The Core Circadian Clock Factor, Bmal1, Transduces Sex-specific Differences in Both Rhythmic and Nonrhythmic Gene Expression in the Mouse Heart

Abstract

It has been well established that cardiovascular diseases exhibit significant differences between sexes in both preclinical models and humans. In addition, there is growing recognition that disrupted circadian rhythms can contribute to the onset and progression of cardiovascular diseases. However, little is known about sex differences between the cardiac circadian clock and circadian transcriptomes in mice. Here, we show that the core clock genes are expressed in common in both sexes, but the cardiac circadian transcriptome is very sex-specific. Hearts from female mice expressed significantly more rhythmically expressed genes (REGs) than male hearts, and the temporal distribution of REGs was distinctly different between sexes. To test the contribution of the circadian clock in sex-specific gene expression in the heart, we knocked out the core circadian clock factor Bmal1 in adult cardiomyocytes. The sex differences in the circadian transcriptomes were significantly diminished with cardiomyocyte-specific loss of Bmal1. Surprisingly, loss of cardiomyocyte Bmal1 also resulted in a roughly 8-fold reduction in the number of all differentially expressed genes between male and female hearts. We highlight sex-specific changes in several cardiac-specific transcription factors, including Gata4, Nkx2-5, and Tbx5. While there is still much to learn, we conclude that cardiomyocyte-specific Bmal1 is vital in conferring sex-specific gene expression in the adult mouse heart.

Graphical Abstract

Open in new tabDownload slide

Introduction

Cardiovascular disease (CVD) is the leading cause of death for both men and women,¹ but distinct sex differences exist in the prevalence and pathology of CVD.² For example, heart failure in elderly populations affects more women than men, yet women typically have better clinical outcomes due in part to differences in the presentation of the disease.³ Similar phenomena have long been observed in preclinical models of CVD, as exemplified by greater cardiac hypertrophy and fibrosis in male rats following chronic pressure overload compared to females.⁴ Unsurprisingly, many of the sex differences observed in CVD have been linked to differences in sex chromosomes as well as sex hormones and their receptors.² However, the specific mechanisms underlying sex-specificity in CVD remain incompletely understood.

More recently, it has been recognized that disruptions to circadian rhythms can contribute to the development of CVD.⁵ The circadian clock mechanism exists in every cell and produces intrinsic circadian rhythms in gene expression, physiology, and behavior. At the molecular level, the circadian clock is described as a transcriptional and translational feedback mechanism in which the transcriptional activators BMAL1 and CLOCK heterodimerize and bind to E-box-containing genes to induce the expression of the negative limb genes, including Per1/Per2, Cry1/Cry2, and Rev-erbs. Observational studies have shown that circadian disruption through shift work is an independent risk factor for CVD,^6,7 likely due to the chronic misalignment of behaviors with the phases of the circadian clocks throughout the body. Preclinical models have also shown that disruption to clock gene expression has detrimental effects on cardiovascular function,^5,8–10 linking the circadian clock to the development of CVD.

Beyond timekeeping, the clock mechanism also contributes to a 24-hour daily gene expression program, termed clock output, which contributes to time-of-day changes in physiology. Sex differences in clock output have recently been reported in mouse liver,^11,12 and a similar result was also reported in a human study in which the authors used genotype-tissue expression with an algorithm to assign circadian phase. Of those rhythmically expressed genes (REGs) identified in the human heart, only 50% were shared between sexes, while others were unique to one sex or showed sex-specific rhythmic patterns.¹³ However, little is known of the sex-specific patterns of the clock and clock output in the heart.

In this study, we used RNAseq to investigate the cardiac circadian transcriptome and the role of cardiomyocyte-specific Bmal1 on the cardiac circadian transcriptome of adult male and female mice. Consistent with other studies, we observed that female hearts express significantly more REGs than male hearts, with a modest overlap in gene identity and functional enrichment. Analysis of the temporal distribution of the REGs, using peak expression time, illustrated that the distribution of REGs throughout the day was distinctly different between sexes. Most REGs in the male heart peak at the transition from rest to active (light to dark) or active to rest (dark to light) phases. In contrast, most REGs in the female heart peak in the middle of the rest (light) or active (dark) periods. Surprisingly, cardiomyocyte-specific loss of Bmal1 resulted in a significant reduction of all sex differences in the cardiac circadian transcriptome. Analysis of all differentially expressed genes (DEGs), including nonrhythmic genes, between males and females showed that differential gene expression between sexes was largely diminished with the loss of cardiomyocyte Bmal1. We conclude that cardiomyocyte-specific Bmal1, and likely the core clock mechanism, may be a vital transcriptional co-factor in conferring a sex-specific gene expression program in the adult mouse heart.

