Abstract
Tubulins, the building block of microtubules (MTs), play a critical role in both supporting and regulating neurite growth. Eukaryotic genomes contain multiple α and β tubulin isotypes, and their dominant missense mutations cause a wide range of neurodevelopmental defects in humans. Thus, identifying the specific functions of each tubulin isotype and evaluating how particular mutations would change those functions during neuronal morphogenesis is fundamental for understanding MT functions in the nervous system. Using the C. elegans touch receptor neurons (TRNs), we analyzed the effects of 67 missense mutations in the mec-12/α-tubulin and mec-7/β-tubulin genes on neurite growth. Three types of mutations emerged: 1) loss-of-function mutations, which are found throughout the molecule and cause mild defects in neurite growth; 2) antimorphic mutations, which map to the GTP binding site and intradimer and interdimer interaction interfaces and significantly reduced MT stability, causing severe defects in the growth of all TRN neurites; and 3) neomorphic mutations, which map to the exterior surface and increased MT stability, causing ectopic neurite growth. Such structure-function analysis revealed a causal relationship between tubulin structure and interactions and MT stability, which in turn affects neuronal morphogenesis. Importantly, we engineered several disease-associated human tubulin mutations into C. elegans genes and examine their impact on neuronal development at cellular level. We also discovered a MT-destabilizing α-tubulin isotype TBA-7, whose loss led to the formation of hyperstable MTs and the generation of ectopic neurites; the lack of potential sites for polyamination and polyglutamination on TBA-7 may be responsible for this destabilization.
Introduction
Microtubules (MTs) play important roles in many aspects of neurite development, including the formation, extension, guidance, and maintenance of neurites (reviewed in Dentet al. 2011; Prokop 2013; Sainath and Gallo 2015). The “search and capture” model (Mitchison and Kirschner 1984) suggests that the dynamic instability of MTs allows them to explore the growth cone periphery until they are captured by stabilized actin filaments or membrane receptors enriched at the side of the growth cone responding to a guidance cue (Tanakaet al. 1995; Challacombeet al. 1996; Schaeferet al. 2008; Quet al. 2013). This capture transiently stabilizes MTs against catastrophe and enables MT elongation in the direction that the growth cone has turned. In addition to providing physical support for the neurite growth that follows the changes in actin dynamics, MTs also play an instructive role for neurite guidance. Since local application of the MT-stabilizing drug paclitaxel (also known as taxol) induced growth cone attraction and of the MT-destabilizing drug nocodazole induced repulsion (Buck and Zheng 2002), signals that act by altering MT stability appear to directly initiate growth cone turning. Indeed, the guidance molecule Wnt can induce growth cone remodeling by changing the organization of MT structure through the inactivation of MT-plus end binding protein Adenomatous Polyposis Coli (Purroet al. 2008). These results indicate that the regulation of MT dynamics is crucial for neurite growth and guidance, but how MT stability is controlled either locally at specific sites of the growth cone or globally in the entire neuron is not well understood.
As the building blocks of MTs, α-and β-tubulins are crucial determinants of MT stability. Eukaryotic genomes contain multiple tubulin genes encoding different isotypes (Sullivan 1988; McKeanet al. 2001) that are expressed in spatially and temporally distinct patterns (Leandro-Garciaet al. 2010). Different tubulin isotypes confer specific dynamic properties on MTs in tubulin polymerization assays in vitro (Pandaet al. 1994) and are not functionally equivalent in vivo in Drosophila (Hoyle and Raff 1990) or in single-celled organisms, such as Tetrahymena thermophile (Pucciarelliet al. 2012). These observations support the “multi-tubulin hypothesis,” which proposes that distinct tubulin isotypes impart specific properties onto MTs, so they can perform particular cellular functions (Fulton and Simpson 1976; Cleveland 1987). Moreover, tubulins also undergo a range of complex post-translational modification, which affect the dynamics of MTs and their interaction with other proteins (reviewed by Song and Brady 2015). In neurons specifically, stable axonal MTs are more detyrosinated, acetylated, and glutamylated, whereas the dynamic MTs in the growth cone are more tyrosinated (Liao and Gundersen 1998; Konishi and Setou 2009). Tubulin isotypes and modifications at the C-terminal tail also control the velocity and processivity of MT motor proteins and microtubule depolymerization rates in vitro (Sirajuddinet al. 2014). Therefore, post-translational modifications on different tubulins by a set of enzymes may generate a “tubulin code” that can be read by microtubule-associated proteins (MAPs), which regulate the organization of MTs and their interaction with other cellular components (Verhey and Gaertig 2007; Yuet al. 2015).
The clinical importance of tubulin genes in the development of the nervous system has been highlighted by the recent finding that patients carrying mutations in α- and β- tubulin genes showed microcephaly, lissencephaly, pachygyria, and other cortical malformations, as well as a range of axon guidance defects, including agenesis or hypoplasia of the corpus callosum, internal capsule, commissural fibers, and corticospinal tracts, depending on the mutations (Tischfieldet al. 2011). So far, based on three summaries (Bahi-Buissonet al. 2014; Liu and Dwyer 2014; Chakrabortiet al. 2016), 60 point mutations in α-tubulin genes (51 in TUBA1A, 1 in TUBA3E, and 8 in TUBA4A), one splicing-affecting intronic deletion in TUBA8, 48 point mutations in β-tubulin genes (2 in TUBB2A, 24 in TUBB2B, 19 in TUBB3, and 3 in TUBB5), and one exonic deletion in TUBB2B are known to cause tubulin-related neurological disorders in heterozygous carriers (Table S4); all mutations, except for the TUBA8 deletion, appear to have dominant-negative effects. Although mutated residues are found throughout the molecules, many are found in region predicted to mediate either GTP binding, heterodimer stability, inter-dimer interaction, or association with motor proteins and other MAPs (Tischfieldet al. 2011). Despite some in vitro studies on the effects of the mutations on tubulin folding, heterodimer assembly, and MT growth (Jaglinet al. 2009; Tianet al. 2010; Tischfieldet al. 2010), very few in vivo studies can systematically test how these different tubulin mutations impact axon guidance and extension in living organisms. Moreover, the complexity of the mammalian nerve system makes the analysis of the developmental consequences of these tubulin mutations very difficult.
Here, we use the morphologically simple and well-defined touch receptor neurons (TRNs) in the nematode Caenorhabditis elegans to model the effects of tubulin mutations on neurite growth. By analyzing a large collection of missense mutations in several tubulin genes, we found that these mutations caused three morphologically distinct defects in TRN neurite outgrowth: 1) the shortening of all TRN neurites; 2) the specific shortening of posteriorly-directed neurites; and 3) the production of ectopic posteriorly-directed neurites. The structural location of the mutated residue correlated with the resulting phenotype. Many tubulin mutations characterized in our study affect the same amino acid residue or region as the disease-causing mutations. We generated several such human mutations in C. elegans tubulin genes through genome editing and found that they also caused distinct neurite growth defects that fall into the above categories. Thus, our system may be used to understand the different effects of the clinically identified tubulin mutations and to facilitate their classification. Moreover, we found that null mutations in two α-tubulin genes led to very different phenotypes, supporting the hypothesis that tubulin isotypes perform specific and often non-overlapping roles in neurite development.
Results
A genetic screen for TRN neurite growth defects
The six mechanosensory TRNs (ALML/R, PLML/R, AVM, and PVM) in C. elegans are a useful model to study axonal outgrowth and guidance because of their well-defined morphology (Chalfie and Sulston 1981). The ALM and PLM neurons are two pairs of embryonically derived, bilaterally symmetric cells, whereas the AVM and PVM neurons arise from postembryonic lineages. All six neurons have a long anteriorly-directed neurite (AN); in addition, the two PLM neurons have a posteriorly-directed neurite (PN), making them bipolar. Except in PVM, the AN branches at its distal end; we refer to this branch as the synaptic branch.
