Dual hit rat models of diabetic cardiomyopathy and nephropathy: pancreatectomy combined with isoprenaline treatment or uninephrectomy

Background Cardiovascular and renal complications are the predominant causes of morbidity and mortality amongst patients with diabetes. Development of novel treatments have been hampered by the lack of available animal models recapitulating the human disease. We hypothesized that experimental diabetes in rats combined with a cardiac or renal stressor, would mimic diabetic cardiomyopathy and nephropathy, respectively. Methods Diabetes was surgically induced in male Sprague Dawley rats by 90% pancreatectomy (Px). Isoprenaline (Iso, 1 mg/kg, sc., 10 days) was administered five weeks after Px with the aim of inducing cardiomyopathy, and cardiac function and remodeling was assessed by echocardiography ten weeks after surgery. Left ventricular (LV) fibrosis was quantified by Picro Sirius Red and gene expression analysis. Nephropathy was induced by Px combined with uninephrectomy (Px-UNx). Kidney function was assessed by measurement of glomerular filtration rate (GFR) and urine albumin excretion, and kidney injury was evaluated by histopathology and gene expression analysis. and

Further insight into the pathogenesis of both DbCM and DN is essential in order to advance clinical management of these major causes of morbidity and mortality worldwide. According to the Diabetic Complications Consortium, a valid rodent model of DbCM should include left ventricular (LV) dysfunction and hypertrophy, interstitial fibrosis, and altered gene expression, but also an increased vulnerability to cardiac stressors [16]. For DN, a valid model should comprise progressive albuminuria and GFR loss, alongside characteristic histopathological changes such as arteriolar hyalinosis, glomerulosclerosis, and tubulointerstitial fibrosis [17].
The most-studied, current rodent models of diabetic complications are often limited in their Page 4 of 32 usefulness as they produce only a few essential features associated with early stages of DbCM and DN progression [18][19][20]. Some of these commonly-used models of diabetes, e.g. streptozotocin-treatment (STZ) and the db/db genetic mouse model, have previously been combined with angiotensin II or uninephrectomy to mimic and accelerate DbCM or DN, respectively [21][22][23].
The use of STZ in this context is however suboptimal, since the toxin also has the potential to induce direct nephrotoxic and cardiotoxic effects, independent of hyperglycemia [24][25][26].
Pancreatectomy (Px), on the other hand, is a well-characterized method of surgically inducing diabetes in rats [27]. An important advantage of this model is that it reflects the isolated effects of a reduced beta cell mass; immediately after surgery Px rats become insulin deficient and hyperglycemic, allowing full control of diabetic onset [28]. To our knowledge, the cardiac and renal pathology has not previously been evaluated in the Px rat, and nor have models of DbCM and DN been superimposed upon this model. The aim of this study was therefore to investigate the development and progression of DbCM and DN in a Px rat model of diabetes. To accentuate the cardiac and renal changes, we combined Px with isoprenaline, a non-selective βadrenoreceptor agonist and well-known cardiac stressor capable of inducing cardiac dysfunction and fibrosis [29][30][31], and uninephrectomy, which is known to accelerate the progression of injury in the remaining kidney [32,33]. We hypothesized that superimposing these cardiac or renal stressors on pre-existing diabetes would result in rat models that recapitulated key features of human DbCM and DN, respectively.
Pancreatectomy. Through a midline abdominal incision, the pancreatic tissue was removed by gentle abrasion with disposable micro-brushes (Plandent). All major blood vessels were left intact. For 60% Px, the tail and body of the pancreas was removed whereas the pancreatic tissue within the duodenal loop, comprising the entire head of the pancreatic tissue, was preserved.
For 90% Px, only the pancreatic tissue between the common bile duct extending to the first loop of the duodenum was preserved.
Sham surgery. Through a midline abdominal incision, the tail and body of the pancreatic tissue was disengaged from the mesentery and gently manipulated before being repositioned to the abdominal cavity.
The rats were fasted until the next day, whereupon 5 g of chow was administered to each animal. Two days after surgery, the animals were fed ad libitum.

Isoprenaline treatment
Isoprenaline hydrochloride (Sigma-Aldrich) was dissolved in saline immediately prior to dosing and administered subcutaneously (1 mg/kg) for ten consecutive days beginning five weeks after sham surgery or 90% Px.

Uninephrectomy
Page 6 of 32 Animals were subjected to uninephrectomy at the time of 90% Px. The right ureter, renal artery and vein were identified and ligated. Subsequently, the right kidney was removed. In shamoperated animals, the right kidney was exposed and gently manipulated.

