July 2009 | Vol 6 | N.º 7 | CNIC-10 [PDF (281K)]

Targeting the ERK/NF-κB/Snail1 pathway as a potential therapeutic strategy to prevent the failure of peritoneal dialysis

Raffaele Strippoli¹, Miguel Foronda¹, Manuel López-Cabrera^2,3 and Miguel A. del Pozo^1,4

SUMMARY
Peritoneal dialysis is frequently used as an alternative to haemodialysis for the treatment of end-stage renal disease. The peritoneum acts in dialysis as a semipermeable membrane across which ultrafiltration and diffusion take place. Continual exposure to bio-incompatible solutions and episodes of peritonitis or haemoperitoneum damage the peritoneal membrane, which progressively undergoes fibrosis and angiogenesis, leading ultimately to ultrafiltration failure. Changes induced by prolonged inflammatory stimuli in peritoneal mesothelial cells are reminiscent of those occurring during epithelial-to-mesenchymal transition (EMT). Thus, the mechanistic regulation of EMT genesis in the peritoneal membrane is relevant from both basic and clinical perspectives.
In our study, we reproduced EMT in vitro by treating primary mesothelial cells with effluent from patients undergoing peritoneal dialysis or by co-stimulation with transforming growth factor (TGF)-β1 and interleukin (IL)-1β. All of the aforementioned stimuli induced a genuine EMT, characterised by reduced E-cadherin and cytokeratin expression, cell scattering, and spindle-like morphology. We tried to identify the molecular mechanisms underlying this phenomenon and concluded that Snail1, induced by both ERK and NF-κB, is a master molecule that induces EMT in this experimental system. Interestingly, the blockage of ERK/NF-κB/Snail1 signalling in ex vivo cultured mesothelial cells from patients undergoing peritoneal dialysis reverted morphological and biochemical EMT in these cells. Modulation of the ERK/NF-κB/Snail1 activation pathway may thus provide a means of counteracting the progressive alteration of the peritoneal membrane and prolong the viability of peritoneal dialysis for the treatment of uremic patients.

BACKGROUND

From peritoneal dialysis to membrane ultrafiltration failure
In recent decades, peritoneal dialysis (PD) has been established as an alternative to haemodialysis for the treatment of end-stage renal disease. The number of patients included in PD programs has increased progressively worldwide, especially in some Asian countries.¹ One of the main factors limiting the increase of this form of renal replacement is the progressive functional deficiency of ultrafiltration, which is linked to structural alterations of the peritoneal membrane. One of the most important challenges in PD, therefore, is the long-term preservation of peritoneal membrane integrity.

The main etiologic factors leading to the functional decline of the peritoneal membrane are the toxic properties of dialysis fluids, such as acidic pH, lactate-based buffering, high glucose, glucose degradation products, and the membrane’s uremic status. These factors induce a chronic peritoneal inflammation that is aggravated periodically by episodes of peritonitis or haemoperitoneum. Mesothelial cells (MCs) play an important role in the regulation of the peritoneal inflammatory response by producing inflammatory cytokines and chemoattractants (such as CCL2), thus enhancing leucocyte recruitment.
The reparative process, which is closely linked to the inflammatory response, is responsible for many of the structural abnormalities of the peritoneal membrane, including loss of the MC monolayer structure, submesothelial fibrosis, angiogenesis and hyalinising vasculopathy.² Such alterations are considered the major causes of ultrafiltration failure and the loss of peritoneal dialytic capacity.

Peritoneal fibrosis and angiogenesis
There are two forms of PD-related fibrosis. The most common is simple peritoneal sclerosis, where the degree of fibrosis is usually mild and is correlated with the length of time that the patient must be treated by PD.³ Encapsulating peritoneal sclerosis is a rare form of fibrosis that progresses rapidly with intense fibrosis, inflammation and fibrin deposits.⁴ It is a life-threatening condition that, in many cases, develops into visceral encapsulation and progresses even if the patient is removed from PD. Besides fibrosis, the peritoneum also shows an increased capillary number (angiogenesis) and vasculopathy.⁵ Increasing evidence shows that fibrosis, angiogenesis and (probably) augmented vessel permeability are key determinants of ultrafiltration and transport dysfunction. The relationship between peritoneal fibrosis and angiogenesis has not been clearly defined. In animal models of PD, it has been shown that fibrosis and angiogenesis may be two separate responses to peritoneal injury. However, in PD patients, it is likely that fibrosis and angiogenesis are intimately and closely related to each other in the response of the peritoneum to continuous injury.
Animal models, especially rat models, have been useful in the understanding of various aspects of PD. More recently, mouse models of PD have been established, thus allowing for the study of PD in genetically modified animals ⁶.

Mesenchymal transformation of mesothelial cells in response to PD
Soon after PD is initiated, peritoneal MC patients progressively lose their epithelial phenotype and acquire myofibroblast-like characteristics reminiscent of an epithelial-mesenchymal transition (EMT).⁷ EMT is a complex physiopathologic phenomenon that occurs during development, cancer and metastasis.⁸ EMT is increasingly being studied in the pathogenesis of chronic fibrotic pathologies of the kidney, lung and liver and in the peritoneum of patients undergoing PD.
EMT starts with the dissociation of intercellular junctions due to down-regulation of adhesion molecules such as E-cadherin, claudins, occludins, zona occludens-1 (ZO-1) and desmoplakin. E-cadherin downregulation is considered the main hallmark of EMT. The subsequent disruption of adherens junctions induces cellular scattering, an early event in EMT. Advanced stages are characterised by the acquisition of a motile and invasive phenotype. The upregulation of matrix metalloproteinases (MMPs), such as MMP2 and -9, allows the cells to degrade type IV collagen, which is a major component of the basement membrane.
Cells that have undergone EMT downregulate other epithelial markers, such as cytokeratins, and upregulate mesenchymal markers such as vimentin, N-cadherin and a-SMA (smooth muscle actin), as well as the production of extracellular matrix (ECM) proteins such as fibronectin and collagen I. In addition to the acquisition of a fibroblastic morphology, EMT involves a major reprogramming of the cell proteome.

EMT can be induced by combinations of a wide spectrum of extracellular stimuli, including high glucose, MMPs, ECM proteins, growth factors and cytokines. Of these, TGF-β1 is considered a master molecule in the genesis of EMT in many cellular models, including peritoneal MCs.
EMT is a reversible process, at least during its early stages. The ‘transformation state’ of the cell can thus be thought of as the result of a balance between signals promoting EMT (like those induced by TGF-β1) and signals promoting EMT reversal (MET). Molecules that were demonstrated to block and reverse EMT in different cellular models include hepatocyte growth factor (HGF), bone morphogenetic protein-7 (BMP-7) and Vitamin D.⁹-¹¹

Signalling pathways and Molecular mechanisms leading to EMT
EMT is a multistep process, and its induction in vivo is the result of repeated and persistent extracellular stimuli. It is difficult to establish a hierarchy among the different signals and extracellular mediators that induce EMT. In addition, because EMT is often cell specific, extrapolating data from different cell systems can be misleading (especially considering that most studies having been carried out in transformed tumour models). Regarding primary peritoneal MCs, very few studies so far have analysed the signalling pathways involved in EMT induction, thus highlighting the novelty of our study (see below).

EMT is triggered by a wide array of extracellular stimuli, including components of the extracellular matrix (such as fibronectin, collagen I and hyaluronic acid). Integrins are the main receptors for ECM proteins and can synergise with growth factors in triggering the activation of a complex network of intracellular effector molecules involved in EMT induction, including Ras/Rho GTPases, Rho-activated kinase (ROCK), the tyrosine-kinase Src, integrin-linked kinase (ILK) and mitogen-activated protein (MAP) kinases.⁸

E-cadherin internalisation, ubiquitination and subsequent proteasomal degradation play a central role in EMT. E-cadherin tonically inhibits EMT-related signalling events. Transcription factors such as catenins and NF-kB, which can be sequestered at adherens junctions, are released upon E-cadherin downregulation and may initiate transcription in the nucleus. Besides increased degradation, E-cadherin expression is directly inhibited by several transcription factors, such as the zinc-finger factors Snail and Slug, SIP1, and the basic helix-loop-helix factors E47 and Twist.¹²

Within the cell, glycogen-synthase kinase (GSK)-3b contributes to the preservation of the epithelial phenotype by phosphorylating b-catenin and Snail, leading to their ubiquitination and degradation via the proteasome. The phosphorylation of GSK-3b by ERK, ILK, Wnt-1 or PI3-K leads to its functional inhibition. As a result, b-catenin and Snail are stabilised and localise to the nucleus, where they can carry out their transcriptional activities.13 A contribution to tumoural EMT has been reported for NF-kB 14, a transcriptional protein complex that translocates to the nucleus in response to inflammatory stimuli. NF-kB is also associated with chronic fibrotic pathologies, including MC EMT (see below).15 One of the main mediators of fibrosis, as demonstrated in animal models, is the transcription factor SMAD3,16 a member of the SMAD family that forms part of the ´classical’ TGF-b-induced pathway.

Extracellular factors that induce EMT reversion (MET) can interfere with EMT-inducing signalling pathways. In particular, HGF has been demonstrated to induce transcriptional co-repressors that interact with the activated SMAD-2/4 complex and block the expression of SMAD-dependent genes, including ILK. BMP-7 treatment induces SMAD-1/5/8 activation, which interferes with TGF-b-activated SMAD-2/3. Finally, a vitamin D analogue has been demonstrated to reverse EMT and subsequent fibrosis by blocking NF-kB signalling in a model of kidney inflammation.¹⁷

Towards a mechanistic prevention and treatment of peritoneal membrane alterations
An understanding of the molecular mechanisms that regulate peritoneal-injury-induced chronic inflammation can bring about the design of precise therapeutic strategies aimed either at counteracting the EMT or at combating its effects (such as cellular invasion, ECM accumulation or angiogenesis).
The possibility of preventing or reversing EMT by using endogenous extracellular factors such as HGF and BMP-7 is particularly interesting. These cytokines have been demonstrated to block glucose-induced EMT, and BMP7 has been shown to ameliorate fibrosis in an animal model of PD.10 BMP7 has been shown to play a role in preventing peritoneal EMT in an in vivo model (Lopez Cabrera, unpublished). Another endogenous factor participating in EMT regulation is vitamin D. Paricalcitol, a synthetic vitamin D analogue, attenuates renal interstitial fibrosis in obstructive nephropathy by inducing E-cadherin and blocking aSMA expression in tubular epithelial cells.¹⁷

Another approach is to target extracellular EMT inducers. Many studies have focused on the renin-angiotensin system, which induces EMT and fibrosis in the kidney.18 Inhibitors of angiotensin-converting enzyme (ACE) and angiotensin II type 1-receptor (AT1) attenuate the production of VEGF in MCs. In addition, the intraperitoneal administration of an ACE inhibitor attenuated the structural and functional alteration of the peritoneum in a rat PD model. Because integrins activate changes in the cytoskeleton and signalling pathways during EMT, their activity has been targeted in many studies. Specific monoclonal antibodies or pharmacological inhibitors of integrin activation modulate EMT and fibrosis in in vivo models.¹⁹, ²⁰

It is also possible to interfere with EMT induction by pharmacological inhibition of the intracellular pathways activated during this process. Inhibition of ROCK, a downstream effector kinase of RhoA, suppressed a-SMA expression and renal interstitial fibrosis in a mouse model of ureteral obstruction.21 Other tyrosine kinase inhibitors have been used to prevent fibrosis and angiogenesis in a variety of experimental systems, including systemic sclerosis and a case of peritoneal dialysis complicated with metastatic cancer. ²², ²³

It is important to remember, however, that EMT is involved in pathophysiological processes (such as wound healing), and, hence, its total blockage could be deleterious for PM repair. Thus, it might make more sense to treat the secondary effects of MC EMT, such as cellular invasion, fibrosis or angiogenesis, while the EMT occurs. In this regard, the upregulated expression of MMP-2 and MMP-9 during EMT allows MCs to degrade the basal membrane and invade the submesothelial stroma.²⁴ MMP inhibitors may thus prevent the accumulation of MC-derived myofibroblasts in the submesothelial compartment, which in turn would diminish the structural and functional alteration of the PM, as already demonstrated in a rat model.²⁵ MCs that have undergone EMT produce higher amounts of ECM components, including fibronectin and collagen I, and display a reduced fibrinolytic capacity due to an unbalanced ratio between tissue plasminogen activator (tPA) and plasminogen activator inhibitor-1 (PAI-1). Thus, intervention in peritoneal fibrosis may be designed either to prevent ECM synthesis or to increase fibrinolysis. Several antifibrotic drugs (pentoxifylline, dipyridamole, troglitazone, diltiazem, emodin) directly inhibit ECM protein synthesis or TGF-b expression and activity in MCs.1 Statin simvastatin is another candidate molecule for the therapeutic treatment of peritoneal fibrosis because it inhibits ECM synthesis and displays fibrinolytic activity by inducing tPA synthesis and inhibiting PAI-1 expression.²⁶ Finally, during the progression of EMT, MCs produce higher amounts of VEGF, which correlates with increased peritoneal transport. Therefore, therapeutic intervention might also be directed at preventing peritoneal angiogenesis and vessel permeability by targeting VEGF expression or its effects on endothelial cells. Two angiogenesis inhibitors, TNP-470 and endostatin, suppress the progression of peritoneal deterioration in animalmodels.²⁷, ²⁸

EXPERIMENTAL PROJECT

Our study aimed at characterising the signalling pathways that control the establishment of EMT in primary peritoneal MCs that were induced to differentiate by treatment with either peritonitis effluent or a cytokine cocktail (TGF-β1 plus IL-1β). Once these pathways were identified, we analysed whether their inhibition can reverse EMT (MET) in cells from the peritoneal effluent of patients undergoing PD.¹⁵

Human MCs were obtained through the digestion of omentum samples from patients undergoing unrelated abdominal surgery. To induce EMT, we stimulated MCs with peritoneal effluent from patients experiencing acute peritonitis during PD. Treatment with peritonitis effluent for 72 hours induced a loss of intercellular junctions, cell scattering, and adoption of a spindled fibroblastic phenotype, all of which are characteristic of EMT. A confocal immunofluorescence analysis showed that peritonitis effluent downregulated the expression of cytokeratin, an epithelial marker that is highly expressed in untreated MCs. The expression of E-cadherin was also markedly reduced. Peritonitis effluent also upregulated fibronectin and N-cadherin, the expression of which is associated with EMT. We confirmed these results in a non-dialysis setting by co-stimulating MCs with TGF-β1 plus IL-1β, as described previously (Fig.1).

Our approach of combining TGF-β1 and IL-1β attempts to reproduce a portion of the complex mixture of proinflammatory and profibrotic stimuli induced during peritoneal EMT in vivo. TGF-β1 alone has been widely reported to induce EMT, including in MCs. However, the combined cytokine treatment has additive effects on morphological and biochemical changes associated with EMT. ⁷,¹⁵ The cytokine doses used in our study were in the range of those detected in PD fluids in the presence of peritonitis and similar to those used in previous studies.

