Modulation of inner junction proteins contributes to axoneme differentiation

Significance

We use the model organism Caenorhabditis elegans to study axoneme differentiation. We find that nematode inner junction protein CFAP-20 is restricted to the middle segment with doublets, and its loss disconnects A and B tubules, whereas another inner junction protein PCRG-1 is absent from most sensory cilia, and its deletion does not disrupt cilia. Furthermore, ectopic introduction of PCRG-1 into cilia generates abnormal MT doublets in the distal segment and reduces intraflagellar transport and animal sensation. Our results suggest that the absence of an inner junction protein prevents B-tubule extension and promotes the differentiation of the ciliary distal segments with singlet microtubules (MTs), adding a previously unrecognized layer of regulation underlying cilium formation, differentiation, and function.

Abstract

Cilia build distinct subdomains with variable axonemal structures to perform diverse functions in cell motility and signaling. In sensory cilia across species, an axoneme differentiates longitudinally into a middle segment with nine microtubule (MT) doublets and a distal segment with nine MT singlets that extends from the A tubules of the doublets. Here, we study axoneme differentiation in Caenorhabditis elegans by analyzing the flagellar inner junction protein FAP20 and PCRG1 that connect A and B tubules in Chlamydomonas. The nematode CFAP-20 is restricted to the middle segment with doublets, and its loss disconnects A and B tubules. However, PCRG-1 is absent from most sensory cilia, and its deletion does not disrupt cilia. Ectopic introduction of PCRG-1 into cilia generated abnormal MT doublets in the distal segment and reduced intraflagellar transport and animal sensation. Thus, the absence of an inner junction protein prevents B-tubule extension, which contributes to axoneme differentiation and ciliary function.

The axoneme is a microtubule (MT)-based subcellular assembly that provides a backbone for protruding a cilium from the plasma membrane (1, 2). As an evolutionarily conserved structure, an axoneme has been widely recognized with the 9+2 arrangement in which nine doublet MTs cylindrically surround a central pair of MTs. However, axonemes also differentiate into distinct subdomains to accomplish their diverse functions in different cell types (3–5). Distinct from motile cilia, sensory cilia build up an additional axoneme segment to detect and transmit signals that regulate cell fate and cell behavior (6–8). Sensory cilia consist of three separate domains: transition zone (TZ), middle segment, and distal segment (Fig. 1A). Immediately distal to the TZ is the middle segment that harbors nine MT doublets contiguous with those from the TZ, and only A tubules extend to form the distal segment characteristic of nine MT singlets (1).

Fig. 1.

Inner junction proteins with tubulin and in situ cryo-electron tomography of the Caenorhabditis elegans doublet MTs. (A) Schematic representation of the longitudinal structure of amphid channel cilia (only 4 cilia shown). Glial socket cells (moderate red) and sheath cells (dark cyan) are shown. (B) Experimental (Left, PDB: 6U42) and AlphaFold2 predicted (Middle) structures of Chlamydomonas flagellar inner junction complex including FAP20, PACRG, and alpha/beta tubulin dimer. The superposition of the two structures is shown on the Right with the RMSD and TM-score metrics. (C) AlphaFold2 predicted Chlamydomonas FAP20/PACRG and C. elegans CFAP-20/PCRG-1 heterodimer structures. (D) AlphaFold2 predicted structures of tubulin dimer with single Chlamydomonas FAP20 or C. elegans CFAP-20. (E) AlphaFold2 predicted structures of tubulin dimer with single Chlamydomonas PACRG or C. elegans PCRG-1. (F) AlphaFold2 predicted structures of tubulin dimer and C. elegans CFAP-20 and PCRG-1. (G, Top) Representative cryo-scanning electron microscopy (SEM) images of C. elegans L1 larvae after cryo-focused ion beam (FIB) milling (scale bar: 10 μm); (Bottom) the doublet MT selected for visualization (scale bar: 20 nm.) (H) Two cryo-ET slices of the doublet MT in G based on their localization for tomographic slices (scale bar: 10 nm.) Schematics are shown below. (I) Successive slices showing the ultrastructure of the C. elegans doublet MT inner junction. Magenta and yellow arrowheads indicate high electron densities and low electron densities at the inner junction locations, respectively. (J) Models of Chlamydomonas and C. elegans inner junction proteins in the axonemal doublet MTs.

The development of the ciliary distal segment appears to be widespread from Tetrahymena motile cilia to the nematode chemosensory, mammalian photoreceptor, and olfactory cilia (9). During sexual reproduction, the Chlamydomonas motile flagella assemble the distal segment with MT singlets to perceive mating signals, indicating that such a differentiated axoneme can be formed in motile flagella, despite transiently (10). Depending on cell type and species, the singlet-based distal segments have various lengths, ranging from 2.5 μm in Caenorhabditis elegans sensory cilia to about 50 μm in human olfactory cilia, and ~200 μm in frog olfactory cilia (11–13). Considering that olfactory receptors are localized within this region, the length of the distal segments is thought to correlate with sensory capacity.