Blood and plasma analyses
Blood glucose. Tail vein blood was collected once weekly from non-fasted or fasted (4 h) rats into heparinized glass capillary tubes and immediately suspended in glucose/lactate system solution buffer (EKF-diagnostics). The glucose concentration was measured immediately using a BIOSEN c-Line glucometer (EKF-diagnostics) according to the manufacturer's instructions.
Oral glucose tolerance test (OGTT). In experiment 1, animals were fasted for 4 h prior to oral glucose administration (2 g/kg body weight, 500 mg glucose/mL, Fresenius Kabi), and blood glucose was measured at 6 timepoints over the course of 240 min post-injection.
Blood and plasma markers. Plasma insulin and C-peptide were measured in duplicate using AlphaLISA platform (Perkin Elmer) according to the manufacturer's instructions. Glycated haemoglobin (HbA1c), plasma creatinine, and urea were measured using commercial assays (Roche Diagnostics) on the Cobas C-501 autoanalyzer, and plasma cystatin C was measured using a commercial assay (R&D Systems) all as per manufacturers' instructions.

Urine collection and biochemical analyses
Rats were single-housed in metabolic cages (Techniplast) with free access to powdered chow and water. The excreted urine was collected for 16 hours. Urine creatinine was quantified using a CREP2 assay (Roche Diagnostics) on the Cobas C-501 autoanalyzer as per manufacturer's instructions. Urine albumin was measured using a rat albumin ELISA (Bethyl Laboratories).
Urine neutrophil gelatinase-associated lipocalin (NGAL), soluble tumor necrosis factor receptors (sTNFR) I and II were measured using ELISA assays (R&D Systems), while urine nephrin and podocalyxin were quantified using ELISA assays from LSBio and CusaBio, respectively. All ELISA assays were run as per manufacturers' instructions and levels of analytes were reported as the analyte-to-creatinine ratio (ACR) in urine. Urine ACR values were log10-transformed before statistical analyses.

Echocardiography
Echocardiographic assessments were performed under isoflurane anesthesia (2-3%). After chest hair removal, the rats were positioned in supine position on a heated pad. Electrocardiogram electrodes (lead II configuration, 3M TM ) were placed and the rectal temperature was measured before and after examination. All examinations were performed using an ultrasound device (Philips, iE33) with a 12 MHz sector array probe. Function and dimensions of the LV were assessed from parasternal short and apical long axis views. The LV internal diameter (LVID), anterior and posterior wall thickness (LVAW, LVPW) were assessed in systole (LVIDs, LVAWs and LVPWs, respectively) and diastole (LVIDd, LVAWd and LVPWd, respectively) using 2D-guided M-mode in a short axis view at the level of the papillary muscles, according to the leading edge to leading edge principle [34]. The ejection fraction (EF) was calculated using the Teichholz formula [35]. To correct for the difference in LV size, all short axis measurements were normalized to body weight. LV filling (E and A velocity and deceleration time) was measured in the apical 4-chamber long axis view with the sample volume placed at the tip of the mitral valve leaflets using pulsed wave Doppler.
Echocardiographic analyses were performed offline using Q-station software (version 3.8.5, Philips). Image acquisitions and analyses were performed by one operator blinded towards identity of the animal.

Glomerular filtration rate measurement
GFR was assessed by a fluorescein isothiocyanate (FITC)-inulin test. Briefly, a 5% solution of FITC-inulin (TdB Consultancy) was prepared in 0.85% saline by heating and overnight dialysis using a Spectra/Por 6 dialysis tube (1 kDa molecular cut-off; Spectrum Labs). The FITC-inulin solution was sterile filtered (0.2 µm syringe filter) before tail vein injections in conscious rats (100 mg/kg). Sublingual blood was collected in heparinized tubes at 8 timepoints over the course of 75 min post-injection. Plasma FITC-inulin concentrations were measured using a plate-reader (CLARIOstar, BMG LABTECH) and GFR was calculated using a twocompartment model from the rate of decay in plasma FITC-inulin as described previously [36] and normalized to body weight.

