Translational Regulation of Pmt1 and Pmt2 by Bfr1 Affects Unfolded Protein O-Mannosylation
Abstract: O-mannosylation is implicated in protein quality control in Saccharomyces cerevisiae due to the attachment of mannose to serine and threonine residues of un- or misfolded proteins in the endoplasmic reticulum (ER). This process also designated as unfolded protein O-mannosylation (UPOM) that ends futile folding cycles and saves cellular resources is mainly mediated by protein O-mannosyltransferases Pmt1 and Pmt2. Here we describe a genetic screen for factors that influence O-mannosylation in yeast, using slow-folding green fluorescent protein (GFP) as a reporter. Our screening identifies the RNA binding protein brefeldin A resistance factor 1 (Bfr1) that has not been linked to O-mannosylation and ER protein quality control before. We find that Bfr1 affects O-mannosylation through changes in Pmt1 and Pmt2 protein abundance but has no effect on PMT1 and PMT2 transcript levels, mRNA localization to the ER membrane or protein stability. Ribosome profiling reveals that Bfr1 is a crucial factor for Pmt1 and Pmt2 translation thereby affecting unfolded protein O-mannosylation. Our results uncover a new level of regulation of protein quality control in the secretory pathway.
1.Introduction
Glycosylation is a major protein modification that includes the addition of a sugar moiety onto a protein [1]. Two types of glycosylation conserved from fungi to humans are N-glycosylation and O-mannosylation [2]. Both essential types of glycosylation start in the endoplasmic reticulum (ER) and share the common mannose donor Dol-P-mannose (Dol-P-Man). O-mannosylation entails the direct transfer of mannose from Dol-P-Man to serine and threonine residues of proteins entering the secretory pathway (herein referred to as secretory proteins) by different types of protein O-mannosyltransferase enzymes. Among those, only the protein O-mannosyltransferase (PMT) family is conserved among eukaryotes [3–7]. Changes in PMT-based O-mannosylation in humans result in genetic disorders calledα-dystroglycanopathies [8] and are also associated with various cancers [9,10]. In the baker´s yeast,S. cerevisiae, (from here on termed simply yeast) O-mannosylation in the ER depends on PMTs only,making it an ideal model to study this crucial protein modification.PMTs are ER membrane glycoproteins that have been shown to associate with the translocon to modify translocating polypeptides [11]. In yeast the redundant PMT family contains seven members, for six of which the O-mannosyltransferase activity has been proven. They are subdivided into three subfamilies referred to as PMT1 (Pmt1, Pmt5), PMT2 (Pmt2, Pmt3, Pmt6), and PMT4 (Pmt4) that show distinct substrate specificities [12]. Pmt1-Pmt2 heterodimers contribute a major part of O-mannosyltransferase activity [13].
Analysis of the yeast O-mannose glycoproteome revealed that around 20% of all ER and Golgi proteins are O-mannosylated, many of those with crucial functions in protein glycosylation, folding, quality control, and trafficking [14]. Hence it is not surprising that transcription of PMTs is enhanced under ER stress conditions [15] and general PMT inhibition induces the unfolded protein response (UPR) [16], a transcriptional response that regulates protein folding capacities of the ER and degradative processes termed ER associated degradation (ERAD) of un- or misfolded proteins [17].While most studies of O-mannosylation focus on the role of this modification during normal protein maturation along the secretory pathway, recently it has been demonstrated that there exists non-canonical O-mannosylation of proteins due to un- or misfolding [18]. This so-called unfolded protein O-mannosylation (UPOM) has been proposed as a molecular timer that is active in the early stages of ER protein quality control to abrogate futile folding cycles and save valuable cellular resources [19]. The consequences of UPOM strongly depend on the substrate proteins which have been shown to be later eliminated by the cell either by ERAD [20], vacuolar degradation [21] or cellular exclusion [22]. To date this modification has been observed for several mutated proteins, however, not for their wild type counterparts [20–25]. Central regulators of UPOM have been shown to be Pmt1 and Pmt2 but not Pmt4 [19,26]. The most prominent UPOM substrate to date is slow-folding green fluorescent protein (GFP) that folds properly in the cytosol, but when targeted to the ER is recognized as a misfolded protein due to its slow folding and therefore gets O-mannosylated [19]. O-mannosylation itself then blocks further folding of the fluorophore, resulting in decreased fluorescence intensity and rendering this protein an adequate reporter to monitor UPOM efficiency.With the exception of Pmt1 and Pmt2 that mediate UPOM this protein quality control system is poorly defined. In the present study we screened for cellular factors that affect UPOM in yeast. To this end we took advantage of ER-targeted slow-folding GFP as a UPOM-reporter and identified brefeldin A resistance factor 1 (Bfr1) as an enhancer of Pmt1 and Pmt2 translation.
