Gut insulin action protects from hepatocarcinogenesis in diabetic mice comorbid with nonalcoholic steatohepatitis

Diabetes is known to increase the risk of nonalcoholic steatohepatitis (NASH) and hepatocellular carcinoma (HCC). Here we treat male STAM (STelic Animal Model) mice, which develop diabetes, NASH and HCC associated with dysbiosis upon low-dose streptozotocin and high-fat diet (HFD), with insulin or phlorizin. Although both treatments ameliorate hyperglycemia and NASH, insulin treatment alone lead to suppression of HCC accompanied by improvement of dysbiosis and restoration of antimicrobial peptide production. There are some similarities in changes of microflora from insulin-treated patients comorbid with diabetes and NASH. Insulin treatment, however, fails to suppress HCC in the male STAM mice lacking insulin receptor specifically in intestinal epithelial cells (ieIRKO), which show dysbiosis and impaired gut barrier function. Furthermore, male ieIRKO mice are prone to develop HCC merely on HFD. These data suggest that impaired gut insulin signaling increases the risk of HCC, which can be countered by restoration of insulin action in diabetes.

The exact P values are provided in Supplementary Data 3.Primary component analysis was conducted by the software invented by human metabolome technologies.The liver tissue samples from mice in their 9 weeks of age were under analysis.SHFD-CON, for control STAM mice.SHFD-INS, for STAM mice with insulin treatment.SHFD-PHZ, for STAM mice with phlorizin treatment.The following groups were also included in the analysis along with drug treatment to evaluate the influence of feeding and streptozotocin (STZ) treatment in this model mice.Normal, for mice fed with normal chow.HFD-nonSTZ, for mice fed with high fat diet (HFD) and without treatment of STZ on birth.STZ-nonHFD, for mice fed with normal chow and treated with STZ on birth.Each group includes 3 samples.The exact P values are provided in Supplementary Data 3.

4 4
Characteristics relating to hepatic metabolomic and fecal metagenomic changes in insulintreated STAM mice.Primary component analysis from LC-MS analysis was conducted by the software invented by human metabolome technologies.LC-MS analysis was conducted from tumor or non-tumor liver sample of 20-week-old STAM mice.STAM-NON-Tumor, for tumor samples from STAM mice without treatment.STAM-INS-Tumor, for tumor samples from insulin-treated STAM mice.STAM-NON-Liver, for non-tumor liver tissue from STAM mice without treatment.STAM-INS-Liver, for non-tumor liver tissue from insulin-treated STAM mice.Normal, for normal-chowfed mice without STZ-treatment on birth.(a) Each group includes 3 samples.The ratio of deoxycholic acid (DCA) to cholic acid (CA) in liver (b).Values of the data are expressed as mean ± SEM (b).Primary component analysis (PCoA) of 16S metagenomics in 9-week-old STAM mice (c).DIO mice n = 4, non-treated STAM mice (STAM-NON) n = 5, insulin-treated STAM mice (STAM-INS) n = 4, and PHZ-treated STAM mice (STAM-PHZ) n = 4 at 9 weeks of age.Values of the data are expressed as mean ± SEM.The exact P values are provided in Supplementary Data 3. Physical barrier dysfunction in STAM mice.Lipopolysaccharide (LPS) concentration in plasma (a).DIO mice (n = 7), STAM mice (n = 21).The concentration of fluorescein isothiocyanatedextran -4kDa (FD-4) in plasma.The comparison was conducted between normal mice (n = 4) and diet-induced-obese (DIO) mice (n = 4) (b), between DIO mice (n = 4) and STAM mice (n = 8) (c), and between STAM mice without treatment (n = 4) and insulin-treated STAM mice (n = 4) (d).Values of the data are expressed as mean ± SEM (a, b, c, d).* P < 0.05, 2-sided Mann-Whitney's u test (a), and 2-sided unpaired t test (b, c, d).The exact P values are provided in Supplementary Data 3. to Figure516S metagenomic signature alteration by insulin use in NASH patients with diabetes Volcano plot of proportion to total reads of fold change by insulin treatment and P value by ANCOM-BC2 test between two groups (a).Species shown here are selected by average proportion over 1% in total reads in the group without insulin treatment.(See also Supplementary Table7.)Proportion of Bacteroides massiliensis (AY126616) (b) in non-insulin-treated (no insulin use, n = 15) and insulin-treated NASH patients (insulin use, n = 5).The logarithm 10 of proportion of Bacteroides massiliensis (AY126616) (c) in non-insulin-treated (no insulin use, n = 13) and insulin-treated NASH patients (insulin use, n = 5). 2 samples in non-insulin-treated group were excluded because of no detection of this bacterium by this method.Values of the data are expressed as mean ± SEM (b, c).Shannon index (d), the number of identified species (e), and loading data from primary component analysis (f) in non-insulin-treated (no insulin use, n = 15) and insulin-treated NASH patients (insulin use, n = 5), compared by 2-sided unpaired t test.Values of the data are expressed as mean ± SEM (b, c).Source data are provided as a Supplementary Data 3.

Supplementary Figure 3 Related to Figure 2
Characteristics relating to hepatic steatosis in insulin-treated STAM mice and PHZ-treated STAM mice.Body weight (a), liver weight (b), hepatic triglyceride (c), hepatic cholesterol (d) at 9 weeks of age.Relative mRNA expression level of Fasn, Acc (e).

, c, d, e, f, g).
Western blotting by eWAT lysate of STAM mice at 9 weeks of age (h).The experiments were repeated independently at least twice.Anti-pHSL, anti-tHSL, anti-β-actin antibodies were used as primary antibodies to detect each molecule.Relative expression level of Ucp1 in eWAT (i), and relative expression of Cidec,

Table 1 ,
Related to Figure4Abundance of metabolites determined by LC-MS in liver tissue of normal mice, non-treated STAM mice (STAM-NON), insulin-treated STAM mice (STAM-INS) at 20 weeks of age."S.E." means standard error.$; "Undetermined" means below the detection limit.#; "N.D." means the value was not calculated because one or more values of samples were below the detection limit.