Introduction

Modern investigation into human disease relies heavily on genetic and genomic approaches, and most assume that human disease is primarily driven by variation in the human genome. However, in the post genomic era we now understand that human genetic variation explains only a small proportion of risk for complex diseases such as obesity, diabetes, and cardiovascular disease (CVD). Instead, in many cases environmental factors are the predominant drivers of cardiometabolic disease pathogenesis. Among all contributing environmental factors, it is clear that dietary patterns, and diet-driven alterations in the gut microbiome, can profoundly impact many common human diseases.^1–4 There are now many examples of diet-microbe-host interactions shaping human disease, but one of the more compelling is the reproducible link between elevated trimethylamine N-oxide (TMAO) levels and cardiovascular disease (CVD) risk.^5–12 TMAO is generated by a metaorganismal (i.e. microbe & host) pathway where dietary substrates such as choline, L-carnitine, and γ-butyrobetaine are metabolized by gut microbial enzymes to generate the primary metabolite trimethylamine (TMA).^5–12 TMA is then further metabolized by the host enzyme flavin-containing monooxygenase 3 (FMO3) in the liver to produce trimethylamine-N-oxide (TMAO).¹⁰ Elevated TMAO levels are associated with many human diseases including diverse forms of CVD^5–12, obesity^13,14, type 2 diabetes^15,16, chronic kidney disease (CKD)^17,18, neurodegenerative conditions including Parkinson’s and Alzheimer’s disease^19,20, and several cancers^21,22. Many of these disease associations have been validated in several large population meta-analyses^23–25 and Mendelian randomization studies^26,27. The emerging body of evidence supports the notion that elevated TMAO levels are causally related to cardiometabolic disease pathogenesis. In further support, therapies aimed at lowering circulating TMAO levels provide striking protection against cardiometabolic disease in animal models^28–36. Several independent studies have now shown that inhibition of either the host TMAO-producing enzyme FMO3 or the microbial TMA-producing enzyme CutC protects against diet-induced atherosclerosis^15,30, heart failure³², thrombosis^29,31, obesity^13,33, liver disease³⁴, insulin resistance^13,33, CKD^35,36, and abdominal aortic aneurysm³⁷.

Even though TMAO lowering therapies are very effective in preclinical animal models, the underlying mechanism(s) linking the metaorganismal production of TMAO to cardiometabolic disease pathogenesis are still incompletely understood. Mechanistic understanding has been hampered by the fact that it is hard to disentangle the pleiotropic effects of dietary substrates (choline, carnitine, γ-butyrobetaine, trimethyllysine, etc.), bacterial production of TMA, and host-driven conversion of TMA to TMAO. At this point, the vast majority of studies have focused on the end product of this pathway TMAO, but it is equally plausible that the primary metabolite TMA may play some role in microbe-host crosstalk related to human disease. There is some evidence that TMAO can promote inflammatory processes via activation of the nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing-3 (NLRP3) inflammasome and nuclear factor κB (NFκB).^38–41 TMAO can also activate the endoplasmic reticulum (ER) stress kinase PERK (EIF2AK3) in hepatocytes to promote metabolic disturbance.²⁸ In parallel, TMAO promotes stimulus-dependent calcium release in platelets to promote thrombosis.⁸ Although the end product of the pathway TMAO clearly impacts cell signaling in the host to impact cardiometabolic disease, these TMAO-driven mechanisms do not fully explain how elevated TMAO levels contribute to so many diverse diseases in humans. In addition to TMAO-driven signaling mechanisms, we recently reported that major components of the TMAO pathway (choline, TMA, FMO3, and TAAR5) oscillate in a highly circadian fashion.³³ Furthermore, gut microbe-targeted drugs that selectively block TMA production alter host circadian rhythms in the gut microbiome and host phospholipid metabolism.³³ It is important to note that disruption of the circadian clock is a common hallmark of almost all diseases where TMAO levels are elevated^42–45. To follow up on the potential links between the TMAO pathway and host circadian disruption here we have used genetic knockout approaches at the level of host sensing of TMA (i.e. Taar5^-/-), gut microbial TMA production (i.e. cutC-null microbial communities), and host TMA oxidation (i.e. Fmo3^-/-). Results here further bolster the concept that TMA production and associated TAAR5 activation shapes host circadian rhythms.

Results

Mice Lacking the TMA Receptor TAAR5 Have Altered Circadian Rhythms in Metabolic Homeostasis and Innate Olfactory-Related Behaviors

We recently demonstrated that drugs blocking gut microbial TMA production protect against obesity via rewiring circadian rhythms in the gut microbiome, liver, white adipose tissue and skeletal muscle.³³ Furthermore, we also showed that blocking bacterial TMA production elicited unexpected alterations in olfactory perception of diverse odorant stimuli.⁴⁶ Therefore, we have followed up here to further interrogate circadian rhythms in the gut microbiome, liver, white adipose tissue (WAT), skeletal muscle, and olfactory bulb in mice lacking the only known host G protein-coupled receptor (GPCR) that senses TMA known as TAAR5.^47,48 In agreement with our previous report showing that Taar5 mRNA is expressed in a circadian manner in skeletal muscle³³, LacZ reporter expression oscillates with peak expression in the dark cycle in skeletal muscle (Figure 1A). However, unlike the striking impact that choline TMA lyase inhibitors have on the core circadian clock machinery in skeletal muscle³³, mice lacking the TMA receptor (Taar5^-/-)also have largely unaltered circadian gene expression in skeletal muscle, with only nuclear receptor subfamily 1 group D member 1 (Nr1d1, Rev-Erbα) showing a modest yet significant delay in acrophase (Figure 1A; Table 1). Instead, Taar5^-/- mice have alterations in the expression of key circadian genes including basic helix-loop-helix ARNT like 1 (Arntl;Bmal), clock circadian regulator (Clock), Nr1d1, Rev-Erbα, cryptochrome 1 (Cry1), and period 2 (Per2) in the olfactory bulb (Figure 1B; Table 1). Taar5^-/- mice exhibited increased mesor for Bmal1, Clock, and Rev-Erbα compared to Taar5^+/+ controls in the olfactory bulb (Figure 1B; Table 1). Whereas, Taar5^-/- mice exhibited advanced acrophase in Cry1 and Per2 compared to Taar5^+/+ controls in the olfactory bulb (Figure 1B; Table 1). There is also some reorganization of circadian gene expression in the liver (Figure 1C) and gonadal white adipose tissue (Figure 1D; Table 1), albeit modest. In the liver, Taar5^-/- mice had normal circadian gene expression, when compared to Taar5^+/+ controls (Figure 1C; Table 1). In gonadal WAT, the amplitude of Bmal1 was increased and acrophase of Bmal1 was delayed in Taar5^-/- mice, whereas only the acrophase of Per2 was advanced in Taar5^-/- mice compared to Taar5^+/+ controls (Figure 1D; Table 1). Given the key role that the TMAO pathway plays in suppressing the beiging of WAT¹³, we also examined the expression of PR/SET domain 16 (Prdm16) and uncoupling protein 1 (Ucp1). Taar5^-/- mice had increased mesor and advanced acrophase of Prdm16, yet no difference was detected in Ucp1, compared to Taar5^+/+ controls (Figure 1D; Table 1). Taar5^-/- mice have unaltered oscillations in body weight (Figure 1 - figure supplement 1; Table 1).

Figure 1.