Previous studies had identified several components needed for proper TRN outgrowth: UNC-6/Netrin (Hedgecocket al. 1990), LIN-44/Wnt proteins (Hilliard and Bargmann 2006), and a few other signaling molecules (Du and Chalfie 2001). To systematically search for genes involved in the regulation of neurite growth, we performed an extensive genetic screen to isolate mutants with defects in TRN neurite extension. We mutagenized a strain (TU4069) carrying the uIs134 transgene, which allows RFP expression from the TRN-specific mec-17 promoter, and screened the progeny of individual F1 animals representing 20,200 haploid genomes. Although we focused on isolating mutants whose TRNs had morphological defects, we also obtained mutants with abnormal numbers of fluorescently labeled cells. This screen yielded 80 mutants. We identified the phenotype-causing mutations for 62 mutants; the remaining mutants either were of low penetrance (7) or did not yield a confirmable mutation (11). The characterized mutations represented 26 genes. Mutations in 8 genes caused the loss of TRN marker expression or its expression in extra cells, and mutations in 2 genes specifically affected the development of the postembryonic AVM and PVM neurons (Table S1); most of these mutant phenotypes were previous described and thus are not discussed here. Mutations in the remaining 16 genes resulted in various TRN neurite extension defects, which were organized into seven phenotypic categories (A-G, Table S2). This screen is likely to be near saturation because 1) the number of haploid genomes examined is 10 times the reciprocal of the average mutation rate (5 × 10−4) for null alleles (Brenner 1974); and 2) nine of the ten genes represented by multiple alleles had three or more alleles.
The 16 genes whose products regulate neurite outgrowth and guidance (Table S2) encode proteins that affect either extracellular signaling to the TRNs, intracellular signaling, or effectors that are directly involved in cytoskeleton rearrangement and growth cone movement. Molecules affecting extracellular signaling include the guidance protein LIN-44/Wnt and its receptor LIN-17/Frizzled, and three proteins, UNC-23, MUA-3, and SUP-26 not acting within the TRNs. PLM neurites in lin-44 and lin-17 mutants failed to navigate towards the anterior (category D) because of the repellent activity of the Wnt signal (Hilliard and Bargmann 2006; Zhenget al. 2015a). unc-23 (category G) and mua-3 (category B) are only expressed in muscle cells (Bercheret al. 2001; Papsdorfet al. 2014), and their mutant phenotypes could not rescued by TRN-specific expression of the wild-type gene. Similarly, mutations in sup-26 (category B), which encodes a RNA-binding protein and is expressed in most somatic cells (Mapeset al. 2010), could not be rescued by TRN-specific expression of the wild-type gene. We have not investigated how these molecules affect neurite outgrowth and guidance in a non-cell autonomous manner.
Genes encoding molecules that are presumably involved with intracellular signaling include unc-51, mec-15, dsh-1, egl-5, unc-73, and tiam-1. unc-51 encodes a serine/threonine-protein kinase; mutation of unc-51 caused general outgrowth and guidance defects (category A; Du and Chalfie 2001). mec-15, which is needed for touch sensitivity (Au and Chalfie, 1989), encodes a F-box protein with WD repeats and so is likely to mediate ubiquitination and protein degradation (category E; Bounoutaset al. 2009b). Mutation of mec-15 caused shortening of both PLM neurites. We previously reported that DSH-1/Dishevelled (category B) negatively modulates the activity of Wnt signaling to allow posterior outgrowth against the Wnt gradients (Zhenget al. 2015a) and that egl-5 (category B), which encodes a Abd-B-like Hox transcription factor, promotes PLM differentiation by inducing the growth of PLM-PN and at least one downstream effector (Zhenget al. 2015b). We also identified guanine nucleotide exchange factors (GEFs) [UNC-73/Trio (category E) and TIAM-1 (category B)], which controls neurite extension towards the anterior and posterior, respectively (Zhenget al. 2016).
In this study, we describe the role of the downstream effectors of directional neurite outgrowth, the presumed targets of extracellular and intracellular signaling: primarily tubulin isotypes (MEC-7/β-tubulin, MEC-12/α-tubulin, and TBA-7/α-tubulin) and kinesin motor proteins (KLP-7 and KLP-11). MEC-7 and MEC-12 are expressed at high levels in the TRNs (Hamelinet al. 1992; Mitaniet al. 1993), which contain large diameter (15-protofilament, 15-p) MTs instead of the typical 11-protofilament (11-p) MTs found in all other cells (Chalfie and Thomson 1979; Savageet al. 1994). Combining the alleles isolated in our screen, created through genome editing, and previously obtained by our lab (Savageet al. 1994) and the million mutation project (Thompsonet al. 2013), we collected in total 78 alleles representing 48 different missense mutations in mec-7 and 19 in mec-12 (Table 1) and found that these mutations led to various outgrowth defects (category A, B, and C defined in Table S2).
mec-7/β-tubulin, but not mec-12/α-tubulin, is needed to form 15-p MTs
Previous studies suggested that the α-tubulin MEC-12 and the β-tubulin MEC-7 form the TRN-specific 15-p MTs that assemble into bundles to fill up the TRN neurites and are required for the mechanosensory functions of TRNs (Chalfie and Thomson 1979; Chalfie and Au 1989; Savageet al. 1994; Bounoutaset al. 2009a). We first examined the MT structures of the mec-12(tm5083) and mec-7(ok2152) deletion alleles using electron microscopy (EM; Figure 1A and S1). mec-12 and mec-7 knockout animals had on average 7 and 6 MTs, respectively, in a cross section of the ALM neurite compared to 31 MTs in the wild type animals (Figure 1B). MTs in mec-12(tm5083) animals had the same diameter as the 15-p MTs in the wild-type animals, whereas the diameter of MTs in mec-7(ok2152) animals was much smaller (16.6 nm; Figure 1). These results suggest that although the abundance of MTs is dependent on both MEC-12 and MEC-7, the formation of 15-p MTs requires MEC-7 but not MEC-12; other α-tubulin genes may compensate for the loss of MEC-12. In the absence of MEC-7, however, the 15-p MTs in TRNs are replaced by smaller, presumably 11-p, MTs.
The mec-7 result is similar to our previous observations on mec-7(e1506) animals (the e1506 mutation alters the start codon and results in no detectable mec-7 mRNA; Chalfie and Thomson 1982; Savageet al. 1994). In contrast, the mec-12 result was unexpected, because the mec-12[(e1607 (G144S)] allele, which was regarded as the null allele in several previous studies (Bounoutaset al. 2009a; Bounoutaset al. 2011; Hsuet al. 2014), had very few (~2) MTs, all of which had the small diameter (Chalfie and Au 1989; also this study). We found that e1607 and several previously characterized missense alleles were in fact antimorphic (anti) gain-of-function alleles and their phenotypes in neurite growth was also different from the mec-12 loss-of-function (lf) mutants (see below).
Loss-of-function mutations in mec-7/β-tubulin but not mec-12/α-tubulin cause defects in posteriorly directed neurite growth
To test whether MEC-12 and MEC-7 were needed for neurite growth, we examined the outgrowth in animals with mec-12 and mec-7 null alleles. No defects in TRN outgrowth were found in animals mutant for four presumably null mec-12 alleles [the knockout allele (tm5083) and three frameshift mutations (u1026, u1027, and u1028; Figure S1A)] (Figure 2). These results suggest that MTs made without MEC-12 can still respond normally to guidance cues and support neurite morphogenesis. Because C. elegans genome contains nine α-tubulin genes, other tubulin isotypes may compensate for the loss of MEC-12. One such candidate is TBA-7/α-tubulin, which is expressed and functions in TRNs (see below); however, mec-12; tba-7 double mutants were still capable of forming and growing normal TRN neurites, suggesting further genetic redundancy.
Null mutations in mec-7 (ok2152, u156, and u440; Figure S1B) only slightly affected the growth of anteriorly-directed neurites but markedly shortened the PLM-PN (Figure 2B). This result suggests that PLM-AN and PLM-PN, which arise at the same time in the embryo, have different tubulin requirements. Normally the PLM-AN extends anteriorly past the vulva to within 50 μm of ALM cell body. Lockhead et al. (2016) reported that the gap between the end of PLM and ALM cell body was wider in mec-7 lf mutants due to the shortening of the PLM-AN (Lockheadet al. 2016). We also found a similar modest widening of the gap in mec-7(ok2152) animals, but 92% of the PLM-ANs still extended beyond the vulva (Figure 2B and D). Moreover, no shortening of the ALM-AN was seen. A more penetrant and striking defect, however, was the fact that 78% of PLM-ANs in mec-7(ok2152) lacked a synaptic branch, which normally arises posterior to the vulva (Figure 2A and S2). In our analysis, 12 missense mutations (Table 1), including the newly isolated u1020 (G34S) allele, caused a similar mec-7 null phenotype.
The above results and EM data suggest that 1) the reduction of MT numbers in mec-12 lf mutants does not affect neurite development; 2) the smaller, generic 11-p MTs in mec-7 lf mutants are capable of supporting general neurite growth in TRNs; 3) the presence of 15-p MTs is essential for the normal extension of PLM-PN.