Axoneme formation requires an MT-based intraflagellar transport (IFT) system that was originally discovered from Chlamydomonas but employed by virtually all the cilia (14–16). During the anterograde IFT, the kinesin-2 family motor proteins ferry the multimeric protein complex called IFT-particles that load the axoneme building blocks, such as α/β tubulin, from the ciliary base to the tip. The anterograde IFT machinery and ciliary turnover products are recycled to the base by dynein-2-powered retrograde IFT (14, 16–19). To assemble sensory cilia on C. elegans chemosensory neurons, two members of the kinesin-2 family, heterotrimeric kinesin-II (KIF3A/B/C in mammals) and homodimeric OSM-3 (KIF17) cooperate to drive IFT. First, the slower moving kinesin-II initiates transport from the basal body to build the middle segment of the axoneme. Subsequently, kinesin-II is gradually replaced by the faster moving OSM-3-kinesin, which alone enters and constructs the distal segment (20, 21). In vertebrates, the KIF17 homolog is responsible for building distal singlets on zebrafish photoreceptors and transporting signaling proteins to sensory cilia, suggesting the wide use of homodimeric kinesin-2 in the formation of the ciliary distal segment (22). Intriguingly, the loss of IFT-particle subunit IFT70/DYF-1 and IFT46/DYF-6 specifically disrupts the distal segment construction (23, 24), and a ciliary-specific α tubulin TBA-5 is enriched at the distal segment, demonstrating a specific pathway that promotes the extension of A tubules from the middle doublets to assemble the distal singlets (25). Equally essential is the inhibition of B-tubule extension from the doublets; otherwise, the distal segment would be formed by MT doublets. In comparison to the progress in understanding A-tubule extension to the distal segment, relatively less is known about the mechanisms that prevent B-tubule elongation. In previous research, Louka et al. identified CHE-12/Crescerin and ARMC9 acted as positive and negative regulators of B-tubule length, respectively, in Tetrahymena (26), and Park et al. reported that CDKL-1 and DYF-18 regulated the ciliary proximal segment length in C. elegans (27). However, functional significance of why the distal segment only forms MT singlets is mysterious.

In the doublet MTs, the inner junction and outer junction attach 13 protofilaments of the A tubule with 10 protofilaments of the B tubule (28–30). Cryo-electron micrography studies reveal that FAP20 (flagellar associated protein 20) and PACRG (Parkin Co-Regulated Gene protein) are arranged alternatively to form the inner junction bridge that links the A1 protofilament in the A tubule to B10 in the B tubule (31–34). Disruption of the inner junction in Chlamydomonas FAP20 or PACRG mutants still permits flagellar formation to normal length but causes abnormal flagellar beating patterns by affecting axonemal dynein–driven MT sliding, the effects of which are conserved in the motile cilia of zebrafish (31, 35, 36). In the sensory cilia of C. elegans, a recent study showed that CFAP-20 maintains inner junction integrity and is involved in sensory-dependent signaling, lifespan, and body development (36). Significantly, human patients carrying CFAP20 mutations exhibit retinal dystrophy, suggesting a pathological mechanism of inner junction proteins in ciliary diseases (36, 37).

C. elegans sensory cilia in the amphid channel perceive and transmit environmental cues regulating animal behavior such as chemotaxis and osmotic avoidance (Osm) (38–40). These cilia are composed of a MT-based axoneme surrounded by a ciliary membrane and feature 9 doublet MTs in the “middle segment” along with 2 to 6 singlet MTs, followed by the distal ciliary segment in which only A tubules extend from the middle doublets (Fig. 2 A and B), making these cilia a valuable model system to understand axoneme differentiation (12, 41, 42). Here, we show that two inner junction proteins CFAP-20 and PCRG-1 play distinct roles in axoneme assembly and that the ectopic gain of PCRG-1 in sensory cilia promotes B-tubule elongation, thereby promoting axoneme differentiation and ciliary function.

Fig. 2.

TEM analysis of ciliary ultrastructure in wild type (WT) and mutants. (A) Schematic of the longitudinal ultrastructure of amphid channel cilia (only 4 cilia shown). Glial socket cells (moderate red) and sheath cells (dark cyan) are shown. (B–D) Representative TEM images (cross-sections) of the middle segment of amphid channel cilia in (B) WT, (C) cfap-20, and (D) pcrg-1 mutant C. elegans. A high magnification of two doublet MTs is shown on the Right (all scale bars: 100 nm). The blue arrowheads indicate broken doublet MTs; the red arrowheads indicate ectopic singlet MTs. (E) Representative TEM images of the TZ of cfap-20 and pcrg-1 mutants (scale bars: 100 nm.) Blue arrowheads indicate the broken doublet MTs. (F) Schematics depicting the traverse ultrastructural phenotypes of doublet MTs from WT and cfap-20 mutants. IJ, inner junction proteins.

Results

AlphaFold2 Predictions of C. elegans Inner Junction Protein CFAP-20 and PCRG-1.

The C. elegans genome encodes cfap-20 and pcrg-1, the corresponding homologs of Chlamydomonas flagellar inner junction proteins FAP20 and PACRG. To gain insight into whether and how C. elegans homologs interact with tubulin dimers, we first assessed the ability of AlphaFold2 to predict the structures of the Chlamydomonas inner junction complex comprised of FAP20 and PACRG, and tubulin dimer. AlphaFold2 structures have a mean pLDDT score of 85.8, indicating good confidence in the prediction (43) (Fig. 1B). We then compared the predicted structures with the experimental structure (PDB: 6U42) (33) and calculated the rmsd (44, 45) and the TM-score (46), two common metrics for evaluating the similarity of protein structures. Remarkably, the predicted structures were highly accurate according to the experimental structures with an RMSD of 1.7 Å, and a TM-score of 0.91 (Fig. 1B). Both experimental and predicted results indicate that Chlamydomonas FAP20 and PACRG form heterodimers, and interact with alpha and beta tubulin, suggesting that AlphaFold2 may provide reasonable predictions for inner junction proteins.