Table 1.

We next examined circadian oscillations in circulating metabolite, hormone, and cytokine levels in Taar5^+/+ and Taar5^-/- mice (Figure 1 – figure supplement 2; Table 1). When we measured substrates for gut microbial TMA production we found that Taar5^-/- mice had normal oscillations in choline and L-carnitine (Figure 1 - figure supplement 2A; Table 1). However, there was a modest yet significant advance in the acrophase of plasma γ-butyrobetaine in Taar5^-/- mice (Figure 1 - figure supplement 2A; Table 1). Taar5^-/- mice also had slightly increased levels of TMA at ZT22, when compared to Taar5^+/+ controls (Figure 1 - figure supplement 2A; Table 1). However, TMA and TMAO levels were not significantly altered in Taar5^-/- mice (Figure 1 - figure supplement 2A; Table 1). Interestingly, Taar5^-/- mice had altered rhythmic levels in some but not all host metabolic hormones. Although, the circadian oscillations of plasma insulin and C-peptide were not significantly altered, Taar5^-/- mice had a delay in acrophase of plasma glucagon, yet advanced acrophase of glucagon-like peptide 1 (GLP-1), compared to Taar5^+/+ controls (Figure 1 - figure supplement 2B; Table 1). Taar5^-/- mice also exhibited increased mesor for plasma leptin levels compare to wild-type controls (Figure 1 - figure supplement 2B; Table 1). Taar5^-/- mice also had modest differences in circulating cytokines including a decrease in the mesor of monocyte chemoattractant 1 (MCP-1) and tumor necrosis factor α (TNFα) (Figure 1 - figure supplement 2B; Table 1) Collectively, these data demonstrate that Taar5^-/- mice have altered circadian-related gene expression that is most apparent in the olfactory bulb (Figure 1; Table 1), and abnormal circadian oscillations in some but not all circulating hormones and cytokines (Figure 1 - figure supplement 2; Table 1).

We next set out to comprehensively analyze the circadian rhythms in behavioral phenotypes in mice lacking the TMA receptor TAAR5. In this line of investigation we took a very broad approach to examine impacts of Taar5 deficiency on metabolic, cognitive, motor, anxiolytic, social, olfactory and innate behaviors. The main rationale behind this in-depth investigation was to allow for comparison to our recent work showing that pharmacologic blockade of the production of the TAAR5 ligand TMA (using choline TMA lyase inhibitors) produced clear metabolic, innate, and olfactory-related social behavioral phenotypes, but did not dramatically impact aspects of cognition, motor, or anxiolytic behaviors.^33,46. Also, it is important to note other groups have recently shown that TAAR5 activation with non-TMA ligands or genetic deletion of Taar5 results in clear olfactory^48–51, anxiolytic⁵¹, cognitive⁵², and sensorimotor^53,54 behavioral abnormalities. Here, it was our main goal to identify whether any behavioral phenotypes that were consistently seen in mice lacking bacterial TMA production^33,46 or host TMA sensing by TAAR5 (studied here) are time-of-day dependent indicating circadian inputs. Given the consistent alterations in innate and olfactory-related phenotypes seen in both mice lacking bacterial TMA production^33,46 and mice lacking host TMA sensing by TAAR5^47–54, we next followed up to perform select innate and olfactory-related behavioral tests in mice at defined circadian time points (Figure 2).

Figure 2.

To study the circadian presentation of phenotype, we carefully controlled the time window of testing for either the olfactory cookie test or marble burying test, both of which have been shown to be altered in mice lacking bacterial TMA synthesis^46,55. When the olfactory cookie test was performed during the mid light cycle (ZT5-ZT7), the latency to find the buried cookie was significantly increased in male Taar5^-/- mice compared to Taar5^+/+ controls (Figure 2A). However, when the same mice performed the olfactory cookie test at the dark-to-light phase transition (ZT23-ZT1), or at the light-to-dark phase transition (ZT13-ZT15), there were no significant differences between Taar5^+/+ and Taar5^-/- mice (Figure 2A). When subjected to the marble burying test, only female Taar5^-/- mice buried significantly more marbles than female wild-type controls only at ZT5-ZT7, but this was not apparent at other ZT timepoints (Figure 2B). Collectively, these data demonstrate that Taar5^-/- mice exhibit highly gender-specific alterations in innate and olfactory behaviors, and these behavior phenotypes are only apparent at certain periods within the light cycle (Figures 2, and S2).

Although Taar5^-/- mice showed time-dependent alterations in the olfactory cookie test, olfactory discrimination towards other diverse single stimuli such as banana, corn oil, almond, water, or social cues were not significantly altered (Figure 2 - figure supplement 1). When we subjected Taar5^-/- mice to a battery of social behavioral tests, there were test-specific alterations that occur in a sexually dimorphic manner. All mouse groups (Taar5^+/+, Taar5^-/-, male and female) displayed no initial chamber bias in the initial trial of the three-chamber test (Figure 2 - figure supplement 2A). In the three-chamber preference test, only male Taar5^-/- mice showed no preference between an inanimate object and a social stimuli interaction (Figure 2 - figure supplement 2B). When subjected to the three-chamber social novelty test box, both male and female Taar5^-/- mice showed no preference between the novel and familiar stimuli (Figure 2 - figure supplement 2C). In the social interaction with a juvenile mouse test, Taar5^-/- females showed no significant difference in interaction time between the initial interaction trial and the recognition trial four days later (Figure 2 - figure supplement 2D). In addition to alterations in social interactions, Taar5^-/- mice also showed sexually dimorphic alterations in several other innate behavioral tests (Figure 2 - figure supplement 3). Female Taar5^-/- mice exhibited a significantly higher startle response at 90, 100, 110, and 120 decibels (Figure 2 - figure supplement 3A), and significantly weaker forelimb grip strength (Figure 2 - figure supplement 3B) compared to Taar5^+/+ controls. When both sexes are combined there is a significant increase in the latency to withdraw during the hotplate sensitivity test in Taar5^-/- mice compared to controls (Figure 2 - figure supplement 3C). Furthermore, both male and female Taar5^-/- mice have slightly increased latency to fall during the rotarod test (Figure 2 - figure supplement 3D). Also, female but not male Taar5^-/- mice exhibit reduced nest building compared Taar5^+/+ mice (Figure 2 - figure supplement 3E).

We next comprehensively examined cognitive, depression, and anxiety-like behaviors in Taar5^+/+ and Taar5^-/- mice (Figure 2 - figure supplements 4 and 5). Both male and female Taar5^-/- mice performed similarly to Taar5^+/+ controls in the open field, elevated plus maze, and Y-maze tests (Figure 2 - figure supplement 4B-D). However, female, but not male, Taar5^-/- mice showed significantly reduced freezing compared to Taar5^+/+ controls in the cued fear conditioning test (Figure 2 - figure supplement 4A). When subjected to the Morris water maze, there were only minor alterations found in Taar5^-/- mice. All mice showed similar latency to the platform. Male Taar5^-/- mice showed increased distance traveled compared to wild-type controls, yet females were more similar to Taar5^+/+ controls (Figure 2 – figure supplement 5). However, female Taar5^-/- mice showed increased velocity compared to Taar5^+/+ mice during the last 3 days of testing (Figure 2 - figure supplement 5). Collectively, the impact of Taar5 deficiency on cognitive, depression, and anxiety-like behaviors was very modest.

