This study establishes that MIFC analysis of budding yeast represents a valid method for determining a complete cell cycle profile in budding yeast. Unlike traditional flow cytometry, MIFC makes it possible to detect minor perturbations in the cell cycle without requiring synchronization by cell cycle arrest and release. Gene expression studies have shown that different methods for cell cycle arrest, such as nocodazole treatment, elutriation, α-factor pheromone treatment and temperature-sensitive mutant strains, yield inconsistent results (23
). Additionally, specific strains are hypersensitive to chemical agents used for synchronization and one arrest technique may not be suitable for analyses with different strains (26
). Thus, one of the primary advantages of MIFC analysis as presented here is it allows a detailed cell cycle profile to be determined without the additional manipulations required for arrest and release. Visual analysis of the cell cycle does not require synchronization, but is very time consuming; obtaining statistically significant verification of minor defects can be difficult, due to limited sample sizes. This technique is also more prone to subjective errors. MIFC provides the means to objectively evaluate large numbers of cells using both flow cytometry and morphological criteria.
We find that quantification of the cell cycle using standard flow cytometry coupled to standard modeling programs overestimates the proportion of cells in the G2/M phases of the cell cycle and underestimates those in G1, relative to both visual analysis and MIFC. Interestingly, overestimation of the predicted proportion of G2/M cells was a systematic error identified in the original assessment of the Watson Pragmatic model (20
). This may be exacerbated in yeast, since a lack of concise separation of late stage G2/M and recently divided cells (doublets) would also lead to an overestimation of G2/M by traditional flow cytometry. Although the analytical parameters differ, both flow cytometry software models fit S-phase to the trough between the G1 and G2/M peaks and attribute some of the area beneath each peak to S-phase. Our findings suggest that both models may assign a proportion of cells to G2/M that are still actively replicating DNA, and should thus be considered in S-phase. Similarly, underestimation of G1 by flow cytometry suggests that a portion of cells beneath the peak corresponding to 1N DNA are being allocated to S phase.
DNA content is measured along a continuum, and thus any algorithm must contain simulated peak boundaries for calculating S phase. This presents a caveat when more complex cell cycle analysis is required, though is not a concern for straightforward analyses in which comparison of relative DNA content (i.e. 1N or 2N) is sufficient. It is important to note that both the Watson Pragmatic model and ModFit LT were designed and evaluated using mammalian cells. Budding yeast exhibit relatively large coefficients of variance (CVs) in fluorescent DNA stain in comparison to mammalian cells (28
). The small genome size of yeast renders it more sensitive to variations in cell number and DNA stain concentration between samples, and the asymmetrical cellular morphology of budding yeast can also contribute to the increased variance observed. Another reason for this is that yeast are particularly sensitive to changes in the sheath pressure during flow, and these fluctuations can cause shifts in the fluorescent peak. (28
). It is therefore feasible that existing cell cycle models based on the mammalian cells could be modified in order to account for these yeast-specific disparities. Other methods for monitoring cell cycle progression by flow cytometry, such as simultaneously tracking protein and DNA content, can also provide better resolution of the cell cycle in heterogeneous yeast populations (30
In comparing the different methods for determining cell cycle, it must be considered that in the MIFC analysis cell populations are not segregated according to DNA content alone, the standard criteria used in traditional flow cytometric cell cycle analysis. Using MIFC we consider multiple criteria, and this provides a more precise demarcation of cell cycle stage. Nuclear elongation is employed here as a marker for the G2/M transition and this is one of a number of morphological events that occur between late S phase and the onset of anaphase. Other events, such as mitotic spindle formation or the migration of the nucleus to the bud neck, could also be easily applied to MIFC analysis.
Using MIFC analysis we confirmed that nap1Δ cells exhibit an almost two-fold increase in the proportion of cells in G2/M of the cycle, indicative of a delayed exit from mitosis, and this is coupled with a shortening of both G1 and S. In cells overexpressing NAP1, this phenotype was less pronounced; the proportion of cells in G2/M increased by about one-third relative to wild-type, and this strain also demonstrated shortened G1 and S phases. These minor perturbations in the cell cycle would not have been detected by traditional flow cytometry, and the small differences would have required very large sample sizes to be statistically significant by purely visual analysis.
The algorithm we developed in order to measure bud length was determined to be a reliable method for evaluating numbers of elongated buds within a population, as tested by previously published criteria. This confirms the utility of MIFC analysis to quantify morphological phenotypes in budding yeast. By categorizing elongated cells as those in which bud length is greater than 1.5 times the width, we determined that nap1Δ cells showed greater than a three-fold increase in elongated buds relative to wild-type yeast whereas in cells overexpressing NAP1, this increase was more than six-fold. This was in agreement with previous unpublished observations that overexpression of NAP1 causes in increase in severity of the elongated bud phenotype relative to wild-type.
It was of particular interest that in cells overexpressing NAP1
, the number of cells with elongated buds was higher than in nap1Δ
cells, yet the G2/M delay was less pronounced. This implies that the switch from polar to isotropic bud growth is more impaired when excess Nap1 is present than in its absence. Conversely, the delay in G2/M is more severe in the absence of Nap1, and yet fewer of the cells in G2/M exhibit defects in isotropic bud growth. Therefore, the regulation of bud growth and mitotic progression are both affected by expression levels of NAP1
, yet the effects are at least somewhat independent. A specific role in mitosis for Nap1 has not been defined, though it interacts with the mitotic cyclin Clb2 (32
). Many of the events of mitosis require the activation of the cyclin-dependent kinase Cdc28 by Clb2, including the regulation of bud growth. Nap1 also interacts with the kinase Gin4, though apparently in an independent complex since Gin4 and Clb2 are not detected together (32
). Interestingly, deletion of the genes for both NAP1
in wild-type cells causes a more severe elongated bud phenotype than either single deletion, whereas in a CLB2
-dependent background the phenotypes of the double and single mutants are equivalent (33
). Taken together, these results imply that Nap1 and Gin4 share both Clb2-dependent and independent functions in regulating bud growth. It is possible that excess Nap1 impairs the regulation of bud growth by Gin4 without significantly affecting mitotic progression (perhaps by outcompeting other Gin4-interacting proteins), whereas total absence of Nap1 is damaging to both processes. Characterization of other mutants involved in the control of bud growth during mitosis should help elucidate this. In summary, this study validates the use of MIFC in performing tandem cell cycle and morphological analyses on budding yeast. As we have demonstrated, this technique proves particularly useful for characterizing cells in which cell cycle defects have become uncoupled from associated morphological phenotypes.