Regulation of polycystin expression, maturation and trafficking

Abstract

The major autosomal dominant polycystic kidney disease (ADPKD) genes, PKD1 and PKD2, are wildly expressed at the organ and tissue level. PKD1 encodes polycystin 1 (PC1), a large membrane associated receptor-like protein that can complex with the PKD2 product, PC2. Various cellular locations have been described for both PC1, including the plasma membrane and extracellular vesicles, and PC2, especially the endoplasmic reticulum (ER), but compelling evidence indicates that the primary cilium, a sensory organelle, is the key site for the polycystin complex to prevent PKD. As with other membrane proteins, the ER biogenesis pathway is key to appropriately folding, performing quality control, and exporting fully folded PC1 to the Golgi apparatus. There is a requirement for binding with PC2 and cleavage of PC1 at the GPS for this folding and export to occur. Six different monogenic defects in this pathway lead to cystic disease development, with PC1 apparently particularly sensitive to defects in this general protein processing pathway. Trafficking of membrane proteins, and the polycystins in particular, through the Golgi to the primary cilium have been analyzed in detail, but at this time, there is no clear consensus on a ciliary targeting sequence required to export proteins to the cilium. After transitioning though the trans-Golgi network, polycystin-bearing vesicles are likely sorted to early or recycling endosomes and then transported to the ciliary base, possibly via docking to transition fibers (TF). The membrane-bound polycystin complex then undergoes facilitated trafficking through the transition zone, the diffusion barrier at the base of the cilium, before entering the cilium. Intraflagellar transport (IFT) may be involved in moving the polycystins along the cilia, but data also indicates other mechanisms. The ciliary polycystin complex can be ubiquitinated and removed from cilia by internalization at the ciliary base and may be sent back to the plasma membrane for recycling or to lysosomes for degradation. Monogenic defects in processes regulating the protein composition of cilia are associated with syndromic disorders involving many organ systems, reflecting the pleotropic role of cilia during development and for tissue maintenance. Many of these ciliopathies have renal involvement, likely because of faulty polycystin signaling from cilia. Understanding the expression, maturation and trafficking of the polycystins helps understand PKD pathogenesis and suggests opportunities for therapeutic intervention.

Introduction

Autosomal dominant polycystic kidney disease (ADPKD) is common monogenic disorder associated with end stage renal disease (ESRD) in late middle age and accounting for 4–10% of ESRD worldwide¹. The major ADPKD genes are PKD1 and PKD2, with PKD1 being a large gene located in a complex genomic region. PKD1 encodes a massive and complex protein, polycystin 1 (PC1, previously called TRPP1 but now no TRP name), that has characteristics of a cell surface receptor, G-protein coupled receptor, and channel subunit, but the unique combinations of domains has complicated determining its function (Figure 1A)¹. The PKD2 protein, polycystin 2 (PC2 or TRPP1, previously TRPP2) is an atypical TRP type channel that complexes with PC1, but the function of this channel and complex is still much debated^{2, 3}. Given the gene and protein complexity and questions about function, determining the expression of the ADPKD genes and proteins, including subcellular localization, has significantly contributed toward understanding this disease. In addition, documenting posttranslational processing and modifications, essential steps to develop a mature, appropriately localized and functional PC1 and PC2, have provided crucial insights into pathogenesis. These studies have highlighted vulnerabilities due to PC1’s tortuous maturation process that have uncovered further inherited forms of cystic disease. Although PC1 and/or PC2 likely have roles on the plasma membrane and/or ER, the primary cilium has emerged as a vital location of the polycystin complex necessary for maintenance of normal tubular structures in the kidney. In this review, we focus attention on understanding pathways critical for proper trafficking of polycystins to the cilium of renal epithelial cells. The ciliary localization has uncovered an additional group of monogenetic diseases associated with defects of cilia components, ciliopathies; syndromic, multiorgan disorders that often have a renal phenotype. Therefore, understanding expression, maturation and trafficking of the ADPKD proteins is central to understanding this disorder and related diseases.

1. Structure of polycystin 1 (PC1) and polycystin 2 (PC2), and analysis of PC1 glycoforms.

(A) PC1 and PC2 are transmembrane proteins that form a complex that is localized to multiple subcellular locations, including the primary cilium. Different conserved domains include the: N-ter, amino terminus; LRR, leucine-rich repeat, and C flank, carboxyl flank, and N flank, amino flank, regions; WSC, Cell wall integrity and stress response component, PKD Repeats, polycystic kidney disease domain repeats; REJ, receptor for egg jelly; GAIN, G protein-coupled receptor autoproteolysis-inducing domains A and B; GPS, G protein-coupled receptor proteolysis site; PLAT, PC1 lipoxygenase, α-toxin domain; TOP, tetragonal opening for polycystins; GPAP, G-protein-activating peptide; CC, coiled coil; C ter, carboxyl terminus; ERR, endoplasmic reticulum retention signal. (B) Glycosylation analysis of PC1 glycoproducts found in wildtype (WT) and Pkd2^−/− mouse embryonic fibroblasts (MEFs), that are untreated (U), or treated with EndoH (+E) or PNGase (+P), and detected with an antibody to the N-terminal region of PC1. In the WT, two N-terminal PC1, glycoforms are seen, resistant to (NTR) or sensitive to EndoH (NTS). PNGase removes all glycosylation and shows that the products are different glycoforms and not different protein isoforms. In the Pkd2^−/− cells the PC1-NTR product is absent and PC1-NTS elevated compared with WT MEFs. Therefore, PC2 is essential for appropriate glycosylation and maturation of PC1. Full length PC1 (not GPS cleaved), does not mature (is EndoH sensitive) and is seen more clearly in the Pkd2^−/− cells, so binding PC2 modestly influences GPS cleavage of PC1. Panel A is modified from Bergmann et.al. 2018¹ and Panel B is taken from Gainullin et. al. 2015⁴⁰

