Non-canonical NFκB activation promotes chemokine expression in podocytes

TNF-like weak inducer of apoptosis (TWEAK) receptor Fn14 is expressed by podocytes and Fn14 deficiency protects from experimental proteinuric kidney disease. However, the downstream effectors of TWEAK/Fn14 in podocytes are poorly characterized. We have explored TWEAK activation of non-canonical NFκB signaling in cultured podocytes. In cultured podocytes, TWEAK increased the expression of the chemokines CCL21, CCL19 and RANTES in a time-dependent manner. The inhibitor of canonical NFκB activation parthenolide inhibited the CCL19 and the early RANTES responses, but not the CCL21 or late RANTES responses. In this regard, TWEAK induced non-canonical NFκB activation in podocytes, characterized by NFκB2/p100 processing to NFκB2/p52 and nuclear migration of RelB/p52. Silencing by a specific siRNA of NIK, the upstream kinase of the non-canonical NFκB pathway, prevented CCL21 upregulation but did not modulate CCL19 or RANTES expression in response to TWEAK, thus establishing CCL21 as a non-canonical NFκB target in podocytes. Increased kidney Fn14 and CCL21 expression was also observed in rat proteinuric kidney disease induced by puromycin, and was localized to podocytes. In conclusion, TWEAK activates the non-canonical NFκB pathway in podocytes, leading to upregulation of CCL21 expression. The non-canonical NFκB pathway should be explored as a potential therapeutic target in proteinuric kidney disease.

Chronic kidney disease (CKD) is one the three causes of death that most increased worldwide from 1990 to 2013 1,2 . CKD is now categorized based on glomerular filtration rate and degree of albuminuria, since albuminuria, a marker of podocyte injury, is a key risk factor for death and for progression of CKD. Proteinuric kidney diseases, such as diabetic kidney disease and chronic glomerulopathies, remain the most frequent causes of CKD, despite the availability of antiproteinuric agents such as renin-angiotensin system blockers 3 . Thus, there is an increasing interest in understanding the molecular mechanisms of podocyte injury or activation to express inflammatory mediators that may contribute to proteinuric kidney disease. Recently, functional in vivo studies identified the TNF superfamily cytokine Tumor necrosis factor-like weak inducer of apoptosis (TWEAK, Apo3L or TNFSF12) as a key contributor to proteinuric kidney disease in the context of immune-complex deposition (lupus nephritis, anti-glomerular basement membrane disease) and in the absence of immune-mediated injury (protein overload) [4][5][6][7][8] . Fn14-knockout MRL-lpr/lpr mice were protected from glomerular injury and glomerular podocytes were preserved 5 . Based on this preclinical data, clinical trials are testing the hypothesis that anti-TWEAK neutralizing antibodies are nephroprotective in human lupus nephritis 4 . However, TWEAK actions on podocytes have been poorly characterized. TWEAK signaling directly damaged barrier function and increased filtration through podocyte monolayers 5 , induced cell motility 9 and promoted inflammatory responses in podocytes 8 . In this regard, the podocyte contribution to glomerular inflammation has recently been emphasized [10][11][12] . Thus, we have now explored further actions of TWEAK in podocytes. Specifically, we have focused on the regulation of the activation of the pro-inflammatory transcription factor nuclear factor-kappa B (NFκ B) in podocytes 13,14 , since unravelling this pathway may provide new possibilities for development of therapeutic approaches for proteinuric kidney disease. Like TNF, TWEAK activates canonical NFκ B signaling in diverse cell types, including podocytes 8,15,16 . Unlike TNF, TWEAK promotes the non-canonical activation of NFκ B in splenocytes and tubular cells 14,17 . In tubular cells, TWEAK-induced non-canonical NFκ B activation led to expression of CCL21, a chemokine previously described as a transcriptional target of non-canonical NFκ B activation in splenocytes 18 . However, little is known of non-canonical NFκ B activation in podocytes (reviewed in ref. 14). Interestingly, podocyte CCL21 was reported to contribute to the podocyte-mesangial cell crosstalk by activating CCR7 receptors in mesangial cells and promoting mesangial cell survival, proliferation, migration and matrix adhesion through activation of glycogen synthase kinase-3 (GSK-3), protein kinase B (PKB/Akt) and integrin-linked kinase [19][20][21] . While CCL21-induced mesangial cell proliferation was initially hypothesized to contribute to glomerular homeostasis, the hypothesis was not tested and several glomerulonephritis are characterized by mesangial proliferation. Furthermore, kidney CCL21 expression was associated to recurrence of original nephropathy among renal transplant patients 22 and to acute rejection in zero-hour biopsies from deceased donor kidneys 23 , and was hypothesized to facilitate alloreactive immune responses in renal transplant recipients 24 . Moreover, anti-CCL21 antibodies reduced kidney fibrosis and inflammation through inhibition of fibrocyte recruitment 25 . Thus, functional in vivo preclinical studies are consistent with a detrimental role of CCL21 in kidney fibrosis and inflammation. CCL21 expression by podocytes could thus be chemotactic or elicit detrimental biological responses in mesangial or tubular cells. However, the factors regulating CCL21 expression in podocytes have not been characterized.
