FK506-binding protein 10 (FKBP10) regulates lung fibroblast migration via collagen VI synthesis

Patients suffering from idiopathic pulmonary fibrosis (IPF), a highly progressive interstitial lung disease, have a median survival prognosis of 2-5 years after diagnosis [1]. The pathogenic processes are not completely understood. It is currently believed that the fibrotic response is caused by repeated micro-injuries of the respiratory epithelium [2] which leads to the release of profibrotic mediators like transforming growth factor β1 (TGF-β1), followed by myofibroblast differentiation, increased fibroblast migration, and, ultimately, excessive deposition of extracellular matrix (ECM) in the alveolar region [3‐6]. More recent evidence suggests that the composition of the ECM strongly affects fibroblast phenotypes and therefore plays a crucial role in disease progression [4, 7, 8].

Aberrant fibroblast adhesion and migration are common features of fibrosis [9‐11] and targeting fibroblast migration, e.g. by inhibition of focal adhesion kinase (FAK) or of integrins, has been proposed as a treatment strategy [12, 13]. For instance, myofibroblasts possess an increased ability to adhere to the ECM, which is mediated by focal adhesions (FA) attaching the actin cytoskeleton to the matrix [14]. Cell attachment to the ECM via clustering of integrins leads to the recruitment of numerous FA proteins with adapter, structural, and enzymatic functions [15, 16]. For instance, structural proteins like talin, vinculin and α-actinin facilitate the connection between integrins and actin fibers and provide the basis for the transmission of mechanical forces between cell and ECM [17]. To enable cell migration, turnover of FA is necessary. Several factors are involved in FA disassembly, like actin dynamics, FAK and Src phosphorylation, and ERK/MAP kinase-mediated activation of calpain-2, a calcium-dependent protease [15].

Finally, migration is strongly influenced by topology and composition of the ECM including integrin ligands like collagen, fibronectin (FN), and laminin [7]. Collagen type VI appears to play a particularly important role in this context, as several studies indicate a central, albeit context-dependent and tissue-specific role of collagen VI for migration and adhesion [18‐20].

FK506-binding protein 10 (FKBP10, also termed FKBP65), a member of the family of immunophilins, is an endoplasmic reticulum (ER) -resident peptidyl prolyl isomerase and a collagen I chaperone [21]. We have previously reported upregulation of FKBP10 in experimental lung fibrosis and IPF, where it is mainly expressed by (myo)fibroblasts [22]. Deficiency of FKBP10 by siRNA-mediated knockdown in primary human lung fibroblasts (phLF) reduced the expression of profibrotic markers like α-smooth muscle actin (α-SMA), FN and collagen I, and suppressed collagen secretion [22].

As properties of the ECM play an important role in adhesion dynamics and FKBP10 has been identified previously as a regulator of collagen biosynthesis in phLF, the aim of this study was to assess the effect of FKBP10 deficiency on adhesion and migration in phLF. To elucidate the underlying mechanisms, we analyzed the effect of siRNA-mediated knockdown of FKBP10 on intracellular and membrane-spanning components of the FA complex, on regulatory events of FA turnover, on proteins involved in actin dynamics, and, finally, on a selection of ECM proteins with important emerging functions in migration.

In this study we demonstrated that FKBP10 deficiency inhibited phLF adhesion and migration. This effect could neither be explained by changes in expression or activation of components of the FA complex, nor by changes in FAK downstream signaling events, nor by changes in regulators of actin dynamics. Instead, we found that the effect of FKBP10 deficiency on migration and adhesion depended on 2-phosphoascorbate, pointing towards a central role of collagen biosynthesis in this context. Loss of FKBP10 downregulated the expression of collagen VI, a collagen type increasingly recognized as a central player in migration, and coating culture dishes with collagen VI completely abolished the effect of FKBP10 deficiency on phLF migration.

Next to excessive ECM deposition by interstitial fibroblasts, aberrant fibroblast adhesion and increased migration are also important features of IPF [6, 9‐11]. Here we show that loss of the collagen chaperone FKBP10, which we previously identified as potential IPF drug target due to its role in ECM synthesis and secretion [22], inhibited wound closure and reduced the mean velocity and adhesion capacity of phLF (Fig. 1). This observation is in line with the very recent report by Liang et al., showing that siRNA-mediated knockdown of FKBP10 led to reduced migration in human hypertrophic scar fibroblasts [32]. Collectively, these studies are confirmative of our concept of FKBP10 as potential drug target in fibrotic disease.

We assessed fibroblast migration both in a conventional scratch assay as well as by tracking individual cells with videomicroscopy in time-lapse experiments, a more accurate approach. Overall, both assays consistently showed reduced migration under conditions of FKBP10 deficiency, even if the scratch assay results in presence of TGF-β1 failed significance (p=0.06). We believe that this minor discrepancy reflects the well-known disadvantages of the scratch assay, most importantly variations in gap width due to the manually applied and therefore often uneven scratch, the fact that ECM substrate is equally scraped off together with the cells, and also the mechanical cell damage which may introduce artifacts. These technical drawbacks may explain the greater variations observed in this assay, leading to results that just fail significance.

