Histone Deacetylase Inhibition by Gut Microbe-Generated Short-Chain Fatty Acids Entrains Intestinal Epithelial Circadian Rhythms

Abstract

Background & aims: The circadian clock orchestrates ∼24-hour oscillations of gastrointestinal epithelial structure and function that drive diurnal rhythms in gut microbiota. Here, we use experimental and computational approaches in intestinal organoids to reveal reciprocal effects of gut microbial metabolites on epithelial timekeeping by an epigenetic mechanism.

Methods: We cultured enteroids in media supplemented with sterile supernatants from the altered Schaedler Flora (ASF), a defined murine microbiota. Circadian oscillations of bioluminescent PER2 and Bmal1 were measured in the presence or absence of individual ASF supernatants. Separately, we applied machine learning to ASF metabolomics to identify phase-shifting metabolites.

Results: Sterile filtrates from 3 of 7 ASF species (ASF360 Lactobacillus intestinalis, ASF361 Ligilactobacillus murinus, and ASF502 Clostridium species) induced minimal alterations in circadian rhythms, whereas filtrates from 4 ASF species (ASF356 Clostridium species, ASF492 Eubacterium plexicaudatum, ASF500 Pseudoflavonifactor species, and ASF519 Parabacteroides goldsteinii) induced profound, concentration-dependent phase shifts. Random forest classification identified short-chain fatty acid (SCFA) (butyrate, propionate, acetate, and isovalerate) production as a discriminating feature of ASF "shifters." Experiments with SCFAs confirmed machine learning predictions, with a median phase shift of 6.2 hours in murine enteroids. Pharmacologic or botanical histone deacetylase (HDAC) inhibitors yielded similar findings. Further, mithramycin A, an inhibitor of HDAC inhibition, reduced SCFA-induced phase shifts by 20% (P < .05) and conditional knockout of HDAC3 in enteroids abrogated butyrate effects on Per2 expression. Key findings were reproducible in human Bmal1-luciferase enteroids, colonoids, and Per2-luciferase Caco-2 cells.

Conclusions: Gut microbe-generated SCFAs entrain intestinal epithelial circadian rhythms by an HDACi-dependent mechanism, with critical implications for understanding microbial and circadian network regulation of intestinal epithelial homeostasis.

Keywords: Circadian Rhythm; Histone Deacetylation; Intestinal Organoids; Microbiome; Short-Chain Fatty Acids; Sulforaphane.

Figure 1:

PER2::LUC enteroid oscillations over 72 hours. (A) 25%mLB and ‘non-shifter’ group of ASF species (ASF360, ASF361, ASF502) show no alterations. (B) ASF ‘shifter’ group (ASF356, ASF492, ASF500, ASF519) induces phase delays in PER2::LUC oscillations represented as a forward shift in waveforms (P<0.05 by Mann-Whitney U test, n=3 samples per taxa, ~50 organoids per plate). ASF500 results in decreased PER2::LUC amplitudes.

Figure 2:

Median phase delay (represented by middle horizontal line of samples) in response to mLB or ASF supernatants calculated as forward shift in PER2::LUC oscillations versus untreated controls. ASF ‘shifters’ were taxa inducing significant phase delays of PER2::LUC oscillation versus 25% mLB control (*P<0.05, Mann Whitney U test).

Figure 3:

A) Eleven of 53 detectable metabolites higher in ASF shifters (356, 492, 500, 519) versus non-shifters (360, 361, 502). Metabolites required median values greater than zero. B) With a 5-fold cross-validation, using 20 repetitions, the accuracy random forests to discriminate between phenotypes was 98% with only three features. C) Discriminating metabolites. Y-axis, normalized metabolite concentration relative to fresh media. Values above zero indicate the metabolite was produced by bacteria in culture. Middle lines represent median values.

Figure 4:

PER2::LUC enteroids displaying oscillations over 72 hours. Acetate, butyrate, isovaleric acid and propionate induced phase delays represented as a forward shift in waveforms (P<0.05 by Mann-Whitney U test, n=3 samples per metabolite). Isovaleric acid enhanced PER2::LUC amplitudes.

Figure 5

A. Quantification of metabolite concentrations in bacterial supernatants. For selected metabolites, concentration ranges were ~0.5mM-7mM; with supernatants supplemented at 10% or 25%, translating to 0.05mM-.7mM and 0.125mM-1.75mM, respectively. B. Median phase delay (represented by middle horizontal line of samples) induced by ASF-generated metabolites calculated as a forward shift in PER2::LUC oscillation compared to an untreated control. Shifting metabolites are those causing statistically significant phase delays (forward shift) or phase advances (backward shifts) in PER2::LUC oscillations versus untreated controls (*P<0.05, Mann Whitney U test; 10mM Butyrate was toxic to cells).

Figure 6:

HDAC inhibitors phase shift PER2::LUC rhythms. (A) HDAC inhibitors TSA and SAHA induce phase delay patterns of PER2::LUC similar to phase delays with 1mM butyrate. (B) HDAC inhibitors LP-411, HDAC1–2, MS275 and sulforaphane (SFN) cause significant phase delays in PER2::LUC abundance versus untreated controls (*P<0.05, Mann Whitney U test).

Figure 7:

Mithramycin A, an inhibitor of HDAC inhibition, abrogates butyrate induced phase delay of PER2::LUC. (A) Representative oscillations showing a 3-hour reduction in butyrate induced phase delay of PER2::LUC abundance in the presence of mithramycin A. (B) Reduction in butyrate-induced median phase delay (represented by middle horizontal line of samples) of PER2::LUC oscillations due to mithramycin A (*P<0.05, Mann Whitney U test). (C) Butyrate modulation of Per2 expression is HDAC3-dependent. Tamoxifen induced deletion of HDAC3 in HDAC3^ΔindKO enteroids abrogates the effects of butyrate induced alterations of Per2 expression (**P<0.01).