Genome-wide Analysis Reveals Novel and Discrete Functions for Tubulin Carboxy-terminal Tails

Summary

Background

Microtubules (MTs) support diverse transport and force generation processes in cells. Both α- and β-tubulin proteins possess carboxy-terminal tail regions (CTTs) that are negatively charged, intrinsically disordered, and project from the MT surface where they interact with motors and other proteins. Although CTTs are presumed to play important roles in MT networks, these roles have not been determined in vivo.

Results

We examined the function of CTTs in vivo using a systematic collection of mutants in budding yeast. We find that CTTs are not essential; however, loss of either α- or β-CTT sensitizes cells to MT destabilizing drugs. β-CTT, but not α-CTT, regulates MT dynamics by increasing frequencies of catastrophe and rescue events. In addition, β-CTT is critical for the assembly of the mitotic spindle and its elongation during anaphase. We use genome-wide genetic interaction screens to identify roles for α- and β-CTTs, including a specific role for β-CTT in supporting kinesin-5/Cin8. Our genetic screens also identified novel interactions with pathways not related to canonical MT functions.

Conclusions

We conclude that α- and β-CTTs play important and largely discrete roles in MT networks. β-CTT promotes MT dynamics. β-CTT also regulates force generation in the mitotic spindle by supporting kinesin-5/Cin8 and dampening dynein. Our genetic screens identify links between α- and β-CTT and additional cellular pathways, and suggest novel functions.

Introduction

Microtubules (MTs) are indispensable components of eukaryotic cells, forming networks that organize the cytoplasm in a variety of contexts. How MT networks are adapted for different contexts is an important question. Evolutionarily distinct MT motors and binding proteins contribute to functional diversity by promoting different activities within the network. Whether MTs themselves contribute to functional diversity is poorly understood.

CTTs of α- and β-tubulins are likely to regulate the complexity of MT functions. CTTs were first distinguished by comparisons of α- and β-tubulin primary sequences. Whereas the majority of α- and β-tubulin sequences are conserved, the 10–20 amino acids at the carboxy-termini are variable and enriched for negatively-charged amino acids, primarily glutamates (Figure S1). Higher eukaryotes possess multiple isotypes of α- and β-tubulin with unique CTT sequences. These unique CTT sequences define isotype classes that are conserved across species. In mammalian cells, the relative abundance of each isotype varies according to cell type [1, 2]. Studies in Drosophila have identified one isotype CTT that has a cell type specific role in the formation of the central pair of axonemal MTs during spermatogenesis [3]. In vitro assays with individual purified isotypes reveal distinct effects on MT dynamics [4]. These findings support a model where isotype CTTs impart functional differences that tune MT networks for different cellular contexts.

Biochemical studies suggest roles for CTTs in MT assembly and interactions with MT binding proteins. CTTs extend from helix 12 on the outer surface of the microtubule, where they are highly dynamic and can be removed by proteolytic digestion with subtilisin (Figure 1A) [5, 6]. Soluble tubulin treated with subtilisin assembles into MTs at lower concentrations than untreated tubulin, and alters protofilament organization in the MT, suggesting that CTTs influence the formation of the MT lattice [7, 8]. CTT removal also alters interactions with binding proteins and motors in vitro. CLIP170[9], EB1 [10], XMAP215 [11], Tau [12], MAP2[7], Ndc80/Hec1 [13], Aurora-B [14], and the Dam1 complex [15] are depleted from subtilisin-treated MTs. Kinesin and dynein exhibit changes in specific interacting states with subtilisin-treated MTs, resulting in diminished processivity or lattice diffusion [16–20]. These interactions between MTs and binding partners are also highly sensitive to salt, consistent with a prominent electrostatic contribution.

Figure 1. α- and β-CTTs contribute to microtubule function.

A) Structure and predicted range of motion for mammalian α- and β-CTTs. The CTT regions of yeast α-tubulin/Tub1 and yeast β-tubulin/Tub2 were modeled on the structure of mammalian α/β tubulin heterodimer [5]. The surface and lumen sides of the microtubule are indicated, as well as the plus and minus ends. Shaded volumes represent the 95% confidence interval of the sampled range of motion for α(blue) and β-CTTs (red), based on molecular dynamics simulations. Volumes of these regions are shown in ų. Tub1 residue A442 and Tub2 residue V430 are labeled in yellow. B) Sequences of CTTs. The carboxy-terminal residues of the α-tubulins, Tub1 and Tub3, and the β-tubulin, Tub2. These regions were selected based on abundance of negatively-charged residues, high interspecies sequence heterogeneity (Figure S1), lack of resolution in structural studies [5], and correspondence to the major fragment produced by subtilisin digestion [6]. Residues highlighted in yellow correspond to residues highlighted in (A). C) CTT mutants are sensitive to benomyl. 10-fold dilution series of indicated strains were spotted to rich media (YPD; “control”) or rich media supplemented with benomyl (10μg/mL). Strains: wild type, yJM0596; tub1-QQQQF, yJM0418; tub1-EEF, yJM0118; tub1-446Δ, yJM0105; tub1-442Δ, yJM0116; tub1-442Δ tub3-442Δ, yJM0212; tub2-445Δ, yJM0583; tub2-438Δ, yJM0565; tub2-430Δ, yJM0282; tub1-442Δ tub2-430Δ, yJM0559; tub1-442Δ tub3-442Δ tub2-430Δ, yJM0581; CTT swap, yJM0551.

