Pbp1 stabilizes and promotes the translation of Puf3-target mRNAs involved in mitochondrial biogenesis

Pbp1 (poly(A)-binding protein - binding protein 1) is a cytoplasmic stress granule marker that is capable of forming condensates that function in the negative regulation of TORC1 signaling under respiratory conditions. How mutations in its mammalian ortholog ataxin-2 are linked to neurodegenerative conditions remains unclear. Here, we show that loss of Pbp1 leads to decreases in amounts of mitochondrial proteins whose encoding mRNAs are targets of Puf3, a member of the PUF (Pumilio and FBF) family of RNA-binding proteins. We found that Pbp1 stabilizes and promotes the translation of Puf3-target mRNAs in respiratory conditions, such as those involved in the assembly of cytochrome c oxidase. We further show that Pbp1 and Puf3 interact through their respective low complexity domains, which is required for Puf3-target mRNA stabilization and translation. Our findings reveal a key role for Pbp1-containing assemblies in enabling the translation of mRNAs critical for mitochondrial biogenesis and respiration. They may further begin to explain prior associations of Pbp1/ataxin-2 with RNA, stress granule biology, mitochondrial function, and neuronal health.


17
Yeast cells are capable of rapidly adapting their metabolism to changes in 18 environmental conditions. When grown in glucose media, yeast cells use glycolysis for 19 energy production and suppress mitochondrial biogenesis. However, in the presence of 20 a non-fermentable carbon source, such as lactate, yeast cells adapt by inducing 21 mitochondrial biogenesis to increase ATP production by oxidative phosphorylation 22 (OXPHOS). The majority of the protein components of the electron transport chain are 23 nuclear-encoded and need to be imported into the mitochondria. This essential process 24 also requires cytosolic protein participants and is tightly regulated according to the cell's 25 metabolic needs (1). 26

27
Puf3 is one such cytosolic protein that plays a key role in the post-transcriptional 28 regulation of mitochondrial biogenesis. Puf3 is a member of the PUF protein family, which 29 are conserved RNA-binding proteins that recognize a consensus motif in the 3' UTR of 30 their target mRNAs (2). Most of the target mRNAs associated with Puf3 are important for 31 OXPHOS, many of which encode for subunits of mitochondrial ribosomes or respiratory 32 complexes (3-8). Depending on glucose availability, Puf3 can switch the fate of its target 33 mRNAs from decay to translation (9). In the presence of non-fermentable sugars that 34 require respiratory metabolism, Puf3 becomes heavily phosphorylated in its N-terminal 35 low complexity region, leading to the stabilization and translation of its bound transcripts. 36 Furthermore, Puf3 has been found in ribonucleoprotein (RNP) granules (9). 37 38 Pbp1 can lead to mitochondrial dysfunction, increased sensitivity to oxidative stress, and 48 cell death (14,15). These findings collectively suggest that Pbp1 plays an important role 49 in promoting mitochondrial function. 50

51
Here we utilized S. cerevisiae to discover a functional relationship between Pbp1, 52 mitochondria, and Puf3. We find that Pbp1 and Puf3 interact through their low-complexity 53 domains and that this interaction facilitates the translation of mRNAs that are targets of 54 Puf3 under respiratory conditions. These findings reveal that Pbp1 supports mitochondrial 55 function via assemblies involved in mRNA translation, which may provide key insights 56 into how ataxin-2 mutations cause neurodegenerative diseases. 57

58
To interrogate a possible role for Pbp1 in mitochondrial function, we assayed 59 mitochondrial protein abundance in response to switching from glycolytic (YPD) to 60 respiratory (YPL) media, conditions that demand mitochondrial biogenesis, in wild type 61 (WT) and pbp1∆ cells. We assessed Por1 (mitochondrial porin) and Cox2 (subunit II of 62 cytochrome c oxidase) protein levels using readily available antibodies by immunoblot in 63 YPD and at several time points after switching to YPL ( Figure 1A). Por1 protein levels 64 increased over time during growth in respiratory conditions in both WT and pbp1∆ cells. 65 Strikingly, Cox2 protein levels were severely decreased in pbp1∆ compared to WT cells 66 at all time points. Moreover, pbp1∆ cells showed a significantly reduced growth rate in 67 YPL compared to WT cells ( Figure 1B), but their growth rate in YPD was normal. These 68 observations are consistent with pbp1∆ cells having compromised mitochondrial function 69 due to reduced amounts of proteins required for respiration, such as Cox2. 70 71 As Pbp1 interacts with Pab1 (poly(A)-binding protein 1) and contains putative 72 RNA-binding domains (10), we tested the possibility that the loss of Pbp1 might affect the 73 abundance of mRNAs involved in mitochondrial function. Analysis of a panel of nuclear-74 and mitochondrial-encoded mRNAs revealed that COX17, COX10, and MRP51 75 transcripts were decreased in pbp1∆ cells ( Figure 1C). By contrast, no significant 76 differences were observed in POR1 and ATP2 transcripts. Notably, the transcripts 77 exhibiting reduced abundance in pbp1∆ mutant cells (COX17, COX10, MRP51) all have 78 Puf3-binding elements in their 3' UTRs and are Puf3-target mRNAs (6,9). For 79 mitochondrial-encoded transcripts, we observed a modest decrease in COX2 transcripts 80 in pbp1∆ cells at later time points in YPL and slight differences in COX1. The translation 81 of COX2 relies on the mitochondrial ribosome, most subunits of which are post-82 transcriptional mRNA targets of Puf3 (e.g., MRP51). Therefore, defects in the translation 83 of mitochondrial-encoded genes, such as Cox2, indicate disruption of Puf3 function (9). 84

