The experimental protocol was approved by the Federal University of Santa Catarina Committee for Ethics in Animal Experimentation (identification no 7696221019) in accordance with the Brazilian National Council for Animal Experimentation Control (CONCEA) and the Guide for the Care and Use of Laboratory Animals (National Research Council, 9th edition).

All mice were maintained in appropriate cages at the Physiological Science Department in a temperature (22–24 °C) and humidity-controlled environment and kept on a 12-h light-dark cycle (lights on 06:00–18:00 h) with access to food (Nuvilab CR1, Brazil) and water (filtered tap) ad libitum. To test our hypothesis we used tissue-specific and temporal-conditional knockout male mice for HNF4α (HNF4αloxP/loxP;Ins1Cre+), generated by crossing the lineage carrying the CRE-recombinase enzyme in β-cells (Ins2.Cre+- #008122, Jackson Laboratory, USA) and the lineage that possesses the gene for the HNF4α FT flanked by the LoxP sequence (HNF4αloxP/loxP - #004665, Jackson Laboratory, USA). The KO mice were homozygous for the loxP-flanked allele and heterozygous for the RIP-CreER, while the control (CTL) mice were homozygous for the loxP-flanked allele but did not carry the RIP-CreER. To determine the genotype of the mice, RT-PCR was performed according to the Jackson Laboratory instructions. To activate the CRE-recombinase enzyme, mice aged 35-40 days received daily intraperitoneal injections of 100 µl of a 20 mg/mL tamoxifen solution (#T5648, Sigma-Aldrich, USA) for five consecutive days (9).

To confirm the CRE-recombinase activation, we evaluated the fasted glycaemia 15 days after the last tamoxifen injection. After that, mice were divided into three groups: Control (CTL) fed with a chow diet, Control fed with HFD, and Knockout fed with HFD. The HFD administration lasted for 20 weeks, and the compositions of the diets are shown in Table 1 .

After euthanasia, the pancreas was injected with Hank’s buffer containing 0.9 mg/ml of type V collagenase. Once removed, the inflated pancreas was incubated for 17 minutes at 37°C and the islet was hand-picked one-by-one. To analyze the HNF4α mRNA, 200 islet were homogenized with Tryzol™ (Invitrogen, MA, USA) and the mRNA was isolated following the manufacture’s guide. After the extraction, 200 ng of mRNA were converted into cDNA with the HighCapacity™ cDNA Transcription Kit (Applied Biosystems, CA, USA). The qPCR was performed with the Fast SYBR™ Green Master Mix (AppliedBiosystem, MA, USA) and the GAPDH gene was used as endogenous control. The primers used are GAPDH Forward: CACATTGGGGGTAGGAACAC; GAPDH Reverse: GCCAAAAGGGTCATCATCTC; HNF4α Forward: GCAAGTGAGCCTGGAGGATT; HNF4α Reverse: TGTCCATTGCTGAGGTGAGA.

Ten-hours fasted mice received an intraperitoneal administration of 1 g/kg of glucose dissolved in saline solution (0.9% NaCl wt/vol). Blood glucose levels were measured before (0 min) and 15, 30, 60, and 120 minutes after glucose administration. Glucose levels were evaluated in blood drops collected from the tail tip and measured by Accu-Chek Performa II glucometer (Roche^®).

One-hour fasted mice received an intraperitoneal administration of 1 IU/kg insulin dissolved in saline solution (0.9% NaCl wt/vol). Blood glucose levels were measured before (0 min) and 5, 15, 30, 45 and 60 minutes after the insulin administration. Glucose levels were evaluated in blood drops collected from the tail tip and measured by Accu-Chek Performa II glucometer (Roche^®). The rate of glucose decay constant (kITT) was calculated from the blood glucose concentrations during the linear decay phase, using the formula 0.693/t1/2.

Before the euthanasia, the mice were 10-hours fasted and the trunk blood samples were collected and placed into microtubes containing anticoagulant heparin. The tubes were centrifuged at 1100 G, 15 min, 4°C, and the plasma was collected and stored at −80°C. Insulin was measured by kit AlphaLISA Detection Kit. (#AL204C, PerkinElmer^®, MA, USA), as described according to the manufacturer’s instructions.

After euthanasia the pancreas were dissected, weighed and fixed in paraformaldehyde 4% w/v for 24 hours at room temperature. Then, the organ was dehydrated and embedded in paraffin. A serial section of five µm-thickness were done and placed into a silanized slide. After deparaffinization, the sections were rehydrated and incubated in 10mM sodium citrate solution (pH 6) at 98°C for 30 min. After the antigenic retrieval, permeabilization and the blocking of unspecific bindings, the slides were incubated with primary antibody anti-insulin A0564, Dako cytomation, CA, USA, and anti-glucagon sc7779R, Santa Cruz Biotechnology (at dilution 1:100 and 1:25 respectively, in PBS 1% bovine albumin) for 4 hours at 4°C in humidity chamber. After this period, the sections were washed in PBS and incubated with secondary anti-goat antibody AlexaFluor 488 (1:500) for 1 hour at room temperature. Then, the sections were washed 3 more times in PBS, incubated with DAPI for five minutes in a moisture chamber and then mounted with coverslips using Vectashield antifading mounting media. The images were obtained by AxioScan (ZEISS^®, Oberkochen, Germany). The islet morphology analysis was done as described before (16). Representative images were obtained using an inverted microscope (OLYMPUS^®; Tokyo, Japan).

