Alpha-tubulin tails regulate axoneme differentiation

Significance

This study systematically deleted the tails of five alpha- and four beta-tubulin genes in C. elegans, finding that the deletion of the alpha-tubulin tail induced ectopic doublet microtubules (MTs) formation. Molecular dynamics simulations suggested that the alpha-tubulin tail hindered the nucleation of the B-tubule on the A-tubule surface. Furthermore, in vitro MTs assembly assay using recombinant tubulin demonstrated that removing the alpha-tubulin tail efficiently promoted doublet MTs assembly. Our results provide in vivo evidence indicating that the tubulin tail, specifically the alpha-tubulin tail, suppresses doublet MTs formation, highlighting its role in maintaining the structural integrity and accuracy of axoneme MT organization, which is essential for proper ciliary differentiation and function.

Abstract

The tubulin tail is a key element for microtubule (MT) functionality, but the functional redundancy of tubulin genes complicates the genetic determination of their physiological functions. Here, we removed the C-terminal tail of five alpha- and four beta-tubulin genes in the C. elegans genome. Sensory cilia typically exhibit an axoneme that longitudinally differentiates into a middle segment with doublet MTs and a distal segment with singlet MTs. However, the excision of the alpha-tubulin tail, but not the beta-tubulin tail, resulted in the ectopic formation of doublet MTs in the distal segments. Molecular dynamics simulations suggest that the alpha-tubulin tail could prevent the B-tubule from docking on the surface of A-tubule. Using recombinant tubulins, we demonstrated that removing the alpha-tubulin tail efficiently promoted doublet MTs formation in vitro. These results reveal the vital and unique contributions of tubulin tails to the structural integrity and accuracy of axoneme MT organization.

Tubulin is crucial for microtubule (MT) assembly and maintenance in eukaryotic cell cytoskeletons (1). It consists of a globular core and a carboxyl-terminal tail (CTT) known as the “tubulin tail,” which undergoes various posttranslational modifications (PTMs) such as detyrosination, polyglutamylation, and polyglycylation. These PTMs affect MT stability, dynamics, and interactions with motor proteins and microtubule-associated proteins (MAPs), thereby regulating cellular functions (2–4). Despite being a small component of the entire protein, tubulin tails exhibit a high degree of sequence variability and adopt disordered structures. These tails protrude from the compact globular core, imparting negative charges to the MT surface. Furthermore, the majority of PTMs on MT are concentrated in this region, altering MT interactions with other proteins (3, 5). Consequently, the tubulin tails play a significant role in shaping the MT behavior and function.

The metazoan genome’s evolutionary path highlights complexity and diversity driven by gene amplification, notably in tubulin genes. These genes have duplicated and diversified, essential for the versatility of multicellular organisms (6). Duplicated tubulin genes may acquire new functions, partition original roles, or maintain primary functions, enhancing genetic network robustness and specialization through adaptation (7). For example, the unicellular budding yeast genome only encodes two alpha-tubulin and one beta-tubulin gene, while Caenorhabditis elegans has expanded to nine alpha- and six beta-tubulin genes, and humans have nine alpha- and nine beta-tubulin genes (2, 3, 8, 9). Tubulin gene duplication generates distinct tubulin isotypes in multicellular organisms, which are expressed in various cell types and developmental stages, contributing to specialized cellular functions, including the axoneme-specific tubulin isotypes for ciliogenesis (10–13). Despite the advantages of tubulin gene duplication, functional redundancy complicates the study of genetically modified tubulin genes. Targeted deletions of the tubulin tail could elucidate resulting cellular phenotypes, yet such genetic dissection of the physiological role of tubulin tails in metazoans remains limited.

In C. elegans, tubulin genes have undergone extensive diversification to generate a large and varied set of tubulin isotypes, which are expressed in specific tissues and developmental stages. This extensive diversification makes C. elegans an ideal model system for studying the functional redundancy and specialization of tubulin isotypes (7, 8). Notably, in C. elegans, different tubulin isotypes contribute to the formation of MTs with distinct protofilament numbers, a feature that is unique to this organism and is crucial for the specialization of MTs in various cellular contexts (7, 8). For example, ciliated sensory neurons in C. elegans express specialized tubulin isotypes such as the alpha-tubulin TBA-5 and beta-tubulin TBB-4. These isotypes contribute to the formation of axonemal MTs, which consist of 13 protofilaments in the A-tubule and 11 protofilaments in the B-tubule. In contrast, nonciliated cells express other tubulin isotypes, such as TBA-1 and TBA-2 (alpha-tubulins) and TBB-1 and TBB-2 (beta-tubulins), which assemble into MTs containing 11 protofilaments. Additionally, C. elegans mechanosensory neurons specifically express the tubulin isotypes MEC-7 and MEC-12, which form MTs with 15 protofilaments, likely contributing to the specialized functions of mechanosensation (7, 8, 14, 15).

The ciliary axoneme provides the track for kinesin- and dynein-mediated intraflagellar transport (IFT), essential for cilia construction and maintenance. During anterograde IFT, kinesin-2 transports IFT particles loaded with tubulins and other ciliary precursors to the ciliary tip, while retrograde IFT, driven by dynein-2, returns ciliary turnover products and anterograde IFT-machinery to the base (16, 17). In vitro studies revealed the role of detyrosination of alpha-tubulin tail for kinesin-2 motility and tyrosination for dynein-dynactin initiation (18, 19). Despite these findings, the in vivo impact of tubulin tails on IFT is not well understood.

Axonemes extend beyond the classic 9+2 configuration with specialized subdomains across cell types. Sensory cilia feature a transition zone (TZ), a middle segment with nine doublet MTs, and a distal segment with singlet MTs crucial for signal detection and cellular function across species from Tetrahymena to mammals (20–22). To differentiate multiple ciliary segments, additional kinesin-2 motor proteins are crucial. In C. elegans chemosensory neurons, heterotrimeric kinesin-II and homodimeric OSMotic avoidance abnormal protein 3 (OSM-3) collaborate in IFT. Initially, kinesin-II begins transport from the base, constructing the axoneme’s middle segment, which is then succeeded by OSM-3-kinesin, exclusively advancing into and building the distal segment (23, 24). In vertebrates, the OSM-3 homolog kinesin-like protein KIF17 (KIF17) plays a vital role in forming distal singlet MTs in zebrafish photoreceptors and transporting signaling proteins to sensory cilia, highlighting the widespread use of homodimeric kinesin-2 in ciliary distal segment formation (25).