Materials and Methods

All animal procedures were performed by the Association for Assessment and Accreditation of Laboratory Animal Care guidelines and approved by the University of Kentucky Institutional Animal Care and Use Committee and the University of Florida Institutional Animal Care and Use Committee.

The inducible cardiomyocyte-specific Bmal1 knockout mice (iCS Bmal1 KO) were bred by crossing the floxed Bmal1 mouse with the cardiomyocyte-specific Myh6-MerCreMer recombinase transgenic mouse as described previously.¹⁴ All mice were born from the same breeders and housed in the same facility for the total duration of the experiment. Cre recombination was induced at 3-4 months of age by intraperitoneal injection of tamoxifen (35 mg/kg/day) for 3 consecutive days to induce the deletion of Bmal1 in adult cardiomyocytes. The control mice were generated by injecting Cre^±  Bmal1f/f mice with vehicle (15% ethanol in sunflower seed oil). Mice remained in their cages for 6-7 weeks post-treatment to wash out potential tamoxifen effects before the time course collection. The average age of the mice at tissue collection was ∼5 months old.

Tissue collection was performed following the circadian analysis guidelines provided by Hughes et al.¹⁵ and is consistent with prior work.¹⁴ Briefly, both vehicle treated and tamoxifen treated mice were housed at 12:12 hour light/dark (7:00 am light on and 7:00 pm light off) schedule for 2 weeks with food and water ad libitum. Before tissue collection, mice were housed in constant darkness for 30 hours (light turned off at 7:00 am) to avoid light effect on endogenous gene expression. The heart tissue was collected every 4 hours for 48 hours (n = 2 per timepoint) starting at 1:00 am (CT18). The mice were euthanized with cervical dislocation under dim red light, lights were turned on for dissection, and tissues were flash-frozen for RNA and protein analysis.

We tested all hearts for Bmal1 recombination as described before.¹⁴ A small piece (∼5 mg) of cardiac muscle tissue was lysed using proteinase K (final concentration 50 ng/µL in lysis buffer (100 mm Tris-HCl, pH 8.5, 5 mm EDTA, 200 mm NaCl, 0.2% SDS) for 2 hours at 55°C. Genomic DNA was isolated by ethanol precipitation. Up to 200 ng genomic DNA was used for each polymerase chain reaction (PCR) reaction. The forward primer (ACTGGAAGTAACTTTATCAAACTG) and the reverse primer of nonrecombined (CTGACCAACTTGCTAACAATTA) yields a 431 bp PCR product, while the forward and the recombined reverse primer (CTCCTAACTTGGTTTTTGTCTGT) yield a 572 bp PCR product.

RNA isolation was performed as described before.¹⁶ Total RNA from frozen heart ventricle (10-15 mg) was isolated using 1 mL TRIzol (Invitrogen, cat#15 596 018) according to the manufacturer’s user guide. Briefly, homogenization was conducted using a Bullet Blender with appropriate amount of sterile stainless-steel beads (0.5-2 mm) (NextAdvance, NY, USA) in 30-second bouts for a total of 3 minutes with 1-minute breaks between each bout. The aqueous phase containing RNA was separated by mixing with chloroform, followed by centrifugation. The RNA was further purified using the RNeasy Mini RNA Extraction Kit (Qiagen, cat#74 106) and subjected to DNase treatment. The integrity of purified RNA was determined using the Agilent Bioanalyzer and all RNA samples had a RIN number above 8.0.

Protein western blot was performed as described before.¹⁷ Briefly, about 5 mg of frozen heart ventricle from CT30 (estimated BMAL1 protein peak) was homogenized with Laemmli sample buffer and protein concentration was measured using RC-DC (Bio-Rad) reagent. For SDS PAGE (10%), 50 ug of protein was loaded. After protein was transferred to polyvinylidene fluoride (PVDF) membrane, the membrane was blocked with 5% non-fat milk, and blotted with anti-BMAL1 Ab (Sigma SAB4300614, rabbit, 1:1000 dilution) or anti-tubulin (Sigma T6557, mouse, 1:1000 dilution). Secondary Ab F(ab’) 2-goat anti-rabbit IgG (H + L) horseradish peroxidase (HRP) or Ab F(ab′)2-goat anti-mouse IgG (H + L) HRP (Invitrogen, 1:10 000 dilution) were used and the result was visualized using a Bio-Rad imager. BMAL1 bands were quantified and normalized to the tubulin loading control using ImageJ software.

