|Home | About | Journals | Submit | Contact Us | Français|
Current models of cell cycle control, based on classic studies of fused cells, predict that nuclei in a shared cytoplasm respond to the same CDK activities to undergo synchronous cycling. However, synchrony is rarely observed in naturally occurring syncytia, such as the multinucleate fungus Ashbya gossypii. In this system, nuclei divide asynchronously raising the question of how nuclear timing differences are maintained despite sharing a common milieu.
We observe that neighboring nuclei are highly variable in division cycle duration and neighbors repel one another to space apart and demarcate their own cytoplasmic territories. The size of these territories increases as a nucleus approaches mitosis and can influence cycling rates. This non-random nuclear spacing is regulated by microtubules and is required for nuclear asynchrony, as nuclei that transiently come in very close proximity will partially synchronize. Sister nuclei born of the same mitosis are generally not persistent neighbors over their lifetimes yet remarkably retain similar division cycle times. This indicates that nuclei carry a memory of their birth state that influences their division timing and supports that nuclei subdivide a common cytosol into functionally distinct yet mobile compartments.
These findings support that nuclei use cytoplasmic microtubules to establish “cells within cells.” Individual compartments appear to push against one another to compete for cytoplasmic territory and insulate the division cycle. This provides a mechanism by which syncytial nuclei can spatially organize cell cycle signaling and suggests size control can act in a system without physical boundaries.
Classic cell fusion experiments by Rao and Johnson from the 1970's showed that when HeLa cells in different phases of the cell cycle were fused together to form multinucleate cells, their nuclei rapidly synchronized [1–5]. Similarly, early Drosophila syncytial embryos orchestrate highly synchronous nuclear division cycles . These findings indicate that nuclear division can be coordinated through sharing a common cytoplasm, likely by exposure to similar levels of key regulators.
Although sharing a common cytoplasm can result in synchronous nuclear division cycles, it is by no means certain. After HeLa cell fusion, nuclear asynchrony may arise in subsequent mitoses . When a multi-nucleated myotubule re-enters the cell cycle, its nuclei do so asynchronously . Similarly, many filamentous fungi display asynchronous division of nuclei in one cell . Therefore, synchronization due to shared cytoplasmic signals can be spatially restricted. Although examples of asynchronous nuclear division within a common cytoplasm have been documented, the mechanisms of asynchrony in syncytia are not well understood. Asynchrony presumably requires timing variability within the nuclear division cycle in addition to a mechanism, such as compartmentalization of the cytoplasm, which would prevent adjacent nuclei from experiencing similar concentrations of regulatory molecules.
There are numerous known molecular sources of cell cycle timing variability, including stochastic differences in gene expression and size control . In Saccharomyces cerevisiae, stochastic expression of cell cycle regulators generates cell-to-cell variability in division timing within a population [11, 12]. In addition, differences in cell size at birth influence the duration of G1 prior to the point of commitment to cell division [11, 13–17]. Smaller cells delay cell cycle progression relative to larger cells, which results in timing variability within the population. In a multinucleate context it is unknown whether nuclei sense and respond to a local volume of cytoplasm or somehow more globally coordinate nuclear division with cell growth.
Progression through the phases of the cell cycle is driven by the cyclin/CDK biochemical oscillator regulated by periodic cyclin accumulation and degradation [18–20]. In a syncytial context where nuclei cycle asynchronously, many out of sync biochemical oscillators coexist and fail to entrain one another. This requires either that the activity of the oscillators (e.g., the capacity to phosphorylate substrates) is completely restricted to nuclei, perhaps due to a cytoplasmic CKI and/or that there are barriers to diffusion of CDK activity. Such barriers would allow each nucleus to have its own segment of cytoplasm driving only its local oscillator. It is unclear what the cell biological basis would be of such cytoplasmic barriers yet the phenomenon of asynchrony likely requires some insulation of nuclei from diffusing signaling molecules from neighboring regions.
