Fibrosis and progression of Autosomal Dominant Polycystic Kidney Disease (ADPKD)

Abstract

The age on onset of decline in renal function and end-stage renal disease (ESRD) in autosomal polycystic kidney disease (ADPKD) is highly variable and there are currently no prognostic tools to identify patients who will progress rapidly to ESRD. In ADPKD, expansion of cysts and loss of renal function is associated with progressive fibrosis. Similar to the correlation between tubulointerstitial fibrosis and progression of CKD, in ADPKD, fibrosis has been identified as the most significant manifestation associated with an increased rate of progression to ESRD. Fibrosis in CKD has been studied extensively. In contrast, little is known about the mechanisms underlying progressive scarring in ADPKD although some commonality may be anticipated. Current data suggest that fibrosis associated with ADPKD shares at least some of the “classical” features of fibrosis in CKD (increased interstitial collagens, changes in MMPs, over-expression of TIMP-1, over-expression of PAI-1 and increased TGFβ) but that there are also some unique and stage-specific features. Epithelial changes appear to precede and to drive interstitial changes leading to the proposal that development of fibrosis in ADPKD is biphasic with alterations in cystic epithelia precipitating changes in interstitial fibroblasts and that reciprocal interactions between these cell types drives progressive accumulation of ECM. Since fibrosis is a major component of ADPKD it follows that preventing or slowing fibrosis should retard disease progression with obvious therapeutic benefits. The development of effective anti-fibrotic strategies in ADPKD is dependent on understanding the precise mechanisms underlying initiation and progression of fibrosis in ADPKD and the role of the intrinsic genetic defect in these processes.

1. Autosomal dominant polycystic kidney disease (ADPKD)

ADPKD is the most common lethal monogenic human disease inherited as a dominant trait. It is estimated to affect at least 10 million individuals world-wide and to account for up to 10% of all patients with end-stage renal disease (ESRD) requiring renal replacement therapy (RRT) by haemodialysis or transplantation. ADPKD is caused by a spectrum of mutations in a single gene, PKD1 in 85% of cases or PKD2 in the remaining 15% of cases, which encode the proteins polycystin-1 (PC1) and polycystin-2 (PC2), respectively. The age of onset of renal functional decline in ADPKD is highly variable and can range from the 1st to the 8th decade [1]. Fifty percent of ADPKD patients reach ESRD by approximately 50 years of age, necessitating RRT, which, despite a few interesting leads to new therapeutic strategies, is currently the mainline treatment for this disease.

ADPKD is characterized by the progressive enlargement of multiple tubular cysts derived from all segments of the nephron. As discussed in detail elsewhere in this special issue, a variety of mechanisms have been identified which contribute to cyst formation including inter alia, increased Epidermal growth factor (EGF)- and cAMP-dependent epithelial cell proliferation, abnormal fluid secretion and apicobasal and planar polarity abnormalities [1–7]. Differences in epithelial responses to growth factors is also evident, for example, the negative regulation of EGF-dependent proliferation by Transforming growth factor (TGF)β seen in normal human renal tubule epithelia, is lost in ADPKD [2]. In addition, apoptosis is frequently seen in the normal tubules of early-stage, pre-dialysis (E-) ADPKD kidneys during the early phases of disease development, when there is active proliferative expansion of multiple small cysts. ADPKD is a progressive disease in which tubular epithelial cysts develop in utero and continue to expand after birth, gradually displacing normal renal parenchyma until >60% of normal nephrons are destroyed, thereafter rapid renal functional decline ensues to end-stage (ES-)ADPKD. The wide variability in age of onset of ESRD in ADPKD patients is thought to be influenced by the particular PKD mutation and the numbers of “second hits” or polymorphisms in disease-associated modifier genes and epigenetic factors [1, 8–10] although there is increasing evidence that haploinsufficiency alone is sufficient to elicit a cystic phenotype [11,12]. In patients with PKD1 mutations, onset of renal failure occurs earlier than in those with mutations in the PKD2 gene. The difference in severity is thought to be due to the development of an increased number of cysts in PKD1 rather than faster cyst growth [13].

Expanding renal cysts in ADPKD are surrounded by widely varying degrees of interstitial fibrosis. Although interstitial fibrosis has been described in models of PKD as well as in other genetic and acquired cystic diseases and has been suggested to have a significant impact on the large variation in the age of onset and rate of subsequent renal functional decline leading to ESRD in ADPKD [1], little is known about the pathogenesis of this component of the disease. Further, it is unclear whether fibrosis in ADPKD shares common mechanisms with fibrosis in chronic kidney disease (CKD) of other etiologies and/or whether there are disease-specific characteristics conferred by the genetic mutation (PKD1 or PKD2).

2. Fibrosis in Chronic kidney disease (CKD)

Progressive scarring or fibrosis of the renal parenchyma leading to organ failure is a characteristic feature of CKD of multiple etiologies [14–17]. Regardless of the initiating insult, CKD presents a common pathology of glomerulosclerosis and tubulointerstitial fibrosis. Importantly, the degree of tubulointerstitial involvement, rather than glomerular injury, provides the best correlation with progression to ESRD and the need for RRT. Tubulointerstitial fibrosis in CKD has been studied extensively and a brief overview is provided here for the purposes of comparison with fibrosis in ADPKD. Fibrosis associated with CKD presents a number of common characteristics including: a persistent inflammatory cell infiltrate; an increase in interstitial cell number with the appearance of myofibroblasts expressing the cytoskeletal protein α-smooth muscle actin (αSMA); tubular atrophy and epithelial apoptosis; obliteration of the peritubular capillaries; and, accumulation of extracellular matrix (ECM). The increase in interstitial cell number is due to both increased proliferation and decreased apoptosis of resident interstitial cells as well as migration of cells into the tubulointerstitium.

