Preclinical Evaluation of Sodium Butyrate’s Potential to Reduce Alcohol Consumption: A Dose-Escalation Study in C57BL/6J Mice in Antibiotic-Enhanced Binge-Like Drinking Model

Alcohol addiction, specified in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), as alcohol use disorder (AUD), has diverse negative health effects with strong implications in the liver and brain [1]. The existing treatment approaches for AUD are limited to behavioral counseling and a few pharmacotherapies with low efficacy and patient compliance [2, 3], enhancing the need in developing newer, more effective strategies to combat various aspects of AUD.

Intestinal dysfunction, present during a number of neuropathologies including AUD, is an area of intense investigation with regard to its potential as a therapeutic target [4]. The gut microbiota, a complex ecosystem of microorganisms dominated by Bacteroidetes and Firmicutes bacterial phyla, play a crucial role for host health including digestion and absorption of nutrients, and functioning of immune and other organ systems [5]. A number of studies have demonstrated shifts in gut bacterial composition during AUD, serving the basis for a role of gut-liver and gut-brain axes in development of AUD [5]. Specific changes largely depend on the clinical or preclinical alcohol exposure paradigms [6]. Overall, alcohol consumption is shown to reduce relative abundances of Firmicutes, Bacteroidetes, and Actinobacteria with an increase in abundances for Proteobacteria, Enterobacteria, and Verrucomicrobia [7, 8]. Increases in gram-negative bacterial species may be the underlying causes of compromised intestinal wall integrity or “leaky gut” and associated endotoxin lipopolysaccharide (LPS)-induced inflammation, known gut-related effects of chronic ethanol exposure [9].

Intestinal dysbiosis, caused by chronic alcohol use, was found to correlate with altered levels of gut bacteria-derived metabolites, such as short-chain fatty acids (SCFAs) [5, 10]. SCFAs are a major class of lipid metabolites resulting from bacterial fermentation of dietary fibers in the colon and have pleiotropic beneficial functions including energy metabolism, anti-inflammation, epigenetic regulation, etc. [11]. An essential SCFA butyrate has been implicated in beneficial responses through modulation of metabolic processes and peripheral immune system function, stimulation of the vagus nerve, and endocrine signaling [12]. Chronic alcohol use has been shown to cause reduction in the abundance of butyrate-producing bacterial species, such as Firmicutes-belonging Clostridium genera [7]. In line with this, our previous work with the use of nonabsorbable antibiotic cocktail (ABX) to induce dysbiosis caused dramatic reduction in butyrate-producing bacterial populations, Lachnospiraceae unclassified, Clostridium Cluster IV, and Clostridium Cluster XIVa, and this was inversely correlated with ethanol consumption levels in mice [13].

Emerging research suggests that interventions aimed at restoring the gut microbiota composition and optimizing the production of beneficial metabolites such as butyrate could have a positive impact during neurological conditions and substance use disorders [14, 15]. Animal studies with the use of various forms of butyrate, such as sodium butyrate and tributyrin (butyrate precursor), have shown a positive impact on the liver, metabolism and cardiovascular health through modulation of gut barrier integrity, prevention of blood brain barrier breakdown, and reduced inflammation [16‒18]. In our previous study, ad libitum supplementation of C57BL/6J male mice with 8 mg/mL SB was able to reverse the increase in ethanol consumption and neuroinflammatory response induced by treatment with ABX but could not lower the baseline drinking levels in voluntary binge-like drinking in the dark (DID) alcohol consumption paradigm [19, 20]. These findings suggested that SB supplementation might offer a novel approach to managing AUD. In the current study, we further evaluated the preclinical potential of higher concentrations of SB (20 and 50 mg/mL in drinking water). The ability of these SB concentrations to protect against chronic alcohol consumption was tested in C57BL/6J male mice in the ABX-enhanced binge-like DID paradigm. Potential adverse effects of these SB concentrations on various parameters, such as food and liquid intake, body and liver weights, and intestinal parameters were investigated. To overcome potential smell/taste aversion of the high concentrations of SB provided in drinking water, a follow-up study was set wherein SB was administered via gavage at 160 mg/kg and 320 mg/kg doses.