Given the TMAO pathway has been linked to the beiging of white adipose tissue, and energy expenditure¹³, we next examined circadian rhythms in energy metabolism during a cold challenge and gene expression in thermogenic brown adipose tissue (BAT) in Taar5^-/- mice (Figure 2 - figure supplement 6). Although male Taar5^-/- mice showed unaltered oxygen consumption at thermoneutrality (30°C), room temperature (22°C), and during cold (4°C) exposure, female Taar5^-/- mice had significantly elevated oxygen consumption that appeared most significant during the light cycle periods (Figure 2 - figure supplement 6A). To follow up we collected BAT from Taar5^+/+ and Taar5^-/- mice at ZT2 (early light cycle) and ZT14 (early dark cycle) to examine potential alterations in circadian gene expression. Both male and female Taar5^-/- mice showed marked upregulation of Arntl1/Bmal1, yet other than increased Per1 expression at ZT14 all other circadian genes were largely unaltered in Taar5^-/- mice (Figure 2 - figure supplement 6B). Taken together, all behavioral data presented here show that mice lacking Taar5 have select sexually dimorphic alterations in olfactory, innate, social, and metabolic phenotypes.

Host TAAR5 Regulates the Circadian Rhythmicity of the Gut Microbiome

Bi-directional microbe-host communication is required for homeostatic control of chronobiology in the metaorganism.^56–60 Our data strongly suggest that gut microbe-derived TMA can shape host circadian rhythms in metabolic homeostasis and behavior^33,46, but we also wanted to test whether host TAAR5 may reciprocally regulate circadian rhythms of the gut microbiome^58,59. The rationale for studying alterations in the gut microbiome stems from that fact that several previous studies show that manipulating the TMAO pathway at different levels (diet, microbe, and host) strongly alters gut microbiota in unexpected ways^7,11,31,33. For example, provision of dietary substrates (i.e. choline or L-carnitine) or exogenous TMAO itself can reorganize gut microbial communities in mice^7,11. Also, small molecule enzyme inhibitors blocking the bacterial production of TMA profoundly alter the cecal microbiome^31,33, and some of the anti-obesity and circadian rhythm altering effects can be transmitted by cecal microbial tranplantation³³. Therefore, we examined the circadian oscillation in the cecal microbiome in Taar5^+/+ and Taar5^-/- mice over a 24-hour period, and found there were clear alterations that were time of day dependent (Figure 3 and Figure 3 – figure supplements 1 and 2; Table 1). At the phylum level, wild-type mice showed a modest decrease in the light cycle (i.e. ZT2-ZT10), and progressive loss of Firmicutes over the dark cycle (i.e. from ZT14-ZT22). In contrast, Taar5^-/- mice showed a reciprocal increase during the light cycle, and more modest dip during the light cycle in Firmicutes (Figure 3 - figure supplement 1; Table 1). When we examined microbiome alterations at the genus level, there were much clearer alterations in specific bacteria (Figure 3; Figure 3 - figure supplement 2; Table 1). In particular, the acrophase of several Lachnospiraceae, Odoribacter, and Dubosiella genera are altered in Taar5^-/- mice when compared to wild-type controls (Figure 3; Figure 3 - figure supplement 2; Table 1). Although there were many microbiome alterations, Taar5^-/- mice showed delayed acrophase for Lachnospiraceae UCG-002, Lachnospiraceae NK4A136, Desulfovibrio, Bacteroides, Dubosiella, Colidextribacter, Alistipes, Turicibacter, Muribaculum, Helicobacter, and Parabacteroides, when compared to Taar5^+/+ mice (Figure 3 - figure supplement 2; Table 1). Other genera such as Lachnospiraceae UCG-006, Odoribacter, Butyricicoccus, Coriobacteriaceae UCG-002, Acetatifactor, and A2 showed advanced acrophase in Taar5^-/- mice (Figure 3 - figure supplement 2; Table 1). It is important to note that previous independent studies have also shown that either blocking bacterial TMA synthesis³³ or FMO3-driven TMA oxidation²⁹ can also strongly reorganize the gut microbiome in mice. Although more work is needed to fully understand the underlying mechanisms, it is clear that both gut microbe-driven TMA production and host sensing of TMA by TAAR5 (Figures 3; Figure 3 - figure supplement 2; Table 1) can strongly impact on circadian oscillations of the gut microbiome.

Figure 3.

Mice Genetically Lacking Either Gut Microbial TMA Production or Host-Driven TMA Oxidation Have Altered Circadian Rhythms

To confirm and extend the idea that gut microbe-derived TMA can shape host circadian rhythms we next performed experiments where we genetically deleted either gut microbial TMA synthesis or host-driven TMA oxidation. First, we used gnotobiotic mice engrafted with a defined microbial community with or without genetic deletion of the choline TMA lyase CutC (i.e. Clostridium sporogenes wild-type versus ΔcutC) to understand the ability of gut microbe-derived TMA to alter host circadian rhythms (Figure 4; Figure 4 – figure supplement 1; Figure 4 – figure supplement 2). Circadian oscillations in plasma TMA and TMAO that peak in the early dark cycle are only detectable in the group colonized with C. sporogenes (WT), but not ΔcutC C. sporogenes (Figure 4A). Other TMAO pathway-related metabolites including choline, L-carnitine, γ-butyrobetaine, and betaine also exhibited modest alterations in circadian rhythms (Figure 4A). Mice lacking choline TMA lyase activity showed modestly reduced mesor in plasma choline, and advanced acrophase in both plasma betaine and γ-butyrobetaine (Figure 4A; Table 1). In a similar manner to Taar5^-/- mice (Figure 1), mice colonized with ΔcutC C. sporogenes show alterations in the expression of some but not all circadian genes in the olfactory bulb (Figure 4B; Table 1). Mice colonized with ΔcutC C. sporogenes had significantly advanced acrophase for Arntl/Bmal1, yet delayed acrophase for Clock, in the olfactory bulb when compared to mice colonized with the control community (Figure 4B; Table 1). However, the oscillatory pattern of other circadian genes were not significantly altered. When we examined circulating hormones and cytokines, there were clear alterations in the circadian oscillation in mice genetically lacking choline TMA lyase activity (Figure 4C; Table 1). Mice colonized with the ΔcutC C. sporogenes community exhibited delayed acrophase for plasma insulin and interleukins 2 (IL-2) and 33 (IL-33) when compared to mice harboring the wild-type community that could produce TMA (Figure 4C; Table 1). Mice lacking choline TMA lyase activity also had an advanced acrophase for GLP-1, leptin, and IL-1β (Figure 2C; Table 1). It is interesting to note that we examined the cecal abundance of the five bacterial strains in our defined community, there were clear alterations that were time of day dependent (Figure 4 - figure supplement 1). Four of the bacterial strains (C. sporogenes, B. theta, B. caccae, and B. ovatus) showed clear circadian oscillations when a zero amplitude test was performed in mice colonized with wild-type community. However, the community lacking cutC lost all significant circadian oscillation of these same four bacterial strains (Figure 4 - figure supplement 1; Table 1), showing that bacterial production of TMA can broadly impact both microbe-microbe and microbe-host interactions in circadian rhythms.

Figure 4.

