Proteolytic Cleavage and Nuclear Translocation of Fibrocystin Is Regulated by Intracellular Ca2+ and Activation of Protein Kinase C*

Fibrocystin, a type I membrane protein of unknown function, is the protein affected in the autosomal recessive form of polycystic kidney disease. Here we show that fibrocystin undergoes regulated proteolysis. Several proteolytic cleavages occur within the predicted ectodomain, whereas at least one cleavage occurs within the cytoplasmic portion. The latter generates a C-terminal intracellular fragment that harbors the nuclear localization signal KRKVSRLAVTGERTATPAPKIPRIT and translocates to the nucleus. Proteolytic cleavage of fibrocystin occurs constitutively in long term cultures of polarized inner medullary collecting duct cells (mIMCD-3). Activation of protein kinase C and release of intracellular Ca2+ are required for proteolysis under these conditions. In short term cultures of human embryonic kidney 293 cells (HEK-293), proteolytic cleavage of fibrocystin can be elicited by stimulation of intracellular Ca2+ release or activation of protein kinase C. These results identify a novel Ca2+-dependent pathway that signals from fibrocystin located in the cell membrane to the nucleus.

Autosomal recessive polycystic kidney disease (ARPKD) 2 is a hereditary cause of kidney failure in infants and children. ARPKD affects 1 in 20,000 individuals and is characterized by aberrant epithelial cell proliferation, which causes cystic dilation of the renal collecting ducts, and abnormal development of intrahepatic bile ducts (1). Affected individuals present with bilateral kidney enlargement, intrauterine kidney failure, and oligohydramnios; the latter causes pulmonary hypoplasia and limb and facial abnormalities. Children who survive the perinatal period or develop ARPKD later in life develop chronic kidney disease and portal hypertension due to congenital hepatic fibrosis.
ARPKD is caused by mutations of the polycystic kidney and hepatic disease gene 1 (PKHD1) located on chromosome 6 (2). The protein encoded by PKHD1 is termed fibrocystin (also called polyductin, or tigmin (3)(4)(5)). Fibrocystin is an ϳ500,000 dalton type I membrane protein comprised of a large N-terminal ectodomain, a single transmembrane segment, and a short C-terminal cytoplasmic domain. The ectodomain contains arrays of IPT (Ig-like, plexins, transcription factors) domains and PbH1 (parallel ␤-helix repeats) domains. The structure of fibrocystin suggests that it may function as a receptor for an as yet unidentified ligand. However, the authentic function of fibrocystin and the mechanism of signal transduction remain unknown. Recent studies suggest that the molecular pathogenesis of PKD may involve primary cilia and associated Ca 2ϩ -dependent signaling (6). Primary cilia are present on the apical membrane of renal tubular epithelial cells, and bending of the cilia in response to fluid flow shear stress elicits an increase in cytosolic Ca 2ϩ concentration ([Ca 2ϩ ]). Polycystin-1 and polycystin-2, which are affected in the autosomal dominant form of PKD, are located in primary cilia and are required for an initial Ca 2ϩ influx that is triggered by fluid flow shear stress (6). Subsequently, the increase in [Ca 2ϩ ] i is amplified by Ca 2ϩ -induced Ca 2ϩ release. Inhibitors of intracellular Ca 2ϩ release abolish the primary cilia-dependent increase in [Ca 2ϩ ] i , which suggests that Ca 2ϩ -induced Ca 2ϩ release is an obligatory step in the primary cilia signaling cascade (6).
Fibrocystin is located in the primary cilium as well as the basal body, which anchors the primary cilium in the cell body (7)(8)(9)(10). Together with some similarities in disease manifestations, the overlapping subcellular localization of fibrocystin and polycystins suggests that they may be involved in a common pathway. However, so far, no such pathway has been described.
Recent studies suggest that ciliary signaling may involve the regulated proteolysis of polycystin-1. Polycystin-1 undergoes regulated intramembrane proteolysis (RIP), which releases a C-terminal fragment that translocates to the nucleus. Disruption of primary cilia formation causes the accumulation of the fragment in the nucleus, where it affects AP-1 activity. The C-terminal fragment of polycystin-1 has also been shown to interact with STAT6 and the coactivator p100 in the nucleus.
