OXPHOS deficiency activates global adaptation pathways to maintain mitochondrial membrane potential

Abstract

Reduction of mitochondrial membrane potential (Δψ_m) is a hallmark of mitochondrial dysfunction. It activates adaptive responses in organisms from yeast to human to rewire metabolism, remove depolarized mitochondria, and degrade unimported precursor proteins. It remains unclear how cells maintain Δψ_m, which is critical for maintaining iron‐sulfur cluster (ISC) synthesis, an indispensable function of mitochondria. Here, we show that yeast oxidative phosphorylation mutants deficient in complex III, IV, V, and mtDNA, respectively, exhibit activated stress responses and progressive reduction of Δψ_m. Extensive omics analyses of these mutants show that these mutants progressively activate adaptive responses, including transcriptional downregulation of ATP synthase inhibitor Inh1 and OXPHOS subunits, Puf3‐mediated upregulation of import receptor Mia40 and global mitochondrial biogenesis, Snf1/AMPK‐mediated upregulation of glycolysis and repression of ribosome biogenesis, and transcriptional upregulation of cytoplasmic chaperones. These adaptations disinhibit mitochondrial ATP hydrolysis, remodel mitochondrial proteome, and optimize ATP supply to mitochondria to convergently maintain Δψ_m, ISC biosynthesis, and cell proliferation.

Oxidative phosphorylation deficiency causes mitochondrial membrane potential (Δψ_m) reduction and activates transcriptional and post‐transcriptional adaptions to maintain Δψ_m.

Introduction

Oxidative phosphorylation (OXPHOS) system consists of electron transport chain (ETC) complexes I–IV and the ATP synthase (complex V). ETC complexes oxidize reducing equivalents (NADH and FADH₂) and transport electrons to oxygen. Energy released by electron transport pumps protons from matrix to inter‐membrane space (IMS) to generate ∆ψ_m, which powers ATP synthesis by complex V. OXPHOS‐linked mitochondrial diseases are caused by mutations of over 150 genes; these genes encode OXPHOS subunits, cofactor synthesis, OXPHOS complex assembly factors, and mitochondrial gene expression system (mitochondrial DNA maintenance, transcription, and translation; Gorman et al, 2016; Craven et al, 2017).

Reduction of ∆ψ_m is a common and dangerous situation associated with OXPHOS impairment because ∆ψ_m drives the import of carriers and pre‐sequence guided mitochondrial precursors, including enzymes for iron‐sulfur cluster (ISC) synthesis (Chacinska et al, 2009; Pfanner et al, 2019). ISCs are essential cofactors that are synthesized in mitochondrial matrix and required for house‐keeping enzymes such as DNA polymerases and ribosome recycling factor (Lill, 2009). Reduction of ∆ψ_m compromises ISC synthesis to cause secondary defects in extra‐mitochondrial ISC‐containing enzymes (Veatch et al, 2009).

Oxidative phosphorylation impairment and the associated reduction of ∆ψ_m activate stress responses in diverse organisms. Early studies in yeast highlight that OXPHOS impairment activates a retrograde signaling pathway, the RTG pathway, to rewire metabolism (to synthesize glutamate; Epstein et al, 2001; Liu & Butow, 2006). Recent studies in yeast show that ∆ψ_m reduction as well as other forms of mitochondrial stresses impairs the import of mitochondrial precursor proteins to cause mPOS (mitochondrial precursor over‐accumulation stress; Wang & Chen, 2015) and activate the UPR^am (unfolded protein response activated by mistargeting of proteins; Wrobel et al, 2015), as well as the mitoCPR (mitochondrial compromised protein import response; Weidberg & Amon, 2018) to activate proteasome, inhibit cytoplasmic protein synthesis, and degrade precursors accumulated in the cytosol and clogged in the TOM import channel. In C. elegans, genetic and chemical disruption of OXPHOS decreases the mitochondrial import of transcription factor ATFS‐1, causing its nuclear translocation to upregulate glycolysis, repress the transcription of mitochondrial OXPHOS subunits, and promote mitochondrial biogenesis and quality control (Nargund et al, 2012; Nargund et al, 2015). In multi‐cellular organisms such as C. elegans, D. melanogaster, and human, ∆ψ_m reduction stabilizes PINK1 kinase on mitochondrial outer membrane (OM) to recruit ubiquitin E3 ligases Parkin to activate proteosomal degradation of mitochondrial proteins and mitophagy (Youle & Narendra, 2011). Recent studies showed that in mammalian cells, OXPHOS impairment activates inner membrane (IM) metalloprotease OMA1 to cleave DELE1, which then translocates to the cytosol to activate integrative stress response (Fessler et al, 2020; Guo et al, 2020).

These progresses are remarkable examples demonstrating how cells deal with the consequence of OXPHOS impairment and ∆ψ_m reduction by removing toxic precursor proteins and degrading dysfunctional mitochondria. But none has directly addressed how to maintain ∆ψ_m. Here, we have characterized the global adaptations of yeast OXPHOS mutants through multiple‐omics approaches. We report the surprising finding that accompanying increasing proteostatic stresses and reduction of ∆ψ_m, OXPHOS mutants progressively activate increasing number of adaptive responses. We report the key pathways and effectors of these adaptations and show they convergently maintain ∆ψ_m.

Results

Decreasing ∆ψ_m and proliferation rates among OXPHOS mutants

To model OXPHOS deficiency, we deleted complex III subunit Cor1 (cor1∆), complex IV subunit Cox9 (cox9∆), ATP synthase stalk subunit Atp7 (atp7∆), and mtDNA (ρ⁰). These cells were incapable of proliferation under respiratory growth condition (YPEG) (Appendix Fig S1A). Because some OXPHOS subunits are required for mtDNA maintenance (Merz & Westermann, 2009), we mated cor1∆, cox9∆, and atp7∆ cells with ρ⁰ cells to examine their mtDNA quality. The diploid cells had normal growth on YPEG plates (Appendix Fig S1A), indicating these mutants have intact mtDNA. For simplicity, we labeled cor1∆, cox9∆, atp7∆ as ∆III, ∆IV, and ∆V. Cells were cultured in synthetic medium with 2% glucose (SCD) to OD600 0.4–0.6, and RNAseq and proteomic analyses were performed within this OD range. At this condition, ρ⁰ cells consumed < 15% of total glucose (Appendix Fig S1B) and maintained exponential growth.

Although these mutants had common impairment of OXPHOS, we were surprised to find that comparing with wild‐type (WT) cells, ∆III and ∆IV cells had slightly decreased ∆ψ_m and proliferation rates, ∆V cells had lower and ρ⁰ cells had the lowest ∆ψ_m and proliferation rates (Fig 1A and B). Two factors may contribute to the increasing difficultness/stresses to maintain ∆ψ_m and proliferation among these mutants. First, mtDNA encodes subunits of complex III, IV, and V, and mtrRNAs, as well as mttRNAs and several other proteins. ρ⁰ cells have multiple disassembled OXPHOS complexes and disassembled mtRibosome, thus more proteostatic stresses as compared to other OXPHOS mutants. Second, the differential ∆ψ_m maintenance mechanisms among these mutants may cause extra stresses in ∆V and ρ⁰ cells (Appendix Fig S1C). In ∆III and ∆IV cells, ∆ψ_m is maintained by the reversal of complex V to hydrolyze ATP and export protons to IMS (Chinopoulos & Adam‐Vizi, 2010). In this way, ∆III and ∆IV cells hydrolyze 1 ATP to export 3.33 protons (yeast complex V contains a c₁₀ ring) and 1 positive charge (exchange of ADP^3‐ with ATP^4‐). ρ⁰ cells lose mtDNA‐encoded F_o components of complex V and maintain ∆ψ_m through F₁‐dependent ATP hydrolysis (Buchet & Godinot, 1998; Appleby et al, 1999). In this way, mitochondria import 1 ATP^4‐ at the exchange of 1 ADP^3‐ through the ATP/ADP antiporter Aac2, thus importing 1 negative charge at the cost of 1 ATP (Appendix Fig S1C). Theoretically, ∆V cells block proton flux into matrix and thus should have elevated ∆ψ_m. However, ∆V cells had intermediate ∆ψ_m between ∆III/IV and ρ⁰ cells (Fig 1A). We observed that ∆V cells had greatly reduced complex IV level (Appendix Fig S1D), similar to other mutants of complex V subunits (Hadikusumo et al, 1988; Marzuki et al, 1989; Paul et al, 1989; Spannagel et al, 1997; Marsy et al, 2008). This indicates that ∆ψ_m of ∆V cells is not maintained by the ETC. Interestingly, ∆ψ_m and the proliferation of ∆V cells were significantly impaired by deleting AAC2 (Appendix Fig S1E and F), indicating ∆ψ_m of ∆V cells is maintained by mechanism similar to that of ρ⁰ cells.

Figure 1. Yeast OXPHOS mutants have progressive reduction of ∆ψ_m and cell proliferation, and increasing activation of transcriptional responses.

1. Decreasing ∆ψ_m from ∆III to ρ⁰ cells. WT and mutant yeast cells were cultured to log phase and stained with TMRM for FACS analysis.
2. Decreasing proliferation rates from ∆III to ρ⁰ cells. Cells were cultured from 0.05 OD600 in SCD. OD600 was measured every 2 h.
3. Transcriptome analysis of OXPHOS mutants. WT and mutant yeast cells were cultured to log phase. RNAseq was performed with three biological replicates per genotype (Dataset EV1). All the transcripts and four categories of transcripts (gene lists in Table EV1B) are shown. Significantly changed genes (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed genes are indicated.
4. qRT–PCR analysis of selected transcripts in ρ⁰ cells.

Data information: Data are mean ± SD from three biological replicates (B, D).

Source data are available online for this figure.