Furthermore, the ciliary-specific alpha-tubulin TBA-5 localizes within the cilium, with an enrichment in the distal segments, where it may contribute to the axoneme structure integrity (11). Recent research has reported roles of inner junction proteins CFAP-20 (Cilia and Flagella Associated Protein 20) and PCRG-1 (Parkin Co-Regulated Gene protein) in axoneme assembly. PCRG-1 presence in sensory cilia leads to B-tubule elongation and atypical doublet MTs formation despite cilia length resembling wild type (WT), leading to reduced chemosensory capabilities (26). This underscores the importance of precise MT patterning for ciliary structure and function.

In vitro, subtilisin-mediated proteolysis of the tubulin tail from assembled A-tubules facilitates B-tubule nucleation, indicating an inhibitory role of the tubulin tail in doublet MTs formation (27–29). Removal of the beta-tubulin CTT promotes MT growth and also increases catastrophe rates, regulating the structure and stability of the MT plus end (30). Duan et al. (31) showed that the tail is required for the essential function of both alpha- and beta-tubulin; intriguingly, the two tails are interchangeable, and cells grow normally with either an alpha or a beta tail on both tubulins. Chen et al. (32) revealed that the alpha-tubulin tail is an inhibitor of MT growth, and their molecular dynamics (MD) simulations indicated that the alpha-tubulin tail transiently occludes the longitudinal MT polymerization interface. However, the effects of genetically removing the tubulin tail on doublet MT formation in live animals remain to be fully elucidated.

In this study, we employed genome editing in C. elegans to delete the CTT from the specific tubulin isotypes associated with the axoneme as an initial step. In order to systematically evaluate the contribution of other tubulin isotypes, we extended our approach to delete the CTT from a total of five alpha-tubulin and four beta-tubulin genes. Transmission electron microscopy (TEM) analysis revealed that deletion of the alpha-tubulin tail, but not the beta-tubulin tail, induced the formation of ectopic doublet MTs in ciliary distal segments, even with a single alpha-tubulin tail deletion. MD simulations suggested that the alpha-tubulin tail might prevent nucleation of the B-tubule on the A-tubule surface. To validate this, we generated recombinant tubulins with or without CTT and conducted in vitro doublet MT assembly assays, demonstrating that removal of the alpha-tubulin tail was more effective than the beta-tubulin tail in promoting doublet MTs assembly. Moreover, both tail deletions significantly enhanced doublet MTs formation compared to single tail deletions. These findings underscore the critical and unique role of the tubulin tail in maintaining the structural integrity and precise organization of axoneme MTs, essential for proper differentiation into distinct ciliary segments with specific MT configurations.

Results

Generation of Tubulin Tail Deletions in C. elegans.

Our initial investigation into tubulin gene expression in ciliated sensory neurons of C. elegans involved examining single-cell RNA sequencing databases and relevant literature (Fig. 1 A and B) (8, 15, 33, 34). We then applied CRISPR-Cas9 genome editing to successfully excise the CTT of five alpha- and four beta-tubulin genes that express in the ciliated neurons (Fig. 1C and SI Appendix, Fig. S1). The growth and fertility of these C-terminal tail deletion (ΔCTT) mutants appeared comparable to the WT N2 strain (N > 100 for each genotype), likely due to the functional redundancy of tubulin genes. Next, we generated multiple ΔCTT mutants through various genetic crosses. For tubulin genes located adjacently on the same chromosome, we used genome editing to remove one tubulin tail in the ΔCTT background of another tubulin gene (Materials and Methods). Alpha-tubulin genes TBA-1 and TBA-2, and beta-tubulin genes TBB-1 and TBB-2 are abundantly expressed across virtually all C. elegans cells including germline, embryos, and neurons. We were unable to create heritable homozygous double ΔCTT mutants for either tba-1; tba-2 or tbb-1; tbb-2 combinations. WT embryos continue to develop 6 h after egg laying; however, the embryos of these double ΔCTT mutants lysed (N > 50 for each mutant), indicating a defect in early embryonic development. Nonetheless, we succeeded in obtaining viable double, triple, quadruple, and quintuple ΔCTT mutant animals (Figs. 2 and 3).

Fig. 1.

Construction of the tubulin ΔCTT mutants. (A) Sequence alignments of C. elegans alpha-tubulins (Left) and beta-tubulins (Right). The yellow lines indicate the cleavage sites as previously reported (27). (B) Heatmap showing the expression levels alpha-tubulins and beta-tubulins in C. elegans ciliated sensory neurons. The plot was generated from the CeNGEN website (https://cengen.shinyapps.io/CengenApp/). The size of each circle corresponds to the proportion of neurons in each cluster expressing a particular gene. TPM, transcripts per million. (C) Schematic design of tubulin ΔCTT mutants. The CTT of five alpha- and four beta-tubulins were excised by CRISPR-Cas9 genome editing.

Fig. 2.

Histograms of IFT velocities in WT and tubulin ΔCTT mutants. (Left) Anterograde IFT at the middle segments (m.s.). (Middle) Anterograde IFT at the distal segments (d.s.). (Right) Retrograde IFT. The plots were fit by the Gaussian distribution. N indicates the number of IFT particles analyzed. Comparisons were performed against WT animals by a nonparametric Mann–Whitney U test. n.s., not significant.

Fig. 3.