Interstitial fibroblasts/myofibroblasts are the major players in fibrosis responsible for the deposition of ECM. Fibrotic myofibroblasts appear to have multiple origins, arising by differentiation of resident interstitial fibroblasts [18] and infiltrating cells including bone marrow-derived progenitor cells (fibrocytes) and inflammatory cells [19]. A burgeoning literature also points to the transdifferentiation of tubular epithelial cells (epithelial-to-mesenchymal transition) as an additional source of myofibroblasts as well as contributing to the loss of tubular structures [20–22]. However, this remains an area of some controversy [23] and may be context-dependent. Recent lineage tracing studies and marker analyses have suggested that a proportion of interstitial fibroblasts derive from pericytes, the supporting cells surrounding peritubular capillaries [24], and from capillary endothelial cells by endothelial-to-mesenchymal transition [25,26]. Transition of pericytes and endothelial cells to myofibroblasts may also explain the loss of peritubular capillaries which is invariably associated with progressive fibrosis and leads to tubulointerstitial hypoxia, a potent driver of the fibrotic process [27].

2.1 Extracellular matrix metabolism in fibrosis

The hallmark of tubulointerstitial fibrosis is the progressive accumulation of ECM. In the normal kidney, the interstitial ECM comprises only a small proportion of the tissue volume. The matrix is composed of a variety of macromolecules including collagens, fibronectin, elastin, laminin, proteoglycans and non-collagenous glycoproteins. In tubulointerstitial fibrosis, expansion of the ECM is marked by accumulation of interstitial collagens and fibronectin. In addition to increased amounts of proteins, different isoforms may be expressed in disease versus normal tissues; for example, the EDA splice variant of fibronectin appears early in the fibrotic response. In addition to changes in the composition of the ECM, modification of proteins, such as crosslinking of collagens by Transglutaminase 2, may alter the mechanical properties of the matrix and render it resistant to normal degradative processes [28]. Compositional and mechanical changes in the ECM will lead to changes in cell-matrix interactions which are mediated, in large part, by cell surface matrix receptors such as integrins and sydecans [29]. Changes in matrix receptor profile in turn leads to alterations in down-stream signalling, gene expression and cell function. Thus changes in ECM may set in train a self-perpetuating cycle whereby alterations in the ECM induce changes is cell behaviour which in turn induce changes in ECM.

In fibrosis, both increased production and decreased turnover of ECM proteins contribute to matrix accumulation. Increased production is regulated primarily at the level of transcription of matrix protein genes while matrix turnover is mainly regulated by two proteolytic cascades, one mediated by the matrix metalloproteinases (MMPs) and their endogenous inhibitors, the tissue inhibitors of metalloproteinases (TIMPs), and the other by the serine proteases plasminogen activators (uPA and tPA) and the plasminogen activator inhibitors (PAIs) [15]. Patterns of expression of MMPs in fibrosis are complex with both up- and down-regulation that may be time-, location- and cell type-dependent. Interpretation of the in vivo effects is complicated: for example, although an increase in MMP-2 might be assumed to be anti-fibrotic by increasing ECM turnover, over-expression of MMP-2 exacerbates fibrosis, at least in part by promoting EMT [30]. Additional complexity comes from the fact that, in addition to their matrix-remodelling activities, MMPs cleave and activate a variety of biological molecules including cytokines and chemokines. Thus, increased MMP activity may increase expression of biologically-active, pro-fibrotic growth factors. The role of the TIMPs, of which 3 of the 4 family members are expressed in the kidney, is similarly complex. Fibrosis in vivo is invariably associated with an increase in TIMP-1 implying a decrease in MMP activity and hence increased ECM accumulation, however genetic deficiency of TIMP-1 does not reduce the fibrotic response [31]. Indeed, genetic deletion of TIMP-3 exacerbates myofibroblast activation [32]. MMP-independent effects of TIMPs on cell proliferation, survival and differentiation may also be relevant to the pathogenesis of fibrosis [33]. Increased expression of PAI-1 is also characteristic of CKD [15,34]. Although it was assumed that the pro-fibrotic effects of PAI-1 were due to its role as an inhibitor of serine proteases this has not proved to be the case, rather PAI-1 promotes fibrosis by increasing macrophage and myofibroblast recruitment to the tubulointerstitium [34]. tPA and uPA are increased in at least some models of fibrosis and tPA has been shown to be pro-rather than anti-fibrotic [34].

2.2 Mechanisms of fibrosis

A number of mechanisms have been implicated in the initiation and pathogenesis of fibrosis in CKD [14,17,22,27,34,35] including i) an imbalance in growth factors and cytokines with up-regulation of pro-fibrotic factors including inter alia, TGFβ1, Connective tissue growth factor (CTGF), platelet-derived growth factor (PDGF), Fibroblast growth factor (FGF)-2, Osteopontin and Endothelin (ET)-1 and concomitant down-regulation of anti-fibrotic factors such as Hepatocyte growth factor (HGF) and Bone morphogenic protein (BMP)-7; ii) dysregulation of vasoactive factors with an increase in vasoconstrictors most notably Angiotensin II (Ang II), and a decrease in vasodilators such as Nitric oxide; iii) hypoxia as a consequence of microvascular injury and obliteration; and iv) disruption of normal homeostatic interactions between adjacent cell populations, in particular, the interactions between tubular epithelial cells and interstitial fibroblasts. Further, the close juxtaposition of epithelial cells, fibroblasts, endothelial cells and immune cells within the tubulointerstitium suggests important but as yet largely unexplored, interactions between the microvascular endothelium of peritubular capillaries and both tubular epithelial cells and fibroblasts as well as between immune cells (notably, monocytes and macrophages) and tubulointerstitial cells [36,37].

