Ubiquitination regulates cytoophidium assembly in Schizosaccharomyces pombe

CTP synthase (CTPS), a metabolic enzyme responsible for the de novo synthesis of CTP, can form filamentous structures termed cytoophidia, which are evolutionarily conserved from bacteria to humans. Here we used Schizosaccharomyces pombe to study the cytoophidium assembly regulation by ubiquitination. We tested the CTP synthase’s capacity to be epigenetically modified by ubiquitin or be affected by the ubiquitination state of the cell, showed that CTPS is immunoprecipitated with ubiquitin, and that ubiquitination is important for the maintenance of the CTPS filamentous structure in fission yeast. We have identified proteins which are in complex with CTPS, including specific ubiquitination regulators which significantly affect CTPS filamentation, and mapped probable ubiquitination targets on CTPS. Furthermore, we discovered that a cohort of deubiquitinating enzymes is significant for the regulation of cytoophidium morphology. Our study provides a framework for the analysis of the effects that ubiquitination and deubiquitination have on the formation of CTPS filaments.


Introduction
CTP synthase (CTPS) is the enzyme that catalyzes the rate-limiting step in the production of CTP from UTP in the de novo synthesis of CTP (Ozier-Kalogeropoulos et al., 1991, Ozier-Kalogeropoulos et al., 1994. Dysregulation of CTP pool homeostasis and increased CTPS levels have been correlated with a number of cancers, rendering CTPS an important target for drug development (Williams et al., 1978, Whelan et al., 1993, Groblewski et al., 1995, Hatse et al., 1999, Verschuur et al., 2000, Verschuur et al., 2001, Verschuur et al., 2002, Furuta et al., 2010, Wise and Thompson, 2010, Nilsson et al., 2014. Notably, CTPS is suggested to play a functional role in cancer metabolism, since knocking down of the enzyme in Drosophila cancer models is correlated with the reduction of tumour formation (Willoughby et al., 2013).
In the past few years, a connection has been gradually established between CTPS filamentation and the ubiquitination system. The ubiquitin E3 ligase activity of the proto-oncogene Cbl has been found to be required for CTPS filamentation, without affecting CTPS protein levels in Drosophila ovaries during endocycles (Wang et al., 2015). It is not yet clear whether this effect of ubiquitination on cytoophidia formation is direct, or whether there is a still unknown component of the CTPS complex whose regulation by ubiquitination affects the filamentation . Furthermore, a degenerate cohort of fission yeast membrane trafficking deubiquitinating enzymes (DUBs) which were found to mediate cell polarity and survival also seem to target CTPS (Kouranti et al., 2010, Beckley et al., 2015, thus pointing to a possible regulatory role of deubiquitination in CTPS filamentation.
Previous studies in S. cerevisiae and mammalian cells have shown that CTPS can be regulated via phosphorylation by kinases (Park et al., 1999, Choi et al., 2003, Han et al., 2005, Chang et al., 2007. We have recently shown that TOR pathway mediates cytoophidium assembly in mammalian and fission yeast cells (Andreadis et al., 2019, Sun andLiu, 2019b). In addition to phosphorylation, ubiquitination seems to have an involvement in the cytoophidium assembly in Drosophila and mammalian cells . In these cases, CTPS ubiquitination is negatively associated with cytoophidia formation, while inhibition of ubiquitination leads to a reduction of cells containing cytoophidia under glutamine deprivation (Peng et al., 2003.
Here, we used S. pombe to study the capability of ubiquitination to affect the regulation of CTPS cytoophidium formation. By using drugs that modulate ubiquitination and deubiquitination efficiency, as well as by using a series of ubiquitin ligase and DUB deletion mutants, we built upon previous results to show that ubiquitination, as a proteasome-independent modification, significantly affects CTPS cytoophidium formation, while improper deubiquitination significantly affects the morphology of CTPS filaments. Further to this, we performed protein interaction analyses to identify the proteins in complex with CTPS, while also focusing on specific CTPS residues, as potential ubiquitination targets. Our data showcase the importance of ubiquitination in regulating CTPS filamentation, while extending the spectrum of organisms in which CTPS filamentation seems to be modulated by ubiquitination.

