Metformin Influence on the Intestinal Microbiota and Organism of Rats with Metabolic Syndrome

Metformin is a first-line drug for DM2 treatment and prevention, but its complex effect on impaired glucose tolerance (IGT), including its influence on myocardial resistance to ischemia-reperfusion injury, is not completely studied. We aimed to evaluate the influence of metformin on the intestinal microbiota (IM), metabolism, and functional and morphological characteristics of myocardium in rats with IGT. IGT was modelled in SPF Wistar rats with a high-fat diet and streptozotocin and nicotinamide injection. Rats were divided into three groups: IGT (without treatment), IGT MET (metformin therapy), and CRL (without IGT induction and treatment). IGT group was characterized by: higher body weight, increased serum glucose and total cholesterol levels, atherogenic coefficient, impairment in the functional parameters of the isolated heart during perfusion, and larger myocardium infarction (MI) size in comparison with the CRL group. IM of IGT rats differed from that of CRL: an increase of Bacteroides, Acinetobacter, Akkermansia, Roseburia, and a decrease of Lactobacillus genera representation. Metformin therapy led to the diminishing of metabolic syndrome (MS) symptoms, which correlated with IM restoration, especially with the growth of Akkermansia spp. and decline of Roseburia populations and their influence on other members of IM. The obtained results allow us to consider from a new point of view the expediency of probiotic A. muciniphila use for MS treatment.


Introduction
Metabolic syndrome (MS) is currently one of the most outstanding (important, or actual) problems of medicine. Importantly, MS with disorders of carbohydrate and fat metabolism can be considered as a set of modifiable risk factors for cardiovascular diseases (CVD) development [1]. MS components, such as obesity, arterial hypertension, dyslipidemia, and glucose metabolism impairment, including impaired glucose tolerance (IGT) and type 2 diabetes mellitus (DM2), not only often take place together but pathogenetically influence each other comprising so-called cardiovascular continuum. One of the drugs for the treatment of DM2 is metformin. It has been shown that this drug reduces gluconeogenesis, glucose absorption, secretion of glucagon-like peptide 1 (GLP-1), increases glucose utilization [2], activates the energy sensor 5 -adenosine monophosphate-activated protein kinase (AMPK) [3,4]. As previously described, the cardioprotective effect of metformin and its anti-diabetic action are connected with: (1) inhibition of complex 1 of the mitochondrial

Body Weight
The results of comparing the weight of rats from different groups are presented in Figure 1. Initially, body weight did not differ significantly between groups. During the first 4 weeks weight gain in rats receiving HFD was slightly higher compared to rats fed with standard chow, though the difference was not significant statistically. IGT induction led to a significant body weight gain, which was thereafter observed in the IGT group up to the end of the experiment. Metformin administration to IGT rats caused a diminishing of body weight gain compared to IGT rats without treatment. Body weight in the IGT MET group was similar to that in the CRL group.
Attention was drawn to the tendency to decrease the weight of the omentum in the IGT MT group of rats. According to this parameter, differences were found only between the control group and the IGT group ( Figure S1).

Isolated Heart Function and Myocardial Infarct Size
Changes in LVP taking place during the global ischemia are shown in Figure 2A. LVP was higher in the IGT group than in the CRL group with the most outstanding differences observed at the beginning of the ischemic period-on the 10th and 15th minutes. Metformin administration did not influence LVP, which was as high in the IGT MET group as in the IGT group.

Isolated Heart Function and Myocardial Infarct Size
Changes in LVP taking place during the global ischemia are shown in Figure 2A. LVP was higher in the IGT group than in the CRL group with the most outstanding differences observed at the beginning of the ischemic period-on the 10th and 15th minutes. Metformin administration did not influence LVP, which was as high in the IGT MET group as in the IGT group.
The baseline LVDP values were similar in all the groups. LVDP in both IGT and IGT MET groups was higher than in the CRL group during the whole reperfusion period. Interestingly, no significant differences were observed between IGT and IGT MET groups (Figure 2A,B).