Cilia and the doublet MT marker in WT animals and tubulin ΔCTT mutants. (A) Amphid and phasmid cilia of the indicated animals were labeled by OSM-6::GFP, while the doublet MTs were labeled by CFAP-20::Scarlet. Arrowheads indicate the ciliary base; the yellow lines indicate the junction between the middle segment and the distal segment in WT cilia. (Scale bar: 5 μm.) (B) The cilium length (mean ± SD) was measured using OSM-6::GFP fluorescence, and the middle segment length (mean ± SD) was measured using CFAP-20::Scarlet fluorescence in the indicated animals. The numbers of cilia analyzed (N) were given on the right of the charts. The numbers of animals analyzed are 21 (WT) 22 (tba-5 ΔCTT); 23 (tbb-4 ΔCTT); 28 (tba-5; tbb-4 ΔCTT); 21 (tba-5; tba-1 ΔCTT); 23 (tba-5; tba-2 ΔCTT); 23 (tbb-4; tbb-1 ΔCTT); 22 (tbb-4; tbb-2 ΔCTT); 24 (tba-5; tba-1; tbb-4; tbb-1 ΔCTT); 19 (tba-5; tba-1; tbb-4; tbb-2 ΔCTT); 21 (tba-5; tba-2; tbb-4; tbb-1 ΔCTT). Error bars in Left and Middle plots represent SD. Comparisons were performed against WT animals by two-tailed Student’s t test. n.s., not significant; ****P < 0.0001.

The Loss of Tubulin Tail Does Not Affect IFT.

To assess whether CTT deletion impacts IFT or cilium length, we genetically introduced a green fluorescent protein (GFP)-tagged IFT marker, IFT52/OSM-6::GFP, into the ΔCTT mutant animals. IFT assays revealed that those ΔCTT mutants aforementioned in the previous section exhibited similar IFT velocities, both anterogradely along the ciliary middle or distal segments and retrogradely, compared to the WT (Fig. 2, SI Appendix, Figs. S2 and S3, and Movie S13). These findings suggest that tubulin tail removal does not influence IFT.

The Ciliary Distal Segments of Alpha-Tubulin ΔCTT Mutants Aberrantly Exhibit the Doublet MT Marker.

To test whether the loss of tubulin tail affects cilia structure, we utilized a well-established method to evaluate cilium length, measuring the distribution distance of OSM-6::GFP reporter GFP fluorescence along the cilia. We observed no significant differences in cilium length between ΔCTT mutants and WT animals (Fig. 3 A and B), corroborating the normal cilia length and unaffected IFT observed in ΔCTT mutants. We then investigated the distribution of axonemal markers in ΔCTT mutant animals. Within doublet MTs, FAP20 and PACRG alternate to form the inner junction bridge, linking the A1 protofilament in the A-tubule to B10 in the B-tubule (35–38). In C. elegans sensory cilia, we utilized Scarlet, a red fluorescent protein, to label endogenous CFAP-20 (26). When coexpressed with IFT-marker OSM-6::GFP, CFAP-20 exhibited concentration in the middle segment containing doublets, but was absent in the distal segment comprising singlet MTs, consistent with its association with the inner junction (Fig. 3A). However, in several ΔCTT mutants, the CFAP-20::Scarlet signal extended aberrantly into the distal ciliary segments of both amphid and phasmid cilia (Fig. 3 A and B). Analysis of genotypic patterns in these mutants revealed that the presence of an alpha-tubulin ΔCTT consistently correlated with this abnormal extension of CFAP-20 into the distal segment. In contrast, we did not observe a statistically significant effect of the beta-tubulin ΔCTT on the distribution of CFAP-20::Scarlet, either in single or double mutants (Fig. 3 A and B). However, we cannot rule out the possibility of a modest effect on the length of the B-tubule, or a potential synergistic interaction between the deletion of the alpha- and beta-tubulin tails. Thus, the CTT deletion in alpha-tubulin, rather than beta-tubulin, causes the ectopic extension of CFAP-20::Scarlet into the ciliary distal domains.

Ectopic Doublet MT Formation in the Ciliary Distal Segments of Alpha-Tubulin ΔCTT Mutants.

To directly visualize the ectopic formation of doublet MTs, we employed TEM to analyze the ultrastructure of amphid cilia in the ΔCTT mutants. Serial sectioning was performed starting from the nematode’s head and advancing in a perpendicular direction along the body axis. The initial 12 µm of sections were examined sequentially, which typically encompassed the entire amphid cilia region. Upon first observing the channel, we determined that we had reached the ciliary region. Then, the sections that came from the following 2.5 µm were inspected, which corresponded to the theoretical distal segment. Sections from the next 4 µm were also examined to find whether ectopic structures were present in the middle segment (Fig. 4A). Our observations confirmed that all ΔCTT mutant animals developed both distal and middle ciliary segments, consistent with our measurements of normal cilium length using the OSM-6::GFP reporter (SI Appendix, Fig. S4). In WT animals, ciliary distal segments typically feature singlet MTs, whereas the middle segments contain nine doublet MTs (Fig. 4 A and B and SI Appendix, Fig. S4). Notably, we observed ectopic formation of doublet MTs in distal segments of the mutants with any alpha-tubulin ΔCTT (Fig. 4 C and E and SI Appendix, Fig. S5A). Moreover, we could occasionally find extra hook-like B-tubules connecting on doublet MTs (SI Appendix, Fig. S5B). Conversely, TEM analysis of mutants with beta-tubulin tail deletions showed hardly any observable doublet MT in their distal segments, corroborating our findings from the CFAP-20::Scarlet reporter (Fig. 4 D and E and SI Appendix, Fig. S5A). Thus, deletion of the alpha-tubulin tail, but not the beta-tubulin tail, resulted in ectopic doublet MT formation in the ciliary distal segments.

Fig. 4.

TEM analysis of ciliary ultrastructure in WT and tubulin ΔCTT mutants. (A) Schematic of the longitudinal ultrastructure of the amphid channel cilia (only four cilia shown). Glial socket cells (magenta) and sheath cells (blue) are shown. (B) Representative TEM images (cross sections) of distal segment of amphid channel cilia in WT animals. High magnification on the right side shows the singlet MTs in one cilium. (C) Representative TEM images (cross sections) of the distal segments of amphid channel cilia in tba-5 ΔCTT, tba-5; tbb-4 ΔCTT, tba-5; tba-1; tbb-4; tbb-1 ΔCTT, and tba-5; tba-1; tbb-4; tbb-2 ΔCTT animals. High magnifications on the right side demonstrate the ectopic doublet MTs in one cilium. (D) Representative TEM images (cross sections) of distal segment of amphid channel cilia in tbb-4 ΔCTT animals. High magnification on the right side shows the singlet MTs in one cilium. (E) The numbers of doublet MTs versus singlet MTs in the distal segment of cilia in WT and mutants. N indicates the cilia examined. [Scale bars in B–D: 200 nm (low magnification), 100 nm (high magnification).]