CTTs are also major sites of posttranslational modification in higher eukaryotes – detyrosination and tyrosination, polyglutamylation, and polyglycylation -- which provide additional MT regulation [21]. α-CTTs are tyrosinated by tubulin tyrosine ligase (TTL), which forms a complex with the αβ heterodimer that guides the α-CTT into the active site of the enzyme [22, 23]. This requires two consecutive acidic residues at the CTT terminus; hence, detyrosinated α-CTTs can be tyrosinated, but β-CTTs cannot [23]. Tyrosination of α-CTT regulates MT function by promoting interactions with CLIP170, p150^Glued, and the depolymerase MCAK [24, 25]. In contrast to tyrosination, polyglutamylation is a heterogeneous modification where glutamate chains of variable lengths are added to genetically encoded glutamate residues in either α- or β-CTT. Selectivity for α- versus β-CTT is determined by different glutamylase enzymes of the TTL-like (TTLL) family [26]. Although the sites of polyglutamylation have been identified, how TTLL enzymes discriminate between CTT sequences is yet unclear. Polyglutamylation is enriched in neurons, axonemes, and microtubule organizing centers; and regulates MT function by promoting the activity of the microtubule-severing enzyme spastin [27], dynein motility in axonemes [28, 29], and perhaps additional undiscovered mechanisms. These findings demonstrate that CTTs in higher eukaryotes are hubs for posttranslational modifications that regulate MT function.

Nevertheless, the roles of CTTs in cells are largely uncharacterized. Previous studies in Tetrahymena found that mutations deleting CTT sequences from α- or β-tubulin are lethal; however, lethality could be rescued by chimeras that replace α-CTT with β-CTT and vice versa [30]. These results suggest essential but undefined roles for CTTs in Tetrahymena. The budding yeast model affords important advantages for defining the roles of CTTs. In contrast to higher eukaryotes, yeast possess a simple repertoire of tubulin genes – two α-tubulin isotypes (Tub1 and Tub3) and a single β-tubulin (Tub2). Tub1 accounts for ~90% of α-tubulin in the cell and Tub3 the remaining 10% [31]. Mutations at the endogenous loci affect all cellular tubulin while maintaining native expression levels. Previous structure/function studies of yeast tubulins identified regions critical for polymer stability, drug tolerance, and protein-binding. Thus far, mutations in α- and β-CTT regions have been shown to cause sensitivity to MT depolymerizing drugs [32, 33].

Here, we generated a series of point mutations that alter or ablate α- and β-CTTs. We defined CTTs as regions that exhibited: 1) abundance of negatively-charged residues, 2) high interspecies sequence heterogeneity, 3) lack of resolution in structural studies, and 4) correspondence to the major fragment produced by subtilisin digestion (Figure S1) [6]. To investigate the roles of each CTT, we used genome-wide screens to map genetic interactions and live-cell microscopy to measure effects on MT dynamics and spindle dynamics. We find that although α- and β-CTTs are not essential for viability, each promotes specific functions within the network, including a roles for β-CTT in regulating MT dynamics and promoting kinesin-5. In addition, we identify novel roles for tubulins that may be regulated by CTTs. Together our results support a model in which CTTs modulate the activities of different pathways in a microtubule network, which may be important for tuning the network for different cellular contexts.

Results

Generating tubulin CTT mutants

CTTs are thought to be disordered regions that extend out from the microtubule surface. We used molecular dynamics to determine the confirmations and dynamics of yeast α- and β-CTTs relative to the known structure of the αβ heterodimer [5]. These simulations demonstrate that CTTs are highly dynamic. Whereas yeast α-CTT sampled a volume of 5500 ų near the longitudinal interdimer interface, the longer yeast β-CTT sampled a volume of 25,000 ų near the intradimer interface (Figure 1A). When compared to simulations of mammalian CTTs, we found that β-CTTs were highly mobile in both species while mammalian α-CTT is longer and samples a larger volume than yeast α-CTT (Figure S1A and B). This suggests that CTTs could influence interactions within the heterodimer, between heterodimers in the lattice, or with binding partners at the microtubule surface.

