Nitrogen availability and TOR signalling are important for preventing catastrophic mitosis in fission yeast

Mitosis is a critical stage in the cell cycle, controlled by a vast network of regulators responding to multiple internal and external factors. The fission yeast Schizosaccharomyces pombe may demonstrate catastrophic mitotic phenotypes due to mutations or drug treatments. One of the factors provoking catastrophic mitosis is a disturbed lipid metabolism, resulting from e.g. mutations in acetyl-CoA/biotin carboxylase (cut6), in fatty acid synthase (fas2/lsd1), or in the transcriptional regulator of lipid metabolism (cbf11) genes, as well as treatment with inhibitors of fatty acid synthesis. It was previously shown that mitotic fidelity in lipid metabolism mutants can be partially rescued by ammonium chloride. In this study we demonstrate that mitotic fidelity can be improved by multiple good nitrogen sources. Moreover, this rescue is not limited to lipid metabolism disturbances but also applies to a number of unrelated mitotic mutants. Interestingly, the rescue is not achieved by restoring the lipid metabolism state, but rather indirectly. We found that the TOR regulatory network plays a major role in mediating such rescue, highlighting a novel role for TOR in mitotic fidelity.


INTRODUCTION
Mitosis is a critical stage in the cell cycle of any eukaryotic cell.Irregularities within this process may have far-reaching consequences, such as aneuploidy, mutations and/or cell death.These, in turn, can result in uncontrolled proliferation and tumour formation in metazoans [1], or cause decreased fitness in populations of unicellular organisms, such as yeasts [2].Because of its significance, the entry into mitosis and progression through mitotic phases are tightly regulated by a network of redundant regulatory pathways, creating a fail-safe mechanism of control [3].This network also integrates numerous internal and external factors that influence the decisions to proceed to the next phase.In yeasts, cell-cycle regulation is strongly affected by the availability of nutrients, such as carbon and nitrogen [4].
In contrast to the open mitosis of higher eukaryotes, many unicellular species undergo closed mitosis.The nuclear envelope (NE) remains intact during the whole process of a closed mitosis, even though substantial NE remodelling is required [5].For example, in the fission yeast Schizosaccharomyces pombe the NE surface area needs to expand by 33% during mitosis to properly accommodate the elongating spindle and segregating chromosomes [6].Notably, a number of abnormal mitotic phenotypes associated with closed mitosis have been described.These include: 'cut' -cell untimely torn, a form of mitotic catastrophe where cytokinesis takes place before nuclear division has been properly resolved, and the mother nucleus gets transected by and trapped in the forming septum (Fig. 1 A) [7] [8]; 'lsd' -large and small daughters, a segregation error where the resulting daughter nuclei are of unequal sizes [9]; and nucleus displacement, which is similar to 'cut', but the forming septum misses the nucleus, resulting in one of the daughter cell being anucleate and the other one containing a diploid nucleus [10].Typically, such events are lethal for at least one of the daughter cells.The 'cut' phenotype has been described for mutations in a range of mitosis-related genes, including the separase (cut1) and securin (cut2) [11], condensin (cut3) [12], anaphase-promoting complex (APC/C; cut4, cut9) [13], and others [14].
Interestingly, perturbations of lipid metabolism can also cause mitotic catastrophe in S. pombe [15].It was shown that mutations in acetyl-CoA/biotin carboxylase (cut6), fatty acid (FA) synthase (fas2/lsd1) [9], or the CSL transcriptional regulator of lipid metabolism (cbf11) genes [16], as well as treatment with FA synthesis inhibitors cerulenin [9] or Cutin-1 [17] lead to the 'cut' and/or 'lsd' phenotypes.It was assumed that decreased supply of precursors of membrane phospholipids (PL) leads to insufficient NE expansion during anaphase and mitotic failure [17].However, we recently found additional factors contributing to decreased mitotic fidelity in cells with perturbed lipid metabolism, as cbf11 and cut6 mutants show altered cohesin dynamics and H3K9 modifications at the centromeric regions [18].
Nitrogen is an important macronutrient required for the synthesis of amino acids, nucleotides, and many other biomolecules, and the availability of nitrogen has a profound effect on the timing of entry into mitosis [19].To coordinate growth and division with available resources the cell employs several nutrient-responsive regulatory pathways.In S. pombe, these include the protein kinase A (PKA), the stress-response MAP kinase Sty1, and, most prominently, the target of rapamycin (TOR) kinases [20].Notably, similar mechanisms operate in mammalian cells as well [21].There are two TOR complexes (TORCs) in S. pombe, each containing a different TOR kinase paralog.TORC1 (Tor2 kinase) is a major regulator of nitrogen metabolism and it stimulates growth and proliferation.On the other hand, TORC2 (Tor1 kinase) is involved in stress responses and maintenance of genome integrity.Importantly, there is crosstalk within the TOR network and the two TORCs operate in an antagonistic manner [22] [23].Moreover, apart from the quantity of available nitrogen, the exact chemical nature (quality) of a nitrogenous substance must be taken into account.Some, like ammonium and glutamate, are classified as 'good' sources that can be utilised easily and efficiently by the cells, while others, like proline, are ranked as 'poor' sources, in spite of them all containing equal amounts of nitrogen atoms per molecule [24] [25].
Intriguingly, we previously demonstrated that the incidence of mitotic catastrophe in the cbf11 and cut6 lipid metabolism mutants is suppressed when cells are grown in the minimal defined EMM medium, which contains abundant ammonium chloride as the nitrogen source [16].A similar rescue effect was achieved by growing mutant cells in the complex (and relatively nitrogen-poor) YES medium supplemented with ammonium chloride [26].We now show that this rescue of mitotic fidelity is indeed nitrogen-dependent and is mediated by the TOR network.Unexpectedly, we found that nitrogen supplementation does not restore the expression of lipid-metabolism genes or lipid composition in Δcbf11 cells, nor does it restore NE expansion during mitosis, indicating an indirect nature of the rescue.Moreover, we demonstrate that nitrogen supplementation can also rescue a range of other, non-lipid 'cut' mutants, suggesting a more general effect of nitrogen on mitotic fidelity.Our results highlight a previously unappreciated role of the TOR network in successful progression through mitosis.

