Multiple E3s promote the degradation of histone H3 variant Cse4

The histone H3-like protein Cse4/CENP-A acts as a key molecular marker that differentiates the special centromeric chromatin structures from bulk nucleosomes. As altered Cse4/CENP-A activity leads to genome instability, it is pivotal to understand the mechanism underlying Cse4 regulation. Here, we demonstrate that four ubiquitin ligases (i.e., Ubr1, Slx5, Psh1, and Rcy1) work in parallel to promote Cse4 turnover in yeast. Interestingly, Cse4 overexpression leads to cellular toxicity and cell cycle delay in yeast cells lacking PSH1, but not in cells lacking UBR1, suggesting different roles of these two degradation pathways. Our findings suggest that various ubiquitin ligases collaborate to keep the Cse4 level in check, providing a basis for further delineating the intricate network involved in Cse4 regulation.

The centromere, a highly condensed and constricted chromosome region, serves as the site of kinetochore assembly, which is essential to chromosome segregation that allows proper distribution of duplicated genetic information to daughter cells 1,2 . The yeast histone H3 variant Cse4 and its human counterpart CENP-A are molecular markers that differentiate the special centromeric structure from bulk nucleosomes [2][3][4][5] . Excessive or nonrestraint Cse4/CENP-A can lead to its incorporation into euchromatic nucleosomes, which may result in aneuploidy and promote tumorigenesis 2, 6-8 . It is therefore critical to understand how Cse4/CENP-A activity is regulated and restricted to the centromere.
Human CENP-A and yeast Cse4 have been shown to be subject to ubiquitin-mediated degradation 2,9,10 . Ubiquitin, a 76-residue polypeptide, often functions as the molecular flag that marks protein for destruction, promoting rapid changes in protein concentration [11][12][13] . The key factors in ubiquitin-mediated events are the ubiquitin ligases (E3s), which specifically recognize substrates. With the help of a ubiquitin-activating enzyme (E1) and a ubiquitin-conjugating enzyme (E2), E3 ligases facilitate the covalent attachment of ubiquitin molecules onto their targets. Ubiquitin-decorated substrates are then transferred to and degraded by the 26S proteasome, a multi-subunit protease [11][12][13] . As E3 ligases confer substrate selectivity and carry out the rate-limiting step of ubiquitylation, the key to understanding the role of ubiquitin-mediated proteolysis in Cse4/CENP-A regulation is identifying ubiquitin ligases that are responsible for selecting Cse4/CENP-A for ubiquitylation and degradation.
The first degradation route for Cse4 was identified ~7 years ago but only accounted for partial Cse4 degradation. Using an affinity purification-mass spectrometry coupled strategy, Psh1 was shown to bind Cse4 directly and promote its ubiquitylation 14,15 . We recently performed a synthetic dosage lethality screen and identified the SCF (Skp1-Cdc53-F box) complex containing the F-box protein Rcy1 as another ubiquitin ligase required for Cse4 turnover 16 . Excessive Cse4 accumulation leads to growth retardation in cells lacking PSH1 or RCY1. However, in the absence of both PSH1 and RCY1, Cse4 is still degraded albeit more slowly 16 , suggesting other degradation pathway(s) remains to be discovered.
To uncover other ubiquitylation enzymes involved in Cse4 regulation, we screened for ubiquitin-conjugating enzymes (E2) involved and found that Cse4 turnover is impaired in yeast cells lacking UBC4. We next demonstrated that a Ubc4-dependent E3 complex composed of Slx5 and Slx8 promotes Cse4 turnover. Moreover, we showed that Ubr1 is the fourth E3 involved in Cse4 ubiquitylation and degradation. Our results suggest that these four ubiquitin ligases likely act in parallel to maintain the Cse4 level. Interestingly, unlike PSH1-defective cells, UBR1 deficiency does not lead to growth retardation and cell cycle delay upon Cse4 overexpression. These findings offer novel insights into the mechanism underlying Cse4/CENP-A regulation.
As an E3 directly interacts with its target, we evaluated whether Slx5 binds Cse4 in vivo by co-immunoprecipitation assays (Fig. 2E). We found that Slx5 and Cse4 reciprocally co-immunoprecipitated ( Fig. 2E), supporting the involvement of Slx5 in Cse4 regulation. Slx5 is known as a SUMO-targeted ubiquitin ligase (STUbL) that recognizes sumoylated proteins as substrates. Slx5 substrates are first sumoylated by the Ubc9 (E2)-Siz1/2 (E3) pathway, then recognized and ubiquitylated by the Ubc4 (E2)-Slx5-Slx8 (E3) pathway for subsequent degradation. We therefore wondered whether sumoylation was involved in Cse4 degradation. Ubc9 is the sole and essential SUMO conjugating enzyme in yeast. When a yeast strain bearing a temperature sensitive mutation in UBC9 was shifted from 30 °C to the non-permissive temperature 37 °C, Cse4 degradation was markedly compromised (Fig. 2F,G), which is consistent with the link between sumoylation and Slx5 in proteolysis.
