TGF-β and Renal Fibrosis

See related article on page 1236

Why kidney injury is sometimes repaired with complete restoration of its structural features, whereas in other instances it leaves behind tubular atrophy and interstitial fibrosis, is a conundrum. There is also considerable controversy regarding the primacy of tubular or interstitial factors in renal disease progression. As recently editorialized in this journal by Cook,¹ there is a need for further research in this area, but carefully performed morphological studies show that tubular pathological characteristics precede interstitial fibrosis in most of the renal diseases; he suggests that fibrosis does not proceed autonomously. The implication is that continuing tubular damage is needed to drive the disease progression, culminating in fibrosis. In a given glomerular disease, tubular damage occurs secondarily in the same nephron and fibrosis develops only around the atrophic tubules.² By using models of tubular injury caused by ischemia or selective ATP depletion by maleate, Geng et al³ and Lan et al⁴ strengthened the case for the primacy of tubular-based factors in the progression of renal fibrosis. Similarly, Grgic et al⁵ used a transgenic approach in mice to elucidate that interstitial fibrosis follows selective damage to proximal tubular cells. Regardless, the precise tubular-based factors that cause fibrosis remain to be identified.

Autocrine TGF-β Signaling by Tubules Causes Fibrosis in the Injured Kidney

In response to injury, kidney tubules increase synthesis of transforming growth factor-β (TGF-β), a profibrogenic cytokine. Spurgeon et al⁶ observed that, after ischemia-reperfusion injury (IRI), regenerating proximal tubules express more TGF-β and TGF-β receptors, suggesting that the mechanisms related to increased autocrine TGF-β signaling may be operative. Interestingly, neutralizing TGF-β antibodies prevented IRI-initiated kidney fibrosis.⁶ Along these lines, Geng et al³ further emphasized the role played by TGF-β–induced dedifferentiation of tubules in fibrosis after IRI, and they showed that inhibition of TGF-β signaling effectively promotes tubular epithelial differentiation, prevents tubular atrophy, and reduces interstitial fibrosis. The phenotype of TGF-β–induced dedifferentiated tubules was defined by Lan et al,⁴ who suggested that atrophic tubules that evolve after IRI show persistent dedifferentiation, depleted phosphatase and tensin homolog protein with growth arrest, accentuated signaling by Jun N-terminal kinase, and increased expression of profibrogenic peptides [ie, platelet-derived growth factor-B (PDGF-B) chain and connective tissue growth factor (CTGF)]. The entire spectrum of complex pathological characteristics affecting this dysfunctional tubular phenotype was ameliorated by prior inhibition of TGF-β signaling.^3,4 Likewise, Yang et al⁷ noted an aberrant persistence of Jun N-terminal kinase signaling and production of TGF-β and CTGF by growth-arrested tubular cells associated with interstitial fibrosis that ensued after IRI. On the other hand, it is still unclear how TGF-β signaling becomes accentuated in tubules of an injured kidney, which then generates fibrogenic signals or effects.

A Winding Road from GPCRs to TGF-β Activation and Profibrogenic Peptide Secretion by Tubular Cells

In this issue of The American Journal of Pathology, Venkatachalam and colleagues⁸ demonstrate the operation of an intricate signaling mechanism in proximal tubular cells that involves the activation of G-protein–coupled receptors (GPCRs) by lysophosphatidic acid (LPA) and subsequent αvβ6 integrin–dependent conversion of latent TGF-β to active peptide. These actions of LPA require LPA2 receptors and Gαq signaling, channeled via a Rho/Rho-associated kinase pathway. The nascent TGF-β produced by this mechanism initiates autocrine signaling that triggers the synthesis and secretion of additional profibrogenic peptides, PDGF-B and CTGF.⁸ The step-by-step signaling connections between LPA2 receptor activation and PDGF-B/CTGF secretion in cultured cells elucidated by these authors could possibly have important implications for understanding the pathogenesis of tubulointerstitial fibrosis. Indeed, their study includes important in vivo observations that suggest that the new signaling paradigm for profibrogenic peptide secretion is relevant to the tubulointerstitial disease process. They provide solid circumstantial, but tantalizing, data to string together the LPA2 and αvβ6 integrin to heighten TGF-β signaling and overexpression of PDGF-B and CTGF in tubules and ensuing fibrosis after IRI. Furthermore, the in vitro culture system data indicated that the expression of epithelial-restricted αvβ6 integrin is rapidly increased after wounding of proximal tubular epithelium and during the reparative phase of tubular epithelial injury; in addition, the LPA2 receptor becomes overexpressed in kidneys after IRI in a manner that is temporally coincident with increased TGF-β signaling and profibrogenic peptide expression in tubules associated with fibrosis. These observations should provide impetus to further pursue research in this field and address certain critical questions. Does TGF-β directly control the production of PDGF-B and CTGF after a tubular injury in vivo, as the studies in cultured cells suggest? How is TGF-β signaling regulated in kidney tubules during health and disease? If, as the authors suggest, there are multiple and possibly redundant pathways to regulate TGF-β signaling in kidney tubules in vivo through GPCRs, what are those cellular events?