In Silico Predictions of Alpha-Tubulin Tails in Preventing Doublet MT Formation.

Next, we employed an in silico approach to investigate the effects of tubulin tails on doublet MTs formation. Previous research has indicated that the initiation of doublet MTs formation begins with the attachment of the B01 tubulin dimer to the internal side of the outer junction (27). To assess the tubulin tail dependency of B01 attachment and the energy state of the B01 tubulin dimer under varying ΔCTT conditions, we developed all-atom MD models of the outer junction based on the cryogenic electron microscopy (cryo-EM) doublet MTs structure (PDB ID: 6U42) (37). Each ΔCTT state underwent 100-ns simulations with three independent MD runs. Consistent with previous findings (27), the B01 tubulin dimer in the model composed of full-length alpha- and beta-tubulins (α/β) detached from the outer junction soon after simulation initiation (Fig. 5 A and B and Movies S1–S3). In contrast, in the model with all alpha- and beta-tubulins without CTT (α-ΔCTT/β-ΔCTT), the B01 tubulin dimer remained firmly attached to the outer junction throughout the simulation (Fig. 5 A and B and Movies S10–S12). We then focused on the individual ΔCTT conditions to evaluate the contributions of alpha- and beta-tubulin tails to the energy barrier that prevents B01 attachment. In the alpha-tubulins CTT deletion/full-length beta-tubulins model (α-ΔCTT/β), the B01 tubulin dimer did not remain stationary, it consistently remained attached to the A10-A11 protofilament patch for the majority of the simulation time (Fig. 5 A and B and Movies S4–S6). Conversely, In the full-length alpha-tubulin/beta-tubulin CTT deletion model (α/β-ΔCTT), the B01 tubulin dimer detached rapidly from the outer junction (Fig. 5 A and B and Movies S7–S9). We also quantified the total contact area between the B01 tubulin dimer and the A10-A11 protofilament patch, which reflects the affinity and stability of the A10-A11/B01 interface (Fig. 5C and SI Appendix, Fig. S6A). During the simulations, the α-ΔCTT/β-ΔCTT model exhibited the most stable contact between A10-A11 and the B01 tubulin dimer. In comparison, the α-ΔCTT/β model showed a slight reduction in contact area, indicating that the presence of the β-tubulin tail did not considerably impede the formation of contacts at the outer junction. Conversely, the α/β and α/β-ΔCTT models failed to maintain sufficient contact area for B01 to stably attach to the A10-A11 protofilament, leading to the detachment of the B01 tubulin dimer.

Fig. 5.

MD simulation suggested that CTT of alpha-tubulin contributes the most energy barrier in preventing doublet MT formation. (A and B) Top views (A) and side views (B) of the representative simulation results. The alpha-/beta-tubulin tails are represented by spheres to indicate the Van der Waals radii, while the remaining parts are depicted using a cartoon representation to highlight the secondary structure. The initial snapshot (0 ns) captures the conformation after NPT equilibrium, followed by the snapshots at the end of 100 ns simulations. (C) The contact area between B01 dimer and A10-A11 patch in different cases. The first 50 ns of the simulations were discarded as the equilibration phase. The contact area was calculated every 0.5 ns during the last 50 ns. Different colors indicate independent runs. Comparisons were performed against the case α/β by a nonparametric Mann–Whitney U test; n.s., not significant; ****P < 0.0001. (D) Free energy landscape inferred from the distribution of center of mass coordinate within the 100 ns simulations of each case of the first MD run (the results from the second and third MD runs are presented in SI Appendix, Fig. S6B). The probability density on the X-Y plane was estimated by kernel density estimation and the free energy was calculated by Boltzmann inversion. The origin of the coordinates is the starting position. Three independent simulations for each case were performed for (A–D).

We then calculated the energy barrier imposed by tubulin tails by projecting the trajectory of the center of mass (COM) of B01 tubulin dimer onto the X-Y plane (Materials and Methods) and estimating the free energy landscapes. As shown in Fig. 5D and SI Appendix, Fig. S6B, the immediate detachment of the B01 tubulin dimer in the α/β and α/β-ΔCTT models suggested a high-energy barrier near its initial position (coordinate origin), primarily attributable to alpha-tubulin tail dynamics. In contrast, the α-ΔCTT/β-ΔCTT and α-ΔCTT/β models displayed a confined range of low-energy regions near the initial position. Together, MD simulation results suggest that alpha-tubulin tails, rather than beta-tubulin tails, impede the attachment of B01 to the A10-A11 protofilament.

In Vitro Reconstitution of Doublet MTs from ΔCTT Tubulins.

We sought to perform in vitro assay to dissect the specific roles of an individual tubulin tail in doublet MT formation. Subtilisin serine proteases secreted from Bacillus subtilis have been extensively employed to cleave tubulin tails in vitro (27, 28). However, these proteases remove both alpha- and beta-tubulin tails indiscriminately, hindering the distinction of each tubulin isotype’s role in doublet MTs formation. To address this, we utilized the baculovirus protein expression system to generate engineered recombinant mouse tubulin variants lacking either alpha-tubulin, beta-tubulin, or both tails (Fig. 6 A and B) (39–41).

Fig. 6.

The deletion of tubulin CTT promoted in vitro doublet MT assembly. (A) Schematic diagram for the purification of recombinant tubulins. (B) Coomassie blue staining of purified WT and ΔCTT tubulin dimers. The table on the Right shows the expected molecular weights of WT and the ΔCTT tubulins. (C) Representative images of doublet MTs assembled. WT and ΔCTT tubulins assembled A-tubules are shown in magenta; A-tubule tips and B-tubules formed by Alexa488-labeled tubulin are shown in green. White arrows indicate the B-tubule assembling on the surface of the A-tubule. (Scale bar: 5 μm.) (D and E) Quantification of doublet MTs assembled in vitro. The percentage of the green fluorescence on the red fluorescence (D), and the length proportion of the green fluorescence to the red fluorescence (E) were shown. Four independent doublet MT assembly assays were performed for (C–E). The numbers of examined MTs are 24 (α/β); 63 (α-ΔCTT/β); 56 (α/β-ΔCTT); 70 (α-ΔCTT/β-ΔCTT). Error bars represent SD in (D). Comparisons were performed with the control with a matching color code by two-tailed Student’s t test (D) or nonparametric Mann–Whitney U test (E); ***P < 0.001, ****P < 0.0001. (F) Representative negative-staining EM images of the MTs assembled. White arrowheads indicate the B-tubules. The numbers of B-tubules observed versus MTs examined are 4/73 (α/β); 58/100 (α-ΔCTT/β); 26/78 (α/β-ΔCTT); 45/64 (α-ΔCTT/β-ΔCTT). [Scale bars: 200 nm (low magnification), 50 nm (high magnification).]