Here we show that fibrocystin undergoes regulated proteolytic cleavage releasing a cytoplasmic fragment that translocates to the nucleus. We identify factors that are necessary and sufficient to trigger proteolytic cleavage, and we define a nuclear translocation signal located within the cytoplasmic portion of fibrocystin.
Cell Culture, Transfection, and Generation of Stable Cell Lines-mIMCD-3 cells were plated at a density of 500,000 cells/ 100-mm dish, and 24 and 48 h later, the cells were transfected with 2 g of pFC-V5 using Effectene (Qiagen, Valencia, CA). Cells were incubated for an additional 72 h and lysed in 100 l of buffer containing Tris-buffered saline, 0.5% Triton X-100, protease inhibitor mixture (Hoffmann-La Roche Inc.). Twenty l of the lysates were analyzed by SDS-PAGE, and immunoblot analysis was performed using horseradish peroxidase-conjugated anti-V5 as described previously (12). mIMCD3/FC and mIMCD3/EGFP cell lines with inducible expression of fibrocystin and EGFP, respectively, were produced by transfecting mIMCD-3 cells with 1.5 g of pSwitch and either 1.5 g of pGeneFC or 1.5 g of pGeneEGFP. Stable transfectants were isolated after 14 days of growth in medium containing hygromycin (350 g/ml) and zeocin (300 g/ml). To test for inducible expression, cells derived from individual clones were treated with mifepristone (10 nM) for 48 h or left untreated, and proteins were analyzed by immunoblotting using horseradish peroxidase-conjugated anti-FLAG. HEK-293 cells stably transfected with the muscarinic acetylcholine receptor M3 were a generous gift from Trevor Shuttleworth (University of Rochester Medical Center) (13).
Subcellular Fractionation-Fractions containing nuclear proteins, cell membrane proteins, or soluble proteins were generated essentially as described previously (15). Successful fractionation was confirmed by the distribution of the membrane protein polycystin-2 and the nuclear protein proliferating cell nuclear antigen. To concentrate nuclear proteins, nuclear extract was precipitated with trichloroacetic acid.

Proteolysis of Fibrocystin Produces a C-terminal Nuclear
Fragment-A 12,225-bp DNA fragment encoding human fibrocystin was assembled from 12 PCR fragments, cloned into the mammalian expression plasmid pcDNA3.1, and verified by sequencing. To facilitate detection of the recombinant protein, a C-terminal V5 epitope tag was added, and the resultant plasmid was termed pFC-V5. Mouse inner medullary collecting duct cells (mIMCD-3), which endogenously express native fibrocystin (14), were transfected with pFC-V5 or empty pcDNA3.1, and cellular proteins were analyzed after 3 days. Immunoblotting utilizing an antibody directed against the C-terminal V5 epitope tag revealed a high molecular weight band corresponding to full-length fibrocystin as well as a series of proteolytic fragments (Fig. 1A).
To further investigate the proteolytic cleavage of fibrocystin, we generated mIMCD3/FC cells in which the expression of recombinant human fibrocystin can be induced by treatment with mifepristone. Immunoblot analysis of cells expressing recombinant fibrocystin revealed the presence of proteolytic fragments similar to those seen in transient transfection experiments (Fig. 1B).
The approximate sites of proteolytic cleavage of fibrocystin were estimated from the molecular weights of the proteolytic fragments. Since the cytoplasmic domain of human fibrocystin including the epitope tag has a calculated molecular mass of 25 kDa, the 21-kDa fragment (Fig. 1B, FCA) is likely produced by cleavage in the cytoplasmic domain close to the transmembrane region. Proteolytic fragments of fibrocystin that retain the membrane-spanning segment as well as the C-terminal V5 epitope tag have a minimal molecular mass of 27.5 kDa, suggesting that fragments named FCC to FCF were generated by proteolytic cleavage in the ectodomain of fibrocystin. The site of cleavage producing the fragment named FCB may be within the transmembrane segment.
