The Ccr4-Not complex regulates TORC1 signaling and mitochondrial metabolism by promoting vacuole V-ATPase activity

The Ccr4-Not complex functions as an effector of multiple signaling pathways that control gene transcription and mRNA turnover. Consequently, Ccr4-Not contributes to a diverse array of processes, which includes a significant role in cell metabolism. Yet a mechanistic understanding of how it contributes to metabolism is lacking. Herein, we provide evidence that Ccr4-Not activates nutrient signaling through the essential target of rapamycin complex 1 (TORC1) pathway. Ccr4-Not disruption reduces global TORC1 signaling, and it also upregulates expression of the cell wall integrity (CWI) pathway terminal kinase Mpk1. Although CWI signaling represses TORC1 signaling, we find that Ccr4-Not loss inhibits TORC1 independently of CWI activation. Instead, we demonstrate that Ccr4-Not promotes the function of the vacuole V-ATPase, which interacts with the Gtr1 GTPase-containing EGO complex to stimulate TORC1 in response to nutrient sufficiency. Bypassing the V-ATPase requirement in TORC1 activation using a constitutively active Gtr1 mutant fully restores TORC1 signaling in Ccr4-Not deficient cells. Transcriptome analysis and functional studies revealed that loss of the Ccr4 subunit activates the TORC1 repressed retrograde signaling pathway to upregulate mitochondrial activity. Blocking this mitochondrial upregulation in Ccr4-Not deficient cells further represses TORC1 signaling, and it causes synergistic deficiencies in mitochondrial-dependent metabolism. These data support a model whereby Ccr4-Not loss impairs V-ATPase dependent TORC1 activation that forces cells to enhance mitochondrial metabolism to sustain a minimal level of TORC1 signaling necessary for cell growth and proliferation. Therefore, Ccr4-Not plays an integral role in nutrient signaling and cell metabolism by promoting V-ATPase dependent TORC1 activation.


Introduction
Eukaryotic cells respond to nutrient flux by modifying signaling through pathways that promote adaptive growth and proliferative responses. These mechanisms allow cells to dynamically adjust their metabolism to coordinate growth and proliferation with the availability of the nutrients that support these processes [1]. This regulation involves coordinated control at the different stages of the gene expression pathway, including at the level of gene transcription, mRNA stability, protein translation, and protein degradation [2]. Oftentimes these regulators function at multiple steps in the gene expression process to ensure efficient and coordinated responses to environmental nutrient flux. Understanding how such shared regulators operate within the context of nutrient signaling is essential for determining how cells adapt to shifts in their nutrient environment.
The evolutionarily conserved Ccr4-Not complex regulates gene expression at the transcriptional and post-transcriptional stages, while it also contributes to translational control and proteostasis maintenance [3]. Its role in these diverse pathways highlights its wide-ranging importance in the cell, which is further underscored by the embryonic lethality caused by loss of individual Ccr4-Not members [4,5]. Yet how Ccr4-Not mechanistically regulates these diverse activities remains unclear. Ccr4-Not is best characterized in budding yeast where it exists as an approximately 1 MDa complex containing the Ccr4, Caf1, Caf40, Caf130, and Not1-5 subunits [3]. The complex has a modular organization with the Ccr4 and Caf1, Caf40 and Caf130, and the Not2-5 subunits forming distinct sub-modules organized onto a Not1 scaffold [6]. Ccr4 is an mRNA deadenylase that degrades polyadenylated mRNA, and it functions as the main eukaryotic deadenylase [7,8]. The Not4 subunit is a RING domain ubiquitin ligase that additionally has an RNA recognition motif (RRM) and C3H1 domain (referred to as the RRM-C domain), which are candidate RNA binding domains [9][10][11]. Whether Not4 binds RNA remains unknown, although the RRM-C domain contributes to Not4 in vivo ligase activity suggesting Ccr4-Not may integrate RNA metabolism with ubiquitin ligation [11]. The Not2, Not3, and Not5 subunits likely contribute regulatory and structural functions to the complex [3]. While the role of the yeast Caf40 and Caf130 submodule remains unknown, the Drosophila and mammalian Caf40 orthologs recruit Not4 ligases into the Ccr4-Not complex [12,13].
Since Ccr4-Not controls every stage of the gene expression pathway, it is an ideal effector for integrating global responses to nutrient flux. Support for this connection has been provided by studies demonstrating that yeast Ccr4-Not acts within both the glucose-activated Ras/PKA, and the nutrient activated target of rapamycin complex 1 (TORC1), pathways [14][15][16].
TORC1 also phosphorylates the regulatory factor Tap42 to inactivate PP2A and PP2A-like phosphatases to prevent dephosphorylation and activation of nutrient-stress responsive transcription factors such as Gln3 [37].
In this report, we demonstrate that Ccr4-Not deficiency dramatically impairs TORC1 signaling, and it also deregulates expression of the Mpk1/Slt2 (hereafter referred to as Mpk1) mitogen activated protein kinase (MAPK) kinase that is a downstream effector of the CWI stress signaling pathway that inhibits TORC1. However, we find that TORC1 inhibition in a Ccr4-Not mutant (ccr4Δ) is independent of Mpk1 deregulation or CWI signaling. Instead, we demonstrate that Ccr4-Not disruption reduces the stability and function of the vacuole V-ATPase, which causes vacuole acidification defects and reduced TORC1 signaling. Bypassing the V-ATPase requirement in TORC1 activation through expression of a constitutively active Gtr1 GTPase fully restores TORC1 activity in ccr4Δ. However, this Gtr1 mutant fails to restore growth of ccr4Δ under TORC1 inhibitory conditions, suggesting Ccr4-Not has additional effects on TORC1 signaling downstream of the V-ATPase. Transcriptome analysis (RNA-seq) and functional studies reveal that Ccr4-Not disruption increases mitochondrial metabolism, which TORC1 normally represses by inhibiting mitochondrial retrograde signaling. Intriguingly, preventing mitochondrial upregulation in ccr4Δ further impairs TORC1 signaling and leads to mitochondrial metabolic defects. These data suggest that the enhanced mitochondrial metabolism in Ccr4-Not mutants functions as an adaptive response required to sustain a minimal level of TORC1 signaling needed for cell growth and proliferation.