Using an established method for doublet MT formation (27), we first polymerized recombinant tubulin mixed with 10% Rhodamine-labeled porcine brain tubulin to assemble A-tubules. Subsequently, we introduced Alexa 488-labeled porcine tubulin to induce B-tubule assembly. Under total internal reflection fluorescence (TIRF) microscopy, we observed the elongation of template A-tubules at both ends in the presence of guanosine triphosphate (GTP). Importantly, we detected patches of green fluorescence from Alexa 488–labeled tubulin on the red fluorescent A-tubules composed of recombinant tubulins lacking the tail of either alpha-tubulin or both tubulins (Fig. 6 C–E and SI Appendix, Fig. S7). To further validate these observations, we examined the same reactions using negative staining TEM and confirmed that these green fluorescent patches indeed corresponded to B-tubules (Fig. 6F). Although we also observed some green fluorescence on the red A-tubules assembled with WT tubulin, the frequency of this observation was much lower than with the tail-less variants (Fig. 6 D and E). We cannot rule out the possibility that tubulin may be associating with the periphery of existing MTs at a low frequency, rather than forming continuous B-tubules. This is supported by our TEM analysis, which shows that WT tubulin does not typically form doublet MTs under these conditions (Fig. 6F). This is also consistent with the notion that WT tubulin alone is insufficient to assemble stable doublet MTs (27–29).

Interestingly, our quantitative analysis of these B-tubules revealed that the frequency of presence and the relative length of B-tubules remarkably increase in the α-ΔCTT/β recombinant tubulins, suggesting that deletion of the alpha-tubulin tail significantly contributes to doublet MT formation (Fig. 6 D and E). Deletion of the beta-tubulin tail (α/β-ΔCTT) also contributes, albeit to a lesser extent but with statistical significance (Fig. 6 D and E). Importantly, the simultaneous deletion of both tails (α-ΔCTT/β-ΔCTT) resulted in an even higher incidence of doublet MT formation compared to the deletion of just the alpha-tubulin tail, indicating an enhancement of doublet MT formation by the deletion of the beta-tubulin tail (Fig. 6 D and E). Furthermore, in our negative-staining TEM experiments, the observed ratios of B-tubules to total MTs were as follows: 4/73 for α/β, 58/100 for α-ΔCTT/β, 26/78 for α/β-ΔCTT, and 45/64 for α-ΔCTT/β-ΔCTT (Fig. 6F), which corroborate our findings from fluorescence microscopy. In conclusion, these results underscore the predominant role of alpha-tubulin tail deletion in promoting doublet MT formation in vitro, while also highlighting the synergistic effect of beta-tubulin tail deletion in this process.

Discussion

The redundancy of tubulin genes has complicated efforts to genetically dissect the physiological roles of the tubulin tail in vivo. Using systematic genome editing, we demonstrated that deletion of the alpha-tubulin tail induces the formation of ectopic doublet MTs in the ciliary distal segment of C. elegans sensory neurons. This provides the initial genetic evidence supporting a role for tubulin tails in axoneme development, specifically in guiding the differentiation into segments with distinct singlet or doublet MT organization. Given the highly conserved nature of tubulin across species, we speculate that the regulatory role of alpha-tubulin observed here may have broad implications for axoneme formation and differentiation across diverse species.

The ciliary axoneme contains a specific subset of tubulin isotypes. For instance, the C. elegans axoneme includes the specific alpha-tubulin TBA-5 and beta-tubulin TBB-4 (11). However, deletion of TBA-5 and TBB-4 only results in minor ciliary defects (11), suggesting that other tubulin isotypes contribute to axoneme formation. Indeed, GFP knock-in strains labeling all tubulin isotypes in C. elegans (15) have revealed that tubulin isotypes TBA-1/2 and TBB-1/2, which are expressed in the germline and other tissues, may also be present in sensory cilia. Our recent tissue-specific labeling of endogenous tubulins provides clear evidence for the presence of TBB-2 in the cilia (42). Based on these findings, we propose that the axoneme is composed not only of cilia-specific tubulin isotypes but also of tubulin isotypes that are widely expressed in the cytoplasm, which helps explain why mutations in any alpha-tubulin ΔCTT lead to the formation of ectopic doublet MTs in the ciliary distal segments.

We observed that deletion of the tail of tba-5 alone allows for doublet MTs formation, suggesting that a segment of the alpha-tubulin tail may be sufficient to inhibit ectopic doublet MTs assembly. Intriguingly, the loss-of-function allele of tba-5 does not result in ectopic doublet MTs formation in the ciliary distal segment. We hypothesize that the absence of tba-5 may be compensated by another alpha-tubulin isotype or multiple isotypes, whose intact tails prevent aberrant doublet MT assembly in the ciliary distal domain. Considering the presence of other alpha-tubulin isotypes in the ciliary distal segment, it is plausible that the remaining alpha-tubulin tails inhibit the nucleation of B-tubules on the surface of A-tubules. This mechanism could explain why we only observed a small proportion of single distal MTs being converted into doublets in the tba-5 ΔCTT mutant allele.

Our findings suggest that the alpha-tubulin tail serves a distinct function from the beta-tubulin tail, which aligns with previous reports indicating that it is the beta-tubulin tail that mediates MT severing by katanin (43). Chen et al. (32) showed that the alpha-tubulin tail inhibits MT growth using recombinant, engineered human tubulins, which suggests that the alpha-tubulin tail not only hinders B01 attachment to the A10-A11 protofilament, but also contributes to the growth and stability of the B-tubule on the A-tubule. This insight is consistent with our observation that the removal of the CTT from alpha-tubulin facilitates doublet MTs formation and may promote more stable and efficient B-tubule assembly.

