Disruption of Polycystin Ciliary Localization and Channel Function by Autosomal Dominant Polycystic Kidney Disease–Causing Polycystin-1 Variants

Visual Abstract

Abstract

Key Points

• We developed assays to measure genetic variant effects on polycystin-1, the protein mutated in most autosomal dominant polycystic kidney disease.

• All tested pathogenic variants disrupted either polycystin-1 ciliary trafficking or channel function.

• Trafficking and channel function of some pathogenic variants was restored by low temperature culture to promote polycystin folding.

Background

Autosomal dominant polycystic kidney disease (ADPKD) is the leading monogenic cause of kidney failure and affects millions of people worldwide. Despite the prevalence of ADPKD, limited mechanistic understanding has hindered therapeutic development. Most ADPKD is caused by loss-of-function variants in polycystin-1 (PC1).

Methods

We developed assays that quantify the effect of nontruncating variants on PC1 ciliary localization, membrane trafficking, and polycystin channel function.

Results

We evaluated 29 nontruncating variants in PC1 and found that pathogenic variants disrupt two molecular phenotypes: (1) localization of PC1 at the primary cilium or (2) polycystin ion channel activity. Ciliary localization of a subset of polycystin variants was restored when cells were cultured at low temperature. A subset of variants with localization restored by low temperature formed functional channels.

Conclusions

This study demonstrated that disruptions in polycystin ciliary trafficking and channel function are common causes of ADPKD. Defects in ciliary trafficking and channel function can be rescued for a subset of pathogenic variants, establishing a foundation for polycystin-targeted therapies in ADPKD.

Introduction

Autosomal dominant polycystic kidney disease (ADPKD) is a common monogenic disease in humans and accounts for approximately 10% of global kidney failure.¹ Because the molecular mechanisms underlying ADPKD remain largely unknown, treatment remains limited. ADPKD is characterized by significant allelic heterogeneity, with over 1200 pathogenic or likely pathogenic variants identified.² Many pathogenic variants are restricted to individual families, even in large cohorts.^3,4 Most ADPKD cases (80%) result from variants in PKD1 encoding polycystin-1 (PC1), while remaining cases (15%) are caused by variants in PKD2 encoding polycystin-2 (PC2).^3,5,6 PC1 and PC2 assemble into a heteromeric transient receptor potential-like ion channel with 1:3 stoichiometry (hereafter called the polycystin complex).⁷

Many molecular functions of PC1 have been proposed in addition to ciliary ion channel function.^8–11 For instance, PC1 has been linked to adhesion G protein–coupled receptor (GPCR) activity,^12–14 mechanosensation of tubule flow,^15,16 signaling to mitochondria^17,18 and the nucleus¹⁵ through a cleaved C-terminus, and as a functional component of extracellular vesicles.¹⁹ In addition, PC2 has been associated with ion channel activity in the endoplasmic reticulum.²⁰ However, whether disruption of these molecular functions plays a significant role in human ADPKD remains unclear.

Several lines of evidence suggest that disrupting ciliary polycystin localization causes ADPKD. Mutations in enzymes necessary for PC1 folding, maturation, and glycosylation disrupt PC1 ciliary localization and cause cystic kidney and liver disease.^21–24 Several pathogenic missense variants in PC1 and PC2 disrupt polycystin localization to the primary cilium.^25,26 Mouse models of PC1 variants with defective ciliary localization or disrupted biogenesis cause PKD.^25,27–29 However, the mechanisms by which pathogenic PC1 variants that localize to the primary cilium cause ADPKD remain unclear.

Like ADPKD, cystic fibrosis is a life-threatening, monogenic disease caused by loss-of-function variants in a multispan membrane protein—the cystic fibrosis transmembrane conductance regulator (CFTR) ion channel. Pathogenic variants in CFTR either disrupt plasma membrane localization, CFTR channel activation, or both.^30–33 Recent targeted therapies rescue CFTR plasma membrane localization or channel function. These therapies have been transformative for the health and lifespan of cystic fibrosis patients.³²

Several lines of evidence suggest that PC1 and PC2 form an ion channel complex.^7,8,11,34 Here, we evaluate the hypothesis that analogously to CFTR, pathogenic variants in PC1 disrupt localization or ion channel function.

Methods

UK Biobank Analysis

UK Biobank participants with at least one International Classification of Diseases-10 code and exome data (415,848) were analyzed. The International Classification of Diseases-10 code Q61.2 (PKD, Adult) was used to identify polycystic kidney disease (PKD) cases. Genotypes were ascertained using population level exome original quality functionally equivalent data (final release) in the project variant call format file format. Variants were normalized (with “bcf tools norm”), variants for which at least 90% of the samples had a nonmissing genotype and at least ten reads (“F_PASS(DP>=10 & GT!=‛mis’)> 0.9”) were retained. Variant annotations were determined using Ensembl Variant Effect Predictor.³⁵ Frameshift and stop gained variants were labeled as truncating variants. For Supplemental Figure 1C, each participant was assigned an allele frequency based on the rarest missense variant they had in PKD1 or an allele frequency of 1 if they had no missense variant in PKD1. Evolutionary scale modeling 1b (ESM1b) scores for all possible missense mutations in PKD1 were downloaded in comma-separated values file format from the ESM Variants portal at https://huggingface.co/spaces/ntranoslab/esm_variants. The training of ESM1b is described in Rives et al.,³⁶ and the methods for predicting variant effect scores with the model are described in Brandes et al.³⁷ Work was conducted under UK Biobank Application 78378.