Our finding that TGF-β1 did not affect fibroblast migration is in agreement with previous studies by others [33‐35]. Increased fibroblast adhesion in presence of TGF-β1 may in part reflect increased expression of β1-integrin (this work, cf. Fig. 2e, f, scr ctrl vs. scr TGF-β1, p=0.0642) and/or β3-integrins [36].

Initially, we sought to assess the effect of FKBP10 deficiency on several regulatory levels of FA turnover to elucidate the mechanism underlying attenuated migration. Cell migration is a complex cyclic process starting with the extension of membrane protusions (lamellipodia) at the leading edge followed by their adhesion to the ECM and retraction of the cell tail [37]. During cell migration assembly and disassembly of FA is dynamically regulated. The process of cell attachment to the ECM is initiated by clustering of integrins on the cell surface, heterodimeric transmembrane receptors consisting of α and β subunits. The intracellular domains of the clustered integrins serve as a platform for FA protein recruitment and ultimately link the ECM via FA to actin stress fibers [38]. In this context, TLN1 mediates the initiation of FA assembly by interacting with the cytoplasmic domain of the integrin β subunit on the one hand and with actin and actin-binding proteins on the other [15]. Another initial event upon integrin clustering is activation of FAK including autophosphorylation of Tyr 397 and Src-mediated phosphorylation of additional tyrosine residues within FAK (e.g. Y 566 and 577), which are essential for full FAK activity. Active FAK interacts with multiple signalling molecules including Grb7, PI3K, paxillin, MLCK and ERK, activating signalling pathways which result in protrusion extension, increased FA turnover, and therefore, ultimately, in increased cell motility [15, 39‐42]. The protease calpain 2, a heterodimer consisting of a catalytic subunit and a regulatory subunit (calpain-4, CAPNS1) plays a central role in this context as it, when recruited by FAK and activated by ERK/MAP kinase, mediates amongst others proteolysis of TLN1 and FAK, considered the rate-limiting step in FA turnover [15, 38, 39, 43, 44]. Finally, directional cell migration is strongly dependent on polarized actin dynamics. Rho-like GTPases like RhoA and Rac1 control cytoskeleton contractility, polymerization, and protrusion. The integral membrane protein caveolin-1 (Cav-1), activated by small kinases like Src, is a central regulator of Rho-like GTPase signaling in this context [45, 46]. Another regulator of actin filament turnover in the lamellipodia during cell migration is the actin binding protein coronin 1C (CORO1C) [47].

FKBP10 deficiency led to several significant changes in expression of FA components on all levels assessed, i.e. upregulation of ITGB1, TLN1, CAPNS1, FAK, CORO1C and downregulation of CAV1 (Fig. 2). Upregulation of ITGB1 may associate with reduced FA turnover and slow migration [48‐50] but impaired migration and adhesion has also been reported as a result of ITGB1 deficiency [51]. Collectively, these findings suggest that manipulation of ITGB1 protein levels in general is unfavorable to cell motility, regardless of direction of regulation. As ITGB1 functions as an adapter protein between the ECM and intracellular FA complexes, it is conceivable that its expression levels must be tightly regulated to allow for functional intermolecular interactions between different interacting partners.

The same may apply to TLN1, an adapter protein linking the cytoskeleton to ITGB1, where reports in the literature are also seemingly controversial: Upregulation of TLN1, but also suppression of TLN1 has been reported to reduce migration and adhesion [52‐54]. Interestingly, downregulation of TLN1 has also been associated with increased migration [52, 55], which is in support of its function as a regulator of FA turnover together with its protease calpain-2 [15, 38, 39, 43, 44]. In our system, however, the simultaneous increase of the regulatory calpain 2 subunit CAPNS1, argues for overall little change in FA turnover, at least in absence of TGF-β1 (Fig. 2c, d). This is consistent with our observation that phosphorylation levels of FAK, Src, and ERK (Fig. 3), central signaling events in the process of FA turnover [39, 56‐58] remained unchanged.

At first, observing inhibition of adhesion and migration in the absence of changes in activation of FAK and related signaling pathways seemed contradictory to us. However, similarly, Asano and colleagues have reported that siRNA-mediated knockdown of α-smooth muscle actin (α-SMA) in phLF led to inhibition of migration without affecting the FAK signaling pathway [59]. This observation suggested that changes in actin dynamics may underlie the observed inhibition of migration and, indeed, from our previous studies, we know that FKBP10 deficiency reduces α-SMA expression in phLF [22]. Also, deficiency of the actin binding protein CORO1C typically results in inhibition of migration; however, here, we observed a moderate increase of CORO1C protein rather than downregulation (Fig. 2i) [60, 61]. Expression of CAV1, deficiency of which typically results in decreased migratory speed in variable cell types [62‐64], was only moderately downregulated in presence of TGF-β1 (Fig. 2g, h). Collectively, these observations do not argue for altered actin dynamics as a major mechanism underlying inhibition of migration in response to FKBP10 deficiency.