Ubiquitination and deubiquitination affect CTPS filamentation
We used a S. pombe strain that we have previously constructed (Zhang et al., 2014), in which CTPS is fused with YFP tag, to study the effects of ubiquitination in fission yeast cytoophidia. The Cts1-YFP strain has the same growth profile as the WT strain (Zhang et al., 2014). In our model system, cytoophidia are formed in exponential phase, in which 92.7% of the cells on average contain cytoophidia ( Figure 1A and 1K), and get degraded in stationary phase, as we have previously shown (Andreadis et al., 2019).
We have interfered with the ubiquitination and deubiquitination systems in exponentially growing cells and monitored the changes in CTPS cytoophidium formation. More specifically, we applied E1-ubiquitin ligase inhibitor PYR-41, proteasome inhibitor MG132, and DUB inhibitor PR619 in logarithmic cultures for 2 hours at a range of concentrations, to study the effect on CTPS filamentation ( Figure 1B-J). When the ubiquitination system was impaired by PYR-41 E1-ubiquitin ligase inhibitor ( Figure 1B-D), we observed a strong and significant reduction of cytoophidia, correlated to the concentration of the inhibitor applied, ranging from 37.9% (50μM PYR-41, p<0.001) to 89.2% (75μM PYR-41, p<0.0001) and 97.4% (100μM PYR-41, p<0.0001), compared to the control strain ( Figure 1K).
Interestingly, prolonged treatment of the cells with E1-ubiquitin ligase inhibitor PYR-41, proteasome inhibitor MG132, and DUB inhibitor Pr619 for 4.5 hours, showed a sufficient recovery of the phenotype ( Figure S1).
Assuming that the activity of the drugs remains unchanged for the duration of the prolonged treatment, this result highlights the dynamic nature of CTPS cytoophidia, as well as their capability to adapt under this type of stress.
After observing the substantial dependency of CTPS filamentation on the ubiquitination and deubiquitination systems in fission yeast, we sought to investigate whether this effect could be attributed to a possible change in the CTPS protein levels. Our expression analyses showed that this was not the case, as the CTPS protein levels did not change significantly upon impairment of ubiquitination and deubiquitination systems or upon inhibition of the proteasome pathway (p>0.15) ( Figure 1M).
Taken together, our data show that the ubiquitination state of the cell plays an important role in maintaining the filamentous structures of CTPS in fission yeast, which are naturally formed under physiological growth conditions.

S. pombe CTPS immunoprecipitates with ubiquitin
To explore whether the effect of ubiquitination on cytoophidium formation is direct, we conducted co-immunoprecipitation (co-IP) experiments to determine whether CTPS can be ubiquitinated. We constructed a fission yeast strain for the co-IP analyses in which CTPS is tagged with the small tags HA and flag. The strain Cts1-flag-3HA strain provided us with the flexibility of conducting IP experiments using small tags, which are non-eukaryotic and do not interfere with the complex assembly or the protein function. Our co-IP experiments showed that CTPS immunoprecipitates with ubiquitin (Figure 2A), and a shift in the Cts1 protein size, corresponding to the small size of ubiquitin (<10kDa) ( Figure 2B). This finding is indicative of a potential direct effect of ubiquitination on CTPS compartmentation, though other proteins in complex with CTPS may also be ubiquitinated.

