α-tubulin regulation by 5′ introns in Saccharomyces cerevisiae

Abstract

Across eukaryotic genomes, multiple α- and β-tubulin genes require regulation to ensure sufficient production of tubulin heterodimers. Features within these gene families that regulate expression remain underexplored. Here, we investigate the role of the 5′ intron in regulating α-tubulin expression in Saccharomyces cerevisiae. We find that the intron in the α-tubulin, TUB1, promotes α-tubulin expression and cell fitness during microtubule stress. The role of the TUB1 intron depends on proximity to the TUB1 promoter and sequence features that are distinct from the intron in the alternative α-tubulin isotype, TUB3. These results lead us to perform a screen to identify genes that act with the TUB1 intron. We identified several genes involved in chromatin remodeling, α/β-tubulin heterodimer assembly, and the spindle assembly checkpoint. We propose a model where the TUB1 intron promotes expression from the chromosomal locus and that this may represent a conserved mechanism for tubulin regulation under conditions that require high levels of tubulin production.

Introduction

Tubulin, the fundamental subunit of the microtubule cytoskeleton, is a heterodimeric protein consisting of α and β monomers. In many eukaryotes, the α and β monomers are encoded by a family of genes at disparate genomic loci. For example, in humans there are 9 well-annotated genes that encode α-tubulin and 8 β-tubulin genes, located across 5 and 7 chromosomes, respectively (Leandro-García et al. 2010; Findeisen et al. 2014; Park et al. 2021). Some of the tubulin genes are clustered together on chromosomes (for example, TUBA1A, TUBA1B, and TUBA1C); however, many are the only tubulin gene encoded on its chromosome. It is unclear how the structure of the tubulin genes impacts tubulin expression.

Our previous work showed that cells must maintain excess α-tubulin for efficient mitotic timing and to prevent the accumulation of monomeric β-tubulin (Wethekam and Moore 2023). Conserved features within coding sequences of α-tubulin genes may provide a point of regulation. One conserved feature of α-tubulin genes is an intron located near the 5′ end of the coding sequence (Fig. 1a). Within the human α-tubulins, the early intron is positioned immediately 3′ of the start codon, while in both budding yeast α-tubulin genes, known as TUB1 and TUB3, the intron is positioned 23 base pairs 3′ of the start codon (ORF; Fig. 1a; Findeisen et al. 2014). The role of the early intron in α-tubulin gene function is unexplored.

Fig. 1.

TUB1 intron promotes α-tubulin expression. a) Diagram of α-tubulin genes from S. cerevisiae, K. lactis, and H. sapiens. All genes are drawn to scale with 20 bp/pixel. For 5 introns the scale was adjusted, // = 40 bp per pixel, /// = 80 bp per pixel. b) Diagram of genotypes used in this figure. c) Example western blot used to determine the number of α-tubulin molecules per cell. Left shows the purified tubulin standards used to build the standard curve. Right shows the estimated number of cells loaded onto the gel used to identify the number of molecules of α-tubulin per cell. d) Quantification of α-tubulin molecules per cell across each genotype. Band intensities were converted to protein mass (ng) and then converted to molecules per cell. Data represent at least 3 independent experiments with 2 biological replicates per experiment. Each dot represents 1 mean molecules per cell measurement for at least 3 dilutions per biological replicate. Dots are colored by experiment. e) Quantification of doubling times for the indicated mutants normalized to the mean of the technical replicates for wild type. For each genotype, at least 3 technical replicates were used in at least 3 independent experiments. Each circle represents a single technical replicate, with triangles representing the mean of all technical replicates. Shapes are colored based on experiment. Bars represent mean ± 95% CI. P-values are from a t-test comparing wild type and mutant; specific P-values are listed in the text. f) Ten-fold dilution series of listed strains spotted onto rich media or rich media supplemented with 5 or 10 µg/mL benomyl. Plates were grown at 30°C for 2 days before imaging.

The conservation of the early intron in budding yeast α-tubulins is particularly notable. While introns are common within metazoan genomes, only 5% of the Saccharomyces cerevisiae genes contain introns (Stajich et al. 2007; Neuvéglise et al. 2011). Of the intron containing genes in S. cerevisiae, a majority encode ribosomal proteins and removing the intron disrupts ribosome function and ribosomal protein expression, particularly under stress (Parenteau et al. 2011). Another example of a gene with a 5′ intron important for regulating protein production in S. cerevisiae is ACT1, the gene encoding g-actin. ACT1 is the only gene that encodes g-actin in S. cerevisiae and must be expressed at high levels to maintain the dynamic actin cytoskeleton (Wertman et al. 1992; Blank et al. 2020). The ACT1 intron has been shown to promote the expression of ACT1 and is especially important when cells are stressed with the actin-depolymerizing drug, latrunculin A (Juneau et al. 2006; Agarwal and Ansari 2016). Removal of the ACT1 intron also reduces the amount of ACT1 mRNA (Juneau et al. 2006; Agarwal and Ansari 2016). Conversely, exogenously inserting the ACT1 intron into another gene can increase the expression of that gene, and this depends on the location of the intron within the coding sequence, with the enhancing capability of the intron decreasing the further it is placed 3′ of the start codon (Agarwal and Ansari 2016; Dwyer et al. 2021). This suggests that the α-tubulin early introns may also promote protein expression through a related mechanism.

In this study, we sought to understand the role of the α-tubulin introns in regulating α-tubulin protein production. We used budding yeast which contains 2 genes for α-tubulin, TUB1 and TUB3, and both genes contain introns. We find that the intron in TUB1 is important for promoting the expression of α-tubulin and resistance to microtubule stress. Furthermore, the TUB1 intron exhibits a stronger effect than the TUB3 intron and partially rescues the stress sensitivity of cells expressing TUB3 only. A genetic screen for potential regulators of TUB1 that act through the intron identified genes encoding RNA-regulating proteins and chromatin modifiers, representing novel regulators of α-tubulin expression. Our screen also identified several genes known to be involved in heterodimer biogenesis and turnover, suggesting that the TUB1 intron is important for the maintaining balance between heterodimer production and destruction.