To examine the mechanisms of asynchrony, we observe nuclear division timing and spatial distributions in the multinucleate fungus Ashbya gossypii. Neighboring Ashbya nuclei can be in different cell cycle stages and their nuclear division cycle times can vary widely . Asynchrony in Ashbya emerges early in G1 and is under genetic control as mutant cells lacking central components of the G1/S regulatory pathway become more synchronous in their division cycles . Components of this pathway control transcription, which is of interest given that transcripts are translated and shared in the common cytoplasm. The importance of this transcriptional regulatory pathway for asynchrony supports the hypothesis that there may be restricted sharing of newly made proteins between neighboring nuclei. Here, we employ live-cell imaging and statistical approaches to investigate how Ashbya nuclei functionally insulate themselves to produce variable nuclear division cycle times within a common cytoplasm.
The positions and divisions of all nuclei in single Ashbya cells were tracked through time based on timelapse imaging of cells expressing histone H4-GFP (HHF1-GFP) to visualize nuclei (Movies S1–S3). Nuclear coordinates were imported into MATLAB to analyze the timing patterns and spatial relationships of individual nuclei (Movies S1–S3). The mean nuclear division time and its standard deviation were 118 and 38 minutes respectively (n=92 division times from 7 independent timelapse movies), consistent with previous observations (Figure 1A, ). Mitoses occur at all locations throughout the cell and are not restricted to a single region, such as a tip (Figure 1B, data not shown) indicating that there are no stable, persistent sites that consistently favor or disallow nuclear division. Additionally, nuclei that divide around the same timepoint in the movie are not generally found to be near each other in the cell suggesting there are not transient, local bursts of mitotic activity (Figure 1B, Figure S1, n=95 mitoses in three independent movies). Nuclear autonomous division occurs regardless of the size of the cell. In fact, even the very first nuclear divisions after germination are asynchronous (Figure 1C, Movie S4, n=25 mitoses in 5 independent movies). These data indicate that nuclear autonomy is not a consequence of the gradual accumulation of timing differences among many nuclei as cells age and become large. Thus, regardless of overall cell size, nuclei are highly variable and asynchronous in division timing even when sharing the same cytosol.
How might nuclei establish functionally autonomous zones in a common cytoplasm? Notably, we see highly regular spacing between neighboring nuclei that is significantly different from what would be expected if they were randomly positioned using a Monte Carlo simulation that maintains the same number of nuclei in the same hyphal geometry (Figure 2A–B, observed mean=4.3 ± 2.1μm, p<0.001, K-S Test and F-Test). This prompted us to ask how non-random spacing is achieved and look at how nuclei move relative to their neighbors. First, we examined nuclear positions in a variety of mutants lacking microtubule motors or having perturbed microtubule length [23–25]. The majority of these mutants show nuclei that are closer together while cells lacking Ase1, a microtubule associated protein (MAP), and the kinesin Kip2 both show larger distances between neighboring nuclei (Figure 2C, Table 1). Importantly, the nuclear spacing in all mutant strains except Kip2 is more variable compared to WT (Table 1). This increased variability is associated with more random nuclear spacing for all mutants compared to the non-random spacing observed in WT (Figure 2D–E, Figure S2). This was assessed by comparing the distribution of spacing observed in the mutant and wild-type strains to multiple iterations of random distributions generated for each mutant data set. By creating distinct random simulations based on mean spacing of each mutant strain we ensure that mutants are compared to a random distribution of the same mean and therefore there are not artifacts of comparing distributions with different means. Wildtype deviates substantially from random while most mutants are more consistent with a random distribution (Figure S2). Thus, regulation of the microtubule cytoskeleton is critical for non-random nuclear spacing.
Next, we looked at how neighboring nuclei move relative to one another to examine how non-random spacing is achieved. To do this, we measured the difference in the distances between a nucleus and its two nearest neighbors (“neighbor offset”) and plotted the offsets through time (Figure 3A–B). In wild-type cells, the distances between individual nuclei and their neighbors typically fluctuate around a mean value however these fluctuations are eliminated when microtubule dynamics are impaired with nocodazole (Figure 3C, left). To assess further how neighboring nuclei change positions through time we plotted the neighbor offset at each timepoint against the change in neighbor offset at the subsequent timepoint (Figure 3C, middle and right). In wild-type cells we see the offsets change substantially each timepoint as indicated by spread of points around zero. In contrast, in the absence of microtubules, nuclei no longer fluctuate in position relative to their neighbors. In cells with depolymerized microtubules, the offsets cluster at a non-zero point indicative of aberrant spacing and importantly there is little change in the offset from timepoint to timepoint such that the points are often highly clustered (Figure 3C, middle and right).