Variant Classification

We defined likely benign variants as those with allele frequency >2×10⁻⁵, reported homozygosity in humans—as loss of PC1 is embryonic lethal,³⁸ and classified as likely benign or benign in the Mayo ADPKD² database or ClinVar.³⁹ We defined likely pathogenic variants as those with allele frequencies <5×10⁻⁶, reports of pathogenicity in the literature, and classification of likely pathogenic or pathogenic in the Mayo ADPKD database² or ClinVar.³⁹ Variant of uncertain significance (VUS) were classified based on clinical genetic testing from a reference laboratory (Natera).

Patients

Patients with VUS in PKD1 provided consent for participation in the description of their histories and variants. Human subjects research was performed under Institutional Review Board 14-15601.

Molecular Biology

Kappa (κ) HA-PC1 was described previously.⁸ All missense mutations in PC1 were introduced using PCR amplification and Gibson assembly (New England Biolabs). PC1 variants were introduced into a plasmid (Addgene 104454) containing a hygromycin selection cassette. PC2 was ligated into the same expression plasmid as PC1 but with a puromycin selection cassette. Sequences were confirmed by whole plasmid sequencing (Primordium Labs).

Antibodies

The antibodies include rat anti-hemagglutinin (Roche, 11867423001), rabbit anti-hemagglutinin (Cell Signaling Technology, C29F4), mouse anti-ARL13b (Antibodies Inc, N295B/66), rat anti-PC1 C-terminus (Kerafast EMD303), and rabbit anti-α-Tubulin (Proteintech, 11224-I-AP).

Immunocytochemistry and Confocal Microscopy

Cells were fixed with 4% formaldehyde, permeabilized with 0.2% Triton X-100, and blocked by 2% FBS, 2% BSA, and 0.2% fish gelatin in PBS. Cells were labeled with the indicated antibody and secondary goat anti-rabbit, anti-rat, or anti-mouse fluorescently labeled IgG (Thermo Fisher) and Hoechst 33342 (Thermo Fisher). Confocal images were obtained using a Zeiss laser scanning microscope 800 laser scanning confocal microscope equipped with a 63×/1.4 oil immersion objective. Images were processed using ImageJ (National Institutes of Health).

Cell Culture, Quantification of Ciliary and Plasma Membrane PC1 and PC2, and Quantification of PC1 GPCR Proteolytic Site Cleavage

Surface localization of PC1/PC2 complexes was quantified with staining for the influenza hemagglutinin (HA) peptide in nonpermeabilized cells. Isogenic clones of a human embryonic kidney cell line (HEK-293) or inner medulary collecting duct cell line (IMCD3) with clustered regularly interspaced short palindromic repeats–ablated endogenous PKD1 expression (gift of F. Qian⁸) expressing PC2 under a doxycycline-inducible promoter were isolated. PC1 variants were introduced with the constructs described above using hygromycin selection. Polycystin complex expression was induced in HEK-293 cell lines with 2 μg/ml doxycycline in complete medium for 48 hours. For experiments of the effect of temperature on surface localization, the last 24 hours of induction was performed at the indicated temperature (37°C or 32°C). Cells were detached with Versene (Thermo Fisher) and incubated on ice in blocking buffer (2% FBS, 2% BSA, and 0.2% fish gelatin in PBS) for 10 minutes. After centrifugation, cell pellets were resuspended in blocking buffer with rabbit anti-HA diluted at 1:100. After incubation on ice for 20 minutes, cells were washed twice with cold PBS and incubated for 20 minutes with 488-conjugated goat anti-rabbit antibodies. Cells were washed 2× with ice-cold PBS. Sytox Blue (Thermo Fisher) was added at 1:1000 5 minutes before sample acquisition on an Attune Nxt flow cytometer, and analysis was performed using FlowJo (v10.8.2). PC1 surface localization was quantified as the fraction of live cells with PC1 surface localization greater than an uninduced control. PC1 surface localization was normalized to wild-type (WT) PC1 surface localization.

To quantify PC1 expression on the ciliary membrane, IMCD3 cell lines were incubated in Opti-minimal essential medium (Thermo Fisher) for 24 hours to initiate ciliogenesis. Cells were further incubated for 48 hours with 0.2 μg/ml doxycycline in Opti-minimal essential medium. Adherent live cells were incubated with L-15 medium containing rabbit anti-HA antibody (1:100) for 25 minutes at room temperature to avoid internalization of antibodies. Cells were washed twice with L-15 medium, fixed with 4% paraformaldehyde, and permeabilized with 0.2% Triton X-100. Surface HA fluorescence was quantified from Z-projections in regions of interest defined by ARL13b staining (cilia marker) using ImageJ.

