PKD2-like proteins target Chlamydomonas PKD2 to distinct subciliary regions

Abstract

Many motile and sensory functions of cilia and flagella depend on the precise localization of transmembrane proteins within the ciliary membrane. Rather than being uniformly distributed, these proteins are often targeted to specific subciliary regions along the length or circumference of cilia. Here, we use genome editing and in vivo imaging to dissect the spatial organization of the channel protein PKD2 in Chlamydomonas cilia. MST3, a PKD2-like protein with five transmembrane helices (5TMH), is localized exclusively to the distal cilium, where it is essential for the localization of PKD2 and the formation of the hair-like mastigonemes. MST3, however, is dispensable for PKD2 assembly in the proximal cilium. In contrast, the PKD2-like 5TMH protein Proximal PKD2 Interactor (PPI) resides in the proximal cilium, where it is required for PKD2 assembly; ppi mutants retain distal PKD2 and mastigonemes. In mst3 ppi double knockouts, both proximal and distal PKD2 are absent, revealing that MST3 and PPI organize PKD2 into distal and proximal regions. Further, PKD2 colocalizes and comigrates with PKD2-like protein 1 (PLP1), another 5TMH protein, in cilia, indicating that they are present together in a complex. While plp1 mutants swim largely normally, mst3 and, to a lesser degree, ppi mutants swim with reduced velocity, indicating a functional specialization of the different ciliary PKD2 complexes. Thus, Chlamydomonas PKD2 associates with one of several 5TMH proteins to form spatially and genetically distinct PKD2 complexes, while also establishing their longitudinal distribution in cilia.

Introduction

Many sensory and motile functions of cilia and flagella depend on transmembrane proteins, including G-protein-coupled receptors, receptor guanylyl cyclases, and gated channels, localized within their membranes. For instance, insect olfaction involves ligand-gated ion channels (1), and mammalian sperm flagellar motility is regulated by several ion channels including the CatSper complex (2). Instead of being randomly distributed in the ciliary membrane, certain transmembrane proteins are targeted to specific subciliary regions, such as the ciliary tip or segments along the length of cilia. For example, the proximal region of human primary cilia is often specialized in transmembrane protein content, forming the inversin compartment (3) and the EvC zone (4). In Drosophila auditory cilia, the heteromeric ion channels formed by Inactive and Nanchung dominate the proximal cilium, whereas the mechanosensitive channel NompC is specifically located in the distal region (5). In addition to longitudinal patterning, transmembrane proteins also display asymmetric positioning around the circumference of cilia. CatSper channels, for example, form four longitudinal rows along the principal piece of sperm flagella (6), and the voltage-gated proton channel Hv1 is positioned in two lines asymmetrically on one side of the sperm flagellum (7). The presence of spatially restricted distributions suggests that the precise location of transmembrane proteins is critical for ciliary function (7). However, the molecular mechanisms that organize transmembrane proteins into subciliary domains and the functional consequences of these arrangements remain incompletely understood.

Polycystin-2 (PKD2), a member of the TRPP (transient receptor potential polycystin) subfamily of ion channels, is widely distributed among eukaryotes and expressed across a variety of tissues in multicellular organisms (8). PKD2 has six transmembrane helices and, in metazoans, forms heterotetrameric complexes by partnering with PKD1 and PKD1-like proteins in a 3:1 ratio (9). PKD1 encompasses 11 transmembrane helices, in which TMHs 6 to 11 are structurally similar to PKD2 (10). PKD1 and related proteins also have large N-terminal ectodomains, which are thought to serve sensory functions to control channel activity. PKD1/PKD2 complexes reside and function within cilia, and variants in PKD1 and PKD2 cause autosomal dominant polycystic kidney disease, a frequently inherited and currently incurable renal disorder, and left-right patterning defects (11–16). Notably, PKD2 often shows an asymmetrical subciliary localization. In the mammalian embryonic node, PKD2 preferentially localizes to the fluid flow-facing side of nodal primary cilia; the asymmetric distribution appears to be regulated by BMP4 (17, 18). The Drosophila PKD2 ortholog, Almost there (Amo), is concentrated near the ciliary tip of sperm flagella (19). A specialized PKD2 subdomain at the ciliary tip, from which PKD2 containing vesicles are released into the environment, is also observed in male-specific ciliated sensory neurons of C. elegans (20).

PKD2 is also present in the motile cilia of Chlamydomonas reinhardtii (21, 22), where it adopts a multipart distribution with distinct proximal and distal regions separated by a gap (23). In the distal cilium, PKD2 is attached to two of the nine axonemal doublet microtubules (doublets 4 and 8), forming two stationary rows, whereas PKD2 in the proximal cilium is mostly mobile (23). The distally localized PKD2 anchors mastigonemes, hair-like polymeric structures, to the ciliary surface. Proximal PKD2 is not associated with mastigonemes as this region of the cilium lacks mastigonemes (23, 24). Mastigonemes encompass the glycoprotein MST1, and in mst1 mutants, mastigonemes and distal PKD2 are lost, while PKD2 is retained in the proximal cilium (23). Furthermore, in zygotic rescue experiments, PKD2 is rapidly incorporated into the proximal part of the initially PKD2-deficient cilia, whereas entry into the distal cilium is sluggish, revealing distinct dynamics of distal and proximal PKD2 (25). These findings suggest the presence of at least two spatially distinct PKD2 complexes within Chlamydomonas cilia.

Although PKD2 is widely distributed across species, most protists, including Chlamydomonas, lack a PKD1 ortholog, raising questions about the composition of PKD2 complexes in these species. We recently identified Small Interactor of PKD2 (SIP) as an essential component of the PKD2-mastigoneme complex in Chlamydomonas (25, 26). SIP is a single-pass transmembrane protein that shares sequence similarity with the amino-terminal portion of Chlamydomonas PKD2, specifically TMH1 and parts of the extracellular TOP domain (25). Genetic deletion of SIP abolishes mastigonemes and dramatically reduces ciliary PKD2 levels. These findings suggest that SIP plays a general role in the assembly and ciliary import of PKD2 complexes rather than a specific role in organizing the distal or proximal PKD2 regions (25).