Expression of PKD1 and PKD2 and PC1 and PC2

The two major ADPKD genes are PKD1 (16p13.3; 46 exons; coding DNA sequence [CDS] 12,909bp) and PKD2 (4q22; 15 exons; CDS 2904bp)^4–7. PKD1 lies in a segmentally duplicated region of the genome, such that the 5’ part of the gene to exon 33 matches six pseudogenes (P1-P6) located ~15Mb further proximal in 16p^{4, 8}. A high level of similarity is seen with the pseudogenes (up to 99%), complicating analysis⁹. Both PKD1 and PKD2 are widely expressed genes with transcripts found in practically all organs, including during embryonic development, and cell types^{4, 5, 10, 11}. RNAseq data from human tissue¹² shows PKD1 expression in all tissues with the highest level in endometrium and lowest in liver (https://www.ncbi.nlm.nih.gov/gene/5310#gene-expression), while for PKD2 the highest is in endometrium and lowest pancreas (https://www.ncbi.nlm.nih.gov/gene/5311#gene-expression). Analysis of the Pkd1 promotor with X-Gal shows widespread kidney expression during development and in the adult, plus expression in bile ducts, pancreas, heart, the vasculature, and skeleton¹³. Common regulator elements have been found in the PKD1 and PKD2 promoters¹⁴, but some expression differences have been suggested during development¹⁵. This ubiquitous expression contrasts with the autosomal recessive polycystic kidney disease (ARPKD) gene, PKHD1, where expression is mainly limited to kidney, liver, pancreas and testis¹⁶ (https://www.ncbi.nlm.nih.gov/gene/5314#gene-expression). Employing various regions of the Pkhd1 promoter to drive Cre, loss of expression can be achieved just in collecting ducts or additionally in intrahepatic bile ducts and the male reproductive tract¹⁷.

Similarly, PC1 and PC2 are wildly expressed at the organ and cell type level when analyzed by western blotting, but PC1 expression seems to be higher during kidney development than in the adult^{11, 18–21}. Global and tissue specific knockouts have indicated roles for PC1 beyond the kidney and liver, including the pancreas, heart, vasculature, developing skeleton, lymphatics, and placenta^{13, 22–28}. Widespread expression is also supported by immunohistochemical and immunofluorescence analysis, with correspondence between the PC1 and PC2^{10, 11, 29–35}. However, this localization, and subcellular mapping is controversial and hampered by limited antibody reagents that work for immunolocalization, especially for PC1. Consistent with the structure of the proteins, they are enriched in membrane preparations¹¹. Possible locations for PC1 are the plasma membrane, including involvement in cell/cell interactions (via desmosomes), and secretion from cells on extracellular vesicles (EV)^{21, 31, 34, 36–40}. Secreted PC1 (and associated EVs) may bind to downstream targets, such as on primary cilia, and mediate “urocrine” signaling^{37, 41}. Apical membrane localization may be a precursor to secretion, or PC1 may act as a receptor at that location. For PC2, a plasma membrane location has been suggested, although the majority appears to be in the endoplasmic reticulum (ER)^{20, 37, 42, 43}. In the ER, interaction with the inositol 1,4,5-trisphosphate receptor (InsP3R) and regulation of the ryanodine receptor (RyR2) have been described, with a proposed role in regulating calcium homeostasis^44–46. PC1 and PC2 interact via their C-terminal tails to form a complex (Figure 1) as illustrated by co-immunoprecipitation, with recent data indicating a 3 PC2: 1 PC1 channel complex^{2, 3, 38, 40, 47, 48}. Compelling evidence indicates that the primary cilium is the likely location of this complex that is linked to the PKD phenotype (see below for details).

Primary cilia and a role in PKD

Cilia are microtubule (MT)-based hair-like organelles that protrude from the surface of most eukaryotic cells and are built and maintained by a MT-based intraflagellar transport (IFT)⁴⁹. Based on their function, cilia are divided into two types: motile or primary. The primary cilium has long been overlooked by mainstream science and considered as a vestigial organelle since mammalian primary cilia were first reported in 1898⁵⁰. Ironically, although mammalian primary cilia were first discovered in rabbit renal tubules⁵⁰, their pathogenic role in kidney diseases was only recognized in the past ~20 years. In 1999, the C. elegans homologues of PC1 (LOV-1) and PC2 (PKD-2) were colocalized on sensory cilia of adult males and found to be required for stereotypical nematode mating behaviors⁵¹. Soon after, the Tg737 gene mutated in orpk recessive PKD mice⁵² was found to encode IFT88, a key component of the IFT complex and necessary for assembly of kidney primary cilia⁵³. Furthermore, evidence that PC1 and PC2 are expressed in the membrane surrounding the primary cilium^{54, 55}, and the genes mutated in other PKD mouse models encode ciliary proteins^{56, 57}, consolidated the theory that PKD is a cilia-related disease, a ciliopathy. Primary cilia are sensory organelles that convert environmental cues into various cellular signaling, most clearly sonic hedgehog (shh), but also likely Wnt, and PKA signaling, that are key pathways involved in vertebrate embryonic development and tissue homeostasis^{58, 59}; and also the polycystin complex. However, the precise role of the polycystin complex is not known. One suggestion is that it acts as a mechanical sensor on the ciliary surface in renal tubules to regulate the proper response to intra-tubular liquid flow, perhaps through triggering a Ca2+ influx⁶⁰. Although, the electrophysical properties of the polycystin complex as well as how the mechanical sensing is transduced from the cilium to the cell body remain hotly debated topics in PKD research⁶¹.