Ccr4-Not activates TORC1 signaling independently of its role in mRNA deadenylation
To further define how Ccr4-Not functions in the TORC1 pathway, we cultured wild-type (WT) cells, as well as cells lacking the Not4 ubiquitin ligase (not4Δ), the Caf40 subunit (caf40Δ), or the Ccr4 mRNA deadenylase (ccr4Δ) to mid-log phase in nutrient rich (YPD) media. As a control for impaired TORC1 signaling, we analyzed a mutant lacking the vacuolespecific V-ATPase subunit Vph1 (vph1Δ) [28]. As expected, vph1Δ reduced TORC1 activity while both ccr4Δ and not4Δ, but not caf40Δ, also decreased TORC1 signaling, which was determined by quantifying S6ph (Fig 1A). Because not4Δ grows poorly, we analyzed the ccr4Δ further to define how Ccr4-Not activates TORC1 since ccr4Δ has only a minor growth phenotype. To confirm the TORC1 signaling defect using an independent readout, WT and ccr4Δ were transformed with a plasmid expressing Myc-tagged Gln3. Gln3 is phosphorylated when TORC1 is active, and it becomes rapidly dephosphorylated upon TORC1 inhibition [37,38]. Cells were grown to mid-log phase, and Gln3 phosphorylation was detected by mobility shift assay using Phostag impregnated SDS-PAGE, which separates proteins based on their phosphorylation state [39,40]. As a control for TORC1 inhibition, WT cells were treated with 200 nM rapamycin (Rap) for 30 minutes before harvesting. Hyperphosphorylated Gln3 (indicated by upper arrow in Fig 1B) was detected in mock-treated WT cells, while rapamycin-mediated TORC1 inhibition increased Gln3 mobility thus indicating decreased Gln3 phosphorylation (denoted by the bottom arrow in Fig 1B). Gln3 mobility from ccr4Δ resembled that from the rapamycin-treated WT cells, thus indicating ccr4Δ has reduced TORC1 signaling ( Fig 1B). Therefore, two independent readouts for TORC1 activity indicate that Ccr4-Not disruption decreases signaling through this pathway.
To determine if Ccr4-Not activates TORC1 via its role in mRNA degradation, WT and ccr4Δ were transformed with either control vector or, in ccr4Δ, with vector overexpressing either FLAG-tagged CCR4 or the catalytically inactive ccr4-1 mutant [8,16]. Cells were grown (A) TORC1 activity was analyzed by performing immunoblot (IB) analysis for S6ph and total S6 from WT, not4Δ, caf40Δ, ccr4Δ, and vph1Δ cell extracts. Numbers reflect the S6ph/S6 ratio with WT set to 1 and the mutant S6ph/S6 ratio normalized to WT. The data also are represented as a histogram, and they are representative of three independent experiments. (B) WT and ccr4Δ were transformed with a Gln3-13XMyc expression plasmid and then cultured in SC selective media to mid-log phase. The WT was then mock treated or treated with 200 nM rapamycin for 30 minutes before harvesting. Total cell extracts were resolved in an 8% SDS-PAGE gel containing 150 μM Phostag reagent and then probed by α-Myc IB to detect Gln3. The arrows indicate hyperphosphorylated (upper) and hypophosphorylated (lower) Gln3, while the asterisk indicates Gln3 cleavage products. Glucose-6-phosphate dehydrogenase (G6PDH) levels were analyzed as a loading control. The results are representative of three independent experiments. (C) Analysis of TORC1 in WT and ccr4Δ transformed with control vector, or vector expressing FLAG-tagged wild-type Ccr4 or the enzymatically inactive ccr4-1 mutant. Samples were analyzed and quantified as described in (A), and the results are representative of four independent experiments. (D) Analysis of TORC1 signaling in WT, ccr4Δ, and various PUF deletion mutants performed as in (A). Data are representative of three independent experiments. (E) RT-qPCR analysis of the indicated TORC1 pathway genes normalized to the SPT15 housekeeping gene. Data are the average and standard deviation (SD) of three independent experiments with pairwise Student's t-test performed to test significance. � -p<0.05; �� -p<0.01. (F) WT in nutrient defined synthetic complete (SC) media lacking uracil (to select for plasmid maintenance), and then they were analyzed for TORC1 activity. The ccr4Δ vector control cells exhibited reduced S6ph relative to WT, while both CCR4 and ccr4-1 restored S6ph levels ( Fig 1C). These data indicate Ccr4-Not activates TORC1 independently of its role in mRNA degradation. To provide further support that the TORC1 signaling defect is independent of Ccr4-Not mediated mRNA decay, we analyzed TORC1 in cells lacking individual Pumilio (PUF) family RNA binding proteins, including Puf3 and Puf4, which recruit Ccr4-Not to many of its mRNA substrates for deadenylation [22,41,42]. While ccr4Δ decreased TORC1 signaling, loss of individual PUF proteins had no effect ( Fig 1D). Collectively, these data further support the conclusion that Ccr4-Not activates TORC1 independently of its role in mRNA deadenylation. We next tested if this TORC1 inhibition was due to reduced mRNA expression for any of the TORC1 components. The mRNA levels for all TORC1 subunits (TOR1, TOR2, LST8, KOG1, and TCO89), and the downstream ribosomal S6 kinase (YPK3), were unaffected or increased slightly (KOG1 and TCO89) by ccr4Δ ( Fig 1E). Therefore, Ccr4-Not loss does not impair TORC1 signaling by reducing the mRNA expression of TORC1 constituents.
Ccr4-Not facilitates co-translational assembly of some large macromolecular complexes [43][44][45], so we next tested if ccr4Δ inhibited TORC1 by reducing TORC1 complex stability. To do so, we evaluated the association of the essential (and unique) TORC1 subunit Kog1 with the Lst8 subunit, which is an essential component shared between TORC1 and the additional TOR kinase complex TORC2 [46]. Kog1 immunoprecipitation (IP), but not the control IP, weakly co-precipitated Lst8 from WT extracts, while the ccr4Δ enhanced this association ( Fig  1F). Intriguingly, the increased Kog1-Lst8 interaction correlated with higher overall Lst8 levels in ccr4Δ (Fig 1F), which is not due to increased LST8 mRNA expression ( Fig 1E). These data suggest ccr4Δ may increase Lst8 expression post-transcriptionally. Although the mechanism explaining this increased Lst8-Kog1 interaction is not yet known, these data indicate that Ccr4-Not loss does not disrupt TORC1 to inhibit its signaling.

