A cold-sensitive γ-tubulin allele of Aspergillus nidulans, mipAD159, causes defects in mitotic and cell cycle regulation at restrictive temperatures that are apparently independent of microtubule nucleation defects. Time-lapse microscopy of fluorescently tagged mitotic regulatory proteins reveals that cyclin B, cyclin-dependent kinase 1, and the Ancdc14 phosphatase fail to accumulate in a subset of nuclei at restrictive temperatures. These nuclei are permanently removed from the cell cycle, whereas other nuclei, in the same multinucleate cell, cycle normally, accumulating and degrading these proteins. After each mitosis, additional daughter nuclei fail to accumulate these proteins, resulting in an increase in noncycling nuclei over time and consequent inhibition of growth. Extensive analyses reveal that these noncycling nuclei result from a nuclear autonomous, microtubule-independent failure of inactivation of the anaphase-promoting complex/cyclosome. Thus, γ-tubulin functions to regulate this key mitotic and cell cycle regulatory complex.
γ-Tubulin is a component of microtubule-organizing centers and is conserved in all eukaryotes. Its role as a nucleator of spindle and cytoplasmic microtubules is well established (for reviews see Wiese and Zheng, 1999; Oakley, 2000; Job et al., 2003), but evidence has emerged from several sources that γ-tubulin and γ-tubulin ring complex proteins have microtubule-independent roles in mitotic regulation (for reviews see Lange, 2002; Cuschieri et al., 2007). Our laboratory has generated a collection of mutant alleles of mipA, the Aspergillus nidulans γ-tubulin gene (Jung et al., 2001). One cold-sensitive allele, mipAD159, disrupts the coordination of late mitotic events such as chromosomal disjunction and mitotic exit at restrictive temperatures (Prigozhina et al., 2004). We reasoned that examining the localization of mitotic regulatory proteins in strains carrying mipAD159 would likely reveal the nature of mitotic misregulation caused by this allele and, thus, clarify the functions of γ-tubulin in mitotic regulation.
Using recent advances in gene targeting techniques in A. nidulans (Yang et al., 2004; Nayak et al., 2006; Szewczyk et al., 2006), we fused GFP, monomeric RFP (mRFP), or mCherry to the C terminus of several mitotic regulatory proteins and studied their localization in strains carrying the wild-type mipA allele (mipA+) and strains carrying mipAD159. Although the localization of some proteins was not altered in an informative way, three proteins, cyclin B, Cdk1, and Ancdc14, showed a dramatic alteration of localization. A. nidulans is coenocytic, and nuclei within a cell go through the cell cycle synchronously, with mitotic regulatory proteins such as cyclin B accumulating and being destroyed at the same time in all nuclei. In strains carrying mipAD159, at restrictive temperatures, cyclin B, Cdk1, and Ancdc14 failed to accumulate in a subset of the nuclei within a cell. Time-lapse imaging revealed that these nuclei were removed from the cell cycle and that the fraction of noncycling nuclei increased with each cell cycle. Extensive experimental analyses indicate that these noncycling nuclei result from mipAD159 causing a nuclear autonomous failure of inactivation of the anaphase-promoting complex/cyclosome (APC/C) in the G1 phase of the cell cycle.
Creation of fluorescent fusion proteins
Unless otherwise mentioned, all imaging was performed with proteins tagged at their C termini by using fusion PCR to create transforming DNA fragments in which the 3′ end of the target gene was fused in frame to a GFP, mCherry, or mRFP coding sequence. The fusion genes were the only copies in the genome. They were under the control of their endogenous promoters, and the fusion proteins appeared to be fully functional, supporting normal growth over a wide range of temperatures (from 20 to 42°C) and causing no apparent mitotic or cell cycle defects. Genotypes of strains are shown in Table S1.
In vivo localization of cyclin B and Cdk1 through the cell cycle
In A. nidulans, only hyphal tip cells are mitotically active, and all observations and experiments involving the cell cycle were performed with tip cells. The localization patterns of cyclin B and Cdk1 were very similar (as expected because Cdk1 is localized to the nucleus through its binding to cyclin B; Nurse, 1990), and we will discuss them together. Cyclin B is the only essential cyclin in A. nidulans, and it is encoded by the nimE gene (Morris, 1975; O’Connell et al., 1992; Osmani et al., 1994). In A. nidulans, there is a single Cdk1 gene, nimX, and it is required for progression through G1 and G2 (Osmani et al., 1994).
We performed time-lapse imaging of strains LO1438 and LO3317, expressing cyclin B–GFP and histone H1–mRFP to allow visualization of chromatin (Nayak et al., 2006). The timing of cell cycle stages in A. nidulans has been determined with precision (Bergen and Morris, 1983), and chromosomes visibly condense in mitosis, so it is possible to correlate our observations with cell cycle stages with reasonable confidence. Cyclin B–GFP was not visible in early G1 nuclei but became visible as nuclei progressed through interphase (Fig. 1 A). In 346 randomly selected interphase tip cells, 61.0% of cells (weighted mean of four experiments; weighted SD = 3.4%) had visible cyclin B–GFP in their nuclei. Thus, cyclin B becomes visible ∼40% of the way through interphase, and this point corresponds to S phase (Bergen and Morris, 1983). Because cyclin B must accumulate to some extent before we can detect it, the point at which cyclin B begins to accumulate in nuclei is earlier, logically at the G1/S boundary because cyclin B and Cdk1 are required for DNA synthesis in A. nidulans (Osmani et al., 1994). Cyclin B–GFP fluorescence was visible in the nucleoplasm and, more intensely, at spindle pole bodies (SPBs). The signal increased as the cell cycle proceeded (Fig. 1 A), which is consistent with previous observations (Wu et al., 1998; De Souza et al., 2009).