AlphaFold2 predicts that the C. elegans CFAP-20 and PCRG-1 can also form heterodimers that interact with tubulin dimers (Fig. 1C); CFAP-20 has an interaction with alpha tubulin, similar to that of FAP20 in Chlamydomonas. Conversely, PCRG-1 alone does not appear to interact with beta tubulin, contrary to its counterpart in algae (Fig. 1 D and E). We searched the C. elegans single-cell RNA sequencing database for the gene expression pattern of cfap-20 and pcrg-1. As expected, cfap-20 is widely expressed in all ciliated sensory neurons, but pcrg-1 is only expressed in five out of 25 ciliated sensory neurons (SI Appendix, Fig. S1), the pattern of which was supported using GFP (green fluorescent protein) transgenic reporters (47, 48). Among the eight sensory neurons in the amphid channel, only the ASE neuron (ciliated neurons that are part of the amphid sensilla) expresses pcrg-1. These results suggest that distinct from the FAP20 and PACRG heterodimer in algae, most of the nematode cilia may use CFAP-20 rather than PCRG-1 to connect A and B tubules in their doublets.

In Situ Cryo-Electron Tomography (ET) of the C. elegans Doublet MTs.

To directly visualize the MT doublet structure of C. elegans cilia at high resolution, we employed a cryo-ET pipeline that we have recently developed (49). By vitrifying the C. elegans L1 larvae with high-pressure freezing, thinning them with cryo-focused ion beam (FIB) milling, followed by cryo-ET, we resolved tubulin subunits within the C. elegans doublet MTs in situ (Fig. 1G). Unlike the model that FAP20 and PACRG alternately align at the inner junction in Chlamydomonas (33), low electron density was periodically observed at the inner junction (Fig. 1 H and I, yellow arrows) in C. elegans MT doublets. These gaps occurred at axial intervals of approximately 4 nm, which is half of the 8-nm length of a tubulin heterodimer, reminiscent of those observed in the PACRG deletion mutant Chlamydomonas (31). In line with the undetectable expression of pcrg-1 in ciliated neurons, we argue that the C. elegans PCRG-1 is not located at the inner junction (Fig. 1J). We predict that, if CFAP-20 alone connects A and B tubules, cfap-20 but not pcrg-1 mutation will break the A and B tubule connection.

Transmission Electron Microscopy (TEM) of cfap-20 and pcrg-1 Mutant Cilia.

To examine the role of CFAP-20 or PCRG-1 in MT doublet assembly, we deleted 842 base pairs (bps) or 795 bps at the N termini of cfap-20 or pcrg-1, respectively, generating the putative null allele of each gene (SI Appendix, Fig. S2A). We analyzed the amphid cilia ultrastructure of the mutants via TEM with high-pressure frozen, freeze-substitution methods. Our observations from serial sections of amphid cilia confirmed that the mutants developed the distal and middle ciliary segments (SI Appendix, Fig. S2B). However, in all 28 examined cilia from two cfap-20 mutant animals, we found significant ultrastructural defects in the middle segment, characterized by the failure of B tubules to connect to A tubules in almost every recognizable doublet MT, forming hook-like structures (Fig. 2C), which is consistent with the lack of A-B connection in another cfap-20 mutant allele (36); In contrast, loss of PCRG-1 produced only mild phenotypes. In 30 cilia from two examined pcrg-1 mutant animals, we only found ectopic singlet MTs that appeared between doublet MTs in two cilia, including an ASK containing one ectopic singlet and an ASH containing two ectopic singlets (Fig. 2D). We further extended our observation to the TZ and found that the number of defective doublet MTs was reduced in the cfap-20 mutant, and all doublets remained intact in the pcrg-1 mutant (Fig. 2E).

Loucks et al. previously reported that pcrg-1 in C. elegans localized in a small subset of ciliated neurons, including two amphid neurons ASE and four outer labial quadrant (OLQ) neurons, and that its absence did not result in significant structural and functional defects, except for the abnormalities in OLQ neurons (48). In this context, we also used TEM to examine the ciliary distal segment, middle segment and dendrite. As shown in SI Appendix, Fig. S3, the ultrastructure of OLQ cilia in WT and pcrg-1 mutants is consistent with the early publication (48). The OLQ in cfap-20 mutant largely retained WT structure, despite several doublets in the middle segment were slightly deviated. Therefore, we concluded that CFAP-20 but not PCRG-1 is essential for connecting A and B tubules in C. elegans doublets (Fig. 2F).

The Loss of CFAP-20 Affects Cilium Length and IFT.

Next, we asked whether the absence of the inner junction compromises cilia length or IFT. By genetically crossing a GFP-tagged IFT marker, IFT54/DYF-11::3×GFP, into cfap-20 and pcrg-1 mutants, we found that cfap-20 mutant animals displayed slightly shorter cilia in comparison to the wild type, while the pcrg-1 mutant exhibited normal cilia length (Fig. 3 A and B). Examination of additional ciliary markers (IFT52/OSM-6 and IFT-dynein/CHE-3) resulted in similar findings, leading us to conclude that the CFAP-20 loss still permits the formation of the ciliary distal segments but modestly affects their length (SI Appendix, Fig. S4 A and B). An IFT assay of GFP-tagged IFT54/DYF-11 or IFT52/OSM-6 showed that anterograde IFT velocities along the middle segment were increased in the cfap-20 mutant, but the anterograde IFT at the distal segment and the retrograde IFT velocities remained unchanged (Fig. 3 C–E and SI Appendix, Fig. S4 C and D). The IFT velocities in the pcrg-1 mutant were unaffected (Fig. 3 C–E), which is consistent with previous report (48). To further understand the increase in IFT velocities along the middle segment, we monitored two IFT-kinesin motors and observed that the OSM-3 kinesin had accelerated in the middle segment (SI Appendix, Figs. S4 C and D). Considering that the phasmid neurons do not express PCRG-1, we suggest that CFAP-20, rather than PCRG-1, regulates ciliogenesis and IFT.

Fig. 3.

