A conserved signal and GTPase complex are required for the ciliary transport of polycystin-1

Abstract

Primary cilia regulate epithelial differentiation and organ function. Failure of mutant polycystins to localize to cilia abolishes flow-stimulated calcium signaling and causes autosomal dominant polycystic kidney disease. We identify a conserved amino acid sequence, KVHPSST, in the C-terminus of polycystin-1 (PC1) that serves as a ciliary-targeting signal. PC1 binds a multimeric protein complex consisting of several GTPases (Arf4, Rab6, Rab11) and the GTPase-activating protein (GAP), ArfGAP with SH3 domain, ankyrin repeat and PH domain 1 (ASAP1) in the Golgi, which facilitates vesicle budding and Golgi exocytosis. A related N-terminal ciliary-targeting sequence in polycystin-2 similarly binds Arf4. Deletion of the extreme C-terminus of PC1 ablates Arf4 and ASAP1 binding and prevents ciliary localization of an integral membrane CD16.7-PC1 chimera. Interactions are confirmed for chimeric and endogenous proteins through quantitated in vitro and cell-based approaches. PC1 also complexes with Rab8; knockdown of trafficking regulators Arf4 or Rab8 functionally blocks CD16.7-PC1 trafficking to cilia. Mutations in rhodopsin disrupt a similar signal and cause retinitis pigmentosa, while Bardet-Biedl syndrome, primary open-angle glaucoma, and tumor cell invasiveness are linked to dysregulation of ASAP1 or Rab8 or its effectors. In this paper, we provide evidence for a conserved GTPase-dependent ciliary-trafficking mechanism that is shared between epithelia and neurons, and is essential in ciliary-trafficking and cell homeostasis.

FIGURE 1:

CD16.7-PC1-truncation and alanine-scanning mutants do not localize to primary cilia. (A) Alignment of the C-terminal 112 amino acids of PC1 across multiple species. Top, amino acid numbering sequence for human full-length PC1; bottom, 112 amino acids tail numbering. Red underlined regions represent conserved domains, x(1,2)-P-[GS]-x(4)-[APTV]-x(2)-[AGP]-x(3)-[ADQ]-x-[DG]-x(3)-[AQT], repeated in the PC1 C-terminus (see Materials and Methods and Table 1). Black underlined text indicates a conserved (Q/R/K)-V-x-P-x ciliary-targeting domain also found in rhodopsin and PC2. Green lines denote PC1 C-terminal truncation and alanine-scanning mutants tested in ciliary localization studies. (B) Truncation and alanine mutants identify critical KVHPSST sequence for ciliary targeting. RCTE cells were transfected 3 d postconfluence for 24 h and analyzed for CD16.7-PC1 fusion proteins expressed in the primary cilia. Analyzed were: CD16.7-PC1-WT, three different truncation mutants +/− KVHPSST sequence, alanine-scanning mutant, or a CD16 chimera bearing either the complete CD7 C-terminus (CD16-CD7) or the complete CD7 C-terminus fused to the KVHPSST sequence (CD16-CD7-KVHPSST). Graph shows the fraction of RCTE cells containing each fusion protein in the primary cilium. Quantification was performed on Z-stacks of 30–50 ciliated cells for each mutant. n = 2; one-way ANOVA, p < 0.001. (C) Representative micrographs with cartoon drawing of construct (CD16, white box; CD7, gray box; PC1, green box) show CD16.7-chimera localization detected with mAb specific for CD16 (green) and cilia labeled with acetylated α-tubulin (red). Four of eight constructs are represented. Yellow arrows indicate location of a primary cilium in each XY panel. Bars 10 μm.

FIGURE 2:

Arf4 interacts with the PC1 C-terminus. (A) In vitro interaction of PC1 and Arf4. Purified GppNHp-activated GST-Arf4 was immobilized on glutathione-Sepharose 4B (top left blot) and incubated with lysates from HeLa cells transfected with CD16.7-PC1-WT, -359, -339, or -296. Immunoblots were performed using antibodies specific for CD16 or GST. Left graph, enrichment of CD16.7-PC1 tails bound by bead-immobilized GST-Arf4 (binding normalized to amount of GST-Arf4 bound to the beads and lysate expression). Right graph, quantification in relative optical density units (OD) of the amount of CD16-PC1 protein expression in lysates. Experiments were performed in triplicate; representative panels from multiple experiments are shown. Input for immunoblots: lysate, 7%; Arf4 in vitro capture, 30%. (B) Cellular interaction of PC1 and Arf4. HeLa cells were cotransfected with Arf4-GFP and CD16.7-PC1-WT or CD16.7-PC1-truncation mutants. An antibody specific for GFP was used to immunoprecipitate Arf4-GFP and PC1 fusion proteins from HeLa lysates. Immunoblots were performed using antibodies specific for CD16 or GFP. Immunoprecipitated CD16.7-PC1 fusion proteins were normalized to corresponding immunoprecipitated Arf4-GFP and CD16-PC1 fusion protein expression in lysates between experiments. Left graph, binding of CD16.7-PC1 mutants to Arf4-GFP was calculated as a relative percentage of the CD16.7-PC-WT binding to Arf4-GFP. Differences in CD16.7-PC1 mutant protein expression levels were taken into consideration by first calculating the fraction coprecipitated with Arf4-GFP as a function of CD16.7-PC1 protein expression in the lysate. Right graph, the results were the same. Graph depicts the means ± SE of three separate experiments. Brackets indicate CD16.7-PC1 chimeric proteins; arrows point to Arf4-GFP or the heavy chain of the immunoprecipitating antibody (as labeled); * refers to nonspecific or unidentified bands. Input for immunoblots: CD16-PC1 lysates, 7%; Arf4-GFP IP blot for CD16, 22%; Arf4-GFP lysates, and IP analyzed for GFP, 1%.