To determine the subcellular localization of the proteolytic fragments, we prepared membrane, cytosolic, and nuclear extracts and analyzed the proteins by immunoblotting. Fragments that were calculated to contain the transmembrane domain (FCC to FCF) were found in the membrane fraction, as expected (Fig. 1B). In contrast, the 21-kDa fragment containing the C terminus of fibrocystin was detected in the nuclear fraction. This result suggests that the cytoplasmic domain of fibrocystin undergoes proteolytic cleavage, releasing a 21-kDa fragment that translocates to the nucleus. In some experiments, FCA appeared as well as doublet, suggesting that it may undergo additional cleavage or posttranslational modification (Fig. 1B, FCA and FCA*). The fragment FCB was found in the nuclear fraction and the membrane fraction but only after very long exposure of the fluorograms, suggesting that FCB is unstable or is lost during subcellular fractionation (data not shown).
To test whether endogenous fibrocystin undergoes proteolytic processing producing a nuclear fragment, nuclear extracts of mIMCD-3 cells were analyzed by immunoblotting. A polyclonal antibody raised against the C-terminal domain of fibrocystin recognized a protein corresponding in size to FCA (Fig. 1C). The 21-kDa fragment was detected using two different antibodies raised against the C-terminal domain of fibrocystin (not shown). Minor, smaller peptides were also detected, raising the possibility that endogenous FCA is subjected to further endoproteolytic processing (Fig. 1C, asterisk).
Cytoplasmic Domain of Fibrocystin Contains a Nuclear Localization Signal-To determine the localization of the cytoplasmic domain of fibrocystin in vivo, kidney sections of 21-day old mice were stained with an antibody specific for the cytoplasmic region of the protein. Antibody staining confirmed that the cytoplasmic domain of fibrocystin was located in the nuclei of renal tubular epithelial cells ( Fig. 2A). The cytoplasmic domain was also found in the cytosol as well as the primary cilia when the cells were observed in a different focal plane (data not shown). Interestingly, the fibrocystin staining in the nucleus exhibited a speckled pattern.
To define the mechanism of nuclear localization, the cytoplasmic domain of fibrocystin was linked to EGFP or DsRed and expressed in MDCK cells. The subcellular localization of the fusion proteins was determined by fluorescence microscopy. A recombinant protein containing the cytoplasmic domain of fibrocystin (amino acids 3876 -4059) and EGFP was located exclusively in the nucleus, whereas EGFP by itself was predominantly in the cytosol (Fig. 2B). Similarly, a fusion protein containing the cytoplasmic domain of fibrocystin and DsRed was located in the nucleus. Like endogenous fibrocystin, the fusion proteins were not diffusely distributed in the nucleus but were concentrated in structures that had a speckled subnuclear distribution. To identity the nuclear structures, cells were transfected with plasmids encoding the cytoplasmic domain fused to DsRed, and the nucleoli were counterstained with an RNA-specific stain. Co-staining demonstrated that the fusion proteins containing the cytoplasmic domain were located in nucleoli (Fig. 2C).
To identify the region within the cytoplasmic domain of fibrocystin that is required for nuclear localization, plasmids  3A). To verify these findings using a different cell type and method, plasmids encoding the V5 epitope-tagged N-terminal portion and central portion of the cytoplasmic domain were transfected into HEK-293 cells. Immunoblot analysis of nuclear and cytosolic fractions revealed that the N-terminal portion but not the central portion was present in the nuclear fraction (Fig.  3B). To further define the sequence mediating nuclear localization, plasmids encoding various regions of the cytoplasmic domain of fibrocystin fused to EGFP were generated, and the subcellular localizations of the fusion proteins were evaluated. These experiments defined the minimal nuclear localization signal (NLS) to be KRKVSRLAVTGERTAT-PAPKIPRIT (Fig. 3C). Further deletions within this sequence abolished nuclear localization.
Intracellular Ca 2ϩ Release Is Necessary and Sufficient to Trigger Fibrocystin Proteolysis-To determine whether the proteolytic cleavage of fibrocystin is affected by changes in intracellular Ca 2ϩ concentration and Ca 2ϩ -induced Ca 2ϩ release, mIMCD3/FC cells were pretreated with thapsigargin prior to induction of fibrocystin expression. Pretreatment with 0.1 M thapsigargin to deplete intracellular Ca 2ϩ stores (16) abolished the generation of the FCA fragment (Fig.  4A). Similarly, depletion of ryanodine-sensitive Ca 2ϩ stores by pretreatment with 5 mM caffeine also prevented the generation of FCA (Fig. 4A). Pretreatment with 200 M ruthenium red, a nonspecific Ca 2ϩ channel inhibitor, or treatment with 75 M dantrolene, which interferes with ryanodine receptor (RyR)-mediated intracellular Ca 2ϩ release, abolished the formation of FCA (Fig. 4A).