To investigate functions of CTTs, we first generated a series of mutations to alter or remove residues at the carboxy-termini of the major yeast α-tubulin, Tub1. We altered the charge of the Tub1 CTT by replacing glutamates with glutamines at residues 443–446 and created truncation alleles that lacked the terminal phenylalanine (446Δ), two of the glutamates (EEF), or all four glutamates and the phenylalanine (442Δ) (Figure 1B). The tub1-442Δ mutation removes the region of negative charge and reduces the predicted range of motion for the α-tubulin carboxy-terminus (Figure 1A). None of these mutations affected growth on solid media under optimal conditions (YPD, 30°C; Figure 1C). However, when we stressed the MT network with a low dose of the MT-destabilizing drug benomyl, mutations that altered α-CTT charge (tub1-QQQQF) or removed the charged region (tub1-442Δ) exhibited poor growth compared to wild-type controls (Figure 1C). Truncation of the CTT from the minor α-tubulin, Tub3, alone had no effect (tub3-442Δ; data not shown). Simultaneous truncation of CTTs from both yeast α-tubulins (tub1-442Δ, tub3-442Δ) exacerbated the benomyl sensitivity phenotype (Figure 1C). Benomyl treatment therefore reveals a charge-dependent function of α-tubulin CTTs.

Next, we turned to the CTT of the yeast β-tubulin, Tub2. We created truncation alleles that lacked the terminal twelve residues (445Δ; identical to the allele tested in [32]), nineteen residues (438Δ), or twenty-seven residues (430Δ; Figure 1B). The tub2-430Δ mutation removes most negatively charged residues, reduces the predicted range of motion for β-CTT, and mimics the primary truncation produced by subtilisin (Figure 1A, S1D)[6]. Similar to α-tubulin mutants, truncation of the β-tubulin CTT did not impair growth under optimal conditions, but it increased sensitivity to benomyl (Figure 1C). Furthermore, simultaneous truncation of CTTs from all α- and β-tubulins (tub1-442Δ tub3-442Δ tub2-430Δ) had no obvious effect under optimal conditions, but the mutants had increased sensitivity to benomyl, compared to mutants lacking α- or β-CTTs alone (Figure 1C).

We then asked whether α- and β-CTTs are functionally redundant, in which case one CTT could rescue the function of the other. To test this, we created chimeric ‘CTT-swap’ alleles where the β-CTT region replaced the endogenous α-CTT and vice versa (Figure S2A). Each CTT-swap allele exhibited benomyl sensitivity similar to truncation mutants, indicating failure to rescue (Figure 1C; S2B). In this assay, α- and β-CTTs are neither functionally redundant nor autonomous. We conclude that α- and β-tubulin CTTs are important but not essential for MT function.

CTT mutations affect microtubule dynamics

The benomyl sensitivity phenotype of CTT mutants is reminiscent of tubulin mutants that affect MT stability and dynamics. We therefore asked whether CTT mutations disrupt MT stability and dynamics. The MT lattice is destabilized at low temperatures in vitro, and several yeast tubulin mutants that exhibit benomyl sensitivity also exhibit poor growth at low temperatures [33, 34]. At low temperature (16°C), α-CTT mutant growth was similar to that of wild-type cells (Figure 2A). In contrast, β-CTT mutants showed improved growth compared to wild type (Figure 2A). To our knowledge, this is the first description of a tubulin mutation improving fitness at low temperature. This result suggests that MTs may be stabilized by the loss of the β-CTT.

Figure 2. β-CTT regulates microtubule stability.

A) β-CTT mutants are cold tolerant. Cells were grown to saturation at 30°C in rich media and a 10-fold dilution series of each was spotted to YPD; plates were grown at 16°C for 5 days. Strains are the same used in Figure 1C. B) α- and β-CTT mutants exhibit longer astral microtubules. Distributions of astral microtubule lengths, with bars indicating mean ±SD. Asterisks indicate statistical significance (p < 0.01) by t-test, compared to wild type. Strains: wild type, yJM0596; tub1-442Δ tub3-442Δ, yJM0212; tub2-430Δ, yJM0282; tub1-442Δ tub3-442Δ tub2-430Δ, yJM0581; CTT swap, yJM0551.

To evaluate stability and dynamics, we investigated the lengths of MTs. In contrast to spindle MTs, individual astral MTs (aMTs) are clearly distinguished by light microscopy, allowing measurement of individual microtubule lengths. We first measured the lengths of aMTs in fixed cells by immunofluorescence. Compared to wild-type cells, mutants lacking α-CTTs exhibited a slight but significant increase in aMT length (Figure 2B; S3; p<0.001). Mutants lacking β-CTTs showed a greater increase in aMT length, which was significantly different from those of wild type or α-CTT mutants (p<0.001). In the absence of both α- and β-CTTs, aMT lengths were similar to wild type (p=0.49). Swapping α- and β-CTTs elicited an intermediate phenotype, with some aMTs longer than wild type but typically shorter than α- or β-CTT truncation mutants (Figure 2B; S3).