In silico structural analyses predict that SIP forms complexes with a family of PKD2-like 5TMH proteins that contain TMHs 2 to 6 of PKD2 but lack the PKD2-like TMH1 and TOP domain helices that correspond to SIP (26). These PKD2-like proteins typically have extended multidomain N-terminal ectodomains and unstructured C-terminal regions. SIP binding effectively reconstitutes the 6TMH architecture of PKD2 (26). The Chlamydomonas genome encodes ten such 5TMH PKD2-like proteins (Table S1), including MST3 (also known as Mstax), whose ectodomain was identified by cryo-EM as forming the central core of mastigonemes where it binds multiple copies of MST1 and helps to establish the polarity and ~800-nm length of the mastigonemes (26, 27).

We reasoned that the circumferential and proximo-distal patterns of PKD2 in Chlamydomonas cilia depend on its interactions with PKD2-like 5TMH proteins. To test this hypothesis, we combined proximity labeling, gene editing, and dual-color live-cell imaging of three PKD2-like proteins: MST3, Proximal PKD2 Interactor (PPI), and PKD2-like Protein 1 (PLP1). Our findings demonstrate that MST3 is required for the localization of PKD2 in the distal cilium and for the formation of mastigonemes, but not for PKD2 localization in the proximal cilium. In contrast, PPI is required for proximal PKD2 but not for the maintenance of distal PKD2 and mastigonemes. Both the distal and proximal PKD2 regions are lost from the cilia of a mst3 ppi double knockout strain. PLP1, another 5TMH PKD2-like protein, colocalizes and comigrates with PKD2 in cilia but is dispensable for the formation of both the distal and proximal PKD2 regions. These results suggest that Chlamydomonas employs several 5TMH partners to establish spatially, and likely functionally, distinct PKD2 complexes within cilia. Although SIP, MST3, PPI and PLP1 lack identifiable counterparts in other organisms, both mammalian and green algal PKD2 partner with proteins with extended N-terminal ectodomains, indicating an evolutionarily conserved design of PKD2 complexes.

Results

Proximity labeling identifies MST3 as a PKD2 interaction partner

In Chlamydomonas cilia, mNeonGreen (mNG)-tagged PKD2 assembles into distinct proximal and distal regions separated by a gap that is largely devoid of PKD2 (Figs. 1A, B) (23). To identify proteins involved in establishing the spatial pattern of PKD2, we applied a proximity labeling approach in which pkd2 mutants were rescued with PKD2 fused to the promiscuous biotin ligase miniTurbo (PKD2-miniT; Fig. S1A) (28). Expression of PKD2-miniT restored mastigonemes to pkd2 mutant cilia, confirming rescue (Fig. S1B). After biotin addition, Western blotting detected an increased number of biotinylated proteins in cilia of the pkd2 PKD2-miniT-TG strain compared to control cilia (Fig. S1C). Biotinylated proteins were enriched from both strains using magnetic avidin beads (Fig. S1D) and subjected to mass spectrometry (Table S2). Out of 396 proteins identified, the highest scoring protein was MST3, which was present in all eight experimental samples obtained from pkd2 PKD2-miniT-TG cilia and absent in all eight control samples. MST3 is a 5TMH PKD2-like protein recently identified as forming the central filament of mastigonemes (26, 27). As shown previously (26), mst3-2^CLiP mutant cilia lacked MST1 and mastigonemes (Fig. S1 E–G). Also identified was the known PKD2-interacting protein SIP (Table S2) (25). These results support the hypothesis that PKD2 can physically interact with the 5TMH protein MST3 in cilia.

Figure 1). MST3 is required for assembly of the distal PKD2 region.

A) Brightfield (BF), TIRFM, and merged images and the corresponding kymograms of a PKD2-mNG cell. The ciliary tips (T) and bases (B) and the gap region (open arrowhead) are marked. (Scale bar, 2 s and 2 μm.)

B) Schematic of a Chlamydomonas cell indicating PKD2 regions (green) and mastigonemes.

(C, D) As in A but for MST3-mNG (C) and mst3KO PKD2-mNG (D) cells.

E) Western blot analysis of cilia isolated from PKD2-mNG, the pkd2 mutant, and ppi, mst3, and mst3 ppi knockouts generated in the PKD2-mNG strain. The membranes were stained with antibodies against MST1, PKD2, SIP, and IC2, as a loading control.

F) TIRFM images of cells expressing endogenously tagged PKD2-mS and MST3-mNG. MST3-mNG is absent from the proximal ciliary region. Bar = 10 μm.

G) Kymograms of PKD2-mS MST3-mNG cilia showing PKD2-mS, MST3-mNG and the merged image. White arrows mark co-transport of both proteins in cilia by IFT. Green arrows mark MST3-mNG transports without detectable associated PKD2-mS. Red arrows mark PKD2-mS transports without associated MST3-mNG. Additional kymograms of this strain are shown in Fig. S2E. The ciliary tips and bases are marked. Bars = 2 s and 2 μm.

PKD2 localization in the distal cilium requires MST3

To localize MST3 within cilia, we used CRISPR/Cas9 gene editing to insert mNG upstream of the MST3 stop codon. The insert included mNG, the RBCS2 terminator sequence, and the aph7″ gene, which confers resistance to hygromycin (29–31). Thus, MST3-mNG is expressed from its endogenous locus but with the heterologous 3’-UTR of RBCS2. Although using a non-native 3’-UTR can affect gene expression, in Chlamydomonas, gene expression is thought to be predominantly controlled by genome context and the promoter region. Employing the RBCS2 3’-UTR allowed us to use the same DNA template to tag the other genes analyzed in this study.

MST3-mNG localized specifically to the distal cilium, forming two irregular rows in a pattern reminiscent of mastigonemes (Fig. 1C, S2A, B). Mastigoneme assembly requires the secreted glycoprotein MST1 (23, 32, 33). To test whether MST3 localization to cilia depends on MST1, we tagged MST3 in the previously characterized, mastigoneme-deficient mst1^CLiP mutant (23). Except for a few mostly stationary MST3-mNG particles, mst1^CLiP cilia were largely devoid of MST3-mNG (Fig. S3A). These findings indicate that the entry of MST3 into cilia and/or its cellular stability depend on MST1. Similarly, PKD2 depends on SIP for proper stability, proteolytic processing, and ciliary import (25).