In addition to neurite growth, MTs are important for many other functions in TRNs, which were studied before using a few mec-12 and mec-7 missense alleles (Bounoutaset al. 2009a; Hsuet al. 2014). In this study, we systematically confirmed their phenotypes using the knockout mutants and also compared the null alleles with gf alleles for those phenotypes (Figure 3). Deletion of either mec-12 or mec-7 resulted in touch insensitivity, defects in the localization of pre-synaptic vesicles, and severe loss of MT acetylation, a mark for stable MTs. Since MEC-12 is the only α-tubulin isotype that contains the acetylation site (lysine 40) in C. elegans, the TRN-specific 15-p MTs are likely the only MTs that can be acetylated. The loss of mec-7 and mec-12 also induced a global reduction in protein levels through a mechanism dependent on the dual leucine zipper-bearing kinase DLK-1 (Bounoutaset al. 2011). Interestingly, mutations in dlk-1 restored the normal protein expression in mec-7 and mec-12 null mutants but failed to rescue the PLM-PN growth defects in mec-7(ok2152) animals (Figure 3D). Therefore, this PLM-PN defect is most likely a direct effect of the loss of MEC-7 and not the result of secondary changes in protein levels.
Several mec-12 partial lf mutations affected only a subset of MT functions. mec-12(e1605) animals carrying the H192Y missense mutation were defective in mechanosensation but had retained normal 15-p MT structures, tubulin acetylation, axonal transport, protein levels, and neurite growth patterns (Figure S3; Bounoutaset al. 2009a). Similarly, mec-12 mutants carrying the u63 (E415K) or gm379 (G416E) alleles were partially touch-insensitive but kept the large dimeter MTs; however, these animals showed mistargeting of synaptic vesicles presumably because alterations in the EEGE (amino acid 414-417) motif increased affinity with the motor dynein (Hsuet al. 2014). u63 and gm379 mutations also caused a slight decrease in tubulin acetylation and a partial reduction in protein production; gm379 allele led to very mild defects in the growth of PLM-PN (Figure S3). These mutations may be useful in understanding some specific aspects of MT functions.
Antimorphic mutations in mec-7/β-tubulin led to severe, general defects in neurite outgrowth
We identified 19 strong mec-7(anti) gain-of-function missense mutations (Table 1) that caused severe shortening of all TRN neurites (Figure 2B); the ALM-AN did not reach the pharynx, the PLM-AN terminated before reaching the PVM cell body and the PLM-PN was shortened (Figure 2D). We also identified 4 weak mec-7(anti) mutants (Table 1), in which ALM-AN did not extend beyond the nerve ring and PLM-AN did not reach the vulva. All of these alleles, except the recessive u430 (A97V) mutation, were either dominant or semi-dominant. Early studies of one of the dominant mec-7 alleles, e1343 (P171L), found that the mutants contained very few (2.8 ± 0.5 in a cross section) MTs (Chalfie and Thomson 1982), which suggests that the anti mutations may block MT polymerization and thereby cause the severe neurite outgrowth defects. In fact, when compared to the mec-7(ok2152) null allele, these anti mutations caused stronger defects not only in neurite outgrowth, but also in touch sensitivity, synaptic vesicle transport, tubulin acetylation, and protein expression levels (Figure 3). Mutations in dlk-1 could block the reduction in global expression but failed to rescue the neurite growth defects (Figure 3D).
The amino acid residues mutated in the antimorphs are located in three regions on the structure of an αβ tubulin dimer determined by electron crystallography (Nogaleset al. 1998): 1) the GTP/GDP binding pocket, 2) the intradimer interface, and 3) the lateral or longitudinal interdimer interface. All of which are important for tubulin polymerization. For example, u430 (A97V), u911 (P171S), u957 (P171L), and u262 (N226Y) mutations alter amino acids in direct contact with the GTP molecule, and u48 (S176F) and u449 (V179A) change amino acids in the B5 (the fifth β-strand)-to-H5 (the fifth α-helix) loop, which is crucial for forming the GTP/GDP binding pocket (Figure 4C). Two other mutations u445 (M300V) and u98 (M300T) substitute M300, which is located in a loop near the GTP binding site and may participate in positioning H7 (the seventh α-helix) with its critical N226, and thus indirectly affect GTP binding.
The second group of mutated residues in mec-7(anti) alleles is located in the intradimer interface, where β-tubulin makes contact with α-tubulin to form the heterodimer (red residues in Figure 4D). u283 (P243L), u958 (G244S), n434 (N247I), and u162 (D249N) mutations all affect the residues on the H7-to-H8 loop, which interacts extensively with residues on H1 and H2 of α tubulin. Two other dominant mutations gk286000 (A314V) and u955 (A352T) changed residues that located on B8 and B9, respectively, which are physically adjacent to the H7-to-H8 loop at the interface. In addition, two prolines mutations, u910 (P357L) and u956 (P358L) possibly disrupted the B9-to-B10 loop (a.a. 356-361) critical for the positioning of B9.
The third group of mec-7(anti) alleles include e1527 (V286D), u18 (A393T), and gk285997 (A393V), which affect residues involved in the lateral (V286) and longitudinal (A393) interdimer interaction between one α/β-tubulin heterodimer and the neighboring one on the MTs. These mutations may interfere with the interaction between tubulin dimers.
Overall, the above β-tubulin antimorphs likely act as dominant-negative mutants by forming a poisonous α/β dimer, whose incorporation into MTs could terminate MT polymerization and induce instability; the mutated β-tubulin either cannot properly bind to GTP (the first group) or form misshaped α/β heterodimers that block the growing end of MTs (the second group) or disrupt the stacking of tubulin dimers (the third group). Therefore, changes in tubulin structure led to compromised MT elongation, which caused severe defects in neurite outgrowth, highlighting the importance of functional MTs in neuronal morphogenesis.
Antimorphic mutations in mec-12/α-tubulin led to specific defects in posteriorly directed neurite growth
Similar to the mec-7(anti) mutations, we identified 5 mec-12(anti) mutations that also mapped to the GTP binding pocket [u76 (D69N), u950 (S140F), and u1021 (G144S); labeled in red in Figure 4B], the intradimer interface [u1016 (E97K); Figure 4D], or the interdimer interface [u1019 (G354E)]. However, these mec-12 antimorphs are mostly recessive, except for u1019, and did not cause strong, general defects in TRN neurite outgrowth. Instead, these mutations led to the specific shortening of PLM-PN, similar to the defects observed in mec-7 null animals. Electron microscopy studies revealed that the mec-12(anti) mutants have very few small diameter MTs (e.g. ALM neurons in u1021 animals had on average 2.7 MTs in a cross section and 15% of the 72 sections examined had no MTs; the diameter of MTs was 17.9 nm on average; Figure 1B). Therefore, mec-12(anti) mutations have distinct phenotypes from the lf alleles, which retained 15-p MTs and had no defects in neurite formation. The e1607 (G144S) allele, which was previously thought to be a null allele, had the same molecular lesion as antimorph u1021 and caused the shortening of PLM-PN (Figure 2E) and the loss of 15-p MTs (Chalfie and Au 1989); these results further confirm that e1607 should not be regarded as a lf allele.
In addition, the mec-12 antimorphs also resulted in touch insensitivity, reduction in tubulin acetylation, synaptic vesicle mistargeting, and decrease in TRN protein levels; these phenotypes were much stronger than the phenotypes of mec-12 null alleles and were comparable to those produced by mec-7 null mutations but not mec-7 anti mutations (Figure 3). One possible explanation for this phenotype is that MEC-12/α-tubulin preferentially or exclusively binds to MEC-7/β-tubulin, and MEC-12 mutant proteins could only sequester and disable MEC-7 but not other β-tubulin isotypes. As a result, either anti mutations in mec-12 or lf mutations in mec-7 could eliminate the specialized, large diameter, stable MTs that require the MEC-12/MEC-7 heterodimer, but neurite growth could still be supported by regular MTs formed with other α- and β-tubulin isotypes. Consistent with this hypothesis, mec-7(lf); mec-12(anti) double mutants did not show additive effects with regards to the neurite growth defects compared to either single mutant (Figure 2E).
