Sleeve Gastrectomy Suppresses Hepatic Glucose Production and Increases Hepatic Insulin Clearance Independent of Weight Loss

Hepatic glucose production is an essential component of the regulation of circulating glucose levels. The liver produces glucose by glycogenolysis and gluconeogenesis according to substrate availability, neuronal cues, hormonal signaling, and cellular redox state, while glucose production is normally suppressed by insulin signaling (1–4). The liver degrades a fraction of the insulin passing through it by internalizing insulin bound to the insulin receptor complex. Insulin is then degraded intracellularly, and the unbound receptor is transported back to the surface. Insulin receptor availability, insulin signaling, and clearance are tightly linked (5–8).

Hepatic steatosis is characterized by the accumulation of lipid droplets in hepatocytes. It is associated with insulin resistance, hyperinsulinemia, and reduced hepatic insulin clearance and may lead to the development of type 2 diabetes (T2D) (5,9–12). The mechanisms coupling hepatic steatosis and insulin resistance are under intense study (10,13), but relatively less is known about the factors linking hepatic steatosis and the reduction in insulin clearance (5–7).

Obesity drives hepatic steatosis and T2D. Bariatric operations that lead to weight loss often alleviate metabolic diseases, such as hepatic steatosis and T2D (14–18), and increase hepatic insulin sensitivity and clearance (19–25). Because bariatric surgeries induce rapid weight loss in the months following surgery and a change in diet, it is difficult to determine to what extent weight loss, change in diet, or the new anatomy contribute to the improvement in hepatic insulin sensitivity and clearance in patients.

We and others have shown in clinical and animal studies that bariatric surgery affects glycemia and hepatic metabolism also independent of weight loss by multiple pathways such as an increase in Glp1 signaling, changes in the gut microbiome, and activation of hepatic peroxisome proliferator-activated receptor-α (PPARα) signaling (26–34). We hypothesized that bariatric surgery reduces hepatic glucose production and increases hepatic insulin clearance also independent of weight loss and change of diet. By measuring glucose and insulin levels in blood vessels feeding and draining the liver, we show here that vertical sleeve gastrectomy (VSG) increases hepatic insulin clearance in obese and lean mice compared with weight-matched sham-operated mice (WMS) and that hepatic glucose production and hepatic steatosis are reduced in VSG-operated mice compared with WMS fed a calorie-rich diet. Together, these results provide evidence that VSG affects hepatic insulin and glucose metabolism also independent of diet and weight loss.

All studies were approved by the Hebrew University of Jerusalem Institutional Animal Care and Use Committee.

Male 60-day-old C57Bl/6OlaHsd mice were fed ad libitum a high-fat high-sucrose (HFHS) diet (Envigo Teklad diets TD.08811) for 115 days before surgery and 40 days after surgery and then euthanized.

Male 70-day-old C57Bl/6OlaHsd mice were fed ad libitum a normal chow diet. Mice were operated at the age of 10 weeks and euthanized 21 days after surgery. Mice were fed ad libitum after surgery.

Male Hsd:ICR mice fed ad libitum normal chow were operated at the age of 10 weeks and remained on ad libitum normal chow diet. Mice were euthanized 12 days after surgery.

Male C57BL/6JOlaHsd mice fed ad libitum on normal chow were operated at the age of 14 weeks and remained on ad libitum normal chow diet. Mice were euthanized 14 days after surgery.

Surgical procedures were described previously (31). Briefly, mice were fasted overnight before surgery. Mice were anesthetized using isoflurane, shaved, and locally disinfected. A 1.5-cm laparotomy was performed, and a large LIGACLIP (Ethicon LT400) was placed parallel to the small curvature of the stomach to form a sleeve from the stomach. The excess stomach was cut, and ∼75% of the stomach was removed. Muscle and skin were closed using Vicryl sutures. The sham procedure was the same except for placing the clip.

All mice were treated daily with meloxicam, 2 mg/kg body wt, 3 days after surgery. The mice were fasted the night before surgery until the day after surgery, after which the presurgical diet was reintroduced. We limited the amount of food provided to weight-matched sham (WMS)-operated mice to 3.5 g per mouse per day to facilitate weight matching until weight matching was achieved. Additional food, up to 5 g per mouse per day, was provided as the VSG mice gained weight. Food was provided at the onset of the dark cycle and removed in the morning in the VSG- and WMS-operated mice to limit the effects of food restriction on circadian rhythms in both groups. VSG-operated mice received ad libitum HFHS diet during the night cycle.