Procurement of adult male C57BL/6J mice, aged 6–8 weeks, was done from Jackson Laboratories (Sacramento, CA, USA). These mice were given a minimum of 2 weeks to adapt to their new environment before being randomly assigned to different treatment groups. The mice were housed in a room where humidity, temperature, and lighting were automatically regulated, following a reversed 12-h light/dark cycle with lights on from 12:00 am to 12:00 pm. Facility-provided standard mouse chow (Laboratory Rodent Diet 5001) was available at all times along with drinking water, unless stated otherwise. The National Institutes of Health Guide for Care and Use of Laboratory Animals was followed, along with adherence to protocols approved by the USC Institutional Animal Care and Use Committee, to ensure the proper treatment of all animals.

Mice had ad libitum access to drinking water containing a cocktail of non-absorbable ABX that do not enter the host circulation and have the ability to reduce a broad range of gut microbiota populations in adult rodents when administered orally [21]. The ABX cocktail was a mixture of 0.5 mg/mL bacitracin (Sigma Aldrich, St Louis, MO, USA), 2.0 mg/mL neomycin (GoldBio, St Louis, MO, USA), and 0.2 mg/mL vancomycin (Thermo Fisher Scientific, Waltham, MA, USA). Pimaricin (1.2 μg/mL, Molekula, Lewisville, TX, USA) was also added as an anti-fungal agent to the ABX cocktail solution in order to control gut fungal overgrowth due to continuous antibiotic use. SB (Sigma) solution was provided to the mice in the drinking water at 20 mg/mL and 50 mg/mL or through oral gavage at 160 mg/kg and 320 mg/kg to measure a dose dependent effect. SB doses used in the gavage study were estimated to produce comparable exposures to SB concentrations used in the oral administration procedure over 24-h period. For ABX and ad libitum SB co-treatment, SB at above mentioned concentrations was mixed with freshly prepared ABX solution. The pH of SB and ABX+SB solutions were adjusted to the pH of the drinking water to avoid any additional effects. All solutions were freshly prepared every 2 days. Experimental designs of both studies, specifying the number of mice used in each condition, are shown on Figures 1a and 4a.

Mice in all groups were started on the DID procedure after respective pretreatment periods, i.e., 2 weeks of SB ad libitum ± ABX or 1 week of SB gavage ± ABX. Mice were provided a single bottle with 20% ethanol (20E) 3 h into the dark cycle for 2 h to replace their treatment bottle Monday-Friday for 4 weeks. Volumes (in mL) of ethanol solution were recorded at the end of each drinking session. This procedure results in consistent high ethanol intake levels in C57BL/6J male mice as shown in our earlier studies [13, 20]. For DID, ethanol intake was calculated as g/kg (g of pure ethanol per kg of body weight; 20E intake = [mL of 20E consumed × 0.158 g/mL]/body weight in kg).

Body weights (g), food (g), and non-DID liquid intakes (mL) were measured every other day (Monday, Wednesday, Friday) throughout the study. After necropsies, the liver and cecum (where warranted) were weighed (g), intestinal lengths (cm) were measured, blood collected through cardiac puncture, samples allowed to clot for 1 hr, and serum obtained by centrifugation at 2,000 g for 10 min. Several 1 cm pieces were collected from the left liver lobe and stored in −80°C for further analysis. Similarly, intestinal tissues were separated into their respective sections, such as the duodenum, ileum, and colon, and were stored in −80°C until further use.

Blood ethanol concentration (BEC) in serum samples were measured using EnzyChromTM Ethanol Assay Kit with colorimetric detection (Bioassay Systems, Hayward, CA, USA). Standards and 10 μL serum samples were mixed with enzymatic mix and substrates in a 96-well plate, and after incubating for 30 min at room temperature, reaction was stopped and the plate was read at 570 nm in Synergy H1 plate reader (Agilent Technologies, Santa Clara, CA, USA). BEC in samples were estimated from the standard curve and presented as g/dL or %.

Serum LPS was measured using PierceTM Chromogenic Endotoxin Quant kit (Thermo). Low range standards and serum samples (50 μL) were heat equilibrated on 37°C plate heater and incubated with the reagent at 37°C for an additional time, after which the reaction was stopped, chromogenic substrate added, and absorbance read at 540 nm. LPS concentrations (EU/mL) were calculated from the standard curve after subtracting the average blank absorbance value.