To determine how CTTs affect the MT dynamics, we tracked Bik1-3GFP in live cells. Bik1 is the yeast homologue of CLIP-170 and associates with polymerizing and depolymerizing aMT plus ends (Figure 3A) [35]. Bik1-3GFP signal at aMT plus ends is decreased in both α- and β-CTT mutants; however, in each case Bik1-3GFP is detectable and sufficient for our measurements (Figure 3B). In wild-type cells we observed intermittent phases of polymerization and depolymerization, with Bik1-labeled aMT ends moving away from and toward the spindle, respectively (Movie 1). The distance between Bik1-3GFP foci and the spindle pole provides readout of aMT length over time, which we used to calculate rates of polymerization and depolymerization, and frequencies of transitions (Figure 3C; S4; Table 1). Loss of the β-CTT produces a striking phenotype. In these mutants, aMTs undergo prolonged phases of polymerization and depolymerization, and sample a broader range of lengths (Figure 3C; S4E; Table 1; Movie 1). Frequencies of catastrophe and rescue are significantly decreased, while rates of growth and shrinking are not altered. In contrast, loss of the α-CTT causes only a slight increase in depolymerization rate (Table 1; Figure S4B). We conclude that β-CTT promotes both catastrophe and rescue events.

Figure 3. Bik1 recruitment and microtubule plus end dynamics.

A) Representative images of wild-type and α-CTT mutant cells labeled with Bik1-3GFP (green) and Spc110-DsRed (red) shown as merge. Cells are classified as G1, pre-anaphase (pre-ana), and mitosis based on bud morphology and spindle morphology. Arrow marks Bik1-3GFP at an astral microtubule plus end. Scalebar = 2μm. B) Fluorescence intensity measurements of Bik1-3GFP foci at MT plus-ends, from G1, pre-anaphase, and mitotic classes. Plot depicts background-adjusted intensity measurements with bars indicating mean ±SD. Single asterisk indicates significant difference compared to wild type, by t-test (p < 0.05). Double asterisk, p < 0.01. C) Representative lifeplots of astral microtubule dynamics in wild type and β-CTT mutants. Astral microtubule length was measured over time by plotting the distance between plus-end associated Bik1-3GFP and the proximal spindle pole. See also Figure S3. Strains: wild type, yJM0424; tub1/3-442Δ, yJM0324; tub2-430Δ, yJM0541.

Table 1.

Microtubule dynamics in CTT mutants.

		Polymerization		Depolymerization
strain	length (μm)	rate (μm/min)	duration (s)	rate (μm/min)	duration (s)	Rescues (events/min)	Catastrophes (events/min)	Pause time (%)
wild type	1.3± 0.1	1.8 ± 0.1	44.3 ± 3.5	2.3 ± 0.1	32.6 ± 1.6	1.5 ± 0.2	0.8 ± 0.1	15.5 ± 3.3
tub1/3-442Δ	1.2 ± 0.1	2.3 ± 0.2	44.2 ± 5.1	3.3 ± 0.2	29.2 ± 2.8	2.2 ± 0.4	0.8 ± 0.1	23.1 ± 5.1
tub2-430Δ	2.9 ± 0.7	2.0 ± 0.2	74.6 ± 8.8	2.6 ± 0.2	61.3 ± 7.9	0.7 ± 0.1	0.5 ± 0.1	14.6 ± 4.3

μm, micron; min, minute; s, second; %, percent of total. Values are mean ± SEM of measurements from10 cells imaged for 576 seconds at 4 second intervals.

Mapping genetic interactions of CTT mutants

We next used a genetic screen to search for physiological roles of CTTs. We reasoned that cells tolerate CTT mutations because of compensation by related pathways. In this case, CTT mutations should exhibit negative genetic interactions with mutations that disrupt compensatory pathways; the fitness of double mutants would be worse than either single mutant alone. To identify genetic interactions, we crossed mutants lacking the major α-CTT (tub1-442Δ) or β-CTT (tub2-430Δ) to a comprehensive set of ~5000 viable haploid null mutants. Heterozygous diploids were then induced to undergo meiosis, and the growth of resulting haploid double mutant progeny was measured quantitatively [36]. We identified 67 negative interactions for the α-CTT mutant and 248 for the β-CTT mutant (Figure 4A). Only 7 genes were identified in both screens – SPT3, BUB3, LGE1, ADH5, AIM13, FCJ1, TNA1. The majority of negative interactions are unique to either the α-CTT or β-CTT mutant screen.

Figure 4. Mapping genetic interactions of CTT mutants.

A) Venn diagram of negative interactions recovered from SGA screens with tub1-442Δ and tub2-430Δ. The tub1-442Δ screen identified 67 negative genetic interactions. The tub2-430Δ screen identified 248. Seven interacting genes were common to both screens. B) Biological process gene ontology terms enriched among negatively interacting genes in tub1-442Δ and tub2-430Δ screens. Blue bars represent genes identified only in the tub1-442Δ screen, red bars represent genes identified only in the tub2-430Δ screen and yellow bars represent genes identified in both screens. All terms are significantly enriched in the respective data set, relative to the S. cerevisiae reference set, determined by Fisher’s exact test (p < 0.05; see Supplemental Experimental Procedures). C) tub2-430Δ screen is enriched for genes related to MT-processes and spindle checkpoints. Spindle checkpoint genes are highlighted in purple. Dynein pathway genes and highlighted in green. Red nodes are genes identified exclusively by the tub2-430Δ screen, yellow node (BUB3) was also identified in the tub1-442Δ screen, and white node (CIN8) is predicted to be related to the network (see Supplemental Experimental Procedures). Black lines denote previously identified genetic interactions with cin8 mutants.