Methods

Yeast manipulation and culturing

Yeast manipulation, media, and transformations were performed by standard methods (Amberg et al. 2000). Deletion mutants were generated by PCR-based, homologous recombination methods (Petracek and Longtine 2002). All gene and intron swap alleles were generated by PCR amplification from plasmid templates described below (Supplementary Table 3) and transformation and homologous recombination into diploid strains where 1 copy of TUB1 is replaced by a hygromycin B resistance marker (yJM0591; Supplementary Table 2; Goldstein and McCusker 1999). Heterozygous diploids with 1 wild-type and 1 mutant allele were then dissected to acquire haploids. All alleles were confirmed by sequencing. To build the tub1Δi allele, the coding sequence from KGY2914 (Burns et al. 2002) was amplified by PCR and transformed into diploid strain yJM0591.

Plasmid construction

To build the plasmid containing a 6xHis tag inserted between codons 43 and 44 in TUB1, genomic DNA from strain yJM1796 was used as a template for a PCR amplicon containing 450 bp of TUB1 UAS, the coding sequence and intron, and 450 bp of UTR including a URA3 marker 280 bp 3′ of the stop codon. This amplicon was cloned into pRS314 at the NotI and KpnI sites to create pJM738. To build the tub1Δi plasmid, QuikChange mutagenesis oligos were generated to remove the intron from pJM738 plasmid. To build the tub1i∷ACT1i and the tub1i∷TUB3i plasmids, a Gibson assembly reaction was used to combine pJM738 with the ACT1i or TUB3i amplified from the genomic DNA from yJM1837. To build the tub1∷TUB3 locus swap plasmid, a Gibson assembly reaction was used to replace the TUB1 coding sequence and intron in pJM738 with the TUB3 coding sequence and intron, which was amplified from pJM886. To build the tub1∷TUB3^TUB1i plasmid, first a Gibson assembly reaction was used to exchange the TUB1 intron into the tub3 locus in plasmid pJM886. Then a second Gibson assembly reaction was used to replace the TUB1 coding sequence and intron in pJM887 with the tub1∷TUB3^TUB1i sequence.

Doubling time measurement

Cells were grown in 3 mL of rich liquid media (YPD) to saturation at 30°C and diluted 1:50 into fresh media. The diluted cultures were then aliquoted into a 96-well plate, with 3 to 6 technical replicates per experiment, and incubated at 30°C while single orbital shaking in a plate reader. We used 2 different instruments for our experiments. A Cytation3 plate reader (BioTek, Winooski, VT) was used for experiments with strains: yJM1837, 1838, 0120, 0121, 4478, 4479, 4611, and 4745, and Epoch2 microplate reader (BioTek, Winooski, VT) was used for experiments with strains: yJM1837, 0120, 5124-5127, and 5158-5163. The OD₆₀₀ was measured every 5 min for 24 h. Doubling time was calculated by fitting the growth curves to a nonlinear exponential growth curve as previously published (Fees and Moore 2018). Each experiment was repeated 3 independent times with wild-type cells included in each experiment as an internal control. P-values are from Student's t-test.

Benomyl sensitivity assay

Cells were grown in rich liquid media to saturation at 30°C, and a 10-fold dilution series of each culture was spotted to either rich media plates or rich media plates supplemented with 5 or 10 µg/mL benomyl (Sigma-Aldrich #381586, St. Louis, MO, USA). Plates were grown at the indicated temperature for the indicated days.

Western blotting

Soluble protein lysates were prepared under denaturing conditions using the method of Zhang et al. (2011). To make lysate, log-phase cells were pelleted and resuspended in 2 M lithium acetate and incubated for 5 min at room temperature. Cells were pelleted again and resuspended in 0.4 M NaOH for 5 min on ice. Cells were pelleted and resuspended in 2.5× Laemmli buffer and boiled for 5 min. Before loading gels, samples were boiled and centrifuged at 6,000 × g for 3 min. Total protein concentration of clarified lysate was determined by Pierce 660 nm protein assay with the Ionic Detergent Compatibility Reagent before blotting (Cat. 1861426 and 22663, Rockford, IL, USA). 0.75µg of total protein was loaded per lane for western blots looking to separate the 2 α-tubulin isotypes. Samples were run on 10% Bis-Tris PAGE gels in NuPAGE MOPS running buffer (50 mM MOPS, 50 mM TrisBase, 0.1% SDS, 1 mM EDTA, and pH 7.7) at 0.04 mAmp per gel for 2.25−2.5 h to separate α-tubulin isotypes or 1.25 h to determine the number of molecules per cell. Gels were transferred to PVDF (Millipore, IPFL85R) in NuPAGE transfer buffer (25 mM Bicine, 25 mM Bis-Tris, 1 mM EDTA, and pH7.2) at 0.33 mAmp for 1 h. Membranes were then blocked for 1 h at room temperature in PBS blocking buffer (LI-COR, 927-70001). Membranes were probed in PBS blocking buffer including the following primary antibodies: mouse anti-α-tubulin (4A1; at 1:100; Piperno and Fuller 1985), mouse anti-β-tubulin (E7; at 1:100; Developmental Studies Hybridoma Bank, University of Iowa), rabbit anti-Zwf1 (glucose-6-phosphate dehydrogenase; Sigma A9521; at 1:10,000) overnight at 4°C. After incubation in primary antibody, membranes were washed once in PBS for 5 min at room temperature and then probed with the following secondary antibodies: goat antimouse-680 (LI-COR 926-68070, Superior, NE; at 1:15,000) and goat antirabbit-800 (LI-COR 926-32211; at 1:15,000) for 1 h at room temperature. After incubation in secondary antibodies, blots were washed twice in PBST (1XPBS, 0.1% Tween-20), once in PBS, and imaged on an Odyssey Imager (LI-COR, 2471).