Polycystin expression and GPCR proteolytic site (GPS) cleavage was assayed by Western blot in IMCD3 cells treated as described above for ciliary quantification. Lysates were generated in radioimmunoprecipitation assay buffer (R0278, Sigma-Aldrich) supplemented with protease inhibitor cocktail (11873580001, Sigma-Aldrich). The cleared supernatant was resolved on a NuPAGE 3%–8% Tris-Acetate gel (EA03755BOX, Fisher Scientific), and wet transferred to a polyvinylidene fluoride membrane. Image acquisition was performed using the ChemiDoc MP Imaging System (Bio-Rad), and signal intensity was quantified using ImageJ.

Electrophysiology

HEK-293 cells stably expressing doxycycline-inducible PC1 and PC2 F604P were generated as described above. To induce polycystin expression, we incubated cells in DMEM/F12 medium supplemented with 10% FBS and 1 μg/ml doxycycline at 37°C for 36-48 hours. To measure the effect of low temperature culture on channel function, we initially incubated cells at 37°C for 24 hours with 1 μg/ml doxycycline, followed by a 24-hour incubation at 32°C in the continued presence of doxycycline.

We performed whole-cell configuration patch clamp using Multiclamp 200B (Axon Instruments) and Digidata 1324A (Axon Instruments). Recording was performed using pClamp software (Axon Instruments). Whole-cell configuration patch-clamp dataset was filtered at 1 kHz and sampled at 10 kHz. The voltage step pulse ranged from −100 mV to +180 mV in 20 mV increments during 150 ms, followed by −80 mV tail pulse during 50 ms. The holding potential was −60 mV for the recordings. We pulled the glass pipettes using P-100 micropipette puller (Sutter instrument). The resistance of pipettes for the whole-cell recordings was within 6-8 MΩ. The tip of the pipette was further polished using a Narishige MF-830 microforge. For patch-clamp experiments, we used an extracellular solution consisting of (mM): 145 Na-gluconate, 5 KCl, 2 CaCl₂, 5 MgCl₂, 10 HEPES, and adjusted to pH 7.4 using NaOH. We used an intracellular solution consisting of (mM): 90 sodium methanesulfonate, 10 NaCl, 5 MgCl₂, 10 HEPES, 5 EGTA, 100 nM free calcium adjusted by CaCl₂, and adjusted to pH 7.4 using NaOH. We used Clampfit10.6 (Axon Instruments/Molecular devices), Origin8 (Originlab), and Prism10.0 (GraphPad) to analyze data from the whole-cell configuration patch clamp. Data are shown as mean±SEM, and n represents the number of tested cells.

Results

Contribution of Missense Variants to ADPKD

It is challenging to assess the pathogenicity of missense variants in PKD1, potentially limiting prior estimates of the fraction of ADPKD explained by missense variants. We examined the incidence of truncating (frameshift and stop-gain) and missense variants in PKD1 in patients with and without a diagnosis of PKD in the UK Biobank. The UK Biobank cohort includes over 400,000 exome-sequenced individuals with linked phenotype data, including PKD diagnostic codes. We observed that 19.6% of individuals with a diagnostic code for PKD (Q61.2) carry a truncating variant in PKD1, whereas 0.02% of controls have such mutations (Supplemental Figure 1A). This difference between the prevalence of truncating variants in cases and controls indicates that approximately 19.6% (19.6%–0.02%) of PKD cases in UK Biobank are attributable to truncating variants.

In total, 15.4% of patients with PKD carry a rare (allele frequency <2×10⁻⁵) PKD1 missense variant, compared with 1.8% of controls who carry a rare missense variant (Supplemental Figure 1A). Rare allele frequency alone is not sufficient to establish variant pathogenicity.⁴⁰ To evaluate the pathogenicity of PKD1 rare missense variants identified in PKD cases, we scored these variants using ESM1b, a deep protein language model that discriminates between benign and pathogenic variants.³⁷ Rare PKD1 missense variants identified in PKD cases without truncating variants had significantly more negative ESM1b scores than missense variants identified in controls (Supplemental Figure 1B), supporting the hypothesis that many rare PKD1 missense variants identified in PKD cases are pathogenic. The difference between the rate of rare missense variants in cases and controls suggests that approximately 13.6% (15.4%–1.8%) of PKD cases in the UK Biobank are attributable to missense variants. This difference between the carrier frequency of rare missense variants in ADPKD cases and controls is robust to the choice of allele frequency (Supplemental Figure 1C). These data suggest that approximately 41% ($\frac{13.6\%}{13.6\%+19.6\%}$) of ADPKD-PKD1 is attributable to missense variants, slightly higher than historical data from clinical cohorts.^3,41,42

The Impact of Pathogenic PC1 Variants on Cell Surface Localization

Besides lack of functional assays to assess pathogenicity of missense mutations, another major hurdle in ADPKD research is the small fraction of cases caused by each variant (allelic heterogeneity). We therefore combined data from ClinVar,³⁹ the Mayo ADPKD variant database,² allele frequency, and published literature to identify probable benign and pathogenic variants throughout PC1 for further study (see Methods). We then generated PC1 expression constructs with pathogenic and benign variants (Supplemental Table 1).