The loss of CFAP-20 affects cilium length and IFT. (A) Amphid and phasmid cilia in WT, cfap-20, and pcrg-1 mutants were labeled by DYF-11::3×GFP. (Scale bar: 5 μm.) (B) The cilium length (mean ± SD) in WT, cfap-20, and pcrg-1 mutants. n.s., not significant; ***P < 0.001. (C) Kymographs illustrate the anterograde and retrograde movement of DYF-11::3×GFP in the middle and distal segments. Representative particle traces are marked with blue, orange, and green lines. The scale bars represent 5 μm (horizontal) and 5 s (vertical). (D) Histogram of DYF-11::3×GFP velocities. (Top) anterograde IFT along the middle segments (Antero. m.s.). (Middle) anterograde IFT along the distal segments (Antero. d.s.). (Bottom) retrograde IFT (Retro.). Each plot was fitted with a Gaussian distribution. (E) Summary of IFT velocities of DYF-11::3×GFP in WT, cfap-20, and pcrg-1 mutants. ***P < 0.001. m.s.: middle segment; d.s.: distal segment. (F) Localization of CFAP-20 in amphid and phasmid cilia. OSM-6 was tagged with GFP. Endogenous CFAP-20 was tagged with Scarlet. (Scale bar: 5 μm.) (G) Localization of CFAP-20 in the phasmid cilia of the indicated animals. OSM-6 and CFAP-20 were labeled as in F. (Scale bar: 5 μm.) (H–I) The cilium length (mean ± SD) was measured using OSM-6::GFP fluorescence, and the middle segment length (mean ± SD) was measured using CFAP-20::Scarlet fluorescence in the indicated animals. Comparisons were performed between WT and the indicated mutants. n.s., not significant; **P < 0.01; ***P < 0.001. (J) CFAP-20::Scarlet fluorescence was photobleached in the middle segment region and analyzed in the recovery region cilia. The schematic at the Upper Left shows the photobleached region (black rectangle). D, dendrite; MS, middle segment; TZ, transition zone. (Scale bar: 5 μm.) (K) The kinetics of FRAP recovery at the middle segment is fitted with a single exponential equation (black line). The fluorescence intensity is normalized to the prebleach. In all panels, arrowheads indicate the ciliary base, and arrows indicate the junction between the middle segment and the distal segment.

We examined the cellular localization of CFAP-20 using the red fluorescence protein Scarlet to label the endogenous CFAP-20. Upon introduction of a GFP-tagged ciliary marker, we showed that CFAP-20 was concentrated in the middle but not the distal segment of amphid and phasmid cilia, consistent with its localization to the inner junction (Fig. 3F). To investigate the dependence of CFAP-20 localization on IFT components, we introduced CFAP-20::Scarlet into various ciliary mutants (Fig. 3G). Defects in IFT-A (che-11), IFT-B (osm-5), and IFT-dynein (che-3) mutant animals severely reduced cilia length, but these mutations did not prevent CFAP-20 from entering the cilia (Fig. 3G). The same was found in IFT-kinesin mutants (klp-11 and osm-3), although the length of the CFAP-20::Scarlet distribution along the middle segment length was partially reduced (Fig. 3 G–I). By examining two ciliary kinase mutants (dyf-5 and dyf-18), we found that the dyf-5 null caused abnormal cilia elongation, but the middle segment remained normal. Conversely, the defect in DYF-18 did not significantly alter the total ciliary length but caused a slight elongation of the middle segment (Fig. 3 G–I). We found that IFT mutants have shorter than normal cilia and that CFAP-20 is still associated with the now shorter middle segment. Thus, the effect of IFT on CFAP-20 distribution is likely to be indirect because cilia are shorter and the CFAP-20 region is sized accordingly.

CFAP-20 Exhibits High Dynamic Behavior in the Middle Segment.

Previous studies revealed that the exchange of tubulin subunits in the middle and distal segments of sensory cilia occurs relatively slowly (25). Given this, we sought to examine the dynamics of CFAP-20, which is located at the junctions between A and B tubules, through the use of fluorescence recovery after photobleaching (FRAP). Our results showed that after photobleaching of the partial middle segment, CFAP-20::Scarlet recovered rapidly in the bleached area (t_1/2= 16.1 ± 7.6 s), exhibiting an almost fivefold faster recovery rate than TBB-4 (t_1/2= 77.2 ± 23.7 s) (Fig. 3 J and K). To validate this result, we utilized the Maple3 photoconversion system by fusing the cfap-20 gene with Maple3 (SI Appendix, Fig. S5A). After global illumination with a 405-nm laser, CFAP-20::Maple3 underwent photoconversion from its green to red state (SI Appendix, Fig. S5B). When using precise illumination at the dendrite and partial middle segment, the converted red state CFAP-20::Maple3 diffused into the unconverted region at a rate (t_1/2= 15.7 ± 10.1 s), comparable to that observed in the FRAP assay (SI Appendix, Fig. S5 C and D).

Our FRAP and photoconversion assays demonstrated that CFAP-20 is highly dynamic in the middle segment. Given that the IFT machinery is responsible for transporting the ciliary cargo molecules and recycling the turnover of ciliary proteins, we examined whether CFAP-20 could be transported by IFT as a cargo. Similar to TBB-4, direct visualization of CFAP-20 movement by IFT assay was not possible due to the high levels of fluorescence resulting from the assembly of CFAP-20 into axonemes. To overcome this challenge, we performed a photobleaching-assisted motility assay, and we did not detect any obvious IFT of CFAP-20 along the C. elegans cilia (SI Appendix, Fig. S5 E and F), which is consistent with the early observation of FAP20 in Chlamydomonas flagella (35).