Strains, media and cultivation
Standard methods and media were used for the cultivation of Schizosaccharomyces pombe strains [27].YES medium was prepared using Bacto Yeast Extract (BD Biosciences) and SP Supplements (Formedium).EMM medium was prepared using EMM Broth without Nitrogen (EMM-N; Formedium).Ammonium chloride (Sigma) or ammonium sulphate (Lachema/Chempol) were added to EMM-N to the final concentration of 93 mM, unless indicated otherwise.L-Glutamic acid monosodium hydrate, L-Proline or Uracil (Sigma) were added to the final concentrations from 20 to 100 mM, as required.S. pombe cell cultures were pre-grown for 8 hours at 32°C (25°C for temperature-sensitive strains) in 5 ml YES, then transferred to the medium with the appropriate supplements or stressors, diluted to OD 600 =0.005 (WT) or ~0.03-0.05(mutants) and incubated overnight to the early exponential phase (OD 600 =0.5) at 32°C or at appropriate semi-restrictive temperature.Cerulenin (Abcam) was added 2 hours before harvesting to the final concentration of 10 μg/ml.Rapamycin (Merck) was added at the time of starting the final culture to the final concentration of 0.3 μg/ml.Routine optical density (OD) measurements of liquid cell cultures were taken using the WPA CO 8000 Cell Density Meter (Biochrom).The strains used in the study are listed in Table S1.Deletion of cbf11 was carried out using the pMP91 or pMP92 targeting plasmid, based on pCloneNAT1 or pCloneHYG1 respectively, as described [28] and confirmed by PCR.All other strains used in this study were constructed by standard genetic crosses.Plasmids and oligonucleotides used in this study are listed in Table S2 and Table S3 respectively.

Microscopy
For nuclear staining, exponentially growing S. pombe cells were pelleted by centrifugation (1000 g, 3 min), fixed in 70% ethanol and stored at 4°C prior to imaging.Then cells were rehydrated in water and stained with 4′,6-diamidino-2-phenylindole (DAPI), 0.1 μg/ml final concentration.Samples were analysed using a Leica DM750 microscope with HC FL PLAN 100x/1.25 OIL objective.For each sample at least ten images (~500-1500 cells total) were acquired.Frequency of catastrophic mitosis was counted manually using our standard scoring criteria [18].
For live-cell microscopy 1 μl of slightly resuspended cell pellet was applied on 2% agarose-YES media solidified in a 2 mm PDMS spacer [29], and covered with a coverslip.Slides were placed into an OKOlab environmental chamber set to 32°C.Time-lapses were acquired using Nikon Ti2 microscope with Plan Apo Lambda 60x Oil objective coupled with Hamamatsu ORCA-Flash4.0camera in 16-bit, 2x2 binning mode.Snapshots were taken at 2 minute intervals.Images were acquired as Z-stacks, 5 to 9 slices with 0.3 μm step.Fluorophores were excited using CoolLED pE-4000 device.GFP fluorophore was filmed using 460 nm excitation at 21% light power, 100 (Cut11-GFP) or 50 ms (Hht2-GFP) exposure time and 510 nm emission filter.mCherry fluorophore was filmed using 580 nm excitation at 20% power, 200 nm exposure time and 590 nm emission filter.