2.Results
To perform a genome-wide screen for identification of cellular factors affecting UPOM we took advantage of the model UPOM substrate, slow-folding ER-GFP [19]. We stably introduced ER-GFP into the innocuous HO locus of pmt1∆, pmt2∆, and pmt4∆ cells. ER targeting of GFP was ensured by an N-terminal Kar2 signal peptide and ER retention by a C-terminal HDEL retention signal (Figure 1A, upper scheme). A fast folding variant of GFP (ER-GFPf) that escapes O-mannosylation and therefore changes in folding and fluorescence served as a negative control [27]. As shown in Figure 1B, ER-GFP shows reduced fluorescence compared to ER-GFPf expressed in wild type cells. In pmt1∆ and pmt2∆ cells reporter fluorescence is considerably enhanced compared to wild type whereas the GFP signal in pmt4∆ is not affected (Figure 1B,C). These results are in line with previously published data in which ER-GFP is expressed from a centromeric plasmid [19]. O-mannosylation of ER-GFP in wild type and PMT deletion mutants was monitored by probing lysates of respective cells for GFP (Figure 1D). ER-GFP detection results in a main GFP signal accompanied by multiple higher molecular weight bands that are not seen in case of ER-GFPf (Figure 1D, compare area designated by the white arrowin lanes 2 and 3). The same GFP pattern is detected in PMT4 deficient (lane 6) but not PMT1 and PMT2 deficient cells (lanes 4 and 5) and correlates with O-mannosylation of ER-GFP. Treatment of immunopurified FLAG-tagged ER-GFP (Figure 1A, lower scheme) with α1-2,3,6 mannosidase that removes O-linked α-mannose [28] confirmed that the signal above the main GFP band emanates from O-mannosyl glycans (Figure 1E). We further examined whether ER-GFP expression that is driven by the strong TDH3 promotor induces ER stress resulting in UPR induction (Figure 1F).
In contrast to ER-GFPf, expression of ER-GFP triggers the UPR as indicated by the significant increase of mRNA levels of the spliced (active) variant (Figure 1F, HAC1s) of the UPR-inducing transcription factor Hac1 and the UPR-targeted Hsp70 chaperone Kar2. This suggests that at least in the case of GFP, slow folding rates rather than protein overexpression constitute the biggest challenge for the ER.As depicted in Figure 2A, the ER-GFP expressing wild type strain was crossed with libraries containing viable deletion strains of non-essential genes and hypomorphic mutants of essential ones, to create new libraries in which each haploid strain expresses the ER-GFP on the background of one mutant allele. The median fluorescence intensities (MFIs) of all viable strains resulting upon crossing are shown in Figure 2B (small diagram on the right) and a detailed listing of all identified targets is available in Supplementary Table S1. Analysis of ER-GFP median intensity frequency distribution for more than 5000 viable mutant strains revealed that approximately 5% displayed fluorescence exceeding the MFI range of ER-GFP in wild type cells (Figure 2B, zoomed in area and green bars in bar diagram). A total of 109 genes exceeded the threshold (median GFP intensity at 187, red dotted line in Figure 2B) and were considered as positive hits (supplementary Table S1). Validity of the screen was confirmed by the presence of PMT1 (position 38) and PMT2 (position 3) among the positive candidates. Further analysis of screening hits was performed by manual assessment of GFP signal localization to the ER. Out of 109 candidates, only spf1∆ cells showed predominant cytosolic GFP fluorescence further confirmed in an independent spf1∆ mutant by fluorescence microscopy (Supplementary Figure S1A).
