Chlamydomonas protein kinase MAK phosphorylates FAP256/CEP104 and regulates axonemal microtubule assembly

Significance

The formation of cilia depends on the assembly of axonemal microtubules, the core structure of the organelle. However, the mechanisms controlling this process remain poorly understood. Here, we show that a Chlamydomonas kinase (CrMAK), homologous to mammalian ICK and MAK, regulates this assembly and, consequently, cilia formation. CrMAK phosphorylates CEP104/FAP256, a microtubule plus-end tracking protein with localization at the ciliary tip, to modulate microtubule dynamics. Our findings reveal that axonemal microtubule assembly is tightly controlled during ciliogenesis in Chlamydomonas, with implications of similar mechanisms in other organisms.

Abstract

The formation of cilia involves the assembly of axonemal microtubules, the structural core of the organelle, from tubulins that are delivered by intraflagellar transport and diffusion. Whether and how this process is controlled are not clear. Here, we show that Chlamydomonas MAK (CrMAK), related to mammalian ICK/MAK, regulates axonemal microtubule assembly by phosphorylating FAP256/CEP104, a microtubule plus-end tracking protein. CrMAK, primarily present in cilia, associates with the axoneme via its C-terminal tail, and is enriched at the ciliary tip during active ciliary assembly. It is constitutively active during this process and acts downstream of a ciliary length-regulatory complex composed of LF1, LF3, and LF2, a CCRK homologue. Loss of CrMAK or its kinase activity leads to aciliated cells. CrMAK interacts with and phosphorylates FAP256/CEP104 and dephosphorylation of FAP256/CEP104 impairs ciliary assembly. These findings reveal a molecular mechanism governing the assembly of axonemal microtubules.

Cilia are ubiquitous microtubule-based cellular organelles with essential functions in cellular motility and signaling. The microtubule core of the cilium, known as the axoneme, consists of nine outer doublet microtubules with (9 + 2) or without (9 + 0) a central pair of microtubules. During de novo ciliogenesis, the docking of the basal body to ciliary vesicles or the apical plasma membrane triggers the establishment of the transition zone (TZ), a ciliary gate that controls the entry and exit of ciliary proteins, which is followed by the assembly of the cilium proper (1–4). The extension of the doublet microtubules at the TZ gives rise to the outer doublet microtubules of the axoneme (1). Additionally, cilia can regenerate after deciliation in which axonemal microtubules are severed distal to the TZ (5, 6). However, the mechanisms regulating axonemal microtubule assembly during ciliogenesis remain poorly understood.

The assembly of the axonemal microtubules requires tubulins that are delivered by intraflagellar transport (IFT) and diffusion (7–10). Loss of FLA10, a subunit of the IFT anterograde motor kinesin-II, completely blocks microtubule assembly above the TZ (11), indicating that the assembly of axonemal microtubules involves distinct mechanisms from those operating at the TZ. Microtubule assembly can be regulated by microtubule-modulating kinesins and microtubule-associated proteins including microtubule plus-end tracking proteins (+TIPs) (12, 13). Several +TIPs have been found to be associated with ciliary tips and functions in ciliary assembly (14, 15). For example, CEP104/FAP256 is associated with the ciliary tip during ciliary assembly in Chlamydomonas and mammalian cells (16, 17). Depletion or null mutation of CEP104/FAP256 reduces the number of ciliated cells and/or cilia length in multiple organisms (16–20). In addition, CEP104/FAP256 mutations are associated with Joubert syndrome, a ciliopathy (21).

Protein kinase RCK (the tyrosine kinase gene v-ros cross-hybridizing kinase) family includes ICK, MAK, and MOK and acts downstream of CCRKs (cell cycle-related kinases) (22). RCKs have been found to regulate ciliogenesis either positively or negatively in a manner depending on cell types and/or organisms. Loss of DFY-18 CCRK or DYF-5 MAK kinases in Caenorhabditis elegans induces cilia elongation (23, 24). Depletion of CCRK or ICK in mammalian cell lines causes primary cilia elongation while overexpression of ICK or MAK leads to shorter cilia (25–28). Conditional knockout of MAK in the mouse leads to longer photoreceptor cilia (29). In contrast, ICK knockout in mice inhibits primary cilia formation in multiple cell types (30). Furthermore, simultaneous knockout of ICK and MAK in mice or knockout of MAK, the sole member of RCK in zebrafish, completely blocks formation of photoreceptor cilia (31, 32). These data suggest that in certain types of cells/organisms ICK and MAK are required for cilia formation; however, the underlying mechanism is not known.

In Chlamydomonas, LF2 (a CCRK), forms a ciliary length regulatory complex (LRC) with LF1 and LF3 to regulate ciliogenesis (33–35). Loss of LF2 or its kinase activity results in cells with stumpy cilia and/or unequal length cilia and the cilia in cells that have them fail to regenerate after deciliation (33). LF4, a homologue of MOK, acts downstream of LF2 and its loss causes extremely long cilia (36, 37). Our bioinformatic analysis shows that Chlamydomonas possesses another member of RCK, CrMAK (Chlamydomonas MAK), which is closely related to ICK/MAK (Fig. 1A). However, its function and working mechanism is not clear. In this study, we showed that CrMAK is essential for ciliogenesis and regulated by LRC. CrMAK associates with the axonemal microtubules via its C-terminal tail and interacts with and phosphorylates CEP104/FAP256 to regulate axonemal microtubule assembly.

