Human ADPKD primary cyst epithelial cells with a novel, single codon deletion in the PKD1 gene exhibit defective ciliary polycystin localization and loss of flow-induced Ca2+ signaling

Chang Xu, Sandro Rossetti, Lianwei Jiang, Peter C. Harris, Ursa Brown-Glaberman, Angela Wandinger-Ness, Robert Bacallao, Seth L. Alper


Autosomal dominant polycystic kidney disease (ADPKD) gene products polycystin-1 (PC1) and polycystin-2 (PC2) colocalize in the apical monocilia of renal epithelial cells. Mouse and human renal cells without PC1 protein show impaired ciliary mechanosensation, and this impairment has been proposed to promote cystogenesis. However, most cyst epithelia of human ADPKD kidneys appear to express full-length PC1 and PC2 in normal or increased abundance. We show that confluent primary ADPKD cyst cells with the novel PC1 mutation ΔL2433 and with normal abundance of PC1 and PC2 polypeptides lack ciliary PC1 and often lack ciliary PC2, whereas PC1 and PC2 are both present in cilia of confluent normal human kidney (NK) epithelial cells in primary culture. Confluent NK cells respond to shear stress with transient increases in cytoplasmic Ca2+ concentration ([Ca2+]i), dependent on both extracellular Ca2+ and release from intracellular stores. In contrast, ADPKD cyst cells lack flow-sensitive [Ca2+]i signaling and exhibit reduced endoplasmic reticulum Ca2+ stores and store-depletion-operated Ca2+ entry but retain near-normal [Ca2+]i responses to ANG II and to vasopressin. Expression of wild-type and mutant CD16.7-PKD1(115–226) fusion proteins reveals within the COOH-terminal 112 amino acids of PC1 a coiled-coil domain-independent ciliary localization signal. However, the coiled-coil domain is required for CD16.7-PKD1(115–226) expression to accelerate decay of the flow-induced Ca2+ signal in NK cells. These data provide evidence for ciliary dysfunction and polycystin mislocalization in human ADPKD cells with normal levels of PC1.

  • autosomal dominant polycystic kidney disease
  • monocilium
  • shear stress
  • protein trafficking
  • fura 2

autosomal dominant polycystic kidney disease (ADPKD) is the most common life-threatening monogenic human renal disease, with a prevalence of between 1:400 and 1:1,000. It is characterized by progressive development and enlargement of fluid-filled cysts originating from only ∼3% of nephrons, leading ultimately to renal failure in 50% of affected individuals. More than 85% of ADPKD cases are caused by mutations in the PKD1 gene, with almost all remaining cases associated with PKD2 gene mutations. The PKD1 polypeptide gene product, polycystin-1 (PC1/TRPP1), is a 4,302-amino acid (aa) polypeptide with an NH2-terminal extracellular domain of ∼3,000 aa, ∼11 transmembrane domains, and a ∼200-aa COOH-terminal cytoplasmic domain interacting with polycystin-2 (PC2/TRPP2) (46, 63), heterotrimeric G proteins, and the regulator of G protein signaling RGS7, among many other proteins. The COOH-terminal tail of PC1 also upregulates several transcriptional pathways, in part by regulated proteolysis (24), and activates endogenous Ca2+-permeable cation channels of 20–30 pS in Xenopus laevis oocytes and HEK-293 cells (64, 65). The PKD2 gene product, PC2, is a 968-aa polypeptide believed to function as a Ca2+-permeable cation channel in the endoplasmic reticulum and/or at the plasma membrane, independently or in complex with PC1 (11, 26, 33) or other proteins. The cellular functions of PC1, PC2, and the PC1/PC2 complex remain incompletely understood, but likely include roles in epithelial cell proliferation, differentiation and tubulogenesis, matrix interaction, Ca2+ signaling, and determination of developmental asymmetry in the embryonic ventral node.

Nearly all polarized epithelial cell types express a central apical monocilium with a “9 + 0” axoneme structure, long considered vestigial but periodically proposed to function as a mechanosensor. Recent findings from diverse fields have converged to suggest a central role for the primary cilia of renal tubular epithelial cells in the cystogenesis of polycystic kidney disease (8, 12, 41). After the Madin-Darby canine kidney (MDCK) cell cilium was shown to respond to mechanical bending and to flow by transducing an increase in cytoplasmic Ca2+ concentration ([Ca2+]i) (42, 43), several genes encoding intraflagellar transport proteins of the green alga Chlamydomonas were noted to encode cystic kidney disease genes and to localize to the renal epithelial cilium (40, 74). These findings promoted the discovery of altered ciliary morphology in the orpk mouse (40) and provided additional insight into ciliary localization of the ADPKD gene homologs lov-1 and pkd2 in sensory neurons of Caenorhabditis elegans, and ciliary or basal body functions of several glomerulocystic disease genes from zebrafish.

Nauli et al. (30) soon proposed that the PC1/PC2 complex functions as a flow-sensing mechanoreceptor in the primary cilia of primary cultures of mouse embryonic renal epithelial cells and showed that cells from pkd1(−/−) mice are deficient in this proposed sensing function. The partially defective flow-sensing in the orpk mouse (21, 55) further supported the proposed central role of defective ciliary sensation of and/or response to tubular flow to the cystogenesis of ADPKD (30). The pkd1(−/−) mouse embryonic renal epithelial cells in which flow-induced signaling defects were observed completely lacked PC1 and PC2 polypeptides. A recent report appearing after completion of the work presented here extended this observation to human ADPKD cyst cells expressing little or no PC1 polypeptide (31). However, human ADPKD is almost always characterized by normal or increased renal levels of apparently full-length PC1 polypeptide (32, 36, 37), despite the significantly truncated proteins encoded by most PKD1 germline mutations (35). Indeed, phenotypically similar murine polycystic kidney disease (PKD) results from knockout and from overexpression of the wild-type pkd1 gene (44, 60).

Therefore, we compared ciliary expression of PC1 and PC2 and flow-sensitive Ca2+ signaling in primary human renal epithelial cells derived from normal kidneys (NK cells) or from ADPKD cysts (PKD cells). We report that NK cells and PKD cells with a novel heterozygous in-frame single codon deletion mutation in the PKD1 gene exhibit equivalent abundance of PC1 and PC2 polypeptides, but differ in their ciliary localization of PC1 and PC2. Exposure to low shear stress increased [Ca2+]i in a minority of NK cells, but shear stress at slightly higher levels and at the still higher levels achieved during diuresis increased [Ca2+]i to higher peak values in most NK cells. In contrast, PKD cyst cells exhibited no flow-sensitive elevation of [Ca2+]i at any level of shear stress. PKD cyst cells were also characterized by reduced endoplasmic reticulum Ca2+ stores, reduced capacitative Ca2+ entry, and near-normal hormone-induced Ca2+ signaling. We also report that aa 115–226 of the PC1 COOH-terminal tail modulated flow-induced NK cell Ca2+ signaling through its coiled-coil domain and contain a coiled-coil domain-independent ciliary localization sequence.


Cell culture.

Human renal cyst epithelial cells (PKD) were harvested from multiple superficial cortical cysts of kidneys resected from ADPKD patients PKD 10–27-98 and PKD 3/14/00, in both of whom disease progression, family histories, and pathological examination suggested germline PKD1 mutations. Cortical tubules were dissected from one freshly harvested cadaveric human kidney not used for transplant (NK 6–1-99) and from the grossly normal lower poles of two human kidneys (NK57M03 and NK 11–7-02) resected for renal cell carcinoma. The epithelial cells from both sources were grown in primary culture and passaged as described (4, 5). Cells for experiments were transferred to glass coverslips coated with Vitrogen collagen (Conhesion Technologies, Palo Alto, CA), fed every other day with Clonetics Renal Epithelial Cell Media (REBM, Clonetics), and grown to confluence 5–6 days after plating. Only cells between passages 2 and 6 were used for experiments. Experiments presented in Figs. 37 and in Supplemental Fig. 2 were replicated with and include cells from all donor lines (the online version of this article contains all supplementary material). PKD cyst cell results presented in Figs. 810 and Supplemental Fig. 1 are from donor PKD 10/27/98. All discarded tissue was harvested according to Committee on Clinical Investigations/Institutional Review Board protocols reviewed and approved at Indiana University School of Medicine and Beth Israel Deaconess Medical Center.

Fig. 1.