Quantifying tubulin concentration

To determine levels of tubulin in the cell, wild-type or mutant cells were grown to log phase in rich media at 30°C. To prepare lysate of 5 × 10⁷ cells in 50 µL samples, cells from each culture were counted on a hemocytometer, and the appropriate volume of cells was determined based on the hemocytometer counts and prepared as described above. To confirm the number of cells per volume of culture, separate samples of cells from the overnight cultures were diluted and plated to rich media at ∼200 cells/plate. After 2 days at 30°C, the number of colonies/plate was counted, and the fraction of cells that formed colonies was determined. Lysate was resuspended in 2.5× Laemmli buffer, and standards of purified yeast tubulin (Wethekam and Moore 2023) were prepared by diluting protein to 2.5 ng/µL in 2.5× Laemmli buffer. Samples containing increasing amounts of cells (3.5, 4.5, 6, and 8 × 10⁶) or purified tubulin (4,10, 15, 30, and 40 ng of total protein heterodimers or 2, 5, 7.5, 15, and 20 ng of α-tubulin) were loaded and blotted as described above. Band intensities were quantified using the gel analysis plug-in in FIJI.

To confirm the amount of cell lysate loaded per lane, we used a method described in Wethekam and Moore (2023). Zwf1 loading control intensity was plotted against expected number of cells loaded, and the r² value was calculated. If the r² value was <0.75, the outlier lane was identified, and we excluded it. If at least 3 lanes from a replicate did not generate an r² value of ≥0.8, then that replicate was removed. The proportionality of Zwf1 signal to cell number was determined by dividing the measured Zwf1 signal intensity per lane by the number of expected cells loaded, and the average of those values represents the Zwf1 signal per cell for that blot. That value was used to recalculate the number of cells in each lane by dividing the Zwf1 band intensity by the average Zwf1 signal per cell.

Finally, we converted band intensities for α-tubulin from cell lysates into estimated nanograms of protein using a standard curve ranging from 2.0 to 20 ng of α-tubulin. Linearity of the signal was assessed as a function of ng loaded for each standard curve and set a cut off of r² < 0.85. In some cases, we found that higher amounts (i.e. 20 ng) of pure tubulin deviated from the linear regression, and in these cases, we limited the standard curves to lower amounts of tubulin. We used these linear regressions to calculate the ng of α-tubulin in each lane of cell lysate for that blot. The calculated ng of α-tubulin was then divided by the estimated number of cells in that lane and converted to molecules/cell with the following:

$$\mathrm{ng}*{\displaystyle \frac{1\mathrm{Da}}{1.66*{10}^{-15}\mathrm{ng}}*{\displaystyle \frac{1\mathrm{kDa}}{1,000\mathrm{Da}}*{\displaystyle \frac{1\mathrm{molecule}}{\mathrm{MW}\mathrm{kDa}}*{\displaystyle \frac{1}{\mathrm{cells}\mathrm{loaded}}={\displaystyle \frac{\mathrm{molecules}}{\mathrm{cell}}}}}}}$$

The molecule/cell values for biological replicates were averaged for a single experiment, and the corresponding α-tubulin was compared to determine the ratio of α-tubulin. P-values are from Student's t-test after a 1-way ANOVA with a Tukey post hoc test for P < 0.05.

RNA preparations and RT-qPCR

Total RNA was isolated from log-phase cultures using a standard phenol chloroform-based extraction method. Briefly, cells were pelleted and resuspended in TES (10 mM Tris, pH 7.5, 1 mM EDTA, and 0.5% SDS). Warm acid phenol was then added, and cells were vortexed for 15 s every 5 min for 30 min at 65°C. Cell lysates were incubated on ice before centrifugation at 13,400 rpm at 4°C for 10 min. The aqueous phase was transferred to a new tube, and chloroform was added before the tube was vortexed and centrifuged. The aqueous layer was transferred to a new tube with 3 M sodium acetate and 100% ethanol, and tubes were stored overnight at −80°C. Pellets were thawed on ice then centrifuged at 13,000 rpm at 4°C for 8 min. Supernatant was discarded, and pellets were washed with 70% ethanol and centrifuged again. Supernatant was discarded, and pellets were air-dried. RNA was resuspended and 50 µg was DNase treated using the TURBO DNA-free kit (Invitrogen, AM1907). Five micrograms of cDNA was synthesized using SuperScript III First-Strand Synthesis System (Invitrogen, 18080051) primed with random hexamers.

For the qPCR, all cDNA was diluted to 5 ng/µl. Primers were designed to amplify the pre-mRNA for wild-type or chimeric TUB1 or mature TUB1 mRNA and the intronless control PGK1. For each reaction, we combined the following: 5 µL of iTaq Universal SYBR Green Supermix (BioRad, 1725121), 0.2 µL of forward and reverse primers, and 4.8 µL of cDNA. Each primer set was done in triplicate per strain. Data were analyzed using the 2^−ΔΔCT method (Schmittgen and Livak 2008). RT-qPCR was completed on 3 independent cDNA preparations.

Deletion collection and SGA screen

The yeast deletion collection was prepared as described in Baryshnikova et al. (2010) (Baryshnikova et al. 2010). Briefly, the deletion collection (Open Biosciences now Horizon Discovery, YSC1053, Winzeler et al. 1999; Gardner and Jaspersen 2014) was thawed from −80°C storage and transferred onto rich media + 50 µg/mL G418 with the Singer RoToR (Singer Instruments, Somerset, UK). Four, 96-well plates were condensed onto 1 plate resulting in 384 colonies and then amplified to quadruplicate to generate an array of 1,536 deletion mutants per plate. These plates were incubated at 30°C and then transferred to 4 different plates with the following conditions: rich media incubated at 30°C, rich media incubated at 15°C, rich media with 2% DMSO incubated at 30°C, and rich media with 10 µg/mL benomyl and 2% DMSO incubated at 30°C. Plates were incubated at 30°C for 2 days, or at 15°C for 10 days. Each plate was scanned on an Epson Perfection V300 Photo scanner and processed using the gitter plugin in R as described on http://gitter.ccbr.utoronto.ca (Wagih and Parts 2014). Plates were manually rotated in FIJI to ensure colony recognition. Once colonies were identified, data were normalized, and scores were calculated using http://sgatools.ccbr.utoronto.ca (Wagih et al. 2013). Since this analysis compares the deletion collection on different conditions, no linkage was taken into account for scoring. Each of the conditions was normalized to the rich media alone, but we also compared the 10 µg/mL benomyl + 2% DMSO data to the 2% DMSO data to identify genes that are specifically sensitive to benomyl and not DMSO. We used a score > −0.08 and P-value < 0.05 as the cutoffs for genes we considered benomyl-sensitive. The scores from that comparison were used for analysis in Fig. 3.