For RNA sequencing, all the procedures were performed by the Interdisciplinary Center for Biotechnology Research at the University of Florida. PolyA messenger RNA (mRNA) was isolated from 500 ng of total RNA for each sample by using the NEBNext poly(A) mRNA magnetic isolation module (E7490L), and the RNAseq library was prepared using the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (E7760L). Paired-end (150 bp) sequencing to a read depth of ∼ 40 m reads was performed.

RNA sequencing data analysis was performed as described before.¹⁶ FastQC and MultiQC were used to check sequencing quality. Sequencing reads were aligned to Mus musculus genome GRCm38 (10 mm) using HISAT2. Gene expression counts were analyzed using HTseq tools.

Analysis of circadian rhythmicity and differential circadian gene pattern were performed as described previously using DiffCircadian (https://github.com/diffCircadian/diffCircadian).^16,18,19 We used a q < 0.01 as the cutoff for selection of mRNAs expressed in a circadian pattern and differential circadian gene pattern between groups. DiffCircadian software utilizes cosinor model fitting across multiple time points, leverages the continuous nature of the time-course data, and provides a robust framework for detecting rhythmic patterns with less samples per time point. Moreover, DiffCircadian detects differential circadian pattern of transcriptomes between different samples.

Nonrhythmic differential expression analysis was performed by using DESeq2 (https://bioconductor.org/packages/release/bioc/html/DESeq2.html) in R Studio.²⁰ Gene expression at each time point was treated as a replicate within a given genotype to assess differential expression independent of time of day. Genes that did not meet a total of 100 counts across all samples were filtered out prior to analysis. We set the statistical cutoff for significance using a false discovery rate (FDR) < 0.05 and |log2FoldChange| ≥ 0.5.

Gene ontology was analyzed using DAVID (david.ncifcrf/gov/). Enriched biological pathways were defined as those with FDR < 0.05.

BMAL1 protein expression was compared between groups using t-test, and gene expression of select genes was compared between sexes and genotypes using two-way ANOVA followed by Turkey’s multiple comparisons test. Analysis was performed using GraphPad Prism 10.4.0. The threshold for statistical significance was P < 0.05. Corresponding symbols to highlight statistical significance are as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Results

Male and Female Hearts Exhibit Differences in Their Circadian Transcriptomes

The circadian clock mechanism regulates a daily program of rhythmic gene expression in every cell, which contributes to time-of-day changes in cell and tissue physiology. To study the circadian transcriptome in male and female hearts, total RNA from both vehicle and iCS Bmal1 KO mice was isolated from ventricular tissue collected every 4 hours for 48 hours under constant darkness from circadian time (CT) 18 to CT62 (n = 2 mice per timepoint; CT represents the timing of the internal circadian clock and is based on the organism’s intrinsic circadian rhythm independent of external stimuli; CT0 refers to the start of subjective day or the rest phase for nocturnal animals such as mice). The sequencing data was analyzed using DiffCircadian to identify REGs. Table S1 provides the number of REGs identified at different P-values or q-values. For all downstream analyses, we defined REGs as those mRNAs with a q < 0.01.

First, we looked to compare the expression of the core clock factors between male and female vehicle hearts. We plotted the gene expression of Bmal1, Clock, Per1, Per2, Cry1, Cry2, Nr1d1, Nr1d2, Rora, and Rorb mRNAs over 48 hours (Figure 1A). Core clock gene expression appeared quite similar between sexes, and further analysis determined that there were no significant differences in the amplitude of expression. However, DiffCircadian analysis revealed some small but significant differences in the time of peak expression for a few core clock genes between male and female hearts. For example, Bmal1 and Nr1d1 gene expression peaks about 1.1 hours and 1.5 hours earlier in females than in males, respectively. In addition, the mean expression levels (MESOR) of Clock, Cry1, and Cry2 genes are higher in females. No other differences were detected in the expression of core clock genes (Table S1).

Next, we analyzed the REGs of the circadian transcriptome and identified 1435 REGs in the male hearts (Table S2) and 3277 REGs in the female hearts (Table S3). We then looked to determine the number of common REGs between the male and female hearts and found that only 421 REGs were shared between males and females. The majority of REGs were sex-specific with 1014 (71%) male-specific and 2856 (87%) female-specific REGs (Table S4, Figure 1B), indicating a large difference of ∼1800 REGs with ∼2× more REGs in the female heart than the male heart. As expected, the heatmap of z-score normalized gene expression over 48 hours from both male- and female-specific REGs showed clear sex-specific time-of-day rhythmic patterns (Figure 1C).

