Research Article

The intestinal microbiota programs diurnal rhythms in host metabolism through histone deacetylase 3

See allHide authors and affiliations

Science  27 Sep 2019:
Vol. 365, Issue 6460, pp. 1428-1434
DOI: 10.1126/science.aaw3134

Microbial entrainment of metabolism

The metabolism of mammals is synchronized to daily cycles relating to sleep and mealtimes. It is not surprising that the gut microbiota, which aids in digestion, should also display daily cycling. Kuang et al. found that the gut microbiota can mediate daily cycles epigenetically (see the Perspective by Bishehsari and Keshavarzian). The microbiota induces rhythmic expression of histone deacetylase 3 (HDAC3) in epithelial cells of the small intestine, but not those of the colon. HDAC3 expression drives oscillations in intestinal metabolic gene expression, especially for nutrient transport and lipid metabolism. HDAC3 also directly activates estrogen-related receptor α, which promotes lipid absorption. Consequently, mice that lack a gut microbiota lack daily regulation of their metabolism and become obese on high-fat chow. Disruption of HDAC3 cycling could be an explanation for human obesity associated with antibiotic damage to the microbiota and with sleep disruption caused by jet lag and nighttime working.

Science, this issue p. 1428; see also p. 1379

Abstract

Circadian rhythmicity is a defining feature of mammalian metabolism that synchronizes metabolic processes to day-night light cycles. Here, we show that the intestinal microbiota programs diurnal metabolic rhythms in the mouse small intestine through histone deacetylase 3 (HDAC3). The microbiota induced expression of intestinal epithelial HDAC3, which was recruited rhythmically to chromatin, and produced synchronized diurnal oscillations in histone acetylation, metabolic gene expression, and nutrient uptake. HDAC3 also functioned noncanonically to coactivate estrogen-related receptor α, inducing microbiota-dependent rhythmic transcription of the lipid transporter gene Cd36 and promoting lipid absorption and diet-induced obesity. Our findings reveal that HDAC3 integrates microbial and circadian cues for regulation of diurnal metabolic rhythms and pinpoint a key mechanism by which the microbiota controls host metabolism.

Mammalian metabolism is acutely sensitive to environmental cues. These cues include circadian day-night light cycles, which govern when food is consumed, and the microbiome, which affects how food is digested. The intestinal microbiota shapes the expression of host metabolic pathways and influences the development of obesity (14). Environmental light signals entrain rhythms in gene expression that are synchronized with the day-night light cycle through the circadian clock (57). This circadian synchronization is fundamental to metabolic processes that must be coupled to diurnal sleep-wake and feeding-fasting cycles. It is becoming clear that there is cross-talk between the microbiota and circadian pathways that affects host metabolism (59). However, the molecular mechanisms by which diurnal light signals and the microbiota converge to regulate metabolism are largely unknown.

The microbiota controls diurnal rhythms in epithelial cell histone acetylation

A key mechanism by which the circadian clock generates rhythms in metabolic gene expression is by regulating the recruitment of histone modifiers to chromatin (1013). We therefore investigated whether the microbiota regulates host circadian gene expression by altering histone modifications. We collected small intestinal epithelial cells (IECs) from conventional (CV) and germ-free (GF) mice across a 24-hour cycle (fig. S1A). We then performed RNA sequencing (RNA-seq) and chromatin immunoprecipitation followed by sequencing (ChIP-seq) of histone H3 lysine 9 acetylation (H3K9ac) and H3K27ac, two histone acetylation marks that indicate active promoters and enhancers. As expected, both acetylation marks localized to the regulatory regions of actively transcribed genes (fig. S1, B and C). Both H3K9ac and H3K27ac showed synchronized diurnal oscillations in CV IECs, with peaks between Zeitgeber time 8 (ZT8) and ZT16 (where ZT0 is light on and ZT12 is light off) and troughs between ZT20 and ZT4 (Fig. 1, A and B, and table S2).

Fig. 1 Histone acetylation in small intestinal epithelial cells exhibits synchronized diurnal rhythmicity that depends on the microbiota.