Ccr4-Not activates TORC1 independently of CWI pathway repression
The CWI pathway is activated by cell wall or membrane stress [47], and complete TORC1 inhibition is a known activator of CWI signaling [47]. CWI activation increases Rho1 GTPase activity, which signals through Pkc1 to ultimately activate the terminal MAPK Mpk1 to induce transcription of stress-responsive genes, including MPK1 itself as well as genes such as FKS2 (outlined in Fig 2A) [48]. Activated Rho1 also transiently inhibits TORC1 until cells adapt to the CWI activating stress [39], suggesting TORC1 and CWI exhibit mutually inhibitory crosstalk signaling. From WT and individual Ccr4-Not mutants grown in nutrient rich (YPD) media, we determined if Mpk1 signaling was affected by monitoring both activated (phosphorylated Mpk1, denoted Mpk1ph) and total Mpk1 [49]. For a comparison, we analyzed cells lacking the non-essential TORC1 subunit Tco89 whose loss decreases, but does not ablate, TORC1 signaling [50,51]. The tco89Δ increased both Mpk1ph and total Mpk1 levels compared to WT, but after normalizing for the increase in Mpk1 expression the tco89Δ resulted only in a minor activation of Mpk1 (increased Mpk1ph/Mpk1 ratio) (Fig 2B). Both ccr4Δ and not4Δ, but not caf40Δ, similarly affected Mpk1ph and Mpk1 levels. While not4Δ increased Mpk1 activation similar to that detected for tco89Δ, the ccr4Δ resulted in a modest decrease in Mpk1 activation while caf40Δ had no effect ( Fig 2B). These data demonstrate that while absolute levels of and ccr4Δ expressing Kog1-6XHA and Lst8-9XMyc were cultured to mid-log phase before harvesting and performing the indicated immunoprecipitations (IPs) and IBs. The arrow indicates the Lst8-specific band and the asterisk denotes the IgG heavy chain. The data are representative of three independent experiments.
https://doi.org/10.1371/journal.pgen.1009046.g001 Mpk1ph and total Mpk1 are deregulated in Ccr4-Not and TORC1 mutants, overall Mpk1 activation remains only modestly increased (in not4Δ and tco89Δ), or it is decreased (in ccr4Δ) ( Fig 2B). Ccr4-Not regulates post-transcriptional stability of MPK1 mRNA via binding of the Ccr4 subunit to MPK1 mRNA [22], so the increase in ccr4Δ of total Mpk1 likely reflects dysregulation of MPK1 mRNA turnover. Indeed, we find that ccr4Δ increased MPK1 mRNA expression ( Fig 2C). However, this mechanism cannot fully account for the effect Ccr4-Not disruption has on Mpk1 expression since Ccr4 also binds HOG1 mRNA (which encodes the p38 stress-inducible MAPK) [22,52], yet ccr4Δ neither affected basal Hog1 protein levels nor its activation after osmostress ( Fig 2D). Collectively, these data suggest that while reduced TORC1 signaling in tco89Δ and Ccr4-Not mutants deregulate Mpk1 expression, they have only minor effects on the overall action of Mpk1.
To further probe how ccr4Δ affects Mpk1 signaling, we analyzed Mpk1ph and total Mpk1 from WT and ccr4Δ vector control cells grown in nutrient defined SC selection media, along with ccr4Δ overexpressing either CCR4 or the ccr4-1 mutant. Surprisingly, in these growth conditions we did not detect an increase in Mpk1ph or total Mpk1 in the ccr4Δ vector control cells ( Fig 2E). However, ccr4Δ overexpressing CCR4 increased Mpk1ph without altering total Mpk1 levels, an effect dependent on its deadenylase activity since the ccr4-1 mutant had no effect ( Fig 2E). These results suggest that overexpressing the WT Ccr4 deadenylase may generate an mRNA degradation stress response that activates Mpk1 signaling. The data also suggest that the cellular growth environment has a significant role in determining if Ccr4-Not disruption affects Mpk1 expression since it does so under nutrient rich (YPD) conditions ( Fig 2B), but not in nutrient defined SC media ( Fig 2E).
TORC1 inhibition enhances Mpk1 signaling to induce expression of proteasomal subunits and chaperones, which enhances protein degradation and maintains cellular homeostasis [53]. Ccr4-Not also interacts with the proteasome [54], and Not4 is essential for correct proteasome assembly, catalytic activity, and cellular proteostasis [10,11,55]. Because Ccr4-Not loss represses TORC1 and deregulates Mpk1 expression, we analyzed the effect it has on proteasome activity by monitoring 20S core particle activity. As previously demonstrated, not4Δ dramatically increased 20S function (Fig 2F) [10,11]. While caf40Δ had no effect, ccr4Δ caused a minor increase in 20S activity that resembled the increase caused by the vacuole V-ATPase mutant vph1Δ, which inhibits TORC1 (Fig 2F and Fig 1A). To test if ccr4Δ activates the proteasome by repressing TORC1, WT and ccr4Δ were mock-treated or treated with rapamycin to completely repress TORC1 activity. TORC1 inhibition increased proteasome activity in both WT and ccr4Δ to similar extents ( Fig  2G). Therefore, the residual TORC1 signaling in ccr4Δ suffices to maintain TORC1-dependent proteasome repression even though Mpk1 expression is deregulated in these cells.
TORC1 incorporates either the Tor1 or Tor2 kinase to mediate signaling, whereas the TORC2 complex utilizes only the Tor2 kinase. Consequently, a tor1Δ is viable and has minimal growth defects since Tor2 functions redundantly within TORC1 [46]. We next determined if combining ccr4Δ with tor1Δ or mpk1Δ synergistically impaired TORC1 signaling. While ccr4Δ inhibited TORC1 as expected, neither the tor1Δ nor mpk1Δ affected TORC1 activity. Furthermore, the tor1Δ also had no effect either on Mpk1 activation or expression ( Fig 3A). Importantly, the ccr4Δ tor1Δ repressed both absolute S6ph levels and total S6 protein ( Fig 3A). Although the S6ph/S6 ratio was similar between ccr4Δ and ccr4Δ tor1Δ (Fig 3A), the effects on total S6ph and S6 levels by ccr4Δ tor1Δ suggests the combinatorial mutant has a synthetic negative effect on TORC1 activity. This observation would be consistent with a previous report comparing the level of 20S activity from each mutant to WT by pairwise Student's t-test. � p<0.05; ��� p<0.005; ns-not significant. (G) As in (F), except the WT and ccr4Δ cultures were mock treated or treated with 200 nM rapamycin for four hours. � p<0.05; ��� p<0.005; ns-not significant.

PLOS GENETICS
The Ccr4-Not complex regulates TORC1 signaling increased total S6 levels, which resulted in a net reduction in S6ph ( Fig 3A). The significance, if any, of this observation currently is unknown. Importantly, the ccr4Δ mpk1Δ does not restore TORC1 signaling compared to ccr4Δ (Fig 3A), thus demonstrating that deregulated Mpk1 expression does not mediate TORC1 inhibition in Ccr4-Not mutants.
The CWI pathway effector Rho1 GTPase transiently inhibits TORC1 when it is bound to GTP (the active signaling state) [39]. To exclude the possibility that ccr4Δ activates Rho1 independently of its role in the canonical CWI pathway to repress TORC1, WT and ccr4Δ cells were transformed with control vector, or a high copy vector overexpressing the SAC7 genomic locus. Sac7 is a GTPase activating protein (GAP) that represses Rho1 signaling (Fig 2A) [48]. As expected, TORC1 activity in the ccr4Δ vector control was reduced relative to the WT control, while SAC7 overexpression failed to restore TORC1 signaling in ccr4Δ ( Fig 3B). Furthermore, SAC7 overexpression also failed to restore ccr4Δ growth under TORC1 inhibitory conditions ( Fig 3C). These data demonstrate that Ccr4-Not disruption does not enhance Rho1 signaling to repress TORC1. In contrast, a Tor1 gain of function mutant (Tor1 I1954V ) expressed in ccr4Δ partially restored TORC1 function ( Fig 3D) [56]. This partial TORC1 rescue is due to competition of the Tor1 I1954V mutant with endogenous WT Tor1 and Tor2 kinases for incorporation into TORC1. However, the modest increase in TORC1 signaling suffices to restore growth in TORC1 inhibitory conditions ( Fig 3E). Collectively, our results demonstrate that neither Mpk1 deregulation, nor Rho1 activation, explain the TORC1 inhibition in the Ccr4-Not mutant.

