaction to maintain tricarboxylic acid cycle flux by providing oxaloacetate. This reaction was reported to be critical for maintaining the oxidative function of mitochondria in skeletal muscle 64,65 . Therefore, Pdk4 and Pcx may together coordinate female-predominant mitochondrial fatty acid β-oxidation through simultaneous induction by E2.
It has been accepted that cardiac muscle fibers of women have a higher activity of fatty acid β-oxidation than those of men 66,67 . This female-biased fatty acid β-oxidation was observed in the mice that developed a hypertrophied heart by exercise 68 . To comprehend the mechanism for the sexually dimorphic metabolism, transcriptomes were obtained from the cardiac muscles of both sexes [69][70][71] . A few genes required for fatty acid utilization were found as femaleenriched genes. Unfortunately, however, none of the studies found Pdk4 as the female-enriched gene, suggesting that distinct mechanism for female-biased fatty acid β-oxidation might work between the cardiac muscle and the type IIB fibers of skeletal muscle. Alternatively, because the estrus cycle was not considered in those studies, experiments to investigate the effects of E2 could uncover the implication of PDK4/Pdk4 in female-biased fatty acid β-oxidation in the cardiac muscle. PFKFB3 mediates the reaction from F-6-P to F-2,6-BP, the latter of which acts as a strong activator of PFKM. b Male (n = 3) and female (n = 3) muscle fibers were prepared from type IIB-enriched regions of the quadriceps muscle ( Supplementary Fig. 3a). The ECARs of cultured muscle fibers were examined, and the data were corrected by the ratio of the mean protein content in the fibers from the two sexes (male/female = 1.10) (Supplementary Fig. 3d). **p < 0.01. c The expressions of glycolytic genes were determined in the 10 experimental mouse groups ( Supplementary Fig. 2) by qRT-PCR (n = 3 each group). Among paralogous genes, if any, the gene showing the highest expression was examined. d The expression of Pfkfb3 was extracted from the transcriptome datasets. e The expression of Pfkfb3 mRNA was determined by qRT-PCR (n = 3 each group). f The amounts of PFKFB3 protein were analyzed by western blotting. Whole proteins prepared from the type IIB-enriched areas of the quadriceps muscles of male (M) and female (F) were used. HeLa cell lysate was used as a control (C). Western blot images for PFKFB3 (upper left) and GAPDH (lower left) are shown. Full blot images are shown in Supplementary Fig. 5a. Three biologically independent samples were analyzed. The data were normalized to GAPDH and are presented as means ± SD (right). **p < 0.01. For c and e, the bars (means ± SD) with the same letter are not significantly different from each other (p < 0.01).
By focusing on gene expression in type IIB fibers in the quadriceps muscle, we unveiled the mechanisms of malepredominant glycolysis and female-predominant fatty acid β-oxidation. Interestingly, it appears that sexually dimorphic metabolism was achieved possibly by two genes, namely Pdk4 and Pfkfb3, through transcriptional regulation by E2 and one or more unknown factor(s), respectively. Considering that skeletal muscle is the largest energy-consuming organ in the human body, our present findings may contribute to understanding the metabolism of both male and female individuals. In particular, our results may provide an insight into the metabolic properties of females, whose E2 concentrations vary throughout life and during the estrus cycle. Our present study evaluated the characteristics of a single type of muscle fiber. However, since skeletal muscles consist of multiple fiber types 4 , additional studies may provide deeper insights into a variety of functional differences, such as between fast and slow-twitch muscles, between the two sexes, and between healthy and pathological conditions.

Methods
Treatment of animals. Male and female C57BL/6J mice (Japan SLC, Inc.) were gonadectomized or sham-operated at 3 weeks after birth. Three mice were kept in one cage and fed with standard CRF-1 chow (Oriental Yeast Co., Ltd., Tokyo, Japan). They had no interaction with individuals of opposite sexes. Treatment with DHT and E2 was performed as summarized in Supplementary Fig. 2. Every experimental group comprised three male and three female mice. Skeletal muscles (gastrocnemius, tibialis anterior, quadriceps, triceps, and soleus) were isolated at 8 weeks after birth and used for further studies. To determine estrus cycle phases in females, a vaginal smear test was conducted 72 . All animal experiment protocols were approved by the Animal Care and Use Committee of Kyushu University. All experiments were performed in accordance with the guidelines.
mRNA sequencing and data processing. The quadriceps muscles isolated from the aforementioned mice were immersed in RNAlater (Qiagen, Venlo, The Netherlands). Approximately 100 individual fibers were prepared from the muscles of 10 experimental groups (sham-operated male (n = 3) and female mice (n = 3) (CTR), pellet implantation male (n = 3) (Cas+P) and female mice (n = 3) (Ovx+P) after gonadectomy, oil injection male (n = 3) (Cas+Oil) and female mice (n = 3) (Ovx+Oil) after gonadectomy, DHT-treated male mice (n = 3) (Cas+DHT) and female mice (n = 3) (Ovx+DHT) after gonadectomy, and E2treated male mice (n = 3) (Cas+E2) and female mice (n = 3) (Ovx+E2) after gonadectomy) (Supplementary Fig. 2) using fine forceps under a SMZ-U Zoom 1:10 stereomicroscope (Nikon, Tokyo, Japan) 73 . RNA was obtained individually from each fiber using TRIzol (Thermo Fisher Scientific, Waltham, MA, USA). cDNA was prepared from a small aliquot of each RNA and then subjected to PCR to distinguish fiber types using primer sets for myosin heavy chains (Supplementary Table 3). After the RNAs of type IIB fibers were collected, ribosomal RNA was removed using a NEBNext rRNA Depletion Kit (NEB, Ipswich, MA, USA). cDNA libraries for mRNA-seq were prepared using a NEBNext Ultra II RNA Directional Library Prep Kit (NEB). After the quality of the cDNA libraries was validated using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA), the libraries were subjected to sequencing (NovaSeq 6000 System: Illumina, San Diego, CA, USA). STAR (version 2.7.3a) 74 and featureCounts (version 2.0.0) 75 were used for alignment and assembly of the sequence reads, respectively. Mus musculus genome assembly (mm10, NCBI) was used as the reference.
