The findings presented here begin to unravel the mechanism by which Gcn4 degradation is regulated: a) inhibition of protein synthesis, even in the absence of starvation, is sufficient to stabilize Gcn4; b) the SCFCDC4 ubiquitination complex is required for Gcn4 degradation, but its activity is not affected by starvation; c) Gcn4 degradation requires the activity of the cyclin-dependent kinase Pho85; d) a specific domain of Gcn4 is required for its starvation-sensitive degradation; e) a specific site within that region, Thr165, is required for Gcn4 degradation; f) phosphorylation of this site depends on Pho85 activity and correlates with the conditions that allow rapid degradation of Gcn4; g) Pho85 is able to phosphorylate Thr165; and h) Pho85 activity toward Gcn4 is reduced in starved cells. The simple model emerging from these results is that Gcn4 phosphorylation by Pho85 on Thr165 transforms it into a substrate of the SCFCDC4 ubiquitination complex (Figure ). The differential degradation of Gcn4 versus other SCFCDC4 substrates is explained by the different kinases required to transform the proteins into effective substrates of the ubiquitination complex. The regulation of Pho85 activity by starvation explains the stabilization of Gcn4 under these conditions.
A model of the proposed pathway of regulation of Gcn4 degradation by starvation. See text for details.
Recognition sites of the SCF complexes are ill defined, beyond the general requirement for phosphorylation. Deletion analysis allowed to define two separate domains, I and II, within the region of Gcn4 required for its rapid degradation (Figure ). We previously described a number of mutations in and around position 105 that stabilize Gcn4 (Kornitzer et al., 1994
). Thr105 is located in domain I. Thr165, which we characterize in the present study, is located in domain II. It is possible that these two residues constitute each the core of partially redundant SCFCDC4
recognition sites. Phosphorylation of the 62–202 fragment of Gcn4 (Figure ), which carries both sites, suggests that in vivo, Pho85 activity is required for Thr165 phosphorylation but that phosphorylation of (an)other site(s) in this fragment, possibly Thr105, does not require Pho85. Conversely, our in vitro phosphorylation data (Figures and ) indicate that this same fragment can be phosphorylated by Pho85 at Thr165 as well as at another site, possibly Thr105. Thus, Pho85 activity is necessary and sufficient only for phosphorylation of Thr165.
Interestingly, contrary to Nishizawa et al. (1998)
, who found that recombinant Pho85-Pcl1 will phosphorylate Sic1 only after being “activated” with a yeast cell extract, we found that phosphorylation of Gcn4 does not require such an activation. Our results are in agreement with the recent findings of Wilson et al. (1999)
, that Gsy2 is efficiently phosphorylated in vitro by recombinant Pho85-Pcl10. Thus, it is possible that in the case of Sic1 phosphorylation, the cell extract confers a specificity factor required for Sic1 recognition, rather than an activation factor.
The Thr165 mutation was not isolated in our previous screen, probably because that screen depended on the enhanced transcriptional activity of Gcn4—the Thr105 mutation yielded a 15-fold increase in transcriptional activity (Kornitzer et al., 1994
), whereas the Thr165 mutation only increased transcription 5-fold (Figure ). Interestingly, Thr105 lies within the region defined by deletion analysis as the Gcn4 activation domain (Hope and Struhl, 1986
; Hope et al., 1988
). More recent mutational mapping showed that the activation domain can be subdivided into two redundant subdomains, the N-terminal and central acidic activation domains (Drysdale et al., 1995
; Jackson et al., 1996
). The Thr105 region lies between these two subdomains. Thus, it is possible that the mutations that were isolated in that region, in addition to stabilizing the protein, increase its specific transcriptional activity by modification of the activation region. In line with the recent findings of Komeili and O'Shea (1999)
, showing that phosphorylation of the transcription factor Pho4 by Pho85 affects its transcriptional activity, it is conceivable that the role of phosphorylation at Thr105 is to modulate the transcriptional activity of Gcn4 more than its stability.
