Danegaptide Prevents TGFβ1-Induced Damage in Human Proximal Tubule Epithelial Cells of the Kidney

Chronic kidney disease (CKD) is a global health problem associated with a number of comorbidities. Recent evidence implicates increased hemichannel-mediated release of adenosine triphosphate (ATP) in the progression of tubulointerstitial fibrosis, the main underlying pathology of CKD. Here, we evaluate the effect of danegaptide on blocking hemichannel-mediated changes in the expression and function of proteins associated with disease progression in tubular epithelial kidney cells. Primary human proximal tubule epithelial cells (hPTECs) were treated with the beta1 isoform of the pro-fibrotic cytokine transforming growth factor (TGFβ1) ± danegaptide. qRT-PCR and immunoblotting confirmed mRNA and protein expression, whilst a cytokine antibody array assessed the expression/secretion of proinflammatory and profibrotic cytokines. Carboxyfluorescein dye uptake and ATP biosensing measured hemichannel activity and ATP release, whilst transepithelial electrical resistance was used to assess paracellular permeability. Danegaptide negated carboxyfluorescein dye uptake and ATP release and protected against protein changes associated with tubular injury. Blocking Cx43-mediated ATP release was paralleled by partial restoration of the expression of cell cycle inhibitors, adherens and tight junction proteins and decreased paracellular permeability. Furthermore, danegaptide inhibited TGFβ1-induced changes in the expression and secretion of key adipokines, cytokines, chemokines, growth factors and interleukins. The data suggest that as a gap junction modulator and hemichannel blocker, danegaptide has potential in the future treatment of CKD.


Introduction
Chronic kidney disease (CKD) is a growing health concern associated with increased risk of cardiovascular disease and morbidity [1]. Estimated to affect 10% of the global population [2], risk factors include age, diabetes, hypertension, dyslipidaemia and obesity [3]. The disease is characterised by a decline in the glomerular filtration rate (GFR), in addition to proteinuria, with glomerulosclerosis, tubular atrophy and tubulointerstitial fibrosis (TIF), which are common histopathological changes [4]. Culminating in the loss of epithelial stability, persistent inflammation and increased deposition of the extracellular matrix [5,6], treatment of TIF and advanced CKD represents an unmet clinical need [3]. Consequently, urgent therapeutic approaches are required.
Altered connexin (Cx) expression and function have been implicated in the pathology of various forms of disease (as reviewed in [7]), including CKD (as reviewed in [8,9]). Connexins are a family of membrane-bound proteins that oligomerise into hexameric assemblies termed connexons [10]. When neighbouring cells align, connexons dock to The current study investigated the efficacy of danegaptide in negating connexinmediated hemichannel ATP release in primary TGFβ1-treated human proximal tubule epithelial cells (hPTECs), ahead of delineating its protective role against the expression and functional changes commonly associated with progression of tubulointerstitial fibrosis. Using techniques to determine changes in expression (qRT-PCR, immunocytochemistry, immunoblotting, antibody arrays) and function (ATP biosensing, dye uptake, transepithelial electrical resistance, antibody arrays), our findings demonstrate that nanomolar concentrations of danegaptide block Cx hemichannel-mediated ATP release and partly attenuate TGFβ1-induced changes in the expression of cell cycle proteins (e.g., p16, p21, cyclin D1) and reno-protective factors (e.g., Klotho). Moreover, danegaptide decreases the changes often observed in tubular injury, including disassembly of adherens junctions (including a loss of E-cadherin) and tight junctions (including a loss of zona occludens 1 (ZO-1) and claudin 2). Importantly, the proteome profiler array demonstrated the ability of danegaptide to restore changes in the expression and secretion of key extracellular matrix proteins, adipokines, chemokines and growth factors induced by TGFβ1. Consequently, our in vitro data suggest that danegaptide (50-100 nM) can successfully block hemichannelmediated ATP release in primary hPTECs and partially protect against changes associated with inflammation and fibrosis, as observed in late-stage CKD.
In panel (C), transepithelial electrical resistance (TER) assessed the consequence of altered adherens and tight junction protein expression on epithelial integrity. HK2 cells were cultured in low glucose (5 mM) on Transwell inserts and transepithelial resistance measured. TGFβ1 reduced TER, an effect partially restored by the addition of danegaptide (100 nM). Data are expressed as the mean ± SEM (n = 3; *** p < 0.001).