Importantly, cell migration is influenced by properties of the ECM, like density of adhesion ligands (collagen, FN), ECM composition, and stiffness [7, 65]. Previously, we have observed downregulation of expression and secretion of type I collagen and FN, both major components of the fibroblast ECM, in response to FKBP10 knockdown [22]. Here we extended this analysis and assessed additional ECM components with important roles in cell migration, namely type VI collagen and FBLN1. Both proteins colocalized with FKBP10 in phLF, as assessed by both immunofluorescence colocalization analysis and proximity ligation assay (Fig. 4g-j), indicating direct interaction with FKBP10 in the ER. Interestingly, we found that loss of FKBP10 significantly increased FBLN1 expression (Fig. 4e, f), but decreased protein levels of COL6A1 (Fig. 4a, b). Notably, COL6A1 deficiency is sufficient to inhibit collagen VI deposition in the ECM, as no triple helical molecules can be formed without the α1(VI)-chain [66]. These results suggest opposing functions of FKBP10 in FBLN1 and collagen VI biosynthesis in phLF. It is tempting to speculate, for instance, that FKBP10 acts as a FBLN1 chaperone, sequestering FBLN1 in the ER, prohibiting packing in vesicles for secretion or targeting FBLN1 for ER-associated protein degradation, while at the same time FKBP10 is likely required for efficient collagen VI triple helix formation, similar to collagen I and III [21, 67, 68]. These aspects will be interesting to explore in future studies.

As to function of these proteins in migration, reduced attachment and decreased migratory speed has been reported for a human breast cancer cell line (MDA MB231) in response to FBLN1 overexpression [27] and siRNA mediated knockdown of FBLN1 in corneal fibroblasts upregulated cell migration [69]. Therefore, taken together, it was plausible that increased FBLN1 levels may underlie the observed inhibition of migration under conditions of FKBP10 deficiency.

While collagen VI begins to emerge as an important regulator of cell migration, reports on its direction of effect, inhibiting or promoting migration, are controversial [18‐20]. For instance, collagen VI-deficient tendon fibroblasts show delayed wound closure, i.e. lower migration speed, in a scratch assay [18], while human dermal fibroblasts displayed higher migration speed on matrices derived from collagen VI-deficient cells [20]. These discrepancies may be a result of the divergent approaches in the mentioned studies (assessment of newly formed ECM versus assessment of ECM deposited within 10 days, respectively), different collagen VI chains assessed (COL6A1 versus COL6A2), but also of the different cell origins, thus possibly reflecting time-, chain-, and cell-specific effects of collagen VI.

Ascorbic acid is a cofactor necessary for proline and lysine hydroxylation during collagen synthesis [68] including collagen VI [30, 70], but not required for FBLN1 synthesis. Therefore, to differentiate between increased FBLN1 or decreased type VI collagen as the underlying mechanism of decreased adhesion and migration, we compared effects of FKBP10 knockdown on adhesion and migration in absence and presence of 2-phosphoascorbate, a stable analogue of ascorbic acid. Notably, neither migration nor adhesion were changed upon loss of FKBP10 when the cell culture medium was ascorbate-deficient (Fig. 5a, b). These results strongly indicated that the effect of FKBP10 deficiency on adhesion and migration was collagen-dependent. While coating with collagen VI abolished the effect of FKBP10 knockdown on migration completely, coating with collagen I only did so marginally (Fig. 6). Overall, this indicated that FKBP10 knockdown inhibits lung fibroblast migration by reduced collagen VI biosynthesis rather than reduced collagen I biosynthesis, an effect of FKBP10 deficiency which we have reported previously [22].

Collagen VI is an important regulator of the ECM organizing the three-dimensional meshwork of collagen and FN fibers [71, 72] and the topography of the fibrillar ECM network guides directional cell migration [73]. Therefore, our observations suggest that loss of FKBP10 results in reduced biosynthesis of collagen VI and, possibly in combination with decreased extracellular collagen I levels, leads to a disorganized ECM with changed adhesion ligand density, stiffness and composition, which may not provide sites for integrin clustering and does not favor directional cell migration. Upregulation of ITGB1, TLN1, CAPNS1, total FAK, and CORO1C may ultimately reflect an attempt of the cells to overcome decreased migration by overcompensation, increasing expression of ECM receptors or components of the FA turnover machinery.

Interestingly, FKBP10 mediates dimerization with collagen lysyl hydroxylase 2 and thus contributes to the generation of collagen hydroxylysines, which act as substrates for extracellular lysyl-oxidase-mediated collagen crosslinking [74‐76]. Extracellular collagen lacking proper crosslinks is prone to proteolytic degradation and may not be able to contribute to the higher ordered organization of the ECM anymore [77‐80]. Therefore, downregulation of FKBP10 may also contribute to disorganization of the ECM by providing less crosslinking sites in type I and type VI collagen.

In summary, we found that loss of FKBP10 in phLF results in decreased adhesion and migration. We found that the main underlying mechanism was reduced collagen VI biosynthesis, as both ascorbate deficiency and coating of cell culture dishes with collagen VI abolished the effect of FKBP10 knockdown on migration. As increased fibroblast migration is a characteristic of IPF, the results are in support of our concept of FKBP10 as a potential drug target for IPF.