Liver and serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT) activities were measured using colorimetric assays from Sigma. Around 50 mg of tissue samples were homogenized using 200 μL of ice-cold AST or ALT assay buffer and centrifuged at 13,000 g for 10 min. AST and ALT activities were determined by incubating 20 μL of cleared homogenates or serum samples with a reaction mix containing enzyme mix, substrate and developer at 37°C and absorbance read at 450 nm every 5 min till it reached plateau. AST and ALT activities were calculated from the following formula AST/ALT activity = B × V × D/(T2-T1) where B is the amount of either glutamate (for AST) or pyruvate (for ALT), obtained from the standard curves, V is the original sample volume, D is sample dilution, and T1 and T2 are the times of the first and the last readings of the linear portion of the curve. Data are presented as mU/mg protein for the liver or as mU/mL for serum.

Liver triglycerides (TGs) were measured using the EnzyChromTM Triglyceride Assay Kit (Bioassay Systems, Hayward, CA, USA). Ten µL of 5% Triton X-100 solubilized samples were incubated in 96-well plates with 100 μL working reagent containing assay buffer, enzyme mix, lipase, ATP, and a dye reagent for 30 min at room temperature. Subsequently, OD was read at 570 nm using Synergy H1 plate reader (Agilent). TG concentrations were calculated using a standard curve and expressed as nM/mg protein.

Duodenal tissue pieces (1 cm) were washed carefully to remove any food particles, lysed using RIPA buffer (10 mm Tris, 140 mm NaCl, 1 mm EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, pH 8.0) with added protease inhibitors (Protease inhibitor cocktail, Sigma). Samples were cleared with centrifugation and BCA assay (Thermo) was performed to determine protein concentrations. For Western blotting, 30 μg of protein was separated using 10% SDS-PAGE on Mini-PROTEAN TGX Gel (Bio-Rad, Hercules, CA, USA), proteins transferred to nitrocellulose membranes using Trans-Blot Turbo Membrane Transfer System (Bio-Rad). After the transfer, nonspecific binding on membranes was blocked using 5% non-fat milk and blotted against occludin (1:1,000; rabbit monoclonal antibody, ab216327; Abcam, Waltham, MA, USA) and GAPDH (1:1,000, mouse monoclonal antibody; 97166S, Cell Signaling; Danvers, MA, USA); incubated overnight at 4°C. The protein bands were visualized following incubations with anti-rabbit or anti-mouse HRP-conjugated secondary antibodies (1:10,000, Bio-Rad) for 1 h at room temperature and incubation with enhanced chemiluminescence substrate (Clarity^TM Western, Bio-Rad). Western blot analysis was performed using Image J software. GAPDH densities were used for normalization of occludin data.

All data were analyzed using GraphPad Prism (GraphPad, San Diego, CA, USA) and are presented as mean ± SEM. One-way or two-way ANOVA was performed for group comparisons with Sidak or Bonferroni posttest multiple comparisons between individual groups, data for TG and occludin were analyzed using an unpaired two-tailed t test. The significance was defined as p < 0.05.

There were significant effects of SB treatments on the final body weights of the mice (F(5, 28) = 58; p < 0.001; one-way ANOVA) (Fig. 1b). Supplementation of mice with the lower SB20 concentration alone and in combination with ABX treatment had a small reduction in body weights with a significance trend compared to the H₂O and ABX groups, respectively (Fig. 2a). Supplementation with the higher SB concentration 50 mg/mL SB (SB50) alone caused a large reduction in body weights compared to the H₂O group (17.7 ± 0.6 vs. 27.8 0.06 g; p < 0.001; Fig. 2a). There was a smaller but significant weight reduction with SB50 when combined with ABX treatment versus ABX alone (25.3 ± 0.08 vs. 28.3 ± 0.4 g, p < 0.05; Fig. 1b). Body weights were not different between the ABX and H₂O treated groups.

Body weight changes throughout the duration of the study showed a similar and a steady increase for mice in H₂O, ABX, and 20 mg/mL SB (SB20) groups (online suppl. Fig. S1A; for all online suppl. material, see https://doi.org/10.1159/000540882). In contrast, mice in SB50 alone group steadily lost weight through the duration of the study (online suppl. Fig. S1A). When combined with ABX, both ABX+SB20 and ABX+SB50 groups had an initial dip in body weights for the first 2 weeks or before the start of the DID, later gaining weight to close the gap toward the end of the study, with ABX+SB20 recovering faster than mice in ABX+SB50 group (online suppl. Fig. S1A). There was no apparent effect of the DID exposure which started at the 3rd week of the study on weight changes of any of the treatment groups.