Novel heterozygous missense mutation in the human PKD1 gene. A: aligned nucleotide and predicted polypeptide sequences from exon 18 of wild-type (wt) and mutant PKD1 alleles in primary cyst epithelial cells. The mutant allele has a 3-nucleotide deletion, producing an in-frame deletion of a single codon, and encodes the polycystin 1 mutant polypeptide polycystin-1 (PC1) ΔL2433. B: aligned genomic DNA sequence traces from a representative wild-type genome (top) and the heterozygous mutant cyst cell genome (bottom). Underscored automated sequence call starts at the deletion site. C: schematic of the PC1 polypeptide locating the Leu2433 deletion within the receptor-for-egg jelly (REJ) domain. TM, transmembrane.

Fig. 2.

Autosomal dominant polycystic kidney disease (ADPKD) cyst cells express normal or elevated levels of PC1 and PC2. A: immunoblot of normal kidney cell (NK) lysates with 3 anti-PC1 antibodies: 7e12, NM005, and LRR. Lanes are from a single blot. Shown is a representative of 6 similar experiments. B: immunoblot of NK and ADPKD (PKD) cell lysates containing equivalent amounts of protein were probed with anti-PC1 antibody NM005. Shown is 1 of 2 similar experiments. C: PC1 polypeptide abundance in human pancreatic adenocarcinoma cells (HPAC) is reduced by transfection of specific PC1 small-interference (si) RNA, whereas GAPDH is unaffected. Transfection of control siRNA had no effect on either PC1 or GAPDH bands (not shown). Shown is a representative of 6 similar experiments. D: PC2 immunoblot of cell lysates separated on 7.5% polyacrylamide gels and probed with anti-PC2 antibody. PC2 was detected in cell lysates prepared in RIPA buffer (lanes 1 and 3) or in SDS buffer (lanes 2 and 4). The same blot was reprobed with antibody to β-actin as a loading control. Shown is a representative of 3 similar experiments.

Fig. 3.

Identification of primary cilia in human renal epithelial cells in culture by localization of N-acetylated α-tubulin in NK (left) and PKD cyst cells (right). Note that the primary cilia appear as dots (arrows) in this out-of-focus view from above the apical cell surface. When the dotlike structure was viewed by linescan through the x-z plane, the entire primary cilium can be seen (as shown in Figs. 4 and 5). Scale bar = 10 μm.

Human pancreatic adenocarcinoma cells (HPAC), HeLa, and HEK 293 human embryonic kidney cells from American Type Culture Collection were grown in DMEM supplemented with 10% fetal calf serum.

Genomic DNA analysis.

Genomic DNA was prepared from NK cells from individual NK57M03 and cyst cells from individual PKD 10/27/98. All coding exons of the PKD1 and PKD2 genes were amplified by PCR from genomic DNA as previously reported (51, 52) or with newly developed primers (available on request). Amplicons (300–600 ng) were analyzed by denaturing high-pressure liquid chromatography (DHPLC) performed at two temperatures on a DNASep Cartridge HT with the Wave System 3500HT (Transgenomic, Omaha, NE) and eluted through a 2.5-min linear gradient of buffer A [5% triethylammonium acetate (TEAA)] and buffer B (5% TEAA and 25% acetonitrile). Samples with abnormal elution chromatograms compared with a normal control were subjected to DNA sequencing. The DNA sequencing allowed identification of heterozygous variation in the gene sequence.

Antibodies and other reagents.

Rabbit polyclonal anti-PC1 antibody NM005 raised against the 223-aa recombinant PC1 COOH-terminal cytoplasmic domain aa 4070–4302 (66) was used as an Ig fraction, with specificity confirmed by fusion protein antigen preadsorption as previously described (50) and by small-interference (siRNA) knockdown as described below. Rabbit polyclonal antibody leucine-rich repeat (LRR) (14) and monoclonal antibody 7e12 (36, 37), both raised against the PC1 leucine-rich repeat domain, were previously described. Rabbit polyclonal antibody NM002 was raised against PC1 aa 3619–3631, affinity-purified by peptide antigen column, and specificity tested by siRNA knockdown. Anti-PC2 antibody (64) raised against a GST-PC2 fusion protein encoding the PC2 COOH-terminal cytoplasmic aa 687–968 was originally obtained from Dr. Oxana Beskrovnaya-Ibraghimova (Genzyme). Anti-CD16 monoclonal antibody (mAb) 3G8 was used as an Ig fraction as described (64, 65). Anti-N-acetylated-α-tubulin was obtained from Sigma, anti-calnexin from Stressgen, and anti-GM130 from BD Biosciences Pharmingen. FITC-labeled lectins were from Vector Laboratories. GsMTx-IV was obtained from Peptides International (Osaka, Japan). All other drugs were from Sigma.


Cells scraped in ice-cold PBS in the presence of Complete protease inhibitor (Roche Diagnostics, Mannheim, Germany) were pelleted, rinsed in the same medium, then suspended and boiled briefly in 1% SDS, 10 mM Tris·HCl, pH 7.4, or in RIPA buffer (150 mM NaCl, 1% NP-40, 0.5% DOC, 0.1% SDS, 50 mM Tris·HCl, pH 7.5). Samples were suspended in Laemmli sample buffer (100 mM Tris·HCl, pH 6.8, 4% SDS, 20% glycerol, 0.1% bromophenol blue, 5% β-mercaptoethanol) and incubated 30 min at 37°C before loading onto Nupage 3–8% polyacrylamide-SDS Tris-Acetate gels (Invitrogen, Carlsbad, CA) or Criterion 5% polyacrylamide-SDS gels (Bio-Rad) for PC1 analysis, and 7.5% acrylamide-SDS gels for PC2 analysis. Proteins separated by electrophoresis were transferred at 100 V for 1 h to nitrocellulose, blocked with 5% milk in BLOTTO buffer (20 mM Tris, 0.9% NaCl, 0.03% Tween 20, pH 7.4) for 30 min at 37°C, and probed with the primary antibody for 1 h at 37°C followed by horseradish peroxidase-conjugated goat anti-rabbit-Ig for 1 h at 20°C. Bound secondary antibody was detected by enhanced chemiluminescence (Boehringer) on SB5 X-ray film (Kodak).

cDNA transfection.

Lipofection was ineffective for transfection of NK and PKD cyst cells. Therefore, cells were trypsinized for 7 min at 37°C, pelleted, and resuspended at 1.5–2.5 × 106 cells/100 μl in basic primary mammalian epithelial cell Nucleofector solution (Amaxa Biosystems, Cologne, Germany). After addition of 10 μg plasmid DNA, the mixture was placed in a 2-mm cuvette and electroporated at room temperature in the Nucleofector device (Amaxa) with program T-05. Five hundred microliters of prewarmed REBM medium containing 10% FBS was immediately added to the cuvette, and the suspension was plated to a 35-mm coverslip in a 60-mm2 culture dish. The transfected cells were incubated at 37°C for 5 days or longer in REBM containing 10% FBS. Apparent transfection efficiency assessed cytologically by GFP fluorescence 6 days postelectroporation was up to 70% in both cell types. Transfection efficiency assessed by expression of CD16.7-PKD1(115–226) was 15% in NK cells and 5% in PKD cyst cells.

Confocal immunofluorescence microscopy.

Cell monolayers grown on coverslips were fixed for 30 min at room temperature with PBS containing 3% (wt/vol) paraformaldehyde. Fixed cells were extensively rinsed with PBS, quenched with 50 mM lysine HCl, pH 8.0, exposed to 1% SDS for 15 min, and blocked for 15 min in PBS with 1% bovine serum albumin and 0.05% saponin. After 4°C overnight incubation with the primary antibody against PC1 or PC2 (1:100), coverslips were incubated with Cy3-conjugated donkey anti-rabbit Ig secondary antibody (1:500) for 2 h at room temperature. Some coverslips immunostained as above for PC1 or PC2 were then costained with antibodies to the ciliary marker N-acetylated α-tubulin (1:500), the endoplasmic reticulum (ER) marker calnexin (1:500), or the Golgi marker GM130 (1:200 dilution). On occasional coverslips (as specified), unfixed cells were preextracted for 5 min with 0.01% saponin in “cytoskeletal buffer” [containing (in mM) 138 KCl, 3 MgCl2, 2 EGTA, and 10 2-N-morpholino-ethane-sulfonate, pH 6.1] before fixation and immunostaining as above.