Isolated Heart Function and Myocardial Infarct Size
Changes in LVP taking place during the global ischemia are shown in Figure 2A. L was higher in the IGT group than in the CRL group with the most outstanding differen observed at the beginning of the ischemic period-on the 10th and 15th minutes. Met min administration did not influence LVP, which was as high in the IGT MET group a the IGT group. The baseline LVDP values were similar in all the groups. LVDP in both IGT and MET groups was higher than in the CRL group during the whole reperfusion period terestingly, no significant differences were observed between IGT and IGT MET gro (Figure 2A,B).
In accordance with the changes observed for LVDP, LVEDP in the IGT and IGT M groups was lower than in the CRL group, with no significant difference between the and IGT MET groups ( Figure 2C). There was a tendency for a slight increase in LVEDP in IGT MET group during the 75th and 90th min of the reperfusion period, but it did not reach value typical for control animals ( Figure 2C). Surprisingly, CFR values, being similar in all gr at baseline, was significantly higher in both IGT and IGT MET groups in comparison with the group throughout the reperfusion period, with no difference between IGT and IGT MET gr ( Figure 2D).
Myocardial infarction size in the IGT group was significantly larger than in the group. Metformin administration to animals with IGT led to a decrease in myocar damage size in comparison with the IGT group, MI size in the IGT MET group did differ from that in the CRL group ( Figure 3). In accordance with the changes observed for LVDP, LVEDP in the IGT and IGT MET groups was lower than in the CRL group, with no significant difference between the IGT and IGT MET groups ( Figure 2C). There was a tendency for a slight increase in LVEDP in the IGT MET group during the 75th and 90th min of the reperfusion period, but it did not reach the value typical for control animals ( Figure 2C). Surprisingly, CFR values, being similar in all groups at baseline, was significantly higher in both IGT and IGT MET groups in comparison with the CRL group throughout the reperfusion period, with no difference between IGT and IGT MET groups ( Figure 2D).
Myocardial infarction size in the IGT group was significantly larger than in the CRL group. Metformin administration to animals with IGT led to a decrease in myocardial damage size in comparison with the IGT group, MI size in the IGT MET group did not differ from that in the CRL group ( Figure 3). . Myocardial infarction size in isolated rat hearts subjected to 30 min of global ischemia followed by 90 min of reperfusion. Notes: Results are presented as median (25%; 75%). CRL-control group, IGT-impaired glucose tolerance group, IGT MET-impaired glucose tolerance + metformin group.

Biochemical Parameters
We analyzed blood biochemical parameters in rats-BGL and lipid profile (TC, TG, HDL, and LDL).
BGL in the IGT and IGT MET groups was significantly higher than in the CRL group . Myocardial infarction size in isolated rat hearts subjected to 30 min of global ischemia followed by 90 min of reperfusion. Notes: Results are presented as median (25%; 75%). CRL-control group, IGT-impaired glucose tolerance group, IGT MET-impaired glucose tolerance + metformin group.

Biochemical Parameters
We analyzed blood biochemical parameters in rats-BGL and lipid profile (TC, TG,  HDL, and LDL).
BGL in the IGT and IGT MET groups was significantly higher than in the CRL group directly after nicotinamide + streptozotocin injection. Thereafter BGL in animals with IGT not receiving treatment remained significantly higher than that in control rats, although not exceeding the normal value. Metformin administration to IGT rats lead to a BGL decrease-there was no difference in BGL in the CRL and the IGT MET groups (Figure 4).

Figure 3.
Myocardial infarction size in isolated rat hearts subjected to 30 min of global ischemia followed by 90 min of reperfusion. Notes: Results are presented as median (25%; 75%). CRL-con trol group, IGT-impaired glucose tolerance group, IGT MET-impaired glucose tolerance + met formin group.

Biochemical Parameters
We analyzed blood biochemical parameters in rats-BGL and lipid profile (TC, TG HDL, and LDL).
BGL in the IGT and IGT MET groups was significantly higher than in the CRL group directly after nicotinamide + streptozotocin injection. Thereafter BGL in animals with IGT not receiving treatment remained significantly higher than that in control rats, although not exceeding the normal value. Metformin administration to IGT rats lead to a BGL de crease-there was no difference in BGL in the CRL and the IGT MET groups (Figure 4).  Dynamics of blood glucose level in rats from different groups. Notes: Results are presented as median (25%; 75%). C-CRL(control group), IGT-impaired glucose tolerance group, IGT METimpaired glucose tolerance + metformin group. * p < 0.05, CRL both groups (IGT and IGT MET groups). in comparison with CRL group, § p < 0.05, in comparison with "IGT" group. Despite observed differences in BGL between the IGT and the IGT MET groups it seems important to emphasize that BGL mostly remained in the reference interval, therefore we can suggest that revealed the cardioprotective effects of metformin cannot be completely explained by glucose profile normalization.
Differences in the level of TG, HDL, LDL, CRI-1 and CRI-2 were not detected. However, total cholesterol (TC) levels ( Figure 5A) and atherogenic coefficient (AC) ( Figure 5B) were lower in group IGT and were fully recovered after metformin therapy.
Despite observed differences in BGL between the IGT and the IGT MET groups it seems important to emphasize that BGL mostly remained in the reference interval, therefore we can suggest that revealed the cardioprotective effects of metformin cannot be completely explained by glucose profile normalization.
Differences in the level of TG, HDL, LDL, CRI-1 and CRI-2 were not detected. However, total cholesterol (TC) levels ( Figure 5A) and atherogenic coefficient (AC) ( Figure 5B) were lower in group IGT and were fully recovered after metformin therapy.

qPCR
The study was carried out by comparing the following microorganisms (the quantitative content of intestinal microbiota representatives): the total number of bacteria, Acinetobacter spp., Citrobacter spp., Escherichia coli, and enteropathogenic E. coli, Proteus spp., Lactobacillus spp., Bifidobacterium spp., Bacteroides thetaiotaomicron, Bacteroides fragilis group, Clostridioides difficile, Clostridium perfringens, Enterococcus spp., Faecalibacteriumprausnitzii, Fusobacteriumnucleatum, Parvimonasmicra, Roseburia inilinivorans, and Akkermansia muciniphila. The statistically significant results and trends obtained when comparing the microbiota of rats from different groups by qPCR are presented in Figure 6. A lower content of lactobacilli and a higher amount of Acinetobacter spp. were found in the IGT group. In the IGT MET group, the content of Akkermansia muciniphila was greater than after IGT induction without further therapy. After the introduction of metformin (IGT MET group), the content of Acinetobacter spp. differed from the control values to a significantly lesser extent. It should also be noted, that in the IGT MET group, the content of lactobacilli significantly less prominently differed from the CRL group compared with the IGT group. Attention was drawn to the tendency of the number of Roseburiainulinivorans to decrease after metformin administration.