Fig. 3.

Genome-wide screen to identify genes that regulate TUB1 through its intron. a) Plot of Cellular Component GO Terms with P < 0.05 and not redundant with other GO terms. b) GeneMANIA network analysis of physical interactions between the 33 genes identified from our screening and filtering.

To screen for genetic interactions with the deletion collection, we generated a tub1Δi allele in strain Y7092 (Baryshnikova et al. 2010). We carried out the synthetic genetic array (SGA) screen following the protocol described in Baryshnikova et al. (2010). 1536-colony array plates containing haploid, double mutants of tub1Δi and the deletion mutants were stamped to 4 different conditions: rich media incubated at 30°C, rich media to be incubated at 15°C, rich media with 2% DMSO incubated at 30°C, and rich media with 10 µg/mL benomyl and 2% DMSO incubated at 30°C. Plates were incubated at 30°C for 2 days, or at 15°C for 10 days. All of the plates were processed as described for the deletion collection alone with the following modifications: (1) for each condition, double mutants were compared to the deletion collection alone; (2) we completed this analysis for both 200 kb of linkage and no linkage (Baryshnikova et al. 2010; Wagih et al. 2013).

GO term analysis and plotting

To obtain enriched GO terms for both hits identified in the genetic screen and the negative genetic interactors pulled from the Saccharomyces Genome Database we used g:Profiler (Raudvere et al. 2019). All settings were left at default with the exceptions of species switched to S. cerevisiae, and all results were displayed. For Fig. 3a, the −log₁₀(P-values) were plotted for GO terms with P < 0.05 that were also nonredundant with other GO terms.

Protein interaction networks

Network map showing previously annotated protein–protein interactions were generated using GeneMANIA (Montojo et al. 2010). Network branches are weighted by Cellular Component GO terms.

α-Tubulin isotype analysis by western blot

To measure the proportional levels of each α-tubulin isotype, we used strains where a 6× histidine tag was inserted between codons 43 and 44 of TUB1, allowing the resolution of Tub1 and Tub3 proteins by western blot. Western blots were then analyzed using the following method. We first rotated the gel image in FIJI to ensure that each lane would be square to our regions of interest (ROIs). To sample the band intensity in each lane, we used the Gel Analysis Plugin to create four 5-pixel × 500-pixel ROIs and place them along the vertical axis of the gel and across the middle of each lane, ∼20, 40, 60, and 80% of the width of the lane. Intensity profiles were plotted, and the background for each ROI was removed by drawing a line at the minimum intensity value. To identify the peak for Tub1 and Tub3, lines were drawn in 3 places: on either side of the α-tubulin signal and then at the minimum intensity point between the Tub1 and Tub3 peaks. We then measured the area within each peak corresponding to the slower-migrating Tub1 with the internal 6× histidine tag and faster migrating Tub3. The fraction of Tub1 reported in Fig. 4c represents the intensity value of Tub1 divided by the sum of Tub1 and Tub3 from the same sample. Each dot in Fig. 4e represents a biological replicate as the mean from the 4 positions in a single lane.

Fig. 4.

Screen hits identify novel regulators of α-tubulin expression. a) Ten-fold dilution series of screen hits with an extra copy of TUB1 or a control plasmid spotted onto rich media or rich media supplemented with 10 µg/mL benomyl. Plates were incubated at 30°C for 3 days before imaging. b) Ten-fold dilution series of screen hits with an extra copy of TUB3 or a control plasmid spotted onto rich media or rich media supplemented with 10 µg/mL benomyl. Plates were incubated at 30°C for 3 days before imaging. c) Ten-fold dilution series of screen hits with the indicated TUB1 allele spotted onto rich media or rich media containing 5 µg/mL benomyl. Plates were incubated at 30°C for 3 days or at 15°C for 17 days before imaging. d) Representative western blot of α-tubulin contribution assay. Blots were probed for α-tubulin and Zwf1 (G6PD) as a loading control. Bands corresponding to Tub1-6xHis and Tub3 are labeled. e) Quantification of α-tubulin contribution assay for the fraction of α-tubulin that corresponds to Tub1-6xHis. Dots and triangles represent different biological replicates. Bars represent mean ± 95% CI. P-values are from a t-test comparing wild type and mutant; specific P-values are listed in the text.

Results

TUB1 intron promotes α-tubulin expression

5′ introns are a common feature among α-tubulin genes. Both the TUB1 and TUB3 genes in S. cerevisiae contain single introns that begin 23 base pairs 3′ of the start codon (Fig. 1a). A similar intron is found in the single α-tubulin of the budding yeast Kluyveromyces lactis, suggesting that the intron predates the whole-genome duplication (Fig. 1a). Furthermore, all human α-tubulin genes exhibit an intron immediately 3′ of the start codon (Fig. 1a). The prevalence of 5′ introns raises the hypothesis that these may represent an important and conserved feature for promoting α-tubulin gene function.

We used 3 experiments to determine if the TUB1 intron is important for promoting α-tubulin expression in S. cerevisiae. First, we tested if the TUB1 intron impacts α-tubulin at the protein level by performing quantitative western blots. To do this, we used purified yeast tubulin to build standard curves of α-tubulin. Band intensities of cell lysates from log-phase cells grown in rich media were compared to the standard curve and converted from nanograms of protein to molecules per cell (see Material and methods; Wethekam and Moore 2023). In cells where we removed the TUB1 intron, the level of α-tubulin across experiments and biological replicates was more consistent and tended to be less than what we observed in wild-type controls (Fig. 1b–d). However, this apparent difference was not significant (P = 0.30). To isolate the output from the intron-less tub1Δi gene, we knocked out the other α-tubulin isotype, TUB3. These tub1Δi tub3Δ cells exhibit ∼50% less α-tubulin than wild-type controls or TUB1 tub3Δ cells (P = 0.01, P = 0.026; Fig. 1b–d). These results suggest that the TUB1 intron promotes α-tubulin protein production and that this role is most apparent when TUB1 is the only source of α-tubulin.