The Hsd:ICR and c57/Bl6 lean VSG mice both underwent the same surgery as described but were provided ad libitum normal chow before and 1 day after the surgery until the end of the experiment.

All mice were euthanized following 6 h of fasting at ∼1–2 p.m.

Body composition was determined by EchoMRI-100H (Echo Medical Systems).

Mice were monitored by the Promethion High-Definition Behavioral Phenotyping System (Sable Instruments) over 24 h after 24 h of habituation, as described previously (35).

Mice were fasted for 6 h and were given d-glucose, 2 g/kg body weight, by 18-gauge gavage needle (Fine Science Tools 18061-50). Blood glucose was measured using Roche Accu-check Performa from the tail. The area under the curve was calculated after subtracting the fasting glucose level (36).

Mice were anesthetized using a low dose of ketamine HCl:xylazine (100 mg:8 mg/kg body wt). Glucose was first measured from the tail. The mice underwent laparotomy, and the inferior vena cava (IVC) was exposed. A surgical endoscopic LIGACLIP 12-L (Ethicon) was placed on the IVC superior to the renal vein. Blood was first collected 30 s later from the portal vein, then the suprahepatic IVC, and finally from the aorta. Blood glucose levels were measured in each location. Finally, glucose was measured from the tail to assess anesthesia or procedure-induced changes in blood glucose.

The liver was fixed overnight in formalin. Hematoxylin–eosin staining were performed as previously described (31). Histological quantification was performed blindly. A hepatic microsteatosis score was defined as 0 for no microsteatosis and 3 for abundant microsteatosis.

Ultrasensitive mouse insulin ELISA (Crystal Chem 90082) was used according to the manufacturer’s instructions.

A mouse glucagon ELISA kit (Crystal Chem 81518) was used according to the manufacturer’s instructions.

We assume linear clearance of hormones within the sinusoids from the periportal to the pericentral pole and analyze the steady-state,

where v is the velocity in the liver lobule, H is hormone, α_H is the clearance rate, and L is the distance from the periportal to pericentral poles.

Data were normalized to the average levels of ln(H_IVC /H_in) in sham-operated mice. We used α_H = 0 in cases where H_IVC > H_in due to measurements errors.

A 30-mg liver biopsy specimen was taken during the sacrifice and directly frozen in liquid nitrogen. Tissue was minced and homogenized using TRI reagent (Sigma-Aldrich T9424) according to the protocol, and ethanol was precipitated using ammonium acetate for additional purification. Then, 1 μg of RNA was reverse transcribed using an iScript cDNA synthesis kit (Bio-Rad cat. no. 1708890). cDNA was diluted and quantitative (q)PCR was done using iTaq Universal SYBR Green Supermix (Bio-Rad cat. no. 1725124). The ΔΔCt method was applied to calculate fold-change expression. qPCR results were normalized to the expression of mTbp and Hprt. The primers for qPCR analysis are provided in Supplementary Table 1.

Data are shown as box-and-whisker plots or bar plots. In bar plots, error bars denote the SE. One- or two-way repeated-measures ANOVA with the Tukey honestly significant difference (HSD) or Student t test were performed to determine statistical significance as indicated in the text. The Bonferroni correction was applied when multiple tests were performed on the same data. Benjamini-Hochberg correction was used to correct for the multiple genes tested via qPCR analysis.

The data sets generated during the current study are available from the corresponding author upon reasonable request.

Male mice, 9 weeks old, were fed the HFHS diet for 115 days and were then randomly assigned to VSG or WMS groups (research design and methods). Two VSG mice did not recover well from surgery and were removed from the study. The remaining mice were followed for 40 days after surgery. There were no significant differences in weight before surgery or once weight matching was achieved on day 20 after surgery (Fig. 1A). At that time, mice had a similar fraction of fat mass (Fig. 1B). Respirometry analysis showed no difference in the respiratory quotient, but a small increase in total energy expenditure was observed in WMS-operated mice that was associated with increased wheel running (Fig. 1C and D and Supplementary Fig. 1). Nonfasting glucose levels did not differ between the groups (Fig. 1E). An oral glucose tolerance test showed that VSG-operated mice experience a sharp increase in glucose levels 15 min after gavage, followed by a reduction in glucose, with no difference in the area under the curve (Fig. 1F and G). There was no difference in fasting blood glucose after 6-h fasting on day 40 after surgery (Fig. 1H).