(A) Genome browser view of the 200-kb region surrounding the Nr1d1 locus, showing ChIP-seq analysis results for H3K9ac and H3K27ac marks in IECs. The analysis was done across a 24-hour cycle in CV and GF mice. Each track represents the normalized ChIP-seq read coverage at a single time point. Examples of peaks showing microbiota-dependent diurnal rhythmicity are highlighted in gray. N = 3 pooled biological replicates per library. (B) Heat maps of H3K9ac and H3K27ac signals (reads at 50-bp windows; logarithmic scale) from −1 to +1 kb surrounding the centers of all cycling H3K9ac and H3K27ac peaks (adjusted P < 0.01 by ARSER). Each peak in the genome is represented as a horizontal line, ordered vertically by signal strength, and the analysis was done across a circadian cycle in CV and GF mice. The number of peaks in the genome is indicated at the bottom. The blue-red gradient indicates the coverage or signal strength (normalized uniquely mapped reads per 20 million reads). (C and D) Intensity (C) and amplitude (D) of H3K9ac and H3K27ac peaks in CV and GF mice. *P < 0.05; ***P < 0.001 by one-tailed paired t test. (E) Enriched GO (gene ontology) categories for H3K9ac- and H3K27ac-targeted genes, determined using the DAVID GO analysis tool.

Although the majority of the H3K9ac and H3K27ac marks were present in both CV and GF mice (fig. S1D), most of the peaks did not oscillate in GF mice as they were constitutively high (Fig. 1, A to D). Average signal intensities of both marks were increased and circadian amplitudes were decreased in GF mice (Fig. 1, C and D). Genes involved in metabolic processes, including nutrient transport and lipid metabolism, were enriched near cycling histone acetylation peaks (Fig. 1E) and coincided with dampened oscillations in the abundance levels of transcripts encoding proteins involved in nutrient transport and lipid metabolism (fig. S2, A to C). Thus, the microbiota drives synchronized diurnal oscillations in histone acetylation in the small intestine, and many of the targeted genes regulate nutrient uptake and metabolism. This behavior is distinct from the effect of the microbiota on histone acetylation in the colon, where the overall genome-wide rhythmicity of acetylation is not lost after antibiotic depletion of the microbiota (5).

The microbiota programs rhythmic histone acetylation through HDAC3

H3K9ac and H3K27ac marks cycled diurnally in CV mice but remained constitutively high in GF mice (Fig. 1, A to D), suggesting a possible mechanism involving a histone deacetylase (HDAC) that is expressed and cycling in CV but not in GF mice (schematic diagram in Fig. 2A). HDACs repress transcription by deacetylating histones but can also target nonhistone proteins. We examined expression of multiple Hdac genes (Hdac1 to Hdac11) and found that Hdac1 and Hdac3 were more highly expressed in IECs than the other Hdac genes (fig. S3A). However, only Hdac3 showed differential expression in CV and GF mice, with lowered abundance of Hdac3 transcripts and HDAC3 protein in IECs from GF mice compared with CV mice (Fig. 2, B and C, and fig. S3, B and C). Hdac3 expression increased in IECs after conventionalization of adult GF mice (fig. S3D) and was selectively triggered by monocolonization of GF mice with the commensal bacterial species Bacteroides thetaiotaomicron (fig. S3E). Consequently, HDAC3 protein levels also increased in GF mice monocolonized with B. thetaiotaomicron (fig. S3F). Elevated Hdac3 expression in CV mice required the Toll-like receptor adaptor protein MyD88 (14) (fig. S3G) and, accordingly, MyD88−/− mice showed dampened rhythms in small intestine histone acetylation (fig. S3H and table S2).

Fig. 2 The intestinal microbiota programs the diurnal rhythmicity of epithelial histone acetylation and gene expression through HDAC3.