The specific shortening of PLM-PN in mec-7 null mutants and mec-12 antimorphs may be caused by higher sensitivity of the posteriorly neurites to changes in microtubule stability compared to the anteriorly directed neurites. The normal 15-p MTs exist in great numbers and form bundles, which may provide stronger support for neurite growth than the few 11-p MTs found in the mutants. In fact, treatment with the MT-destabilizing drug colchicine (1mM) led to a similar shortening of the PLM-PN, phenocopying these mutations (Figure S4).
Neomorphic mutations in mec-7 and mec-12 resulted in the growth of ectopic neurites
A third category of mec-7/β-tubulin mutations resulted in novel phenotypes, different from the effects of either mec-7 lf or anti mutations; these neomorphic (neo) alleles caused the growth of an extra posteriorly directed neurite in ALM neurons, as well as the overextension of the PLM-PN. Three such alleles [u170 (E407K), u278 (C303Y), and ky852 (P220S)] were previously reported (Savageet al. 1994; Kirszenblatet al. 2013), and three more [gk895768 (T149I), gk286003 (R162Q), and u1017 (L377F)] were identified in this study. L377 and E407 are located on the H11 and H12 helices, respectively, both of which are on the exterior side of MTs and may, therefore, bind motor proteins and other MAPs (Figure 4E). R162 on the H4-to-B5 loop and C303 on the H9-to-B8 loop are also exposed on the MT surface, whereas T149 and P220 are located inside the tubulin structure and may be important for the folding of the protein. Among these six alleles, u278 (C303Y) produced the strongest phenotype; the ectopic posterior neurite of ALM neurons often extended posteriorly to the PLM cell body (Figure 2B). Interestingly, the C303S mutation in allele gk286001 did not lead to similar excessive growth (Table 1), suggesting that tyrosine’s bulky phenolic ring may be responsible for the phenotype.
We also found four mec-12/α-tubulin neo alleles [gk170196 (P32S), gk915672 (E196K), u917 (V260I), gk515972 (V323I)], which generated similar but less severe phenotypes than the mec-7(neo) alleles, since the ectopic ALM-PN was shorter in the mec-12 mutants than in the mec-7 mutants (Figure 2F). This observation supports the notion that MEC-12 may be less critical than MEC-7 in controlling MT dynamics and neurite growth. E196 and V260 are located on loops on the exterior surface, whereas P32 and V323 may be involved in protein folding (Figure 4F).
The production of the ectopic ALM-PN requires both mec-7 and mec-12. The mec-12 null alleles suppressed the ALM posterior outgrowth in mec-7(neo) mutants, and mec-7 lf mutation similarly suppressed the effect of mec-12(neo) alleles (Figure 6A). We also conducted a suppressor screen to search for mutations that can suppress the strong phenotype of ALM-PN growth in mec-7(u278 neo) mutants. After screening 16,000 haploid genomes, we isolated three suppressors: 1) an intragenic mutation in mec-7 [u1040 (G98E)], 2) a lf mutation in mec-12 [u1041 (G246E)], and 3) a lf mutation in mec-15 [u1042 (R27*)]. The identification of mec-7 and mec-12 mutations as suppressors suggests that the incorporation of the MEC-7/MEC-12 heterodimers into the MTs is required for the generation of the excessive neurite; the role of mec-15, which codes for an F-box protein with WD repeats, is described below.
Neomorphic mutations in mec-7 and mec-12 increase MT stability
These neomorphic tubulin mutations appeared to increase the stability of MTs, as first suggested by Kirszenblatet al. (2013) for mec-7. Supporting this hypothesis, pharmacological stabilization of MTs with paclitaxel induced the growth of ALM-PN, although at a low frequency (Kirszenblatet al. 2013), and destabilizing MTs with colchicine can partially suppress this ectopic growth in mec-7(ky852) mutants (Kirszenblatet al. 2013) and in mec-7(u278), mec-7(u1017), and mec-12(u917) animals (Figure 6A). Moreover, these mutants also showed increased tubulin acetylation (Figure 3C), an indication of stable MTs (Song and Brady 2015). mec-7(u278) and mec-7(u170) mutants had large diameter MTs (slightly larger than in wild-type cells), but fewer of them (Figure 1B). Moreover, the MT bundles in the mutants are more closely spaced than the wild type and are not surrounded by the electron-lucent region seen in the wild type (Savageet al. 1994); we quantified these differences by showing that the distance between two closest center points of MTs is smaller in mec-7 neomorph compared to the wild type and that MTs occupied a bigger proportion of the cross-sectional area of the neurite in the mutants (Figure 1B). In fact, the TRN neurite in these mutants is about 50% thinner than the wild type animals. The observation that tightly packed 15-p MT bundles filled up most of the space in the TRN neurites is consistent with increased MT stability, which presumably allowed the excessive neurite growth towards the posterior, overcoming the normal inhibition of this growth in the wild-type ALM neurons.
Another line of evidence indicating higher stability of MTs in mec-7 and mec-12 neomorphs came from their increased resistance to colchicine. This MT-destabilizing drug caused the similar level of reduction in both touch insensitivity and PLM-PN length at a higher concentration in the neo mutants than in the wild type animals (Figure S4). Because the majority of the altered residues in the neo alleles are located on the exterior surface of MTs, these mutations may interfere with the binding of motor proteins or MAPs to the MTs and therefore cause changes in MT stability.
The mec-7 and mec-12 neo alleles also caused some of the same phenotypes that lf mutations do, such as touch insensitivity, synaptic vesicle mistargeting, and reduced TRN protein levels, suggesting that an optimal stability of MTs is needed for their proper functions. Furthermore, the fact that these phenotypes can occur in conjunction with either impaired (in anti mutants) or excessive (in neo mutants) neurite growth support to the hypothesis that the role of MTs in regulating neurite development is genetically separable from other MT functions.
Modeling the effects of tubulin mutations using the TRN neurites
Our genetic analysis shows that missense mutations in tubulin genes can have distinct effects on MT stability and neurite growth pattern, and those different phenotypes appear to correlate with the positions of the altered residues in the tubulin structure. Such structure-function analyses were carried out before for yeast α-tubulin TUB1 and for Drosophila testis-specific β-tubulin bTub85D (Fackenthalet al. 1995; Richardset al. 2000) but not for tubulins in the nervous system. Moreover, recent discovery of over a hundred missense mutations in tubulin genes in patients with a range of neurological disorders prompted us to examine the cellular impact of specific tubulin mutations on neurons. Combining the convenience of genome editing in C. elegans and the ease of observing TRN morphology as the readout of MT stability and neurite growth, we could systematically study the effects of those human tubulin mutations. To provide proof of concept, we generated five clinically observed β-tubulin mutations (S172P, P173L, A302T, P380S, and E410K) in mec-7 through CRISPR/Cas9-mediated gene editing (Materials and Methods; Figure 5A) and found that these mutations indeed caused distinct phenotypes on TRN morphogenesis.
S172P and P173L mutations in human TUBB3 were found in heterozygous patients with microcephaly, polymicrogyria, cortical dysplasia, and agenesis of the corpus callosum, which are defects in neuronal migration and axon growth (Jaglinet al. 2009; Poirieret al. 2010; Bahi-Buissonet al. 2014). In vitro studies found that tubulin heterodimer containing the β-tubulin S172P mutant could not be incorporated into MTs (Jaglinet al. 2009), indicating that the mutated protein is nonfunctional. Consistent with these findings, the u1056 (S172P) mutation of mec-7 led to a recessive loss-of-function Mec-7 phenotype in the TRNs, which showed significantly shortened PLM-PN and slightly shortened PLM-AN with branching defects (Figure 5B, i). Surprisingly, the u1057 (P173L) allele did not cause any TRN morphological or functional defects (Figure 5B, ii). The difference between the mutant phenotypes in humans and worms is puzzling, since both S172 and P173 are located on the GTP binding B5-to-H5 loop (Figure 4C), which is identical in C. elegans MEC-7 and human TUBB3 (Figure S5). Perhaps the differences arise from differences in MT structure and organization: the C. elegans TRNs have 15-p MTs that are bundled, whereas human neurons have 13-p MTs that are not bundled. In fact, when 15-p MTs were converted to 13-p MTs and the bundle was disrupted in mec-17 (α tubulin acetyl-transferase) mutants (Topalidouet al. 2012), both MEC-7 S172P and P173L mutations led to decreased MT stability and the loss of ALM-PN (Figure 5B, iii). This result suggests that P173L mutation indeed compromises MT stability in TRNs under sensitized conditions.