To test whether MST3 is necessary for the assembly of the PKD2 complex in the distal cilium, we used CRISPR/Cas9 to disrupt MST3 in a strain expressing endogenously mNG-tagged PKD2 (PKD2-mNG). Cilia of the resulting mst3^KO PKD2-mNG strain lacked MST1 and mastigonemes, similar to previously characterized mst3^CLiP mutant strains (Figs. 1E, S4B, C) (26). In this strain, the distal rows of stationary PKD2-mNG were lost, but some PKD2-mNG particles entered the distal cilium, moving by diffusion and intraflagellar transport (IFT), the dedicated bidirectional protein shuttle of cilia and flagella (Fig. 1D, S5A, B). Thus, MST3 is required for the assembly of PKD2 rows in the distal cilium.

Next, we generated a double-tagged PKD2-mS MST3-mNG strain by mating the individual single-tagged strains. In this double-tagged strain, both proteins colocalized in the distal cilium, whereas the proximal PKD2-mS region lacked an MST3-mNG signal (Fig. 1F, Movies S1 and S2). High-resolution in vivo imaging revealed that PKD2 and MST3 largely colocalize in the distal cilium as two irregular rows (Pearson coefficient 0.83, SD 0.05, n=10 cilia; Fig. S2A). A minor subset of cilia (<5%) possessed only one PKD2-mS/MST3-mNG row (Fig. S2B).

The two proteins were frequently co-transported at velocities characteristic of IFT (anterograde: 2.3 μm/s, SD 0.29 μm/s, n=50; retrograde: 2.95 μm/s, SD 0.56 μm/s, n=16; Figs. 1G, S2E). However, instances of apparent independent transport of PKD2-mS and MST3-mNG were also observed (Figs. 1G, S2E). For MST3-mNG, this behavior could result from the limited photostability of mS, which may cause comigrating PKD2-mS to go undetected. In contrast, the independent transport of PKD2-mS may suggest the existence of MST3-independent PKD2 complexes within cilia. Our findings indicate that MST3 is the organizer of PKD2 in the distal cilium.

PKD2 localization in the proximal cilium requires Proximal PKD2 Interactor (PPI)

In mst3 knockout strains, reduced amounts of PKD2 and SIP were detected in Western blots of isolated cilia, and PKD2-mNG remained present in the proximal cilium (Fig. 1D, E), suggesting the presence of an MST3-independent PKD2 complex in this region. To identify proteins specific for the assembly of proximal PKD2, we took a candidate approach. Besides MST3, the Chlamydomonas genome encodes nine additional PKD2-like 5TMH proteins (Table S1, Fig. S6) that are all potentially capable of interacting with SIP and PKD2 based on structural predictions with AlphaFold2-Multimer (26). Previous proteomic studies of cilia showed that three of the gene products (Cre12.g539650, Cre09.g400850, and Cre13.g569550) are present in wild-type cilia and reduced in pkd2 and sip mutant cilia, indicating a possible interaction with PKD2 (26). Here, we refer to Cre12.g539650, Cre09.g400850, and Cre13.g569550 as Proximal PKD2 Interactor (PPI), and PKD2-like Protein 1 (PLP1) and 2 (PLP2), respectively. Each protein has a large ectodomain encompassing multiple lectin-like, polyproline, immunoglobulin-like, and cysteine-rich domains (Table S1, Fig. S6).

Using CRISPR/Cas9, we knocked out the three corresponding genes in a strain expressing endogenously tagged PKD2-mNG (Fig. 2, S2D). TIRFM showed a largely normal distribution of PKD2-mNG in plp1^KO and plp2^KO cilia, and negative-stain TEM revealed the presence of mastigonemes (Figs. 2A, B, S4B, C). In contrast, knockout of PPI largely eliminated PKD2-mNG from the proximal cilium (Fig. 2C). The ppi^KO PKD2-mNG strain, however, retained PKD2-mNG and mastigonemes in the distal cilium, and Western blots revealed the presence of MST1 in isolated cilia (Figs. 1E, 2C, S4). Thus, PPI is necessary for the assembly of PKD2 in the proximal cilium.

Figure 2). Assembly of the proximal PKD2 region requires PPI.

A–C) Brightfield (BF), TIRFM, and merged images and the corresponding kymograms of plp1^KO PKD2-mNG (A), plp2^KO PKD2-mNG (B), and ppi^KO PKD2-mNG (C) cells. PKD2-mNG is largely absent from the proximal region in ppi^KO PKD2-mNG. In the kymograms, the ciliary tips and bases are marked; cilia are numbered consistently in the images and kymograms. Bars = 2 s and 2 μm.

To determine the localization of PPI, we fused mNG to its C-terminus in a wild-type strain (Figs. 3A, B). The PPI-mNG particles were concentrated in the proximal cilium and were either stationary or moving by IFT and slow diffusion, resembling the behavior of PKD2-mNG in this region (Fig. S5A, C). Occasionally, a few PPI-mNG particles were observed distal to the gap, in the most proximal part of the distal region. This distribution distinguishes PPI-mNG from residual PKD2-mNG in mst3 cilia, which frequently moved by IFT and diffusion into the distal part of the mastigoneme-deficient mutant cilia (Figs. 1D, S5B). To determine whether PPI-mNG is confined to the proximal cilium because PKD2-mastigoneme complexes prevent it from entering the distal cilium, we generated a mst3^KO PPI-mNG strain. In this strain, PPI-mNG remained restricted to the proximal cilium, suggesting that it is confined to this region independently of the PKD2-mastigoneme complexes (Fig. S3B). The PKD2-mNG particles observed in the distal cilium of mst3 mutants could represent an additional PKD2 complex lacking both MST3 and PPI.

Figure 3). PPI and PKD2 colocalize in the proximal ciliary region.

A) Brightfield (BF), TIRFM, and merged images and the corresponding kymograms of PPI-mNG cells; cilia are numbered consistently in the images and kymograms. Occasionally, individual PPI-mNG particles were observed in the proximal part of the distal cilium (green arrows near cilium 4). White arrow indicates IFT of PPI-mNG. In the kymograms, the ciliary tips and bases are marked. Bars = 2 s and 2 μm.