(A) Schematic diagram summarizing the diurnal histone acetylation patterns in CV and GF mice shown in Fig. 1. The patterns suggest a loss of HDAC function in GF mice. (B) Reverse transcription–quantitative polymerase chain reaction (RT-qPCR) analysis of Hdac3 transcripts in IECs from CV and GF mice at ZT8. n = 3 mice per group. (C) Western blot of HDAC3 and lamin B (loading control) in IECs from protein in IECs from CV and GF mice at ZT8. Bands were quantified by scanning densitometry and normalized to the lamin B band intensity. (D) RT-qPCR analysis of Hdac3 transcript abundance in epithelial cells from GF and CV mice across a 24-hour cycle. n = 3 mice per group. (E) Coimmunoprecipitation of endogenous HDAC3 and NCoR1 in IECs from GF WT mice and CV Hdac3fl/fl and Hdac3ΔIEC mice. n =3 pooled biological replicates per lane. (F and G) Relative abundance of H3K9ac and H3K27ac marks (F) and bound HDAC3 (G) at the promoters of Enpep, Epas1, and Slc25a45 in CV and GF, determined by ChIP-qPCR analysis. n = 3 mice per time point per group. (H) RT-qPCR analysis of Enpep, Epas1, and Slc25a45 transcript abundance in IECs from CV and GF mice or Hdac3fl/fl and Hdac3ΔIEC mice across a 24-hour cycle. n = 3 mice per group. (I) Relative abundance of bound HDAC3 at the promoters of Enpep, Epas1, and Slc25a45 in IECs from WT and Nr1d1−/− mice, determined by ChIP-qPCR analysis. n = 3 mice per time point per group. (J) Genome browser view of the 200-kb region surrounding the Nr1d1 locus; ChIP-seq analysis showed H3K9ac and H3K27ac marks in IECs from Hdac3fl/fl and Hdac3ΔIEC mice. N = 3 pooled biological replicates per library. (K) Heat maps of H3K9ac and H3K27ac signals (reads at 50-bp windows; logarithmic scale) from −1 to +1 kb surrounding the centers of all cycling H3K9ac and H3K27ac peaks. (L) Heat map showing diurnal gene expression patterns in IECs from Hdac3fl/fl and Hdac3ΔIEC mice across a 24-hour cycle, determined by RNA-seq. N = 3 pooled biological replicates per library. (M) Absolute [fragments per kilobase of transcript per million mapped reads (FPKM)] or relative oscillating amplitudes of transcripts in GF WT mice and CV Hdac3fl/fl and Hdac3ΔIEC mice. n = 3 mice per group. (N) Enriched GO categories of genes with decreased cycling amplitudes, determined using DAVID. *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant by two-tailed t test. Means ± SEM (error bars) are plotted.

We reasoned that if HDAC3 is essential for microbiota-driven diurnal rhythms in epithelial histone acetylation, then HDAC3 expression, genome recruitment, or enzymatic activity should be diurnally rhythmic. Hdac3 transcript abundances showed only a modest diurnal oscillation (Fig. 2D), suggesting that the rhythmicity in IEC histone acetylation was not driven by rhythmic Hdac3 expression. In liver and adipose tissue, rhythmicity in histone acetylation is generated when HDAC3 complexes with the nuclear receptor co-repressor (NCoR) (12, 1518), which is recruited to chromatin in a diurnally rhythmic manner by components of the circadian clock (12). We therefore postulated that rhythms in IEC chromatin acetylation might be generated by rhythmic chromatin recruitment of HDAC3 through binding to NCoR.

In support of this idea, we found that HDAC3 complexed with NCoR in IECs and that binding was markedly reduced in IECs from GF mice (Fig. 2E). Thus, the microbiota facilitates formation of HDAC3-NCoR complexes in IECs. We next assessed whether HDAC3 was rhythmically recruited to target genes by studying three genes, Enpep, Epas1, and Slc25a45, that showed pronounced microbiota-dependent rhythms in H3K9ac and H3K27ac marks (Fig. 2F and fig. S4A). As predicted, HDAC3 was recruited to each of these genes in a diurnally rhythmic manner that depended on the microbiota (Fig. 2G), and expression of each gene was higher in GF mice than in CV mice (Fig. 2H). Expression was also higher in mice with an IEC-specific deletion of Hdac3 (Hdac3ΔIEC) than in wild-type (WT) mice (Fig. 2H), consistent with the canonical function of HDAC3 as a co-repressor through histone deacetylation (18). Conventionalization of GF mice also reduced histone acetylation signals at these genes and repressed their expression (fig. S4, B and C). Thus, the microbiota is required for diurnally rhythmic recruitment of HDAC3 to target gene promoters.

To further understand the molecular basis for rhythmic HDAC3 recruitment, we focused on a circadian clock repressor, REV-ERBα (encoded by the Nr1d1 gene), which recruits the HDAC3-NCoR complex to the mouse liver genome in a diurnal manner (12). Binding of HDAC3 to target genes was reduced in the absence of REV-ERBα (Fig. 2I), and genome-wide diurnal changes in H3K9ac marks were lost in Nr1d1−/− mice (fig. S3I and table S2). Thus, the microbiota enhances formation of the HDAC3-NCoR complex, which is recruited rhythmically to target genes by REV-ERBα. This generates diurnal rhythms in histone deacetylation in IECs.