Heterozygous A302T and R380C mutations in human TUBB3, which affect residues on the external surface of the MTs, were associated with moderate congenital fibrosis of the extraocular muscles 3 (CEFOM3), anterior commissure hypoplasia, and corpus callosum hypoplasia. Mutation of A302 and R380 in yeast led to the formation of highly stable, benomyl-resistant MTs (Tischfieldet al. 2010). Consistent with these observations, we found that the same mutations of MEC-7 produced a neo phenotype (the generation of an ectopic ALM-PN; Figure 5B, iv and v).
TUBB3 E410K mutation was also found in patients with severe CEFOM3 and hypoplasia of anterior commissure and corpus callosum, but these patients also suffer from facial weakness and progressive axonal sensorimotor polyneuropathy (Tischfieldet al. 2010). In yeast, MTs containing β-tubulin with the E410K substitution were less stable and less resistant to benomyl and had markedly decreased plus-end accumulation of kinesin-like motor proteins, compared to MTs with A302T and R380C mutations (Tischfieldet al. 2010). In C. elegans, MEC-7(E410K) produced a distinct neurite growth phenotype: ALM-AN and PLM-AN, but not PLM-PN were significantly shortened, which indicates reduced MT stability; but the PLM-PN was not affected, making the phenotype also different from that of the mec-7 antimorphs, which cause general defects in MT polymerization (Figure 5B, vi). Since E410 is also exposed to the exterior of MTs like A302 and R380 (Figure 4E), these results suggest that the cellular impact of the missense mutations would depend on the specific amino acid change instead of the location of the affected residues alone.
The loss of tba-7/α-tubulin also caused excessive posterior neurite growth
C. elegans genome contains nine α-tubulin genes (mec-12, tba-1, tba-2, and tba-4 through tba-9) and six β-tubulin genes (ben-1, mec-7, tbb-1, tbb-2, tbb-4, and tbb-6). Our screen yielded mutants in only three of these genes: mec-7, mec-12, and tba-7. The missense tba-7 allele u1015 (G92D) caused the growth of an ectopic ALM posterior neurite, resembling the phenotype of mec-7 and mec-12 neo alleles. This tba-7 allele is likely a lf allele, because it failed to complement with gk787939 (Q230*), which is presumably a null and produced similar phenotype as u1015 (Figure S1C). We also examined mutants of other tubulin isotypes, and did not find morphological defects in the TRNs (we did not test for redundancy among the genes; Table S3).
tba-7 was expressed in the TRNs, and the excessive growth phenotype of tba-7(u1015) could be rescued by expressing tba-7(+) from the TRN-specific mec-17 promoter, indicating that TBA-7 acts cell-autonomously in the TRNs (Figure S6). tba-7 lf mutants did not exhibit touch insensitivity, synaptic vesicle mistargeting, or reduced protein levels, suggesting that the general function of TRN MTs was not affected and that the primary function of TBA-7 is to regulate neurite growth (Figure S7). We hypothesized that the loss of TBA-7 led to the formation of hyperstable MTs, similar to the mec-7 and mec-12 neo mutations. Supporting this hypothesis, electron microscopy studies revealed that, like mec-7(neo) mutants, tba-7(u1015) animals retained the large diameter 15-p MTs and have closely packed MT bundles, which occupied a larger-than-normal area of a neurite cross-section (Figure 1). tba-7 lf mutants also retained normal tubulin acetylation levels (Figure S7C) and had increased resistance to colchicine (Figure S4), which confirmed the presence of stable MTs. Treatment with 1 mM colchicine fully suppressed in the growth of ALM-PN in tba-7(u1015) animals, and either lf or gf mutations in mec-7 or mec-12 could do so as well (Figure 6A).
The amino acids sequences of α-tubulin TBA-7 and MEC-12 are 82% identical and 94% similar; they mainly differ in a N-terminal region and the C-terminal tail (Figure 6C). Domain swapping experiments showed that changing L31 of TBA-7 to Q found in MEC-12 or replacing the C-terminal DANDNGD (a.a. 435-441 in TBA-7) sequence with MEDNGEEG (a.a. 440-447 in MEC-12) partially impaired the function of TBA-7 proteins to restrict excessive neurite growth; and TBA-7 proteins with both L31Q mutation and the C-terminal replacement completely lost the ability to rescue the tba-7 lf phenotype (Figure 6D). Q31 in MEC-12 is a potential site for polyamination (Songet al. 2013) and E445 of the GEE motif (a.a. 444-446 in MEC-12, but absent in TBA-7) is a target for polyglutamination (Eddeet al. 1990). Post-translational modifications at these two sites could increase the stability of neuronal MTs (Song and Brady 2015). Interestingly, replacing the a.a. 35- 47 region of TBA-7 with the sequence from MEC-12 that contains lysine 40 (the site for α-tubulin acetylation) did not affect the TBA-7 function (Figure 6D), suggesting that the absence of MT acetylation site in TBA-7 was not responsible for its activity in preventing ectopic neurite growth.
Since TBA-7 lacks some modification sites (e.g. Q31 and E445) that could stabilize MTs, our results suggest that TBA-7 may be a MT-destabilizing tubulin isotype, as compared to MEC-12. In wild-type animals, TBA-7 incorporation into the 15-p MTs with MEC-7 and MEC-12 may reduce MT stability; when TBA-7 is not available, more MEC-12 and possibly other α-tubulin isotypes replace TBA-7 and this replacement alters MT dynamics to produces hyperstable MTs.
The tba-7(u1015 lf); mec-7(u1017 neo) double mutant had an ALM-PN that was similar in length to that in mec-7(u1017) single mutants (Figure 6A), but the double mutant additionally produced up to 4 and 5 short ectopic neurites sprouting from the ALM and PLM cell bodies, respectively; this phenotypes was rarely observed in mec-7(u1017) animals (Figure 6E and F). These results suggest that an additive effect occurs in the double mutant. Treating tba-7(u1015) and mec-7(u1017) single mutants with paclitaxel did not produce the double mutant phenotype, probably because the effect of paclitaxel on neurite growth is weak (Kirszenblatet al. 2013).
Mutations in MT-associated Kinesins affect TRN neurite growth
Since motor proteins and MAPs regulate MT dynamics (Akhmanova and Steinmetz 2015), we expected mutations in MAP genes to cause similar phenotype to those of some of the tubulin mutations. Indeed, Ghosh-Royet al. (2012) found and we have confirmed that the loss of the MT-depolymerizing kinesin klp-7 induced the growth of an ectopic posterior neurite in ALM neurons (Figure 6B). KLP-7 belongs to the kinesin-13 family of catastrophe factors that bind to MT plus-ends and promotes depolymerization, thus generating dynamic microtubules (Hanet al. 2015). Importantly, as with the tubulin neomorphs, we found that the growth of the ectopic ALM-PN in klp-7 mutants was suppressed by mec-7 and mec-12 lf mutations but not enhanced by mec-7 neo mutations. These results suggest that the mutated MEC-7 proteins may render MTs hyperstable by reducing their interaction with KLP-7 or by making the MTs insensitive to the action of KLP-7.
Our screen yielded a mutation in a second kinesin gene, klp-11. The putative klp-11 null allele u1024 (Q53*) caused a moderate shortening of ALM-AN and PLM-AN without affecting PLM-PN; in the mutants, ALM-AN failed to extend beyond the metacorpus of the pharynx and PLM-AN did not reach the vulva (Figure S8). A klp-11 deletion allele, tm374, gave the same phenotype. klp-11 codes for a homolog of human KIF3B, a subunit of the plus end-directed kinesin-II motor complex. Although Kinesin-II motor mainly drives intraflagellar transport in the cilia, studies have also shown that Kinesin-II can function in fast, anterograde axonal transport of vesicles in Drosophila (Takedaet al. 2000) and is required for axon growth and regeneration in mouse (Gumyet al. 2013). These data are consistent with KLP-11 being a positive regulator of neurite growth in TRNs. The selectivity of the klp-11 effect on the ANs and not the PNs may relate to the polarity of the MTs they contain. MTs in the anterior neurites of TRNs have uniform polarity with all the plus ends oriented outward from the cell body, whereas the MTs in the PLM-PN have mixed polarity, containing both minus-end-out and plus-end-out MTs (Chalfie and Thomson 1979; Hsuet al. 2014). Thus, the use of minus end-directed motor proteins could support the growth of PLM-PN in the absence of KLP-11. Moreover, the moderate shortening of the ANs also suggests redundancy among the kinesin motor proteins. These results support the hypothesis that MTs promote TRN neurite growth by serving as tracks for the transport of membrane-bound vesicles.