LPA Is One of the Several Possible Ligands to Activate the GPCR–αvβ6 Integrin–TGF-β Axis

Studies in lung epithelial cells by Jenkins and colleagues^9,10 have previously demonstrated that LPA and thrombin engage GPCRs [LPA2 and protease activated receptor (PAR)-1] to activate signaling via a Gαq and RhoA/Rho-associated kinase pathway that leads to αvβ6 integrin–dependent activation of latent TGF-β, steps that are identical to those reported for kidney epithelium by Venkatachalam and colleagues.⁸ Furthermore, these investigators correlated LPA-mediated TGF-β activation with lung fibrosis and convincingly demonstrated in their studies the spatial and temporal associations of increased epithelial expression of LPA2 and αvβ6 integrin in the regions of fibrosis in lungs that were experimentally injured by bleomycin administration and in human lung tissue with idiopathic pulmonary fibrosis.¹⁰ In addition, their studies indicated that thrombin stimulates αvβ6 integrin–mediated activation of TGF-β through PAR-1 receptors and a RhoA/Rho-associated kinase pathway in lungs; they thoroughly documented the dependence of experimentally induced acute lung injury on αvβ6 integrin– and PAR-1–dependent mechanisms.⁹ The similarity of these robust data of experimental and human lung diseases to those reported by Venkatachalam and colleagues is striking.⁸ The data of these studies are persuasive to indicate that the GPCR–αvβ6 integrin–TGF-β signaling axis has an important role to play in diverse contexts of a fibrotic disease process. Another serum lipid, sphingosine-1-phosphate, a less well-studied GPCR ligand that Geng et al⁸ discovered, was also capable of triggering TGF-β signaling. Thus, the GPCR signaling that activates TGF-β could conceivably have several signaling inputs in the early stages of the disease, while making the actual totality of signaling involved a complex pathobiological process.

The Case for Integrin-Dependent Epithelial TGF-β Activation in Fibrosis

Experimental evidence restricted to integrin-dependent activation of TGF-β is even more substantial than the detailed and intricate connections between GPCR activation, αvβ6 integrin, and TGF-β signaling, as alluded by Jenkins et al⁹; this was further extended by Geng et al⁸ to GPCR–αvβ6 integrin–TGF-β–dependent production of PDGF-B and CTGF. Munger et al¹¹ first established αvβ6 integrin–dependent activation of latent TGF-β as a major mechanism for spatially restricted synthesis of active TGF-β on the plasmalemmal surface of pulmonary epithelial cells, and they proposed this to be an underlying mechanism for experimental lung fibrosis. Interestingly, they also reported that β6 integrin–null mice exhibit exaggerated inflammation because of the lack of active TGF-β1 production but were protected from the development of fibrosis.¹¹ Similarly, Wang et al¹² showed that the β6 integrin–null phenotype conferred protection from TGF-β–mediated hepatic fibrosis after bile duct ligation without affecting the inflammatory component. Thus, by eliminating the effects of αvβ6 integrin–dependent TGF-β1 signaling on inflammation versus fibrosis, these investigators demonstrated that the profibrogenic signals of TGF-β1 generated on epithelial surfaces occur directly and are independent of intermediary inflammation steps.