We assessed hepatic glucose production directly in fasting animals by measuring glucose levels in the blood vessels feeding and draining the liver. Because the hepatic veins draining the liver in mice are very small, we ligated the IVC above the renal junction in an anesthetized mouse (Fig. 2A). We waited 30 s and then measured blood glucose levels in the suprahepatic IVC below the diaphragm (Fig. 2B). At this time, blood in the IVC is predominately supplied from the liver. We then exposed the portal vein and measured blood glucose levels (Fig. 2C) and finally measured glucose in the descending abdominal aorta (Fig. 2D). This terminal procedure lasted 2 min, and the blood collection strategy is summarized in Fig. 2E. Ketamine-xylazine anesthesia can increase blood glucose levels (37). To control for this effect, we dosed the anesthesia such that glycemia would be stable throughout blood collection and compared blood glucose levels measured from the tail before laparotomy and after collection from the aorta in VSG-, WMS-, and lean sham (LS)-operated mice that were fed ad libitum with normal chow for 40 days after surgery (Fig. 2F). Mice in which this difference was >40 mg/dL were excluded from the analysis.

Using the strategy described above, we measured glucose levels in the hepatic IVC, portal vein, and aorta in VSG, WMS, and LS-operated mice. Glucose levels in the aorta, portal vein, hepatic IVC, and the tail tip were lower in LS compared with VSG and WMS (Fig. 3A–D). There was no difference in blood glucose levels between VSG and WMS in the aorta, portal vein, or tail tip (Fig. 3A, B, and D). In contrast, the suprahepatic IVC glucose level of WMS mice was elevated compared with VSG (Fig. 3C). Blood glucose levels of LS-operated mice were lower in all blood vessels than in VSG- and WMS-operated mice. We assessed the hepatic glucose production as the difference in blood glucose entering and leaving the liver by calculating:

ΔGlucose_Liver was significantly lower by 55 mg/dL in the VSG than in the WMS group (Fig. 3E). There was no difference in ΔGlucose_Liver between LS and the WMS or between LS and VSG groups (Fig. 3E). Assuming that the contribution of the portal vein to total hepatic input was 80% or 70% did not affect the results qualitatively.

We were able to collect sufficient plasma from the suprahepatic IVC, aorta, and portal vein to measure insulin levels in VSG- and WMS-operated mice. The insulin levels of the two groups were similar in the suprahepatic IVC and the aorta but were higher in the portal vein of VSG- compared with WMS-operated mice (Fig. 4A–C). We defined the hepatic fractional insulin clearance (FIC) by:

Assuming again that 75% of the blood feeding the liver is supplied by the portal vein. Using this calculation, we found that nearly 45% of the insulin is removed in the livers of WMS mice. Strikingly, the livers of VSG-operated mice removed ∼70% of the incoming insulin, pointing to an increase in hepatic insulin clearance in the livers of VSG mice (Fig. 4C). We modeled hepatic insulin clearance (research design and methods) and estimate that the insulin clearance rate is twofold higher in VSG compared with WMS-operated mice (Fig. 4E). qPCR analysis of genes associated with insulin clearance showed a significant increase in the transcript levels of Ide1, Ceacam1, and Prcke. There was a significant increase in the mRNA levels of exon 19 and exon 11 of Insr. Exon 19 is transcribed in both isoforms of the insulin receptor, and exon 11 is transcribed only in isoform B of the insulin receptor. There was no difference in the levels of Irs1 and Igf1r in WMS and VSG mice (Fig. 4F).

We analyzed livers of VSG and WMS sacrificed 40 days after surgery. Even though the weight was equivalent in both groups, hepatic steatosis was less prevalent in VSG- compared with WMS-operated mice, with a reduction in both large lipid droplets and degree of microsteatosis (Fig. 5A–D). We did not detect inflammation or fibrosis in both groups.