qPCR
The study was carried out by comparing the following microorganisms (the quantitative content of intestinal microbiota representatives): the total number of bacteria, Acinetobacter spp., Citrobacter spp., Escherichia coli, and enteropathogenic E. coli, Proteus spp., Lactobacillus spp., Bifidobacterium spp., Bacteroides thetaiotaomicron, Bacteroides fragilis group, Clostridioides difficile, Clostridium perfringens, Enterococcus spp., Faecalibacteriumprausnitzii, Fusobacteriumnucleatum, Parvimonasmicra, Roseburia inilinivorans, and Akkermansia muciniphila. The statistically significant results and trends obtained when comparing the microbiota of rats from different groups by qPCR are presented in Figure 6. A lower content of lactobacilli and a higher amount of Acinetobacter spp. were found in the IGT group. In the IGT MET group, the content of Akkermansia muciniphila was greater than after IGT induction without further therapy. After the introduction of metformin (IGT MET group), the content of Acinetobacter spp. differed from the control values to a significantly lesser extent. It should also be noted, that in the IGT MET group, the content of lactobacilli significantly less prominently differed from the CRL group compared with the IGT group. Attention was drawn to the tendency of the number of Roseburiainulinivorans to decrease after metformin administration.

Metagenome (16S rRNA) Study
The representation of Firmicutes was greater in the IGT group and was equal to normal values after metformin administration ( Figure 7A). The relative abundance of Actinobacteria was greater in the IGT MET group than in other groups ( Figure 7B). Results of the metagenome study are presented at the genera (

Metagenome (16S rRNA) Study
The representation of Firmicutes was greater in the IGT group and was equal to normal values after metformin administration ( Figure 7A). The relative abundance of Actinobacteria was greater in the IGT MET group than in other groups ( Figure 7B). Results of the metagenome study are presented at the genera (

Metagenome (16S rRNA) Study
The representation of Firmicutes was greater in the IGT group and was equal to normal values after metformin administration ( Figure 7A). The relative abundance of Actinobacteria was greater in the IGT MET group than in other groups ( Figure 7B). Results of the metagenome study are presented at the genera (   The representation of Akkermansia spp. was higher in both study groups (IGT, IGT MET) than in the CRL group. However, the use of metformin led to an even greater increase in these bacteria populations ( Figure 8A). The relative abundance of Bacteroides spp. increased in both IGT and IGT MET groups, but in the IGT MET group was significantly less prominent than in the IGT group ( Figure 8B). The representations of Roseburia spp. ( Figure 8C) and Lachnospiraceae family (Supplement materials, Figure S2) was higher in both study groups (IGT and IGT MET) compared to the CRL group, being lower in the IGT MET group than in the IGT group.

Correlation Analysis between Intestinal Microbiomecontentand Other Parameters
The summary effects of IGT induction and metformin administration on metabolic and functional parameters and intestinal microbiome are presented in Tables S1 and S2 as well as in Figures S3 and S4 (Supplementary Materials Section). In the correlation analysis, we focused on those bacterial taxa, whose quantity and representation changed after metformin treatment, as well as those biological parameters that are MS components and changed after treatment with this drug.

Correlation Analysis between Bacterial Taxa
Correlation analysis of the microbiota components with Akkermansia spp. revealed a positive correlation between the abundance of this bacteria and Lactobacillus spp., Faecalibacterium sp. and Bifidobacterium spp., as well as Actinobacterium relative abundance (Figure 9A). At the same time, there was a negative correlation between the taxa A. muciniphyla and Acinetobacter spp. quantity and Lachnosperaceae family in fecal samples. Correlation analysis of the microbiota components with Roseburia spp. revealed a positive correlation between this bacteria and the quantity of Faecalibacterium sp. and a negative correlation with Actynobacteria filum, Bifidobacteroim, Lactobacillus, and Bacteroides genera relative abundance ( Figure 9B). The representation of Akkermansia spp. was higher in both study groups (IGT, IGT MET) than in the CRL group. However, the use of metformin led to an even greater increase in these bacteria populations ( Figure 8A). The relative abundance of Bacteroides spp. increased in both IGT and IGT MET groups, but in the IGT MET group was significantly less prominent than in the IGT group ( Figure 8B). The representations of Roseburia spp. ( Figure 8C) and Lachnospiraceae family (Supplement Materials, Figure S2) was higher in both study groups (IGT and IGT MET) compared to the CRL group, being lower in the IGT MET group than in the IGT group.