Ciliopathies and cystic disease

Since primary cilia exist ubiquitously on cell surfaces, it is not surprising that ciliary dysfunction causes a wide spectrum of human genetic disorders, ciliopathies. At least 35 human disorders (with 187 causal loci cloned and more suspected) have been characterized as ciliopathies. Many of these ciliopathies have renal phenotypes, including: nephronophthisis (NPHP), Meckel syndrome (MKS), Joubert syndrome (JBTS), Bardet-Biedl syndrome (BBS), and orofaciodigital syndrome (OFD)^{57, 62–64}. Ciliopathies are often genetically complex, as an example, BBS is recessively inherited with at least 21 known causal loci with even suggestions of digenic inheritance^65–67. These syndromic disorders have a wide range of phenotypes, including: polycystic kidneys/tubulointerstitial nephritis/renal fibrosis, congenital hepatic fibrosis, craniofacial and digital defects, retinal degeneration, abdominal skeletal disorders, cardiovascular diseases, obesity/diabetes, and central nervous system anomalies. The range of organ involvement reflects the importance of ciliary signaling for tissue development and maintenance, although it is usually unclear why defects to specific ciliary components have the particular disease phenotype. A majority of characterized ciliopathy proteins are involved in ciliary trafficking⁶³, and a reasonable assumption is that the often seen cystic kidney phenotype is due to mistrafficking and reduced ciliary level of the polycystin complex (and possibly ARPKD protein, fibrocystin)⁶⁸.

Cilia trafficking and ciliary targeting sequence (CTS)

The fact that apical glycosylphosphatidylinositol (GPI)-anchored proteins fails to diffuse from the plasma membrane to the ciliary membrane in kidney tubule cells suggests a diffusion barrier for membrane proteins at the base of the cilium⁶⁹. Therefore, distinct trafficking mechanisms are likely required to strictly control vesical trafficking, ciliary import, intraciliary trafficking, and the removal of various sensory receptors and signaling proteins from cilia, to make the cilium a distinct functional entity.

To efficient direct polarized trafficking to a privileged subcellular compartment like the cilium, ciliary proteins would be expected to possess a discrete ciliary targeting sequence (CTS). However, the characteristics of a CTS remain poorly defined. One candidate CTS is the VxPX motif identified in the rhodopsin C-terminus, which when recognized by ARF4 may direct trafficking from the Golgi to the rod outer segment, a modified cilium of the retinal photoreceptor^{70, 71}. VxPx also seems to be required to target the retinol dehydrogenase (RDH8)⁷² to the outer segment and CNGB1b to olfactory cilia⁷³. It has been suggested that a similar N-terminal RVxP motif in PC2⁷⁴ and a C-terminal targeting sequence KVHPSST motif in PC1⁷⁵ might act as CTSs. However, other studies failed to reproduce the requirement for the VxPx motif to target PC1, CRMP-2, or NPHP3 to cilia^76–78. The predictive value of VxPx as a CTS is further questioned since the VxPx motif exists in 58% of mouse proteins and Arf4 knockout mice show no retinal degeneration or cystic kidneys⁷⁹. Subsequently, other CTSs have been suggested, including an AxEGG motif in the cystin N-terminus⁸⁰, an Ax(S/A)xQ sequence in the third intracellular loop of ciliary G-protein coupled receptors (SSTR3, HTR6, and MCHR1)⁸¹, and an RHKVRFEG motif in the PC1 C tail⁸². At this time, it seems that ciliary trafficking is governed by diverse mechanisms involving different CTSs, divergent recognition of such sequences, and even CTS-independently. As an example, ciliary targeting of some proteins can be accomplished through lipid modifications, such as palmitoylation and myristylation^{80, 83, 84}.

PC1 processing and post translational modification in the endoplasmic reticulum (ER)

PC1 is a highly glycosylated protein with 61 possible N-linked glycosylation sites⁶. N-linked glycosylation of membrane and secreted proteins occurs by a complex process shortly after translocation into the ER and before the protein is folded (Figure 2). In an iterative process, oligosaccharide precursor subunits are generated, added to an asparagine residue, and then trimmed to form the final glycan sidechain (Figure 2)^85–87. Each step in the process is facilitated by a specific enzyme and processing of the glycan sidechain is intimately associated with the appropriate folding of the protein in the calnexin/calreticulin cycle, with rejected, unfolded proteins unable to be trafficked onto to Golgi and exported for degradation (Figure 2)⁸⁸. Therefore, this N-glycan regulated protein folding is a crucial quality control step ensuring that only properly folded proteins are exported to the cell surface and secreted⁸⁵. Specific enzymes that remove all glycans, PNGase F, or remove just immature glycans, as found on ER resident proteins, endoglycosidase H (EndoH), can be used to trace proteins through the maturation process.

2. Diagram showing the ER biogenesis pathway for membrane/secreted proteins, highlighting proteins associated with cystic disease.

On the left are shown the steps involved in generating the Glc3Man3GlcNAc2-PP-Dol precursor (only limited steps outside of the ER are shown, dashed arrow) that is added to the nascent polypeptide following translation by the ribosome and entry into the ER. The enzymes catalyzing each of the glycan modification steps are shown. The process for generating the dolichol phosphate-linked mannose (Dol-Man) glycans that are added to grow the oligosaccharide, is shown bottom left, outside the ER. Phosphomannomutase 2 (PMM2) catalyzes the isomerization of mannose 6 phosphate (−6-P) to mannose 1-phosphate (−1-P). Mannose 1-phosphate is a precursor to GDP-mannose necessary for the synthesis of dolichol-P-oligosaccharides. The nascent polypeptide is imported into the ER through the SEC61 translocon, that also includes SEC62 and SEC63. The oligosaccharyltransferase (OST) transfers the Glc3Man3GlcNAc2 onto the peptide, with the chaperone proteins binding immunoglobulin protein (BiP) and ERdj3 (encoded by DNAJB11) in attendance. The oligosaccharide structure is further processed by glycosidase I (GI) and GII, that consists of α (encoded by GANAB) and β (encoded by PRKCSH) subunits. The glycoprotein then enters into the calnexin/calreticulin (CNX/CRT) cycle for protein folding. Alternate action of GII and UDP-Glc:glycoprotein glucosyltransferase (UGGT) remove and add the α1,3-linked glucose residue, with the residue required to bind the calnexin and calreticulin chaperones, until the protein is correctly folded or fails to fold. When the protein is correctly folded it is exported to the Golgi. If folding ultimately fails, the protein is de-mannosylated and becomes a substrate for the ER associated degradation (ERAD) pathway. BiP and ERdj3 can bind misfolded proteins. Enzymes and other proteins associated with cystic disease are shown in red.

