|Home | About | Journals | Submit | Contact Us | Français|
Microtubules are polymers composed of α-β tubulin heterodimers that form 12 to 15 protofilaments and assemble into microtubules . Microtubules are dynamic structures that have periods of both growth and shrinkage by addition and removal of subunits from the polymer . Microtubules stochastically switch between periods of growth and shrinkage, termed dynamic instability . Dynamic instability is coupled to the GTPase activity of the β-tubulin subunit of the tubulin heterodimer . Microtubule dynamics are regulated by microtubule associated proteins (MAPs), that interact with microtubules to regulate dynamic instability . MAPs in budding yeast have been identified that bind microtubule ends (Bim1), that stabilize microtubule structures (Stu2), that bundle microtubules by forming cross-bridges (Ase1), and that interact with microtubules at the kinetochore (Cin8, Kar3, Kip3) [6-10]. IRC15 was previously identified in four different genetic screens for mutants affecting chromosome transmission or repair [11-14]. Here we present evidence that Irc15 is a microtubule associated protein, localizing to microtubules in vivo and binding to purified microtubules in vitro. Irc15 regulates microtubule dynamics in vivo and loss of IRC15 function leads to delayed mitotic progression, due to failure to establish tension between sister kinetochores.
IRC15 (conserved in closely related senso stricto yeast) has strong sequence similarity to LPD1 in S. cerevisiae. LPD1 encodes a dihydrolipoamide dehydrogenase, a protein involved in glycolysis . IRC15 lacks key catalytic residues (C45Y and C50A) that Lpd1 requires for enzymatic function and therefore Irc15 is not a redundant dihydrolipoamide dehydrogenase . LPD1 has sequence similarity to a large class of conserved enzymes (pyridine nucleotide disulfide oxidoreductases) that includes glutathione reductase, mercuric reductase, thioredoxin reductase and trypanothion reductase . It is possible that IRC15-like function has been conserved in humans within this large family of Lpd1-related proteins.
We analyzed the distribution of cells in the cell cycle to determine if a loss of Irc15 function alters cell cycle progression. irc15Δ and BY4741 (wild type) cells were grown asynchronously and budding morphology was scored. 52% of irc15Δ cells were large budded (Figure 1A) compared to 35% in wild type, suggesting an increased proportion of cells in mitosis. 44% of irc15Δ cells were large budded with undivided nuclei compared to 23% in wild type cells (Figure 1B). These data suggest that irc15Δ cells delay mitosis prior to anaphase. We analyzed cell cycle progression in synchronous cells by flow cytometry, Pds1 (securin) stability and spindle elongation. Cells were arrested with α-factor, released into the cell cycle and α-factor was re-added after cells budded to limit the analysis to a single cell cycle. Cell cycle progression was delayed 15 minutes in irc15Δ cells as determined from the proportion of cells with a 2C DNA content compared to wild type cells (Figure 1C and D). Pds1 levels persisted 15 minutes longer in irc15Δ cells suggesting a pre-anaphase delay. We determined the timing of anaphase in a separate experiment using synchronous cells expressing TUB1-GFP. Wild types cells elongated their spindles at approximately 72 minutes and irc15Δ cells had elongated spindles at approximately 90 minutes confirming a pre-anaphase mitotic delay in irc15Δ cells (Figure 1E).
We determined the localization of Irc15-GFP by anti-GFP immunofluorescence and observed Irc15-GFP co-localized with microtubules at all phases of the cell cycle (Figure 2A). We tested the ability of Irc15 to bind microtubules in vitro using whole cell extracts from cells expressing epitope tagged Irc15 and incubating the extract with taxol stabilized microtubules. Irc15 was recovered in the pellet fraction indicating a physical interaction with microtubules (Figure 2B). Depolymerizing microtubules, by adding nocodazole, shifted tubulin and Irc15 to the supernatant. To determine if Irc15 interacted with microtubules directly, we purified bacterially expressed recombinant Irc15 and tested its ability to directly bind taxol stabilized microtubules in vitro. Recombinant Irc15 bound to taxol stabilized microtubules demonstrating a direct interaction (Figure 2C).
We determined the hydrodynamic properties of Irc15 to determine if it was a homodimer like dihydrolipoamide dehydrogenase . We fractionated protein extracts from cells expressing epitope tagged Irc15 using size exclusion chromatography and velocity sedimentation (Figure 2D and E). The Stokes’ radius was 42.5Å and the Svedberg coefficient was 7.2S. We calculate a native molecular weight of 125 kDa for Irc15. Conceptual translation of the IRC15 DNA sequence predicts a protein molecular weight of 54.1 kDa, suggesting that it exists as a homodimer in vivo. Purified recombinant Irc15 also behaved hydrodynamically as a homodimer (data not shown). These results suggest that Irc15 can bind microtubules as a homodimer.
