Abstract

   The circadian clock regulates tissue-specific homeostasis, and its
   disruption is associated with metabolic disorders. Both host metabolic
   processes and the gut microbiota exhibit diurnal regulation, and both
   contribute to the maintenance of glucose homeostasis (Thaiss et al.,
   2014; Bishehsari et al., 2020; Frazier et al., 2023) [1–3]. However,
   how the gut microbiota and the circadian rhythm interplay to control
   host glucose homeostasis is not fully understood. Here, we identified
   gut microbiota-derived extracellular vesicles (MEV) as a potential
   peripheral Zeitgeber (time cue) for the hepatic circadian clock,
   controlling hepatic gluconeogenesis. Host feeding patterns influence
   the gut microbiota, driving the diurnal production of MEV. Gut MEV
   levels coincide with the activity of hepatic gluconeogenesis, with
   overnight fasting associated with increased production of MEV by gut
   bacteria. MEV directly activates hepatic gluconeogenesis and chronic
   increase in MEV exposure impairs glucose homeostasis in vivo. Our
   finding highlights a mechanism by which the gut microbiota has
   co-evolved with the host to support its glucose needs during periods of
   energy demands (such as during fasting or starvation). On the contrary,
   an abnormal increase in MEV production, leading to dysregulated
   gluconeogenesis, may underlie various glucose-associated disorders,
   such as type 2 or gestational diabetes. Together, our data reconcile
   the gut microbiota and circadian rhythm in the control of host glucose
   homeostasis and metabolic health.

   Keywords: Gut microbiota, Extracellular vesicles, Circadian rhythm,
   Gluconeogenesis, Glucose homeostasis, Metabolism

Highlights

     * •
       Gut microbiota extracellular vesicle production is regulated by
       host feeding status.
     * •
       Gut microbiota extracellular vesicle activates hepatic
       gluconeogenesis.
     * •
       Gut microbiota extracellular vesicle influence the hepatic
       circadian clock.

1. Introduction

   Glucose is a major energy source critical for the optimal function and
   survival of most cellular organisms. However, the availability of
   glucose fluctuates constantly, influenced by factors like nutrient
   availability and the organism's feeding state. Gluconeogenesis is a
   critical metabolic process that evolved to maintain euglycemia during
   periods of low glucose availability. In mammals, it occurs
   predominantly in the liver and is activated in response to low blood
   glucose levels, such as during fasting. This process is hormonally
   regulated and is primarily stimulated by cortisol and glucagon produced
   by the adrenal glands and the pancreas, respectively. The secretion of
   these hormones is also regulated by circadian rhythm [[39]1],
   anticipatory to the rhythmic transition between feeding and fasting
   states. Indeed, gluconeogenic genes like phosphoenolpyruvate
   carboxykinase (PEPCK) and glucose-6-phosphatase (G6P) exhibit diurnal
   regulation, aligning with the activity of the master transcription
   factors CLOCK and BMAL1 [[40]2], which govern the mammalian circadian
   clock. Recent evidence has highlighted a role for the gut microbiota in
   regulating host metabolism and metabolic health [[41]3]. Notably, the
   absence of a gut microbiota led to lower blood glucose levels under
   both fed and fasted states [[42]4], while microbial-derived metabolites
   like butyrate can directly stimulate hepatic gluconeogenesis [[43]5].
   Dysregulation of blood glucose levels is associated with metabolic
   disorders like type 2 diabetes [[44]6] and gestational diabetes
   [[45]7]. Modern lifestyle factors are linked to detrimental shifts in
   gut microbiota composition and their function [[46]8,[47]9]. This
   underscores the importance of understanding how gut microbes can
   influence host glucose homeostasis, and ultimately, metabolic health.
   Furthermore, little is known about how the gut microbiota regulates
   peripheral clocks, given the well-established consequence of circadian
   rhythm disruption to the development of metabolic disorders [[48][10],
   [49][11], [50][12]]. Current identified microbial factors that
   influence gluconeogenesis act indirectly by optimising intracellular
   hepatic metabolism [[51]4], or directly by acting as substrates for
   gluconeogenesis [[52]5]. While many of these pathways are rhythmically
   regulated due to host dietary behaviours, they do not directly affect
   circadian rhythm per se, with no reported effects on the mammalian
   clock system. Emerging evidence indicates that gut microbiota-derived
   extracellular vesicles (MEVs) play a significant role in the host-gut
   microbiota axis [[53][13], [54][14], [55][15]]. MEV are nanosized
   membrane structures produced primarily from the budding of the plasma
   membrane and can carry a wide range of products including DNA, protein,
   and lipids. There are evidence that bacterial-derived extracellular
   vesicles (EV) can influence host metabolic outcomes. For example,
   orally administered Pseudomonas panacis-derived EVs induce insulin
   resistance and impairs glucose metabolism in skeletal muscles [[56]16]
   while Akkermansia muciniphila EVs improved the metabolic profile of
   high-fat induced diabetic mice [[57]17]. It is unclear what role EVs
   produced by resident gut microbiota play on host gluconeogenesis and
   hence glucose homeostasis.