Correlation Analysis between Intestinal Microbiomecontentand Other Parameters
The summary effects of IGT induction and metformin administration on metabolic and functional parameters and intestinal microbiome are presented in Tables S1 and S2 as well as in Figures S3 and S4 (Supplementary Materials Section). In the correlation analysis, we focused on those bacterial taxa, whose quantity and representation changed after metformin treatment, as well as those biological parameters that are MS components and changed after treatment with this drug.

Correlation Analysis between Bacterial Taxa
Correlation analysis of the microbiota components with Akkermansia spp. revealed a positive correlation between the abundance of this bacteria and Lactobacillus spp., Faecalibacterium sp. and Bifidobacterium spp., as well as Actinobacterium relative abundance ( Figure 9A). At the same time, there was a negative correlation between the taxa A. muciniphyla and Acinetobacter spp. quantity and Lachnosperaceae family in fecal samples. Correlation analysis of the microbiota components with Roseburia spp. revealed a positive correlation between this bacteria and the quantity of Faecalibacterium sp. and a negative correlation with Actynobacteria filum, Bifidobacteroim, Lactobacillus, and Bacteroides genera relative abundance ( Figure 9B).

Correlation Analysis between Bacterial Taxa and Metabolic Symptoms Body Weight
Of great interest is the strong positive correlation between the body weight and both content ( Figure 10) and relative abundance of Roseburia inulinivorans. This interrelation can partly explain the weight loss that accompanied the metformin administration and the downward trend in the considered bacterial taxa. Metagenomic analysis revealed a positive correlation between body weight and the phylum Bacteroidetes and genus Bacteroides.  Figure S5).

Correlation Analysis between Bacterial Taxa and Metabolic Symptoms Body Weight
Of great interest is the strong positive correlation between the body weight and both content ( Figure 10) and relative abundance of Roseburia inulinivorans. This interrelation can partly explain the weight loss that accompanied the metformin administration and the downward trend in the considered bacterial taxa. Metagenomic analysis revealed a positive correlation between body weight and the phylum Bacteroidetes and genus Bacteroides. As expected, we revealed a negative correlation between weight and relative abundance of Lactobacillus spp. As expected, we revealed a negative correlation between weight and relative abundance of Lactobacillus spp. Figure 10. Coefficients of correlation between the quantity and relative abundance/quantitative content of bacterial taxa in intestinal microbiome and body weight at the end of the experiment. Notes: total data for all animals, the results of the study are presented using p < 0.05. A search for correlations between the studied parameters was performed using Spearman's test. r.a.-relative abundance. The quantitative content of taxa was analyzed in other causes.
The correlation analysis revealed patterns for interrelationships between intestinal microbiome representatives and BGL only in the metagenome study. A significant positive correlation was found between BGL and Roseburia spp. representations ( Figure 11). Figure 11. Coefficients of correlation between relative abundances of bacterial taxa in intestinal microbiome and BGL at the end of the experiment. Notes: total data for all animals, the results of the study are presented, using p < 0.05. A search for correlations between the studied parameters was performed using Spearman's test.
The myocardial infarction size was directly related to the abundance of Roseburia spp. At the same time a negative correlation was detected between myocardial infarction size and bacteria belonging to the genus Akkermansia (r = −0.7), Streptococcus (r = −0.4), and Butiricimonas (r = −0.4) (Figure 12). Figure 10. Coefficients of correlation between the quantity and relative abundance/quantitative content of bacterial taxa in intestinal microbiome and body weight at the end of the experiment. Notes: total data for all animals, the results of the study are presented using p < 0.05. A search for correlations between the studied parameters was performed using Spearman's test. r.a.-relative abundance. The quantitative content of taxa was analyzed in other causes.
The correlation analysis revealed patterns for interrelationships between intestinal microbiome representatives and BGL only in the metagenome study. A significant positive correlation was found between BGL and Roseburia spp. representations ( Figure 11). Figure 11. Coefficients of correlation between relative abundances of bacterial taxa in intestinal microbiome and BGL at the end of the experiment. Notes: total data for all animals, the results of the study are presented, using p < 0.05. A search for correlations between the studied parameters was performed using Spearman's test.
The myocardial infarction size was directly related to the abundance of Roseburia spp. At the same time a negative correlation was detected between myocardial infarction size and bacteria belonging to the genus Akkermansia (r = −0.7), Streptococcus (r = −0.4), and Butiricimonas (r = −0.4) (Figure 12).

Figure 12.
Coefficients of correlation between relative abundances bacterial taxa in the intestinal microbiome and myocardial infarction size at the end of the experiment. Notes: total data for all animals, the results of the study are presented using p < 0.05. A search for correlations between the studied parameters was performed using Spearman's test.

