In budding yeast, dynein activity is both spatially and temporally regulated to achieve correct spindle positioning within the dividing cell. Before mitosis, dynein is asymmetrically loaded onto the daughter-bound SPB and its associated aMTs (Grava et al., 2006 ; Segal and Bloom, 2001 ). This localization ensures unidirectional spindle movement toward the bud and is governed by Cdk1, spindle pole components, and bud neck kinases (Grava et al., 2006 ). Furthermore, analysis of dynein mutants showed that (1) dynein is necessary for establishing proper spindle position during, but not before, anaphase; and (2) cortical MT-sliding events mediated by dynein motor activity occur only during anaphase (Adames and Cooper, 2000 ). These data indicate that dynein activity is strictly limited to anaphase.

To test directly whether She1p plays a role in orienting the spindle, we monitored pre-anaphase spindle movements in wild-type and she1Δ mutant cells expressing GFP-Tub1 (α tubulin) fusion protein. Spindle movement was quantified by measuring the distance between the daughter-bound spindle pole body (dSPB) and the bud neck over time (see Figure 1I for schematic). In wild-type cells, spindles stayed relatively fixed near the bud neck in the mother cell and were oriented perpendicular to the plane of division (Figure 1, C and D, and Supplemental Movie S1). However, in she1Δ cells, spindles exhibited dramatic movements. These motile spindles were lead by long aMTs that appeared to glide around the cell cortex, reminiscent of dynein-directed cortical aMT sliding normally seen during anaphase. Additionally, ≈29% of these spindles also traveled back and forth between the mother and daughter cells (Figure 1, C and E, and Supplemental Movie S2). To test whether dynein activity was responsible for these dramatic spindle movements, we monitored spindle position in cells lacking both She1p and the dynein heavy chain subunit (Dyn1). Indeed, pre-anaphase spindles in the she1Δ dyn1Δ mutant resembled wild-type spindles, lacking any dramatic movement (Figure 1F). We also noticed that she1Δ pre-anaphase spindles were longer than wild-type pre-anaphase spindles. Depletion of dynein in she1Δ cells restored pre-anaphase spindle length to normal (she1Δ : 1.75 ± 0.4 μm, n = 32; wild-type: 1.34 ± 0.3 μm, n = 19; she1Δ dyn1Δ : 1.28 ± 0.3 μm, n = 15), suggesting that ectopic dynein activity can “stretch” the spindle. Because dynein ordinarily is inactive before anaphase in budding yeast, these results suggest that She1p represses dynein activity until anaphase.

Premature spindle migration between the mother and the daughter cell has been reported in kar9Δ cells, raising the possibility that She1p could be an activator of the Kar9 pathway (Yeh et al., 2000 ). We found that spindle movements in kar9Δ cells (Figure 1G) were not as dramatic as those in she1Δ cells (Figure 1E), and that addition of the she1Δ mutation enhanced the spindle movements seen in kar9Δ cells (Figure 1H; mean maximum displacement = 2.93 ± 1 μm [she1Δ, n = 7], 1.66 ± 0.8 μm [kar9Δ, n = 8], 3.63 ± 0.6 μm [she1Δ kar9Δ, n = 7]). Also, we observed no synthetic interactions between the she1Δ and dyn1Δ alleles (unpublished data). Traditionally, null mutations in all known Kar9p pathway components show synthetic lethal/sick interactions with the dyn1Δ mutation (Lee et al., 2003 ; Grava et al., 2006 ). These results support the notion that She1p is not an activator of the Kar9p pathway but rather a repressor of the dynein pathway.

The results above indicate that She1p inhibits dynein activity before anaphase. We next asked whether She1p also negatively regulates dynein after anaphase. Once the cell completes anaphase, it must suppress dynein activity to permit proper spindle orientation in the next cell cycle. The absence of aMT-sliding on the cell cortex from late anaphase onward suggests that dynein suppression occurs before the next cell cycle begins (Adames and Cooper, 2000 ). If She1p is necessary to inactivate dynein during late anaphase, then cortical aMT sliding events may be observed during that time in she1Δ cells. To test this hypothesis, we monitored GFP-Tub1–expressing cells undergoing anaphase. In all wild-type cells, the spindle entered the bud, elongated until the spindle poles reached the ends of the dividing cell, and summarily disassembled (n = 7; Figure 2A and Supplemental Movie S3). In no case did cortical MT sliding occur once the spindle poles reached the cortex. In contrast, in 16 of 19 she1Δ cells, the spindle elongated properly but was then pulled around the cortex, likely by aMTs sliding along the cortex. These forces consequently bent the spindle to create a distinctive curled morphology (Figure 2B and Supplemental Movie S4). Although the presence of aMT-sliding suggests that spindle curling is caused by dynein activity, it is still possible that spindle curling is a side effect of spindle overextension. However, in all she1Δ dyn1Δ cells observed, late anaphase spindles remained straight throughout extension, implicating ectopic dynein activity as the cause of spindle curling in she1Δ cells (n = 7; Figure 2C). These results, combined with the observation that kar9Δ cells do not display spindle curling (unpublished data), further support the conclusion that She1p is an inhibitor of dynein and not an activator of the Kar9p pathway.

