Inhibition of TNF-α Improves the Bladder Dysfunction That Is Associated With Type 2 Diabetes

The floxed Irs1, Irs2 mice were generated as previously reported (10,11,13). To generate liver-specific knockout mice, floxed Irs1 and Irs2 mice were crossed with albumin-Cre mice (The Jackson Laboratory). The Irs1KO, Irs2KO and their corresponding control floxed mice were maintained on a C57/BL6 and 129Sv mixed genetic background. To avoid any potential obstructive effect of the prostate on the bladder, we used female DKO animals and their female littermates as control for these studies. All animal experiments were performed according to procedures approved by the Institutional Animal Care and Use Committee at Massachusetts General Hospital, Children’s Hospital Boston.

General anesthesia was induced by intraperitoneal injection of ketamine (100 mg/kg) and xylazine (10 mg/kg), the animal’s bladder was surgically exposed, and a 25-gauge needle was introduced into the dome of the bladder and connected via a three-way adapter to a fluid-filled pressure line at one end and an injection pump (Harvard Apparatus) at the other. The pressure line was connected to a physiologic pressure transducer (MLT844 AD Instruments), and the bladder was filled with sterile saline at a constant rate (15 μL/min). The output signal was amplified and recorded using the Windaq system (Dataq Instruments) (14,15).

Mice were placed individually in metabolic cages with available food and water and allowed to adapt to their new environment for 24 h. Urine output was measured by evaluating the surface area of the stained VSOP paper (Bio-Rad) 3 h after the mice were placed in the metabolic cages. Collected papers were imaged under ultraviolet light to visualize the urine and analyzed by using the edge-detection function of ImageJ software (National Institutes of Health [NIH]) to identify the surface area of individual voided urine spots. The volumes of individual voids were calculated based on a calibration curve relating surface area to fluid drops of known volume (14,15).

Full-thickness longitudinal detrusor muscle strips (0.7 to 1 × 3.5 to 5 mm) were prepared and transferred to 5 mL tissue baths containing Kreb’s solution bubbled with a mixture of oxygen (95%) and carbon dioxide (5%). One end of each strip of tissue was tied by silk suture to a fixed hook and the other to a force transducer. Bladder tissue was stretched to a passive tension of 0.5 g and allowed to equilibrate for 60 min before experiments were performed. The frequency and amplitude of spontaneous activity was determined over a 5-min interval after the equilibration period. In addition, the contractile response to stimulation of bladder tissue by potassium chloride (KCl) (120 mmol/L) and electrical field stimulation (EFS; 2, 4, 8, 16, 32, 64 Hz; 40 V; 0.5 ms pulse duration for 10 s) was measured. The dose-response curve for carbachol was generated by adding cumulative concentrations of the agonist (10⁻⁹ to 10⁻⁵ mol/L) to the bath. Isometric force was continuously recorded with a computerized data acquisition program at a sampling rate of 20 Hz. At the end of experiments, the weight, length, and width of each strip was measured to normalize the force data (16).

The Affymetrix GeneChip analysis was performed as previously described (10,11). Briefly, total RNA was extracted from bladder smooth muscle tissues using the RNeasy Mini Kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Microarray analysis was conducted by the Harvard Partners HealthCare Center for Personalized Genetic Medicine Microarray Facility using 15 μg RNA from each sample, with each group in triplicate. Gene expression was analyzed by using Mouse Genome 430 2.0 array system (Affymetrix). The Database for Annotation, Visualization and Integrated Discovery (DAVID Bioinformatics Resources 6.7, NIH) was used to categorize the most significantly changed genes (upregulated 1.5-fold or more relative to age-matched control) by biologic functions or disease processes.

Rat BSMCs were isolated from 1-week-old Sprague-Dawley female rats, as described previously (17). The cell pellet was resuspended in medium 199 (Gibco) and used for experiments up to passage 3. For gene suppression experiments, BSMCs were transfected with small interfering (si)RNA at 40% confluency using silMPORTER transfection reagent (Millipore), according to the manufacturer’s instructions.