Discussion
We, like many other researchers, turned to the problem of treating MS, which is manifested by disorders of carbohydrate and fat metabolism, obesity, and the inevitable occurrence of cardiovascular pathology. A feature of this study was the use of the IGT model, which made it possible to assess the risks of myocardial infarction and attempts to link changes in the body that occur against the background of the use of the multifunctional drug metformin with changes in the composition of the intestinal microbiota.
In this work, a comprehensive experimental study of IGT itself and the metformin effect on the organism and its intestinal microbiota in IGT conditions was carried out. Both our team [29] and other authors [6,30] used this model of HFD and nicotinamide + streptozotocin injection for IGT modeling. Importantly, the use of this model led to the development of lipid metabolism disorders and body weight gain thus making this model relevant for MS representation. Moreover, IGT modelling caused functional and structural changes in the myocardium during ischemia-reperfusion injury similar to those observed in type 2 diabetes mellitus, proving the hypothesis that even prediabetes provokes myocardial damage and increases cardiovascular risk [29]. Metabolism impairment started to develop on the 2nd-3rd day after nicotinamide + streptozotocin injection. It should be noted, that BGL in the IGT group was significantly higher than in the CRL group until the end of the experiment. Lipid metabolism disorders were confirmed in the IGT group by an increase in total cholesterol and AC at the end of the experiment. At the same time, functional and morphological disorders of the myocardium were revealed in ischemiareperfusion injury of the isolated heart of IGT rats. Similar changes in the myocardium have already been noted earlier when using models of antibiotic-associated dysbiosis [31] and obesity [32,33]. The changes that the authors attributed mainly to changes in the microbiota (excessive growth of opportunistic enterobacteria), an increase of C reactive protein, and pro-inflammatory cytokines level in blood serum were revealed [32]. So it was Figure 12. Coefficients of correlation between relative abundances bacterial taxa in the intestinal microbiome and myocardial infarction size at the end of the experiment. Notes: total data for all animals, the results of the study are presented using p < 0.05. A search for correlations between the studied parameters was performed using Spearman's test.