Danegaptide Reduces TGFβ1-Evoked Changes in the Expression of Adipokines, Chemokines, Growth Factors and Interleukins from hPTECs
The inflammatory response in and around proximal tubules involves both the activation of multiple cell types and the secretion of numerous inflammatory markers. Specifically, soluble chemokines, cytokines and growth factors recruit and activate infiltrating immune cells and stimulate resident fibroblasts. Sustained activation of these cells mediates tubulointerstitial fibrosis. We used the proteome profiler array to determine whether danegaptide negates TGFβ1-induced changes in the expression and secretion of key proinflammatory mediators. Primary hPTECs were cultured, as described above, and treated with TGFβ1 (10 ng/mL) ± danegaptide (100 nM) for 48 h. A list of changes in lysate ( Figure 6) and supernatant (Figure 7) are provided for 31 candidate proteins grouped by primary function. Figure 6. Danegaptide prevents TGFβ1-evoked changes in the expression of adipokines, chemokines, growth factors and interleukins. Primary hPTECs were cultured in low glucose (5 mM) ± TGFβ1 (10 ng/mL) ± danegaptide (100 nM) for 48 h. An inflammation antibody array was used to assess the expression of 31 candidate inflammatory proteins in hPTEC lysates. Results are representative of 3 separate experiments and presented as the mean ± SEM (n = 3), with key significances indicated (* p < 0.05, ** p < 0.01, *** p < 0.001; one-way ANOVA and Tukey's post-test).