2.3 Growth factors and signalling pathways activated in fibrosis

Although numerous fibrogenic factors have been reported, TGFβ1 is generally regarded as the pre-eminent fibrogenic cytokine which is up-regulated in every form of CKD both in humans and in animal models and activation of the downstream Smad signalling pathway is a hallmark of fibrosis [38]. TGFβ also signals though a variety of non-Smad pathways including ERK and JNK/p38 MAP kinase, small GTPAse and PI3K/Akt pathways [39] many of which have been shown to be activated in fibrosis. Other intracellular pathways stimulated in fibrosis include the Ras pathway which drives fibroblast proliferation [40], integrin-linked kinase (ILK) [41] and Wingless/Int (Wnt) [42].

2.4 Epigenetics and fibrosis

Recent studies suggest a role for epigenetic changes in the pathogenesis of fibrosis. Epigenetic changes are heritable changes in gene expression which do not involve changes in DNA sequence and include DNA methylation and histone modification. Recent exciting findings suggest an important role for DNA methylation in the perpetuation of the scarring process [43]. Further, prolonged exposure of fibroblasts to TGFβ leads to hypermethylation and epigenetic changes which perpetuate the fibrotic phenotype [43]. This study highlights the importance of non-genetic DNA modifications in the pathogenesis of disease

2.5 microRNAs and fibrosis

Another area of rapidly expanding interest in fibrosis is that of the role of microRNAs (miRNAs). miRNAs are short double-stranded RNA sequences (20–23 nucleotides) which, in the main, regulate gene expression at the translational level either by repressing mRNA translation or reducing mRNA stability [44] although recent studies suggest that miRNAs may also enhance mRNA translation in response to cellular stress [45]. miRNAs have been implicated in the control of diverse biological functions relevant to fibrosis including proliferation, differentiation and apoptosis and a number have been shown to be up- or down-regulated in fibrosis [46]. Several miRNAs are highly expressed in the kidney compared to other organs including miR-192, -194, -204, -215 and -216 [47,48]. Expression of miR-192 is up-regulated in the mouse model of unilateral ureteral obstruction (UUO), the rat remnant kidney [49] and in diabetic mouse glomeruli and mesangial cells treated with TGFβ [50]. In contrast, Krupa et al [51] reported decreased expression of miR-192 in human diabetic nephropathy. In TGFβ-treated mesangial cells, increased collagen α1(2) expression is mediated by miR-192 repression of the E-box repressor, Zeb2. Down-regulation of Zeb2 induced miR-216a and -217 which also contributed to increased collagen expression via suppression of PTEN and activation of Akt [52]. miR-377 is also up-regulated in the diabetic mouse kidney and increases expression of fibronectin via down-regulation of p21-activated kinase and superoxide dismutase [53]. Although this is still a very new field, it seems likely that miRs will become targets for novel anti-fibrotic therapies.

3. ADPKD and progression

Although, in a small subset of PKD patients, very early onset ESRD has been associated with the location (N-terminal) of PKD1 mutation [54], in the vast majority of cases, no predictive link between progression and mutation has been found. Onset of ESRD in patients with a PKD2 mutation is delayed compared to PKD1 mutations [13]. Similar to the correlation between tubulointerstitial fibrosis and progression of CKD, in ADPKD patients the degree of fibrosis has been identified as the most significant and variable manifestation that is clearly associated with an increased rate of progression to ESRD [54–56]. However, relatively little is known about the mechanisms underlying fibrosis in ADPKD. Unlike in CKD where kidney size is unchanged or reduced, ADPKD kidneys enlarge dramatically and progressively throughout the disease course even into end-stage disease apparently as a result of both cystic epithelial and interstitial expansion [1,57,58]. Although some commonality in the mechanisms underlying fibrosis in ADPKD and CKD may be anticipated, it seems likely that the underlying genetic defect will also contribute unique disease- and stage-specific processes which may provide novel targets for therapeutic intervention.

3.1 Inflammation and fibrosis

One of the hallmarks of fibrosis is a persistent inflammatory infiltrate [36,37]. Inflammmatory cells have also been observed in ADPKD kidneys often in association with expanding cysts. Gene expression profiling of end-stage ADPKD kidneys revealed expression of genes associated with immune or inflammatory responses [59]. Increased expression of PAI-1 in cystic epithelial may also suggest an increased inflammatory response [34].

3.2 Extracellular matrix (ECM)

Pathological examination of human ADPKD kidneys compared to age-matched controls [2, 57,60–63] as well as phenotypic and genotypic animal models of PKD [64,65], show abnormalities in peri-cystic ECM deposition in basement membranes and the interstitium as well as increased numbers of interstitial cells. Early-stage, pre-dialysis ADPKD kidneys are characterized by proliferative expansion of cyst-lining epithelial cells and alterations in the ECM composition and thickness of the basement membranes (Figure 1b versus a). Correlative in vivo/in vitro analyses of human ADPKD (PKD1 mutant) cyst epithelia versus age-matched normal tubular epithelial cells identified specific abnormalities of ECM composition and turnover including alterations in collagen types I, III and IV, fibronectin and in heparan sulphate proteoglycan (HSPG) turnover, with the resultant increase in cystic epithelial proliferation [7,60]. In end-stage (ES-)ADPKD kidneys, a variable degree of expansion of the interstitial compartment is seen due to progressive accumulation of fibrogenic cells (fibroblasts) and collagen fibrils (Figures 1c–e). Gene expression analyses identified global increases in interstitial collagens type I and type III as well as in the basement membrane collagen type IV. Immunofluorescence analysis localized these increases to the basement membrane structures surrounding small cysts as well as in the fibroblasts deep in the interstitium of ES-ADPKD kidneys (Figure 1f). The development of renal fibrosis in ADPKD is characterized by focal changes. For instance, tenascin, which is seen as a thin linear, homogeneous lining between tubules in normal human kidneys is concentrated in focal areas of matrix accumulation in close association with the basement membranes of epithelia lining small cysts that are particularly prevalent in E-ADPKD kidneys (Figure 1g). Consistent with findings in CKD increased EDA-FN expression was recently reported in Pkd1-null mice [66]. Fascinating recent findings in zebrafish suggest the polycystins may directly regulate collagen production and suggest that over-production of collagen may be a consequence of polycystin loss of function [67].