Ccr4-Not regulates V-ATPase stability and function to activate TORC1
Ccr4-Not mutants previously were identified in a genetic screen for vacuole regulators [57]. Because the vacuole has a crucial role in TORC1 activation [23, 28], we tested if ccr4Δ altered vacuole function. WT and ccr4Δ grown in YPD were stained with the vacuole-specific dye FM 4-64, analyzed by confocal microscopy, and the vacuole number per cell was enumerated. WT cells had 1-2 vacuoles per cell (~85% of total cells), with a smaller number of cells (~15%) having three or more vacuoles (Fig 4A and Fig 4B). In contrast, ccr4Δ had no more than two vacuoles per cell, with most cells containing only a single large vacuole ( Fig 4A and Fig 4B). Cells containing single large vacuoles indicate defects in vacuole acidification, which is a process that requires functional vacuole V-ATPase activity [58]. To test if ccr4Δ altered vacuole acidification, WT and ccr4Δ were stained with 6-carboxyfluorescein diacetate (6-CFDA) that is imported into the vacuole where it fluoresces as a function of decreasing pH. This fluorescence was quantified using a ratiometric assay that allows a relative assessment of vacuole acidity [59]. As a control, we analyzed a vma3Δ mutant, which ablates all cellular V-ATPase activity [60]. The vma3Δ resulted in a higher fluorescence ratio relative to WT, which indicates defective vacuole acidification, while the ccr4Δ had an intermediate effect (Fig 4C). These data demonstrate that ccr4Δ increases vacuole pH, which suggests that Ccr4-Not disruption impairs V-ATPase activity to inhibit TORC1.
A V-ATPase defect in Ccr4-Not mutants could inhibit TORC1 indirectly by increasing vacuole pH and reducing the level of free intra-vacuole amino acids required to activate TORC1 [31]. Alternatively, the V-ATPase interacts with Gtr1 in the EGO complex to activate TORC1 [28], so V-ATPase defects could impair TORC1 activation directly by reducing Gtr1-dependent activation. Because growth of V-ATPase mutants in acidic media partially suppresses V-ATPase phenotypes [61], we initially tested if an acidic growth environment could restore TORC1. WT and ccr4Δ were cultured in YPD media buffered to pH 4.7, or standard YPD, which is pH~6.0. The acidic environment had no discernible effect on WT TORC1 signaling, while the acidic pH failed to restore TORC1 signaling in ccr4Δ (Fig 4D). Therefore, an acidic growth environment is insufficient to restore TORC1 activity in the Ccr4-Not mutant. Next, we determined if Ccr4-Not disruption altered V-ATPase stability by immunoprecipitating (IP) the V 0 Vph1 subunit from WT and ccr4Δ cells expressing the V 1 subunit Vma1 as a GFP fusion (Vma1-GFP). The α-Vph1 and α-IgG control IPs were split into two and either probed with α-GFP to detect Vma1 co-association, or probed with α-Vph1 to assess IP efficiency. The α-Vph1 IP from WT extracts co-precipitated two specific Vma1 bands, with the uppermost band likely being a post-translationally modified Vma1 as it is extensively modified [62][63][64]. The ccr4Δ dramatically reduced Vma1 co-association with Vph1, indicating that V-ATPase V 0 -V 1 interactions are less stable in ccr4Δ (Fig 4E). The Vph1-specific antibody has poor sensitivity, and Vph1 is undetectable in the input samples, so the α-Vph1 IB confirmed the IP efficiency between the samples ( Fig 4E). Intriguingly, we do find that α-Vph1 IP from WT extracts results in multiple Vph1-specific bands, while ccr4Δ selectively reduces the presence of the uppermost Vph1-specific band (denoted by the red arrow in Fig 4E). As with Vma1, Vph1 is extensively post-translationally modified, including by ubiquitination, which promotes Vph1 turnover and vacuole membrane remodeling under TORC1 inhibitory conditions [62,[64][65][66]. These data indicate that Ccr4-Not promotes the V 0 -V 1 interactions necessary for V-ATPase activity that is key for vacuole acidification and TORC1 activation. They also suggest the possibility that Ccr4-Not may promote V-ATPase stability by affecting the post-translational modification state of core V-ATPase subunits.
If ccr4Δ impairs V-ATPase function to inhibit TORC1, then bypassing the V-ATPase in ccr4Δ to activate TORC1 should restore signaling through this pathway. To directly test this possibility, we used the constitutively active Gtr1 Q65L (Gtr1 GTP ) mutant that bypasses the requirement for the V-ATPase in TORC1 activation [27,28]. WT and ccr4Δ were transformed with control or Gtr1 GTP -HA expression vectors, and then TORC1 signaling was analyzed. Gtr1 GTP in WT cells had no obvious effect on TORC1 activity, but Gtr1 GTP expression restored TORC1 signaling in ccr4Δ ( Fig 4F). However, unlike the Tor1 I1954V mutant that partially restored ccr4Δ TORC1 activity and fully rescued its growth in the presence of rapamycin ( Fig  3C and 3D and Fig 4G), Gtr1 GTP expression failed to restore ccr4Δ growth under these conditions ( Fig 4G). Gtr1 acts at the vacuole surface upstream of TORC1 to activate TORC1 in response to nutrient sufficiency, while the Tor1 I1954V directly incorporates into the TORC1 complex. Therefore, these data suggest that Ccr4-Not may activate TORC1 by promoting V-ATPase function, and that it also may function at a step downstream of the V-ATPase to regulate TORC1 directly. This possibility would be consistent with the observation that ccr4Δ may affect the TORC1 complex directly since it increases Lst8-Kog1 interactions (Fig 1E).

Ccr4-Not disruption remodels the transcriptome to enhance mitochondrial activity
To further delineate how Ccr4-Not contributes to TORC1 regulation, we performed transcriptome sequencing (RNA-seq) from triplicate cultures of WT and ccr4Δ cells grown in nutrient rich YPD media. Since Ccr4 binds large numbers of mRNA transcripts to affect gene expression globally [22], we utilized a stringent 2-fold change in gene expression between WT and ccr4Δ to define those genes most affected by Ccr4-Not disruption. The expression of 499 total genes in ccr4Δ met these criteria, including 157 downregulated and 342 upregulated genes that are represented as a heatmap (Fig 5A, S1 Fig and S1 File). Gene ontology (GO) analysis of the downregulated genes revealed the single broad category "intracellular ribonucleoprotein complex" to be overrepresented (p<0.027 by chi-square analysis) (S2 File). GO categories related