2. Results

2.1. Gut microbiota-derived extracellular vesicles directly activate hepatic
gluconeogenesis

   To elucidate the potential role of MEV in host glucose homeostasis, we
   first determined whether MEV produced by gut bacteria can cross the
   intestinal barrier and translocate systemically. MEV was purified from
   cecal content and stool from control animals ([58]Fig. S1a–b),
   fluorescently-tagged and orally administered to germ-free (GF) mice and
   translocation was tracked by ex vivo tissue imaging 12 h later. GF mice
   were used to eliminate potential competition from endogenous MEV
   produced by the gut microbiota for translocation. We found that MEV
   predominantly concentrated in the liver ([59]Figure 1A), and to a
   lesser extent in the kidneys ([60]Fig. S1c). To understand the impact
   of MEV on the liver, we administered MEV to GF mice orally twice weekly
   for 3 weeks and performed transcriptomic analysis on the liver.
   Principal component analysis (PCA) revealed clear separation of liver
   transcriptomes between MEV-treated and PBS-treated groups ([61]Figure
   1B). Genes related to gluconeogenesis including Pck1, G6pc, Foxo1 and
   Ppargc1a were differentially upregulated in MEV-treated animals,
   compared to PBS-treated animals ([62]Figure 1C). This was confirmed by
   pathway enrichment analysis, with the FOXO mediated transcription
   pathway significantly enriched ([63]Figure 1D). FOXO1 is a key
   transcription factor that induces the expression of the key
   gluconeogenic enzymes PEPCK and G6P. Indeed, the gluconeogenesis
   pathway was also enriched in MEV-treated mice ([64]Figure 1D),
   corresponding to higher Pck1 and G6pc expression as validated by qPCR
   ([65]Fig. S1d). Differentially expressed genes are provided
   ([66]Supplemental Table 1). To validate this phenotype, we orally
   administered 1 × 10^10 MEV to specific-pathogen free (SPF) mice daily
   for 5 weeks, and then quantified liver gene expression of Pck1 and
   G6pc. In our hands, 50–100 mg of stool contains approximately 6.3 ×
   10^10 to 1.26 × 10^11 MEV, indicating that 1 × 10^10 MEV is not an
   excessive dose and likely falls within the lower end of physiological
   ranges. Compared to PBS-treated animals, MEV-treated mice had increased
   liver expression of both Pck1 and G6pc ([67]Figure 1E). We further
   verified that MEV induced hepatic gluconeogenesis, upregulating the
   expression of genes encoding for PEPCK and G6P, under both ex vivo
   conditions with mouse primary hepatocytes ([68]Figure 1F), and under in
   vitro conditions with the human hepatocyte cell line HEPG2 ([69]Figure
   1G). These results indicate that MEV act directly on hepatocytes to
   induce gluconeogenesis. We found that the effect of MEV on the
   expression of gluconeogenic genes in HEPG2 cells occurred at a
   threshold dose of approximately 3 × 10^9 particles, with G6PC
   expression increasing at higher doses ([70]Fig. S1e), with maximal
   effect at 10 × 10^9 particles (data now shown). Finally, to determine
   the physiological significance of MEV on the host, SPF mice were
   administered PBS or MEV for 5 weeks, as described above. Pyruvate
   tolerance test was then performed, a gold standard method to assess the
   conversion of exogenous pyruvate to glucose in vivo, as a readout of
   gluconeogenesis. Indeed, MEV-treated mice had higher blood glucose
   levels following fasting and administration of the gluconeogenic
   substrate pyruvate ([71]Figure 1H). Together, these results demonstrate
   that MEV exposure increases liver gluconeogenesis both in vivo and in
   vitro, and that MEV acted directly on hepatocytes to induce
   gluconeogenesis.

Figure 1.