Figure 1. Progressive fibrosis in ADPKD.

a–d. Toluidine blue-stained 1µm plastic sections of: a. Normal human kidney with thin basement membranes and little interstitial space; b. E-ADPKD kidney with thickened basement membranes and increased numbers of interstitial cells; c. ES-ADPKD kidney with 2+ interstitial fibrosis between cystic epithelial dilations; d. ES-ADPKD kidney with 3+ interstitial fibrosis between epithelia of large cysts; e. Transmission electron micrograph of ES-ADPKD kidney with 4+ interstitial fibrosis showing interstitial fibroblasts and collagen fibrils underlying a flattened cyst lining epithelia cell [57]; f. Immunofluorescence of collagen type I underlying cyst epithelia and in interstitial fibroblasts in ES-ADPKD kidney (unpublished data); g. Avidin-biotin-peroxidase-aminoethylcarbazole immunohistochemistry of focal accumulation of tenascin in E-ADPKD kidney [2]. Note: Fibrosis scores: 1+ less than 20%; 2+ 2–40%; 3+ 40–60%; 4+ >80% on analysis of 10 random pathology sections.

3.3 ECM turnover

The integrity and amount of ECM is maintained, at least in part, by regulation of matrix turnover. A number of studies have reported alterations in MMPs and TIMPs in animal models of PKD [68] and present a complicated and often apparently conflicting picture. In a rat model of PKD, epithelial expression of MMP-2 decreased [69] while MMP-14 (MT1-MMP) increased [70]. MMP-14 expression was also increased in fibroblasts in the vicinity of both cystic and non-cystic tubules in this model. In contrast to the findings in the rat, tubules from the Cpk mouse (a model of ARPKD in which cilial cystein is absent) in vitro secreted increased MMP-2 and also MMP-9 [71]. Recently Hassane S et al reported increased MMP-2 and MMP-14 gene expression in the kidneys of Pkd1-mutant mouse models [66]. Data on matrix remodelling enzymes and their inhibitors in human ADPKD is sparse. Compared to controls, serum levels of MMP-1, MMP-9 and TIMP-1 were reported to be increased in ADPKD patients [72]. A comparative analysis of MMP-2, MMP-9, stromelysin (MMP-3), TIMP-1 and-2 and PAI-1 in tissue sections from age-matched normal, E- and ES-ADPKD kidneys demonstrated a consistent pattern of increased MMP expression in the epithelia of small cysts in E-ADPKD human kidneys (Figure 2b versus a) and in the interstitial fibroblasts in ES-ADPKD kidneys (Figure 2c). TIMP-1 was also increased and localized to interstitial fibroblasts in ES-ADPKD kidneys (Figure 2d) while increased PAI-1, was characteristic of ES-ADPKD cystic epithelia (Figure 2e) (our unpublished data). PAI-1 expression was also reported to increase in Pkd1-deletion mouse models [66]. The recent demonstration of fragments of collagen I and III in urine of young, dialysis-independent APDKD patients highlights the importance of ECM remodelling in disease and may suggest disease-specific changes in proteases [73]. Interestingly, there was both up-regulation and down-regulation of particular fragments suggesting a complex interplay of accumulation of some matrix components and turnover of others. A detailed time-course of protease activity, expression of protease inhibitors and matrix proteins in slowly progressive models which closely mimic the human disease may provide insight into the role of ECM metabolism in disease.

Figure 2. Immunohistochemistry of MMP-2, TIMP-1 and PAI-1 in human ADPKD.

a. MMP-2 expression in normal human adult kidney showing little reaction product (red); b. E-ADPKD showing strong staining of MMP-2 in epithelia lining small cysts; c. ES-ADPKD showing MMP-2 staining in interstitial cells; d. TIMP expression in interstitial fibroblasts in ES-ADPKD; e. PAI-1 in epithelia lining large cysts in ES-ADPKD (unpublished data).

3.4 ECM receptors

Cells and ECM interact in a reciprocal fashion to regulate proliferation, differentiation and function. These cell-matrix interactions are mediated via specialized matrix receptors; the integrins and proteoglycan-containing syndecans, which act in concert with growth factors and their receptors [74]. At points of cell-matrix interaction, integrins and proteoglycans are clustered into focal adhesion complexes [75] that provide a dynamic nucleation site for the activation of multiple intracellular signaling pathways to regulate gene expression and cell function [75,76]. The components of focal adhesions are diverse and include scaffold proteins, actin-binding proteins, GTPases, kinases such as focal adhesion kinase (FAK), src and fyn, phosphatases and lipases. The interdependence of matrix, receptors, signalling and gene expression means that changes in one component will inevitably lead to altered cell function. Changes in matrix receptor expression in a variety of cell types have been described in CKD and are implicated in the pathogenesis of progressive renal fibrosis [29]. In ADPKD, the changes identified to date are confined to the cystic epithelial cell and show increased expression and syndecan-4 [62,63] (our unpublished data), integrin α8 [77], αv [78] and β4 [79]. Furthermore, recent studies have also identified high levels of expression and alterations in the polarized distribution of integrin-associated phospho-FAK in E-ADPKD cystic epithelia suggesting activation of matrix receptor-mediated intracellular signalling. To date no detailed analysis of fibroblast matrix receptor expression in ADPKD has been performed but pathogenically-relevant changes might be anticipated.