We next tested whether the TUB1 intron is important for cell fitness by comparing doubling time in cells with or without the TUB1 intron. Cells were grown to saturation, diluted 500-fold into rich media, and the OD₆₀₀ was recorded every 5 min for 20 h. We used these measurements to estimate the doubling time during the exponential phase of growth. In cells where we removed the TUB1 intron, there is no distinguishable difference in doubling time between wild-type and tub1Δi cells (P = 0.15; Fig. 1b and e). We also find no distinguishable difference in tub3Δ compared to wild type (P = 0.44; Fig. 1b and e). In contrast, we see a significant increase in the doubling time in tub1Δi tub3Δ cells, where the only source of α-tubulin is tub1Δi (P = 0.006; Fig. 1b and e). Together these results indicate that the TUB1 intron promotes fitness.

As a final test, we determined if the TUB1 intron was important for tolerance of microtubule stress. Benomyl is a fungicide that creates microtubule stress by inhibiting tubulin polymerization (Gupta et al. 2004; Rathinasamy and Panda 2006). We predicted that if removing the TUB1 intron weakens α-tubulin production, then these cells should exhibit increased sensitivity to benomyl. Consistent with this prediction, tub3Δ cells that lack the minor α-tubulin isotype proliferate slower than wild-type control cells at low concentrations of benomyl (5 µg/mL; Fig. 1b and f). tub1Δi cells are also hypersensitive to benomyl but to a lesser degree than tub3Δ cells (Fig. 1b and f). In tub1Δi tub3Δ, double-mutant cells exhibit a level of sensitivity that is similar to tub3Δ single-mutant cells (Fig. 1b and f). We also find that providing an additional copy of TUB1 rescues the benomyl sensitivity of these mutants (Supplementary Fig. 2). These results suggest that the TUB1 intron is important for tolerance of microtubule stress and that the role of the TUB1 intron may be to promote the production of Tub1 protein.

TUB1 intron is not rescued by the ACT1 intron

If the intron promotes TUB1 function, we predicted that this function might be rescued by substituting an alternative intron that is known to promote expression of other genes. We tested this prediction by replacing the TUB1 intron with the ACT1 intron sequence at the endogenous TUB1 locus, creating tub1i∷ACT1i (Fig. 1b) and used the 3 experiments described above. We find that tub1i∷act1i cells hold ∼33% less α-tubulin than wild type and tub1i∷ACT1i tub3Δ double mutants hold 54% less α-tubulin than wild type (P = 0.08 and 0.02, respectively; Fig. 1b–d). In the doubling time assay, tub1i∷ACT1i cells grow 20% slower than wild-type controls or tub1Δi cells (157.3 min vs 206.4 min; P = 0.02; Fig. 1b and e). tub1i::ACT1i tub3Δ proliferates >100% slower than tub3Δ cells (356.5 min compared to 157.3 min; P = 0.0001; Fig. 1b and e). In the benomyl assay, tub1i∷ACT1i cells exhibit greater sensitivity to benomyl than the wild-type controls or tub1Δi cells (Fig. 1b and d). This is sensitivity is exacerbated in tub1i∷ACT1i tub3Δ double-mutant cells (Fig. 1b and d). The benomyl sensitivity of both strains is rescued by providing an additional copy of TUB1 (Supplementary Fig. 2). Together, these results indicate that the ACT1 intron cannot rescue the TUB1 intron and suggest that the sequence of the TUB1 intron contributes a distinct function.

Comparing the introns of TUB1 and TUB3 isotypes

The introns in both TUB1 and TUB3 begin at the same position in the ORF but differ in sequence and total length (Figs. 1a and 2a). The TUB1 intron is 118 bp long, and the TUB3 intron is 300 bp long (Fig. 2a and Supplementary Fig. 1). To test whether the TUB1 and TUB3 introns are functionally equivalent, we generated 3 chimeric alleles at the TUB1 locus: (1) tub1i::TUB3i, which has the coding sequence of TUB1 and the intron of TUB3; (2) tub1Δ::TUB3, which has the coding sequence and intron of TUB3; and (3) tub1Δ::TUB3^TUB1i which has the coding sequence of TUB3 and the intron of TUB1 (Fig. 2b). Replacing the TUB1 intron with that of TUB3 showed a moderate level of benomyl sensitivity that matches the sensitivity of tub1Δi cells (Fig. 2c). Consistent with prior work, tub1Δ∷TUB3 cells are hypersensitive to benomyl (Fig. 2b and c; Nsamba et al. 2021), but that benomyl sensitivity is partially rescued in tub1Δ::TUB3^TUB1i cells (Fig. 2b and c). This benomyl sensitivity is fully rescued by providing an additional copy of TUB1 (Supplementary Fig. 2). This suggests that the TUB1 intron provides a higher level of function than the TUB3 intron, even when combined with the coding sequence of TUB3. To confirm the partial rescue in a separate assay, we measured doubling time in the presence of a different microtubule-destabilizing drug, nocodazole. Whereas the chimeric alleles show no effect in DMSO controls, in 5 µM nocodazole tub1Δ∷TUB3 cells show a strong increase in doubling time (248.7 min), and tub1Δ∷TUB3^TUB1i cells show an intermediate effect (210.5 min), compared to wild-type controls (163.5 min; Fig. 2d). Together, these results suggest that the TUB1 and TUB3 introns are not functionally equivalent and that the TUB1 intron may promote a higher level of α-tubulin function.

Fig. 2.

TUB1 intron protects cells from microtubule stress. a) Diagram of both α-tubulin introns and the ACT1 intron. The last base pair in exon 1 is listed before the 5′ splice site sequence. The length from the end of the 5′ splice site to beginning of the branch point sequence is listed and drawn to scale. Length from the end of the branch point sequence to the start of the 3′ splice site is also drawn to scale. The base pair of the exon 2 is listed immediately following the 3′ splice site. Scale is 1 bp per pixel. b) Diagram of the genotypes used in the benomyl sensitivity assay. c) Ten-fold dilution series of listed strains spotted onto rich media or rich media supplemented with 5 µg/mL benomyl. Plates were grown at 30°C for 2 days before imaging. d) Quantification of doubling time for each of the indicated strain under indicated concentration of nocodazole. For each genotype 2 biological replicates were used in at least 2 independent experiments. Three technical replicates were used in each independent experiment. Circles represent 1 biological replicate, and triangles represent the other biological replicate. Circles or triangles with boarders represent the means of the technical replicates; all other circles or triangles represent technical replicates. e) Quantification of relative TUB1 pre-mRNA expression normalized to wild type. Error bars represent SD. Each dot is the relative expression for 3 technical replicates. P-values based on 2-tailed t-test compared to wild type. f) Quantification of relative TUB1 mature mRNA expression normalized to wild type. Error bars represent SD. Each dot is the relative expression for 3 technical replicates. P-values based on 2-tailed t-test compared to wild type.