An early step in the maturation of PC1 is cleavage at the GPS domain just before the first transmembrane of PC1 that results in N-terminal and C-terminal products (Figure 1A)⁸⁹. GPS cleavage occurs by a cis-autoproteolytic process and aided by the adjacent GAIN domain, but the products remain attached after cleavage by a non-covalent process such that the products can be co-immunoprecipitated^{90, 91}. The uncleaved PC1 remains immature and largely in the ER and is absent, or only seen at very low levels, in western analysis of tissue or cell lines, and not localized to cilia (Figure 1B)^{19, 40, 92–95}. GPS cleavage of PC1 occurs in the absence of PC2, although may occur less efficiently^{40, 96}. However, since a mouse PC1 cleavage mutant (p.Thr3041Val), Pkd1^V/V, survives to birth with just a few cysts, compared to Pkd1^−/− embryos that develop cysts from ~E13.5 and die embryonically, uncleaved PC1 may play a role during embryonic development^{22, 92}. Consistent with this, PC1^V/V, when expressed with PC2 in xenopus oocytes, was found to form a functional channel³. Although of note, a second cleavage mutation (p.Leu3040His) was not able to make a functional channel with PC2. PC1 cleavage to give rise to an 80–100kDa product (P100) corresponding to the region homologous to PC2 (the final 6 transmembrane domains), has also been described⁹⁷, with possible C-terminal tail cleavages discussed elsewhere in this issue of Cellular Signalling (Caplan review).

Analysis of PC1 glycoforms in membrane fractions of cells or tissue with an N-terminal antibody and EndoH or PNGase digestion shows the presence of an N-terminal mature form (resistant to EndoH; NTR) and N-terminal immature form (sensitive to EndoH; NTS), likely representing plasma membrane (mature) and ER (immature) forms of the protein (Figure 1B)^{38, 40, 93}. Immunoprecipitation shows that both PC1-NTR and PC1-NTS interact with the corresponding PC1-C-terminal products (CTR and CTS) and with PC2, and this interaction is required for the protein to exit the ER^{38, 40, 95}. A smaller, ~330kDa product found in humans but not mice may be a truncated PC1 resulting from inefficient splicing over the complex polypyrimidine tract in PKD1 exon 21⁹⁸. In EVs, only the mature, PC1 products are found⁹³. For PC2, only the EndoH sensitive form of the protein can generally be detected in tissue and cells, consistent with mainly an ER localization^{38, 40, 42}, with the possibility that the PC2 interacting with PC1 on the plasma membrane is in the ER membrane⁴⁰. However, a mature form of PC2 has been identified when the protein is isolated from cilia (that is also evident in a whole cell lysate IP), with a mature form of PC1 also found in a ciliary preparation⁹⁵.

Without PC2, PC1 is unable to mature staying as the PC1-NTS form (Figure 1B), and in PC1 null cells, PC2 is unable to translocate to cilia, indicating an interdependence of these proteins for appropriate trafficking to cilia, and for PC1 to the plasma membrane^{19, 40, 95, 96}. However, it is not clear that these proteins always traffic together with evidence that PC2 can traffic to cilia independently of PC1⁷⁴. Furthermore, it has been suggested that an N-terminal PC1 (PC1^deN) product may localize to the plasma membrane when unattached to the C-terminal product and to PC2⁹⁴. In Pkd2^+/− cells, the quantity of PC1-NTR is ~75% of normal, and so if the amount of PC1 on cilia is key to pathogenesis in ADPKD, this lesser decrease in PC1-NTR compared to in the Pkd1^+/− setting, may explain why PKD1 associated cystic kidney disease is more severe than that generally associated with PKD2⁴⁰.

Golgi to cilia trafficking of polycystins

The Golgi apparatus is the major sorting compartment of the cell with cargoes sorted not only to distinct plasma membrane domains: basolateral or apical, but also to endosomes and lysosomes, or back to the ER. ER-derived cargos traverse the Golgi from the cis to the trans cisternae (Figure 3). As integral membrane proteins, mature PC1 and PC2 glycoforms are transported through ER–Golgi shuttle vesicles to the trans-Golgi network (TGN) and then sorted and directed to their final destination, cilia. Various linear sorting signals are required to efficiently move membrane cargos from the Golgi to their final membrane compartment, with most of this selection occurring in the TGN^99–101. Sorting at the TGN depends on cooperation between clathrin and distinct adaptors such as the heterotetrameric adaptor protein (AP) complexes or Golgi-localizing, γ-adaptin ear homology domain (GGA) proteins^{101, 102}. Clathrin-coated cargo-bearing vesicles pinch off of the TGN and travel along MT tracks to the next set of membranes⁹⁹. For PC1, the C-terminus directly binds to RABEP1, a known GGA1 interactor⁹⁵. After the PC-complex leaves the ER, RABEP1 couples it to a GGA1/ARL3 module and then GGA1 assembles the clathrin coat to form PC-complex-bearing vesicles that are sorted along the MT network to the ciliary base (Figure 3)⁹⁵. Of note, inhibiting the kinase activity of CK1-δ, a kinase with its interactor A-kinase anchor protein 450 (AKAP450), both implicated in ciliogenesis by regulating Golgi organization and Golgi-derived MT nucleation^{103, 104}, significantly reduces PC2 signaling around the centrosome¹⁰⁵. Other studies suggest that PC2 may take distinct trafficking routes to cilia and the plasma membrane since specific variants in cell surface localized truncated PC2 (p.Lys572Ala and p.Phe576Ala, in the 4th transmembrane loop) specifically abolish trafficking to the plasma membrane but not the cilium¹⁰⁶.