Collagen lattice was carefully prepared in 24-well plate by collagen type 1 from rat tail (Sigma) (18). A total of 25,000 BSMCs were layered onto each collagen lattice and left to settle for 2 h at 37°C. To allow contraction, each collagen lattice was detached from the bottom of the well with a small spatula and left overnight at 37°C. Contraction was stopped by fixing the lattices in PBS-4% paraformaldehyde, and the plate was photographed. Image analyses of the gel area were performed by using ImageJ software (NIH).

Immunoblotting was performed as previously described (19). To stimulate the phosphorylation of MLC, bladder smooth muscle tissues were treated with 80 mmol/L KCl for 15 s and then quickly frozen in liquid nitrogen (20), and cultured BSMCs were treated with 80 mmol/L KCl for 5 min. Total protein was quickly extracted on ice (21), and the protein extracts were subjected to SDS-PAGE and blotted. Membranes were probed with target antibodies.

As described previously (22), sections (5 μm) from paraffin-embedded tissue underwent antigen retrieval with proteinase K (Dako) and were incubated with primary and secondary antibody. Cells grown on glass coverslips were fixed in 4% paraformaldehyde, permeabilized with 0.2% Triton X-100/PBS, and stained with antibodies (23).

Activation of ROCK1 or ROCK2 was evaluated by immunoprecipitating ROCK1 or ROCK2, followed by an in vitro kinase activity assay (24). Cells were lysed using cell lysis buffer (Cell Signaling), and cleared lysates were incubated for 1 h at 4°C with 2 μg isotype control IgG, anti-ROCK1, or anti-ROCK2 (Santa Cruz). The mixture was incubated with protein A/G agarose beads at 4°C overnight. The beads were pelleted, washed, and incubated for 30 min at 30°C with kinase reaction buffer containing 200 μmol/L ATP (Cell Signaling) and 0.5 μg purified myosin phosphatase target subunit 1 (MYPT-1; Millipore), a ROCK-specific substrate. The reaction was terminated by boiling the samples in SDS. The amount of phosphorylated MYPT-1 or immunoprecipitated ROCK1 or ROCK2 was examined using immunoblot.

ROCK enzymatic activity was determined with a commercially available Cyclex Rho kinase Assay Kit-96 (MBL International) (25).

Cytokine productions were analyzed with a Milliplex Map Kit for mouse cytokine assay (Millipore) according to the manufacturer’s instructions, which allowed the simultaneous quantification of the following mouse cytokines: granulocyte-macrophage colony-stimulating factor, interferon-γ, interleukin (IL)-1β, IL-2, IL-4, IL-5, IL-6, IL-10, IL-12(p70), and TNF-α. In anti-TNF receptor (TNFR) treatment experiments, the level of TNF-α in serum was determined using commercial mouse TNF-α ELISA Kit (eBioscience) according to the manufacturer’s protocol.

Blood glucose levels were measured using a portable glucometer (Bayer). For glucose tolerance tests (GTT), mice were fasted overnight and injected intraperitoneally with d-glucose (2 g/kg body weight), and blood glucose levels were measured at indicated time points, as previously described (10).

