Hyperbranched polyglycerol is superior to glucose for long-term preservation of peritoneal membrane in a rat model of chronic peritoneal dialysis

Physioneal™ 40 solution (denoted as PYS here) was purchased from Baxter Healthcare Co. (Deerfield, IL, USA). Primary antibodies were: mouse anti-vascular endothelial growth factor (VEGF) antibody (Cat#: NB100-664, Novus Biologicals, Littleton CO, USA), mouse monoclonal anti-α-smooth muscle actin (SMA) (clone 1A4, Sigma-Aldrich, Oakville, ON, Canada), and mouse monoclonal anti-MAC387 antibody (Clone MAC387, Santa Cruz Biotech Inc., Dallas, TX, USA), and mouse monoclonal anti-β-actin (clone AC-40, Sigma-Aldrich, Canada). Secondary horseradish peroxidase (HRP)-conjugated anti-mouse IgG HRP (sc-2314) was from Santa Cruz Biotech.

Male outbred Wistar rats (~400 g bodyweight, 12–14 weeks old) were purchased from the Charles River Laboratories International, Inc. (Wilmington, MA, USA), and maintained in the animal facility of the Jack Bell Research Centre at the University of British Columbia (UBC) (Vancouver, BC, Canada). Animal experiments were performed in accordance with the Canadian Council on Animal Care guidelines under protocols approved by the animal use subcommittee of UBC.

HPG (1 kDa) was synthesized by anionic ring opening multi-branching polymerization, and its characteristics were verified as described previously [18, 19].

HPG solution (denoted as HPG, approximately 402 mOsmol/kg, pH 7.4) was prepared by dissolving of HPG (1 kDa, 6% w/v) in PD buffer solution similar to PYS solution containing sodium chloride (NaCl, 538 mg/100 mL), sodium lactate (NaC₃H₅O₃, 168 mg/100 mL), calcium chloride dehydrate (CaCl₂ 2H₂O, 18.4 mg/100 mL), magnesium chloride hexahydrate (MgCl₂ 6H₂O, 5.1 mg/100 mL) and sodium bicarbonate (NaHCO₃, 210 mg/100 mL). The osmolality and pH of HPG solution were approximately the same as those of PYS (2.27% glucose, 395 mOsm/L, pH 7.4). The electrolye solution (denoted as Control, 251 mOsmol/kg, pH 7.4) had the same composition as HPG solution except that the HPG polymer was omitted. Both HPG and the electrolyte solutions were filter-sterilized prior to use in rats.

Chronic rat model of PD—the Wistar rats receiving once-daily intraperitoneal instillation of 10 mL of a PD solution was established as previously described [20‐22]. In brief, after anesthesia with isoflurane, 10 mL of a pre-warmed PD solution (HPG, PYS or Control) were slowly injected to the peritoneal cavity at the abdominal midline. Animals awoke within 1–2 min after the procedure, and had free access to food and tap water. Animals in Sham group were treated in the same manner including the needle puncture but without fluid injection.

After 3 months of daily exposure to the PD solutions, both bodyweight gain (BWG) and PM ultrafiltration (UF) were determined. BWG in each rat was determined as follows: BWG (%) = (final bodyweight—initial bodyweight)/initial bodyweight. The PM function was examined using UF. In brief, the UF was performed by intraperitoneal injection (IP) of 30 mL of PYS. After 4 h of dwelling time, the peritoneal fluid was recovered using a syringe, and its volume was used for calculating the UF capacity.

After UF measurement, two strips of parietal peritoneum from each rat were collected, one from left lumbar abdominal region and the other from the right region. The histological analysis of the submesothelial thickness was performed in hematoxylin and eosin (H&E) stained tissue section as described previously [18]. The number of the blood vessel and capillaries per millimeter of the peritoneum section was counted longitudinally in the PM (both submesothelial and muscle-associated connective tissue layers) of the whole tissue sections. In addition to H&E stain, the tissue sections were also stained with Masson’s trichrome method for the examination of collagen fibers.