85
The specific decrease in mRNA levels of Puf3-target mRNAs in pbp1∆ cells 86 suggests that Pbp1 may help Puf3 stabilize and promote the translation of its target 87 mRNAs. To test this possibility, we examined the extent of translation of a Puf3-binding 88 element-containing mRNA following the shutoff of its transcription. The reporter consists 89 of a mitochondrial targeting sequence (MTS), the GFP coding sequence, and the 3' UTR 90 of the mitochondrial ribosomal gene MRP51, which contains a Puf3-binding element 91 (P3E) (Figure 2A) (9). The construct was integrated into a strain expressing a chimeric 92 GAL4DBD-ER-VP16 transcription factor enabling inducible expression by the addition of 93 ß-estradiol (16). This reporter allows monitoring of GFP transcript and protein amounts 94 following a pulse of expression in the background of WT or pbp1∆ cells. Moreover, 95 including a mutant reporter in which the P3E contains a four-nucleotide mutation allows 96 the assessment of reporter translation dependent on an intact P3E. 97 98 WT and pbp1∆ cells were grown in YPD media, and ß-estradiol was added for 45 99 min before washout and then switched to YPL respiratory media. We observed the 100 highest reporter mRNA expression at 30 min, followed by an expected, gradual decrease 101 in mRNA due to the absence of any inducer ( Figure 2B). Strikingly, levels of the reporter 102 mRNA were significantly lower in pbp1∆ cells, strongly suggesting that Pbp1 is required 103 to stabilize Puf3-target mRNAs. Interestingly, mRNA expression levels of the mutant P3E 104 reporter were similar in WT and pbp1∆ cells, indicating that the reduced reporter 105 expression in pbp1∆ cells is mediated by Puf3 and an intact P3E sequence in its target 106 mRNAs. We next tested whether there might be a genetic interaction between Pbp1 and 120 Puf3. We generated puf3∆, pbp1∆, and pbp1∆puf3∆ cells and examined different 121 mitochondrial protein levels by immunoblot following growth in YPD and YPL media 122 ( Figure 3A). As expected, Cox2, Por1, and Atp2 (subunit of the mitochondrial ATP 123 synthase) protein levels all increased upon switching to respiratory conditions. However, 124 Cox2 protein levels, which are dependent on the translation of Puf3-target mRNAs, 125 decreased in pbp1∆, puf3∆, and pbp1∆puf3∆ cells compared to WT cells. Por1 and Atp2 126 protein levels, which do not depend on Puf3-targets, were similar amongst the tested 127 strains. Interestingly, the double knockout pbp1∆puf3∆ exhibited a similar reduction in 128 Cox2 as puf3∆ cells, suggesting that Puf3 is epistatic to Pbp1. In addition, we confirmed 129 that Pbp1 levels were not affected by the absence of Puf3, whereas Puf3 levels were only 130 modestly decreased in the absence of Pbp1 but only in lactate media ( Figure S1). Taken 131 together, these data suggest that Pbp1 and Puf3 function together in the same pathway 132 and are consistent with the hypothesis that Pbp1 regulates Puf3. Notably, a decrease in 133 each of the surveyed mitochondrial protein levels can also be observed in pbp1∆ cells in 134 YPD, which is suggestive of a general impairment in mitochondria even in glucose 135 conditions. However, for this study, we focus on the role of Pbp1 in respiratory conditions 136 due to the exacerbated phenotype of reduced Cox2 levels and Puf3-target mRNA 137 destabilization, specifically observed in YPL media. 138