We first evaluated whether the DKO mice exhibit alterations in bladder function and found that DKO mice exhibited significantly higher amplitudes of spontaneous activity at age 6 and 12 weeks relative to age-matched controls (Fig. 1A and B). In contrast, 16- and 20-week-old DKO mice demonstrated lower amplitudes of spontaneous activity than age-matched controls (Fig. 1A and B). Meanwhile, the frequency of spontaneous activity in the bladders did not vary (Fig. 1C). The contractile responses of bladder tissue to KCl (120 mmol/L; Fig. 1D) and carbachol stimulation (10⁻⁹ to ∼10⁻⁵ mol/L; Fig. 1E and F) were significantly greater in young DKO mice (age 6–12 weeks), but significantly diminished in 20-week-old DKO mice compared with age-matched control animals. However, the response to carbachol stimulation in control mice was not different among age groups (Fig. 1F), which ruled out the possibility of age-dependent changes in cholinergic receptor sensitivity. We further induced neurogenically mediated bladder contractions by EFS (26,27) and found that the contractile response to EFS was augmented in bladder tissue of 6- and 12-week-old DKO mice but diminished in older mice at ages 16 and 20 weeks (Supplementary Table 1).

To gain further insight into the development of bladder dysfunction associated with type 2 diabetes, we evaluated the urinary voiding patterns in the DKO mice in vivo by performing cystometry and VSOP analysis. During cystometry, the voiding pattern in DKO mice was markedly altered. Although the micturition frequency did not change significantly (Fig. 2A and B), 12-week-old DKO animals exhibited clear signs of bladder overactivity, characterized by increased frequency of nonvoiding contractions and significantly reduced voided volumes (Fig. 2C and E). For 20-week-old DKO animals, however, the amplitude of bladder contractions during voiding was lower than for age-matched controls (Fig. 2D). In comparison, we found a normal pattern of bladder contraction and voiding in 12- or 20-week-old control mice (Fig. 2A), which is in line with other reports (14,15).

In the VSOP test, freely moving floxed control animals showed a normal voiding pattern: voiding occurred around the edge of the cage, and few voided urine spots appeared in the middle of the cage (Fig. 2F) (15). However, the voided stains for the DKO mice were much more dispersed, with multiple urine spots away from the edge, suggesting an abnormal voiding pattern (15) (Fig. 2F). Notably, the micturition volumes were lower and the frequency of small volume voids (<10 μL) was higher in DKO mice than in age-matched controls, especially for 12-week-old DKO mice (P < 0.01) (Fig. 2G and H). Taking the cystometric and VSOP data together, DKO animals developed abnormal voiding behavior with increased frequency of nonvoiding contractions and significantly reduced voided volumes at the age 12 weeks, which exhibited clear signs of bladder overactivity. For 20-week-old DKO animals, the low amplitude of bladder contractions in tissue and cystometric tests demonstrated decreased contractility at this stage.

Using the Affymetrix GeneChips, we compared 45,000 genes in bladder smooth muscle between the diabetic DKO and control animals. We found 30 genes that were upregulated 1.5-fold or greater in 12-week-old DKO animals relative to age-matched controls (Supplementary Table 2). Of these, 20 genes were inflammatory mediator genes that belonged to the TNF family and were also persistently elevated in 20-week-old mice. Because inflammation is known to play an important role in the pathogenesis of diabetic organ damage (28), we wished to determine whether TNF-α plays any role in DBD. We found that the serum level of TNF-α was elevated in 12- and 20-week-old DKO animals compared with age-matched controls (Fig. 3A), although the protein levels of 10 cytokines in urine and homogenized bladder tissue were below the limit of detection by the Milliplex kit assay. Western blot analysis also showed that TNF-α and TNFR1 were significantly upregulated in DKO bladder tissue, whereas TNFR2 expression did not change (Fig. 3B and C).

Immunohistochemistry staining for the macrophage-specific marker F4/80 further showed that F4/80⁺ macrophages significantly increased in DKO animal bladders in an age-dependent manner (Fig. 3D and E). Interestingly, immunoreactivity for macrophages localized in the lamina propria (i.e., junction between smooth muscle and epithelial layers; Fig. 3D). Together, these data suggest that circulating and bladder-localized inflammatory mediators are upregulated in DKO animals.