After euthanasia, the liver was extracted and fixed in paraformaldehyde 4% w/v for 24 hours at room temperature. After this period, the organ was dehydrated and embedded in paraffin. Five µm-thickness sections were done and placed in silanized slide. After the deparaffinization, the five µm-thickness sections were hydrated and stained with Ehrlich’s hematoxylin and eosin, acid and basic differentiator, dehydration and mounted with Entellan. Images were obtained using an inverted microscope (OLYMPUS^®; Tokyo, Japan).

After euthanasia both perigonadal fat pad, right and left, was removed, weighed and the values was recorded.

Groups were compared by one-way ANOVA using the unpaired Turkey’s post-hoc or Kruskal Wallis Test, with Dunn comparisons, when data were not parametric. Data are presented as the mean ± S.E.M. All data were considered statistically different if P value was ≤0.05.

To validate the HNF4α knockout model we analyzed the mRNA for the transcription factor HNF4α in isolated islet from CTL and KO mice. As expected, we found a reduction near to 50% in the HNF4α mRNA expression in KO mice ( Figure 1A ). To confirm the CRE-recombinase activation, we assessed the fasting glycaemia and, it was lower in KO compared to CTL mice ( Figure 1B ).

During the 20-week HFD regiment, HFD-fed mice consumed less amount of food than CHOW-fed mice ( Figures 2A, C ). Daily caloric intake was similar across groups ( Figure 2D ), but HFD-fed mice showed greater energy efficiency, leading to increased body mass and perigonadal fat ( Figures 2E, B, F ). Liver analysis showed higher lipid accumulation in KO/HFD mice, with larger lipid droplets, Mallory hyaline, and hepatocyte bulging, suggesting more pronounced steatosis and fibrosis compared to CTL mice ( Figure 3 ).

KO mice fed with HFD had significantly higher fasting blood glucose levels than CTL mice and was not different from CTL/HFD, this may represent an additive effect of diet and genotype ( Figure 4A ). Also, KO/HFD mice presented greater glucose intolerance compared with CTL and CTL/HFD mice ( Figures 4B, C ). Insulin sensitivity was reduced in both CTL/HFD and KO/HFD group ( Figure 4E ), but fasting insulinemia was lower in KO/HFD mice compared to CTL/HFD mice ( Figure 4F ).

The HFD-induced increase in β-cell mass expansion was only observed in CTL/HFD mice ( Figure 5 ). Additionally, islet diameter was significantly smaller in KO/HFD mice compared to CTL/HFD mice ( Figure 5 ).

Since the number and/or the functionality of the pancreatic β-cells is the root of Diabetes mellitus, regardless of the etiological classification (DM1, DM2) of this disease, any kind of treatment that restores a functional β-cell mass will benefit diabetic patients. The Transcription Factor HNF4α is pointed out as a relevant target that could be manipulated (6, 9) to aim β-cell regeneration. So, in this scenario and based on the fact that HNF4α may participate in the metabolic-induced pancreas adaptation, we decided to investigate the HNF4α role on the pancreatic β-cells and metabolism adaptation to an HFD-induced obesogenic environment.

Before exposing the mice to an obesogenic environment, we validated the mice model that lack HNF4α expression specifically in pancreatic β-cell. As expected, we found a reduction near 50% in the HNF4α mRNA expression in KO mice compared to CTL mice ( Figure 1A ). It is worth to point that the expression of HNF4α is not limited to β-cells and our analysis was done in the whole islet.

Pancreatic β-cells are a great sensor of glycemia; therefore, small changes in physiological glucose concentration influence insulin secretion. This ability is due to the function of several enzymes and proteins, such as glucose transporter 2 (GLUT2), glucokinase (GCK), and ATP-sensitive potassium channel (KATP). Once glucose enters β-cells via GLUT2, it is phosphorylated by glucokinase (GCK) to form glucose-6-phosphate. This molecule then undergoes glycolysis to produce pyruvate, which enters the tricarboxylic acid (TCA) cycle, ultimately increasing ATP production (18). The resulting rise in the ATP/ADP ratio leads to the closure of KATP channels, whose functionality depends on the proper expression of the Kir6.2 subunit, regulated by the HNF4α gene. Closure of the KATP channels causes membrane depolarization which, in turn, opens voltage-dependent L-type Ca2+ channels, allowing a rapid influx of Ca2+ into the cell. The increase in intracellular cytosolic Ca2+ is the primary mediator of insulin exocytosis (19).