We tested whether pathogenic missense variants in PC1 impair cell surface localization using our previously published PC1 expression construct.⁸ This construct includes a modified PC1 in which the endogenous signal peptide was replaced with the 12 amino acid IgG leader sequence followed by an HA tag.⁸ When coexpressed with PC2, this construct enabled quantification of PC1 at the ciliary membrane of kidney-derived IMCD3 cells by staining live cells with an anti-HA antibody.⁸ In nonciliated HEK-293 cells, these expression constructs enabled recording of polycystin channels at the plasma membrane.⁸

We generated stable HEK-293 cells coexpressing PC1 missense variants together with PC2 and quantified plasma membrane localization of the polycystin complex using the HA-tag on PC1 by flow cytometry. As PC1 and PC2 are expressed under the control of a doxycycline-inducible promoter, cells cultured in the absence of doxycycline served as controls. Flow cytometry detected robust cell surface localization of WT PC1 (Figure 1B). Similar to WT PC1, the pathogenic PC1 variant L4139P⁴¹ also localized to the cell surface. By contrast, the pathogenic variant W139C⁴¹ exhibited markedly reduced cell surface levels. Thus, pathogenic PC1 variants can have markedly different effects on polycystin membrane localization.

Figure 1.

Many pathogenic ADPKD variants disrupt polycystin ciliary localization. (A) Human benign and pathogenic variants in PC1 were assayed for their effect on ciliary localization or channel function to test the hypothesis that these molecular phenotypes are critical for preventing cystic kidney disease. (B) Cell surface localization of WT PC1 or PC1 with pathogenic variants in nonciliated HEK-293 cells measured by flow cytometry. PC1 expression is doxycycline-inducible; cells cultured without doxycycline serve as a control. (C) Quantification of cell surface localization of PC1 by flow cytometry for benign and pathogenic variants in HEK-293 cells. Each point represents an independent experiment; surface localization is normalized to WT PC1 surface localization from the same experiment. Significance was determined with a one-way ANOVA with Dunnett multiple comparison test comparing pathogenic variants to the benign variant with lowest surface localization (R80W). (D) Cell surface localization of WT PC1 or PC1 with pathogenic variants in ciliated IMCD3 cells visualized by immunofluorescence. ARL13B is stained to identify primary cilia. (E) PC1 variant expression on the ciliary membrane of IMCD3 cells was quantified by immunofluorescence. Median ciliary expression is indicated with horizontal bars in the violin plots. Significance was determined with the Kruskal–Wallis test with Dunn multiple comparison test comparing pathogenic variants to the benign variant with lowest ciliary expression (R80W). Each point represents a cilium (WT n=785, F3066 n=349, R4276W n=553, I3167F n=1366, R80W n=1059, L4139P n=525, R3719Q n=955, G3651S n=373, N77S n=158, W139C n=505, R2215W n=273, R2220W n=448, T3135M n=993). ADPKD, autosomal dominant polycystic kidney disease; AU, arbitrary units; HEK, human embryonic kidney cell line; IMCD3, inner medulary collecting duct cell line; PC1, polycystin-1; WT, wild-type.

We next measured cell surface localization in HEK-293 cells of 11 benign and seven pathogenic PC1 variants (Figure 1C). All benign missense variants localized robustly at the cell membrane. By contrast, four of seven tested pathogenic variants (N77S, W139C, R2215W, T3135M) attenuated plasma membrane localization (Figure 1C).

Polycystins localize to the primary cilium of kidney epithelial cells.^43,44 We therefore asked whether pathogenic variants in PC1 affect ciliary polycystin localization. Consistent with plasma membrane localization in HEK-293 cells, all benign PC1 variants robustly localized to the primary cilia of IMCD3 cells (Figure 1, D and E); all pathogenic PC1 variants that compromised plasma membrane localization in HEK-293 cells also failed to localize to primary cilia of IMCD3 cells (Figure 1, D and E). Similarly, pathogenic variants that had localized normally to the plasma membrane in nonciliated HEK-293 cells exhibited robust ciliary localization in ciliated IMCD3 cells (Figure 1, D and E). Consistency between defects in plasma membrane localization in nonciliated cells and ciliary localization in ciliated cells suggests that many pathogenic variants compromise general mechanisms underlying PC1 biogenesis rather than specific mechanisms underlying ciliary trafficking.

To begin to investigate the mechanism underlying disrupted ciliary localization of pathogenic PC1 variants, we evaluated expression levels of PC1 variants (Supplemental Figure 2, A–C). Defective ciliary localization was not explained by a decrease in total PC1 expression. This was confirmed by immunofluorescence of membrane localized and total PC1, which revealed abundant intracellular expression of the pathogenic PC1 R2215W variant, but failure of this variant to localize to the ciliary membrane (Supplemental Figure 3).

PC1 undergoes autoproteolytic cleavage at a GPS, and variants that disrupt this cleavage can impair ciliary localization.^27,45 We therefore asked whether pathogenic missense mutations that disrupt PC1 ciliary localization affected GPS cleavage. Two pathogenic variants that disrupt ciliary localization, R2215W and R2220W, impaired GPS cleavage, consistent with previous studies of R2220W and its mouse ortholog R2216W, and the reported role of the receptor egg jelly domain on GPS cleavage^28,29,45 (Supplemental Figure 2, A and D). By contrast, three other variants that disrupt ciliary localization (N77S, W139C, T3135M) were cleaved comparably to WT PC1, indicating that alternative mechanisms disrupt the ciliary localization of these variants (Supplemental Figure 2, A and D).