To determine whether HDAC3 drives genome-wide rhythmic histone acetylation in IECs, we detected H3K9ac and H3K27ac marks by ChIP-seq in IECs from Hdac3ΔIEC mice. Both marks showed diurnal changes in Hdac3fl/fl IECs but remained constitutively high in the Hdac3ΔIEC IECs (Fig. 2, J and K, and table S2). We also performed RNA-seq of Hdac3fl/fl and Hdac3ΔIEC IECs across a circadian cycle. Rhythmic oscillations in the abundances of 2729 transcripts were dampened in Hdac3ΔIEC IECs, with amplitudes approximating those seen in GF mice (Fig. 2, L and M). Gene transcripts showing dampened oscillations in Hdac3ΔIEC mice were enriched for nutrient transport and lipid metabolism pathways (Fig. 2N), similar to the enrichment of these pathways among genes with oscillating acetylation marks (Fig. 1E). Rhythmic expression of the clock genes Bmal1 (Arnt1), Per2, and Nr1d1 was maintained in Hdac3ΔIEC IECs (Fig. 3, A and B, and table S3), indicating that the core clock mechanism remains intact. Thus, the intestinal microbiota regulates genome-wide diurnal rhythms in IEC acetylation and gene expression through HDAC3.

Fig. 3 Intestinal epithelial HDAC3 regulates the diurnal rhythmicity of nutrient uptake.

(A and B) Diurnal expression patterns of circadian clock genes in Hdac3fl/fl and Hdac3ΔIEC mouse IECs are represented by a heat map (A), with amplitudes shown in (B). (C and D) Diurnal expression patterns of transporter genes in Hdac3fl/fl and Hdac3ΔIEC mouse IECs are represented by a heat map (C), with amplitudes shown in (D). (E and F) Diurnal oscillations of serum tyrosine, inositol, and riboflavin in Hdac3fl/fl and Hdac3ΔIEC mice (E), with amplitudes shown in (F). n = 3 mice per group. *P < 0.05; **P < 0.01 by two-tailed t test.

Epithelial HDAC3 regulates rhythmic nutrient uptake

The RNA-seq analysis identified several nutrient transporter genes with diurnally rhythmic expression that was dampened in Hdac3ΔIEC mice (Fig. 3, C and D, and table S3). These genes included Slc5a11, Slc52a3, and Slc16a10, which transport aromatic amino acids, inositol, and riboflavin, respectively (1922). Accordingly, Hdac3ΔIEC mice showed dampened daily rhythms for serum concentrations of several metabolites, including tyrosine (an aromatic amino acid), inositol, and riboflavin (Fig. 3, E and F; fig. S5; and table S4). This suggests that epithelial HDAC3 regulates nutrient uptake in the small intestine.

Epithelial HDAC3 controls intestinal lipid absorption

Genes with functions in lipid metabolism also showed dampened diurnal expression rhythms in the Hdac3ΔIEC mice. Seventeen of these genes showed increased overall expression in the Hdac3ΔIEC mice, including Acsl3, Cpt1a/b, Ppara, and Agpat2, which participate in fatty acid β-oxidation and lipid biosynthesis (Fig. 4, A and B, and table S3). In contrast, 18 genes that regulate lipid uptake and processing had decreased overall expression in the Hdac3ΔIEC mice. These genes included Cd36, encoding a fatty acid transporter (23); Fabp4/5, encoding fatty acid binding proteins (24); and Scd1, encoding a stearoyl coenzyme A desaturase (25) (Fig. 4, A and B, and table S3). Consistent with reduced Cd36 expression, Hdac3ΔIEC mice had lower IEC total lipid concentrations and increased fecal lipid concentrations (Fig. 4, C and D). Our finding of decreased lipid uptake differs from that of a prior report of increased lipid stores in IECs of Hdac3ΔIEC mice (26). However, we note that the mice in the prior study were fed a defined safflower lipid diet that differed from the high-fat diet used in our studies (TestDiet AIN-76A), and this might explain the discordant findings. We also observed lower concentrations of serum triglycerides and nonesterified fatty acids (Fig. 4E) and dampened diurnal rhythms in serum triglyceride concentrations in the Hdac3ΔIEC mice (Fig. 4, F and G). Thus, HDAC3 enhances expression of the lipid transporter CD36, stimulates lipid uptake by IECs, and promotes diurnal rhythmicity in serum triglyceride concentrations.