The localization pattern of cyclin B in mitosis in living cells has been described recently (De Souza et al., 2009), and our observations are largely consistent with these data. At mitotic entry, when the nuclear pore complex (NPC) partially disassembles (De Souza et al., 2004), much of the cyclin B rapidly exited the nuclei, but a fraction remained in the nucleoplasm and at the SPB (Fig. 1 B). As mitosis proceeded (Fig. 1 B), SPBs separated, and cyclin B–GFP was concentrated at the SPBs and between the SPBs (i.e., on the spindle) but was visible at a lower intensity throughout the nucleus. The concentration of cyclin B at the SPB and on the mitotic spindle was reduced to the levels in the general nucleoplasm by 30–90 s before anaphase onset (data from 21 nuclei observed by dual-wavelength time-lapse imaging with z-series stacks [0.4-µm step size] collected at 30-s intervals at 25 ± 1°C). The remaining cyclin B disappeared gradually. In 9/21 cases, it was undetectable by the end of anaphase A, in 9/21 cases, it persisted into anaphase B/telophase, and in 3/21 cases, it was present in anaphase A but gone by the earliest anaphase B time point 30 s later.
A. nidulans cells are multinucleate, and the nuclei within a cell go through the cell cycle in synchrony. Nuclei at one end of the cell enter mitosis first, and a wave of mitosis sweeps rapidly through the cell. Cyclin B–GFP fluorescence was essentially identical in all nuclei within a tip cell in interphase, and nuclei showed the same pattern of cyclin B–GFP destruction as the mitotic wave progressed through the cell.
In A. nidulans, multiple rounds of mitosis occur before the first cell division (septation) occurs. After each subsequent mitosis, septation occurs asymmetrically, dividing the hypha into a mitotically active tip cell and mitotically inactive subapical cells. We found that cyclin B does not accumulate in the nuclei in subapical cells (Fig. 1 A), indicating that they are blocked in G1. Cyclin B did accumulate again in subapical cells when a side branch with a growing tip formed, restarting the cell cycle.
The Cdk1-GFP localization pattern was very similar to that of cyclin B. Simultaneous time-lapse imaging of cyclin B–GFP and Cdk1-mCherry (strain LO1851) revealed that the two proteins appeared in the nucleus at the same time, increased in brightness as the cell cycle progressed, and disappeared from nuclei during mitosis. A surprising difference, however, was that Cdk1 consistently exited from mitotic nuclei ∼1.5 min before cyclin B–GFP had completely disappeared (Fig. 2). This suggests that a portion of Cdk1 is released before cyclin B is destroyed. Release of Cdk1 from cyclin B has been studied surprisingly little, perhaps because the assumption has been that the destruction of cyclin B by the proteasome causes release of Cdk1. However, the available evidence (Nishiyama et al., 2000; Chesnel et al., 2006) from in vitro studies suggests that the proteasome actually causes release of Cdk1 by a mechanism that precedes and is independent of proteolysis. Our observations provide in vivo results supporting these data.
Cyclin B, Cdk1, and Ancdc14 fail to accumulate in a subset of nuclei in a γ-tubulin mutant strain
To determine whether cyclin B localization was altered by mipAD159, we observed a strain (LO1439) carrying cyclin B–GFP, histone H1–mRFP, and mipAD159 at a restrictive temperature of 25°C. Many nuclei divide normally at 25°C in strains carrying mipAD159, whereas others exhibit chromosomal segregation abnormalities (Prigozhina et al., 2004). We found that cyclin B localization patterns in mitosis were normal in many nuclei, but we noted two alterations in nuclei undergoing abnormal mitosis. First, cyclin B remained at the poles for an extended period in some such nuclei, and chromatin did not decondense until cyclin B had disappeared. This presumably reflects the fact that cyclin B is involved in the monitoring of the successful completion of mitosis and that destruction of cyclin B is important for mitotic exit (Murray et al., 1989; Nurse, 1990; for review see Pines and Rieder, 2001). Second, cyclin B was often localized into clumps (Fig. S1 A), a localization pattern that was never seen in mipA+ controls. The cyclin B in these clumps disappeared over time, coincident with chromatin decondensation, but mitosis was usually not completed successfully.
However, the most striking defect was that a subset of nuclei failed to accumulate cyclin B in interphase, whereas other nuclei in the same cell accumulated it normally and passed through the cell cycle successfully (Fig. 3). The presence of such cyclin B–negative (CB−) nuclei was quite surprising in that, as mentioned, nuclei within a single-tip cell normally pass through the cell cycle in synchrony and behave identically with respect to the accumulation of cyclin B.
Cdk1-GFP had a similar localization pattern to cyclin B in a strain carrying mipAD159 (LO4031) at the same restrictive temperature and was completely absent from some nuclei in cells in which most nuclei had accumulated Cdk1 (Fig. S1 B). We imaged a strain carrying mipAD159, cyclin B–GFP, and Cdk1-mCherry (LO2586) at the restrictive temperature and found that no cyclin B–positive (CB+) nuclei were Cdk1 negative and vice versa. Thus, we infer that all Cdk1-negative nuclei are also CB−.
