Altered gut microbiota and metabolites profile are associated with reduced bone metabolism in ethanol‐induced osteoporosis

Abstract Objective Chronic heavy drinking causes ethanol‐induced osteoporosis (EIO). The present study aimed to explore the role of GM in EIO. Material and Methods A rat EIO model was established by chronic ethanol intake. Taking the antibiotic application as the matched group of dysbacteriosis, an integrated 16S rRNA sequencing and liquid chromatography–tandem mass spectrometry‐based metabolomics in serum and faeces were applied to explore the association of differential metabolic phenotypes and screen out the candidate metabolites detrimental to ossification. The colon organoids were used to track the source of 5‐HT and the effect of 5‐HT on bone formation was examined in vitro . Results Compared with antibiotics application, ethanol‐gavaged decreased the BMD in rats. We found that both ethanol and antibiotic intake affected the composition of GM, but ethanol intake increased the ratio of Firmicutes to Bacteroidetes. Elevated serotonin was proved to be positively correlated with the changes of the composition of GM and faecal metabolites and inhibited the proliferation and mineralization of osteogenesis‐related cells. However, the direct secretory promotion of serotonin was absent in the colon organoids exposed to ethanol. Conclusion This study demonstrated that ethanol consumption led to osteoporosis and intestinal‐specific dysbacteriosis. Conjoint analysis of the genetic profiles of GM and metabolic phenotypes in serum and faeces allowed us to understand the endogenous metabolite, 5‐HT, as detrimental regulators in the gut‐bone axis to impair bone formation.


| INTRODUCTION
Chronic heavy ethanol consumption leads to gradual damages to multiple systems, 1 which represents one of the most common causes of mortality worldwide. 2 Recent studies have indicated that chronic heavy ethanol consumption can be positively associated with bone impairments and an increased risk for bone fracture. 3 A large-scale case control study in Denmark reveals an alcoholism rate of 7.1% in patients with fractures versus only 2.5% in control subjects without fractures. 4 Although accumulating evidence shows that ethanol is an important risk factor for osteoporosis, 5,6 it remains largely unknown how bone loss and osteopenia occur as parts of many unwanted consequences of ethanol consumption. It is recently reported that chronic heavy ethanol consumption directly impairs gut microbiota (GM) composition, which might be a pivotal pathogenic factor or aggravate pre-existing illnesses. 7 Notably, studies have repeatedly shown that a gut-bone regulatory axis exists, by which GM composition could indirectly regulate bone metabolism. 8 However, it remains largely unknown whether the osteoporosis caused by chronic heavy ethanol consumption is related to the impairments of GM composition.
Bone is a dynamic tissue throughout life. The maintenance of healthy bone in human adults depends on the balance of bone resorption and formation, which is called bone remodelling. 9 The bone loss and osteoporosis following chronic heavy ethanol consumption are mainly due to the imbalance of bone remodelling. This imbalance eventually results in osteopenia, an established risk factor for osteoporosis. Results of human and animal experiments indicate a direct inhibitive effect of ethanol on bone-forming cells. [10][11][12] Biochemical and histological evaluation in patients with ethanolic bone disease reveals a marked impairment in bone formation in the face of relatively normal bone resorption. A well-defined rat model of heavy drinking shows that ethanol-induced bone loss rests with antiproliferative effects on osteoblasts, ultimately impairing bone remodelling and mineralization. 13 Although there is increasing evidence that ethanol plays roles in membrane perturbation and transmembrane signal pathways, the specific subcellular mechanisms whereby inhibit cell proliferation are unknown. Notably, Chronic, heavy ethanol consumption has been proved to be detrimental to every tissue of the body, which means, despite the direct impairments, ethanol-induced bone loss could be secondary by disrupting metabolic homeostasis of hormones, endogenous metabolites, and nutrient absorption, which are critical for bone forming and growth. 14,15 Several observations have previously suggested that circulating serotonin, as an endogenous metabolite, could be playing a significant role in the regulation of bone mass. 16,17 The Lrp5 gene deficient mice are shown to have high serotonin levels and a low bone mass phenotype. 16 While it is still unknown that whether the gutderived serotonin levels change in EIO. People with chronic heavy ethanol consumption show lower serum vitamin D levels, indicating mineralization disorders. 13 A closer examination of these indirect factors may contribute to developing interventional strategies for the treatment of ethanol-induced bone diseases.
Recent studies, however, reveal that chronic heavy ethanol consumption not only affects the gastrointestinal tract but also induces changes of microbiota composition in the gastrointestinal tract (GIT). 18 The GM is referred to as the second gene pool of the human body and a resident microorganism, both symbiotic and pathogenic, living in our gastrointestinal tract. Accumulating evidence highlights the importance of GM dysbiosis, which is postulated to be a major factor in human disorders. All clinical and preclinical data suggest that the quantitative and qualitative dysbiotic changes in the GM may contribute to ethanolrelated disorders. 19 Chronic heavy ethanol consumption may be associated with increased GIT inflammation and intestinal hyperpermeability, resulting in elevated serum levels of lipopolysaccharide (LPS), 20 systemic inflammation, and eventually tissue damage. 21 Sjogren et al. for the first time discover the relation between microbiota and bone development and demonstrate the higher bone mass formation in germ-free mice than that in normal mice. 22 Subsequently, recent findings, however, provide substantial evidence for the existence of a GM-bone axis, by which GM influences the skeletal homeostasis via affecting the host metabolism, immune function, and hormone secretion. [23][24][25] Alterations in the GM may serve as biomarkers or therapeutic targets for glucocorticoid-induced osteoporosis. 26,27 It is noteworthy that ethanol-related disorders are associated with quantitative and qualitative changes of GM dysbiotic, which may contribute to the intestinal hyperpermeability to luminal bacterial products and influence the skeletal homeostasis via the GM-bone axis.
Although such epidemiologic analyses demonstrate the underlying GM-bone axis mechanism in bone loss, present evidence mainly focuses on the direct damage on the bone-forming cell by ethanolinduced oxidative stress, instead of the relevance of GM dysbiosis to bone homeostasis. Hence, it is meaningful to study GM-related phenotype alterations in ethanol-induced osteoporosis for understanding the modulating pathways involving in the GM-bone axis, raising the possibility of therapeutic strategies for ethanol-induced bone disorders.
In this study, a rat model of chronic heavy ethanol consumption was used to investigate the relevance of ethanol-induced GM dysbiosis to the metabolic phenotype alterations in the serum and faecal using 16S rRNA gene sequencing and liquid chromatography-tandem mass spectrometry (LC/MS)-based metabolomics.