PLOS GENETICS
The Ccr4-Not complex regulates TORC1 signaling to mitochondria, ribosomal synthesis, DNA replication, and components of the spindle pole body, were significantly overrepresented in the ccr4Δ upregulated gene set (Fig 5B and S3  File). These data confirm results from older microarray studies demonstrating that ccr4Δ increases expression of genes involved in metabolism, oxidative phosphorylation, and mitochondrial function [19,20].
To confirm the integrity of the RNA-seq results, we selected from the dataset two highly induced (ARO9 and FKS2) and repressed (PLB2 and TPO1) genes, and their expression was verified by reverse transcription coupled with quantitative PCR (RT-qPCR). Ccr4 loss altered the expression of these genes as predicted by the transcriptome analysis (Fig 5C-5F), thus confirming the validity of our RNA-seq analysis. Although GO categories for the vacuole or V-ATPase were not overrepresented in ccr4Δ, two V-ATPase genes, VMA6 and VMA13, were identified in the RNA-seq dataset as downregulated 2-fold by ccr4Δ (S1 File). To test if their reduced expression explained the TORC1 signaling defects, we cloned their genomic loci into high copy vectors and transformed both into WT and ccr4Δ. Simultaneous high copy expression of both V-ATPase subunits had no effect on TORC1 in WT or ccr4Δ (S2 Fig). Therefore, the ccr4Δ TORC1 signaling defects cannot be explained by reduced expression of these V-ATPase subunits.
The enhanced mitochondrial gene expression, combined with the TORC1 and V-ATPase defects described above, suggests Ccr4-Not may induce a nutrient stress state that requires a compensatory increase in mitochondrial metabolism for cellular adaptation. TORC1 repression activates retrograde signaling to transcriptionally induce, via the transcription factors Rtg1 and Rtg3, nuclear-encoded mitochondrial genes including the gene CIT2 [67,68]. We analyzed CIT2 levels by RT-qPCR and found that ccr4Δ significantly increased CIT2 expression (Fig 6A), a result confirmed in the RNA-seq dataset (S1 File.). These data suggest ccr4Δ activates retrograde signaling to increase mitochondrial activity. To determine if this was indeed the case, we quantified the relative amount of mitochondrial DNA in WT and ccr4Δ by performing qPCR with primer sets to three distinct mitochondrial genome-encoded genes (COB, ATP9, and COX1) normalized to the signal for the nuclear encoded gene SPT15. The ccr4Δ increased the relative amounts of all three mitochondrial genes, which suggests ccr4Δ increases mitochondrial content (Fig 6B). We tested this directly by staining WT and ccr4Δ with Mitotracker and then analyzing cells by confocal microscopy. WT mitochondria exhibited their characteristic tubular morphology [69], while ccr4Δ mitochondria also were tubular but they stained more intensely and appeared thicker, which further supports the observation that they have increased mitochondrial content (Fig 6C). Dihydroethidium (DHE) staining to quantify reactive oxygen species (ROS) revealed that ccr4Δ increased ROS levels, which is consistent with their increased mitochondrial content (Fig 6D). Although excessive ROS causes toxicity, moderate ROS levels have signaling roles and promote hormetic responses that facilitate stress adaptation [70]. To determine if the ROS in ccr4Δ affects the cellular response to TORC1 stress, we transformed WT and ccr4Δ with control vector or vector overexpressing superoxide dismutase 1 (SOD1), which detoxifies oxygen radicals [71]. Cells were plated onto control plates, or plates containing a low (5 nM) rapamycin concentration. While SOD1 had no effect on WT cells, SOD1 overexpression in ccr4Δ further sensitized cells to TORC1 inhibition ( Fig 6E). Therefore, the enhanced mitochondrial metabolism and ROS in ccr4Δ may contribute to a metabolic adaptive response required for coping with decreased TORC1 activity. is upregulated [72]. Relative to WT, ccr4Δ decreased HXK2 and increased HXK1 mRNA expression (Fig 6F), thus further indicating that ccr4Δ enhances mitochondrial metabolism. To test if the increased mitochondrial metabolism is due to TORC1 repression, WT and ccr4Δ carrying control or GTR1 GTP expression vectors were stained with Mitotracker. WT cells expressing GTR1 GTP drastically reduced the mitochondrial signal compared to WT control vector cells but, surprisingly, we detected no obvious differences in Mitotracker staining between the WT control, ccr4Δ control, or ccr4Δ expressing Gtr1 GTP (S3 Fig). This result is in direct contrast to the same experiment performed in WT and ccr4Δ cells grown in YPD ( Fig 6C). Although we currently cannot explain these different results, we believe they reflect differences due to the cellular nutrient environment since these experiments were performed in SC media (to select for plasmids), which is less nutrient rich than YPD. Regardless of these discrepancies, under these conditions Gtr1 GTP expression in both WT and ccr4Δ reduced mitochondrial ROS production relative to their respective controls. Importantly, GTR1 GTP expression reduced ROS in ccr4Δ to levels that resemble the WT vector control (Fig 6G). These data indicate that the increased ROS in ccr4Δ is a consequence of their reduced TORC1 signaling.

Ccr4-Not loss upregulates mitochondrial metabolism to sustain TORC1 signaling
Metabolic stress activates the Snf1/AMPK kinase that represses TORC1, and it inhibits TORC1 in part by promoting formation of Kog1/Raptor bodies [73]. To test if ccr4Δ increased Snf1/AMPK activation, WT and ccr4Δ Snf1-6XHA expressing cells were mock treated or treated with the mitochondrial ATP synthase inhibitor oligomycin before assessing TORC1 activity. Mpk1 activation and Snf1/AMPK activation (phosphorylated Snf1/AMPK) also were analyzed. As demonstrated above, TORC1 signaling was impaired in the mock-treated ccr4Δ, while Mpk1 expression also was deregulated. Importantly, basal Snf1/AMPK phosphorylation was lower in ccr4Δ, which indicates decreased Snf1/AMPK activation (Fig 7A). This reduced Snf1/AMPK activation likely is a consequence of their enhanced mitochondrial metabolism and energy production. Although the oligomycin concentration used in these experiments failed to significantly activate Snf1/AMPK or repress TORC1 in either WT or ccr4Δ, it did reduce Mpk1ph in both suggesting mitochondrial respiration contributes to Mpk1 activation ( Fig 7A). Snf1/AMPK also represses TORC1 through a separate mechanism involving the PAS kinase Psk1 that phosphorylates the stress granule factor Pbp1. Pbp1 then sequesters TORC1 into cytoplasmic stress granules, which inhibits TORC1 signaling [74]. Because a pbp1Δ ccr4Δ suppresses some ccr4Δ phenotypes [75], we tested if Ccr4-Not disruption inhibits TORC1 through Pbp1 by analyzing TORC1 in WT, ccr4Δ, pbp1Δ, and ccr4Δ pbp1Δ. The ccr4Δ pbp1Δ failed to restore TORC1 activity as would be predicted if ccr4Δ caused sequestration of TORC1 into stress granules (S4 Fig). Additionally, the double mutant had no effect on Mpk1ph, demonstrating that the deregulated Mpk1 expression in ccr4Δ is independent of the stress granule pathway (S4 Fig). Therefore, TORC1 inhibition in Ccr4-Not mutants cannot be explained by activation of either Snf1/AMPK dependent inhibitory pathways.
Mitochondria can serve as a metabolic source of amino acids since the TCA cycle intermediate α-ketoglutarate is used to synthesize glutamate via NADPH-dependent glutamate experiments with significance determined by Student's t-test. ��� -p<0.005. (C) WT and ccr4Δ were grown to mid-log phase in YPD, stained with Mitotracker, and analyzed by confocal microscopy. Scale bar indicates 5 μm. Data are representative of three independent experiments. (D) ROS levels were quantified by flow cytometry of WT and ccr4Δ DHE stained cells grown to mid-log phase in YPD. Data represent the average and SD three independent experiments with Student's t-test determining significance. ��� -p<0.005. (E) Equal numbers of overnight cultures of WT and ccr4Δ transformed with control or high copy vector overexpressing SOD1 were five-fold serially diluted and spotted to control or 5 nM rapamycin plates. Plates were incubated at 30˚C for two days. (F) RT-qPCR analysis of HXK1 and HXK2 expression in WT and ccr4Δ. Data are the average and SD of three independent experiments with significance determined by Student's t-test. � -p<0.05; �� -p<0.01. (G) ROS quantification of DHE stained mid-log phase WT and ccr4Δ cells transformed either with control or GTR1 GTP expression vector. Data represent the average and SD of three independent experiments, and significance was determined by Student's t-test. �� -p<0.01. https://doi.org/10.1371/journal.pgen.1009046.g006