RNA-seq
Samples were prepared from 3 biological replicates.Cells were cultured to exponential phase, 10 mL were harvested (600 g, 2 min) and the cell pellet was flash-frozen with liquid nitrogen.Total RNA was isolated using hot acidic phenol method followed by phenol-chloroform extractions and precipitation [33].Extracted RNA was treated with TURBO DNase (Thermo Fisher Scientific) and purified using RNeasy columns (Qiagen).RNA quality was assessed on Bioanalyzer 2100 (Agilent).Detailed sample preparation protocol is available at ArrayExpress database (see "Data availability").
WT and Δcbf11 samples were processed at the Institute of Clinical Molecular Biology, Christian-Albrechts-University Kiel, Germany.Libraries were prepared using the Illumina TruSeq stranded mRNA Library (poly-A), and sequenced in the pair-end mode on an Illumina NovaSeq 6000 instrument with the NovaSeq 6000 SP Reagent Kit with 100 cycles.

Lipid analysis
Samples were prepared from 3 biological replicates.Cells were cultured to exponential phase in YES, 50 mL were harvested (1000 g, 3 min).Cell pellets were flash-frozen with liquid nitrogen and freeze-dried using HyperCOOL 3055 device (Gyrozen).
Total lipids were extracted using a method previously reported with minor modifications [47].Briefly, dry cell weight (DCW) was determined gravimetrically prior to lipid extraction.Freeze-dried cells (approximately 10-15 mg) were mixed with 100 µL of ice-cold water followed by addition of 1 mL chloroform and methanol mixture (2:1, v/v).Cells were disrupted by FastPrep disintegrator (MP Biomedicals) with glass beads (diameter 0.4 mm, 3x40 s at the highest speed, with 5 min cooling on ice between cycles).Lipids were extracted by incubating in chloroform/methanol/water (1:2:0.8,v/v) and subsequently adjusting the mixture proportion to 2:2:1.8 (v/v) at room temperature.The organic phase containing the lipids was separated by centrifugation and dried under a stream of nitrogen.The resulting dry lipids were dissolved in 100 µL of chloroform and methanol mixture (2:1, v/v).
For fatty acid analysis total lipid extracts were transmethylated with 5% Na-OCH 3 in methanol.Fatty acid methyl esters (FAME) were then extracted using n-hexane as described previously [48].The analysis of FAME involved injecting 1 μL aliquots into a GC2010Plus gas chromatography (GC) apparatus (Shimadzu) equipped with a BPX70 capillary column (30 m × 0.25 mm × 0.25 µm, SGE Analytical Science) as described previously [47] [49].Identification of individual FAMEs was accomplished by comparing them with authentic standards of a C 4 −C 24 FAME mixture (Supelco).The quantification of individual fatty acids was conducted using heptadecanoic acid methyl ester as an internal standard (Sigma Aldrich).
For the thin layer chromatography (TCL) an aliquot of lipid extract corresponding to 8 mg of DCW was applied to silica gel TLC plates (Merck) by a Linomat 5 semiautomatic sample applicator (Camag).Neutral lipids were separated by a two-step TLC solvent system using a method described previously (first step: petroleum ether/diethyl ether/acetic acid, 70:30:2; second step: petroleum ether and diethyl ether, 49:1) [50].Individual lipid spots were visualised by charring the plates as previously reported [51].Individual lipid spots were identified using lipid standards.Phospholipids were separated by the solvent system (chloroform/methanol/acetic acid/water, 75:45:3:1) as described previously [52].