Progressive activation of transcription responses in OXPHOS mutants

We performed RNAseq analysis to understand how OXPHOS mutants respond to mitochondrial stresses (Fig 1C and Dataset EV1). mRNA level changes of selected transcripts in ρ⁰ cells were confirmed by qPCR analysis (Fig 1D). Apparently, there were increasing transcriptional changes from ∆III to ρ⁰ cells. In ρ⁰ cells, there were 173 downregulated transcripts and 867 upregulated transcripts reaching the significance threshold (P < 0.05, fold change > 1.5 or < 0.67). Gene ontology analysis (Ashburner et al, 2000) of transcripts with P < 0.05 in ρ⁰ cells showed that the three most downregulated biological processes all relate to ribosome biogenesis (Fig EV1A) and mainly consist of transcripts encoding ribosome subunits and ribosome processing and assembly (Ribi) factors. The most‐upregulated processes are mitochondrial translation, mitochondrial gene expression, and carbohydrate metabolic process (glucose transporters and glycolysis genes) (Fig EV1A). We further analyzed transcripts encoding mitochondrial proteins (gene list in Table EV1A) with P < 0.05 in ρ⁰ cells and showed the upregulation of mitochondrial gene expression, mitochondrial organization, and mitochondrial translation, and the downregulation of ATP metabolic process and oxidative phosphorylation (both consist of OXPHOS subunits) (Fig EV1B). We thus collected four gene lists for further analysis in all the OXPHOS mutants, these genes encode: nuclear‐encoded OXPHOS subunits, mitochondrial proteins other than OXPHOS subunits, ribosome subunits and Ribi factors, as well as glycolysis enzymes and glucose transporters (gene lists in Table EV1B).

Figure EV1. Transcriptome analysis of yeast OXPHOS mutants. Related to Fig 1 .

1. Gene ontology analysis of significantly changed transcripts in ρ⁰ cells (P value < 0.05).
2. Gene ontology analysis of significantly changed transcripts encoding mitochondrial proteins in ρ⁰ cells (P value < 0.05).
3. Heat map plots of the mRNA level changes of genes encoding selected functional groups of mitochondrial proteins (gene list in Table EV1C) in the OXPHOS mutants. Gray boxes: deleted genes (COR1, COX9, and ATP7).

We observed interesting patterns of gene expression changes in these mutants (Fig 1C): (i) Nuclear transcripts encoding OXPHOS subunits were commonly downregulated in all the mutants. (ii) Nuclear transcripts encoding mitochondrial proteins except OXPHOS subunits were progressively upregulated in ∆V and ρ⁰ cells. We selected gene lists encoding several important pathways/functions of mitochondrial proteins (gene list in Table EV1C). The heat map plots of the mRNA level changes of these genes showed the increasing upregulation of mitochondrial proteostasis, import, organization, OXPHOS complex assembly, translation, and metabolism in ∆V and ρ⁰ cells (Fig EV1C). (iii) Transcripts encoding ribosomal subunits and Ribi factors were progressively downregulated in ∆V and ρ⁰ cells; (iv) Transcripts encoding glucose transporters and glycolysis enzymes were progressively upregulated in ∆V and ρ⁰ cells.

Proteome remodeling in OXPHOS mutants

To examine whether transcriptional changes remodel mitochondrial proteome, we performed stable isotope labeling using amino acids in cell culture (SILAC) analysis of ∆IV and ρ⁰ cells (Dataset EV2). Equal number of isotope‐labeled WT cells and isotope‐unlabeled mutant cells were mixed together. Mitochondrial fraction was purified for mass spectrometry analysis.

Our results showed that ∆IV cells had small changes of mitochondrial proteome with 17 downregulated proteins and 28 upregulated proteins reaching the significance threshold (P < 0.05, fold change > 1.5 or < 0.67). In contrast, 251 proteins were significantly upregulated and 84 proteins were significantly downregulated in ρ⁰ cells (Fig 2A). Integrating RNAseq and proteome results showed correlative changes of most mRNA and protein levels in ∆IV and ρ⁰ cells with two exceptions (Fig EV2). First, in ρ⁰ cells, some mtRibosome subunits were upregulated at both mRNA and protein levels, whereas the others were upregulated at mRNA but downregulated at protein levels. Because mtDNA encodes mtrRNAs, the scaffolds for mtRibosome assembly and function, we speculate the downregulated subunits may be destabilized by mtrRNA depletion. Second, in ρ⁰ cells, about half of the carriers showed discordant changes at mRNA and protein levels. The underlying reason remains unclear.

Figure 2. Remodeling of mitochondrial proteome and cytoplasmic ribosome in the OXPHOS mutants.

1. SILAC analysis of mitochondrial proteome of ∆IV and ρ⁰ cells. Equal amount of WT cells labeled with heavy amino acids and unlabeled mutant cells were mixed to extract mitochondrial fraction for mass spectrometry analysis. Three biological replicates were prepared for each pair of strains (Dataset EV2). 568 high‐fidelity mitochondrial proteins (gene/protein list in Table EV1A) in ∆IV cells, and 572 such proteins in ρ⁰ cells were quantitatively measured. Significantly changed proteins (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed proteins are indicated.
2. Western blot analysis of representative mitochondrial proteins endogenously tagged with HA. The selection markers were removed by the Cre‐loxP system. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot.
3. Polysome fractionation analysis of OXPHOS mutants. Polysomes were fractionated by sucrose gradients (10%–50%), and the gradients were separated and measured at A₂₅₄.

Source data are available online for this figure.

Figure EV2. Correlation between mRNA and protein level changes of selected mitochondrial proteins. Related to Fig 2.

Heat map plots of the mRNA and protein level changes of genes encoding selected functional groups of mitochondrial proteins (gene list in Table EV1C) in the ∆IV and ρ⁰ cells.

We then tagged endogenous mitochondrial proteins with an HA tag and deleted selection marker with the Cre‐loxP system to preserve endogenous gene regulation. We confirmed the upregulation of mitochondrial import, ISC synthesis, and F₁ assembly factors in ρ⁰ cells (Fig 2B). Importantly, the upregulation of mitochondrial proteins specifically occurred in ∆V and ρ⁰ cells but not in ∆III and ∆IV cells (Fig 2B), matching the RNAseq and SILAC results.

Ribosome biogenesis was progressively downregulated at transcript level (Fig 1C). Fractionation analysis showed that ∆III and ∆IV cells had similar polysome profiles as WT cells, but ∆V and ρ⁰ cells had significantly reduced ribosome contents, especially in the 40S, 60S, and 80S fractions (Fig 2C), correlating with transcript changes.

Transcriptional repression of Inh1, the ATP synthase inhibitor, maintains ∆ψ_m and proliferation of ρ⁰ cells

Downregulation of the transcript level of OXPHOS subunits happens in all the mutants (Fig 1C). We noticed the most downregulated one is Inh1 (Fig 1C), a conserved inhibitor of complex V (ATPIF1 in mouse and human) (Garcia‐Bermudez & Cuezva, 2016). Deletion of Inh1/ATPIF1 disinhibits the ATPase activity of F₁ to maintain ∆ψ_m in both yeast and human cells (Clark‐Walker, 2007; Lefebvre et al, 2013; Chen et al, 2014; Martinez‐Reyes et al, 2016). But it is unclear whether Inh1/ATPIF1 is regulated by mitochondrial adaptive response.

We confirmed Inh1 downregulation at mRNA and protein levels in all the OXPHOS mutants (Fig 3A and B). Deleting Inh1 in ρ⁰ cells benefited cell proliferation at 37°C (Fig 3C). We then used copper‐inducible promoter to inducibly overexpress Inh1 (Fig 3D). Inh1 overexpression selectively inhibited ∆ψ_m and cell proliferation of ρ⁰ cells but had no effect in WT, ∆III, ∆IV, or ∆V cells (Fig 3E and F). To examine the detrimental effect of losing ∆ψ_m, we genetically tagged endogenous Nfs1 and Yah1, two ISC‐synthesis enzymes in mitochondrial matrix, and DNA polymerase Pol3, an essential nuclear ISC‐containing protein required for DNA replication (Lill, 2009), and examined their protein levels by Western blot. Inh1 overexpression in ρ⁰ cells greatly downregulated mitochondrial ISC‐synthesis enzymes Nfs1 and Yah1 and also downregulated nuclear ISC‐containing protein Pol3 (Fig 3G, lane 8 vs. 7). These results indicate ∆ψ_m reduction causes import defects to compromise ISC synthesis and destabilize ISC‐containing proteins. Taken together, OXPHOS mutants commonly downregulate Inh1 but this response is specifically critical for ρ⁰ cells to maintain ∆ψ_m and ISC synthesis, as well as cell proliferation.

Figure 3. Transcriptional repression of Inh1 maintains ∆ψ_m, ISC synthesis, and proliferation of ρ⁰ cells.

1. qRT–PCR analysis of INH1 mRNA levels. Data are mean ± SD from three biological replicates.
2. Western blot analysis of Inh1 protein levels. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot.
3. Deletion of INH1 benefits the proliferation of ρ⁰ cells at 37°C. Serial dilutions (tenfold dilution) of the indicated strains were analyzed on SCD plates at 30°C and 37°C for 2 days.
4. Western blot analysis of copper‐inducible overexpression of Inh1. A cassette expressing Inh1 under the control of CUP1 promoter was inserted into the HO locus. Inh1 overexpression was induced by 100 μM CuSO₄ for 12 h.
5. Overexpression of Inh1 decreases ∆ψ_m in ρ⁰ cells. Cells were cultured in SCD or SCD plus 100 μM CuSO₄ for 12 h to mid‐log phase. Cells were then stained with 125 nM TMRM for FACS analysis.
6. Overexpression of Inh1 represses the proliferation of ρ⁰ cells. Serial dilutions (tenfold dilution) of the indicated strains were analyzed on SCD or SCD plus 100 μM CuSO₄ plates at 30°C for 2 days.
7. Overexpression of Inh1 reduces the protein levels of mitochondrial ISC biosynthesis proteins (Nfs1 and Yah1) and nuclear ISC‐containing protein Pol3. Cells were cultured in SCD or SCD plus 100 μM CuSO₄ for 12 h to mid‐log phase. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot. Nfs1 and Yah1 were endogenously tagged with FLAG, and Pol3 was endogenously tagged with HA.
8. Cartoon illustration of the translational repression of Hap4 and the subsequent transcriptional downregulation of Inh1 to maintain ∆ψ_m in ρ⁰ cells.

Data information: The red highlight two cell lines for comparison (E, F).

Source data are available online for this figure.