The Impact of Pathogenic PC1 Variants on Polycystin Channel Function

We next characterized the subset of pathogenic ADPKD variants that localize to the ciliary membrane at levels equivalent to benign variants (G3651S, R3719Q, L4139P; Figure 1E). This suggests that these variants might functionally compromise PC1 through a mechanism other than impaired localization, such as impaired channel function. To test this hypothesis, we coexpressed these pathogenic PC1 variants together with a gain-of-function mutation in PC2 (F604P, called PC2-GOF) in HEK-293 cells and measured polycystin currents in whole-cell recordings (Figure 2, A and B).^8,46 As we evaluate channel function at the membrane, we did not further evaluate variants that did not localize to the membrane. As previously observed, cells coexpressing WT PC1 with PC2-GOF generated robust outwardly rectifying currents (71.00±6.60 pA/pF, n=21, Figure 2, C and D). By contrast, cells coexpressing PC1 G3651S (7.27±0.99 pA/pF, n=11), PC1 R3719Q (8.65±1.07 pA/pF, n=12), and PC1 L4139P (14.67±2.13 pA/pF, n=13) with PC2-GOF exhibited channel activity indistinguishable from background currents (Figure 2, C and D). Together, these experiments suggest that a subset of pathogenic PC1 variants impair channel activity. These data are consistent with a model in which nontruncating pathogenic PC1 variants cause ADPKD through their disruption of polycystin ciliary localization or polycystin channel function.

Figure 2.

Pathogenic variants with normal ciliary localization disrupt polycystin channel function. (A) Whole-cell patch clamp was used to measure polycystin channel function in the plasma membrane of HEK-293 cells. (B) Step pulse protocol applied for the whole-cell patch clamp. The voltage step was given from −100 to +180 mV in 20 mV increments. Each pulse was given for 150 ms followed by a 50 ms −80 mV tail pulse. Holding potential was maintained at −60 mV. (C) Representative whole-cell patch-clamp recordings for PC1 WT, G3651S, R3719Q, and L4139P. (D) Mean current density (pA/pF) of WT and pathogenic variants. The current density was obtained 140 ms after the +180 mV step was applied. Each point represents a cell (WT n=20, G3651S n=12, R3719Q n=11, and L4139P n=13). Significance was determined with a one-way ANOVA with Dunnett multiple comparison test.

Characterization of PC1 Pathogenic Variants in Previously Functionalized Protein Domains

Many molecular functions have been described for PC1, including GPCR signaling, activation of signaling pathways through the PC1 C-terminus, and signaling transduction through the cleavage and translocation of the PC1 C-terminus.^{12,14,15,17,18,47,48} To further test the hypothesis that PC1 ciliary localization and channel function are critical mediators of human ADPKD phenotypes, we evaluated human pathogenic variants associated with alternate PC1 molecular functions. Specifically, we tested pathogenic PC1 variants in the region of the C-terminus that has been linked to translocation and function in the nucleus or mitochondria^{15,17,18,47–49} and variants linked to adhesion GPCR activity^45,50,51 (Supplemental Table 1).

A common feature of adhesion GPCRs is auto-proteolysis at a GPS motif. We characterized three pathogenic variants within the GPS domain—PC1 G3052R—reported to disrupt adhesion GPCR-like signaling by PC1⁵⁰ and two unstudied pathogenic variants: C3015W and F3018S. Both C3015W and F3018S severely disrupt ciliary localization of PC1, consistent with the previously characterized role of the GPS on PC1 ciliary trafficking (Supplemental Figure 4A).⁵² By contrast, G3052R localizes to the primary cilium but disrupts polycystin channel function (Supplemental Figure 4).

The polycystin C-terminus has been shown to bind and signal through G proteins and other signaling mediators as well as translocate and function in the nucleus and mitochondria.^{14,15,17,18,49,53} The pathogenic PC1 variant L4139P is in the previously described G protein–binding region, nuclear localization sequence, and mitochondrial localization sequence.^{14,15,17,18,47,49} We found that this variant has minimal effect on PC1 ciliary localization but severely disrupts polycystin channel function (Figures 1, D and E, and 2, C and D). To expand on these findings, we characterized six additional pathogenic variants in the PC1 C-terminus. PC1 L4132del was previously shown to disrupt GPCR signaling by the PC1 C-terminus.^54,55 PC1 L4132del localized to the primary cilium; however, PC1 L4132del severely impaired polycystin channel function consistent with a prior report (Supplemental Figure 4).⁵⁴ Four additional pathogenic PC1 C-terminal variants displayed similar molecular phenotypes to PC1 L4139P and L4132del (Supplemental Figure 4). These results suggest that the PC1 C-terminus plays a critical role in polycystin channel function and is consistent with the model that polycystins prevent PKD through ciliary ion channel function.

Classification of Variants of Uncertain Significance in PC1

There are no prevalent alleles in ADPKD; this allelic heterogeneity makes variant interpretation challenging and limits the utility of genetic testing in ADPKD. We therefore asked whether we could functionally classify PC1 missense VUS carried by individuals with ADPKD (Figure 3, A and B).

Figure 3.