We used RT-qPCR experiments to determine whether the TUB1 intron promotes α-tubulin function by increasing levels of TUB1 mRNA. We designed oligonucleotide primers to bind to either intronic and/or coding sequence of TUB1, enabling the measurement of pre-mRNA or spliced, mature mRNA (Supplementary Table 5). Compared to wild-type cells, levels of TUB1 pre-mRNA and mature mRNA are significantly increased in in the tub3Δ (P = 0.03 and 0.003; Fig. 2e and f). This suggests that when TUB3 is knocked out, cells respond by overproducing TUB1 mRNA to increase α-tubulin protein levels. tub1Δi cells show no pre-mRNA expression, as expected (Fig. 2e), and decreased levels of mature mRNA compared to wild type (P = 0.13; Fig. 2f). This result suggests that the intron promotes α-tubulin production by increasing the abundance of mature TUB1 mRNA. tub1i∷ACT1i cells show increased levels of pre-mRNA compared to wild-type controls but decreased mature mRNA, (P = 0.12 and 0.023; Fig. 2e and f). This is consistent with the decreased α-tubulin levels seen by western blot (Fig. 1d) and suggests that the ACT1 intron promotes pre-mRNA levels but may be less efficient at promoting mature TUB1 mRNA levels. Finally, tub1i∷TUB3i cells show slightly decreased levels of pre-mRNA (P = 0.15; Fig. 2e) and significantly decreased levels of mature mRNA (0.0004; Fig. 2f). This result is consistent with the benomyl sensitivity of tub1i∷TUB3i cells. Together these data demonstrate that the native TUB1 intron promotes high levels of mature α-tubulin mRNA.

Identifying genes that act through the TUB1 intron

Our results above indicate that the TUB1 intron promotes α-tubulin expression. To elucidate the underlying mechanism and identify extrinsic regulators that might act in the same pathway with the intron, we used a genetic interaction screen with the collection of ∼5000 nonessential gene deletion strains (see Materials and Methods). We first predicted that loss of a gene that promotes TUB1 expression would create hypersensitivity to benomyl, similar to cells that lack the TUB1 intron (Fig. 1d). We compared the growth of the haploid deletion collection on rich media supplemented with 10 µg/mL benomyl and 2% DMSO to rich media with 2% DMSO alone and used a previously published image analysis method to measure and score the growth of 4 technical replicates of each strain (Wagih et al. 2013; Wagih and Parts 2014). This analysis identified 649 gene deletions that exhibit hypersensitivity to benomyl (Supplementary Table 1). To narrow this list and identify genes that may act in a pathway with the TUB1 intron to promote α-tubulin expression, we focused on genes that are known to exhibit a negative genetic interaction with a tub3Δ, since loss of TUB3 exacerbates the tub1Δi allele in our experiments. We identified 150 genes listed as negative genetic interactors with tub3Δ in the Saccharomyces Genome Database; 42 of these genes are also benomyl-sensitive in our assay (Supplementary Table 2). Finally, we predicted that genes acting through the TUB1 intron would exhibit a positive genetic interaction with tub1Δi. In other words, combining a mutant allele that ablates gene function with a TUB1 mutant allele that lacks its intron would exhibit a level of benomyl sensitivity that is equivalent or better than either single mutant alone. To identify this set of genes, we generated double mutants by SGA with the complete haploid deletion collection (see Materials and Methods) and quantitatively compared the growth of double mutants on 10 µg/mL benomyl with 2% DMSO to that of single mutants under the same conditions. We identified 33 genes where combining the deletion allele with the tub1Δi allele exhibits a similar level of benomyl sensitivity or improves benomyl sensitivity compared to the deletion allele alone (Supplementary Tables 2 and 5).

We performed a GO term analysis to determine if the 33 genes we identified are known to act in shared pathways or processes (see Materials and Methods). We identified significantly enriched Cellular Component GO Terms associated with the Swr1 complex (ARP6, SWC3, SWR1, TAF15, VPS71, and YAF9), prefoldin complex (GIM4, GIM5, PAC10, and YKE2), microtubules (CIN1, MAD2, NIP100, PAC2, TMA19, and TUB3), the bub1–bub3 complex (BUB1 and BUB3), and the mitotic checkpoint complex (BUB3 and MAD2; Fig. 3a). To determine if any of our identified genes encode proteins that work together in complexes, we used GeneMANIA to map physical interactions between these gene products (Fig. 3b; Montojo et al. 2010). This analysis identified components of the Swr1/Ino80 complex that replaces dimers of H2A–H2B histones for Htz1-H2B dimers, the GimC/prefoldin complex that folds nascent α- and β-tubulin monomers, the tubulin binding cofactors (TBCs) that assemble tubulin monomers into heterodimers, and the spindle assembly checkpoint complex that prevents anaphase onset in the presence of mitotic spindle errors (Rudner and Murray 1996; Lewis et al. 1997; Hansen et al. 1999; Tian et al. 1999; Mizuguchi et al. 2004; Vainberg et al. 1998; Fig. 3b). This analysis suggests that genes identified in our screen are likely involved in promoting α-tubulin expression through altering the chromatin landscape or impacting tubulin folding and heterodimer assembly.