| Animal management and faecal sampling
In this study, a total of 27 adult (7-8 weeks old) male Sprague-Dawley (SD) rats (purchased from Beijing Vital River Laboratory Animal Technology Co.) were used, weighing 250-300 g. Animals were housed in groups of n = 3-4 (medium density) rats per cage on a reverse 12 h light/dark cycle under standard temperature and humidity conditioned at the conventional SPF Animal Care Facility. Standard food and tap water were available ad libitum in the home cages. Animals were maintained under constant conditions for 10 days before the start of the experiments and under daily surveillance by veterinary staff and/or experimenters.
All animals were divided into three groups, which including saline (S) group, ethanol (E) group and antibiotic (A) group (n = 9 in each group). In ethanol group, rats were gavaged with water containing 20% (vol/vol) ethanol (10 ml/kg, 6 times/week, 1 time/day, 16 weeks). The method was slightly modified based on existing articles. 28,29 The antibiotic-treated rats were given antibiotic gavage containing 1 g/L metronidazole and 0.2 g/L ciprofloxacin (10 ml/kg, 6 times/week, 1 time/day, 16 weeks). The rats in the S group were given the same amount of saline by gavage as the E group. To avoid contamination, fresh faeces were collected from the terminal rectum of each rat. A total of 27 faecal samples from the rectum were collected on ice, immediately frozen in liquid nitrogen, and then stored at À80 C until microbiome and metabolome analysis.

| Micro-computed tomography scanning
All the right tibia bones were dissected, cleaned, and fixed in 4% formalin fixative. Then, they were scanned by micro-computed tomography (micro-CT) (SCANCO, Switzerland) with a spatial resolution of 20 μm and the X-ray tube potential of 90 kV and 480 μA. The tibiae analyses were performed on trabecular bone defined as beginning proximal (a distance of 1% of total bone length) to the growth plate and then extending 10% of total bone length toward the diaphysis, excluding cortical bone. Different microstructural parameters were analysed, which included trabecular bone mineral density (BMD), trabecular bone volume/tissue volume ratio (BV/TV), trabecular number (Tb. N), trabecular bone separation (Tb. Sp) and trabecular bone thickness (Tb. Th). Analyses were carried out by an operator, blinded to the treatment assignment of samples. This technology, previously described further in detail, 30 was considered the gold standard for assessing three-dimensional analysis of bone microstructure.