To quantitatively measure fibrocystin cleavage, we developed a luciferase assay similar to the ones employed to quantify the proteolytic cleavage of amyloid precursor protein (APP) and low density lipoprotein receptor-related protein-1 (LRP1) (17,18). A plasmid (FC-GV) encoding the synthetic transcription factor Gal4-VP16 fused to the C terminus of fibrocystin was generated. Proteolytic cleavage of the fibrocystin fusion protein releases Gal4-VP16, which translocates to the nucleus and activates a Gal4responsive luciferase reporter gene (pG5-luc). Therefore, luciferase expression correlates with fibrocystin cleavage. Transfection of mIMCD-3 cells with a plasmid encoding FC-GV stimulated luciferase activity 2.6-fold when compared with cells expressing LDLR-GV (low density lipoprotein receptor fused to Gal4-VP16), which is not subjected to proteolysis. The increase in luciferase activity was abolished in the presence of dantrolene, verifying that RyR activity was required for fibrocystin cleavage (Fig. 4B).
To test whether intracellular Ca 2ϩ release is not only required but is also sufficient to induce fibrocystin proteolysis, we employed a short term cell culture system of human embryonic kidney cells (HEK-293). These cells express RyR and exhibit an RyR-mediated increase of intracellular Ca 2ϩ upon treatment with caffeine (19). Under basal conditions, no proteolytic fragments of transiently transfected fibrocystin were detected. However, the addition of caffeine induced dose-dependent proteolysis of fibrocystin (Fig. 4C). Caffeine-induced fibrocystin proteolysis was prevented by treatment with the protein kinase C (PKC) inhibitor calphostin C. This latter result indicated a possible role of PKC in the Ca 2ϩ -induced cleavage of fibrocystin.
To quantify fibrocystin proteolysis in response to intracellular Ca 2ϩ release, the luciferase reporter assay was employed. HEK-293 cells were cotransfected with the plasmids pFC-GV and pG5-luc and incubated with either 5 mM caffeine or vehicle (as a control). Treatment with caffeine increased luciferase activity 10-fold, confirming that pharmacological stimulation of intracellular Ca 2ϩ release was sufficient to induce fibrocystin cleavage (Fig. 3D).
Next, we tested whether fibrocystin cleavage could be induced by activation of a cell surface receptor that stimulates intracellular Ca 2ϩ release. These experiments utilized HEK-293 cells stably expressing the muscarinic acetylcholine receptor m3 (HEK-293(m3)), which respond to treatment with the agonist carbachol with activation of phospholipase C and inositol 1,4,5-trisphosphate-mediated Ca 2ϩ release (20). Exposure to 5 mM caffeine triggered fibrocystin proteolysis in both HEK-293 cells and HEK-293(m3) cells, whereas treatment with 100 M carbachol induced fibrocystin cleavage only in HEK-293(m3) cells (Fig. 4E). Taken together, these results demonstrate that stimulation of intracellular Ca 2ϩ release is sufficient to trigger fibrocystin cleavage.
Activation of PKC Is Necessary and Sufficient for the Generation of FCA-The inhibition of fibrocystin proteolysis in the presence of the PKC inhibitor calphostin C suggested an involvement of members of the group of conventional PKC (PKC␣, PKC␤1, PKC␤2, or PKC␥). To test whether activation of PKC is sufficient to induce fibrocystin cleavage, HEK-293 cells were transfected with pFC-V5 and exposed to activators of protein kinases. Treatment with PMA, an activator of PKC, produced a marked increase of proteolytic cleavage, which was prevented by pretreatment with calphostin C (Fig. 5A). In contrast, no cleavage was observed when cells were left untreated or treated with the protein kinase A activator 8-bromo-cAMP. However, besides inducing proteolysis, PMA also increased the total expression of fibrocystin, controlled by the cytomegalovirus promoter, which made it difficult to unequivocally correlate PKC activation with fibrocystin cleavage.