SPF1 encodes an ER membrane P-type ATPase important for maintenance of Ca2+ homeostasis and normal lipid composition of intracellular membranes [29,30]. Among the residual 108 candidates, stress pathway components (e.g., oca1∆ and oca2∆ involved in oxidative stress response; sln1∆, ptc1∆ and sic1∆ encoding for functional components of the high osmolarity glycerol (HOG) pathway) and components of N-glycosylation and quality control (e.g., ost3∆ (Supplementary Figure S2) and cwh41∆) were present. Analysis of the O-mannosylation status of the canonical Pmt1-Pmt2 client Hsp150 revealed that the vast majority of the mutants do not severely affect O-mannosylation in general, judging by the prevalence of the molecular mass of Hsp150 upon the gene deletions. However, in a substantial number of mutants, we observed the presence of subspecies of Hsp150 that likely result from general maturation defects (Supplementary Table S1). Among those are for example ost3∆ and pop2∆ (Supplementary Figure S1B) that affect N-glycosylation and mRNA catabolism, respectively, and for which general defects in protein homeostasis have been reported previously [31,32]. Since we were especially interested in candidates that directly affect glycosylation of the UPOM-reporter, systematic analysis of candidate genes was performed by determining ER-GFP O-mannosylation by Western blot. This revealed that for most of the tested mutant strains increased GFP fluorescence did not correlate with significantly reduced O-mannosylation (Supplementary Table S1). Next to pmt1∆ and pmt2∆, only two additional mutants were found to abrogate ER-GFP O-mannosylation: bfr1∆ (Figure 2C,D) and psa1DAmP (Supplementary Figure S3A). PSA1 is an essential gene encoding for the enzyme GDP-mannose pyrophosphorylase that is responsible for the synthesis of GDP-mannose, the mannose donor in Dol-P-Man synthesis [33] (Supplementary Figure S3B). Since decreased expression of Psa1 in the psa1DAmP most likely limits availability of the mannose donor Dol-P-Man thereby affecting PMT activity, we decided to herein focus on BFR1 whose role remains unknown. Enhanced ER-GFP fluorescence upon BFR1 deletion was confirmed via flow cytometry in several independent mutants (Supplementary Figure S4).
BFR1 was identified in a genetic screen as a multicopy suppressor of brefeldin A induced lethality in yeast [37]. It is associated with mRNA metabolism as it was shown to interact with the RNA binding protein Scp160 in polyribosome associated mRNP complexes [38]. Since then mRNA related functions of Bfr1 have gained increasing attention: Bfr1 was shown to affect P-body formation [39,40] and to bind hundreds of mRNAs despite the fact that it lacks canonical RNA binding domains [41,42].Considering the role of Bfr1 in mRNA metabolism and the recent finding that Bfr1 binds PMT1and PMT2 transcripts [42], we hypothesized that Bfr1 could affect UPOM by modulating Pmt1 and Pmt2 protein levels. We therefore analyzed Pmt1 and Pmt2 protein abundance in wild type versus bfr1∆ cells (Figure 3A,B, left panels). Our results show that Pmt1 and Pmt2 protein levels are markedly reduced in BFR1 deficient versus wild type cells. This holds true under ER stress conditions caused by the ER-GFP reporter in the screening strain background (compare lanes 2 and 4 in Figure 3A,B) with Pmt1 and Pmt2 levels increased in response to UPR (compare lanes 1 and 2 in Figure 3A,B), as well as in the absence of ER-GFP in an independent strain background (compare lanes 1 and 3 in Figure 3A,B). Quantification of PMT protein levels reveals a significant 2-fold reduction for both PMTs (Figure 3A,B, right panels) in bfr1∆ versus wild type cells.Since Bfr1 binds to numerous mRNAs, we investigated the effect of BFR1 deletion on proteinlevels of representative Bfr1 interactors [42] involved in protein import such as the main translocon subunit Sec61 [43], quality control such as the Hsp70 chaperone Kar2 [44] and N-glycosylation such as oligosaccharyl transferase (OST) subunits Ost3 and Wbp1 [45,46]. We also analyzed protein levels of the GPI-anchored protein Gas1 that is highly O-mannosylated [47] (Figure 3C).