PLOS GENETICS
The Ccr4-Not complex regulates TORC1 signaling dehydrogenase. This mitochondrial-derived glutamate can be combined with ammonia via glutamine synthetase to also generate glutamine that is used as a nitrogen source for amino acid synthesis [76,77]. A poorly understood glutamine-dependent, but EGO-independent, mechanism activates TORC1 in yeast, while glutamine also activates mammalian mTORC1 independently of the orthologous Ragulator complex [33,78]. We next tested if the enhanced mitochondrial metabolism in ccr4Δ compensates for defective V-ATPase activity to activate TORC1. The ccr4Δ was combined with a rtg1Δ to prevent retrograde activation, and then WT, ccr4Δ, rtg1Δ, and ccr4Δ rtg1Δ grown in YPD were stained with Mitotracker. As seen previously, ccr4Δ increased mitochondrial content compared to WT, while rtg1Δ dramatically reduced overall mitochondrial number. Importantly, the ccr4Δ rtg1Δ completely prevented the mitochondrial increase mediated by ccr4Δ (Fig 7B). ROS production also reflected these mitochondrial changes since rtg1Δ reduced ROS levels to below that of WT, while the ccr4Δ rtg1Δ completely suppressed the increased ROS due to ccr4Δ (Fig 7C). We next assessed whether retrograde activation contributed to either TORC1 signaling or Mpk1 regulation in these cells. As described above, the ccr4Δ reduced TORC1 activity and deregulated Mpk1 expression, while the rtg1Δ had no effect on TORC1 but it did reduce total Mpk1 expression ( Fig 7D). Importantly, the ccr4Δ rtg1Δ decreased TORC1 signaling to a level substantially below that of ccr4Δ without affecting Mpk1 expression (Fig 7D). These results clearly indicate that preventing mitochondrial upregulation in Ccr4-Not deficient cells suppresses TORC1 signaling further. To probe this metabolic connection in vivo, we assessed growth of WT, ccr4Δ, tor1Δ, mpk1Δ, rtg1Δ, and the ccr4Δ combined with each of these mutants on both control (YP + 2% glucose, YPD media) or YP + 3% glycerol media. Glycerol is a non-fermentable carbon source that requires mitochondrial metabolism to support cell growth [79]. Relative to WT and the individual gene deletions, each of the double mutants grows less robustly on control media suggesting they have modest synthetic sick phenotypes (Fig 7E). Importantly, on glycerol media the ccr4Δ caused a mild growth phenotype relative to WT (indicated by reduced colony size), while growth of the tor1Δ, mpk1Δ, and rtg1Δ was similar to WT. Importantly, both the ccr4Δ tor1Δ and ccr4Δ rtg1Δ, but not the ccr4Δ mpk1Δ, were unable to grow on glycerol media (Fig 7E). These data indicate that Ccr4-Not, TORC1, and the Rtg1-activated retrograde pathway have partially overlapping roles in the in vivo regulation of mitochondrial metabolism. Although Mpk1 expression is upregulated in the absence of functional Ccr4-Not, the mpk1Δ had no effect on this mitochondrial metabolic response (Fig 7E). Collectively, we propose a model where decreased Ccr4-Not activity reduces the stability and function of the V-ATPase to impair vacuole-mediated amino acid dependent TORC1 activation. Consequently, this reduction in Ccr4-Not activity causes an adaptive increase in mitochondrial metabolism that provides a vacuole-independent source of amino acids required to sustain a minimal level of TORC1 signaling for cell growth and proliferation (Fig 7F).