Fig. 1.

Isolation and characterization of Chlamydomonas mak mutants. (A) Phylogenetic analysis of CCRKs and RCKs, constructed using Phylogeny (https://www.phylogeny.fr). The numbers indicate branch support values in percent. Hs, Homo sapiens; Mm, Mus musculus; Dr, Danio rerio; Ce, C. elegans; Cr, Chlamydomonas reinhardtii; Lmx, Leishmania mexicana. (B) Gene structure diagram of CrMAK, highlighting the AphVIII DNA fragment insertion site in mak-1 mutant. This insertion occurs between nucleotides 2495 and 2496. Black boxes, exons; lines, introns. (C) Representative differential interference contrast (DIC) images of WT, mak-1, and mak-1;CrMAK:HA-TG (rescue) cells. An average of 100 cells were counted in three experiments to determine the percentage of aciliated cells in the mak-1 mutant. (Scale bar, 5 μm.) (D) Immunoblot analysis of whole cell lysates from the indicated strains using antibodies against CrMAK, HA, and α-tubulin. Equal amounts of proteins (10 μg) were loaded. (E and F) TEM micrographs of WT and mak-1 mutant in longitudinal (E) and cross sections (F) showing defects of the mutant in axoneme formation above the TZ. (Scale bars, 100 nm.)

Results

Chlamydomonas MAK Is Essential for Cilia formation.

Blast search and phylogenetic analysis in Chlamydomonas has identified two homologues of RCKs: LF4 and MAK that are closely related to mammalian MOK and ICK/MAK, respectively (Fig. 1A). To study the function of Chlamydomonas MAK (hereafter CrMAK), we have isolated two insertional mak mutants (Fig. 1B and SI Appendix, Fig. S1A). The mutant cells, being palmelloid, had no cilia or only cilia buds as examined after cell release from the mother cell wall by autolysin treatment (Fig. 1C and SI Appendix, Fig. S1B). CrMAK encodes a protein of 585 residues with a predicted molecular weight of 63.33 kDa, which was not detectable by immunoblotting in both mutants (Fig. 1D and SI Appendix, Fig. S1C). Ectopic expression of CrMAK-HA in the mutants rescued the ciliary phenotype (Fig. 1 C and D and SI Appendix, Fig. S1 B and C), indicating that CrMAK is the causal gene for the defects in ciliogenesis. Further analysis of the mutants by transmission electron microscopy (TEM) showed that the assembly of axonemal microtubules above the TZ was disrupted, whereas the main structures of TZ and basal bodies appeared normal (Fig. 1 E and F). Please note, we have also observed singlet microtubules in TZ in the mutant (Fig. 1F), indicating that the assembly of microtubules in TZ may also be affected.

CrMAK Is Enriched in Cilia, Associated with the Axoneme, and Phosphorylated at the Kinase Activation Loop.

Immunostaining analysis showed that CrMAK-HA was localized in both cell body and cilia (Fig. 2A). CrMAK-HA colocalized with the cytoplasmic microtubules in the cell body. CrMAK-HA was also distributed along the length of cilia with enrichment at the ciliary tip and the ciliary base. To pinpoint the localization of CrMAK-HA at the ciliary base, deciliated cells were coimmunostained with anti-HA and anti-centrin antibodies followed by visualization through structured illumination microscopy (SIM) (Fig. 2B). Centrin showed a typical localization at the nucleus basal body connector and distal striated fibers that connect two cilia at the distal end of the basal bodies (38). CrMAK-HA exhibited two dots proximal to the striated fibers, indicating that it is associated with the basal body. Immunoblotting analysis of whole cells, cell bodies, and cilia with equal protein loading revealed that CrMAK-HA was highly enriched in cilia relative to the cell body (Fig. 2C). Further fractionation of isolated cilia showed that CrMAK was primarily associated with the axoneme (Fig. 2D). These data are consistent with its colocalization with cytoplasmic microtubules, indicating that CrMAK is a microtubule associated protein.

Fig. 2.

Cellular localization of CrMAK and pCrMAK. (A) Immunofluorescence staining of WT and mak-1;CrMAK:HA-TG cells using antibodies against HA and α-tubulin. (Scale bar, 5 μm.) (B) SIM of mak-1;CrMAK:HA-TG cells immunostained with antibodies against HA and centrin. Images shown are single optical sections. (Scale bar, 2 μm.) (C) CrMAK is enriched in cilia as revealed by immunoblotting of whole cells, cell bodies, and cilia with equal amounts of proteins being loaded. (D) CrMAK is primarily associated with the axoneme. Cilia were fractionated into membrane matrix (M + M) and axonemal fractions, followed by immunoblotting with equal proportions of samples loaded. FMG1 and RIB72 were used as markers for the M + M and axonemal fractions, respectively. (E and F) pCrMAK is enriched in cilia and the axonemal fractions. Immunoblot analysis of whole cells, cell bodies, and cilia (E) or cilia fractions (F) was performed as described in (D), using the indicated antibodies.