The F-box containing protein MEC-15 is required for tubulin neomorphic phenotype
The putative mec-15 null allele u1008 (Q194*) was isolated from our main screen because the mutation caused the shortening of both PLM-AN and PLM-PN. Another null allele u75 (Q118*) caused the same phenotype (Figure 7A). mec-15 encodes a protein that contains a F-box at the N-terminus and four WD-40 repeats in the middle. Previous work from our lab found that mec-15 is required for touch sensitivity and chemical synapse development in TRNs (Bounoutaset al. 2009b). We found that the ALM-AN and PLM-AN in mec-15 mutants could not fully extend the synaptic branch (Figure 7A), which may have caused the defects in synaptic development.
From the screen for suppressors of mec-7(u278) neomorph, we identified another mec-15 null allele u1024 (R27*), which strongly suppressed the growth of ectopic ALM-PN in mec-7(u278 neo) animals. This suppression is cell-autonomous and depends on the F-box of MEC-15; the expression of the wild-type MEC-15 from the TRN-specific mec-18 promoter in mec-15(lf); mec-7(u278 neo) double mutants restored the ectopic ALM-PN, but the expression of MEC-15 proteins that lacked or had a truncated the F-box domain could not (Figure 7C and D). Since F-box is the substrate recognition subunit of the Skp, Cullin, F-box containing E3 ubiquitin ligase complex (known as the SCF complex), our results suggest that ubiquitination is involved in stabilizing the MTs in mec-7(neo) mutants. One hypothesis is that some unidentified MT-destabilizing molecule is normally ubiquitinated and targeted for degradation in a MEC-15-dependent manner. This unknown protein would be abnormally accumulated in TRNs to reduce MT stability, which caused the shortening of the main neurites and their branches in mec-15(lf) single mutants and the suppression of ALM-PN in mec-15(lf); mec-7(u278 neo) double mutants; the double mutants no longer have the hyperstable MTs. We also observed similar suppression of the ectopic growth of ALM-PN in mec-15(lf); mec-7(u1017 neo), mec-15(lf); tba-7(lf), and mec-15(lf); klp-7(lf) double mutants (Figure 7D), supporting the hypothesis that MEC-15-dependent ubiquitination is needed for the increase of MT stability in the regulation of neurite growth.
Tubulin mutations induce excessive neurite growth independent of MEC-17/α-TAT
Our lab previously reported that the α-tubulin acetyl-transferase (α-TAT) MEC-17, which acetylates MEC-12 in the MTs, is required for the formation of 15-p MTs, as well as MT bundling and organization (Topalidouet al. 2012). Although the loss of MEC-17 does not significantly reduce MT acetylation due to the redundancy with its paralog ATAT-2, mec-17 null mutants showed defects in touch sensitivity, neurite growth and morphology, and vesicle transport; these defects were rescued by expressing a catalytically inactive version of MEC-17, indicating that non-enzymatic activity of MEC-17 is responsible for normal neurite formation (Topalidouet al. 2012; Neumann and Hilliard 2014).
Although the ALM neurons in mec-17(lf) mutants also grew an ectopic posterior neurite, the Mec-17 phenotype is quite different from the phenotypes of the mec-7 or mec-12 neo mutants in the following ways. 1) the ALM-PN in mec-17 animals grows beginning at the L4 stage, whereas it was present as early as the L1 stage in mec-7 and mec-12 neo mutants, and tba-7 and klp-7 lf mutants; 2) the ALM-PN in mec-17 null mutants was longer at 25 °C than at 15 °C, whereas the opposite was seen in tubulin neomorphs, which have a longer ALM-PN at 15 °C than at 25 °C (Figure 8A); 3) excessive swelling and looping of the TRN neurites were observed in mec-17 mutants (Topalidouet al. 2012) but not the other mutants; 4) mec-17 mutants had twice as many actively growing MTs as the wild-type animals, indicating increased dynamics (Neumann and Hilliard 2014), whereas the other mutants presumably caused the formation of hyperstable MTs. Interestingly, treatment with paclitaxel, which decreases MT dynamics, partially suppressed the growth of ALM-PN in mec-17(lf) mutants but not in mec-7 and mec-12 neo mutants or tba-7 or klp-7 lf animals (Figure 8B). Moreover, when we combined the mec-17(ok2109 lf) allele with the mec-7(u1017 neo) allele or tba-7(u1015 lf) allele, additive effects were observed for the length of both ALM-PN and PLM-PN (Figure 8C). Together, our data indicate that the loss of MEC-17 and the mutations in tubulin isotypes may induce excessive, posteriorly directed neurite growth through different mechanisms.
One such difference in mechanism is the differential dependency on the guanine nucleotide exchange factor TIAM-1, which promotes the growth of a posteriorly directed neurite in the bipolar PLM neurons (Zhenget al. 2016). tiam-1 null mutations suppressed the growth of ALM-PN in mec-17(lf) mutants but not in mec-7 and mec-12 neo mutants and tba-7(lf) mutants (Figure 8D). Because TIAM-1 promotes neurite extension through the Rac1 GTPase, which presumably regulates the dynamics of actin cytoskeleton (Zhenget al. 2016), the loss of MEC-17 may activate TIAM-1 and induce actin remodeling, whereas the MT-stabilizing tubulin mutations do not evoke TIAM-1 to promote posterior outgrowth.
Discussion
Neuronal morphogenesis requires an optimal MT stability
MTs both support and regulate neurite growth. As the building block of MTs, α- and β-tubulins determine the structural properties of MTs and also mediate their interaction with motor proteins and MAPs. As a result, any alteration in the tubulin proteins can potentially lead to changes in MT dynamics, which then affects neurite formation, guidance, and extension. In this study, we analyzed the effects of 67 tubulin missense mutations on neurite development in C. elegans touch receptor neurons. Based on the phenotypes, we categorized these mutations into three classes: loss-of-function (lf), antimorphic (anti), and neomorphic (neo) mutations. 1) lf mutations cause only mild neurite growth defects by reducing MT stability moderately; 2) anti mutations block MT polymerization and cause severe neurite growth defects; 3) neo mutations cause excessive neurite growth by inducing hyperstable MTs (Figure 9). The directionality of the excessive neurite in ALM reflects its potential to grow towards the posterior. In fact, about 30% of the adult ALM neurons have a very short (less than one cell body length) posterior protrusion, and manipulation of cellular signaling (e.g. activating the small GTPase Rac or altering the recycling endosome pathway) can induce the production of ALM-PN (Zhenget al. 2015b; Zhenget al. 2016). Thus, the ectopic growth of ALM-PN may serve as an indication for altered cytoskeletal organization.
The correlation between the amount of neurite growth and the stability of MTs suggests that MT structure and dynamics are major determinants of neurite growth. Given the great number of proteins involved in both positively and negatively regulating MT assembly and disassembly (Mimori-Kiyosue 2011), our results support that an optimal level of MT stability is required for proper neuronal morphogenesis. This hypothesis is further supported by the finding that negative regulators of MT stability, such as the α-tubulin isotype TBA-7 and the MT-destabilizing kinesin-13 KLP-7, prevent excessive TRN neurite growth.
We suspect that other MT destabilizers are also important for TRN neurite development and that these destabilizing proteins are, in turn, regulated by the F-box containing protein MEC-15, perhaps by its involvement in ubiquitination and protein degradation. Loss of mec-15 results in TRNs with shorter ANs and PNs. Since mec-15(lf) is epistatic with regard to neurite growth to the mec-7(neo), mec-12(neo), tba-7(lf), and klp-7(lf) mutations, MEC-15 is likely to reduce the amount of, as yet, unknown MT destabilizers.
Structure-function analysis of tubulin mutations
The large number of missense mutations analyzed in this study allowed us to perform structure-function analysis of neuronal tubulin proteins in living animals. Mapping the altered amino acid residues onto the Bos taurus tubulin α/β heterodimer structure 1JFF (Nogaleset al. 1998), we found that lf mutations mostly affect residues located in the interior of the structure and possibly disrupt protein folding. anti mutations affect residues that participate in GTP binding, intradimer interaction, or interdimer interactions. neo mutations mostly alter residues on the exterior surface and probably perturb the association of motor proteins and MAPs with MTs (Figure 4 and 9).