To follow up on our previous findings that FMO3 suppresses both beige and brown fat-induced cold induced thermogenesis¹³, we also wanted to examine circadian patterns in subscapular brown adipose tissue (BAT) of ΔcutC C. sporogenes-colonized mice (Figure 4 - figure supplement 2; Table 1). Although there were no statistically-significant alterations in circadian gene expression in the BAT isolated from ΔcutC C. sporogenes-colonized mice, the acrophase of phosphatidylethanolamine methyltransferase (Pemt) was delayed in mice colonized with the ΔcutC community compared to mice colonized with the control wild-type community (Figure 4 - figure supplement 2A; Table 1). We next wanted more comprehensively quantify a variety of other well-known metabolites that are known to originate from bacterial sources (Figure 4 - figure supplement 2B). The rationale for performing microbe-focused metabolomics is that we have found if we alter one gut microbe-derived metabolite in other gnotobiotic mouse studies, there can be unexpected alterations in other distinct classes of microbe-derived metabolites⁶¹. This is likely due to the fact that complex microbe-microbe and microbe-host interactions work together to define systemic levels of circulating metabolites, influencing both the production and turnover of distinct and unrelated metabolites. Although the targeted deletion of cutC in C. sporogenes prevents the production of TMA from choline, there were alterations of several other gut microbe-derived metabolites in mice colonized with the ΔcutC C. sporogenes community (Figure 4 - figure supplement 2). Bacterially-derived aromatic amino acid metabolties such as hippuric acid, indole-3-propionic acid, and indole acetic acid were significantly reduced in mice colonized with the ΔcutC community (Figure 4 - figure supplement 2). Also, the bacterially-derived phenylalanine metabolites phenylacetic acid and phenylacetylglycine gained rhymicity with marked increases in their peak amplitude in mice colonized with the ΔcutC community when compared to control mice (Figure 4 - figure supplement 2).

Finally, we wanted to examine whether host co-metabolism of TMA can also alter circadian rhythms (Figure 4 – figure supplement 2). Although gut microbes are the sole source of TMA, circulating levels are also shaped by abundant conversion of TMA to TMAO by the host liver enzyme FMO3^13,29. To test whether the hepatic conversion of TMA to TMAO by FMO3 may alter circadian rhythms, we necropsied female Fmo3^+/+ and Fmo3^-/- mice at ZT2 (early light cycle) or ZT14 (early dark cycle). Fmo3^-/- mice have reduced levels of TMAO at both ZT2 and ZT14, as well as reduced expression of key circadian genes Arntl and Per1 at ZT2 (Figure 4 – figure supplement 2). Taken together, our data suggest that bacterial production, FMO3-driven metabolism of TMA, as well as sensing of TMA by the host receptor TAAR5, converge to shape circadian rhythms in gene expression, metabolic hormones, gut microbiome composition, and innate behaviors.

Discussion

The metaorganismal TMAO pathway is strongly associated with many human diseases^5–29, which has prompted the rapid development of drugs intended to lower circulating levels of TMAO.^30–37 As the rapid drug discovery advances towards human studies, it will be extremely important to understand the diverse mechanisms by which the primary metabolite TMA and/or the secondary metabolite TMAO promotes disease pathogenesis in the human metaorganism. There is compelling evidence that the end product of the pathway TMAO can promote inflammation, ER stress and platelet activation via activation of NFκB, the NLRP3 inflammasome, PERK, and stimulus-dependent calcium release, respectively.^28,38–41 However, these TMAO-driven mechanisms only partially explain the links to so many diverse human diseases. Here, we provide new evidence that in addition to these TMAO-driven mechanisms, the primary gut microbe TMA can in parallel shape circadian rhythms through the host GPCR TAAR5. The major findings of the current studies are: (1) Mice lacking the TMA receptor TAAR5 have abnormal oscillations in core circadian genes, particularly in the olfactory bulb, (2) Compared to wild-type controls, Taar5^-/- mice have altered circulating levels of cytokines, metabolic hormones, and metabolites at certain times of the day, (3) Taar5^-/- mice have altered innate and repetitive behaviors that emerge only in a time of day-dependent manner, (4) The normal oscillatory behavior of the cecal microbiome is dysregulated in Taar5^-/- mice, (5) Genetic deletion of gut microbial choline TMA lyase activity, using defined cutC-null microbial communities in vivo, results in rewiring of circadian rhythms in gene expression, metabolic hormones, cytokines, and metabolites, (6) The bacterial strains in the defined microbial used in gnotobiotic mouse study oscillate differently depending on the presence of cutC, and (7) Mice lacking the host liver TMA-to-TMAO converting enzyme FMO3 likewise have altered circadian gene expression in the olfactory bulb. Collectively, our findings suggest that therapeutic strategies designed to limit gut microbial TMA production (i.e. TMA lyase inhibitors), host liver TMA oxidation (i.e. FMO3 inhibitors), or sensing of TMA by the host GPCR TAAR5 (i.e. TAAR5 inhibitors) will need to be carefully evaluated for pleiotropic effects on circadian rhythms. This body of work also demonstrates that diet-microbe-host interactions can powerfully shape chronobiology, and provides one of the first examples of a microbial metabolite being sensed by a host GPCR to rewire circadian rhythms.

Currently, the only known host receptor that senses gut microbe-produced TMA is the volatile amine receptor TAAR5, which allows for species-specific recognition of the “fishy” odor intrinsic to TMA but not TMAO^,47–49 In general, the trace amine-associated receptor (TAAR) subfamily of GPCRs primarily function as olfactory receptors in vertebrates.^49,50 Ligand-dependent activation of each individual TAAR receptor occurs in a unique sensory neuron population primarily localized to olfactory cilia, and when activated typically elicit TAAR-specific olfactory-related behavior responses.^49,50 TMA-dependent activation of TAAR5 was first shown to promote attraction and social interaction between females and male mice in a very species and strain-specific manner.⁴⁸ This seminal study by Liberles and colleagues showed that the TMA-TAAR5 olfactory circuit can powerful shape olfactory-driven social interaction, and FMO3-driven conversion of TMA to TMAO suppresses TAAR5 activation.⁴⁸ In parallel, we recently performed a comprehensive study to assess cognitive, motor, anxiolytic, social, olfactory and innate behaviors in mice treated with small molecule choline TMA lyase inhibitors to block bacterial TMA production.⁴⁶ Much like the olfactory and social phenotypes seen in Taar5^-/- mice⁴⁸, we found that pharmacologic blockade of bacterial TMA production significantly altered olfaction and olfactory-related social behaviors, but did not significantly alter cognitive, motor, anxiolytic behaviors⁴⁶. It is interesting to note that choline TMA lyase inhibitors altered olfactory perception of several odorant stimuli beyond TMA itself including a cookie, almond, vanilla, corn oil, and coyote urine.⁴⁶ This indicates that drugs blocking TMA-TAAR5 signaling could impact olfactory neurogenesis to impact sensing of diverse odorant cues that are well beyond the rotten fish smell intrinsic to TMA. Given that anosmia is a common occurrence in several TMAO-related diseases including CKD⁶², obesity, diabetes⁶³, neurodegenerative diseases^64,65, and COVID-19⁶⁶, it is tempting to speculate that TMA-lowering therapeutics may hold promise to potentially restore olfactory perception to diverse stimuli in these disease conditions. However, additional studies are needed to formally test this hypothesis. Given the clear links between the TMA-TAAR5 signaling axis and olfactory-related behaviors in mice^46,48, it will be important to carefully evaluate effects of TMAO-lowering drugs on olfaction in future human studies.