139
Having observed genetic interactions and phenotypic consequences on Puf3-140 targets in pbp1∆ mutant cells, we next investigated whether Pbp1 and Puf3 interact by 141 co-immunoprecipitation analysis using C-terminal, epitope-tagged versions of Pbp1 142 (Pbp1-FLAG) and Puf3 (Puf3-HA) ( Figure 3B). Pbp1-Flag co-IP experiments revealed 143 that Pbp1 was indeed able to pull down Puf3. In addition, we performed the reciprocal 144 Flag co-IP and obtained corresponding results, demonstrating that Pbp1 and Puf3 145 interact. We observed that Puf3 protein levels increase over time in response to the switch 146 to respiratory conditions ( Figure 3B and S1). Interestingly, we observed an increased 147 interaction between Pbp1 and Puf3 over time in YPL media ( Figure 3B). This suggests 148 that the Pbp1-Puf3 interaction is responsive to the cellular metabolic state. These Previous studies have shown that Pbp1 is a negative regulator of TORC1 (14). We 153 next tested whether the decreased Cox2 protein amounts in pbp1∆ cells might be due to 154 increased TORC1 activity by treating cells with rapamycin. No difference in Cox2 levels 155 was observed in pbp1∆ cells following rapamycin addition, indicating that the Cox2 protein 156 reduction in pbp1D cells is independent of TORC1 signaling and instead likely due to its 157 interaction with Puf3 ( Figure S2). 158

159
To determine which region within Pbp1 is required for binding Puf3, we engineered 160 a series of Pbp1 truncations ( Figure 4A). Using co-immunoprecipitation, we observed that 161 deletion of the mid-section (aa 299-570) or the low-complexity domain (aa 570-722) 162 largely abolished the ability of Pbp1 to pull down Puf3 ( Figure 4B). We then asked if the 163 Pbp1-Puf3 interaction facilitates the translation of Puf3-target mRNAs by assessing Cox2 164 and Por1 protein levels in these mutants. We observed reduced Cox2, but not Por1, 165 protein levels in Pbp1 mid∆ and Pbp1 LCD∆ cells, while no differences were observed in 166 cells expressing the other Pbp1 truncation constructs ( Figure 4C). Consistent with these 167 effects on protein levels, qRT-PCR analysis showed reduced mRNA levels of Puf3-target 168 transcripts COX17 and COX10, but not POR1 or ATP2, in the corresponding truncation 169 mutants ( Figure 4D). 170 171 We previously found that Pbp1 self-associates through its low complexity domain, 172 which is mediated by key methionine residues in the LCD (14,15). Replacement of eight 173 methionine residues in the LCD by serine (M8S) leads to weaker self-assembly, 174 increased TORC1 signaling, and reduced autophagy. In contrast, the M8F and M8Y 175 mutants show stronger self-assembly and slightly increased autophagy (14,15). 176 Therefore, we asked if these key methionine residues in the LCD responsible for the 177 ability of Pbp1 to form condensates might also alter its interaction with Puf3 ( Figures 5A  178 and 5B). Interestingly, the M8S mutant showed reduced interaction with Puf3, in line with 179 its compromised ability to phase separate and form assemblies. In contrast, the M8F and 180 M8Y mutants showed increased interaction with Puf3, which is consistent with their ability 181 to form stronger assemblies. 182 183 Furthermore, Cox2 protein expression in these mutants, but not Por1, was affected 184 in a manner that correlated with their ability to interact with Puf3. Decreased Cox2 protein 185 levels were observed in the M8S mutant ( Figure 5C), whereas increased Cox2 protein 186 levels were observed in the M8F and M8Y mutants, compared to WT. These data suggest 187 a strong correlation between the ability of Pbp1 to self-associate, the extent of its 188 interaction with Puf3, and resulting amounts of Cox2 protein.

