Autosomal dominant polycystic kidney disease (ADPKD) is the most common inherited human renal disease and is caused by mutations in two genes, PKD1 (85%) and PKD2 (15%). Cyst epithelial cells are characterised by a complex cellular phenotype including changes in proliferation, apoptosis, basement membrane composition and apicobasal polarity. Since polycystin 1 (PC1), the PKD1 protein, has been located in the basolateral membrane of kidney epithelial cells, we hypothesised that it might have a key role in mediating or stabilising cell-cell interactions. In non-ciliated L929 cells, stable or transient surface expression of the PC1 extracellular domain was sufficient to confer an adhesive phenotype and stimulate junction formation. In MDCK cells, we found that PC1 was recruited to the lateral membranes coincident with E-cadherin within 30 minutes after a `calcium switch'. Recruitment of both proteins was significantly delayed when cells were treated with a PC1 blocking antibody raised to the PKD domains. Finally, PC1 and E-cadherin could be coimmunoprecipitated together from MDCK cells. We conclude that PC1 has a key role in initiating junction formation via initial homophilic interactions and facilitates junction assembly and the establishment of apicobasal polarity by E-cadherin recruitment.
Autosomal dominant polycystic kidney disease (ADPKD) is the most common inherited human renal disease (incidence 1 in 1000 live births) and is caused by mutations in two genes, PKD1 (85%) and PKD2 (15%) (Ong and Harris, 2005; Torres and Harris, 2006). ADPKD is an important cause of end-stage renal failure, accounting for ∼10% of patients on renal replacement therapy. The disease is characterised by the formation of fluid filled cysts in both kidneys of affected individuals, which ultimately result in end-stage renal failure. Other extrarenal manifestations of the disease include hypertension, cardiac valve abnormalities and cerebral aneurysms (Calvet and Grantham, 2001; Wilson, 2004).
The two proteins involved in ADPKD, polycystin 1 (PC1; also known as PKD1) and polycystin 2 (PC2; also known as PKD2) have been shown to function as a heterodimeric complex (Hanaoka et al., 2000; Newby et al., 2002), activating a number of key signalling pathways, which in turn regulate diverse cellular functions including proliferation, apoptosis, tubulogenesis and fluid secretion. This is consistent with the largely overlapping renal and extrarenal phenotypes of PKD1 and PKD2 patients. PC1 and PC2 are likely to function together in many systems but there is evidence to suggest they can also function independently (Ong and Harris, 2005). Both proteins have been located in several subcellular structures including primary cilia and the basolateral membrane. Functional evidence that the polycystins can transduce a mechanosensitive Ca2+ current and mediate cell adhesion has been reported (Ibraghimov-Beskrovnaya et al., 2000; Nauli et al., 2003; Streets et al., 2003).
PC1 is a large (>460 kDa) heavily glycosylated integral membrane protein with a predicted large N-terminal extracellular domain (∼2500 aa), 11 transmembrane domains and a short C-terminal cytoplasmic tail (Hughes et al., 1995). The extracellular region appears to have a modular structure suggesting the presence of potential functional motifs. These include two leucine-rich repeats (LRR), a C-type lectin, a LDL-A receptor motif and a large region (∼1000 residues) with strong homology to the sea urchin receptor for egg jelly (REJ) protein. The major part of the N-terminal region, however, consists of 16 novel repeats (80-90 aa) with low sequence homology to immunoglobulin domains. These so called PKD domains or repeats are arranged in tandem (II-XVI) except for domain I, which is present between the LRR and lectin modules. The extracellular domain of PC1 has been shown to be cleaved at a conserved G-protein-coupled receptor (GPCR) proteolytic site (GPS) (position T3049) resulting in N-terminal and C-terminal fragments tethered to each other at the cell surface (Qian et al., 2002) This proteolytic event is thought to be essential for the function of PC1 in the mature kidney (Yu et al., 2007).