To address this issue, we used two additional approaches: (i) replacement of the PKC-responsive cytomegalovirus promoter and (ii) genetic activation of PKC. mIMCD3/FC cells, in which fibrocystin is expressed under the control of a mifepristoneresponsive promoter, were treated with calphostin C 24 h after induction, and cleavage was analyzed 30 h later. Treatment with calphostin C did not alter expression levels of fibrocystin but produced a dose-dependent inhibition of fibrocystin proteolysis (Fig. 5B). Next, we employed a constitutively active pseudosubstrate mutant of PKC␣ (PKC␣A/E) (21). Cotransfection of a plasmid encoding PKC␣A/E resulted in fibrocystin proteolysis and generation of FCA, whereas cotransfection of empty expression plasmid did not affect processing (Fig. 5C). The presence or absence of fetal calf serum did not markedly affect cleavage.
To quantify the cleavage of fibrocystin stimulated by PKC activation, luciferase reporter assays were performed. HEK-293 cells were cotransfected with pFC-GV and pG5-luc, and PKC activity was stimulated by treatment with PMA or expression of the PKC␣A/E mutant. Fibrocystin proteolysis was evaluated by measuring luciferase activity. Treatment with PMA increased luciferase activity 27-fold, whereas the addition of 8-bromo-cAMP did not significantly affect luciferase activity. Co-expression of PKC␣A/E stimulated luciferase activity 19-fold but had no effect on cells expressing the negative control LDLR-GV (Fig. 5D). Taken together, these results demonstrate that fibrocystin-specific proteolysis is PKC-dependent.

DISCUSSION
The data presented here indicate that fibrocystin is subjected to site-specific proteolytic cleavage. At least six different fragments that contain the C terminus of fibrocystin are generated. Although the abundance of the larger proteolytic fragments varied under different experimental conditions, a 21-kDa fragment (FCA) was consistently observed. The 21-kDa fragment contains most of the cytoplasmic domain of fibrocystin and was primarily observed in the nucleus. In contrast, full-length fibrocystin has been localized primarily in the primary cilium, basal body region, and apical plasma membrane, and to a lesser degree, in the cytoplasm. These findings suggest that fibrocystin undergoes regulated proteolysis releasing a cytoplasmic C-terminal fragment that translocates from the cilium, apical  NOVEMBER 10, 2006 • VOLUME 281 • NUMBER 45 membrane, or cytoplasm to the nucleus. The protease that mediates fibrocystin cleavage remains unidentified, but inhibitor studies suggest that it is not ␥-secretase or calpain (data not shown).

Regulated Proteolysis of Fibrocystin
Similar to other proteins that are subjected to regulated proteolysis, only a small fraction of fibrocystin appears to be cleaved. In particular, the pattern of proteolytic cleavage (close proximity to the transmembrane domain, nuclear translocation of the resulting cytoplasmic fragment) resembles RIP of other type I membrane proteins. Nuclear signaling of cytoplasmic fragments generated by RIP is most extensively described for Notch but also found in the case of APP, ErbB4, NRG-1, CD44, N-cadherin, and lipoprotein receptor-related protein (22). RIP controls vital signaling mechanisms that are crucial for regulatory processes and disease pathogenesis. Often, the cleavage releasing the nuclear fragment is preceded by an obligatory proteolytic event in the ectodomain of the protein. It is conceivable that the proteolytic cleavage occurring in the ectodomain of fibrocystin is also required for a subsequent step that releases the cytoplasmic fragment.
Polycystin-1, a protein affected in the autosomal dominant form of PKD, was found to be subjected to proteolysis in a manner consistent with RIP. Proteolysis of polycystin-1, similar to proteolysis of fibrocystin, generates a C-terminal cytoplasmic fragment that translocates to the nucleus upon release. The C-terminal cytoplasmic fragment of polycystin-1 has been shown to regulate AP-1 activation, Na,K-ATPase activity, and the mammalian target of rapamycin (mTOR) pathway, the latter via direct interaction with tuberin. In addition, the polycystin-1 C-terminal cytoplasmic fragment interacts with STAT6 and the coactivator p100 and stimulates STAT6-dependent transcription. However, overexpression of the C-terminal cytoplasmic domain of fibrocystin had no effect on AP-1-or STAT-dependent reporter genes (data not shown), which suggests that at least some of the functions of the cytoplasmic domains of polycystin-1 and fibrocystin are distinct.