Although similar structure-function relationship studies were conducted before in yeast α-tubulin TUB1 and β-tubulin TUB2 and Drosophila testis-specific β2-tubulin (Reijoet al. 1994; Fackenthalet al. 1995; Richardset al. 2000), our work provides another model to study the properties of tubulin proteins. In fact, our findings are consistent with those previous studies. For example, yeast TUB1 mutations that caused supersensitivity to the MT-destabilizing drug benomyl mainly altered amino acids involved in GTP binding (e.g. D70) and intradimer (e.g. E98) and interdimer interaction; the change of equivalent residues in MEC-12 (D69 and E97) were identified in mec-12 anti mutants. Similarly, some Drosophila β2-tubulin mutations (S25L and G96E) that made MTs less stable (Fackenthalet al. 1995) affected the same or the adjacent residue in mec-7 antimorphs [u319 (S25F) and u430 (A97V)]. On the other hand, yeast TUB1 mutations that conferred benomyl-resistance mostly affect the residues exposed to the outer surface of the MTs (Richardset al. 2000), and mec-7 neo mutations similarly led to increased resistance to colchicine. Therefore, comparative studies using the three different organisms can help identify key residues in the tubulin structure responsible for various MT functions.
Modeling the effects of neuronal tubulin mutations
We have used the C. elegans TRNs as a model to study neuronal MTs, which are much more stable and have different types of post-translational modification than the MTs in dividing cells (Baaset al. 2016). Clinical studies in the past five years identified more than 100 missense mutations in tubulin genes that cause a wide spectrum of neurodevelopmental disorders, such as microcephaly, lissencephaly, polymicrogyria, and cortical malformation, along with severe defects in axon guidance and growth, including the complete or partial absence of corpus callosum, defects in commissural fiber tracts, and degeneration of motor and sensory axons (Tischfieldet al. 2011; Chakrabortiet al. 2016). Patients carrying different tubulin mutations often display distinct symptoms, but the complexity of the human nervous system makes it difficult to analyze the impact of specific mutations at the cellular level.
Combining CRISPR/Cas9-mediated genome editing with in vivo examination of TRN morphology, we were able to create several disease-causing human β-tubulin mutations in the C. elegans mec-7 gene and examined their effects on neurite growth at a single neuron resolution. Our phenotypic evaluation was largely consistent with previous characterization of those mutations, suggesting that the TRN system could be instrumental to model the effects of human tubulin mutations identified in the clinic.
In fact, many disease-causing tubulin mutations alter amino acids that are physically adjacent to or located in the same region as the ones affected in our mutants, suggesting that they may affect MT stability in similar ways. By mapping 51 TUBA1A, 24 TUBB2B, and 19 TUBB3B mutations that were clinically identified onto the tubulin structure, we could correlate the location of some of the affected residues with the resulting clinical manifestation (Table S4). For example, 7/8 (87.5%) mutations that alter residues involved in GTP binding caused complete agenesis of corpus callosum, an indication of severe defects in axonal growth, whereas only 5/23 (21.7%) changes of residues in the interior of the structure and 10/32 (31.2%) mutations of residues involved in MAP binding did so (Table S4).
However, position of the mutated residues alone could not determine the effects of the missense mutation; each specific mutation should be evaluated individually. For example, the B5-to-H5 loop of β-tubulin (from V170 to V179) is in contact with the GTP molecule and four strong mec-7 anti mutations (P171S, P171L, S176F, and V179A) were mapped to this loop. Unexpectedly, S172P mutation led to a lf phenotype and P173L mutation caused no defects in TRNs. One possible explanation is that S172 and P173 are further away from the GTP binding site and thus are less important than P171 and S176.
The identity of the residue replacing the wild type one also plays a critical role. For instance, u278 (C303Y) was the strongest mec-7 neomorph, but gk286001 (C303S) had no effect at all; since both tyrosine and serine have the polar hydroxyl group, the bulky phenyl group of the tyrosine may be responsible for disrupting some MT functions. Moreover, the u1058 (A302T) mutation also caused a strong mec-7(neo) phenotype. A302 and C303 on the H9-to-B8 loop are exposed to the exterior of MTs but are not located on the major landing surface (H11 and H12) for motor protein and MAPs. Our results suggest that A302 and C303 may represent a previously unrecognized site of interaction between the MTs and their associated proteins. Even for residues located on H11 and H12, the effects of their mutation can be different. u1059 (R380S) is a mec-7 neomorph, whereas u1060 (E410K) allele produced an antimorph-like phenotype, suggesting that the impact of the missense mutation depends on how the interaction of MTs with the motor proteins or MAP is affected by the amino acid change.
MT-destabilizing tubulin isotype and “multi-tubulin hypothesis”
The presence of multiple tubulin genes in eukaryotic genomes, their subtle differences in the protein sequences, and their functional differences led to the “multi-tubulin hypothesis.” In multicellular organism, the tubulins may have distinct expression patterns, e.g. Drosophila β2-tubulin is only expressed in the male germline (Kemphueset al. 1980); human TUBB3B is highly enriched in the nervous system, whereas TUBB2B is expressed in many tissues (Sullivan and Cleveland 1986). In single-cell organisms like Tetrahymena, different β-tubulin isotypes are spatially separated; some are only used to make the nuclear MTs of the mitotic apparatus, and some are only detected in the MTs of somatic cilia and basal bodies (Pucciarelliet al. 2012). Our studies, however, found that two distinct α-tubulin isotypes (MEC-12 and TBA-7) could have different functions when both incorporated into the same MTs in the same neurons; MEC-12 promotes stability and TBA-7 promotes instability.
The abundance of large diameter 15-p MTs in the TRNs requires MEC-12, but the role of TBA-7 is to increase MT dynamics and to prevent the ectopic growth of posteriorly directed neurites in ALMs. Thus, TBA-7 is a MT-destabilizing tubulin isotype, and this destabilizing activity depends on the absence of potential polyamination and polyglutamination sites that could increase MT stability. Our results are consistent with early in vitro experiments suggesting that TUBB3 (also known as βIII) also promotes MT dynamics; removal of TUBB3 from the brain extract resulted in a tubulin mixture that assembled much more rapidly than the unfractionated control (Banerjeeet al. 1990), and MTs assembled from the purified αβIII heterodimers were considerably more dynamic than MTs made from the αβII and αβIV dimers (Pandaet al. 1994). However, the long-standing question is whether some tubulin isotypes do destabilize MTs in vivo. Our findings suggest the answer is yes. A balanced incorporation of multiple tubulin isotypes into the same MT structure is critical to generate MTs with the optimal stability.
The absence of post-translational modification sites in tubulin isotypes may lead to functional diversity. In addition to TBA-7, among the other eight C. elegans α-tubulin isotypes, TBA-8 and TBA-5 also lack the Q31 for polyamination, and TBA-6 lacks the E445 for polyglutamination; all nine human α-tubulins have Q31, but TUBAL3 has a short C-terminal tail and does not contain potential polyglutamination sites (Figure S9). The incorporation of those tubulin isotypes may be a general regulatory mechanism to control MT stability.
Moreover, mec-12 and tba-7 do not appear to be redundant, because the tba-7 lf mutation alone produced uncontrolled growth. Moreover, both mec-12 lf mutants and mec-12; tba-7 double lf mutants retained the ability to grow out a normal amount of neurites, suggesting strong genetic redundancy among the α-tubulins with regards to supporting neurite growth.
Different contributions of tubulin isotypes to MT dynamics were also observed in C. elegans embryogenesis (Hondaet al. 2017). The spindle MTs required two β-tubulins (TBB-1 and TBB-2), but TBB-2 was incorporated into the MTs twice as much as TBB-1; the loss of TBB-2 caused a dramatic decrease in MT growth rate and an increase in catastrophe frequency, leading to highly unstable MTs, whereas the loss of TBB-1 only slightly reduced growth rate. These data support the “multi-tubulin concept” that different tubulin isotypes have distinct functions in the same cells.