We sorted negative interactions by gene ontology (GO) annotations and found that the results of the two screens are enriched with different GO terms (Figure 4B). The β-CTT mutant screen identified negative interactions with MT-related genes, including kinesin-14/KAR3, kinesin-5/KIP1, kinesin/KIP2, CLIP-170/BIK1, tubulin-specific chaperone A/RBL2, and genes required for the dynein pathway (Figure 4C). The β-CTT mutant also interacted with components of the Spindle Assembly Checkpoint (SAC) and Spindle Position Checkpoint (SPC), which detect errors in spindle function and delay anaphase or mitotic exit, respectively (Figure 4C). We used tetrad analysis to confirm these negative genetic interactions with the β-CTT truncation mutant; as expected, the fitness of haploid double mutant progeny was impaired (Figure S5). In addition, both the α-CTT and β-CTT mutants screens are enriched for genes that are not obviously related to MT function, and instead suggest diverse roles for both CTTs. These results indicate that α- and β-CTTs support different functions.

Distinct role for β-CTT in the mitotic spindle

Our genetic interaction screens suggest that β-CTT is important for spindle function. To determine the role of β-CTT, we used the GeneMANIA server to predict genes that are related to the 26 MT-based process and spindle checkpoint hits from the β-CTT screen. This identified CIN8, which encodes a kinesin-5 homologue. Kinesin-5/Cin8 is a microtubule motor that slides antiparallel MTs apart during spindle assembly and anaphase spindle elongation [37, 38]. Null mutants in cin8 exhibit a genetic interaction profile that is highly similar to the β-CTT mutant, including negative interactions with 24 of the 26 MT-based process and spindle checkpoint genes identified in our screen (Figure 4C). These results suggest that Cin8 may act in a common pathway with β-CTT. Although CIN8 was not identified in the β-CTT screen, we observed a negative genetic interaction between cin8 null and β-CTT truncation mutations by tetrad analysis (Figure S5). Therefore, Cin8 appears to retain some level of function that becomes essential in the absence of β-CTT.

To test this hypothesis, we evaluated spindle morphology and dynamics in β-CTT mutants. Cells expressing Spc110-tdTomato to label spindle poles and Dad1-GFP to label spindle MTs were released from G1 arrest and imaged at 15-minute intervals (Figure 5A). We observed several aberrant features. First, β-CTT mutants delayed the assembly of bipolar spindles (Figure 5B). Second, we rarely observed β-CTT mutants with long mitotic spindles, even 2.5 hours after release from G1 arrest (Figure 5C). Despite these defects, β-CTT mutants did not delay the arrival of one spindle pole in the bud; the frequency of bipolar spindles with one pole in the bud was not diminished relative to wild type (Figure 5D). Movement of one spindle pole into the bud requires either pushing the spindle poles apart (spindle elongation) or translocating the entire spindle; therefore, the β-CTT mutant may be defective in one but not both of these processes.

Figure 5. β-CTT is important for spindle assembly and elongation.

A) Wild-type cell labeled with Spc110-tdTomato (red) and Dad1-GFP (green), shown as merge with DIC image. Scalebar = 2μm. B) Spindle assembly. G1-arrested cells were released into fresh media and samples were collected at 15min intervals, fixed, and imaged. Percentage of cells with bipolar spindles (Spc110 foci linked by Dad1 signal) was determined for each timepoint. Wild-type cells show a decrease in the percentage of bipolar spindle at 120 min as many cells have completed mitosis and disassembled their spindles. The value increases at 150 min as cells begin another S-phase. At least 96 cells were assayed for each timepoint. Error bars are standard error of proportion. Strains: wild type, yJM0694; tub2-430Δ, yJM0695; tub2-430Δ td-dyn1, yJM0696. C) Spindle length. Pole-to-pole distance was measured for each bipolar spindle, and an average was determined per timepoint. Wild-type cells show decreased mean length at 135 min as many cells have completed mitosis and begun the assembly of short spindles for a new S-phase. At least 20 cells were assayed for each timepoint. Error bars are 95% confidence interval. D) Spindle movement. Percentage of bipolar spindles with at least one pole in the daughter cell was determined for each timepoint. Values for wild-type and tub2-430Δ cells decrease as cells complete mitosis and disassemble spindles. At least 20 cells were assayed for each timepoint. Error bars are standard error of proportion. E) Wild-type cell labeled with Spc110-GFP. Scalebar =2μm. F) Spindle elongation kinetics. Pole-to-pole distance was measured over time from 3D images of synchronized cells expressing Spc110-GFP. Plots are aligned at the timepoint prior to sustained spindle elongation (t=0s). Lines depict the mean pole-to-pole distance for at least 15 cells per strain. Error bars are 95% confidence interval. Strains: wild type, yJM0165; tub1/3-442Δ, yJM0215; tub2-430Δ, yJM0330. G) Dynein is essential in the absence of β-CTT. Cells were grown to saturation at 25°C in rich media and a 10-fold dilution series of each was spotted YPD; plates were grown at either 25°C or 37°C for 2 days. Strains: wild type, yJM0596; td-dyn1, yJM0604; tub1-442Δ td-dyn1, yJM0625; tub2-430Δ td-dyn1, yJM0623; tub1::βCTT tub2::αCTT td-dyn1, yJM0626; tub2::αCTT td-dyn1, yJM0629; tub1::βCTT tub2-430Δ td-dyn1, yJM0633.