   [72]Figure 1
   [73]Open in a new tab

   Gut microbiota-derived extracellular vesicles induce hepatic
   gluconeogenesis in vivo and in vitro. (a) Germ-free animals were orally
   administered 80 μg of DiD-labelled microbiota-derived extracellular
   vesicles (MEV) (n = 2) or PBS as control (n = 1). 12 h later, liver
   tissue was collected and fluorescence quantified ex vivo using the IVIS
   (In vivo imaging system). Fluorescence represented as Epi-fluorescence.
   Data representative of n = 2 independent experiments (b) Germ-free
   animals were orally administered 80 μg MEV in PBS (GF + MEV), or PBS
   alone (GF + PBS), twice-weekly for 3 weeks. Whole liver was then
   collected and RNAseq performed and differences in individual
   transcriptome represented on a PCA plot (n = 5 per group). (c) Volcano
   plot illustrating differentially down- and up-regulated genes in GF +
   MEV vs. GF + PBS mice. (d) Enrichment plot for GSEA Reactome pathway
   analysis of liver RNA transcriptome data, with significant enrichment
   in pathway associated with ‘FOXO mediated transcription’ and
   ‘Gluconeogenesis’ in GF + MEV livers (e) Specific-pathogen free (SPF)
   mice were administered 1 × 10^10 MEV (MEV), or PBS as control (PBS)
   daily for 5 weeks, and liver gene expression of Pck1 and G6pc
   quantified by qPCR (n = 6 per group). RNAseq experiment representative
   of n = 1 experiment (f) Primary hepatocyte cell isolated from control
   C57BL/6 male mice was cultured for 18 h in the absence (PBS) or
   presence of 3 × 10^9 MEV (MEV), and gene expression level of Pck1 and
   G6pc quantified by qPCR (n = 5 PBS, n = 6 MEV). Data representative of
   n = 2 independent experiments (g) HEPG2 human hepatocyte cell line was
   cultured for 18 h in the absence (PBS) or presence of 3 × 10^9 MEV
   (MEV), and gene expression level of PCK1 and G6PC quantified by qPCR (n
   = 6 per condition). Data representative of n = 5 independent
   experiments (h) SPF mice were administered 1 × 10^10 MEV (MEV), or PBS
   as control (PBS) daily for 5 weeks, and then pyruvate tolerance test
   performed, with blood glucose level measured to 2 h following
   intraperitoneal injection of pyruvate at a dose of 2.5 g/kg body weight
   following overnight fasting (n = 8 PBS and n = 6 MEV). Data
   representative of n = 1 experiment. All data are presented as mean ±
   SEM. ∗p < 0.05 and ∗∗p < 0.01 by Mann–Whitney test. For PTT experiment,
   overall difference was determined by two-way ANOVA.

2.2. Gut microbiota-derived extracellular vesicles promote hepatic
inflammation and metabolic syndrome

   Next, we sought to determine how increased exposure to MEV affects host
   metabolic health. Treatment with MEV for 5 weeks did not alter food
   intake ([74]Fig. S2a), nor affect body weight gain ([75]Figure 2A). MEV
   treatment did not affect inguinal white adipose tissue (WATi) weight
   but increased epididymal white adipose tissue (WATe) weight ([76]Fig.
   S2b). Notably, it significantly increased liver weight ([77]Figure 2B),
   a change often associated with inflammation [[78]18]. This was
   confirmed by Ingenuity pathway analysis of our transcriptomic data
   ([79]Figure 2C), revealing enrichment of pro-inflammatory pathways
   relating to IL-6 and acute phase response signalling, as well as HMGB1
   signalling pathways. Genes common to these pathways like Jun, Cebpb and
   Tnfrsf1b are generally considered pro-inflammatory and were
   significantly upregulated in MEV treated animals ([80]Figure 2D).
   Histological analysis revealed mild signs of liver inflammation in
   MEV-treated animals ([81]Figure 2E–F), particularly increased
   hepatocyte ballooning ([82]Fig. S2c), while no fibrosis was detected
   (data not shown). This was not accompanied by evidence of elevated
   systemic inflammation, with no differences in circulating levels of
   IL-1β, TNF, IL-6, MCP-1, KC and IL-12 ([83]Fig. S2d), suggesting that
   this effect was local. Excessive gluconeogenesis is a feature of
   obesity, and chronic low-grade inflammation is a feature of metabolic
   syndrome. To determine the physiological implication of increased
   hepatic gluconeogenesis and hepatic inflammatory state, we tested the
   impacts of MEV on glucose metabolism. We performed glucose tolerance
   and insulin tolerance test on mice treated with PBS or MEV for 5 weeks,
   which revealed that MEV administration induced mild but significant
   insulin resistance ([84]Figure 2G) and glucose intolerance ([85]Figure
   2H). Indeed, mouse primary hepatocyte that were pre-treated overnight
   with MEV had decreased Akt phosphorylation following stimulation with
   insulin, highlighting that MEV can directly induce insulin resistance
   ([86]Figure 2I). Altogether, these results demonstrate that increased
   MEV exposure can lead to a mild but significant disruption to metabolic
   health by promoting glucose intolerance and insulin resistance. This
   phenotype was not attributed to increased systemic inflammation.

Figure 2.