Antibodies to the PKD domains of PC1 have been shown to disrupt cell-cell adhesion in subconfluent canine, murine and human kidney epithelial cells (Ibraghimov-Beskrovnaya et al., 2000; Streets et al., 2003). These results support a key role for PC1 in the regulation of cell adhesion. However, it is possible that the role of PC1 is dependent on the function of other adhesion molecules such as E-cadherin or desmosomal cadherins. In this regard, a role for PC1 in E-cadherin recruitment has also been reported (Charron et al., 2000). PC1 has also been shown to colocalise and coimmunoprecipitate with E-cadherin and the catenins in human pancreatic adenocarcinoma cells (Huan and van Adelsberg, 1999). In human cystic cells, the absence of surface PC1 is associated with concomitant loss of surface E-cadherin expression and its replacement by N-cadherin (Streets et al., 2003; Roitbak et al., 2004; Russo et al., 2005). A putative role for PC1 in desmosome function has also been postulated because of its immunolocalisation to desmosomal junctions, the mislocalisation of desmosome junction proteins in cystic cells and the reported interaction of the PC1 C-terminus with intermediate filaments (Scheffers et al., 2000; Xu et al., 2001; Russo et al., 2005; Silberberg et al., 2005). However, two studies reporting the timing of PC1 and desmoplakin recruitment after a `calcium switch' conflict with one study showing that PC1 precedes desmoplakin and another showing that it follows desmoplakin (Scheffers et al., 2000; Silberberg et al., 2005).
In this study, we present data demonstrating that expression of the extracellular domain of PC1 is sufficient to induce cell aggregation and junction formation in L929 cells and that it is involved in E-cadherin recruitment during MDCK junction reassembly. Our data support a proximal role for PC1 in the establishment of cell junctions and the subsequent acquisition of apicobasal polarity.
Cell surface expression of the extracellular domain of PC1 as a GPI fusion protein (NT1-GPI)
To investigate whether PC1 is sufficient to mediate cell-cell adhesion, we expressed PC1 in L929 mouse fibroblasts which lack classical cell adhesion molecules. Previous studies have utilised this non-adhesive cell line to demonstrate the function of exogenous cell adhesion molecules including E-cadherin (Nagafuchi et al., 1987), and desmosome proteins (Marcozzi et al., 1998). In addition, we confirmed that these cells do not express primary cilia or detectable PC1 (not shown). Therefore we reasoned that this model system would also enable us to determine whether PC1 could function as a cell adhesion molecule independently of ciliary expression.
Because our initial attempts to generate stable L929 clones expressing full length PC1 were unsuccessful, an alternative strategy utilising a GPI anchor signal sequence (Bernasconi et al., 1996) to direct the expression of the extracellular domain of PC1 (aa M1-R2961) to the cell surface was devised (Fig. 1A). This site is N-terminal of the GPS cleavage site (located at T3049) and was chosen so as to exclude the possibility of proteolytic cleavage of the ectodomain (Qian et al., 2002). Surface expression of the NT1-GPI fusion protein was easily detectable in transiently transfected HEK293 cells by cell surface biotinylation (Fig. 1B) and immunofluorescence (not shown). Of note, transient expression of NT1-GPI was sufficient to induce the formation of cell-cell contacts in L929 cell with the protein concentrated at these regions (Fig. 1C). By contrast, L929 cells transfected with the empty vector showed no cell-cell contacts and no PC1 staining (Fig. 1C). We proceeded to generate a total of 20 stable L929 clones expressing NT1-GPI or the empty vector by screening G418-resistant cells by RT-PCR. Examples of two such clones are shown in Fig. 1D: c2 (a low expressor) and c13 (a high expressor). Using a previously characterised N-terminal PC1 antibody (7e12), we confirmed the expression of NT1 in both clones by immunoblotting (Fig. 1D) and surface expression by FACS analysis (Fig. 1E). Of interest, NT1-GPI was expressed as a doublet band possibly because of post-translational modification such as glycosylation. The absence of the GPS cleavage site excludes this as a possible cause.