The expression of VEGF, α-SMA and MAC387 in the tissue sections of six rats randomly selected from each group was examined using a routine immunohistochemical method as described by our lab [23].

Four rats were randomly selected from each group except of three in Sham. The parietal peritoneum tissues were collected as described in above “Histological Analyses” section, and were snap-frozen with liquid nitrogen first, and then stored in −80 °C until use. Total RNA was extracted by using mirVana™ isolation kit (Ambion, Austin, TX, USA) from the PM that was carefully pealed off from the frozen peritoneum tissue. The RNA samples with ≥8 of RNA Integrity Number (RIN) were selected for microarray analysis.

Microarray analysis was performed in the Laboratory for Advanced Genome Analysis at Vancouver Prostate Centre (Vancouver, BC, Canada) following a standardized protocol as described previously [24]. Only the genes with p ≤ 0.05 (t test) and ≥ twofold change were uploaded into IPA software (Ingenuity Systems, Redwood City, CA, USA) to examine gene enrichment pathways.

The signaling pathway search was performed using the keywords related to inflammation, such as “B cell development”, “complement”, “granulocyte/agranulocyte adhesion and diapedesis”, “antigen presentation pathway”, and so on, to identify all the genes related to these cellular functions. After that, the genes associated with these pathways were separately imported into the “Canonical Pathway” frame of the IPA software to determine whether or not the signaling pathways were significantly regulated by each treatment (HPG, PYS or Control) compared to Sham or HPG or PYS compared to Control.

A comprehensive metabolic panel of 14 blood substances, including alanine aminotransferase (ALT), albumin (ALB), alkaline phosphatase (ALP), amylase (AMY) total calcium (Ca²⁺), creatinine (Cre), globulin (Glob), glucose, phosphorus (Phos), potassium (K⁺), sodium (Na⁺), total bilirubin (TBIL), total protein (TP), and urea nitrogen (BUN), in heparinized whole blood samples were determined in blood using the VetScan^® Comprehensive Diagnostic Profile reagent rotor with the VetScan Chemistry Analyzer VS2 (Abaxis, Inc., Union City, CA, USA) in the Animal Unit Hematology Diagnostic Laboratory at Jack Bell Research Centre (Vancouver, BC, Canada).

Data were presented as mean ± standard derivation (SD) of each group. Both Student’s t test with two-tailed distribution and analysis of variance (ANOVA) (GraphPad Prism software, Inc., La Jolla, CA, USA) were used as appropriate for data analyses. A p value of ≤0.05 was considered significant.

The overall health status of rats was determined by both BWG and changes in a comprehensive metabolic panel of 14 blood substances that represent the basic metabolism (glucose) and electrolytes (Na⁺, K⁺, Ca²⁺, and Phos), and markers of both kidney and liver functions. After 3 months of daily treatment with different solutions, there was no significant difference in the BWG in Control (46.04 ± 15.88%, n = 7), PYS (39.75 ± 15.66%, n = 7) and HPG (35.01 ± 5.95%, n = 8) as compared to Sham (50.67 ± 12.42%, n = 3) using one-way ANOVA with Dunnett’s multiple comparison test (p = 0.246). The blood chemistry test was performed once every month during 3 months of the treatment. Overall, the changes of the blood substances in response to daily injection of HPG solution versus electrolyte Control (without HPG) or Sham were not significantly different (Additional file 1: Table S1). As compared to the initial levels (prior to the treatment), the treatment procedure nonspecifically (regardless of Sham or injection of any of Control, PYS and HPG solutions) induced an increase in Na⁺, TBIL and TP, and a decrease in Phos (Additional file 1: Table S1). However, as compared to Control or HPG, IP injection of PYS significantly changed the levels of Cre, ALP, ALB and Glob (including ALB/Glob ratio) during 3 months of treatment (Sham: n = 3; Control: n = 7; PYS: n = 7; HPG: n = 8). As shown in Fig. 1 or in Additional file 1: Table S1, unlike HPG, PYS increased serum levels of Cre significantly in comparison to the Control or HPG (PYS vs. Control: p = 0.0231, HPG vs. Control: p = 0.8439, HPG vs. PYS: p = 0.0269, two-way ANOVA) (Fig. 1a), ALP (PYS vs. Control: p = 0.0074, HPG vs. Control: p = 0.4325, HPG vs. PYS: p = 0.0204, two-way ANOVA) (Fig. 1b) and of Glob (PYS vs. Control: p = 0.0141, HPG vs. Control: p = 0.7534, HPG vs. PYS: p = 0.0171, two-way ANOVA) (Fig. 1c). The levels of ALB in PYS but not HPG decreased as compared to that in Control (PYS vs. Control: p = 0.003, HPG vs. Control: p = 0.7796, HPG vs. PYS: p = 0.0403, two-way ANOVA) (Fig. 1d), and more significant differences were seen from the calculation of ALB/Glob ratio among these groups (PYS vs. Control: p = 0.0022, HPG vs. Control: p = 0.5636, HPG vs. PYS: p = 0.0114, two-way ANOVA) (Fig. 1e).

