A germline-specific role for unconventional components of the γ-tubulin complex in Caenorhabditis elegans

The γ-tubulin complex (γTuC) is a highly conserved microtubule (MT) nucleator that accumulates at microtubule-organizing centers (MTOCs) such as centrosomes in animal cells and spindle pole bodies in yeasts. There are two types of γTuCs with different sizes: the γ-tubulin small complex (γTuSC) and the γ-tubulin ring complex (γTuRC) (Kollman et al., 2011). The γTuSC consists of two molecules of γ-tubulin and one molecule each of GCP2 and GCP3 (also known as TUBGCP2 and TUBGCP3, respectively); the two γ-tubulin molecules bind to the C-terminal region of GCP2 and GCP3, forming a Y-shaped structure (Guillet et al., 2011; Kollman et al., 2008). The γTuRC comprises four to six γTuSCs and several additional γTuRC-specific components, including GCP4, GCP5 and GCP6 (also known as TUBGCP4, TUBGCP5 and TUBGCP6, respectively; referred to collectively as GCP4–6), and the small protein MOZART1 (also known as MZT1). The γTuRC forms a lock-washer-shaped structure to provide a template for 13-protofilament MTs (Zheng et al., 1995; Moritz et al., 1998; Oegema et al., 1999; Keating and Borisy, 2000; Wiese and Zheng, 2000; Murphy et al., 2001; Kollman et al., 2010). GCP2 to GCP6 share common structures, containing two conserved regions called grip1 and grip2 motifs; the grip2 motif directly binds γ-tubulin, whereas grip1 is involved in lateral interaction between GCP proteins (Gunawardane et al., 2000; Guillet et al., 2011; Kollman et al., 2008). GCP4–6 are proposed to be integrated into the γTuRC as a spoke similar to γTuSC assembly at the distal seam of the ring (Liu et al., 2019; Wieczorek et al., 2020; Consolati, et al., 2020).

Among the γTuC components, the requirement of GCP4–6 for γTuC function varies depending on the tissue and organism (Anders et al., 2006; Colombie et al., 2006; Venkatram et al., 2004; Fujita et al., 2002). In particular, Saccharomyces cerevisiae lacks GCP4–6 (Teixido-Travesa et al., 2012; Kollman et al., 2010; Geissler et al., 1996; Knop and Schiebel, 1997), but this yeast's γTuSCs can form a stable ring-like structure with 13-fold symmetry by associating with Spc110 (Kollman et al., 2010, 2015). In Drosophila, the GCP4 ortholog Grip75 and the GCP5 ortholog Grip128, which correspond to γTuRC-specific components, are dispensable for viability but required for spermatogenesis (Vogt et al., 2006). Thus, the functions of GCP4–6 have diverged during evolution to play various roles in different cell types.

The nematode Caenorhabditis elegans is another organism in which the genes encoding GCP4–6 have not been identified in the genome. The C. elegans gene tbg-1 is the only γ-tubulin gene (Bobinnec et al., 2000; Strome et al., 2001; Hannak et al., 2002), and GIP-2 (ortholog of GCP2), GIP-1 (ortholog of GCP3) and MZT-1 (ortholog of MOZART1) have been identified based on their sequence similarities (Hannak et al., 2002; Sallee et al., 2018). Whether the C. elegans γTuC contains additional components remains unknown.

The γTuC in C. elegans localizes to various sites in a cell type-specific or tissue-specific manner. In mitotic cells, centrosomal localization of the γTuC is dependent on SPD-5, which is a scaffold protein of the pericentriolar matrix (PCM) and a functional homolog of Drosophila Centrosomin (Hamill, et al., 2002; Nakajo, et al., 2022). In some tissues, the γTuC localizes at non-centrosomal MTOCs to organize specific MT arrays (Bobinnec et al., 2000; Zhou et al., 2009; Wang et al., 2015; Sallee et al., 2018; Sanchez et al., 2021). For example, in embryonic intestinal cells, the γTuC localizes to the apical membrane with the tissue-specific protein WDR-62 but does not require MZT-1 (Sallee et al., 2018; Sanchez et al., 2021). In the adult hermaphrodite germline, the γTuC along with the ninein homolog NOCA-1 localizes at the germ cell membrane, and this localization is required to maintain the germline morphology (Green et al., 2011; Wang et al., 2015). The tissue-specific localization and functions of the γTuC imply that certain unidentified components or associated proteins might be involved in targeting of the γTuC to specific intracellular sites in different tissues.

Here, we report the identification and characterization of two γ-tubulin-associated proteins, namely GTAP-1 and GTAP-2, in C. elegans. These proteins interact with γTuC components and are predicted to have a partial structural similarity to GCP protein components of the γTuC. Both proteins contribute to the efficient recruitment of γ-tubulin to centrosomes in early embryos and the plasma membrane in the germline. GTAP-1, but not GTAP-2, plays a crucial role in the organization of non-centrosomal MTs in the germline. We propose that GTAP-1 and GTAP-2 are highly diverged GCP protein components of the γTuC in C. elegans that play distinct roles in centrosomal and non-centrosomal MTOCs in a tissue-specific manner.

To identify new components of the γTuC in C. elegans, we biochemically analyzed proteins that associate with γ-tubulin. An antibody specific for FLAG was used for immunoprecipitation of an extract of worm embryos expressing GFP::γ-tubulin::FLAG. The immunoprecipitates were comprehensively analyzed using mass spectrometry (Fig. 1A,B; full mass spectrometry results are available via Figshare, https://doi.org/10.6084/m9.figshare.23301683). The known components of the γTuSC, namely GIP-1 and GIP-2, efficiently co-precipitated with γ-tubulin, as expected (Fig. 1A,B). Two uncharacterized proteins, encoded by ZK1248.13 and ZK632.5, were identified in the top-ranked list of proteins that co-precipitated with γ-tubulin; the proteins were named gamma-tubulin-associating protein 1 (GTAP-1) and gamma-tubulin-associating protein 2 (GTAP-2), respectively.

GTAP-1 and GTAP-2 were found to lack any significant amino acid sequence similarity to known proteins based on BLAST (Basic Local Alignment Search Tool) searches, with the exception of apparent orthologs in nematodes of the genus Caenorhabditis. However, the overall predicted structure of GTAP-1 modeled using AlphaFold (Jumper et al., 2021) resembled that of human GCP proteins and C. elegans GIP-1 (Fig. 1C). A region of GTAP-2 showed only modest sequence similarity in a part of the grip2 motif, which corresponds to a γ-tubulin-binding site conserved among GCP2–6 (Fig. S1), but similarity with the grip1 motif was not detected. The local sequence similarity and predicted structures implied that GTAP-1 and GTAP-2 might have diverged from GCP proteins during evolution.

The subcellular localization of GTAP-1 and GTAP-2 was examined using embryos exogenously expressing GFP::GTAP-1 or GFP::GTAP-2 along with mCherry::γ-tubulin (also referred to herein as mCherry::TBG-1) and mCherry::histone H2B. In early embryos, GFP::GTAP-1 and GFP::GTAP-2 colocalized with mCherry::γ-tubulin in the PCM throughout the cell cycle (Fig. 1D,E; images of single endogenous GFP::GTAP-1 or GFP::GTAP-2 are presented in Fig. S2). The foci areas and relative fluorescence intensities of GFP::GTAP-1 and GFP::GTAP-2 signals at centrosomes increased synchronously with γ-tubulin during centrosome maturation. At the end of telophase, the majority of GFP::GTAP-1 and GFP::GTAP-2 had dispersed to the cytoplasm, but small amounts remained at centrosomes as small foci during interphase, coinciding with γ-tubulin (Fig. 1D,E). Thus, GTAP-1 and GTAP-2 colocalized with γ-tubulin throughout the cell cycle.