3. Dynamics and major regulators of ciliary trafficking of the polycystin complex.

In renal epithelial cells, the membrane trafficking of polycystin-bearing vesicles is likely divided into multiple routes. Via the exocytosis pathway, newly synthesized polycystins are translocated into the ER where they are folded (Figure 2). ER-derived polycystin cargos then enter the cis-Golgi complex and moves through the TGN (1). Proteins involved in the TNG sorting are listed (2). It remains debatable whether there is a noncanonical cis-Golgi-to-apical-membrane route for polycystin-bearing vesicles (?). After exiting the TGN, polycystin-bearing vesicles are likely sorted to intermediate vesicular compartments, such as early or recycling endosomes, and then transported to the ciliary base (3). Some evidence suggests that polycystin-bearing vesicles might first dock on transition fibers before they fuse with the periciliary membrane (4). The membrane bound polycystin complex undergoes facilitated trafficking through the diffusion barrier of the transition zone and then enters the cilium. How the ciliary localization/retention of polycystin complex is regulated is poorly understood, but possible players in transporting the polycystins in cilia are listed (5). The ciliary polycystin complex can be ubiquitinated and removed from cilia by internalization at the ciliary base (6). The endosome system likely determines the subsequent fate of the internalized polycystins, sending it either back to the plasma membrane for recycling or to lysosomes for degradation. An alternative destination of polycystin-bearing vesicles not illustrated in this figure is their exocytic fusion with late endosomes with intraluminal multivesicular bodies (MVBs), which ends in the release of specific EVs, intraluminal vesicles (also known as exosomes), into the extracellular milieu.

Phylogenetically conserved IFT machinery specifically mediates the bidirectional movement of IFT cargos that are required for the biogenesis, maintenance, and signaling of cilia^{49, 107}. The IFT particle is composed of two subcomplexes (IFT-A and IFT-B). In a simple model, anterograde transport is regulated by kinesin-2, whereas dynein regulates retrograde transport. Interestingly, IFT20, a component of the ciliary IFT-B subcomplex, was discovered in the Golgi complex, anchored by the Golgin resident protein GMAP210/TRIP11, that facilitates transport of PC2 to the ciliary base^{108, 109}. As well as IFT20, RAB11, a small GTPase functioning in membrane trafficking from the TGN and recycling endosomes, has been implicated as a key regulator in TGN-to-cilia trafficking and ciliogenesis (Figure 3)^110–114. PC1 binds to a multimeric protein complex consisting of RAB11, ARF4, ARL6, and ASAP1, that is important for vesicle budding and Golgi exocytosis⁷⁵. Knockdown of trafficking regulators ARF4 or RAB8 functionally blocks the ciliary localization of an integral membrane CD16.7-PC1 chimera⁷⁵, although the exact role of factors such as RAB11 in regulating polarized trafficking of polycystins remains to be characterized.

The ciliary import of polycystins

Mounting evidence suggests that gating mechanisms regulate the selective ciliary entry of membrane proteins^115–117. Transition fibers (TFs, generated from the distal appendages of the mother centriole¹¹⁸) at the distal end of the basal body, and the transition zone (TZ, the proximal part of the axoneme that contains Y-links) at the most-proximal segment of the axoneme, are likely two structurally distinct and highly conserved functional compartments of the proposed ciliary gate (Figure 4)^{49, 119}. The TZ is characterized by Y-links that connect the axoneme to the ciliary membrane. Loss of TZ integrity results in loss of the diffusion barrier to cilia import in vertebrates, but does not significantly affect IFT or ciliogenesis in C. elegans^119–121. Immediately proximal to the TZ, TFs anchor the basal body to the apical membrane and constitute the first visible physical barrier between the cytoplasm and the cilium (Figure 4)¹¹⁸. Only a few proteins (including CEP164, CEP83, CEP89, SCLT1, LRRC45 and FBF1) have been characterized as genuine TF components in vertebrates^122–129, with only FBF1 having a clear homolog from C. elegans to humans¹³⁰. Many TF components or regulators are associated with human ciliopathies^131–133. For example, mutations in CEP164 and CEP83 cause NPHP-related disease^{131, 132} and SCLT1 and two centriole proteins critical for appendage formation, OFD1 and C2CD3, are linked to OFD^133–135. The homolog of human ciliopathy protein HYLS1 regulates TF formation and cilia gating in C. elegans¹³⁶.

4. Regulation of ciliary import of polycystin complexes at the ciliary base.

At the ciliary base, polycystin-bearing vesicles dock along the transition fibers (TF) by an unknown mechanism. The docking step likely serves to direct the fusion of polycystin-bearing vesicles with the periciliary membrane at specific sites, which are likely defined by cilia-specific fusion machineries. After fusing with the periciliary membrane, the polycystin complexes are actively transported to pass through the diffusion barrier of the transition zone (TZ).