We distinguished between these possibilities by examining movies of live cells at high time resolution. We measured elongation by tracking the distance between GFP-labeled spindle poles over time (Figure 5E; Movie 2). Consistent with published results, wild-type cells exhibited two phases of spindle elongation – an initial fast phase followed by a slow phase (Figure 5F). In contrast, β-CTT mutants exhibited a slow initial rate of elongation and reached a short terminal length (Figure 5F; S6). This is reminiscent of cin8Δ null mutants [37]. α-CTT mutants were indistinguishable from wild-type cells in this assay (Figure 5F; S6).

To evaluate spindle translocation, we examined spindles labeled with Dad1-GFP in live cells. We found that short bipolar spindles were highly mobile in β-CTT mutants (Movie 3). During these movements, spindles often traversed back-and-forth through the mother-bud junction, the bud neck. In contrast, short bipolar spindles in wild-type cells moved less often and over shorter distances (Movie 3). These results suggest that β-CTT mutants selectively impair pathways involved in spindle assembly and elongation, and enhance spindle translocation.

If β-CTT mutants are selectively impaired for spindle elongation, these cells might depend on spindle translocation pathways to deliver a spindle pole and associated genome to the bud. Consistent with this hypothesis, our genetic interaction screen reveals negative genetic interactions between the β-CTT mutant and mutants that disrupt dynein (Figure 4C). Dynein generates tension on aMTs from the yeast cell cortex to move the spindle into the bud neck [39]. Normally, dynein null mutations are not lethal; however, dynein mutants exhibit negative genetic interactions with mutants that disrupt spindle stability and elongation, including cin8 null mutants [40, 41]. To confirm the negative genetic interaction, we created a conditional allele to deplete dynein heavy chain in a temperature dependent manner (td-dyn1). β-CTT mutants carrying the td-dyn1 allele did not grow when dynein was depleted (37°C), confirming the results from our screen (Figure 5G). In contrast, cells expressing wild-type tubulins or α-CTT mutants grew when dynein was depleted (Figure 5G). We then asked whether β-CTT could rescue the genetic interaction with dynein mutants from a different site on the tubulin heterodimer. To test this, we used CTT swap alleles. Indeed, transferring β-CTT to α-tubulin was sufficient to rescue viability of tub2-430Δ mutants in the absence of dynein (Figure 5G). Transferring β-CTT to α-tubulin also rescued the fast phase of spindle elongation (Figure S6). Thus β-CTT is necessary for spindle elongation and becomes essential in the absence of dynein; but the presence of β-CTT at either carboxy-terminus in the heterodimer is sufficient to rescue.

To determine whether dynein is required to deliver one spindle pole to the bud in the absence of β-CTT, we examined spindles in β-CTT mutants after dynein depletion. Similar to β-CTT mutant alone, these cells failed to form long mitotic spindles (Figure 5C). Spindle lengths in β-CTT mutants were not significantly different after dynein depletion, indicating that dynein does not promote spindle elongation in β-CTT mutants. Spindle translocation, however, was ablated when dynein was depleted from β-CTT mutant cells (Figure 5D; Movie 4). We conclude that in the absence of dynein, β-CTT mutants are inviable because they fail to deliver a spindle pole and associated genome to the daughter cell. This indicates that β-CTT and dynein promote discrete aspects of spindle function.

We then asked whether β-CTT is important for the function of kinesin-5/Cin8. First we examined the localization of a functional fusion of Cin8 to 3 tandem copies of GFP, expressed from its native locus. Prior to anaphase, Cin8 is enriched on spindle MTs in wild-type cells (Figure 6A and B). In contrast, we found that Cin8 is significantly depleted from preanaphase spindles in β-CTT mutants. Transferring β-CTT to α-tubulin partially restored Cin8 localization to preanaphase spindles. During early anaphase in wild-type cells, when the spindle elongates, Cin8 becomes concentrated in a region of overlapping interpolar MTs known as the spindle midzone (Figure 6C and D; [42]). In β-CTT mutants, however, Cin8 does not accumulate at the center of the anaphase spindle and is instead enriched near the spindle poles (Figure 6C and D). Transferring β-CTT to α-tubulin improved Cin8 localization in some cells, but this difference was not significant across the population. These results suggest that β-CTT is important for the proper localization of Cin8 during spindle assembly and elongation.

Figure 6. β-CTT promotes the activity of kinesin-5/Cin8.