Male and Female Hearts Exhibit Differences in the Functional Enrichment and Temporal Distribution of REGs

To identify the functional pathways of the shared and sex-specific REGs, we performed gene ontology analysis using DAVID.^21,22 Using a statistical cutoff of FDR < 0.05, we identified 8 biological pathways from the shared REGs, which primarily included pathways related to circadian rhythm and its regulation (Figure 2A, Table S4). The male-specific REGs were enriched for 12 biological pathways compared to 25 enriched biological pathways from the female-specific REGs. We note that there were 2 common pathways enriched by male-specific and female-specific REGs: DNA damage response and protein transport. Most other functional pathways were sex-specific with the male-specific REGs being enriched for translation, aerobic respiration, mitochondrial translation, and response to oxidative stress. In contrast, the biological pathways enriched by female-specific REGs included protein folding, collagen fibril organization, angiogenesis, vasculogenesis, GTPase activity, chromatin organization, and apoptotic processes. These data suggest that the divergent circadian transcriptomes are largely comprised of sex-specific REGs, which contribute to sex-specific functional pathways.

A unique feature of REGs is that their expression peaks at a specific time of day. The DiffCircadian analysis provides the expected time of day at which a REG peaks in expression and defines this value as peak time. By plotting the number of REGs peaking at specific times across the day, we can observe their temporal distribution and identify the times of day when most genes peak in both male and female hearts. We found that the temporal distribution of the 421 shared REGs was noisy and not clearly different between males and females (Figure 2B). In contrast, we noted distinctly different distributions for the sex-specific REGs (Figure 2C). The majority of male REGs peak just before the transition from the rest/light to active/dark phase and before the transition from the active/dark to rest/light phase. In contrast, the temporal patterning of the REGs in females exhibited a bimodal pattern with broader groups of genes peaking in the middle of the rest/light and active/dark phases of the day. This large temporal difference in the distribution of the circadian transcriptome between males and females cannot be simply explained by the small differences in core clock gene expression observed above, suggesting that there are sex-specific co-factors modulating the REG expression patterns.

We next asked if there was any expression pattern in the sex hormone receptors in the heart that could be implicated in the patterns of the male and female REGs.^23,24 Specifically, we asked whether differences in the expression of the androgen receptor (Ar) and estrogen receptor (Esr1) may provide a clue for the sex differences observed in the cardiac circadian transcriptome. Interestingly, we observed no differences using DiffCircadian in the magnitude or pattern of Ar and Esr1 expression between male and female hearts over 48 hours (Figure S1, Table S1). This suggests that sex-specific clock output in the heart cannot be explained by differences in the expression of the core clock factors or the sex hormone receptor mRNAs.

The Biological Pathways Enriched by the Cardiac Circadian Transcriptome Is Largely Sex- and Time-of-Day-Specific

The sex-specific temporal pattern of peak REG expression was striking and suggested that there could be some time-of-day-specific biological processes. To identify the temporal pattern of biological processes, we binned all REGs by their peak time into six 4-hour phases assigned to their respective zeitgeber times (ZT). For reference, ZT0-4 hours, early light phase; ZT4-8 hours, middle light phase; ZT8-12 hours, late light phase; ZT12-16 hours, early dark phase; ZT16-20 hours, middle dark phase; and ZT20-24 hours, late dark phase. Once the REGs were binned into a time-of-day phase, we performed gene ontology analysis to assign biological pathways to each phase (FDR < 0.05). As shown in Figure 3A, the biological pathways in the male heart reflect the distribution of the male REGs observed in Figure 2 (Tables S5 and S6), with the largest cluster of significantly enriched pathways at the late rest/light phase and secondarily at the late active/dark phase. In females, the distribution of enriched biological pathways (Table S8) is also consistent with the REGs (Table S7) but demonstrates the largest cluster of pathways in the middle dark/active phase with small but broader pathway coverage throughout the light/rest phase (Figure 3B). Most pathways identified were sex-specific and unique to the time of day. In male hearts, these included mitochondrial ATP synthesis, mitochondrial translation, and aerobic respiration enriched in the late light/rest phase. While in female hearts, protein folding, protein stabilization, and mRNA processing were enriched in the middle dark/active phase (Figure 3C). This temporal pathway analysis also identified biological pathways that were shared between sexes, but interestingly, these shared pathways peaked at different phases of the day (Figure 3C) and were enriched by sex-specific REGs. For example, protein transport was enriched in the middle rest/light phase in the female hearts (Figure 3C, F4-8) and the late active/dark phase in male hearts (Figure 3C, M20-24). Exemplary of this difference is the expression of Exoc1 and Exoc4, both being genes involved in protein transport. Interestingly, Exoc1 is rhythmically expressed only in females and peaks in middle rest/light phase (ZT4.9). On the other hand, Exoc4 expression oscillates only in males and peaks in late active/dark phase (ZT22.4) (Figure S2). Similarly, RNA splicing was enriched at the early active/dark phase in female hearts (Figure 3C, F16-20) versus the late rest/light phase in male hearts (Figure 3C, M8-12). Snrnp25 and Snrnp48 are genes involved in RNA splicing. Snrnp25 is rhythmically expressed only in males and peaks at late light/rest phase (ZT9.7). In contrast, Snrnp48 expression is rhythmic only in females and peaks in the middle dark/active phase (ZT18.7) (Figure S2). These observations highlight an important point that there are some common biological pathways regulated by the clock mechanism in male and female hearts, but the timing of their expression appears offset in phase.