The production of recombinant tubulin is known to have several challenges, including the potential effects of affinity tags or the “scar” residues left after tag removal. In our study, we used 6×His for alpha-tubulin and Flag for beta-tubulin as affinity tags during two-step purification. These tags were not removed after purification. Although the recombinant tubulin generated the expected doublet MTs in vitro, we observed that the B-tubules were not continuous, suggesting a limited efficacy in forming doublet MTs compared to those in the ciliary axoneme. Future studies should focus on improving recombinant tubulin production by generating scarless versions that only alter the designed region.

Our in vitro experiments successfully generated B-tubules; however, we observed that they formed small patches rather than continuous B-tubules along the A-tubules. Because of the limited yield of recombinant tubulin, we followed the published protocols to assemble MTs by mixing recombinant tubulin with WT brain tubulin. This mixture might have hindered the formation of continuous B-tubules, as the tail from the full-length brain tubulin could not support the formation of B-tubules.

We also noticed that the in vitro assembled doublet MTs are smaller than those observed in cilia, a finding also reported in earlier studies using tail-less tubulin to assemble doublet MTs (27). One explanation for the smaller size is that the assembled B-tubule may be incomplete or unstable, containing fewer than 11 protofilaments. Additionally, we propose that other factors—such as accessory proteins or regulatory mechanisms—may be required to stabilize doublet MTs in vivo and facilitate their continuous growth in the axoneme.

Recently, a MAP MAPH-9 has been reported to exclusively localize to the doublet MTs in C. elegans sensory cilia (44). By examining the recruitment of MAPH-9 to the cilia in different ΔCTT mutants, we did not find any apparent changes in MAPH-9 localization in these mutant cilia, suggesting that tubulin tail may be dispensable for MAPH-9 localization (SI Appendix, Fig. S8). Distinct from CFAP-20, MAPH-9 does not extend to the ciliary distal segments that contain ectopic doublet MTs in the ΔCTT mutants, which suggests that these ectopic doublets may not fully resemble the WT ones. Future studies will examine additional ciliary protein localization with doublet MTs to explore further insights into doublet MT formation in vivo. We speculate that certain MT inner proteins (MIPs) interact with and displace the alpha-tubulin tail from the A11 protofilament, thereby promoting the formation of B-tubules. Examination of cryoelectron tomography images of axonemes may help identify potential candidate MIPs involved in this process (37, 38, 45–48).

Materials and Methods

C. elegans Strains and Culture.

C. elegans strains were cultured on nematode growth medium plates seeded with Escherichia coli strain OP50 at 20 °C. All animal experiments were conducted in compliance with governmental and institutional guidelines. A summary of all strains and plasmids used in this study is provided in SI Appendix, Table S1.

Molecular Biology.

To generate tba-1, tba-2, and tbb-2 ΔCTT mutants, we inserted the CRISPR/Cas9 target sequence into the pDD162 vector (Addgene #47549) by PCR linearization. Approximately 100-nucleotide oligonucleotides were used as homology recombination. The CRISPR/Cas9 target and oligonucleotide sequences used are listed in SI Appendix, Table S1. CRISPR/Cas9 constructs and HR templates, along with rol-6[su1006] markers, were microinjected into the germline of young adult WT hermaphrodites at a concentration of ~50 ng/μL. The mutant strains were confirmed by PCR and Sanger sequencing. We injected the overexpression constructs into WT N2 or tba-5, tbb-4, tba-9, tba-6, ben-1 ΔCTT (PHX5594) strain at 10 ng/μL for Pdyf-1::osm-6::Scarlet or Pdyf-1::klp-11::GFP with the rol-6 marker at 25 ng/μL.

Live-Cell Imaging.

Young adult C. elegans hermaphrodites were anesthetized with 0.1 mmol/L levamisole in M9 buffer, mounted on 5% agar pads, and maintained at 20 °C. Agar pads were prepared 1 d before the experiment to ensure optimal immobilization of the worms. Images were captured using a spinning disk confocal system, which includes an Axio Observer Z1 microscope (Carl Zeiss) equipped with a 100×, 1.46 numerical aperture (NA) objective lens, an EMCCD camera (iXon+ DU-897D-C00-#BV-500; Andor Technology), and the 405 nm, 488 nm, and 568 nm lines of a Sapphire CW CDRH USB Laser System (Coherent) with a spinning disk confocal scan head (CSU-X1 Spinning Disk Unit; Yokogawa Electric Corporation). The confocal system was controlled by μManager (https://www.micro-manager.org), and time-lapse images were acquired using a 200 ms exposure time.

Image Processing and Analysis.

The images were processed using the ImageJ software (http://rsbweb.nih.gov/ij/) to generate kymographs and quantify cilium length and IFT velocity. To measure cilium length, z-stack images were acquired and processed using maximal intensity projection. To analyze IFT velocity, we employed the established method (49). Briefly, separated anterograde and retrograde kymographs were generated with the KymographClear toolset plugin in ImageJ by manually drawing lines along cilia at the width of 3 pixels. IFT velocities were calculated by determining the total moved distance of particles in the given time period. Then, 80 to 120 particles from about 10 strains were analyzed. We performed a nonparametric Mann–Whitney U test to compare the velocity distributions for data that do not follow a normal distribution.

Expression and Purification of Mouse Recombinant ΔCTT Tubulins.