B) Still images and corresponding kymograms, for the cilia marked 1 and 2, of PKD2-mS PPI-mNG cells. Bars = 2 s and 2 μm.

C) Kymograms showing co-transport of PKD2-mS and PPI-mNG. Bars = 2 s and 2 μm.

D–G) BF, TIRFM, and merged images and the corresponding kymograms of PKD2-mNG (D), mst3^KO PKD2-mNG (E), ppi^KO PKD2-mNG (F), and mst3^KO ppi^KO PKD2-mNG (G) cells. Both the proximal and distal PKD2-mNG region are lost in the mst3 ppi double knockout cilia (see Fig. S5D). Bars = 2 s and 2 μm.

In a PKD2-mS PPI-mNG dual-color strain generated by mating, PPI-mNG colocalized with PKD2-mS in the proximal region, whereas only PKD2-mS was detected in the distal region (Fig. 3B, Movies S1 and S2). Occasionally, we observed co-transport of PPI-mNG and PKD2-mS by IFT (n=2; Fig. 3C). The rarity of such events was likely due to the high signal density in the crowded proximal region.

To test whether PKD2 and PPI are in a complex, PKD2-mNG was immunoprecipitated from cilia isolated from the mst3 PKD2-mNG strain (Fig. S3D), which lacks the abundant PKD2-mastigoneme complex and thus likely offers a better chance to identify other PKD2 interactors. Mass spectrometry identified PKD2, SIP, and PPI in the immunopurified eluates of the mst3 PKD2-mNG strain but not those of the wild-type control strain CC-124, providing biochemical evidence for a ciliary PKD2/PPI/SIP complex (Table S3). To conclude, genetic, in vivo imaging, and biochemical data indicate that a PKD2/PPI/SIP complex resides in the proximal cilium.

MST3 and PPI organize the distal and proximal PKD2 regions

To further test the role of PPI and MST3 as organizers of discrete ciliary PKD2 regions, we generated an mst3 ppi PKD2-mNG double knockout strain (Fig. 3D–G). In Western blots of isolated cilia of this strain PKD2-mNG and SIP were strongly reduced and MST1 was not detected (Fig. 1E). TIRFM imaging revealed the presence of a variable number of residual stationary and mobile PKD2-mNG particles, but the proximodistal organization characteristic of PKD2-mNG in control cilia was absent in the mst3 ppi double knockout cells (Figs. 3G, S5D). In conclusion, the distal and proximal PKD2 regions in Chlamydomonas cilia are established by interactions with MST3 and PPI, respectively.

PLP1 forms a complex with PKD2

The observation of residual PKD2-mNG in mst3^KO ppi^KO cilia suggests the presence of additional ciliary PKD2 complexes, lacking MST3 or PPI. Previous proteomic analysis showed that PLP1 and PLP2, which are expendable for the formation of the distal and proximal PKD2-mNG regions (Figs. 2A, B, 4A), are severely reduced in pkd2 and sip mutant cilia (26), suggesting that their presence in cilia depends on PKD2 and SIP and implying that they are possible interactors of PKD2. To determine their distribution in cilia, both proteins were tagged with mNG. PLP2-mNG, which was tagged in the PKD2-mS background, was abundant at the ciliary base, and a variable number of bright particles were observed in cilia (ranging from 0 to 6/cilium; Fig. S3C a, b). Occasionally, we observed IFT of PLP2-mNG particles (Fig. S3C). The PLP2-mNG aggregates did not colocalize with PKD2-mS, indicating that either PLP2 does not partner with PKD2-mS or, more likely considering the proteomics data, that the tag impairs PLP2-mNG distribution.

In contrast, PLP1-mNG localized preferentially to the distal cilium, where it appeared organized into two narrowly spaced rows of stationary particles, resembling the distribution of tagged MST3 (Figs. 4B, S2C, S5F). The quantity and distribution of PLP1-mNG particles varied between cells and, in a subset of cells, PLP1-mNG was also present in the proximal cilium (Fig. S5F). In double-tagged cells, PLP1-mNG and PKD2-mS partially colocalized in cilia and were observed to travel together by IFT (Fig. 4C, D). These observations suggest that PLP1 forms a complex with PKD2 in Chlamydomonas cilia. Interestingly, a PKD2-mNG population, corresponding in distribution to PLP1-mNG, was absent from cilia of the mst3 ppi double mutant, suggesting that the PKD2/PLP1 complexes may depend on the PKD2/MST3 and/or PKD2/PPI complexes for assembly into cilia.

Figure 4). PLP1-mNG colocalizes and comigrates with PKD2-mS.

A) Brightfield (BF), TIRFM, and merged images and the corresponding kymograms of plp1^KO PKD2-mNG cells; cilia are numbered consistently in the images and kymograms. The ciliary tip and base are marked. Bars = 2 s and 2 μm.

B) BF, TIRFM, and merged images and the corresponding kymograms of PLP1-mNG cells; cilia are numbered consistently in the images and kymograms. Bars = 2 s and 2 μm.

C) BF, TIRFM, and merged images showing PLP1-mNG and PKD2-mS. The arrowheads mark colocalizing particles. Bars = 2 s and 2 μm.

D) Individual and merged kymograms of a PLP1-mNG PKD2-mS cilium. Individual transports of PLP1-mNG and PKD2-mS and co-transports of both proteins are marked with green, red, and white arrowheads, respectively. Bars = 2 s and 2 μm.