Costaining with two rabbit polyclonal antibodies against PC2 and calnexin (supplemental Fig. 3) was performed by the microwave denaturation method (62). After completion of PC2 staining, coverslips were microwaved 10 min in 10 mM citrate, pH 6.0, to denature bound antibody molecules and prevent cross-reaction during subsequent calnexin staining with the rabbit polyclonal antibody and FITC-coupled goat anti-rabbit Ig secondary antibody. The same microwave procedure was used to costain with two mAbs against CD16 and N-acetylated α-tubulin (see Fig. 9). In both cases, control incubations with secondary antibody post-microwave treatment alone confirmed successful denaturation of previously bound Ig (not shown). Immunostained cells were imaged with a Bio-Rad MRC-1024 laser-scanning confocal immunofluorescence microscope. Pixel fluorescence intensity of cilia and cell bodies in x-z sections of single cells was measured with Bio-Rad software.

RNA knockdown.

Six human PC1 siRNA sequences of 21 nt in length complementary to several PC1 domains of the 14,135-pb coding sequence of PKD1 (GenBank NM000296) were selected with Ambion's “siRNA Target Finder tool” ( The most effective sequence targeted nt 584–605 within the LRR domain. A BLAST search confirmed the lack of significant homology with other human genes. Sense and antisense oligodeoxynucleotides (Integrated DNA Technologies, Coralville, IA) were used to generate templates and synthesize siRNA according to the manufacturer's instructions. PC1 siRNA, GAPDH siRNA, or control scrambled siRNA (Ambion control template set 4800) were separately transfected into 50% confluent HPAC (50–100 pmol/well of a 6-well plate) using Lipofectamine 2000 (Invitrogen) per the manufacturer's instructions. Knockdown efficacy was evaluated by immunoblot of cells lysed in SDS-PAGE sample buffer containing 5% 2-mercaptoethanol and resolved on precast 4–15% gels (Bio-Rad Criterion), using the anti-PC1 antibody NM005 or monoclonal anti-GAPDH antibody (Ambion). Chemiluminescent signal was quantitated with a PhosphoImager (Molecular Dynamics) using ImageQuant software. The fold-decrease was calculated relative to the scrambled siRNA control. Maximal knockdown was observed 48 h posttransfection.

Measurement of [Ca2+]i.

NK and PKD cyst cells cultured to confluence on collagen-coated 35-mm glass coverslips were loaded with 5 μM fura 2-AM (Molecular Probes) in HEPES-buffered (pH 7.2) at 37°C for 30 min. Extracellular fura 2-AM was removed by washing twice with HEPES-HBSS. The coverslip was then mounted into the bottom of a parallel-plate flow chamber (GlycoTech) 0.5 cm in width and 0.0254 cm in depth and perfused at room temperature with a Harvard Syringe pump. In some experiments, cells were perfused at 37°C using a calibrated WPI in-line heater, with monitoring of inflow and outflow temperatures. Perfusion medium composition was (in mM) 127 NaCl, 5.4 KCl, 1.27 CaCl2, 1 MgCl2, 5.6 glucose, and 11.6 HEPES, final pH 7.2.

[Ca2+]i was measured by fluorescence ratio imaging with a Metafluor digital imaging system (Universal Imaging, West Chester, PA), equipped with an Olympus IMT-2 inverted microscope, and a CoolSNAP CCD camera (Photometrics, Tucson, AZ). Fura 2 emission ratio images were monitored at 510 nm with alternating excitation at 340 and 380 nm. Fura 2 fluorescence ratio values determined by in situ calibration in immortalized epithelial cells did not differ from values determined by in vitro calibration for [Ca2+]i (29). Therefore, fura 2 fluorescence ratios were calibrated in vitro (29) with the same experimental settings for the imaging system, using the Ca2+ calibration buffer kit no. 2 (Molecular Probes) with concentrations between 36 nM and 4 μM [Ca2+]. The minimal fluorescence ratio (Rmin) was determined at “zero Ca2+” (free Ca2+ < 10 nM) and the maximal fluorescence ratio (Rmax) at 4 μM total Ca2+. The equilibrium constant (Kb) was determined by fitting experimental fluorescence ratio R values at various free [Ca2+] with the equation [Ca2+]free = Kb (Sf2/Sb2)[(R − Rmin)/(Rmax − R)], where the factor Sf2/Sb2 corrects for fura 2 ion selectivity at 380 nm.

For each coverslip, one visual field was selected as a region of interest, recorded before and during imposition of a uniform rate of fluid flow.

Shear stress (τw) was calculated as τw = 6μQ/a2b, where μ = apparent viscosity of superfusate (1.00 for H2O at 20°C; 0.70 at 37°C), Q = volumetric flow rate (ml/s), and a and b = flow chamber depth and width, respectively.

All cells within the visual field were analyzed as a single region of interest. On some coverslips, individual cells were analyzed separately as noted. Fluorescence ratio emission values were calculated on a pixel-by-pixel basis from ratio images and processed with Metamorph software (Universal Imaging). The statistical significance of differences in [Ca2+]i between groups was evaluated by Student's paired or unpaired t-test or by Fisher's test. These comparisons between mean [Ca2+]i measured from all cells within the visual field were corroborated by application of the Wilcoxon rank order test to comparisons of means of single-cell [Ca2+]i from all cells within the visual fields of two experimental groups (not shown).

In addition to analysis of all cells within a coverslip's visual field, the subset of “responsive” NK cell coverslips was also analyzed separately. An NK cell coverslip was defined as “unresponsive” if the mean fura 2 fluorescence emission ratio of three-to-five arbitrarily selected subregions of the recorded visual field did not increase during flow to a peak value significantly higher than the ratio recorded before imposition of flow (paired Student's t-test). Coverslips on which such subregions did show significantly increased fluorescence ratio during flow were deemed responsive coverslips (see Fig. 6, A and B, for an analysis of all NK coverslips and Fig. 6, C and D, for data from only responsive coverslips).

For each coverslip of fura 2-loaded, cDNA-transfected cells, individual dsRed-positive and dsRed-negative cells in a visual field were preselected as regions of interest, then recorded before and during fluid flow. Pharmacological characterization (see Fig. 8A, C, and D) was carried out in a coverslip chamber of 1-ml volume to which 100 × drug stock solution was added by pipette under “no-flow conditions,” which in the absence of drug elicited no increase in [Ca2+]i. Ca2+ readdition (see Fig. 8D) was performed by superfusion at 0.75 dyne/cm2, or (as noted) by gentle manual replacement of half the cell chamber volume.


A novel ADPKD mutation.

Genomic DNA was prepared from cyst cells of individual PKD10–27-98 and from normal cortical tubular epithelial cells of individual NK57M03. No mutations of the PKD1 or PKD2 genes were detected in the NK cell sample. In contrast, the PKD cyst cell genome revealed a heterozygous in-frame deletion of a tgc trinucleotide in exon 18 (7509_7511delTGC), encoding the predicted mutant polycystin-1 polypeptide PC1 ΔL2433 (p.Leu2433del) lacking a leucine in the receptor-for-egg jelly (REJ) domain (Fig. 1). The deleted Leu residue encoded by the human gene is identical in chicken, X. laevis, and Fugu, and is conserved as Val in mouse and rat. Two additional noncoding polymorphisms were also detected in cis: IVS31–38C-G and IVS44+22delG. Since neither the coding deletion mutation nor the two IVS mutations were present in 100 chromosomes from 50 normal control individuals, the deletion is likely the germline disease mutation. However, the anonymity constraints under which kidneys were obtained prevented confirmation by genotyping of family members. No PKD2 mutations were found in this patient.

PC1 and PC2 polypeptide expression in NK and PKD cyst cells.

PC1 was detected by immunoblot of confluent NK cell lysate with three distinct antibodies (7e12, NM005, and LRR) as a polypeptide of Mr >400 kDa (Fig. 2A). The total cell abundance of PC1 in confluent PKD cells was no less than that in confluent NK cells (Fig. 2B). Immunoblot specificity of the previously characterized NM005 PC1 antibody (50) was confirmed by the 85% reduction in the >400 kDa PC1 immunoblot band in PC1 siRNA-treated HPAC cells (Fig. 2C). A similar PC1 band of >400 kDa was detected by all three antibodies also in HEK 293 cells, HeLa cells, and HPAC cells (not shown). PC1 immunolocalization with the NM005 antibody revealed a predominantly intracellular distribution in both NK cells and PKD cells, with some PC1 detectable at lateral cell membranes in NK and PKD cells as previously described (50) (Supplemental Fig. 1).

PC2 migrated with Mr ∼110 kDa and was detected at equal abundance in NK and PKD cyst cells (Fig. 2B). PC2 localized predominantly to the ER in NK and PKD cells, as evidenced by colocalization with calnexin (Supplemental Fig. 2), and consistent with previous reports (2, 18).