The homologue of Saccharomyces cerevisiae cdc14p exhibited a similar accumulation defect. This protein plays a critical role in mitotic regulation in some organisms (Jaspersen et al., 1998; Visintin et al., 1998; for review see Stegmeier and Amon, 2004). The A. nidulans cdc14 homologue has been identified (Son and Osmani, 2009) and designated Ancdc14. It is not essential for mitosis (Son and Osmani, 2009).
Time-lapse observations of a strain (LO1517) carrying Ancdc14-GFP and histone H1–mRFP revealed that Ancdc14 (the product of the Ancdc14 gene) begins to be visible in S phase, slightly after Cdk1 is detectable. Interestingly, Ancdc14-GFP does not localize to the nucleolus in A. nidulans (Fig. 4) as its homologues do in S. cerevisiae and Schizosaccharomyces pombe (for review see Stegmeier and Amon, 2004). It increases in intensity within nuclei as they pass through interphase, displays a complex localization in mitosis, and localizes to the forming septum after mitosis is completed. Time-lapse imaging of a strain carrying mipAD159, Ancdc14-GFP, and histone H1–mRFP (LO1532) revealed that, at a restrictive temperature, Ancdc14-GFP accumulated normally at septa but, like cyclin B and Cdk1, failed to accumulate in a subset of nuclei (Fig. 4).
To determine whether the nuclei that fail to accumulate Ancdc14 are the same as those that lack Cdk1 and, by inference, cyclin B, we imaged Ancdc14-GFP and Cdk1-mCherry in a strain carrying mipAD159 (LO4322). Because Cdk1 is detectable earlier in the cell cycle than Ancdc14, we only scored nuclei in cells in which Ancdc14-GFP was present in some nuclei. Of 139 Ancdc14-GFP–positive nuclei scored, only two (1.4%) were Cdk1-mCherry negative. Nearly all nuclei that accumulate Ancdc14 also accumulate Cdk1, and we can infer that nearly all Ancdc14-negative nuclei also fail to accumulate Cdk1 and cyclin B.
Analysis of the failure of cyclin B accumulation by time-lapse imaging
To investigate the failure of cyclin B/Cdk1 and Ancdc14 accumulation, we focused our attention on cyclin B because it is essential for cell cycle progression and because Cdk1 localization depends on cyclin B (for review see Nigg, 1995). We imaged strain LO1439 (cyclin B–GFP, histone H1–mRFP, and mipAD159) incubated at a restrictive temperature of 25°C from the time of germination (11 h after inoculation) through as many as five rounds of mitosis (20 h of imaging per sample). We first asked whether CB− nuclei resulted from disappearance of cyclin B at an inappropriate time or from failure of cyclin B to accumulate in interphase. Observations of 359 mitoses from many time-lapse datasets revealed that the latter was the case. Nuclei went through mitosis, often showing no mitotic abnormalities, and one or, less frequently, both daughter nuclei failed to accumulate cyclin B in the subsequent interphase (Fig. 5). Although the first nuclear division upon germination rarely resulted in CB− daughter nuclei, the percentage of daughter nuclei that became CB− increased with each successive round of mitosis until the majority of mitoses produced at least one CB− daughter nuclei (Fig. 6). Other than the increase over time, there was no obvious pattern with respect to nuclei becoming CB−.
From the same time-lapse datasets, we found that CB− nuclei did not go through mitosis or accumulate cyclin B in subsequent cell cycles. In effect, the CB− nuclei were removed from the cell cycle. When other nuclei in the same cell entered mitosis, releasing cyclin B into the cytoplasm, a small amount of cyclin B entered CB− nuclei (because of partial disassembly of the NPC; see the following sections) but quickly disappeared. The chromatin partially condensed as the cyclin B entered and decondensed as it disappeared. Observations of histone H1–mRFP revealed that CB− nuclei did not become grossly polyploid over time as other nuclei in the same cells passed through multiple cell cycles, although we cannot rule out the possibility that some DNA replication may have occurred. The combination of increasing percentages of daughter nuclei becoming CB− after each round of mitosis and the fact that once nuclei become CB− they stay CB− results in an increase in CB− nuclei over time (Fig. S2) that must contribute significantly to growth inhibition.
In combination, these data indicate that mipAD159 causes a failure of accumulation of cyclin B in interphase at restrictive temperatures, and this implies that γ-tubulin has an important, previously undiscovered function in the nuclear accumulation of cyclin B. Because different nuclei in the same cell exhibit different behaviors with respect to the accumulation of cyclin B, this phenomenon is nuclear autonomous and does not lend itself to bulk biochemical analysis. Consequently, we have performed observational and molecular genetic experiments to try to understand this phenomenon.
Failure of cyclin B accumulation is not simply because of abnormal mitosis
MipAD159 causes multiple mitotic defects at restrictive temperatures, including chromosomal nondisjunction (Prigozhina et al., 2004). Consequently, we asked whether the failure of nuclei to accumulate cyclin B was simply a result of defective chromosomal segregation. As mentioned above, time-lapse observations of LO1439 revealed that when nuclei underwent apparently normal mitosis, one of the two daughter nuclei often failed to accumulate cyclin B in the subsequent interphase (Fig. 5). Conversely, many aneuploid nuclei accumulated cyclin B normally (Fig. S3). We scored 127 nuclei that were obviously aneuploid (displaying a dramatically different amount of chromatin from mipA+ controls, as judged by histone H1–mRFP fluorescence) after incubation at a restrictive temperature of 25°C for 23 h, and 63 of them accumulated cyclin B. Thus, the failure of accumulation of cyclin B is not an obligate consequence of abnormal chromosomal segregation.