3.5 Growth factors and signalling pathways

A number of growth factors implicated in fibrosis are up-regulated in ADPKD. EGF and EGF family ligands have been shown to be important mediators of cystic epithelial proliferation [2, 5]. TGFβ, the major fibrogenic cytokine, is highly expressed in cystic epithelia in human ADPKD kidneys [2,59], Han:SPRD-Cy/+ rats [80] and Pkd-1 deletion mouse models [66]. Studies in rodent models suggest that increased TGFβ expression and signalling correlates with later stages of the disease implicating it in progression (and fibrosis) rather than cyst initiation [66]. Levels of FGF-1 are also increased in end-stage human ADPKD [81].

As discussed elsewhere in this special issue, a number of intracellular signalling pathways are activated in cystic lining epithelial cells in ADPKD many of which are also activated in fibrosis. Preliminary immunohistochemical analysis showed increased phospho-JNK expression in interstitial cells of ES-ADPKD kidneys (our unpublished data), however the specific pathways involved in fibrosis associated with ADPKD remain to be defined.

3.6 Myofibroblast differentiation

An increase in interstitial cell number and the appearance of myofibroblasts expressing cytoskeletal αSMA, are cardinal features of fibrosis and these cells are thought to play a major role in ECM accumulation [19]. Immunohistochemical staining of αSMA in human ADPKD kidneys showed the focal appearance of αSMA staining in some interstitial cells in early disease with more extensive and marked expression in ES-APDKD, consistent with accumulation of myofibroblasts and interstitial ECM expansion in the later stages of the disease (Figure 3). Increased numbers αSMA-positive cells were also observed in Pkd1-deletion mouse models where they appeared as a layer around cysts [66]. Studies on expression profiling of end-stage APDKD kidneys and kidneys of Han:SPRD-Cy/+ rats also reported increased expression of smooth muscle proteins indicative of myofibroblast expansion [59,80]. Calcium is reported to play an important role in myofibroblast differentiation [82], given that PC2 acts as an endoplasmic reticulum calcium channel, it is interesting to speculate that mutations in this protein and alterations in intracellular Ca levels may have pathologically-relevant effects on fibroblast differentiation. As mentioned in the context of CKD, the origin of myofibroblasts in fibrosis remains controversial with a variety of cell sources potentially contributing to the increased number of αSMA-positive cells. In ADPKD, the derivation of these cells is unknown though it seems likely that increased proliferation and differentiation of interstitial fibroblasts will account for at least some of the increase [81]. Expression profiling studies of end-stage ADPKD kidneys and kidneys of the phenotypic ADPKD model, the Han:SPRD Cy/+ rat, showing increased expression of gene sets associated with EMT suggest this may be a pathogenic factor in ADPKD [59,80]. Marker studies in the genotypic ARPKD Pck rat model also suggest that cystic epithelial cells may express mesenchymal markers [83]. TGFβ expression is elevated in ADPKD and TGFβ readily induces EMT in normal tubular epithelial cells in vitro. However, based on immunohistochemical staining for Fsp1 (a fibroblast marker) and αSMA, a recent study in mice with Pkd1-deletion concluded that if EMT contributes to fibrosis, the contribution is likely to be small [66]. Two other lines of evidence mitigate against a major contribution from EMT. Many of the demonstrations of EMT rely on changes in expression of epithelial and mesenchymal markers rather than functional changes. For bona fide EMT cells must leave the tubular structure and migrate into the tubulointerstitium a process that requires the dissolution of the basement membrane however in ADPKD the basement membrane is frequently thickened rather than rarefied [57] and in vitro studies suggest cystic cells adhere more strongly to the ECM and are less migratory than their normal counterparts [84].

Figure 3. Myofibroblast accumulation in ADPKD.

Immunohistochemistry of αSMA (red deposits). a. Normal human adult kidney showing no reaction; b. E-ADPKD showing focal, light reaction product in the interstitium; c. ES-ADPKD showing strong reaction product in the fibrotic interstitium (unpublished data).

Expression studies comparing normal human proximal tubular cells, early and end-stage ADPKD epithelial cells in vitro showed increased expression of MMP-2, αSMA and heat shock protein (HSP) 47, a chaperone protein involved in collagen metabolism and up-regulated in fibrosis, in E- and ES-ADPKD. Levels of E-cadherin were decreased in ES- but not E-ADPKD epithelia (our unpublished data). While these data may suggest EMT can occur in cystic epithelial cells these changes may also represent de-differentiation of the epithelia. Whether fibrocyte-, endothelial-, pericyte- or inflammatory cell-to-mesenchymal transdifferentiation occurs in ADPKD to increase the number of interstitial cells has not been examined. Lineage tracing studies will be required to define the origin(s) of the interstitial cells in ADPKD.

3.7 Expression of PC1 and PC2

The PKD gene products Polycystin-1 (PC1) and Polycystin-2 (PC2) are localised to a number of cellular sites including on the plasma membrane at sites of cell-cell and cell-matrix interactions, in the endoplasmic reticulum, the nucleus and most importantly, the primary cilium where the polycystin complex is thought to act as a flow sensor for epithelial cells [85].