Changes in liver weights had overall similar patterns to body weights (Fig. 1c). There was a significant effect of SB treatments on the liver weights (F(5, 28) = 8.5; p < 0.001; one-way ANOVA). SB20 did not cause a significant change in the liver weights while SB50 treatment had a significant reducing effect. ABX treatment did not cause a significant change compared to the H₂O group. Small but significant reductions of liver weights were observed at both SB concentrations when combined with ABX (Fig. 1c). Due to similar patterns of body and liver weights, we also calculated the ratios of the 2 parameters. As expected, these ratios showed no significant differences between individual treatment groups (Fig. 1d). However, significant reduction in the liver and body weight ratios was noted when comparing data of all ABX groups to no ABX groups (0.036 ± 0.0009 vs. 0.045 ± 0.0017; p < 0.001).

We evaluated food and liquid intake in all experimental groups to see whether effects on body weights were due to differences in those indices. There were significant effects of treatment (F(5, 476) = 34.2; p < 0.001), time (F(16, 476) = 10.2; p < 0.001) and their interaction (F(80, 476) = 1.9; p < 0.001) for overall food intake, analyzed for the whole duration of the study and confirmed with two-way ANOVA. Food intake was not different between the SB20 and H₂O groups; however, it was significantly increased for SB50, and all the ABX groups compared to the H₂O group (Fig. 2a). Food intake was similar between ABX and ABX+SB groups at both SB concentrations.

Because of the observed reductions in body weights for SB50, ABX+SB20, and ABX+SB50 groups before the start of the DID (See online suppl. Fig. S1A), we analyzed changes in food intake for 2 time periods, i.e., before DID (weeks 1 and 2) and during DID (weeks 3–6). Food intake was not significantly different between individual groups for the weeks 1–2 (online suppl. Fig. S1B). However, during DID period (weeks 3–6) food intake significantly increased in SB50, ABX, ABX+20, and ABX+SB50 groups compared to the H₂O group, and these changes mirrored the overall food intake (Fig. 2a). This was confirmed with two-way ANOVA with significant treatment (F(5, 325) = 35.4; p < 0.001) and time effects (F(11, 325) = 6.0; p < 0.001).

We also analyzed the overall liquid intake for the whole duration of the study. Two-way ANOVA revealed significant effects of treatment (F(5, 569) = 61.7; p < 0.001), time (F(18, 569) = 13.5; p < 0.001) and their interaction (F(90, 569) = 3.0; p < 0.001) for this parameter. Overall liquid intake was not different between the ABX and H₂O groups (Fig. 2b). There was a small and significant increase in overall liquid intake in SB20 versus H₂O group, whereas it was significantly reduced in the SB50 group (Fig. 2b). There were also significant reductions in liquid intake for the combined effects of SB and ABX in both ABX+SB20 and ABX+SB50 groups versus ABX alone. The patterns for liquid intake were similar for non-DID and DID-periods (online suppl. Fig. S1C) and resembled that of overall liquid intake (Fig. 2b). The pattern for overall body and liver weights agreed well with the overall liquid intake (Fig. 1b, c vs. 2b) but not the food intake (Fig. 1b, c vs. 2a). These data suggested that there was a smell/taste aversion to higher concentrations of SB.

ABX treatment led to an increase in overall ethanol (20E) intake calculated for the whole duration of DID (Fig. 2c), which was similar to findings from previous studies [13, 20]. SB20 alone or in combination with ABX treatment did not affect the overall 20E intake, while SB50 alone caused a significant increase compared to water treatment. This effect was consistent throughout the 4 weeks of DID (online suppl. Fig. S2A). In combination with ABX, SB50 also caused a significant increase in 20E intake only during the 1st week of DID (online suppl. Fig. S2A). In addition, the increase in 20E uptake caused by ABX treatment were significant in weeks 3 and 4 of DID.

Blood was obtained from mice at the end of the study to analyze the effects of treatments on BECs (Fig. 2d). One-way ANOVA did not show any significant differences between the groups, despite some visible increases for SB50. The BEC increase with SB50 was significant when considering the liver weights (online suppl. Fig. S2B).