The PM function was determined using UF capacity after 3 months of daily exposure to Control, PYS or HPG solution. As shown in Fig. 2, there was no significant difference in the UF of the PM between Sham (26.33 ± 3.51 mL, n = 3) and Control (26.17 ± 3.06 mL, n = 7). Statistical analysis using two-tailed t test indicated that as compared to Control group, no significant difference from HPG group was found in this limited number of animals (17.64 ± 9.34 mL, n = 8) (p = 0.0568), while the rats in PYS group lost the PM function (8.64 ± 5.14 mL, n = 7) as compared with Control (p < 0.0001) or HPG (p = 0.0453). After 4 h of dwell time only a few milliliters of dialysate were recovered in some rats, and were heavily “contaminated” with tissue debris and the blood, so that we were not able to determine the PM function using other parameters, such as the removal of the urea, creatinine, or sodium in these samples.

To investigate the pathological changes in the PM after daily exposure to different solutions (particularly PYS and HPG), histology analysis of the PM was performed after UF determination. In H&E stained sections (Fig. 3a, top panel), the number of small blood vessels and capillaries in the deeper loose adipose connective tissues, and the thickness of the submesothelial compact zones were examined. As showed in Fig. 3b, rats in PYS group had the thickest submesothelial compact zones (98.25 ± 32.73 µm, n = 14 sections of seven rats), which were significantly larger than those in Control group (53.57 ± 20.2 µm, n = 14 sections of seven rats) (p = 0.0007, two-tailed t test) or in HPG group (66.68 ± 21.11 µm, n = 16 sections of eight rats) (p = 0.0016, two-tailed t test). Also, there was no statistical difference in the thickness of the submesothelial layer between the Control and HPG groups (p = 0.1184, two-tailed t test). Similar to the changes in the submesothelial layer thickness in these groups, more blood vessels and capillaries in PYS group (3.17 ± 1.18 per mm, n = 14 sections of seven rats) were seen than those of Control group (1.47 ± 0.75 per mm, n = 14 sections of seven rats) (p < 0.0001, two-tailed t test) or HPG group (2.36 ± 1.32 per mm, n = 16 sections of eight rats) (p = 0.0454) (Fig. 3c). Furthermore, the trichrome stain indicated the deposition of collagen (blue color) within the PM. As shown in Fig. 3a (bottom panel), there was clearly an increase in collagen accumulation in all of 3 groups (Control, PYS and HPG) as compared to that in Sham, but it was difficult to quantitatively compare the differences in the collagen accumulation between Control and PYS or HPG because the integrity of collagen deposition might be affected by the test of UF.