We have now explored whether TWEAK modulates CCL21 expression in podocytes and the role of non-canonical NFκ B activation in the process. Furthermore, we have explored the expression of the TWEAK receptor and CCL21 in experimental proteinuric kidney disease.

TWEAK increases CCL21 expression in cultured podocytes and this response is independent
from canonical NFκB activation. The time-course of TWEAK-induced chemokine expression differed between cultured podocytes and tubular cells 15,17 . This may imply differential involvement of different NFκ B activation pathways. TWEAK-induced mRNA expression of the non-canonical NFκ B target CCL21 progressively increased over 24 h in podocytes (Fig. 1A) as was observed for tubular cells 17 . TWEAK also increased the expression of CCL19 mRNA, which is also a target of non-canonical NFκ B activation in splenocytes 26,27 (Fig. 1B). However, unlike the similar time-course observed for TWEAK-induced CCL21 and CCL19 expression in tubular cells 17 , the time-course of CCL19 and CCL21 expression differed in podocytes. Thus, CCL19 expression peaked at 6 h and was already decreasing at 24 h (Fig. 1B). TWEAK also increased RANTES mRNA expression progressively for up to 24 h (Fig. 1C). This is in contrast to tubular cells, in which TWEAK-induced RANTES expression peaks at 6 h, as expected for a canonical NFκ B target, and then decreases 15 . Increased protein levels in cell lysates was also noted (Fig. 1D).
Parthenolide inhibits Iκ Bα degradation and RelA nuclear translocation and thus, canonical NFκ B activation 15,28 . In previous studies, we observed that parthenolide inhibited RelA nuclear translocation induced by TWEAK in podocytes as well as the expression of the canonical RelA target gene MCP-1 8 . However, parthenolide did not prevent CCL21 mRNA or protein up-regulation induced by TWEAK ( Fig. 2A,B) suggesting that RelA does not mediate CCL21 transcription. By contrast, parthenolide did prevent the upregulation of CCL19 mRNA in podocytes (Fig. 2C), suggesting the involvement of different pathways for NFκ B activation in the regulation of both chemokines in podocytes, in contrast to prior reports in splenocytes 18,26 . Parthenolide prevented the early, but not the late increase in RANTES mRNA expression (Fig. 2D), suggesting the contribution of canonical but also of parthenolide-resistant pathways to TWEAK-induced RANTES upregulation in podocytes. Similar results were observed using the NF-κ B inhibitor BAY 11-7085 (Suppl. Fig. 1).

TWEAK induces non-canonical NFκB activation in cultured podocytes.
In previous studies we observed that TWEAK induced a sustained increase in NFκ B DNA-binding activity in podocytes that peaked at 24 h as assessed by electrophoretic mobility shift assay (EMSA) 8 . This sustained increase is consistent with non-canonical NFκ B activation, since canonical NFκ B is a transient phenomenon, but non-canonical NFκ B activation in response to TWEAK in other epithelial cells, such as tubular cells, increases progressively up to 24 h 13 . The non-canonical NFκ B pathway requires activation of NFκ B-inducing kinase (NIK), that phosphorylates IKKα and serves as a docking molecule that recruits IKKα to NFκ B2 p100, facilitating NFκ B2 p100 ubiquitination and subsequent proteasomal processing into the mature NFκ B2 p52 subunit, allowing RelB/p52 complexes to enter the nucleus 14 . TWEAK induces NFκ B2 p100 processing to NFκ B2 p52 (Fig. 3A) as well as p52/RelB nuclear translocation in cultured podocytes in a time-dependent manner (Fig. 3B,C). These results indicate that TWEAK induces non-canonical, sustained NFκ B activation in podocytes. To explore the role of non-canonical NFκ B activation in the regulation of TWEAK-induced chemokine expression, we silenced NIK using a specific siRNA (Fig. 4A). NIK silencing prevented CCL21 upregulation in cultured podocytes (Fig. 4B,C) but did not modulate CCL19 or RANTES expression (Fig. 4D,E). This suggests that CCL21 is a non-canonical NFκ B pathway target in podocytes, but other chemokines are not.