CTPS cytoophidium formation is affected by ubiquitin ligases
We used the Cts1-flag-3HA strain and performed a series of FLAG-HA tandem affinity purification assays (TAP assays), combined with analyses by liquid chromatography and mass spectrometry (LC-MS), in order to characterise the fission yeast CTPS protein complex. We chose to use the HA-FLAG TAP assay because it involves a double sequential elution which takes advantage of the double tag on the protein, offering higher specificity and pull-down of fewer contaminants, compared to other systems. This strategy offers a very low rate of cross-reactivity with non-targeted endogenous proteins, while the tags do not sterically affect the bait proteins' interactions. Our analyses showed a number of proteins that are in complex with CTPS. Based on the biological processes in which they participate, these proteins are either structural, or involved in metabolism, transport or stress (Table S1). Further to this, we used Cytoscape software (Cline et al., 2007) to visualise the potential networks in which the proteins in complex with CTPS participate, along with the biological networks gene ontology tool (BiNGO) (Maere et al., 2005) to determine the gene ontology (GO) terms that are significantly overrepresented in our group of proteins ( Figure S2). This analysis, not only mapped S. pombe CTPS interactome, but also provided us with a mechanistic insight into the regulation of cytoophidia formation by ubiquitination, by the identification of Mug30 and Ubr11 ubiquitin ligases in complex with CTPS.
We constructed deletion mutant strains for the Mug30 and Ubr11 ubiquitin ligases in a Cts1-YFP background in order to monitor the effects of the lack of these enzymes in the filamentation of CTPS. We performed a detailed confocal microscopic analysis ( Figure 3A-C) and found that the abundance of cytoplasmic cytoophidia in mug30Δ mutant strains was significantly reduced by 21% (p<0.0001) compared to the Cts1-YFP reference strain ( Figure 3D). The average length of CTPS cytoophidia was significantly reduced in both mutants, compared to the Cts1-YFP strain.
More specifically the average length was reduced by 21.3% in mug30Δ (p<0.01), and by 22.8% in ubr11Δ strains (p<0.01) ( Figure 3E). We then grouped the cytoophidia of each strain into three groups based on their length (<1μm, between 1-2μm, >2μm) and conducted a more in-depth analysis on how their length changed. We found a strong shift from longer to shorter cytoophidia in all mutants, compared to the control strain ( Figure 3F).
We then examined whether these trends were related to any changes in the protein levels of CTPS in these mutants and found that there was no significant change ( Figure 3G).
Collectively, our data introduce some of the protein players involved in the regulation of S. pombe CTPS compartmentation by ubiquitination, by identifying specific ubiquitin ligases in complex with CTPS, the action of which significantly affects the abundance and the length of CTPS cytoophidia.

CTPS filamentation is disrupted in K430R and K264R mutants
Next, we screened S. pombe CTPS amino acids based on their ability to become ubiquitinated. We have initially utilised UbiPred software to predict which lysine residues are more probable to be modified by ubiquitin (Radivojac et al., 2010). We have then modelled the three-dimensional conformation of S. pombe CTPS, based on the already solved CTPS structure in mammalian cells, by using Phyre2 software (Kelley et al., 2015).
After combining the results of the two analyses, we have identified two lysine residues, namely K430 and K264, which, based on their position in the threedimensional reconstruction of CTPS, are freer to interact with other proteins and have higher probability to become ubiquitinated ( Figure 4A).
Following this, we proceeded with mutating each of these lysine residues into arginine, rendering them incapable of being ubiquitinated.
Subsequently, we constructed the fission yeast Cts1-YFP strains bearing either K430R or K264R point mutations on CTPS and studied how this could affect the regulation of cytoophidia formation by performing confocal microscopy ( Figure 4B-D). We found that, compared to the reference Cts1-YFP strain, the abundance of cytoophidia was significantly reduced in both K430R and K264R mutants, by 16.7% in (p<0.001), and by 28.1% in (p<0.0001), respectively ( Figure 4E).
The effect of the mutations on cytoophidia formation was more evident on their average length, which was reduced by 40.3% in K430R cells (p<0.0001) and by 58.3% in K264R cells (p<0.0001), compared to the Cts1-YFP reference strain ( Figure 4F). Furthermore, when we grouped the CTPS cytoophidia into three groups based on their length (<1μm, between 1-2μm, >2μm), our results showed a significant 4.3-fold increase of smaller cytoophidia (<1μm) in K430R strain (p<0.0001) and an equally significant 6.7-fold increase of smaller cytoophidia in K264R strain (p<0.0001) as compared to the Cts1-YFP strain ( Figure 4G). In agreement with the trend of the reduction of cytoophidia length in the point mutation strains, we observed an 8.2-fold reduction in large cytoophidia (>2μm) in K430R cells, (p<0.01), and a 13.2-fold reduction in K264R cells (p<0.0001), as compared to the reference Cts1-YFP strain ( Figure 4G). The phenotypes exhibited by the CTPS cytoophidia in the K430R and K264R mutant strains was not due a change in the protein levels of CTPS, since these were found to be nonsignificant ( Figure 4H).
Taken together, our results strongly indicate that not only the ubiquitination state of the cell affects cytoophidia formation, but there are specific ubiquitin ligases in complex with CTPS, as well as particular CTPS residues which are involved in the regulation of CTPS compartmentation.
Our in silico analyses further showed that, apart from K430 and K264 residues, there are additional candidates which could be potential ubiquitination targets, albeit with a lower degree of confidence, nevertheless worth-exploring in a future study ( Figure S4).