Fig. 4 Intestinal epithelial HDAC3 controls lipid absorption in the intestine.

(A and B) Diurnal expression pattern of lipid metabolic genes in Hdac3fl/fl and Hdac3ΔIEC mouse IECs is represented by a heat map (A) with amplitudes shown in (B). (C) Oil red O staining of lipids in small intestines of Hdac3fl/fl and Hdac3ΔIEC mice fed an HFD. Intestines were harvested at ZT8. Black arrows indicate epithelial cells. Scale bars = 100 μm. (D) Quantification of total lipids (at ZT8) in IECs and neutral lipids in feces from Hdac3fl/fl and Hdac3ΔIEC mice fed an HFD. n = 6 or 5 mice per group. ***P < 0.001 by two-tailed t test. (E) Quantification of triglycerides and free fatty acids (at ZT4) in serum from Hdac3fl/fl and Hdac3ΔIEC mice fed an HFD. n = 6 or 5 mice per group. *P < 0.05; **P < 0.01 by one-tailed t test. (F and G) Diurnal oscillations of serum triglyceride concentrations in Hdac3fl/fl and Hdac3ΔIEC mice fed a chow diet (F) with amplitudes shown in (G). n = 3 mice per group. *P < 0.05 by one-tailed t test. Statistical significance levels for circadian rhythms: P = 0.002 for Hdac3fl/fl mice and P = 0.287 for Hdac3ΔIEC mice. (H and I) Weights (H) and body fat percentages (I) of Hdac3fl/fl and Hdac3ΔIEC mice fed an HFD with or without antibiotics for 10 weeks. n = 5 mice per group. *P < 0.05; **P < 0.01; ns, not significant by one-tailed t test. (J) Weight gain in mice fed an HFD with or without jet lag. Jet lag was induced by an 8-hour light cycle shift every 3 days. n = 5 mice per group. *P < 0.05 two-tailed t test. Means ± SEM (error bars) are plotted.

The lowered lipid uptake into Hdac3ΔIEC IECs suggested that the Hdac3ΔIEC mice might be protected against high-fat diet–induced obesity. Indeed, when placed on a high-fat, high-sugar Western-style diet (HFD) for 10 weeks, the Hdac3ΔIEC mice maintained lower body weights and had lower body fat percentages than Hdac3fl/fl mice (Fig. 4, H and I, and fig. S6A); these findings are in accordance with those of a prior report (26). The Hdac3ΔIEC mice also had increased glucose tolerance, decreased insulin resistance, smaller epididymal fat pads, and less liver fat accumulation (fig. S6, B to D). The metabolic differences between Hdac3fl/fl and Hdac3ΔIEC mice were not due to altered physical activity, energy utilization (fig. S7A), food intake (fig. S7, B and C), or altered microbiota taxonomic composition (fig. S8, A and B), all of which were similar between the two groups. The HFD caused marked changes in the microbiota, but the changes were not HDAC3 dependent (fig. S8, C and D). Thus, epithelial HDAC3 enables HFD-induced obesity in mice, likely by enhancing intestinal lipid absorption.

The microbiota is essential for diet-induced obesity, and thus GF mice have lower body fat percentages and are protected from HFD-induced obesity compared with CV mice (2). The lowered expression and dampened rhythmic recruitment of HDAC3 to epithelial chromatin in GF mice suggest a causal role for epithelial HDAC3 in microbiota-dependent obesity. Depletion of the microbiota through antibiotic treatment produced lowered body weights and body fat percentages in the Hdac3fl/fl mice, resulting in body compositions that did not differ significantly from those of Hdac3ΔIEC mice (Fig. 4, H and I). Thus, epithelial HDAC3 is necessary for the microbiota to promote HFD-induced obesity in mice.

Chronic circadian light cycle disruptions exacerbate HFD-induced obesity in mice (8). Consistent with prior findings (8), Hdac3fl/fl mice subjected to experimental jet lag (an 8-hour shift every 3 days) gained weight faster than Hdac3fl/fl mice on a normal day-night light cycle (Fig. 4J). However, Hdac3ΔIEC mice were protected from the metabolic effects of jet lag and gained weight at a similar rate as Hdac3ΔIEC mice exposed to a normal light cycle. Their overall weight gain was also lower than that in Hdac3fl/fl mice that were either jet-lagged or exposed to a normal light cycle. Thus, epithelial HDAC3 worsens jet lag–induced obesity in mice.

