TNF has been implicated in glomerular diseases, but its actions on podocytes are not well understood. Endogenous TNF expression is markedly increased in mouse podocytes exposed to sera from patients with recurrent focal segmental glomerulosclerosis, and TNF is able to increase its own expression in these cells. Exposure of podocytes to TNF increased phosphorylation of NF-κB p65-RelA followed by increased tyrosine phosphorylation of STAT3. STAT3 activation was blocked by the NF-κB inhibitor JSH-23 and by the STAT3 inhibitor stattic, whereas TNF-evoked NF-κB activation was not affected by stattic. TNF treatment increased nuclear accumulation of nuclear factor of activated T cells (NFAT)c1 in podocytes, a process that occurred downstream of STAT3 activation. TNF also increased expression of cyclin D1 but had no effect on cyclin-dependent kinase 4, p27kip, or podocin. Despite its effects on cyclin D1, TNF treatment for up to 72 h did not cause podocytes to reenter the cell cycle. TNF increased total expression of transient receptor potential (TRP)C6 channels through a pathway dependent on NFATc1 and increased the steady-state expression of TRPC6 subunits on the podocyte cell surface. TNF effects on TRPC6 trafficking required ROS. Consistent with this, La3+-sensitive cationic currents activated by a diacylglycerol analog were increased in TNF-treated cells. The effects of TNF on NFATc1 and TRPC6 expression were blocked by cyclosporine A but were not blocked by the pan-TRP inhibitor SKF-96365. TNF therefore influences multiple pathways previously implicated in podocyte pathophysiology and is likely to sensitize these cells to other insults.
- transient receptor potential cation channel, subfamily C, member 6
- tumor necrosis factor
tumor necrosis factor (TNF) is a proinflammatory cytokine released from macrophages, monocytes, dendritic cells, T lymphocytes, and many other cell types (70). Within the kidney, TNF is produced by podocytes (58), mesangial cells (12), and tubules (56). TNF has been implicated in several diseases affecting glomerular function, including crescentic glomerulonephritis (13), lupus nephritis (7), and idiopathic membranous nephropathy (7). TNF signaling has also been implicated in the progression of diabetic nephropathy (51), acute kidney injuries (55, 75), and podocyte damage in Alport's syndrome (60).
TNF has also been implicated in primary podocytopathies. For example, case reports have described remissions from recurrent focal segmental glomerulosclerosis (FSGS) induced by anti-TNF therapies (14, 42), and elevated serum TNF has been observed in patients with primary nephrotic syndromes (65, 54, 40). TNF secretion is markedly increased from monocytes isolated from children with recurrent FSGS (15, 10). Agents that absorb free TNF or that block TNF receptors can reduce the effects of serum from recurrent FSGS patients on the cytoskeleton of cultured podocytes (14), and the addition of exogenous TNF increases the albumin permeability of isolated glomeruli (45). Massive nephrosis has been reported in a patient bearing a mutation in TNF receptors, which was successfully treated by inhibition of TNF receptors (20). Another case report described transfer of a nephrotic syndrome from a mother to a newborn child, which appeared to be mediated in part by TNF (9). Elevated circulating TNF has also been associated with minimal change nephrotic syndromes, which, in some cases, were observed before the detection of associated lymphomas (50).
Podocytes are the earliest targets in primary FSGS and minimal change disease (11, 17, 18). However, relatively little is known about cellular effects of TNF on podocytes, in contrast to other renal cell types (4). The purpose of the present study was to identify signaling pathways activated by TNF in podocytes and to establish their hierarchy. The pathways established here are novel in the context of TNF signaling, but they include components such as STAT3, nuclear factor of activated T cells (NFAT)c1, and transient receptor potential (TRP)C6 channels, which have already been implicated in the pathogenesis of podocytopathies.
MATERIALS AND METHODS
Cell culture protocols, transfection, patient sera, and chemicals.