We next assessed the time scale of the fluctuations characteristic of wild-type cells. Autocorrelation analysis of nuclear offset over time reveals that nuclei are repelling one another on a time scale of ~2.2 minutes (Figure 3D). The autocorrelation function fits well to an exponential decay. This is consistent with an Ornstein-Uhlenbeck model for Brownian motion subject to a restoring force in a highly damped media . Thus, nuclei use microtubule based-motion to repel their neighbors to produce a non-random spatial distribution.
We hypothesize that repulsive forces between nuclei help them to subdivide the cytoplasm into “territories” or zones of influence. How then is regular nuclear spacing maintained while the cytoplasm expands due to cell growth? Importantly all expansion is happening at hyphal tips and so most nuclei are tens to hundreds of micrometers away from the actual site where cell size is increasing yet nuclear spacing remains consistent at all locations. We hypothesized that nuclear dynamics may also promote the distribution of spacing in a cell with asymmetric growth and each nucleus would locally control the amount of space it accrues. In this model, the sizes of individual nuclear territories may change analogous to how a uninucleate cell grows through its cell cycle.
By tracking how neighbor spacing changes over the division cycle, we found an increase in local spacing occurs gradually in anticipation of mitosis (Figure 4A, r=0.91). This growth in territory size is characteristic for most of the population as r>0.5 for a linear fit of territory size versus time in over 59% of nuclei (Figure S3A–B) and clearly nuclei are significantly farther from their neighbors prior to mitosis than throughout the remainder of the cell cycle (Figure 4B, p<0.001, K-S Test). Importantly because all cell growth is restricted to the hyphal tip, the increase in local cytoplasm size prior to mitosis is due to repositioning of nuclei and not local intercalary growth. Furthermore, because nuclei can bypass one another and migrate far from their birth location in the cell (Figure S3C), the local decrease in nuclear spacing that must occur for each mitosis is quite transient and not creating an artifactual increase in space throughout the entire cell cycle. The link between the division cycle and territory size increase further supports that spacing is controlled at the level of the individual nucleus.
In most uninucleate cells, cell size is inversely proportional to cycle time so we next asked whether division timing is affected by the amount of space a nucleus inhabits at birth. However, nuclear spacing in the early period of the cycle only very weakly correlates with overall division cycle length (Figure S3D, r=−0.19). Additionally, nuclear spacing immediately preceding mitosis has a modest correlation with division timing (Figure S3E, r=0.28). Therefore, the amount of space a nucleus is born into and ultimately possesses seems to only modestly influence overall division timing. As an additional method to investigate how territory size may be linked to division time, we artificially increased the local spacing around nuclei using nocodazole, which arrests the nuclear cycle while still allowing normal cell growth (Figure 4C–D). Upon release from the arrest, nuclei re-enter the division cycle and cell growth continues at hyphal tips at normal rates (data not shown). Remarkably, we see that nuclei swiftly recover to near wild-type spacing even while the cells continue to grow during the release period (Figure 4C–D). Therefore, the nuclear cycle must speed up in response to the increased amount of cytoplasm per nucleus to restore the original spacing. This argues that the division cycle is able to respond to alterations in the amount of cytoplasm per nucleus but this reaction is not readily detectable unless spacing is substantially perturbed.