We assayed microtubule behavior in irc15Δ cells to determine if Irc15 regulates microtubule dynamics. Dynamics of cytoplasmic microtubules were measured in cells expressing TUB1-GFP (online supplemental movies 1 and 2). Images from live cell microscopy (Figure 3A) are shown for wild type (top) and irc15Δ (bottom) cells. Individual microtubules (arrows) are dynamic in wild type cells but are more stable in irc15Δ cells. Rates of microtubule growth and shrinkage, as well as the number of rescue and catastrophe events were determined from plots of microtubule lengths at each time point from the live cell movies (Figure 3B). The growth and shrinkage rates of microtubules in irc15Δ cells were slower and the frequencies of rescue and catastrophe were decreased compared to the wild type cells. The wild type cells displayed a more dynamic range of lengths over the time course. Overall, irc15Δ cells had a 2.4-fold decrease in the rate of microtubule growth and shrinkage as well as a 2.4-fold decrease in rescue and catastrophe events (Figure 3C). Microtubules from irc15Δ cells also had a 5.2-fold increase in the time they spent paused. Overall, irc15Δ cells had a global reduction in microtubule dynamics in vivo.
This phenotype is reminiscent of the tubulin mutants tub2-V169A and tub2-C354A. These mutants have predicted defects in GTP hydrolysis and dimer-dimer interactions respectively, and both globally reduce microtubule dynamics in vivo . Both tub2-V169A and tub2-C354A cells have similar decreases in microtubule dynamics to irc15Δ cells, yet are predicted to affect the microtubule polymer in different ways. We crossed irc15Δ to tub2-V169A and tub2-C354A mutants and isolated double mutants. irc15Δ tub2-C354A double mutants are inviable but irc15Δ tub2-V169A double mutants are not. This result suggests that IRC15 lies in the same pathway as the tub2-V169A mutant, and therefore may have a role in the GTP to GDP cycle on the tubulin dimer.
We used recombinant Irc15 to examine the effect on microtubule dynamics in vitro. GMPCPP microtubule seeds (used as nucleation sites for microtubule assembly) and purified bovine tubulin were incubated with Irc15 and microtubule counts and lengths were measured every 2 minutes for 12 minutes. Reactions incubated with Irc15 had longer microtubules at each time point (Supplemental Figure 1A). The distribution of microtubule lengths in Irc15-containing reactions favored a greater proportion of longer microtubules than seen in the control. The distribution of lengths were different at each of the time points (students t-test p<0.0001 for 4-12 minutes and p<0.02 for 2 minutes using Mann-Whitney test due to small sample size). Irc15 also stimulated microtubule nucleation as there were more microtubules per field than in untreated samples (Supplemental Figure 1B). Therefore, Irc15 promotes nucleation and growth of microtubules at early phases of microtubule assembly in vitro.
Dynamic microtubules promote the attachment of microtubules to kinetochores to establish chromosome bi-orientation on the spindle . We tested irc15Δ cells and did not see significant defects in a variety of assays including benomyl sensitivity, mating, nuclear positioning and binucleate formation (data not shown). We visualized kinetochores labeled with Ndc80-GFP and spindle pole bodies labeled with Spc42-DsRed to examine kinetochore-microtubule attachments in irc15Δ cells. When kinetochores are attached to the spindle in a bipolar fashion there is tension between sister kinetochores, producing two GFP lobes . Cells were synchronized in mitosis with cdc13-1, which activates the DNA damage checkpoint and arrests cells in metaphase . In wild type cells 90% of kinetochores (n=36) had a distinctive bi-lobed appearance (Figure 4A panel I), indicating that kinetochores were attached in a bipolar fashion to the spindle and tension was being generated between sister kinetochore pairs. In irc15Δ cells we observed a bi-lobed kinetochore appearance in 32% of cells (n=50). We also observed cells with a signal GFP lobe (Figure 4A panel III) in 38% of cells and 30% of cells with disorganized kinetochore localization (Figure 5B), defined here as cells with more than two GFP foci (Figure 4A panels II and IV). The spindle length (pole to pole distance) in irc15Δ cells was also decreased in comparison to wild type cells (2.03+/−0.6μm to 2.9+/−0.8μm, respectively). These data suggest that irc15Δ cells do not effectively establish bipolar spindle microtubule-kinetochore attachments, or that they lack the ability to generate sufficient tension between sister kinetochore pairs to achieve separation.