Translational downregulation of HAP4 represses Inh1 transcription

To understand how Inh1 and OXPHOS subunits are downregulated at transcript level, we analyzed the HAP complex, consisting of Hap2‐5, which is responsible for the transcription of OXPHOS subunits (Boos et al, 2019; Mao & Chen, 2019). In ρ⁰ cells, most HAP subunits were upregulated at both mRNA and protein levels with one exception that Hap4 mRNA was increased twofold but its protein level was decreased (Fig EV3A and B). Downregulation of Hap4 protein was observed in all the OXPHOS mutants (Fig EV3C). In ρ⁰ cells, we did not detect change of Hap4 protein half‐life (Fig EV3D). We then performed the ribosome footprinting to measure genome‐wide changes of translation efficiency (Ingolia, 2010; Ingolia et al, 2012). ρ⁰ cells specifically downregulated the translation efficiency of Hap4 but not the other three HAP transcripts (Fig EV3E, Dataset EV3). The decrease of Hap4 translation efficiency exceeds the increase of its mRNA level, thus may cause a net reduction of Hap4 protein.

Figure EV3. Translational downregulation of Hap4 represses transcription of INH1. Related to Fig 3.

1. qRT–PCR analysis of HAP complex components in ρ⁰ cells. Data are mean ± SD from three biological replicates. The horizontal line indicates fold change of 1.
2. Western blot analysis of Hap proteins. Hap proteins were endogenously tagged with HA. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot.
3. Western blot analysis of Hap4‐FLAG level in OXPHOS mutants. Hap4 was endogenously tagged with FLAG. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot.
4. Protein half‐life of Hap4‐FLAG is not changed in ρ⁰ cells as compared to WT cells. Hap4 was endogenously tagged with FLAG. Cells were treated with cycloheximide (CHX) to block protein synthesis for the indicated time. Whole cell lysates were extracted for Western blot analysis. Protein samples of ρ⁰ cells were loaded twofold the amounts of samples of WT cells to equalize Hap4‐FLAG amount at 0 min of CHX treatment. Relative Hap4‐FLAG/G6PDH ratio at 0 min of CHX treatment was normalized as 1. Data are mean ± SD from three biological replicates.
5. Genome‐wide translation efficiency changes in ρ⁰ cells versus WT cells (Dataset EV3). The translation efficiencies of Hap proteins are indicated.
6. qRT–PCR analysis of HAP4 and INH1 mRNA levels upon Hap4 overexpression in the indicated strains. A cassette expressing Hap4 under the control of CUP1 promoter was inserted into the HO locus. Hap4 overexpression was induced by 100 μM CuSO₄ for 12 h. Data are mean ± SD from three biological replicates.
7. Western blot analysis of Inh1 level upon Hap4 overexpression in the indicated strains. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot.
8. Deletion of HAP4 benefits the proliferation of ρ⁰ cells at 37°C. Serial dilutions (tenfold dilution) of the indicated strains were analyzed on SCD plates at 30°C and 37°C for 2 days.
9. Overexpression of Hap4 inhibits the proliferation of ρ⁰ cells, but not the proliferation of ρ⁰ inh1∆ cells.
10. Overexpression of Inh1 inhibits the proliferation of both ρ⁰ and ρ⁰ hap4∆ cells.

Data information: Serial dilutions (tenfold dilution) of the indicated strains were analyzed on SCD or SCD plus 100 μM CuSO₄ plates at 30°C for 2 days (I, J).

Source data are available online for this figure.

To examine whether Hap4 level determines Inh1 transcription, we used the copper‐inducible CUP1 promoter to drive Hap4 expression in ρ⁰ cells. Without copper supplementation, we noticed the upregulation of HAP4 mRNA (Fig EV3F, left panel, highlighted in red) and the corresponding upregulation of INH1 mRNA (Fig EV3F, right panel, highlighted in red) and Inh1 protein (Fig EV3G, lane 3 vs. 2). This could be due to the tracing amount of copper in the synthetic medium. Copper supplementation further induced INH1 at mRNA (Fig EV3F, highlighted in green) and protein (Fig EV3G, lane 4 vs. 2) levels. Thus, Hap4 level is a key determinant of Inh1 transcription.

Functionally, deleting Hap4 benefited ρ⁰ cell proliferation at 37°C (Fig EV3H) and the leaky expression of Hap4 was sufficient to suppress ρ⁰ cell growth (Fig EV3I). Furthermore, deleting Inh1 completely abolished the growth suppression effect of Hap4 in ρ⁰ cells (Fig EV3I). Overexpression of Inh1 suppressed proliferation of ρ⁰ hap4∆ cells (Fig EV3J). These epistatic analyses demonstrate that Inh1 is the major downstream effector of Hap4. Taken together, ρ⁰ cells downregulate Hap4 translation to repress Inh1 transcription and thus to disinhibit the ATP hydrolysis activity of F₁ component (Fig 3H).

Puf3 promotes mitochondrial biogenesis to maintain ∆ψ_m and proliferation in ρ⁰ cells

Next, we focused on the global mitochondrial biogenesis response in ∆V and ρ⁰ cells. Puf3 is a cytoplasmic mRNA‐binding protein that mainly associates with mRNAs encoding mitochondrial proteins (Gerber et al, 2004; Lapointe et al, 2015; Lapointe et al, 2018). It promotes mRNA degradation under fermentative condition (Olivas & Parker, 2000; Houshmandi & Olivas, 2005) and gets phosphorylated to promote mRNA translation under respiratory growth condition (Lee & Tu, 2015). We combined two studies to generate a list of Puf3‐associated mRNAs (Table EV1D) (Gerber et al, 2004; Lapointe et al, 2015). We found that most‐upregulated transcripts encoding mitochondrial proteins in ∆V and ρ⁰ cells overlapped with transcripts known to associate with Puf3 (Fig 4A).

Figure 4. mRNA‐binding protein Puf3 upregulates Mia40 and global mitochondrial biogenesis to maintain ∆ψ_m .

• A
  The changes of Puf3‐associated transcripts encoding mitochondrial proteins in OXPHOS mutants. The RNAseq results in Fig 1C (Dataset EV1) are re‐analyzed. Transcripts encoding high‐fidelity mitochondrial proteins (gene list in Table EV1A) and the Puf3‐associated transcripts encoding mitochondrial proteins (gene list in Table EV1D) are shown. The significantly changed transcripts (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed transcripts are indicated.
• B, C
  Deletion of PUF3 upregulates mitochondrial transcripts in WT cells (B), whereas downregulates mitochondrial transcripts in ρ⁰ cells (C). WT and mutant yeast cells were cultured to log phase. RNAseq was performed with three biological replicates per genotype (Dataset EV4). All the transcripts and the Puf3‐associated transcripts encoding mitochondrial proteins are shown. The significantly changed transcripts (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed transcripts are indicated.
• D
  Deletion of PUF3 in ρ⁰ cells downregulates mitochondrial proteins. SILAC analysis of mitochondrial fractions from ρ⁰ and ρ⁰ puf3∆ cells are shown. Equal amount of ρ⁰ cells labeled with heavy amino acids and unlabeled ρ⁰ puf3∆ cells were mixed to extract mitochondrial fraction for mass spectrometry analysis. Three biological replicates were prepared for each pair of strains (Dataset EV5). Mitochondrial proteins encoded by Puf3‐associated mRNAs are highlighted in red. Mia40 substrates are highlighted in green (gene list in Table EV1E). Mia40 substrates encoded by Puf3‐associated transcripts are highlighted in green with red circle. The significantly changed proteins (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed proteins are indicated.
• E
  Correlation plots of fold changes (ρ⁰ puf3∆ versus ρ⁰) between mitochondrial proteins and their transcripts. Puf3‐associated transcripts encoding mitochondrial proteins are highlighted in red. Mia40 substrates are highlighted in green. Mia40 substrates encoded by Puf3‐associated transcripts are highlighted in green with red circle.
• F
  Deletion of PUF3 decreases ∆ψ_m in ρ⁰ cells.
• G
  Deletion of PUF3 decreases the proliferation of ρ⁰ cells. Serial dilutions (tenfold dilution) of the indicated strains were analyzed on SCD plates at 30°C for 2 days.
• H
  Schematic illustration of the consensus Puf3‐binding motif (Lapointe et al, 2015) and the motif at Mia40 3’‐UTR. Mutating the Puf3‐binding motif of MIA40 mRNA abolishes Mia40 upregulation in ρ⁰ cells. Mia40 was endogenously tagged with FLAG, and its 3’‐UTR was edited as indicated. The selection markers were removed by the Cre‐loxP system. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot. RC: reverse complement. DE: deletion.
• H
  Schematic illustration of the consensus Puf3‐binding motif (Lapointe et al, 2015) and the motif at Mia40 3’‐UTR. Mutating the Puf3‐binding motif of MIA40 mRNA abolishes Mia40 upregulation in ρ⁰ cells. Mia40 was endogenously tagged with FLAG, and its 3’‐UTR was edited as indicated. The selection markers were removed by the Cre‐loxP system. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot. RC: reverse complement. DE: deletion.
• I
  Mutating the Puf3‐binding motif of MIA40 mRNA decreases ∆ψ_m of ρ⁰ cells.
• J
  Mutating the Puf3‐binding motif of MIA40 mRNA impairs the proliferation of ρ⁰ cells. Cells were cultured from 0.05 OD in SCD, and OD600 was measured every 2 h.
• K
  Overexpression of MIA40 rescues ∆ψ_m of ρ⁰ puf3∆ cells. Single copy plasmid expressing Mia40 driven by the MIA40 promoter was transformed into the indicated strains.
• L
  Overexpression of MIA40 rescues the proliferation of ρ⁰ puf3∆ cells. Strains were prepared as in (K). Cells were cultured from 0.05 OD in SCD, and OD600 was measured every 2 h.

Data information: Cells were cultured to mid‐log phase and stained with 125 nM TMRM for FACS analysis (F, I, K). The red highlight two cell lines for comparison (F, G, I, K). Data are mean ± SD from three biological replicates (J, L).

Source data are available online for this figure.