To further understand the pathways by which PYS induced severe pathological changes in the peritoneum compared to HPG, the expression of α-SMA (a marker of myofibroblast differentiation), MAC387 (a marker of macrophages) and VEGF (a pro-angiogenic growth factor) was examined using immunohistochemical stain of peritoneal tissue sections. As shown in Fig. 4a, there were more cells stained positively with α-SMA in tissue sections of PYS group than those in HPG or Control group (15.0 ± 6.95 in PYS versus 2.71 ± 1.87 in HPG, p = 0.0019, two-tailed t test, n = 6). In addition to the positive stain of α-SMA in smooth muscle cells of the blood vessels, myofibroblasts were seen in the submesothelial layer as well as on the surface of the peritoneum in PYS group, whereas in other groups fewer of these cells were observed on the peritoneal surface. Also, there were more infiltrating macrophages in PYS than in HPG group (20.83 ± 14.03 in PYS versus 7.58 ± 3.75 in HPG, p = 0.0494, two-tailed t test, n = 6), which is consistent with more severity of peritoneal injury in PYS than that in HPG. In addition to that, the macrophages in the PYS group were localized in both inside the submesothelial layer and on its surface (Fig. 4b).

To further confirm the increase of the neoangiogenesis in the PM after chronic exposure to PYS compared to HPG, the expression of VEGF, a neoangiogenesis factor, in PM sections was examined using immunohistochemical staining. As shown in Fig. 4c, there were more VEGF-expressing cells in the submesothelial membranes in PYS group than that of HPG or Control group (58.83 ± 31.34 in PYS versus 22.8 ± 10.88 in HPG, p = 0.0307; 58.83 ± 31.34 in PYS versus 3.33 ± 0.88 in Control, p = 0.0019; two-tailed t test, n = 6 rats), which was in parallel with the most blood vessels found in the PYS group (Fig. 3c).

To further understand the molecular signaling pathways mediating the effects of daily injection of electrolyte Control, PYS or HPG solution on PM structure and function, the transcriptome-based pathway analysis was performed. Compared to Sham as a baseline there were more gene alterations in HPG group (785 up-regulated, 857 downregulated) than those in both PYS group (606 up-regulated, 507 downregulated) and Control (529 up-regulated, 439 down-regulated) (Fig. 5, upper panel). Also when HPG or PYS was compared with Control, a similar trend (HPG > PYS) was noted but with less affected gene expression (Fig. 5, bottom panel). Pathway analysis of these gene expression profilings showed that in comparison with Control, PYS or HPG with Sham, more pathways related to inflammation were significantly induced (p < 0.05) in the PM of rats receiving PYS than those in HPG or Control (Table 1), which included “Role of hypercytokinemia/hyperchemokineemia in the pathogenesis”, “IL-6 signaling”, “Toll-like signaling”, “HMGB1 signaling”, and so on. Also the p values of the pathways activated in all of three groups, such as “Granulocyte or agranulocyte adhesion and diapedesis” and “Acute phase response signaling” reached a higher statistical significance in PYS group than those in HPG or Control group (Table 1). The higher activation of the inflammatory pathways in PYS group was consistent with a significant up-regulation of several well-known pro-inflammatory mediators, such as TNF, IL-1β, CCL2 and MMP9, which were not seen in either HPG or Control group (Tables 2, 3, 4). When we compared PYS or HPG with Control, there were still more inflammatory pathways being significantly affected by glucose-based PYS than those seen in HPG group (nine in PYS versus five in HPG) (Table 5), which included “Acute phase response signaling”, “Inhibition of matrix metalloproteinase” and IL-6 signaling”. Similarly, the highly up-regulated TNF and MMP9 were found in PYS but not in HPG when Control was used as a reference in analysis (Tables 6, 7). All these data implied that TNF and/or MMP9-dependent pathways were specifically activated in the PM by chronic exposure to PYS.

In addition to the pathways mediating inflammation as listed in Table 1, the analysis of global canonical pathways showed the function of “Triacylglycerol degradation”, CDP diacylglycerol biosynthesis I”, “Phosphotidylglycerol biosynthesis II (non plastidic)” and “Triacyglycerol biosynthesis” was found only in the MP of rats receiving HPG (Table 8), which may be related to the local metabolism of this polymer.