Discussion
We, like many other researchers, turned to the problem of treating MS, which is manifested by disorders of carbohydrate and fat metabolism, obesity, and the inevitable occurrence of cardiovascular pathology. A feature of this study was the use of the IGT model, which made it possible to assess the risks of myocardial infarction and attempts to link changes in the body that occur against the background of the use of the multifunctional drug metformin with changes in the composition of the intestinal microbiota.
In this work, a comprehensive experimental study of IGT itself and the metformin effect on the organism and its intestinal microbiota in IGT conditions was carried out. Both our team [29] and other authors [6,30] used this model of HFD and nicotinamide + streptozotocin injection for IGT modeling. Importantly, the use of this model led to the development of lipid metabolism disorders and body weight gain thus making this model relevant for MS representation. Moreover, IGT modelling caused functional and structural changes in the myocardium during ischemia-reperfusion injury similar to those observed in type 2 diabetes mellitus, proving the hypothesis that even prediabetes provokes myocardial damage and increases cardiovascular risk [29]. Metabolism impairment started to develop on the 2nd-3rd day after nicotinamide + streptozotocin injection. It should be noted, that BGL in the IGT group was significantly higher than in the CRL group until the end of the experiment. Lipid metabolism disorders were confirmed in the IGT group by an increase in total cholesterol and AC at the end of the experiment. At the same time, functional and morphological disorders of the myocardium were revealed in ischemia-reperfusion injury of the isolated heart of IGT rats. Similar changes in the myocardium have already been noted earlier when using models of antibiotic-associated dysbiosis [31] and obesity [32,33]. The changes that the authors attributed mainly to changes in the microbiota (excessive growth of opportunistic enterobacteria), an increase of C reactive protein, and pro-inflammatory cytokines level in blood serum were revealed [32]. So it was shown, that the changes in myocardium could be associated not only with the MS complications but also with changes in the microbiota caused by the administration of antibacterial drugs: ampicillin, metronidazole, azithromycin, tetracycline, leading to intestinal dysbiosis [31,32,34]. A significant contribution of the microbiota to the development of CVD was proved by the effectiveness of probiotic bacteria (enterococci, saccharomycetes, and lactobacilli) used for dysbiosis correction [31]. Microbiota normalization in these experimental models correlated with a decrease in inflammation.
The features of the intestinal microbiota revealed by us in experimental IGT were in many respects similar to those previously described in animals and humans with DM2 by other studies. However, in many ways, they differed from them. It could be both due to the peculiarity of experiment design and the study of the initial stages of development of more severe pathology (DM2).
Unlike other reports, describing the microbiota in DM2an increase [13,18,23,35] or decrease [21,22] in the representation of the phylum Bacteroidetes were not detected. At the same time, an increase in the representation of the Firmicutes phylum was revealed, as noted earlier [22]. We can suspect that such parameter as Bacteroidetes/Firmicutes ratio is not reliable enough for assessing pathognomonic shifts in MS, in particular in IGT. At the same time, the increase in the representation of the genus Bacteroides attracted our attention. This contradicts data received earlier, evidenced by decrees of these bacteria populations in MS [21,22]. It should be noted, that the decline in Bacteroides spp. representation in the intestinal metagenome could be associated with insulin resistance and dyslipidaemia [36]. In our experiment, such a correlation was not established.
The greatest surprise was an increase in the content of butyrate-producing Roseburia spp. and Akkermansia spp., the representation of which is usually reduced in DM2 [13,17,26,27]. This phenomenon might be connected with compensatory changes in the microbiota in response to the development of more serious pathology than IGT.
Our attention was drawn (to the increase in the quantitative content of opportunistic Acinetobacter spp., which can be considered as a potential cause of so-called "low-grade inflammation" development that often accompanies MS [37]. The signs of low-grade inflammation detected when using this model earlier [31] were not evaluated in this experiment. There was also a confirmed microbiota profile typical for DM2 [18,38] a decrease in the lactobacilli content. It was unexpected, because Lactobacillus spp. can normalize weight, lipid metabolism (decrease LDL/HDL and cholesterol),and glucose level [39].
We administrated metformin orally because not intravenous metformin administration has glucose-lowering properties [40]. This phenomenon suggests that the intestines might be one of the main target organs for metformin action [41]. Metformin realizes both its direct and pleotropic, including cardiovascular, effects by means of different mechanisms. In particular, it is known that metformin can modulate glucagon-like peptide-1 (GLP-1) receptor expression in the pancreas and increase plasma GLP-1 concentrations [42,43], possibly through enteroendocrine L-sells stimulation inside the intestines. The clinical and experimental trials revealed cardioprotective and anti-inflammatory properties of glucagonlike peptide-1 (GLP-1) analogs like liraglutide (Lira). demonstrated improved survival in lipopolysaccharide (LPS)-induced endotoxemia by inhibition of GLP-1 degradation [44].
Moreover, metformin can realize its actions in close connection with gut microbiota compound changes. In our study, the changes in intestinal microbiome coincided with the data obtained by other researchers [10,45] only with respect to the increase of Akkermansia spp. population and the decrease of members of the Enterobacteriaceae family.
We have shown in the IGT model for the first time a complete recovery of Firmicutes phylum and a sharp decrease in Acinetobacter sp. and a seemingly illogical representation of Bacteroides and Roseburia genera additionally.
The increase in the akkermansia population in the IGT model after metformin turned out to be important. It was unexpected for us. Positive metabolic and pleotropic effects of Akkermansia muciniphila were established earlier. A. muciniphila abundance is reduced in individuals with irritable bowel disease and other chronic diseases like obesity, as well as DM2 and prediabetes [38]. Thus, akkermansia growth is associated with body weight reduction, as well as hepatic steatosis, inflammation decrease, and even dyslipidemia reduction leading to atherosclerosis risk decrease. Akkermansia spp. also improves insulin sensitivity thus positively influencing type 2 diabetes mellitus pathogenesis [46]. Akkermansia muciniphila growth has been observed in obese people with healthier metabolic status and less prominent cardiovascular risks [47]. At the same time, the mechanisms of action of Akkermancia spp. are currently not entirely clear.
Akkermansia spp. are commensal Gram-negative bacteria, and are members of the phylum Verrucomicrobia [28]. A. muciniphila being Gram-negative bacteria, A. muciniphila also contains lipopolysaccharide endotoxin in its outer membrane of the cell wall, which differs from that of Escherichia coli [48]. Moreover, it was shown in a mouse model, that the presence of these bacteria in the intestine does not increase, but even reduces LPSendotoxinemia in animals with a long-term fat diet [45]. These bacteria break down mucins and convert them into short-chain fatty acids (SCFAs), including acetate and propionate. Acetate is used by other beneficial bacteria, such as Firmicutes, to produce butyrate, a vital energy source for the cells that line the intestinum [41]. The latter gastrointestinal mucins that are able to protect the epithelium from pathogens [49].
In our t study a negative correlation between the quantitative content of A. muciniphila and Acinetobacter spp. was revealed, which suggested the presence of mechanisms for suppressing the reproduction of opportunistic bacteria in the organism by these bacteria.
The quantitative content of A. muciniphila correlated directly with the number of bifidobacteria, lactobacilli, which are usually considered as obligatory intestinal microbiota members and active against pathogens. This was confirmed by the presence of a negative correlation between Lactobacillus spp. and Bifidobacretium spp. with Acinetobacter spp. quantity.
Indirectly, the beneficial role of A. muciniphila against the background of metformin therapy IGT was supported by a negative correlation with glucose levels and an increase of resistance to myocardial ischemia. A negative correlation with the level of serum glucose was also established for the Actinobacteria phylum (bifidobacteria and other), the growth of which could be stimulated by A. muciniphila.
The direct and indirect effect of akkermansia on the level of glucose has been proven phenomenologically and can be explained by the stimulation of the growth of bacteria utilizing glucose, influence on GLP-1 expression and regulation of the absorption process at the level of the intestinal mucous membrane.
If we consider the patterns of changes in the composition of the microbiota and the symptoms of MS (weight loss, glucose levels, and resistance to cardiac ischemia), then we should consider another bacterial genus. Roseburia, belonging to the Lachnospiraceae family, which demonstrated a negative correlation with the representation of Akkermansia spp. Unlike akkermansia, the decrease in the quantity and representation of these bacterium correlated with an increase in the populations of lactobacilli, representatives of the phylum Actinobacteria, Bifidobactrium spp. and Propionibacterium spp. as well as a decrease in the population of Fecalibacterium sp. They also had a negative correlation with the weight of the animals, the level of glucose, and the infarction size. Consequently, roseburia, in its influence, in fact, could be an antagonist of akkermansia and in the case of an increase of its population, could counteract the akkermansia action.
Roseburia spp. are commensal bacteria, members of the phylum Firmicutes, Lachnospiraceae family. They produce short-chain fatty acids, especially butyrate, affecting colonic motility, immunity maintenance, and anti-inflammatory properties [36,50]. It was shown, that the increased abundance of Roseburia is associated with weight loss and reduced glucose intolerance in mice [51]. Apparently, other bacteria, in particular, Akkermansia spp., KM decrease in the intestinal microbiome.
Separately, we should dwell on the previously discovered [29] cardioprotective effect of metformin confirmed in our study by the decrease of myocardial infarction size in risolated rat hearts subjected to global ischemia followed by reperfusion. The mechanisms responsible for metformin cardioprotective effect realization may include influence on lipid profile, anti-atherosclerotic action, vascular endothelium protection, and influence on vascular smooth cells [52]. In our study, for the first time, causal relationships between the presence of the cardioprotective effect of metformin with an increase in individual representatives of the microbiota, primarily akkermansia were identified. However, the nature of this positive influence is not yet clear and only opens the way for further research.
Despite the fact that type 2 diabetes mellitus is accompanied by a severe increase in cardio-vascular risk, leading to increased incidence of myocardial infarction and even cardiovascular death. On the other hand, data concerning IGT's influence on cardiovascular morbidity and mortality is even less presented in the scientific literature. Moreover, IGT's impact on myocardial damage size is even less studied. Moreover, recently much attention has been paid to the hypothesis that pleotropic cardioprotective effects of metformin in diabetes mellitus could be connected with metformin influence on microbiota compounds. On the other hand, a similar effect in IGT conditions remains the subject of investigation, which, in our opinion, forms the novelty of our study. Our study was the first to investigate correlation links between myocardial damage and microbiota compound in IGT conditions which helps to elucidate potential mechanisms of metformin cardioprotective effect.
The strong side of our study is the correlation analysis between microbiota features and MS symptoms, which opens up new areas of research already more complete, taking into account the dynamics of MS symptoms changes, the impact on the digestive, cardiovascular systems, immune and may be other (nerves and endocrine) systems.