The Ectopic Expression (EE) of PCRG-1 Generates Abnormal MT Doublets in the Ciliary Distal Segments and Inhibits the Animal Sensation.

The Chlamydomonas axoneme consists of MT doublets connected by the alternating arrangement of FAP20 and PACRG, whereas C. elegans only has CFAP-20 in the middle segment with doublets. These differences stimulated us to ask whether the ectopic gain of PCRG-1 in sensory cilia promotes B tubule elongation from the middle doublets into the distal segments. To test this idea, we chose the Pdyf-1 promoter that turns on gene expression in ciliated neurons to express PCRG-1::Scarlet (Fig. 4A). DYF-1 is a conserved IFT70 protein homologue and plays an essential role for cilium formation across species. Our early study using a Pdyf-1::gfp reporter showed that this gene is expressed in all ciliated neurons in the amphid and phasmid (23), which is confirmed by the recent single-cell RNA-seq results (SI Appendix, Fig. S6A). As expected, we detected the red fluorescence of PCRG-1 within the amphid or phasmid channel cilia, where most PCRG-1 does not localize in WT animals. Unlike CFAP-20::Scarlet, PCRG-1::Scarlet was able to enter and concentrate in the distal segment (Fig. 4 B and C). Using TEM, we observed 31 cilia from two worms and found 11 cilia developed doublets in the ciliary distal segments (Fig. 4D), suggesting that B tubules could be abnormally extended from the middle to the distal when the sensory cilia ectopically gained PCRG-1. We detected an increased number of singlet MTs in the middle segment and the TZ (Fig. 4E). Statistical data demonstrate that an EE of pcrg-1 results in an increase in the number of cilia with more than 6 singlet MTs in the axonemes, indicating that PCRG-1 promotes MT assembly along cilia (Fig. 4F). We also examined the OLQ cilia in the pcrg-1 EE animal, and found the cilia retained the ultrastructure like that in the WT animal (SI Appendix, Fig. S3).

Fig. 4.

EE of pcrg-1-generated doublets MTs in the ciliary distal segments. (A) Schematic of plasmids used to label cilia (Pdyf-1::osm-6::GFP) and overexpress cfap-20 (Pdyf-1::cfap-20::Scarlet) or ectopically express pcrg-1 (Pdyf-1::pcrg-1::Scarlet). (B) Amphid and phasmid cilia in WT (sp2101, GFP tagged OSM-6 transgenic line), cfap-20 overexpression, and pcrg-1 EE animals were labeled by OSM-6::GFP. Arrowheads indicate the ciliary base, and arrows indicate the junction between the middle segment and the distal segment. (Scale bar: 5 μm.) (C) Representative fluorescence intensity profiles along the cilium. All fluorescence-intensity profiles are normalized to their maximum. (D) Representative TEM image (cross-sections) of the distal segment of amphid channel cilia (Left, scale bar: 200 nm.) Magnified images corresponding to the boxed regions are shown on the Right to show ectopic doublet MTs (scale bars: 100 nm.) (E) Representative TEM images of the middle segment and TZ showing supernumerary singlet MTs (scale bars: 100 nm.) (F) Statistical data on the number of singlet MTs in the ciliary middle segments of WT (N = 68 cilia in 8 worms), pcrg-1 EE animals (N = 42 cilia in 4 worms).

We then wondered whether the promotion of MT assembly by PCRG-1 relies on the presence of CFAP-20. To address this question, we expressed pcrg-1 in cfap-20::Scarlet animals. By examining fluorescence in 29 worms from two independent transgenic lines, we found that PCRG-1 specifically localizes to the distal segment whereas CFAP-20 remains restricted to the middle segment in all examined animals (SI Appendix, Fig. S6B). We next expressed pcrg-1 in cfap-20 mutant animals, and we did not detect any difference in the PCRG-1 fluorescence between WT and cfap-20 mutant animals, which suggests that the function of PCRG-1 does not depend on CFAP-20 (SI Appendix, Fig. S6 C and D).

To examine the impacts of pcrg-1 EE on ciliary formation and function, we measured the cilium length and performed the IFT assay. Our results revealed that the expression of PCRG-1 decreased the cilium length (Figs. 4B and 5A and SI Appendix, Fig. S6 B and G), and reduced the IFT velocities in both the middle and distal segments (Fig. 5 B–D). By introducing KLP-20::GFP KI into animals ectopically expressing PCRG-1, we did not detect any KLP-20 fluorescence in the ciliary distal segments, indicating that the decreased IFT velocity in the distal is not resulted from abnormal entrance of kinesin-II into this region (SI Appendix, Fig. S6 E and F).

Fig. 5.

EE of pcrg-1 reduces cilium length and IFT velocities, and disrupted animal Osm behavior. (A) The cilium length (mean ± SD) in WT, cfap-20 overexpression (OE), and pcrg-1 EE animals. n.s., not significant; ***P < 0.001. (B) Kymographs show the anterograde and retrograde movement of OSM-6::GFP in the middle segment and distal segment. Representative particle traces are marked with blue, orange, and green lines. The scale bars represent 5 μm (horizontal) and 5 s (vertical). (C) Histogram of OSM-6::GFP velocities. (Top) anterograde IFT along the middle segments (Antero. m.s.). (Middle) anterograde IFT along the distal segments (Antero. d.s.). (Bottom) retrograde IFT (Retro.). Each plot was fitted by a Gaussian distribution. (D) IFT velocities summary of OSM-6::GFP in WT, cfap-20 OE, and pcrg-1 EE animals. *P < 0.05; ***P < 0.001. m.s.: middle segment; d.s.: distal segment. (E) Analysis of the Osm behavior in the WT, osm-3 null, cfap-20 overexpression (OE), and pcrg-1 EE animals. N = 5 trials with 30 to 35 animals tested per trial. n.s., not significant, ***P < 0.001 by Student’s t test. Error bars represent SEM. (F) Analysis of the Osm behavior in the WT, osm-3 null, cfap-20 null, and pcrg-1 null animals. N = 3 trials with 30 to 35 animals tested per trial. *p < 0.05, ***P < 0.001 by Student’s t test. Error bars represent SEM.