Discussion
Previous work has indicated that Ccr4-Not functions within the TORC1 pathway and that it also represses mitochondrial metabolism [14,16,[18][19][20]. Yet whether these processes are linked, and how Ccr4-Not contributes to their regulation, has remained elusive. In this report we provide evidence that Ccr4-Not integrates these activities by promoting vacuole V-ATPase dependent TORC1 activation. In the absence of functional Ccr4-Not, V-ATPase V 0 -V 1 interactions are less stable, which causes defective vacuole acidification and decreased TORC1 signaling. How Ccr4-Not promotes V-ATPase activity remains unclear, although our observations suggest one possibility could be that Ccr4-Not affects post-translational modification of V-ATPase subunits like Vph1. V-ATPase subunits are extensively modified, including many subunits that are ubiquitinated [62][63][64]. Although the contribution of ubiquitination to V-ATPase activity and/or stability remains mostly uncharacterized, a recent study determined that TORC1 inhibition increases Vph1 ubiquitination and turnover to promote vacuole membrane remodeling during nutrient stress adaptation [65]. While our data reveal that total Vph1 levels remain similar between WT and ccr4Δ (Fig 4E), there is a clear decrease in the highest mobility Vph1-specific band in ccr4Δ, suggesting Ccr4 loss may affect Vph1 modification. Because ccr4Δ and not4Δ both reduce TORC1 activity, TORC1 regulation involves the Ccr4-Not complex and not just the Ccr4 subunit alone. The not4Δ ubiquitin ligase mutant causes a well-established disruption in global ubiquitin homeostasis due to defective proteasome assembly and regulation [10,11]. While the ccr4Δ effect on the proteasome is not as extensive as not4Δ, our data do indicate the ccr4Δ causes a minor increase in proteasome activity, which could indicate modest proteasome dysfunction. Therefore, one explanation for our observations could be that Ccr4 has a minor role in Not4 regulation of the proteasome. Loss of Ccr4 then might affect proteasome-dependent ubiquitin recycling that reduces V-ATPase subunit ubiquitination, especially since ubiquitin levels are tightly regulated [80]. If V-ATPase ubiquitination promotes its stability and/or activity, Ccr4-Not mutants that impair ubiquitin homeostasis could decrease vacuole acidification and TORC1 activation by altering V-ATPase subunit ubiquitination.
Additionally, not4Δ exhibits synthetic sick and/or lethal phenotypes with mutations in several ubiquitin ligases that regulate the endocytic compartment [81]. These genetic data suggest Not4 may ubiquitinate endocytic-specific substrates, a definite possibility since few Not4 substrates have been identified. V-ATPase activity is critical for function of the endocytic pathway [82]. If Not4 influences V-ATPase subunit ubiquitination, then Ccr4-Not could have broader roles in the regulation of the endocytic compartment. Since Ccr4 is positioned near Not4 in the Ccr4-Not complex [83], Ccr4 might participate in substrate selection by Not4 that has a role in endocytic regulation. Such a possibility would explain why ccr4Δ impairs V-ATPase activity and inhibits TORC1. Alternatively, if the V-ATPase is co-translationally assembled, then Ccr4-Not disruption may destabilize the V-ATPase by inhibiting this process since Ccr4-Not controls co-translational assembly of other large macromolecular complexes [43,44].
Regardless of how Ccr4-Not promotes V-ATPase activity, V-ATPase disruption is a known TORC1 inhibitor [28]. Expressing Gtr1 GTP , which bypasses the V-ATPase to activate TORC1, fully restored TORC1 signaling in ccr4Δ. This result further supports our model that Ccr4-Not activates TORC1 through V-ATPase regulation. Ccr4-Not regulation of the V-ATPase also explains why some Ccr4-Not mutants have vacuole morphology defects since V-ATPase mediated vacuole acidification regulates vacuole morphology [57]. Our data do not support the alternative possibilities that Ccr4-Not loss activates known TORC1 inhibitory pathways, including those mediated by Mpk1, Rho1, or Snf1/AMPK signaling. Because the V-ATPase also activates PKA signaling through the Arf1 GTPase [28], Ccr4-Not may promote V-ATPase activity to coordinate both TORC1 and PKA signaling, a possibility that will need to be explored in future studies. If this occurs, then Ccr4-Not could be a key regulatory node for integrating nitrogen and glucose-dependent metabolic regulation.
Although Gtr1 GTP expression restored TORC1 signaling in ccr4Δ, it failed to rescue ccr4Δ growth under TORC1 inhibitory conditions. These data suggest that Ccr4-Not also acts downstream of the V-ATPase to regulate TORC1. This control may be at the level of TORC1 structure and/or TORC1 subunit availability, a possibility for which we have circumstantial evidence since ccr4Δ increases both Lst8 expression and its interaction with TORC1. Structural studies of the human mTORC1 complex reveal that mLst8 binds to the ATP-binding cleft to limit ATP binding and prevent mTORC1 from phosphorylating non-specific substrates [84]. The increased Lst8 association with Kog1 we detect in ccr4Δ might further restrict ATP binding, which could reduce TORC1 kinase activity. How Ccr4-Not disruption increases Lst8 protein levels currently is not understood, but it likely occurs post-transcriptionally since ccr4Δ does not increase LST8 mRNA expression. Intriguingly, both Lst8 and Kog1 are WD40-repeat containing proteins, and the WD40 repeats in Kog1 bind ubiquitin under stress conditions to stabilize the complex [85]. Whether yeast Lst8 binds ubiquitin is unknown, but K63-linked polyubiquitination of mammalian mLst8 does promote preferential incorporation of mLst8 into mTORC1 over mTORC2 [86]. Therefore, some precedent for ubiquitin signaling is known to drive both yeast TORC1 and mammalian mTORC1 signaling. If, as discussed above, Lst8 binding to ubiquitin affects its stability or TORC1 incorporation, then Ccr4 loss could alter Not4-dependent ubiquitin homeostasis to more directly control TORC1 signaling. More detailed future studies will be required to refine our understanding of how Ccr4-Not regulates TORC1 activity.
TORC1 inhibition in Ccr4-Not mutants is predicted to generate a nutrient stress response that activates retrograde signaling to increase mitochondrial metabolism [67,68], a result we demonstrate does occur. Retrograde activation in Ccr4-Not mutants likely functions as an adaptive mechanism to compensate for reduced V-ATPase dependent TORC1 activation since blocking retrograde signaling further reduces TORC1 activity. We believe Ccr4-Not mutants enhance mitochondrial metabolism to sustain a minimal level of TORC1 signaling required for growth and proliferation. Such a mechanism most likely involves diverting some of the mitochondrial α-ketoglutarate into glutamate and glutamine biosynthetic pathways. These vacuole-independent amino acid pools then could activate TORC1 independently of the V-ATPase, or perhaps enhance the ability of the impaired V-ATPase to activate TORC1. Evidence exists for vacuole EGO-independent TORC1 activation mechanisms in yeast [33], and an analogous process also occurs in mammals [78]. This mechanism also would be consistent with the increased sensitivity ccr4Δ has to TORC1 inhibition when the SOD1 superoxide dismutase is overexpressed. Sod1 enzymatically stabilizes the casein kinase I-gamma (CK1γ) isoforms Yck1 and Yck2 (Yck1/2) to promote glucose and amino acid metabolism, and CK1γ also represses mitochondrial respiration [87]. Increasing Sod1 levels in ccr4Δ likely enhances CK1γ signaling to repress the mitochondrial upregulation required for ccr4Δ to bypass V-ATPase dependent TORC1 activation. The increased ROS in ccr4Δ also may allow these cells to adapt to TORC1 stress through other means as well, since ROS increases stress resistance mechanisms [70].
Although Ccr4-Not clearly represses Mpk1ph and total Mpk1 levels, the relative level of Mpk1 activation remains only modestly affected in Ccr4-Not mutants. Most importantly, we confirmed this Mpk1 deregulation does not repress TORC1 since the combined loss of both Ccr4 and Mpk1 failed to restore TORC1 signaling. Ccr4-Not disruption likely deregulates Mpk1 in part by preventing post-transcriptional turnover of MPK1 mRNA since its steadystate levels increase in ccr4Δ, and Ccr4 binds MPK1 transcripts to promote their degradation [22]. However, Ccr4-Not likely regulates Mpk1 through additional mechanisms that may include nutrient signaling since Ccr4-Not disruption deregulates Mpk1 expression in nutrient rich, but not nutrient defined, media. The mechanisms underlying this difference, and their impact on Mpk1 regulated stress pathways will need to be delineated in future studies.
Collectively, we provide evidence that Ccr4-Not stimulates TORC1 signaling through the vacuole V-ATPase that activates TORC1 in response to amino acid sufficiency. Upon Ccr4-Not inactivation, cells upregulate mitochondrial metabolism in part to compensate for this defect in TORC1 activity that then sustains a minimum level of TORC1 signaling to promote cell growth and proliferation. The V-ATPase activates mTORC1 in mammalian cells [31], and Ccr4-Not has poorly understood roles in mammalian cellular metabolism [88][89][90]. Our studies suggest the possibility that these mechanisms could be conserved in human cells to allow Ccr4-Not to activate mTORC1. Such a possibility could explain why Ccr4-Not has such critical roles in cell growth and proliferation, and why its disruption plays an important role in a wide variety of diseases and developmental disorders.