RESULTS
The mitotic defects of Δcbf11 cells can be rescued by a good nitrogen source.We have shown previously that mitotic defects in fission yeast lipid metabolism 'cut' mutants (Fig. 1 A) can be rescued by the addition of ammonium chloride to culture media [26].As the first step toward understanding this phenomenon we compared the frequency of mitotic catastrophe in Δcbf11 cells in the standard complex YES medium, and in the minimal defined EMM medium supplemented with ammonium chloride or ammonium sulphate.We found that the magnitude of the rescue effect was similar between the two ammonium salts (Fig. 1 B), indicating that the previously observed rescue had been caused by the ammonium, not the chloride.
Different nitrogen-containing compounds can be utilised by the cell with varying ease and efficiency ('good' vs 'poor' nitrogen sources) [24] [25].Therefore, we next tested several nitrogen-containing compounds for their ability to suppress mitotic defects in Δcbf11 cells.We found that ammonium, a prototypical good nitrogen source, had the most profound effect, suppressing mitotic defects even at reduced concentrations (Fig. 1 C).Glutamate, another good nitrogen source, could also improve mitotic fidelity, albeit only at higher concentration (Fig. 1 D).On the other hand, the poor nitrogen sources proline and uracil did not suppress the mitotic defects of Δcbf11 cells at the concentrations tested (Fig. 1 E).Thus, the magnitude of the rescue effect correlates with the preferability of a given compound as a nitrogen source to fission yeast cells [24] [25].
Mitotic catastrophe can also be triggered by chemical inhibition of the fatty acid synthase (FAS).This is thought to be caused by insufficient supply of membrane building blocks for NE expansion during the anaphase of closed mitosis [6] [17].We found that the frequency of mitotic catastrophe in wild-type (WT) treated with the FAS inhibitor cerulenin [9] can also be partially rescued by ammonium supplementation (Fig. 1 F).This finding thus seems to be compatible with the hypothesis of membrane building block shortage being responsible for mitotic defects upon lipid metabolism disturbances [6] [17].

Ammonium does not suppress perturbations of lipid metabolism.
To investigate whether ammonium supplementation suppresses mitotic defects via stimulating production of new membranes, we performed measurements of NE expansion during mitosis using live-cell microscopy.We used a fluorescently tagged nuclear pore protein, Cut11-GFP, as a NE marker [53].When comparing Δcbf11 and WT cells we found that the mutant cells have smaller NE cross-section, both at the start of mitosis and at the moment of maximum NE expansion.Importantly, we did not observe any differences between cultures grown with or without ammonium supplementation (Fig. 2).This indicates that the nitrogen-dependent rescue of mitotic fidelity is not achieved by enhancing the anaphase NE expansion.Indeed, we previously showed that ammonium supplementation did not correct the aberrant lipid droplet (LD) content of Δcbf11 cells [26].Also, we demonstrated that the defects in Δcbf11 mitosis manifest already before anaphase [18].
To corroborate our new findings, we performed an RNA-seq analysis of a panel of lipid-metabolism mutants to assess the effect of nitrogen on the transcriptome.Our panel included the transcriptional regulators Cbf11 (Δcbf11 and cbf11DBM, a mutant with abolished binding to DNA [54]), and Mga2 (Δmga2) [55], and the FA synthesis rate-limiting enzyme Cut6 (Pcut6MUT showing 50% reduction of cut6 transcript levels [56]).As a control, we treated WT cells with the FA synthesis inhibitor cerulenin.We found that a number of lipid metabolism genes were downregulated in the Δcbf11, Δmga2, and Δmga2 Δcbf11 double mutant, as well as in the cbf11DBM strain (Fig. 3).These genes include e.g.cut6, fas1/2, the acyl-coA desaturase ole1, and the long-chain-fatty-acid-CoA ligases lcf1/2.This is in agreement with the previously published studies of Δcbf11 and Δmga2 transcriptomes [16] [55].In contrast to the regulator mutants, targeted inhibition of FA synthesis (Pcut6MUT, cerulenin treatment) resulted in modestly increased expression of lipid metabolism genes, likely as a feedback reaction to FA shortage (Fig. 3).Importantly, lipid gene expression was not restored in regulator mutants grown in a medium supplemented with ammonium chloride (Fig. 3).This shows that the rescue of mitotic fidelity by a good nitrogen source is unlikely to be achieved by boosting the expression of lipid metabolism genes.
To further test for any potential effects of a good nitrogen source on the lipid metabolism in mitotic mutants, we analysed the total lipid content of Δcbf11 and WT cells using thin layer chromatography (TLC) and FA composition by gas chromatography.The results showed that the total amount of FA is decreased in Δcbf11 cells (Fig. 4 A), which correlates well with our previous data on decreased number of storage LDs in this strain [16].Also, total FA saturation level is higher in Δcbf11 cells compared to WT (Fig. 4 B), which may be caused by the decreased expression of the ole1 desaturase gene (Fig. 3).This tendency towards higher FA saturation is also visible at the level of individual FA species (Fig. 4 C; compare C18:1 with C18:0 and C16:0).Finally, the TLC analysis revealed that the content of squalene and sterol esters is markedly increased in Δcbf11 cells (Fig. 4 D, Fig. S1), suggesting failed coordination between the sterol and the triglyceride lipid metabolism pathways [57].Notably, none of the detected changes in the Δcbf11 FA and lipid composition were reversed by ammonium supplementation (Fig. 4, Fig. S1, Fig. S2).Taken together, our results strongly suggest that a good nitrogen source does not rectify the disturbed lipid metabolism in Δcbf11 cells, and therefore it likely brings about the mitotic rescue in a different, indirect way.