The main difference between the requirements of Gcn4 degradation compared with that of the cell cycle substrates of SCF is the specific kinase involved. Gcn4 degradation requires Pho85 activity, whereas the other known SCF substrates require Cdc28 activity. The known functions of Pho85 in association with specific cyclins relate to metabolic regulation, e.g., phosphate assimilation (cyclin: Pho80; Kaffman et al., 1994
), or glycogen synthesis (cyclins: Pcl8 and -10; Huang et al., 1998
). In this respect, a function in Gcn4 degradation and therefore, indirectly, in amino acid metabolism, fits this pattern. The fact that the degradation of the cell cycle substrates of the SCFCDC4
complex depends on the phase of the cell cycle, whereas Gcn4 degradation is constitutive during the cell cycle, can be accounted for by difference in the kinase involved. Indeed, the nine Cdc28-associated cyclins, which confer its substrate specificity, are each present only during a limited window of each cell cycle. In contrast, most, but not all, of the 10 Pho85 cyclins are present throughout the cell cycle (Andrews and Measday, 1998
). Strikingly however, Pcl1 is one of the Pho85 cyclins that is strongly cell cycle regulated, at least at the level of transcription (Measday et al., 1997
). This apparent contradiction would be resolved if different Pcls were able to promote Gcn4 phosphorylation. Two observations indicate that this is in fact the case: a) if Pcl1 were the only Pcl able to promote Gcn4 phosphorylation, then Gcn4 should be stabilized in a pcl1Δ
mutant. However, a number of single and multiple combinations of pcl
mutants that we tested, including pcl1Δ
, are unaffected in Gcn4 degradation (our unpublished results); and b) preliminary data indicate that at least one other Pcl, Pho80, is able to promote Gcn4 phosphorylation on Thr165 by Pho85 in vitro (our unpublished results). These two findings indicate that there may be a high redundancy in the Pcls able to activate Pho85 for Gcn4 phosphorylation.
How is Gcn4 phosphorylation by Pho85 regulated by starvation? Starvation could induce an inhibitor of Pho85 activity. Previous studies demonstrated that inhibition of an amino acyl-tRNA–charging enzyme also leads to Gcn4 stabilization (Kornitzer et al., 1994
), suggesting that uncharged tRNA constitutes the signal for Gcn4 stabilization. An elevated concentration of uncharged tRNA is also thought to constitute the primary signal for the translational control of Gcn4 (Wek et al., 1995
). In the current study, we found that inhibition of translation by cycloheximide is sufficient to stabilize Gcn4. Under these conditions, charged
rather than uncharged tRNA accumulates in the cell. It should be noted that this does not contradict the earlier results obtained with the aminoacyl-tRNA–charging enzyme mutant; indeed, inhibition of one of the 20 aminoacyl tRNA synthetases, although resulting in an increase in the concentration of the uncharged form of its cognate tRNAs, would also, via inhibition of translation, increase the concentration of the charged
form of all the other tRNAs. However, there is no direct evidence that Pho85 activity is regulated by the charged tRNA concentration or by any other signal generated by the stalled cellular biosynthetic machinery. An alternative possibility is that continued protein synthesis per se is required for maintaining Pho85 activity; therefore, starvation, or direct inhibition of protein synthesis, would result in reduced kinase activity.
Limitation of protein synthesis is known to arrest the cell cycle in G1 (Unger and Hartwell, 1976
; Shilo et al., 1978
; Pardee, 1989
). The fact that the SCFCDC4
complex is required for the G1/S transition and for Gcn4 turnover raised the possibility that amino acid starvation coordinately inhibits cell cycle progression and Gcn4 degradation via inhibition of the SCFCDC4
complex (Kornitzer et al., 1994
). However, our data do not support this hypothesis. Cdc6 and Sic1, two other substrates of the SCFCDC4
complex, are not stabilized upon starvation. Rather, our data indicate that Gcn4 degradation is regulated via modulation of Pho85 activity. Might modulation of Pho85 activity coordinately regulate Gcn4 turnover and cell cycle progression? Recent reports indicate a role for the G1 cyclin Cln3 in the cell cycle arrest in face of reduced protein synthesis (Polymenis and Schmidt, 1997
; Danaie et al., 1999
). However, it cannot be excluded that modulation of Pho85 activity also participates in this regulation. Pho85, although not essential for cell cycle progression, displays cell cycle–related phenotypes. For example, a deletion of the Cdc28 cyclins CLN1
is synthetically lethal with a deletion of the Pho85 cyclins PCL1
or of PHO85
itself (Espinoza et al., 1994
; Measday et al., 1994
). Thus, it is possible that Pho85 is one of the transducers of physiological signals, such as amino acid starvation, that need to be integrated by the cell cycle machinery.