There were no differences in the activity of liver AST between all treatment groups (online suppl. Fig. S3A). Liver ALT activity with SB50 and ABX+SB50 treatments was visibly elevated but these shifts did not reach significance (online suppl. Fig. S3B). The liver AST/ALT ratio was significantly reduced for SB50 treatment group (Fig. 3a; p < 0.05). We also analyzed serum AST and ALT activities as part of measures of liver health. When combined with ABX, both SB concentrations reduced serum AST activity compared to that of ABX only group (online suppl. Fig. S3C), and serum ALT was not different between groups (online suppl. Fig. S3D). The reducing effect for ABX+SB20 and ABX+SB50 was also evident in serum AST/ALT ratios (Fig. 3b). Reduction of liver AST/ALT suggested that SB50 treatment might have induced steatosis in liver. Based on that, we analyzed liver triglyceride (TG) levels. Indeed, TG levels were elevated in the SB50 group (Fig. 3c); however, they showed only a significance trend when applying an unpaired t test (p = 0.084). There were no effects of SB when combined with ABX (Fig. 3c).

We analyzed intestinal parameters as ABX and SB treatments are known to alter microbiome and intestinal epithelial integrity [22‒24]. SB treatments without and with ABX did not affect the intestinal length at the time of necropsy including small intestine and colon (Fig. 4). Intestinal length significantly increased with ABX treatment and was consistent with treatments combined with SB (Fig. 4a). To find out whether this effect was associated with intestinal wall leakiness, we analyzed serum LPS levels and found no significant differences between groups (Fig. 4b).

We also analyzed the tight junction protein occludin in the duodenum. ABX treatment resulted in a significant decrease in the expression of the monomeric forms and significant increase in occludin dimer (Fig. 4c). When applied alone, SB20 significantly increased the expression of the monomers but not the dimer, and when combined with ABX it increased both the monomers and the dimer of occludin, but these increases did not reach significance. SB50 treatment caused an increase in the monomeric form with a significance trend but did not affect the dimer expression, whereas combined with ABX it significantly upregulated the expression of both forms of occludin (Fig. 4c). It appeared that SB and not ABX was the dominant factor for the increase in the monomeric form of occludin (2.0 ± 0.02 for SB vs. 0.81 ± 0.08 no SB groups, p < 0.0001; 1.6 ± 0.3 for ABX vs. 1.5 ± 0.16 for no ABX groups, p = 0.77). On the contrary, the changes in the expression of the dimeric occludin were governed by ABX (2.36 ± 0.3 for ABX vs. 1.05 ± 0.11 for no ABX groups, p < 0.0001), while SB treatment was without effect (1.9 ± 0.3 for SB vs. 1.3 ± 0.12 for no SB, p = 0.14).

In a separate study, SB was administered through a gavage procedure to test our assumption that SB caused a compensatory increase in ethanol intake due to taste/smell aversion that mice had when SB was supplemented in drinking water. All the mice had access to ABX and a gavage of saline or 2 doses of SB (160 and 320 mg/kg) was performed on mice every other day throughout the duration of the study including 2 weeks of pretreatment and 3 weeks of DID. SB administered through gavage caused reduction in 20E intake in the DID procedure reaching significance with a higher dose (Fig. 5b; 5.1 ± 0.55 for Gavage-SB320 group vs. 6.2 ± 0.62 g/kg in Gavage-Saline group, p < 0.05). There were no significant differences in several bodily parameters, i.e., body, liver, cecum weights, intestinal length (Fig. 5c), which matched well with no differences in the overall food and liquid intake (Fig. 5d).

The current study was set to evaluate the pre-clinical potential of SB to confer AUD. Specifically, the safety and efficacy of increasing concentrations of SB administration in drinking water was tested in the ABX-enhanced voluntary binge-like ethanol consumption paradigm in C57BL/6J male mice. In our previous studies, ad libitum supplementation of C57BL/6J male mice with 8 mg/mL SB was able to reverse the increase in ethanol consumption and neuroinflammatory response induced by treatment with ABX without any adverse effects on mouse health [19, 20]. In the current study, increasing concentrations of SB to 20 and 50 mg/mL induced negative effects on several parameters, such as liquid intake and bodily indices of mice with prominent effects at the higher 50 mg/mL concentration. Importantly, there was a loss of ability to protect against chronic alcohol consumption at these higher SB concentrations. These effects were most likely due to smell/taste aversion to the SB solution at the tested concentrations. This idea was supported by the findings of the follow-up proof-of-concept study where SB was provided to mice via a gavage needle. The higher gavage dose (320 mg/kg) was able to significantly reduce ethanol consumption without causing any adverse effects on any tested parameters.

