In this study we have investigated the cell cycle specific interaction networks of the yeast 26S proteasome by applying the QTAX strategy38,39
to the analyses of samples from synchronized cell populations. A total of 677 PIPs have been captured, identified and quantified from three cell cycle phases (G1, S, and M). In comparison to the results from unsynchronized cells,38,39
266 additional proteins were identified. To further understand the identified proteasome interactions, we have clustered the PIPs by the characteristics of their cell cycle phase dependent SILAC ratios using a profile vector-based clustering approach.61–63
This method separates PIPs based on a set of constraints that we have defined rather than by using the known heuristic clustering algorithms such as hierarchical and k-means methods.64,65
Although both the hierarchical and k-means clustering methods produced PIP clusters with assorted enrichments, the profile vector-based method produced more mid to high density clusters and had significantly more enrichments, suggesting that more biological relevant information was extracted by this approach. In total, we have identified 36 function, 9 complex, and 26 pathway enrichments from 20 total clusters, including notable functional enrichments in signal transduction and cell fate as well as complexes involved in intracellular transport and transcription. Among the 20 clusters, 3 different PIP clusters were found to be enriched with biological functions and pathways pertaining to cell cycle ( and ).
In this study, it is noted that the synchronization efficiency for S phase cells was limited. This is due to the fact that we had to release the cells from G1 phase arrest to obtain the S phase samples since the common chemical to induce S phase arrest, that is, hydroxyurea, can not be used due to its interference with formaldehyde cross-linking in QTAX experiments. In contrast to S phase arrest using cell cycle checkpoint induction by hydroxyurea, the arrest-release strategy allows for obtaining an unperturbed S phase population. However, the arrest-release strategy has limitations as it is necessary to compromise between complete release from G1 and progression of cells too far into S-phase or G2. Despite the limited cell synchronization in S phase, our results remain valid since we have clustered proteins based on changes in their interactions with proteasome among three different phases. Since all samples were subjected to the same synchronization, proteins with similar interaction changes should still be grouped/clustered together based on their SILAC ratio profiles. Although better synchronization might increase the ratio changes of a protein between two phases, proteins interacting similarly with proteasomes at three different phases would still have similar SILAC ratio profiles and thus allow them being clustered together.
In this work, our analysis identified clusters enriched with a total of 71 functions/complexes/pathways (i.e., 36 function, 9 complex, and 26 pathway enrichments) that have p-values below 0.1 ( and ). Among these enrichments, 43 (60.5%) have p
-values lower than 0.05, 19 (26.8%) have p
-values in the 0.05–0.075 range, and only 9 (12.7%) have p
-values in the 0.075–0.1 range. We use p
-value <0.1 as the threshold for cluster enrichment in functions/complexes/pathways due to the fact that the biological experiments and the PPI network data used in this study are not perfect and do contain some noise. In addition, many genes and their protein products do not have functional/complex/pathway annotations in current databases. Therefore, the lower the p
-value threshold used, the more likely one is to omit results that are of biological importance but are not statistically significant. Moreover, this threshold allows increased flexibility and balance between removing those enrichments that are very likely to occur at random and keeping potentially biologically interesting but statistically nonsignificant results. This is important, since it has been shown that statistically significant results may not be scientifically or biologically significant and that nonsignificant results may turn out to be very important.66
One of the interesting clusters identified is the defined ratio cluster [G1_S]#3, which has enriched functions in cell cycle and DNA processing (). Five PIPs in this cluster, that is, Fus3, Rvs161, Rvs167, Sst2, and Tpd3, are known to be involved in the yeast mating signaling pathway that induces cell cycle arrest.58,59
Upon pheromone binding to the G-protein-coupled-receptor (GPCR) on the cell-surface a series of events occur, which ultimately lead to up- and down-regulation of transcription of many genes, arrest in G1 phase, and bud tip formation and elongation.58
In the presence of a suitable mating partner, the fusion of the two plasma membranes and nuclei also occurs.58,67
There are many proteins involved in this signal cascade including GPCRs and several effector kinases such as Fus3.58,59