In embryos with RNAi-mediated knockdown of γ-tubulin [tbg-1(RNAi) embryos], GFP::GTAP-1 (5 of 6 embryos) and GFP::GTAP-2 (17 of 18 embryos) signals were undetectable at centrosomes (Fig. 1F,G). Therefore, the centrosomal localization of GTAP-1 and GTAP-2 was dependent on γ-tubulin, similar to that of the γTuSC components GIP-1 and GIP-2.

The loss-of-function phenotypes of gtap-1 and/or gtap-2 in early embryos were examined using RNAi-mediated knockdown. The gtap-1(RNAi), gtap-2(RNAi) and gtap-1(RNAi); gtap-2(RNAi) embryos were viable, but the amount of centrosomal γ-tubulin was reduced (Fig. 2A,B). The fluorescence intensity of centrosomal mCherry::γ-tubulin at metaphase in embryos at the one-cell stage was reduced to 51% (P<0.0001, n=12) in gtap-1(RNAi) embryos, 32% (P<0.0001, n=11) in gtap-2(RNAi) embryos and 64% in gtap-1(RNAi); gtap-2(RNAi) embryos (P<0.001, n=10) relative to levels in control embryos (n=33) (Fig. 2B). Western blotting revealed that the level of γ-tubulin was not reduced in these embryos (Figs S3 and S8), indicating that GTAP-1 and GTAP-2 did not affect the cellular level or stability of γ-tubulin. We also examined the centrosomal amount of another component of the γTuSC, GIP-2, using exogenous GFP::GIP-2. The level of centrosomal GFP::GIP-2 was also reduced to 62% of the control level in gtap-1(RNAi) embryos and 52% of the control level in gtap-2(RNAi) embryos, comparable to the reduction in centrosomal mCherry::γ-tubulin (Fig. 2C). The fact that the double depletion did not further reduce the amount of γ-tubulin at centrosomes compared to the effect of single depletion suggests that GTAP-1 and GTAP-2 likely function in the same pathway for the efficient recruitment of γ-tubulin to centrosomes in embryos (Fig. 2B). These results suggest that GTAP-1 and GTAP-2 are involved in the efficient localization of the components of γTuCs at the centrosomes.

Although the level of centrosomal γ-tubulin was significantly reduced, the amount of centrosomal MTs in one-cell embryos was unaffected by gtap-1(RNAi) (101% of mean of control, n=12, P>0.1), gtap-2(RNAi) (110% of mean of control, n=11, P>0.1), or gtap-1(RNAi); gtap-2(RNAi) (97% of mean of control, n=10, P>0.1) treatments compared with the control (n=33) (Fig. 2A,D). This discrepancy between the amount of γ-tubulin and the amount of MTs at centrosomes could be explained if a large portion of centrosomal γ-tubulin is inactive for MT nucleation. To test this hypothesis, γ-tubulin was partially depleted using RNAi [tbg-1(partial RNAi)], and γ-tubulin and MTs surrounding centrosomes were quantified using exogenous mCherry::γ-tubulin and GFP::β-tubulin (also referred to herein as GFP::TBB-2), respectively (Fig. 2E,F). In tbg-1(partial RNAi) embryos, even when centrosomal γ-tubulin was reduced to ∼30% of the level in control embryos, centrosomal MT content was not substantially lower (89% of mean of control, n=48, P>0.05) than that in control embryos (n=45). Moreover, bipolar spindle formation and chromosome segregation were not affected in these embryos (100%, n=48). Thus, centrosomes in one-cell embryos contain excessive γ-tubulin, and the reduced levels of γ-tubulin upon the loss of GTAP-1 and GTAP-2 are sufficient for MT nucleation to assemble functional mitotic spindles.

In gtap-2(RNAi) one-cell embryos, the level of centrosomal GFP::GTAP-1 at metaphase was reduced to 10% (P<0.01, n=11) of that in control embryos (n=10) (Fig. 2G). Similarly, in gtap-1(RNAi) one-cell embryos, centrosomal GFP::GTAP-2 at metaphase was reduced to 40% (P<0.01, n=7) of that in control embryos (n=6) (Fig. 2H). Western blotting revealed that the amount of GTAP-1 and GTAP-2 was not reduced in gtap-2(RNAi) and gtap-1(RNAi) embryos, respectively (Figs S3 and S8). Thus, GTAP-1 and GTAP-2 are partially dependent on each other for correct centrosomal localization, although GTAP-1 localization is more dependent on GTAP-2 than GTAP-2 localization is dependent on GTAP-1.

Although the amount of centrosomal MTs was barely affected in gtap-1(RNAi), gtap-2(RNAi) and gtap-1(RNAi); gtap-2(RNAi) embryos, their interaction with the PCM seemed to be altered. In wild-type embryos at the end of telophase, the dynein-dependent cortical MT pulling force mediates fragmentation of spindle poles (Grill et al., 2001, 2003; Labbé et al., 2004; Pecreaux et al., 2006; Kozlowski et al., 2007), and γ-tubulin in the PCM scatters into the cytoplasm. In gtap-1(RNAi), gtap-2(RNAi) and gtap-1(RNAi); gtap-2(RNAi) embryos, fragmentation of the spindle pole occurred earlier and more drastically than expected (Fig. 3A; Fig. S4). Concurrently, PCM, as visualized using exogenous GFP::SPD-5, was also precociously fragmented (Fig. 3B). In tbg-1(partial RNAi) embryos, this fragmentation of the spindle pole was not observed; instead, the signal of the spindle pole gradually decreased (Fig. 3A; Fig. S4). These observations indicated that exaggerated fragmentation of the spindle pole in GTAP-1- and/or GTAP-2-depleted embryos was not simply caused by a reduction in the level of centrosomal γ-tubulin; rather, depletion of GTAP-1 and GTAP-2 seemed to affect PCM integrity, by, for example, affecting the interaction between γTuC and PCM scaffold components. An alternate possibility is that the dynamics of MTs that generate the cortical pulling force might differ between the MTs nucleated from an intact γTuC and those nucleated from an incomplete γTuC without GTAP-1 and GTAP-2.

To examine the developmental phenotypes, null mutants of gtap-1 and gtap-2 were constructed using CRISPR/Cas9. In each of the null mutants, namely gtap-1(tj84) (which has a stop codon at Arg51) and gtap-2(tj92) (which has amino acids 93–400 replaced by the HygR hygromycin resistance protein), centrosomal γ-tubulin was reduced by ∼50% in one-cell embryos, as was the case in the gtap-1(RNAi) and gtap-2(RNAi) embryos (Fig. S5), and the mutants could be maintained as homozygotes, suggesting that GTAP-1 and GTAP-2 are not essential for development and fertility. However, the fecundity of gtap-1(tj84) worms was severely reduced after 24 h (Fig. 4A). Correspondingly, the embryonic lethality of the gtap-1(tj84) mutant increased with age, rising to ∼50% after 48 h (Fig. 4B). The variability of the size and shape of gtap-1(tj84) embryos also increased with age [average length of the major axis and minor axis changed from 48.7±4.3 µm (day 1; mean±s.d.) to 56.5±8.6 µm (day 2) and from 34.0±1.6 µm (day 1) to 32.3±3.6 µm (day 2), respectively] (Fig. 4C; Table S1). In contrast, the gtap-2(tj92) mutant did not show embryonic lethality (Fig. 4B), and the proportion of embryos having abnormal size and shape was lower than that for gtap-1(tj84) (Fig. 4C; Table S1).