Classifying the effects of variants of uncertain significance on polycystin surface localization and channel function. (A) Clinical and genetic characteristics of patients with ADPKD and variants of uncertain significance in PC1. (B) Abdominal magnetic resonance images from patients with ADPKD and variants of uncertain significance in PC1. (C) Reduced cell surface localization of variants of uncertain significance in PC1 identified in patients with ADPKD. Cell surface localization was quantified by flow cytometry in HEK-293 cells and is normalized to WT PC1 from the same experiment. Each point represents an independent experiment. (D) Representative whole-cell patch-clamp recording for Y420F PC1. (E) Mean current density (pA/pF) of WT PC1 and PC1 Y420F. The current density was obtained at 140 ms after the +180 mV step pulse was applied. Each point represents a cell (WT n=12, Y420F n=11). Significance was determined with a Student's t test. F, female; M, male.

We identified patients in our PKD center who have a clinical diagnosis of ADPKD, a missense variant in PKD1 classified as VUS on clinical genetic testing, and no other causal variants for cystic kidney disease on a kidney disease gene panel (Figure 3, A and B). We note that one of these variants (PC1 L3565R) has been previously reported in the literature in one individual with PKD.⁵⁶ To characterize these VUS, we generated three HEK-293 cell lines, each coexpressing one of the three PC1 VUS together with PC2 and quantified cell surface localization. PC1 C129Y and L3565R did not localize to the cell surface (Figure 3C). By contrast, PC1 Y420F exhibited decreased but measurable cell surface localization (Figure 3C). Cells expressing PC1 Y420F/PC2-GOF had dramatically impaired channel activity compared with cells expressing WT PC1/PC2-GOF (Figure 3, D and E). Taken together, our results suggest that interrogation of membrane localization and ion channel function of the polycystin complex are promising tools to determine PC1 variant pathogenicity.

Restoring Ciliary Localization of PC1 Pathogenic Variants

Since several pathogenic ADPKD variants impair polycystin surface localization, pharmacologic interventions to restore surface localization may be therapeutically beneficial. This principle was successfully established in cystic fibrosis.³² Before the identification of small molecule correctors of CFTR localization, it was recognized that culture at low temperature improved membrane localization of the common pathogenic CFTR F508del variant.⁵⁷ Lower temperature can stabilize folding intermediates that are destabilized by temperature-sensitive variants.⁵⁸ In support of this hypothesis, one PKD1 hypomorphic allele has been reported as being temperature sensitive.²⁸

To investigate whether ciliary localization and function of pathogenic PC1 variants could be broadly restored by interventions that facilitate protein folding, we cultured IMCD3 cells expressing PC1 variants together with PC2 at 37°C and 32°C and assessed PC1 ciliary localization. Ciliary localization of WT PC1 at 37°C and 32°C was equivalent (Figure 4, A and B). By contrast, four of five variants that disrupted ciliary localization demonstrated temperature sensitivity, with improved ciliary localization following culture at 32°C (Figure 4, A and B).

Figure 4.

The effect of temperature on surface localization of pathogenic PC1. (A) Ciliary localization of WT PC1 and pathogenic R2215W PC1 in IMCD3 cells cultured at 37°C and 32°C. (B) Effect of temperature on ciliary localization of pathogenic PC1 variants. Violin plots are shown with the median represented by a black line. Significance was determined with a one-way ANOVA on ranks with a Dunn multiple comparison test. Each point represents a cilium (WT 37°C n=2565, 32°C n=847, N77S 37°C n=158, 32°C n=265, W139C 37°C n=506, 32°C n=432, R2215W 37°C n=273, 32°C n=241, R2220W 37°C n=448, 32°C n=173, T3135M 37°C n=235, 32°C n=281). (C) Representative whole-cell patch-clamp recordings for cells expressing the indicated PC1 variant cultured at 37°C (black) or 32°C (blue). (D) Mean current density of cells expressing the indicated PC1 variant cultured at 37°C or 32°C shown as violin plots with median indicated. WT 37°C n=21, 32°C n=8, R2215W 37°C n=9, 32°C n=11, R2220W 37°C n=15, 32°C n=15, T3135M 37°C n=10, 32°C n=13. Significance was determined with a one-way ANOVA with the Holm–Šidák multiple comparison test.

If temperature sensitivity of PC1 variants is a result of improved folding, culture at lower temperature should also restore membrane trafficking of the modified polycystin constructs in nonciliated cells. Indeed, in HEK-293 cells expressing PC1 variants together with PC2, we observed improved plasma membrane localization of temperature-sensitive variants (Supplemental Figure 5). Together, these data suggest that many pathogenic PC1 variants cause disease by disrupting PC1 ciliary localization, potentially through their effects on PC1 folding.

The most common pathogenic cystic fibrosis variant disrupts both membrane localization and channel function.^32,59 Therefore, effective therapy for patients with this mutation relies on distinct molecules that improve CFTR membrane localization and channel function.^60–62 To evaluate whether restoring membrane localization of pathogenic PC1 variants restored channel function, we evaluated the effect of culture at low temperature on channel function of pathogenic PC1 variants.