Dependency on αvβ6 integrin for TGF-β1–mediated fibrosis has been shown in kidney diseases. Ma et al¹³ reported that renal fibrosis after unilateral ureteral obstruction is decreased in β6 integrin–null mice. Hahm et al¹⁴ described increased αvβ6 integrin expression in renal tubular epithelial cells in a variety of human chronic kidney diseases associated with fibrosis and in experimental Alport's syndrome in mice. This increase of αvβ6 integrin has been ascribed to the effects of enhanced TGF-β signaling.¹⁴ Conceivably, that would mean that the increased αvβ6 integrin expression and enhanced active TGF-β production form a self-sustaining loop of positively reinforced TGF-β signaling, and that αvβ6 integrin–neutralizing antibodies may be useful for dampening the fibrogenic response in chronic kidney diseases.¹⁴ Along these lines, several other integrin species have been described that can activate latent TGF-β.¹⁵ However, such integrins, although having activities similar to that of αvβ6 (ie, mediating the release of active TGF-β from latent precursor through actin cytoskeleton–dependent conformational changes), operate in mesenchymal cells, rather than in epithelia.¹⁵ One exception may be MDCK cells, in which αvβ3 has activated TGF-β.¹⁶ αvβ8 Integrin also activates TGF-β, but via a protease-mediated mechanism.¹⁵ Thus, αvβ6 integrin, restricted to epithelia, is uniquely well positioned to respond to signals in a manner alluded to by Jenkins et al⁹ and Venkatachalam and colleagues⁸ for the generation of active TGF-β.

Because TGF-β Is Profibrogenic, What Is the Need for Later Steps Involving PDGF-B and CTGF?

Regardless of the significance of GPCR-αvβ6 integrin signaling for latent TGF-β activation, a plausible mechanism, the findings of Venkatachalam and colleagues⁸ raise certain other intriguing questions. TGF-β acts independently profibrogenic because it increases the synthesis of extracellular matrix while decreasing its degradation, which potentially is mediated by paracrine stimulation of pericytes and fibroblast progenitors in the adjacent interstitium. It is intriguing to determine the significance of TGF-β–dependent synthesis and the secretion of two other profibrogenic peptides, PDGF-B and CTGF, by proximal tubule cells in the current scenario. The clues may lie in the unique and stringently controlled mechanisms by which unused nascent TGF-β becomes rapidly deactivated. After active TGF-β is formed on epithelial cell surfaces by the actions of αvβ6 integrin, ligand binding to TGF-β receptors is followed by endocytosis of the signaling complex and eventual degradation. Unused ligand left on the plasmalemmal surface is quickly deactivated by reconversion to latent TGF-β, and its bonding with latent TGF-β–binding proteins would likely form large inactive complexes.^11,15,17 The present consensus is that an entire TGF-β activation-deactivation machinery is located on the cell surface. This implies that active TGF-β, formed in vivo on proximal tubular cell surfaces by the αvβ6-mediated mechanism, is available within a short time frame for binding to its receptors on the same cell or those present in its immediate vicinity, but not to interstitial cells that lie beyond the barrier of the tubular basement membrane (TBM). Although latent TGF-β could diffuse across the TBM, it has to be activated by interstitial pericytes/fibroblast progenitors if signaling is to be initiated through their receptors. However, it is more likely that latent TGF-β does not exist as such in the extracellular matrix but becomes immobilized through binding to latent TGF-β binding proteins. Therefore, the generation of active TGF-β on proximal tubular cell surfaces in vivo may be expected to trigger autocrine TGF-β signaling that increases the production of TBM matrix by the same cells in an autologous manner; this results in thickened TBMs, a consistently observed pathological feature in kidneys with increased TGF-β signaling, but not in fibrosis, as conventionally understood. The latter involves (myo)-fibroblast proliferation in the interstitium and increased expression of types I and III collagens. It is in this context that TGF-β–dependent secretion of PDGF-B and CTGF by proximal tubular cells becomes pertinent. Both PDGF-B and CTGF do not require activation or become deactivated as active TGF-β does; therefore, they are able to diffuse across TBMs and then induce paracrine stimulation of fibroblast progenitors and pericytes in the renal interstitium. These considerations do not exclude the possibility of independent activating mechanisms for TGF-β signaling by interstitial cells in kidney disease. However, as previously considered, such processes that operate autonomously of tubules are unlikely to cause kidney fibrosis, except in contexts in which the primary site of disease initiation is the renal interstitium.¹ In sites where tubular damage is primary, a signaling pathway such as that proposed by Venkatachalam and colleagues⁸ is likely to be an initiating step that produces interstitial pathological characteristics via the generation of secreted paracrine peptides (PDGF-B and CTGF) that diffuse across the TBMs into the interstitium. In such a scenario, the evolution of progressive disease can only be explained by repetitive or relentless tubular injury that leads to comparable and parallel repetitive or sustained signaling stimuli, which ultimately would be a trigger for ensuing interstitial fibrosis.