A commonality of many bacterially-derived volatile metabolites is that they have a pungent odor which is sensed by olfactory neurons, and these microbe-host olfactory circuits shape many diverse behavioral and metabolic responses in the human host. For example, strong bacterially-derived odors such as trimethylamine (i.e. smells like rotting fish), hydrogen sulfide (i.e. the rotten egg smell), polyamines including putrescine and cadaverine (i.e. the smell of rotting cadavers/roadkill) can serve as an effective “don’t eat me” signal for us to avoid consuming spoiled food. Within the human metaorganism, the gut microbiome and select gut microbe-derived metabolites play an important role in shaping not only food intake but also other aspects of host energy metabolism, insulin sensitivity, immunology, and the circadian clock.^56–60 It is important to note that there is bi-directional communication between gut microbes and the host in maintaining the circadian rhythms in metabolism and immunity within mammalian metaorganisms.^56–60 First, central and peripheral clock machinery in the host clearly shape circadian oscillations in the gut microbiome primarily through innate immune mechanisms.^56,57 At the same time, circadian oscillations intrinsic to the gut microbiome itself can have profound impacts on host physiology and disease.^58–60 Although there is extensive knowledge on how the host circadian clock can impact the oscillatory behavior of the gut microbiome^56,57, much less is known regarding mechanisms by which the gut microbiome can instruct the host circadian clock. Our data support the notion that gut microbe-derived TMA and TAAR5 activation can impact circadian rhythms in gene expression, metabolic hormones, gut microbiome composition, and innate behaviors. Using both pharmacologic³³ and genetic approaches, we have demonstrated that blocking either gut microbial TMA production or sensing by the host TMA receptor TAAR5 results in striking rearrangement of circadian rhythms in the gut microbiome that are linked to metabolic homeostasis and olfactory-related behaviors in mice^33,46. In further support of the connections between the TMAO pathway and host circadian disruption, a recent paper showed that FMO3 inhibitors blocking host TMA oxidation improved learning and memory deficits driven by chronic sleep loss⁶⁷. Other recent studies in humans also show that TMAO is elevated with sleep deprivation⁶⁸, and elevated TMAO levels are associated with an “evening chronotype” that is linked to increased risk of cardiometabolic disease.⁶⁹ Collectively, the metaorganismal TMAO pathway plays an important in integrating dietary cues and microbe-host interactions in circadian rhythms. Given TMAO lowering drugs are rapidly advancing towards human studies, it will be of great interest to determine whether these drugs can improve circadian rhythms and potentially synergize with other developmental chronotherapies to improve human health.

Limitations

First, it is critically important to point out that the TMA-TAAR5 signaling axis has evolved to elicit species-specific behavioral responses that differ quite substantially across rodent species and humans.^48,49 When the “stinky” fish odor of TMA is sensed by rat or human TAAR5, the result is typically aversion from consuming what is perceived as rotting food.^48,49 However, select strains of mice have concentration-dependent attraction to low levels of TMA or aversion to high levels of TMA.⁴⁸ It has been theorized that this concentration-dependent behavioral response to TMA in mice may have evolved as a kairomone signal that allows mice to sense approaching prey animals.^48,49 Clearly the human TMA-TAAR5 signaling circuit evolved for different reasons, where predator-prey signaling circuits are less prominent. Therefore, additional studies are needed to better understand what diverse roles TMA-driven TAAR5 activation plays in the circadian rhythms in human metabolism and behavior. The recent discovery of humans with loss-of-function TAAR5 genetic variants⁷⁰ may provide a unique population to confirm and extend the findings we report here in mice. Another limitation in our current understanding is that drugs blocking gut microbial TMA production^30–37 or host TMA oxidation^13,15,67 have only been tested in mice. Although these drugs show clear benefit in mice models, the human gut microbiome is quite different than mice and the relative affinity of TMA for TAAR5 is ∼ 200-fold less in humans when compared to rodent forms of the receptor.^47–49 Therefore, it is impossible to extrapolate finding in mice to humans. As TMAO lowering drugs move into human trials, there is strong potential that human-specific phenotypes may emerge. Another limitation is that we have not rigorously studied diet-induced obesity, effects on food intake, or cardiovascular disease/atherosclerosis in this study. Additional studies are needed to understand whether TMA-driven TAAR5 activation impacts cardiometabolic disease. A final limitation here is that here we have studied mice globally lacking TAAR5 in all tissues and cell types. Although TAAR5 has primarily been studied in olfactory neurons in the brain, TAAR5 is also expressed in immune cells⁷¹ as well as skeletal and cardiac muscle³³. Given the expression of TAAR5 outside of the central nervous system, future studies utilizing cell-restricted Taar5 knockout mice will provide clues into the cell autonomous roles for TMA-TAAR5 signaling. Despite these limitations, a growing body of evidence links TMA and TAAR5 activation to circadian regulation of metabolism and olfactory-related behaviors.

STAR★Methods

Key resources table

Experimental procedures

Animal Studies (Mice)

Taar5^-/- mice [Taar5^{tm1.1(KOMP)Vlcg}] used here were created from ES cell clone 10675A-A8 and originated on the C57BL/6N background by Regeneron Pharmaceuticals, Inc. The successfully targeted ES cells were then used to create live mice by the KOMP Repository and the Mouse Biology Program at the University of California Davis. It is important to note that this KO/reporter mouse line has the removal of the neomycin selection cassette and critical exon(s) leaving behind the inserted lacZ reporter sequence knocked into the Taar5 locus. Given the well appreciated differences in metabolic homeostasis between C57BL/6N and C57BL/6J substrains^72,73, we backcrossed this line > 10 generations into the C57BL/6J congenic strain and backcrossing was confirmed by mouse genome SNP scanning at the Jackson Laboratory (Bar Harbor, ME). Heterozygous mice on a pure C57BL/6J background were bred to generate littermate Taar5^+/+ and Taar5^-/- mice. Global Fmo3^-/- mice on a pure C57BL/6J background have been previously described.⁷⁴ All mice were fed standardized control chow diet with defined sufficient level of dietary choline (Envigo diet TD:130104) that was twice irradiated for sterilization. For studies of circadian rhythms, 9-week-old C57Bl6/J male mice were adapted for 2 weeks with a strict 12:12 light:dark cycle. After 7 days, plasma and tissue were collected every 4 hours over a 24-hour period. All dark cycle necropsies were performed under red light conditions to avoid light-driven circadian cues. For the CutC mutant studies shown in main Figure 4, 7-8 week-old germ-free B6/N mice were purchased from Taconic and placed on a sterile chow. Two weeks later, the mice were given an oral gavage of our 6 member community that included B. caccae, B. ovatus, B. theta, C. aerofaciens, E. rectale, and either wild-type or mutant C. sporogenes suspended in sterile PBS + 20% glycerol as previously described.⁵⁵ Gnotobiotic colonized mice were maintained using the Allentown Sentry SPP Cage System (Allentown, NJ) on a 14 hr:10 hr light:dark cycle. Mice were transferred to a room equipped with 12:12 light-dark cycling while still housed on Allentown Transport Carts. After one week of acclimation, plasma and tissue collection were performed every 4 hours over a 24-hour period. All dark cycle necropsies were performed under red light conditions. For all studies plasma was collected by cardiac puncture, and liver, GWAT, BAT and olfactory bulb were collected, flash-frozen, and stored at –80°C until the time of analysis. All mice were maintained in an Association for the Assessment and Accreditation of Laboratory Animal Care, International-approved animal facility, and all experimental protocols were approved by the Institutional Animal Care and use Committee of the Cleveland Clinic (Approved IACUC protocol numbers 00001941, 00002609, and 0002866).