The mouse podocyte MPC-5 cell line, originally provided by Dr. Peter Mundel (Harvard Medical School), was cultured and differentiated as previously described (6, 31, 32, 59) In one set of experiments, STAT3 was knocked down using a panel of small interfering (si)RNAs directed against STAT3 or nontargeted siRNAs obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and transfected into differentiated podocytes using Lipofectamine 3000 (Invitrogen) in serum-reduced medium, according to the manufacturer's directions (1, 31). Analysis of cell proteins was carried out 48 h after transfection with siRNAs. SKF-96365, cyclosporine A, stattic, and the NFATc1 inhibitor 11R-VIVIT were purchased from Tocris Bioscience (Minneapolis, MN). TNF was obtained from R&D Systems (Minneapolis, MN). The NF-κB inhibitor 4-methyl-N1-(3-phenylpropyl)-1,2-benzenediamine (JSH-23) was purchased from Sigma-Aldrich. Serum from an adult male patient with recurrent FSGS (patient 1) was collected at the time he presented in relapse in March 2009. Serum was also prepared from blood taken from patient 1 after three LDL apheresis treatments using a Miltenyi TheraSorb column (Miltenyi Biotec, Gladbach, Germany) with a sheep antibody against the B-100 component of LDL. These samples were stored at −80°C. Serum from other recurrent FSGS patients was collected while the patients were in relapse. Serum samples were collected after ethical review by the Institutional Review Board of the University of Cologne. No information that would allow identification of the patients was provided.
Immunoblot analysis and cell surface biotinylation assays.
Immunoblot analysis and cell surface biotinylation assays were carried out as previously described (1). After treatments with sera or TNF, cells were washed repeatedly and lysed, and proteins were separated on SDS-PAGE and transferred to membranes. Blots were probed with a 1:1,000 dilution of primary antibodies: anti-phospho-STAT3 Y705 and S727 (Gene Tex, Irvine, CA), anti-STAT3, anti-phospho-NF-κB p65, anti-NF-κB p65, anti-β-actin, anti-cyclin D1, anti-CDK4, anti-p27kip (all from Cell Signaling), anti-podocin (Santa Cruz Biotechnology, Santa Cruz, CA), anti-TRPC6 (Alomone, Tel Aviv, Israel), and anti-NFATc1 (Biolegend, San Diego, CA). Membranes were washed repeatedly and then probed with a 1:10,000 dilution of horseradish peroxidase-conjugated secondary anti-rabbit or anti-mouse antibodies (Cell Signaling Technology). Signals were visualized by chemiluminescence and analyzed by densitometry using ImageJ software (version 1.46, National Institutes of Health, Bethesda, MD) to confirm the effects (1). All experiments were repeated three to five times, and examples shown are representative blots. Cell surface biotinylation assays were carried out as previously described (31, 32). NFATc1 abundance in podocyte nuclei was measured by immunoblot analysis of podocyte nuclear extracts prepared using a kit from Active Motif (Carlsbad, CA). In this assay, we also used a mouse monoclonal anti-histone H1 antibody (EMD Millipore, Billerica, MA) as a loading control for nuclei.
Immortalized podocytes were starved in serum-free RPMI medium for 24 h, after which 10 ng/ml TNF was added for the indicated times (24, 48, and 72 h). Cells were then harvested by trypsinization, washed with PBS, and fixed in 70% ethanol overnight at 4°C. Cells were resuspended in staining solution containing 50 μg/ml RNAse A, 50 μg/ml propidium iodide, and 3 mM EDTA and analyzed for DNA content in a FACS Aria II cell sorter (BD Biosciences, San Jose, CA). The collected data were analyzed with FlowJo v10 cell cycle analysis software (Tree Star, Ashland, OR).
Whole cell recordings were made at room temperature, as previously described (5, 6, 59), using fire-polished borosilicate glass microelectrodes (4–6 MΩ) and an Axopatch 1D amplifier (Molecular Devices, Foster City, CA). Up to 80% of series resistance was compensated using the circuits in this instrument. Cells were held at −40 mV, and ramp voltage commands (−80 to +80 mV over 2.5 s) were applied periodically. TRPC6 was activated by bath solutions containing 100 μM 1-(9Z-octadecenoyl)-2-acetoyl-sn-glycerol (OAG; Avanti Polar Lipids, Alabaster, AL), a membrane-permeable analog of diacylglycerol. This was prepared as a stock solution by sonication in DMSO and stored at −80°C. OAG-evoked currents were then recorded in the presence and absence of 50 μM La3+, which blocks TRPC6 but not TRPC5 channels (5, 6, 28, 59, 67).