Knockout of SIP eliminates most PKD2 from cilia

AlphaFold2-Multimer modeling indicates that the PKD2-like 5TMH proteins partner with SIP to form PKD2-like transmembrane regions (26). Previously, we showed that PKD2 is largely absent from cilia of the sip^CLiP mutant, indicating that SIP is a critical component of the ciliary PKD2 complexes (25). Here, we generated a sip^KO PKD2-mNG strain by inserting a hygromycin-resistance cassette into exon 2 of SIP in the PKD2-mNG reporter strain. The resulting strain lacked mastigonemes and, in Western blots of isolated cilia, SIP and MST1 were undetectable or severely reduced (Figs. 5A, S4 B, C). In contrast to cilia isolated from the mst3^KO ppi^KO PKD2-mNG double mutant, which contained reduced amounts of PKD2-mNG as monitored by staining with anti-mNG, only traces of PKD2-mNG were detected in sip^KO PKD2-mNG cilia (Fig. 5A). In agreement with the Western blot data, PKD2-mNG was not detected in 40 of 77 sip^KO PKD2-mNG cilia analyzed by TIRFM; in the remaining cilia, PKD2-mNG particles were rare with a maximum of 6 particles/cilium and an average of 2.3 particles/cilium (SD 1.47, n=37 cilia; Fig. 5B a). The particles occasionally moved by IFT and bleached in multiple steps, indicating that they encompass several PKD2-mNG proteins (Fig. 5B b and c). The composition of these residual PKD2 particles (e.g., SIP-free PKD2/MST3 heterotetramers, PKD2-mNG homotetramers, etc.) and whether such particles are present in wild-type cilia or only form when the assembly of SIP-dependent PKD2 complexes is impaired remain unclear. We conclude that SIP is an essential component of nearly all ciliary PKD2 complexes.

Figure 5). SIP is an essential component of ciliary PKD2 complexes.

A) Western blot of cilia isolated from the PKD2-mNG (control), mst3^KO PKD2-mNG, sip^KO PKD2-mNG, and mst3^KO ppi^KO PKD2-mNG strains probed with the antibodies indicated. Note the near absence of PKD2-mNG in sip^KO cilia.

B) a) BF, TIRFM, and merged images and the corresponding kymograms showing PKD2-mNG in the sip mutant background. Cilia are numbered consistently in the images and kymograms; the ciliary tips and bases are marked. Bars = 10 μm (still images) and 2 s and 2 μm (kymograms). b) Kymogram (top) and gray value profile plot of a PKD2-mNG particle. Arrowheads in a and b indicate bleaching steps. Bar = 2 μm. c) Kymogram showing PKD2-mNG undergoing IFT (arrowheads). Bars = 2 s and 2 μm.

C) Bar graphs showing the swimming velocity of the indicated strains. More than 200 cells of each strain were analyzed; the standard deviations are indicated. The data shown were collected in three sessions and represent one experiment out of one (sip^KO), two (ppi^KO and plp1^KO), and three repeats (control, pkd2, PKD2-mNG, mst3^KO, and mst3^KO ppi^KO). ****p≤0.0001, n.s. not significant as compared to wild type (top) or pkd2 (bottom) by one-way ANOVA with multiple comparisons test.

The ciliary PKD2 complexes contribute differently to high-velocity swimming

The two rows of PKD2-mastigoneme complexes in the distal cilium form a fan-like superstructure oriented roughly perpendicular to the plane of the ciliary beat (23, 34). Mastigonemes are absent from cilia of the pkd2, sip, mst1 and mst3 mutants and mastigoneme-deficient cells swam slower than control cells, indicating a motility-related function of the PKD2-mastigoneme complex (23, 25–27, 32). The ppi mutant retains the PKD2-mastigoneme complexes in the distal cilium, allowing us to analyze how the loss of PKD2/PPI complexes from the proximal cilium impacts swimming velocity (Fig. 5C).

The PKD2-mNG reporter strain swam at an average velocity of 95 μm/s (SD 34.3 μm/s, n=274 cells), which is within the range reported for the previously characterized pkd2 PKD2-mNG-TG rescue strain (23, 25) but somewhat lower than that typically observed for wild-type Chlamydomonas (114 μm/s, SD 26 μm/s, n=231, Fig. 5C). As previously reported for the mst3^CLiP strains (26), the swimming velocity of the mst3 PKD2-mNG mutant was reduced by ~35% to an average of 62 μm/s (SD 20.1 μm/s, n=232 cells). A more moderate ~14% reduction in swimming velocity was observed for ppi PKD2-mNG (82 μm/s, SD 28.1 μm/s, n=207 cells). The data suggest that the PKD2-mastigoneme complex and to a lesser degree, the PKD2/PPI complex contribute to high-velocity swimming of Chlamydomonas. The mst3 ppi PKD2-mNG double knockout cells swam as slowly as the pkd2 mutant (48.1 μm/s, SD 17.8 μm/s, n=256 and 48.3 μm/s, SD 17.5 μm/s, n=234, respectively), and an average swimming velocity of 61.7 μm/s (SD 25 μm/s, n=256) was measured for the sip^KO PKD2-mNG strain. In contrast, a near-normal swimming velocity (90 μm/s, SD 30 μm/s, n=265 cells) was observed for the plp1^KO cells, suggesting that the PKD2/PLP1 complex is expendable for high-velocity swimming (Fig. 5C).

Mammalian PKD2 functions as a calcium-permeable, non-selective cation channel and loss of PKD2 disrupts calcium signaling in primary cilia (35–37). To test whether PKD2 contributes to calcium homeostasis in the motile cilia of Chlamydomonas, we expressed the calcium indicator GCaMP6 fused to the axonemal protein DRC4 (38) in control and pkd2 cells and used TIRFM to analyze ciliary calcium spikes in adhered cells (39, 40) (Fig. S7A). Calcium spikes occurred with similar frequency and intensity in the pf2 DRC4-GCaMP6 control and the pkd2 pf2 DRC4-GCaMP6 strains (Fig. S7B). Thus, PKD2 is not essential for bulk intraciliary Ca²⁺ dynamics in Chlamydomonas cilia, which likely result from the concerted activity of several calcium channels present in these cilia (41).

Besides waveform and beat frequency, calcium ions regulate other cilia-based behaviors in Chlamydomonas. Previously, we reported that phototaxis, photoshock and cilia shedding, which require calcium influx into cilia, are apparently normal in pkd2 mutants (23). Calcium also regulates gliding motility (39, 40, 42), a form of whole-cell locomotion (43). During gliding, cells attach via their cilia to the substrate and recruit the retrograde IFT motor dynein-2 into stationary ciliary adhesions (39, 40). Stepping of IFT dynein immobilized in such adhesions toward the base of the axoneme in the leading cilium pulls the cell along the substrate, while calcium spikes dissolve adhesions in the trailing cilium to initiate gliding. We observed that gliding velocity, the distance moved by gliding in one direction, and the frequency with which cells switch the gliding direction were similar for control and pkd2 cells, indicating that the loss of PKD2 does not affect gliding behavior (Fig. S7D–F). Interestingly, PKD2-mNG was recruited into the stationary ciliary adhesions of pkd2 PKD2-mNG cells and, when gliding continued, was occasionally shed from the tip of the trailing cilium and the base of the leading cilium (Fig. S7G, H). Thus, PKD2 shedding from cilia, previously observed in C. elegans (44), also occurs in Chlamydomonas.