The morphologically abnormal eggs and reduction of brood size in the gtap-1(tj84) mutant implied defects in oogenesis. In adult hermaphrodites, the distal gonad is syncytial, and each meiotic nucleus is compartmentalized, with the membranes organized in a honeycomb-like structure. Each compartment is partially open and connected to the large cytosolic region called the rachis (Fig. 4D). The organization of the adult hermaphrodite germline was analyzed using a strain in which the germ cell membrane was labeled using a GFP-tagged pleckstrin homology domain (GFP::PH) (Fig. 4E). In the gonad of gtap-1(tj84) adults, the honeycomb-like structure of the germ cell membrane was disorganized, and the size of the compartment was highly variable. By contrast, the honeycomb-like structure of the gtap-2(tj92) adult germline was only slightly perturbed. The gtap-1(tj84); gtap-2(tj92) double mutant had more severe and wider ranging phenotypes than either single mutant (a small brood size of only 6.3 eggs per worm, 49.3% embryonic lethality, n=11; low frequency of dumpy phenotype or rupture of adult worms). These results indicate that, whereas GTAP-1 plays a more important role in the adult germline, GTAP-1 and GTAP-2 function redundantly in various tissues.

Next, the subcellular localization of GTAP-1 and GTAP-2 in the hermaphrodite germline was analyzed using strains in which each endogenous GTAP protein was tagged with GFP by CRISPR/Cas9 (Fig. 5A). Both GFP-labeled proteins localized at centrosomes and at the plasma membrane of the adult germline, and GFP::GTAP-2 was also detected in the cytoplasm. These localization patterns at centrosomes and the plasma membrane in the adult germline were similar to that of γ-tubulin (Fig. 5A,B). The γ-tubulin at the plasma membrane is required for the nucleation of MTs involved in positioning of germ cell nuclei in each compartment of the syncytial gonad (Zhou et al., 2009). Therefore, we next examined whether plasma membrane localization of γ-tubulin was affected in the gtap-1(tj84) and gtap-2(tj92) mutants using strains exogenously co-expressing mCherry::γ-tubulin and mCherry::histone H2B. The amount of γ-tubulin at the plasma membrane was quantified based on the relative intensity of mCherry::γ-tubulin using mCherry::histone H2B as the internal control (Fig. 5B). In the gtap-1(tj84) and gtap-2(tj92) mutants, the relative intensity of γ-tubulin was significantly reduced to 49% and 59% of that in the control, respectively (Fig. 5B), suggesting that GTAP-1 and GTAP-2 contribute to the efficient recruitment of γ-tubulin to the plasma membrane in the adult germline.

Because the amount of γ-tubulin at the plasma membrane in the germline was decreased in each of the gtap-1(tj84) and gtap-2(tj92) mutants, MTs in the germline were observed by live imaging using exogenous mCherry::β-tubulin (also referred to herein as mCherry::TBB-2). In wild-type and gtap-2(tj92) adult hermaphrodites, MTs were highly organized around germ cell nuclei located at the periphery of the syncytial gonad (Fig. 5C). In the gtap-1(tj84) mutant, however, the distribution of MTs surrounding nuclei was uneven, the location of nuclei was perturbed, and some nuclei were detached from the peripheral compartments and detected in the rachis (Fig. 5C). Thus, even though both GTAP-1 and GTAP-2 contribute to the recruitment of γ-tubulin to the plasma membrane in the germline, GTAP-1 plays a much larger role than GTAP-2 in the organization of MT arrays.

Because the embryonic and germline subcellular localizations and phenotypes strongly indicated a close link between GTAP-1, GTAP-2 and γ-tubulin, the interaction of GTAP-1 and GTAP-2 with γTuSC components was examined using immunoprecipitation (Fig. 6A). The immunoprecipitation was performed with an anti-GFP antibody using extracts of embryos of the control strain N2 and of strains expressing either GFP::γ-tubulin, GFP::GTAP-2 or GFP::GTAP-1 as a transgene. As expected, from the extract of the strain expressing GFP::γ-tubulin, GTAP-1 and GTAP-2 co-immunoprecipitated along with GIP-1 and GIP-2. Conversely, from the extracts of strains expressing GFP::GTAP-1 or GFP::GTAP-2, all components of the γTuSC (γ-tubulin, GIP-1 and GIP-2), as well as GTAP-2 or GTAP-1, respectively, were co-precipitated. These results indicate that GTAP-1 and GTAP-2 interact with the γTuC in vivo.

Endogenous GTAP-1 was not detected among the proteins that co-immunoprecipitated with GFP::GTAP-1; similarly, endogenous GTAP-2 was not detected among proteins that co-immunoprecipitated with GFP::GTAP-2. Thus, we speculated that the γTuC associates with one molecule of GTAP-1 and one molecule of GTAP-2.

To further examine whether γTuCs in C. elegans contain GTAP-1 and/or GTAP-2, embryo extracts were subjected to sucrose gradient sedimentation. Two distinct peaks of γ-tubulin-containing fractions were detected (fractions 1–9 and 12–16) (Fig. 6B), whose sizes roughly corresponded to that of the γTuSC and γTuRC in other organisms (estimated S values were 4–15S and 25–35S, respectively) (Stearns and Kirschner, 1994; Zheng et al., 1995; Murphy et al., 1998; Oegema et al., 1999; Anders et al., 2006). However, the separation of the two peaks was less clear than has been observed in experiments with other organisms, and both γ-tubulin peaks contained GIP-1, GIP-2, GTAP-1 and GTAP-2. We speculated that γTuCs in C. elegans are heterogeneous, including various numbers of γTuSCs with GTAP-1 and/or GTAP-2.

Because GTAP-1 and GTAP-2 were found to associate with the γTuC in vivo, the interaction of GTAP-1 and GTAP-2 with γTuSC components was examined using yeast two-hybrid assays (Fig. 6C; Table S2). GTAP-2 interacted strongly with GIP-2 and weakly with γ-tubulin (Fig. 6C). GIP-2 interacted strongly with γ-tubulin as well as GTAP-2 (Fig. 6C). Interactions between GTAP-1 and other components were not detected (Fig. 6C).

Because the predicted interactions of GIP-1 with other γTuSC components (γ-tubulin and GIP-2) were not detected in this assay, we speculated that some interactions might require more than two γTuC components. Therefore, interactions between three components were examined using a modified yeast two-hybrid assay. GCP proteins, including GIP-1/GCP3 and GIP-2/GCP2, have grip1 and grip2 motifs, which interact with adjacent GCP proteins and γ-tubulin, respectively. In subsequent assays, the core region of GIP-1 (amino acids 271–891, termed GIP-1_ΔN) containing the grip1 and grip2 motifs was used. When γ-tubulin was co-expressed, GIP-1_ΔN interacted with GIP-2 (Fig. 6D). GIP-2 lacking the grip2 motif (GIP-2_1–360) still interacted with GIP-1_ΔN in a γ-tubulin-dependent manner, whereas GIP-2 lacking the grip1 motif (GIP-2_317–642) did not. Thus, this ternary complex is likely to require interactions between γ-tubulin and GIP-1_ΔN and between GIP-1_ΔN and the grip1 motif of GIP-2 (Fig. 6D).

In a modified yeast two-hybrid assay with γ-tubulin, GTAP-1 interacted with GIP-2 and GIP-2_1–360, whereas interaction between GTAP-1 and GTAP-2 was not detected (Fig. 6E; Table S2). This result indicates that the GTAP-1–GIP-2–γ-tubulin ternary complex requires interactions between γ-tubulin and GTAP-1 and between GTAP-1 and the grip1 motif of GIP-2. These interactions are similar to those observed for the GIP-1–GIP-2–γ-tubulin ternary complex.