Direct vesicle trafficking of cilia membrane proteins through TFs is unlikely due to space limitation between individual fibers (~60 nm). Thus, cilia bound, post-Golgi vesicles likely traffic to a privileged domain below TFs, and then fuse and diffuse laterally along the ciliary membrane^{49, 137, 138}(Figure 4). Loss of TFs abrogates basal-body-to-membrane docking in vertebrates¹²⁷ and affects the selective import of ciliary membrane proteins^{127, 129, 139–141}. In C. elegans, TF dysfunction does not affect the docking of the basal body but specifically compromises cilia import of both membrane (including PKD-2) and soluble proteins^{122, 130, 136}. The PC2 p.Glu442Gly point mutation specifically disrupts its ciliary localization but not channel activity^142–144. Homozygous mice carrying the PC2^E442G missense variant are prenatally lethal with embryonic kidney cystogenesis, suggesting the importance of the ciliary trafficking of polycystins to prevent PKD¹⁴². Intriguingly, PC2^E442G exhibits a unique ring-like cluster pattern that co-localizes with TF component CEP164, suggesting that TFs likely serve as a docking/sorting site for polycystin-bearing vesicles before they fuse with the privileged membrane at the ciliary base (Figure 4)¹⁴².

Polycystin-bearing vesicles at the ciliary base may come from various trafficking routes, including recycling endosomes. In this pathway, Golgi-originated polycystin-bearing vesicles either fuse with the apical membrane and then endocytose to the recycling endosome, or directly fuse with the recycling endosome. Disruption of regulators of the recycling endosome, PI3K-C2α¹⁴⁵, SDCCAG3¹⁴⁶, or the BLOC-1 complex¹⁴⁷, reduce ciliary PC2 and/or cause its accumulation in the recycling endosome (Figure 3). RAB8 and its activator RAB11 are both localized to the TGN and recycling endosomes and regulate vesicular transport from the TGN and the recycling endosomes to the plasma membrane and cilia¹⁴⁸. Hence, RAB8 is a key player in the selective entry of ciliary components, including PC2^{106, 137, 145}, with RAB11 controlling the localization and function of RAB8^{113, 114}, and both RAB11- and RAB8-positive vesicles docking at the cilia base^{145, 149}.

The exocyst is an octameric protein complex involved in vesicle trafficking, specifically the tethering and spatial targeting of post-Golgi vesicles to the plasma membrane prior to vesicle fusion¹⁵⁰. Exocyst subunits or the exocyst regulator CDC42 have been detected in primary cilia^151–154. PC2 appears to directly interact with exocyst subunit SEC10¹⁵⁵, and knockdown of CDC42 disrupts ciliary trafficking of PC2¹⁵⁶; depletion of CDC42 or SEC10 in zebrafish produces PKD-like phenotypes^153–155, and CDC42 deficiency in mice leads to cystic kidneys¹⁵⁷. CC2D2A, a ciliopathy protein, regulates the fusion of ciliary-directed vesicles at the periciliary membrane by maintaining the proper localization of EXOC4 and t-SNAREs (Figure 3)¹⁵⁸.

Intraciliary regulation of polycystin localization

The TZ consists of a number of multimeric protein complexes, including the NPHP, MKS and the newly identified CEP290 complex, that regulate the integrity of Y-links, and entry of ciliary membrane proteins (Figure 4)^{68, 159–164}. MKS complex deficiency leads to reduced ciliary PC2 and many other ciliary membrane proteins, such as GPR161, SMO, and ARL13B, indicating a central role of the TZ in controlling ciliary membrane protein composition^{68, 160, 165–167}. The tubby family protein 3 (TULP3) seems essential for the ciliary localization of PC1 and PC2; reduced ciliary polycystin probably accounts for the cystic phenotype in Tulp3 knockout mice^168–170. A trafficking model is that TULP3, the IFT-A subcomplex, and membrane protein cargo form a ternary complex at the ciliary base to facilitate passing of the cargo through the TZ into cilia¹⁷⁰. Recently, a TZ component, DZIP1L, that interacts with SEPTIN 2, a protein implicated in the maintenance of the TZ diffusion barrier, was identified as a ciliopathy gene with an ARPKD-like phenotype¹⁷¹. Loss of DZIP1L function results in PC1 and PC2 accumulation proximally in the cilium, probably reflecting a defect in TZ barrier function¹⁷¹.

After crossing the TZ, the bidirectional IFT machinery located between the ciliary membrane and the underlying axonemal MTs likely spreads membrane receptors along the cilium. Time-lapse analyses demonstrate that TRPV channels (including OSM-9 and OCR-2) are transported bi-directionally by IFT in nematode cilia¹⁷². However, while kinesin-2, the motor driving anterograde IFT, physically associates with PC2 in mammalian cells¹⁷³, polycystin motility inside cilia could not be detected¹⁷². An IFT-independent mechanism for cilia import of the polycystins is further supported by IFT machinery deficiencies not abolishing polycystin ciliary localization, but causing polycystin accumulation inside both C. elegans and mice cilia^{55, 174}.

The deficiency of small ubiquitin-like modifier (SUMO)ylation of the JBTS associated ARL-13/ARL13B compromises the normal ciliary targeting of PC2 in both worms and mammalian cells. SUMOylation is a reversible post-translational process that generates diverse molecular consequences, usually in the nucleus^{175, 176}, and so it is unexpected that SUMOylation modification regulates polycystin ciliary localization. Another posttranslational modification, axoneme polyglutamylation, also determines polycystin ciliary localization. MT polyglutamylation is most abundant on stable MT structures, such as ciliary axonemes^177–179; PC2 accumulates in hyperglutamylated cilia in both C. elegans¹⁸⁰ and renal epithelial cells¹⁴⁹. Depletion of a cilia-enriched deglutamylase, CCP5, effectively restores the ciliary dosage of polycystins in hypomorphic Pkd1 mammalian cells, highlighting targeting the glutamylation machinery as a potential therapeutic strategy in ADPKD¹⁴⁹. Direct association of polyglutamate chains and polycystins is not possible because of their distance apart, so an adapting mechanism is required to simultaneously recognize the polyglutamylation signal and anchor the receptors. Tubulin polyglutamylation may modulate binding of MAPs^{181, 182}, or enhance the processivity of kinesin motors^{180, 183}. Intriguingly, a unique kinesin-3 member (KLP-6) has been proposed to tether PKD-2 between the ciliary membrane and MT axoneme in C. elegans¹⁸⁴. KLP-6 abnormally accumulates in hyperglutamylated cilia¹⁸⁰, and could facilitate polycystin ciliary diffusion by tethering them to the polyglutamylated axoneme.