SB supplementation in drinking water with 20 and 50 mg/mL SB caused adverse effects on bodily parameters, such as body and liver weights. Specifically, body weights plummeted early on in the study (1–2 weeks) with both concentrations of SB when applied with ABX. There was a slow but not complete recovery toward the end of the study for ABX+SB conditions which occurred faster in the ABX+SB20 group compared to the ABX+SB50 group. In addition, both SB20 and SB50, when applied alone, caused reductions in body weights with SB20 inducing a smaller effect. SB50 alone caused a large reduction in the body weights which steadily worsened throughout the study duration reaching ∼60% of initial numbers toward the end of the study. Further, effects on liver weights measured at the end of the study were similar to those of the final body weights. While SB20 did not affect the liver weights, SB50 treatment had a large reducing effect (20%). Smaller effects were also found for ABX+SB20 and ABX+SB50 combinations (∼10% for both groups). A number of studies have demonstrated the reducing effect of SB supplementation on all metabolic symptoms including body weight gain, adiposity, and liver steatosis in models of obesity and diabetes, and these have been viewed as beneficial [24]. However, no changes in body weight gain and other metabolic parameters with SB supplementation were found in rodents fed standard diet. In most of these studies, SB was provided as a dietary supplement (1 or 5% w/w) [25‒28]. In a study, SB was provided in drinking water at 0.1 m, equivalent to 11 mg/mL concentration, with no effect on standard diet fed animals [29]. In agreement with this finding, in our previous studies, we did not observe changes in body weights when using a comparable 8 mg/mL concentration of SB provided ad libitum [20].

Differences in body and liver weights obtained at the end of the study agreed well with the overall liquid but not food intake. It appears that SB50 alone and both SB concentrations combined with ABX caused an aversion to the liquid intake, while mice still tolerated the smell and taste of SB20 alone, with even a small increase in the intake. SB is notorious for its pungent smell and taste. In our experience, mice have a slight preference for SB when provided at lower concentrations (8 mg/mL) in drinking water [20]. However, increasing SB concentrations may have caused smell (and potentially taste) aversion precluding mice to drink liquid from SB containing bottles resulting in dehydration and other adverse effects.

The assumption on SB smell/taste aversion was further supported by the findings on ethanol consumption. Our expectation was that higher SB concentrations will have a stronger protection against ethanol intake in the DID paradigm. However, SB20 and SB50 were not able to reduce ethanol intake enhanced by ABX treatment (ABX+SB20 and ABX+SB50 vs. ABX). Moreover, SB50 alone group consumed significantly more ethanol compared to the water treated group which resulted in higher BEC levels, especially when taking into consideration liver weights. Because mice tended to drink less liquid during the non-DID period due to aversion to SB smell/taste, they may have compensated for the liquid intake during the DID period, when a single bottle with ethanol and not containing SB was provided to all the experimental groups. This effect may have masked the potential beneficial effects of SB on ethanol consumption. We then set up a follow up proof-of-concept study to bypass the self-administration of SB and apply SB doses through a gavage procedure. SB at the higher gavage dose (320 mg/kg) was able to significantly reduce the ethanol consumption over the ABX only treatment in the DID procedure. These findings highlighted the importance of the proper form of the administration of SB and added more support for the beneficial effect of SB in reducing binge-like ethanol consumption.

SB doses administered via gavage procedure did not cause adverse effects on any of the parameters tested, i.e., food and liquid intakes, body and liver weights, intestinal measures. On the contrary, administration of SB50 alone caused an acute liver damage as suggested by the <1.0 ALT/AST ratio due to increased ALT and not AST. Steatosis may have accounted for this effect, as supported by the increased TG levels with SB50 alone treatment. In addition, lower serum AST/ALT ratios were found for combinations of SB treatments with ABX as a result of reduced serum AST levels, most probably due to extrahepatic causes. SB treatments alone or in combination with ABX did not affect the intestinal parameters such as the intestinal length or serum LPS. The latter would indirectly suggest changes in gut permeability and leakiness. In fact, both SB treatments alone or in combination with ABX-induced increases in duodenal occludin (monomeric), an intestinal tight junction protein. The effect is largely supported by earlier published work regarding the effects of SB on the expression of intestinal tight junction proteins including occludin in various disease models [29, 30]. Combined with finding in AST/ALT ratios, these findings suggested that the acute liver damage with SB50 was not due to intestinal leakiness and inflammation rather caused by dehydration effects due to smell/taste aversion.