Because GTAP-1, GTAP-2 and GIP-1 were found to interact with GIP-2, we next investigated whether these three proteins compete for binding to the same site in GIP-2 using a modified yeast two-hybrid assay in which γ-tubulin and GIP-2 were co-expressed in addition to the bait and prey proteins. In the presence of γ-tubulin and GIP-2, the interactions (either direct or indirect) between GTAP-1 and GTAP-2 or between GTAP-1 and GIP-1 were detected, but the interaction between GTAP-2 and GIP-1 was not (Fig. 6F). These results indicate that GTAP-2 competes with GIP-1 for binding to GIP-2, whereas GTAP-1 does not compete with either GIP-1 or GTAP-2.

MOZART1 is a broadly conserved small protein involved in the γTuC and is essential for spindle assembly in many organisms. C. elegans MOZART1, MZT-1, accumulates in the PCM and is required for the interaction between the N-terminus of SPD-5 and the γ-TuSC (GIP-1, GIP-2 and γ-tubulin) (Sallee et al., 2018; Ohta, et al., 2021). To compare the roles of GTAP-1 and GTAP-2 with that of MZT-1, we examined localization and loss-of-function phenotypes in early embryos using RNAi-mediated knockdown. As previously reported (Sallee et al., 2018), exogenous GFP::MZT-1 accumulated in the PCM, and its recruitment was dependent on γ-tubulin and GIP-1 (Fig. 7A). Whereas the depletion of GTAP-1 and/or GTAP-2 did not cause a significant defect in chromosome segregation (Fig. 2), depletion of MZT-1 resulted in severe defects in both spindle formation and chromosome segregation (10/16 embryos) (Fig. 7B). The PCM localization of γ-tubulin and GIP-1, but not of SPD-5, was significantly reduced in MZT-1-depleted embryos (Fig. 7C), and the amount of centrosomal γ-tubulin was reduced to ∼6% of the control (P<0.0001, n=15), resulting in a significant reduction of centrosomal MTs (72% of the control, P<0.001, n=15) (Fig. 7D,E). Thus, MZT-1 in early embryos plays a more crucial role than GTAP-1 or GTAP-2 in the centrosomal localization of the γTuC.

The localization of MZT-1 was unaffected by depletion of GTAP-1 or GTAP-2 (Fig. 7F), suggesting that GTAP-1 and GTAP-2 affect the recruitment of γ-tubulin and GIP-2 independently of MZT-1.

In Arabidopsis and other organisms, MOZART1 orthologs bind the N-terminal region of GCP3 orthologs, with the exception of the human MOZART1 ortholog, which likely binds the N-terminal region of GCP2, GCP3, GCP5 and GCP6 (Janski et al., 2012; Nakamura et al., 2012; Dhani et al., 2013; Cota et al., 2017). A yeast two-hybrid assay revealed that C. elegans MZT-1 bound to the N-terminal region of GIP-1 but not GIP-2, GTAP-1 or GTAP-2 (Fig. 7G). Because MZT-1 was required for the centrosomal localization of GIP-1 (Fig. 7C), this interaction might be essential for γTuSC localization to centrosomes.

As described above, GTAP-2 contains a region that has only modest sequence similarity to part of the grip2 motif conserved in GCP2–6 proteins in other organisms. The predicted protein structure of GTAP-1 is similar to the common GCP protein structure, even though the sequence similarity was undetectable (Fig. 1C). These findings implied the possibility that GTAP-1 and GTAP-2 are highly divergent versions of GCP4–6. To understand how the composition of the γTuC has evolved, the phylogeny of each γTuC component was analyzed for 28 species (12 metazoan species including seven nematodes, 3 protozoan species, 9 fungi and 4 plants) (Fig. 8A), which led to the following findings.

First, all eukaryotes that were examined were found to contain γTuSC components (γ-tubulin, GCP2 and GCP3), and at least some species in all four groups (metazoans, protozoans, fungi and plants) contain GCP4–6, suggesting that the canonical γTuRC composition was established at the emergence of eukaryotes, before divergence of these four groups. Second, at least twice during evolution, some of the γTuRC-specific components were lost from the genome. S. cerevisiae and two other fungi that belong to Saccharomycotina (Candida albicans and Kluyveromyces lactis) do not have GCP4–6, whereas the majority of other fungi have the complete set. Also, within nematode species, whereas Trichinella spiralis has the complete set of GCP4–6, species in Chromadoria (Brugia malayi, Loa loa and four Caenorhabditis species including C. elegans) lack some of them. GTAP-1 and GTAP-2 were found to be highly conserved in Caenorhabditis species although their phylogenetic relationship with GCP4–6 is unclear. Third, in the species that had lost some of GCP4–6, the evolution of the remaining components was accelerated (Fig. 8B). Whereas the phylogenetic trees for α- and β-tubulins are consistent with the phylogeny of the organisms, in the phylogenetic tree for γTuSC components, all organisms that lack some of GCP4–6 are separated from those with the complete sets. These data imply that structural constraints of the γTuSC components are released by the loss of GCP4–6, resulting in adoption of a new stable state of the γTuRC as a whole.

In this study, we have demonstrated that C. elegans has an unconventional composition of the γTuC by identifying two γTuC components, namely GTAP-1 and GTAP-2, which are predicted to be highly divergent versions of GCP4–6 in other organisms. GTAP-1 and GTAP-2 contribute to the efficient recruitment of γ-tubulin and GIP-2 to centrosomes in early embryos and to the plasma membrane in the germline. Unlike another γTuC component MZT-1, however, GTAP-1 and GTAP-2 are not essential for embryogenesis (Fig. 8C). GTAP-1 plays a critical role in establishing and maintaining the organization of germline cells in adults, whereas GTAP-2 plays a minor role in this process (Fig. 8C). We propose that GTAP-1 and GTAP-2 contribute to the organization of centrosomal and non-centrosomal MTs in a tissue-specific manner.

Our results demonstrate that GTAP-1 and GTAP-2 interact with GIP-2. As such, we predict that the structure of the C. elegans γTuC – which includes GTAP-1 and GTAP-2 – will be similar to that of the γTuC region encompassing GCP4–6 in other organisms, ultimately mirroring the canonical ring-like structure of the γTuC (Fig. 8C). Structural analyses of vertebrate γTuRCs have indicated that two molecules of GCP4 and one molecule each of GCP5 and GCP6 are incorporated in the half region of the γTuRC (Liu et al., 2019; Wieczorek et al., 2020; Consolati, et al., 2020). The structure of the region that contains GCP4–6 is open and more flexible than the rest of the γTuRC. A phylogenetic analysis implies that an ancestral nematode species of the genus Caenorhabditis lost at least one of the conventional γTuRC-specific components (GCP4–6), which led to the rapid co-evolution of the remaining γTuC components. Because the amino acid sequences of GTAP-1 and GTAP-2 and of GCP4–6 have minimal similarity, it is unclear which GCP proteins correspond to GTAP-1 and GTAP-2.

C. elegans lacks several γTuRC-interacting proteins, such as NEDD-1 (also known as GCP-WD) and the Augmin complex (Uehara et al., 2009), that are conserved in other organisms that have a conventional γTuRC. One possibility is that these proteins might have been lost as a consequence of the alteration of the composition and structure of the γTuC in C. elegans. Further structural analysis will be needed to understand how the C. elegans γTuC containing GTAP-1 and GTAP-2 differs in a structural sense from the structure of conventional γTuRCs.