Histology and stereology
Pancreas. At termination, pancreatic samples from sham, 60% and 90% Px rats were removed en-bloc and incubated in 10% neutral buffered formalin until further processing. Pancreatic tissue was carefully dissected, weighed, and processed as previously described [37]. Briefly, pancreas was rolled tightly into strips of gauze before infiltration with paraffin in an automated tissue processor (VIP5, Sakura). The pancreas was then cut into 7-9 systematic uniform random tissue slabs with a razor blade fractionator and embedded in paraffin blocks with the cut surface down. Subsequently, 5 µm thick sections were cut from each block on a microtome and collected on microscope slides. Immunohistochemistry against beta cells and non-beta cells were performed as a double staining using standard procedures. Briefly, after deparaffinization and microwave oven pretreatment in Tris-EGTA buffer (pH 9), sections were stained for nonbeta cells using an antibody cocktail consisting of rabbit anti-glucagon (Phoenix, H-028-02, 1:5000), rabbit anti-somatostatin (Dako, 0566, 1:7500), and rabbit anti-pancreatic polypeptide (Europroxima, B32-1, 1:5000). The antibody cocktail was detected using Envision+ anti-rabbit HRP-coupled polymer system (Dako, K4002) and developed with DAB-Nickel as chromogen.
Heart and kidney. One half of the sagittally-divided LV or left kidney were fixed in 10% neutral buffered formalin for 24 h at room temperature. The fixed tissues were cut into 5-7 systematic uniform random tissue slabs with a razor blade fractionator and embedded in paraffin blocks with the cut surface down. For Picro Sirius Red staining, a set of 3 µm sections were cut on a microtome from each block and collected on microscope slides. For Periodic acid-Schiff (PAS) staining, another set of sections were cut in pairs with a dissector distance of 30 µm. Picro Sirius Red staining: After deparaffinization, sections were incubated in Wiegert's iron hematoxylin (Sigma-Aldrich) and then stained in Picro Sirius Red (Sigma-Aldrich) before they were cover slipped with Pertex. For the PAS staining, sections were deparaffinized and oxidized with 0.5% periodic acid solution followed by incubation with Schiff's reagent.
Sections were counterstained in Mayer's hematoxylin and cover slipped with Pertex.
Podocin/type IV collagen double fluorescent immunohistochemical staining of kidney sections was performed as previously described [22].
Image analysis and stereological quantification. Stereological quantification of beta cell and non-beta cell mass was calculated as an area fraction (area of beta cells versus total pancreatic area) multiplied by the dissected pancreas mass as previously described in [38]. Heart and kidney collagen content was determined as Picro Sirius Red area fraction using image analysis.
Stained slides were scanned under a 20x objective in an Aperio Scanscope AT slide scanner and imported into an image analysis module in Visiopharm Integrator System (Visiopharm).
A Bayesian classifier was trained to detect Picro Sirius Red positive collagen versus other Page 9 of 32 tissue components. The collagen area fraction was calculated as the Picro Sirius Red area divided by the total tissue area. Intra-glomerular collagen type 4 content and stereological estimation of kidney compartmental volumes were determined as previously described [22].
The mean glomerular volume was calculated as total glomerular volume divided by the mean number of glomeruli estimated using the physical dissector on serially cut reference and lookup sections with a dissector distance of 30 µm [39].

RNA sequencing
Transcriptome analysis was performed by sequencing of RNA extracts from the LV and renal cortex. RNA was purified from homogenized tissue using the NucleoSpin RNA Plus kit (Macherey-Nagel GmbH). The RNA quantity was measured using Qubit (Thermo Scientific, Eugene, OR) and RNA quality was determined using a bioanalyzer with RNA 6000 Nano kit (Agilent). Purified RNA from each sample was used to generate a cDNA library using the  [40]. Differential gene expression analysis was performed with DEseq2 [41].

Statistical analysis
All data were analyzed using GraphPad Prism software (version 8.02). Normal distribution of data was assessed by inspection of QQ plots and by using the Shapiro-Wilk test. Normally distributed data are shown as mean ± standard error of mean (SEM). In experiment 1, repeated measures two-way analyses of variance (ANOVA) with Bonferroni's post-hoc test were used for time-course data and one-way ANOVA with Tukey's post-hoc test was used for insulin and beta cell mass data. In experiment 2, two-way ANOVA was applied and in addition to evaluating the main effects of Px and isoprenaline treatment, the interaction between the two was included to evaluate a potential synergistic effect. Unpaired t-test and repeated measurements two-way ANOVA with Bonferroni's Multiple Comparisons test were used for experiment 3. A p-value <0.05 was considered statistically significant.