Mouse recombinant tubulins’ expression and purification were adapted from refs. 41 and 50. To generate ΔCTT tubulins, the plasmid (pFastBac mTuba1a(His)-mTubb2a(flag)) was subjected to PCR to remove the CTT (refer to Fig. 1A). We used the Bac-to-Bac system (Life Technologies) to generate recombinant baculovirus. SF9 cells (Life Technologies) were cultured in Sf-900™ II SFM (Thermo Fisher 10902088) supplemented with penicillin and streptomycin. For protein expression, 1L of SF9 cells was grown to a density of 2.0 × 10⁶ cells/mL and infected with P2 viral stocks. Cells were maintained in suspension at 28 °C and harvested at 60 h after infection. The purification process was performed at 4 °C. First, cells were collected and lysed by sonication in 50 mL lysis buffer [80 mM PIPES pH 6.9, 100 mM KCl, 1 mM MgCl2, 1 mM ethylene glycol tetraacetic acid (EGTA), 0.1 mM GTP, 0.5 mM ATP, 1 mM phenylmethylsulfonyl fluoride (PMSF), and one EDTA-free protease inhibitor tablet (Roche)]. The lysate was centrifuged at 503,000 g for 1 h at 4 °C using a Type 70 Ti Rotor. The supernatant was loaded onto a 1 mL nickel-nitrilotriacetic acid column pre-equilibrated with lysis buffer. The column was washed with 30 mL wash buffer (lysis buffer supplemented with 25 mM imidazole). Primary tubulin was eluted with 10 mL elution buffer [1× BRB80 (80 mM PIPES, 1 mM MgCl2, 1 mM EGTA), 300 mM imidazole, 0.1 mM GTP, pH 7.0]. The eluate was diluted with an equal volume of BRB80 buffer supplemented with 0.1 mM GTP, and mixed with 1 mL anti-Flag antibody-conjugated beads for 2 h. After incubation, the beads were washed with 10 mL [1× BRB80, 150 mM imidazole, 0.1 mM GTP, pH 7.0], followed by 10 mL [1× BRB80, 50 mM imidazole, 0.1 mM GTP, pH 7.0], and 10 mL [1× BRB80, 0.1 mM GTP, pH 7.0]. Flag-tagged tubulin was eluted with 5 mL BRB80 buffer supplemented with 0.2 mg/mL 3× Flag peptide. Tubulin was concentrated and desalted using an Amicon Ultracel-30K filter (Millipore) with BRB80 and 0.1 mM GTP. The concentration of the tubulin dimer was determined by Bradford assay. Small aliquots of tubulin were frozen in liquid nitrogen and stored at −80 °C. The purified recombinant ΔCTT tubulins were analyzed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel followed by Coomassie blue staining. The affinity tags (6×His and Flag) were not removed after purification.

Doublet MT Assembly Assay.

A two-rounds MT polymerization assay was performed to generate doublet MTs. In the first-round polymerization, 4.5 μL of 5 mg/mL purified recombinant tubulins were mixed with 0.5 μL of 5 mg/mL rhodamine-labeled porcine brain tubulin. GTP (final concentration 1 mM), MgCl₂ (final concentration 4 mM), and dimethyl sulfoxide (DMSO) (final concentration 4%) were added to the mixed tubulin and preincubated for 5 min on ice. MTs were then polymerized at 37 °C for 30 min. Following polymerization, the MTs were stabilized in BRB80-Taxol buffer (20 μM taxol in BRB80) and centrifuged at 20,000 g using a tabletop centrifuge for 15 min at room temperature. The MTs pellet was resuspended in 20 μL BRB80-Taxol buffer.

In the second-round polymerization, 4.5 μL of MTs generated in the first-round polymerization were mixed with 0.5 μL of 5 mg/mL Alexa488-labeled porcine brain tubulin. GTP (final concentration 1 mM), MgCl₂ (final concentration 4 mM), and DMSO (final concentration 4%) were added to the mixed tubulin and preincubated for 5 min at room temperature. Then the polymerization, centrifugation and resuspension were performed as the first-round polymerization.

TIRF Microscopy and Analysis.

Flow chambers were prepared following the established protocol (51). Channels were treated with anti-FLAG antibody solution (1:100 dilution in PBS) for 5 min, followed by 10 min incubation with 1% Pluronic F127. The channels were washed with 100 μL BRB80 buffer. Subsequently, two-rounds polymerized MTs were flowed into the channels. Images were captured using a TIRF system, consisting of an Olympus IX83 microscopy equipped with a 150× (NA 1.45 oil, Olympus) objective lens, an ORCA-Flash4.0V3 camera, 488 nm and 561 nm laser. The system was controlled by Micro-Manager 2.0.

To quantify the doublet MTs assembly in vitro, images from four times individual experiments were analyzed. For the percentage of doublet MTs assembled from WT and ΔCTT tubulins, rhodamine-labeled MTs were counted as total MTs, and Alexa488+rhodamine double-labeled MTs were counted as doublet MTs. The length of rhodamine-labeled MT was measured as A-tubule and the length of double-labeled MT was measured as B-tubule of doublet MT.

TEM of C. elegans.

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

TEM analysis focused on the amphid sensory cilia in the nematode’s head. Serial sectioning was performed starting from the nematode’s head, perpendicular to the body axis, and the sections were examined in sequence. Upon first observing the channel, we determined that we had reached the ciliary region. Then, the sections that came from the following 2.5 µm were inspected, which corresponded to the theoretical distal segment. Sections from the next 4 µm were also examined to find whether ectopic structures were present in the middle segment.

Negative Staining of Polymerized MTs.

Samples (5 μL) from different reconstruction systems were applied to glow-discharged carbon grids and negatively stained with 2% uranyl acetate for 3 × 1 min. The specimens were examined in an FEI Tecnai G2 Spirit (120 kV) electron microscope (Thermo Fisher Scientific Co.).

MD Simulations.

Molecular models of GTP–tubulin dimer docking onto the A10-A11 protofilament patch were based on 6U42 PDB structures (37). The structures of TBB-4 and TBA-5 tubulin binding to Mg²⁺-GTP predicted by AlphaFold 3 (www.alphafoldserver.com) (53) were used as alpha- and beta-tubulin. The tubulins were then aligned to the outer junction of the 6U42 model. Protonation state of ionizable amino acid residues was predicted by PDBFixer (54). To simplify the model, the ligands on A10-A11 protofilament were removed. These initial models are available in Zenodo (55).