Ammonium has a general positive effect on mitotic fidelity.
We previously showed that the duration of mitotic phases in Δcbf11 cells is typically longer and more variable compared to WT cells.The delays first manifest well before the anaphase and they accumulate during the whole mitotic duration [18].We now investigated whether mitotic timing is affected by a good nitrogen source.We performed live-cell microscopy of strains with fluorescently tagged histone H3 (Hht2-GFP) and alpha-tubulin (mCherry-Atb2) to visualise the chromatin and mitotic spindle, respectively (Fig. 5 A).We found that ammonium reduced the overall length of mitosis (Fig. 5 B), and this effect manifested both during the prophase and metaphase (Fig. 5 C), as well as anaphase stages (Fig. 5 D) in Δcbf11 cells, bringing them close to WT values.Also, mitotic timing became more uniform among individual Δcbf11 cells (Fig. 5).Note that telophase duration was not included in the overall mitosis length due to technical limitations of our experimental setup.
Our results so far indicated that nitrogen may have a more general effect on mitotic fidelity.To test this hypothesis, we determined the effect of ammonium on a diverse panel of mutants, none of them being related to lipid metabolism, which are prone to develop the 'cut' phenotype with varying severity.Remarkably, ammonium supplementation significantly decreased the frequency of catastrophic mitotic events in many, but not all, of those mutants (Fig. 6 A-C).The group of rescued 'cut' mutants consisted of condensin (cut3), cohesin (psm3), and the SMC5/6 complex (smc6, nse3), a nuclear proteasome tethering factor (cut8) and a nuclear import factor (cut15).On the other hand, the mutants without significant rescue were separase (cut1) and securin (cut2), and APC/C subunits (cut4, cut9).It should be noted though that the magnitude of the rescue varies between the ammonium-responsive mutants, and we suggest possible explanations for this behaviour in the Discussion.Taken together, the ammonium-mediated rescue of mitotic defects is not limited to cells with perturbed lipid metabolism, and seems to operate early in the cell cycle, prior to anaphase.

The nitrogen-dependent rescue of mitotic fidelity is mediated by TOR.
The evolutionarily conserved TOR signalling network is a major hub for controlling nitrogen sensing and utilisation.Therefore, we tested whether and how TOR was involved in the nitrogen-dependent improvement of mitotic fidelity.First, we suppressed the activity of TOR by treating cells with the TOR inhibitor rapamycin [58].Notably, we found that rapamycin rescued mitotic fidelity in Δcbf11 cells to an extent similar to that of ammonium.Additionally, a combined treatment showed that the effects of rapamycin and ammonium were non-additive (Fig. 7 A).These results strongly indicate that both chemicals affect the same target pathway or process.
Although known primarily as a Tor2/TORC1 inhibitor [59][60] [61], rapamycin has a more general effect in S. pombe, inhibiting certain functions of Tor1/TORC2 as well [62] [63].Our next objective thus was to determine which branch of the TOR network was responsible for the rescue.To address this we first employed mutants of the non-essential tor1 gene (TORC2).We found that the temperature-sensitive tor1-D allele [22] increased mitotic fidelity of the Δcbf11 background when cells were grown at the semi-restrictive temperature of 34°C.Moreover, an even stronger rescue effect was achieved by introducing a complete deletion of tor1 into Δcbf11 cells (Fig. 7 B).It is important to note that the two TOR complexes are antagonistic and the inhibition or depletion of Tor1/TORC2 boosts the activity of Tor2/TORC1 [22] [23].
Next, we focused on the role of the essential Tor2 kinase.First, we decreased Tor2/TORC1 kinase activity by combining the Δcbf11 mutation with the temperature sensitive tor2-S allele [59].We did not observe any significant change in the mitotic fidelity at semi-restrictive temperature compared to Δcbf11 alone (Fig. S3).Then, to increase Tor2 activity we took advantage of the fact that the activity of Tor2/TORC1 is negatively regulated by the AMP-activated protein kinase (AMPK) complex [21].Therefore, we deleted the ssp2 gene, which encodes the AMPK catalytic subunit [64] [65].Indeed, the deletion of ssp2 in Δcbf11 cells led to a decrease in catastrophic mitotic events (Fig. 7 C).Collectively, these results strongly suggest that the nitrogen-dependent rescue of mitotic defects is mediated by Tor2/TORC1, and indicate a role for Tor2/TORC1 in ensuring successful progression through (closed) mitosis.