PKD cyst cells form primary cilia devoid of PC1.

PC1 and PC2 have been colocalized, along with other ciliary gene products linked to cystic kidney disease, to the primary cilium in mouse kidney, in primary cultured mouse kidney cells, and in several mammalian kidney cell lines (58, 66, 73). Human kidney cells in culture also exhibit primary cilia (40, 31). Figure 3 shows the presence of cilia in both NK and PKD cyst cells as detected by immunostaining of the ciliary axoneme marker N-acetylated α-tubulin. γ-Tubulin is also localized to cilia in both cell types (not shown). Neither cell type expressed a primary cilium at day 1 or 2 after splitting (<50% confluence), but by day 3 NK cells within confluent islands expressed short primary cilia of 2.7 ± 0.1 μm in length (n = 32). At confluence 6 days after splitting, PKD cyst cell ciliary length was 3.1 ± 0.1 μm (n = 73), slightly shorter than that of NK cells (4.2 ± 0.1 μm; n = 81; P < 0.01).

PC1 polypeptide in NK cells colocalized in the central cilium with N-acetylated-α-tubulin in all 37 NK cells examined but in none of the 38 PKD cyst cells immunostained with anti-PC1 antibody NM005 (Fig. 4). Ciliary immunostaining intensity exceeded that of the cell body. Use of anti-PC1 antibody NM002 similarly revealed ciliary PC1 localization in all 24 NK cells examined, but in none of 17 additionally examined PKD cyst cells (not shown). PC2 polypeptide also colocalized with the ciliary marker in all 41 cilia examined in NK cells (Fig. 5A), with ciliary PC2 immunofluorescence intensity again exceeding that of the cell body. However, PKD cyst cells exhibited two patterns of ciliary PC2 expression. Although bright ciliary PC2 staining was evident in 11 of 30 PKD cyst cells (Fig. 5C), ciliary PC2 was undetectable in 19 of the 30 PKD cyst cells examined (Fig. 5B). The length of PC2-positive cilia in these 30 PKD cyst cells was 2.8 ± 0.2 μm, whereas PC2-negative cilia were consistently shorter (2.2 ± 0.1 μm; P < 0.05), of greater width, and less orthogonal in the fixed state than PC2-positive cilia (Fig. 5C). These data demonstrate localization of PC1 and PC2 to primary cilia in human NK cells, consistent with the recent observations of Nauli et al. (31). In contrast, PKD cyst cells with normal PC1 abundance lack ciliary PC1, and PC2 is absent from most cilia of cyst cells.

Fig. 4.

Colocalization of PC1 with N-acetylated α-tubulin is altered in PKD cells. Confocal x-z plane reconstructions show immunofluorescence colocalization of PC1 with N-acetylated α-tubulin in NK cells (A) but not in PKD cyst cells (B). PC1 was detected in NK cell cilia (white arrows) by antibody NM005. PC1 was absent from the PKD cyst cell cilium but present in the cell body. Cells were preextracted with saponin before fixation (see methods), reducing nuclear staining of PC1. Scale bar = 10 μm.

Fig. 5.

Colocalization of PC2 with N-acetylated α-tubulin is altered in PKD cyst cells. Confocal x-z plane reconstructions showing immunofluorescence colocalization of PC2 with N-acetylated α-tubulin in NK cells (A) and in PKD cyst cells (B and C). PC2 colocalizes with acetylated α-tubulin in the cilia (white arrows) of NK cells (A) but in only 30% of PKD cyst cells (B). In the other 70% of PKD cyst cells, PC2 was not detected in cilia (C). Scale bar = 10 μm.

ADPKD cyst cells lack the flow-induced [Ca2+]i increase observed in NK cells.

Confluent, ciliated MDCK cells (43) and confluent primary mouse embryonic kidney cells (30) responded to low-level shear stress with increased [Ca2+]i. However, primary mouse embryonic kidney cells from Pkd1(del34/del34) mice lacked this response to fluid flow. Although flow-sensitive Ca2+ signaling has been measured in isolated, perfused rabbit (71) and mouse (21) collecting ducts, flow-sensitive Ca2+ signaling had until recently (31) not been examined in human renal epithelial cells.

In NK cells, room temperature basal [Ca2+]i without flow was 141 ± 4 nM (n = 141 coverslips). Abrupt application of 0.75 dyne/cm2 shear stress (Fig. 6A) increased NK cell [Ca2+]i by 19 ± 5 nM in 66 coverslips studied (P < 0.01). Cells on 43 of these 66 coverslips exhibited no flow-induced [Ca2+]i increase and were considered unresponsive (see methods). Cells on the remaining 23 responsive coverslips (34%), considered separately, increased [Ca2+]i 56 ± 10 nM above basal levels in response to this level of flow (Fig. 6C). An increase in shear stress from 0 to 2.3 dyne/cm2 (Fig. 6A) increased NK cell [Ca2+]i by 14 ± 4 nM in 40 coverslips studied (P < 0.01). Twenty-seven of these 40 coverslips were unresponsive (see above), and the 13 responsive coverslips (33%), considered separately, elevated [Ca2+]i 48 ± 9 nM above basal levels (Fig. 6C). Cells on responsive coverslips exhibited [Ca2+] elevations which peaked at ∼10 s and returned to baseline within 40 s while flow was maintained (Fig. 6B).

Fig. 6.

Graded increases in shear stress induce graded, transient increases in cystosolic Ca2+ concentration ([Ca2+]i) in NK cells (AD) but not in PKD cyst cells (E and F), as measured by fura 2 fluorescence excitation ratio. A and B: [Ca2+]i increase in all the NK coverslips studied during initiation of superfusion with low calculated shear stresses of 0.75 or 2.3 dyne/cm2 (A) and high shear stress of 10 or 35 dyne/cm2 (B) in the presence (filled symbols) or absence of added perfusate Ca2+ (open symbols). C and D: [Ca2+]i increase in the responsive subset of NK coverslips during initiation of superfusion with low calculated shear stresses of 0.75 or 2.3 dyne/cm2 (C) and with high shear stress of 10 or 35 dyne/cm2 (D). E and F: [Ca2+]i increase in PKD coverslips during initiation of superfusion with low calculated shear stresses of 0.75 or 2.3 dyne/cm2 (E) and high shear stress of 10 or 35 dyne/cm2 (F). Parentheses indicate number of coverslips. G: time sequence of fluorescence ratio images of a representative NK coverslip subjected to 10 dyne/cm2 shear stress, as in C. Magnification ×20. Pseudocolor scale is at left in all panels.

Flow-induced [Ca2+]i increase in NK cells was abolished in nominally Ca2+-free medium at both 0.75 and 2.3 dyn/cm2 (Fig. 6A, open symbols; P < 0.01 compared with Ca2+-containing medium, Fisher's exact test applied to all coverslips studied). The flow-sensitive and flow-insensitive states of NK cells on responsive and unresponsive coverslips could not be explained by differences in resting, basal [Ca2+]i since, as shown in Supplemental Table 1, basal NK cell [Ca2+]i did not differ between responsive and unresponsive coverslips at any tested shear stress. In 2 of 10 coverslips of NK cells perfused at 1.6 dyne/cm2 at 37°C, [Ca2+]i increased by 53 and 60 nM, but cells on the other 8 coverslips were unresponsive (not shown). Thus at these low levels of shear stress, neither the magnitude nor rate of flow-induced [Ca2+] increase, nor the proportion of responsive coverslips, was higher at 37°C than at 20°C.

Shear stress values of 20 dyne/cm2 or more have been estimated in rat collecting ducts during maximal diuresis (summarized in Ref. 3). As shown in Fig. 6B (•), increasing shear stress to 10 dyne/cm2 increased [Ca2+]i in NK cells by 160 ± 53 nM (n = 16 coverslips, P < 0.01) within 20 s of initiation of flow. After 60 s the falling [Ca2+]i level remained 45 ± 17 nM above the initial basal level. Among these 16 coverslips, 13 were responsive to flow, and 3 were unresponsive. The responsive subset increased [Ca2+]i by 207 ± 65 nM and after 60 s maintained a plateau [Ca2+]i of 63 ± 22 nM above baseline values (Fig. 6D). An abrupt increase in shear stress to the high level of 35 dyne/cm2 increased [Ca2+]i by 200 ± 43 nM (n = 19 coverslips, P < 0.01). The responsive subset of 15 coverslips exhibited a flow-induced [Ca2+]i of 252 ± 43 nM within 10 s, and [Ca2+]i remained 64 ± 12 nM above the basal level after 60 s of perfusion (Fig. 6D). At both these higher shear levels of 10 and 35 dyn/cm2, 80% of tested NK cell coverslips exhibited flow-induced [Ca2+]i elevations. High shear stress-induced elevations of [Ca2+]i were similarly abolished in the nominal absence of extracellular Ca2+ (Fig. 6B, □; P < 10−4 compared with Ca2+-containing medium, Fisher's exact test applied to all coverslips studied). The NK cell [Ca2+]i response to flow at 20°C exhibited a refractory period of >30 min following cessation of flow at all tested shear stress values.

