Hepatocellular glycogenotic foci after combined intraportal pancreatic islet transplantation and knockout of the carbohydrate responsive element binding protein in diabetic mice

Aims The intraportal pancreatic islet transplantation (IPIT) model of diabetic rats is an insulin mediated model of hepatocarcinogenesis characterized by the induction of clear cell foci (CCF) of altered hepatocytes, which are pre-neoplastic lesions excessively storing glycogen (glycogenosis) and exhibiting activation of the AKT/mTOR protooncogenic pathway. In this study, we transferred the IPIT model to the mouse and combined it with the knockout of the transcription factor carbohydrate responsive element binding protein (chREBP). Methods C57BL/6J Wild-type (WT) and chREBP-knockout (chREBP-KO) mice (n = 297) were matched to 16 groups (WT/ chREBP-KO, experimental/control, streptozotocine-induced diabetic/not diabetic, one/four weeks). Experimental groups received the intraportal transplantation of 70 pancreatic islets. Liver and pancreatic tissue was examined using histology, morphometry, enzyme- and immunohistochemistry and electron microscopy. Results CCF emerged in the liver acini downstream of the transplanted islets. In comparison to WT lesions, CCF of chREBP-KO mice displayed more glycogen accumulation, reduced activity of the gluconeogenic enzyme glucose-6-phosphatase, decreased glycolysis, lipogenesis and reduced levels of the AKT/mTOR cascade members. Proliferative activity of CCF was ∼two folds higher in WT mice than in chREBP-KO mice. Conclusions The IPIT model is applicable to mice, as murine CCF resemble preneoplastic liver lesions from this hepatocarcinogenesis model in the rat in terms of morphological, metabolic and molecular alterations and proliferative activity, which is diminished after chREBP knockout. chREBP appears to be an essential component of AKT/mTOR mediated cell proliferation and the metabolic switch from a glycogenotic to lipogenic phenotype in precursor lesions of hepatocarcinogenesis.


Diabetes induction
Diabetes was induced with a single intraperitoneal dose of streptozotocin (180 mg/kg bodyweight, Zanosar®, Sigma-Aldrich, Darmstadt, Germany). Mice with an intended blood glucose level > 20 mmol/l after five days were considered diabetic. Transplantation was performed at least after 1 week.

Transplantation
Pancreatic islets were isolated from male donor mice from the same genotype that were killed under anaesthesia and perfused with neutral red solution (Roth, Karlsruhe, Germany). Pancreatic tissue was dissected, reduced mechanically to small pieces and exocrine tissue digested with collagenase/albumin (2.0/1.0 mg per individual pancreas, Serva, Heidelberg, Germany) for 9 minutes. Pure vital islets were collected manually with a pipette under the stereomicroscope and were stored shortly on ice (5-8°C). Recipient WT or chREBP-KO mice received an intraportal transplantation of 60-70 isolated, isologous pancreatic islets into the liver under anesthesia (50-100 mg/kg bodyweight ketamine, 10 mg/kg bodyweight xylazine) via cannulization of the portal vein with a 27 gauge needle, connected with a thin flexible tube system filled with islet particles in 0.1 ml Hanks-solution (Sigma-Aldrich, Darmstadt, Germany).
As intended, the diabetic mice remained hyperglycemic despite islet transplantation, because the number of transplanted islets was low. Continuous hyperglycemia is necessary in this model, so that the islets are stimulated for permanent maximal insulin synthesis and secretion. As a result, local hyperinsulinism and simultaneous hyperglycemia is obtained at the downstream liver acini. Body weight and blood glucose level were measured at two timepoints -just before transplantation and before killing.

Viability tests
For dye exclusion test with Trypan blue, 100-200 isolated pancreatic islets in Hank`s solution were slowly centrifugated (500 rpm for 5 minutes) and diluted in Trypan blue solution (0.4 %, 1:1, Thermoscientific, Waltham, MA, USA), incubated for 1-2 minutes in a hemocytometer, and evaluated under the microscope (five times in replicate). Blue stained cells were classified as not viable, unstained cells as viable.
For the Nicotinamide-Adenin-Dinucleotide (NADH)-Diaphorase test, another 100-200 isolated islets were frozen via methylbutane (Roth, Karlsruhe, Germany) in -120°C liquid nitrogen and shortly stored at -80°C (three times in replicate). Cryostat sections of 10μm were stained for enzyme activity of NADH-Diaphorase according to Noguchi et al., 2006 [45]. Cells with blue cytoplasmic staining were considered as viable cells.

Increase of transplanted islet number
To increase the frequency of CCF in wildtype mice, we conducted transplantation of higher numbers of 120 and 200 isolated islets, mainly in wildtype mice for one and four weeks, and as a control also in one set of chREBP-KO mice (Supplementary Table 1).