The failure of cyclin B accumulation is not a general nuclear transport defect
To determine whether nuclei fail to accumulate cyclin B because of a general nuclear transport defect, we constructed a strain (LO2589) carrying mipAD159, cyclin B–GFP, and the stuA nuclear localization signal fused to DsRed fluorescent protein (NLS-DsRed; Suelmann et al., 1997) under the control of the constitutive gpd promoter (Toews et al., 2004). At the restrictive temperature, NLS-DsRed accumulated in all nuclei, including those that failed to accumulate cyclin B–GFP (Fig. 7). Nuclei that are defective in cyclin B accumulation are, thus, capable of nuclear transport.
These data indicate that NPCs are functional in CB− nuclei. To observe the behavior of NPCs directly, we created a strain (LO1668) carrying mipAD159, cyclin B–GFP, and a functional fusion of the A. nidulans homologue of the NPC protein nup49 with mCherry (Osmani et al., 2006; De Souza and Osmani, 2007). Nup49-mCherry was present at the nuclear envelope of both CB+ and CB− interphase nuclei (Fig. S4 A). When cells entered mitosis, nup49-mCherry dissociated from CB− nuclei at the same time that it dissociated from normal nuclei. After mitosis, nup49-mCherry relocalized to the nuclear envelopes of nuclei that remained CB− through the following cell cycle. Thus, nuclear pore assembly and disassembly appear to be normal in CB− nuclei. These data cast light on the control of nuclear pore disassembly at mitotic onset. Nuclear pore disassembly could, in principle, be controlled by an internal signal generated by G2 nuclei as they enter mitosis. Because CB− nuclei do not cycle, it is unlikely that they generate a normal nuclear pore disassembly signal. Rather, these results suggest that nuclear pore disassembly is controlled by a cytoplasmic signal that passes through the cell.
Cyclin B is normally exported from nuclei via a CRM1-mediated pathway (Kornbluth et al., 1992; Hagting et al., 1998; Yang et al., 1998). To determine whether CB− nuclei fail to accumulate cyclin B because they constitutively export it, we created strain LO2994 that carries cyclin B–GFP, mipAD159, histone H1–mRFP, and the crmAT525C allele. The crmA gene of A. nidulans encodes a CRM1 homologue, and the mutation crmAT525C makes the protein sensitive to leptomycin B such that 10 ng/ml of the compound prevents CRM1-mediated export (Todd et al., 2005). We grew LO2994 at 25°C to allow the accumulation of CB− nuclei, then added leptomycin B and observed CB− nuclei by time-lapse microscopy to determine whether they accumulated cyclin B–GFP. CB− nuclei did not accumulate cyclin B–GFP in treatments with 10, 20, or even 100 ng/ml leptomycin B (Fig. S4 B). Thus, constitutive export via the CRM1 pathway is not responsible for CB− nuclei.
The cyclin B accumulation defect is caused by failure of APC/C inactivation
Many of the defects caused by mipAD159 at restrictive temperatures (Prigozhina et al., 2004) are consistent with mipAD159 causing misregulation of APC/C. The APC/C is a ubiquitin ligase that ubiquitinates key mitotic regulatory proteins, targeting them for destruction. Activation of the APC/C is required for exit from mitosis, and it remains active through G1 to prevent accumulation of cyclins and other substrates (for review see van Leuken et al., 2008). The APC/C is normally inactivated at the G1/S transition and stays inactive until all kinetochores are attached to spindles in a bipolar fashion, and the spindle assembly checkpoint (SAC) is inactivated (Buschhorn and Peters, 2006; for review see Acquaviva and Pines, 2006). Cyclin B is an APC/C target, and CB− nuclei could, in principle, be caused by nuclear autonomous failure of APC/C inactivation.
The APC/C targets cyclin B for destruction by recognizing a conserved motif near the N terminus called the destruction box (Glotzer et al., 1991; Glotzer, 1995). N-terminal truncations deleting the destruction box result in a form of cyclin B that binds to Cdk1 and drives the cell cycle, but because it is no longer targeted by the APC/C, it is not destroyed in mitosis and causes a mitotic arrest (Murray et al., 1989; Glotzer et al., 1991; Glotzer, 1995; De Souza et al., 2009; for review see Acquaviva and Pines, 2006). A. nidulans cyclin B has a canonical destruction box (RAALGDVSN) near its N terminus. We created a truncated version of cyclin B–GFP lacking the N-terminal 73 amino acids that contain the destruction box–deleted cyclin B–GFP (dbΔ–cyclin B–GFP). Because destruction of cyclin B is essential for the completion of mitosis, expression of dbΔ–cyclin B–GFP should prevent mitotic exit and thus be lethal. To overcome this problem, we placed dbΔ–cyclin B–GFP under the control of the highly regulatable alcA promoter (Waring et al., 1989). We integrated the construct into the wA (white spore color) locus of a strain carrying histone H1–mRFP, leaving the wild-type copy of the nimE (cyclin B) gene intact. The resulting strain, LO1847, grew like the parent on alcA-repressing media, but under inducing conditions, growth was severely inhibited (Fig. 8 A), indicating that expression of dbΔ–cyclin B–GFP was efficiently repressed on glucose and efficiently induced under inducing conditions.