Bmal1 Knockout in Cardiomyocytes Blunts the Circadian Transcriptome in Male and Female Hearts

The focus of the next experiments was to address how disruption to the cardiomyocyte circadian clock mechanism, through cardiomyocyte-specific deletion of Bmal1, impacts the rhythmic expression of genes in male and female hearts. The circadian clock mechanism is a cell-autonomous transcription-translation negative feedback loop consisting of several genes.²⁵  Bmal1 is the only nonredundant core clock gene and loss of Bmal1 results in loss of circadian clock function.^26,27 Initial recombination specificity in iCS Bmal1 KO mice was assessed by PCR as previously described.¹⁴  Figure S3 shows that Bmal1 mRNA was blunted in male and female iCS Bmal1 KO hearts. We determined that BMAL1 protein expression was not significantly different between sexes in either the vehicle (Figure S4A and B) or iCS Bmal1 KO hearts (Figure S4C and D). But as expected, BMAL1 protein levels were blunted in iCS Bmal1 KO hearts compared to the vehicle in both sexes (Figure S4E-H) with a ∼40%-50% reduction (Figure S4I). Since cardiomyocytes make up ∼35%-40% of the cells in the heart,²⁸ the remaining Bmal1 mRNA and protein expression is largely attributed to contributions from other cell types. As expected, we found that the expression of other core clock genes was significantly altered in iCS Bmal1 KO hearts, yet no significant sex differences were observed (Figure S3). These results confirmed that Bmal1 expression is significantly decreased in our mouse model and is consistent with previous reports.¹⁴

Next, we analyzed the circadian transcriptome in the iCS Bmal1 KO heart. As shown in Table S1 and in agreement with other studies,^8,14 the number of REGs significantly decreased in both male and female iCS Bmal1 KO hearts compared to vehicle-treated hearts. In males, the loss of Bmal1 decreased the number of REGs by ∼79% with only 307 REGs identified in male iCS Bmal1 KO hearts (Figure 4A, Table S9). Of the 307 REGs, we observed 218 REGs unique to the iCS Bmal1 KO hearts (Table S10); these genes were not rhythmic in the vehicle but gained rhythmicity in the iCS Bmal1 KO hearts as seen in Figure 4B.

We then asked whether loss of cardiomyocyte Bmal1 altered the temporal distribution of REGs over 24 hours. In contrast to the pattern seen with the REGs in the male vehicle-treated hearts, the REGs in the iCS Bmal1 KO hearts were more evenly distributed throughout the day (Figure 4C). The largest changes occurred within the clusters of genes peaking at the late dark/active phase and late light/rest phase in the iCS Bmal1 KO hearts. This indicates that these clusters of rhythmic genes are largely regulated by the cardiomyocytes and that the temporal expression pattern of these genes requires cardiomyocyte-specific Bmal1. With so few rhythmic genes, there are limited biological pathways enriched in the male iCS Bmal1 KO, but they can be found in Figure S5 and Tables S10 and S11.

The effect of iCS Bmal1 KO on the cardiac circadian transcriptome was next examined in female mice. Surprisingly, we identified 1799 REGs (Table S12), reflecting a ∼45% decrease compared to the vehicle-treated heart, and which is much smaller compared to the ∼79% reduction in the male iCS Bmal1 KO hearts (Figure 4D). We found that 58% (1042) of the REGs in the female iCS Bmal1 KO heart were shared with the vehicle-treated mice (Figure 4D and E). This outcome is quite striking, so we queried expression of Arntl2 and Npas2 to ask if there were any compensatory changes in the female iCS Bmal1 KO hearts compared to the male, but we did not find any significant difference between sexes (Figure S6, Table S1). This raises an interesting possibility that a novel factor can compensate for the loss of Bmal1 in the female cardiomyocytes or that there is a larger contribution to cardiac clock output from noncardiomyocyte cells in the female heart. We also identified 757 iCS Bmal1 KO-specific REGs (Table S13) that were not detected as rhythmic in the vehicle-treated heart but gained rhythmicity in the iCS Bmal1 KO hearts (Figure 4D and E and Table S13).