The data thus far are consistent with the model that nuclear movement generates non-random spacing and this functions to maintain a constant balance of nuclei to cytoplasm. We hypothesized that controlled spacing may provide insulation between neighboring nuclei to promote autonomous nuclear division. To investigate if spacing influences the degree of asynchrony, we took advantage of a sub-population of nuclei that undergo “by-passing” events where they change places with one of their neighbors (Figure 5A). Such encounters require nuclei to pass <0.25μm from one another and thus would represent the greatest opportunity for intermingling of cell cycle regulators. We asked if nuclei that came close together like this were subsequently synchronized such that they would tend to divide at the same timepoint of the movie. In fact, we see that nuclei that partake in a by-passing event are significantly more likely to divide at a similar moment in time than nuclei from the same datasets that are randomly paired irrespective of spacing (Figure 5B). Whereas persistent nuclear neighbors, defined as those spending at least 30 minutes between 2 and 5μm of each other yet do not bypass (Figure 5A), are more similar to the random distribution (Figure 5B; ANOVA p<0.05; median Δtprandom= 49min; median Δtp2–5μm= 41min; Δtp0.25μm= 30min;). This suggests that nuclear spacing is important to insulate nuclei from their neighbors and allow for nuclear cycle timing autonomy.
If controlled nuclear spacing insulates nuclei, we predict that randomly spaced nuclei would divide more synchronously than wildtype. Indeed, we see a significant increase in local synchrony in cells with randomized and closer nuclear spacing due to diminished dynein expression (Figure 5C, Table 2, Movies S5–7). Mitoses are more likely to be adjacent to one another even though the overall proportion of dividing nuclei of this mutant strain is not substantially different from WT (Table 2). In these cells, multiple neighboring nuclei are seen to divide at the same time and we observe “runs” of mitoses within a few timepoint (Figure 5C). Immunofluorescence quantification shows that there are frequent runs of nuclei in the same cell cycle phase, with lengths of up to 12 synchronized nuclei in a row. This is in contrast to wild-type cells that have fewer runs of synchronized nuclei and the frequency of these runs are consistent with what is expected by chance (Figure 5D). These data suggest that non-random nuclear spacing is a key component of asynchronous nuclear division.
Nuclear spacing functions to promote autonomy in nuclei yet there is still a large amount of variation in timing across the overall population (Figure 1A). We next sought to address if the source of timing variability is stochastic or systematic in nature. If there is some heritable or systematic source of timing variation, it should reveal itself as a timing relationship within lineages of related nuclei. Alternatively, if timing is purely stochastic, then related nuclei will have no relationship in their timing. Statistical analysis of division timing differences within and between nuclear lineages can be used to identify the existence of inherited sources of division timing variability (Figure 6A).
We examined division timing in 32 pairs of sister nuclei, born of one mitosis, from 7 movies (Figure 6B, Movies S1–S3). To examine sister timing relationships we compared the difference in nuclear division cycle durations of sister nuclei to a distribution of randomly selected pairs of nuclei. The mean difference between sister nuclear cycle durations was only 23±22min compared to 41±37min for the randomized control indicating a significant degree of inherited variation, as sisters were more similar in timing than expected by chance (Figure 6B, Table 3, p<0.05, 2 sample t-test and K-S test). As an alternative method, we used a non-parametric rank-based statistical test to determine how sisters are related in division times and find a clear positive association (Table 3; ZR=2.83, p<0.01). This also indicates that sister division times are more similar to one another than compared to the entire population of division times. Although statistically similar, sisters still have different absolute times indicating they are not perfectly synchronized but rather more similar to one another in timing than other unrelated nuclei. These data support that a proportion of timing variability arises from a trait that is inherited by sister nuclei at mitosis and differs among the population of mitotic nuclei.
To assess if the timing association between sisters decays as they travel far apart, we examined how far sister nuclei move away from one another after mitosis. We found that most sisters traveled away from one another with greater than 53% of sister pairs at least 10μm (~2–3 nuclei) apart 75 minutes after their birth (Figure 6C). Additionally, sisters that travel far apart are comparably related in timing as sisters that stay in relatively close proximity indicating that the sister timing relationship is not strongly related to how close together they are in their next cycle at least on the scale of tens of microns (t-test, time difference of sister pairs <10μm compared to sisters >10μm apart, p<0.18). Thus, inherited similarity in division timing is retained even when sisters are physically far apart. This inherited timing similarity is an especially robust timing determinant because it is detectable in spite of the fact that sisters undergo several bypassing events with nuclei of other lineages that would presumably diminish their association in timing (Figure 5B). This suggests that nuclei inherit regulators of division and the fact that timing is carried across distances supports the model that each nucleus creates functionally insulated territories within a seemingly continuous cytoplasm.