Tests of flow sensitivity at 37°C shortened both the time-to-peak values of [Ca2+]i and the refractory period without an increase in peak magnitude. Ten of 11 coverslips of NK cells exposed to 10 dyne/cm2 shear stress at 37°C exhibited flow-stimulated increases in [Ca2+]i to 107 ± 6 nM over baseline, peaking between 5 and 10 s in 9 cells, and by 20 s in 1 cell. Rechallenge of these 10 responsive coverslips after 30 min of stasis elicited identical elevations in [Ca2+]i in response to the same shear stress. Similar magnitudes and kinetics of [Ca2+]i increase were elicited in eight additional coverslips by exposure to 24 dyne/cm2 shear stress at 37°C. Thus increasing the temperature from 20 to 37°C increased the rate of rise in flow-induced [Ca2+]i and reduced the refractory period of the Ca2+ signal to <30 min but did not increase signal magnitude (not shown).

PKD cyst cells exhibited basal (static) [Ca2+]i of 135 ± 4 nM (n = 82), not different from the basal [Ca2+]i level in NK cells (141 ± 4 nM, n = 141). Imposition of shear stress at 0.75 or 2.3 dyne/cm2 produced no significant elevation of [Ca2+]i in PKD cyst cells, with respective values of Δ[Ca2+]i after 10-s perfusion of 3 ± 2 (n = 25) and 2 ± 2 nM (n = 18) (Fig. 6E). These minimal increments differed significantly from the peak flow-induced [Ca2+] elevations in NK cells subjected to the same shear stress (P < 10−4, Fisher's exact test applied to all coverslips studied). At the higher shear stress levels of 10 and 35 dyne/cm2, PKD cyst cell [Ca2+]i increased −1 ± 1 nM (n = 14) or 4 ± 2 nM (n = 25), respectively, (Fig. 6F), and again differed from flow-induced [Ca2+]i elevations in NK cells at the same shear stress values (P < 10−6, Fisher's exact test applied to all coverslips studied). Thus [Ca2+]i signaling in PKD cyst cells was uniformly unresponsive to flow across a wide range of shear stress in the conditions tested.

Flow-induced Ca2+ signaling in NK cells requires both Ca2+ influx and Ca2+ release from ryanodine-sensitive Ca2+ stores.

We characterized the inhibitor pharmacology of flow-induced Ca2+ signaling at the high shear stress of 35 dyne/cm2. Praetorius and Spring (42) showed that bending of the MDCK cell primary cilium with a micropipette led to [Ca2+]i elevation, reflecting both Ca2+ entry and release from inositol-1,4,5-triphosphate (IP3)-sensitive stores. However, Nauli et al. (30) found that flow-induced, cilium-dependent elevation of [Ca2+]i in embryonic mouse collecting duct epithelial cells was insensitive to inhibitors of phospholipase C and IP3 receptors.

In NK cells, the phospholipase C inhibitor U73122 (10 μM) was without effect on flow-induced elevation of [Ca2+]i (Fig. 7A), with a peak flow-induced [Ca2+]i increase of 132 ± 20 nM (n = 5) in U73122-treated cells and 146 ± 22 nM (n = 6) in untreated cells. In contrast, 30 μM ryanodine nearly abolished the flow response, supporting a role for ryanodine receptor-regulated Ca2+ stores (Fig. 7C) similar to that proposed for embryonic mouse collecting duct epithelial cells (30). The flow-induced elevation of [Ca2+]i (224 ± 34 nM, n = 5) was also inhibited nearly completely (to 12 ± 8 nM, n = 8) by 20 μM 2-aminophenylborate (2-APB; Fig. 7B). Since 2-APB inhibits not only IP3 receptors but also store-operated Ca2+ entry channels, we examined additional inhibitors of Ca2+ entry. The NK cell peak response to initiation of flow (236 ± 18 nM, n = 4) was reduced to 28 ± 24 nM (n = 9) by 10-min preincubation with 3 μM GsMTx-IV (Fig. 7D), a potent blocker of mechanosensitive channels isolated from tarantula venom (57). In contrast, the nonspecific cation channel inhibitor SKF-96365 (50 μM before and during flow) attenuated and delayed both activation and inactivation of the flow-induced Ca2+ signal (Fig. 7E). After a 30-s lagtime following onset of flow, an increase in [Ca2+]i in a few cells gradually propagated throughout the NK population, increasing after ∼150 s to a modest peak of 67 ± 14 nM by ∼150 s, with a greatly prolonged decay time (n = 6). In contrast, the rapid increase in [Ca2+]i in untreated NK cells (186 ± 25 nM, n = 4) was almost completely reversed by the time [Ca2+]i elevation was evident in cells exposed to SKF-96365.

Fig. 7.

Pharmacology of Ca2+ signaling in NK cells exposed to calculated shear stress of 35 dyne/cm2. A: fura 2-loaded cells were preincubated for 30 min in the absence (control, ○) or presence of 10 μM U73122, then subjected to flow in the continued absence or presence of drug. BE: Flow-sensitive Ca2+ signaling was similarly tested in the absence or presence of 20 μM 2-aminophenylborate (2-APB; B), 30 μM ryanodine (C), or 50 μM SKF96365 (E). D: cells were also preincubated in the absence (control) or presence of 3 μM Grammastola spatulata toxin IV (GsMTx) for 30 min, then subjected to fluid shear stress in the absence of the toxin. Numbers in parentheses indicate “responsive coverslips” (80% of total studied) in A and E and control coverslips (○) and drug-treated coverslips studied (•) in BD.

Figure 8 shows that PKD cyst cells sustained intact or partial activities of other Ca2+ signaling pathways, despite their lack of flow-responsive Ca2+ signaling. The peak response to 10 μM thapsigargin was smaller in PKD cyst cells (148 ± 29 nM) than in NK cells (648 ± 245 nM, n = 8, P < 0.01) (Fig. 8A), suggesting that releaseable ER Ca2+ stores of PKD cyst cells are reduced. “Store-operated” or “capacitative Ca2+ entry” (CCE) was also evaluated in the two cell types (Fig. 8B). Cells were pretreated for 10 min with 10 μM thapsigargin in a nominally Ca2+-free bath to deplete intracellular Ca2+ stores, reducing [Ca2+]i in NK and PKD cells (n = 11) to the similar levels of 111 ± 11 nM and 90 ± 3 nM, respectively (P > 0.05). Superfusion of this Ca2+-free bath at 0.75 dyne/cm2 for an additional 2 min (during which [Ca2+]i was unchanged; see Fig. 6A) was followed at t = 0 by addition of 10 mM CaCl2 to the superfusate (Fig. 8B). The peak CCE response of PKD cyst cells was reduced (64 ± 9 nM, n = 11) compared with the larger peak in NK cells (180 ± 23 nM, n = 11; P < 0.01). The peak [Ca2+]i values in PKD cyst cells exposed (under no-flow conditions) to 1 μM angiotensin II (Fig. 8C) and to 1 μM arginine vasopressin (Fig. 8D) were equivalent in magnitude to those of NK cells but exhibited slightly slower activation and substantially slower decay rates. Thus the nearly complete abrogation of flow sensitivity in PKD cyst cells did not reflect a global loss of Ca2+ signaling responses.

Fig. 8.

Comparison of other Ca2+ signaling pathways in fura 2-loaded NK (○) and PKD cyst cells (•). A: cells were exposed at t = 0 to 10 μM thapsigargin (TG), gently added in “no-flow conditions” from a 100-fold concentrated stock. Vehicle alone was without effect (not shown). B: after a 10-min preincubation with 10 μM TG in the nominal absence of Ca2+, cells were exposed to 10 mM extracellular Ca2+ to elicit capacitative Ca2+ entry (CCE). C and D: cells were exposed at t = 0 to 10 μM ANG II (C) or 10 μM arginine vasopressin (AVP; D). Numbers in parentheses are total number of coverslips studied.