Deubiquitinating enzymes affect CTPS cytoophidia morphology
Previous studies focused on cell polarity and survival have incidentally demonstrated, by employing comparative proteomics, biochemistry and microscopy, that a series of five degenerate deubiquitinating enzymes target CTPS (Beckley et al., 2015). These genes encode four ubiquitin C-terminal hydrolases (Ubp4, Ubp5, Ubp9, Ubp15) and one ubiquitin-specific protease (Sst2). We wanted to examine whether the deubiquitinating function of these proteins is necessary for the CTPS compartmentation and whether it affects in any way the CTPS filament formation.
To address this question, we constructed a series of single and multiple mutants of the deubiquitinating genes and studied the effects on CTPS filamentation by confocal microscopy ( Figure 5A-H). We did not observe any significant change of the percentage of cells containing cytoophidia in sst2Δ,ubp4Δ,ubp5Δ,ubp9Δ,ubp15Δ,sst2Δ ubp4Δ ubp9Δ ubp15Δ,, as compared to the control strain ( Figure 5A). The changes in the abundance of cytoophidia were insignificant ( Figure 5I). However, the average length of cytoplasmic cytoophidia in the series of single and multiple deubiquitinating mutants was significantly reduced. More specifically, while the average length of cytoplasmic cytoophidia in the WT (Cts1-YFP) was 1.77μm (±0.13μm), our calculations showed a significant change in the average cytoophidia length which was 1.46μm (±0.03μm) in sst2Δ (p<0.0001), (p<0.05), and 1.55μm (±0.19μm) in sst2Δ ubp4Δ ubp5Δ ubp9Δ mutant strain (p<0.001) ( Figure 5J).
We further examined this phenotype by grouping the cytoophidia in three categories based on their length (<1μm, between 1-2μm, >2μm), and subsequently checking how they are distributed in these categories in all mutant and control strains ( Figure 5K). A trend of increasing percentage of shorter (<1μm) cytoophidia in ubp5Δ (p<0.05) and decreasing percentage of ubp5Δ, ubp9Δ, and ubp15Δ (p<0.05) was observed, as compared to the wild type strain ( Figure 5K). Notably, this effect was not related to any changes in the CTPS protein levels. Our expression analyses showed no significant change of the CTPS protein levels in any of the DUB mutant strains ( Figure 5L).
Our results demonstrate the importance of the deubiquitination process in retaining the proper formation of CTPS cytoophidia. Our overall data suggest that ubiquitination is a key post-translational modification affecting the process of filamentation of CTPS, be it directly or indirectly, while its reversibility further contributes to the fine-tuning of the CTPS cytoophidium formation in the cells.