Microarray analysis of bladders of 12-week-old DKO animals demonstrated increased expression in the genes encoding ATPase, Rho GTPase, and Rho kinase, which are known as important regulators of cellular metabolism and contractility, but decreased expression in 20-week-old DKO mice (Supplementary Table 2). Therefore, we next evaluated Rho kinase protein expression in bladder smooth muscle tissue. Western blot analysis showed that both ROCK isoforms, ROCK1 and ROCK2, were significantly upregulated in 12-week-old mice. In 20-week-old DKO animals, ROCK2 was significantly downregulated compared with age-matched controls (Fig. 4A and B). We measured ROCK activity in bladder homogenates and further observed that the kinase activity was elevated in 12-week-old but reduced in 20-week-old DKO animals (Fig. 4D). Consistent with these changes, the level of phosphorylated MLC₂₀ (pMLC₂₀) was increased in 12-week-old DKO mice but decreased in 20-week-old DKO mice (Fig. 4C). These findings suggest that the temporal changes in expression levels of ROCK1 and ROCK2 correlated with the hyperactivity and hypoactivity states of bladders in young and old DKO animals, respectively.

We next addressed whether TNF-α directly contributed to the bladder dysfunction. Growing BSMCs from DKO and control mice in primary tissue cultures was challenging because we could not obtain an adequate number of cells for in vitro experiments. Therefore, we used primary rat BSMCs, which we were able to passage in culture. Administration of 1–30 ng/mL TNF-α for 2.5 h dramatically promoted the contraction of BSMC in a dose-dependent manner (Fig. 5A) and upregulated pMLC₂₀ and the phosphorylation of calmodulin (pCalm; Fig. 5B and C). Interestingly, high glucose (22 mmol/L) treatment alone, even for 24 h, did not promote cellular contraction (Fig. 5D). Immunostaining with pMLC₂₀ antibody further confirmed that TNF-α stimulated the phosphorylation of MLC₂₀ (Supplementary Fig. 1A and B). However, pretreatment of cells with MLCK inhibitor P18 or ROCK inhibitors (Y27632 and Fasudil) for 30 min before administration of TNF-α significantly inhibited the ability of TNF-α to promote cell contraction (Fig. 5D and E) and suppressed the level of pMLC₂₀ (Supplementary Fig. 1C and D). Our results suggest that TNF-α may augment cellular contraction through activation of the calcium-dependent Rho kinase–MLCK–pMLC pathway.

To further investigate the signaling pathway affected by TNF-α in BSMCs, we immunoprecipitated ROCK1 and ROCK2 from BSMCs and evaluated their activity through in vitro kinase activity assay using purified MYPT-1 as a substrate. Antibodies against ROCK1 or ROCK2 efficiently immunoprecipitated their presumed target (Fig. 6A). ROCK1 and ROCK2 were able to phosphorylate MYPT-1 in the kinase assay, and this phosphorylation could be abolished by Y27632, the Rho kinase inhibitor (Fig. 6B). Compared with untreated cells, the activity of ROCK1 and ROCK2 was increased by TNF-α stimulation between 5 min and 3 h, indicating that both ROCK isoforms are activated by TNF-α (Fig. 6C and D). However, downregulating of ROCK1 and ROCK2 with ROCK1 siRNA or ROCK2 siRNA alone, or combined, prevented the TNF-α stimulated increase in phosphorylation of MLC₂₀, especially at 6 h (Fig. 6E–I). Together, these data suggest that ROCK1 or ROCK2 is necessary to induce MLC₂₀ phosphorylation in response to TNF-α stimulation.

On the basis of these findings, we wished to investigate whether anti–TNF-α treatment would be a target for treating diabetes-related bladder dysfunction. Systemic inhibition of TNF-α activity with soluble TNFR1 (TNFRI) in DKO mice resulted in a significantly decreased frequency of nonvoiding contractions and an increase in voided volumes compared with untreated DKO mice (Fig. 7A, C, and D), although the frequency of micturition and amplitude of voiding contractions were not changed significantly (P > 0.05, Fig. 7B and E). Administration of TNFRI also improved the voiding behavior of freely moving animals, as reflected by a more controlled and organized voiding pattern, less dispersed urine spots, larger micturition volume, and lower frequency of small volume voids (<10 μL; Fig. 7F–H).