CD16.7-PKD1(115–226) localizes to cilia of both NK and PKD cyst cells.

Heterologous expression of the tripartite fusion protein CD16.7-PKD1(115–226) increases endogenous Ca2+-permeable cation channel activity in X. laevis oocytes (64) and in EcR-293 cells (65). CD16.7-PKD1(115–226) contains a coiled-coil domain that can bind Ca2+-permeable cation channel PC2. Therefore, we tested the hypothesis that heterologous expression of CD16.7-PKD1(115–226) in NK and in PKD cyst cells might through this mechanism modulate flow-sensitive Ca2+ signaling in either cell type. Five days after transfection, unfixed cells were stained for cell surface expression of the CD16 epitope, then fixed. After microwave denaturation of bound anti-CD16 mAb 3G8, coverslips were stained for N-acetylated α-tubulin and studied by confocal immunofluorescence microscopy. CD16.7PKD1-(115–226) was detected not only on the surface membrane but throughout the cilium in all nine transfected NK cells examined (an example is shown in Fig. 9, AF) and was similarly present throughout the cilium of all six transfected PKD cyst cells examined (Fig. 9, MR). We tested the dependence of this localization on the integrity of the PC1 COOH-terminal tail coiled-coil domain, which is required for interaction with PC2 (63, 46). The PC1 coiled-coil domain mutant CD16.7PKD1-(115–226)L152P (64) was similarly expressed at the cell surface and throughout the cilium in 6 of 6 transfected NK cells examined (Fig. 9, GL) and in 16 of 16 transfected PKD cyst cells examined (Fig. 9, SX). Thus CD16.7PKD1(115–226) can accumulate not only in plasma membrane but also throughout the primary cilia of both NK and PKD cyst cells. Neither general cell surface localization nor ciliary localization requires integrity of the PC1 COOH-terminal coiled-coil domain.

Fig. 9.

A: COOH-terminal cytoplasmic tail fragment of PC1 localizes to cilia. Transiently transfected CD16.7-PKD1(115–226) is expressed in monocilia of NK cells (AF) and in monocilia of transiently transfected PKD cyst cells (MR). Transiently transfected CD16.7-PKD1(115–226)L152P is similarly expressed in monocilia of both NK cells (GL) and PKD cyst cells (SX). Unfixed cells (AD, GJ, MP) were stained for surface expression of CD16 with mAb 3G8 (red), then fixed, microwave-denatured as described in methods, and stained for N-acetylated α-tubulin (green). Cells in SX were fixed before staining with monoclonal antibody (mAb) 3G8. Large panels show confocal immunofluorescence x-y images of ciliary planes, and small panels immediately below show x-z images of the cells immediately above. Arrows mark cilia. Insets in A, G, and M: infraciliary x-y plane images of transfected cells (aligned with the higher x-y plane ciliary image of the same cell within each panel) which demonstrate generalized CD16 surface staining not restricted to cilia. Scale bars = 10 μm in x-y images; x-z images are magnified in the z plane.

Expression of CD16.7-PKD1(115–226) in NK cells shortens the duration of flow-induced [Ca2+]i increase.

Five to six days after cotransfection of NK or PKD cyst cells with CD16.7-PKD1(115–226), the consequences to flow-induced Ca2+ signaling were assessed. Transfected and nontransfected cells on single coverslips were identified by the respective presence or absence of fluorescence from cotransfected dsRed (Fig. 10E, top left) which cosegregated with cell surface CD16 expression (not shown). Expression of DsRed alone in NK cells changed neither resting [Ca2+]i (170 ± 6 nM in 21 single transfected cells vs. 186 ± 7 nM in 141 untransfected cells), peak flow-induced [Ca2+]i increase (212 ± 36 nM in 21 DsRed cells vs. 140 ± 12 nM in 141 untransfected cells), nor the rate of postpeak decline in [Ca2+]i (Fig. 10C).

Fig. 10.

A: COOH-terminal fragment of PC1 modulates flow-induced Ca2+ signaling. Transient expression of CD16.7-PKD1(115–226), but not of the mutant CD16.7-PKD1(115–226)L152P, modulates the decay rate of the flow-induced increase in [Ca2+]i elicited by imposition of 35 dyn/cm2 shear stress. A, Time course of [Ca2+]i response in CD16.7-PKD1(115–226)-transfected and untransfected NK cells on 8 coverslips. Transfected cells were identified by dsRed expression. B: time course of [Ca2+]i response in CD16.7-PKD1(115–226)L152P-transfected and untransfected NK cells on 5 coverslips. Transfected cells were identified by dsRed expression. C: time course of [Ca2+]i response in dsRed-transfected and untransfected NK cells on 3 coverslips. D: [Ca2+]i response of CD16.7-PKD1(115–226)-transfected and untransfected PKD cyst cells on 7 coverslips. Numbers in parentheses are total numbers of single cells evaluated. E: fura 2 fluorescence ratio image time course of a representative coverslip of NK cells cotransfected with dsRed and CD16.7-PKD1(115–226). dsRed-expressing cells are outlined in white. Magnification ×20. Pseudocolor scale is at left in all panels.

Resting [Ca2+]i in CD16.7-PKD1(115–226)-transfected (164 ± 9 nM, n = 41) and untransfected NK cells (167 ± 3 nM, n = 219) was indistinguishable, as in EcR-293 cells expressing this construct (65). The flow-induced peak [Ca2+]i increase in CD16.7-PKD1(115–226) expressing NK cells (124 ± 39 nM) also did not differ from that of untransfected cells (187 ± 21 nM; Fig. 10A). However, 25 s after initiation of flow, [Ca2+]i in NK cells expressing CD16.7-PKD1(115–226) had fallen to lower values (66 ± 13 nM) than in untransfected cells (116 ± 11 nM, P < 0.05). In contrast, expression in NK cells of the coiled-coil domain mutant CD16.7-PKD1(115–226)L152P changed neither the peak flow-induced [Ca2+]i increase (237 ± 76 nM vs. 179 ± 17 nM in untransfected cells) nor its rate of postpeak decrease (Fig. 10B). Expression of CD16.7-PKD1(115–226) failed to rescue flow-sensitive [Ca2+]i signaling in PKD cyst cells (Fig. 10D). Thus heterologous expression of the terminal 112 aa of the PC1 COOH-terminal cytoplasmic tail accelerates postpeak decay of flow-induced elevation in [Ca2+]i in NK cells by a mechanism that requires integrity of the PC1 coiled-coil domain.


In the present study, we have shown that confluent primary cultures of normal human renal cortical tubular (NK) cells elevate [Ca2+]i in response to fluid flow. In contrast, confluent primary cultures of human ADPKD cyst epithelial (PKD) cells harboring a novel heterozygous in-frame single codon deletion in the PC1 gene completely lack this response. The defect is not generalized, since PKD cyst cells retain near-normal Ca2+ signaling induced by angiotensin II and by vasopressin, as well as reduced CCE and a reduced thapsigargin response. The flow-induced [Ca2+]i elevation in NK cells requires extracellular Ca2+ and release from ryanodine-sensitive intracellular stores. The NK and PKD cyst cells studied here express equivalent whole-cell levels of PC1 and PC2, and both develop monocilia on achievement of confluency. However, cilia of confluent PKD cyst cells lack detectable PC1, whereas PC1 and PC2 are both present in cilia of confluent NK cells. PC2 is expressed in cilia of only 30% of confluent PKD cyst cells. The COOH-terminal PC1 fusion protein CD16.7-PKD1(115–226) localizes to the cilium of both NK and PKD cyst cells by a mechanism independent of the PC1 coiled-coil domain. However, accelerated decay of the flow-induced Ca2+ signal in NK cells associated with overexpression of CD16.7-PKD1(115–226) requires integrity of that coiled-coil domain. These results confirm and extend the recent study by Nauli et al. (31) published after completion of these experiments.

A novel in-frame single codon deletion in PC1 is associated with loss of ciliary PC1 and (in some cells) of ciliary PC2 without reduction in total PC1 or PC2.

The novel heterozygous PKD1 gene mutation ΔL2433 reported here is likely a germline mutation, since the epithelial cells from which genomic DNA was isolated were harvested from multiple renal cysts of a single ADPKD donor. The failure to detect additional mutation(s) in the PKD1 gene is consistent with the second-hit hypothesis of cystogenesis (47). The deleted residue L2433 resides in the REJ domain of PC1. Overexpression in MDCK cells of engineered PC1 missense mutants in the REJ domain has been associated with loss of AP-1 activation and of tubulogenesis induced by wild-type PC1 overexpression (45).