Results show no major changes in protein levels for any of these Bfr1 targets in wild type versus bfr1∆ cells (compare lane 1 with 3 and 2 with 4 in Figure 3C), suggesting that Bfr1 binding to mRNA alone is not sufficient to affect protein abundance.To further substantiate the finding that O-mannosylation defects observed upon BFR1 deletionresult directly from decreased protein levels of Pmt1 and Pmt2, we performed a functional rescue experiment by overexpressing Pmt2. As shown in Figure 4A, Pmt2 overexpression restores O-mannosylation of ER-GFP in BFR1 deficient cells. In agreement Pmt2 overexpression significantly reduces GFP fluorescence detected in bfr1∆ cells, however, not to wild type levels (Figure 4B). In BFR1 deficient cells Pmt2 protein levels are markedly decreased compared to wild type (Figure 4C, compare lanes 1 and 3), even upon Pmt2 overexpression (compare lanes 2 and 4) and irrespective of ER stress caused by ER-GFP expression (compare lane 5 with 7 and 6 with 8). Inability to restore native Pmt2 levels as well as reduced levels of Pmt1, may explain why full complementation of ER-GFP O-mannosylation could not be gained. Taken together, our data show that the aberrant O-mannosylation of ER-GFP in bfr1∆ cells is a direct consequence of decreased Pmt1 and Pmt2 protein levels and that Bfr1 affects UPOM by controlling the abundance of these enzymes.In addition, total cell extracts from wild type cells expressing fully functional HA-tagged Bfr1 and bfr1∆ cells were fractionated on a sucrose step gradient. Analysis of the RNA content of 20 collected fractions showed enrichment of ribosomes in fractions F10 and F16 (Figure 6C). The respective control experiment was performed with EDTA supplemented lysates and resulted in the shift of both Absorbance260 peaks observed for F10 and F16 to soluble fractions in line with ribosomal disassembly (Supplementary Figure S5).
Analysis of specific marker proteins within F5, F10, and F16 reveals efficient separation of cytosolic and membrane fractions (Figure 6D, compare lanes 2 and 4). All analyzed fractions contain ribosomes as assessed by the ribosomal protein Rpl5, however, to different extents. F10 represents the cytoplasmic polyribosome fraction whereas F16 contains ER membrane bound polysomes (Figure 6D, compare Sec61 in lanes 3 and 4). Bfr1 was found throughout fractions consistent with reports of this cytosolic protein being associated with polyribosomes. Analysis of mRNA content in ribosome containing fractions F10 and F16 in wild type cells showed strong engagement of PMT1, PMT2, and SEC61 mRNA with ER membrane associated ribosomes, whereas only minor amounts of these mRNAs were detectable in the cytoplasmic fraction F10 (Figure 6E, lanes 1 and 2 respectively). In BFR1 deficient cells distribution of neither PMT1 and PMT2 mRNAs nor SEC61 and ACT1 mRNAs was changed compared to wild type cells. In combination our data show that PMT2 transcripts are equally distributed between the cytosolic and ER membrane bound polysomal fraction and that PMT1 and PMT2 mRNAs preferentially colocalize with membrane bound polyribosomes irrespective of Bfr1 presence.Next, we analyzed translation dynamics in wild type versus bfr1∆ cells by ribosome profiling, which provides a quantitative and high-resolution profile of in vivo translation and is based on deep sequencing of ribosome protected mRNA fragments [54]. Protein synthesis rates are derived from average ribosomal density along mRNAs based on two fundamental assumptions: That all ribosomes complete translation and that elongation rates are similar among different mRNAs [55].