We then analyzed the temporal distribution of the REGs in the female iCS Bmal1 KO heart over 24 hours. Like the changes in the male heart, we found that the 2 clusters of REGs that peak in the middle of the rest/light phase and active/dark phase in the female vehicle were no longer distinct in the iCS Bmal1 KO hearts (Figure 4F and Table S14). The pattern was a broader distribution with a larger number of REGs that peak over the transition from the rest/light to active/dark phase and fewer REGs that peak in the middle of the rest/light and active/dark phases. This significant change in the temporal distribution of REGs confirms, like the male hearts, that Bmal1 in the female cardiomyocytes is necessary for the time-of-day distribution of REGs. This also suggests that the distribution of REGs in the female heart is driven by clock output from both cardiomyocytes and other types of cells, but that the loss of Bmal1 in cardiomyocytes results in a shift toward the detection of noncardiomyocyte clock output in this tissue analysis. In contrast to the males, female iCS Bmal1 KO REGs are enriched for many biological processes, which can be found in Figure S7 (Tables S13 and S14).

Inducible Cardiomyocyte-specific Bmal1 Knockout Results in a Loss of Sex Differences in the Cardiac Circadian Transcriptome

Due to the large changes observed in the cardiac circadian transcriptome in both male and female iCS Bmal1 KO hearts, we next asked whether loss of cardiomyocyte Bmal1 resulted in changes to the sex-specific gene expression observed in the vehicle-treated hearts. As seen in Figure 5A, we found 209 REGs out of a total of 307 male iCS Bmal1 KO REGs were shared with the female iCS Bmal1 KO REGs, representing a ∼68% overlap, much larger than 30% overlap between the vehicle-treated hearts. The 209 shared REGs were enriched for biological pathways, including rhythmic process and circadian rhythm. For the female-specific iCS Bmal1 KO REGs, gene ontology analysis identified 23 enriched biological pathways, with no pathways significantly enriched by the 98 male-specific REGs (Table S15, Figure 5B). Female-specific processes included pathways in common with vehicle hearts such as mRNA processing, chromatin organization, and lipid metabolic processes. The large number of REGs and enriched pathways in the female iCS Bmal1 KO heart compared to the male further suggests that female hearts harbor additional mechanisms independent of Bmal1 that contribute to rhythmic gene expression and/or there is a greater contribution from other noncardiomyocyte cells to the cardiac circadian transcriptome in females.

Next, we compared the temporal distribution of the male and female iCS Bmal1 KO REGs. As shown in Figure 5C, the distribution of shared REGs in the male and female iCS Bmal1 KO hearts was more similar than what was seen in the vehicle-treated hearts (Figure 2C). The female REGs exhibited a peak cluster at a slightly advanced phase compared to the male, but overall, the pattern was much more similar. With the sex-specific REGs (Figure 5D), the sex difference was driven primarily by the number of REGs rather than the time-of-day pattern. Compared to those obtained with the vehicle-treated REGs, these temporal patterns demonstrate a significant loss of sex differences in the cardiac circadian transcriptome after cardiomyocyte-specific deletion of Bmal1.

Beyond REGs, Differential Gene Expression Between Sexes Is Significantly Diminished in iCS Bmal1 KO Mice

The significant loss of sex-specific patterns in the REGs of male and female hearts of the iCS Bmal1 KO mice prompted us to ask whether loss of cardiomyocyte Bmal1 impacted sex-specific nonrhythmic gene expression in the heart. To address this question, we performed differential gene expression analysis with DESeq2 using our dataset as described in the methods.²⁰ In the vehicle-treated mice, we identified 3080 DEGs between male and female hearts (Figure 6A; Table S16). In contrast, in the iCS Bmal1 KO mice, we observed only 369 DEGs (Table S16) between male and female hearts, which reflects an 82% decrease in sex-specific DEGs compared to the vehicle hearts (Figure 6A).

Gene ontology analysis of the DEGs in vehicle hearts identified pathways such as the inflammatory response and cytoplasmic translation as being enriched by DEGs upregulated in males with cell adhesion and signal transduction pathways being enriched by DEGs upregulated in the females (Figure 6B). Looking at the DEGs between sexes in the iCS Bmal1 KO hearts found no pathways enriched for the 150 DEGs upregulated in males, while ossification was a newly identified pathway for the 219 DEGs upregulated in females. The inclusion of the ossification pathway appears to largely reflect enrichment from genes related to WNT signaling. Consistent with the female vehicle hearts, DEGs contributing to cell adhesion and signal transduction continued to be enriched in female iCS Bmal1 KO hearts (Figure 6C). The dramatic drop in sex-specific gene expression in iCS Bmal1 hearts was striking and argues that the cardiac transcriptome becomes less different between sexes following iCS Bmal1 KO, indicating that cardiomyocyte Bmal1 and the circadian clock have a role in influencing sex-specific gene expression in the mouse heart.