We measured polycystin channel activity in whole-cell patch-clamp recordings of HEK-293 cells cultured at either 32°C or 37°C. Culture at 32°C did not affect currents measured from WT PC1/PC2-GOF (32°C: 59.03±11.91 pA/pF, n=8, Figure 4, C and D). As described above, culture at 32°C restored ciliary localization of the R2215W and R2220W variants. Consistent with this rescue in localization, PC1 R2215W/PC2-GOF and PC1 R2220W/PC2-GOF generated robust currents only when cells were grown at 32°C (37°C: 7.74±1.63 pA/pF, n=10; 32°C: 73.06±14.55 pA/pF, n=11 for R2215W; 37°C: 9.84±0.85 pA/pF, n=15; 32°C: 72.85±12.01 pA/pF, n=15 for R2220W), suggesting that increasing R2215W and R2220W surface localization is sufficient to restore function. By contrast, while lowering temperature restored cell surface localization of PC1 T3135M, it did not restore current densities (37°C: 10.21±1.38 pA/pF, n=13; 32°C:12.48±2.61 pA/pF, n=12), indicating that the T3135M variant disrupts both surface localization and channel function (Figure 4, C and D). These experiments indicate that restoring surface localization of a subset of PC1 pathogenic variants is sufficient to restore channel function, whereas restoration of function of other PC1 pathogenic variants may require both restoring surface localization and potentiating channel function.

Discussion

We measured the prevalence of PKD1 truncating variants and rare missense variants in individuals with PKD and controls in the UK Biobank. Comparing the prevalence of these two types of variants suggests that missense variants contribute to approximately 41% of ADPKD-PKD1. This proportion is slightly higher than previous studies in ADPKD-specific cohorts, which used completely different strategies to classify variants and patients and did not include control populations.^3,4,41 This high proportion of ADPKD attributable to missense variants may in part be explained by differences in individuals ascertained through clinical cohorts and diagnostic codes. However, our results suggesting that a greater proportion of ADPKD is caused by missense variants than previously indicated also likely reflects the challenge of classifying PKD1 variants. This variant classification challenge results in disease-causing missense variants being labeled as VUS, thereby underestimating the true burden of disease caused by missense variants and limiting utility of clinical genetic testing. The challenge in classifying variants underscores the importance of functional assays for ADPKD pathogenicity.

The total fraction of PKD explained by PKD1 variants in the UK Biobank (approximately 35%) is less than is seen in ADPKD clinical cohorts (approximately 80%), likely because many individuals with a PKD diagnostic code do not have ADPKD. However, because the rate of truncating and rare missense variants is low in individuals without ADPKD (Supplemental Figure 1), our estimates are robust to the limited specificity of PKD diagnostic codes.

Interrogating polycystin localization and channel activity allowed us to characterize molecular mechanisms that may underlie ADPKD. In total, we characterized 29 variants including ten benign variants, 16 pathogenic variants, and three VUS. Within this group of 19 pathogenic variants and VUS identified in patients with ADPKD, all variants severely disrupted at least one of two molecular phenotypes: PC1 ciliary localization and polycystin channel function. We also identified variants that severely disrupted both molecular phenotypes, such as PC1 T3135M.

We currently have only a limited understanding of how pathogenic variants in PC1 affect localization of the PC1/PC2 heterotetramer to the ciliary membrane. All tested pathogenic variants that disrupted ciliary localization also disrupted nonciliated cell plasma membrane localization in nonciliated cells. We infer, therefore, that many pathogenic variants disrupt ciliary localization through their effect on polycystin biogenesis or stability rather than a specific ciliary trafficking defect. As reduced temperatures improved ciliary localization of a subset of pathogenic PC1 variants, we hypothesize that these variants disrupt protein folding^57,63,64; however direct assays of PC1 protein folding and membrane stability are needed to further test this hypothesis.

Prior work has highlighted pathogenic missense variants that disrupt PC1 autoproteolytic cleavage and localization.⁵² We found that only a subset of variants that disrupt ciliary localization affect autoproteolytic cleavage, suggesting that distinct variants may disrupt different steps in PC1 biogenesis. Determining whether nontemperature-sensitive variants predominantly affect protein folding, protein–protein interactions essential for biogenesis, or protein stability may help reveal therapeutic strategies to restore localization and function of these variants.

A second group of pathogenic variants localized to the primary cilium but significantly attenuated channel activity. These “channel dead” mutants span a broad range of functional domains within PC1, including the PC1 tetragonal opening for polycystins (TOPs) domain, the PC1 C-terminal fragment stalk,⁵⁰ and the intracellular PC1 C-terminus. The TOP domain is also present in PC2, and pathogenic variants in the PC2 TOP domain exhibit right-shifted voltage-dependent channel activation (inhibition).⁶⁵ High-resolution structures of the polycystin tetramer that resolve the intracellular PC1 C-terminus and provide insight into channel dynamics are required to understand how variants outside the TOP domain disrupt channel function. In addition, it will need to be determined whether channel activity of channel-dead PC1 variants can be restored with small molecules—as has been pioneered with CFTR-directed therapies in cystic fibrosis.^32,66,67

Prior work indicated that the polycystins can function as ion channels,¹¹ atypical GPCRs,^47,48,68 or that their cleavage products can act as signaling mediators.^18,69 The relevance of these functions for human ADPKD remains unclear. This work, along with recent work on PC2 variants, supports the hypothesis that polycystin ion channel function is critical to ADPKD pathogenesis.^65,70 Although all pathogenic PC1 variants tested in this study disrupted either polycystin ciliary localization or channel function, our study does not exclude the possibility that non-channel function roles of polycystins are relevant to ADPKD pathophysiology. For instance, our study does not address the hypothesis that PC1 GPCR signaling is important for polycystin channel function⁵⁴ or the possibility that signaling by the PC1 C-terminus is dependent on polycystin channel function. In addition, characterization of even larger numbers of pathogenic variants may reveal variants that do not affect ciliary localization or channel function. These variants would provide human genetic evidence for additional functions of polycystins that inhibit kidney cystogenesis in humans.