Metabolic characterization of the pancreatectomized rat
Page 10 of 32 In experiment 1, 60% Px did not affect body weight, whereas 90% Px resulted in a lower rate of body weight gain over 6 weeks compared with sham-operation (p<0.05, fig. 2A). One week after surgery, 90% Px rats exhibited hyperglycemia (non-fasting blood glucose > 10 mmol/L) whereas 60% Px and sham rats were normoglycemic (p<0.05, fig. 2B). Among the 90% Px rats, 50% remained hyperglycemic throughout the six-week study period (defined as '90% Px responders') whereas 50% returned to normoglycemia three weeks after surgery (defined as '90% Px non-responders'). Correspondingly, blood glucose was increased for 15-120 min after an oral glucose bolus during the OGTT in 90% Px responders compared to sham (p<0.05, fig.   2C), but only for 15-30 minutes in 90% Px non-responders and in 60% Px. Both 60% and 90% Px resulted in significantly lower plasma insulin compared with sham (p<0.05, fig. 2D). While the total beta cell mass in the remaining pancreatic tissue was reduced in 60% and 90% Px rats  2F).

The pancreatectomized and isoprenaline-treated rat as a model of DbCM
In experiment 2, one rat was excluded due to lack of hyperglycemia two weeks after Px, five rats died from post-operative complications, and five Sham-Iso rats died after the first administration of isoprenaline (isoprenaline-induced mortality rate: 29.4%).
No significant interaction between Px and isoprenaline treatment was found for any investigated outcome. Instead, irrespective of isoprenaline treatment, a main effect of Px was found for several measures reflecting glucose metabolism and LV remodeling. Specifically, plasma C-peptide and HbA1c were increased post-Px, plasma insulin was decreased, and animals subjected to Px did not gain body weight to the same extent over the following ten weeks as sham-operated animals (Table 1; all p<0.001). Heart rate and peak E wave velocity were decreased (Table 2; p<0.001) and relative LV mass (p<0.05), wall thickness and internal dimensions in systole and diastole (Fig. 3, Table 2; p<0.001) was increased in Px rats compared with sham-operated counterparts. Furthermore, LV collagen area fraction was modestly decreased by Px (Table 1; p<0.05). Owing to the frequent finding of fused E-and A-waves, calculation of E/A ratios was not possible during echocardiography.
In a manner distinct from Px, isoprenaline treatment decreased indices of systolic LV function; LVIDs and EF (Fig. 3, Table 2; both p<0.001) and increased LV collagen area fraction (Table   1, p<0.001). were upregulated (p<0.001, fig. 4). In contrast, isoprenaline treatment per se did not affect gene expression of these selected genes compared to vehicle treatment other than upregulating GLUT1 (p<0.01). Further analysis of expression levels of genes involved in adrenergic signaling revealed that β2-receptors were downregulated by Px (p<0.01) while β1-receptors were unaffected (data not shown). Moreover, expression of downstream mediators of βreceptor signaling, adenylate cyclase (AC) 6 and phosphodiesterase 3A (PDE3A) were upregulated by Px (p<0.001 and p<0.05, respectively) while PDE3A was downregulated by isoprenaline (p<0.05).

The pancreatectomized and uninephrectomized rat as a model of DN
In experiment 2, renal changes induced by Px were observed including increased left kidney weight (1.7±0.1 g vs.1.4±0.1 g, p<0.05) and urine ACR (2549±1447 µg/mg vs. 64 ±10 µg/mg, p<0.01) in Px-Veh compared to Sham-Veh ten weeks after surgery. This indicates renal pathology in the Px rats. In experiment 3, the progression of DN in Px rats was therefore investigated alongside UNx, a known accelerator of disease progression in rodent models of DN. As expected, Px-UNx caused an increase in blood glucose and a decrease in plasma insulin compared to sham (both p<0.05, table 4). Urinary excretion of albumin, NGAL, and sTNFR-II was increased in Px-UNx two weeks after surgery and sustained at six and nine weeks postsurgery (all p<0.001, table 3). Urine sTNFR-I and podocalyxin excretion was increased by week six and nine after surgery (both p<0.001). At termination, eleven weeks after surgery, plasma urea was increased in Px-UNx (p<0.05), whereas plasma creatinine was reduced (p<0.05). Cystatin C was not significantly different, but GFR was increased eight weeks after surgery in Px-UNx compared to sham (p<0.05).
Renal hypertrophy was observed in Px-UNx rats compared with sham-operated animals with increased kidney weight and renal cortex and medulla volumes (p<0.05, fig. 5A-C). Px-UNx rats also displayed significant glomerular hypertrophy as assessed by stereology (p<0.05 vs. Renal cortex gene expression analysis of selected pathways involved in the pathogenesis of diabetic nephropathy is shown in Figure 6A. Several genes involved in TGF-beta signaling, fibrosis and inflammation, and angiogenesis were upregulated, while the expression of genes involved in the renin-angiotensin-aldosterone system was downregulated ( fig. 6A). Genes of the insulin signaling pathway were significantly altered in Px-UNx compared to sham ( fig.   6A). Expression of podocyte markers was either down-regulated (nephrin) or unchanged (podocin), while tubular damage markers (neutrophil gelatinase-associated lipocalin (NGAL) and Kidney Injury Molecule-1 (KIM-1)) were significantly upregulated (p<0.05, fig. 6B).
Finally, drug-related targets including sodium-glucose transport protein 2, glucagon-like peptide 1 receptor, and angiotensin I converting enzyme were differentially regulated in Px-UNx compared to sham ( fig. 6B).