We next examined the effects of dbΔ–cyclin B–GFP microscopically. Upon induction, dbΔ–cyclin B–GFP gradually accumulated in nuclei. Consistent with recent results using a slightly different destruction box–deleted cyclin B construct (De Souza et al., 2009), nuclei in cells expressing dbΔ–cyclin B–GFP entered mitosis and progressed through anaphase, but the dbΔ–cyclin B–GFP signal was visible at spindle poles for an extended period, and nuclei were blocked in telophase.
To determine the effects of mipAD159 on dbΔ–cyclin B–GFP accumulation, we created a strain (LO1849) that carries dbΔ–cyclin B–GFP, histone H1–mRFP, and mipAD159. We also created control strains LO1943 (wild-type cyclin B–GFP under the control of the alcA promoter integrated at the wA locus, histone H1–mRFP, and mipAD159), LO1439 (mipAD159, cyclin B–GFP, and histone H1–mRFP), and LO1438 (mipA+, cyclin B–GFP, and histone H1–mRFP).
Conidia were incubated in medium containing glucose to repress dbΔ–cyclin B–GFP expression at a permissive temperature of 37°C for 12 h. The repressing medium was washed out and replaced with inducing medium, and the culture was immediately shifted to a restrictive temperature of 25°C. Cells were imaged 1–3 h after the shift. This time window was chosen because it is less than one cell cycle after induction but long enough after induction to allow cyclin B–GFP or dbΔ–cyclin B–GFP to accumulate in nuclei. As shown in Fig. 8 B, deletion of the destruction box allowed cyclin B to accumulate in nearly all nuclei. The accumulation was not because of alterations of expression levels by the alcA promoter because cells expressing wild-type cyclin B–GFP under control of this promoter contained many CB− nuclei.
To verify these results, we created a strain (LO1910) that carries mipAD159 and dbΔ–cyclin B–GFP under the control of the alcA promoter along with wild-type cyclin B–mCherry under the control of its endogenous promoter. We performed a temperature-shift dbΔ–cyclin B–GFP induction experiment as above and found that all nuclei accumulated dbΔ–cyclin B–GFP, but a subset of these nuclei did not accumulate cyclin B–mCherry (Fig. 8 C). Thus, two separate sets of experiments reveal that the removal of the destruction box targeted by the APC/C overrides the cyclin B accumulation defect caused by mipAD159. It follows that the cyclin B accumulation defect caused by mipAD159 is because of inappropriately active APC/C. γ-Tubulin, thus, plays an important role in inactivating the APC/C in interphase, an activity that is compromised in mipAD159.
The activity of the APC/C can be reset in CB− nuclei
Our data indicate that mipAD159 causes a nuclear autonomous failure of APC/C inactivation. If this is correct, inactivation of the APC/C in those nuclei should cause them to be reset to become CB+. Treatment with antimicrotubule agents activates the SAC, which, in turn, inactivates the APC/C (Hoyt et al., 1991; Li and Murray, 1991; for review see Musacchio and Salmon, 2007). It follows that if mipAD159 causes constitutive activation of the APC/C, prolonged treatment with an antimicrotubule agent might inactivate the APC/C, allowing some nuclei to reset and accumulate cyclin B.
Benomyl at a concentration of 2.4 µg/ml completely depolymerizes microtubules (Horio and Oakley, 2005). When wild-type cells treated with benomyl enter mitosis, the mitotic spindle does not form, and nuclei are arrested in mitosis for an extended period with an activated SAC. Eventually, the SAC becomes inactivated, and nuclei exit mitosis and enter the next cell cycle (unpublished data). These nuclei accumulate cyclin B normally in interphase and are able to synthesize DNA and increase in ploidy but do not divide during mitosis. We grew a strain carrying mipAD159 at the restrictive temperature of 25°C to accumulate CB− nuclei. After addition of benomyl, we imaged hyphae over 5–10 h to determine whether CB− nuclei regained the ability to accumulate cyclin B (Fig. 9). Because benomyl depolymerizes microtubules that are required for nuclear movement, nuclei moved slowly, and this allowed us to follow them over long periods. When cells entered mitosis, a portion of the cyclin B exited from CB+ nuclei, and the chromosomes condensed. During mitotic arrest, CB− nuclei accumulated some cyclin B, reflecting the fact that the nuclei were blocked in a semi-open conformation with much of the NPC disassembled (De Souza and Osmani, 2007). Eventually, cyclin B disappeared completely from all nuclei (presumably caused by APC/C activation at the end of the block), and nuclei exited mitosis. We followed 71 nuclei by time-lapse microscopy through interphase, mitosis, and at least one subsequent interphase. 23 nuclei (32%) that were CB+ before mitosis remained CB+ through the following interphase. 20 nuclei (28%) that were CB− became CB+, indicating that CB− nuclei can indeed be reset to be CB+ by a prolonged mitotic arrest. 16 nuclei (23%) that were CB− remained CB−. Interestingly, 12 nuclei (17%) that were CB+ became CB−, showing that mipAD159 can cause nuclei to become CB− in the absence of microtubules. Importantly, the fact that nuclei can become CB− when microtubules are depolymerized appears to rule out the possibility that CB− nuclei are related to any potential microtubule nucleation defects of mipAD159. It is also worth noting that no nuclei changed from CB− to CB+ or vice versa until the cell had entered and exited mitosis. In control experiments with a mipA+ strain, the great majority of nuclei (80/85; 94.1%) remained CB+ throughout the experiment, but a small fraction (5/85; 5.9%) became CB−. Although this number is much lower than in the strain carrying mipAD159, it is intriguing and worthy of further study.