   [87]Figure 2
   [88]Open in a new tab

   Gut microbiota-derived extracellular vesicles promote low-grade hepatic
   inflammation and insulin resistance. (a) SPF mice were administered 1 ×
   10^10 MEV (MEV), or PBS as control (PBS) daily for 5 weeks. Graph
   represents % body weight change over the 5 weeks period (left) and
   scatter plot showing the percentage change at endpoint (right) (n = 10
   mice per group). Data representative of n = 3 independent experiments.
   (b) Scatter plot showing whole liver weight following 5 weeks oral
   administration of MEV or PBS (n = 10 mice per group). Data
   representative of n = 2 independent experiments. (c) Germ-free animals
   were orally administered 80 μg MEV in PBS (GF + MEV), or PBS alone (GF
   + PBS), twice-weekly for 3 weeks. Whole liver was then collected and
   RNAseq performed, and Qiagen Ingenuity pathway analysis performed, with
   (d) heatmap with key genes involved in the pathways shown. (n = 5 per
   group). Data represent of n = 1 experiment. (e) SPF mice were
   administered 1 × 10^10 MEV (MEV), or PBS as control (PBS) daily for 5
   weeks. Representative liver hematoxylin and eosin staining. Yellow
   arrows indicate hepatocyte ballooning. Scale bar represents 50 μm and
   (f) combined NAFLD activity score of liver histology samples. Data
   representative of n = 1 experiment and scores represent average scores
   of n = 2 independent scoring. (g) SPF mice were administered 1 × 10^10
   MEV (MEV), or PBS as control (PBS) daily for 5 weeks, and then oral
   glucose tolerance test performed, with blood glucose level measured to
   90min following oral administration of glucose at a dose of 2 g/kg body
   weight following 6 h of fasting (n = 10 mice per group). (h) SPF mice
   were administered 1 × 10^10 MEV (MEV), or PBS as control (PBS) daily
   for 5 weeks, and then insulin tolerance test performed, with blood
   glucose level measured to 90 min following intraperitoneal
   administration of insulin at a dose of 0.5U/kg body weight following 4
   h of fasting (n = 10 mice per group). Data represent of n = 1
   experiment. (i) Primary hepatocyte cell isolated from control C57BL/6
   male mice was cultured for 18 h in the absence (PBS) or presence of 3 ×
   10^9 MEV (MEV), and then stimulated with 10 nM insulin for 0, 5 or
   30min and protein levels of Akt and p-Akt quantified by Western blot
   (each condition pooled from n = 3 replicates of n = 1 mice). Data
   representative of n = 2 independent experiments. All data are presented
   as mean ± SEM. ∗p < 0.05 by Mann–Whitney test. For oGTT and ITT
   experiments, overall difference was determined by two-way ANOVA.

2.3. Host feeding patterns control the diurnal production of MEV to regulate
hepatic gluconeogenesis

   MEV can promote hepatic gluconeogenesis via numerous pathways. We
   previously found that MEV modulated host immunity through the
   activation of TLRs, such as TLR4, via LPS present on the MEV surface
   [[89]19]. LPS has also been shown to promote hepatic gluconeogenesis
   via the activation of the transcriptional regulator P300
   [[90]20,[91]21]. We found that MEV did not induce hepatic
   gluconeogenesis through TLR signalling, with MEV still able to
   upregulate PCK1 and G6PC expression in the presence of the NF-κB
   inhibitor BAY11-7082 ([92]Fig. S3a), and the P300 inhibitor C646
   ([93]Fig. S3b), suggesting that mechanisms independent of LPS were
   involved. We further confirmed that MEV-associated LPS was not the
   mediator of hepatic gluconeogenesis, as EV derived from both
   Escherichia coli (gram-negative, enriched in LPS) and Enterococcus
   faecalis (gram-positive, lacking LPS) could induce PCK1 and/or G6PC
   expression in HEPG2 cell line ([94]Figure 3A). It was also not due to a
   general increase in proinflammatory signalling pathways, mediated by
   JNK ([95]Fig. S3c). Other pathways known to induce gluconeogenesis,
   including ROS activation ([96]Fig. S3d–f) and SIRT1 ([97]Fig. S3g) were
   also not involved.

Figure 3.

   [98]Figure 3
   [99]Open in a new tab

   Gut microbiota-derived vesicles are produced rhythmically in response
   to host nutritional status and activate the circadian clock to promote
   gluconeogenesis. (a) HEPG2 hepatocyte cell line was cultured for 18 h
   in the absence (PBS) or presence of 3 × 10^9 gram-negative EV (G- EV)
   from Escherichia coli or gram-positive EV (G+ EV) from Enterococcus
   faecalis, and gene expression level of PCK1 and G6PC quantified by qPCR
   (n = 6 per condition). Data representative of n = 2 independent
   experiments. (b) Feces MEV concentration was quantified by Nanoparticle
   tracking analysis from the feces of control mice collected at Zeitgeber
   time 0 (start of light cycle), 6, 12 (start of dark cycle) and 18. Data
   represent of n = 1 experiment (c) Cecal MEV concentration was
   quantified by Nanoparticle tracking analysis from the feces of control
   mice that was fed ad libitum, or fasted overnight (18 h). Data
   representative of n = 1 experiment. (d)E. faecalis was grown in vitro
   at decreasing concentration of nutrient and EV was isolated and
   quantified by Nanoparticle tracking analysis. Data representative of n
   = 2 independent experiments. (e) Germ-free animals were orally
   administered 80 μg MEV in PBS, or PBS alone, twice-weekly for 3 weeks.
   Whole liver was then collected and RNAseq performed, and enrichment
   plot for the GSEA Reactome ‘Circadian Clock’ pathway analysis shown.
   Data represent of n = 1 experiment. (f) SPF mice were administered 1 ×
   10^10 MEV (MEV), or PBS as control (PBS) daily for 5 weeks, and then
   liver expression of Per1, Ciart1 and Rora was quantified by qPCR. Data
   representative of n = 2 independent experiments. (g) HEPG2 hepatocyte
   cell line was reverse transfected with CLOCK-siRNA or control
   negative-siRNA and cultured for 18 h in the absence (PBS) or presence
   of 3 × 10^9 MEV (MEV), and gene expression level of PCK1 and G6PC
   quantified by qPCR (n = 6 per condition). Data representative of n = 3
   independent experiments. All data are presented as mean ± SEM. ∗p <
   0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 by Mann–Whitney test (c,
   f), ordinary one-way ANOVA followed by Tukey's multiple comparisons
   test (a, b, d) or two-way ANOVA followed by Tukey's multiple
   comparisons test (g).