An interesting secondary outcome of this work concerns the effects of ABX on the intestinal parameters. All the ABX groups, without and with SB supplementation, caused an increase in the whole intestinal length. This is different from previous studies which reported no changes in this parameter with a similar protocol of ABX treatment over 4 weeks [31, 32]. Notably, we found that ABX treatments caused a shift from monomeric to dimeric occludin when tested in the duodenum. It is possible that dimerization of this tight junction protein (and possibly other ones) can lead to the thickening of the intestinal wall and in the intestinal length. While others have demonstrated reduction in intestinal occludin with ABX treatment [32], we are the first to report a shift from the monomeric to the dimeric form of this intestinal tight junction protein. The increases in intestinal length also agreed well with significantly reduced liver/body weight ratios with ABX treatment groups suggesting an association between these parameters. ABX-induced dramatic alterations in microbiota, especially in the cecum, are known to cause malabsorption of nutrients (vitamins, lipids, amino acids) and hence reduction in white adiposity, and organ and body weights. Previous studies are consistent regarding reduced fat content [33, 34] but vary in their findings on body weights, with some reporting reduction [31, 34] and others no change with ABX treatment [35, 36]. In addition, ABX-induced increase in body water content and enlarged ceca, a well-established consequence of ABX treatment [13, 31, 37], may compensate for reductions in adiposity and organ weights. In our study, ethanol consumption may also be another factor influencing the final liver weights of normal versus ABX treated mice, thus affecting the liver/body weight ratios. Further in-depth studies are needed to dissect specific changes in organ versus body weights and bring more support for the conclusions driven from our current findings.

This study has number of limitations. We could have envisioned the negative impact of the smell/taste of increasing concentrations of SB in drinking water, thus considering a different form for SB supplementation at the inception of the study. Further, the SB administration in the gavage study was performed only 3 days a week, every other day. We chose this approach to reduce the extra stress to the mice, caused by the gavage procedure. However, this may have reduced the true effects of SB on ethanol intake and precluded proper comparisons of data between the gavage doses and SB concentrations provided in drinking water. In addition, histological analyses of the liver for steatosis or testing the kidney function would add more support for the conclusions driven from our findings especially on SB50-induced adverse effects. These and other factors will be considered in our future preclinical studies.

Adverse effects found with high SB concentrations provided to mice in drinking water, including dramatic drop in body and liver weights, reduced overall liquid intake, acute liver toxicity, and loss in the protective effect against ethanol consumption were most probably due to aversion to the smell and taste to the high concentrations of SB. This hypothesis was supported by the findings of the follow-up proof-of-concept study with the intragastric administration of SB. In our future preclinical studies, we will consider other forms of administration of SB including dietary supplementation [28], intraperitoneal injection [38], the use of microencapsulated SB [39], a more palatable SB analog N-(1-carbamoyl-2-phenyl-ethyl butyramide [40]), or butyrate precursor tributyrin [18].

We would like to thank undergraduate students Albright Batdorj, Gwyneth Do, Kevin Frost, Nane Kaymakamyan, and Isabella Santoyo for help with animal studies, tissue collection, running biochemical assays and Western blotting; Simon Deng for help with language editing of the manuscript.

This study protocol was reviewed and approved by the USC Institutional Animal Care and Use Committee, Approval No. 21090.

The authors have no conflicts of interest to declare.

This work was supported in part by National Institute of Alcohol Abuse and Alcoholism (NIAAA)/National Institute of Health (NIH) AA022448 (to D.L.D.), Rose Hills Foundation and Titus Fund Research Awards (USC; to L.A.), USC Mann School of Pharmacy and Pharmaceutical Sciences and USC Good Neighbor.

G.C.H. participated in conceiving and conducting the experiments, data analyses and interpretation, preparation of the original draft and references; A.T.C.T. and S.V. participated in the gavage study, tissue collection, data analysis, and editing the text. D.L.D. contributed to manuscript review and funding acquisition; L.A. was responsible for study concept, design, interpretation of data, writing oversight and final formatting of the manuscript, funding acquisition, and project administration. All authors have read and agreed to the published version of the manuscript.

All data supporting findings of this study are available within the paper and the online supplementary material.