While our work was in progress, Ohkuni et al. also demonstrated the involvement of Slx5 and Slx8 in Cse4 degradation 18 . Using a different approach (i.e., post-translational modification of Cse4), they established Cse4 sumoylation in vivo and in vitro via the SUMO ligases Siz1 and Siz2, which then attracts the Slx5-Slx8 E3 complex for subsequent substrate ubiquitylation and degradation 18 . Our results complement their findings and further reveal an in vivo involvement of the E2 enzymes Ubc4 and Ubc9 in Cse4 regulation (Figs 1 and 2).
Cse4 degradation is compromised in cells lacking Psh1, Slx5 and Rcy1. As both double mutants psh1Δ slx5Δ and psh1Δ rcy1Δ exhibit a slower Cse4 degradation rate than single mutants (i.e., psh1Δ, rcy1Δ, slx5Δ) 16,18 , Psh1 is thought to operate independently of Slx5 and Rcy1.We also found that Slx5 and Rcy1 function in separate pathways (see below in Fig. 6B). Next, we examined whether Cse4 degradation would be abolished in yeast cells missing all three of these ubiquitin ligases Psh1, Slx5 and Rcy1. Instead of using shorter time points up to 40 minutes, which help to capture moderate effects, we extended the chase time of Cse4 stability experiment to 3 hours (Fig. 3A). Interestingly, Cse4 turnover was severely impaired but not entirely eliminated in the psh1Δ rcy1Δ slx5Δ triple mutant (Fig. 3A,B), suggesting that there are other degradation pathways for Cse4. Consistent with the observation of compromised Cse4 turnover in the psh1Δ rcy1Δ slx5Δ mutant, Cse4 ubiquitylation was also significantly reduced in the triple mutant as well (Fig. 3C).

Figure 1.
The involvement of Ubc4 E2 and Ubc4-associated E3s in Cse4 degradation. (A) Cse4 turnover is impaired in ubc4Δ cells. Cse4 stability was examined in wild-type and E2-deficient cells by a protein expression shut-off assay. A plasmid bearing HA-tagged Cse4 was transformed into wild-type cells and 10 E2 mutants. Yeast cells expressing HA-Cse4 were grown to an OD 600 of ~1, and samples were collected after expression was turned off by cycloheximide (indicated as Chase (min). Extracts were analyzed by immunoblotting with HA antibody. Equal amounts of extracts were used and verified by western blotting with Rpt5 antibody in all of protein stability experiments (lower panels). (B) Efficient Cse4 degradation requires Slx5 and Slx8. Cse4 turnover in wild-type cells and 12 Ubc4-associated E3 mutant cells was determined as above.

Ubr1 E3 promotes Cse4 ubiquitylation and degradation. To identify other ubiquitin ligases
involved, we re-examined the list of candidate Cse4-binding proteins that were previously isolated using an immunoprecipitation-mass spectrometry approach 15 . Besides Psh1 15 , Ubr1 was the only other E3 obtained in the screen but its potential role in Cse4 regulation had not been characterized. Ubr1 was the first E3 ligase identified in ubiquitin-mediated processes and has many substrates localized in the nucleus and cytosol 19,20 . First, we performed co-immunoprecipitation to validate the Cse4-Ubr1 interaction. We found that Flag-tagged Ubr1 and HA-tagged Cse4 immunoprecipitated each other (Fig. 4A), supporting binding between Ubr1 and Cse4. Moreover, Cse4 degradation and ubiquitylation are impaired in yeast cells lacking UBR1 (Fig. 4B,C,D), suggesting that Ubr1 acts as another E3 enzyme for Cse4 turnover. One concern is that Cse4 was overexpressed in these experiments. Therefore, we next employed a strain expressing Cse4 tagged with GFP at its endogenous locus. Due to its lower expression level, GFP-Cse4 was enriched by immunoprecipitation with a GFP antibody in these experiments (Fig. 4E). We found that degradation of endogenously expressed GFP-Cse4 was impaired in cells lacking UBR1 (Fig. 4E,F), supporting that Ubr1 also participates in Cse4 regulation.