   We hypothesized that MEV may possibly control hepatic gluconeogenesis
   by regulating the circadian clock, given the rhythmic expression of
   PCK1 and G6PC observed in hepatocytes [[100]2]. This hypothesis was
   supported by several observations. We found that the quantities of MEV
   produced in the gut peaked at similar Zeitgeber times that hepatic Pck1
   and G6pc expression are reported to be at the highest [[101]22]
   ([102]Figure 3B). Notably, MEV concentration was lowest during the
   start of the feeding phase (dark cycle, Zeitgeber time 12), while it
   was highest during the start of the fasting phase (light cycle,
   Zeitgeber time 0), when glucose demand is highest. This demonstrates
   that MEV are produced in a diurnal pattern, and this directly aligns
   with host glucose requirements. To further demonstrate the relationship
   between host nutritional status and MEV production, we subjected mice
   to starvation via overnight fasting. Starvation was linked to a
   dramatic increase in MEV concentrations in the gut, with approximately
   3 times more MEV produced by the gut microbiota following overnight
   fasting compared to ad libitum fed animals ([103]Figure 3C). In
   addition to starvation, we previously found that host dietary
   composition can affect MEV production, with a high-protein diet
   associated with higher gut MEV concentrations [[104]19]. Indeed, we
   found that high protein content in diet was associated with higher
   hepatic expression of Pck1 ([105]Fig. S4a). Next, to directly test the
   relationship between nutrient status and bacterial EV production, we
   cultured E. faecalis, a common gut commensal microbe, at decreasing
   concentration of nutrients in growth media. We found that the
   quantities of EV produced was increased with decreasing nutrient
   concentration ([106]Figure 3D), highlighting that nutrient depletion is
   associated with increased EV production. Interestingly, we found that
   gram-positive bacteria generally produced higher quantities of EV than
   gram-negative bacteria ([107]Fig. S4b). To apply this to another
   physiological setting, we also found pregnant mice had increased MEV
   concentration during the third trimester of gestation ([108]Fig. S4c),
   the period where both energy demands [[109]23] and maternal
   gluconeogenesis [[110]24] are reported to be at their highest.
   Additionally, our hepatic transcriptome data revealed enrichment in the
   circadian clock pathway in MEV treated animals ([111]Figure 3E). The
   activation of circadian clock (via CLOCK/BMAL1) leads to the expression
   of clock-controlled genes, which includes the regulators period (PER)
   and cryptochrome (CRY) that negatively regulate CLOCK/BMAL1 activity
   [[112]25]. It also induces the expression of other regulators such as
   Rora and Ciart1. Accordingly, we found that the expression of these
   genes was differentially upregulated in MEV-treated mice in our
   transcriptomic dataset ([113]Supplemental Table 1). Consistent to this,
   SPF mice treated with MEV for 5 weeks also had upregulation of Per1,
   Rora and Ciart1 compared to PBS-treated animals ([114]Figure 3F), and
   this phenotype was also replicated in HEPG2 cell line treated with MEV
   in vitro ([115]Fig. S4d). Interestingly, MEV administration did not
   induce gluconeogenesis in the kidneys ([116]Fig. S4e), another major
   site of gluconeogenesis, suggesting that this effect is specific to the
   liver. This aligns with a previous report indicating that kidney
   gluconeogenesis is not regulated by circadian rhythm, unlike the liver
   [[117]2]. We also did not see significant changes in the small
   intestine ([118]Fig. S4f), with gluconeogenesis known to be triggered
   more strongly by nutrient availability at this site [[119]26]. Finally,
   siRNA knockdown of CLOCK in HEPG2 cell line ([120]Fig. S4g) lowered
   basal G6PC expression and abrogated MEV-induced upregulation of this
   gene ([121]Figure 3G). We found that transfected cells had expression
   of PCK1 at background levels (data not shown). Together, our data
   indicate that the MEV plays a role in supporting host glucose needs by
   enhancing hepatic gluconeogenesis, and this is associated with changes
   to the liver clock system.