This unexpected finding led us to ask whether differences in the expression of transcription factors, including nuclear receptors related to sex differences (Esrra, Esrrb, Esrrg, Esr1, Ar, and Gr)²⁹ and known cardiac transcription factors (Gata4, Mef2a, Mef2c, Nkx2-5, Srf, Tbx5, and Tead1),³⁰ exhibited any patterns that would be consistent with the loss of sex-specific DEGs. We compared the expression of each gene between sexes and genotypes. Interestingly, as shown in the heatmaps in Figure 6D and E, expression of many of these genes appeared different between male and female vehicle-treated hearts, including Esrrb, Esrrg, Gr, Gata4, Mef2a, Nkx2-5, Srf, Tbx5, and Tead1. In general, we noted that the expression of these factors is largely higher in females compared to males with the exception of the androgen receptor (Ar). When we queried the expression of these factors in the iCS Bmal1 KO hearts, we found that cardiomyocyte-specific loss of Bmal1 resulted in significant changes to their expression. In particular, the expression of many of these factors, including Esrrb, Esrrg, Gata4, Mef2a, Nks2-5, Srf, Tbx5, and Tead1, lost sex-specificity and was much more similar in iCS Bmal1 KO hearts. Uniquely, however, Ar became differentially expressed in iCS Bmal1 KO hearts with expression significantly increasing in females and decreasing in males in the absence of cardiomyocyte Bmal1. These observations were tested statistically through 2 approaches, 1 with DESeq2 and secondarily with two-way ANOVA; these results are provided in Table S16 and Figure S8. A significant interaction effect (P < 0.05) was observed for Esrrb, Esrrg, Gata4, Mef2a, Nks2-5, Srf, Tbx5, and Tead1 with no interaction effects for Esr1, Esrra, and Mef2c. This analysis suggests that that loss of cardiomyocyte Bmal1 in males results in a greater change in the expression of these transcription factors compared to females. We note that the loss of Bmal1 function occurred in the adult mouse and was targeted to cardiomyocytes, which would not be predicted to impact circulating levels of sex hormones. While there is still much to learn, these results indicate that Bmal1, and maybe the core clock, are upstream modulators of the sex-specific expression of important transcription factors in the heart.

Discussion

In this resource paper, we determined that the circadian transcriptome of the mouse heart exhibits significant sex differences. This was apparent in the quantity, identity, and temporal distribution of REGs and the biological processes enriched by those REGs. The sex-specificity of cardiac REGs is more distinct than the sex differences reported in the mouse liver.¹² In particular, while there were more REGs in the female liver compared to the male, the difference was much more modest with a difference of only 295 compared to the 1842 differential REGs in the heart. In addition, there were many more REGs shared between sexes in the liver (∼1350) than in the heart (∼420). These observations indicate that sex-specific gene expression is more prominent in the mouse heart with implications for downstream physiology.

This work also identified that the temporal patterning of REG expression between sexes is quite different with the majority of the male REGs peaking around the rest/light to active/dark transition periods, while in the females this occurs in the middle of the rest/light and active/dark phases. Among the shared rhythmic genes, many of them exhibited a difference in phase of about ∼8 hours between males and females. This phase difference in the heart is very similar to what was recently reported in the mouse liver,¹² yet the functional significance of this phase difference is yet be discerned. However, since the expression core clock factors is largely the same between sexes, these data point to the likely existence of sex-specific transcription factors cooperating with the core clock factors to induce sex differences in rhythmic gene expression. To confirm these observations were not unique to the statistical approach we used, we also performed rhythmic gene expression analysis using MetaCycle¹⁹ and found that the majority of the REGs identified were the same as those found with DiffCircadian. Similarly, MetaCycle analysis identified more REGs in females than males and displayed a temporal distribution of REGs similar to that observed with DiffCircadian (data not shown).