qRT-PCR. Three biologically independent samples were used for qRT-PCR using a CFX96 real-time PCR system (Bio-Rad, Hercules, CA, USA) and SYBR Select Master Mix (Thermo Fisher Scientific). The primer sets used are listed in Supplementary Table 3. The data were standardized using Actb (β-actin) and are presented as means ± standard deviation (SD). Statistical analysis was performed by one-way ANOVA followed by the post hoc Tukey HSD test 76 or the Student's t-test. Significant differences (p < 0.01) are indicated in the figures.
Preparation of muscle fibers from the quadriceps muscle. Type IIB fiberenriched areas in quadriceps muscles were confirmed by an immunofluorescence study. In these regions, approximately >95% fibers were type IIB ( Supplementary  Fig. 3a). Living muscle fibers were prepared from the areas and cultured as described by Kitajima et al. 79    To study the effect of Pfkfb3 knockdown, muscle fibers were transfected with siPfkfb3 for 6 h after 18-h culture in a standard medium. After transfection, the fibers were further cultured in a standard medium for another 12 or 24 h. After 1-h culture in XF base medium, they were subjected to ECAR measurement. Approximately 10% of the fibers died during the transfection, and the numbers of living fibers varied among wells. Thus, the wells containing at least living 13 fibers by the end of the transfection were subjected to ECAR measurement. Since fiber sizes differed between males and females, the amount of protein in each fiber was determined ( Supplementary Fig. 3d) and ECARs were corrected using the ratio of the mean protein content in the fibers of the two sexes (male/female = 1.10). Three biologically independent samples were used. Data are presented as means ± SD and were analyzed by one-way ANOVA followed by the post hoc Tukey HSD test or the Student's t-test.
OCR measurement. The OCR was measured using a Seahorse XFe96 Analyzer (Agilent Technologies) basically according to the manufacturer's protocol. Fifteen muscle fibers were prepared from each of the following groups: male mice (n = 3), oil-injected female mice in diestrus (n = 3), and female mice treated with E2 for 24 h (n = 3). All fibers were plated on a 96-well plate precoated with Matrigel matrix (Corning Incorporated), then incubated in 200 μl of standard medium for 18 h. Before OCR measurement, the medium was changed to XF base medium supplemented with 1 mM pyruvate, 2 mM glutamine, and 10 mM glucose, and the muscle fibers were incubated for 1 h without CO2. To determine the fatty aciddependent OCR, 4 μM Etomoxir (Seahorse XF Mito Fuel Flex Test Kit, Agilent Technologies), an inhibitor of carnitine palmitoyl-transferase, was used. The degree to which the OCR was decreased by the inhibitor was defined as the fatty aciddependent OCR.
To study the effect of Pdk4 knockdown, fibers were transfected with siPdk4 or control siRNA after 18-h incubation in a standard medium. After the transfection, wells containing at least 13 living fibers were used for OCR measurement. The OCRs of fibers from males and females were corrected by the ratio of the mean protein content in the fibers of the two sexes (male/female = 1.10) (Supplementary Fig. 3d). Three biologically independent samples were used. Data are presented as means ± SD and were analyzed using one-way ANOVA followed by the post hoc Tukey HSD test.
Western blotting. Whole protein lysate was prepared from the type IIB fiberenriched area in quadriceps muscles. The fibers were lysed using RIPA buffer (Sigma-Aldrich), followed by sonication (Branson Ultrasonics TM S-250A Model Sonifier TM Analog Cell Disrupter, Branson, Brookfield, CT, USA). Mitochondria were isolated from the muscle fibers as described by Garcia-Cazarin et al. 83 and lysed using RIPA buffer. The protein concentration was determined using BCA Protein Assay Kit (Thermo Fisher Scientific). 30 μg whole lysate or 10 μg mitochondrial proteins were subjected to SDS-polyacrylamide gel electrophoresis, followed by western blotting. Anti-PFKFB3 (1:2000 78 . Data (means ± SD) obtained from three biologically independent samples were analyzed by one-way ANOVA followed by the post hoc Tukey HSD test (Figs. 5b, 6d, g, and h) or the Student's t-test (Fig. 4f).
Statistics and reproducibility. Statistically significant differences between two groups were calculated using Student's t-test. Significant differences between multiple groups were calculated using one-way ANOVA followed by the post hoc Tukey HSD test. All experiments were performed with three biologically independent samples.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
mRNA-seq data have been deposited in DDBJ under the accession code DRA010793 (https://ddbj.nig.ac.jp/DRASearch/). All source data underlying graphs in main figures are provided in Supplementary Data 1-5. All other data are available from the corresponding author on reasonable request.