When determining whether the observed bladder dysfunction was associated with the concentration of circulating TNF-α, we found the serum TNF-α level in DKO mice was significantly reduced after 6 weeks of treatment and approached a level similar to that of the floxed control animals (Fig. 7I). In addition, the elevated ROCK activity as well as ROCK1 and ROCK2 protein expression in DKO bladder smooth muscle homogenates were suppressed (Fig. 7J). Consistent with these effects, anti–TNF-α treatment not only suppressed the TNF-α expression in bladder smooth muscle but also lowered the expression of pMLC₂₀, pCalm, ROCK1, ROCK2, and Rho A (Fig. 7K). Meanwhile, anti–TNF-α treatment alone did not lower the blood glucose level in DKO mice (Fig. 7L–N). However, administration of the antidiabetes drug, metformin, alone or combined with TNFRI, significantly suppressed the blood glucose level after 6 weeks of treatment as assessed by the GTT assay (Fig. 8A–C). More important, TNFRI and metformin synergistically decreased the serum TNF-α level (Fig. 8D) and improved the bladder dysfunction after 6 weeks of treatment, as shown by a significant decrease in frequency of nonvoiding contractions during urodynamic testing (Fig. 8E) and a decreased number of small-volume voids in VSOP analysis (Fig. 8F). Our results demonstrate the significance of targeted molecular treatment for correcting or preventing the secondary complications of diabetes.

DBD affects 80% of diabetic patients; however, the molecular mechanisms underlying DBD remain elusive due to the lack of appropriate animal models. Bladder contraction is mediated by cholinergic and purinergic pathways (26,29). Here, we found phenotypic differences in the behavior of isolated bladder strips in DKO animals. In particular, the spontaneous activity in early-stage DKO mice (age 6 and 12 weeks) was significantly increased, but decreased at the late stage (age 20 weeks). These temporal changes in spontaneous smooth muscle activity and contractile responses to EFS, KCl, and carbachol stimulation are consistent with a compensation–decompensation progression of bladder function in this animal model of type 2 diabetes, analogous to urodynamic findings in patients with type 2 diabetes (2,30). Because all tension data were normalized for the length, width, and weight of the muscle strips, the response to stimuli reflects an inherent change in the bladder smooth muscle. Notably, we observed a parallel alteration in response to EFS and direct receptor activation by carbachol, suggesting a myogenic basis for the abnormal contraction of DKO bladder smooth muscle. However, our findings do not exclude a neuropathic change in bladder innervation that may play a role in DBD in DKO animals.

Previous studies using STZ-induced type 1 diabetes models in rats and mouse, or the Zucker type 2 diabetic and obese rat model, have shown that diabetes in these animal models is associated with increased bladder weight, prolonged duration of bladder muscle contraction, and increased residual rine volume (31). Work from Dr. Damaser’s Urological Biomechanics Laboratory (Cleveland Clinic) has further shown that the voiding dysfunction was only evident in Zucker obese diabetic rats but not in Zucker nonobese diabetic rats (31). In our study, however, we did not find any statistical difference in the bladder-to-body weight ratio between DKO and age-matched control mice. We observed clear signs of bladder overactivity for the early stage (6- and 12-week-old) DKO mice, and hypoactivity in older (20-week-old) DKO animals. The different mechanism in DKO mice versus Zucker rats might cause this functional discrepancy in diabetic animal bladders. In the Zucker rat model, fibrosis of the external urethral sphincter, bladder wall edema, and vasculopathy attributed to the bladder dysfunction in the obese animals (31). In our DKO model, however, inflammation was the main finding, especially elevated TNF-α systemically and locally in the bladder, but without any identifiable gross histologic changes, changes that are more compatible with human patients with type 2 diabetes (32).