The PKD1 gene encoded PC1 is a large (~460kDa) trans-membrane protein with a large extracellular region containing a variety of domains associated with protein-protein and protein-carbohydrate interactions, 11 transmembrane domains and a short cytoplasmic tail. Overall the structure of PC1 suggests a receptor or adhesion molecule and interactions with a large number of proteins including PC2, have been reported (reviewed in [86] and elsewhere in this special issue). PC1 is highly expressed in the developing kidney, predominantly in the ureteric bud-derived collecting tubules. In normal kidneys it is down-regulated after birth and restricted to inner medullary collecting ducts where it has been localized to the apical-lateral membranes and is thought to play a role in the maintenance of normal differentiation, including lumen size [1,58,87]. Since all ADPKD patients are heterozygous for a germ-line PKD1 or PKD2 mutation and second somatic “hits” may also occur, it is not surprising that analysis of numerous (>200) ADPKD (PKD1) kidneys has shown a wide variation in expression of PC1 protein in cystic epithelia [84]. In ADPKD, low, moderate or high levels of PC1 expression are seen in cystic epithelia [84], although this is often associated with abnormal cellular distribution including nuclear localisation of the C-terminal tail [88,89] and, presumably, abnormal protein function. In general, higher levels of PC1 are seen in E-ADPKD epithelia lining small cysts than in ES-ADPKD cyst-lining epithelia. Interestingly, although interstitial PC1 in normal kidneys has occasionally been demonstrated in vascular smooth muscle cells surrounding small vessels this has only been achieved after extreme antigen unmasking procedures [90], while in ES-ADPKD kidneys, PC1 was detected in fibroblasts within the fibrotic interstitium (our unpublished data). PC2 is a much smaller (~110kDa) protein with 6 transmembrane domains and short N- and C-terminal tails both of which are intracellular. PC2 functions a non-selective cation channel that transports calcium from the endoplasmic reticulum whose function is thought to be to regulate intracellular calcium [85,91 and reviewed elsewhere in this special issue]. The pathological relevance of expression of interstitial expression of polycystins remains to be established but the recent link between loss of polycystin function and increased collagen production [67] raises fascinating questions about the potential role of these proteins in fibrosis.

3.8 Vascular remodelling, hypoxia and fibrosis

Loss of peritubular capillaries and tissue hypoxia is a characteristic of tubulointerstitial fibrosis in CKD and is regarded as a final common pathway to ESRD. However, in this setting, peritubular capillary injury is a consequence of glomerular injury since the peritubular network is fed by the glomerular outflow. This scenario is not reiterated in ADPKD where cystic development and expansion are independent of glomerular alterations. However ADPKD cyst formation in human and in animal models is accompanied by vascular abnormalities [1,92,93] and morphological analyses provide evidence of disruption of the normal peritubular microvasculature. ADPKD cyst walls contain an extensive but abnormal capillary network and suggest that endothelial cells express MMP2, integrin αvβ3 and Vascular endothelial growth factor receptor (VEGFR)-2 but not VEGFR-1 [94]. In addition to the renal defects, ADPKD is associated with serious extra-renal vascular complications including hypertension, intra-cerebral aneurysms, aortic dissections, sclerotic vascular changes and vessel fragility. Hypertension is the most common presenting clinical manifestation of ADPKD and can be detected up to two decades prior to ESRD, it is thought to be due to abnormalities in the renin-angiotensin system (RAS;[95,96]). A body of evidence suggests that endothelial dysfunction in ADPKD, which precedes hypertension in these patients, is associated with altered vasoconstriction as a result of changes in the RAS, up-regulation of ET-1 and decreased nitric oxide (NO) release apparently due to diminished constitutive nitric oxide synthase (cNOS) activity [97–100]. A similar imbalance of vasoactive factors is a contributory factor in fibrosis in CKD. Other elements of the vasculature are also affected in ADPKD, including abnormalities in calcium regulation which have been implicated in functional changes of ADPKD vascular smooth muscle cells [101,102].

Studies in humans and in rodent models provide evidence of hypoxia in APDKD kidneys. Hypoxia-inducible transcription factors (HIFs) were differentially expressed in cystic epithelial cells (HIF-1α) and pericystic stromal cells, inflammatory cells and endothelium (HIF-2α) [103]. Gene profiling of PKD1 cysts displayed a rich profile of gene sets associated with hypoxia [59]. While hypoxia in ADPKD is posited as a driver of pericystic hypervascularity providing expanding cysts with nutrients it may also drive a profibrotic response particularly in the interstitial cells [27].

3.9 Epigenetic changes, miRNA and fibrosis

As alluded to earlier, two rapidly evolving areas of fibrosis research are the role of epigenetic changes and the role of miRNAs in progressive scarring. The role of both these processes in ADPKD fibrosis is at this stage, largely speculative but worthy of investigation. Are epigenetic changes in ADPKD fibrosis the same as those identified in CKD where TGFβ induces methylation of target genes to perpetuate the fibrotic phenotype [43]? Increased expression of TGFβ in ADPKD may suggest this is likely to be the case particularly as ADPKD fibroblasts retain the fibrotic phenotype in culture in the same way as other fibrotic fibroblasts. Alternatively or in addition, there may be disease-specific epigenetic changes.

While conclusive evidence of miR involvement in APDKD is currently lacking it seems likely that identification of pathologically-relevant up- and down-regulation of different miRNAs will be forthcoming. A recent publication [104] implicated miR15a in hepatic cystogenesis in a rat model of PKD; decreased miR15a was associated with up-regulation of its target cell division cycle (Cdc) 25A, increased proliferation and cyst growth. The question here too is whether miRNA-induced changes in ADPKD fibrosis will be the same as in other forms of CKD or are there disease-specific alterations?

4. ADPKD fibroblasts in vitro

4.1 ADPKD vs normal fibroblasts

Numerous in vitro studies have compared normal human renal tubular epithelial cells with ADPKD cyst epithelia and have identified a variety of functional differences in cystic epithelial cells compared to their normal counterparts including increased proliferation, altered sensitivity to growth factors, increased adhesion of ECM, decreased migration, abnormal apical-basal polarity and altered secretion [2,5,60,85].