Our results suggest that the role of GTAP-1 and GTAP-2 in the centrosomal recruitment of γ-tubulin is distinct from that of MZT-1. MZT-1 is required for centrosomal targeting of the γTuC by mediating the interaction between the γTuSC and phosphorylated SPD-5 at the mitotic PCM (Sallee et al., 2018; Ohta, et al., 2021). Although depletion of MZT-1 resulted in a ∼95% reduction of centrosomal γ-tubulin, depletion of GTAP-1 and/or GTAP-2 caused a 50–70% reduction, indicating that MZT-1 plays an essential role in centrosomal recruitment of the γTuSC but GTAP-1 and GTAP-2 do not. We speculate that, although the γTuSC can be recruited to the PCM via the interaction between MZT-1 and SPD-5, γTuCs containing GTAP-1 and GTAP-2 might be more stable than γTuCs without them, which might contribute to the efficient accumulation of γTuCs in the PCM.

Although GTAP-1 and GTAP-2 are dispensable for γTuC recruitment to the PCM, our data indicate that their absence affects the dynamics of both centrosomal MTs and the PCM at telophase. One possibility is that GTAP-1 and GTAP-2 contribute to strengthening the connections among γTuCs and PCM proteins at centrosomes, which in turn contributes to the integrity of the PCM. Alternatively, the MTs formed from a γTuC lacking GTAP-1 and/or GTAP-2 might result in distinct dynamics at MT ends or altered interaction with MT-binding proteins or motors such as dynein, which are involved in pulling forth from the cell cortex (Grill et al., 2001, 2003; Labbé et al., 2004; Pecreaux et al., 2006; Kozlowski et al., 2007).

In the adult germline, the γTuC localizes to the gonadal membrane of syncytial gonads, and MTs emanating from the membrane are required for maintaining the germline nuclei in each honeycomb-like cell compartment (Zhou et al., 2009). GTAP-1 and GTAP-2 contribute to the efficient recruitment of the γTuC to the plasma membrane in the germline. The gtap-1(tj84) mutant had severe morphological defects in the organization of the honeycomb-like gonadal compartment, which is similar to what has been observed upon depletion of ZYG-12, which localizes to the nuclear envelope and anchors MTs emanating from the germline membrane (Zhou et al., 2009). We speculate that the γTuC containing GTAP-1 assembles MT arrays on the plasma membrane, and anchoring these MTs to the nuclear envelope via ZYG-12 is crucial for the organization of meiotic nuclei in the compartments of the germline.

Although the amounts of γ-tubulin at the plasma membrane in the gonads was reduced to a similar extent in both the gtap-1(tj84) and gtap-2(tj92) mutants, only gtap-1(tj84) had a severe phenotype. This indicates that the role of GTAP-1 is not limited to the efficient recruitment of the γTuC to the plasma membrane in the germline, and indeed it might be involved in the structural and/or functional regulation of the γTuC.

The phenotype of gtap-1(tj84) is also similar to that observed upon depletion of NOCA-1, which is another protein that localizes to the minus end of MTs (Green et al., 2011; Wang et al., 2015). Unlike GTAP-1 and GTAP-2, however, the localization of the γTuC at the germline membrane is not dependent on NOCA-1, although the localization of NOCA-1 is partially dependent on the γTuC (Wang et al., 2015). This similarity of phenotypes indicates that GTAP-1 and NOCA-1 might cooperatively regulate the γTuC to assemble the MT array on the germline membrane. Further analysis will be needed to understand the interactions between the γTuC, GTAP-1, GTAP-2 and NOCA-1 with regard to MT organization in the germline.

Our study demonstrates that the requirement of GTAP-1 and GTAP-2 for γTuC recruitment might differ between mitotic centrosomes and the non-centrosomal MTOC (ncMTOC) at the plasma membrane in the germline. Although GTAP-1 is dispensable for embryogenesis and has nearly equivalent function to GTAP-2 with regard to centrosomal targeting of the γTuC, it plays a more crucial role than GTAP-2 in germline organization. The severe post-embryonic phenotypes of the gtap-1(tj84); gtap-2(tj92) double mutant indicate that GTAP-1 and GTAP-2 are required in various somatic tissues. These distinct requirements of GTAP-1 and GTAP-2 in various developmental stages and tissues do not contradict our finding that the γTuCs in C. elegans are heterogeneous.

These findings indicate that certain γTuC components play different roles between the centrosomal MTOC and the ncMTOC. Similarly, human GCP6 plays a role in the localization of the γTuC to keratin fibers in epithelial cells in addition to its ubiquitous role in γTuRC assembly (Oriolo et al., 2007; Liu et al., 2019; Wieczorek et al., 2020; Consolati, et al., 2020). In Drosophila, the composition of the γTuRC during spermatogenesis differs from that at the centrosomal MTOC in other cells: Grip84 (GCP2 ortholog), Grip91 (GCP3 ortholog) and Grip128 (GCP5 ortholog) have testis-specific isotypes (Alzyoud et al., 2021), and Grip75 (GCP4 ortholog) and MOZART1 are specifically required for spermatogenesis (Vogt et al., 2006; Tovey et al., 2018). Thus, some components of γTuCs are linked specifically to the ncMTOC to control tissue-specific MT organization.

Because the composition and function of γTuC components vary in the ncMTOC of different tissues, the mechanism by which the γTuC is recruited to the ncMTOC might also differ from that governing its recruitment to centrosomes. In embryonic intestinal epithelial cells in C. elegans, WDR-62 recruits the γTuC to the apical ncMTOC, but WDR-62 is not involved in γTuC recruitment to centrosomes. On the other hand, MZT-1 is essential for centrosomal recruitment of the γTuC but dispensable for its recruitment to the apical ncMTOC (Sallee et al., 2018; Sanchez et al., 2021). Thus, the composition and recruitment mechanism of the γTuC appears to be diverse in mitotic centrosomes and ncMTOCs in various tissues.

In the C. elegans germline, the mechanism by which the γTuC is recruited to the ncMTOC on the plasma membrane remains unknown. Further studies of the tissue-specific roles and interactors of GTAP-1 and GTAP-2 will facilitate better understanding of the spatiotemporal regulation of ncMTOC positioning and function.

The unconventional composition of the C. elegans γTuC – with GTAP-1 and GTAP-2 instead of the typical GCP4–6 – might correlate with certain unique features of MTs in this organism. First, C. elegans embryos have unusually high γ-tubulin-independent MT assembly. Whereas loss of γ-tubulin results in a drastic reduction of MTs in the majority of organisms, up to ∼40% of MTs in C. elegans early embryos are assembled in a manner that is independent of γ-tubulin yet dependent on AIR-1 (the C. elegans Aurora kinase A ortholog) (Hannak et al., 2002; Motegi et al., 2006; Toya et al., 2011). Electron microscopic analysis has revealed that C. elegans embryos have open-ended MTs and capped-ended MTs, and that the latter are likely to be assembled with the γTuC (O'Toole et al., 2003, 2012). One possibility is that an altered γTuC composition in ancestral Caenorhabditis nematodes might have reduced the ability to nucleate MTs, which led to the emergence of a compensatory γ-tubulin-independent (and AIR-1-dependent) MT assembly mechanism. Second, the fact that the usual number of MT protofilaments in most organisms is 13, corresponding to the 13-fold symmetrical structure of the γTuC, also has implications for the unconventional γTuC symmetry in C. elegans. The fundamental role of the γTuC is to serve as a template for MTs, and a typical γTuC has 13-fold symmetry, which is consistent with the 13 protofilaments of canonical MTs in most eukaryotic cells. In contrast, MTs in C. elegans neurons and embryos generally have 11 profilaments (Chalfie and Thomson, 1982; Chaaban et al., 2018). It is tempting to speculate that the unconventional composition of the C. elegans γTuC (i.e. containing GTAP-1 and GTAP-2) might help restrict the number of MT protofilaments to 11 in vivo. Further structural studies will be needed to determine whether the C. elegans γTuC provides the structural seed to form 11-protofilament MTs.