Difficulties Ahead

Pandora's Box Has Too Many Surprises

As attractive as the signaling pathway delineated by Venkatachalam and colleagues⁸ might be, much remains to be done to place it in the proper context of in vivo disease process and to establish the importance of this mechanism in the development of fibrosis. This will be difficult in view of the many variables inherent to the complex signaling steps between LPA ligation of its receptors and the secretion of PDGF-B/CTGF, and also because of many potential redundancies with respect to both plausible GPCR ligands and alternative pathways to activate TGF-β. Geng et al,⁸ in their article, indicated that angiotensin II, sphingosine-1-phosphate, and thrombin are also candidate ligands for GPCR to exert their biological effects in acute kidney injury. In this regard, long-term infusion of angiotensin II has induced renal fibrosis via an epidermal growth factor receptor–mediated extracellular signal–regulated kinase–mitogen-associated protein kinase–dependent pathway that increases the TGF-β expression in proximal tubules.¹⁸ Whether angiotensin II increases TGF-β signaling by mechanisms similar to those described by Geng et al⁸ was not investigated. Despite these imponderable variables, LPA is a strong and viable candidate for relevant pathobiological signaling processes in states of kidney injury, including those settings that are conducive to fibrosis.

LPA is normally present in low concentrations (1 to 5 μmol/L) in serum and biological fluids and in higher levels in damaged tissue.¹⁹ Platelet aggregation can occur in blood circulating through the microcirculation, leading to the release of latent TGF-β, and in LPA and sphingosine-1-phosphate from platelets at the site of tissue injury. Locally increased LPA in tissue may also be derived as a metabolic by-product, by which activation of several enzymes leads to de novo synthesis and release of LPA. The renal content of LPA has significantly increased subsequent to unilateral ureteral obstruction, which is usually accompanied by fibrosis. Interestingly, the renal fibrosis is significantly attenuated in LPA1^−/− mice,²⁰ suggesting that LPA1 is involved in obstruction-induced tubulointerstitial fibrosis. How this may relate to the signaling paradigm described by Geng et al⁸ is unclear.

In summary, first, regardless of obvious merits, the study by Geng et al⁸ lacks direct in vivo evidence for LPA-induced LPA2 receptor–mediated αvβ6 integrin–dependent latent TGF-β activation and PDGF-B/CTGF secretion as they may relate to kidney fibrosis. Although the relevance of αvβ6 integrin in kidney fibrosis has been well described in the ureteral obstruction and Alport's syndrome models,^13,14 no such evidence exists for the role of the LPA2 receptor. LPA2 receptor–null mice are available, and it may be instructive to test whether they respond to renal injury in a manner that would be meaningful for addressing the questions under discussion. Second, participation of other latent TGF-β activators (eg, thrombospondin-1 and other integrins) cannot be excluded in states leading to fibrosis. Third, although Geng et al⁸ provide clear-cut evidence for the proposed mechanism in cultured cells, the cells targeted by LPA in vivo need to be identified and the LPA-mediated mechanism needs to be defined in that appropriate context. Thus, as is usual for studies that report a new paradigm to explain complex biological effects, their study generates more questions than answers. However, the potential ramifications of the intriguing and complex, but plausible, signaling pathway reported herein should provide impetus for addressing these challenging questions. Finally, by identifying LPA as a possible ligand that initiates/propels TGF-β signaling, their study raises an inevitable and mystifying question: given that LPA is ubiquitously present and, therefore, available to uninjured cells, what is the injury-related signal that modifies the LPA2 receptor such that it becomes permissive for LPA-mediated activation?