Discussion
In 2017, the global incidence of CKD was at 697.5 million cases, with the condition identified as the 12th-leading cause of death worldwide [53]. To date, there are no curative therapeutic options for end-stage renal disease (ESRD). Furthermore, with low dialysis survival rates and a global shortage of kidney donors, there is a pressing need to develop therapies that could prevent CKD progression and improve the patient's quality of life.
In CKD, the severity of inflammation in the proximal tubule dictates how quickly a kidney will fail. Contributed to by the activation of multiple cell types, it appears, in part, to involve connexin-mediated cell-to-cell communication [9]. Connexins facilitate direct and local paracrine-mediated cell communication through their ability to oligomerise into hexameric connexons. When neighbouring cells align, connexons dock to form gap junctions [10]. These continuous channels provide a direct route for information transfer, allowing cells to lock into a particular frequency and synchronise activity. In contrast to gap junctions, which typically open under physiological conditions, undocked connexons, also termed hemichannels, have a low open probability and open in response to injury [10]. Altered hemichannel activity is associated with the pathophysiology of multiple disease states, with evidence linking altered cell-to-cell communication to increased senescence [54], inflammation [34,55,56] and fibrosis [18][19][20][21]. Data from our own lab demonstrate that the predominant connexin isoform in the proximal tubule (Cx43) is up-regulated in advanced CKD and correlates with elevated levels of TGFβ1 and severity of fibrosis and inflammation [24]. Moreover, this altered expression translates to a loss of gap junction intercellular communication, which is accompanied by increased hemichannel-mediated ATP release and changes indicative of early tubular injury. Consequently, blocking Cx43-mediated hemichannel ATP release may represent a viable target for treatment of the late-stage damage that develops in individuals with CKD.
Danegaptide is a small therapeutic peptide that until recently has been used in its capacity to restore gap junction coupling in multiple models of injury, including ischaemia [43,44] and retinopathy [49]. Danegaptide has been shown to block hemichannel activity and dye uptake in C6 glioma cells [43]; however, a role for the compound in blocking hemichannel-mediated ATP release remains to be confirmed, and no studies have yet evaluated the effects of danegaptide on injured kidney tubules. In the current study, we presented compelling evidence that danegaptide blocks TGFβ1-induced hemichannelmediated ATP release in primary human proximal tubule epithelial cells. Furthermore, the compound restored TGFβ1-evoked changes in the expression and secretion of proteins linked to inflammation and fibrosis. In CKD, tubulointerstitial fibrosis develops in response to various morphological and phenotypic changes, including epithelial-to-mesenchymal transition (EMT), inflammatory cell infiltration, fibroblast activation and extracellular matrix (ECM) remodelling [57]. Recent evidence suggests that cellular senescence may play a key role in the progression of chronic kidney disease [58], with senescence linked to EMT, a proinflammatory secretome, and extracellular matrix deposition [59][60][61][62]. Senescence denotes irreversible proliferative growth arrest with associated changes in chromatin organisation, gene transcription and protein secretion [63,64]. Consequently, senescent cells are known to exhibit increased expression of cyclin-dependent kinase (CDK) inhibitors (CKIs), including p21 Cip1 (p21) and p16 Ink4a (p16), and altered expression of reno-protective Klotho [65][66][67][68]. In the current study, mRNA expression of CDK inhibitors p16, p21 and cyclin D1 was significantly elevated in TGFβ1-treated tubular epithelial cells, whilst the reno-protective anti-aging protein Klotho demonstrated reduced expression compared to controls. Interestingly, the expression of p16, p21 and cyclin D1 was ameliorated when cells were co-incubated with danegaptide. The importance of this observation is supported by numerous in vivo studies [69,70], including recent work in the p16 INK4a double-knockout mouse model. When induced with UUO to exhibit advanced interstitial inflammation and fibrosis, these mice exhibited decreased apoptosis, senescence, diminished levels of TGFβ1/Smad signalling and a reduction in inflammatory cell infiltration as compared to wild-type animals [70]. Moreover, proximal tubule cells isolated from the UUO model exhibited increased levels of p21, suppression of which is paralleled by a reduction in markers commonly associated with EMT [71,72]. A direct link to cell senescence and induction of EMT has already been established [73].
The loss of E-cadherin-mediated cell adhesion is an initiating trigger of EMT, with a concomitant increase in N-cadherin (the cadherin switch), accompanied by disassembly of the adherens junction and acquisition of proteins commonly associated with a fibroblast phenotype, e.g., vimentin and fibroblast-specific protein [74]. In the kidney, it is well established that tubular injury evokes a number of morphological and phenotypic changes characteristic of partial, if not full, EMT. Our recent studies demonstrated that glucoseevoked changes in TGFβ1 mediate disassembly of the adherens and tight junction complex by regulating changes in adherens (ECAD, NCAD, β-catenin) and tight junction (ZO-1) protein expression. Moreover, in vitro administration of non-hydrolysable ATP downregulated E-cadherin expression in proximal kidney cells, the loss of which was paralleled by a reduction in intercellular ligation forces, decreased tether rupture events and cytoskeletal remodelling [51]. Mediated by P2X7R, the effects were reversed in the heterogenous Cx43 +/-UUO model [12]. Building on these findings, the effectiveness of danegaptide underscores a protective role for blocking hemichannel-mediated ATP release in preventing altered expression of key epithelial (ECAD, β-catenin, ZO-1, claudin 2) and mesenchymal (NCAD, vimentin) proteins, the implications of which are of therapeutic interest.
The adherens junction assembles when the cytoplasmic tail of E-cadherin binds to β-catenin, mediating attachment to the cytoskeletal network and ensuring that cell polarity and architecture are maintained [75]. Despite its role in maintaining cell polarity, disassembly of these junctions allows for the release of β-catenin into the cytosol, which when activated by either Wnt proteins or other upstream regulators, e.g., integrin-linked kinase (ILK), can translocate into the nucleus and regulate cell-specific effects. Integrin-linked kinase is an intracellular serine/threonine protein kinase that plays a fundamental role in the regulation of cell adhesion, survival, proliferation and extracellular matrix (ECM) deposition [76]. Importantly, inhibition of ILK attenuates renal fibrosis in multiple models of CKD [77]. In the current study, TGFβ1-induced increases in ILK1 expression were, in part, restored when cells were co-incubated with danegaptide, an effect that paralleled a decrease in β-catenin expression and ECM-related proteins. Interestingly, whilst danegaptide failed to significantly alter the TGFβ1-induced increase in β-catenin expression, it is important to note that the absence of a change in expression does not reflect the absence of its function within the cell. The canonical Wnt pathway involves the nuclear translocation of β-catenin and activation of target genes via a group of transcription factors called the T cell factor/lymphoid enhancer factors (TCF/LEF) [78]. Normally, Wnt/β-catenin