Strains of C. elegans were cultured using standard methods (Brenner, 1974) at 20°C (N2) or 24°C (all fluorescing strains). The strains constructed in this study are listed in Table S3. Strain SA250 {tjIs54[pie-1p::gfp::tbb-2; pie-1p::2xmCherry::tbg-1; unc-119(+)]; tjIs57[pie-1p::mCherry::histone H2B(his-48); unc-119(+)]} (Toya et al., 2010) was used for microscopic analysis monitoring β-tubulin, γ-tubulin and histone H2B.

The strains that expressed GFP::TEV::γ-tubulin::FLAG, GFP::GTAP-1, GFP::GTAP-2, GFP::SPD-5, GFP::GIP-1 and GFP::GIP-2 were constructed by high-pressure microparticle bombardment (Praitis et al., 2001) of DP38 [unc-119(ed3)] worms (obtained from the Caenorhabditis Genetics Center) with plasmids pMTN1G_TEV::tbg-1(γ-tubulin)_FLAG, pMTN1G_gtap-1, pMTN1G_gtap-2, pMTN1G_spd-5, pMTN1G_GIP-1, and pMTN1G_GIP-2, respectively. To construct these plasmids, full-length γ-tubulin with TEV and FLAG sequences, GTAP-1 and GTAP-2, SPD-5, GIP-1 and GIP-2 cDNAs was amplified by PCR and cloned into the entry vector pDONR201 (Thermo Fisher Scientific). Thereafter, the inserts were transferred to plasmid pMTN1G (Toya et al., 2010) by the LR reaction (Thermo Fisher Scientific).

To construct strain GFP::MZT-1 (W08G9.8), the MosSCI transposon method (https://wormbuilder.org/; Frøkjær-Jensen et al., 2008) was used. In brief, the cDNA fragment encoding W08G9.8 was amplified and inserted into pCFJ350 (Addgene #34866, deposited by Erik Jorgensen) with the gene gfp, the pie-1 promoter, and the pie-1 3′ untranslated region (UTR) by Gibson assembly (Gibson et al., 2009). The plasmid was injected along with a plasmid containing transposase (pCFJ601, Peft-3::Mos1 transposase; Addgene #34874, deposited by Erik Jorgensen) and co-injection markers (pGH8, Addgene #19359; pCFJ90, Addgene #19327; pCFJ104, Addgene #19328; all deposited by Erik Jorgensen) into strain EG6699 [ttTi5605 II; unc-119(ed3) III; oxEx1578] (obtained from Caenorhabditis Genetics Center) to integrate into the ttTi5605 site.

To construct the strain that expressed GFP::PH and mCherry::TBG-1::SL2::mCherry::histone H2B (SA1425), we used the miniMos single-insertion method (https://wormbuilder.org/; Frøkjær-Jensen et al., 2014). In brief, the PCR-amplified fragment of pie-1p::gfp::PH::pie-1 3′ UTR was inserted into pCFJ1662 (Addgene #51482, deposited by Erik Jorgensen). Plasmid mixtures containing pCFJ1662_PH, pCFJ601, pMA122 (peel-1; Addgene #34873, deposited by Erik Jorgensen) and co-injection markers were injected into N2 young adult worms, and transgenic strains were obtained by hygromycin selection. To co-express mCherry::TBG-1 (genomic DNA) and mCherry::histone H2B, plasmid pNH33 was generated by connecting two genes with an SL2 trans-splicing sequence and inserting this into pCFJ910 (Addgene #44481, deposited by Erik Jorgensen,) with the pie-1p and pie-1 3′ UTR. Plasmid mixtures containing pNH33, pCFJ601, pMA122 and pBN41 (Addgene #86716, deposited by Peter Askjaer) were injected into N2 young adult worms, and transgenic strains were obtained by NeoR selection.

To construct the gtap-1(tj84) null mutant, the CRISPR/Cas9 system was used with purified Cas9 proteins and single guide RNAs (sgRNAs) generated by in vitro transcription as described by Honda et al. (2017). In brief, Cas9 (2.5 µM final concentration) was injected with 10 µM sgRNA (target site: 5′-CCTAGAGATCCGTTTCCAGCTGT-3′; sequence in italics indicates PAM sequence) and 30 ng/µl ssODN (5′-GACATCTCACGTGCTGATTCTGCAGTATGTCTCACGACACGTTGATATCCGTTTCCAGCTGTTCTACAGGTAATTGAAGATGAACGAAAACTTTTTATA-3′; sequence in italics indicates mutation sites), which introduced a stop codon (at the Arg51 codon) and a cleavage site for EcoRV. For screening, two sgRNAs (2.5 µM each; 5′-CCGATGAGCATGGGATCCAGCCT-3′, 5′-CCAGCCTGATGGAACTTATAAGG-3′; sequence in italics indicates PAM sequence) were co-injected to introduce a mutation in ben-1 to cause benzimidazole resistance (Driscoll et al., 1989). After ∼3–4 days of incubation on the 7.5 µM benzimidazole-containing NGM plates, healthy F1 worms were isolated and checked for the gtap-1 allele by PCR.

To generate the gtap-2(tj92) mutant, 1632 bp of the gtap-2 coding region was substituted with the Prps27::hygR::unc-54 3′ UTR fragment using the CRISPR/Cas9 system. To construct the template plasmid pNH32, ∼500-bp homology arms were amplified that were adjacent to the double digestion sites within the gtap-2 coding region. Plasmids for sgRNAs (pTK73_gtap-2#3 and pTK73_gtap-2#4) were constructed by inserting the target sequences (5′-CCCGGTTGTCAGATGAGGATTT-3′ and 5′-TTCGATTCATTGTGAAGTAGGG-3′, respectively; sequence in italics indicates PAM sequence) into pTK73 as described by Obinata et al. (2018). Plasmids pNH32 (20 ng/µl), pTK73_gtap-2#3, pTK73_gtap-2#4 (50 ng/µl), pDD162 (50 ng/µl, to express Cas9; Addgene #47549, deposited by Bob Goldstein) (Dickinson et al., 2013), and co-injection markers were injected and screened with 4 mg/ml hygromycin B. The endogenously GFP-tagged GTAP-1 and GTAP-2 strains were constructed as described by Dickinson et al. (2015). In brief, homology arms for gtap-1 (upstream, 542 bp; downstream, 761 bp) and gtap-2 (upstream, 588 bp; downstream, 668 bp) that had been amplified from the N2 genomic DNA were inserted into the ClaI-SpeI-cleaved pDD282 (Addgene #66823, deposited by Bob Goldstein; Dickinson et al., 2015) by Gibson assembly. To generate the endogenously GFP-tagged GTAP-1 strain, a plasmid mixture containing pDD162 (50 ng/µl), pDD282_gtap-1 (50 ng/µl) and pTK73_gtap-1_F+52 (30 ng/µl) targeted to the sequence 5′-GGAATTCGATCAATGCACGCAGG-3′ (sequence in italics indicates PAM sequence) and co-injection markers were injected. To generate the endogenously GFP-tagged GTAP-2 strain, a mixture containing Cas9 proteins (2.5 µM), pDD282_gtap-2#1 (50 ng/µl) and a synthesized sgRNA (2.5 µM; Integrated DNA Technologies) targeted to the sequence 5′-ATTAAGCTTTCTAATACATGGGG-3′ (sequence in italics indicates PAM sequence) was injected. All strains were backcrossed with N2 and confirmed by DNA sequencing of the PCR-amplified modified region. The mutant strains gtap-1(tj84) and gtap-2(tj92) were crossed with SA1393, SA1425, and SA772 (Sumiyoshi et al., 2015), respectively. Endogenously GFP-tagged GTAP-1 and GTAP-2 strains were crossed with blos-2(jpn17) to decrease autofluorescence (Niwa et al., 2017).