FIGURE 3:

Arf4 colocalizes and forms a complex with PC1 in the Golgi. Note: Because each channel in (A) and (B) was rendered in three dimensions using separate palettes in Voxx2 software, regions of colocalization do not appear yellow. (A) A pAb against PC1 detects a pool of PC1 in the Golgi apparatus and in cilia. Top row, human RCTE cells were grown 3 d postconfluence on filter supports and labeled with rabbit pAb against PC1 (NM002, green) and mouse mAb against acetylated α-tubulin (red). Bottom row, human RCTE cells were transfected with ST tyr isoform. Cells were immunostained for PC1 with NM002 and rhodamine-conjugated secondary antibody, and for myc-tagged α2,6-sialyl-transferase with a specific mAb directed against myc and a fluorescein-conjugated secondary antibody. Scale bar: 10 μm. (B) Arf4 and PC1 colocalize in the Golgi. MDCKII cells expressing GFP-Arf4 (green) were fixed and immunostained for PC1 (NM002, ocher), and a single transfected cell within the monolayer is outlined (single confocal Z-slice). See also Supplemental Movie 1. (C) Arf4-GFP colocalizes with Golgin-97. (D) Arf4 and PC1 form a complex that can be reciprocally coprecipitated. PC1 antibody NM002 specifically immunoprecipitated PC1 and Arf4 from RCTE lysate compared with control rabbit IgG (top panels). Equivalent results were found using MDCKII cells transfected with Arf4-GFP (unpublished data). PC1 and Arf4 protein levels were enriched in PC1-specific immunoprecipitation lanes (detailed in the text) and calculated as fraction of lysate protein immunoprecipitated with PC1 antibody relative to the fraction of lysate protein precipitated with control IgG (n = 3). Endogenous Arf4 was immunoprecipitated and immunoblotted for endogenous PC1 using RCTE (or MDCKII, unpublished data) lysates and two different rabbit pAbs against PC1 (NM002 and NM005, bottom panels). (E) Arf4 and PC1 form a complex in vitro. Purified GppNHp-activated GST-Arf4 was incubated with MDCKII lysate at 4°C or 37°C. Blots were probed for PC1; relative protein migration was 250 kDa due to matrix metalloprotease cleavage in the presence of magnesium. See Figures S3F and S8. (F) Arf4-GFP interacts with endogenous PC1, but does not interact with nonciliary, apical plasma membrane protein gp114. Arf4-GFP MDCKII lysate was immunoprecipitated in parallel with antibodies specific for PC1 or gp114. Blots were probed with antibodies specific for gp114 and GFP. (G) Cartoon representation of the predicted interactions between PC1 and Arf4 in specialized Golgi exit sites.

FIGURE 4:

The Arf GAP, ASAP1, binds Arf4 and is part of the PC1-Arf4 complex. (A) ASAP1 interacts with Arf4 in vitro. Purified GppNHp-activated GST-Arf4 was incubated with MDCKII lysate at 4°C or 37°C. Blots were probed for ASAP1. (B) Arf4 binds ASAP1 in cells. Endogenous Arf4 was immunoprecipitated from MDCKII cell lysates. Immunoprecipitates were resolved by SDS–PAGE and immunoblotted for endogenous ASAP1. (C) PC1 binds ASAP1 in cells. Endogenous PC1 was immunoprecipitated from RCTE (and MDCK, unpublished data) cell lysates and immunoblotted for ASAP1. PC1 antibody NM002 specifically immunoprecipitated ASAP1 from RCTE lysate compared with control rabbit IgG (top panels). ASAP1 protein levels were enriched in PC1-specific immunoprecipitation lanes (detailed in the text) and calculated as fraction of lysate protein immunoprecipitated with PC1 antibody relative to the fraction of lysate protein precipitated with control IgG (n = 3). RCTE (and MDCKII, unpublished data) cell lysates were immunoprecipitated for endogenous ASAP1 and immunoblotted for PC1 using two different pAbs (NM002 and NM005). Cartoon represents predicted interaction between PC1, Arf4, and ASAP1. (D) Truncation mutants of CD16.7-PC1-WT block ASAP1 binding. mAb directed specifically against CD16 was immobilized on protein G agarose and incubated with untransfected HeLa or RCTE lysates or lysates from HeLa or RCTE cells transfected with CD16.7-PC1-WT or CD16.7-PC1-359. Samples were immunoblotted with antibodies specific for CD16 and ASAP1. Percent input shown in upper left corner of each blot. * marks nonspecific or unidentified band.