Asynchrony in division timing is a universal property of cultured cells and can be observed even among cycling nuclei in a common cytoplasm of certain syncytia. Variation and autonomy in division timing may protect against external stress, e.g. by limiting the number of nuclei during the sensitive state of DNA replication. Additionally, they may serve to maintain a consistent nuclear to cytoplasmic ratio. Therefore, there are likely active mechanisms that promote asynchrony in multinucleate contexts. We analyzed nuclear asynchrony in Ashbya by considering two different division cycle timing relationships: 1) between nuclei that are neighbors and 2) between sister nuclei that are born of one mitosis and then move far apart from one another. We hypothesized that nuclei may generate independent compartments of cytosol that foster division autonomy. Consistent with this idea, we found that nuclei in the Ashbya syncytium are non-randomly distributed and actively repel neighbors to generate nuclear territories or “cells within cells.” Cytoplasmic territories respond to the local nuclear division cycle and are likely a mechanism by which nuclear density is coordinated with overall cell growth. Thus, active nuclear positioning promotes nuclear autonomy and asynchronous division of neighboring nuclei.
Nuclear spacing is critical in many organisms for diverse cellular functions and cellular organization . We have evidence that nuclei actively control their spacing relative to neighbors in Ashbya using nuclear repulsion and microtubules (Figure 2–3). Previous work in Ashbya has shown that nuclei utilize the microtubule cytoskeleton to fluctuate and bypass within the common cytoplasm [23, 25, 28, 29]. In addition, small datasets have shown neighboring nuclei moving close to one another and then rapidly moving apart and being carried by cytoplasmic flow . Our analyses of large populations of dynamic nuclei support these findings and indicate that individual nuclei create territories by controlling local cytoplasmic spacing. We observe nuclear positions fluctuate about a mean position and nuclei are pushed back to the mean on a time scale of ~2.2 minutes by a microtubule-dependent mechanism (Figure 3). Similar nuclear fluctuations, which begin after nuclear fusion in meiosis in S. pombe, are controlled by dynein motors. The asymmetric loading of dynein and the dynamic redistribution of dynein on microtubules due to load forces facilitate these oscillatory movements. These horsetail oscillations occur with a ~10 minute period and span a 10μm distance which is a relatively similar time and length scale as we see in Ashbya . We speculate that spatially variable regulation of dynein localization and activity in Ashbya is likely the basis of nuclear repulsion and non-random nuclear positioning.
The observation that local nuclear spacing increases with progression through the nuclear division cycle suggests that the nuclear division cycle is in fact able to act on cytoplasmic targets to regulate local nuclear crowding. Local nuclear division cycle regulation of microtubule-associated proteins or motors may function to alter local nuclear spacing as nuclei progress through their cycle. Overlapping microtubules emanating from neighboring nuclei may be responsible for nuclear repulsion . Short cytoplasmic microtubules are thought to generate forces that may be resisted by some other component of the cytosol that perhaps changes stiffness with the cell cycle and allows for MT motors to work [23, 29, 32]. Interestingly, there is evidence that aster-aster interaction zones, such as those seen in the early divisions of large cells such as Zebrafish and Xenopus embryos, are also spaced apart depending on when in the cell cycle the asters meet. This spacing is speculated to be based on dynein activity from molecules anchored on the cytosol [33, 34]. Regulation of motor activity may enhance the pushing apart of neighboring nuclei in preparation for mitosis in Ashbya and several motors examined in our study have consensus CDK-phosphorylation sites. The identity of the cytosolic substrates reacting to this regulation in Ashbya and how this regulation would be spatially restricted to a zone of single nucleus requires further investigation.