Differential responses of psh1Δ and ubr1Δ cells to Cse4 overexpression. Cse4 accumulation has been
shown to cause severe growth inhibition in yeast cells lacking PSH1 or RCY1 [14][15][16] . Interestingly, Cse4 overexpression has little effect on the growth of ubr1Δ or slx5Δ cells, unlike mutant cells missing PSH1 or RCY1 (Fig. 5A) [14][15][16] . Consistent with the normal growth observed in slx5Δ upon Cse4 induction, cells lacking UBC4 were not sensitive to Cse4 overexpression (Fig. 5A) as Ubc4 and Slx5 function in the same pathway 18 . To further understand the toxicity triggered by Cse4 accumulation in psh1Δ cells, we monitored cell cycle progression by FACS analysis since Cse4-mediated chromosome segregation is a crucial cell cycle event. Wild-type or mutant yeast cells bearing a plasmid for Cse4 overexpression or a vector control were arrested at G1 phase by α-factor and then released to resume normal cell growth. In the absence of Cse4 overexpression, largely similar cell cycle progression kinetics was observed in wild-type, psh1Δ (Fig. 5B), and ubr1Δ cells (data not shown). However, compared with wild-type cells, psh1Δ cells overexpressing Cse4 exhibited a significantly slower cell cycle, showing ~100 minutes delay (Fig. 5B). In contrast, ubr1Δ cells overexpressing Cse4 showed only a minor cell cycle delay compared with wild-type cells (Fig. 5B). Four E3s work in parallel to promote Cse4 turnover. We then evaluated the relationship between Ubr1 and the other three E3s. Slx5 and Psh1 define two distinct pathways for Cse4 turnover based on a previous study 18 . Therefore, we compared the degradation kinetics of Cse4 in wild-type cells and psh1Δ slx5Δ, and psh1Δ slx5Δ ubr1Δ mutants using an expression shut-off assay. The triple mutant psh1Δ slx5Δ ubr1Δ showed greater impairment of Cse4 turnover than the psh1Δ slx5Δ double mutant (Fig. 6A,C), suggesting that Ubr1 likely works in a pathway separate from Psh1 and Slx5.
We previously demonstrated that Rcy1 and Psh1 act in different pathways to promote Cse4 destruction 16 . We found that the triple mutants, psh1Δ rcy1Δ slx5Δ and psh1Δ rcy1Δ ubr1Δ, showed more compromised Cse4 turnover than the psh1Δ rcy1Δ double mutant (Fig. 6B,D), suggesting that Ubr1 and Slx5 work separately, but in parallel with Psh1 and Rcy1. Combined, our results suggest that these E3s define four distinct routes of Cse4 degradation.
We further assessed whether these four E3s account for all Cse4-related degradation pathways. Cse4 was significantly stabilized in the quadruple mutant psh1Δ rcy1Δ slx5Δ ubr1Δ, but residual degradation remained (Fig. 7A,B), suggesting that additional ubiquitin ligase(s) or a small pool of stable Cse4 may exist. Consistent with this finding, Cse4 ubiquitylation is largely but not entirely eliminated in psh1Δ rcy1Δ slx5Δ ubr1Δ mutant cells (Fig. 7C). In contrast to the impaired Cse4 turnover observed in the quadruple mutant psh1Δ rcy1Δ slx5Δ ubr1Δ, the degradation of two proteasomal substrates Pex29 and Dbf4 was unaltered in cells lacking these E3s (Fig. 7D,E), indicating that global proteolysis is not affected in the quadruple mutant cells.
We also examined the growth of the quadruple mutant psh1Δ rcy1Δ slx5Δ ubr1Δ under various conditions (Fig. 7F). Interestingly, the quadruple mutant is more sensitive than single E3 mutants to DNA-damaging agents (e.g., UV and HU) and oxidative stress triggered by H 2 O 2 (Fig. 7F), supporting the involvement of these E3s in genome maintenance.

Discussion
Cse4 plays a crucial role in establishing the centromeric domain, which is key to chromosome segregation [2][3][4][5] . Unrestrained Cse4 activity leads to aberrant chromosome-related processes and genome instability 2, 6-8 . Here, we show that at least four E3 ligases (i.e., Psh1, Ubr1, Slx5, Rcy1) work together to regulate the Cse4 level. Although one E3 ligase may be sufficient to adjust the concentration of its targets, multiple ubiquitin ligases have   increasingly been found to be involved in modulating a substrate in response to various stimuli. For example, five E3s (i.e., Tom1, Snt2, Pep5, Hel1 and Hel2) have been implicated in the degradation of histone H3 21 . The use of multiple E3s in substrate turnover allows for more regulation and better control. Instead of simply acting like an "on/off " switch, E3s serve more like a dimmer-type control to modulate protein concentration and activity.