signalling is silent, but it is reactivated after injury in a number of different models of chronic kidney disease [79,80]. Moreover, aberrant Wnt/β-catenin signalling is highly associated with the initiation of senescence, an effect that parallels the loss of expression of Klotho, an anti-aging protein that acts as a negative regulator of the canonical Wnt pathway [81], and in the current study was downregulated in TGFβ1-treated cells (Figure 3). Consequently, sustained activation of Wnt/β-catenin signalling is linked to the progression of fibrosis in both the kidney and other tissue types [79,82,83]. Once translocated into the nucleus, β-catenin is associated with increased expression of the transcription factor SNAIL and the matrix metalloproteinase MMP7. Interestingly, these events are associated with transcriptional repression and extracellular domain membrane shedding of E-cadherin, the latter of which sees an increase in the cytoplasmic pool of free β-catenin [84]. Accordingly, Wnt signalling is a key activator of EMT in conditions of injury [85]. Moreover, with evidence that Wnt/β-catenin signalling promotes the expression of numerous genes, including fibronectin, FSP1, Snail1, MMP7 and cyclin D1, it is not surprising that attenuation of this signalling cascade is associated with improved renal function, reduced EMT [86] and diminished inflammation [87] and fibrosis [80,88]. Consequently, it could, in part, account for the TGFβ1-induced increased expression and secretion of extracellular matrix proteins ( Figure 5), proinflammatory cytokines and chemokines observed in this study ( Figures 6 and 7). Increased ECM deposition and secretion of proinflammatory molecules are hallmarks of tubulointerstitial fibrosis, with different cell types, including renal tubular epithelial cells known to secrete a large number of factors that are collectively defined as the CKD-associated secretory phenotype (CASP). Characterised by numerous molecules, including interleukins, extracellular matrix proteins and chemokines, this CASP bares striking similarities to the senescence-associated secretory phenotype (SASP), with components of the CASP having been strongly associated with the pathology of tubulointerstitial fibrosis [89]. When released, these proinflammatory factors act upon neighbouring healthy cells in a paracrine fashion, thereby driving the progression of fibrosis in CKD. As many factors associated with the CASP are known to induce fibrosis in the kidney (e.g., TGFβ, interleukin (IL)-1, and interleukin (IL)-6), targeting upstream of these inflammatory signalling intermediates might prove an effective alternative strategy for CKD treatment. Using a proteome profile array, the cell supernatant from TGFβ1±danegaptide-treated cells was incubated on a cell membrane preabsorbed with antibodies raised against key inflammatory proteins. As highlighted in Figures 6 and 7, co-incubation of cells with TGFβ1 and danegaptide restored the expression and secretion of a number of candidate proteins whose role in kidney disease is well established as either detrimental (tumour necrosis factor-alpha, interleukin 1-alpha, interferon-gamma) or protective (hepatocyte growth factor). Moreover, in CKD, soluble chemokines, adhesion molecules and growth factors recruit and activate infiltrating immune cells and resident fibroblasts to mediate inflammation and fibrosis in the injured kidney. Our array data confirm that modulating Cx43 and blocking hemichannel-mediated ATP release in tubular epithelial cells negates, either in part or fully, secretion of many inflammatory mediators, including chemokines, monocyte chemo-attractant protein (MCP1) and Regulated on Activation, Normal T Cell Expressed and Secreted (RANTES), both of which are involved in monocyte recruitment. Elevated in both human and experimental kidney diseases, MCP1 secretion is triggered by interleukin-1, tumour necrosis factor-alpha [90] and interferon-gamma [91], all of which were induced in our model by TGFβ1, yet blocked when co-applied with danegaptide. Moreover, danegaptide was also observed to prevent TGFβ1-induced secretion of intercellular cell adhesion molecule (ICAM1) [92], granulocyte-macrophage colony-stimulating factor (GM-CSF) [93] and epithelial-neutrophil-activating peptide (ENA78) [94]. Collectively, these have been linked to the progression of CKD and with other key pathogenic proteins, including dipeptidyl peptidase 4 (DPPIV) [95] and vascular endothelial growth factor (VEGF) [96], and have been identified as potential therapeutic targets. Despite these observations, array analysis identified a number of proteins at 48 h, where TGFβ1induced regulation appeared to independent of hemichannel-mediated ATP release. As evidenced by the expression of the FLT3 ligand in the cell lysate, danegaptide failed to significantly alter TGFβ1-induced changes in whole-cell expression. The FLT3 ligand is initially synthesised as a membrane-bound protein, which must be cleaved to become a soluble growth factor. In TGFβ1-treated cells, an increase in membrane-bound FLT3L was observed compared to controls in the cell lysate, a response that was unaffected when cells were co-incubated with danegaptide. Although little is known about the enzyme involved in the proteolytic cleavage and release of FLT3L, a study by Horiuchi et al. demonstrated that shedding of FLT3L and release from the membrane are metalloprotease dependent and that this effect in fibroblasts is dependent on the TNFα-converting enzyme (TACE) [97]. Whilst we can only speculate, it is possible that blocking Cx43-mediated ATP release may blunt the activity of a protein(s) required for FLT3 membrane shedding and, thus in the absence of FLT3L release, FLT3L levels in the supernatant are significantly less in TGFβ1+danegaptide-treated cells as compared to TGFβ1-treated controls. Interestingly, whilst danegaptide did not appear to reverse the TGFβ1-induced increase in the FLT3L cell lysate, it did have an effect on the secreted form of the protein.
The soluble Receptor for Advanced Glycation Endproducts (RAGE) is a potential biomarker of inflammation and oxidative stress [98]. It acts as a decoy receptor that prevents advanced glycation end products binding to membrane-bound RAGE and RAGE-related detrimental effects. In the current study, we observed a 70% increase in the secretion of sRAGE with TGFβ1, an effect restored to control by danegaptide. sRAGE is functional and able to induce an effect when secreted. The potential of danegaptide to significantly negate this secretion is of clear therapeutic interest. Interestingly, however, we did not observe an increase in the RAGE lysate at 48 h with TGFβ1 ± danegaptide. Whilst these experiments have been performed in human primary proximal tubule cells, the limitations of our model may have cell-, time-and concentration-dependent effects. Consequently, whilst a better understanding of the mechanism of action of danegaptide is needed, findings from this study provide important initial evidence of the benefits of using danegaptide to negate TGFβ1-induced changes in markers of inflammation and tubular injury via blockade of hemichannel-mediated ATP release. We concede that our in vitro data provide a minimalistic model of the multifactorial events that give rise to tubulointerstitial fibrosis, and recommend caution in translating these novel findings to the in vivo situation, especially where studies on other species and models of injury have demonstrated variable effects [41,44,[46][47][48][49]. Further studies are now required to determine the selectivity of these hemichannels and the capability of the drug in preclinical models of chronic damage, ahead of testing efficacy in human clinical trials.