Stable expression of NT1-GPI in L929 cells causes cell aggregation inhibitable by an anti-PC1 antibody
To extend the findings seen in transiently transfected L929 cells, we carried out aggregation assays on c13, the highest NT1-GPI-expressing L929 clone. After a 4 hour incubation, cells expressing the NT1-GPI fusion protein formed visible large aggregates in suspension. By contrast, control mock-transfected cells remained as a single cell suspension after the same treatment (Fig. 2A). To confirm that PC1 was directly responsible for the increase in cell aggregation, we incubated the cells simultaneously with anti-IgPKD, a previously characterised PC1 inhibitory antibody. As shown in Fig. 2A, this PC1 antibody completely inhibited cell aggregation whereas non-immune serum had no effect at the same dilution. These results confirm that PC1 can specifically confer a cell-cell adhesive phenotype mediated by homophilic interactions of the IgPKD domains in the absence of cell-matrix cues. In addition, we found no difference between c13 and mock-transfected L929 clones in adhesion, spreading or migration assays (not shown), confirming that expression of NT1-GPI confers an increase in cell-cell but not cell-matrix adhesion.
Inducible expression of NT1-GPI in L929 cells causes cell aggregation
To further demonstrate that PC1 can directly mediate cell adhesion, we subcloned the NT1-GPI fusion protein into a tetracycline-inducible vector, pEBTetD (Bach et al., 2007) and expressed it transiently in L929 cells. Following a 16 hour induction with tetracycline, but not in its absence, expression of the NT1-GPI fusion protein was detectable by immunoblotting (Fig. 2B). As shown for the c13 line, induced cells showed significant cell aggregation after 4 hours compared with non-induced or wild-type cells (Fig. 2B).
Formation of cell junctions in NT1-GPI expressing L929 cells by electron microscopy
The structural nature of the cell-cell contacts formed in C13 NT1-GPI-expressing L929 cells following cell aggregation was investigated further by electron microscopy. Under lower magnification, cells can be seen to adhere to each other (Fig. 2Ci). With higher magnification, adhesion plaques, visible as regions of increased membrane density, could be seen (Fig. 2Cii). By contrast, wild-type L929 cells did not adhere and showed no signs of plaque formation (not shown).
PC1 regulates E-cadherin recruitment to intercellular junctions
The finding that NT1-GPI could induce structural junctions led us to study the role of PC1 during junction reassembly in polarised epithelial cells. To do this, we exploited the well-established `calcium switch' assay in MDCK cells (Nigam et al., 1992). In fully polarised cells grown to confluence, typical cell junctional proteins, E-cadherin (adherens junctions), desmoplakin (desmosomes) and ZO-1 (tight junctions) were appropriately located by immunofluorescence (Fig. 3A). In addition, endogenous PC1 detected with a previously well characterised antibody (anti-LRR) (Ibraghimov-Beskrovnaya et al., 1997; Ibraghimov-Beskrovnaya et al., 2000) could be clearly seen at primary cilia (not shown) and the lateral plasma membrane where it colocalised with E-cadherin and desmoplakin.
Following overnight incubation in low Ca2+ medium (<5 μM Ca2+), cell-cell contacts were lost and cells appeared to round up. Immunofluorescence staining revealed that E-cadherin, desmoplakin, ZO-1 and PC1 were all redistributed from the cell surface into cytoplasmic vesicles (Fig. 3A). When normal Ca2+ levels (1.8 mM Ca2+) were restored, intercellular junctions reformed over a period of 2 hours. Of the junctional proteins, E-cadherin was recruited earliest to the lateral membrane by 30 minutes as shown by immunofluorescence (Fig. 3A) and cell surface biotinylation (Fig. 3D). Lateral staining of ZO-1 and desmoplakin was only detectable after 2 hours (Fig. 3A). Interestingly PC1 was also recruited to the lateral membrane within 30 minutes of the calcium switch where analysis of Z-sections showed that it colocalised with E-cadherin for the 2-hour duration of the assay. The temporal and spatial coexpression of PC1 with E-cadherin indicated that PC1 is likely to be crucial to E-cadherin recruitment and the formation of adherens junctions (Fig. 3B). These findings were confirmed in paraformaldehyde-fixed MDCK cells using a second PC1 antibody (BD3) raised to the PC1 C-terminus (Fig. 3C). PC1 and E-cadherin were recruited to the cell membrane after 30 minutes. At this timepoint, no recruitment of desmoplakin was observed.