Discussion

Distinct PKD2-like partners direct spatial organization of PKD2

In Chlamydomonas cilia, PKD2 is organized into proximal and distal regions, each with specific properties and dynamics (23, 25). Here, we show that the PKD2-like 5TMH proteins MST3 and PPI colocalize with PKD2 in the distal and proximal cilium, respectively, and are required for the assembly of PKD2 in those regions (Fig. 6). Additionally, another 5TMH, PLP1, colocalizes and comigrates with PKD2 in cilia, suggesting that they are present together in a complex. Thus, Chlamydomonas PKD2 partners with several distinct 5TMH proteins and the resulting complexes are targeted to specific subciliary regions.

Figure 6). A model for the heteropentameric organization of Chlamydomonas PKD2 complexes.

A) TIRFM image showing live cells expressing PKD2-mS in combination with either MST3-mNG or PPI-mNG. Cells of the two strains were mixed prior to imaging. Bar = 10 μm.

B) Schematic representation of the predicted PKD2-containing ciliary complexes in side and top views, illustrating the arrangement of transmembrane helices (TMH). Each complex is depicted as a heteropentamer composed of SIP, a 5TMH protein, and three PKD2 subunits.

While the precise structures of these complexes remain to be determined, the available data allow for estimates of the composition of the ciliary PKD2 complexes in Chlamydomonas. PKD2 is almost absent from sip mutant cilia, indicating that this small protein is an essential part of the ciliary PKD2 complexes (25). PKD2, SIP, and MST3 mutually depend on each other for assembly into the distal cilium and the formation of mastigonemes, supporting the notion that these proteins form a complex. Green algal mastigonemes are solitary structures, i.e., single hairs, each encompassing one MST3, emerging from a specific point on the ciliary surface (26). Further, PKD2-mNG particles residing in the distal cilium photobleach in an average of three discrete steps, indicating the presence of three copies of PKD2-mNG (23). Thus, Chlamydomonas PKD2–mastigonemes complexes likely consist of three copies of PKD2, one MST3 with an extended ectodomain binding multiple MST1, and one SIP, complementing MST3 to form a PKD2-like transmembrane and TOP domain. Similarly, the proximal PKD2 complex encompasses and depends on PKD2, SIP and PPI. We propose that Chlamydomonas PKD2 forms heteropentamers by partnering with SIP and one of several 5TMH proteins in a 3:1:1 stoichiometry (Fig. 6). Mammalian PKD2 assembles into 3:1 heterotetramers with PKD1 or PKD1L proteins and both mammalian and Chlamydomonas PKD2 require the heterogeneous subunits of the complexes for targeting and assembly into cilia.

Targeting of the PKD2 complexes could involve transport into specific subciliary regions by IFT, capture by proteins of the axoneme that define such regions upstream of the PKD2 complexes, or intraciliary membrane barriers that prevent the free movement of complexes in cilia (5).

How are PKD2-5TMH protein complexes targeted to specific subciliary regions?

The 5TMH proteins are likely the unique component of each ciliary PKD2 complex, and we propose that they direct the complexes to specific subciliary regions. Mechanistically, targeting could involve transport into specific subciliary regions by IFT, capture by proteins of the axoneme that define such regions upstream of the PKD2 complexes, or intraciliary membrane barriers that prevent the free movement of complexes in cilia (5). The receptor guanylate cyclase GCY-22 in C. elegans, for example, is concentrated in the distal region of primary cilia by anterograde IFT, which captures GCY-22 along the cilium and returns it to the tip (45). We previously showed that the organization of PKD2-mNG in distal and proximal regions is maintained at the restrictive temperature in fla10, a fast-acting conditional kinesin-2 mutant (46, 47), indicating that active IFT is not required to maintain the distribution of Chlamydomonas PKD2 in cilia (25). Rather, we propose that the intracellular, unstructured C-terminal and loop regions of the 5TMH proteins function as readers, identifying marks on the axonemal microtubules, allowing, for example, the PKD2/MST3 complexes to attach specifically to doublet microtubules 4 and 8 (48). Supporting this model, cryo-electron tomography revealed that the PKD2-mastigoneme complex connects via a thin linker to doublet microtubules 4 and 8 near their outer junction between the A- and B-tubules (23). Similarly, PKD2/PPI might concentrate in the proximal cilium by transiently binding to sites specific to the proximal axoneme. It is currently unknown whether additional proteins are involved in anchoring PKD2 complexes to the axoneme. A potential candidate is Cre17.g697500, an 896-residue protein with EF-hand motifs and P25 microtubule-binding domain, which was strongly reduced in pkd2 and sip cilia based on mass spectrometry (26).

Interestingly, PPI rarely entered the distal cilium even though it was observed to move by diffusion and by short-distance IFT. This suggests that processive IFT and free diffusion of PPI into the distal cilium are restricted. A hypothetical barrier or filter, possibly located in the PKD2-free gap region, could corral the PKD2/PPI complexes in the proximal cilium. Examples of such intraciliary barriers are the ciliary dilation of fly chordotonal neuronal cilia, which separates proximal and distal ciliary regions with distinct TRP channel content (5), and the septin-containing annulus that separates the midpiece and principal piece of mammalian sperm flagella (49). However, evidence derived from structural analysis for such a specialized barrier along the length of Chlamydomonas cilia is lacking. Experiments deleting and exchanging domains of the 5TMH proteins combined with live imaging could elucidate how these proteins guide PKD2 complexes to different subciliary regions.