To determine whether Puf3 is required for the upregulation of transcripts encoding mitochondrial proteins, we deleted PUF3 in both WT and ρ⁰ cells. RNAseq analysis (Dataset EV4) showed that deleting PUF3 modestly upregulated its associated transcripts in WT cells as previously reported (Olivas & Parker, 2000; Fig 4B), but significantly downregulated its associated transcripts (with 82 transcripts reaching the significance threshold) in ρ⁰ cells (Fig 4C), indicating Puf3 switches its role from promoting mRNA degradation in WT cells to mRNA stabilization in ρ⁰ cells.

Stable isotope labeling using amino acids in cell culture analysis of mitochondrial proteome showed that deleting PUF3 downregulated mitochondrial proteins in ρ⁰ cells with 116 proteins reaching significance threshold, among which 102 proteins are encoded by Puf3‐associated mRNAs (Fig 4D and Dataset EV5). Similar to ρ⁰ cells, ∆V cells also required Puf3 to upregulate mitochondrial mRNAs and proteins (Appendix Fig S2A and B).

Deleting PUF3 had no effect on ∆ψ_m and cell proliferation in WT, ∆III, ∆IV, or ∆V cells, but decreased both ∆ψ_m and cell proliferation in ρ⁰ cells (Fig 4F and G). Thus, Puf3 promotes a global mitochondrial biogenesis response in ∆V and ρ⁰ cells, but is only required in ρ⁰ cells to maintain ∆ψ_m and cell proliferation.

Puf3 may promote mitochondrial biogenesis through mRNA stabilization and/or promoting mRNA translation. Ribosome footprinting (Dataset EV3) showed that there was no general upregulation of the translation efficiency of Puf3‐associated transcripts in ρ⁰ cells (Fig EV4A and B), and deleting PUF3 in ρ⁰ cells did not affect the translation efficiency of its associated transcripts (Fig EV4C and D). Polysome fractionation experiments also showed that Puf3 did not enter into the polysome fractions after mtDNA depletion (Fig EV4E). Thus, Puf3 has minimal effect on the translation efficiency of its associated mRNAs in ρ⁰ cells, which is mechanistically different from the glucose deprivation condition, in which Puf3 promotes translation of its associated mRNAs (Lee & Tu, 2015).

Figure EV4. Ribosome footprinting and polysome analysis. Related to Fig 4.

1. Genome‐wide translation efficiency changes of ρ⁰ cells versus WT cells. The translation efficiencies of Puf3‐associated mitochondrial transcripts are highlighted in red. Translation efficiency was determined by ribosome footprinting (Dataset EV3).
2. Violin plot of translation efficiency changes of ρ⁰ versus WT cells.
3. Genome‐wide translation efficiency changes of ρ⁰ puf3∆ versus ρ⁰ cells. Translation efficiency was determined by ribosome footprinting (Dataset EV3).
4. Violin plot of translation efficiency changes of ρ⁰ puf3∆ cells versus ρ⁰ cells.
5. Polysome analysis of Puf3‐FLAG. Equal volume of lysates from the indicated fractions was loaded for Western blot analysis.

Source data are available online for this figure.

Mia40 is the critical mitochondrial protein upregulated by Puf3 to maintain ∆ψ_m and proliferation in ρ⁰ cells

It is intriguing that among the hundreds of mitochondrial mRNAs/proteins upregulated by Puf3, many of them, such as the mtRibosome subunits, seem to aggravate the folding stress inside mitochondrial matrix because many subunits cannot be properly assembled and are presumably degraded in ρ⁰ cells (Fig EV2). Similarly, upregulation of most OXPHOS assembly factors and mitochondrial tRNA synthetases in ρ⁰ cells seems to be a waste. Then, what factors upregulated by Puf3 help ∆ψ_m maintenance?

Because ∆ψ_m maintenance in ρ⁰ cells requires matrix F₁ component and inner membrane (IM) carriers, enhancing the import pathway seems to be a meaningful response. In our RNAseq results, Mia40 was the most‐upregulated mitochondrial import factor in ρ⁰ cells (Fig EV1, highlighted in red in the “import machinery” category). Mia40 is an essential mitochondrial protein that mediates the import and oxidative folding of IMS proteins (Herrmann & Riemer, 2012; Peleh et al, 2016; Stojanovski et al, 2012). Its substrates include phospholipid carriers and small TIMs (Tim8, Tim9, Tim10, Tim12, and Tim13) critical for the biogenesis of OM β‐barrel proteins and IM carrier proteins (Koehler, 2004; Modjtahedi et al, 2016). Inactivation of Mia40 activates the UPR^am (Wrobel et al, 2015). Thus, Mia40 has profound effect on mitochondrial proteome and phospholipids. MIA40 mRNA contains a Puf3‐binding site in its 3’‐UTR (Fig 4H). In our SILAC results, Mia40 and its substrates (gene list in Table EV1E) were among the significantly downregulated mitochondrial proteins upon PUF3 deletion in ρ⁰ cells (Fig 4, D and E, highlighted in green), supporting Mia40 as a substrate upregulated by Puf3. More importantly, many Mia40 substrates, including small TIMs, were downregulated more at protein levels as compared to mRNA levels in ρ⁰ puf3∆ cells (Fig 4E, highlighted in green), indicating Puf3 upregulates Mia40 to promote the import of Mia40 substrates in ρ⁰ cells.

We mutated the Puf3‐binding site in MIA40 mRNA 3’‐UTR by introducing reverse complement (RC) sequence or deletion (DE) (Fig 4H). Both manipulations did not affect the basal protein level of Mia40 in WT cells but selectively blocked Mia40 upregulation in ρ⁰ cells (Fig 4H). Abolishing Puf3 association with Mia40 mRNA (3’‐UTR‐RC and 3’‐UTR‐DE) had no effect on ∆ψ_m and the proliferation of WT cells, but decreased ∆ψ_m and the proliferation of ρ⁰ cells (Fig 4I and J). We further overexpressed Mia40 in ρ⁰ puf3∆ cells, in which hundreds of Puf3‐associated mitochondrial mRNAs remain downregulated. Strikingly, Mia40 overexpression rescued ∆ψ_m and the proliferation of ρ⁰ puf3∆ cells to the level of ρ⁰ cells (Fig 4K and L). These results strongly suggest Mia40 as the major effector upregulated by Puf3 to maintain ∆ψ_m and cell proliferation in ρ⁰ cells.

Puf3 hyper‐phosphorylation upregulates Mia40 and mitochondrial biogenesis

Then how does Puf3 switch from repressing to promoting mitochondrial biogenesis in ∆V and ρ⁰ cells? We generated PUF3‐FLAG knock‐in cells and observed that Puf3‐FLAG prepared from WT cells migrated as a slightly smeared band in Western blot and Puf3‐FLAG from ρ⁰ cells migrated as a wider and upshifted band (Fig 5A), indicating post‐translational modifications. Treatment with λ‐phosphatase collapsed Puf3‐FLAG from both WT and ρ⁰ cells into a single band with lower molecular weight (Fig 5A), indicating Puf3 is a phosphor protein in both WT and ρ⁰ cells. We thus run samples with the Phos‐tag gel, in which the migration of phosphor protein is impeded (Kinoshita et al, 2009). The result showed that Puf3 is a heavily phosphorylated protein in WT cells and gets hyper‐phosphorylated in ρ⁰ cells (Fig 5A). Furthermore, Puf3 hyper‐phosphorylation was induced by treating with complex V inhibitor oligomycin and in ∆V and ρ⁰ cells but not affected by genetic inactivation or chemical inhibition of complex III and complex IV (Fig 5B). Thus, Puf3 hyper‐phosphorylation correlates with the upregulation of mitochondrial biogenesis in ∆V and ρ⁰ cells.

Figure 5. Puf3 hyper‐phosphorylation upregulates Mia40 and mitochondrial biogenesis to maintain ∆ψ_m .

1. Puf3 is phosphorylated in WT cells and is hyper‐phosphorylated in ρ⁰ cells. Puf3‐FLAG was immunoprecipitated from the indicated strains and treated with/without λ‐phosphatase, and analyzed by regular and Phos‐tag SDS–PAGE.
2. Puf3 phosphorylation in response to OXPHOS inhibitors and in OXPHOS mutants. WT cells were treated with antimycin (10 µM), KCN (10 µM), or oligomycin (10 µM) for 3 h, or shifted from SCD to SCEG medium for 3 h. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot. Treatments and strains with Puf3 hyper‐phosphorylation was highlighted in red. Glucose depletion was used as a positive control. Asterisk indicates a non‐specific band.
3. Puf3‐15A mutation inhibits Puf3 hyper‐phosphorylation in ρ⁰ cells. A cassette expressing WT or mutant Puf3‐FLAG (11A or 15A) under the control of PUF3 promoter was inserted into the HO site in puf3∆ cells. The 15A mutant that blocks Puf3 hyper‐phosphorylation is highlighted in red. Puf3‐FLAG was immunoprecipitated from the indicated strains and analyzed by regular and Phos‐tag SDS–PAGE.
4. Puf3‐15A mutation inhibits Puf3 hyper‐phosphorylation upon oligomycin treatment and in ∆V and ρ⁰ cells. WT cells were treated with antimycin (10 µM), KCN (10 µM) or oligomycin (10 µM) for 3 h, or shifted from SCD to SCEG medium for 3 h. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot. Asterisk indicates a non‐specific band.
5. Puf3‐15A mutation downregulates Puf3‐associated mitochondrial transcripts in ρ⁰ cells. ρ⁰ and ρ⁰ puf3‐15A cells were cultured to log phase. RNAseq was performed with three biological replicates per genotype (Dataset EV7). All the transcripts and the Puf3‐associated transcripts encoding mitochondrial proteins (gene lists in Table EV1D) are shown. The significantly changed transcripts (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed transcripts are indicated.
6. Puf3‐15A mutation inhibits the upregulation of Mia40 in ∆V and ρ⁰ cells. Mia40 was endogenously tagged with FLAG. The selection markers were removed by the Cre‐loxP system. Whole cell lysates were extracted, and equal amount of proteins was loaded for Western blot.
7. Puf3‐15A mutation decreases ∆ψ_m in ρ⁰ cells. Cells were cultured to mid‐log phase and stained with 125 nM TMRM for FACS analysis. The red highlight two cell lines for comparison.
8. Puf3‐15A mutation has no effect on the proliferation of WT cells.
9. Puf3‐15A mutation inhibits the proliferation of ρ⁰ cells.
10. Mia40 overexpression rescues the proliferation defect of ρ⁰ puf3‐15A cells. Single copy plasmid expressing Mia40 driven by MIA40 promoter was transformed into the indicated strains.
11. Cartoon illustration that Puf3 hyper‐phosphorylation in ∆V and ρ⁰ cells upregulates Puf3‐associated mitochondrial transcripts, especially Mia40, to enhance mitochondrial biogenesis and ∆ψ_m maintenance.

Data information: Cells were cultured from 0.05 OD in SCD, and OD600 was measured every 2 h; Data are mean ± SD from three biological replicates (H, I, J).

Source data are available online for this figure.

We performed SILAC analysis of Puf3 phosphorylation (Dataset EV6). We reached 70% peptide coverage and identified 43 phosphorylation sites in the N‐terminal unstructured region of Puf3 (Fig EV5A). Phosphorylation of 34 sites could be reliably quantified (Fig EV5B and Dataset EV6). We progressively mutated the sites with enhanced phosphorylation to alanine and reconstituted the mutant PUF3‐FLAG into puf3∆ cells. We found that mutation of 15 sites (Puf3‐15A, sites highlighted in red in Fig EV5B) abolished most of the Puf3 hyper‐phosphorylation in ρ⁰ cells without affecting the basal Puf3 phosphorylation in WT cells (Fig 5C). Puf3 hyper‐phosphorylation upon oligomycin treatment and in ∆V cells were also inhibited by the 15A mutation (Fig 5D).

Figure EV5. SILAC analysis of Puf3 phosphorylation. Related to Fig 5.

1. Peptide coverage map of Puf3. Peptide sequences covered by mass spectrometry are shown in red. Underlined residues are phosphorylated. Blue underlined residues: phosphorylation sites quantitatively analyzed in Fig EV5B. Green underlined residues: phosphorylation sites not quantitatively analyzed due to low phosphorylation level or low total peptide number. Mass spectrometry data are shown in Dataset EV6.
2. Quantitative analysis of Puf3 site‐specific phosphorylation (ρ⁰ versus WT). The localization of phosphorylation sites at the N terminus of Puf3 is illustrated. Residues mutated in Puf3‐15A are highlighted in red. Residues mutated in Puf3‐11A are highlighted in red and underlined. The horizontal line indicates fold change of 1.

Puf3‐associated mitochondrial transcripts were decreased by the Puf3‐15A mutation in ρ⁰ cells (Fig 5E, and Dataset EV7). Importantly, Puf3‐15A mutation did not affect the protein level of Mia40 in WT cells but selectively prevented Mia40 upregulation in ∆V and ρ⁰ cells (Fig 5F). Accordingly, Puf3‐15A mutation did not affect ∆ψ_m and cell proliferation in WT cells but decreased both ∆ψ_m and cell proliferation in ρ⁰ cells (Fig 5G–I). Mia40 overexpression rescued the proliferation defect of ρ⁰ puf3‐15A cells (Fig 5J). Taken together these findings, Puf3 gets hyper‐phosphorylated in ∆V and ρ⁰ cells to upregulate mitochondrial biogenesis, particularly Mia40 (Fig 5K). This response is critical for ρ⁰ cells to maintain ∆ψ_m and cell proliferation.

Snf1/AMPK represses ribosome biogenesis and upregulates glycolysis to maintain ∆ψ_m and proliferation in ∆V and ρ⁰ cells

∆V and ρ⁰ cells consume glycolytic ATP in an expensive way to maintain ∆ψ_m. Transcriptional upregulation of glucose transporters and glycolysis genes and repression of ribosome biogenesis genes in these two mutants (Fig 1C) seem to enhance glycolytic ATP production and reduce ATP expenditure. Ribosomes compose approximately 30% of cellular mass and ribosome biogenesis consumes about 80% of newly synthesized nucleotides (Lunt & Vander Heiden, 2011). In yeast, about 60% of total transcription is to transcribe rRNA and 50% of RNA polymerase II transcription and 90% of mRNA splicing are devoted to ribosomal proteins (Warner, 1999).

We found that the stress‐responsive kinase Snf1, the yeast AMPK (Hedbacker & Carlson, 2008), mediates these two changes. snf1∆ cells were inviable after mtDNA depletion. Therefore, ρ⁰ snf1∆ cells were generated by acute mtDNA depletion through EB treatment. RNAseq (Dataset EV8) showed that deleting SNF1 downregulated transcripts encoding glucose transporter and glycolysis genes in both WT and ρ⁰ cells, and specifically upregulated ribosome and Ribi transcripts in ρ⁰ cells (Fig 6A, highlighted in red). Polysome fractionation experiment showed that deleting SNF1 in WT cells upregulated monosome and polysomes (Fig 6B, solid red line vs. solid black line), indicating Snf1 represses ribosome biogenesis through post‐transcriptional mechanisms in WT cells. SNF1 deletion in ρ⁰ cells significantly upregulated 40S, 60S, and 80S ribosomes but had minimal effect on polysomes (Fig 6B, dashed red line vs. dashed black line). The production of excessive ribosomes in ρ⁰ snf1∆ cells is apparently uneconomic for energy metabolism. Accordingly, SNF1 deletion reduced ∆ψ_m and impaired cell proliferation in ∆V and ρ⁰ cells but not in WT, ∆III, or ∆IV cells (Fig 6C and D).

Figure 6. Snf1/AMPK upregulates glycolysis and represses ribosome biogenesis to maintain ∆ψ_m .

1. Snf1 maintains glycolysis and glucose transporter gene expression in both WT and ρ⁰ cells, and represses ribosome and Ribi gene expression in ρ⁰ cells. ρ⁰ snf1∆ cells were prepared by acute EtBr (25 µg/ml) treatment of snf1∆ cells in SCD for 2 days. RNAseq was performed with three biological replicates per genotype (Dataset EV8). All the transcripts and two categories of transcripts (gene lists in Table EV1B) are shown. Significantly changed genes (P value < 0.05; fold change > 1.5 or < 0.67) are highlighted with dashed lines; numbers of significantly changed genes are indicated.
2. Snf1 represses ribosome biogenesis in WT and ρ⁰ cells. Polysomes of the indicated strains were fractionated by sucrose gradients (10%–50%), and the gradients were separated and measured at A₂₅₄.
3. Deletion of SNF1 decreases ∆ψ_m in both ∆V and ρ⁰ cells.
4. Deletion of SNF1 decreases the proliferation of both ∆V and ρ⁰ cells. Serial dilutions of the indicated strains were analyzed on SCD plates at 30°C for 2 days.
5. Deletion of MIG1 in ρ⁰ snf1∆ cells rescues ∆ψ_m.
6. Deletion of MIG1 in ρ⁰ snf1∆ cells partially rescues cell proliferation. A plasmid overexpressing MIG1 under the control of MIG1 promoter was transformed into the indicated strains. Serial dilutions (tenfold dilution) were analyzed on SCD plates at 30°C for 2 days. V: empty vector.
7. Combining MIG1 deletion and NOG2 overexpression in ρ⁰ snf1∆ cells fully rescues cell proliferation. A plasmid overexpressing NOG2 under the control of TEF1 promoter was transformed into the indicated strains. Serial dilutions (tenfold dilution) were analyzed on SCD plates at 30°C for 2 days. V: empty vector.
8. Cartoon illustration that Snf1 signaling in ∆V and ρ⁰ cells upregulates glycolysis and represses ribosome biogenesis to optimize energy metabolism.

Data information: Cells were cultured to mid‐log phase and stained with 125 nM TMRM for FACS analysis (C, E). The red highlight two cell lines for comparison (C, D, E).

To determine the contributions of Snf1’s two downstream pathways, we tried to upregulate glycolysis and repress ribosome biogenesis in ρ⁰ snf1∆ cells. Snf1 phosphorylates transcription repressor Mig1 and induces its nuclear export to turn on gene transcription (Treitel et al, 1998; Hedbacker & Carlson, 2008). RNAseq showed that deleting MIG1 in ρ⁰ snf1∆ cells upregulated glucose transporter and glycolysis genes and modestly repressed ribosome and Ribi genes (Appendix Fig S3A and Dataset EV9). Functionally, deleting MIG1 in ρ⁰ snf1∆ cells partially restored ∆ψ_m and proliferation (Fig 6E and F). Nog2 is a GTPase that binds to and inhibits the nuclear export of pre‐60S subunit and its overexpression inhibits ribosome biogenesis (Matsuo et al, 2014; Wang & Chen, 2015). We overexpressed Nog2 in ρ⁰ snf1∆ cells and observed no rescuing effect (Fig 6G). However, Nog2 overexpression in ρ⁰ snf1∆ mig1∆ cells fully restored cell proliferation to the level of ρ⁰ cells (Fig 6G). These results suggest glycolysis upregulation plays a major role and ribosome biogenesis repression plays a minor role downstream of Snf1 activation. Activation of both pathways fully mimics the effect of Snf1 activation (Fig 6H).

Three adaptive responses, namely Inh1 downregulation, Puf3‐mediated mitochondrial biogenesis, and Snf1 activation, are activated in ∆V but only Snf1 is required in ∆V cells. We examined if Snf1 and the other two adaptive responses have synergistic effect in ∆V cells. In ∆V snf1∆ cells, deleting either one or both of the responses exacerbated proliferation (Appendix Fig S3B, highlighted in red). As control, deleting all the three responses in WT cells had no effect on proliferation (Appendix Fig S3C).

Upregulation of cytosolic chaperones maintains ∆ψ_m

We observed that accompanying increasing stresses in WT, ∆III, ∆IV, ∆V, and ρ⁰ cells, there was gradual induction of chaperones, ubiquitin‐proteasome system, and autophagy genes (gene lists in Table EV1F). These quality control/degradation genes were further induced in Inh1‐overexpressing ρ⁰ cells (Fig 7A and Dataset EV10). These genes are most likely induced in response to the accumulation of mitochondrial precursor proteins under import arrest (Wrobel et al, 2015; Boos et al, 2019). Proteasome activation has been shown to alleviate proteostatic stress caused by precursor accumulation (Wrobel et al, 2015).

Figure 7. Upregulation of chaperones maintains ∆ψ_m .

1. Heat map plots of the mRNA level changes of chaperones, ubiquitin‐proteasome pathway, and autophagy genes (gene list in Table EV1F) in the indicated strains (data from Dataset EV1 and Dataset EV10). Selected genes for growth analysis in B are highlighted in red.
2. Summary of the growth results of the indicated strains. −: no effect on growth. +: growth inhibition. ++: strong growth inhibition. Genes that are preferentially required in ρ⁰ puf3∆ cells are highlighted in red.
3. Deletion of HSC82 selectively impairs the proliferation of ρ⁰ puf3∆ cells. Cells were cultured from 0.05 OD in SCD, and OD600 was measured every 2 h. Data are mean ± SD from three biological replicates.
4. Deletion of HSC82 selectively decreases ∆ψ_m in ρ⁰ puf3∆ cells. Cells were cultured to mid‐log phase and stained with 125 nM TMRM for FACS analysis. The red highlight two cell lines for comparison.
5. Cartoon illustration of the progressive activation of adaptive responses to convergently maintain ∆ψ_m in OXPHOS mutants.

Source data are available online for this figure.

To understand whether chaperones and autophagy have a role under low ∆ψ_m condition, we selected a representative group of genes and deleted them in WT, ρ⁰, and ρ⁰ puf3∆ cells. These genes include chaperones Hsp70 and Hsp90, disaggregase Hsp104, antiaggregase Hsp26 and Hsp42, membrane chaperone Hsp12 (Verghese et al, 2012), autophagy gene Atg8, and mitophagy receptors Atg32 and Atg33 (Okamoto et al, 2009; Kanki et al, 2009a; Kanki et al, 2009b). As summarized in Fig 7B, disrupting autophagy, mitophagy, disaggregase, and antiaggregase individually or in combination had no effect on cell proliferation. Interestingly, deleting HSC82, the dominant isoform of Hsp90, impaired growth selectively in ρ⁰ puf3∆ cells, but not in WT and ρ⁰ cells (Fig 7B and C). Similarly, ρ⁰ puf3∆ cells relied more on Hsp70 family members Ssa1 and Ssa2 to maintain cell proliferation as compared to WT and ρ⁰ cells (Fig 7B). Deleting HSC82 in ρ⁰ puf3∆ cells decreased ∆ψ_m (Fig 7D), indicating Hsp90 activity becomes critical for ∆ψ_m maintenance under very low ∆ψ_m condition.

Discussion

Progressive activation of mitochondrial stress responses in response to increasing stresses among OXPHOS mutants

Through extensive omics analyses and characterizations of pathways and effectors, we summarize the global stress responses in OXPHOS mutants and show how these responses maintain ∆ψ_m (Fig 7E). Briefly, in ∆III/∆IV cells, a common response, Inh1 downregulation, is activated to release mitochondrial ATP hydrolysis activity. In ∆IV and ρ⁰ cells with increased stresses, two additional responses are induced. One response is the activation of Snf1/AMPK to upregulate glycolysis and repress ribosome biogenesis. This response enhances ATP supply and reduces ATP expenditure, thus optimizing ATP supply to mitochondria. The other response is the phosphorylation of Puf3 to upregulate global mitochondrial biogenesis. We identified Mia40, the import receptor of IMS proteins, as the key factor upregulated by Puf3. The exact mechanism of ∆ψ_m maintenance by Mia40 remains unclear, but may link with its broad effect on mitochondrial proteome and lipid metabolism. Finally, when ∆ψ_m gets further decreased in ρ⁰ cells, cells upregulate chaperones and degradation pathways. Chaperones bind precursor proteins and may escort substrates to import or degradation pathways. The requirement of Hsp90 to maintain ∆ψ_m in ρ⁰ puf3∆ cells (Fig. 7D) indicates its role in mitochondrial protein import. These diverse cytoplasmic and mitochondrial responses convergently maintain ∆ψ_m (Fig 7E). The conserved effectors mechanistically connect mitochondrial stress responses with mitochondrial repair/compensation, filling a major gap in our understanding of mitochondrial adaptive responses.

Caveats and remaining questions of the current study are also evident. First, what aspects of mitochondrial dysfunction are being sensed and how the initial signals are activated remain unclear. Reduction of ∆ψ_m is an attractive inducer, but may not be the only one, especially in ρ⁰ cells with extensive proteostatic stresses. Second, whether these diverse pathways are activated by a common upstream molecule or by different sensor molecules are unclear. Third, other types of changes, e.g., protein translocation from mitochondria to other subcellular locations cannot be revealed by our approach. Future studies are thus warranted.

Evolutionary conservation of the stress responses and implication for mitochondrial diseases

The yeast strategy to cope with OXPHOS dysfunction provides a unique angle to consider human mitochondrial diseases. First, our results highlight the importance of removing Inh1/ATPIF1 to maintain ∆ψ_m. ATPIF1 knockout increases ∆ψ_m in OXPHOS‐defective mammalian cell lines(Lefebvre et al, 2013; Chen et al, 2014). Whether ATPIF1 knockout can have beneficiary effects in mitochondrial diseases remains to be determined. Second, yeast activates Snf1 to upregulate glycolytic ATP supply and suppress ATP expenditure processes, thus channeling glycolytic ATP to mitochondria; yeast also activates mitochondrial biogenesis through Puf3 hyper‐phosphorylation. Interestingly, mammalian AMPK upregulates glycolysis, represses ATP consumption processes, and also promotes mitochondrial biogenesis (Herzig & Shaw, 2018). Thus, AMPK activation in mammalian cells achieves two major adaptations in yeast, and AMPK is an attractive target for mitochondrial diseases. AMPK activation, glycolysis upregulation, and mitochondrial biogenesis have been documented in mouse and C. elegans OXPHOS disease models (Gioran et al, 2019; Haynes et al, 2007; Kuhl et al, 2017; Nargund et al, 2015; Viscomi et al, 2011; Wredenberg et al, 2002). Importantly, pharmacological activation of AMPK by AICAR partially corrects COX deficiency (Viscomi et al, 2011). New generation of AMPK agonists with better pharmacological properties as compared to AICAR may provide a better therapeutic opportunity (Cokorinos et al, 2017; Myers et al, 2017).

Materials and Methods

Reagents or resources used in this paper are listed in Appendix Table S1.

Yeast strains and media

The yeast strains used in this study are listed in Table EV2. All the strains were derived from the wild‐type strain W303‐1A. To generate stable ρ⁰ cell lines, corresponding WT cells were treated with 25 µg/ml EtBr in SCD for 2 days and then plated on YPD plates. Single clones were picked up after 2–3 days. The ρ⁰ clones were streaked on YPEG plates to verify the absence of respiratory growth and were confirmed by DAPI staining and Western blot detecting mtDNA‐encoded Cox2.

Gene deletions were performed by PCR‐based homologous recombination to replace the target gene with appropriate selection marker (Longtine et al, 1998; Gueldener et al, 2002). C‐terminal tagging was similarly performed by PCR‐based homologous recombination to replace the endogenous stop codon with cassettes containing appropriate tags and selection markers. To preserve the intact 3’‐UTR for Puf3 regulation, C‐terminal tagging of Puf3‐associated genes was performed by using plasmids pUG27‐CYC1t‐3 × HA or pUG27‐CYC1t‐3 × FLAG containing tags and floxed selection markers, and then, the selection markers were removed by plasmid pSH47 expressing Cre recombinase (Gueldener et al, 2002). MIA40 3’‐UTR mutants were generated by replacing MIA40 3’‐UTR ~150 bp with mutated sequences in which the Puf3‐binding motif (CATGTAAATA) was deleted or reverse complemented. INH1 or HAP4 overexpression strains were generated by introducing a cassette expressing INH1 or HAP4 under the control of CUP1 promoter into the HO locus. PUF3‐15A strain was generated by introducing a cassette expressing PUF3‐15A under the control of endogenous PUF3 promoter (1 kb upstream of ORF) into the HO locus after disrupting endogenous PUF3.

Media used in this study included SCD (0.67% yeast nitrogen base without amino acids, 0.079% complete supplement mixture, and 2% glucose), and SCEG (0.67% yeast nitrogen base without amino acids, 0.079% complete supplement mixture, 3% ethanol, and 3% glycerol), YPD (1% yeast extract, 2% peptone, and 2% glucose), YPEG (1% yeast extract, 2% peptone, 3% ethanol, and 3% glycerol). Strains were grown at 30°C if not indicated.

Plasmids

The plasmids pUG27‐CYCt‐3 × HA or pUG27‐CYCt‐3 × FLAG were generated as follows: CYC1 terminator was inserted into pUG27 (Gueldener et al, 2002) at BglII site by Gibson assembly (Gibson et al, 2009) to generate pUG27‐CYCt; a 3 × HA or 3 × FLAG tag was inserted into pUG27‐CYCt at HindIII‐SalI sites.

Plasmid overexpressing MIA40 was generated as follows: a cassette including MIA40 promoter (1 kb upstream of ORF), ORF, and downstream 1 kb was inserted into p417‐TEF1 at SacI‐BamHI sites (TEF1 promoter was removed by restriction enzyme digestion; Wu & Tu, 2011).

Plasmid expressing PUF3‐15A mutant was made as follows: a fragment including PUF3 promoter (1 kb upstream of ORF), ORF, a 3 × FLAG tag and downstream 1 kb was generated by overlap extension PCR, and inserted into HO‐natNT2 (KanMX4 in HO‐KanMX4 was replaced by natNT2) at SalI‐BglII sites (Voth et al, 2001). Mutations of PUF3 phosphorylation sites were introduced by QuikChange site‐directed mutagenesis (Agilent Technologies).

Plasmid expressing MIG1 was generated as follows: a cassette including MIG1 promoter (1 kb upstream of ORF), ORF, and downstream 1 kb was inserted into pRS416 at KpnI‐SacI sites(Sikorski & Hieter, 1989).

Plasmid expressing NOG2 was generated by inserting NOG2 ORF into p417‐TEF1‐natNT2 (KanMX4 in p417‐TEF1 was replaced by natNT2) at EcoRI‐SalI sites.

Growth test

Growth curves were performed by measuring absorbance at 600 nm (A₆₀₀) in biological triplicates. Strains were cultured from 0.05 OD, and A₆₀₀ was measured every 2 or 3 h for 12 h at the indicated temperature.

For drop dilution assays, strains were grown in SCD media to mid‐log phase (OD ~0.4–0.6), washed with ddH₂O, and then diluted to 10⁻¹, 10⁻², 10⁻³, and 10⁻⁴ OD. Drop dilutions on the indicated plates and incubated at the indicated temperatures for 2 days.

Mitochondrial membrane potential measurement

Strains were cultured to mid‐log phase in SCD and washed once with washing buffer (10 mM HEPES, pH 7.4, 5% glucose). 0.2 OD cells were stained for 20 min in 1 ml staining buffer (10 mM HEPES, pH 7.4, 5% glucose, 125 nM TMRM). The florescence of 20,000 cells was analyzed by BD FACSAria II (BD Biosciences), and results were plotted with FlowJo V10. Mitochondrial localization of TMRM signal was confirmed by microscope.

Glucose consumption assay

Glucose concentration was measured by Glucose Assay Kit (Sigma). Strains were cultured from 0.05 OD, and media were collected at 0.4, 0.8, and 1.5 OD. The media were diluted 40 times and measured following the manufacturer’s instruction. Glucose concentration of the original SCD media was set as 100%.

Total RNA isolation

Strains were grown to mid‐log phase in SCD, and 5–10 OD cells were collected, snap‐frozen in liquid nitrogen, and stored at −80°C. Total RNA was isolated using hot phenol method as described previously (Lyne et al, 2003). Cells were resuspended in 600 µl TES buffer (10 mM Tris–HCl, pH 7.5, 10 mM EDTA, pH 8, 0.5% SDS), and immediately added with 600 µl acidic phenol‐chloroform‐isoamylalcohol (25:24:1). Samples were incubated at 65°C for 1 h and mixed thoroughly every 10 min. RNA was extracted with acidic phenol‐chloroform‐isoamylalcohol (25:24:1, two times) and chloroform‐isoamylalcohol (24:1, one time). Then, RNA was precipitated by adding 1 volume isopropanol and 0.1 volume 3 M NaAc (pH 5.2), and dissolved in RNase‐free water. RNA integrity was assessed by 1% agarose gel electrophoresis.

qRT–PCR

Total RNA was digested with DNase I (NEB) and extracted again. Complementary DNA was synthesized with 1 µg RNA by PrimeScript™ RT Reagent Kit (Takara). qRT–PCR was performed by using CFX96 Touch Real‐Time PCR Detection System (Bio‐Rad) in biological triplicates. PCRs were prepared with TB Green Premix Ex Taq (Takara). Primers used for qPCR were listed in Table EV3, and ACT1 was selected as reference gene for normalization. Data were analyzed by 2^‐∆∆Ct method, and arithmetic mean ± SD were shown in plots.

RNAseq and data processing

mRNA was purified from total RNA by Oligo d(T)₂₅ Magnetic beads (NEB). mRNA was fragmented by incubating with magnesium ion at 94°C for 5 min. Reverse transcription was performed by SuperScript II Reverse Transcriptase (Invitrogen). Complementary DNA was synthesized by DNA polymerase Ⅰ (Takara) and Random primers (Takara). Sequencing libraries were prepared by NEBNext Ultra II DNA Library Prep Kit for Illumina (NEB). The quality of libraries was assessed by 2100 Bioanalyzer Instrument (Agilent) and Qubit (Thermo Fisher Scientific). Libraries were quantified by KAPA Library Quantification Kit (Roche) and sequenced by Hiseq2500 (Illumina). Three biological replicates were prepared for each strain.

All the raw data were assessed by FastQC and showed good quality. Alignment to S. cerevisiae genome (R‐64‐1‐1) was performed by STAR (2.6.1a) with default parameters. Output BAM files were indexed using SAMtools (1.9). Reads counts for each gene were quantified to the Ensembl R64 GTF annotation file by Rsubread package (1.34.7) with default parameters. Normalization to RPKM (reads per kilobase of exon model per million mapped reads) and differential expression analysis was executed by edgeR (3.26.8). TMM (trimmed mean of M‐values) normalization was applied to dataset to account for compositional difference between the libraries. Then the common dispersion and tagwise dispersions were estimated, and the quasi‐likelihood F‐tests was performed. All the non‐coding RNAs and mtDNA‐encoded mRNAs were removed from results due to lack of polyadenylation.

Gene function classification

Gene functional categories were generated according to literatures or Saccharomyces Genome Database (SGD) and listed in Table EV1. High‐fidelity nuclear‐encoded mitochondrial gene list was selected from gene ontology mitochondrion (GO: 0005739) category after manual checking of the SGD database for protein annotation and localization. OXPHOS subunits, glycolysis genes, and glucose transporter genes are from SGD. Gene list of cytosolic ribosome subunits is from Steffen et al, (2012) and gene list of ribosome processing and assembly (Ribi) is from Woolford and Baserga (2013). Gene list of Puf3‐associated transcripts was combined from Gerber et al, 2004; Lapointe et al, (2015), and gene list of Puf3‐associated transcripts encoding mitochondrial proteins was selected according to high‐fidelity mitochondrial gene list. Gene list of Mia40 substrates is from Herrmann & Riemer, 2012; Mordas and Tokatlidis (2015). Gene list of “Proteostasis” is from Quiros et al, (2015). Gene list of “Import machinery” is from Schmidt et al, (2010). Gene list of “OXPHOS assembly factors” was combined from Ndi et al, (2018), Ruhle and Leister (2015), and Soto et al, (2012). Gene list of “mitoRibosome” was combined from Amunts et al, (2014) and Desai et al, (2017). Gene list of “Iron‐sulfur cluster synthesis” is from Lill (2009). Gene list of “Phospholipid trafficking and synthesis” is from Tatsuta et al (2014). Gene list of “Coenzyme Q” is from Stefely and Pagliarini (2017). Gene list of “Carriers” is from Ogunbona and Claypool (2019). Gene list of “cytoplasmic chaperone” is from Verghese et al, (2012). Gene list of proteasome subunits is from Boos et al, (2019), and gene list of Cdc48 cofactors is from Wu et al, (2016). Gene list of “Autophagy” is from Farre and Subramani (2016), Reggiori and Klionsky (2013), and Farre and Subramani (2016).

Protein isolation and Western blot

5–10 OD yeast cell pellets were resuspended in 300 μl yeast lysis buffer (50 mM NaCl, 50 mM NaF, 100 mM Tris–HCl, pH 7.5, 1 mM EDTA, 1 mM EGTA, 1% Triton X‐100, 10% glycerol, 14 mM 2‐mercaptoethanol, 2 mM PMSF, 5 μM pepstatin A, and 10 μM leupeptin). After adding ∼100 μl of glass beads, cells were lysed via three rounds of bead‐beating (40 s of beating followed by 1 min of cooling on ice), and centrifuged at 18,407 g for 10 min at 4°C to collect the supernatants. Proteins were separated by 4–12% ‐gradient gel and then transferred onto nitrocellulose membrane (GE healthcare). Immunodetection was performed by Western Lightning Plus‐ECL (PerkinElmer).

SILAC analysis of mitochondrial proteome

Strains used for SILAC mass spectrometry were in lys2∆ arg4∆ CAN1 + background. Strains were cultured to mid‐log phase in SCD (0.67% yeast nitrogen base without amino acids, 0.079% complete supplement mixture lacking arginine and lysine, 2% glucose) supplemented with 50 mg/l [¹³C₆/¹⁵N₄] L‐arginine and 50 mg/l [¹³C₆/¹⁵N₂] L‐lysine (Cambridge Isotope Laboratory) or non‐labeled L‐arginine and L‐lysine. The average incorporation efficiency of labeled amino acids was detected by LC‐MS/MS, reaching 98.0%.

Three biological replicates were prepared for each strain. Equal amounts of heavy‐isotope and light‐isotope labeled cells (~150 OD) were harvested and mixed together. Mitochondria were isolated following a previously described method with some modifications (Diekert et al, 2001). Briefly, cells were washed once with water, resuspended in TD buffer (10 mM DTT and 100 mM Tris‐SO₄, pH 9.4) and incubated for 15 min at 30°C. Cells were then pelleted and washed once with SP buffer (1.2 M sorbitol and 20 mM potassium phosphate, pH 7.4) and treated with Zymolyase 100T (MP Biomedical) for 40 min at 30°C to generate spheroplasts. After two times of washes with SP buffer, the spheroplasts were resuspended in SHE buffer (0.6 M sorbitol, 20 mM HEPES‐KOH, pH 7.4, 1 mM EGTA, pH 8.0, and 2 mM MgCl₂) supplemented with protease inhibitors and BSA, and homogenized by a French press (EmulsiFlex‐C3, AVESTIN Inc.) at pressures in the range of 1,000–1,500 psi. The mitochondria‐enriched fraction was obtained by differential centrifugation, snap‐frozen by liquid nitrogen, and stored at −80°C until use.

Mitochondria pellets were lysed in 100 µl 1% SDS lysis buffer (50 mM Tris–HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% SDS, EDTA‐free protease inhibitor cocktail (Roche)) and boiled at 98°C for 10 min. After centrifugation at 18,407 g, 4°C for 10 min, the supernatant was collected and mitochondrial proteins were precipitated overnight at −20°C by adding 8 volumes of ice‐cold acetone. The protein pellets were redissolved in 8 M urea in 50 mM ammonium bicarbonate (pH 8.0). Aliquot of 100 μg protein was reduced in 2 mM DTT at 56°C for 30 min followed by alkylation in 5 mM iodoacetamide at dark for 1 h. Around 2.5 μg of sequencing grade modified trypsin was added to each protein solution and incubated at 37 °C overnight. The resulting tryptic peptides were desalted and fractionated into eight fractions with Thermo Scientific™ Pierce™ High pH Reversed‐Phase Peptide Fractionation Kit (Thermo Fisher Scientific).

Each fraction of peptides was separated by a home‐made analytical capillary column (50 μm × 15 cm) packed with 5 μm spherical C18 reversed‐phase material (YMC). A Waters nanoAcquity UPLC system was used to generate the following HPLC gradient: 0–30% B in 75 min, 30–70% B in 15 min, 70–90% B in 5 min (A = 0.1% formic acid in water, B = 0.1% formic acid in acetonitrile). The eluted peptides were sprayed into a Q Exactive mass spectrometer (Thermo Fisher Scientific) equipped with a nano‐ESI ion source. The mass spectrometer was operated in data‐dependent mode with one MS scan followed by ten HCD (High‐energy Collisional Dissociation) scans for each cycle.

Data were analyzed using MaxQuant (1.6.17.0) and searched against UniProt_Scerevisiae protein database. The search parameters were set as follows: variable modification included “Oxidation (M)”, “Acetyl (Protein N‐term” and “Carbamidomethyl (C)”; digestion enzyme was chosen as “Trypsin/P”, and 2 missed cleavage sites were allowed; protein quantification was based on at least one unique peptide with minimal length of 7 amino acids; protein identification was based on at least one peptide, and false discovery rate at peptide spectrum level and protein level was 0.01; “First search peptide tolerance” was 20 ppm, and “Main search peptide tolerance” was 4.5 ppm. Other parameters were set by default if not indicated. More than 3000 proteins were identified in each experiment, and only high‐fidelity mitochondrial proteins with 3 quantification results were kept for analysis. The ratios of proteins were log₂‐transformed and loaded into Perseus (1.6.14.0) to calculate P value (two‐tailed t‐test). Protein groups with P < 0.05, and fold change > 1.5 or < 0.67 were considered as significantly changed in ∆IV or ρ⁰ cells.

Puf3 protein immunoprecipitation, phosphatase treatment, and Phos‐tag gel analysis

Yeast cells were cultured in SCD to mid‐log phase, and 50 OD cells were harvested. Cell pellets were washed once with 5% TCA, flash frozen in liquid nitrogen, and stored at −80°C until use. Cells were resuspended in 1 volume 1 M Tris–HCl (pH 8.0) plus 2 volumes SDS buffer (100 mM Tris–HCl, pH 7.5, 300 mM NaCl, 4% SDS, 8% BME), and lysed by six rounds of bead beating (20 s beating followed by 1 min of cooling on ice). The cell lysates were diluted with 20 volumes Triton buffer (50 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1% Triton), and incubated with 8–10 µl anti‐FLAG agarose beads (Sigma) at 4°C for 5–6 h. The beads were washed three times with Triton buffer for 10 min and eluted with FLAG peptides (2 mg/ml) overnight at 4°C.

For phosphatase treatment, the beads with immunoprecipitates were treated with 400 U Lambda phosphatase (NEB) at 30°C for 20 min. The reaction was stopped by adding 4× sample buffer (0.2 M Tris–HCl, pH 6.8, 8% SDS, 28% glycerol, 8% BME, 0.08% Bromophenol Blue) and heated at 98°C for 10 min.

Puf3‐FLAG immunoprecipitates were loaded onto 6% Phos‐tag gel, run at 80 V for 1 h, 120 V for 30 min, then at 150 V till the leading front dye run out the gel. Gel was rinsed once with ddH₂O, washed three times with transfer buffer (25 mM Tris, 192 mM Glycine) supplemented with 10 mM EDTA for 8 min, and then washed once with transfer buffer for 8 min. Transfer proteins to PVDF membrane at 140 V, 300 mA for 3 h at 4°C.

SILAC analysis of Puf3 phosphorylation sites

PUF3‐FLAG and PUF3‐FLAG ρ⁰ cells were cultured to mid‐log phase in SCD supplemented with labeled or unlabeled arginine and lysine. About 350 OD cells were harvested for each strain and mixed together. The protein extraction and beads incubation were performed as described above. After that, anti‐FLAG agarose beads were washed four times with yeast lysis buffer and eluted overnight with FLAG peptides at 4°C. The sample was separated by 4–12% gradient gel and stained with coomassie blue. Protein bands on the gradient gel were de‐stained and then reduced in 10 mM DTT at 56°C for 30 min followed by alkylation in 55 mM iodoacetamide at dark for 1 h. The protein bands were isolated and in‐gel digested with sequencing grade trypsin (10 ng/μl trypsin, 50 mM ammonium bicarbonate, pH 8.0) overnight at 37°C. Peptides were extracted with 5% formic acid/50% acetonitrile and 0.1% formic acid/75% acetonitrile sequentially and then concentrated to ~20 μl. The extracted peptides were separated by an analytical capillary column (50 μm × 15 cm) packed with 5 μm spherical C18 reversed‐phase material (YMC). A Waters nanoAcquity UPLC system (Waters) was used to generate the following HPLC gradient: 0–30% B in 40 min, 30–70% B in 15 min (A = 0.1% formic acid in water, B = 0.1% formic acid in acetonitrile). The eluted peptides were sprayed into a LTQ Orbitrap Velos mass spectrometer (Thermo Fisher Scientific) equipped with a nano‐ESI ion source. The mass spectrometer was operated in data‐dependent mode with one MS scan followed by four CID (Collision‐Induced Dissociation) and four HCD (High‐energy Collisional Dissociation) MS/MS scans for each cycle. Database searches were performed on an in‐house Mascot server (Matrix Science Ltd.) against yeast Puf3 protein sequence. SILAC quantification of the phosphorylated peptides was performed by Mascot Distiller. The search parameters are 7 ppm mass tolerance for precursor ions; 0.5 Da mass tolerance for product ions; three missed cleavage sites were allowed for trypsin digestion; and the following variable modifications were included: oxidation on methionine, carbamidomethylation on cysteine, phosphorylation on serine, threonine, and tyrosine. The tandem mass spectra of matched phosphorylated peptides were manually checked for their validity. Phosphorylation level of each specific amino acid sites was quantified manually.

Ribosome footprinting and data analysis

Ribosome footprinting was performed as previously described (Ingolia, 2010; Ingolia et al, 2012). Briefly, yeast cells were cultured to mid‐log phase in SCD, treated with 100 μg/ml cycloheximide for 2 min. 500 OD cells were harvested by vacuum filtration and snap‐frozen in polysome lysis buffer (20 mM Tris–HCl, pH 8, 140 mM KCl, 1.5 mM MgCl₂, 100 μg/ml cycloheximide, and 1% (v/v) Triton). 250 μl aliquot of cell lysate was digested by 7.5 μl RNase Ⅰ (Ambion) at room temperature for 1 h, and digestion was stopped by adding 5 μl SUPERase•In (Ambion). The samples were loaded on sucrose gradients (10%–50%) and centrifuged for 3 h at 209,230 g, 4°C with SW41 Ti rotor. The gradients were separated by Piston Gradient Fractionator™ (Biocomp), and A₂₅₄ values of gradients were continually monitored by Triax^TM Flow Cell (Biocomp). Fractions representing the monosome peak were collected. RNA was isolated by TRIzol reagent (Invitrogen) and separated by 15% polyacrylamide TBE‐urea gel. 26‐nt to 34‐nt region demarcated by the marker oligos was carefully excised. After gel extraction, rRNA was removed by Ribo‐Zero rRNA Removal Kit (Epicentre). The RNA fragments were end‐repaired by T4 Polynucleotide Kinase (NEB) and ligated with Universal miRNA Cloning Linker (NEB) by T4 RNA Ligase 2 (NEB). Reverse transcription was performed with SuperScript III (Invitrogen), and then, the products were circularized by CircLigase (Epicentre). The libraries were amplified by Phusion High‐Fidelity DNA Polymerase (NEB) and sequenced by Hiseq2500 (Illumina). Alignment to yeast genome (R‐64‐1‐1) was performed by using TopHat (2.1.1) with library‐type set as fr‐secondstrand. RPKM was calculated by Rsubread package (1.34.7) and edgeR (3.26.8).

Polysome fractionation analysis

Polysome fractionation analysis was performed as previously described (Lee & Tu, 2015). Yeast cells were cultured to mid‐log phase in SCD, treated with 100 µg/ml cycloheximide for 15 min, and 75 OD cells were harvested. Cell pellets were resuspended in 500 μl lysis buffer (10 mM Tris–HCl, pH 7.5, 100 mM NaCl, 30 mM MgCl₂, 200 μg/ml cycloheximide, 200 mg/ml heparin, 0.1% BME, EDTA‐free protease inhibitor cocktail (Roche), 1 mM EDTA, 1 mM PMSF, 5 mM pepstatin A, 10 mM leupeptin, 0.2 mM Na‐orthovanadate, 10 mM β‐glycerolphosphate, 10 mM NaF, and 20 U/ml SUPERase•In) and mixed with 250 μl glass beads. The cells were lysed by six rounds of bead beating (20 s beating followed by 1 min of cooling on ice). The supernatant was collected after centrifugation at 18,407 g, 4°C for 10 min. 200 μl cell lysate was loaded onto sucrose gradients (10–50%), and fractionated by centrifugation at 209,230 g, 4°C for 3 h with SW41 Ti rotor. The gradients were separated and monitored by Piston Gradient Fractionator™ and Triax^TM Flow Cell. For Puf3 polysome analysis, fractions representing the polysome peak were collected. Proteins were precipitated by 20% TCA and boiled at 98°C for 10 min with 50 µl 4× sample buffer.

Author contributions

Conception of the project, study supervision, and manuscript writing: HJ. Most of the experiments, data analysis, figure preparation, and manuscript editing: SiL. SNF1‐related experiments and data analysis: ShL. Puf3 phosphorylation analysis: BH. Generation of part of the strains: LaL. MS experiments and data analysis: LiL and SC. RNAseq and data analysis: JW and TC.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Table EV1

Table EV2

Table EV3

Dataset EV1

Dataset EV2

Dataset EV3

Dataset EV4

Dataset EV5

Dataset EV6

Dataset EV7

Dataset EV8

Dataset EV9

Dataset EV10

Source Data for Expanded View

Review Process File

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 7

EMBO Reports (2021) 22: e51606.

Data availability

The deep‐sequencing raw data have been deposited in NCBI's Gene Expression Omnibus (Edgar et al, 2002) and are accessible through GEO Series accession number GSE162197 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE162197). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository (Perez‐Riverol et al, 2019) with the dataset identifier PXD022805 (http://www.ebi.ac.uk/pride/archive/projects/PXD022805).