To investigate the possibility that polycystin 1 was crucial for E-cadherin recruitment to the plasma membrane and formation of adherens junctions, after the calcium switch we treated MDCK cells with the PC1 antibody, anti-IgPKD. As shown in Fig. 3B, antibody treatment following calcium switch resulted in a delay in the recruitment of both PC1 and E-cadherin to the lateral membrane at 30 minutes. This was confirmed by cell surface biotinylation experiments showing a reduction in surface E-cadherin expression in antibody-treated cells compared with cells treated with non-immune serum (Fig. 3D). The effect was, however, transient since by 120 minutes after the reintroduction of Ca2+, PC1 and E-cadherin could be seen at the lateral membranes in both antibody treated and untreated cells. Cilia expression was unchanged by antibody treatment (data not shown). These findings support the idea that PC1 regulates E-cadherin surface expression during junctional assembly.
Mapping the interaction domains between PC1 and E-cadherin
We next investigated the possibility that PC1 and E-cadherin form a complex by studying their association in native and heterologous cells. In MDCK cells, endogenous PC1 can be coimmunoprecipitated with E-cadherin (Fig. 4A). We confirmed this interaction by reciprocal immunoprecipitation of transiently transfected heterologous PC1 and E-cadherin in HEK293 cells (Fig. 4C). Surprisingly we found that a naturally occurring PKD1 mutant, R4227X, in which the coiled-coil domain (responsible for interacting with PC2) has been deleted, could still bind to E-cadherin in this assay. Transfected E-cadherin was detectable as a prominent doublet band as previous reported (Lickert et al., 2000). The upper band represents an immature precursor form which is also detectable for the endogenous protein (Fig. 3D).
To further define the interacting domains between the two proteins, we expressed the entire PC1 C-terminus as an epitope-tagged fusion with CD8 (CT1-CD8-Myc) and found that it was sufficient to pull-down E-cadherin. Previous studies had shown that phosphorylation of a serine-rich domain (840-855) in the C-terminus of E-cadherin is important for β-catenin binding (Lickert et al., 2000). Indeed, an E-cadherin mutant with a deletion in this region (S840X) lost the ability to bind endogenous β-catenin (Fig. 4E) but was still capable of binding CT1, albeit with a weaker signal than a mutant (S855X) retaining this domain (Fig. 4D). These findings indicate that the binding domain for PC1 in the C-terminus of E-cadherin must lie upstream of S840. Nevertheless, it is possible that the serine-rich domain might act as a regulatory site for binding affinity, probably through phosphorylation at specific residues. It is apparent however that PC1 can bind to E-cadherin independently of PC2 or β-catenin.
In vitro binding assays of purified bacterial fusion proteins containing the C-termini of PC1 (MBP-CT1) and E-cadherin were performed to investigate whether the two proteins could interact directly. Using this assay, we were unable to demonstrate a direct interaction between PC1 and E-cadherin (Fig. 4F). By contrast, a direct interaction between the C-termini of PC1 and PC2 was clearly seen. Our findings indicate that PC1-E-cadherin interaction is likely to be indirect and mediated through other, yet to be identified, adapter proteins. Alternatively, their direct interaction may require an essential post-translational modification absent in the bacterial proteins.
In this paper, we provide the first direct evidence that PC1 can act as a cell adhesion molecule by expressing the entire extracellular region of PC1 as a GPI-anchored protein (NT1-GPI) in L929 cells. Using a specific inhibitory antibody, we confirm that this is dependent on trans-homophilic interactions of the PKD extracellular domains. Previous studies had demonstrated the importance of this region for PC1 mediated cell-cell adhesion but were unable to distinguish between its homophilic and heterophilic interactions (with other cadherins) in renal epithelial cells (Ibraghimov-Beskrovnaya et al., 2000; Streets et al., 2003). In L929 cells, NT1-GPI also induced the formation of structural junctions when the cells were induced to aggregate in suspension implying that PC1 might play a crucial role in junction formation.