Epithelial HDAC3 controls rhythmic expression of Cd36

We next investigated the molecular mechanisms by which HDAC3 regulates the rhythmic expression of lipid metabolic pathways in the small intestine epithelium. Canonically, HDAC3 acts as a co-repressor through its histone deacetylation activity. The elevated expression of genes involved in fatty acid β-oxidation and lipid biosynthesis in Hdac3ΔIEC mice (Fig. 4A) suggested that HDAC3 functions canonically as a co-repressor of these genes. Accordingly, H3K9ac marks at Cpt1a, Acsl3, Ppara, and Agpat2 oscillated diurnally in conventional mice but were constitutively elevated and nonrhythmic in GF mice and conventional Hdac3ΔIEC mice (Fig. 5, A and B; fig. S9, A and B; and table S5), consistent with HDAC3 promoting histone deacetylation that in turn suppresses transcription of these genes.

Fig. 5 Epithelial HDAC3 regulates expression of the lipid transporter gene Cd36 by coactivating ERRα through PGC1α.

(A and B) Genome browser view of the Cpt1a locus, showing H3K9ac marks in IECs from CV and GF WT mice (A) and in CV Hdac3fl/fl and Hdac3ΔIEC mice (B). Statistical significance for circadian rhythms: P = 0.0007 for CV, P = 0.504 for GF, and P = 0.00012 for Hdac3fl/fl and Hdac3ΔIEC mice. (C and D) Diurnal expression of Cd36 transcript (C) and protein (D) in IECs from Hdac3fl/fl CV mice, Hdac3ΔIEC CV mice, and GF WT mice. n = 3 mice per group. (E) ChIP-qPCR analysis of HDAC3 binding at the Cd36 promoter in CV and GF mice across a day-night cycle. (F) Relative abundance of bound HDAC3 at Cd36 in IECs from WT and Nr1d1−/− mice, determined by ChIP-qPCR analysis. n = 3 mice per time point per group. (G) ChIP-qPCR analysis of H3K9ac and H3K27ac marks at Cd36 in IECs from Hdac3fl/fl and Hdac3ΔIEC mice (at ZT14). (H) Coimmunoprecipitation of endogenous HDAC3 and PGC1α in IECs from Hdac3fl/fl and Hdac3ΔIEC mice. N = 3 pooled biological replicates per lane. (I) ChIP-qPCR analysis of HDAC3, PGC1α, and ERRα shows colocalization at the Cd36 promoter in IECs from CV WT mice. Binding region refers to a known binding site for each protein at the Cd36 locus in adipose tissue (17). n = 3 mice per group. (J) Luciferase reporter assay results for transcription driven by a Cd36 promoter, demonstrating combinatorial effects of HDAC3, PGC1α, and ERRα. Empty vector was used in the control group. n = 3 per group. (K) Model showing how the intestinal microbiota regulates diurnal rhythms in epithelial metabolic pathways through HDAC3. *P < 0.05; **P < 0.01; ns, not significant by two-tailed t test. Means ± SEM (error bars) are plotted.

By contrast, genes that control lipid uptake, transport, and processing, including Cd36, Scd1, and Fabp4, showed lowered expression in Hdac3ΔIEC IECs (Fig. 4A). This suggested that HDAC3 regulates expression of these genes either indirectly or through a noncanonical coactivation mechanism. Several of the genes in this group (including Cd36, Scd1, and Fabp4) are induced by the microbiota through the circadian transcription factor NFIL3 (7), suggesting that HDAC3 might regulate their expression by controlling Nfil3 expression. However, Nfil3 expression was not markedly altered in Hdac3ΔIEC mice and remained rhythmic (Fig. 3, A and B), arguing against a role for NFIL3.