Studies comparing fibroblasts from normal and fibrotic tissues have shown that cells retain the fibrotic phenotype in vitro. Although studies on ADPKD fibroblasts are limited, comparison of primary cultures of age-matched normal human kidney and ADPKD (PKD1 mutations) fibroblasts from E- and ES-ADPKD kidneys show disease- and stage-specific differences [81]. Compared to age-matched normal cells, both E- and ES-ADPKD fibroblasts in vitro show an enhanced proliferative response to selected growth factors including FGF-1, PDGF, TGFβ, Insulin-like growth factor (IGF)I and IGFII but not EGF. ES-ADPKD fibroblasts are capable of growth in soft agar [81]. ADPKD fibroblasts synthesized and secreted more FGF-1 [81] and FGF-1 elicited a more rapid and persistent tyrosine phosphorylation of intracellular proteins in this cell type compared to normal renal fibroblasts. ADPKD fibroblasts secreted high levels of uPA which was further enhanced by FGF-1 [81]. In addition to increased FGF-1 expression, ES-ADPKD fibroblasts express increased levels of mRNAs for TGFβ, HSP47, MMP-2 consistent with a profibrotic phenotype. Expression of αSMA was also up-regulated (unpublished data). Both normal and ADPKD fibroblasts expressed PKD1 mRNA and PC1 protein, however PKD1 mRNA was increased in ES-ADPKD fibroblasts compared to age-matched, normal counterparts (unpublished data).

4.2 Fibroblasts and cilia

Polycystins are identified as focal adhesion, cell-cell adherens junction and ciliary proteins and ciliary defects in cyst epithelial cells have been implicated in the pathogenesis of ADPKD [1,85,86,91]. Less well recognized is the observation that fibroblasts, including human kidney fibroblasts, also possess a primary cilium (Figure 4). Cilia are thought to act as flow-, chemo- and mechano-sensors and together with focal adhesions and cell-cell adherens junctions, play an important role in integrating signaling pathways [105–107]. It has recently been shown that in ORPK (IFT88(Tg737Rpw)) mouse fibroblasts in which ciliary assembly is defective, responses to PDGFAA are abnormal and wound healing is defective [108]. Also of interest is the observation that, cilia are reduced in cells treated with TGFβ. The effect of PKD1 or PKD2 mutations on cilia and adult fibroblast function in the context of fibrosis is a question of considerable interest particularly as signalling systems that may be co-ordinated by the cilium include PDGFRα, Hedgehog and Wnt pathways which are also implicated in fibrosis. Some insight into the effects of genetic deletion of fibroblasts may be gained from studies in mouse embryonic fibroblasts (MEFs). Pkd1−/− MEFs have been reported to show increased proliferation and increased cell size [109]. Other investigators have reported reduced migration in Pkd1-null MEFs [110]. Also relevant to the pathogenesis of ADPKD is the demonstration of cilia in other mesenchymal cells including vascular smooth muscle cells [111,112] and changes in cilia have been implicated in vascular abnormalities in ADPKD.

Figure 4. Quiescent normal human kidney fibroblasts in vitro express primary cilia.

a. Scanning electron micrograph showing primary cilium (arrowhead); b. Immunofluorescence of nuclei stained with DAPI (blue) and cilia (arrowhead) stained with FITC-labelled antibody to acetylated tubulin (green), a marker of cilia; low power field showing multiple nuclei each with an associated cilium; c. High power detail (unpublished data).

5. Epithelial-fibroblast interactions, ADPKD and fibrosis

One of the recurrent themes that emerges in looking at fibrosis in ADPKD is that cystic changes in the epithelium are associated with local changes in adjacent fibroblasts. This is particularly striking in detailed chronological studies in animal models, for example, in the Han:SPRD rat model of ADPKD [113] increased interstitial cell proliferation and matrix deposition is associated with sites of increased epithelial cell proliferation and thickened basement membrane. Similarly, in Pkd2 mutant mice as well as in human ADPKD there was a strong correlation between cyst proliferation and fibrosis score [114]. Likewise early alterations in synthesis and secretion of MMPs and TIMPs in ADPKD kidneys appear to precede interstitial changes [115]. Collectively these data suggest an epithelial cell-initiated mechanism for fibrosis in ADPKD and imply an important role for epithelial-fibroblast interactions in the pathogenesis of this disease.

Classical embryological experiments first demonstrated the essential nature of reciprocal, sequential interactions between epithelial and mesenchymal cells for the normal growth and differentiation of the developing kidney [116,117]. In the adult organ, epithelial–stromal interactions, mediated by direct cell contact, the ECM and diffusible paracrine factors [118–120], continue to play an important role in maintaining normal tissue homeostasis and in tissue repair following injury. Dysregulation of epithelial-stromal interactions has been implicated in the pathogenesis of a variety of diseases including various cancers. In fibrosis it is proposed that dysregulated function of, and impaired communication between, epithelial and fibroblastic cells prevents the resolution of the wound-repair response following injury, leading to an on-going scarring process. Furthermore, in CKD, disruption of interactions between tubular epithelial cells and the underlying fibroblasts may be an early event in the pathogenesis of fibrosis with tubular cell injury induced by, for example, by proteins and cytokines in the filtrate from damaged glomeruli, leading to activation of the underlying fibroblasts which in turn exacerbate tubular damage. However, since ADPKD cysts close off from the nephron of origin rapidly after their initiation [1], this mechanism is unlikely to be significant in ADPKD. While altered tubular epithelial-interstitial fibroblast interaction is an attractive mechanism to explain how glomerular injury is transmitted to the tubulointerstitial compartment relatively few studies have examined the mechanisms of interaction between these two cell types [119–121]. Interestingly, our previous studies identified a paracine, PDGF–mediated, regulatory loop between inner medullary collecting duct epithelial cells and medullary fibroblasts [119] that was not apparent between proximal tubular epithelial cells and cortical fibroblasts, highlighting not only the importance of tubular epithelial-interstitial fibroblast interactions but raising the possibility of segment-specific regulatory interactions. Reciprocal effects of fibroblasts on epithelial cells to drive a fibrotic response have been demonstrated in other models, for example Wnt-4-producing fibroblasts (αSMA and Col1a1 positive) placed under the renal capsule induce tubular destruction [42].