FIGURE 5:

PC1 interacts specifically with Rab6, Rab8, and Rab11. (A) PC1 interacts with Rab GTPases in cells using two different methods. Left blots, endogenous PC1 was immunoprecipitated from postconfluent MDCKII cell lysates using covalently conjugated NM002-affinity resin. Coisolated proteins were resolved by SDS–PAGE and analyzed by immunoblotting for PC1, Rab6, Rab8, and Rab11. Uncoated (bead) resin served as a nonspecific binding control. Right blots, endogenous PC1 was immunoprecipitated from RCTE cell lysates; proteins were immunoblotted for individual Rab GTPases as indicated and detected using TrueBlot secondary antibody. Immunoprecipitations performed with preimmune rabbit serum served as a control. Graph represents quantification of multiple experiments represented by right-hand blots. n ≥ 3 experiments, *p ≤ 0.05, **p ≤ 0.005. (B) Rab8 and Rab11 antibodies reciprocally immunoprecipitated PC1 detected with two different anti-PC1 antibodies (NM002 and NM005). (C) PC1 does not interact with endocytic GTPases, Rab5 and Rab7. Endogenous PC1 was immunoprecipitated from postconfluent MDCKII cell lysates using covalently conjugated NM002-affinity resin. Coisolated proteins were resolved by SDS–PAGE and analyzed by immunoblotting for PC1, Rab5, and Rab7. (D) Truncation mutants of CD16.7-PC1-WT reduce Rab8 and Rab11 binding. mAb directed specifically against CD16 was immobilized on protein G agarose and incubated with untransfected HeLa or RCTE lysates or lysates from HeLa or RCTE cells transfected with CD16.7-PC1-WT or CD16.7-PC1-359. Samples were immunoblotted with antibodies specific for Rab8 and Rab11 and detected using TrueBlot secondary antibody. Confirmation of CD16 immunoprecipitation was conducted as in Figure 4 (unpublished data). Background in untransfected lane may be due to combination of nonspecific binding to protein G agarose and comigrating antibody light chains, detection of which cannot be completely eliminated for some antibodies, even with the TrueBlot system. Graphs show quantification of representative experiments. n = 2, *p ≤ 0.05, ** p ≤ 0.005. (E) Confocal microscopy of cultured NK cells shows colocalization (arrowheads) of Rab11 (red), PC1 (green), and ASAP1 (blue) in select punctae on the Golgi ribbon identified by PC1 staining. Scale bar: 20 μm. See also Figure S5 and Supplemental Movie 2.

FIGURE 6:

ShRNA-mediated depletion of Arf4 or Rab8 impairs ciliary trafficking. RCTE cells were sequentially transfected with pGIPZ-GFP shRNA plasmids (GAPDH, Arf4, or Rab8) for 72 h, followed by transfection and expression of CD16.7-PC1-WT for an additional 20 h. Samples were labeled for CD16 and acetylated α-tubulin. Micrographs, confocal stacks of samples were collected and individual cells (CD16.7-PC1-WT transfected only, or CD16.7-PC1-WT + shRNA GFP [labeled as GAPDH, Arf4, or Rab8 shRNA]) were analyzed for cilia (Cy5), shRNA transfection (GFP coexpression), and CD16.7-PC1-WT (Alexa 546) expression within cilia. Graph, p values represent comparison of counts of dually transfected cells relative to the counts of cells transfected with CD16.7-PC1-WT alone (labeled as “None”). Comparisons of Arf4 or Rab8 dually transfected cells to GAPDH dually transfected cells yielded the same results as comparisons to CD16.7-PC1-WT “None” control. Statistical analyses based on > 250 cells per shRNA condition. (Immunoblot) For biochemical studies, RCTE cells were sequentially transfected with pGIPZ-GFP shRNA plasmids (GAPDH, Arf4, or Rab8) for 72 h. ShRNA-transfected cells were isolated by flow-sorting based on GFP coexpression and average shRNA-mediated reduction in Rab8 and Arf4 was quantified by immunoblot analyses (n = 2).

FIGURE 7:

PC1 ciliary-trafficking pathway depends on a conserved K/R/QVxPx sequence and Arf4, Rab6, Rab11, Rab8, and ASAP1-trafficking machinery, and is distinct from the endocytic-trafficking route regulated by Rab5 and Rab7.