Materials
Clonal human kidney 2 (HK2) epithelial cells and primary human proximal tubule epithelial cells (hPTECs) were purchased from the American Type Culture Collection (ATCC) (LGC Standards). Tissue culture supplies were purchased from Invitrogen (Paisley, UK). The Immobilon-FL PVDF membrane was from Millipore (Watford, UK), and Odyssey blocking buffer and secondary fluorescent antibodies were purchased from LI-COR (Cambridge, UK). Antibodies for E-cadherin, N-cadherin, ILK1, MMP-3, β-catenin, vimentin and ZO-1 were obtained from Cell Signalling Technologies (Hertfordshire, UK), whilst claudin-2, laminin, collagen I and collagen IV antibodies were obtained from ABCAM (Cambridge, UK). Fibronectin antibody was purchased from Santa Cruz (Santa Cruz, CA, USA).
Recombinant hTGFβ1 was obtained from Sigma (Poole, UK), as were all other general chemicals. Danegaptide was provided by Zealand Pharmaceuticals. ATP biosensors were from Sarissa Biomedical Ltd. (Coventry, UK) and fluorodishes from WPI (Hertfordshire, UK). Transwell filters were purchased from Corning (Nottinghamshire, UK). The Proteome Profiler Human Cytokine Array Kit was from R&D Systems (Oxfordshire, UK).