Novel regulators of α-tubulin expression

We selected several genes from our screen for further investigation, based on their reported roles in RNA binding (NCL1) or chromatin regulation (SWC3 and VPS71; Fig. 3a). We also included GIM5 since it has a well-established role in α-tubulin protein folding (Vainberg et al. 1998; Lacefield and Solomon 2003). To confirm that loss of these genes diminishes α-tubulin expression, we tested whether adding low copy number plasmids expressing either TUB1 or TUB3 can suppress the benomyl sensitivity phenotype of the null mutants. Transformants were grown in media selective for the plasmid and then spotted to rich media or rich media containing benomyl. In control experiments, additional copies of TUB1 or TUB3 confer benomyl resistance to wild-type cells and rescue the benomyl sensitivity of tub3Δ mutants (Fig. 4a and b). Additional copies of TUB1 or TUB3 also rescue the benomyl sensitivity of the ncl1Δ, swc3Δ, and vps71Δ mutants but do not rescue the benomyl sensitivity of gim5Δ mutants (Fig. 4a and b). These results demonstrate that increasing the copy number of α-tubulin genes can rescue the sensitivity of ncl1Δ, swc3Δ, and vps71Δ mutants to microtubule stress.

To confirm that these genes specifically promote the function of the TUB1 intron, we used epistasis experiments to test whether the sensitivity of the null mutants to microtubule stress would be rescued by the tub1Δ∷TUB3 allele, but not tub1Δ∷TUB3^TUB1i. We find that the benomyl sensitivity of ncl1Δ mutants was not rescued by either allele; in fact, it is exacerbated (Fig. 4c). In contrast, the benomyl sensitivity of both swc3Δ and vps71Δ is rescued by tub1Δ::TUB3, but not by tub1Δ∷TUB3^TUB1i (Fig. 4c). We find similar results for growth at low temperature (15°C), which represents a different microtubule-destabilizing stress (Fig. 4c). These results are consistent with SWC3 and VPS71 operating in a pathway with the TUB1 intron and suggest that NCL1 operates in a separate pathway.

Finally, we asked whether these genes and the intron selectively promote the expression of the Tub1 protein. Delineating the relative protein levels of the Tub1 and Tub3 isotypes on a denaturing gel is a challenge, since Tub1 contains 447 amino acids and Tub3 contains 445 amino acids. To improve resolution, we inserted a 6xHis tag between codons 43 and 44 of TUB1, allowing us to clearly separate the 2 isotypes on a 10% acrylamide gel (Fig. 4d). Lysate from tub3Δ cells exhibits a single, slower-migrating band, confirming that the slower-migrating band corresponds to Tub1-6xHis (Fig. 4d). This method shows that in wild-type cells, 67% of α-tubulin is Tub1, which is consistent with previous data (Fig. 4d and e; Bode et al. 2003; Gartz Hanson et al. 2016). When the TUB1 intron is replaced with the ACT1 intron, the amount of Tub1-6xHis is reduced to 50% of total α-tubulin (P = 0.0002; Fig. 4d and e). Removal of the TUB1 intron had a weaker effect on the fraction of α-tubulin that is Tub1-6xHis (P = 0.079; Fig. 4d and e). This result suggests that the intron promotes the expression of Tub1 protein and may play a role in regulating the balance between the 2 α-tubulin isotypes.

With this method, we next asked if the genes identified in our screen maintain the ratio of Tub1-6xHis to Tub3 protein. ncl1Δ mutant cells show a small but significant reduction in the proportion of α-tubulin that is Tub1-6xHis, from 67 to 61% (P = 0.044; Fig. 4b and c). Neither swc3Δ nor vps71Δ mutants show a difference in the ratio of the α-tubulin isotypes, compared to wild type (P = 0.79 and P = 0.22; Fig. 4b and c). A null mutant in CDC40, which is known to disrupt the splicing of the TUB1 intron, also failed to show a significant change in the ratio of the α-tubulin isotypes. As expected, gim5Δ mutant cells show a decrease in total α-tubulin but no significant decrease in the ratio of the α-tubulin isotypes (P = 0.114; Fig. 4d and e). Together these results suggest that some of the genes we identified through our screen may promote the expression of Tub1-6xHis and not Tub3, while other genes may be less important for isotype balance.

Discussion

α-Tubulin genes commonly contain 5′ introns, but how these introns impact expression and ensure sufficient α-tubulin production has not been established. In this study, we find that the intron within the budding yeast α-tubulin TUB1 promotes α-tubulin expression and is important during microtubule stress. While we find that cells can survive without the intron, disrupting the intron sequence diminishes α-tubulin mRNA and protein levels and response to microtubule stress. Finally, we performed an unbiased screen for genes that could promote α-tubulin expression through the TUB1 intron. Our screen results also suggest a key role for α-tubulin introns in maintaining the balance between heterodimer production and turnover (Fig. 5).

Fig. 5.

Proposed model for how the TUB1 intron regulates α-tubulin expression. Diagram of α-tubulin biogenesis. Includes the genes we identified as acting through the TUB1 intron for promoting α-tubulin expression and where we would expect they act in the pathway.

The TUB1 intron appears to function differently from the ACT1 intron. The ACT1 intron is known to promote transcription (Furger et al. 2002; Moabbi et al. 2012; Agarwal and Ansari 2016), and simply inserting this intron into the coding sequence of another gene has been shown to boost gene expression in several cases (Agarwal and Ansari 2016). While the ACT1 intron does promote high levels of TUB1 pre-mRNA, it does not rescue the TUB1 intron function in our experiments and instead decreases the expression of TUB1 mature mRNA and protein (Figs. 1c–e, 2e and f, and 4d and e). We propose that the TUB1 intron works together in a locus-specific manner that depends on either the promoter or the coding sequence and promotes efficient splicing to create the mature TUB1 mRNA. Consistent with this hypothesis, we find that combining the TUB1 intron with the TUB3 coding sequence is more resistant to benomyl than full replacement by TUB3 but less resistant than wild-type TUB1. It remains to be determined how secondary structure elements in the TUB1 intron contribute to this regulation and how these structures may differ in TUB3. These results suggest synergy between the TUB1 promoter and the TUB1 intron to promote α-tubulin expression, perhaps by the intron recruiting chromatin regulators to enhance transcription activation, which has been observed for introns in other S. cerevisiae genes (Moabbi et al. 2012).