The primary defect in ADPKD is caused by alterations in the development and differentiation of renal tubule epithelial cells leading to inappropriate, persistent proliferation of these cells after birth and consequent cystic expansion. Many factors can influence proliferation of epithelial cells including the secretion of growth factors and accessibility to their functional receptors as well as the composition of the basement membrane and ECM from which epithelial cells receive instructive signals. A number of studies have shown that cyst-lining ADPKD epithelia are hyper-proliferative in vivo and parallel in vitro studies have shown this to be due to hyper-responsiveness to EGF and cAMP as well as to a variety of ECM proteins. Autocrine and paracrine signals generated by proliferating cystic epithelial differ from those produced by normal, quiescent tubular epithelial cells and are likely to induce different, potentially pathologic, responses in the underlying fibroblasts which in turn may have deleterious effects on the cystic epithelia.

Fibrosis develops secondary to cystic changes suggesting that if the epithelial abnormality is corrected fibrosis will be obviated. However evidence derived mostly from the developmental and cancer literature implies that fibroblasts can induce changes in epithelial phenotype and function [122]. Interestingly, preliminary co-culture experiments combining normal or ADPKD fibroblasts and normal proximal tubular epithelial cells suggest that compared to normal fibroblasts, ADPKD fibroblasts suppressed the expression of an epithelial differentiation marker E-cadherin (unpublished data) which is also associated with EMT [21] and is reduced in ADPKD epithelial cells compared to their normal counterparts where it is replaced by N-cadherin [123]. These data may suggest that correction of the epithelial defect alone may not be sufficient to prevent fibrosis.

6. Fibrosis in APDKD: Conclusions

Taken together the current data indicate that fibrosis associated with ADPKD shares at least some of the “classical” features of fibrosis in CKD (increased interstitial collagens, changes in MMPs, over-expression of TIMP-1, over-expression of PAI-1 and increased TGFβ) but that there are also some unique and stage-specific features. Epithelial changes appear to precede and to drive changes in the interstitium leading to the proposal that development of fibrosis in ADPKD is a biphasic process with alterations in the cystic epithelia followed by changes in the interstitial fibroblasts and that reciprocal interactions between these cell types precipitates a progressive accumulation of ECM in the interstitial compartment (Figure 5). If there is sequential induction of a fibrotic cascade it might be possible to intervene and prevent the process before the establishment of a reciprocal “vicious cycle” of fibrosis. The role of PC1 and PC2 in fibroblasts; the effect of mutations in these genes on fibroblast function; and, whether mutation per se leads to alterations in fibroblast phenotype or render the fibroblasts susceptible to pro-fibrotic signals produced by epithelial cells all remain to be determined.

Figure 5. Schematic of the proposed model of fibrosis in ADPKD.

Normal kidney structure and function is maintained by dynamic, homeostatic interactions between adjacent cell populations. In ADPKD mutations in PKD1 or PKD2 lead to epithelial alterations including: αβ integrin profile, downstream signalling including P-FAK, P-MAPK, P-JNK; MMP-2 secretion; altered response to TGFβ; changes in PC1 distribution/function. Epithelial changes in turn induce changes in underlying fibroblasts including: synthesis/secretion collagens I, III, MMPs, TIMPs, FGF-1, uPA; differentiation to myofibroblasts expressing αSMA; induction of PC1. Dynamic reciprocal interactions between the two cell types (arrows) dictate the overall pathological phenotype and may depend upon the PKD mutation.

7. Anti-fibrotic therapies in ADPKD

Slowing cystic expansion and the increase in kidney size in ADPKD, thereby sparing normal renal tubule destruction, is key to delaying the need for RRT. Since fibrosis is a major component of the increase in kidney size in ADPKD it follows that preventing or slowing fibrosis should retard disease progression with obvious therapeutic benefits. While work in CKD may help identity potential targets for anti-fibrotic therapies [17,27,124] the development of effective anti-fibrotic strategies in ADPKD is dependent on understanding the precise mechanisms underlying the initiation and progression of fibrosis in this disease and the role of the intrinsic genetic defect in these processes. These studies are in their infancy but it seems likely that future therapeutic strategies for ADPKD will be combinatorial, targeting not only the proliferation and secretion defects in cystic epithelial cells but also the pro-fibrotic responses of the interstitial fibroblasts. For example, it has recently been shown that inhibition of the KCa3.1 potassium channels suppresses cyst secretion and growth and, in the UUO model of CKD, attenuates fibrosis [125,126].

Research Highlights.

• In ADPKD, expansion of cysts is accompanied by progressive fibrosis leading to end-stage renal failure.

• In CKD of diverse etiologies, tubulointerstitial fibrosis provides the best correlation with progression to organ failure.

• Fibrosis in CKD has been studied extensively and presents a number of common characteristics.

• Similar to CKD, fibrosis in ADPKD is associated with a more rapid decline in renal function.

• The mechanisms of ADPKD fibrosis are largely unexplored.

• Although fibrosis in ADPKD shares some features of CKD fibrosis but there are unique, stage-specific features.

• Anti-fibrotic therapies may prove an effective adjunct to treatment of ADPKD.