Young adult worms expressing GFP::TEV::γ-tubulin::FLAG (SA303) were grown synchronously on EPP plates [25 mM potassium phosphate (pH 6.0), 20 mM NaCl, 2% (w/v) peptone, 1 mM MgSO₄, 5 µg/ml cholesterol, 2.5% (w/v) agar]. Approximately 3 million embryos were collected by bleaching (Epstein and Shakes, 1995), washed in lysis buffer [50 mM HEPES pH 7.4, 1 mM EGTA, 1 mM MgCl₂, 100 mM KCl, 10% (v/v) glycerol, 0.05% (v/v) NP40, and 0.1 mM GTP], and frozen in liquid nitrogen. Embryos were suspended in 1 ml lysis buffer containing a protease inhibitor cocktail (Roche) supplemented with 1 mM phenylmethylsulfonyl fluoride and lysed by sonication. After centrifugation at 20,000 g for 10 min, 20 µg anti-FLAG preincubated with 50 µl protein G-conjugated Dynabeads (Thermo Fisher Scientific) was mixed with the supernatant for 3 h at 4°C. Immunoprecipitates were collected using magnets, washed three times with 1 ml lysis buffer, and immunoprecipitated proteins were eluted from Dynabeads by incubating with 200 µM FLAG peptides (Sigma-Aldrich) for 90 min. Elutes were subjected to precipitation with acetone, lysed in SDS–PAGE loading buffer, subjected to gradient SDS–PAGE (5–20% polyacrylamide) and visualized by silver staining. Bands specifically detected in the extract of strain SA303 were excised from the gel and analyzed by mass spectrometry. For a comprehensive analysis, each lane was dissected into 15 pieces using a razor blade, and each gel piece was analyzed by mass spectrometry. Full mass spectrometry results are available at Figshare (https://doi.org/10.6084/m9.figshare.23301683).

The number of peptides for each protein detected by mass spectrometry was compared between the GFP::γ-tubulin::FLAG-expressing and control strains. Proteins with two or more identified peptides were selected to ensure the reliability of the results. Thereafter, proteins were selected if they were enriched by ≥3-fold in strain SA303 compared with the control strain. Finally, proteins that were abundant in embryos and unlikely to have a functional relationship with γ-tubulin (e.g. vitellogenin) were excluded. The remaining proteins were considered as potential GFP::γ-tubulin::FLAG interactors.

The immunoprecipitation against GFP-tagged proteins (TBG-1, GTAP-1 and GTAP-2) was carried out using agarose conjugated with rat anti-GFP antibody (06083, Nacalai Tesque). The 50 µl embryo pellets prepared by bleaching were suspended in 350 µl lysis and rinse buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 1% NP-40, 2.5 mM EDTA with protease inhibitors as described above) and sonicated. After centrifugation for 15 min at 20,400 g, the supernatant was incubated with the anti-GFP antibody-conjugated agarose for 3 h at 4°C. After five washes using lysis and rinse buffer, the samples were eluted with 1 M NaCl, 1% SDS and SDS–PAGE sample buffer for western blotting analysis (uncropped images of western blots are presented in Fig. S6).

To raise rabbit polyclonal antibodies specific for GIP-1 and GTAP-1, a cDNA fragment corresponding to amino acid residues 771–891 of GIP-1 or the full-length GTAP-1 cDNA were inserted into pDEST17 (Thermo Fisher Scientific). The resultant fusion proteins were expressed in Escherichia coli, purified using nickel affinity columns, and used to inoculate rabbits (carried out by Medical and Biological Laboratories). To generate anti-GIP-2, full-length GIP-2 cDNA with a C-terminal V5-His6 tag was inserted into pColdI (TAKARA), and recombinant protein was used to inoculate rabbits and rats (carried out by Medical and Biological Laboratories). Antibodies to GIP-1, GIP-2 and GTAP-1 were affinity purified. Rabbit anti-GTAP-2 was generated by inserting residues 591–823 of GTAP-2 into vector pColdI. Serum from GTAP-2-immunized rabbits was fractionated to obtain an IgG-enriched fraction (carried out by Medical and Biological Laboratories). Rat anti-γ-tubulin was prepared as described by Toya et al. (2011). The secondary antibodies for western blotting were horseradish peroxidase–conjugated anti-rabbit IgG and horseradish peroxidase–conjugated anti-rat IgG (711-035-152 and 712-035-153; Jackson ImmunoResearch).

Yeast two-hybrid analysis was performed using the ProQuest Two-Hybrid system (Thermo Fisher Scientific). Briefly, PCR-amplified cDNA fragments encoding each protein were subcloned in-frame downstream of the GAL4 DNA-binding domain of pDEST32 or the GAL4 activation domain of pDEST22. The constructed bait and prey vectors were confirmed by DNA sequencing.

For expression of γ-tubulin or GIP-2 as the ‘third’ protein – i.e., in addition to the DNA-binding domain-fused and activation domain-fused proteins – the ADH1 promoter was inserted at the HindIII-SmaI site in plasmid pRS426, which contains the URA3 gene; then, the γ-tubulin or GIP-2 coding sequences were inserted downstream of the ADH1 promoter. For expression of γ-tubulin as a ‘fourth’ protein, pAUR112 (TAKARA) carrying the Aureobasidin A resistance gene was digested with BstBI and SmaI to delete the URA3 gene, and then the ADH1 promoter and the γ-tubulin or MZT-1 coding sequence were inserted into the KpnI-SacI site.

For typical yeast two-hybrid assays, yeast strain Mav203 (Thermo Fisher Scientific) was co-transformed with a bait vector and a prey vector and incubated on Sc-Leu-Trp plates (according to manufacturer's procedure). For each pair of vectors transformed, four transformants were picked and characterized on selection plates for 3–4 days at 30°C.

For expression with γ-tubulin as the ‘third’ protein, Mav203 was transformed with three plasmids and spread onto Sc-Leu-Trp-Ura plates. For co-expression with the ‘third’ and ‘fourth’ proteins, Mav203 was transformed with four plasmids and spread onto Sc-Leu-Trp-Ura plates containing 0.5 µg/ml Aureobasidin A (TAKARA).

The strength of protein interactions was accessed by the growth of each yeast strain on the following plates with various concentrations of 3AT (3-amino-1,2,4-triazole; Sigma-Aldrich), an inhibitor of HIS3: Sc-Leu-Trp-His (yeast two-hybrid), Sc -Leu-Trp-Ura-His (yeast three-hybrid), and Sc-Leu-Trp-Ura-His with Aureobasidin A (yeast four-hybrid).