Evidence for a functional diversification of PKD2 complexes in Chlamydomonas

In Chlamydomonas cilia, PKD2 interacts with at least three different 5TMH proteins (i.e., MST3, PPI, and PLP1). In agreement with most previous studies (23, 25–27, 32, 33, 50), the swimming velocity of the mst3 and sip KO strains, which lack mastigonemes, was considerably reduced. The reduction in swimming velocity observed for the mastigoneme-deficient mutants suggests that the PKD2-mastigoneme complexes possess a structural role, increasing the ciliary surface area and the efficiency of the ciliary beat (23, 34). A more moderate reduction was observed for the ppi mutant, which retains PKD2-mastigoneme complexes. The PKD2/PPI complexes are located in the proximal cilium, where ciliary bending is initiated and where bending is the strongest (51). While channel activity has not been demonstrated for Chlamydomonas PKD2 complexes, it is possible that the PKD2/PPI complexes could sense the deformation of the ciliary membrane in this region to fine-tune bend initiation and promote high-velocity swimming. A mechanosensory role is proposed for mammalian PKD1/PKD2 complexes in nodal and renal cilia, which generate calcium currents in response to fluid flow and ciliary bending (17, 35). Finally, plp1 mutants, which retain both PKD2-mastigoneme and proximal PKD2/PPI complexes, swam with largely normal velocity, suggesting that PLP1 is expendable for high-velocity swimming. The differences in swimming velocity between wild-type, mst3 and ppi cells, while moderate, suggest a possible functional specialization of the ciliary PKD2 complexes.

In addition to the ciliary 5TMH proteins (MST3, PPI, PLP1, and likely PLP2), the Chlamydomonas genome encodes six additional PKD2-like 5TMH proteins, expression of which is supported by expressed sequence tag (EST) data. Assuming that these proteins also partner with PKD2 and SIP, the function of all 5TMH protein complexes should be affected in pkd2 and sip mutants. However, the only currently known defect of pkd2 and sip mutants is the reduction of swimming velocity, which can be largely attributed to mastigoneme loss, raising the question of what roles the other putative PKD2/SIP/5TMH complexes perform. Identifying the physiological functions of the distinct PKD2 complexes remains an important direction for future studies.

A conserved design for ciliary PKD2 complexes?

SIP and the PKD2-like 5TMH proteins are restricted in their distribution to Chlamydomonas and its close relatives, most of which also possess mastigonemes (25). PKD1 and PKD1-like proteins show a similarly limited distribution, confined to metazoans. PKD2, in contrast, is widely distributed across eukaryotes. This combination of a broadly conserved channel with lineage-specific partner proteins raises the possibility that other phylogenetic groups have evolved their own as-yet-undescribed PKD2 partner proteins. Despite this divergence in partner identity, the overall organization of PKD2 complexes appears to be conserved. In both Chlamydomonas and vertebrates, PKD2 associates with multiple partners rather than a single dedicated one. In Chlamydomonas, PKD2 pairs with one of several 5TMH proteins, whereas vertebrate PKD2 partners with several PKD1-type proteins that are expressed in a tissue- and cell-type-specific manner (52, 53)}. These partners share a common architecture: both the 11TMH PKD1 proteins and the Chlamydomonas 5TMH proteins couple a PKD2-like transmembrane region to a large extracellular N-terminal region built from a diverse combination of domains. The two systems nevertheless differ in their stoichiometry: Chlamydomonas PKD2 appears to form pentameric complexes, whereas metazoan PKD1/PKD2 complexes assemble as tetramers (10). Thus, although the specific partner proteins and oligomeric states differ, PKD2 complexes appear to share a common design principle: the association of PKD2 with PKD2-like proteins bearing extended ectodomains.

MATERIALS AND METHODS

Strains, culture conditions and genotyping

The Chlamydomonas reinhardtii strains used in this study are listed in Table S4. Cells were grown in modified minimal (M) medium (https://www.chlamycollection.org/methods/media-recipes/minimal-or-m-medium-and-derivatives-sager-granick/) and maintained at 22°C with a 14/10 h light/dark cycle. Large cultures (>800 mL) used for cilia isolation were aerated with air enriched with 0.5% CO₂.

Cilia isolation and fractionation

To isolate cilia, the concentrated cells were washed in 10 mM HEPES (pH 7.4), resuspended in 10 mL of HMS (10 mM HEPES, pH 7.4, 5 mM MgSO₄, 4% sucrose), and deciliated immediately by adding 2.0 mL of dibucaine-HCl (25 mM in H₂O; Sigma-Aldrich) and vigorous pipetting (54). After addition of 20 mL 0.7 mM EGTA in HMS, the cell bodies were removed by centrifugation (1150 × g, 3 min, 4°C; Sorvall Legend XTR, Thermo Fisher Scientific). The supernatant was transferred to a fresh centrifuge tube and underlaid with a sucrose cushion (10 mL of 25% sucrose in HMS) and the remaining cell bodies were removed by centrifugation (1700 × g, 4°C, 10 min). Cilia in the upper phase were sedimented by centrifugation (27,000× g, 4°C, 20 min; Beckman Coulter, Avanti JXN-26), and resuspended in HMEK (30 mM HEPES, 5 mM MgSO₄, 0.5 mM EGTA and 25 mM KCl) supplemented with 1% protease inhibitor cocktail (Sigma-Aldrich, P9599). Cilia were lysed for 20 min on ice by addition of Triton X-100 (1% final concentration). The axonemes were separated from the membrane plus matrix fraction by centrifugation (27,000 × g, 4°C, 15 min).

Gene editing using CRISPR-Cas9

Gene knockouts and tag insertions were generated using CRISPR/Cas9 as previously described (30). CHOPCHOP (https://chopchop.cbu.uib.no) was used to select sgRNAs (Sigma; Table S5). For gene tagging with fluorescent proteins, the repair template was generated by PCR using primers encompassing a 20-25 base pair 5′-overhang homologous to the genomic sequence of the gene of interest and a plasmid encoding mNeonGreen or mScarlet-I fused to the RBCS2 3′UTR, followed by a selectable marker gene (i.e., aph7″ or aphVIII conferring resistance to hygromycin and paromomycin, respectively; Table S6). To generate gene knockouts, CRISPR/Cas9 was used to insert a selectable maker gene into an exon.