Phosphorylation of RCKs on the TxY motif at the kinase activation loop is essential for its full activity and can be used to assess the kinase activity (22, 39). Anti-pCrMAK, a rabbit polyclonal antibody against phosphorylated T156 at the activation loop, specifically recognized CrMAK with T156 phosphorylation (pCrMAK) (SI Appendix, Fig. S2). Immunoblot analysis revealed that pCrMAK was highly enriched in cilia and associated with the axoneme (Fig. 2 E and F), similarly to CrMAK. These data suggest that CrMAK in cilia exists as a phosphorylated and active kinase.

CrMAK Phosphorylation at the Kinase Activation Loop Requires Ciliary LRC.

Chlamydomonas possesses five LF (long-flagella) mutants, designated lf1 to lf5 (33–36, 40). LF2 is an ortholog of CCRK and forms LRC in the cytoplasm with LF1 and LF3, two novel proteins with unknown functions (33), and regulates the phosphorylation of LF4, an ortholog of MOK while LF5 is a CDK-like kinase (37, 40). We examined the levels of pCrMAK in lf1 to lf5 mutants by immunoblotting with anti-pCrMAK antibody (Fig. 3A). CDPK3, a calcium-dependent kinase (41), was used as a loading control. pCrMAK was not detected in lf1-1, lf2-6, and lf3-6 mutants though CrMAK was present in these cells, suggesting that CrMAK phosphorylation is tightly regulated by LRC.

Fig. 3.

Regulation of CrMAK phosphorylation by LF genes. (A) The LRC, composed of LF1, LF2, and LF3, mediates the phosphorylation of CrMAK. Cell lysates containing equal amounts of proteins from the indicated cells were analyzed by immunoblotting using the indicated antibodies. (B) Representative DIC images of the indicated cells. Please note that the double mutant lf4-3;mak-3 lacks cilia and forms palmelloids while ectopic expression of CrMAK-HA restores the long cilia phenotype of lf4-3. The percentages of cells forming clusters or bearing long cilia are reported as the average of three experiments, each with 100 cells counted. (Scale bar, 5 μm.) (C) Immunoblot analysis showing the loss of LF4 and CrMAK in lf4-3;mak-3 double mutants. Equal amounts of proteins from the cells shown were probed with the antibodies as indicated. An arrow marks the band of CrMAK-HA in DKO;CrMAK:HA-TG cells.

The lf2-6 null mutation has been reported to cause stumpy cilia and unequal ciliary length, including very short cilia (<2 μm) and longer cilia. Mutants predicted to reduce the kinase activity, such as K41R and G21V, are capable of partially rescuing the null phenotype. However, the double mutant fails to do so entirely, indicating that the kinase activity is essential for the protein’s function (33). Because CrMAK and LF4 are the only two RCKs in Chlamydomonas, and their mutants exhibited no cilia and long cilia, respectively (Fig. 3B), we hypothesized that the phenotype of lf2-6 might be due to the inactivation of both LF4 and CrMAK. To test this hypothesis, CrMAK was disrupted in lf4-3 mutants using CRISPR/Cas9 (SI Appendix, Fig. S3A). Immunoblot analysis confirmed that lf4-3;mak-3 double knockout (DKO) cells lost both LF4 and CrMAK (Fig. 3C). Unexpectedly, the DKO mutant cells did not form cilia, similarly to the mak-1 mutants (Fig. 3B). To further confirm this result, transformants that expressed CrMAK-HA in the DKO mutants exhibited the long cilia phenotype of lf4-3. These results suggest that LF2 has additional targets beyond LF4 and CrMAK.

The Kinase Activity of CrMAK and Its Association with Axonemal Microtubules Via the C-Terminal Tail Are Required for Cilia Assembly.

CrMAK possesses a N-terminal kinase domain, containing a conserved TxY motif as well as a conserved K33 residue predicted to be critical for ATP catalysis (SI Appendix, Fig. S2A), and a long C-terminal tail (Fig. 4A). To characterize the functions of the kinase activity and the C-terminal tail, we have generated CrMAK mutants with C-terminally tagged HA including kinase dead mutant (K33R), phosphorylation dead (PD) mutant (T156A), phosphorylation mimic mutant (T156E), and the C-terminal truncation mutant (ΔC), which were transformed into mak-1 mutant cells, respectively. Unlike wild-type CrMAK-HA, K33R mutant failed to rescue mak-1 ciliary phenotype and all the cells were aciliated (Fig. 4B). In vitro kinase assay confirmed that K33R mutant lost its kinase activity (Fig. 4C). However, T156 phosphorylation was not abolished (Fig. 4D), supporting that T156 was phosphorylated by an upstream kinase, namely LF2. As expected, T156E and T156A mutants were not phosphorylated at T156 (Fig. 4E). T156E but not T156A mutants rescued the mak-1 ciliary phenotype. All T156E mutant cells had normal cilia, while all T156A mutant cells were aciliated (Fig. 4B). These data indicate that the kinase activity of CrMAK rendered by T156 phosphorylation is required for its function.

Fig. 4.