RNA and Realtime PCR Methods

RNA isolation and quantitative polymerase chain reaction (qPCR) was conducted using methods previously described with minor modifications.^13,33,46 To extract RNA, the Monarch Total RNA MiniPrep Kit (New England BioLabs, Inc.) was employed, and 325 ng of cDNA was utilized for Realtime PCR on a 384-well plate, using Applied Biosystem’s Power SYBR Green PC mix in Roche LightCycler 480 I1. The levels of induced mRNAs were standardized by the ΔΔCT method and were normalized to cyclophilin A. The specificity of the primers was verified by analyzing the melting curves of the PCR products. Primer information is available in the Star Methods table.

Quantification of Plasma Metabolites Using Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS)

Stable isotope dilution high-performance LC-MS/MS was used for quantification of levels of TMAO, TMA, choline, carnitine, and γ-butyrobetaine and all other reported circulating plasma metabolites as previously described performed on a Shimadzu 8050 triple quadrupole mass spectrometer.^13,33,46 A subset of samples were also analyzed on a Thermo Vanquish liquid chromatograph coupled with a Thermo TSQ Quantiva mass spectrometer. 0.2% of formic acid in water (LC-MS grade) was used as mobile phase A and 0.2% formic acid in acetonitrile was used as mobile phase B. The separation was conducted using the following gradient: 0 min, 5% B; 0-2 min, 5% B; 2-8 min, 5%-100%B; 8-16 mins, 100%B; 16-16.5 mins, 5%B; 16.5-25 mins, 5%B. The flow rate was set at 0.2 ml/min. Samples were injected at 2 μl onto a Phenomenex Gemini C18 analytical column (2.0×150 mm, 3 μm). Column temperature was set at 25°C. The acquired data were processed by Thermo Xcalibur 4.3 software to calculate the concentrations. Both methods used multiple reaction monitoring of precursor and characteristic products as we have previously described.^13,33,46

Plasma Hormone and Cytokine Quantification

Plasma hormones and cytokine levels were quantified using U-PLEX (product # K15297K) and V-PLEX (K15297K) assays per the manufacturer’s instructions (Meso Scale Diagnostics, Rockville, Maryland, USA).

Olfactory Cookie Test

To broadly understand olfactory perception, we performed the olfactory cookie test using methods previously described.⁴⁶ Briefly, one day prior to testing, food was removed from the home cage. The following day, a clean cage was filled with ∼1.5 inches of clean bedding and a small portion of peanut butter cookie (Nutter Butter, Nabisco®) was placed randomly ∼1 cm under the bedding. A test mouse was placed into the cage and given 10 minutes to find the hidden cookie. Latency to find the cookie was recorded.

Marble Burying Test

To test for innate/repetitive behavioral alterations, mice underwent a marble burying test. A clean mouse cage was filled with 2.5-3 inches of bedding. 20 black marbles were placed evenly in 5 rows of 4 across the bedding. A test mouse was placed into the cage and was allowed to bury marbles for 30 minutes. After 30 minutes the number of marbles buried was counted. A marble was defined as buried when < 25% of the marble was visible.

Olfactory Discrimination Tests

These tests were performed as described previously⁴⁶. Briefly, a set of non-social odors (almond, banana, corn oil, and water) and two social odors were prepared. Three 6” cotton-tipped applicators containing a single scent were prepared for each of the five odors. Each odor was presented 3 times in a row for 2 min. The cotton applicator was tapped to the wire cage lid on an empty clean home cage so that the cotton tip was hanging just below the wire lid. All trials were video recorded and the amount of time each mouse spent exploring the cotton tip was recorded.

Three-Box Chamber Tests

Preference

3-chamber social interaction test was performed using a three-chambered box as described previously⁴⁶. This test consisted of three 10-min trials. During the first trial, the mouse was allowed to explore the three-chamber box in which each end-chamber contained an empty cage (upside down pencil holder). This test was used to determine if mice had a clear preference for a chamber.

Social

In the second trial, the three-chamber box contained a novel stimulus mouse under a cage in one of the end-chambers and an inanimate object (ping pong ball) under a cage in the opposite end-chamber. The test mouse was free to choose between a caged inanimate object and a caged, social target.

Novel Object

For the third trial, the test mouse was free to choose between a caged novel social target (novel mouse) versus the same caged mouse in trial 2 (familiar social target). Locations of inanimate targets and social targets were counterbalanced, and mice were placed back into the home cage for very brief intervals between trials. Total duration in chamber and interaction zone was recorded using video-tracking and automated Noldus EthoVision XT 14 (Leesburg VA).

Social Interaction with Juvenile

Social interaction with a juvenile was performed in a novel empty home cage under dim white light as described previously⁴⁶. Briefly, following a 15-min habituation, a test mouse was placed in a novel empty cage with a juvenile stimulus mouse (BALB/cJ juvenile mouse; 3 weeks of age; Stock #000651 Jackson Labs) and allowed to directly interact for 2 min. Interaction was scored by observing the duration and number of times the test mouse-initiated contact with or sniffed the juvenile mouse. Contact was considered as any part of the boy touching the other mouse. Three days later the same test mouse and juvenile mouse were paired again in a novel clean cage for 2 min and scored in the same manner.

Startle Test

Mice underwent a startle threshold test using methods previously described.⁴⁶ Mice were placed inside holding cylinders that sit atop of a piezoelectric accelerometer that detects and transduces animal movement (SR-LAB Startle Response System; SD Instruments; San Diego, CA). Acoustic stimuli were delivered by high-frequency speakers mounted 33 cm above the cylinders. Mouse movements were digitized and stored using computer software supplied by San Diego Instruments. For the startle reactivity test, mice were subjected to eight presentations of six trial types given in pseudorandom order: no stimulus, 80, 90-, 100-, 110-, or 120-dB pulses. The mean startle amplitudes for each condition were calculated. Chambers were calibrated before each set of mice, and sound levels were monitored using a sound meter (Tandy).

Forepaw Grip Strength Test

Grip strength was measured using a digital grip meter (Chatillon Guage with Mesh bar; Columbus Instruments, Columbus, OH) in which a mouse was suspended by its tail and allowed to grab onto a stainless-steel grid bar that is attached to a force transducer. The mouse was gently pulled away from the meter in a horizontal plane. The force applied to the bar immediately before the mouse released its grip was recorded in grams force by the digital sensor. Mice completed 10 trials with an intertrial interval of a least 30 seconds. The mouse’s reported grip strength was the average of all 10 trials.

Hotplate Sensitivity Test

For the hotplate sensitivity test, each mouse received a single trial that lasted a maximum of 30 seconds, and the behavioral response was scored live. To assess sensitivity to hot stimuli, a hot plate (Ugo Basile, Italy) was warmed to 52°C and a mouse was placed on the plate and a tall plexiglass cylinder was placed around the mouse to prevent the animal from leaving. A foot pedal was pressed to start the timer on the hotplate. As soon as the mouse showed a response to the stimuli such as jumping or licking their paws, the foot pedal was pressed again to stop the timer and the mouse was removed from the plate. Latency to respond was recorded.