We labeled the centromere of Chromosome IV with GFP to examine attachment of a single kinetochore to the spindle . Cells labeled at a single kinetochore have one GFP spot prior to DNA replication, but after replication and bipolar attachment the sister centromeres are separated and two GFP spots are visible. The separation of sister centromeres indicates bipolar attachment of microtubules to CEN4. Live cell imaging of GFP labeled CEN4 in mitotic cells was performed over ten minute intervals (online supplemental moves 3 and 4). Bipolar attachment generated tension and sister centromeres were separated in wild type cells (Figure 5A panels I and II). In contrast, sister centromeres remained as a single focus of GFP in the majority of irc15Δ cells (Figure 5A panels III and IV), indicating a lack of bipolar attachment or a lack of tension. In wild type cells, 98% of centromere pairs (n=42) were two distinct foci (Figure 5B), compared to 27% in irc15Δ cells (n=63). Over the 600 second time course, 29 of 63 centromere pairs in irc15Δ cells remained as a single GFP focus for the entire period. A representative sister centromere pair (Figure 5C) from wild type cells spend more time separated and the distance between sister centromeres (0.77+/−0.33μm) oscillates over time. In contrast, centromeres in irc15Δ cells spend less time separated and the distance they separate (0.11±0.15μm) is shorter than observed in wild type cells.
Kinetochores in irc15Δ cells either lack bipolar attachment or lack tension between sister chromatids. To distinguish between these alternatives we determined the movement of individual centromeres relative to the spindle axis and poles. When kinetochores are attached to microtubules, they remain near the spindle axis. When kinetochores establish bipolar attachments they are separated and positioned approximately half way between the spindle pole and spindle mid-zone . When kinetochores are mono-oriented (syntelic or monotelic attachments) they are unseparated and localized adjacent to the spindle pole. A representative tracing showing the behavior of one of the two centromeres in a wild type cell and the unseparated centromeres in a irc15Δ cell is shown in Figure 5D. The separated centromeres in wild type cells and the unseparated centromeres in irc15Δ cells behaved similarly and remained mostly on the spindle axis or within 0.4μm when off the axis suggesting that kinetochores in irc15Δ cells are bound to microtubules. As predicted, the centromeres in wild type cells had an average relative spindle position of 0.37 ± 0.12 (n=5). The centromeres in irc15Δ cells had an average relative position of 0.46 ± 0.09 (n=5), remaining near the spindle mid-zone, suggesting that kinetochores are attached to the spindle in a bipolar fashion but are unable to create enough tension to separate sister chromatids.
Most centromere pairs observed in irc15Δ cells did not separate, this lack of separation of centromere pairs explains the appearance of single lobes of Ndc80-GFP seen in irc15Δ cells (Figure 4A). Individual centromeres in irc15Δ cells behaved differently from centromeres in mutants, such as ndc80-1, that do not attach kinetochores to microtubules. Centromeres in ndc80-1 cells move randomly in the nucleus averaging between 1-1.5μm off the spindle axis . Centromeres in irc15Δ cells also behaved differently from centromeres in mutants, such as ipl1 that do not establish bipolar orientation of chromatids on the spindle. Unseparated centromeres of ipl1 cells remain on the spindle axis but very close to one pole . Centromeres in irc15Δ cells remain on the spindle axis and are broadly distributed between the poles distinguishing irc15Δ cells from ipl1 cells and suggesting that centromeres in irc15Δ cells are attached to the spindle in a bipolar fashion. The absence of sister separation seen in irc15Δ cells is therefore a defect in generating tension between sister chromatids.
In order to test if decreased microtubule dynamics and defective kinetochore-microtubule attachments were affecting genomic stability in irc15Δ cells, we assayed cells for chromosome transmission fidelity using a nonessential yeast chromosome fragment (CFIII) . irc15Δ cells had an 377-fold greater CFIII loss rate (6.75 × 10−3) compared to wild type cells (1.78 × 10−5).
The ORF of IRC15 overlaps 26 base pairs CTF19 which encodes a kinetochore protein and irc15Δ truncates Ctf19 by 8 amino acids. The phenotypes we observed could be due to truncation of CTF19 or IRC15 loss of function. To determine which mutation was responsible for the phenotypes we observed, we transformed irc15Δ cells with plasmids containing CTF19 and IRC15. CTF19 failed to complement CFIII loss, but IRC15 (pBK2) complemented the phenotype (Supplemental Figure 2C). pBK2 also complemented the budding morphology and the kinetochore phenotype (Supplemental Figure 2A and B). The phenotypes described here are due to a loss of IRC15 function and not due to the truncation of CTF19.
Irc15 is a novel microtubule associated protein that directly interacts with microtubules and regulates their dynamics. In addition, Irc15 plays an important role in generating tension between sister chromatids and in chromosome segregation. irc15Δ cells are viable and have only a slight reduction in growth rate (not shown). The relatively mild effect on growth belies the dramatic affect on microtubule dynamics and chromosome segregation. This suggests that non-essential genes may make large contributions to essential processes and more detailed analysis of the yeast collection of mutations in non-essential genes is warranted to fully understand processes such as mitosis.
We thank Tim Sterns and Phil Heiter for pTS990 and pKH7 plasmids. We also thank Arshad Desai for advice on microtubule assembly assays. Special thanks to Michael Emanuele and Todd Stukenberg for expertise with biochemistry and Anne Knowlton for helpful comments on the manuscript.
Experimental Procedures can be found in the Supplemental Data.