Ribosomal densities along transcripts show active translation and provide a snapshot of protein synthesis within the cell, independent of transcript levels.Ribosome profiling was performed in duplicate for both wild type and BFR1 deficient cells.Replicates showed high correlation of reads per million mapped reads (RPM) values (r2 = 0.99 and r2 = 0.97 for wild type and bfr1∆ cells, respectively) (Supplementary Figure S6A; Supplementary Table S2). RPM values of wild type and bfr1∆ cells also showed high correlation (r2 = 0.97) (Supplementary Figure S6B; Supplementary Table S2) ruling out a generalized effect on translation. Statistical analysis revealed comparable subsets of genes significantly up- or downregulated at 0.01 false discovery rate (FDR) (red dots on Figure 7A). For Pmt1 and Pmt2 ribosome profiling data demonstrate a bfr1∆ to wild type ratio of averaged RPMs of 0.581 and 0.596 respectively that corresponds to a significant 1.7-fold decrease of ribosomal footprint density and therefore active protein synthesis in BFR1 deficient cells. This decrease in active translation correlates with the approximate 2-fold decrease in PMT protein abundance detected in bfr1∆ cells (Figure 3A,B). In line with this observation, active translation of representative Bfr1 targets whose expression levels did not change upon BFR1 deletion (Figure 3C), remain unaffected with the exception of Kar2 (wild type/bfr1∆ = 0.582) (Supplementary Table S2). Since PMT1 and PMT2 transcript levels do not change between wild type and mutant cells whereas ribosomal density is 1.7-fold lower, these results reveal Bfr1 as a translational enhancer of Pmt1 and Pmt2. Furthermore, we combined our ribosomal footprint data with the Bfr1 mRNA interactome unraveled by Lapointe et al. [42]. The 174 strongest mRNA interactors (Figure 7B, class A) include 104 mRNAs encoding for proteins of the secretome (filled dots) defined by Ast et al. [56]. Translation of 35 mRNAs, all encoding secretome proteins, is significantly reduced in absence of BFR1, suggesting that Bfr1 preferentially affects translation of ER-targeted proteins. Intriguingly, GO functional annotation clustering identified among those targets, protein glycosylation (PMT1, OST1,PMT2, PMT3, PMT4, KTR1, STT3, ALG12) and ergosterol biosynthesis (ERG24, ERG3, NCP1, ERG4, ERG11) as major functional clusters, pointing to Bfr1 as an important factor governing these processes.
3.Discussion
In recent years, protein O-mannosylation proved to be critically important for ER protein quality control. O-mannosylation affects ER protein homeostasis at different levels. On one hand, stress sensors as well as other crucial components of protein folding and quality control machineries carry O-mannosyl glycans which may directly impact on their function [14,58]. On the other hand, un- or misfolded proteins receive O-mannosyl glycans which label them for ER clearance [59]. In a first effort to identify factors that affect UPOM, the Pmt1-Pmt2 complex proved to be a central hub for ER protein quality control. Among our screening hits we find several mutants that probably impact on ER protein folding but do not directly affect O-mannosylation of the UPOM-reporter (Supplementary Table S1). An example is INO2 that encodes for a transcription factor responsible for derepression of phospholipid biosynthetic genes [60]. Membrane phospholipid perturbations have been linked to chronic ER stress in S. cerevisiae [61]. The presence of INO2 as well as SPF1 that was reported to cause ergosterol deficiency in the ER [62] further emphasizes the importance of ER membrane integrity to maintain the ER as a robust folding compartment in general. Most of the candidates, however, are linked to protein quality control as components of stress related pathways such as sln1∆, ptc1∆ and sic1∆ that encode functional components of the high osmolarity glycerol pathway (HOG pathway), as well as oca1∆ and oca2∆ that are involved in oxidative stress. Basal activity of the HOG pathway was shown to contribute to UPR induced accumulation of glycerol and thereby mediates resistance towards the ER stress inducing agent tunicamycin in S. cerevisiae [63]. Osmolytes such as glycerol are often referred to as “chemical chaperones” and have been shown to increase protein stability and restore ER homeostasis [64].