We identified the minimal nuclear localization sequence within the FCA fragment of fibrocystin to be KRKVS-RLAVTGERTATPAPKIPRIT. This sequence mildly resembles a bipartite NLS, which is characterized by two stretches of basic amino acids that are usually separated by a 10 -12-aminoacid-long cluster (23). Interestingly, FCA appears not to be uniformly distributed in the nucleus but accumulates specifically in the nucleolus. Nucleoli are the site of rDNA transcription, assembly of ribosomes, and biogenesis of small ribonucleoprotein particles and influence cell cycle progression and transcription by sequestering specific trans-acting factors (24). The nucleolar localization of FCA suggests that fibrocystin may play an important role in regulating these processes.
Several lines of evidence indicate that the release of intracellular Ca 2ϩ is a prerequisite for fibrocystin proteolysis. Constitutive cleavage of fibrocystin observed in a long term cell culture system could be prevented by pharmacological inhibition of Ca 2ϩ -induced Ca 2ϩ release. No cleavage of fibrocystin was observed under basal conditions in a short term cell culture system, but cleavage and nuclear translocation could be induced by pharmacological or receptor-mediated stimulation of intracellular Ca 2ϩ release. These findings suggest that signaling cascades that affect [Ca 2ϩ ] i might act as central regulators or mediators of fibrocystin cleavage.
One Ca 2ϩ -dependent signaling cascade that is considered to be central in the development of polycystic kidney disease  1, 4, and 6). After 2 h, media were replaced with media lacking serum but containing the same supplements, and in addition, 10 nM mifepristone. Cell lysates were prepared after 24 h and analyzed by immunoblotting using anti-V5 antibody. B, mIMCD-3 cells were transfected with 0.2 g of pG5-luc and either 0. involves primary cilia. Primary cilia located on the apical surface of renal tubular epithelial cells are involved in sensing luminal fluid flow. Shear stress, caused by the flow of fluid across the epithelial cell layer, bends the cilium and triggers an increase in [Ca 2ϩ ] i . Proteolytic cleavage of polycystin-1 appears to involve primary cilia. The loss of primary cilia or obstruction of fluid flow results in the accumulation of the polycystin-1 C-terminal cytoplasmic fragment in the nucleus. It is possible that primary cilia are also involved in the regulation of fibrocystin proteolysis. Fibrocystin proteolysis was observed in long term cultures of mIMCD-3 cells, which form primary cilia, and no cleavage was observed in short term HEK-293 cell cultures, which lack primary cilia. However, no effect of fluid flow on fibrocystin cleavage has been detected (data not shown). Therefore, further experimentation will be required to clarify the role of primary cilia and the flowdependent increase in [Ca 2ϩ ] i on fibrocystin cleavage.
Cleavage of fibrocystin is stimulated by activation of PKC. Studies using PKC activators, the PKC-specific inhibitor calphostin C, and a constitutively active PKC mutant indicate that activation of PKC is necessary and sufficient to cause fibrocystin proteolysis. Activation of PKC can circumvent the requirement for release of intracellular Ca 2ϩ , indicating that PKC activation is likely to be downstream to Ca 2ϩ release. Conventional PKCs are activated by intracellular Ca 2ϩ release and represent potential downstream effectors of Ca 2ϩ -induced Ca 2ϩ release. Although it is not clear how activation of PKC triggers fibrocystin cleavage, one possibility is that PKC regulates the trafficking of intracellular vesicles, allowing fibrocystin to interact with the protease. Interestingly, PKC activation is also involved in the processing of APP. Modulators of PKC activity have been shown to alter APP processing and reduce the levels of ␤-amyloid in animal models (25). Similarly, alteration of fibrocystin proteolysis achieved by pharmacological modulation of PKC activation or intracellular Ca 2ϩ release might represent a possible therapeutic strategy to treat PKD.
In summary, we propose a model in which cytosolic Ca 2ϩ concentration and activation of PKC control a proteolytic process that leads to the release of the cytoplasmic C-terminal fragment of fibrocystin. Because of the presence of a NLS, the C-terminal fragment (FCA) is translocated to the nucleus. A role of fibrocystin in a signaling cascade is suggested by mechanistic similarities with other proteins subjected to RIP.