Increased podocyte CCL21 in experimental proteinuric kidney disease. We next explored the expression of the chemokine identified as a non-canonical NFκ B target, CCL21, during podocyte injury and by podocytes in vivo. For assessment of CCL21 expression in proteinuric kidney disease we chose a non-immunological model of podocyte injury, rat PAN nephrosis, since it is directly induced by a podocyte toxin, does not require activation of the immune system and is representative of human focal segmental glomerulosclerosis, a common human proteinuric nephropathy 29,30 . Systemic PAN administration causes podocyte injury in rats, leading to increased urinary protein excretion by day 2 and full-blown nephrotic syndrome at day 10. Increased whole kidney TWEAK receptor (Fn14) mRNA (Fig. 5A) and protein (Fig. 5B,C) expression was noted in PAN-injected rats 2 and 10 days post-injection. Moreover, CCL21 mRNA (Fig. 5D) and protein (Fig. 5C,E) was also increased, following a similar time-course. Immunohistochemistry confirmed Fn14 and CCL21 protein expressing cells co-localized with the podocyte-specific maker WT-1 in glomeruli of PAN-injected rats ( Fig. 6 and Suppl. Fig. 2).
Scientific RepoRts | 6:28857 | DOI: 10.1038/srep28857 To better understand the relationship between podocyte injury and expression of inflammatory molecules, we tested the regulation of Fn14 and CCL21 expression by PAN or by high glucose in cultured podocytes. Either a cytotoxic concentration of PAN or high glucose levels increased Fn14 and CCL21 mRNA (Suppl. Fig. 3) expression in cultured podocytes.

Discussion
TWEAK had previously been shown to have a proinflammatory effect dependent on canonical NFκ B activation in podocytes, leading to the synthesis of the MCP-1 chemokine 8,31 . We now show that TWEAK also activates the non-canonical NFκ B pathway in podocytes, promoting the expression of the chemokine CCL21 that, as Fn14, is upregulated in vivo in the course of nephrotoxic podocyte injury.
NFκ B activates the transcription of different genes with specificity and kinetics that vary in a gene-, stimulusand cell-specific manner. Delayed kinetics of gene transcription may be due to involvement of the non-canonical NFκ B pathway or to decreased DNA accessibility to the canonical NFκ B pathway. The later has been described for RANTES in tubular cells and other cell types 13,15 . We have now focused on understanding the regulators and targets of non-canonical NFκ B activation in podocytes in response to TWEAK, a cytokine that promotes glomerular injury in experimental animals and is currently the target of clinical trials in human glomerular disease.
Some actions of TWEAK in podocytes had been previously characterized, including damage to the barrier function 5 , motility 9 and inflammatory responses as a consequence of canonical NFκ B activation leading to MCP-1 synthesis 8 . These observations may underlie the causative role of TWEAK in glomerular injury 4,32 . Interestingly, in podocytes TWEAK induced sustained NFκ B activation as assessed by EMSA 8 , although the molecular basis for the sustained NFκ B activation was not explored. We now provide evidence that in podocytes TWEAK recruits sequentially the canonical and non-canonical pathways of NFκ B activation, resulting in the transcription of different sets of chemokines. Initially canonical activation of RelA promotes the release of MCP-1 8 , and as identified here, the gene expression of RANTES and CCL19, followed by activation of the non-canonical NFκ B pathway requiring NIK, NFκ B2 p100 processing and migration of NFκ B2 p52/RelB to nuclei that promotes synthesis of CCL21. The transient nature of chemokine protein expression in cell lysates (3 h for CCL19, 3-6 h for RANTES) may be due to either posttranscriptional regulatory mechanisms or secretion of the chemokine into the cell culture medium. Thus, CCL21 is identified as a TWEAK-and non-canonical NFκ B pathway-regulated gene in podocytes. In this regard, podocytes differ from splenocytes, which were reported to require non-canonical NFκ B activation by LTβ R ligation to upregulate both CCL21 and CCL19 33 .
TWEAK induced non-canonical NFκ B activation in fibroblasts 34 and in kidney tubular cells 17 , but it was unknown whether the pathway is active in podocytes. Information on non-canonical NFκ B targets has been detailed for lymphoid cells. Thus, LTβ R activation of the non-canonical NFκ B pathway induces CCL21, CCL19, CXCL12, CXCL13, and TNFSF13b (BAFF/BLYSS) expression in the spleen 33 . Of these potential NFκ B2 targets, we chose CCL19 and CCL21 for further study in podocytes, together with RANTES, a representative of genes with a delayed response to canonical NFκ B activation in other cell types. Only CCL21 was regulated by the non-canonical NFκ B target in podocytes. TWEAK-induced increased CCL21 mRNA was delayed with respect to other chemokines and persistent. This contrasts to the earlier peaks of MCP-1 8 , RANTES and CCL19 which were abolished by RelA inhibitors. The delayed induction of CCL21 is dependent on the delayed activation of the non-canonical NFκ B pathway that requires NIK. To the best of our knowledge CCL21 is the first non-canonical NFκ B2 target identified in podocytes. The in vivo observation of increased CCL21 and Fn14 expression in experimental proteinuric kidney disease suggests that this might be a clinically relevant observation.
Interestingly, differences were apparent between the pathways for NFκ B activation and the time-course of chemokine expression between podocytes and tubular cells and splenocytes 15,17,26,27 (Fig. 7). This argues for podocyte-specific mechanism regulating chemokine expression and for the need to understand the precise molecular mechanisms involved in the regulation of podocyte inflammatory responses, since not all observations in other renal cell types apply to podocytes. This offers the opportunity for differential manipulation of inflammatory responses in a cell type-specific fashion in the kidneys.
In conclusion, TWEAK is an activator of the non-canonical NFκ B pathway in podocytes, and the chemokine CCL21 is activated through this pathway while CCL19 or RANTES are not. To our knowledge this is the first time that stimuli modulating non-canonical NFκ B activation and gene targets have been described in podocytes. Activation of the non-canonical NFκ B pathway may contribute to the deleterious effect of TWEAK in proteinuric kidney disease, as Fn14 knockout mice were protected from proteinuric kidney injury of immune and non-immune origin 5,8 and should be explored as a therapeutic target itself.

Material and Methods
Cells and reagents. Conditionally immortalized mouse podocytes were a kind gift by Peter Mundel and were cultured as described 40 . Podocytes were propagated on type I collagen (Biochrom, Berlin, Germany) at 33 °C in the presence of 10 U/mL mouse recombinant interferon (IFN-γ ) (Immugenex, Los Angeles, CA) (permissive conditions) to enhance expression of a thermosensitive T antigen. Once cells had reached 70 to 80% confluence, differentiation and a quiescent phenotype was induced by culture under "non-permissive conditions" at 37 °C without IFN-γ for >12 days, resulting in disappearance of T antigen. Culture medium was RPMI 1640 (GIBCO, Grand Island, NY) with 10% heat-inactivated fetal bovine serum (FBS), 2 mM glutamine, 100 U/mL penicillin and 100 μ g/mL streptomycin. For experiments cells were cultured in serum-free media 24 hours previous to stimulation and throughout the experiment.
Recombinant human soluble TWEAK (Alexis, Läufelfingen, Switzerland) was used at 100 ng/ml based on prior TWEAK dose-response studies in renal cells 8 15,17 . TWEAK-induced RANTES expression is inhibited by parthenolide, suggesting canonical NFκB activation. By contrast, CCL21 expression is prevented by NIK siRNA, but not by parthenolide. While not specifically studied by functional inhibitors, the time-course of CCL19 expression and the fact that CCL19 was upregulated by TWEAK but not by TNF suggests that CCL19 is a transcriptional target for non-canonical NFκ B activation in tubular cells. (B) Podocytes. Note the different time-course of TWEAK-induced RANTES and CCL19 mRNA expression between tubular cells and podocytes. The increased CCL19 mRNA and the early increase in RANTES mRNA is inhibited by parthenolide, suggesting dependence on canonical NFκ B activation. CCL21 is the only chemokine that depends on non-canonical NFκ B activation in splenocytes, tubular cells and podocytes. The late increase of RANTES mRNA expression in podocytes could not be inhibited by targeting the canonical or non-canonical pathways suggesting NFκ B-independence or activation of alternative NFκ B pathways 13,15 .
Scientific RepoRts | 6:28857 | DOI: 10.1038/srep28857 podocytes in culture. For high glucose experiments, glucose was added in the media to reach a final concentration of 700 mg/dl versus control media with 200 mg/dl glucose 41 .
Quantitative reverse transcription-polymerase chain reaction. One μ g RNA isolated by Trizol (Invitrogen UK) was reverse transcribed with High Capacity cDNA Archive Kit and real-time PCR was performed on a ABI Prism 7500 PCR system (Applied Biosystems, Foster City, CA) using the DeltaDelta Ct method. Expression levels are given as ratios to GAPDH. Pre-developed primer and probe assays were from Applied 41 . siRNA transfection. Cells were seeded at a 3 × 10 5 in 6-wells plates and transfected with 25 nM NIK siRNA (Santa Cruz), Opti-MEM I Reduced Serum Medium and Lipofectamine 2000 (Invitrogen) 42 . After 18 h, transfected cells were washed and cultured for 24 hours in complete medium, stimulated with 100 ng/mL TWEAK for 24 h and harvested for analysis. This time point was selected from a time-course of decreasing NIK protein expression in response to siRNA. A negative control scrambled siRNA provided by the manufacturer did not reduce NIK protein.