C. elegans eggs (100 µl, ∼30,000/µl, stored at −80°C) were collected by bleaching and suspended in 300 µl lysis buffer containing 50 mM HEPES pH 7.5, 1 mM MgCl₂, 1 mM EGTA, 1 mM β-mercaptoethanol, 100 mM NaCl, 0.1 mM GTP, 1 mM phenylmethylsulfonyl fluoride and protease inhibitors (Roche). A crude extract was prepared by sonicating the eggs six times for 10 s each, followed by centrifugation at 20,000 g for 15 min at 4°C. The supernatant (200 µl) was loaded onto a 3.8-ml, 4–40% sucrose gradient prepared in lysis buffer and centrifuged at 150,000 g for 4 h at 4°C in a MLS-50 rotor (Beckman Coulter). Thereafter, 20 fractions (200 µl) were collected and analyzed by western blotting using affinity-purified rat anti-γ-tubulin (1:1000), rabbit anti-GIP-1 (1:1000), rabbit anti-GTAP-1 (1:300), and rat anti-GFP (1:1000). GIP-2 was probed using rat anti-GIP-2 serum (1:1000) (uncropped images of western blots are presented in Fig. S7). The sucrose gradients were performed three times independently and three replicates were analyzed by western blots.

RNAi was carried out using the soaking method. The following cDNA clones (gifts from Professor Yuji Kohara, National Institute of Genetics, Mishima, Japan) were used as templates to synthesize double-stranded RNAs (dsRNAs): yk1562g08 (tbg-1), yk330f6 (gtap-1) and yk1443f05 (gtap-2). The cDNA inserts were PCR-amplified using primer sets containing vector and T7 primer sequences. Primers Cmo422 and T7 were used to amplify gtap-1, whereas primers T7-ME774 and T7-ME1250 were used to amplify tbg-1 and gtap-2, respectively. The primer sequences were as follows: Cmo422, 5′-GCGTAATACGACTCACTATAGGGAACAAAAGCTGGAGCT-3′; T7, 5′-GTAATACGACTCACTATAGGGC-3′; T7-ME774, 5′-TAATACGACTCACTATAGGGCTTCTGCTCTAAAAGCTGCG-3′; and T7-ME1250, 5′-TAATACGACTCACTATAGGGTGTGGGAGGTTTTTTCTCTA-3′. For mzt-1 and gip-1, the cDNA templates were amplified from a cDNA library (Invitrogen) using gene-specific primers containing the T7 promoter sequence at each 5′ end. Primers for mzt-1 were T7_mzt-1_F (5′-TAATACGACTCACTATAGGGATGAGCGACCCAAAGAAACAC-3′) and T7_mzt-1_R (5′-TAATACGACTCACTATAGGGTCATGACAATGCATTTTCCCG-3′), and primers for gip-1 were T7_gip-1_F (5′-TAATACGACTCACTATAGGGATGCGTCGACAAGGCAGCGAA-3′) and T7_gip-1_R_270AA (5′-TAATACGACTCACTATAGGGCACAGATGTATTCAATAGGTGG-3′). Then, dsRNAs were synthesized in vitro using the RiboMAX T7 Express System (Promega) and purified using the phenol chloride.

Worms at the L4 stage were soaked in 2 mg/ml dsRNA solution at 24°C for 12–24 h. After removal from the dsRNA solution, worms were cultured at 24.5°C for 24 h, then observed. For partial RNAi of tbg-1, worms were soaked in 0.3 mg/ml tbg-1 dsRNA solution at 24.5°C for 2 h. After removal from the dsRNA solution, worms were cultured at 24.5°C for 10–18.5 h and observed. The knockdown efficiencies of GTAP-1 and GTAP-2 were confirmed by staining RNAi-treated embryos with an antibody specific for each target protein (data not shown).

For time-lapse microscopy, embryos expressing fluorescent proteins were mounted on 2% agarose pads. The specimens were imaged using a CSU-X1 spinning disc confocal system (Yokogawa Electric Corp.) mounted on an IX71 inverted microscope (Olympus) controlled by MetaMorph software (Molecular Devices). Images were acquired as described by Toya et al. (2010). Briefly, images were acquired using an Orca-R2 12-bit/16-bit cooled charge-coupled device camera (Hamamatsu Photonics) and a 60×1.30 NA UPlanSApo silicon objective lens without binning and with streaming. To obtain Z-sectioned images, 7–25 Z sections at 1-µm steps were acquired using a 300–500-ms exposure (camera gain, 255) for each wavelength. For time-lapse recording, images were acquired every 30 s. To analyze colocalization, three Z-sectioned images of GFP and mCherry at each Z position were obtained at an interval of 30 s without streaming. To assess the dependency of GTAP-1 and GTAP-2 localization on γ-tubulin, 25 Z-sectioned images were obtained for tbg-1(RNAi) embryos to cover the Z axis of the entire embryo, and single Z-sectioned images of centrosomes were obtained in control embryos to avoid a time lag between the 488-nm and 568-nm excitation.

To quantify the fluorescence intensities of TBG-1, GTAP-1 and GTAP-2 in the centrosomal region, projected images of seven Z-sectioned images with a Z interval of 1 µm around centrosomes were generated using the SUM algorithm in ImageJ software (National Institutes of Health). The average intensity was measured in a circular region of radius 3 µm around each centrosome in the projected images, and the average fluorescence intensity in a cytoplasmic ring-shaped region around centrosomes was subtracted. The cytoplasmic region had an inner radius of 3 µm and an outer radius of 4.2 µm in one-cell embryos. To quantify MTs, the average fluorescence intensity was measured in a circular region of radius 6 µm around each centrosome, and the average fluorescence in a circular region of radius 6 µm outside embryos was subtracted as background. Fluorescence intensities were normalized by the average fluorescence intensity in control embryos. Projection and quantification were performed using ImageJ/Fiji software (https://imagej.net/software/fiji/). Statistical analysis was performed using GraphPad Prism software.

For live-cell imaging of the GFP::PH domain and endogenously tagged GFP::GTAP-1 and GFP::GTAP-2 in the adult gonads, adult worms were treated with 0.0025 mM levamisole in polystyrene polybeads (Polysciences) and mounted on 5% agar pads. For imaging, the same microscopic instruments were used as described above, with a 60×1.3 NA UplanSApo silicon objective lens for the GFP::PH domain and mCherry::TBB-2, and a 100×1.35 NA UplansSApo silicon objective lens for endogenously tagged GFP::GTAP-1 and GFP::GTAP-2. Image projection and tiling were performed using ImageJ/Fiji software.

For imaging mCherry::histone H2B and mCherry::γ-tubulin in adult gonads, the same microscopic instruments were used as described above, with a 60×1.3 NA UplanSApo silicon objective lens except that the camera used was ORCA-Flash 4.0 sCMOS (Hamamatsu Photonics). To quantify the amount of γ-tubulin at the plasma membrane, the fluorescence intensities in the band regions (0.85×8.5 µm) were measured using Prot Profile in Fiji software, and the mCherry::γ-tubulin signals across the plasma membrane were normalized using that of mCherry::histone H2B.

Orthologs of tubulins and the components of the γTuC were searched using BLAST (Altschul, et al., 1990), PSI-BLAST (position-specific iterated BLAST) (Altschul, et al., 1997) and DELTA-BLAST (domain enhanced lookup time accelerated BLAST) (Boratyn, et al., 2012) against human orthologs. The sequence similarity between orthologs was calculated and phylogenetic trees were drawn using ClustalW (EMBL-EBI).

We are grateful to Professors Masakado Kawata and Takashi Makino (Graduate School of Life Sciences, Tohoku University, Sendai, Japan) for discussion pertaining to molecular evolution; Professor Yuji Kohara (National Institute of Genetics, Mishima, Japan) for providing cDNA clones; and Ms Hiroko Sugawara, Ms Makiko Sasaki, Dr Kenji Tsuyama, Dr Satoshi Namai and Ms Yuki Hoshi for helping construct the plasmids and worm strains. Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440).

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260922.reviewer-comments.pdf.