The heterozygous PC1 ΔL2433 mutant allele is associated with normal cellular levels of PC1 and PC2 (Fig. 2). Expression of PC1 and PC2 polypeptides in the cell body of all PKD cyst cells as noted in the present study resembles human ADPKD kidney and Pkd1(+/−) mouse kidney (66, 72) but differs from the mosaic (all-or-none) PC2 expression observed in renal cysts of the cy/+ rat, and from the mosaic or complete loss of PC2 expression in the cy/cy mouse (34) and in the Pkd2(+/−) and Pkd1(+/−)/Pkd2(+/−) mice (71). These findings suggest that the PC1 ΔL2433 mutant polypeptide may be present in PKD cyst cells.

PC1 antigen has been detected at tight junctions, adherens junctions, desmosomes, focal adhesions, intracellular cytoplasmic vesicles (summarized in Refs. 56 and 50), and nuclei (24), as well as in plasma membrane and cilia. These many PC1 localizations may reflect examination of varied cell types and differentiation states, as well as posttranslational processing or covalent modification of PC1, and (at least in rodents) alternative splicing of PC1 transcripts. PC2 is predominantly localized in the ER (2). The presence of the heterozygous PC1 ΔL2433 mutation is associated with complete loss of ciliary PC1 expression and by loss of ciliary PC2 expression in 70% of cells (Figs. 4 and 5). Our results thus show that ciliary localization of PC2 does not require immunocytochemically detectable levels of ciliary PC1, consistent with recent findings on ciliary targeting of PC2 (10). The continued ciliary expression of PC2 in 30% of cells with the novel PC1 ΔL2433 mutation contrasts with the reported absence of ciliary PC2 in all SV40 large T-transformed, DBA-positive human 9–12 cyst cells and in (genetically uncharacterized) primary human ADPKD cyst epithelial cells (31). Ciliary localization of PC1 and PC2 was not described in a study of tsSV40LgT-immortalized human ADPKD cells with the E1537X germline mutation, which express normal PC2 levels with very low levels of PC1 (56).

Ciliary length was slightly shorter in PKD cyst cells than in NK cells of equivalent confluency, and cyst cilia lacking PC2 tended not to be orthogonal in orientation. This ciliary phenotype of human cyst epithelial cells in primary culture is intermediate in severity between the severely shortened, dysmorphic cilia of cells cultured from the orpk mouse with a hypomorphic polaris mutation (40) and the wild-type ciliary length in cells from the pkd1del34/del34 mouse (30). Reported wild-type ciliary length has varied widely: between 8 (42) and 2–4 μm (75) in confluent MDCK cells, between 2 and 4 μm in immortalized juvenile mouse collecting duct cells (75) and 12 μm in immortalized embryonic mouse collecting duct cells, and between 3 and 5 μm in wild-type mouse and rat renal cortical tubules in situ (39). Thus the ciliary length of NK cells in the current study falls within the wide range of previously reported ciliary lengths and matches the in situ ciliary length in rodent tubules.

The flow-sensitive Ca2+ signaling response of NK cells is absent from PKD cyst cells which retain other Ca2+ responses.

Flow-sensitive Ca2+-signaling is a property of isolated, perfused rabbit and mouse collecting ducts (22, 21) and has also been reported in isolated, perfused mouse medullary thick ascending limb (19). In the orpk mouse model of recessive PKD, flow-evoked Ca2+ signaling in the cortical collecting duct (CCD) remained normal during the first postnatal week but was slightly reduced compared with wild-type CCD at age 2 wk (21), a difference replicated in CCD cells grown in primary culture (55).

Our observation of the complete absence of flow-sensitive Ca2+ signaling in human primary cultures of PKD cyst cells resembles those previously reported in immortalized CCD cells from the pkd1del34/del34 mouse and in immortalized and primary human cyst cells (30, 31). Our results extend this earlier work in cyst cells lacking PC1 polypeptide by showing that the lack of flow-induced intracellular Ca2+ signaling in human cyst cells occurs also in the presence of normal total cell polypeptide levels of PC1 and PC2, including the 30% of cyst cells which retain PC2 expression in the cilium. The results further show that human cyst cells fail to elevate [Ca2+]i in response to flow at all tested levels of shear stress, and at both 20 and 37°C. These results comparing adult NK cells with PKD cyst cells isolated from end-stage ADPKD kidneys do not allow conclusions about the centrality of defective ciliary flow sensing to early cystogenesis or cyst growth in ADPKD.

The current results also present an initial pharmacological characterization of other Ca2+ signaling responses of human PKD cyst cells, indicating that the absence of flow-induced elevation in [Ca2+]i in PKD cyst cells does not represent a global Ca2+ signaling defect. Thus thapsigargin-induced Ca2+ release and CCE are both preserved in PKD cyst cells, although decreased in magnitude compared with NK cells (Fig. 8). These diminished signals resemble those observed in aortic vascular smooth muscle cells of pkd2(+/−) mice (48). Unlike the reduced thapsigargin and CCE responses, PKD cyst cell CaMath elevations in response to angiotensin II and to AVP are of normal magnitude, although slightly delayed in onset. The components of the local renin-angiotensin system are overexpressed in human ADPKD (23), with possible consequences to flow-induced regulation of apical exocytic insertion of AT1a receptors in proximal tubular cells (16). The prolonged rate of Ca2+ signal decay in PKD cyst cells treated with angiotensin II or vasopressin recalls that produced in ATP-stimulated M1 CCD cells by overexpression of a fusion protein containing the PC1 COOH-terminal cytoplasmic tail (69), attributed by those authors to prolongation of Ca2+ entry. This prolonged rate of Ca2+ signal decay may represent the converse of the accelerated Ca2+ signal decay following ATP stimulation of MDCK cells overexpressing full-length PC1, attributed to enhanced ER Ca2+ reuptake with inhibition of CCE (13). CCE inhibition by either overexpression or loss of PC1 function in PKD cyst cells parallels the ability of both underexpression and overexpression of PC1 to cause PKD in mice (60).

Range and uniformity of flow sensitivity of NK cell flow-induced Ca2+ signaling.

Resting [Ca2+]i measured in the absence of flow at room temperature and at 37°C was indistinguishable in confluent, serum-replete NK and PKD cells and was equivalent to that reported in confluent embryonic CCD cells from wild-type and pkd1del34/del34 mice and in human normal and cyst cells by Nauli et al. (30, 31). Both sets of these resting [Ca2+]i values substantially exceed those reported by Yamaguchi et al. (73) in subconfluent primary human epithelial cells subjected to 48 h of progressively increasing degrees of serum starvation. In such low-serum conditions, favorable for testing effects of growth regulators, human cyst cell [Ca2+]i at 57 nM was significantly lower than the 77 nM measured in normal human cells (73). However, thapsigargin-induced depletion of CaMath stores in nominally Ca2+-free medium unmasked a similarly lower value of [Ca2+]i in our PKD cyst cells compared with NK cells.

The current results with human primary NK cells differ somewhat in flow sensitivity from those reported for immortalized mouse CCD cells, or for immortalized human cells with lectin-staining properties consistent with collecting duct origin and primary human renal cortical epithelial cells of unspecified segment of origin. Mouse CCD cells elevated [Ca2+]i in response to shear stress of 0.75 dyne/cm2 but failed to respond to the higher shear stress of 15 dyne/cm2 (30). Immortalized normal human renal cortical tubular epithelial cells (RCTE) responded optimally to shear stress of 1.2 dyne/cm2 (31), results which we have reproduced with near-uniform responsiveness of coverslips (Xu and Alper, unpublished observations). However, only 5 of 8 tested coverslips of primary kidney cortical tubular epithelial cells were reported to exhibit a flow response (31). NK cells from three individuals in the current study responded to shear stresses of 0.75 or 2.3 dyne/cm2 with modest elevations in [Ca2+]i in only 33% of coverslips. At 10 dyne/cm2 or at higher values, the robust [Ca2+]i responses observed in 80% of tested coverslips were of a magnitude comparable to those reported by Nauli et al. (31). Factors distinguishing the NK cells of flow-responsive and of flow-unresponsive coverslips remain unclear and must include mechanisms other than ciliary localization of PC1 and PC2. However, at none of the shear stress values tested did responsive and unresponsive NK cell coverslips differ in baseline [Ca2+]i.