| Biochemical analysis of serum parameters
Blood samples were allowed to clot, and the serum was collected by centrifugation. The levels of serum diamine oxidase (DAO) (Elabscience, China), D-lactate (D-LA) (Elabscience) and serum calcium (Nanjing Jiancheng Bio.) were measured by using commercial kits according to the manufacturer's protocols.

| Real-time PCR
The total RNA of the frozen colonic tissues and colonic organoids were After sequencing, the obtained reads were analysed using the QIIME2 pipeline for taxonomic classification. Then, by using DADA2, we obtained feature table and feature sequence. Alpha diversity was applied to measure the species diversity for samples through 2 indices, including Chao1 index and Good's Coverage. Beta diversity was visualized by principal coordinate analysis (PCoA) based on Unweighted Unifrac and analysis of similarities (ANOSIM Analysis). Blast was for sequence alignment, and the feature sequences were annotated with SILVA database for each representative sequence. Other diagrams were implemented using the R package (v3.5.2).

| hiPSCs maintenance and colonic organoid differentiation
To induce definitive endoderm formation, hiPSCs were cultured on feeders to 70% confluence, and then plated at a density of 6000 clumps per well in a matrigel-coated (24-well plate). For accutase split cells, 10 μM Y27632 compound (Sigma) was added to the media for the first day. After the first day, media was changed to mTESR1 and cells were grown for an additional 24 hours. Cells were then treated with 100 ng/mL of Activin A for 3 days and treated with hindgut induction medium (RPMI 1640, 2 mM L-glutamine, 2% decomplemented FBS, penicillin-streptomycin and 100 ng/mL Activin A) for 4 days with 500 ng/ml FGF4 (R&D) and 3 μM Chiron 99021 (Tocris) to induce formation of mid-hindgut spheroids.

| Cell viability assay
Cells were seeded at a density of 2 Â 10 4 cells/well in 96-well plates. After the cell confluence reached 80%, the cells were treated with 5-HT from 10 À9 to 10 À3 M for 1 day and 3 days, respectively.

| Association analysis and statistical analysis
All data were expressed as mean values ± standard deviation (SD) for independent experiments. For comparison between two groups, a paired t-test was performed. Statistical differences groups were analysed by one-way ANOVA followed by the Tukey test. Spearman's correlation was used to show the relations between parameters of flora and metabolites. Statistical analysis was performed using R3.5.1 and SPSS 25.0 (IBM). p < 0.05 was considered statistically significant.  Figure 1C,G), and a similar trend was found in trabecular number (Tb.N) (p < 0.01) and trabecular thickness (Tb.Th) (p < 0.05, Figure 1D,E). In contrast, trabecular separation (Tb.Sp) increased significantly in ethanol group (p < 0.01, Figure 1F). Otherwise, the decreasing calcium in serum also reflected an imbalance in bone metabolism (p < 0.01; Figure 1H), even though the phosphate levels is not change significantly. Taken together, these results verified that the animal model of osteoporosis successfully established, which caused by chronic heavy drinking. Meanwhile, the antibiotic group did not cause bone quality changes.

| Ethanol impairs the gut epithelial barrier integrity
In order to observe the pathological changes of colon tissues treated with ethanol, the histological sections were evaluated by H&Estaining. The colonic tissue of Ethanol group showed obvious infiltration of inflammatory cells, decrease of goblet cells and the muscularis externa were widened (Figure 2A). We determined the relative expression of tight junction genes, including Ocln and Cdh1, and found significantly decreased expression in the E group compared with the S group (p < 0.01, Figure 2B). In Western Blot, we also saw that the expression levels of connexin, E-cadherin and Occludin in the colon tissue of E group were lower than those of the S groups ( Figure 2C). Furthermore, gut leakage biomarkers in rat serum were also determined. Our results showed that serum diamine oxidase (DAO) and D-lactose (D-LA) were significantly enriched in the E group (p < 0.01, Figure 2D,E). These results demonstrated that the function of the gut epithelial barrier was impaired by ethanol, which might cause intestinal microorganisms or their toxins leaking into the body.