Previous studies and case reports have provided evidence that TNF is involved in nephrotic syndromes (10, 14, 15, 20, 40, 42, 50, 54, 65) and is increased in the renal cortex in animal models of nephrosis (49). We therefore examined if endogenous TNF production could be induced in podocytes. Figure 1A shows TNF expression in immortalized mouse podocytes (MPC-5 cells) exposed to sera of three different patients with recurrent FSGS. Serum samples from patient 1 were collected during a relapse of his nephrotic syndrome. A second serum sample from patient 1 was collected after three LDL apheresis treatments using an antibody directed against B-100. This procedure removes oxidized LDL, among other circulating proteins. We also examined serum samples taken from two other patients with recurrent FSGS in relapse. Podocytes were cultured with these serum samples (final concentration of 10%, replacement of FBS in the normal cell culture medium) for 24 h. Immunoblot analysis showed large increases in TNF in podocytes cultured in serum sampled while patient 1 was in relapse compared with what was detected in cells cultured with control serum or serum from patient 1 taken after LDL apheresis. An increase in TNF was also observed when podocytes were exposed to sera from two other patients taken during relapse of recurrent FSGS (Fig. 1A). The identity of the factor that causes this effect is not known. However, we note here that TNF can increase its own expression in podocytes (Fig. 1B) as it does in other cell types (47, 64). Thus, exposure of podocytes to 10 ng/ml TNF for 24 followed by an extensive rinse resulted in increased abundance of TNF in these cells, as detected by immunoblot analysis.
We have examined several additional cellular effects of recombinant TNF on immortalized mouse podocytes. We confirmed that TNF (10 ng/ml) evoked a marked increase in the activation of NF-κB by examining levels of p65-RelA phosphorylated at S536 (Fig. 2A). Marked increases of phosphorylated p65-RelA relative to total p65-RelA expression were observed in podocytes exposed to TNF for 0.5–24 h. Given recent evidence of a role for STAT3 in collapsing glomerulopathies (19, 22) and in the regulation of the podocyte cytoskeleton (2), we examined whether TNF could activate STAT3 signaling. We observed that exposure to TNF for times of ≥1 h caused marked increases in the levels of tyrosine-phosphorylated STAT3 (at Y705) but had no effect on levels of serine-phosphorylated STAT3 (at S727) or total STAT3 (Fig. 2B). Previous studies in other cells have shown that translocation of phosphorylated p65-RelA to the cell nucleus is maximal around 1 h after the initiation of TNF treatment (34, 68). Application of 10 μM stattic, an inhibitor of STAT3 activation (62), had no effect on p65-RelA phosphorylation (Fig. 2C). In contrast, STAT3 signaling was almost completely blocked when TNF was applied in the presence of 20 μM JSH-23, an inhibitor of NF-κB activation (Fig. 2D) (63). This pattern is consistent with the shorter latency for p65-RelA phosphorylation compared with phosphorylation of STAT3 and indicates that STAT3 is downstream of NF-κB.
NFATc1 is a transcriptional regulatory protein implicated in inflammation (3) and podocyte pathophysiology (52, 61, 71). We observed that TNF caused a marked increase in the total abundance of NFATc1 in podocytes relative to podocin or actin (Fig. 3A). TNF did not cause a marked change in podocin abundance relative to actin. TNF also evoked a robust increase in the expression of cyclin D1 (Fig. 3B). In contrast, TNF did not evoke any changes in total expression of cyclin-dependent kinase 4 (Cdk4) or p27kip. TNF effects on NFATc1 and cyclin D1 require STAT3 activation. Thus, total NFATc1 and cyclin D1 were markedly reduced in podocytes in which total STAT3 was reduced by treatment with a siRNA that specifically targets STAT3 (Fig. 4A). Control cells were treated with a nontargeting siRNA. In addition, TNF effects on NFATc1 and cyclin D1 were blocked by 10 μM stattic (Fig. 4B). Therefore, the abundance of NFATc1 and cyclin D1 is regulated by processes that occur downstream of STAT3 signaling. Consistent with this, we observed that TNF effects on STAT3 phosphorylation persisted in cells treated with 10 μM 11R-VIVIT, a membrane-permeable NFAT inhibitor (Fig. 5A). Exposure to 11R-VIVIT had no effect on basal cyclin D1 expression but appeared to reduce TNF-induced expression of cyclin D1 slightly (Fig. 5A). However, this effect was very small compared with the effect of stattic, which caused substantial reductions of both basal and TNF-evoked cyclin D1 expression. Despite its stimulation of cyclin D1, TNF treatment did not cause MPC-5 cells to reenter the cell cycle at 24, 48, or 72 h of continuous exposure (Fig. 6), consistent with its lack of effect on Cdk4 or p27kip (Fig. 3B).