Yeast strains and culture conditions
Yeast strains are derivatives of the BY4741 background and are listed in S1 Table. All yeast growth media components were purchased from US Biologicals and Research Products International. Cells were grown to mid-log phase (OD 600 = 0.8-1.2) for all experiments. Experiments performed in nutrient rich media utilized YPD (1% yeast extract, 2% peptone, 2% dextrose), while those experiments where selection for plasmid maintenance was required were performed in synthetic complete (SC) media (0.2% yeast nitrogen base, 0.5% ammonium sulfate, 2% dextrose, 0.19% amino acid dropout mix). To make the appropriate SC selective media, all nutrients lacking in the dropout mix were added back except the nutrient that allowed for plasmid selection. Yeast genetic manipulations to generate integrated epitope tags or gene deletions were performed following standard procedures as previously described [91].

Cloning
All plasmids and PCR primers used in this study are listed in S2 Table and S3 Table. Plasmids overexpressing SAC7, VMA6, and VMA13 were generated by amplifying 300 base pairs upstream of their translational start site and 100 base pairs downstream of the translational stop using Q5 high fidelity DNA polymerase (New England Biolabs). PCR fragments were then cloned into high copy vectors pRS425 or pRS426. The Gtr1GTP expression vector was generated by amplifying Gtr1 Q65L as a C-terminal mono-HA tag fusion from plasmid pMB1394 [27], and then the PCR fragment was cloned into plasmid p416ADH [92]. The Ccr4 and ccr4-1 Cterminal mono-FLAG expression plasmids were generated using Q5 enzyme with plasmid templates pADHCCR4 and pADHccr4-1, and the resulting PCR fragments then cloned into p415ADH [92]. All plasmids were sequenced to confirm cloned products were correct.

Whole cell extract preparation and immunoprecipitations
Yeast total cell extracts were prepared as described previously [93]. Briefly, cells were harvested by centrifugation, washed once with ddH 2 0, and then the pellets frozen at -80˚C. For standard SDS-PAGE analysis, cell pellets were lysed by bead beating in lysis buffer (300 mM NaCl, 10% glycerol, 10 mM Tris pH 8.0, and 0.1% NP-40) containing protease and phosphates inhibitors and 1 mM DTT. All standard immunoblot analyses were performed using 30 μg total cell extract. For the immunoprecipitation experiments, extracts were prepared in the same buffer but containing 150 mM NaCl instead. Lysed extracts were clarified by centrifugation at 4˚C for 15 minutes, and then the supernatants were quantified by Bradford assay. Immunoprecipitations were performed using 750 μg-1 mg total cell extracts for four hours to overnight incubation at 4˚C with rotation. Immune complexes were isolated with Protein A-conjugated agarose beads, washed extensively with extraction buffer, and then resuspended in 2X SDSloading buffer and resolved by SDS-PAGE. For input samples, 30 μg cell extracts were analyzed. Immunoblots were quantified using ImageJ. All experiments were independently repeated a minimum of three or more times.

Cell staining and confocal microscopy
For analysis of ROS, cells were cultured overnight to OD 600 = 0.8 and 1 mL of culture was harvested, centrifuged to remove the media, and then resuspended in 200 μL of PBS containing 20 μM DHE for 15 minutes at 30˚C. Samples then were analyzed on a BD Accuri C6 flow cytometer. Vacuole staining was performed by culturing cells overnight to OD 600 = 0.8, and then removing 1 mL of culture and adding FM 4-64 to 1 mg/mL for 15-30 minutes before analyzing by confocal microscopy. To detect mitochondria, cells were cultured and harvested as described immediately above, but then they were resuspended in 10mM HEPES buffer (pH 7.4) with 5% glucose. Mitotracker Red FM was added to a final concentration of 100 nM and cells were incubated in a shaker for 15-30 minutes at 30˚C. Stained cells then were visualized on a Zeiss confocal microscope using a 63X oil objective with Z-stacks (4-5) taken for each image. To analyze vacuole pH, overnight cell cultures at OD 600 = 0.8 were harvested, media was removed, and cells washed in ddH 2 0. Cells then were resuspended in YPD containing 50 mM citric acid adjusted to pH 3.5 and 6-CFDA was added to a final concentration of 20 μm for 30 minutes in a 30˚C shaker. Cells were cooled on ice before centrifuging and removing the media. The cell pellet was washed twice using pre-cooled YPD, and then each sample was resuspended to the same OD 600 in a 96 well plate. The plate was analyzed on a Molecular Device plate reader using excitations at 444 nm and 485 nm with emissions captured at 538 nm.

RNA sequencing and bioinformatics analysis
Total RNA was isolated by hot phenol extraction from mid-log phase WT and ccr4Δ cultures grown in YPD in triplicate, and RNAs then were treated with DNase I to get rid of contaminating genomic DNA. The purified RNA samples were submitted to the University of Tennessee Molecular Resource Center where they were analyzed on an Agilent Bioanalyzer to assess the RNA quality with RIN numbers meeting a minimum �8.0. After passing this initial screening, 250 ng of total RNA was used to prepare libraries for sequencing using the Qiagen Qiaseq Fas-tSelect rRNA depletion kit to eliminate rRNA. The libraries were then prepared using the NEB Ultra II directional RNA-seq kit for Illumina and then amplified for 15 cycles as the final step of library preparation. Libraries were quantified using data from the Agilent Bioanalyzer and a Qbit fluorometer, and then they were pooled in equimolar amounts. Following pooling, the library pools were sized to a target range of 300 bp-700 bp using AMPpure beads. The sized libraries were examined on an Agilent High Sensitivity DNA chip, quantified using a Q-bit fluorometer, and used for sequencing on the Illumina NextSeq platform. All fastq files then were gathered from the sequencer, and quality assurance was performed using FASTQC. Reads were trimmed to remove any nucleotide with a PHRED score < Q20, and the trimmed FASTQ files were aligned to the yeast coding fasta reference library using RNA STAR. Once aligned, the SAM files were collected and mined for the read count information of each gene present in the reference file. Read counts were normalized using the Counts per Million (CPM) method across the entire experiment. Principle component analysis and Pearson's coefficient plots were performed on the normalized transcriptome profile. A Wilcoxon's t test was used to determine significance between conditions, and all genes that failed to yield a pvalue greater than 0.05 were removed. Benjamini and Hochberg false discovery rate was performed on the trimmed gene list, and all genes failing to yield a false discovery rate of less than 0.05 were discarded. The final significant differential gene list was loaded into R to generate heatmaps. The gene targets were loaded into GO term finder (version 0.86) for biological function and process categorical analysis. The summary statistics for all graphical data are presented in S4 File.