| Ethanol alters the balance of intestinal bacterial diversity
The 16S rRNA gene sequencing data were obtained from the three group, 27 stool samples, with a mean of 66,965 ± 11,179 sequences per specimen. A total of 1,808,057 high-quality reads were obtained after noise dropout and error correction. According to the results of the Kruskal-Wallis rank sum test and linear discriminant analysis, the phylum abundance of Proteobacteria, Tenericutes, Actinobacteria and Firmicutes was increased and the abundance of Bacteroidetes was decreased in the E group and A group as compared to the S group. . Values represent as bar graph with mean ± standard deviation (SD). *p < 0.05, **p < 0.01 Among them, the change level of Firmicutes/Bacteroidetes ratio in E group was significantly (p < 0.05, Figure 3A,B).
To check whether the sequencing data was sufficiently, rarefaction analysis (Chao1 and Good's Coverage index) was performed for these samples. Chao1 estimator represented the level of bacterial community richness and evenness, which was significantly lower in the E group and A group than the S group ( Figure 3C). Good's Coverage index was significantly higher in the A group than the S group and E group ( Figure 3D). Meanwhile, there was no statistic difference in terms of the Good's Coverage index between E group and S group (eventhough the average value of S group was slightly lower than that of E group). The value in each group was near saturation (saturation value = 1), which suggested that the sequencing data was sufficiently robust and few new species was undetected. We further performed Unweighted UniFrac-based principal coordinates analysis (PCoA) and Analysis of similarities (ANOSIM) based on Feature abundances and found significant differences among the three groups. In detail, PCoA revealed a distinct clustering of the GM composition in each group ( Figure 3E). ANOSIM showed that the E group was significantly different from S and A groups ( Figure 3F), suggesting that ethanol-induced osteoporosis as well as GM dysbiosis.
At the genus level, the relative abundances of Catabacter, Peptostreptococcaceae, Lactobacillus, Turicibacter, Erysipelotrichaceae_UCG-003 and Romboutsia belonging to the Firmicutes phylum, were significantly higher in the E group than the other two groups ( Figure 4A, B). Furthermore, the relative abundances of Bifidobacterium, Lachnospiraceae, Muribaculaceae and Roseburia were significantly higher in the S group ( Figures 3A and 4A). These species-specific changes may be highlight candidates in the pathogenesis of EIO.

| Serum and faecal metabolism profiles in ethanol-induced osteoporosis
The serum and faecal metabolic profiles were analysed by LC/MS. Differences metabolite profiles of the three groups were revealed by OPLS-DA. A total of 322 and 374 differential metabolites were identified in serum and faeces, respectively, between the S group and E group ( Figure 5A). These metabolite features were largely distinguished subjects with osteoporosis from those non-osteoporosis, which indicated of broad metabolic differences between the two groups. Compared with the S and A group, ethanol-gavaged rats had higher concentrations of (2S,2 0 S)-Oscillol, 2-(Formylamino) benzoic with Capecitabine. We also found that delta-Tocotrienol was negatively correlated with many high levels of serum metabolites, including (2S,2 0 S)-Oscillol, MPTP and 5-HT. It had already been reported that F I G U R E 3 Effect of ethanol and antibiotic gavaged on gut-microbiota composition (n = 9). (A) Annotation of phylum, family and genus level for the three groups. (B) Firmicutes/Bacteroidetes ratio (F/B ratio) at phylum level. Alpha diversity analysis, Chao1 (C) and goods coverage (D) represent the difference in diversity within the three groups. Beta diversity analysis, PCoA plot (E) and ANOSIM plot(F), represent the difference in diversity among the three groups. * p < 0.05, ** p < 0.01 delta-Tocotrienol was associated with osteoblast differentiation and promoted alkaline phosphatase synthesis. 32 Interestingly, delta-Tocotrienol also has significant negative correlations with g_Oxyphotobacteria. In the faecal metabolism, osteoporosis-reduced Lipoxin A4 correlated negatively with g_Oxyphotobacteria and g_Ralstonia, which had reported that it could inhibit the activity of osteoclasts and reduced the bone loss caused by ovariectomized rats. 33 In addition, we found that osteoporosis-enriched N-Methyl-Lglutamic acid and alpha-Phosphoribosylpyrophosphoric acid in the faecal were positively associated with the same five genera. These five genera included g_Muribaculaceae, g_Ralstonia, g_Catabacter, g_Mollicutes_RF39 and g_Papillibacter ( Figure 6A). Next, we chose serotonin (5-HT), which had been reported in the literature, as the core target to observe whether it affected the balance of bone metabolism. 34 Through correlation analysis, we found that four kinds of serum metabolites, five kinds of faecal metabolites and seven kinds of different bacterial genera had a clear correlation with the change levels of 5-HT ( Figure 6B). Altogether, these results indicated that the distinguishing metabolites were closely related to GM variation and the distinguished metabolites and GM were related to osteoporosis, even though it remains to be explored whether these metabolites are directly produced by the intestinal bacteria.