The NK cell response to graded increases in shear stress with further increased elevation of [Ca2+]i, rather than by diminution or loss of responsiveness, corresponds better to the MDCK cell response to flow (42) than to the mouse and human cortical epithelial cell responses to flow reported by Nauli et al. (30, 31). The responses of human NK cells in the current study were not altered in magnitude or shear sensitivity by increasing temperature from 20 to 37°C. However, the higher temperature accelerated the time-to-peak [Ca2+]i and shortened the refractory period.

Thus NK cells in the current study were weakly responsive to low shear stress and exhibited more robust responses to shear values associated with diuresis (3, 68) or with shear stress values close to those present in the central axis of the initial S1 proximal tubule. These differences in flow sensitivity may reflect differences in ciliary length (8–12 μm vs. 4 μm in the current study), in conditions of initial tubule cell outgrowth or of subsequent primary cell culture, and likely reflect heterogeneity of the cultured population, and/or in nephron segment of origin. Nearly all NK cells in the current study were Lotus tetragonolobus agglutinin (LTA) positive, whereas only 10% of NK cells expressed Dolichos biflorus agglutinin (DBA). Although all NK cells expressed E-cadherin, strong staining characterized only 10%. Similarly, most ADPKD cyst cells in the current study were LTA positive, whereas none expressed DBA. Thus proximal tubular markers predominated among both the NK cells and ADPKD cyst cells used in the present study.

Marker studies and cyst fluid composition have suggested possible proximal tubular origin of up to 30–44% of closed cysts in human ADPKD (reviewed in Refs. 61 and 70). Moreover, late-onset renal cysts in the Pkd1(+/−) mouse expressing LTA were twice as frequently observed as cysts expressing DBA, although most cysts expressed neither lectin marker (25). The laminar shear stress predicted for the central axis of a human initial S1 proximal tubule of 22 μm diameter is predicted to be 7.6 dyn/cm2 for a 125 ml/min glomerular filtration rate (with the oversimplifying assumptions of an inelastic tubule without brush border). For a vertical cilium with a length of 4.2 μm, the experienced shear stress under parabolic flow in this 22-μm-diameter tubule would be 4.7 dyne/cm2 at the ciliary tip. These values might plausibly increase twofold in the oligonephronia proposed in essential hypertension (15) or after uninephrectomy. Thus the shear stresses of 7–10 dyne/cm2 at which our NK cells exhibit a maximal peak elevation of [Ca2+]i are not far outside the physiological range. Different nephron segments of origin may thus explain most of the difference in flow sensitivity between NK cells and the cells expressing collecting duct markers studied by Nauli et al. (30, 31).

Pharmacological properties of flow-induced elevation of [Ca2+]i in NK cells.

The pharmacology of flow-induced Ca2+ signaling in human NK cells resembled mouse CCD and MDCK cells in some ways and differed in others. The flow response in all these cell types required both extracellular Ca2+ and Ca2+-induced Ca2+ release from internal stores. Although IP3-sensitive Ca2+ stores were implicated in MDCK cells (13) and in the perfused rabbit CCD (22), ryanodine-sensitive stores seemed to predominate in the flow response in human NK cells and in mouse embryonic CCD cells. However, ryanodine receptor inhibitors can reduce IP3-mediated Ca2+ signaling in colonic smooth muscle cells (27). The reported ability of PC2 to modulate IP3 receptor function by direct interaction (20) appeared not to contribute to NK cell flow sensitivity. Whereas 2-APB (10 μM, 45 min) was without effect in mouse embryonic kidney cells, the NK cell flow response in NK cells was completely inhibited (20 μM, 30 min). However, interpretation of this inhibition is complicated by the drug's dual inhibitory effects on IP3 receptor and TRP channels, and by its ability to activate TRPV1–3 channels (6). The novel NK cell response to SKF96365, with reduced peak [Ca2+]i and slowed kinetics of activation and decay, may represent weak agonist activity for Ca2+ store release (7) rather than atypical inhibition of Ca2+ entry.

Flow-activated Ca2+ entry clearly requires integrity of the PC1/PC2 complex, but the identity of the Ca2+ entry pathway in NK cells remains unknown. The density and diversity of ion channels in monocilia may be very high (49). Complete block of flow-induced Ca2+ entry in NK cells by 3 μM GsMTx-IV, an inhibitor of stretch-activated cation channels, may provide a path toward channel identification, but a less specific lipid bilayer intercalation effect remains possible (57). TRPC1, previously shown to bind to PC2 in vitro, has been recently implicated as a mechanosensitive channel in X. laevis oocytes (28). TRPV4 colocalizes with PC2 (59) and may bind and modulate its activity (17). Since 30% of PKD cyst cells retain normal ciliary localization of PC2, the presence of PC2 in cilia apparently does not suffice for normal flow-induced Ca2+ signaling in the absence of ciliary PC1. However, primary cultures of cells isolated from a mature cyst of end-stage ADPKD kidney may be dedifferentiated compared with NK cells in ways not specifically related to ADPKD.

The pharmacological properties of flow-induced Ca2+ signaling in normal human kidney tubular epithelial cells in primary culture were not reported by Nauli et al. (31). Nevertheless, the similarities between our NK cells and the wild-type mouse CCD cells studied by Nauli et al. (30) include a requirement for extracellular Ca2+ entry, sensitivity to ryanodine, the kinetics of Ca2+ signal onset and decline, and the time course for recovery from the post-flow refractory period at 37°C. Thus even between cells expressing markers suggesting different nephron segments of origin, some properties of flow-induced [Ca2+]i elevation are shared.

Overexpression of the PC1 COOH-terminal tail in NK cells alters the decay kinetics of the flow-induced Ca2+ signal.

CD16.7-PKD1(115–226) overexpressed in NK cells accelerated decay kinetics of the flow-induced Ca2+ signal, in contrast to prolongation of the ATP-induced Ca2+ signal in M1 cells by a similar fusion protein (69). A disease mutation disrupting the coiled-coil domain and blocking cation current activation in oocytes and EcR-293 cells prevented the accelerated decay of the flow-induced Ca2+ signal in NK cells. This result suggests that CD16.7-PKD1(115–226) interacts with an endogenous protein to modulate flow-induced Ca2+ signaling. The inhibitory effect of CD16.7-PKD1(115–226) might therefore reflect competition with mechanosensitive full-length PC1 for binding to or regulation of PC2 channel activity or signaling, as proposed by Low et al. (24) to explain cystic dilation of the zebrafish pronephric duct. However, the mechanism of mechanosensation by the ciliary apical PC1/PC2 complex exposed to fluid flow may differ from that by basolateral PC1 or the PC1/PC2 complex exposed to a neighboring cell or to matrix. Moreover, Ca2+ entry pathways induced by flow in NK cells may differ from those induced by ATP in M1 cells.


Flow sensing has long been thought to contribute to proximal tubular perfusion-absorption balance, to tubuloglomerular feedback (9), to CCD K+ secretion (71) and Na+ reabsorption (53), and to nitric oxide release by the thick ascending limb of Henle's loop (38) and the inner medullary collecting duct (3). The cilium, with its apparent concentration of receptors and signaling molecules, is an attractive candidate to integrate these signals controlling tubular epithelial cell differentiation and function, perhaps through [Ca2+]i -mediated regulation of B-raf (73), regulation of mTOR (54), or other pathways. However, the role of ciliary flow sensing in the prevention of cystogenesis in the normal state remains in question, as evidenced in orpk mice by failure of polaris transgene rescue to prevent cystic disease even with the normalization of ciliary structure and correction of left-right asymmetry (1). Thus comparative studies of flow sensitivity in human renal cells and PKD cyst epithelial cells should play an important and continuing role in defining the place of defective ciliary mechanosensation in the pathogenesis of dysregulated growth and secretion in human ADPKD.


This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants F32 DK-69049 to C. Xu, R01-DK-57662 to S. L. Alper, R01-DK-58816 to P. C. Harris, R01-DK-50141 to A. Wandinger-Ness, and a Polycystic Kidney Disease Foundation award to R. Bacallao.


We thank Boris E. Shmukler, David H. Vandorpe, and Vince Carone for helpful discussion, Oxana Ibraghimova-Beskrovnaya for antisera to PC1 (LRR) and to PC2, Wayne Lencer for anti-GM130, Carrie Phillips for pathology support, Genevieve Philips for analytic assistance, and Elsa Romero and Alan Stuart-Tilley for expert technical assistance.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract