Mitochondrial respiration promotes Cdc37-dependent stability of the Cdk1 homolog Cdc28

The eukaryotic cell cycle is controlled by cyclin-dependent kinases (CDKs). In Saccharomyces cerevisiae, a single CDK, Cdc28, is necessary and sufficient for cell cycle regulation. Cdc28 (Cdk1 in mammals; Cdc2 in S. pombe) targets more than 200 substrates, and its activity and specificity are provided by the binding to one of nine cyclin subunits, divided in two groups, the G1 cyclins and B-type cyclins. The regulation of the different phase-specific cyclins is tightly controlled by oscillatory mechanisms that provide alternate periods of low and high levels of the different cyclin–Cdc28 activities, ensuring orderly progression through the cell cycle (Spellman et al., 1998; Orlando et al., 2008). In addition to cyclins, Cdc28 activity is regulated by the presence of Cdc28 inhibitors and by post-translational modifications (PTMs) such as phosphorylation (Mendenhall and Hodge, 1998). Transcriptional and translational regulation of Cdc28 seems to be of minor importance, given that the amount of Cdc28 is normally present at constant levels (and in excess) throughout the cell cycle (Mendenhall and Hodge, 1998). Cdc28 steady-state levels are promoted by Cdc37 (Farrell and Morgan, 2000), an Hsp90 co-chaperone that specifically targets protein kinases (Kimura et al., 1997). Cdc37 promotes Cdc28 folding and/or stabilization, and mediates Cdc28 activation by promoting formation of the Cdc28–Cln and Cdc28–Clb2 complexes (Mendenhall and Hodge, 1998), and is thus critical for proper cell cycle progression.

Cdc28 controls the timing of mitotic commitment, DNA replication and repair, spindle formation, chromosome separation and cytokinesis (Mendenhall and Hodge, 1998). It also affects processes not directly associated with the cell cycle, but coordinated with it, such as cell growth and bud morphogenesis (McCusker et al., 2007), lipid synthesis (Kurat et al., 2009), carbohydrate metabolism (Ewald et al., 2016) and vacuole inheritance (Peng and Weisman, 2008). Cdc28 has also been reported to impact on mitochondrial DNA (mtDNA) stability (Devin et al., 1990; Koltovaya et al., 1998) and to favour mitochondrial protein import and bioenergetic activity at G2/M phase (Harbauer et al., 2014).

Mitochondria play a number of key roles in eukaryotic cells, including ATP synthesis by oxidative phosphorylation (OXPHOS). The OXPHOS system is composed of four respiratory chain complexes, which are responsible for electron transport and generation of a proton gradient across the inner mitochondrial membrane, and the ATP synthase uses this proton gradient to drive the aerobic synthesis of ATP. S. cerevisiae is tolerant to mutations that inactivate the electron transport chain or OXPHOS due to its good fermenting capacity. In fact, this species can tolerate a partial deletion or complete loss of mtDNA when grown on a fermentable carbon source. Even so, the mitochondrial compartment is essential for yeast cell growth as it is required for essential biosynthetic pathways, such as amino acid synthesis and iron-sulfur cluster biogenesis (Lill et al., 2012). It is therefore not surprising that severe defects in mitochondrial inheritance trigger a checkpoint that inhibits cytokinesis (García-Rodríguez et al., 2009). Plus, cells that have lost their mtDNA (rho0 cells) exhibit defects in G1 to S progression (Crider et al., 2012). Taken together, these pieces of evidence point to a two-way communication between mitochondrial function and cell cycle progression.

Here, we report that mitochondrial OXPHOS mutations inhibit growth in Cdc28 conditional mutants by impairing the activity of the Hsp90 co-chaperone Cdc37 and further compromising Cdc28 levels. This work presents mitochondrial dysfunction as a new factor affecting Cdc37 activity, expanding our knowledge on the complex networks involved in cell division decisions.

In this study, to improve our understanding of the interplay between cell cycle regulation and mitochondrial function, we investigated the genetic interaction between CDC28 and ATP2, which encode a key cell cycle regulatory kinase and the catalytic subunit of the F_oF₁-ATP synthase, respectively. In the presence of glucose, yeast cells generate ATP primarily by glycolysis, and the presence of an intact OXPHOS is not essential for growth (Warburg, 1956). As such, to allow growth of atp2Δ cells, strains were cultivated on glucose-based media (fermenting conditions) throughout this work. Given that the CDC28 gene is essential, we used a Cdc28 degron system (cdc28td), based on a C-terminal fusion with a heat-inducible ubiquitin ligase-target peptide (Dohmen et al., 1994), to deplete Cdc28 by shifting cells from 23°C to 37°C. Cdc28 levels after 3 h at the temperatures used in this study are shown in Fig. 1A. In fermenting conditions, wild-type (wt), atp2Δ, cdc28td and atp2Δcdc28td cells cultured at the permissive temperature of 23°C grew similarly (Fig. 1B). However, at the semi-permissive temperature of 35°C, for which only a mild growth defect is apparent for cdc28td cells, a growth lethality was observed in the atp2Δcdc28td double mutant. Reintroducing wt Cdc28 (from its natural promoter in a low-copy number vector) reverted the atp2Δcdc28td mutant growth defects (Fig. S1). Growth defects in the atp2Δcdc28td mutant were already evident for temperatures as low as 26°C, in which no growth defects are observed for cdc28td cells (Fig. 1B). At 26°C, although cdc28td cells exhibit normal growth, cells exhibited an altered morphology (Fig. 1B). This indicates that the cdc28td protein activity is suboptimal, and that potential conformational alterations in the mutant due to the presence of the degron tag not being compensated for by the increased protein levels (Fig. 1A). Absence of Atp2 aggravated the morphological alterations of cdc28td cells, leading to abnormally large, elongated cells even at 26°C (Fig. 1C). As growth of the atp2Δ single mutant is not affected under fermenting conditions, irrespective of the growth temperature, these results suggest a synergistic interaction between CDC28 and ATP2.

An analysis of DNA content by flow cytometry showed that at the permissive temperature all strains showed normal distribution in the cell cycle. Although depletion of Cdc28 is expected to arrest the cell cycle at any given cell cycle phase (Papagiannakis et al., 2017), the cdc28-td strain showed a mild enrichment in cells at S phase after 4 h at the restrictive temperature (37°C). Notably, we found this S phase arrest was further aggravated by the loss of Atp2 (Fig. 1D).

Since distinct temperature-sensitive (ts) alleles of Cdc28 are available that cause arrest at specific cell cycle phases, such as G1 (cdc28-1) or G2/M (cdc28-1N) (Neiman et al., 1990; Surana et al., 1991), we tested whether lethality of ATP2 disruption might be related to a specific cell cycle phase. We found that ATP2 deletion also promoted a severe slow growth phenotype in cdc28-1 and cdc28-1N cells (although to a lesser degree), at the respective semi-permissive conditions (Fig. 2A). Our results are consistent with Atp2 being critical for cell proliferation upon low Cdc28 activity both in G1 (Start) and mitosis. In addition, loss of Atp2 did not lead to a growth arrest at a specific cell cycle phase as it aggravated each of the cdc28 mutant distinctive cell cycle phase arrest profiles at the permissive temperature of 26°C (Fig. 2B).

To investigate whether ATP2 genetically interacts with additional unrelated cell cycle mutants, we introduced in atp2Δ cells ts mutations for the yeast Aurora kinase Ipl1 and the anaphase-promoting complex/cyclosome (APC/C) adaptor subunit Cdc20. Absence of ATP2 did not affect the growth of either ipl1-1 or cdc20-1 cells at respective semi-repressive temperatures (Fig. 3A). We also found no genetic interaction between ATP2 and the polo-like kinase-encoding gene CDC5 (Fig. S2). These data indicate that the growth defect caused by ATP2 deletion in cdc28 mutants is not a phenotype shared with other cell cycle mutants. Atp2, as an essential subunit of an OXPHOS component, plays a vital role in ATP production. Yet, during growth in glucose medium, the glycolytic flux is high and the mitochondrial respiratory chain is repressed and does not contribute significantly to the cellular ATP levels (Warburg, 1956), as evidenced by the normal growth of atp2Δ single mutants. In addition, Atp2 might play cellular role(s) unrelated to ATP synthesis (Chen and Clark-Walker, 1999). To support the hypothesis that the negative effect of ATP2 deletion when combined with cdc28td allele is related to the role of Atp2 in OXPHOS, genes encoding other ATP synthase subunits, namely ATP1 (F1 alpha subunit) and ATP4 (F_o subunit b) were deleted in cdc28td cells. Deletion of both ATP1 and ATP4 strongly inhibited cdc28td proliferation (Fig. 3B), indicating this is not a β-subunit-specific effect. We next questioned whether this phenotype was restricted to ATP synthase, or common to other OXPHOS complexes. For that, we tested the deletion of QCR2, a core subunit of Complex III (cytochrome bc1), and COX4, which encodes for a subunit of complex IV (cytochrome c oxidase), in cdc28td cells. Both qcr2Δcdc28td and cox4Δcdc28td strains phenocopied the atp2Δcdc28td double mutant lethality (Fig. 3B). Ablating mitochondrial DNA in a strain carrying the cdc28td allele (cdc28td rho0) also resulted in a similar phenotype (Fig. S3). Overall, these results suggest that mitochondrial respiration in general is essential for growth of cells with reduced Cdc28 levels.

The observed aggravating genetic interaction between OXPHOS-encoding genes and CDC28 suggest that they have a functional relationship. Given that previous studies have indicated that Cdc28 plays a role in mitochondrial function (Devin et al., 1990; Harbauer et al., 2014), we started by addressing whether loss of Cdc28 affected mitochondrial bioenergetics (respiration and mitochondrial membrane potential) and morphology of OXPHOS mutants. For mitochondrial respiration, wt, atp2Δ, cdc28td and atp2Δcdc28td strains cells were cultured in glucose medium to mid-log phase and incubated at the semi-permissive temperature (35°C) for 3 h. The basal oxygen consumption rate (OCR) of intact cells was then measured polarographically. The OCR was strongly reduced in atp2Δ and cox4Δ cells (2.4 and 0.8 nmol/min, respectively) compared with that found for wt cells (10 nmol/min), as anticipated. The cdc28td mutant, likewise cdc28-1 or cdc28-1N cells (Fig. S4), exhibited an increase in OCR (16.5 nmol/min) (Fig. 4A). In addition, OCR was higher in atp2Δcdc28td versus atp2Δ cells, indicating some level of mitochondrial uncoupling in the double mutant. As expected, absence of COX4 drastically reduced the OCR in the cdc28td mutant since it is essential for respiration.

Next, we analysed the mitochondrial morphology of these mutants expressing a mitochondria-targeted green fluorescent protein (mtGFP), and visualized by fluorescence microscopy. In agreement with previous observations, the vast majority (97%) of mitochondria in wt cells displayed a filamentous morphology. Similar results were observed in single mutants (88% for atp2Δ and 93% for cdc28td mutant). In contrast, the predominant phenotype of the atp2Δcdc28td double mutant was punctate mitochondrial morphology (69%) indicating fragmentation (Fig. 4B; Fig. S5). We also assessed mitochondrial membrane potential (Δσm), as it reflects OXPHOS activity and is essential for mitochondrial function and biogenesis (Martin et al., 1991). The Δσm was evaluated by flow cytometry using the cationic lipophilic dye DiOC6(3), in cells incubated for 3 h in repressive conditions (35°C). Addition of the FCCP protonophore, which dissipates the Δσm, led to a substantial reduction in the DiOC6(3) uptake by all strains (Fig. S6). We used the FCCP responsive potential, eliminating unspecific binding, in the quantification of the Δσm shown in Fig. 4C. The Δσm was also normalized to the mitochondrial mass evaluated using preSU9-GFP, a mitochondrial presequence with low sensitivity to the membrane potential (Vowinckel et al., 2015). We found a significant increase (2.7-fold) of Δσm in atp2Δcdc28td mutants, compared with that in wt cells (Fig. 4C). For the single mutant strains, no significant differences were observed. The mitochondrial localization of the fluorochrome was checked by epifluorescence microscopy and corroborated the flow cytometry results (Fig. S6). These data revealed that Cdc28 impacts on mitochondrial respiration, and loss of Cdc28 in atp2Δ cells leads to mitochondrial dysfunction, as indicated by the mitochondrial hyperpolarization, fragmentation and increase of uncoupled respiration.

Mitochondrial dysfunction can trigger an adaptive programme named the mitochondrial retrograde response (RTG), a pathway of communication from mitochondria to the nucleus (Butow and Avadhani, 2004). Previous studies have shown that activation of the RTG prevents cell cycle progression by upregulating Swe1, a Cdc28 inhibitor (Chen et al., 2010), and contributes to cell death during prolonged S/G2 arrest (Zyrina et al., 2015), indicating a link between RTG signalling and cell cycle. This prompted us to examine the possibility that mitochondrial dysfunction in the atp2Δcdc28td mutant could be triggering the RTG, resulting in cell growth arrest.

The classical marker of RTG in yeast is the upregulation of CIT2, which encodes a glyoxylate cycle isoform of citrate synthase (Liao et al., 1991). CIT2 induction was evaluated using a CIT2-LacZ reporter that was genomically integrated in the strains under study and in a rho0 background that served as a positive control. We found that ATP2 deletion (or COX4 deletion, data not shown) did not induce CIT2-LacZ activation, either alone or in combination with the cdc28td allele. Similar results were observed in cdc28-1 and cdc28-1N mutants lacking Atp2 (Fig. 4D). Intriguingly, LacZ activity was even lower in the cdc28td and cdc28-1N single mutants or in combination with Atp2 loss compared to that in wt, but no changes were observed in cdc28-1 cells. Alterations in CIT2 basal expression may be caused by the distinct cell cycle phases at which the mutants arrest, given that CIT2 expression varies with cell cycle phases (Campbell et al., 2020; Barberis and Mondeel, 2022). Yet, as CIT2-LacZ activation was identical for cdc28 single and atp2Δcdc28 double mutants, it seems unlikely that the RTG pathway is involved in the growth defects of the atp2Δcdc28 double mutant. In agreement, downregulation of the RTG pathway by deleting RTG3 did not rescue the atp2Δcdc28td mutant growth defects (Fig. 4E). The RTG pathway can be regulated by the levels of glutamate in the medium, whereby low glutamate levels activate and high glutamate levels repress RTG-dependent gene expression (Liu et al., 2001). Corroborating previous results, we found that growth in medium lacking or containing excess glutamate had no impact on the atp2Δcdc28td mutant growth (Fig. S7). Overall these results show that activation of the RTG signalling pathway does not play a significant role in the lethal phenotype of the atp2Δcdc28td mutant at semi-restrictive temperature.

Given that the cell morphology and cell cycle arrest phenotypes of atp2Δcdc28 mutants resemble an aggravated cdc28 phenotype, we considered whether Cdc28 steady-state protein levels could be affected in the OXPHOS mutants. We evaluated wt Cdc28 and cdc28td levels in the absence of Atp2, Qcr2 and Cox4, in cells grown at the permissive temperature (26°C). No alterations in wt Cdc28 levels were observed in the absence of the OXPHOS proteins. However, cdc28td levels were noticeably lower in these cells compared with controls (Fig. 5A). This phenotype was also observed in the ts mutant cdc28-1, which was strongly destabilized by the absence of Atp2 (Fig. 5B). A 120-min time course analysis of cdc28td levels at restrictive conditions showed that Cdc28 levels decreased overtime both in cdc28td and atp2Δcdc28td strains. However, in the absence of Atp2, as the Cdc28 protein level was already lower at t0, the detection limit was achieved earlier, after 1 h of incubation (Fig. 5C). Given that Cdc28 is essential for proliferation, our data suggest that the destabilization of cdc28 upon OXPHOS disruption is the primary cause of the aggravated phenotype in the double mutant.

Cdc28 levels are usually constant during cell division. However, Cdc28 folding and/or stabilization requires the co-chaperone Cdc37 (Farrell and Morgan, 2000), a subunit of the chaperone Hsp90 that specifically targets protein kinases (Kimura et al., 1997). Given that Cdc37 loss-of-function mutations can lead to decreased Cdc28 levels (Gerber et al., 1995), we investigated whether defects in this chaperone system could underlie the phenotype of the OXPHOS mutants carrying the cdc28td allele. For that, we tested the sensitivity of the strains to geldanamycin (GA), an Hsp90 inhibitor that prevents the association of Hsp90–Cdc37 complexes with client protein kinases, inhibiting their maturation (Bandhakavi et al., 2003). As shown in Fig. 6A, atp2Δcdc28td and cox4Δcdc28td double mutants were hypersensitive to 100 µM GA at the permissive temperature, suggesting that Hsp90–Cdc37 activity must be affected in OXPHOS/cdc28 double mutants. Interestingly, OXPHOS and cdc28td single mutants exhibited a sensitivity to GA similar to wt cells.

The Hsp90–Cdc37 complex controls the folding of a large proportion of protein kinases (Mandal et al., 2007). We investigated the levels of two Hsp90–Cdc37 client kinases involved in distinct signalling pathways, Slt2 and Yck2, identified by a reduced accumulation in cells expressing a Cdc37^S14A mutant protein (Mandal et al., 2007) that exhibits low Cdc37 activity (Bandhakavi et al., 2003). The steady-state levels of Slt2 and Yck2 were determined in wt, atp2Δ, cdc28td and atp2Δcdc28td cells cultured at 30°C. Our results show that Slt2 and Yck2 accumulate at reduced levels in the atp2Δcdc28td double mutant, but not in single mutants (Fig. 6B,C). This result indicates that the combination of respiratory and cdc28 defects impacts on the stability of multiple kinase substrates of the Hsp90–Cdc37 system. We confirmed the reduced accumulation of these kinases in the cdc37^S14A cells (Fig. 6B).

We also investigated the levels of two kinases, Snf1 and Ypk1, whose levels were reportedly unchanged in cdc37^S14A cells (Mandal et al., 2007). As they are not likely Hsp90–Cdc37 targets, we predicted they would not be affected in the strains in study. Unexpectedly, Snf1 protein levels were also strongly decreased in atp2Δcdc28td cells in comparison with wt and single mutant strains (Fig. 6B,C). Interestingly, though not statistically significant, both Yck2 and Snf1 exhibit a tendency to accumulate at higher levels in the single atp2Δ mutant. Unlike Snf1, the endogenous Ypk1 levels were unchanged in atp2Δcdc28td cells, as expected (Fig. 6B,C).

To assess which of the Hsp90–Cdc37 complex components might be affected in the atp2Δcdc28td cells, we overexpressed Cdc37 and Hsp90 (encoded by HSC82 and HSP82 in S. cerevisiae) independently. Overexpression of Cdc37, the component providing kinase specificity, but not Hsc82 or Hsp82, supressed the atp2Δcdc28td mutant growth defects at 29°C (Fig. 7A), and had a partial protective effect at higher temperatures (32°C). Notably, the growth rescue promoted by Cdc37 overexpression in the atp2Δcdc28td mutant correlated with a partial restoration of Cdc28 and Snf1 levels (Fig. 7B). These results suggest that Cdc37 activity is affected in the atp2Δcdc28td mutant, underlying its growth defects. Notably, Cdc37 protein levels were not altered in OXPHOS mutants (single or in combination with cdc28td) (Fig. S8; Fig. 7B). Thus, it seems post-translational regulation of Cdc37 might be compromised in these mutant cells given that PTMs are vital for its activity (Bandhakavi et al., 2003; Shao et al., 2003). As phosphorylation on Ser14/Ser17 is essential for Cdc37 function by promoting its protein kinase-binding activity (Bandhakavi et al., 2003; Shao et al., 2003), we evaluated growth of atp2Δcdc28td cells transformed with either empty vector or plasmids expressing wt Cdc37, nonphosphorylatable Cdc37-S14A/S17A (Cdc37-AA) or phosphomimetic Cdc37-S14E/S17E (Cdc37-EE). Cdc37, but not Cdc37-AA or Cdc37-EE, rescued atp2Δcdc28td growth defects in semi-repressive conditions (Fig. 7C), suggesting that the phosphoregulation of Cdc37 is important for its protective effects.

Mitochondria are not only at the centre of aerobic energy production and essential biosynthetic activities, but are also tightly interconnected to other cellular processes and signalling cascades important for cell proliferation and survival. The present work point to a role of mitochondria in cell cycle progression by revealing that combined disturbance of mitochondrial respiration and Cdc28 activity inhibits cell proliferation. Importantly, this effect is due to inhibition of Cdc37, which in turn leads to a further decrease in Cdc28, underscoring more complex interactions between these proteins than previously anticipated. Cdc28 plays a major role in cell division and its activity is carefully controlled by checkpoint mechanisms that monitor cues from the environment and their own internal state (Mendenhall and Hodge, 1998). Our results suggest that under less than optimal conditions, when Cdc28 activity is reduced, mitochondrial fitness will be decisive for cell cycle progression decisions.

The aggravated phenotype in the OXPHOS/cdc28 double mutants (mitochondrial fragmentation and hyperpolarization) suggests that Cdc28 might affect a pathway in common with mitochondrial respiration. Previous reports have already given evidence that Cdc28/Cdk1 is required for normal mitochondrial function in both yeast and mammals (Genta et al., 1995; Harbauer et al., 2014; Taguchi et al., 2007). In mammalian cells, Cdk1 favours mitochondrial fragmentation during mitosis by phosphorylating the dynamin-related GTPase Drp1 (yeast Dnm1) (Taguchi et al., 2007). Whether Cdc28 also targets the mitochondrial fusion/fission machinery in yeast is still unknown. Curiously, given that a repressible mutant of Cdc37 also leads to abnormal fragmentation (Altmann and Westermann, 2005), we cannot discard the hypothesis that the observed mitochondrial morphology defects in the OXPHOS/cdc28 double mutants is a consequence, and not a cause, of Cdc37 inhibition. Apparently, mitochondrial function might affect more and be more affected by the cell cycle machinery than previously anticipated.

A remaining question is how the combined impairment of mitochondrial respiration and Cdc28 activity leads to an inhibition of Cdc37. In this work, we have discarded the hypothesis that the aggravated mitochondrial dysfunction in the OXPHOS/cdc28 mutants triggers the RTG response. Given that we observed that Cdc37 protein levels are not altered in this strain, Cdc37 inhibition might be associated with changes at the PTM level. Cdc37 PTMs can occur in several amino acid residues, many of unknown physiological relevance, as listed in the SGD database (Hellerstedt et al., 2017). A prominent PTM is phosphorylation on S14 and/or S17 (residue S13 in the human protein), as this modification is considered essential for Cdc37 function (Bandhakavi et al., 2003; Shao et al., 2003). However, the expression of a phosphomimetic Cdc37-EE version, comparing to the phosphoresistant Cdc37-AA, did not have a protective effect on the atp2Δcdc28td mutant, arguing against this hypothesis. Plus, Snf1 stability decreased in the atp2Δcdc28td mutant (our results) but not in cdc37^S14A cells (Mandal et al., 2007). In our conditions, Snf1 even accumulated at higher levels in cells expressing the Cdc37-S14A mutant. Phosphorylation of S14 is normally crucial for the stabilization of Cdc37 client kinases, with exceptions. For example, Cak1 accumulates at higher levels upon loss of Cdc37-S14 phosphorylation (Millson et al., 2014). It is conceivable that PTMs other than S14 phosphorylation might impact on the Cdc37 activity towards specific kinases.

Expanding our knowledge on the complex interactions between mitochondrial dysfunction and impaired kinase activity will improve our understanding of the multifactorial interactions controlling proliferation and potentially disease. In humans, a prominent number of Cdc37 targets are oncoproteins (Miyata and Nishida, 2004) and aberrant CDK activity also underpins proliferation of tumour cells (Malumbres and Barbacid, 2009), making research on the regulation of these proteins of considerable interest. Plus, mitochondrial metabolism is altered in cancer cells and OXPHOS inhibitors are gaining attention as therapeutic approaches (Ashton et al., 2018). Given that inhibition of both Cdc28 and OXPHOS have a strong growth inhibitory effect in our model, we postulate that the synergistic use of CDK and OXPHOS inhibitors might also be advantageous in antitumour therapeutics.

The S. cerevisiae strains used in this study are all BY4741 derivative and are listed in Table S1. The atp1Δ, atp4Δ, qcr2Δ and cox4Δ strains were constructed by replacing the open-reading frame (ORF) of each gene with the auxotroph marker HIS3, using PCR-mediated homologous recombination. To construct double mutant strains, the cassette containing cdc28td:KanMX4, cdc28-1:KanMX4, cdc20-1:KanMX4 and ipl1-1:KanMX4 were amplified by PCR and transformed in indicated deletion strains. To construct the cdc28-1N-expressing strains, the cdc28-1N sequence was amplified from the US102 strain (Loy et al., 1999) and fused to the KanMX6 cassette, using overlap extension PCR. This product was then integrated into wt and atp2Δ strains. pBL101-pCIT2-LacZ (Liu and Butow, 1999) was linearized by digestion with SmaI and integrated into the genomic CIT2 locus. Correct deletions/insertions were confirmed by PCR. For Hsp90 and Cdc37 overexpression, the auxotroph marker HIS3 in the atp2Δcdc28td strain was replaced by URA3 by PCR-mediated recombination, and cells were transformed with empty vector, pAG423GAL-HSC82, pAG423GAL-HSP82 or pAG423GAL-CDC37 (Lin et al., 2021). For kinase levels assays, wt, atp2Δ, cdc28td, atp2Δcdc28td cells were transformed with YEp352-SLT2-3HA (Madden et al., 1997), pRS316-SNF1-HA (Ye et al., 2008) and p416-ADH-yeGFP-YCK2. To generate p416-ADH-yeGFP-YCK2, the yeGFP sequence was inserted into the XbaI site of the p416ADH plasmid followed by insertion of the YCK2 sequence (BamHI-SalI fragment) amplified. Strains TM141 cdc37Δ::HIS3 [Ycplac111 CDC37-HA] and cdc37Δ::HIS3 [Ycplac111 CDC37-S14A-HA] (Hawle et al., 2007) were also transformed with the above plasmids and used as control. For growth assays, Ycplac111-CDC37, CDC37S14/17A and CDC37S14/17E were isolated from TM141 strains (Vaughan et al., 2008) and plasmids used to transform cdc28td and atp2Δcdc28td cells.

Strains were transformed by the standard lithium acetate procedure (Gietz and Schiestl, 2007). To avoid clonal variability, at least two different colonies were analysed in all experiments.

Cells were grown in YPD rich medium [1% (w/v) yeast extract, 2% (w/v) peptone and 2% (w/v) glucose; Thermo Fisher Scientific]. YPGal/Raf containing 2% (w/v) galactose (Biosynth) plus 2% (w/v) raffinose (Biosynth) instead of glucose as a carbon source, was used for the induction of gene expression from plasmids with the GAL1 promoter. Cultures were routinely grown in an orbital shaker at 140 rpm.

For clone selection, strains were grown in YPD agar supplemented with 250 µg/ml geneticin (Santa Cruz Biotechnology) or in synthetic complete medium [SC; 0.67% (w/v) Bacto-yeast nitrogen base without amino acids (BD Difco), 2% (w/v) glucose and 0.2% (w/v) dropout mix; Sigma] lacking histidine or uracil, as appropriate.

For geldanamycin (GA; MedChemExpress)-supplemented medium, GA was dissolved in dimethyl sulfoxide and added to a final concentration of 100 µM to warm YPD agar medium before pouring.

All strains are available upon request.

For the β-galactosidase assay, yeast cells expressing pCIT2-LacZ were grown in YPD to exponential phase. Cells were incubated for 3 h at the restrictive temperature to inactivate cdc28td (37°C), cdc28-1N (37°C) and cdc28-1 (35°C). The β-galactosidase activity was measured in a liquid assay using o-nitrophenyl-β-D-galactoside (ONPG) as previously described (Almeida et al., 2008), using 60 µg of total protein.

For cell cycle analysis, strains were cultured overnight in YPD at 24°C to mid-log phase. A portion of the cultures was then shifted to 37°C for 3 h, while the remaining was maintained at 24°C for the same time and used as the control. Cells were fixed overnight and DNA content was analysed by flow cytometry using 10 µM SYTOX Green (Molecular Probes), basically as described in Haase and Reed (2002). Cells were analysed using the FL1 detector in a BD Accuri C6. 20,000 events were analysed per sample and cell cycle distribution was estimated using FlowJo v10 software version.

To assess the mitochondrial membrane potential (ΔΨm), cells were incubated for 30 min at 30°C with 40 nM of 3,3 dihexyloacarbocynine iodide [DiOC6(3); Molecular Probes; Pereira et al., 2014]. When used, the uncoupler carbonyl-cyanide 4-(trifluoromethoxy)-phenylhydrazone (FCCP, Sigma; 10 μM) was added 10 min before loading with DiOC6(3). At least 20,000 cells were analysed per sample using the FL1 detector in a BD Accuri C6 flow cytometer. The collected data was analysed using FlowJo v10 software version to determine the mean green fluorescence intensity.

The oxygen consumption rate was measured polarographically in whole cells (10⁸ in PBS buffer) from cultures grown in YPD medium to late log phase, using a Clark-type oxygen electrode coupled to an Oxygraph plus system (Hansatech). Data were analysed using the OxyTrace⁺ software. The respiratory rate in the presence and absence of the inhibitor potassium cyanide (700 µM; Merck) was obtained by dividing the oxygen consumed per min by the number of cells used in the experiment.

For the analysis of mitochondrial morphology and volume, samples from strains transformed with a pVT100mtGFP fusion gene (Westermann and Neupert, 2000) were imaged by fluorescence microscopy. Images were acquired by epifluorescence in a Zeiss Axio Imager Z1 microscope fitted with Nomarski optics with an Axiocam MR3.0 camera and analysed using the Axiovision 4.7 software. Images were collected at 0.4 μm z-intervals and average projected to produce the final image. The mitochondria morphology types were analysed manually in ∼100 cells per strain, and the mitochondrial area in the cell was measured using the tools included in the ImageJ software (NIH).

Yeast total protein lysates were obtained by boiling in SDS sample buffer (without β-mercapthoethanol) at 95°C (5 min). Protein concentration was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) and β-mercapthoethanol was then added to the lysate. Proteins (40 µg) were separated by SDS-PAGE, electroblotted onto a nitrocellulose membrane (Hybond-C, GE Healthcare) and probed with antibodies using standard protocols. Primary antibodies used were anti-Cdc28 (1:150; sc-515762, Santa Cruz Biotechnology), anti-Cdc37 (1:15,000; ab188280, Abcam), anti-Ypk1 (1:500; sc-12051, Santa Cruz Biotechnology), anti-Snf1 (1:100; sc-15622, Santa Cruz Biotechnology), anti-Pgk1 (1:25,000; 22C5D8, Thermo Fisher Scientific), anti-GFP (1:6000; Clontech), and anti-HA (1:5000; sc-7392, Santa Cruz Biotechnology). Secondary antibodies used were anti-mouse IgG-HRP (1:7000, Molecular probes), anti-rabbit IgG-HRP (1:7000, Sigma) and anti-goat IgG-HRP (1:5000, Santa Cruz Biotechnology). Membranes were washed, incubated with WesternBright ECL (Advansta), exposed to LucentBlue X-ray film (Advansta), and scanned on a Molecular Imager GS800. Protein band intensities were quantified using Quantity One 1-D Analysis Software version 4.6 (Bio-Rad).

Full-length blots corresponding to the blots displayed in various figures and showing molecular mass markers are provided in Fig. S9.

We are very grateful to Uttam Surana (A*STAR, Singapore) for the cdc28-1N strain, Zhengchang Liu (University of New Orleans, LA, USA) for the CIT2-LacZ plasmid, John Pringle (Stanford University School of Medicine, CA, USA) for the pFA6a-kanMX6-PGAL1-3HA plasmid, and Martin L. Duennwald (University of Western Ontario, Canada) for the CDC37, HSC82 and HSP82 overexpression plasmids. The Cdc37 (WT), cdc37S14A, cdc37S14/17A, and cdc37S14/17E strains were generously provided by Marco Siderius (Vrije Universiteit Amsterdam, Netherlands) and Mehdi Mollapour (Upstate Medical University, NY, USA). The authors acknowledge the support of the i3S Scientific Platforms Translational Cytometry and Advanced Light Microscopy (member of the national infrastructure PPBI - Portuguese Platform of Bioimaging PPBI-POCI-01-0145-FEDER-022122).

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.260279.reviewer-comments.pdf