Greater fractional hepatic insulin uptake following VSG can be attributed to the improvement of hepatic steatosis, to a response to VSG, or restricted feeding. We performed VSG or sham surgery on normoglycemic CD1 male mice consuming a normal chow diet. LS- and lean VSG (LVSG)-operated mice had similar weight and glucose levels 12 days after surgery (Fig. 6A and B). We collected blood from the suprahepatic IVC, descending aorta, and portal vein as before.

Insulin levels in the portal vein of LVSG mice were higher than those of LS-operated mice, while insulin levels in the suprahepatic IVC and aorta were not different (Fig. 6C–E). Correspondingly, hepatic FIC was higher in the LVSG-operated group (Fig. 6F and G). Insulin levels and clearance were low, as expected during fasting in lean mice. Glucagon levels were significantly higher in the portal vein, aorta, and suprahepatic IVC of LVSG-operated mice. Yet, there was no difference in hepatic fractional glucagon uptake or normalized clearance rate between the two experimental groups (Fig. 6H–L).

Finally, we repeated the experiment on 14-week-old C57Bl6 male mice fed a normal chow diet. In this case, LVSG mice weighed 2 g less than LS mice 2 weeks after surgery, with no significant difference in fasting glucose levels. As with the CD1 mice and HFHS-fed C57Bl6 mice, we detected an increase in portal insulin levels and an increase in insulin clearance. Glucagon levels were higher in the portal vein, aorta, and suprahepatic IVC of LVSG mice, with no difference in hepatic glucagon clearance.

VSG was shown to increase hepatic insulin sensitivity, reduce hepatic glucose production, and increase insulin clearance in patients (24,25). These results may be explained in part by weight loss that follows surgery, leading to improvement in hepatic steatosis and reduction in the level of obesity-associated systemic inflammation. Insulin clearance was also shown to be affected by diet (20). Here we provide evidence that changes in hepatic steatosis, insulin clearance, and hepatic glucose production stem also from VSG per se, while diet and weight loss are controlled (12,38,39).

We have previously shown using hyperinsulinemic-euglycemic clamps that glucose production is repressed after VSG in obese db/db mice even though these mice remain obese after surgery (31). Our findings here support and expand these findings using different methods and models. We show using a dietary obesity model that even after controlling for weight loss, VSG is more effective in reducing hepatic glucose production and hepatic steatosis. Moreover, portal insulin levels and hepatic insulin clearance are higher following VSG. These results are obtained directly by measuring glucose and insulin levels entering and leaving the liver under fasting conditions and not during a hyperinsulinemic clamp. Notably, fasting glucose levels and area under the curve following glucose challenge are similar following VSG or food restriction (40), but the dynamics of glucose and underlying regulation are different. We further show that an increase in portal insulin and hepatic insulin clearance is observed in two strains of lean mice. Together, these results show that VSG itself affects insulin metabolism and hepatic physiology, independent of weight loss, in obese and lean animals, under different diets, and in several mouse strains.

Our results are in line with previous rodent studies that have shown that VSG has weight loss-independent effects that contribute to glucose homeostasis. VSG leads to enhanced insulin and glucagon secretion in vivo and ex vivo (41), increases secretion of gastrointestinal hormones, and affects the gut microbiome and levels of bile acids and other effectors of glucose metabolism (42–45). This study contributes previous knowledge by taking an anatomical approach with a focus on the liver and by analyzing obese and lean mice to directly measure surgically induced changes in glucose and insulin metabolism in vivo.

WMS mice fed an HFHS diet had higher hepatic glucose production than VSG-operated mice. There was no difference in glucose levels in the aorta, portal vein, or tail tip between VSG- and WMS-operated mice, suggesting an increase in glucose utilization in WMS mice, which may correspond to the higher total energy expenditure in WMS-operated mice and a trend toward more wheel-running activity. We did not find evidence for induction of intestinal gluconeogenesis in VSG mice, which was reported in Roux-en-Y gastric bypass surgery in rodents (46,47).

Fasting insulin levels are not different between patients who had bariatric surgery or consumed a drastic diet (48). We have obtained similar results in both normal chow- and HFHS-fed mice models. However, we did detect an increase in insulin levels in the portal vein in both lean models and the HFHS mouse model after VSG. Fasting insulin levels were higher in HFHS-fed mice compared with lean mice of the same strain fed normal chow in both surgical groups as HFHS-fed mice display some hepatic steatosis and have increased adiposity compared with chow-fed mice.