Occasionally, during spindle curling in she1Δ cells, an aMT pulled one spindle pole far enough to penetrate the other cell, resulting in the formation of one cell with two spindle poles and one without any after cytokinesis (2 of 16 events; Supplemental Figure S1A and Supplemental Movie S5). Because chromosomes stay very closely attached to the spindle pole throughout the entire cell cycle in yeast, we suspected that spindle curling could cause unequal distribution of chromosomes. We tested this possibility by arresting haploid yeast in G1 and visualizing fluorescently marked chromosome III (Chr III). Wild-type cells possessed only one Chr III (99% with one GFP “dot”), whereas she1Δ cells frequently possessed two Chr IIIs (18% with two GFP dots; Supplemental Figure S1B). Further analysis revealed that she1Δ cells correctly segregated Chr III in early anaphase, indicating that the unequal distribution of Chr III seen in G1 cells was not a result of chromosome nondisjunction (unpublished data). These data suggest that She1p is required to inactivate dynein at the end of anaphase to ensure equal distribution of spindle poles and their associated chromosomes between the mother and daughter cell.

We next addressed how She1p affects dynein activity. The prevailing model for dynein function proposes that dynein is loaded onto aMTs, targeted to the plus ends, and then off-loaded to the cortex. Once anchored at the cortex, dynein uses its minus-end-directed motor activity to pull the attached aMT and the connected spindle toward the site of dynein anchorage (Lee et al., 2003 ; Li et al., 2005 ). Because She1p localizes along MTs, we first asked if She1p affects dynein activity indirectly by modifying aMT dynamics. However, there was no significant difference in aMT length, growth rate, and shrinkage rate between she1Δ and wild-type cells (Supplemental Table S1). Next, we tested whether She1p interferes with dynein loading onto aMTs or recruitment to aMT plus ends. Dynein is found on aMTs throughout the cell cycle, and, although its localization to aMT plus ends does increase as the cell enters mitosis, at least 50% of all cells retain dynein at aMT plus ends regardless of cell cycle stage (Sheeman et al., 2003 ). Therefore, it is unlikely that loading of dynein onto aMTs or its recruitment to the plus end contributes to the cell cycle–dependent regulation of its activity. Nevertheless, we studied the localization of dynein in wild-type and she1Δ cells using a 3GFP-tagged version of the dynein heavy chain (Dyn1-3GFP). Consistent with previous reports (Lee et al., 2003 ; Grava et al., 2006 ), we observed Dyn1-3GFP at SPBs, aMT plus ends, and the cell cortex in wild-type cells. This localization was unchanged in she1Δ cells (Supplemental Figure S2A). We also found that the percentage of cells with dynein localized to aMT plus ends increased slightly as the cells entered anaphase. Only a small, statistically insignificant change in the amount of plus end-localized dynein was observed in she1Δ cells (G1: 67.9 ± 8.9% [wt] vs. 85.2 ± 5.2% [she1Δ] P = 0.14; pre-anaphase: 84.5 ± 6.3% versus 87.3 ± 0.4% P = 0.59; anaphase: 90.3 ± 2% versus 92.3 ± 10.9% P = 0.82; Supplemental Figure S2B). Loss of She1p induces premature dynein activity without affecting dynein localization in pre-anaphase cells, suggesting that She1p inhibits dynein activity by a mechanism other than restricting its loading onto aMTs or recruitment to aMT plus ends.

Another possibility is that She1p regulates a known enhancer of dynein motor function. One such candidate is the multi-subunit dynactin complex, which is essential for dynein activity but dispensable for dynein recruitment to aMTs in yeast (Schroer, 2004 ; Sheeman et al., 2003 ; Moore et al., 2008 ). We tested whether She1p regulates dynactin function by monitoring the localization of four prominent dynactin subunits: the p150^glued ortholog Nip100p, the actin-related protein Arp1p, the dynamitin ortholog Jnm1p, and the p24 ortholog Ldb18p. For visualization, we tagged the endogenous copies of each protein with 3GFP at the C terminus. Haploid cells expressing the Jnm1-3GFP, Ldb18-3GFP, and Nip100-3GFP fusions displayed normal spindle positioning and were viable when Kar9p was depleted, indicating that the fusion proteins were functional. However, the Arp1-3GFP fusion was partially functional (see Materials and Methods). In addition, we coexpressed each 3GFP fusion protein with the microtubule marker mCherry-Tub1. We noticed that reducing the ratio of modified tubulin to wild-type tubulin enhanced dynactin localization to concentrated foci. Hence, we studied dynactin localization in homozygous TUB1/TUB1 diploid cells with only one copy of the mCherry-TUB1 allele integrated at the URA3 locus (see Materials and Methods). Consistent with previous reports (Moore et al., 2008 ), in wild-type cells, all four dynactin subunits were found at aMT plus ends, near SPBs, and the cell cortex (Figures 3 and ​and44).