Animals
The study was carried out in 38 male Specific pathogen-free (SPF) Wistar rats weighing 150-250 g (6-7 weeks old) obtained from the Animal Breeding Center, (Puschino, Russia).
The animals were maintained in fixed light mode, 12.00:12.00 h (light:dark), with no more than 5 animals per cage with free access to food and water. The temperature was maintained within the range of 22-25 • C, and the relative humidity-50-70%.
The duration of quarantine (acclimatization period) for all animals was 14 days. During the quarantine, every animal was examined daily. The color of the skin and visible mucous membranes, behavior, the motor activity, the presence of seizures, changes in respiratory movements, and tail position were assessed. The weighing was carried out upon arrival of the animals and during the quarantine period-at least once a week. Animals with deviations in weight, general condition, or behavior were not included in the experiment.

Ethics Approval
All experimental procedures were performed in accordance with the Guide for the Care and Use of Laboratory Animals (NIH publication No. 85-23, revised 1996)
After the first 4 weeks of HFD, a solution of nicotinamide 230 mg/kg was injected intraperitoneally as a pancreatic protector, followed by an intraperitoneal injection of a solution of streptozotocin 60 mg/kg as a pancreatic toxin 15 min later [53].

IGT Verification
To verify the sufficiency of IGT modelling, blood glucose levels (BGL) in IGT and IGT MET groups were measured on the 2nd and 3rd day after nicotinamide and-streptozotocin injection. Nicotinamide (Sigma-Aldrich, St. Louis, MO, USA) 230 mg/kg and streptozocin (Sigma-Aldrich, St. Louis, MO, USA) 60 mg/kg were injected intraperitoneally.
Glycemic values of 3.3 to 7.8 mmol/L were considered normal-these rats were excluded from the experiment. If BGL reached 11.1 mmol/L DM2 during two consequent days was diagnosed [54] and these animals also were not included in the future experiment. If BGL on the second and third day was in the range of 7.8-11.0 mmol/L an oral glucose tolerance test was performed. BGL was measured initially (fasting) and 15, 30, and 60 min after gastric administration of a 40% glucose solution 3 g/kg. We diagnosed IGT if BGL was in the interval of 7.8-11.0 at least at one time point (15, 30, or 60 min after glucose solution administration) and was not equal and did not exceed 11.1 mmol/L ever [55].

Study Design
After the acclimatization period, the following experimental groups were formed. Control (CRL) group of rats (n = 20) fed with standard chow for 16 weeks. The impaired glucose tolerance (IGT) group of rats (n = 6) was kept without treatment for 12 weeks after the induction of IGT; IGT MET group of rats (n = 12) was treated with metformin (8-weeks) for 4 weeks after the induction of IGT. Metformin powder (Metformin hydrochloride Sigma-Aldrich, St. Louis, MO, USA) was dissolved in distilled water and given per os by gastric tube 200 mg/kg of body weight.
The animals from IGT and IGT MET groups were receiving HFD throughout the entire experiment. The study design is presented in Table 1. Once every two days during the entire experiment, the animals were weighed, and the weight of the chow consumed in 2 days was determined.
Glucose measurement was performed using StatStrip Xpress Glucose/Ketone Meter (Nova Biomedical, Waltham, MA, USA). The glucose assessment in all groups was performed at the end of the 4th week, 8th week, and once a week during the weeks 9-16 at the same time of time (not after fasting).