How these E3s collaborate remains far from clear. The delayed cell cycle progression observed in psh1Δ cells (Fig. 5B) further suggests that sustained Cse4 may activate cell cycle checkpoints to ensure proper chromosome segregation. Our findings raise many challenging issues that would be important for future studies, including how these E3s specifically recognize Cse4, how they work with other cellular cues and pathways (e.g., casein kinase 2, Siz1-and Siz2-mediated sumoylation, SWI/SNF remodeling enzymes, the FACT complex, and the proline isomerase Fpr3) 17,18,[22][23][24][25] , and what the functional role of each degradation pathway is.
The four ubiquitin ligases involved in Cse4 turnover have been identified via different approaches, including immunoprecipitation-mass spectrometry, a synthetic lethality screen, and a systematic analysis for specific E2s and related E3s [14][15][16]18 . Since Cse4 degradation is not entirely abolished in the quadruple mutant (Fig. 7), it is possible that a small pool of Cse4 may be spared from degradation or additional ubiquitin ligases are involved in Cse4 turnover. Key to understanding the physiological functions of the ubiquitin system is to delineate the  functional relationships between ubiquitin E3 ligases and cognate targets. However, due to the transient nature of substrate-E3 binding and the technical challenges involved, it can be a significant undertaking to identify a relevant E3 involved in the modification of a given substrate. For example, more than 20 years after the discovery of the ubiquitylated form of histone H2B, Bre1 was identified as the ubiquitin ligase in H2B regulation 26,27 . Identifying cellular targets of an E3 can be equally frustrating as illustrated by the time it took (i.e., ~15 years) to discover a physiological substrate (i.e., Scc1) of the N-end rule E3 Ubr1, the first E3 ligase identified in ubiquitin-mediated events 19,28 . Given the critical role of Cse4 in centromere function, it may be worthwhile to comprehensively analyze Cse4 stability in all yeast mutants that are defective in E3 activity.
The ubiquitin ligase(s) involved in human CENP-A degradation remain unclear 2 . While Psh1 does not seem to have a mammalian counterpart, Ubr1, Slx5 and Rcy1 are known to have human homologues. It will be of interest to evaluate CENP-A turnover in mammalian cells deficient for these homologues and also to determine whether the human homologues of these E3s are altered in CENP-A-related cancer cells. The knowledge gained will significantly advance our understanding of CENP-A mediated centromere function and related diseases, and may lead to the development of new avenues to control its activity.
The plasmids pMB1458 bearing 3HA-tagged Cse4 has been previously described 30 . The pRS423-based plasmids expressing wild-type SLX5 or a RING finger mutant (C561S, C564S) of SLX5 were kindly provided by Dr. Hochstrasser (Yale U.). The pNT-Flag-UBR1 plasmid bearing Flag-tagged UBR1 was obtained from Dr. Varshavsky (Caltech). Ubiquitylation Assay. Yeast cells expressing a GAL1-regulated substrate HA-tagged Cse4 or a vector control were grown to log phase in SR medium, and 2% galactose was added to induce protein expression for 4 h. The cells were harvested by centrifugation and lysed with glass beads. The lysates were immunoprecipitated with tandem ubiquitin binding entities (TUBEs) agarose gel (UM401, LifeSensors) for ubiquitylated species at 4 °C for 6 h. The immunoprecipitates were resolved by SDS-PAGE, transferred to PVDF membranes, and immunoblotted with a HA antibody (MMS-101P, BioLegend) to detect ubiquitylated Cse4, which were detected by anti-mouse HRP conjugates and ECL reagents. The samples were also analyzed by anti-HA for Cse4 and anti-Rpt5.
Expression shut-off assay. As previously described 31 , yeast cells expressing HA-tagged Cse4 were grown at 30 °C to an OD 600 of ~1 in synthetic medium SR lacking uracil. The expression of HA-Cse4 was induced by galactose for 2 hr 31 . Protein synthesis was turned off by the addition of cycloheximide (150 μg/ml). Samples were collected at the indicated time points. Proteins were extracted by glass bead lysis of cells and processed for immunoblotting with anti-HA (BioLegend), followed by detection with goat anti-mouse HRP conjugate using ECL reagents (GE Healthcare). The stable protein Rpt5 was used as a loading control.