TNF treatment also increased the amount of NFATc1 that could be detected in nuclei isolated from cultured podocytes, which was quantified by comparison to amounts of total histone H1 detected in the same fractions (Fig. 5B). These effects were blocked by cotreatment with the NFAT inhibitor 11R-VIVIT or by the STAT3 inhibitor stattic (Fig. 5B). The TNF-evoked increase in nuclear NFATc1 was also blocked by cyclosporine A, an inhibitor of the phosphatase calcineurin (Fig. 5C). In contrast, nuclear translocation of NFATc1 was not affected by 10 μM SKF-96365, an inhibitor of TRP family cation channels (Fig. 5C).
TRPC6 channels are Ca2+-permeable cationic channels that have been implicated in familial (24, 25, 57, 73) and acquired (46) forms of glomerular disease. We observed that 24 h of exposure to TNF caused a marked increase in the amount of TRPC6 protein detected on the podocyte cell surface (Fig. 7A). TNF also produced a consistent but modest increase in the total amount of TRPC6 protein that could be detected relative to the amount of podocin present in the same cells (Fig. 7A). This observation is not surprising because NFATc1 is a known transcriptional regulator of TRPC6 in podocytes (52, 61) and other cell types (39). TNF also evoked an increase in the generation of ROS, which was eliminated by inhibition of NF-κB (with JSH-23) or STAT3 (using stattic; Fig. 7B). Previous work has shown that steady-state surface expression of podocyte TRPC6 channels is increased by ROS, which contribute to podocyte signaling mechanisms that converge on TRPC6 (6, 27, 31, 33, 59). Consistent with this, we observed that the membrane-permeable ROS quencher tempol markedly attenuated effects of TNF on the surface expression of TRPC6 (Fig. 7C). TNF-evoked mobilization of TRPC6 was also attenuated by JSH-23 and stattic (Fig. 7, D and E). The results of electrophysiological experiments were consistent with results of biochemical investigations (Fig. 8). In these experiments, we examined control and TNF-treated podocytes using whole cell recordings. We used a membrane-permeable diacylglycerol analog (OAG) to evoke activation of podocyte channels containing TRPC6 subunits, as previously described (5). Currents were recorded during the application of ramp voltage commands in the presence and absence of 100 μM OAG applied by gravity-fed bath perfusion. After the responses to OAG stabilized, we applied 50 μM La3+, which blocks TRPC6 channels but does not block TRPC5 channels (28, 67). We observed significantly (P < 0.05) increased responses to OAG in TNF-treated cells compared with controls. The OAG-evoked currents in both groups were abolished by 50 μM La3+ (Fig. 8). Finally, we examined pathways required for TNF-evoked stimulation of TRPC6 abundance (Fig. 9, A and B). We observed that the TNF-evoked increase in total TRPC6 was blocked by NFAT inhibition (Fig. 9, C and D). However, SKF-96365 did not reduce TNF stimulation of total TRPC6 (Fig. 9, C and D). This suggests that TNF can engage alternative pathways that lead to increases in total amounts of TRPC6 without actually needing to activate Ca2+ influx through TRPC6.
In the present study, we identified multiple signaling pathways activated by TNF in podocytes, some of which are novel in the context of TNF signaling. We have also established the hierarchy of these pathways, as shown in Fig. 10. All of the biochemical events studied here occur downstream of the activation of NF-κB, a nearly universal component of TNF signaling (4). NF-κB activation, in turn, leads to the activation of the transcription factor STAT3. After these initial steps, several parallel pathways are engaged, which include an increase in the expression and activation of the transcription factor NFATc1, an increase in cyclin D1, and an increase in the number of functional TRPC6 channels on the podocyte cell surface, driven in part by increased ROS generation.