Studies have shown that diabetes-induced diuresis might contribute to the bladder dysfunction by altering the nerves and vasculature of the urinary bladder (7,9). We did not evaluate diuresis because the voiding frequency was equivalent in diabetic and nondiabetic animals and the voided volume was even reduced in DKO diabetic mice. In addition, absence of a diuretic response in our model could explain why the bladder-to-body weight ratios were not changed in our model. However, there is significant diuresis in type 1 diabetic animal models that can partly explain the bladder hypertrophy and increased bladder-to-body weight ratios (33).

The initiation of smooth muscle contractility is predominantly controlled by a calcium-dependent increase in MLC₂₀ phosphorylation (34). Activation of Rho A may be of particular importance and lead to subsequent activation of Rho kinase in BSMCs (35). Here we observed that TNF-α not only promoted the contraction of BSMCs but also activated ROCK1 and ROCK2 and induced the phosphorylation of MLC₂₀, which suggests a close link between TNF-α and the ROCK-signaling pathway in DBD. This association was further confirmed by anti–TNF-α treatment, which not only improved bladder function but also suppressed the TNF-α level and ROCK pathway. Generally, these observations support an important role for TNF-α in DKO bladder dysfunction and indicate that TNF-α may be responsible for the observed phenotypic differences in isolated bladder tissue and in vivo.

Given that TNF-α plays a key role in the pathogenesis of DBD, anti-TNF therapy was expected to provide protection against the toxicity of TNF-α (36). Anti–TNF-α reagents significantly improved insulin resistance and reduced leukocyte adherence in retinal blood vessels of diabetic rats (37–39). In humans, however, the importance of this mechanism is much debated because limited studies of anti–TNF-α reagents have shown little or no effect on the insulin-resistant state (40). Our experimental findings match the clinical findings of Ofei et al. (40), who showed no significant alteration of hyperglycemia with anti–TNF-α treatment. However, we observed significant improvements of the parameters associated with secondary contractions of diabetes as measured by bladder function (Fig. 7). Moreover, our data further suggest that the benefit of anti–TNF-α therapy on DBD might be achieved through blockade of TNF-α–induced ROCK-MLC signaling pathway and improvement of abnormal BSMC contractility. Notably, our findings further indicate that targeted inhibition of TNF-α combined with the oral hypoglycemic agent, metformin, can synergistically improve the abnormalities associated with DBD.

In conclusion, our study provides the first demonstration that DKO mice represent a novel animal model to investigate DBD associated with type 2 diabetes. We demonstrate that the TNF-α–ROCK–pMLC pathway may represent a new target to treat patients with DBD dysfunction and provide proof of principle that pharmacologic inhibition of TNF-α can mitigate the urinary tract complications of type 2 diabetes without affecting serum glucose levels. Furthermore, the synergistic benefit of metformin and TNFRI in reversing DBD has significant clinical implications in long-term management of patients with type 2 diabetes.

This work was funded by National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases grant (Animal Models of Diabetic Complications Pilot & Feasibility Project, 09MCG72) and the American Urological Association/Pfizer Pharmaceutical Company competitive grants to A.F.O. No other potential conflicts of interest relevant to this article were reported.

Z.W. designed and performed research, analyzed the data, developed the animal model, and wrote the manuscript. Z.C. and M.P.S. performed research, analyzed the data, and wrote the manuscript. V.C. performed research and analyzed the data. J.L., X.X., P.G., E.G., and K.S. performed research. R.G. analyzed the data. R.M.A. analyzed the data and wrote the manuscript. M.F.W. developed the animal model. A.F.O. designed the research and wrote the manuscript. Z.W. and A.F.O. are the guarantors of this work, and as such, had full access to all the data in the study and take responsibility for the integrity of data and the accuracy of the data analysis.