DISCUSSION
The general availability and the particular quality of a nitrogen source have long been recognized as important factors regulating progression through the cell cycle.The absence of a nitrogen source (nitrogen starvation) or a shift to a less preferable nitrogen source (nitrogen stress) inhibit cell growth, and trigger quiescence or sexual differentiation, or accelerate the entry into mitosis, respectively [4] [66].However, we recently demonstrated that nitrogen also plays a role in mitotic fidelity in cells with perturbed lipid metabolism, with ammonium chloride being able to partially rescue catastrophic mitosis phenotypes [16] [26].Intriguingly, our current data revealed that this nitrogen-dependent rescue of mitosis is not accompanied by corrections of the aberrant FA and lipid composition, hinting at an indirect rescue mechanism.Indeed, we found that the rescue effect is of a more general nature, not limited to mutants in lipid metabolism.Moreover, we showed that nitrogen availability also affects the progression through mitosis, not just the G2/M transition (mitotic entry).While the mechanistic details remain to be elucidated, we demonstrated that the effect of nitrogen is mediated by the TOR regulatory network, namely the growth-promoting Tor2/TORC1 complex.
It was previously suggested that the mitotic defects of fission yeast cells with chemically or genetically perturbed lipid metabolism are caused by insufficient NE expansion due to a shortage of membrane precursors [6] [17].It is also possible that these mitotic defects could be related to altered mechanical properties of cell membranes that are important for spindle pole body integration and other crucial steps of the closed mitosis [53].We have indeed found a number of aberrations in the composition of lipids in Δcbf11 cells grown in the complex YES medium compared to WT.These aberrations included an overall lower content of FA, which are required for the production of new membranes.However, we did not observe any notable corrections in the mRNA levels of lipid metabolism genes (Fig. 3) or in composition of FA and lipids (Fig. 4, Fig. S1, Fig. S2) in Δcbf11 cultures grown in an ammonium-supplemented YES medium, where mitotic defects are suppressed.Neither did we observe any significant improvement in NE expansion during mitosis in YES+ammonium (Fig. 2).These results are consistent with our recent findings that, in addition to any issues with the production of new membranes, mitotic fidelity in lipid metabolism mutants is affected by changes in centromeric chromatin structure and cohesin dynamics [18].While we did not detect any nitrogen-dependent improvement in mitotic NE dynamics and FA and lipid composition composition, we found that ammonium supplementation clearly normalised the timing of mitotic progression, restoring the duration of individual mitotic phases close to their WT state (Fig. 5).Strikingly, we also found that the phenomenon of nitrogen-mediated rescue of mitotic fidelity is not limited to lipid metabolism-related problems, as mitotic defects in multiple unrelated 'cut' mutants could also be suppressed by ammonium supplementation (Fig. 6).Importantly, not all tested 'cut' mutants showed responsiveness to nitrogen availability, and among those which did, the magnitude of the rescue effect varied.Notably, the rescuable strains are typically mutants in genes involved in pre-anaphase processes, while all ammonium-insensitive 'cut' mutants are related to anaphase (securin/separase, APC/C) (Fig. 6).Thus, this specificity of the rescue effect may be dictated by the particular time during which the respective 'cut' genes perform their mitosis-related functions.Taken together, the ammonium-mediated rescue of mitotic defects is not limited to cells with perturbed lipid metabolism, and it seems to operate in cell cycle phase(s) prior to anaphase.
The TOR network is known to be one of the key cell-cycle regulators, promoting or inhibiting mitotic onset according to nutrient availability [19] [67].Interestingly, a previous report hinted that TOR may also have a role later in mitosis, as the viability of several temperature-sensitive separase and securin mutants (including those used in our study) was improved by treating cells with the TOR inhibitor rapamycin or by introducing the tor2-S mutation that impairs TORC1 kinase activity.However, the authors did not specifically test whether mitotic fidelity was also improved [22].In our hands, the occurrence of mitotic defects in the cut1-206 and cut2-447 mutants was not suppressed by ammonium supplementation, a treatment we showed to have an impact similar to rapamycin treatment (Fig. 6).Neither did we observe any differences in mitotic fidelity between Δcbf11 and Δcbf11 tor2-S cells (Fig. S3).It is therefore possible that any impact of TOR on the fidelity of anaphase events is regulated separately, by mechanism(s) different from those mediating the nitrogen-dependent rescue that we report here.Indeed, the authors themselves reported that the phenotype of the tor2-S mutant manifested clearly only in the peptone-containing YPD medium and not in the ammonium-containing EMM2, and they observed the cut1/cut2 suppression under suboptimal nitrogen availability in YPD [22].