Bioactive signaling phospholipids may also be involved in ciliary protein localization. They have a hydrophobic lipid tail that inserts into membrane compartments and a distinctively phosphorylated inositol head that binds to specific phosphoinositide (PI) effectors. By adjusting the subcellular localization, conformation, and/or function of unique PI effectors, PIs contribute to a wide range of cellular processes, especially protein trafficking^185–188. Thus, the spatiotemporality of individual PI species, often controlled by specific PI kinases or phosphatases, is important for precise localizations. Dysfunction of multiple PI enzymes, both phosphatases (INPP5E^{189, 190} and OCRL1^191–193) and kinases (PI3KC2α¹⁹⁴ and PIPKIγ¹⁹⁵), have been connected to ciliary defects and/or ciliopathies. In C. elegans, CIL-1, a PI 5-phosphatase, regulates the ciliary localization of PKD-2 by controlling PI(3)P levels¹⁹⁶, while PI(4)P and PI(4,5)P2 levels in cilia regulate ciliary protein trafficking and subsequent signaling^197–200. Notably, two recent studies described a role of INPP5E in cilia signaling by maintaining a unique PI(4)P-high, PI(4,5)P2-low environment in the ciliary membrane^{197, 199}. Loss of INPP5E leads to an increase of the ciliary PI(4,5)P2 and, subsequently, the ciliary accumulation of PI(4,5)P2-binding protein, TULP3, and TULP3-binding membrane cargoes, including PC2^{170, 197, 199}.

Removal of polycystins from cilia

Downregulation of activated membrane receptors to the degradative pathway has been commonly used to spatiotemporally fine-tune signaling intensity. Improper downregulation of ciliary receptors may cause sensory transduction anomalies and ciliopathies. In C. elegans cilia, a phosphorylation/dephosphorylation balance of PKD-2, regulated by Casein Kinase 2 and phosphatase calcineurin, is critical for its ciliary homeostasis²⁰¹. LOV-1 physically associates with the STAM-Hrs complex, which directs internalized LOV-1 and PKD-2 on endosomes at the ciliary base for lysosomal degradation²⁰².

Ciliary polycystins are retrogradely transported to the ciliary base and endocytosed before they can be sorted to the lysosome. One key player implicated in this regulation is the BBSome. Eight evolutionarily conserved BBS genes (BBS1, 2, 4, 5, 7, 8, 9 and BBIP10) encode proteins that form the BBSome complex^{66, 203, 204}. The BBSome shares similar structural elements with clathrin coats, COPI and COPII, that polymerize and recognize ciliary membrane proteins²⁰⁵. However, its action may differ between organisms/systems, for example in mammalian systems, depletion of BBSome components compromise the ciliary import of several sensory receptors^{156, 205, 206}, while in Chlamydomonas reinhardtii, BBSome loss causes ciliary accumulation of several membrane-associated signaling proteins^{207, 208}. Interestingly, abnormal ciliary exit of SSTR3, Smoothened (Smo), Patched 1 and the dopamine D1 receptor (D1R) was also observed upon depletion of BBS2, BBS4, or BBS7 in mammalian cells^{206, 209–212}. The accumulation of ciliogenic proteins in BBSome-deficient cilia is believed to be caused by compromised retrograde IFT transport^213–217. Co-depletion of BBS4 and BBS5 disrupts lysosome-targeted degradative sorting of polycystins in either C. elegans or human kidney cells²¹⁸. It is likely that the BBSome also facilitates ciliary exit of activated and ubiquitinated polycystins through the TZ diffusion barrier and at the periciliary membrane, and then the downregulated polycystins are endocytosed into intermediate endosomes. In C. elegans, CAV-1 mediates endocytosis of PKD-2 at the periciliary membrane, then RABS-5 and VPS-45 coordinate the fusion of endocytic vesicles with early endosomes²¹⁹. The ubiquitin motif on the downregulated polycystins is then recognized by degradative sorting machinery enriched around the endocytic-machinery-containing periciliary membrane compartment, such as STAM/HRS²⁰², RAB5²⁰², Clathrin and AP2²²⁰. However, it remains unclear whether a similar endocytic pathway also regulates the turnover of ciliary polycystins in mammalian kidneys.

PC1/PC2 trafficking and the mutational mechanism in ADPKD

Although the majority of PC2 is in the ER, exogenously or transgenically expressed PC2 truncated in the C-terminal tail (amino acid 703) is found to escape the ER and go to the cell surface^42,221. Despite these trafficking defects, for many PKD1 and PKD2 mutations the disease mechanism is probably reduced functional polycystin complex on the cilium. For PKD1, truncated protein can often be found in mutant cells, sometimes expressed at a higher level than the normal protein, but judging from glycosylation analysis they do not mature and traffic out of the ER⁴⁰. This suggests that truncating mutations do not act by a dominant negative disease mechanism. PC2 truncating mutations also disrupt trafficking of PC1 to cilia and the apical membrane in polarized cells⁹⁶. For many missense and other inframe mutations to PC1 and PC2, similar to the cystic fibrosis protein (CFTR), the basic problem may often be defects in folding and trafficking of the proteins to the cell membrane and cilia1^{9, 40, 222}. Because of the requirement for PC2 for appropriate localization of PC1 (and vice versa), a PC2 missense change can result in reduced surface localized PC1-NTR, and lack of PC2 localization to cilia for a PC1 change^{19, 40}. The mechanism of some PKD1 mutations is illustrated by the missense change p.Arg3277Cys, a variant found in homozygosity in a family with adult onset ADPKD, and to modify the disease to very early onset when found in trans with a truncating PKD1 mutation²²³; phenotypes mirrored in a knock-in mouse model with this hypomorphic variant, Pkd1^{RC 93}. Analysis of urinary EVs from this mouse showed just ~40% of the mature, PC1-NTR, product compared to wildtype. Since this level can be increased to close the 100% in EVs secreted by Pkd1^RC/RC renal epithelial cells if incubated at 33°C compared to 37°C, this indicates that p.Arg3277Cys is a folding and trafficking mutation⁹³. Furthermore, it indicates some PC1 mutations may be amenable to chaperone treatment to aid folding and trafficking of the mutant protein²²⁴, as has successfully been applied for CFTR with corrector treatment²²⁵.