Pre-operative insulin treatment
Furthermore, a possible glucolipotoxicity [40] could directly inhibit islet engraftment in portal vein branches immediately after transplantation. In a further additional experimental setting (Supplementary Table 1), diabetic C57Bl/6J wildtype mice received a pre-operative treatment for five days with Neutral Protamin Hagedorn (NPH) Insulin (Protaphane ®, Novo Nordisk, Mainz, Germany; 40IU/kg body weight, subcutaneous [1]), to achieve normoglycemia of 5-16 mmol/l. At the day of transplantation, insulin treatment was discontinued. Diabetic wildtype mice received an intraportal transplantation of 200 isolated islets. Blood glucose level was measured daily prior to transplantation, and 1 day, 3 days and then weekly after transplantation. Control wildtype mice remained without islet transplantation.

Application of the nucleoside analog 5-Bromo-2`deoxyuridine (BrdU)
Seven days before sacrifice, half of the animals of each group was anaesthetized and osmotic minipumps (Osmotic Pump Model 2001, Charles River Laboratories, Sulzfeld, Germany) filled with 2.5mg of BrdU (Sigma Aldrich, Heidelberg, Germany) were surgically implanted subcutaneously between the scapulae. These pumps continuously delivered BrdU until the animals were sacrificed. The other half of each group received BrdU (8mg/kg bodyweight) 24, 12 and one hour before killing.
The liver was removed and cut into slices of 1-2 mm thickness. Liver lesions of > 1mm and/or transplants were detected and documented under the stereomicroscope. Liver and pancreas tissue was embedded in paraffin or fixed in glutaraldehyde and embedded in glycidether (Serva, Heidelberg, Germany).
Paraffin slides of 1-2 μm thickness were cut and stained by H&E and the periodic acid Schiff (PAS) reaction.

Enzyme histochemistry
Pieces from frozen liver tissue of two WT and chREBP-KO mice were frozen onto the same tissue holder, and serial sections were cut simultaneously in a cryostat (Microm HM 550, Thermoscientific, Waltham, MA, USA), mounted onto the same slide, stained with H&E, PAS and corresponding 16 μm slices for glucose-6-phosphatase activity (lead method [33]) and Glucose-6-Phosphate-dehydrogenase (Tetrazolium salt MTT method [34]).
Immunohistochemical signal intensity in CCF was estimated semi-quantitatively by comparing CCF with corresponding surrounding unaltered liver tissue. Negative controls were stained without a primary antibody. Pancreas BrdU stainings were double-stained with Insulin using an automated immunostainer (Leica Biosystems, Wetzlar, Germany), the UltraVision LP-kit (Thermoscientific, Waltham, MA, USA), and the AP-Polymer and Fast Red as chromogen substrate.
Detailed information about primary antibodies used for immunohistochemistry is listed in Supplementary  Table 2.

Morphologic and proliferation kinetic investigations
CCF were identified in the liver as lesions of enlarged hepatocytes with pale cytoplasm in H&E staining due to extensive glycogen storage (positive in the PAS reaction) and were accompanied by lipid droplets in WT mice, but not in chREBP-KO mice. The corresponding lesions in the enzyme-and immunostained sections were detected by comparison with H&E stained sections. Only well-demarcated foci of clear cell phenotype that were easily distinguished from the surrounding liver parenchyma were determined as CCF. BrdU labeling indices of CCF and extrafocal liver tissue were evaluated in representative sections of mice livers containing CCF. If less than 100 hepatocytes were detectable in the CCF, the liver was excluded from the evaluation. Counts of 2000 extrafocal hepatocytes served as internal control. BrdU labeling indices of pancreatic islets were evaluated for all islet cells in one parenchyma section of about 1cm 2 , and additionally after double-staining with insulin for β-cells. Levels of proliferation are expressed based on BrdU labeling index (number of positive hepatocyte/ total number of hepatocyte nuclei x 100 %, or number of positive islet (or β-) cells/total number of islet (or β-) cell nuclei x 100 %, respectively). Quantitative data are expressed as mean ± standard error of the mean (S.E.M.).

Ultrastructural analysis
Specimens of 2 mm 3 liver containing CCF and pancreatic tissue were cut with a razor blade, fixed in 2.5 % glutaraldehyde, embedded in Glycidether 100, cut with diamond knifes (Science Servcie GmbH, Munich, Germany) with a Leica ultratome Leica EM UC7 (Leica Biosystems, Wetzlar, Germany) to 500 and 750 nm thick semi-thin slides and stained with H&E, PAS and according to Richardson [48]. Ultrathin sections of 70-90 nm were stained with uranyl acetate (Merck, Darmstadt, Germany) and lead citrate (Sigma-Aldrich, Darmstadt, Germany) and examined with a Libra 120 electron microscope from Carl Zeiss (Jena, Germany).

Statistical analysis
Differences in body weight, blood glucose level, proliferative activity of CCF using Student`s t test. Differences of frequency of CCF were tested using Fisher`s exact test. Differences were considered significant if p < 0.05.

Supplementary Figure 10: Expression of TNFα and iNOS in clear cell foci (CCF) of diabetic chREBP-KO mice, four weeks after intraportal pancreatic islet transplantation (TX).
In comparison to WT mice without any inflammatory reaction, CCF (H&E) of chREBP-KO mice reveal an upregulation of TNFα and iNOS in leucocytes between parenchyma cells, but also focally in hepatocytes. Length of the lower edge: H&E 0.6mm; immunostainings (corresponding to black square in H&E) 0.35mm.