Fus3 is involved in several processes including control of pheromone induced gene expression by phosphorylation of transcription factors, induction of cell cycle arrest in the G1 phase by phosphorylating the cyclin-dependent kinase inhibitor Far1, and pathways related to membrane fusion.58,68–70
Reciprocal co-IP and Western blot analyses have confirmed the cell cycle specific interaction of Fus3 with the proteasome during G1 and S phase (). Rvs161, an amphiphysin homologue, is localized at the shmoo tip during mating and is involved in actin cytoskeleton organization, regulation of cell polarity, bud formation, and cell fusion.71–73
Rvs161 interacts and functions with another amphiphysin like protein, Rvs167, another PIP in this cluster.72,74,75
Additionally, the stability of Rvs161 is largely dependent upon the presence of Rvs167.72
Interestingly, Rvs167 is also a Fus3 substrate subsequent to pheromone signaling.76
Another Fus3 substrate, Sst2, is also a member of this cluster. Sst2 localizes to the plasma membrane and acts as a GTPase activating protein, thus desensitizing the cell to pheromone signaling.77
Tpd3 is a component of the yeast PP2A Ser/Thr phosphatase and has been shown to be involved in mediating G1 arrest in response to ceramide.78
Its presence in this cluster suggests that Tpd3 may be associated with α-factor induced G1 phase arrest. Taken together, these results have provided the first physical evidence directly linking the proteasome to pheromone signaling pathway in yeast and indicated that the proteasome may regulate the cell cycle through this signaling pathway. However, whether the proteasome is the target of pheromone signaling pathway or conversely a potential regulator of this pathway needs to be further investigated.
Another interesting cluster is the [G1_S_M]all_high cluster, in which the PIPs have high ratios in all three phases, the same characteristic SILAC ratios of proteasome subunits,39
suggesting their highly specific binding to the proteasome particles. In addition to the known ubiquitin receptors (Rad23, Dsk2 and Ddi1), a shuttling protein Cdc48 and its adaptor protein, Shp1, are grouped into this cluster. Shp1 has both a UBA (ubiquitin -associated) and a UBX (ubiquitin regulatory X) domain and has been shown to bind to ubiquitinated substrates in vivo
and to be involved in proteasome dependent degradation.79
The [G1_S_M]all_high cluster also contained Skp1, a core component of SCF ubiquitin ligases.28,80
However, other components of SCF ligases were not among the members of this cluster, suggesting that the Skp1 interacted independently of its SCF ligase binding partners. Interestingly, Skp1 has also been identified as a component of the yeast kinetochore independently of its function in the SCF ubiquitin ligase complex,81
and another protein linked to the kinetochore, Bub3, was placed in the [G1_S_M]all_high cluster. Bub3 is involved in the mitotic spindle checkpoint, and was identified as a “high” ratio PIP from unsynchronized cells previously.39,82,83
The other four PIPs in this cluster are Cmd1, Sok1, Grx1, Tsl1, and Tom70, which do not have confirmed physical interactions with any of the proteasome subunits except Sok1. Sok1 was identified as a high ratio PIP in our previous study using unsynchronized yeast cells.39
Although the biological functions of these PIPs’ (i.e., Cdm1, Bub3, Sok1, Grx1, Tsl1, and Tom3) interactions with the proteasome has yet to be determined, the fact that these PIPs have grouped to this cluster that harbors all known ubiquitin receptors and several key players in the proteasomal degradation pathway strongly suggests that these high ratio PIPs most likely play critical roles in assisting proteasomal degradation.
Although proteasome subunits including Rpn11 have stable expression throughout the cell cycle, it has been suggested that about 15% of budding yeast genes are subjected to transcriptional regulation during the cell cycle based on several genome-wide transcript measurements.84–86
Therefore, some of the identified PIPs whose cell cycle-dependent dynamic interactions with the proteasome may be the results of their cell cycle regulated expression changes. However, direct comparison of our data and the reported transcriptome data84–86
is not feasible due to differences in experimental conditions and the fact that protein expression level does not correlate well with its mRNA level.87
Future studies targeting at complete proteomic profiling of cell cycle regulated protein abundance changes will make it possible for us to further characterize these PIPs.
In summary, the QTAX approach has proven to be a robust and effective strategy for studying proteasome interacting proteins throughout the cell cycle. These results describe the first extensive proteomic analysis of the cell cycle specific proteasome interaction networks. Protein network analysis combined with cluster analysis has led to the direct physical connection between the proteasome and the pheromone induced signaling pathway. This study represents the first step toward understanding of how dynamics of the proteasome complex itself is involved in regulating cell cycle progression.