The kinase activity and the C-terminal tail of CrMAK are required for ciliogenesis. (A) Domain organization of CrMAK and its mutants. The green box marks the kinase domain. Numbers indicate amino acid positions. (B) DIC images of WT and mak-1 mutants expressing the indicated mutant constructs. Note that cells expressing K33R, T156A, and ΔC formed palmelloids and were released from the mother cell wall by autolysin treatment. (Scale bar, 5 μm.) (C) K33R mutant lacks kinase activity as shown by in vitro kinase assay. Cell lysates from CrMAK-HA and CrMAK^K33R-HA cells were subjected to immunoprecipitation using anti-HA antibodies, followed by in vitro kinase assay. Phosphorylation was detected using an anti-thiophosphate ester antibody (anti-Thio). (D) Phosphorylation at the kinase activation loop does not require K33. Cell lysates with equal amounts of proteins from mak-1, CrMAK-HA, and CrMAK^K33R-HA cells were immunoblotted with anti-pCrMAK and anti-HA antibodies. *, nonspecific bands. (E) Immunoblot analysis of cell lysates from the indicated cells using the indicated antibodies. Note that T156E and T156A mutants were not recognized by the anti-pCrMAK antibody. (F and G) Immunoblot analysis of cell lysates from mak-1, CrMAK-HA, and CrMAK^ΔC-HA cells using anti-HA (F) and anti-pCrMAK (G) antibodies. (H) CrMAK^ΔC-HA is barely associated with the axoneme. Isolated cilia from CrMAK^ΔC-HA cells were fractionated into membrane-matrix and axonemal fractions, followed by immunoblotting. Equal proportions of the samples were loaded.

Expression of CrMAK^ΔC-HA failed to fully rescue mak-1 ciliary phenotype (Fig. 4 B and F). 94% of the cells completely lacked cilia, while the remaining 6% displayed short cilia (an average of 100 cells scored across three experiments). The average length of the short cilia was 3.81 μm ± 0.20 (an average of 50 cilia measured across three experiments). These observations suggest that the truncated form retains partial functionality. However, CrMAK^ΔC-HA was still phosphorylated at T156 (Fig. 4G). It was interesting how the C-terminal tail of CrMAK affects its ciliary function. Because CrMAK is a microtubule-associating protein (Fig. 2A), we speculated that the C-terminal tail might mediate the interaction of CrMAK with the axonemal microtubules. To test this, CrMAK^ΔC-HA was expressed in wild-type cells. As expected, immunoblotting of ciliary fractions showed that CrMAK^ΔC-HA was primarily present in the membrane matrix but not the axonemal fractions (Fig. 4H). These data suggest that CrMAK functions in ciliary assembly dependent on its kinase activity as well as association with the axonemal microtubules.

CrMAK Mediates Axonemal Microtubule Dynamics.

After deciliation, Chlamydomonas cilia are rapidly regenerated with a deceleratory kinetics and attain the original length within ~120 min (42, 43), enabling examination of the relationship between CrMAK and ciliogenesis. The level of pCrMAK did not change during cilia assembly (Fig. 5A), indicating that CrMAK is constitutively activated. Immunostaining analysis showed that CrMAK-HA became concentrated at the ciliary base after deciliation. At 15 min after deciliation, it was highly enriched at the ciliary tip and decreased at 30 min when cilia started to reduce growth rate (Fig. 5B and SI Appendix, Figs. S4 and S5), which suggests that CrMAK is intimately associated with the earlier stages of ciliogenesis.

Fig. 5.

CrMAK affects the dynamics of the axonemal microtubules. (A) CrMAK is constitutively phosphorylated at the activation loop during cilia regeneration. Cells were deciliated to allow cilia regeneration. Samples from predeciliated cells (pdc) and cells at the indicated times postdeciliation were subjected to immunoblotting. The pCrMAK-to-CrMAK ratio is shown below each lane, normalized to the levels in predeciliated cells. (B) Immunostaining of CrMAK-HA cells during cilia regeneration. (Scale bar, 5 μm.) (C) Immunoblot analysis of mak-1 and mak-1;CrMAK:HA-TG cells with varying levels of CrMAK expression. CDPK3, loading control. The ratio of CrMAK-HA to CDPK3 is shown below each lane, normalized to the levels in strain #14. (D and E) Quantification of ciliated cells (D) and ciliary length (E) of the indicated strains. Data are presented as mean ± SD, n = 3. For quantification of ciliated cells, an average of 200 cells were counted across three experiments. For ciliary length measurement, 50 cilia from 50 cells were assessed in each of three experiments. (F and G) Cabazitaxel treatment partially rescues the ciliary phenotype of mak-1. Shown are DIC images of mak-1 cells after treatment with cabazitaxel (F) and quantification of mak-1 and fla10-1 cells as a control (G). For each sample, an average of 100 cells were counted in three experiments. Data are presented as mean ± SD, n = 3. (H) TEM analysis of cilia of WT and mak-1 cells treated with cabazitaxel. (Scale bar, 100 nm.)

To learn more about how CrMAK regulates ciliogenesis, we analyzed ciliary phenotypes of mak-1 cells expressing different levels of CrMAK-HA (Fig. 5C). It was found that reducing the amount of CrMAK increased the number of aciliated cells while the cilia that were formed were of normal length (Fig. 5 D and E), indicating that CrMAK is critical in initiating ciliogenesis but not in regulating ciliary length.