To confirm the activation of the CRE-recombinase enzyme, we evaluated the fasting blood glucose levels of the mice 15 days after the final tamoxifen administration. As expected (9), KO mice showed lower fasting blood glucose levels compared to CTL mice ( Figure 1B ). Since KATP channels are an important key in the β cell’s glucose sensing ability and that the Kir6.2 gene, a subunit of the KATP channel, is controlled by the HNF4α, the KO mice β-cells present an inability to secrete insulin in a regulated manner, resulting in lower insulin secretion under basal blood glucose conditions (fasting glycemia).

In addition, HNF4α also contributes to the expression of mRNAs for genes involved in glucose metabolism and glucose-stimulated insulin secretion (GSIS). We have previously shown that HNF4α KO mice present higher insulinemia before (0’) and 60 min after glucose injection in the ipGTT. In addition, these KO animals present and increased GSIS that explains the lower fasting glycemia observed here and it confirms our HNF4α KO model (9).

Upon the HNF4α KO model confirmation, we feed the mice HFD for twenty weeks. During this period, the mice were weighed, and their diet consumption was documented, wherein we noticed that the HFD-fed group had lower dietary intake during the course of the experiment compared to the CHOW-fed group. ( Figures 2A, B ).

Gastrointestinal hormones stimulated by fatty meals, such as Cholecystokinin (CCK), contribute to energy homeostasis, and some of these hormones may influence the brain by serving as satiety signals to inhibit food intake. Peripheral CCK is secreted by intestinal I cells in response to fatty meals and is involved in intestinal modulation, motility, stimulating pancreatic enzyme secretion, and regulating appetite (20). Intraperitoneal administration of purified CCK in fasted rats reduced food intake amount but not meal frequency, classifying CCK as a satiety peptide (21). Thus, in line with the literature and considering that it is physiological that the rodents in the food ingestion based on the energetic density of the diet (22), the HFD induced greater satiety than CHOW, resulting in animals consuming less food over the 20-week period.

Furthermore, despite Figure 2C ( Figure 2C ) suggests that the daily calorie consumption of both groups was similar, Figure 2D shows that HFD had a more pronounced energy efficiency than CHOW, justifying the greater body mass gain and increased perigonadal fat in mice fed an HFD ( Figures 2E, F ). In addition, the HFD-fed mice developed obesity, since they gained more weight and the excess energy was stored as fat.

The high-fat diet (HFD) also affects liver tissue. Our observations indicate that while HFD induces an increase in lipid storage in the liver regardless of genotype ( Figure 3 ), knockout mice fed with HFD showed a significantly higher amount of lipid droplets in the liver ( Figures 3B, C ), along with alterations in Mallory hyaline and bulging hepatocytes, indicating a state of steatosis (23, 24). Additionally, the HFD-induced increase in liver parenchymal fibrosis was greater in animals lacking HNF4α compared to CTL mice ( Figure 3 ).

These HFD outcomes, observed here, confirm that we have an obesity mice model and lead us to believe that the specific deletion of TF HNF4α in the β-cells of the pancreas may trigger systemic changes that are important for an adaptive response to obesity, probably due to the lack of the expected physiological amount circulating insulin in the knockout mice.

To deeply investigate the metabolic adaptive response to obesity we evaluated the glucose tolerance, insulin sensitivity, and insulinemia after 20 weeks on the HDF feeding program. The HFD-induced increase in fasting glycaemia was greater in the KO mice than in the CTL mice ( Figure 4A ). Similarly, the glucose intolerance observed in the HFD fed mice was more prominent in the mice which lack the HNF4α in the β-cells. These results are in accordance with the literature (6) and show that the β-cell ability to respond to a glucose challenge is diminished, suggesting that KO mice are more susceptible to the effects of HFD, exhibiting, in addition to a significant increase in fasting blood glucose, a profound glucose intolerance ( Figures 4B, C ).

As expected, exposure to an HFD reduced insulin sensitivity in these mice, regardless of the genotype ( Figures 4D, E ). It is well established that any reduction in peripheral insulin sensitivity is adaptively compensated by increased pancreatic β-cell function. As expected, CTL mice fed with HFD presented an increase in fasting insulinemia, and such an expected outcome was not observed with the same magnitude in KO/HFD mice ( Figure 4F ).

When β-cells cannot adjust to the increased insulin demand imposed by factors that increase metabolic demand, such as glucocorticoids (25), pregnancy (6), or obesity (26) fasting and/or fed hyperglycemia may occur. Here, we showed that the lack of the TF HNF4α, specifically in β-cells, impaired the proper metabolic adaptation in an obesogenic environment due to a failure in the β-cells ability in respond to this scenario.

To investigate if the inability of the β cell lacking HNF4α in response to obesity was due to a structural adaptation failure we analyzed the islet architecture and observed that the HFD-induced increase in β-cell mass expansion was not noticed in KO/HFD mice. Moreover, it is visible that the islet diameter was lower in the KO/HFD when compared with CTL/HFD group ( Figure 5 ). These results corroborate the ones published by Gupta (6) and our group (9) and reinforce the evidence that the TF HNF4α is relevant for the ability of the β-cells to adapt themselves to respond to a scenario where there is an increase in the metabolic demand.