Rotarod

This was a 2-day task in which mice were placed onto a nonrotating rod (3 cm in diameter) that had ridges for a mouse to grip and was divided into 4 sections that allowed 4 mice to be tested at the same time (Rotamex-5; Columbus Instruments; Columbus, OH). Each mouse was subjected to 4 trials per day with an inter-trial interval of at least 30 minutes to prevent its performance from being impaired by fatigue. For each trial, a mouse was placed onto the rod and allowed to balance itself. When mice were balanced on the rod, the rod started rotating and accelerated from 4 to 40 rpm over a 5-minute period. Latency to stay onto the rod was assessed with the use of an infrared beam just above the rod and was broken once the mouse fell off the rod. Each mouse was subjected to a total of 8 trials over a 2-day period (4 trial/day). Latency to fall (seconds) was recorded.

Nesting

Nesting behavior was performed in a well-lit room by placing a mouse in a novel home cage with a cotton nestlet (5.5 x 5.5 x 0.5 cm) but no bedding. Height and width of the nests were measured at 30, 60, and 90 min.

Fear Conditioning

Fear conditioning (FC) consisted of a training period and a testing period. For the training period, mice were placed in the fear conditioning boxes (Med Associates; Fairfax VT), with a gridded floor, an inside white light and NIR light, and a fixed NIR camera for video recording. Mice were placed inside the chamber for 2 minutes, and then a 30 second, 90 dB acoustic conditioned stimulus (CS; white noise) co-terminated with a 2 second 0.6 mA foot shock (US). Mice received a total of 3 US-CS pairing, with each pairing separated by 1 minute. The mouse remained in the chamber for 30 seconds after the pairings before returning to its home cage. The testing period occurred 24-hrs after the training period ended. The mouse’s freezing behavior (motionless except for respirations) was monitored during testing to access memory. Prior to cue FC testing, the inside of the FC chamber was modified using white plastic sheets, interior lighting, and vanilla extract on a paper towel placed in the bedding chamber to make the chamber look and smell differently than the training period. Mice were placed into this modified chamber and allowed to habituate for 3 minutes. Following 3 minutes the same white noise tone used in training was presented to the mouse for 3 minutes. During the tone presentation mouse freezing behavior was recorded. Total percent freezing for cue FC was recorded.

Elevated Plus Maze

The elevated plus maze task was conducted essentially as described previously.⁴⁶ Mice were placed in the center of a black, plexiglass elevated plus maze (each arm 33 cm long and 5 cm wide with 25 cm high walls on closed arms) in a dimly lit room for 5 minutes. Automated video tracking software from Noldus Ethovision XT13 (Leesburg VA) was used to track time spent in the open and closed arms, number of open and closed arm entries, and number of explorations of the open arm (defined as placing head and two limbs into open arm without full entry).

Y-Maze

Spontaneous alteration in Y-Maze was assessed as previously described.⁴⁶ Briefly, a mouse was placed into one arm of the Y-maze facing the center (14”). Video-tracking was used to record the spontaneous behavior of each mouse for 10 min. Zone (i.e. arm) alteration was later analyzed using Noldus EthoVision XT14, to determine when a mouse entered 3 consecutive different arms of the maze. Spontaneous alteration % was calculated as follows: Alteration % = #spontaneous alterations/ total number of arm entries −2 x 100.

Open Field

The open field was conducted as described previously⁴⁶. Briefly, mice were placed along the edge of an open arena (44 x 44 x 44 cm) and allowed to freely explore for 10 min. Time spent in the center of the arena 15 x 15 cm as well as locomotor activity were measured. Mice were monitored using Noldus EthoVision XT14 (Leesburg VA).

Morris Water Maze

A 48” diameter, white, plastic, circular, heated pool was filled to the depth of 23.5” with 22°C + 1°C water made opaque with gothic white, nontoxic, liquid tempera paint in a room with prominent extra-maze cues. Mice were placed in one of the four starting locations facing the pool wall and allowed to swim until finding a 15 cm diameter clear platform for 20 s before being removed to the home cage. If mice did not find the platform within 60 s, they were guided to the platform by the experimenter and remained on the platform for 20 s before being removed to the home cage. Latency to reach the platform, distance traveled to reach the platform, swim speed, and time spent in each of four quadrants were obtained using automated video tracking software from Noldus (Ethovision XT13, Leesburg VA). Mice were trained with 4 trials/day with an intertrial interval of 1–1.5 min for 10 consecutive days between 10 am and 3pm. A probe trial (free swim with the submerged platform removed) was performed as the first trial of the day on days 11. Percent time spent in the target quadrant was calculated. Latency to platform, distance to platform, swim speed, and time spent in the target quadrant was analyzed.

Indirect Calorimetry and Metabolic Cage Measurements During Cold Challenge

Cold-induced metabolic responses were quantified using the Oxymax CLAMS system (Columbus Instruments) as previously described.¹³ Briefly, weight matched Taar5^-/- and Taar5^+/+ mice were acclimated to metabolic cages for 48 hours. Thereafter, physical activity, oxygen consumption (VO₂), carbon dioxide production (VCO₂), and respiratory exchange ratio (RER) were continually monitored for 24 hours at thermoneutrality (30°C), 24 hours at room temperature 22°C, and 24 hours in the cold (4°C). Data represent the last 6 a.m. to 6 a.m. period after adequate acclimation.

16S rRNA gene amplicon sequencing and bioinformatics

16S rRNA gene amplicon sequencing and bioinformatics analysis were performed using our published methods using mouse cecum samples.³³ Briefly, raw 16S amplicon sequence and metadata, were demultiplexed using split_libraries_fastq.py script implemented in QIIME2.⁷⁵ Demultiplexed fastq file was split into sample-specific fastq files using split_sequence_file_on_sample_ids.py script from QIIME2. Individual fastq files without non-biological nucleotides were processed using Divisive Amplicon Denoising Algorithm (DADA) pipeline.⁷⁶ The output of the dada2 pipeline [feature table of amplicon sequence variants (an ASV table)] was processed for alpha and beta diversity analysis using phyloseq⁷⁷, and microbiomeSeq (http://www.github.com/umerijaz/microbiomeSeq) packages in R. We analyzed variance (ANOVA) among sample categories while measuring the of α-diversity measures using plot_anova_diversity function in microbiomeSeq package. Permutational multivariate analysis of variance (PERMANOVA) with 999 permutations was performed on all principal coordinates obtained during CCA with the ordination function of the microbiomeSeq package. Pairwise correlation was performed between the microbiome (genera) and metabolomics (metabolites) data was performed using the microbiomeSeq package.

16S Statistical Analysis

Differential abundance analysis was performed using the random-forest algorithm, implemented in the DAtest package (https://github.com/Russel88/DAtest/wiki/usage#typical-workflow). Briefly, differentially abundant methods were compared with False Discovery Rate (FDR), Area Under the (Receiver Operator) Curve (AUC), Empirical power (Power), and False Positive Rate (FPR). Based on the DAtest’s benchmarking, we selected lefseq and anova as the methods of choice to perform differential abundance analysis. We assessed the statistical significance (P < 0.05) throughout, and whenever necessary, we adjusted P-values for multiple comparisons according to the Benjamini and Hochberg method to control False Discovery Rate.⁷⁸ Linear regression (parametric test), and Wilcoxon (Non-parametric) test were performed on genera and ASVs abundances against metadata variables using their base functions in R (version 4.1.2; R Core Team, 2021).