Our results also confirm that PC1 localisation is dynamic in MDCK cells and alters temporally with junctional assembly and the reacquisition of cell polarity. Using the well established calcium switch assay, we demonstrate that PC1 is localised best with desmoplakin in fully polarised cells but overlaps more with E-cadherin during repolarisation. This may explain some of the differences observed in previous work in which PC1 expression was found to colocalise either to adherens junctions or desmosomes (Scheffers et al., 2000; van Adelsberg, 2000). Differences in confluency, differentiation, polarity or cell type could have contributed to some of the observed differences (Ong, 2000).
The formation of a polarised kidney tubule relies on a series of sequential events including cell-cell association, the formation of epithelial junctions, the acquisition of apicobasal polarity and finally the imposition of planar polarity. A large body of work has shown that the assembly of epithelial junctions is a hierarchical process in which, E-cadherin plays a primary role (Adams and Nelson, 1998; Braga, 2002; Jamora and Fuchs, 2002). For instance, MDCK cells depleted of E-cadherin by siRNA show a marked delay in re-establishing cell polarity in a calcium switch assay (Capaldo and Macara, 2007). However, once formed, cell junctions are maintained even when E-cadherin was subsequently depleted, indicating that the role of E-cadherin is likely to be in the initial phase of junction assembly but is not essential for their maintenance. During junction assembly, E-cadherin is recruited early to cell-cell contact sites, forming a scaffold for other proteins such as the α-catenin and β-catenin, other actin-binding proteins and signalling molecules (Vasioukhin and Fuchs, 2001; Perez-Moreno et al., 2003; Bershadsky, 2004). The ectodomains of E-cadherin also mediate Ca2+-dependent homophilic interactions between adjacent cells thus strengthening cell-cell associations. Tight junction (TJ) proteins are initially recruited to cadherin contacts possibly through α-catenin-mediated recruitment of ZO-1 and only later move away to form the TJ (Muller et al., 2005). The desmosomal cadherin complex is then recruited to complete the formation of the intercellular junctions and the establishment of apicobasal polarity (Gumbiner et al., 1988; Yin and Green, 2004).
Our results suggest that PC1 plays a proximal role in the process of junction assembly in MDCK cells both by mediating initial cell-cell association and by regulating E-cadherin recruitment. In previous studies, we and others have shown that cell-cell association in a number of renal epithelial cells including MDCK can be disrupted by an antibody to the PKD domains (IgPKD) or soluble GST recombinant protein (Ibraghimov-Beskrovnaya et al., 2000; Streets et al., 2003). Of interest, the cell-cell dissociating effect of the antibody was reduced by the transgenic expression of PC1 and was absent in the PKD1 null cystic line (Streets et al., 2003). Taken together with our data from L929 cells, it seems highly probable that PC1 mediates cell-cell associations in the earliest phase of junction assembly.
During the calcium switch, PC1 and E-cadherin membrane localisation overlap temporally and spatially and can be delayed by pre-incubation with an antibody to PC1. The delay is, however, transient and can be overcome 2 hours after the reintroduction of Ca2+. These findings are corroborated by a recent study in which the rate of E-cadherin recruitment was reported to be enhanced in MDCK cells over-expressing PC1 following calcium switch (Markoff et al., 2007). Our results are very similar to those described for E-cadherin-depleted MDCK cells where junction formation was delayed but not prevented (Capaldo and Macara, 2007). These data indicate that PC1 and E-cadherin are important in the establishment of junctions but dispensable for their maintenance. Human cystic epithelial cells have been reported with variable levels of E-cadherin but with neo-expression of N-cadherin (Streets et al., 2003; Roitbak et al., 2004; Russo et al., 2005). Since these cells are primarily derived from large surface cortical cysts, we suggest that the loss of E-cadherin expression (rather than its early mislocalisation) and concomitant N-cadherin transcription are likely to occur as later events in cystogenesis.