There is mounting evidence for cytoplasmic organization within many syncytia. Crosses of Neurospora crassa “banana” mutants generate one large multinucleate ascospore with a genetically mixed population of nuclei. When banana mutant crosses involve one parent strain expressing GFP, those nuclei that encode GFP have increased GFP localization compared to nuclei from the parent without the fluorescent label encoded even though the nuclei reside in a common cytoplasm. This pattern is accentuated after several mitotic divisions resulting in a gradient of GFP intensity from one end of the spore to the other, a distance of approximately 100μm . Nuclear-based cytoplasmic organization has also been observed in the Drosophila syncytial embryo that contains thousands of nuclei in one cytoplasm and, prior to cellularization, endomembranes (ER and Golgi) are organized to create functionally distinct units around individual nuclei [36–38]. These examples, when combined with this study, support that nuclei can insulate themselves within a common cytoplasm.
The notion that individual nuclei in a syncytium may be in conflict or competition, which we observe as the repulsion of nuclear neighbors from one another, is well documented on the genetic level. Individual genomes in filamentous fungi can give rise to new individuals through asexual spore formation and there is substantial thought and interest considering the role of nuclear migration and position in the potential for intracellular genome competition . Our ability to detect independence of nuclear movement coupled to functional insulation adds support to the idea that there can be functionally relevant genome competition even in a common cytoplasm.
While mitosis in Ashbya is not restricted in time and space and overall division time is highly variable (Figure 1), sister nuclei have related division timing even when travelling apart after their birth (Figure 6). These surprising results suggest that timing variability arises from a trait that is inherited by sister nuclei at mitosis and differs among the population of mitotic nuclei. Due to limitations of phototoxicity and tracking, it is not possible to determine from these data if similar division times persist in a lineage over many generations. However, if timing were inherited consistently over multiple generations, we predict that the mean division timing for a population of nuclei would decrease as the cells age. In fact, we observe that mean division timing increases as the cells age, suggesting that the persistence of division timing within a lineage of related nuclei is short (data not shown). The molecular basis for the shared timing behavior is not yet clear but could lie in the transcriptional state, ploidy, and/or the distribution of nuclear pore complexes which are all traits that are known to vary between nuclei in Ashbya (unpublished data).
As in uninucleate cells, one of the sources of timing variation in Ashbya nuclear division cycles is the size of the local cytoplasm [17, 40–44]. There is evidence for local size control working in the multinucleate context of fused onion root cells, where the nuclei with persistent access to a larger area of the cytoplasm progress through prophase earlier than those that are more crowded . The modest correlation between nuclear spacing and cell cycle progression we observe in Ashbya suggests that territory size may have some influence on division timing (Figure S3). It is likely that local nuclear spacing is more clearly important for the duration of specific phases of the cell cycle; our data analyzes complete division cycle times because G1 and G2 durations are unknown in this dataset. Thus strong evidence for size control of cell cycle progression, particularly early in the division cycle, may be obscured by sources of timing variability acting in other phases of the cycle.
Importantly, we found that large alterations in nuclear spacing clearly resulted in an increased mitosis rate (Figure 4C,D). This suggests that the cell is able to sense and respond to the amount of cytoplasm associated with each nucleus. Prior to this work, genetic evidence for controlling the amount of cytoplasm per nucleus in Ashbya included the fact that the internuclear distances get smaller in certain cell cycle mutants (such as whi5) known to accelerate G1 . This change in nuclear spacing in Ashbya mutants is analogous to budding and fission yeast mutants altering G1/S and G2/M control, respectively, leading to overly small or large cells [46–49]. While the relative amount of cytoplasm around a nucleus can contribute to nuclear division cycle timing, the mechanism for such size control is unclear. Some mechanisms proposed for budding yeast to measure size, such as the measurement of local protein synthesis rate may be applicable in both uni-nucleate and syncytial cells .
Nuclear positioning allows Ashbya to create “cells within cells” to foster autonomy. We observed that altering the nuclear spacing results in increased synchrony across the cell. When nuclei are more randomly spaced, neighboring nuclei are more likely to be in the same cell cycle stage and are seen to undergo mitosis at the same time (Figure 5, Table 2, Movies S5–7). This suggests that nuclei are no longer able to compartmentalize themselves relative to their neighbors and are potentially more able to share diffusing signaling molecules. Supporting this hypothesis, even in wild-type cells we see that nuclei that come very close together are more likely to divide at the same moment of time than nuclei that are spaced apart (Figure 5). Thus, non-random nuclear spacing is critical for cell cycle independence within the syncytium.