It was also reported that TOR is linked to growth phase-related changes in lipid metabolism.The switch between membrane phospholipid and storage triacylglycerol production is regulated by the lipin phosphatidic acid phosphatase [68].In the budding yeast Saccharomyces cerevisiae, TOR controls lipin activity to ensure sufficient supply of phospholipids for new membrane production during active proliferation [69].On the other hand, TOR/lipin-dependent overproduction of FA and endoplasmic reticulum membrane leads to mitotic defects and formation of micronuclei in mammals [70].Nevertheless, we did not observe any major changes in phospholipid composition in S. pombe cells grown in ammonium supplemented YES medium compared to plain YES (Fig. S2).
We showed increased mitotic fidelity in Δcbf11 cells when the stress-response branch of the TOR network was suppressed, either by ablation of Tor1/TORC2 or by boosting the activity of the pro-growth Tor2/TORC1 branch (Fig. 7).These data are in agreement with previous findings that Tor2/TORC1 inhibition mimics nitrogen starvation [71] [72].And vice versa, Tor2/TORC1 hyperactivation delays the response to nitrogen starvation [73], highlighting that the two branches of the TOR network are antagonistic and act in a negative feedback loop.Our results indicate that interventions providing cells with more, well-utilisable nitrogenous compounds (i.e.ammonium supplementation), or merely triggering internal signals mimicking the state of such nitrogen availability (i.e.boosting Tor2/TORC1 activity by ablating TORC2 or AMPK), make the mitotic defects in cells prone to catastrophic mitosis less severe.This could mean that the signalling of availability of a good nitrogen source is by itself more important for mitotic fidelity than the actual physical presence of the nutrients.However, the signalling of nutrient availability could in return affect the uptake and/or utilisation of nutrients by the cell.Additionally, it should be noted that ammonium-dependent improvements in mitotic dynamics manifested very early on during the mitosis of Δcbf11 cells (Fig. 5).It is therefore also possible that the TOR network does not act on mitosis directly during the process of nuclear division, but rather earlier in the cell cycle helps establish conditions more favourable for smooth mitotic progression and increased fidelity further down the road.
In any case, the exact mechanism of the nitrogen-mediated rescue of mitotic fidelity remains to be characterised in detail, including the regulatory level(s) on which the observed effect is achieved.Since the TOR proteins are kinases, it is likely that the rescue is mediated, at least in part, at the post-translational level by phosphorylation of downstream effector proteins.These could be identified, for example, by screening of a knock-out library [74] combined with the Δcbf11 mutation, as removal of these effectors should abolish the positive effects of nitrogen on mitotic fidelity and/or overall viability.Such studies could also explain the general character of the rescue, which occurs in a functionally diverse group of 'cut' mutants (Fig. 6).
TOR is linked to diabetes, cancer and other age-related diseases in humans [75].Therefore, it would be interesting to see whether the novel mitotic role of TOR is conserved in metazoan cells, where both TORC1 and TORC2 complexes contain the same mTor kinase [76], and where open mitosis is employed.Currently, anti-cancer drugs targeting the human TOR are designed to inhibit the kinase activity.Unfortunately, such broad-impact TOR inhibition has so far shown limited applicability in cancer treatment (reviewed in [75][77]).Perhaps a more selective approach to targeting the TOR network may prove more suitable.In summary, we revealed a novel role of the TOR regulatory network as a factor influencing mitotic fidelity, thereby linking nutritional stimuli and metabolic state to the successful progression and completion of mitosis.We also revisited the previously identified importance of sufficient NE expansion for the fidelity of closed mitosis [6] [17], and showed that it is not critical for the nitrogen-dependent mitotic rescue.cells and its effect is not additive with ammonium supplementation.The Δcbf11 mitotic defects can also be rescued, to varying degrees, by the introduction of a tor1-D temperature-sensitive allele, tor1 deletion (B), or deletion of ssp2 (C).Cells were grown to exponential phase in the indicated media and temperature, fixed, stained with DAPI and subjected to microscopy.Mean values ± SD from 3 independent experiments are shown.NA -not analysed.To determine statistical significance one-way Student's t-test with Holm correction was applied.