We extended these findings by first showing that PC1 and E-cadherin can be coimmunoprecipitated from MDCK cells confirming previous findings in human pancreatic adenocarcinoma (HPAC) cells and normal human kidney cells (Huan and van Adelsberg, 1999; Roitbak et al., 2004). We show for the first time that both proteins can associate via their C-termini independently of previously described binding domains. In particular the PC1 mutant R4227X, which lacks a coiled-coil domain essential for PC2 binding (Qian et al., 1997) was still able to coimmunoprecipitate E-cadherin. For E-cadherin, it had been shown that a serine-rich domain between amino acids 840 and 855 is essential for β-catenin binding (Lickert et al., 2000). Significantly, binding affinity between E-cadherin and β-catenin could be altered by casein kinase II (CKII) and glycogen synthase kinase-3 β (GSK-3β) phosphorylation at specific serine residues (Lickert et al., 2000). Although deletion of this domain abolished the binding of endogenous β-catenin, PC1 association was still present, albeit reduced. We found no evidence of direct binding between PC1 and E-cadherin using in vitro binding assays. This could indicate the importance of key mediatory adapter proteins or the crucial role of post-translational modifications (including phosphorylation) in mediating and regulating this process. Future work will seek to address these questions.
In summary, we have demonstrated the ability of PC1 to promote strong aggregating ability in non-adherent L929 cells via homophilic interactions of the PKD domains. Moreover, its temporal, spatial and biochemical association with E-cadherin in MDCK cells indicates a probable crucial role in junction initiation rather than maintenance. This model could explain why the mislocalisation of some basolateral proteins such as Na+-K+-ATPase and epidermal growth factor receptor (EGFR) are often but not universally observed in cysts (Wilson et al., 1991; Du and Wilson, 1995) and why the deletion of Pkd1 in adult kidney tubules has a mild phenotype compared with that in the foetus (Lantinga-van Leeuwen et al., 2007). If the major role of PC1 is in junction initiation, a disturbance in its function (leading to cyst initiation) is most likely to be manifest under conditions of high junctional turnover such as during tubular elongation in the developing kidney or during tissue repair following injury in the mature kidney.
Materials and Methods
All chemicals were purchased from Sigma (Poole, UK) unless otherwise stated. MDCK II cells were a gift from N. Simmons (University of Newcastle, Newcastle, UK) and L929 cells were purchased from the ECACC (Salisbury, UK). Antibodies to PC1 (IgPKD, LRR and BD3) and desmoplakin were obtained from Oxana Ibraghimov-Beskrovnaya (Genzyme) and D. Garrod (University of Manchester, Manchester, UK), respectively. A rat monoclonal antibody to ZO-1 was obtained from the Developmental Studies Hybridoma Bank (University of Iowa, USA) and antibodies to E-cadherin and calnexin were purchased from BD Transduction Labs (Cowley, Oxford, UK). The following cDNA expression constructs were kind gifts from the following: PCMUIV-CD8α (S. Ponnambalam, University of Leeds, Leeds, UK), pBAT-EM2 (D. Garrod), PCR-3/GPI (N. Fasel, University of Lausanne, Lausanne, Switzerland) and pEBTetD (D. Grundemann, University of Cologne, Cologne, Germany).
A full-length PKD1 FLAG-tagged cDNA was generated by exon-linking (C.J.W., unpublished data). We modified this cDNA by replacing the C-terminus FLAG epitope with a Myc epitope (PKD1-Myc) using PCR. Similarly, an epitope-tagged PKD1 truncation mutant (PKD1-R4227X-HA) was generated by introducing a premature stop codon preceded by an HA-epitope tag sequence. To create NT1-GPI, the entire extracellular domain of PC1 (M1-R2961) was removed from full-length PKD1 cDNA by restriction digestion with EcoRI and BamHI and ligated in frame with a Thy-1 GPI anchor signal sequence in PCR-3/GPI. A tetracycline inducible NT1-GPI construct was generated by subcloning the NT1-GPI insert into the pEBTetD vector (Bach et al., 2007). Full length HA-tagged E-cadherin was generated in pcDNA3 by PCR from the plasmid pBATEM2, which contains the entire coding region of E-cadherin (Nose et al., 1988). HA-tagged E-cadherin truncation mutants were similarly constructed with stop codons replacing S840 and S855, respectively. The PKD1 C-terminus (aa 4107-4303) containing a Myc epitope tag replacing the CD8 cytoplasmic tail was cloned in frame with the preceding CD8α sequence to create the CT1-CD8-Myc fusion construct (Ponnambalam et al., 1994). All constructs were verified by sequencing.
Cell culture and transfection
HEK293, MDCK and L929 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. Transient transfection was carried out on cells cultured to 90% confluency using Lipofectamine 2000 (Invitrogen, Paisley, UK) according to the manufacturer's instructions. L929 cells stably expressing NT1-GPI were generated by selecting resistant clones with 800 μg/ml G418 for a period of 3 weeks. Protein expression in cells expressing a pEBTetD NT1-GPI inducible vector was induced by addition of 2 μg/ml tetracycline to the culture media for 16 hours.
RNA was extracted from stable clones using Trizol reagent, cDNA was synthesised using a reverse transcriptase kit (Ambion, Applied Biosystems, UK) and expression of the NT1-GPI construct was determined by RT-PCR. A forward primer specific to the N-terminus of PC1 and a reverse primer specific to the Thy-1 signal sequence were used.
Immunoprecipitation and immunoblotting
Total cell lysates were prepared and processed for immunoprecipitation and western blotting as previously described (Newby et al., 2002).
Cells were grown on filters and fixed with acid-ethanol at 4°C for 1 hour or 4% paraformaldehyde for 5 minutes followed by permeabilisation with PBS, 0.1% Triton X-100. Blocking was carried out for 1 hour with 5% milk powder in PBS, and primary antibodies were incubated overnight at 4°C in 3% BSA-PBS. Controls included cells stained without the primary antibody, an irrelevant IgG1 mAb (Serotec, Kidlington, UK) or a non-immune rabbit IgG fraction (Dako, Ely, UK). Antibody binding was visualised using FITC-conjugated goat ant-mouse IgG and Alexa-Fluor-568-labelled goat anti-rabbit secondary antibodies. Slides were viewed using an Olympus Imaging Systems inverted IX71 microscope configured for multi-fluorescence image capture. Images were acquired and analysed using SimplePCI imaging software (Compix). Z sections were acquired and analysed using a Zeiss LSM 510 META Axiovert 200M confocal microscope and Zeiss LSM image analysis software.
Cell surface biotinylation
Cell surface biotinylation was performed as previously described (Streets et al., 2006). Samples were analysed by SDS-PAGE and western blotting using the antibodies described.
Fluorescence-activated cell sorting (FACS)
Cells cultured to confluency in 10 cm culture dishes were detached using cell dissociation buffer (Invitrogen) and resuspended in ice cold 3% BSA-PBS at 1×107 cells/ml. They were incubated with 1/50 dilution of anti-PC1 7e12 antibody or mIgG1 negative control antibody (Serotec, Kidlington, UK) for 30 minutes and washed three times by centrifugation at 400 g for 5 minutes in PBS. Secondary antibody incubation was with a 1/300 dilution of goat anti-mouse FITC for 30 minutes. After washing cells were resuspended in PBS and analysed by flow cytometry.
Cell aggregation assay
Cells grown to confluence were detached using cell dissociation buffer (Invitrogen, UK). A single cell suspension was generated by gently passing the detached cells through a 21 gauge needle and this was verified by light microscopy. DNaseI (1 μg/ml) was added to prevent clumping as a result of release of DNA from dead cells. Cells were gently mixed at 100 r.p.m. in 12-well dishes pre-coated with 1.5% BSA for 4 hours. To test the effect of the anti-PC1 blocking antibody IgPKD on cell aggregation, IgPKD or non-immune rabbit serum was added at 1/50 for the same period of time. Images were taken and particle numbers quantified from six random fields using ImageJ (NIH). Results were expressed as Nt/N0 where Nt is the number of particles at 4 hours and N0 is the number of particles at the start (Hordijk et al., 1997).
Calcium switch assay
Calcium switch assay was carried out essentially as described previously (Nigam et al., 1992) with the following modifications. Cells were grown on filters to full confluence in complete medium. They were rinsed three times and then incubated overnight in Spinner's modified Eagle's medium (SMEM; low Ca2+ media: <5 μM Ca2+) with 5% dialysed FCS. The cells were then switched to complete medium for different lengths of time ranging from 0 minutes to 2 hours either in the presence of non-immune rabbit sera or 1/50 IgPKD antibody, after which cells were fixed as described above.
Following an aggregation assay, cells in suspension were fixed with an equal volume of 3% phosphate-buffered glutaraldehyde in PBS for 1 hour and then gently sedimented by centrifugation at 1000 g for 5 minutes. The supernatant was then removed and the cell pellet stored in 3% glutaraldehyde for a further 24 hours. The cells were then resuspended in molten 2% agarose. On setting, the agarose block was chopped into 1 mm cubes and further fixed in 3% phosphate-buffered glutaraldehyde. This was followed by further fixation in 1% aqueous osmium tetroxide, processing through graded alcohols, propylene oxide and into medium grade epoxy resin. Following heat polymerisation, resin blocks were sectioned using a Reichert ultracut E ultramicrotome at 0.6 μm, stained using 1% Toluidine blue in 1% sodium tetraborate and the sections examined by light microscopy. The most appropriate block was selected and thin-sectioned using a Diatome diamond knife at 85 nm. Thin sections were then stained in saturated uranyl acetate in 99% ethanol and Reynold's lead citrate. Sections were examined on a Philips 400 transmission electron microscope and photographed using large format Kodak EM negatives.
In vitro binding assays
Constructs encoding the PC1 C-terminal domain and the truncated PC1 C-terminal domain (CT1R4227X) as maltose-binding protein (MBP) fusions or MBP alone were made in pMAL-c2 (New England Biolabs). The C-terminus of E-cadherin or PC2 (CT2) were cloned in-frame with histidine (His) and thioredoxin (Thio) in the pet32a+ vector (Novagen) and the constructs verified by sequencing. Fusion proteins were induced in BL21-DE3 with 1 mM isopropyl β-D-thiogalactopyranoside for 2 hours and purified from cleared bacterial lysates using amylose (New England Biolabs) or Nickel His-Select beads (Sigma) as described by the manufacturer. Affinity-purified recombinant proteins were visualized on Coomassie blue stained gels and immunoblotting with MBP and thioredoxin (Invitrogen) antibodies. Binding assays were performed by mixing 2 μg of MBP, MBP-CT1, MBP-CT1R4227X fusion protein attached to amylose beads with either 2 μg of His-Thio E-cadherin CT or His-Thio CT2 in binding buffer (5 mM Hepes, 150 mM NaCl, 1.5 mM MgCl2, 0.1 mM EDTA, 0.1% NP-40). Samples were rotated for 1.5 hours at 4°C, pelleted by centrifugation and washed three times with bead-binding buffer prior to elution.
This work was funded by the Wellcome Trust (GR071201). We thank O. Ibraghimov-Beskrovnaya, D. Garrod, N. Simmons, S. Ponnambalam, N. Fasel and D. Grundemann for kind gifts of reagents, and G. Howell (Leeds Bioimaging Facility) for technical assistance with confocal imaging. Portions of this work were presented at the American Society of Nephrology 2007 meeting, in San Francisco and the British Society for Cell Biology meeting on `Epithelial Morphogenesis and Diseases' (London, 8-10 October 2008). A.J.S. is a Research Councils UK Academic Fellow and A.C.M.O. is a Wellcome Trust Research Leave Senior Fellow. Deposited in PMC for release after 6 months.
- Accepted January 19, 2009.
- © The Company of Biologists Limited 2009