Ciliary regeneration initiates with the assembly of axonemal microtubules, which requires tubulin delivery from the cytoplasm by IFT and/or diffusion (7–10). ICK in mammalian cells and DYF-5 in C. elegans have been implicated in IFT regulation in terms of IFT speeds and/or turnaround at the ciliary tip (23, 29, 30, 32, 44). In addition, in C. elegans, it was proposed that phosphorylation of IFT74 mediated by DYF-5 is involved in tubulin loading and unloading, thereby regulating ciliogenesis (45). Thus, it raises a possibility that CrMAK functions through IFT to control initiation of ciliogenesis. We found that loss of CrMAK appeared not to disrupt IFT complex formation as revealed by sucrose gradient analysis combined with immunoblotting (SI Appendix, Fig. S6A). Immunostaining analysis revealed no apparent difference in the enrichment of IFT proteins at the ciliary base between the mutant and wild-type cells (SI Appendix, Fig. S6B). Furthermore, IFT74 phosphorylation was not altered in mak-1 mutants as analyzed by immunoblotting with an anti-pan phospho-serine/threonine antibody (SI Appendix, Fig. S6C). These data indicate that CrMAK is unlikely in regulating IFT for the initiation of ciliogenesis.

DYF-5 in C. elegans and MAK in mice have a function in the stability of axonemal microtubules (24, 29). We speculate that CrMAK may also modulate the dynamics of axonemal microtubules. To test this, mak-1 mutant cells were treated with cabazitaxel. As expected, this treatment induced cilia formation though the axonemal microtubules were improperly assembled as shown by TEM (Fig. 5 F–H). In contrast, treatment of fla10-2 mutant, which lacks the IFT motor subunit FLA10/KIF3A (11), barely induced cilia formation (Fig. 5G). These data suggest that in the absence of CrMAK axonemal microtubules can be assembled via tubulins delivered by IFT but are unstable to sustain cilia growth.

CrMAK Phosphorylates FAP256 to Regulate Ciliogenesis.

The effect of CrMAK on microtubule dynamics is likely mediated by its substrates and/or interacting partners. An attempt with immunoprecipitation did not identify obvious candidates, likely due to transient and/or weak interactions between a protein kinase and its partners. The ascorbate peroxidase 2 (APEX2)-based proximity labeling strategy facilitates the identification of transient protein interactions (46). A codon optimized APEX2 fused to the C-terminus of CrMAK-HA effectively rescued the cilia phenotype of mak-1 cells, as confirmed by light microscopy. Thus, the mak-1;CrMAK:HA:APEX2-TG strain was used for proximity labeling with the strain mak-1;CrMAK:HA as a control. To facilitate identification of the interacting proteins by mass-spectrometry, we employed comparative proteomics with ¹⁵N labeling (47). Cells from the APEX2 strain were labeled with ¹⁵N by culturing the cells in medium supplemented with ¹⁵NH4¹⁵NO3 as the sole nitrogen source while the control cells were labeled with ¹⁴N or vice versa. After proximity labeling, equal amounts of proteins from the APEX2 and control cells were mixed, followed by affinity purification and mass spectrometry.

Among the proteins identified in two experiments, FAP256/CEP104 was the only known microtubule-associated protein implicated in the regulation of microtubule dynamics. (SI Appendix, Table S1). It contains a tubulin-binding TOG (tumor overexpressed gene) domain, which has microtubule-polymerizing activity and tracks microtubule plus end in vitro (17, 19, 48). Its depletion or genetic inactivation results in defects in ciliogenesis in mammalian cells, zebrafish, Tetrahymena, and Chlamydomonas (16–20). The majority of cells from roc22, an insertional mutant of FAP256 that could form cilia in steady state cells, is defective in regenerating cilia after deciliation (16). To confirm this finding, we generated three fap256 mutants at three different genomic loci using CRISPR/Cas9 techniques (SI Appendix, Fig. S3B). In contrast to the wild-type, cilia formation was not observed in all the three mutants at 15 min after deciliation. Although the number of mutant cells forming cilia gradually increased after 30 min and about 30% of cells still could not form cilia even at 180 min (SI Appendix, Fig. S7A). Ectopically expression of FAP256-HA in the mutants rescued the ciliary phenotype (SI Appendix, Fig. S7A), indicating that the ciliary defects are caused by loss of FAP256. Consistent with previous findings (16), FAP256-HA localized to the basal bodies as well as at the ciliary tip in full-length and growing cilia as shown by immunofluorescence microscopy (SI Appendix, Fig. S7 B and C).

To demonstrate whether FAP256 interacts with CrMAK, FAP256-HA was immunoprecipitated with anti-HA antibody followed by immunoblotting with anti-HA and anti-CrMAK antibodies. As expected, CrMAK was coimmunoprecipitated (Fig. 6A). Next, we examined whether FAP256 was phosphorylated by CrMAK. CrMAK was knocked out in the strain fap256-1;FAP256:HA-TG (SI Appendix, Fig. S3C). To examine the phosphorylation status of FAP256, we employed phos-tag SDS-PAGE, followed by immunoblotting, which could detect protein phosphorylation based on gel-shift (49, 50). It showed that the loss of CrMAK resulted in dephosphorylation of FAP256-HA (Fig. 6B), a phenomenon consistent with the effects observed following protein phosphatase treatment in wild-type cells (Fig. 6C). These data suggest that FAP256 is a substrate of CrMAK.