Lactate Dehydrogenase Assay
The release of lactate dehydrogenase (LDH) into media as a result of plasma membrane damage is commonly used to evaluate cell death or cytotoxicity. HK2 cells were seeded in 96-well plates and cultured in low-glucose DMEM/F12 (5 mM) for 48 h prior to overnight serum starvation. Cells were then stimulated for 48 h with TGFβ1 (10 ng/mL) ± danegaptide (50-1000 nM). The LDH cytotoxicity assay kit II (Abcam) was used to quantify LDH according to the manufacturer's instructions.

Crystal Violet Assay
This simple assay is used to measure the relative density of adhered cells to multi-well dishes. Crystal violet stains DNA and can be quantified colourimetrically after solubilisation. HK2 cells were seeded in 12-well plates and cultured in low-glucose DMEM/F12 (5 mM) for 48 h, prior to overnight serum starvation, and then subsequently stimulated for 48 h with TGFβ1 (10 ng/mL) ± danegaptide (50-1000 nM). The assay has been described previously [99]. Briefly, cells were fixed using paraformaldehyde for 10 min, washed with phosphate buffered saline (PBS) and incubated for 10 min at room temperature (RT) in a 0.1% crystal violet solution. After several more washes, the stain was solubilised using 1% SDS, and absorbance was measured by a plate reader.

Western Blotting
Preparation of cytosolic protein from HK2 cells and hPTECs, their separation by SDSgel electrophoresis and transfer onto Immobilon-FL PVDF membranes have been described previously [12]. Membranes were blocked using Odyssey blocking buffer (LI-COR) and then probed overnight with antibodies against E-cadherin

ATP Biosensing
ATP biosensors (Sarissa Biomedical, Coventry, UK) were used in a simultaneous dual-recording amperometric set-up, as described previously [12]. A null biosensor was used to account for non-specific electro-active artefacts and subtracted from the ATP trace. Glycerol (2 mM) was included in all recording solutions to enable ATP detection. HK2 cells were seeded on glass coverslips (10 mm diameter) in low-glucose DMEM/F12 (5 mmol/L) for 48 h prior to overnight serum starvation. The cells were then incubated with TGFβ1 (10 ng/mL) ± danegaptide (100 nM) for 48 h. The coverslips were transferred to a chamber containing Ca 2+ -containing BSS perfused at 6 mL/min (37 • C) and left for 10 min to acclimatise. ATP and null biosensors were bent and lowered so that the electrode lay parallel to the cellular monolayer. Once a stable baseline occurred, perfusion of Ca 2+ -free BSS stimulated hemichannel opening. After ATP release, Ca 2+ -containing BSS was used to close hemichannels, followed by a calibration solution of ATP (10 mM). Recordings were acquired at 4 Hz with a Micro CED (Mark2) interface using Spike (v8.03) software.

Inflammation Antibody Array
An inflammation antibody array (RnD Systems) assessed TGFβ1-induced regulation of 31 inflammatory markers. The array was performed by following the manufacturer's instructions. Briefly, hPTECs were cultured in low-glucose DMEM/F12 (5 mM) for 48 h, prior to overnight serum starvation, and then subsequently stimulated for 48 h with TGFβ1 (10 ng/mL) ± danegaptide (100 nM). The cell lysates and supernatant were collected and incubated overnight with pre-blocked membranes spotted in duplicate with capture antibodies. An 800 CW fluorescent streptavidin/biotinylated cocktail mixture was used to visualise expression protein/antibody complexes by using Odyssey FC and semi-quantified using ImageStudio (v5.2, LI-COR).

Analysis
For all experiments, the low-glucose control (5 mM) was normalised to 100%, and all other conditions were compared to their respective controls. Statistical analysis was performed using a one-way ANOVA test with Tukey's multiple-comparison post-test. Data are expressed as the mean ± SEM, with n denoting the sample number. A p-value of ≤0.05 signified statistical significance.  Informed Consent Statement: Primary human proximal tubule cells (hPTECs) were commercially purchased from the American Tissue Type Collection (ATCC). All tissues used for isolation by the ATCC are obtained under informed consent and conform to HIPAA standards to protect the privacy of the donor's personal health information.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to product development related to the research being reported, please see conflicts of interest statement for U.M.