The disruption of polycystin surface localization and channel function by ADPKD pathogenic variants is reminiscent of cystic fibrosis. Today, the combination of small molecules that rescue CFTR localization or channel function treat approximately 90% of all cystic fibrosis patients, dramatically ameliorating the disease severity.³² Our work suggests that similarly correctors and potentiators for polycystins hold promise for ADPKD.

CFTR correctors are molecular chaperones that improve surface localization by promoting folding of pathogenic CFTR variants. We found that lower temperature improved ciliary localization of a subset of pathogenic ADPKD variants. These results suggest that folding and ciliary localization of these variants could be restored by molecular chaperones that, like low temperature, stabilize PC1-folding intermediates. In line with our findings, folding of a mouse orthologue of a human hypomorphic PC1 variant and folding and localization of a PC2 pathogenic variant were found to be temperature sensitive.^28,71 Moreover, promoting folding of a mouse ortholog of the temperature-sensitive R2220W variant studied here reduced cystogenesis in a mouse model.²⁹ Together, these data suggest that molecular chaperones for polycystins may be broadly applicable to a significant subset of patients with ADPKD.

Our work has important limitations. We evaluated ciliary localization and channel function of a small subset of the human variation in PC1. Given our approach to choosing variants, we anticipate that our results are representative of the association between pathogenicity, ciliary localization, and channel function; however, additional studies will be important to determine whether there are smaller subsets of pathogenic variants that do not affect ciliary localization or channel function. Second, our assays for channel function and ciliary localization were performed in kidney cell lines with overexpressed polycystins rather than in vivo with polycystins expressed at endogenous levels. Future work enabling assessment of ciliary localization and channel function of polycystins expressed at endogenous levels will be an important advance.

In summary, our results reveal an approach to classify PC1 missense variants and suggest the potential for ADPKD therapies that promote polycystin surface localization and channel function.

Supplementary Material

jasn-37-1661-s001.pdf ^{(1.4MB, pdf)}

jasn-37-1661-s002.xlsx ^{(12.9KB, xlsx)}

jasn-37-1661-s003.pdf ^{(653.4KB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/F679.

Author Contributions

Conceptualization: Willow Coyote-Maestas, Markus Delling, Gabriel B. Loeb, Meyeon Park.

Data curation: Kotdaji Ha, Gabriel B. Loeb.

Formal analysis: Yukako Akiyama, Nadav Brandes, Mohona Gupta, Gabriel B. Loeb.

Investigation: Yukako Akiyama, Jillian Argiris, Nadav Brandes, Markus Delling, Mohona Gupta, Kotdaji Ha, Gabriel B. Loeb, Christine Haewon Park, Meyeon Park, Aide Pinedo, Fnu Ritu, Chun Jimmie Ye.

Methodology: Markus Delling.

Project administration: Jeremy F. Reiter.

Supervision: Markus Delling, Gabriel B. Loeb.

Writing – original draft: Markus Delling, Kotdaji Ha, Gabriel B. Loeb, Meyeon Park.

Writing – review & editing: Meyeon Park, Jeremy F. Reiter.

Funding

K. Ha: National Institute of Diabetes and Digestive and Kidney Diseases (K99DK131361). G.B. Loeb: National Institute of Diabetes and Digestive and Kidney Diseases (K08DK143339, T32DK007219), ASN Carl W. Gottschalk Research Scholar Grant, and UCSF Physician Scientist Scholars Program. M. Delling: National Institute of Diabetes and Digestive and Kidney Diseases (R01DK127277). Y. Akiyama: National Institutes of Health (5TL1DK139565).

Declarative Statements

This study includes clinical experimentation and received Institutional Review Board or Ethics Committee approval. All patients provided written informed consent. This study includes clinical experimentation and complies with the Declaration of Helsinki. This research was posted on a preprint server. 10.1101/2023.12.04.570035.

Data Availability Statements

Original data generated for the study will be made available upon reasonable request to the corresponding author. Data Type: Image Data; Raw Data/Source Data; Software Executable Code. Reason for Restricted Access: The data consist of large imaging and electrophysiology datasets.

Supplemental Material

This article contains supplemental material online, published as provided by the authors, at http://links.lww.com/JSN/F680, http://links.lww.com/JSN/F681.

Supplemental Figure 1. The prevalence of rare PKD1 missense variants in individuals with PKD.

Supplemental Figure 2. Effect of PC1 missense variants on PC1 expression and processing.

Supplemental Figure 3. A subset of pathogenic PC1 variants selectively disrupts ciliary membrane localization.

Supplemental Figure 4. PC1 variants in protein domains previously linked to alternate cellular functions disrupt polycystin ciliary localization or channel function.

Supplemental Figure 5. Effect of temperature on cell surface localization of pathogenic PC1 variants.

Supplemental Table 1. Table of genetic variants examined in this study.