One of the most unexpected outcomes of this study was that the loss of Bmal1 in cardiomyocytes led to an over 8-fold decrease in differential gene expression between male and female hearts. We expected a potential impact on sex-specific cardiac REGs, but the 88% decline in the number of DEGs between the male and female hearts was striking. To begin to address the potential mechanisms that link Bmal1, and potentially the core circadian clock, with sex-specific gene expression, we queried specific nuclear receptors linked to sex-specific gene expression as well as a group of well-known cardiac-specific transcription factors. This analysis identified that the expression of many nuclear receptors, such as Esrrb, as well as cardiac transcription factors, such as Tbx5 and Gata4, exhibit strong sex differences. In general, these factors are more highly expressed in the hearts of females compared to male mice. The differential expression of these factors between sexes was lost when cardiomyocyte Bmal1 was knocked out, suggesting that the difference in expression at the tissue level is largely coming from the cardiomyocytes. However, future studies including single nuclei RNA sequencing will be necessary to make conclusions. These results also suggest that these sex-specific transcription factors may partner with Bmal1 and the circadian clock mechanism or act downstream of Bmal1 to contribute to sex-specific gene expression. Prior work has found that many commonly expressed transcription factors are not differentially expressed between males and females but rather display sex-specific activity.^31,32 This has put forward that the transcriptional differences between different cell types and sexes are driven by an organized transcription factor network.³³ Our results suggest that Bmal1, and potentially the molecular clock components, is a critical transcriptional node within a network of key transcription factors for sex-specific activity in the heart. We suggest that the sex-specific expression of these factors modulates the timing and output of the transcriptional machinery, resulting in differential gene expression between sexes.^34–36 This would make Bmal1 and the core clock mechanism a key hub for sex-specific gene expression in the heart.

This sex-specific cardiac gene expression analysis may also provide new insight into sex differences observed in cardiovascular pathologies. For example, sudden cardiac death (SCD) happens more frequently in women during the night.³⁷ A prolonged QT interval is a risk factor for SCD, and genetic association analysis has revealed relationships between various ion channels and a prolonged QT interval.^38,39 Specifically, mutations in Kcnq1, Scn5a, Kcne1, Kcnh2, and Kcnj2 are associated with long QT syndrome.³⁹ Here, we found Kcnq1, Scn5a, and Kcne1 oscillate only in females, Kcnj2 oscillates only in males, and Kcnh2 oscillates in both sexes. Interestingly, Kcnq1, Scn5a, and Kcne1 all peak during the rest/light phase in females, while Kcnj2 and Kcnh2 peak during the active/dark phase in males (Tables S2 and S3). Additionally, we found that Kcne1 is more highly expressed in females compared to males as determined by DESeq2 (Table S16). We propose that the differential rhythmicity of genes between sexes could contribute to the sex-specific phenotypes of cardiovascular events such as SCD, and future studies should examine this relationship further.

In conclusion, differential gene expression between sexes has been shown to contribute to sexually dimorphic phenotypes and diseases.³¹ Here, we identified sex differences in the circadian transcriptome in mouse hearts and showed that cardiomyocyte Bmal1, and likely the core clock mechanism, is indispensable for sex-specific gene expression. These findings provide new clues into the mechanisms directing sex-specific gene expression programs and provide important reference data to support our understanding of sex differences in cardiovascular health and disease.

Conflict of Interest

KAE holds the position of Editorial Board Member for Function and is blinded from reviewing or making decisions for the manuscript.

Acknowledgments

We are grateful for the work of Quenten Lakey and Gabriela Ares in isolating RNA for this study. Graphical abstract created with BioRender.com.

Author Contributions

Xiping Zhang (Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing—original draft, Writing—review & editing), Spencer B. Procopio (Data curation, Formal analysis, Investigation, Methodology, Visualization, Writing—original draft, Writing—review & editing), Haocheng Ding (Data curation, Formal analysis, Methodology, Validation), Maya G. Semel (Data curation, Formal analysis, Investigation), Elizabeth A. Schroder (Conceptualization, Investigation, Methodology, Project administration, Resources, Supervision, Writing—review & editing), Mark R. Viggars (Formal analysis, Investigation), Tanya S. Seward (Investigation, Methodology, Resources, Supervision), Ping Du (Investigation, Methodology, Validation, Visualization), Kevin Wu (Investigation, Methodology), Sidney R. Johnson (Investigation), Abhilash Prabhat (Investigation), David J. Schneider (Investigation), Isabel G. Stumpf (Investigation), Ezekiel R. Rozmus (Investigation), Zhiguang Huo (Data curation, Methodology, Software, Supervision, Writing—review & editing), Brian P. Delisle (Conceptualization, Formal analysis, Funding acquisition, Project administration, Resources, Supervision, Writing—review & editing), and Karyn A. Esser (Conceptualization, Formal analysis, Funding acquisition, Project administration, Resources, Supervision, Writing—review & editing)

Funding

This study was supported by National Institutes of Health (NIH) grant R01-HL153042 to K.A. Esser and B.P. Delisle, as well as NIH grant R01-HL141343 to B.P. Delisle.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author. Bulk RNA-seq data are available in the GEO database under accession number GSE262714.

References

Author notes