| The effect of ethanol on the colonic organoid
In vitro, we successfully cultured and induced iPSCs cells into colonic organoids. In the organoid supernatant test for five consecutive days, we found that the release content of 5-HT by colonic organoids in the ethanol group was slightly lower than that in the control group, while there was no statistical significance ( Figure 7A). The accumulation content of 5-HT in the organoid supernatant for 3 days was still no significant difference between the two groups ( Figure 7B). We also found the levels of the Cdh1 genes depression (p < 0.05, Figure 7C).
Through immunofluorescence detection, we found that the expression of E-cadherin and ZO-1 protein on the surface of colon organoids decreased after ethanol intervention ( Figure 7D).  Figure 8D). Otherwise, we also saw that 5-HT had an inhibitory effect on the bone formation and mineralization ability of BMSCs ( Figure 8E, F).

| DISCUSSION
There is solid evidence that chronic heavy ethanol consumption has detrimental effects on bone health and increases the risk of osteoporosis. 35 Although heavy drinking negatively impacts bone formation, the underlying mechanisms by which ethanol affects bone turnover are poorly understood. Alcoholic injury is multi-systemic, predominantly the alcoholic liver disease (ALD). In ALD patients, alcohol impairs the gut epithelial barrier and aggravates ALD through activation of inflammation induced by bacterial endotoxin and lipopolysaccharide. Moreover, the changes in gut microbiota composition occur before ALD development. These results demonstrate that alterations F I G U R E 5 Faecal and Serum Metabolomic Signatures in osteoporosis (n = 9). (A) Volcano plot represents the difference changes in serum and faecal metabolites after ethanol-gavaged. (B) Removing the metabolites that coexisting both in the Ethanol group and Antibiotic group, the top 5 and bottom 5 differential metabolites caused by ethanol were screened out. (C) Enrichment analysis for differential metabolite pathways. The abscissa represents the Rich factor corresponding to each pathway, the ordinate represents the pathway name, and the colour of the point represents p-value. The colour of the points represents the significance of the enrichment. The size of the dot represents the number of different metabolites enriched in the gut microbiome are recognized as a major factor in the ALD progression. 36 Studies in the amount and diversity of bacterial populations of patients with osteoporosis indicate that osteoporotic adults appear to have reduced diversity of GM, which reveals that GM is also a central regulator of bone homeostasis and the pathogenesis of osteoporosis, 37,38 in what is now being called the gut-bone axis. However, the mechanism underlying this connection how ethanol affects bone remodelling via GM composition is still somewhat enigmatic.
To validate the correlation between GM dysbiosis and osteoporosis in this study, GM composition by 16S rRNA gene seq and bone density by micro-CT are tested in rats with chronic ethanol intake. Meanwhile, antibiotics are applied in rats for 16 weeks as a GM dysbiosis matched group, which has been proven to reduce taxonomic richness and diversity and influence the abundance of bacterial taxa in faeces. 39 Previous studies have shown that the bone strength is not affected by long-term low-glycemic diet containing antibiotics, but GM is significantly changed. 40 Here, we report that the bacterial community richness and even- increased serum endogenous metabolite, 5-HT, were significantly correlated with the GM dysbiosis, the changes of seven differential gut microbiota, including g_catabacter, g_Mollicutes_RF39, g_Ralstonia and so on. Further, we find that the elevated 5-HT level impairs the bone formation in vitro, which reaffirms that the GM dysbiosis might be an indirect pathogenic factor to disrupt the bone formation in the ethanol-induced osteoporosis. Although the specific subtypes of GM that impairs bone formation are unknown for lack of a proof-ofconcept study, these results indicate that maintaining the GM haemostasis might be a viable treatment strategy for serotoninrelated gut-bone axis disorders.