The APC/C is a potential source of nuclear autonomy
In the absence of benomyl, once nuclei become CB−, they remain CB−. If CB− nuclei are caused by a constitutively active APC/C, it follows that the APC/C in nuclei should not exchange freely with APC/C in the cytoplasm nor leave the nucleus completely at any time in the cell cycle. To determine whether the localization pattern of the APC/C is consistent with a constitutively active APC/C causing CB− nuclei, we wished to tag and image at least one APC/C component. It is notoriously difficult to obtain functional tagged versions of APC/C components (Acquaviva et al., 2004), but we were able to create a functional tagged version of the A. nidulans APC3 homologue BimA (O’Donnell et al., 1991; Ye et al., 1998). The localization of BimA has previously been studied by immunofluorescence microscopy (Mirabito and Morris, 1993), where it was reported to localize to SPBs or, if overexpressed, to the nucleus. Although we did see an apparent SPB localization in some nuclei, the predominant localization of BimA-GFP was with chromatin. BimA was nuclear in interphase and early mitosis, and during anaphase, BimA-GFP localized to the separating chromosomes and the spindle between the separating chromosomes (Fig. 10). It then disappeared from the spindle but remained associated with the decondensing chromatin. Thus, BimA is not freely diffusible, and its continuous association with chromatin suggests that it could be a source of nuclear autonomy.
A question that arises is why the localization pattern we have observed is different from the pattern observed by immunofluorescence. BimA is an essential gene (O’Donnell et al., 1991), and our bimA-GFP construct appears fully functional in that it is the only copy in the genome and supports normal growth and doesn’t cause any obvious mitotic abnormalities. It is expressed from its endogenous promoter and, thus, should be present at normal levels. Therefore, we are confident that the localization pattern we have observed accurately reflects the distribution of BimA in the cell.
We next asked whether BimA-GFP localizes to all nuclei in a strain carrying mipAD159 at the restrictive temperature of 25°C. For this, we grew strain LO3935 at 25°C for 23 h. Under these conditions, 39.5% of nuclei did not accumulate cyclin B. In contrast, 96.6 ± 1.3% of nuclei contained visible BimA-GFP (mean and SD of three experiments; total n = 634). These data indicate that most if not all CB− nuclei contained BimA and, by inference, APC/C.
At restrictive temperatures, in strains carrying the γ-tubulin mutation mipAD159, a subset of nuclei fails to accumulate the key regulatory proteins cyclin B and Cdk1 and the phosphatase Ancdc14, whereas other nuclei in the same cells accumulate them normally. Because Cdk1 localization depends on cyclin B, we presume that the failure of accumulation of Cdk1 is a consequence of the failure of accumulation of cyclin B. In the absence of the Cdk1–cyclin B complex, which is required for progression through both S and M phases in A. nidulans (Osmani et al., 1994), nuclei are effectively removed from the cell cycle. Additional nuclei become CB− in every cell cycle, resulting in an increase over time in the percentage of nuclei that are CB−, thus contributing to growth inhibition.
Our results strongly indicate that CB− nuclei result from a nuclear autonomous failure of inactivation of the APC/C, resulting in the continuous targeting of cyclin B for proteolysis. The lack of Ancdc14 in CB− nuclei is also consistent with a permanently active APC/C because Ancdc14 has a canonical destruction box and is thus likely to be an APC/C target. Because a γ-tubulin mutation causes a failure of inactivation of the APC/C at restrictive temperatures, it follows that γ-tubulin normally plays an important role in the inactivation of the APC/C.
The APC/C, which is active in mitosis, is inactivated before S phase, allowing the accumulation of APC/C substrates (such as cyclin B) that are required for DNA replication and cell cycle progression (for review see Acquaviva and Pines, 2006). Our observations show that CB− nuclei result from formerly CB+ nuclei that enter and exit mitosis but do not subsequently accumulate cyclin B. It follows that the failure of APC/C inactivation occurs after mitotic APC/C activation but before S when cyclin B accumulation normally becomes apparent.
Our benomyl data allow us to draw two important conclusions. The first is that because some CB− nuclei are reset in benomyl to become CB+, CB− nuclei are not biochemically dead. They have the potential to cycle but are prevented from doing so by the constitutively active APC/C. The second is that because CB+ nuclei can become CB− in the absence of microtubules, the failure of inactivation of the APC/C by mipAD159 is independent of microtubules and is not a consequence of any microtubule nucleation defect that might be caused by mipAD159. One question that arises is why some nuclei are reset to become CB+, whereas others are not. Our data do not allow us to answer this question definitively. One possibility is that the APC/C is not completely inactivated in some nuclei for reasons that are not clear but might, for example, relate to the amount of active APC/C in each particular nucleus. Another is that the APC/C is inactivated in mitosis in each nucleus but subsequently becomes reactivated in some nuclei in a probabilistic fashion after mitotic exit such that cyclin B does not accumulate. Some nuclei that are CB− would thus remain CB− even though the APC/C had been transiently inactivated during mitotic blockage.