With PCRG-1 EE, a subset of the amphid channel cilia developed abnormal doublets in the ciliary distal, likely impairing sensory perception or signal transduction, thereby resulting in the observed behavioral changes. According to our Osm test that measures the animal's ability to perceive environmental cues, most wild-type animals escaped from osmotic stress, but about 30% of the PCRG-1 EE animals failed to do so, comparable to the phenotype of the Osm variant osm-3(p802) null allele (Fig. 5E). Intriguingly, we found that the overexpression of CFAP-20 in sensory cilia had no apparent effect on ciliary structure or function, disfavoring the possibility that an excessive amount of the inner junction protein affects cilia. We also found that cfap-20 mutant displayed Osm defect (Fig. 5F). These results suggest that modulation of pcrg-1 gene expression contributes to the differentiation of axonemes into subdomains with distinct ultrastructure and physiology.

Discussion

This study shows that the inner junction protein CFAP-20 and PCRG-1, originally discovered from the motile cilia of Chlamydomonas, are employed differently to promote axoneme differentiation in C. elegans sensory cilia. Based on our data and recent findings in C. elegans (36), CFAP-20 appears to be the sole inner junction protein that bridges the A and B tubules in the axoneme of the ciliary middle segment. Conversely, PCRG-1 does not enter cilia in most neurons of WT animals, and EE of pcrg-1 in sensory neurons abnormally introduced this protein into cilia, which allows B-tubule extension into the distal segment, thereby forming abnormal MT doublets and inhibiting the sensory function. These results show that PCRG-1 absence from sensory cilia is required for developing two distinct ciliary segments characteristic of the MT doublets or singlets. Along with the early findings of how the OSM-3 kinesin and its regulators extend A tubules into the distal segment, this work offers a glimpse of how to inhibit B-tubule elongation, which is also essential for assembling MT singlets in the distal, providing additional molecular insights into understanding axoneme differentiation.

Why do motile cilia and sensory cilia develop different axoneme MT assemblies? Motile cilia perform cell motility, which is resulted from the sliding of axonemal dynein and axonemal MTs, and the mechanical force leads to the MT bending. Thus, axoneme stability from the MT doublets is perhaps higher than that from the singlets. As demonstrated from early studies, the mutant axonemes defective in the inner junction have weakened inter-doublet connections, and flagellar beating caused these axonemes liable to disintegrate (35). In addition to stabilizing the A-B tubule bridge, recent studies show that PACRG and FAP20 are also involved in the assembly of inner-arm dynein and the beak-MIP structures and suggest the formation of an inner junction hub that modulates dynein-driven MT sliding and assembles additional components to coordinate flagellar beating (31).

The immotile sensory cilia harbor signaling receptor molecules on the ciliary membrane, and the longer cilia likely correlate with higher sensory capacity. However, the cilia length cannot be unlimited, and different ciliary length control models assume that the amount of ciliary tubulin, the axoneme building block, is an essential factor for length control. From this point of view, the formation of singlets may only use about half the amount of tubulin molecules as the building block and can form longer cilia than forming doublets, perhaps increasing sensory capacity. For example, the frog olfactory cilia develop a 20-micron-long middle segment with doublets but extend another 200-micron-long distal segment (11). On the other hand, the MT singlets must be more dynamic than doublets, making ciliary turnover more efficient in the distal segment, which is probably also important for ciliary signaling processing and transmission.

The use of only one inner junction protein CFAP-20 in sensory cilia may have advantages to build two distinct ciliary domains. Our cryo-ET, genetic, and TEM data suggest that sensory cilia may have CFAP-20 alternatively arranged to bridge A and B tubules in the doublets along the middle segment. Why such a pattern cannot be continued into the distal segment? We speculate on two possibilities. One possibility is the lack of PCRG-1 in sensory cilia. AlphaFold2 predicts that the C. elegans CFAP-20 and PCRG-1, similar to their Chlamydomonas homologs can form a heterodimer that interacts with tubulin dimers (Fig. 1F). If both proteins were expressed in sensory cilia, the formation of distal doublets had been detected (Fig. 4D). The other nonexclusive possibility is that CFAP-20 might not be delivered or even diffused into the distal. Upon overexpression of PCRG-1::GFP in CFAP-20::Scarlet knock-in animals, CFAP-20::Scarlet remained restricted to the middle segment (SI Appendix, Fig. S6B), indicating that animals that gain ectopic PCRG-1 in the sensory cilia and form abnormal distal doublets have very little CFAP-20 signal in ciliary distal segments, indicating that the ciliary entrance of CFAP-20 is probably under tight control. Overexpression of CFAP-20::GFP led to a reduction in the entry of endogenous CFAP-20::Scarlet into the cilia, suggesting that the quantity of ciliary CFAP-20 is regulated (SI Appendix, Fig. S6H).

How can the EE of pcrg-1 promote B-tubule extension of doublet MTs? As shown in SI Appendix, Fig. S6B, the ectopically expressed PCRG-1 protein enters the ciliary distal segments, but CFAP-20 does not. The lack of colocalization of these two proteins indicates that they do not interact in the distal segments that form doublets. We argue that PCRG-1 can promote doublet formation in the ciliary distal segment without interacting with CFAP-20. We cannot exclude the possibility that ectopic PCRG-1 interacts with CFAP-20 in the ciliary middle segments because of their overlapped fluorescence; however, CFAP-20 alone is sufficient to build doublets in the middle segments without PCRG-1 in WT cilia. In support of these imaging and genetic results, the in vitro biochemical data show that the human PACRG proteins bind to MTs and recruit tubulin (50), suggesting that PCRG-1 alone may promote axoneme doublet formation.