Figure 2 .
Figure 2. Small nuclear size in mitotic Δcbf11 cells is not rescued by ammonium.WT and Δcbf11 cells expressing the Cut11-GFP marker of NE were grown to exponential phase in the indicated media and subjected to time-lapse microscopy.The area of nuclear cross-section at the beginning (A) and end (B) of anaphase for 41-45 cells per condition is shown.Boxplots display medians (thick line), Q 1 and Q 3 quartiles (the box) and ± 1.5x interquartile ranges (whiskers).To determine statistical significance one-way Student's t-test with Holm correction was applied.

Figure 3 .
Figure 3. Decreased expression of lipid metabolism genes in cbf11 and mga2 mutants is not rescued by ammonium.RNA-seq results are shown as log2 of expression fold change values normalised to WT. Gene expression in the Pcut6MUT mutant and in WT treated with cerulenin are shown for comparison.

Figure 4 .
Figure 4. Lack of cbf11 leads to pronounced changes in FA and lipid composition which are not affected by ammonium supplementation.WT and Δcbf11 cells were grown to exponential phase in the indicated media, and their lipid composition was analysed.(A) Δcbf11 cells have lower FA content per unit of dry cell weight compared to WT. (B) Degree of FA saturation is higher in Δcbf11 cells.(C) Abundance of selected FA species in total lipid samples.Two different y-axis scales are shown to better visualise the full range of FA abundances.Mean values ± SD from 3 independent experiments are shown in panels A-C.To determine statistical significance one-way (A) or two-way (B, C) Student's t-test with Holm correction was applied.(D) Thin layer chromatography (TLC) analysis of neutral lipids.Only a section of the TLC plate is shown (see Fig. S1 for the full TLC plate).SQ -squalene; SE-1, SE-2 -sterylester species.

Figure 5 .
Figure 5. Ammonium rescues prolonged mitotic duration in Δcbf11 cells.Examples of mitotic phases as they are observed under the microscope.Green -chromatin (Hht2-GFP), magenta -microtubules (mCherry-Atb2) (A).The duration of the whole mitosis (prophase, metaphase and anaphase combined) (B), as well as prophase+metaphase (C) or anaphase (D) separately are all prolonged in cells lacking cbf11.This aberrant timing does not occur in the presence of ammonium.Data for 31-41 cells per condition are shown.Barplot (B) displays mean values ± SD together with individual data points.Boxplots (C, D) display medians (thick line), Q 1 and Q 3 quartiles (the box) and ± 1.5x interquartile ranges (whiskers), together with individual data points.To determine statistical significance one-way Student's t-test with Holm correction was applied.Only cells which successfully completed mitosis were analysed.

Figure 6 .
Figure 6.Ammonium supplementation rescues a range of cut mutants.(A-C) Cells were grown to exponential phase in the indicated media, fixed, stained with DAPI and subjected to microscopy.Cultivation temperatures were chosen to provide semi-restrictive conditions for temperature-sensitive strains.Mean values ± SD from 2-3 independent experiments are shown.To determine statistical significance one-way Student's t-test with Holm correction was applied.(D) Mutants responsive to ammonium supplementation tend to be associated with pre-anaphase ("pre-A") rather than anaphase ("A").Grey background -no rescue.

Figure 7 .
Figure7.The TOR network is critical for the ammonium-mediated rescue of Δcbf11 mitotic defects.(A) Rapamycin treatment partially suppresses the mitotic defects of Δcbf11 cells and its effect is not additive with ammonium supplementation.The Δcbf11 mitotic defects can also be rescued, to varying degrees, by the introduction of a tor1-D temperature-sensitive allele, tor1 deletion (B), or deletion of ssp2 (C).Cells were grown to exponential phase in the indicated media and temperature, fixed, stained with DAPI and subjected to microscopy.Mean values ± SD from 3 independent experiments are shown.NA -not analysed.To determine statistical significance one-way Student's t-test with Holm correction was applied.