Fig. 6.

CrMAK phosphorylates FAP256 to regulate ciliogenesis. (A) CrMAK interacts with FAP256. Cell lysates (input) and anti-HA immunoprecipitates from the indicated cells were analyzed by immunoblotting using anti-HA and CrMAK antibodies. (B) Dephosphorylation of FAP256-HA following the loss of CrMAK. Cell lysates from the indicated cells were analyzed by phospho-tag immunoblotting. IC2, an intermediate chain of axonemal dynein, was used as a loading control. (C) Phosphatase treatment results in FAP256-HA dephosphorylation in the WT background. Cell lysates from the indicated cells, with or without phosphatase treatment, were analyzed by phospho-tag immunoblotting. (D) CrMAK directly phosphorylates FAP256. Bacterially expressed FAP256 and FAP256PD were incubated with or without CrMAK in an in vitro kinase assay, followed by Phos-tag SDS-PAGE and Coomassie Blue staining. (E) FAP256PD-HA migrates similarly to the dephosphorylated form of FAP256-HA, as shown by phospho-tag immunoblotting. (F) Quantification of ciliation rate of the indicated cells during cilia regeneration. Cells were deciliated by pH shock to allow ciliary assembly. Predeciliated cells (pdc) and cells at different times postdeciliation were analyzed. Data are presented as mean ± SD, n = 3. An average of 200 cells were counted across three experiments at each time point for each strain. (G and H) FAP256-HA is diminished at the ciliary base in the absence of CrMAK. The cells shown were immunostained with the indicated antibodies (G; Scale bar, 5 μm) and analyzed by immunoblotting with equal protein amounts (H). Note, cells in WT background were deciliated for immunostaining to allow comparison with aciliated mak mutant cells. (I–K) FAP256PD-HA is reduced at the ciliary base and tip. The cells shown were immunostained with the indicated antibodies (I; Scale bar, 5 μm), statistically analyzed (J), and analyzed by immunoblotting with equal protein amounts (K). Data in (J) are mean ± SD. ***P < 0.001.

A preliminary comparative phosphoproteomics had identified that four residues (S356, S383, S386, T406) of FAP256 were unphosphorylated in mak-1 mutant. These four residues plus another four conserved serine/threonine residues (T75, S104, S111, and S120) were all mutated to alanine to make a PD mutant (SI Appendix, Fig. S8). To determine whether FAP256 is directly phosphorylated by CrMAK and whether these residues are involved in phosphorylation mediated by CrMAK, we performed an in vitro phosphorylation assay using bacterially purified proteins. As expected, CrMAK phosphorylated FAP256, but not the PD mutant (Fig. 6D). FAP256PD-HA expressed in fap256-1 mutants had a lower molecular weight that is comparable to that of FAP256-HA following the loss of CrMAK (Fig. 6E), which further demonstrates that FAP256PD likely represents the unphosphorylated form in mak mutants. Examination of ciliary regeneration after deciliation revealed that FAP256PD cells had a similar phenotype to fap256-1 mutant (Fig. 6F), suggesting that FAP256 phosphorylation mediated by CrMAK is required for cilia regeneration. To learn whether FAP256 phosphorylation could rescue the aciliated phenotype of the mak mutant, all eight predicted phosphorylation sites were replaced with phosphomimetic substitutions (Ser to Asp; Thr to Glu) to create a phosphomimetic variant, FAP256PM. Although FAP256PM was successfully expressed, it failed to restore cilia formation in mak-1 cells while expression of FAP256PM could rescue the phenotype of fap256-1 (SI Appendix, Fig. S9). Taken together, we have shown that FAP256 is a direct substrate of CrMAK and its phosphorylation mediates its function during cilia regeneration.

The next question is how loss of phosphorylation of FAP256 perturbs its function. First, we examined the cellular localization of FAP256-HA after loss of MAK by immunostaining. In contrast to the control, FAP256-HA was barely detectable at the ciliary base in the absence of CrMAK although the protein levels were similar in the two cell types (Fig. 6 G and H). Similarly, the PD mutant of FAP256-HA was diminished at both ciliary base and ciliary tip in contrast to the control (Fig. 6 I–K). Thus, we conclude that CrMAK phosphorylates FAP256 to regulate its proper cellular localization, particularly at the ciliary tip, thereby controlling ciliogenesis.

Discussion

Cilia formation depends on the assembly of axonemal microtubules, which initiates from the doublet microtubules at the TZ. Whether and how this process is controlled is not clear. In this work, we show in Chlamydomonas that this assembly requires CrMAK, a kinase related to ICK/MAK in mammalian cells. Furthermore, we reveal that FAP256/CEP104, a microtubule plus-end tracking protein enriched at the ciliary tip, is phosphorylation target of CrMAK. Loss of CrMAK or disruption of CrMAK mediated FAP256/CEP104 phosphorylation leads to mislocalization of FAP256/CEP104 from the ciliary tip and impairs ciliogenesis. Thus, this work provides molecular mechanisms underlying the assembly of axonemal microtubules that is coupled with ciliogenesis.