DISCUSSION
Here we characterized two preclinical rat models of DbCM and DN by superimposing isoprenaline treatment and UNx, respectively, on Px. Px led to marked alterations in glucose homeostasis and induced morphological and functional changes related to cardiovascular and renal endpoints. While no synergistic effect was revealed by superimposing the cardiac stressor, isoprenaline, to Px, superimposing UNx to Px in rats led to exacerbated DN as measured by alterations in albuminuria and GFR.
In agreement with previous findings [42], we demonstrated that 90%, but not 60%, Px induced permanent, stable hyperglycemia, concomitant with a surgically induced reduction in beta cell mass and thereby lower plasma insulin levels. Interestingly, we found that despite having undergone 90% Px, normoglycemia was restored three weeks after surgery in a subset (50%) of rats, termed non-responders, in experiment 1. These rats were able to maintain adequate insulin production to support normoglycemia. Hereafter, the Px procedure was further refined, which resulted in a reduction to 2% of non-responders in subsequent studies. Evidently, a certain threshold in critical beta cell mass exists.
Besides demonstrating a pronounced and sustained diabetic phenotype, including decreased Cpeptide as well as increased HbA1c, the Px rat also exhibited mild cardiac hypertrophy and remodeling indicated by increased relative LV mass, LV wall thickness and LVIDd. Although EF was preserved, early systolic dysfunction was revealed by increased LVIDs. In DbCM, diastolic dysfunction is known to precede systolic dysfunction but, in this study, unfortunately, Page 13 of 32 several indices of diastolic function, including the late filling of the LV, could not be derived due to low temporal resolution and thus E/A fusion during the echocardiographic examination.
Although the LV filling pattern could not be fully evaluated, the finding of a decreased early filling, corresponding to the peak E wave mitral inflow velocity, may be indicative of impaired relaxation and thus early diastolic dysfunction in Px rats. Furthermore, Px intervention altered the expression of selected genes involved in the pathogenesis of DbCM as suggested by the Diabetes Complications Consortium [16]. For instance, the major cardiac glucose transporter, GLUT4, was downregulated in Px. Interestingly, impaired insulin signaling, including decreased recruitment of GLUT4 to the plasma membrane, has been implicated in promoting cardiomyocyte stiffness in the diabetic heart [43]. In support of an attenuated contractile function, a shift from the 'fast' isoform α-MHC to the slower β-MHC along with decreased SERCA-2 expression was found in LV tissue.
An important advantage of the Px model is that it reflects the isolated effects of reduced beta cell mass, without the potential multiple-organ toxic effects of STZ [24] or the confounding effects of obesity and impaired leptin signaling found in e.g. ZDF rats. A disadvantage, however, is that not only beta cells, but also exocrine acinar cells are excised during 90% Px.
The lack of pancreatic enzymes following Px [44] may thus act as a confounder by potentially inducing malnutrition and growth retardation. In future studies, supplementation with pancreatic enzymes may therefore be considered to avoid such potential effects.
Isoprenaline is a β-adrenergic agonist promoting cardiac hypertrophy, fibrosis, and systolic dysfunction in rats when administered in supraphysiological dosages [31,45,46]. In the present study, isoprenaline treatment per se did not induce cardiac hypertrophy or remodeling but did induce increased LV fibrosis and reduced systolic function. A possible explanation for the lack of isoprenaline-induced LV hypertrophy is that the cardiac remodeling may be reversible as indicated previously by Golomb et al. [47], demonstrating reversal just seven days after cessation of isoprenaline. In the present study, LV changes were assessed nearly four weeks after isoprenaline treatment and in comparison to previous studies where a single high-dose regimen has been used [30,45], a relatively low isoprenaline dosage was applied.
Unexpectedly, the combination of Px and isoprenaline treatment did not accentuate or accelerate LV morphological or functional changes. Clinical studies of patients with diabetes have previously shown an increased susceptibility to hemodynamic stressors [48,49] and we thus hypothesized that the cardiac changes induced by isoprenaline would be accelerated and Page 14 of 32 accentuated when superimposed on Px. Isoprenaline induced a nearly 30% mortality rate in sham-operated rats after the first exposure, whereas no isoprenaline-related deaths were observed in Px animals. Mortality following isoprenaline treatment has been reported previously using higher doses in non-diabetic rats [30,31,45]. In the present study, the expression of β2-receptors was slightly downregulated by Px whilst PDE3A was upregulated, which could attenuate responsiveness to β-receptor stimulation by isoprenaline [50], and hence partly explain the lack of synergy, but also the increased mortality rate. Conversely, AC6, known to improve cardiac function and responsiveness to β-receptor stimulation [51], was upregulated by Px. Another explanation could be related to the decreased heart rate in Px rats.
Lower heart rates will have longer refractory periods and this phenomenon could perhaps protect against isoprenaline-induced tachycardia and arrhythmias. However, further experiments are needed to elucidate this subject further.
UNx is known to accelerate DN progression in rodents [18]. Along this line, we observed that the combination of Px and UNx induced several hallmarks of DN. Px-UNx rats displayed profound renal and glomerular hypertrophy, which together with increased GFR are indicative of renal hyperfiltration. These phenomena are also observed in patients during the early phase of DN disease progression [52]. Importantly, Px-UNx animals presented with persistent albuminuria detectable as early as two weeks post-surgery. Similarly, urine NGAL excretion was increased, which is indicative of tubular damage and supported by increased RNA expression levels of the clinical biomarkers, NGAL and KIM-1 [53,54]. Urine sTNFR I and II excretion was increased; a finding that corresponds to reports of diabetic patients with diabetic kidney disease [55]. Finally, urine podocalyxin excretion was increased in line with observations of early podocyte injury in human patients [56].
Histopathological evaluation of kidneys from the Px-UNx model did not reveal evidence of renal fibrosis or glomerulosclerosis as measured by intra-glomerular type IV collagen.
However, stereological assessment of the kidney revealed massive renal and glomerular hypertrophy. Together with the measured GFR, these data demonstrate that the Px-UNx rat represents a model of early stage DN in human patients. Longer term studies, beyond the ten weeks reported here, are likely required for progression of DN with resulting GFR loss, glomerulosclerosis, and tubulointerstitial fibrosis.
Although non-diabetic uninephrectomized animals were not included as a control in this study, our data suggest that UNx accelerates DN in the Px rat. Compared with Px rats from experiment