Isolated Heart Perfusion According to Langendorff
After reaching the surgical stage of anesthesia (tiletamine hydrochloride 30 mg/kg and zolazepam hydrochloride 30 mg/kg IM, then xylazine hydrochloride 6 mg/kg IM), thoracolaparotomy was performed to remove the heart. Then the heart was connected to a modified Langendorff apparatus (Supplement Materials, Figure S6). Retrograde perfusion was performed through the ascending aorta with modified Krebs-Henseleit buffer solution, consisting of the following (in mmol/L): NaCl 118.5; KCl 4.7; NaHCO 3 25.0; KH 2 PO 4 1.2; MgSO 4 1.2; glucose 11.0; CaCl 2 1.5) at a constant pressure of 80 mm Hg and at +37 • C [57]. After the end of the stabilization period (5 min), the following parameters were registered: left ventricular systolic pressure (LVSP), left ventricular end-diastolic pressure (LVEDP) that were evaluated iso volumetrically using a nonelastic polyethylene balloon introduced into the left ventricle, left ventricular developed pressure (LVDP) was calculated as the difference between LVSP and LVEDP. Coronary flow rate (CFR) was evaluated by measuring the time for the collection of perfusate outflow.
Global 30-min normothermic myocardial ischemia followed by 90-min reperfusion was induced by the reversible shutdown of perfusion. During the ischemic period, intraleft ventricular pressure (LVP) was recorded every 5 min in order to assess the severity of ischemic contracture. Ischemic contracture was determined as at least a threefold increase in LVP at any time during the ischemic period compared to the LVP after five minutes of ischemia. During the reperfusion period, functional parameters were recorded every 15 min, as done previously [29,58].

Myocardium Infarction Size Measurement
At the end of reperfusion, the volume of the irreversibly damaged myocardium was measured using the method of histochemical staining of transverse heart sections with 1% triphenyltetrazolium chloride solution. The sections were incubated in the indicated solution for 15 min and the viable myocardium was stained bright red, while areas of irreversibly damaged myocardium remained unstained. The ischemia size was recorded according to the method described earlier [29]. Infarction size was expressed as a percentage of total ventricular area minus the cavities.

Metagenome (16S rRNA) Study
Fecal samples frozen on the day of the material's collection were used for the metagenome analysis. Libraries of hypervariable regions, V3 and V4, of the 16S rRNA gene were analyzed on MiSeq (Illumina, San Diego, CA, USA). DNA was isolated from feces using the kit Express-DNA-Bio (Alkor bio, St Petersburg, Russia). The standard method recommended by Illumina based on employing two rounds of PCR was used to prepare the libraries. The first round was the amplification of a 16S rRNA gene fragment, adding to the target fragment of the adapter nucleotide sequences included in the PCR primers [60].
Fastqc was used to assess the quality of raw reads. CD-HIT-OTU-Miseq was used for OTU retrieval with the following parameters: lengths of high-quality parts of R1 and R2 reads of 200 and 180 bp respectively, 97% read similarity for clustering cutoff, and 0.00001 for abundance cutoff. OTUs were annotated using Greengenes database version 13.5 [61]. CD-HIT-OTU-Miseq allows you to select OTU by the sequence of end sections without the need to merge paired sequences. R package phyloseq was used to assess class abundances and beta diversity.

Statistical Analysis
The normality of the data distribution was determined using the Kolmogorov-Smirnov criterion. The nonparametric criteria were used on this basis. Statistic data processing was performed using the software package IBM SPSS Statistics-22 (IBM, Armonk, NY, USA) and Statistica-10 (StatSoft Inc., Tulsa, OK, USA). The presence of statistically significant differences among groups was determined using the Mann-Whitney U-criterion, adjusted for multiple comparisons by the Benjamini-Hochberg method. Wilcoxon's t-test was also used for paired samples. Comparative analysis was conducted using the a posteriori test honestly significant difference (HSD) for unequal N in the Statistica-8 software (StatSoft, Tulsa, OK, USA). Search for correlations between the studied parameters was performed using Spearman's test using the software package Statistica 10.0 (StatSoft, Tulsa, OK, USA). Differences in p < 0.05 were considered significant.

Conclusions
The use of metformin in experimental IGT models led to several positive effects: weight loss, a decrease of glucose and total cholesterol levels in the blood, increased resistance to myocardial ischemia (decreased MI size), and was accompanied by the changes in the intestinal microbiota. The most important changes in the intestinal microbiome were the following: an increase in Akkermansia spp. and a decrease in Roseburia spp. quantity content and relative abundance. These findings allow suggesting, that the positive influence of metformin on the condition of patients with MS is related to the changes in microbiota, especially in Akkermansia and Roseburia genera populations. This should be taken into account when determining indications for the administration of probiotics based on these bacteria for the treatment of MS and prevention of its cardiovascular complications.

Conflicts of Interest:
The authors declare no conflict of interest.