Discussion
Our study has demonstrated a link between CTPS cytoophidium formation and the ubiquitination/deubiquitination processes. We showed that proper, uninhibited ubiquitination and deubiquitination are key to the assembly and maintenance of CTPS filaments. Inhibition of either process lead to a disruption of the physiological formation of CTPS cytoophidia in S. pombe cells. We have constructed a putative map of the proteins in complex with fission yeast CTPS, which included particular ubiquitin ligases. By in silico analyses we identified specific lysine residues, potential ubiquitination targets, which we found to be important for the proper compartmentation of CTPS. Furthermore, we showed that the absence of specific deubiquitinating enzymes affects the morphology of cytoophidia by reducing their average length.
Our results demonstrate that the system of ubiquitination/deubiquitination could be a conserved means of regulation of cytoophidium filamentation.
Previous studies have shown that ubiquitination regulates CTPS activity by promoting CTPS filament formation in Drosophila ovaries during endocycles (Wang et al., 2015). We have previously shown that in mammalian cells the CTPS compartmentation is negatively associated with CTPS ubiquitination, since cytoophidia formation blocked the CTPS ubiquitination and prolonged its half-life (Sun and Liu, 2019a). However, it has also been shown that ubiquitination is required for CTPS filament formation in Drosophila and mammalian cells, since cytoophidia were completely abolished or strongly reduced, respectively, when E1 ubiquitin ligase was inhibited by PYR41 drug . These opposing roles that ubiquitination seems to have on cytoophidia formation could be the result of differential effects of ubiquitin, which could be targeting both CTPS and CTPS-regulating proteins, affecting the CTPS filamentation by direct ubiquitination, or indirectly, by modifying the factors which regulate the filaments. This is in agreement with our data showing that in the absence of specific ubiquitin ligases, which are in complex with CTPS, or after inhibition of ubiquitination by drugs, S. pombe cytoophidia are disrupted. The distinct, probably opposing roles that ubiquitination seems to play on cytoophidia formation could explain our current data in S. pombe DUB deletion mutants, in which cytoophidia become smaller ( Figure 5).
It is known that in Drosophila ovary the presence of cytoophidia is correlated with efficient production of CTP to be used in fast synthesis of DNA and phospholipids (Strochlic et al., 2014, Wang et al., 2015.
Furthermore, there are studies indicative of cooperative incorporation of the newly synthesised CTPS into the cytoophidium (Barry et al., 2014, Wang et al., 2015, suggesting that CTPS filamentation could serve as a reservoir for rapid activation, as has been previously proposed . To this end, our results here show that ubiquitination is an important regulator of the filamentous CTPS conformation, contributing to the dynamic nature of the structure. Data from structural studies on human and bacterial cells revealed that CTPS cytoophidia consist of stacked tetramers of CTPS which undergo a conserved conformational cycle regulated by substrate and product binding (Lynch et al., 2017). Our study provides insights into the determinants of the process that leads to the formation of the CTPS filaments. Our results are in agreement with a model in which both possible direct effects of ubiquitination on CTPS protein itself, and indirect effects via regulation of CTPS filament regulators by ubiquitination, can determine the cytoophidium formation. We propose that the observed cytoophidia in S. pombe consist of bundles of selfassembled filaments which come together and form larger bundles, giving to the membraneless cytoophidia their final form that is observed in exponentially growing cells, a process that is governed by ubiquitination and deubiquitination ( Figure 6).
If CTPS filaments can be ubiquitinated, then this modification could trigger the assembly of larger bundles ( Figure 6).  (Noree et al., 2010.