We found that the localization of Nip100-3GFP and Arp1-3GFP varied dramatically with the cell cycle. Nip100-3GFP and Arp1-3GFP were largely absent from SPBs or aMTs until anaphase: both proteins were found on SPBs or aMTs in ≈25% of G1 and pre-anaphase cells and in ≈80% of anaphase cells, representing an ≈3.2-fold increase (Figure 3, A, C, E, and F). When they did appear, Nip100-3GFP and Arp1-3GFP predominately localized to plus ends (unpublished data; Moore et al., 2008 ). However, in >80% of she1Δ cells, Nip100-3GFP and Arp1-3GFP appeared on aMTs regardless of cell cycle stage (Figure 3, B, D, E, and F), suggesting that She1p governs the cell cycle–dependent recruitment of Nip100p and Arp1p to aMTs. We also considered the possibility that the cell cycle–dependent appearance and disappearance of Nip100p and Arp1p is regulated by changes in protein degradation or expression. However, we found that the presence of She1p had no detectable impact on Nip100p or Arp1p protein levels in asynchronous, G1, and pre-anaphase cells (Supplemental Figure S3A and unpublished data). Yet, the ratio of Nip100-3GFP and Arp1-3GFP fluorescence detected in MT-associated foci versus the cytoplasm was ≈50% higher in anaphase she1Δ cells than in anaphase wild-type cells (Figure 3G), supporting the idea that more Nip100p and Arp1p are recruited to aMTs from the cytoplasmic pool in the absence of She1p. We then asked whether dynactin recruitment to aMTs occurs via direct binding to the microtubule or indirectly through dynein. We observed that Nip100-3GFP localization to SPBs and aMTs was almost entirely eliminated in both dyn1Δ and dyn1Δ she1Δ cells, suggesting that dynactin is recruited to aMTs through its interaction with dynein (Figure 3H). This result is consistent with the report that a mutant version of p150^glued lacking its MT-binding domain still localizes to MTs in Drosophila melanogaster S2 cells (Kim et al., 2007 ). Overall, these data suggest that She1p precludes the interaction of dynein and dynactin before anaphase.

Interestingly, the behavior of Jnm1p and Ldb18p differed from the behavior of the other dynactin subunits. Unlike Nip100-3GFP and Arp1-3GFP, Jnm1-3GFP and Ldb18-3GFP almost always localized in foci associated with SPBs or aMTs throughout the cell cycle in wild-type cells (Figure 4, A, C, E, and F). In G1 and pre-anaphase cells, the majority of Jnm1-3GFP and Ldb18-3GFP foci localized at or very near SPBs (Figure 4, A and C, and unpublished data). However, similar to Arp1-3GFP and Nip100-3GFP, the frequency of Jnm1-3GFP and Ldb18-3GFP on aMTs increased ≈3.4-fold as cells entered anaphase (Figure 4, A, C, G, and H). Again, loss of She1p eliminated the dramatic cell cycle-dependent change in Jnm1-3GFP and Ldb18-3GFP localization and enhanced the presence of both subunits on aMTs (Figure 4, B, D, G, and H). These results suggest that She1p inhibits Jnm1p and Ldb18p recruitment to aMTs until anaphase.

The difference in localization patterns of Jnm1p, Ldb18p, Arp1p, and Nip100p suggests that the dynactin complex could exist in subcomplexes. Indeed, only ≈60% of Jnm1-tdTomato foci colocalized with Nip100-3GFP foci in asynchronous wild-type cells (n = 114; Figure 5A). The presence of dynactin subcomplexes is further supported by previous cosedimentation and coimmunopurification experiments that identified large pools of Jnm1p and Ldb18p unassociated with Nip100p and Arp1p (Moore et al., 2008 ; Amaro et al., 2008 ). Given these findings and our observations that Jnm1p and Ldb18p localize to SPBs throughout the cell cycle but appear on aMTs only during anaphase, it is possible that the complete dynactin complex is assembled during anaphase from two spatially segregated subcomplexes: (1) an SPB-localized subcomplex that contains Jnm1p and Ldb18p, and (2) a cytoplasmic subcomplex that contains Nip100p and Arp1p. However, roughly equal amounts of Jnm1-3HA coimmunoprecipitated with Nip100-3GFP in G1, pre-anaphase, and asynchronous wild-type cells, suggesting that a complete version of the dynactin complex exists throughout the cell cycle (Figure 5B). In addition, time-lapse microscopy following a cell entering anaphase showed that Jnm1p suddenly appeared at the aMT plus end, rather than being transported there from the SPB (Figure 5C). Furthermore, after photobleaching SPB-localized Jnm1-3GFP signal, we observed an increase in GFP fluorescence on the aMT but no recovery at the SPB (Figure 5D), ruling out the possibility that aMT-localized Jnm1-3GFP foci result from transport and concentration of undetectable amounts of SPB-localized Jnm1-3GFP. The fact that we never observed recovery of fluorescence at the SPB indicates that the Jnm1p localized there comprises a static structure (Figure 5E). Therefore, the Jnm1p that appeared on the aMT was recruited from the cytoplasm, most likely in complex with the other dynactin subunits. In total, these results suggest that the dynactin complex comprises complete and incomplete varieties, and that She1p specifically hinders recruitment of the complete version to aMTs until anaphase.