Materials and Methods
Strains and genetic screens
C. elegans wild type (N2) and mutant strains were maintained at 20 °C as previously described (Brenner 1974). Alleles listed in Table S1 and S2 were isolated by visually screening for mutants with TRN differentiation defects using TU4069, which carries uIs115 [mec17p::RFP] for the visualization of the TRNs, as the starter strain and ethyl methanesulfonate as the mutagen. Mutants were outcrossed with wild type, and the phenotype-causing mutations were identified as previously described by whole-genome resequencing (Zhenget al. 2013) or by complementation tests with reference alleles. mec-7 alleles e1343, e1505, e1527, e1522, n434, u10, u18, u22, u48, u58, u98, u127, u129, u162, u170, u223, u225, u278, u234, u249, u262, u275, u283, u305, u319, u428, u429, u430, u433, u449, and u445, and mec-12 alleles e1605, e1607, u50, u63, u76, and u241 were previously isolated (Chalfie and Sulston 1981; Chalfie and Au 1989; Savageet al. 1994) and re-examined in this study.
u1056 to u1060 alleles were created through CRISPR/Cas9-mediated homologous recombination (Dickinsonet al. 2013). Guide RNAs were designed according to the target sequence closest to the desired codon change (Figure 5A). Recombination templates were created by cloning the mec-7 coding region and 466 bp 3’UTR sequence into the SalI and BamHI sites of pBlueScript II SK (+) vector and then introducing the desired missense mutation (red in Figure 5A), along with synonymous mutations (blue in Figure 5A) that change the guide RNA target sites, using the Q5 Site-Directed Mutagenesis Kit from New England Biolabs (NEB; Ipswich, MA). pDD162 constructs expressing Cas9 and specific guide RNAs were injected together with the corresponding recombination template and marker myo-2p::RFP into TU4069 animals. F1s expressing RFP in the muscle were singled out and genotyped to identify heterozygotes with successful edits, and F2s carrying homozygous mutations were then isolated and examined for TRN morphology.
The gk alleles of mec-7 and mec-12 listed in Table 1 were generated in the million mutation project (Thompsonet al. 2013) and obtained from the Caenorhabditis Genetics Center, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). klp-7(tm2143), klp-11(tm374), and mec-12(tm5083) were generated by the National Bioresource Project of Japan, and mec-12(gm379) was kindly provided by Dr. Chun-Liang Pan at the National Taiwan University.
Additional mec-12 null alleles (u1026, u1027, and u1028) were generated by CRISPR/Cas9-mediated genome editing targeting 5’-GAAGTAATTTCGATTCACATCGG-3’ in exon 2 of mec-12 as described above (Dickinsonet al. 2013). Frameshift-causing mutations were identified by sequencing the mec-12 locus (Figure S1).
A second screen was conducted using TU4879 [mec-7 (u278); uIs115 (mec-17p::RFP)] as the starter strain, and mutants with significantly shortened ALM-PN were isolated. mec-7 [u1040 (G98E)], mec-12 [u1041 (G246E)], and mec-15 [u1042 (R26*)] were obtained from this screen as suppressors of mec-7[u278 (C303Y) neo] mutants.
Constructs and Transgenes
A tba-7::GFP reporter (TU#1632) was made by cloning a 1.2 kb tba-7 promoter and the entire coding region from the genomic DNA to wild-type genomic DNA into the Gateway pDONR221 P4-P1r vector; the resulting entry vector, together with the pENTR-GFP, pENTR-unc-54-3’UTR, and destination vector pDEST-R4-R3 were used in the LR reaction to create the final expression vectors. mec-17p:: tba-7(+) (TU#1629) was created by cloning a 1.9 kb mec-17 promoter and the tba-7 coding sequence into pDONR221 P4-P1r and pDONR221 vectors, respectively, and assembling these entry vectors with pENTR-unc-54-3’UTR and the destination vector. Details about the Gateway cloning method by Life Technologies can be found at www.invitrogen.com/site/us/en/home/Products-andServices/Applications/Cloning/Gateway-Cloning.html. tba-7(+) locus, including the 1.2 kb promoter, coding region, and a 911-bp 3’UTR, was cloned using SalI and NotI into pBlueScript II SK(+) vector; NEB Q5 site-directed mutagenesis kit was then used to generate constructs expressing the TBA-7 variants shown in Figure 6D.
To rescue the mec-15 mutants, we amplified the mec-15(+) locus that includes a 3.2 kb promoter, coding region, and a 951 bp downstream region and injected the purified PCR product to perform the rescue experiments. We used previously described TU#891 [mec-18p::mec-15(+)] construct (Bounoutaset al. 2009b) to perform cell-specific rescue and to generate constructs that express truncated MEC-15 proteins shown in Figure 7C.
Transgenes uIs31[mec-17p::GFP] III, uIs115[mec-17p::RFP] IV, and uIs134[mec-17p::RFP] V were used to visualize TRN morphology (Zhenget al. 2015a). jsIs821[mec7p::GFP::RAB-3] was used to assess synaptic vesicle localization (Bounoutaset al. 2009b).
Electron microscopy
We re-analyzed previously collected cross-section images of ALM neurites from wild type animals (N501, N631, N933, and N934 from the Hall lab collection) and from mec-7 neomorphic mutants u170 and u278 (Savageet al. 1994). The wild type animals had been fixed by either chemical immersion fixation without tannic acid (samples N501, N631; Hall 1995) or by high-pressure freezing/freeze substitution (HPF/FS) including tannic acid (samples N933, N934; Topalidouet al. 2012). The mec-7 neomorphic alleles had also been fixed by chemical immersion, without tannic acid.
For this study, we fixed mec-7 lf (ok2152), mec-12 lf (tm5083), and tba-7 lf (u1015) adults using an HPF/FS protocol that included a first fixation in 0.5% glutaraldehyde + 0.1% tannic acid and a second fix in 2% osmium tetroxide + 0.1% uranyl acetate (UAc), both at −90 °C, followed by staining in 1% UAc in acetone at 0 °C. We also fixed adults of mec-12(anti) mutants, u950, u76, and u1021, using HPF/FS using a similar protocol, but without tannic acid. Eighty-nanometer transverse sections were collected at multiple positions along the ALM anterior neurite and post-stained with uranium acetate and lead citrate. A Philips CM10 electron microscope with an attached Morada digital camera (Olympus) was used to acquire the images.
For each animal, we counted the number of MTs in the ALMs and measured the MT diameter on sections collected from at least five different positions along the ALM-AN; data from ALML and ALMR were combined. For each strain, we examined sections from at three different animals. For wild-type animals, tba-7 lf mutants, and mec-7 neomorphs, we used at least ten sections to measure the distance between two closest MT centers and the proportion of MT-occupied area to the cross-sectional area of the neurite.
Phenotype Scoring and Statistical Analysis
We measured the length of TRN neurites in at least 30 fourth stage larvae or young adults grown at 20° C, except where otherwise stated. Relative length of posteriorly directed neurites was calculated by dividing the neurite length by the diameter of the cell body. Defects in TRN anterior neurite length were assessed by counting the percentage of cells whose neurites failed to reach the vulva (for PLM) or the posterior pharyngeal bulb (for ALM); at least 50 cells were examined for each strain.
Fluorescence intensity of the RFP expressed from the uIs134 transgene in the ALM cell body was used to measure TRN protein levels; intensity was calculated after manual background subtraction using ImageJ as previously described (Chen and Chalfie 2015). Synaptic vesicle localization was quantified by calculating the percentage of animals in each of the four phenotypical categories: 1) normal localization to the PLM synaptic branch; 2) reduced intensity of the synaptic marker at the synapses as partial defect; 3) complete loss of normal localization to the synapse as complete defect; 4) localization to the PLM-PN as mistargeting. At least 50 adult animals were examined.
Acetylated α-tubulin staining using antibody [6-11B-1] (abcam, Cambridge, MA) was performed and analyzed as previously described (Topalidouet al. 2012). For colchicine treatment, animals were grown in standard NGM agar plates containing different concentrations (from 0.06 mM to 2 mM) of colchicine before phenotypic analysis as described before (Bounoutaset al. 2009a). Similarly, 100 nM paclitaxel was added to the NGM agar plates, L1 animals were placed onto the plates, and young adults were examined for TRN morphology. PyMOL (Schrodinger 2015) was used to view the structure of α/β tubulin dimer (1jff.pdb; Nogaleset al. 1998) and to label affected residues in mec-7 and mec-12 mutants,.
For statistical analysis, ANOVA and the post hoc Dunnett’s test or Tukey-Kramer test were used to identify significant difference between the mutants and wild type animals in multiple comparisons. Student’s t test was used to find significant difference between two samples in paired comparisons. Single and double asterisks indicated p < 0.05 and p < 0.01, respectively. The χ2 test was used for the categorical data to find significant difference between different strains.
Acknowledgements
We thank Dan Dickinson, Chun-Liang Pan, and Andrew Chisholm for sharing reagents. This work was supported by NIH Grant GM30997 (to M.C.) and OD 010943 (to D.H.H.). Core facilities for electron microscopy were supported by NICHD P30 HD71593 for the RFK-IDDRC at Albert Einstein College of Medicine.