190
To further characterize the interaction between Pbp1 and Puf3, we used different 191 truncation mutants of Puf3 ( Figure 6A). Puf3 contains a PUF domain that binds to the 3' 192 UTR of mRNAs encoding mitochondrial proteins as well as a low-complexity N-terminal 193 (Nt) region (7,9). It also contains a stretch of glutamine residues (polyQ) amid the protein. Puf3 PUF∆ and Puf3 Nt∆ proteins were expressed at lower amounts than WT Puf3 or 195 Puf3 polyQD proteins ( Figures 6B, C). Deleting the Nt region abolished the ability of Puf3 196 to pull down Pbp1. The Puf3 PUF∆ mutant also showed reduced interaction with Pbp1. 197 In addition, Cox2 and Por1 protein levels were assayed in these mutants ( Figure 6C). We In this study, we show that Pbp1 supports mitochondrial function by promoting the 210 stabilization and translation of Puf3-target mRNAs that are involved in mitochondrial 211 biogenesis. Both Pbp1 and Puf3 harbor low complexity domains, and their association is 212 required for normal Puf3 function. We speculate that Pbp1 self-associates under 213 respiratory conditions to recruit Puf3 to the vicinity of mitochondria, where Puf3 promotes 214 the translation of mRNAs central for mitochondrial biogenesis. Consistent with this 215 hypothesis, we observed that Pbp1's capacity to self-assemble correlated with its 216 interaction with Puf3 and its ability to boost Cox2 protein amounts. Interestingly, Pbp1 217 has been suggested to function as a redox sensor (15). Reactive oxygen species can 218 accumulate during mitochondrial dysfunction, underlining the potential for a key sensor  A. A protein (Cox2) whose synthesis is dependent on Puf3-target mRNAs is 267 decreased in pbp1∆ cells compared to WT. Cells were grown to log phase in 268 glucose (YPD) media and then washed and resuspended in lactate (YPL) media. 269 Samples were collected before and after switch to YPL, quenched, followed by 270 protein extraction.  analysis of reporter mRNA levels by qRT-PCR for GFP and normalized to ACT1. 300 The increase in mRNA levels observed after medium switch is due to an 301 unavoidable washing step to remove b-estradiol. Note that Puf3 is not absolutely 302 required for the translation of its target mRNAs. Error bars represent SD; n=3 and 303 a two-tailed two-sample t-test, assuming equal variance, was used to determine 304 statistical significance. p>0.05 (n.s.), *p<0.05, **p<0.01.

305
C. Pbp1 promotes the translation of the reporter containing an intact P3E. Reporter 306 translation was assayed using an anti-GFP antibody, and G6pdh was used as a 307 loading control. * denotes the slightly larger, unprocessed form of GFP containing 308 a MTS. 309 310 Figure 3. Increased amounts of Puf3 interact with Pbp1 under respiratory 311 conditions 312 313 A. Only proteins whose synthesis is dependent on Puf3-target mRNAs (Cox2) are 314 decreased in puf3∆, pbp1∆, and pbp1∆puf3∆ cells as compared to WT. Cells were 315 grown to log phase in YPD, then washed and resuspended in YPL. Samples were 316 collected before (0 h) and after switch to YPL (3 h), quenched, followed by protein 317 extraction. Equal amounts of protein were assayed by immunoblot for Cox2, Por1, 318 Atp2, Rpn10 and G6pdh levels. A. Schematic representation of Pbp1 and the deletion mutants used.

329
B. Interaction between Puf3 and Pbp1 lacking its mid-section (mid∆) or C-terminal 330 LCD (LCD∆) is decreased in respiratory conditions. Cells with epitope-tagged 331 Pbp1 and Puf3 were grown in YPD and then washed and resuspended in YPL.

332
Samples were collected at the indicated time points, after which Flag 333 immunoprecipitation was performed. -denotes negative control lacking a Flag-334 tag, but including HA-tag. * denotes non-specific band.

335
C. Cox2 protein levels are decreased in Pbp1 mid∆, LCD∆, and pbp1∆ in respiratory 336 conditions. Cells were grown to log phase in YPD, then washed and resuspended 337 in YPL. Samples were collected after 3 h in YPL, quenched, followed by protein 338 extraction. Equal amounts of protein were assessed by immunoblot for Cox2, Por1, 339 Atp2, Pbp1, Rpn10, and G6pdh levels.

Yeast strains, growth, and media 385
The prototrophic Saccharomyces cerevisiae CEN.PK strain (30) was used in all 386 experiments. All strains used in this study are listed in Table S1. Gene deletions were 387 performed using standard PCR-based strategies to amplify resistance cassettes with 388 appropriate flanking sequences and replace the target gene through homologous 389 recombination (31). C-terminal tags were similarly made using PCR to amplify resistance  Table S2.    quenched, and extracted for immunoblot analysis of the indicated proteins. Note that 505 rapamycin treatment did not restore protein amounts of Cox2 in pbp1∆ cells. 506 507 Supplemental Figure 3. 508 509 Several mRNAs involved in mitochondrial function were assayed by qRT-PCR in the 510 indicated Puf3 mutant strains at the indicated time points after switch from YPD to YPL. 511 The abundance of the indicated transcripts was normalized to ACT1.