The total and fractional insulin clearance and normalized clearance rate were increased in mice fed the HFHS diet that underwent VSG compared with WMS-operated mice. This increase was accompanied by a decrease in hepatic steatosis and hepatic glucose production. We also observed an increase in the mRNA levels of genes associated with insulin clearance: Insr, Ceacam1, Ide, and Prkce. In a previous study, the hepatic expression of the A isoform of Insr was reduced after gastric bypass surgery in patients with T2D (49). In our data, the total and the B isoform were elevated to the same extent after surgery. These differences may be attributed to differences in surgery type, species, and weight matching, which was not possible in the human study. Ide is an important enzyme for the intracellular transport of the insulin-Insr complex, degradation of insulin, and recycling of Insr (50,51). This process is regulated by phosphorylation of Ceacam1 (8). Inhibition of hepatic protein kinase Cε (PKCε), encoded by Prkce, was shown to reduce insulin clearance, and upregulation of this kinase may contribute to the observed enhanced insulin clearance (52). An elevation in Prkce, Ceacam1, Insr, and Ide levels may explain the observed increase in insulin clearance (5,49,53,54). Histologically, there were fewer lipids droplets in the liver of VSG- compared with WMS-operated mice, despite the same weight and glycemia.

Improvement in insulin clearance can be attributed to an improvement in hepatic steatosis following VSG or a hepatic adaptation to restricted feeding imposed by weight matching. Insulin clearance, however, was increased in lean VSG-operated mice compared with LS-operated mice fed ad libitum, indicating that the new anatomy imposed by VSG may lead to an increase in hepatic insulin clearance. The increase in portal insulin levels and possibly higher insulin signaling, as reflected by lower glucose production in HFHS, mice may drive the increase in hepatic insulin clearance, which results in comparable levels of insulin outside the liver in VSG- and WMS-operated mice. We conclude that surgery, diet, and a reduction in hepatic steatosis all contribute to an increase in hepatic insulin clearance.

Glucagon levels were higher in all three blood pools in lean mice that underwent VSG. An elevation in peripheral plasma glucagon levels following VSG was reported in patients and rodents (23,41,44), yet the mechanism underlying this paradoxical increase is not understood. Insulin levels were increased only in the portal blood of lean VSG-operated mice, and accordingly, insulin but not glucagon clearance was elevated following VSG. VSG therefore leads to an increase in portal insulin and glucagon and exposes the liver to a different endocrine environment than sham-operated mice.

An important limitation of our study is that blood collection was performed during anesthesia, which affects glycemia (37), and the procedure itself can activate neuronal and hormonal signaling that increase hepatic glycemic control. While we verified that glucose levels were stable during blood collection, they were high in all groups of mice and in all blood pools.

Because VSG surgery is accompanied by a drastic weight loss and change of diet in patients, it is difficult to decouple the effects of surgery from those of weight loss and changes in hepatic steatosis. In this study, we used a mouse model of VSG to show directly that VSG reduces hepatic steatosis and glucose production and increases insulin clearance compared with weight and diet-matched sham-operated mice. We therefore propose that the new anatomy, which affects hormonal signaling, has a positive effect on hepatic metabolism beyond weight loss. Understanding the mechanisms underlying metabolic improvements after bariatric surgery may offer new modalities to treat obese and lean patients suffering from hepatic steatosis and T2D.

Funding. This study was supported by H2020 European Research Council Starting Grant (803526) awarded to D.B.-Z. and an Israel Science Foundation (ISF) grant (158/18) to J.T. D.B.-Z. is a Zuckerman STEM faculty fellow.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. R.B.-H.S., A.A.-K., B.M., A.P., H.I., A.B., Y.A., L.H., M.B., A.H., S.A., I.M., and D.B.-Z. performed the experiments. R.B.-H.S., B.M., A.P., J.T., and D.B.-Z. analyzed the data. R.B.-H.S., R.G., and D.B.-Z. conceived the study. R.B.-H.S. and D.B.-Z. wrote the manuscript. J.T. edited the manuscript. D.B.-Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented as an abstract at the 57th Annual Meeting of the European Association for the Study of Diabetes, 27 September–1 October 2021.