Increased fluorescence of ER-GFP in oca1∆ and oca2∆ mutants might be explained by the recent finding that yeast UPR is inhibited by oxidative stress [65]. With important components of the oxidative stress response missing, yeast UPR could be more efficient in folding of the UPOM-reporter. However, general activation of UPR such as in erj5∆ [66] and erv25∆ [67] or hrd1∆ mutants where ERAD is affected [68], do not impact on ER-GFP folding [19] (Supplementary Table S1), suggesting a more specific role of stress related UPR for proper reporter folding. Our screening further revealed unexpected links between ER-GFP, per se a non N-glycosylated protein, and N-glycosylation such as cwh41∆ and ost3∆ (Supplementary Table S1). CWH41 encodes for α-glucosidase I that is responsible for trimming of the outermost glucose of N-glycans within the calnexin-calreticulin cycle, thereby creating a time window before Mns1 and Htm1 mannosidases target the protein for degradation [69]. Ost3 is one out of nine subunits of the yeast OST complex that together with Ost6 determines functionally distinct OST complexes [70]. Ost3 was recently reported to be necessary for N-glycosylation of Pmt2 [71] but no direct evidence of impaired Pmt2 enzymatic activity was obtained in vivo. However, ER-GFP oligomers that are indicative of ER-GFP misfolding [19] were significantly reduced in ost3∆ cells suggesting more efficient folding in the absence of Ost3 (Supplementary Figure S2C). In addition to Pmt1 and Pmt2, the strongest factors identified in the screen directly affecting O-mannosylation of ER-GFP are Psa1 and Bfr1 (Figure 2B).
Psa1 catalyzes biosynthesis of GDP-mannose, the common sugar donor for Dol-P-Man production. Intriguingly, a second enzyme that contributes to GDP-mannose synthesis, the glucose-6-phosphate isomerase Pgi1 (Supplementary Figure S3B), is found at immediate proximity to the screening threshold (Supplementary Table S1), suggesting that GDP-mannose availability might indeed be important for PMT activity. That carbohydrate donor levels affect PMT activity has been also suggested in studies performed in S. cerevisiae [72] and Trichoderma reesei [73] in which manipulation of GDP-mannose levels affects glycosylation. These preliminary data suggest a so far unknown link between carbohydrate metabolism and UPOM. Bfr1 regulates Pmt1 and Pmt2 translation and therefore impacts on UPOM. Bfr1 is a cytoplasmic protein without any common RNA interacting motifs that was described as a component of polyribosome associated mRNP complexes in S. cerevisiae [38]. Further, Bfr1 mediates localization of certain mRNAs to P-bodies [39] and prevents P-body formation under normal conditions [40] further supporting a function for Bfr1 in mRNA metabolism. P-bodies are dynamic ribonucleoprotein complexes where mRNA storage, translational repression or degradation occurs [74]. Recent RNA binding studies that imply the presence of far more RNA binding domains than known to date [75] in combination with multiple approaches that identify hundreds of different mRNAs interacting with Bfr1 [41,42,76], suggest a role for Bfr1 as an RNA binding protein and translational regulator itself.