3. Discussion

   We show for the first time that the gut microbiota, via EVs, can
   activate hepatic gluconeogenesis. MEV production exhibit diurnal
   regulation, which is linked to host nutritional status, and appears to
   have evolved to support host glucose demands ([122]Figure 4). MEV
   mediated this effect by acting as a potential peripheral Zeitgeber for
   the hepatic circadian system, and we hypothesize that this is a new
   pathway for host-gut microbe interactions. Indeed, greater quantities
   of MEV were found to be produced during periods of high glucose demands
   (i.e., fasting and pregnancy). This peripheral mode of regulation is
   applicable across mammalian species, regardless of whether they are
   nocturnal or diurnal, unlike central clock signalling which is
   regulated by the suprachiasmatic nucleus via light inputs. Our finding
   highlights several implications. Firstly, does different gut microbiota
   composition have different capacity for MEV production, and does this
   contribute in part to interindividual differences in baseline and
   fasting glucose levels? A minor increase in blood glucose level, along
   with other factors like genetics and dietary habits, might be
   sufficient to predispose to metabolic disorders. In our hands, we found
   that gram-positive bacteria generally produced higher levels of EVs
   than gram-negative bacteria, at least under in vitro conditions
   ([123]Fig. S4b). This may explain the observation found in some studies
   that obese individuals have a higher Firmicutes:Bacteroidetes ratio
   [[124]27]. Despite this, the fact that both gram-negative and
   gram-positive-derived EVs can activate gluconeogenesis suggests that
   the mechanism is not due to surface membrane components per se, but
   more likely driven by the delivery of intracellular cargo to
   hepatocytes. Host dietary composition can also affect MEV production.
   For example, we have previously shown that a high-protein diet was
   linked to greater quantities of gut MEV [[125]19]. This may partly
   explain the established effects of a high-protein diet on host
   gluconeogenesis with a high-protein, carbohydrate-free diet linked to
   increased gluconeogenesis in humans [[126]28]. We also validated this
   observation in mice utilizing 10 diets with different ratio of
   macronutrient composition, which revealed a positive association
   between liver Pck1 expression and protein content in diet, with a minor
   interaction with carbohydrate content ([127]Fig. S4a). As well as diet
   composition, irregular food intake common in modern societies may also
   impair the normal control of host glucose homeostasis by disrupting MEV
   production. While MEV act as a cue to support host glucose demands
   under normal conditions, it is possible that physiological disruption
   to MEV production (such as altered diet composition, irregular feeding
   and possibly medication etc.) may underlie increased metabolic
   disorders associated with a Western lifestyle.

Figure 4.

   [128]Figure 4
   [129]Open in a new tab

   Model of gut microbiota-derived extracellular vesicles induction of
   host hepatic gluconeogenesis. Schematic of the interplay between host
   nutritional status, gut microbiota production of extracellular
   vesicles, and host hepatic gluconeogenesis. Host nutrient status
   mediates gut microbial stress, with fasting leading to bacterial stress
   and increased production of microbiota-derived extracellular vesicles
   (MEV). MEV can translocate to the liver, where it influence the hepatic
   circadian clock, and is associated with increased expression of PCK1
   and G6PC, and the upregulation of hepatic gluconeogenesis. Under normal
   conditions, the diurnal regulation of MEV associates with host glucose
   requirements, leading to homeostasis. However, the disruption of MEV
   production, such as improper diet or other factors, can stimulate
   gluconeogenesis despite the host being euglycemic, which may lead to
   metabolic disruption and predisposition to metabolic disorders. Created
   in BioRender. Tan, J. (2025) [130]https://BioRender.com/qaenmir.

   Secondly, the fact that MEV directly activate gluconeogenesis in vitro
   indicates that MEV act independent of the diurnal actions of cortisol
   and glucagon on hepatic gluconeogenesis. This feature probably allows
   for the fine-tuning of hepatic gluconeogenesis, allowing the organism
   to adapt to different levels of hypoglycaemia. Indeed, the relative
   production in quantities of MEV was much greater in mice that were
   fasted overnight ([131]Figure 3C) compared to the difference observed
   during normal rhythmic oscillation ([132]Figure 3B). Aligned to this,
   we found that MEV acts on the expression of PCK1 and G6PC, and in a
   dose-dependent fashion, at least with G6PC in vitro ([133]Fig. S1e).
   The production of MEV was related to the nutritional status of the gut
   microbiota, directly linked to the host's nutritional intake. Stress
   (such as antibiotics) has been shown to promote bacterial EV production
   [[134]29], and in this study, we provide evidence that nutrient
   availability is another factor regulating bacterial EV production.

   In conclusion, our study provides new fundamental insights into the
   regulation of host glucose homeostasis. Assessing gut MEV profile holds
   significant clinical potential as a biomarker or risk factor for
   metabolic disorders, including gestational diabetes. Additionally, our
   study provides a mechanism for the control of host glucose homeostasis,
   via the alteration of MEV production. Although our short-term
   experiments revealed no overt systemic inflammation, chronically
   elevated MEV levels over the longer-term may trigger low grade
   inflammation and predispose to hepatic pathologies like non-alcoholic
   fatty liver disease and hepatocellular carcinoma, often associated with
   metabolic dysfunction. Inhibition of hepatic gluconeogenesis has been
   effective in ameliorating type 2 diabetes [[135]30]. Thus, developing
   strategies aimed at reducing gut MEV production, i.e. by shaping
   dietary composition, have incredible potential in helping alleviate
   hyperglycaemia, possibly by tuning the liver clock system. This could
   form part of a strategy for the control or treatment of metabolic
   disorders like type 2 diabetes, by reducing hepatic gluconeogenesis.