We then analysed the molecular mechanisms underlying this event, focusing on ERK and NF-kB pathways. ERK activation can affect many steps of MC EMT: it enhances cell migration and invasion through myosin-light chain kinase phosphorylation and by inducing the expression of MMP-2 and -9 (and aSMA, unpublished) and is a potent inducer of Snail via induced EGR-1 expression.²⁹ We used inhibitors of MEK1/2 (a direct ERK activator) such as PD98052 and U0126. The results obtained with these inhibitors have been recently confirmed using another compound (CI 1040) suitable for in vivo use (Strippoli, Foronda and del Pozo, unpublished).³⁰ The involvement of NF-kB in EMT has been demonstrated primarily in tumors.¹⁴NF-kB activation has been linked to E-cadherin downregulation and Snail family induction. NF-kB can be activated by angiotensin II and AGEs (advanced glycation end products), which appears to be a common pathway in the induction of EMT and fibrosis.³¹ By using ERK pathway inhibitors and a retroviral vector blocking NF-kB nuclear translocation (IkBα super-repressor), we found that both of these pathways are able to counteract the progression of EMT in MCs (Fig. 1). We also identified Snail expression, which can be induced by both pathways, as a central mediator of EMT in this experimental system. Moreover, our findings placed ERK upstream of NF-kB in the pathway of MC EMT induction.
Given the role of ERK and NF-κB signalling in cytokine-induced EMT in culture MCs, we wondered whether this pathway might control the maintenance of the mesenchymal phenotype in PD-effluent-derived MCs that have already undergone EMT in vivo. To investigate this, we obtained MCs from the effluents of more than 20 patients undergoing PD. The use of these cells, which were obtained by centrifugation of dialysis effluents, is particularly interesting because they detach directly from the peritoneum of patients and can be used to monitor the status of the PM. Data obtained in vitro and ex vivo indicate that MCs collected from PD effluents are representative of the whole MC population. Extensive use of these cells is limited by the number of cells collected from each bag of dialysis effluent (around 25,000) and their weak proliferative ability.³² MCs derived from PD effluents show one of three major morphologies: cobblestone-like (similar to normal omentum), transitional or fibroblast-like. More recently, these cells have been classified into two groups: epithelioid and non-epithelioid. The parameters used to evaluate the different stages of transdifferentiation of these cells were both morphological (epithelial-like or non-epithelioid) and biochemical (reduced levels of E-cadherin and cytokeratins, increased expression of VEGF and vimentin), as described in published studies.⁷, ³² Compared with control omentum samples, untreated PD effluent-derived MCs had significantly increased levels of active ERK.¹⁵ PD-effluent-derived MCs were treated with U0126 or infected with a retrovirus encoding the IκBα super-repressor. Microscopy of non-epithelioid effluent-derived MCs showed that U0126 caused cells to revert to an epithelial morphology. The expression of E-cadherin in both epithelial-like and non-epithelioid effluent-derived MCs was restored in both U0126-treated and IκBα super-repressor-expressing cells. Furthermore, confocal immunofluorescence showed that U0126 treatment increased the levels of cytokeratin in non-epithelioid MCs from PD effluents. U0126 also impaired NF-κB nuclear translocation. Finally, U0126 treatment of non-epithelioid MCs also downregulated Snail1 mRNA expression and upregulated E-cadherin levels. The reversal of EMT by blocking ERK or NF-κB in these experiments strongly supports a role for the ERK/NF-κB activation pathway in the maintenance of the mesenchymal phenotype in the peritoneum of patients undergoing PD.

Conclusions

Our results, obtained using pharmacological inhibitors and retroviral vectors, demonstrate that an ERK/NF-κB signalling pathway controls morphological changes and downregulation of E-cadherin and cytokeratin during EMT in MCs. Moreover, the blockage of this signalling pathway in transdifferentiated MCs isolated from PD effluents reverses EMT. These results provide a rationale for the design of new drugs aimed at preventing or counteracting the changes in the peritoneal membrane of patients undergoing PD, which are linked to the decline of dialytic capacity.

Acknowledgements.
This work was supported by the MICINN (Spanish Ministry of Science and Innovation) through grants SAF2008-02100 and RD06/0020/1033 (RTICC, Instituto de Salud Carlos III) to MAdP, by EUROHORCS (European Heads Of Research Councils) and the European Science Foundation (ESF) through a EURYI (European Young Investigator) award to MAdP, by the EMBO Young Investigator Programme (to MAdP). Raffaele Strippoli is supported by a Río Hortega Contract (Instituto de Salud Carlos III, MICINN). Editorial assistance was provided by Simon Bartlett. The CNIC is supported by the Institute of Health Carlos III (MICINN) and the Pro-CNIC Foundation.