Our genetic screen may provide clues to distinguish between these proposed functions and reveal key regulators. In our screen, we identified putative regulators of TUB1 expression through the intron, using the criteria of hypersensitivity to microtubule stress (benomyl) and to loss of the alternative α-tubulin isotype (tub3Δ), and positive genetic interaction when combined with the TUB1 allele that lacks the intron (tub1Δi). Our list of genes encompasses both genes that show no additive sensitivity to microtubule stress and genes where benomyl sensitivity is rescued when combined with tub1Δi. While there are various, previously reported functions among this set of 34 genes, we did find several RNA binding proteins, components of chromatin remodeling complex, and known regulators of tubulin biogenesis and turnover (Fig. 3). SWC3, VPS71, and other members of the Swr1 complex are likely to promote TUB1 transcription either basally or in response to microtubule stress. Our data do not distinguish between these possibilities but do demonstrate clear specificity for the TUB1 intron, consistent with the intron acting synergistically with the TUB1 promoter (Figs. 4 and 5). CDC40, a spicing factor, is known to be required for efficient splicing of TUB1 pre-mRNA into mature mRNA (Fig. 5; Burns et al. 1999, 2002). We propose that NCL1, a tRNA methyltransferase, may impact the translation of nascent α-tubulin through either posttranscriptional modifications of tRNAs or through unidentified modification of the TUB1 mRNA itself (Fig. 5). Together, this network of genes acts with the TUB1 intron to promote α-tubulin function, which may be particularly important during microtubule stress.

Our screen also identified genes where ablating gene function appears to suppress the benomyl sensitivity caused by loss of the TUB1 intron (25/33 genes; Supplementary Table 2). Within this set are genes involved in tubulin biogenesis and turnover (Fig. 3b). Specifically, we identified 3 members of the Gim/prefoldin complex that help fold nascent α- and β-tubulin monomers (GIM4, PAC10, and YKE2) and 3 tubulin-binding cofactors (TBCs) that assemble and turn over α/β-tubulin heterodimers (CIN1, CIN4, and PAC2; Gao et al. 1993, 1994; Tian et al. 1996; Kortazar et al. 2007). In particular, the human homologs of Cin1 and Pac2 are important for driving heterodimer turnover (Tian et al. 1999). How might loss of tubulin biogenesis and turnover ameliorate the effect of diminished α-tubulin expression by tub1Δi? We speculate that this may be explained slowing the production of folded β-tubulin and/or delaying the turnover of a diminished α/β-heterodimer pool. Our previous work established that yeast cells must maintain higher levels of α-tubulin compared to β-tubulin to prevent toxic accumulation of the latter (Wethekam and Moore 2023). If diminished α-tubulin expression in tub1Δi cells creates an imbalance between α- and β-tubulin, then loss of β-tubulin folding by the Gim/prefoldin complex could help restore proper balance by decreasing β-tubulin levels. In addition, if diminished α-tubulin expression by tub1Δi creates a smaller pool of tubulin heterodimers, then turning down the rate of heterodimer turnover could increase the tubulin pool by extending the lifetime of tubulin proteins (Fig. 5). Our data point toward a model where tubulin production and heterodimer turnover must be properly balanced to meet the cell's demand for tubulin (Fig. 5). This model will require further testing, including a better understanding of the mechanism(s) of tubulin turnover.

Finally, our study demonstrates different levels of activity for the 5′ introns in TUB1 and TUB3, indicating that 5′ introns could be a point of regulation for creating blends of α-tubulin isotypes. We find that the tub1Δ∷TUB3^TUB1i allele that replaces the TUB3 intron with the intron from TUB1 improves resistance to microtubule stress (Fig. 2). This result suggests that while differences in amino acid sequence between Tub1 and Tub3 proteins may contribute to benomyl sensitivity, differences in intron sequence also contribute. In contrast, when we replace the TUB1 intron with the TUB3 intron, we see mild benomyl sensitivity and decreased levels of mature mRNA suggesting that the TUB3 intron does not fully replace the TUB1 intron (Fig. 2c). This level of phenotype is not different from that observed for cells lacking the TUB1 intron (Fig. 2c and f). Comparing the 2 intron sequences highlights several differences that could determine activity (Fig. 2a). The first difference is that the TUB3 intron is almost 3× as long as the TUB1 intron (300 bp for TUB3 vs 118 bp for TUB1) (Fig. 2a; Schatz et al. 1986). Longer introns have been correlated with increased mRNA and protein expression; for example, the ACT1 intron is 310 bp, and ACT1 is highly expressed across the cell cycle (Juneau et al. 2006; Blank et al. 2020). Despite its longer intron, we find that there is less Tub3 protein in cells compared to Tub1 protein (Fig. 4f; Bode et al. 2003; Juneau et al. 2006; Gartz Hanson et al. 2016). While intron length may influence function, it is not sufficient to determine expression levels in this case. The second key difference in the intron architecture is the branch point to 3′ splice site length (Fig. 2a). For both TUB1 and ACT1, the branch point to the 3′ splice site is 41 and 42 bp, respectively, and is consistent with the majority of introns in other yeast genes (Fig. 2a; Cellini et al. 1986). TUB3 on the other hand has a branch point to 3′ splice site length of 139 bp (Fig. 2a). Longer branch point to 3′ splice site lengths increase the likelihood for alternative 3′ splice sites and require secondary structures to ensure appropriate 3′ splice site selection and efficient splicing (Cellini et al. 1986; Gahura et al. 2011). Within the TUB3 intron, we identified several possible alternative 3′ splice sites, including 1 that is closer to the branch point (Supplementary Fig. 1). Splicing at these sites would lead to frameshifts and disrupt Tub3 translation; there is some evidence that an alternative 3′ splice site is used in TUB3 (Kawashima et al. 2014). The relevance of these sites or other potential sites in regulating the production of Tub3 or total α-tubulin is still unknown. Our findings establish 5′ introns as important regulators α-tubulin gene function, and a better understanding of their molecular functions may shed new light on conserved mechanisms that balance tubulin isotype expression.

Data availability

All yeast strains and plasmids constructed for this study are listed in Supplementary Tables 3 and 4, respectively. These are available upon request. SGA score data from the screen are included in Supplementary Files 1 and 2.

Supplemental material available at GENETICS online.

Funding

This work was supported by National Institutes of Health R35 GM 136253 to JKM. LCW was supported by T32 GM136444 and the University of Colorado Molecular Biology Training Program Bolie Scholar Award.