In addition to Pmt1 and Pmt2, Bfr1 significantly affects active translation of all PMTs and of additional 322 genes, from which nearly half show reduced translation in absence of Bfr1 (Figure 7A; Supplementary Table S2). Among those we find the sterol reductase Erg4 that catalyzes the final step in ergosterol biosynthesis [77] and that was recently described to be translationally regulated by Bfr1 [78]. We combined our data with Bfr1 interacting transcripts from Lapointe et al. [42] who reported Bfr1 targets to be highly enriched for mRNAs translated at the ER. In this “RNA Tagging” approach, Bfr1 interacting mRNAs were tagged with varying numbers of uridines by the poly(U) polymerase fused to Bfr1, depending on the strength of the interaction. Targets were classified into four groups based on the number of targeted RNAs and the length of the U-tag (class A encloses the strongest interactors). Crossing these datasets shows that Bfr1 controlled targets are enriched in classes A and B, which contain the strongest and most reliable Bfr1 binders. Among class A secretory proteins are Pmt1-4 and Erg4, as well as the OST subunits Ost1, Ost5, and Stt3 that form one out of two subcomplexes during OST complex assembly [79]. Given that these subcomplexes are intermediates that protect subunits from degradation, they might play a decisive role in dynamics of OST complex formation and N-glycosylation. In addition, class A secretory proteins harbor several components of ergosterol biosynthesis (Erg3, Erg4, Erg11, and Erg24) and two iron homeostasis genes (Ftr1 and Smf3). This finding is particularly intriguing given the importance of iron for ergosterol biosynthesis and for Ire1 clustering and UPR activation [80]. A summary of all classified targets is available in Supplementary Table S2. Although a more detailed analysis of strong Bfr1 binders will be necessary to define the biological impact of Bfr1 mediated translation, our data strongly suggest a function of Bfr1 as a local translation factor at the ER membrane.
How does cytoplasmic Bfr1 regulate translation at the ER membrane? Our data strongly suggest that Bfr1 is not a prerequisite for PMT transcript recruitment to the ER, in agreement with similar observations for the Bfr1 target Erg4 [78]. For Bfr1 this suggests two possible scenarios: Bfr1 could be targeted to the ER membrane via bound mRNAs as suggested for Erg4 [78] or Bfr1 could be associated with ER bound ribosomes before respective mRNAs reach the ER. It remains a challenging question for the future whether Bfr1 binds to mRNAs before or after their recruitment to the ER. In a wider context our data together with transcriptomic data from others [15] reveal that ER stress is an important determinant of Pmt1-Pmt2 abundance (Figure 3A,B; Figure 4C), that is additionally controlled on a translational level by Bfr1 (Figure 7). Interestingly Bfr1 is also a target of the UPR (Supplementary Figure S7; [15]) suggesting that the function of Bfr1 is relevant to maintain protein homeostasis in the ER. Maximal Pmt1-Pmt2 expression depends on both, transcriptional activation of Pmt1-Pmt2 under cell stress conditions as well as elevated translation efficiency mediated by Bfr1. The fine-tuned coordination of Pmt1-Pmt2 protein abundance with ER stress further implies that O-mannosylation and protein folding must be balanced to ensure functionality of canonical target proteins and unfolded protein O-mannosylation, the latter being more sensitive to subtle changes of Pmt1-Pmt2 protein levels. Exactly adjusting Pmt1-Pmt2 activity to ER protein load most likely enables O-mannosylation of highly diverse protein substrates without unintentionally interfering with protein folding.
4.Materials and Methods
S. cerevisiae strains used in this study are listed in Table 1. Strains derived from genetic libraries are underlined.
Yeast cultures were grown in yeast extract-peptone-dextrose (YPD) or synthetic defined (SD) medium at 30 ◦C. For auxotrophic selection corresponding amino acids were excluded from SD medium. For antibiotic-based Brefeldin A selection cultures were supplemented with 400 µg/mL geneticin (#11811-031, Invitrogen; Waltham, MA, USA) or 100 mg/L nourseothricin (#96736-11-7; Werner BioAgents; Jena-Cospeda, Germany).