Given that the cell cycle machinery acts in the cytoplasm, as evidenced by spacing increasing with nuclear progression, how are nuclear territories supporting division autonomy? What is the basis for individuality of nuclear compartments within a single cell? Proteins must be translated in the common cytoplasm and yet be restricted to act in or near individual nuclear territories. We have evidence that some cyclin transcripts are preferentially sequestered near nuclei and this could lead to local translation and influence over the most proximal nucleus . The cell biological basis of nuclear autonomy is fascinating and future work will assess the degree to which proteins and transcripts can be shared among neighboring nuclei, as well as additional mechanisms that are acting within nuclear territories to promote nuclear autonomy.
Ashbya gossypii media, culturing, and transformation conditions were performed as described previously [51, 52]. Details on strain construction and preparation of cells for imaging are provided in supplemental methods.
Time-lapse imaging was performed using a Zeiss Axioimage-M1 upright light microscope (Carl Zeiss, Jena, Germany) equipped with a Plan-Apochromat 63X/1.4NA oil objective, an Exfo X-Cite 120 lamp in conjunction with the following filters: Zeiss 38HE (GFP), Chroma 41002B (TRITC), and Zeiss 49 (Hoechst). Images were acquired on an Orca-AG charge-coupled device camera (C4742-80-12AG; Hamamatsu, Bridgewater, NJ) driven by OpenLab 5 (Improvision, Lexington, MA) or μManager (NIH, ). Acquisition and processing details are in supplemental methods.
The position of each nucleus at each timepoint was tracked and coordinates were recorded in Excel for three timelapse movies using the measurements tool in Volocity 5 (Improvision). Four additional movies were tracked for lineages. Subsequent spatial analyses were done in Excel and in MATLAB (see below). All statistical tests were performed in Excel or R (Version 2.12.2). Data plotting was all done in R.
To determine non-random spacing of nuclei, the internuclear distance was measured for nuclei in each strain. The null, `randomized' model was created by a bootstrap procedure: using the measured distances, the same number of nuclei was randomly simulated 100 times per strain in MATLAB within the same cell area and the randomized nuclear distances were calculated. To indicate the degree to which the observed nuclear distances differ from random, a metric was calculated for each random simulation. This was based on the total area between the observed CDF and the simulated random CDF (randA), as well as the total area between the observed CDF and a CDF representing constant spacing (constA). The utilized metric was then calculated as 100*randA / (randA + constA) to yield a measure between 0 and 100%. Values near 100% indicate more uniform spacing, while values near 0% indicate spacing that is closer to random.
To determine if close proximity between nuclei resulted in similarity in the timepoint of division, the distance between all tracked nuclei was calculated over time. By-passing nuclei were those that came within 0.25μm or less within their lifetime. For each observed mitosis, all by-passing nuclei were identified and the average division timepoint of those nuclei was calculated. The absolute value of that average and the observed division timepoint of the reference nucleus was calculated. Neighboring nuclei were considered those that spent more than 30 timepoints within 2–5μm of each other, excluding those nuclei that underwent by-passing events. The same analyses were done for by-passing nuclei were done using neighboring nuclei.
Description of methods used for strain construction, image processing, nocodazole arrest and release, nuclear tracking, calculating the synchrony index and statistical analysis of sister division time relationships can be found in the supplemental methods.
We would like to thank Hany Farid for computational assistance, the Gladfelter lab, Christine Field and Tim Mitchison for discussion and Jamie Moseley and Roger Sloboda for useful comments on the manuscript. We are grateful to MicroVideo Instruments (MVI) and Nikon for supporting our instrument needs at MBL in Woods Hole, MA. This work was supported by NIH R01-GM081506 (ASG), Lemann and Colwin fellowships (ASG) from the Marine Biological Laboratory in Woods Hole, NIH award T32GM008704 (CAA), NIH RO1-GM092925 (JS) and the Burroughs Wellcome Fund (JS).