Consent for publication
All authors have declared their consent for this publication.

Availability of data and materials
All data and materials are available upon request.

Funding
The salary of LT is partly funded by Innovation Fund Denmark (grant no: 7038-00070B). RTL received a PhD studentship from the British Heart Foundation (grant no: FS/14/59/31282).  Body weight (A) and non-fasting blood glucose (B) measured in rats subjected to pancreatectomy (Px), where either 60% Px or 90% Px was removed, or sham surgery throughout the study period. Blood glucose levels from an oral glucose tolerance test performed three weeks after surgery (C). Non-fasting plasma insulin and total beta-cell mass determined by stereology six weeks after surgery (D-E) and representative images of sections from sham and Px rats stained for beta cells (orange-brown; insulin-immunoreactive) and nonbeta cells (black; pancreatic polypeptide, somatostatin and glucagon-immunoreactive) (F). Data is presented as mean + SEM. n=10,10,5,5. Two-way ANOVA with Bonferroni's posthoc test was applied to blood glucose data and one-way ANOVA with Tukey's post-hoc test to insulin and beta cell mass data. *p<0.05, **p<0.01, ***p<0.001 vs Sham.   Table 3. Urine markers in pancreatectomized-uninephrectomized and sham rats. Measurements were performed after two, six, and nine weeks in sham operated or pancreatectomized-uninephrectomized (Px-UNx) rats. Data is presented as mean ± SEM. Two-way repeated measures ANOVA with Bonferroni's Multiple Comparisons post-hoc test; **p<0.01, ***p<0.001 vs. Sham.