These data raise the question of why the SAC does not inactivate the APC/C in CB− nuclei when they enter mitosis in the absence of benomyl. One possibility is that the activation of the SAC requires Cdk1 activity (D’Angiolella et al., 2003; Wong and Fang, 2007), which is missing in CB− nuclei. Another is that one or more SAC proteins are destroyed by the constitutively active APC/C. In either case, a prolonged mitotic block would likely allow enough protein exchange to restore SAC activity.
Why do only a fraction of the nuclei become CB− after each round of nuclear division? MipAD159 is a recessive allele (Jung et al., 2001). It is therefore likely to be a loss of function allele, but loss of a subset of γ-tubulin functions rather than all functions because microtubules are nucleated in abundance in strains carrying mipAD159 (Jung et al., 2001). The simplest explanation is that the function of γ-tubulin in inactivating the APC/C is reduced but not eliminated in mipAD159, and this leads to a stochastic loss of APC/C inactivation in each G1 rather than a certain loss.
γ-Tubulin, the centrosome, and cell cycle regulation
Our data provide the first evidence that γ-tubulin plays a role in inactivation of the APC/C after mitotic exit. Evidence is accumulating that the γ-tubulin complex and polar microtubule-organizing centers, the SPB and the centrosome, play an important role in cell cycle regulation (Vardy and Toda, 2000; Sluder, 2005; Müller et al., 2006; Mikule et al., 2007; Cheng et al., 2008; for review see Cuschieri et al., 2007). It is also becoming clear that cyclin B destruction is spatially regulated in phylogenetically distant organisms and that polar microtubule-organizing centers may play a role in this spatial regulation (Fig. 1 B; Huang and Raff, 1999; Wakefield et al., 2000; De Souza et al., 2009). Also note that the patchy distribution of cyclin B that we see in some mipAD159 mitotic nuclei (Fig. S1) may reflect abnormal spatial regulation of cyclin B destruction. Given these facts and the fact that the γ-tubulin encoded by the mipAD159 allele localizes normally to the SPB (Jung et al., 2001), it is reasonable to assume that the function of γ-tubulin in inactivating the APC/C occurs at the SPB. For example, it could hypothetically be involved in a signal transduction pathway that senses mitotic and cell cycle events and inactivates the APC/C after mitosis. However, our data are also consistent with γ-tubulin functioning away from the SPB. For example, a small amount of γ-tubulin (or γ-tubulin small complexes) independent of the SPB could, in principle, interact directly with the APC/C and be involved in its inactivation.
BimA localization and nuclear autonomy
In some fungi, all nuclei in a cell go through mitosis in synchrony, whereas in others, mitosis and the cell cycle are nuclear autonomous. Our data reveal that although nuclei in A. nidulans normally go through the cell cycle in synchrony, when perturbed by mipAD159, they can behave in an autonomous fashion. Moreover, the localization pattern of BimA suggests that the APC/C is a potential source of nuclear autonomy.
Materials and methods
Table S1 shows a complete list of the strains used in these experiments and their genotypes.
For imaging, conidia were inoculated into liquid minimal medium (6 g/liter NaNO3, 0.52 g/liter KCl, 0.52 g/liter MgSO4 7H2O, 1.52 g/liter KH2PO4, 10 g/liter D-glucose, and 1 ml/liter of a trace element solution [Vishniac and Santer, 1957], pH adjusted to 6.5, with NaOH before autoclaving, supplemented as appropriate for nutritional markers) in four- or eight-chamber cover glasses (Lab-Tek; Thermo Fisher Scientific). We used three systems for imaging. Two were inverted microscopes (IX71; Olympus) equipped with shutters, filter wheels, z-axis drives (Prior Scientific), and mercury light sources. One was equipped with an ORCA ER camera (Hamamatsu Photonics) and the other with an ORCA ERAG camera (Hamamatsu Photonics). We used Semrock GFP/DsRed2X2M-B dual-band Sedat filter sets (459–481-nm bandpass excitation filter for GFP and a 546–566-nm excitation filter for mCherry and mRFP), dual-reflection band dichroic (457–480- and 542–565-nm reflection bands; 500–529- and 584–679-nm transmission bands), and two separate emission filters (GFP, 499–529 nm; mCherry and mRFP, 580–654 nm) were used. Images were acquired with a 60× 1.42 NA Plan Apochromatic objective (Olympus) using Slidebook (Intelligent Imaging Innovations) or Volocity software (PerkinElmer) installed on computers (PowerMac; Apple). For time-lapse two-channel imaging of live cells, z-series stacks were collected at each time point, and maximum intensity projections from all time points were combined to generate videos using Slidebook or Volocity software. The third system was a spinning-disk confocal system (UltraVIEW VoX; PerkinElmer) mounted on an inverted microscope (IX71; Olympus). It was equipped with an environmental chamber to maintain stable temperatures and a software-controlled piezoelectric stage for rapid z-axis movement. Solid-state 488- and 561-nm lasers were used for excitation. A 60× 1.42 NA objective (Olympus) was used (in some cases with a 1.6× Optovar), and images were collected with an ORCA ERAG camera. The system was controlled by a computer (PowerMac) running Volocity software.