A virtual three-dimensional cubic reaction volume filled with TIP3P water with periodic boundary conditions was used for the simulation. The size of the reaction volume was set in such a way that the distance from the protein surface to the nearest box boundary was not initially less than two nanometers. The ionic strength of the solution was set at 100 mM by adding K⁺ and Cl⁻ ions and the total charge of the system was zero. The whole system was constructed as above, containing about 600,000 atoms in each case. Simulations were performed using the GROMACS 2024.2 software package with the CHARMM36m force field (56). The parameters of the GTP were generated using CHARMM General Force Field (CGenFF) by CHARMM-GUI (57, 58).

After preparing each of the tubulin systems as described above, steepest descent algorithm was used to minimize the energy of the whole system. First, 100 ps long simulations were conducted with V-rescale thermostat (τ = 0.1 ps, T = 300 K). Second, we carried out another 100 ps simulations with the same V-rescale thermostat and C-rescale barostat (τ = 2.0 fs, compressibility = 4.5 × 10⁻⁵ bar⁻¹, pressure = 1 bar). Position restrain was applied on protein heavy atoms during the two steps above. The production simulation runs were carried out in NPT ensemble at 300 K, using V-rescale thermostat and C-rescale barostat for a duration of 100 ns each. In the simulations, positional restraints were applied to all Cα atoms, except for the tubulin tails (refer to Fig. 1A) of the tubulins in the A10-A11 protofilament patch, to mimic the stable state of the A-tubule as a MT wall fragment. No positional restraints were applied to B01 dimer so that the interaction between B01 and A10-A11 could be observed. The Particle Mesh Ewald method was used to treat the long-range electrostatics. H-bond LINCS constraints allowed MD with 2 fs time step. PyMOL (version 2.0 Schrödinger, LLC) was used for visualization.

Free Energy Landscape Estimation.

A Cartesian three-dimensional coordinate system was established: The Z-axis was oriented along the MT, pointing toward its plus end. The X-axis lay tangential to the MT, while the Y-axis was perpendicular to the X-axis. The origin of the coordinate system was defined as the initial COM of B01. During the 100 ns simulation, the COM coordinates of B01 was projected to the X-Y plane. The probability density was estimated by Kernel Density Estimation with Gaussian kernel. Bandwidth was determined. The probability density of (x, y) coordinate on the X-Y plane can be estimated as

$$P\left(x,y\right)=\frac{1}{n}{\displaystyle \mathbf{\sum }_{i=1}^n}\frac{1}{2\pi w_x{w}_y}\exp \left(-\frac{(x-x_i{)}^2}{2w_x^2}-\frac{(y-y_i{)}^2}{2w_y^2}\right).$$

Here, the bandwidth w is determined by Scott’s method:

$$w=\sigma n^{-\frac{1}{d+4}},$$

and the free energy was inferred from the probability density by Boltzmann inversion:

$$E\left(x,y\right)=-k_{B}TlnP(x,y).$$

Contact Surface Area Calculation.

The total contact area between B01 and the A-tubule was calculated by estimating the Solvent Accessible Surface Area (SASA) of each component in the complex. We define $A_{a-b}$ as the contact area between components a and b, and $S_a$ as the SASA of component a. The total contact area was computed using the following equation:

$$\begin{array}{c}{A}_{\mathit{total}}=A_{A10.11-B01}-A_{\text{tails}-B01}\hfill \\ =(S_{A10.11}+S_{B01}-S_{\text{complex}})-(S_{\text{tail}}+S_{B01}-S_{\text{tail}-B01 \text{complex}})\\ =S_{A10.11}+S_{\text{tail}-B01 \text{complex}}-S_{\text{complex}} -S_{\text{tail}}\end{array}.$$

SASA values were computed using the GROMACS SASA module with default parameters.

Supplementary Material

Appendix 01 (PDF)

Movie S1.

Molecular dynamics simulation of the α/β model (top view and side view, 100 ns, run 1). Note the B01 tubulin dimer detaches from the A10-A11 protofilament patch.

Movie S2.

Molecular dynamics simulation of the α/β model (top view and side view, 100 ns, run 2). Note the B01 tubulin dimer detaches from the A10-A11 protofilament patch.

Movie S3.

Molecular dynamics simulation of the α/β model (top view and side view, 100 ns, run 3). Note the B01 tubulin dimer detaches from the A10-A11 protofilament patch.

Movie S4.

Molecular dynamics simulation of the α-ΔCTT/β model (top view and side view, 100 ns, run 1). Note the B01 tubulin dimer remains attached to the A10-A11 protofilament patch.

Movie S5.

Molecular dynamics simulation of the α-ΔCTT/β model (top view and side view, 100 ns, run 2). Note the B01 tubulin dimer remains attached to the A10-A11 protofilament patch.

Movie S6.

Molecular dynamics simulation of the α-ΔCTT/β model (top view and side view, 100 ns, run 3). Note the B01 tubulin dimer remains attached to the A10-A11 protofilament patch.

Movie S7.

Molecular dynamics simulation of the α/β-ΔCTT model (top view and side view, 100 ns, run 1). Note the B01 tubulin dimer detaches from the A10-A11 protofilament patch.

Movie S8.

Molecular dynamics simulation of the α/β-ΔCTT model (top view and side view, 100 ns, run 2). Note the B01 tubulin dimer detaches from the A10-A11 protofilament patch.

Movie S9.

Molecular dynamics simulation of the α/β-ΔCTT model (top view and side view, 100 ns, run 3). Note the B01 tubulin dimer detaches from the A10-A11 protofilament patch.

Movie S10.

Molecular dynamics simulation of the α-ΔCTT/β-ΔCTT model (top view and side view, 100 ns, run 1). Note the B01 tubulin dimer remains attached to the A10-A11 protofilament patch.

Movie S11.

Molecular dynamics simulation of the α-ΔCTT/β-ΔCTT model (top view and side view, 100 ns, run 2). Note the B01 tubulin dimer remains attached to the A10-A11 protofilament patch.

Movie S12.

Molecular dynamics simulation of the α-ΔCTT/β-ΔCTT model (top view and side view, 100 ns, run 3). Note the B01 tubulin dimer remains attached to the A10-A11 protofilament patch.

Movie S13.

IFT processes in phasmid cilia in WT and different tubulin ΔCTT animals.

Data, Materials, and Software Availability

Dataset data have been deposited in Zenodo (DOI: 10.5281/zenodo.14785723) (55). All other data are included in the article and/or supporting information.