To further understand the mechanism of gene activation by HDAC3, we studied the fatty acid transporter gene Cd36. Cd36 expression in the intestinal epithelium oscillated diurnally in conventional Hdac3fl/fl mice, but the oscillations were dampened and overall transcript and protein levels were lowered in GF mice and CV Hdac3ΔIEC mice (Fig. 5, C and D). This finding supports the idea that the microbiota promotes Cd36 expression and hence lipid absorption through HDAC3. HDAC3 was recruited rhythmically to Cd36 in a microbiota- and REV-ERBα–dependent manner (Fig. 5, E and F), but H3K9ac and H3K27ac marks at Cd36 were not markedly altered in Hdac3ΔIEC mice (Fig. 5G and fig. S10). These data suggest that HDAC3 regulates Cd36 expression directly but not through histone acetylation.

In adipose tissue, HDAC3 functions as a coactivator of estrogen-related receptor α (ERRα) (17). Transcriptional activation by ERRα requires the proliferator-activated receptor gamma coactivator 1α (PGC1α), which is in turn activated by HDAC3-mediated deacetylation (17, 27). We therefore considered whether a similar mechanism underlies HDAC3 activation of Cd36 expression in IECs. Supporting this idea, analysis of published genome-wide ChIP-seq data from adipose tissue showed simultaneous recruitment of HDAC3 and ERRα to multiple Cd36 promoter and enhancer sites (17) (fig. S11A). Further, HDAC3 coimmunoprecipitated with PGC1α in IECs, indicating a direct physical interaction between the proteins in epithelial cells (Fig. 5H). HDAC3, PGC-1α, and ERRα colocalized at Cd36 in IECs (Fig. 5I), and HDAC3 markedly increased the transcriptional activity of ERRα at Cd36 (Fig. 5J and fig. S11B). This increase occurred with WT PGC1α ut not with a mutant PGC1α that cannot interact with ERRα (17). Thus, the microbiota promotes diurnally rhythmic expression of intestinal Cd36 through HDAC3 and its function as a coactivator of ERRα (Fig. 5K).

Outlook

Circadian rhythmicity is a defining characteristic of mammalian metabolism that coordinates expression of cellular metabolic machinery with environmental light cycles. Our finding that the intestinal microbiota programs the daily rhythmic expression of small intestine metabolic networks illuminates an essential role for the microbiota in regulating host metabolism and indicates that the microbiome, the circadian clock, and the mammalian metabolic system have tightly coevolved. We identified epithelial HDAC3 as a key mechanism that integrates inputs from the microbiota and circadian light cycles and conducts these signals to host metabolic genes. The microbiota–circadian clock interaction mediated by HDAC3 regulates intestinal lipid uptake and likely evolved as a mechanism to enhance the efficiency of dietary lipid absorption. However, in the context of a high-fat, high-sugar Western-style diet, this interaction enables obesity and the worsening of obesity by circadian clock disruption. Our results also suggest how disruption of microbiota-clock interactions, such as through antibiotic treatment or chronic circadian disruptions including jet lag, could worsen human metabolic disease. Finally, these findings point to new avenues for treating metabolic disease by chemical or microbiological targeting of HDAC3.

Supplementary Materials

science.sciencemag.org/content/365/6460/1428/suppl/DC1

Materials and Methods

Figs. S1 to S11

Tables S1 to S5

References (2841)

References and Notes

Acknowledgments: We thank B. Hassell and M. Robinson for assistance with mouse experiments. Funding: This work was supported by NIH grant R01 DK070855 (L.V.H.), Welch Foundation grant I-1874 (L.V.H.), the Helen D. Bader Center for Research on Arthritis and Autoimmune Diseases (L.V.H.), and the Howard Hughes Medical Institute (L.V.H.). Z.K. was supported by NIH T32 AI005284 and NIH K99 DK120897. C.Y. was supported by NIH K99 GM129415. Author contributions: Conceptualization, Z.K. and L.V.H.; Methodology, Z.K., Y.W., C.Y., and L.V.H.; Investigation, Z.K., Y.W., Y.L., C.Y., K.A.R., and C.L.B.; Resources, Z.K., E.N.O, and L.V.H.; Writing – original draft, Z.K. and L.V.H.; Writing – review & editing, Z.K., Y.W., and L.V.H.; Visualization, Z.K. and Y.W.; Supervision, Z.K. and L.V.H.; Project administration, Z.K. and L.V.H.; Funding acquisition, Z.K., C.Y., and L.V.H. Competing interests: The authors have no competing interests. Data and materials availability: All data and code to understand and assess the conclusions of this research are available in the main text, supplementary materials, and via the Gene Expression Omnibus repository, with accession number GSE134303.
View Abstract

Stay Connected to Science

Navigate This Article