Following the delivery of tubulins by IFT and diffusion to the distal end of the TZ (8, 10, 11), axonemal microtubules are assembled. We demonstrate that genetic inactivation of CrMAK severely disrupts this process, leading to failed cilia formation. Taxols stabilize microtubule nuclei and lower the critical concentration of tubulins for microtubule polymerization (51). Treatment of mak mutants with taxol stimulates formation of axonemal microtubules and cilia, indicating that CrMAK functions in regulating microtubule dynamics and assembly. It also associates with ciliary tips and mediates ciliary tip localization of FAP256/CEP104. Thus, this finding highlights that axonemal microtubule assembly is tightly regulated. Loss of MAK in zebrafish and double knockout of MAK and ICK in mice completely block the assembly of photoreceptor cilia (31, 32). Furthermore, knockout of ICK in mice inhibits primary cilia formation in multiple tissues (30). It is proposed that the homologs of CrMAK may utilize similar mechanisms in mammalian cells and zebrafish.

We have identified FAP256/CEP104 as a substrate of CrMAK. FAP256/CEP104 is a microtubule plus-end tracking protein that contains a tubulin binding TOG (tumor overexpressed gene) domain and possesses microtubule polymerizing activity (17, 19, 48). It also associates ciliary tips in Chlamydomonas, Tetrahymena, and mammalian cells (16, 19, 20). Consistent with a previous study in Chlamydomonas (16), FAP256/CEP104 knockout results in failed cilia formation in significant number of cells during cilia regeneration. We further showed that FAP256/CEP104 is phosphorylated by CrMAK and the PD mutant of FAP256/CEP104 exhibits a similar phenotype, indicating that the function of FAP256/CEP104 depends on its phosphorylation mediated by CrMAK. In mammalian cells and Tetrahymena, genetic inactivation of FAP256/CEP104 also reduces the number of ciliated cells (16–18, 20). Thus, it is possible that FAP256/CEP104 is regulated in a similar manner in these organisms while this hypothesis remains to be tested. Loss of CrMAK or loss of phosphorylation of FAP256/CEP104 abolishes the enrichment of FAP256/CEP104 at the ciliary tip as well as the basal body. The underlying mechanisms are not known. One speculation is that the interaction of FAP256/CEP104 with microtubules and/or microtubule associated proteins may be affected by its phosphorylation.

We showed that at relatively low levels of CrMAK expression, the portion of cells capable of forming cilia exhibit normal ciliary length, suggesting that CrMAK is critical in the initiation of cilia formation. Given that CrMAK is implicated in regulating microtubule dynamics and stability, we propose that during the early stages of ciliogenesis, CrMAK modulates microtubule regulatory factors at the site of axonemal microtubule assembly to facilitate axoneme formation. Though we showed that FAP256/CEP104 is a substrate of CrMAK; however, knockout of FAP256/CEP104 does not fully copy the ciliary phenotype of mak mutants. mak mutants are basically aciliated. In contrast, fap256/cep104 mutants are only defective in cilia formation during cilia regeneration, implying the presence of additional targets of CrMAK. Because CrMAK regulates axonemal microtubule dynamics and requires its association with the axoneme, its target may include other axonemal microtubule-associated proteins especially those with enrichment at the ciliary tip. EB1 is localized to the tip of Chlamydomonas cilia although its function is not clear (52). In mammalian cells, depletion of EB1 or EB3 suppresses primary cilia formation (53, 54). Interestingly, loss of DYF-5 in C. elegans results in EB protein decoration along the axonemal microtubules accompanied with cilia elongation (24). Thus, should CrMAK regulate EB proteins, it would act in a different mode from that of DYF-5. Three additional microtubule-associated proteins, namely CSPP1, Crescerin, and ARMC9, have been shown to localize to the ciliary tips (14). Our blast search has failed to identify homologues of CSPP1 and ARMC9 in Chlamydomonas. A Crescerin homologue SHF1 in Chlamydomonas, whose null mutant exhibits short cilia, and is proposed to bind tubulin dimers and transport them into cilium for cilia assembly (55). Future work is necessary to identify additional substrates of CrMAK beyond FAP256/CEP104 for a deeper understanding of the assembly of axonemal microtubules.

CCRK acts upstream of RCKs that include three members including ICK, MAK, and MOK (22). LF2, the Chlamydomonas ortholog of CCRK, forms a ciliary LRC with LF1 and LF3 (33). Our analysis showed that there are two members of RCKs in Chlamydomonas including CrMAK and LF4, a MOK homologue. In contrast to LF4, whose phosphorylation necessitates LF2 and LF3 but not LF1 (37), CrMAK phosphorylation requires all three LRC components, suggesting a divergent regulatory mechanism. LF2 may have additional phosphorylation substrates beyond LF4 and CrMAK because double knockout mutants fail to replicate the ciliary phenotypes of lf2 null mutants. Elucidation of additional targets of LF2 would reveal how LRC functions in ciliogenesis.

Materials and Methods

Methods are described in more detail in SI Appendix.

Strains and Culture Conditions.

Chlamydomonas reinhardtii strains 21gr (mt+; CC-1690, wild-type), lf1-1 (mt+; CC-1678), lf2-6 (mt+; CC-803), lf3-6 (mt+; CC-4189), lf5-2 (mt+; CC-4560), and fla10-2 (mt+; CC-4180) are available from the Chlamydomonas Resource Center (University of Minnesota, Minneapolis, MN). Strain lf4-3 (mt−) was provided by Dr. Pete Lefebvre (University of Minnesota, Minneapolis, MN). Cells were cultivated on 1.5% agar plates or in liquid M medium at 23 °C with aeration under a 14/10 h light/dark cycle (56). For transformation, cells were cultured in liquid tris-acetate-phosphate (TAP) medium with shaking under continuous light (41).