Yeast Strains Construction and culture conditions
Cts1-YFP strain (JLL005S), was constructed as previously described (Zhang et al., 2014), after JLL003S wild type strain was transformed with linearized (SpeI) pSMUY2-Ura4-cts1-YFP plasmid. The Cts1-YFP strain has the same growth profile as the wt strain (Zhang et al., 2014). Haploid deletion strains containing Cts1-YFP were obtained by PCR-based gene targeting using pFA6A-KanMX6 or pFA6A-hphMX6 plasmids (depending on the selection marker needed) as templates for specially-designed gene-specific 80bp-long primers, as previously described (Bahler et al., 1998), followed by transformation of Cts1-YFP strains. Cts1-flag-3HA strain was constructed using the same PCR-based gene targeting method, after using pFA6A-3HA-KanMX6 plasmid as template and incorporating the flag tag in the genespecific 80bp-long primers. The strain was verified by PCR and protein expression analyses confirming the expression of both tags. Cts1 point mutation strains were constructed by incorporating the mutated sequence leading to the K-to-R amino acid mutation into the forward primer used in the PCR-based gene targeting, using genomic DNA of a Cts1-YFP strain as template, followed by transformation of a wild type strain and colony screening. The strain bearing the point mutation was verified by sequencing.
The DUB quadruple mutants were constructed after sequential mating between haploid mutants, followed by random spore analysis as previously described (Ekwall andThon, 2017a, Ekwall andThon, 2017b

Western Blot Analyses
Extraction of proteins was done by using the trichloroacetic acid (TCA) method on equal number of cells for each sample, and Western blotting was performed as previously described (Ralph et al., 2006).

Tandem affinity purification (TAP) and Liquid chromatography -Mass spectrometry (LC-MS) Analyses
We used a Cts1-flag-3HA strain for our TAP and LC-MS analyses. A wild type strain was used as reference. For the TAP assay we used the FLAG-HA tandem affinity purification kit manufactured by Sigma (TP0010) and performed the protocol according to the manufacturer's specifications.
Briefly, Cts1-flag-3HA cells were cultured in rich media (YE4S) until they reached exponential phase. 50x10 7 cells were collected and the cell lysate was extracted, as described in the co-IP protocol. The cell lysate was then incubated with anti-FLAG M2 resin rotating overnight at 4 0 C, ensuring efficient binding. The supernatant was then removed carefully, and the resin was washed with RIPA buffer (Sigma, R0278) containing protease inhibitors (Sigma, P8215), in order to remove any unbound protein. The first elution of the protein complex bound on the resin was done by using 3XFLAG peptide, and in a following step the eluate was bound to anti-HA resin slurry. In the second elution of the protein complex, the anti-HA slurry was washed with TBS (50mM Tris-Cl, ph=7.5; 150mM NaCl), to remove any unbound protein.
The final elution was done by using TBS with 100mM ammonium bicarbonate, and the sample was subsequently digested with trypsin overnight.
Peptides were then separated and analyzed on an Easy-nLC 1000 system coupled to a Q Exactive HF (both -Thermo Scientific). About 2 µg of peptides were separated in an home-made column ( Reversed database searches were used to evaluate false discovery rate (FDR) of peptide and protein identifications. Two missed cleavage sites of trypsin were allowed. Oxidation (M), Acetyl (Protein N-term), deamidation (NQ) and GGE (K) were set as variable modifications. The FDR of both peptide identification and protein identification is set to be 1% (Elias and Gygi, 2007). The option of "Second peptides", "Match between runs" and "Dependent peptides" was enabled. Label-free quantification was used to quantify the difference of protein abundances between different samples Mann, 2008, Cox et al., 2011).

3D modelling of CTPS and ubiquitination target prediction
Based on the resolved CTP synthase three-dimensional conformation, the mammalian CTP synthase amino acid sequence was used as a template in Phyre2 software (Kelley et al., 2015) to obtain the PDB formatted model of the S. pombe CTP synthase, which was subsequently studied with EzMol software (Reynolds et al., 2018), in order to model the three-dimensional conformation of S. pombe CTPS and predict the space occupied by the residues of interest. For the prediction of potential ubiquitination-targeted amino acids on CTPS we used the fission yeast CTPS protein sequence as a template in Ubipred software (Radivojac et al., 2010).

Gene Ontology (GO) Analyses
For the Gene Ontology analyses, the Cytoscape software platform (Cline et al., 2007) was utilised to visualise the potential networks in which the proteins in complex with CTPS participate. In order to determine the gene ontology (GO) terms that are significantly overrepresented, the biological networks gene ontology (BiNGO) tool (Maere et al., 2005) was used.

Funding
This work was supported by ShanghaiTech University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability
The