In addition to connecting MT doublets, C. elegans CFAP-20 may play additional roles. We observed a small but statistically significant increase in anterograde IFT velocity in the ciliary middle segment in cfap-20 mutant cilia. Considering that the anterograde IFT trains move on the outside of the doublet in Chlamydomonas (51), and that kinesin-II and OSM-3 undergo sidestepping on the doublet in C. elegans middle ciliary segments (52), the observed increase can be explained that incomplete doublets may prevent the IFT-kinesins from sidestepping, thereby increasing the IFT velocity in the ciliary middle segment of cfap-20 mutant. This suggests that CFAP-20 protein may modulate the motility of the fast-moving OSM-3 kinesin to a modest level or slow the handover of IFT-particles from the slow-moving kinesin II to OSM-3. This additional role is in line with its dynamic property: Our FRAP and photoconversion results indicate that CFAP-20 has a higher dynamic turnover rate than axoneme MTs. Similarly, PCRG-1 may also have other roles than promoting B-tubule extension. We detected a marked decrease in anterograde IFT speeds in both middle and distal segments in PCRG-1 ectopic gained cilia, which suggests that PCRG-1 down-regulates the motility of both IFT-kinesins. In either case, we cannot exclude the possibility that the altered motor velocities have resulted from changed MT tracks rather than the direct consequences from the inner junction proteins. Nevertheless, CFAP-20 and PCRG-1 have recently been proposed to form an inner junction harbor that not only connects A and B tubules in the doublets but also perhaps has additional functions essential for the physiology of different types of cilia.

Materials and Methods

C. elegans Strains and Culture.

C. elegans strains were cultured on nematode growth medium (NGM) plates seeded with Escherichia coli strain OP50 at 20°C. All animal experiments were conducted in compliance with governmental and institutional guidelines.

The following C. elegans strains were ordered from SunyBiotech (https://www.sunybiotech.com): cfap-20 (syb3841), pcrg-1 (syb3843) and cfap-20::Scarlet (PHX4746). The cfap-20 (syb3841) was generated by CRISPR-Cas9 and deleted 842 bp of genomic DNA in the N-terminal region, which contains the 335-bp promoter, the first exon, and a portion of the first intron. The pcrg-1 (syb3841) was also generated by CRISPR-Cas9 and deleted 795 bp of genomic DNA in the N-terminal region, which includes the first to fifth exons. The cfap-20::Scarlet (PHX4746) was generated by CRISPR-Cas9 to label the endogenous cfap-20 using the red fluorescent protein Scarlet with a GGGGTS linker. A summary of all strains and plasmids used in this study is provided in SI Appendix, Table S1.

Molecular Biology.

The CFAP-20::Maple3 expression fragment was generated by SOEing PCR of the cfap-20 genomic sequence (including a 1,014-bp promoter and cfap-20 coding sequence) with Maple3 and linked by an AGSGSG linker. In the constructs shown in Fig. 4A and SI Appendix, Fig. S5 A, a 458 bp dyf-1 promoter was used. The coding sequences for cfap-20 and pcrg-1 were PCR amplified from genomic DNA. Plasmids were constructed by assembling the dyf-1 promoter, cfap-20 or pcrg-1 coding sequence, and the fluorescent protein into the pDONR vector via the In-Fusion Advantage PCR cloning kit (Clontech, cat. no. 639648).

Microinjection and Transgenesis.

Plasmids for microinjection were prepared with the AxyPrep Plasmid Purification Miniprep Kit (Axygen, #AP-MN-P-250). The PCR product and plasmids were purified with the QIAquick PCR purification kit (Qiagen, 28104) and coinjected into the gonads of young adult worms with rol-6 (su1006) plasmid, except for worms used in the Osm assay. The concentration of each DNA construct for microinjection was 50 ng/μL. Microinjections were performed in either the wild-type N2 or cfap-20::Scarlet (PHX4746) strain, depending on the experiment. F1 progenies with the rol phenotype were selected to test the transmission rate. For worms used in the Osm assay, plasmids were injected without rol-6 marker, and transgenic individuals were selected based on the fluorescent reporter under the fluorescent dissecting microscope (Olympus, MVX10). At least two independent transgenic lines with a transmission rate above 50% were used for subsequent experiments. For the EE of pcrg-1, we also used the MosSC1 method to integrate single-copy transgene as previous described (53).

Live-Cell Imaging.

Young adult C. elegans hermaphrodites were anesthetized with 0.1 mmol/L levamisole in M9 buffer, mounted on 3% agar pads, and maintained at 20°C. Agar pads were prepared 1 d before the experiment to ensure better immobilization of the worms. Images, excluding those acquired in FRAP and photoconversion assays, were captured using a spinning disk confocal system, which contains an Axio Observer Z1 microscope (Carl Zeiss) equipped with a 100×, 1.46 NA objective lens, an EMCCD camera (iXon+ DU-897D-C00-#BV-500; Andor Technology), and the 405-nm, 488-nm and 568-nm lines of a Sapphire CW CDRH USB Laser System (Coherent) with a spinning disk confocal scan head (CSU-X1 Spinning Disk Unit; Yokogawa Electric Corporation). The confocal system was controlled by μManager (https://www.micro-manager.org), and time-lapse images were obtained using a 200-ms exposure time.