A variety of glomerular diseases are caused by injuries or inflammatory processes that initially affect cells located within the glomerular endocapillary compartment. Several potentially irreversible degenerative processes are initiated once podocytes or parietal cells are involved (38). This can occur as a result of prolonged hyperfiltration, which can physically drive the detachment of the podocytes and hypertrophy of those that remain, or if there is a break in the glomerular basement membrane. In addition, podocyte damage accompanied by chronic inflammatory conditions within the endocapillary compartment can, over time, cause podocytes and/or parietal cells to inappropriately reenter the cell cycle, resulting in the collapse of the entire capillary tuft (11, 38, 72). Inflammation can also increase the sensitivity of podocytes to other pathological processes, such that frank disease is manifested in response to multiple “hits” on podocytes. The data in the present study suggest that TNF may increase the vulnerability of podocytes to essentially any insult and thereby contribute to the initiation of vicious cycles.
There are numerous potential sources of TNF capable of acting on podocytes during glomerular diseases. There is evidence that circulating TNF is elevated in primary nephrotic syndromes owing to increased secretion from monocytes and possibly other cell types (10, 15, 65). There is evidence that TNF contributes to effects of serum and plasma of patients with nephrotic syndromes on podocyte structure (14) and can increase the albumin permeability of isolated glomeruli (45). Other sources of TNF may be important. We observed here that TNF expression is increased in podocytes exposed to sera from patients with relapsed recurrent FSGS, suggesting that circulating factors may stimulate local production of TNF in glomeruli. In this regard, pathological processes that lead to NF-κB overactivation in glomeruli correlate strongly with the expression of TNF within podocytes in vivo (76). In addition, during glomerular diseases, various immune cells that express and secrete TNF may migrate into glomeruli (16). It is worth noting that TNF is capable of inducing its own expression in podocytes, as is seen in other cell types (47, 64).
NF-κB activation is a nearly universal feature of TNF signal transduction (4). Full engagement of the other pathways examined here occurred downstream of this event, and we observed maximum tyrosine phosphorylation of STAT3 around 1 h after the application of TNF, which corresponds to the delay between p65-RelA phosphorylation and its translocation to the nucleus seen in other cell types (34, 68). The mechanism whereby NF-κB activation leads to tyrosine phosphorylation of STAT3 in podocytes is not known. However, these two factors are frequently upregulated in parallel (44), and they can physically interact (41, 74).
The activation of STAT3 in podocytes subsequently leads to increases in the expression of other transcription factors and nuclear proteins. NFATc1 is of special interest because it has been observed in the inflammatory milieu in many other cell types (3). Transgenic mice in which NFATc1 is selectively overexpressed in podocytes develop glomerulosclerosis and proteinuria (71). The direction of signaling in this case is unambiguous and one way, as inhibition or knockdown of STAT3 eliminated increases in NFATc1. In contrast, inhibition of NFATc1 activation reduced nuclear localization of NFATc1 but had no effect on TNF-evoked STAT3 phosphorylation.
TNF-evoked activation of NFATc1 requires calcineurin, as the process was eliminated by cyclosporine A. We were somewhat surprised that TNF-evoked NFATc1 activation persisted in the presence of SKF-96365, a potent inhibitor of all members of the TRP family of cation channels, including TRPC6 and TRPC5 (67). This is in marked contrast, for example, to the effects of ANG II, in which NFATc1 activation is largely dependent on TRPC6 activation (52). The effectiveness of SKF-96365 in podocytes has been verified in several electrophysiological and biochemical investigations in these cells (6, 31, 32, 59). TNF signaling in podocytes is characterized by a robust increase in the total amounts of NFATc1, and not just an increase in the amounts of this protein that localize in nuclei. This could cause NFATc1 to have a wider spatial localization than it does in unstimulated cells. Under these conditions, it is possible that TNF can lead to mobilization of intracellular Ca2+ stores (26, 36, 48) or inhibition of Ca2+ sequestration (29, 69), as has been observed in several other cell types, thereby leading to widespread activation of calcineurin.