Cells grown in TAP medium (https://www.chlamycollection.org/methods/media-recipes/tap-and-tris-minimal/) were harvested by centrifugation, resuspended in in 6 mL autolysin at a density of 2-3 × 10⁷ cells/mL, and incubated for 1 h. The de-walled cells were then subjected to heat shock at 40°C for 30 min with gentle agitation. During this time, ribonucleoprotein (RNP) complexes were assembled by combining 1 μL Cas9 (Thermo Fisher Scientific), 2.8 μL Duplex Buffer (IDT), and 1.2 μL sgRNA; the mixture was incubated at 37°C for 30 min. The heat-shocked cells were resuspended in TAP supplemented with 40 mM sucrose at a density of 1 × 10⁸ cells/mL and stored on ice. Aliquots of 110 μL cells were mixed with 5 μL RNPs and 1 μg repair template DNA and subjected to electroporation using two pulses at 2,300 V with a width of 3 ms (Invitrogen Neon^™ Transfection system). After electroporation, the cells were allowed to recover at 16 °C for 1 h and then incubated in dim light for 20-24 h on a rocker with gentle agitation. Finally, cells were plated on TAP plates containing the appropriate antibiotic and incubated under constant light until colonies appeared.

Antibodies and Western blotting

Whole-cell samples, isolated cilia, or ciliary fractions were incubated at 95°C for 5 min in Laemmli SDS sample buffer. Proteins were separated by SDS-PAGE using Bio-Rad TGX precast gels and transferred onto a PVDF membrane. Membranes were blocked in blocking buffer (TBS supplemented with 0.05% Tween 20, 3% bovine serum albumin, and 3% fish gelatin). Immunostaining followed standard protocols: membranes were incubated overnight with primary antibodies in blocking buffer at 4°C with agitation (primary antibodies used in this study are listed in Table S7). After repeated washes in TBS-T, the membranes were incubated with secondary antibodies (anti-mouse IgG, 1:3,000 or anti-rabbit IgG, 1:4,000, conjugated to horseradish peroxidase; Invitrogen 31432/AB_228302 and 31460/AB_228341, respectively) for ~60–90 min at room temperature. For visualization, membranes were incubated with chemiluminescence substrate (SuperSignal West Pico PLUS or Atto; Thermo Fisher Scientific) and images were captured using a Bio-Rad ChemiDoc MP imaging system with Image Lab software (Bio-Rad). The source files for the Western blot images are available at Zenodo (DOI: 10.5281/zenodo.17716532).

Total internal reflection fluorescence microscopy (TIRFM)

TIRFM was performed using a Nikon Eclipse Ti-U microscope equipped with a 60× NA 1.49 objective (55). Through-the-objective illumination was provided by a 40 mW 488-nm and a 75 mW 561-nm diode laser (Spectra-Physics), using a Triple Band LaserSet (ZET405/488/561m/x/rpc; Chroma) excitation filter. The emission signals were cleaned up and separated with 520/35 nm BrightLine single-bandpass filter and a 560 nm edge BrightLine single-edge dichroic beam splitter (both Chroma) using a Hamamatsu W-VIEW Gemini image splitting device. Images were recorded at 10 frames per second (fps) using an Andor iXon 887 or 888 EMCCD controlled by the NIS-Elements Advanced Research software (Nikon). Live imaging of C. reinhardtii was performed in custom observation chambers, consisting of a 24×60 mm No. 1.5 coverslip and a 22×22 mm No. 1.5 coverslip, separated by ring of petroleum jelly. In the chamber, 15 μL of cell suspension was mixed with 15 μL of immobilization buffer (5 mM HEPES, pH 7.3, supplemented with 3 mM EGTA). Imaging was performed at room temperature through the larger coverslip and videos were recorded using NIS-Elements. Image processing and analysis were conducted using FIJI. Kymograms were generated with the KymoResliceWide plugin (https://imagej.net/KymoResliceWide). Colocalization of MST3-mNG and PKD2-mS in the distal cilium was calculated using the Fiji Colocalization tool. Adobe Photoshop was used for contrast and brightness adjustments. Final figures were prepared using Adobe Illustrator.

Analysis of swimming velocity

Cells were collected by low-speed centrifugation, resuspended in fresh M medium, and placed in a chambered plastic slide (Fisherbrand, 14-377-259). Swimming cells were imaged with a Nikon TMS microscope equipped with a 4× objective and a red-light filter to avoid phototactic behavior. Images were captured with a 1-s exposure time with a MU500 camera and its associated Toupview software (Amscope). The resulting cell trajectories were analyzed using the Length Analysis Plugin for ImageJ (https://github.com/Varsha1303/Length-Analysis-Tool-for-Slow-Cells). These measurements were performed at a lower temperature than in our previous studies (~22°C vs. ~24°C). The lower temperature, the use of a red filter, and the revised software could explain why the average velocities were lower than we previously reported. Graphs and statistical analyses (one-way ANOVA with multiple comparisons) were generated using GraphPad Prism.

Proximity labeling, functional assays, NSPARC imaging, negative-stain electron microscopy and mating are described in SI Appendix, Supporting Information Text.

Significance Statement.

Chlamydomonas PKD2 assembles into specific subciliary regions along the length and circumference of cilia. To establish this spatial pattern, PKD2 partners with several PKD2-like 5-transmembrane helix (5TMH) proteins. MST3 organizes PKD2 and the hair-like mastigonemes in the distal cilium, and Proximal PKD2 Interactor (PPI) organizes PKD2 in the proximal cilium. Further, PKD2 partners with PKD2-like protein 1 (PLP1), revealing the presence of spatially and genetically distinct PKD2 complexes in Chlamydomonas cilia. Like the Chlamydomonas 5TMH proteins, PKD1, the partner of PKD2 in metazoans, possesses a PKD2-like transmembrane region and an extended N-terminal ectodomain. The data suggest a conserved design of ciliary PKD2 complexes, encompassing heterogeneous partners to direct PKD2 complexes to cilia and subciliary regions.

Abbreviations:

BF

brightfield

KO

knockout

mNG

mNeonGreen

mS

mScarlet-I

IFT

intraflagellar transport

NSPARC

Nikon spatial array confocal

TG

transgenic strain

TIRFM

total internal reflection fluorescence microscopy

TMH

transmembrane helix

TEM

transmission electron microscopy

Data, Materials, and Software Availability

Study data are included in the article and/or supporting information. The source data for the Western blots and the tables from the mass spectrometry experiments have been deposited in zenodo (https://zenodo.org/records/21980524) (56).