4. Methods

4.1. Animals

   Male C57BL/6 mice (6–8 weeks of age) were purchased commercially from
   Animal BioResources and housed at the Charles Perkins Centre animal
   facility. Animals were maintained under specific-pathogen-free
   conditions, with a standard 12 h light–dark cycle, at 25 °C and 50%
   humidity. Animals were fed ad libitum on AIN93G control diet, with free
   access to drinking water. Germ-free animals were purchased commercially
   from the Walter and Eliza Hall Institute of Medical Research and were
   housed at the Charles Perkins Centre germ-free animal facility. For
   biodistribution study, 80 μg of MEV was orally administered following a
   2-hour fasting period. For other experiments, the administration of MEV
   were performed in the late afternoon, and faeces and cecal content were
   collected in the morning, unless otherwise stated. All animal
   experiments were performed under protocols approved by the University
   of Sydney Animal Ethics Committee (AEC).

4.2. Extracellular vesicle isolation

   For the isolation of extracellular vesicles, cecum content or faecal
   pellets was homogenized in 0.02 μm filtered PBS and EV isolated by
   differential centrifugation, as described previously with slight
   modifications [[136]19]. Briefly, the homogenate was centrifuged at
   10,000×g for 20min at 4 °C followed by 18,000×g for 45min at 4 °C,
   before filtration through a 0.22 μm filter and centrifuged at 100,000×g
   for 2hr at 4 °C. EV pellet was resuspended in 0.02 μm filtered PBS. EV
   particle size and concentration were assessed by Nanoparticle Tracking
   Analysis using the Particle Metrix Zetaview, and analyzed with the
   Zetaview software (8.05.16 SP3) or the Nanosight NS300 (Malvern
   Instruments Limited), and analyzed with the NTA 3.2 software. MEV
   particle size distribution typically ranges from 133 to 425 nm (95%
   interval), with a median size of 211 nm. For fluorescent labelling of
   MEV, MEV was labelled with DiD dye at a final concentration of 5 μM for
   30min at room temperature in the dark, quenched with ice-cold 0.02 μm
   filtered PBS, and then washed by ultracentrifugation as above.

4.3. RNA sequencing and data analysis

   Total RNA was extracted from snap-frozen liver tissues using TRI
   Reagent (Sigma Aldrich) following the manufacturer's instructions, and
   RNA was purified using the RNeasy MinElute Cleanup Kit (Qiagen).
   Library preparation and Illumina HiSeq sequencing was done commercially
   (Azenta Life Sciences, previously Genewiz). Raw data had an average of
   22,232,315 reads per sample with 94.02% of reads with quality of Q30 or
   above. Sequences were aligned to the GRCm38 mouse reference genome
   using STAR v2.7.3a with two-pass mapping [[137]31]. Gene count was
   quantified using HTSeq and genes with less than 20 counts across all
   samples were filtered out prior to analysis. DESeq2 (1.42.0) was used
   for differential expression analysis [[138]32] using R software
   (4.3.1). DESeq2 normalised counts and statistical outputs were used for
   downstream pathway analysis using Gene Set Enrichment Analysis
   [[139]33] and Ingenuity Pathway Analysis (Qiagen). Raw and processed
   sequencing data are deposited in the NCBI GEO repository
   ([140]GSE295797).

4.4. Quantitative PCR

   Total RNA was extracted using TRI Reagent (Sigma Aldrich) following the
   manufacturer's instructions and cDNA generated using the High-Capacity
   cDNA Reverse Transcription Kit (ThermoFisher Scientific). qPCR was
   conducted using the Power SYBR™ Green PCR Master Mix (ThermoFisher
   Scientific) on a LightCycler® 480 Instrument II (Roche Life Science)
   and LightCycler® 480 v1.5.0 software. A list of primers used, and their
   sequences are provided in [141]Supplemental Table 2.

4.5. Histological assessment

   Liver tissues were fixed in 10% formalin and stored in 70% ethanol
   before processing and staining with H&E as previously described
   [[142]34]. NAFLD Activity Score (Steatosis, Inflammation and Ballooning
   index) was used to score liver inflammation and damage, as previously
   described [[143]35].