Metagenomic Sequencing to Detect Oscillatory Patterns Occuring in Gnotobiotic Mice Colonized With Defined Synthetic Communities With or Without CutC

Microbial community DNA extraction and sequencing was performed using methods published by our group earlier⁸³. Briefly, DNA extraction from cecum was performed using Qiagen’s DNeasy PowerSoil Pro kit. To assess sample integrity, an e-gel (electronic agarose gel) was run. The DNA concentration of each sample was then quantified using a Qubit Fluorometer, which measured the values in ng/μL. Using this concentration data, we prepared each sample for Illumina’s Nextera XT library preparation kit, targeting a fragment size of 300–500 bp. After library construction, the fragment size distribution of each library was checked using Agilent’s TapeStation 4200. The libraries were then re-quantified with the Qubit Fluorometer. Next, the libraries were pooled in a manner that ensured an even distribution of sequencing reads across all samples. The pooled library was quantified again using both a Qubit Fluorometer and qPCR to determine its molarity in nM. Finally, the pooled library was loaded onto the sequencer for sequencing.

Data Analyses for Circadian Rhythmicity (Cosinor Analyses)

A single cosinor analysis was performed as previously described.^79,80 Briefly, a cosinor analysis was performed on each sample using the equation for cosinor fit as follows:

where M is the MESOR (midline statistic of rhythm, a rhythm adjusted mean), A is the amplitude (a measure of half the extent of the variation within the cycle), Φ is the acrophase (a measure of the time of overall highest value), and τ is the period. The fit of the model was determined by the residuals of the fitted wave. After a single cosinor fit for all samples, linearized parameters were then averaged across all samples allowing for calculation of delineralized parameters for the population mean. A 24 hour period was used for all analysis. Comparison of population MESOR, amplitude, and acrophase was performed as previously described.³³ Comparisons are based on F-ratios with degrees of freedom representing the number of populations and total number of subjects. All analyses were done in R v.4.0.2 using the cosinor and cosinor2 packages.^80–82

Standard Statistical Analyses

Data are expressed as the mean ± standard error of the mean (SEM). All data were analyzed using either one-way or two-way analysis of variance followed by Student’s t tests for post hoc analysis. Differences were considered significant at p <0.05. All analyses were performed using GraphPad Prism software (version 10.2.2; GraphPad Software, Inc).

Figure supplements

Figure 1 – figure supplement 1.

Figure 1 – figure supplement 2.

Figure 2 – figure supplement 1.

Figure 2 – figure supplement 2.

Figure 2 – figure supplement 3.

Figure 2 – figure supplement 4.

Figure 2 – figure supplement 5.

Figure 2 – figure supplement 6.

Figure 3 – figure supplement 1.

Figure 3 – figure supplement 2.

Figure 4, figure supplement 1.

Figure 4 – figure supplement 2.

Acknowledgements

This work was supported in part by National Institutes of Health grants R01 DK130227 (J.M.B.), P01 HL147823 (J.M.B.), P50 AA024333 (J.M.B.), RF1 NS133812 (J.M.B.), and an American Heart Association Postdoctoral Fellowship 24POST1178494 (S.D.). We are grateful to Dr. Federico Rey (University of Wisconsin – Madison, USA) for providing bacterial strains used in the mouse gnotobiotic studies performed here. We are also grateful to Dr. Stan Hazen for providing access to mass spectrometer instruments dedicated to targeted metabolomic methods necessary to quantify diverse gut microbe-derived metabolites.

Additional information

Data and code availability

The 16S microbiome data from wild-type and Taar5 knockout mice are publicly available at the following link: DOI 10.5281/zenodo.15802940.

Code used for cosinor analyses:

1. Sachs, M.C. (2014) cosinor: Tools for estimating and predicting the cosinor model, version 1.1. R package: https://CRAN.R-project.org/package=cosinor.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact (Dr. J. Mark Brown) upon request

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, J. Mark Brown (brownm5@ccf.org)

Materials availability

All the data and materials that support the findings of this study are available within the article. Additional details can be aquired via a direct request to the corresponding author.

Author Contributions

Conceptualization, K.M. and J.M.B.; methodology, K.M., W.J.M., D.O., A.L.B., T.J., A.C.B., A.J.H., S.D., M.M., N.M., V.V., L.J.O., X.Y., D.M.Y., N.Z., R.H., R.B., P.L., D.L., A.M.H., N.S., M.D., J.A.B., and G.R.S., Z.W; investigation, K.M., W.J.M., D.O., A.L.B., T.J., A.C.B., A.J.H., S.D., M.M., N.M., V.V., L.J.O., X.Y., D.M.Y., N.Z., R.H., R.B., P.L., D.L., N.S., M.D., J.A.B., and Z.W; validation, K.M., W.J.M., A.C.B., A.J.H., S.D., M.M., L.J.O., R.B., P.L., D.L., N.S., M.D., J.A.B., and Z.W; formal analysis, K.M., W.J.M., N.S., T.G., G.R.S., Z.W., J.M.B; writing – original draft, K.M. and J.M.B.; writing – reviewing and editing, K.M., W.J.M., D.O., A.L.B., T.J., A.C.B., A.J.H., S.D., M.M., N.M., V.V., L.J.O., X.Y., D.M.Y., N.Z., R.H., R.B., P.L., D.L., A.M.H., N.S., M.D., J.A.B., G.R.S., and Z.W.; funding acquisition, J.M.B.; supervision, J.M.B.

Abbreviations

• Arntl: basic helix-loop-helix ARNT like 1 or Bmal1

• BAT: brown adipose tissue

• CKD: chronic kidney disease

• Cry1: cryptochrome 1

• Cry2: cryptochrome 2

• CVD: cardiovascular disease

• EIF2AK3: eukaryotic translation initiation factor 2-alpha kinase 3

• ER: endoplasmic reticulum

• FMO3: flavin containing monooxygenase 3

• GPCR: G protein-coupled receptor

• IFNγ: interferon gamma

• IL: interleukin

• MCP-1: monocyte chemoattractant protein-1

• NFκB: nuclear factor κB

• NLRP3: nucleotide-binding domain, leucine-rich-containing family, pyrin domain-containing-3

• Nr1d1: nuclear receptor subfamily 1 group D member 1 or REV-ERBalpha

• Pemt: phosphatidylethanolamine N-methyltransferase

• PERK: eukaryotic translation initiation factor 2-alpha kinase 3

• Per1: period 1

• Per2: period 2

• Taar5: trace amine associated receptor 5

• TMA: trimethylamine

• TMAO: trimethylamine oxide

• WAT: white adipose tissue

• ZT: zeitgeber time

Funding

HHS | NIH | National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (R01 DK130227)

• Jonathan Mark Brown

HHS | NIH | National Heart, Lung, and Blood Institute (NHLBI) (P01 HL147823)

• Jonathan Mark Brown

HHS | NIH | National Institute on Alcohol Abuse and Alcoholism (NIAAA) (P50 AA024333)

• Jonathan Mark Brown

HHS | NIH | National Institute of Neurological Disorders and Stroke (NINDS) (RF1 NS133812)

• Jonathan Mark Brown

American Heart Association (AHA)

https://doi.org/10.58275/AHA.24POST1178494.pc.gr.190863

• Sumita Dutta