TNF caused a marked increase in total expression of cyclin D1, which was also dependent on STAT3 activation but did not require the activation of NFATc1. While cyclin D1 is best known as a regulator of cell proliferation, our observations indicate that TNF did not cause podocytes to reenter the cell cycle. Other key cell cycle regulatory proteins, such as Cdk4 and p27kip, were not affected by TNF, which may explain the lack of proliferation seen here. In this regard, increased NF-κB in glomeruli occurs in diseases in which there is no obvious proliferation of glomerular cells (8, 72, 76), and therefore proliferation is not an inevitable consequence of signaling through TNF and related cytokines. However, it is possible that a second stimulus that could activate Cdk4 or suppress p27kip may then allow for podocyte proliferation in dysregulative glomerulopathies, such as human immunodeficiency virus nephropathy (22) and nephrotoxic serum-induced glomerulonephritis (19). It bears noting that STAT3 knockdown has proven to be beneficial in animal models of these collapsing glomerulopathies (22, 23). However, cyclin D1 activation may have significant effects on other aspects of podocyte physiology, even if there is no proliferative response. Thus, cyclin D1 has been reported to regulate Rho signaling and cell migration, transcriptional activation, mitochondrial function, and cell growth (43, 53). Thus, changes in cyclin D1 could conceivably contribute to podocyte hypertrophy and changes in foot process morphology during glomerular diseases.
TNF treatment evokes an increase in total and cell surface expression of TRPC6 channels. Excessive activation or gain of function of these channels has been previously implicated in familial forms of FSGS (24, 25, 57, 73). Consistent with biochemical measurements, we observed an increase in OAG-evoked La3+-sensitive cationic currents in TNF-treated cells (which could include TRPC3, TRPC6, and TRPC7). TRPC6 and TRPC5 are the most abundant members of the TRP family of channels in podocytes (21, 67). TRPC6 channels (as well as TRPC3 and TRPC7 channels) are activated by OAG and blocked by 50 μM La3+, whereas TRPC5 does not respond to OAG and is actually activated by micromolar La3+ (28, 67). The effect of TNF on overall TRPC6 abundance was dependent on NFATc1 activation. In addition, the increased surface expression of TRPC6 was blocked by tempol, indicating a role for ROS, as we and others have seen with several other treatments that increase the activation of podocyte TRPC6 channels (5, 27, 31, 59). There are several possible sources of ROS in podocytes. We have observed that TNF can increase expression of NADPH oxidase 4 (data not shown), but TNF can stimulate mitochondrial ROS generation in other cell types (30). In this regard, cyclin D1 is known to affect many aspects of mitochondrial function and can bind to outer mitochondrial membranes, leading to a loss of mitochondrial membrane potential and increased mitochondrial ROS generation (66), which could contribute to mobilization of TRPC6.
In summary, we have shown that TNF induces a highly pleiotropic signaling cascade in podocytes that engages proteins and signals already implicated in pathophysiology of these cells, including NF-κB, STAT3, NFATc1, ROS, and TRPC6. These suggest a cellular rational for possible use of anti-TNF therapy in glomerular diseases.
This work was supported in part by a grant from the Juvenile Diabetes Research Foundation.
Work in the Dryer laboratory was supported by a contract from Pfizer.
Author contributions: M.A., E.Y.K., H.R., F.N., and C.T. performed experiments; M.A., E.Y.K., H.R., F.N., and C.T. analyzed data; M.A., E.Y.K., F.N., C.T., and S.E.D. prepared figures; M.A., H.H., T.B., and S.E.D. edited and revised manuscript; M.A., E.Y.K., H.R., F.N., C.T., H.H., T.B., and S.E.D. approved final version of manuscript; E.Y.K., F.N., C.T., H.H., and S.E.D. interpreted results of experiments; H.H., T.B., and S.E.D. conception and design of research; S.E.D. drafted manuscript.
The authors are grateful to Dr. Peter Mundel (Harvard Medical School) for providing the MPC-5 immortalized mouse podocyte cell line.
- Copyright © 2015 the American Physiological Society