To prepare figures, images were exported directly from the imaging software after adjustment of minimum and maximum intensity levels. No γ adjustments were made. Graphic Converter was used to change the file format of some images. Figures were drawn and/or assembled with CorelDraw software.
Gene targeting and transformation
Transformation procedures and generation of linear transforming molecules by fusion PCR were performed as described previously (Nayak et al., 2006; Szewczyk et al., 2006). To create C-terminal fluorescent protein fusions, the linear transforming molecule consisted of ∼500- or 1,000 bp of the coding sequence of the target gene fused in frame to a glycine–alanine linker (Yang et al., 2004), which was in turn fused in frame with the fluorescent protein coding sequence (either GFP, mCherry, or mRFP). The fusion protein coding sequence was followed by a selectable marker (A. fumigatus pyrG [AfpyrG; Weidner et al., 1998], A. fumigatus riboB [AfriboB; Nayak et al., 2006], or A. fumigatus pyroA [AfpyroA; Nayak et al., 2006]), which was in turn followed by ∼500 or 1,000 bp of 3′ untranslated region flanking the target gene. Integration of such fragments by homologous recombination resulted in a single copy of the gene (the only copy in the genome) under the control of its native promoter and fused in frame to the fluorescent protein coding sequence. The GFP variant we used was a plant-adapted GFP (Fernández-Abalos et al., 1998) that works particularly well in A. nidulans. mRFP and mCherry were used as previously described (Campbell et al., 2002; Shaner et al., 2004; Toews et al., 2004). Correct gene targeting was facilitated by the use of nkuAD strains (Nayak et al., 2006).
The dbΔ–cyclin B–GFP construct was created as follows. A fragment of the cyclin B gene extending from nucleotide 316 to the stop codon (nucleotide 1,672) was amplified by PCR. This fragment excludes the N-terminal 73 amino acids that contain the destruction box (bp 285–312). The primers used to amplify this fragment were designed with a tail at the 5′ end that annealed to the alcA promoter, including a start codon, and a tail at the 3′ end that annealed to a GFP-AfpyrG cassette. We also amplified regions flanking the wA gene (5′, 996 bp; 3′, 1,007 bp). One of the two primers used to amplify the 5′ flank had a tail that allowed annealing to the alcA promoter fragment. One of the primers used to amplify the 3′ wA fragment was designed with a tail that allowed annealing to the GFP-AfpyrG cassette. The fragments were combined, and a single fragment was amplified by fusion PCR. A similar strategy was used to construct a full-length cyclin B gene fused to GFP-AfpyrG under control of the alcA promoter with wA flanking sequences.
The alcA promoter is repressed by glucose and induced by several compounds, including threonine. The inducing medium for our observations was 6 g/liter NaNO3, 0.52 g/liter KCl, 0.52 g/liter MgSO4 7H2O, 1.52 g/liter KH2PO4, 9 g/liter fructose, and 1 ml/liter of a trace element solution (Vishniac and Santer, 1957) and 6.25 mM threonine. Repressing medium was identical except that it contained 10 g/liter D-glucose instead of fructose and no threonine.
Diagnostic PCR and Southern hybridizations
A. nidulans genomic DNA of transformants was prepared as described previously (Lee and Taylor, 1990). Positive transformants were first confirmed through diagnostic PCR using primers flanking the target sequence. Correct integrations and the absence of extra integrations of transforming DNAs were also verified by Southern hybridizations in dried agarose gels as described previously (Oakley et al., 1987). Radioactively labeled full-length transforming DNA fragments were used as probes.
LO2994 was grown in minimal medium for 18 h at 25°C before leptomycin B was added to a final concentration of 10 ng/ml. Cells with CB− nuclei were imaged over a period of 3 h after addition of leptomycin B at the restrictive temperature of 25°C. We repeated the experiment with 20 and 100 ng/ml of leptomycin B, imaging up to 15 h after leptomycin B addition, and obtained similar results. Benomyl treatments were performed as previously described (Horio and Oakley, 2005). In brief, benomyl was maintained as a 240 µg/ml solution in 95% ethanol. An appropriate volume of the stock solution was added directly to chambered cover glasses using a micropipeter to give a final concentration of 2.4 µg/ml.
Online supplemental material
Fig. S1 shows abnormal distribution of cyclin B and Cdk1. Fig. S2 shows the accumulation of CB− nuclei in a strain carrying mipAD159 over time at a restrictive temperature. Fig. S3 shows aneuploid CB+ nuclei, and Fig. S4 demonstrates that CB− nuclei are not the result of defects in nuclear pore localization or export. Table S1 lists the strains used for these experiments
We thank Elizabeth Oakley for technical and editorial assistance, Dr. Roger Tsien for the mCherry gene, and Drs. R.B. Todd and M.J. Hynes for the crmAT525C mutant.
This work was supported by the National Institutes of Health (grants GM031837 to B.R. Oakley and GM042564 to S.A. Osmani) and the Kansas University Endowment.
cyclin B negative
cyclin B positive
- dbΔ–cyclin B–GFP
destruction box–deleted cyclin B–GFP
nuclear localization signal fused to DsRed fluorescent protein
nuclear pore complex
spindle assembly checkpoint
spindle pole body
T. Nayak’s present address is Dept. of Pathology and Cell Biology, Columbia University College of Physicians and Surgeons, New York, NY 10032.