A) Preanaphase wild-type and tub2-430Δ cells labeled with Cin8-3GFP (green) and Spc110-tdTomato (magenta), shown as merge. Preanaphase spindles were defined based on spindle length (see Supplemental Experimental Procedures). Scalebar = 2μm. B) Distribution of Cin8-3GFP in preanaphase spindles. Values are mean ± SEM of 38 half spindles from 19 wild-type cells (black), 50 half spindles from 25 tub2-430Δ cells (red), and 28 half spindles from 14 tub2-430Δ tub1::β-CTT cells (green). Asterisk indicates significant difference compared to tub2-430Δ, by t-test (p < 0.05). Strains: wild type, yJM0987; tub2-430Δ, yJM1024 and 1025; tub2-430Δ tub1::β-CTT, yJM1002 and 1003. C) Early anaphase wild-type and tub2-430Δ cells labeled with Cin8-3GFP (green) and Spc110-tdTomato (magenta), shown as merge. Early anaphase spindles were defined based on spindle length (see Supplemental Experimental Procedures). Scalebar = 2μm. D) Distribution of Cin8-3GFP in early anaphase spindles. Values are mean ± SEM of 12 half spindles from 6 wild-type cells (black), 40 half spindles from 20 tub2-430Δ cells (red), and 20 half spindles from 10 tub2-430Δ tub1::β-CTT cells (green). E) Spindle assembly is not rescued by CIN8 overexpression. Synchronized cells were analyzed as described in Figure 5B. Percentage of cells with bipolar spindles (two Spc110 foci) was determined for each timepoint. CIN8 overxpression was induced 1hr prior to release from G1 arrest, and maintained after release. At least 131 cells were assayed for each timepoint. Error bars are standard error of proportion. Strains: wild type, yJM0165; tub2-430Δ, yJM0330; tub2-430Δ [PGAL-CIN8], yJM1050. F) Spindle elongation rate in the presence or absence of CIN8 overexpression. Cells were synchronized as described in B; however, galactose was added at release from G1 arrest. Pole-to-pole distance was measured over time from 3D images of synchronized cells expressing Spc110-GFP. Elongation events were defined as the initial elongation event for a cell beginning from pre-anaphase length and exhibiting increasing pole-pole distance for at least 80 seconds (16 frames with Pearson’s correlation coefficient >0.9). P-values were determined by t-test. Strains: wild type [PGAL-CIN8], yJM1046; others are identical to B.

If β-CTT promotes Cin8 function in the spindle, we reasoned that overexpressing CIN8 might rescue spindle phenotypes in the β-CTT mutant. We synchronized populations of β-CTT mutant cells carrying a galactose inducible CIN8 expression plasmid, released them from G1 arrest into galactose media, and measured spindle assembly and elongation by imaging labeled spindle poles. Overexpressing CIN8 in β-CTT mutants did not rescue the delay in bipolar spindle assembly (Figure 6E). In contrast, overexpressing CIN8 did rescue the rate of fast spindle elongation and promoted the formation of long anaphase spindles (Figure 6F, data not shown). Furthermore, elongation rates were indistinguishable for β-CTT mutants and wild-type cells during CIN8 overexpression (Figure 6F). Interestingly, we found that spindle elongation during CIN8 overexpression was followed by different outcomes -- spindles would elongate for up to several minutes and then either maintain the elongated state or collapse back to a preanaphase length. We observed this behavior in both wild-type and β-CTT mutant cells during CIN8 overexpression (Movies 5 and 6). Collapse events may represent transient spindle elongation during S-phase, which was previously observed during CIN8 overexpression [43, 44]. In addition, both wild-type and β-CTT mutant cells exhibited highly mobile spindles during CIN8 overexpression; the reason for this is unclear (Movies 5 and 6). We also tested whether increasing the levels of other spindle regulators could rescue anaphase spindle elongation in the β-CTT mutant; however, overexpressing ASE1, STU1, and KIP3 did not promote the formation of long anaphase spindles (data not shown). We conclude that β-CTT supports the activity of Cin8.

Discussion

Our findings provide new insight into the regulation of MT networks. We show that α- and β-CTTs have disparate effects on MT dynamics and spindle function, and our genetic screens link CTTs to diverse cellular processes. These results demonstrate that CTTs are important regulators of MT networks and suggest roles in novel functions.