An epistatic phenotype is seen in Pkd1 models when Pkhd1 is also removed^{226, 227}. However, this does not seem to be because Pkd1 or Pkd2 are decreased or the amount of PC1-NTR is lower in the Pkhd1 null setting^{227, 228}. Instead, these diseases seem to affect different pathways, although both interact at the level of the primary cilium²²⁷. Mice with reduced PC1-NTR or both reduced Pkd1 and loss of Pkhd1 have longer cilia, but it is not clear if this is a direct effect of reduced protein levels or compensation for reduced polycystin signaling^{93, 227}. In contrast, it has been suggested that cystic disease associated with mutation to the tuberous sclerosis gene, Tsc1, is because TSC1 regulates the biogenesis of PC1 and the trafficking of the polycystin complex to cilia²²⁹. This may have a bearing on the severe renal phenotype when the adjacent PKD1 and TSC2 genes are contiguously deleted²³⁰. An alternative explanation is that the enhanced mTORC1 signaling (from TSC gene mutation) and disrupted polycystin signaling have a combined enhancement of cystogenesis²³¹.

Defects in PC1 trafficking: a cause of PKD and/or polycystic liver disease (PLD)

There is emerging data that defects in ER biogenesis of membrane proteins is associated with cyst development in the liver and the kidney²³². Initial evidence came from the finding that autosomal dominant polycystic liver disease (ADPLD), a disease associated with severe PLD but no or mild PKD, results from monoallelic mutations to genes encoding ER protein biogenesis components, SEC63 and PRKCSH, which encodes the glucosidase IIb (GIIβ) subunit (Figure 2)^233–235. The SEC63 protein is part of the SEC61 translocon complex that transports membrane proteins across the ER membrane, while GII is involved in trimming the glycan sidechain, and triggers quality control assessment of glycoprotein folding through the calnexin and calreticulin cycle (Figure 2)⁸⁷. Subsequently, monoallelic mutations to GANAB, encoding the GIIα subunit, were associated with a mild ADPKD-like or ADPLD phenotype^{228, 236}. ADPLD is also caused by heterozygous mutations to the gene encoding SEC61B, part of the SEC61 translocon, and ALG8, that is involved in early glycan generation (Figure 2)²²⁸. Recently, monoallelic ALG9 mutations, a protein also involved in early glycan generation, have been associated with renal cystic disease²³⁷. Single mutations to the gene DNAJB11, that encodes ERdj3, a glycoprotein in the ER lumen that acts as a co-chaperone with BiP, a heat shock protein associated with protein folding in the ER, and that binds some misfolded proteins, also cause cystic disease (Figure 2)^{238, 239}. In this case, the phenotype is of mild cystic disease without renal enlargement, with increasing fibrosis and declining renal function in older individuals, overlapping with the autosomal dominant tubulointerstitial kidney disease (ADTKD) phenotype²⁴⁰. Analyses of PC1 maturation in cells null for these various genes involved in the quality control of glycoprotein folding shows loss or a significant reduction of the PC1-NTR (and PC1-CTR) product^{226, 228, 236–238}. Analysis shows that PC1 is more vulnerable to defects in this pathway than the other tested membrane proteins. Specific biallelic mutations to PMM2, that encodes phosphomannomutase, an enzyme necessary for the synthesis of GDP-mannose, a subunit of oligosaccharide development (Figure 2), generates a phenotype of hyperinsulinemic hypoglycemia and PKD (HIPKD), with an ARPKD-like renal phenotype. Recessive PMM2 mutations are usually associated with Type 1a Congenital Disorder of Glycosylation, but if at least one of the mutations is a specific promoter mutation that inhibits binding of ZNF143, HIPKD results²⁴¹. Overall, these various cystic disorders show that PC1 maturation and trafficking is particularly vulnerable to defects in the ER glycosylation/protein folding pathway.

Concluding thoughts

Major technical advances and reagent development in the past two decades have helped define the localization and related processing to maturation of the polycystin complex. Cellular compartments and the connecting pathways involved in the trafficking of polycystins through the ER to their final function destination, the primary cilium, have been identified. The link of PKD to cilia, and parallel discovery of the disease associated, pleotropic consequences of defectively functioning cilia have brought this organelle out of the shadows so that ciliary structural and functional studies are now center place in disease biology. Processes such as IFT and vesicular trafficking to cilia have been characterized but many details of polycystin maturation and trafficking remain to be addressed. Major questions include what is the function of the polycystin complex on cilia, what roles do PC1 and PC2 elsewhere play alone or complexed with other proteins, and are these other functions associated with PKD? No doubt, other disease loci associated ER biogenesis defects and with ciliary dysfunction will be identified, solving the etiology of additional human diseases and providing further insights into polycystin trafficking and ciliary function. Present knowledge of the importance of polycystin dosage in ADPKD is spurring attempts to increase functional polycystin expression and appropriate localization using knowledge of its expression, maturation, and trafficking, with future understanding of these processes likely to present new therapeutic options. Although the field has made great strides in understanding where the polycystins are and how they got there, there is still much to discover before we fully understand PKD pathogenesis, knowledge that can drive the development of logical treatments for this important human disease.