4.6. Cell culture

   HEPG2 was cultured in DMEM supplemented with 10% FBS, 10 mM HEPES, 2 mM
   l-glutamine, 1 mM sodium pyruvate, 100units/ml penicillin and 100 μg/ml
   streptomycin. For experiments, HEPG2 cells were seeded into 6-well
   plate and grown until ∼80% confluence. For siRNA experiment, knockdown
   of CLOCK was performed using Silencer pre-designed siRNA
   (GGAACAAUAGACCCAAAGGtt and CCUUUGGGUCUAUUGUUCCtc) from Ambion (Life
   technologies) with Lipofectamine RNAiMAX (ThermoFisher) via reverse
   transfection, and experiments performed 48 h later. C57BL/6 mouse
   primary hepatocytes were isolated as previously described [[144]36].
   Briefly, liver was perfused with 25 ml of HBSS containing 5 mM calcium
   chloride and 0.05% collagenase IV maintained at 37 °C, and then
   hepatocytes teased out by gentle shaking on a petri dish. Hepatocytes
   were then centrifuged at 50×g for 30min at 22 °C, then resuspended in
   37.5% Percoll solution and centrifuged for 500×g for 15min at 22 °C
   with minimal brake. RBC lysis was then performed, and hepatocytes
   resuspended in DMEM/F12 supplemented with 10% FBS, 5 mM HEPES, 2 mM
   l-glutamine, 2 mM sodium pyruvate, 1 μM dexamethasone, 100units/ml
   penicillin and 100 μg/ml streptomycin, and allowed to rest for 4 h on
   collagen coated plates (5 μg/cm^2) before media removed and cell
   resuspended in fresh media without FBS and dexamethasone for
   experiments.

4.7. Metabolic assessments

   For glucose tolerance test, mice were fasted for 6 h and were then
   given an oral bolus of glucose at a dose of 2 g/kg body weight. For
   insulin tolerance test, mice were fasted for 4 h, then 0.5 U/kg insulin
   was given by intraperitoneal injection. For pyruvate tolerance test,
   mice were fasted overnight, then given sodium pyruvate at a dose of 2.5
   g/kg body weight. Blood glucose level was measured using a glucometer
   (AccuChek Advantage).

4.8. Bacterial cell culture

   Frozen glycerol scrapping of Enterococcus faecalis (NCTC 775) was first
   inoculated in Brain Heart Infusion (BHI) broth and allowed to rest
   overnight, 37 °C shaking 200rpm. The culture was then diluted to OD0.1
   in BHI broth and grown for 18 or 24 h as indicated. Escherichia coli
   (K-12 MG1655) was grown in Luria–Bertani (LB) broth instead. For some
   experiments, E. faecalis were grown in BHI broth diluted with salt
   solution (4.5 g/L NaCl and 2.5 g/L sodium phosphate dibasic).

4.9. Western blots

   Samples were lysed by probe sonication (15s) in RIPA buffer
   supplemented with protease (Roche Mini Protease Inhibitor Cocktail) and
   phosphatase (Roche PhosSTOP) inhibitors before centrifuging (4 °C,
   10min, 16,000×g). Protein concentration was quantified with
   bicinchoninic acid (BCA) assay (Sigma–Aldrich). Lysates were prepared
   for polyacrylamide gel electrophoresis under reduced conditions.
   Proteins were resolved by western blot using 4–12% Bis-Tris gels
   (BioRAD) and transferred to a nitrocellulose membrane (BioRAD). After
   blocking with 5% BSA in TBST, protein expression was detected using
   specific primary antibodies by incubating overnight at 4 °C (Cell
   Signalling Technology: phospho-AKT #4058; AKT #4685; β-actin #4970).
   All antibodies were detected using secondary rabbit IgG-horse
   radishperoxidase (Cell Signalling Technology #7074) and visualised by
   enhanced chemiluminescence using the Pierce ECL Western Blotting
   Substrate (ThermoFisher Scientific).

4.10. Statistical analysis

   For comparison of two independent groups, the Mann–Whitney tests were
   used. For comparison of three or more groups, a one-way ANOVA with
   Tukey's multiple comparison test was used. For comparison of
   differences among group means over time, two-way ANOVA was used. For
   comparison of paired samples, non-parametric paired t-test was used.
   Mixture modelling was performed as previously reported
   [[145]19,[146]37]. P value < 0.05 were considered statistically
   significant with ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.005; ∗∗∗∗p < 0.001.

CRediT authorship contribution statement

   Jian Tan: Writing – review & editing, Writing – original draft,
   Validation, Investigation, Data curation, Conceptualization. Jemma
   Justine Taitz: Writing – review & editing, Investigation, Data
   curation. Duan Ni: Writing – review & editing, Validation,
   Investigation, Data curation. Camille Potier-Vilette: Writing – review
   & editing, Validation, Investigation, Data curation. Gabriela Pinget:
   Writing – review & editing, Investigation. Tamara Pulpitel: Writing –
   review & editing, Investigation, Data curation. Dragana Stanley:
   Writing – review & editing, Investigation. Ralph Nanan: Writing –
   review & editing, Investigation. Laurence Macia: Writing – review &
   editing, Supervision, Funding acquisition, Conceptualization.

Declaration of competing interest

   The authors declare the following financial interests/personal
   relationships which may be considered as potential competing interests:
   Laurence Macia reports financial support was provided by Australian
   Research Council. Laurence Macia reports a relationship with Sanofi
   that includes: employment. L.M. is a current employee of the
   Translational Science Hub Global Sanofi Vaccines R&D Brisbane,
   Australia. Her contribution to this work was done when she was an
   employee of the University of Sydney. The other authors declare no
   competing interests. If there are other authors, they declare that they
   have no known competing financial interests or personal relationships
   that could have appeared to influence the work reported in this paper.

Acknowledgments