CTTs regulate specific processes within MT networks. Previous biochemical studies have identified electrostatic interactions between CTTs and numerous MT motors and binding proteins. These interactions are thought to enhance motor processivity and MT-kinetochore coupling, among other functions. We find that loss of β-CTT in cells impairs MT dynamics, spindle assembly and anaphase spindle elongation (Figures 3 and 5). Our genetic interaction screens identify relationships between β-CTT and select MT pathways, indicating that β-CTT mutant phenotypes may be attributed to disruption of specific processes (Figure 4). One process that requires β-CTT is the function of kinesin-5/Cin8. The genetic interactions and spindle phenotypes of β-CTT mutants are reminiscent of cin8Δ null mutants (Figure 4C). Our localization experiments show that β-CTT mutants have decreased Cin8 in the spindle prior to anaphase and in the midzone during anaphase (Figure 6A–D). Increasing CIN8 expression in β-CTT mutants fails to rescue the spindle assembly defect, but does rescue the fast phase of anaphase spindle elongation (Figure 6E and F). These results are consistent with β-CTT acting upstream of Cin8 during anaphase, and suggest that the mechanism may differ during spindle assembly. Interestingly, anaphase spindle elongation and the requirement for dynein are rescued by transferring β-CTT to α-tubulin; however, this mutant does not fully restore Cin8 localization. Furthermore, overexpressing ASE1 does not rescue spindle elongation in β-CTT mutants. These data suggest that β-CTT does more than recruit Cin8 to the spindle. We speculate that β-CTT may directly promote Cin8 motor activity.

In addition to promoting Cin8, our results suggest that β-CTT may negatively regulate dynein. β-CTT mutants require dynein to move one spindle pole into the daughter cell, and exhibit excessive preanaphase spindle movement indicative of hyperactive dynein (Figure 5; Movie 3). β-CTT may, therefore, have different effects on evolutionarily distinct motors -- promoting Cin8 and dampening dynein. Such differential regulation would position β-CTT as a key regulator of force balance in the MT network.

The sequence and charge of CTTs are critical for specific functions. We find that replacing β-CTT with α-CTT does not rescue the genetic interaction with dynein mutants; however, moving the β-CTT to the carboxy-terminus of α-tubulin does rescue (Figure 5G). The β-CTT sequence is therefore necessary and sufficient for this function. This implies that differences in CTT sequences or posttranslational modifications could have profound effects on function. An important question is whether differences in CTT sequences across isotypes promote differences in MT behavior. To answer this question it will be important to compare the functions of CTTs from different isotypes. The yeast α-CTT is shorter and samples a smaller volume than mammalian α-CTT in our simulations; therefore, mammalian α-CTT may interact with binding partners or regions of the heterodimer or MT lattice that yeast α-CTT does not (Figure S1A and B). In contrast, the yeast β-CTT is longer, samples a larger volume in our simulations, and is more negatively charged than the genetically encoded mammalian β-CTTs (Figure S1A and B). How these changes in charge and sequence affect function, together with the added complexity of posttranslational modifications in higher eukaryotes, is an important question.

Given the biochemical evidence for CTT interactions with various binding partners, one might expect that disrupting CTTs in cells would be lethal. To the contrary, we find that yeast mutants lacking CTTs do not exhibit impaired fitness under optimal conditions, but are sensitized to MT depolymerizing drugs or other mutations (Figure 1C; 4). This contrasts previous studies in Tetrahymena, where deletion of either α- or β-CTT is lethal [30]. One possible explanation for this difference is that CTTs are necessary in the context of axonemal MTs. Cilia are important for motility and feeding in Tetrahymena; however, yeast cells do not form cilia and lack homologues of many axonemal proteins. Another possibility is that posttranslational modifications of CTTs may be essential in Tetrahymena. Consistent with this notion, Tetrahymena mutants that ablate glutamylation and glycylation sites in CTTs while preserving charge are lethal, but can be rescued by adding modification sites to the alternate tubulin CTTs [30, 45]. These modifications are enriched in axonemal MTs, where they are critical for function, but have not been identified in yeast CTTs. Although this hints that CTT modifications may have evolved along with axonemes, non-axonemal MTs such as spindle MTs also exhibit CTT modifications, suggesting additional functions [21]. Hence our findings in yeast provide a basis for understanding roles of CTTs that have expanded in higher eukaryotes to include essential functions.

In addition to roles in regulating the microtubule network, our genetic screens hint at novel roles that may be exclusive to α- or β-CTT. The α-CTT screen was enriched for genes related to mitochondrion organization, DNA damage response, and peroxisome organization, among others (Figure 4B). The β-CTT screen was enriched for genes involved in microtubule function and cell cycle checkpoints, but also genes related to cytosolic ribosomes and histone modification. These negative genetic interactions suggest that α- or β-CTT provides compensatory function related to these processes. Elucidating these functions will be the goal of future studies.

Experimental Procedures

Chemicals and reagents were from Fisher Scientific (Pittsburgh, PA) and Sigma-Aldrich (Saint Louis, MO), unless stated otherwise.

Yeast Strains and Manipulation

General yeast manipulation, media and transformation were performed by standard methods [46]. Mutant alleles of TUB1, TUB2, and TUB3 were generated at the endogenous chromosomal loci. Details of strain construction and a list of strains are provided in the supplemental material.

Genetic screen

Synthetic Genetic Array (SGA) analysis with tub1-442Δ and tub2-430Δ mutants, generated in the SGA strain background, was conducted as described [47].

Highlights.

• α- and β-tubulin CTTs promote unique, non-essential functions in yeast.

• β-tubulin CTT promotes microtubule dynamics.

• β-tubulin CTT supports kinesin-5 activity.

• Genetic interactions suggest novel roles for tubulin CTTs.