Insertional Mutagenesis, Gene Cloning, and Transformation.

Insertional mutants were generated by electroporating AphVIII fragments (57, 58). The insertion sites were identified and confirmed by sequencing following a previous method (59). The CrMAK gene was cloned from genomic DNA while the FAP256 gene was cloned from cDNA. The point mutations were generated by PCR and confirmed by sequencing. Electroporation was used for transformation (41).

Generation of Knockout Strains Using CRISPR/Cas9.

Gene knockout using CRISPR/Cas9 was carried out according to a previous report (60). A ribonucleoprotein complex was assembled in vitro prior to transformation. The resulting transformants were subsequently screened for genotyping using PCR. The nucleotide sequences for the sgRNAs and donor DNAs were described in SI Appendix, Table S2.

Cell Fractionation, Cilia Regeneration, Ciliary Length Measurement, and DIC Cell Imaging.

Cilia and cell bodies were isolated postdeciliation using pH shock (43). For ciliary fractionation, purified cilia were solubilized with 0.5% NP-40 and the resulting supernatant and pellet were categorized as membrane and matrix (M + M) and axonemal fractions, respectively. Cells were deciliated using pH shock to allow ciliary regeneration. Cells were fixed with 1% glutaraldehyde before DIC imaging and ciliary length measurement.

Primary Antibodies.

Rabbit polyclonal anti-CrMAK was generated by immunizing with polypeptide 330 to 585 aa (ABclonal, Wuhan, China). Rabbit polyclonal anti-pCrMAK antibody was produced against the phosphopeptide C-SRPPY (T-p) DYVSTR (ABclonal, Wuhan, China). Both antibodies were affinity purified. Details of other antibodies used for immunofluorescence and immunoblotting are described in SI Appendix, Table S3.

SDS-PAGE, Immunoblotting, Immunoprecipitation, and Sucrose Gradient Analysis.

SDS-PAGE, immunoblotting, and immunoprecipitation analyses were performed as previously described (61, 62). For the analysis of IFT complexes, cell lysates were subjected to sucrose density gradient centrifugation according to established protocols (60, 63).

Immunofluorescence and TEM.

Immunofluorescence analyses were performed according to a previously published protocol (62). Confocal imaging was carried out using a Zeiss LSM780 Meta Observer Z1 Confocal Laser Microscope. For superresolution microscopy, an N-SIM system (Nikon, Tokyo, Japan) was utilized. Electron microscopy was conducted following established protocols (47).

Cabazitaxel Treatment.

mak and fla10 mutant cells after removing cell walls by autolysin treatment were resuspended in M medium (1 × 10⁷ cells/mL) supplemented with cabazitaxel at final concentrations of 25 μM (MedChemExpress, Catalog # HY-15459), fixed after 8 h with 1% glutaraldehyde followed by DIC imaging.

In Vitro Kinase Assay.

The in vitro kinase assays for CrMAK and CrMAK^K33R were conducted as described in previous studies (64, 65). Immunoprecipitated CrMAK-HA or CrMAK^K33R-HA was incubated with ATP-γS in kinase assay buffer. Thiophosphorylation was detected by immunoblotting using an anti-thiophosphate ester antibody (Abcam, Catalog #ab92570). To evaluate the phosphorylation of FAP256 by CrMAK in vitro, we followed a previously established protocol (45). Bacterially expressed recombinant proteins were incubated with ATP, and phosphorylation was assessed by Phos-tag SDS-PAGE, followed by Coomassie Brilliant Blue (CBB) staining.

Phosphatase Treatment and Phos-tag SDS-PAGE.

For dephosphorylation assay, cell lysates were treated with protein phosphatase followed by Phos-tag SDS-PAGE and immunoblotting (49).

APEX2-Based Proximity Labeling and Comparative Proteomic analysis.

To identify interaction partners of CrMAK, we have used Apex2-based proximity labeling combined with comparative proteomics. CrMAK-HA and CrMAK-HA-APEX2 expressing cells were metabolically labeled with ¹⁵N by culturing in medium containing ¹⁵NH₄¹⁵NO₃ or left unlabeled in regular medium, as previously described (47). For the proximity labeling assay (46), dewalled cells were incubated with biotin-phenol for 1 h, followed by deciliation. H₂O₂ was then added 5 min postdeciliation and incubated for 2.5 min to initiate biotin-phenol labeling. Biotinylated proteins were then affinity-purified using streptavidin sepharose beads, separated by SDS-PAGE, and subjected to mass spectrometry analysis. Data analysis was performed using the pFind algorithm with default settings (66).

Quantification and Statistical analysis.

All data were independently verified two or more times. The results are presented as mean ± SD. Statistical significance was determined using unpaired two-tailed Student’s t test analysis, with P-values < 0.05 considered statistically significant. Symbols *, **, and *** denote significance levels of P < 0.05, P < 0.01, and P < 0.001, respectively.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

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