Image Processing and Analysis.

The images were processed using the ImageJ software (http://rsbweb.nih.gov/ij/) to generate kymographs and quantify cilium length and IFT velocity. To measure cilium length and middle segment length, z-stack images were acquired and processed by maximal intensity projection. We used various ciliary markers (DYF-11::3×GFP, OSM-6::GFP, and 3×GFP::CHE-3) for cilium length measurement and CFAP-20::Scarlet fluorescence to determine the middle segment length. To analyze IFT velocity, we employed the established method as previously described (23).

Cryo-ET of C. elegans.

Cryo-ET was performed according to our previous protocol (49). In short, C. elegans L1 larvae were suspended in the M9 buffer and dropped onto a glow-discharged cryo-EM grid. The grid was then immediately transferred into the 100-μm-deep cavity of 6-mm aluminum carriers (Beijing Wulundes Biotech Ltd.) filled with 2-methyl pentane (Sigma, cat. no. M65807), covered with 0.12-mm-thick sapphire discs, and frozen by HPM100 HPF machine (Leica Microsystems). The assembly was transferred to FC6 cryo-ultramicrotome chamber (Leica Microsystems) at −150 °C to wait for the filler to sublimate. The cryo-EM grid was then picked out from the assembly and stored in liquid nitrogen. Cryo-FIB milling was performed using a Helios NanoLab G3 UC dual-beam microscope (Thermo Fisher Scientific Co.) equipped with the Quorum PP3010T cryo-transfer system. Rough milling of the parallel pattern from two sides was performed by using an ion-beam current of 2.5 to ∼0.23 nA until the lamella thickness was ∼1 to 2 μm. An ion-beam current of 80 to ∼24 pA was used for fine polishing until the lamella’s thickness was about 150 to 200 nm. Scanningelectron microscopy images were acquired at the accelerating voltage of 2 kV and beam current of 0.2 nA by an Everhart–Thornley Detector. After milling, the grid was transferred to 300 KV Titan Krios Microscopy (Thermo Fisher Scientific Co.) equipped with a Cs corrector, GIF Quantum energy filter (Gatan), and K2 Summit direct electron detector (Gatan) to collect Cryo-ET data. All the tilt series were recorded from 60° to −60° with the SerialEM software. The recording state was at a nominal magnification of 33,000× in counting mode with a pixel size of 3.421 Å/pixel. Each stack was exposed for 2.4 s with an exposure time of 0.3 s per frame and recorded as a movie of eight frames, resulting in the total dose rate of ∼1.927 electrons per Å2 for each stack. The defocus ranged from −3 to −7 μm.

TEM of C. elegans.

TEM experiment was performed as an early protocol (52). Briefly, adult worms were loaded onto a 50-μm deep specimen carrier with a pipette and frozen with HPM100 HPF machine (Leica Microsystems). Frozen specimens were then transferred into cryovials containing 1-mL fixative comprised of 1% osmium tetroxide and 0.1% uranyl acetate (Electron Microscopy Sciences), and processed in an AFS2 machine (Leica Microsystems) using a standard substitution and fixation program: ~90 °C for 48 h, ~60 °C for 24 h, ~30 °C for 18 h, and finally to 4 °C. Fixed specimens were washed three times with pure acetone and infiltrated with SPI-PON 812 resin (Structure Probe Inc.), and were embedded in a flat mold and polymerized at 60 °C for 3 d. Then, 90-nm ultrathin sections were obtained with UC7 Ultramicrotome (Leica Microsystems) and picked on 200 mesh copper grids. Sections were poststained with 2% uranyl acetate and Reynold’s lead citrate and imaged on FEI Tecnai G2 Spirit (120 kV) electron microscope (Thermo Fisher Scientific Co).

FRAP and Photoconversion Assay.

FRAP and photoconversion assays were performed using an Airyscan confocal microscope (LSM900, Carl Zeiss) equipped with a 63 × 1.4 NA objective. For the FRAP experiment in Fig. 3J, a 561-nm laser at 100% power was used to photobleach the selected region, and images were acquired at 0.5% power every 10 s. For the FRAP experiment in SI Appendix, Fig. S4E, the same power laser was used for photobleaching, but images were acquired with a 488-nm laser and a 561-nm laser both at 0.5% power every 272 ms to visualize the IFT movement. The data were processed as in the previous study (25). Fluorescence intensity was normalized to the fluorescence before bleaching. The recovery curve was fitted using an exponential equation: F(t) = F₀ +(F_inf − F₀)(1−e^−kt), where F(t) represents the total fluorescence at time t after the bleaching, k is a constant describing the rate of recovery, F₀ is the fluorescence immediately after the bleaching, and F_inf is the maximum recovered fluorescence. The recovery half-time was calculated by t_1/2 = ln2/k.

For the photoconversion assay, a 405-nm laser was used to illuminate the selected region to convert the Maple3 protein. And images were required with a 488-nm laser and a 561-nm laser both at 0.5% power every 10 s. To quantify the photoconversion data, the red fluorescence intensity ratio of the unconverted/converted middle segment region at the same time point was used. The curve was fitted similarly to the FRAP assay.

Osm Assay.

The assay was carried out according to the protocol established before (52). The osmotic barrier ring was printed at the center of the plate by using a 15-mL centrifuge tube. The open end of the tube was immersed in a 4 M NaCl solution and then transferred to the NGM plate to allow the solution to soak into the medium. After 5 to 10 min, worms were washed in M9 buffer three times and transferred to the center of the osmotic ring. About 30 worms were tested in each parallel experiment. The trial started when the worms began to move. After 10 min, worms that escaped the osmotic ring were considered nonavoiders. The ratio of nonavoider/total worms was used to score each trial.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.