Discovery of a nucleocytoplasmic O-mannose glycoproteome in yeast

Significance

Nucleocytoplasmic dynamic cycling of N-Acetylglucosamine (GlcNAc) on serine and threonine residues (O-GlcNAcylation) and phosphorylation coregulate important cellular processes in all eukaryotic organisms except yeast, including Saccharomyces cerevisiae and Schizosaccharomyces pombe. The lack of an equivalent nucleocytoplasmic O-glycosylation system in yeast has been difficult to explain given that O-GlcNAcylation is an essential modification in higher organisms. Here, we reveal that yeast use O-linked mannose to modify nucleocytoplasmic proteins on evolutionary-conserved regions and sites normally occupied by O-GlcNAc in higher eukaryotes. The results presented in this study open new avenues for exploration of nutrient sensing and signaling events based on nucleocytoplasmic O-glycosylation in yeast.

Abstract

Dynamic cycling of N-Acetylglucosamine (GlcNAc) on serine and threonine residues (O-GlcNAcylation) is an essential process in all eukaryotic cells except yeast, including Saccharomyces cerevisiae and Schizosaccharomyces pombe. O-GlcNAcylation modulates signaling and cellular processes in an intricate interplay with protein phosphorylation and serves as a key sensor of nutrients by linking the hexosamine biosynthetic pathway to cellular signaling. A longstanding conundrum has been how yeast survives without O-GlcNAcylation in light of its similar phosphorylation signaling system. We previously developed a sensitive lectin enrichment and mass spectrometry workflow for identification of the human O-linked mannose (O-Man) glycoproteome and used this to identify a pleothora of O-Man glycoproteins in human cell lines including the large family of cadherins and protocadherins. Here, we applied the workflow to yeast with the aim to characterize the yeast O-Man glycoproteome, and in doing so, we discovered hitherto unknown O-Man glycosites on nuclear, cytoplasmic, and mitochondrial proteins in S. cerevisiae and S. pombe. Such O-Man glycoproteins were not found in our analysis of human cell lines. However, the type of yeast O-Man nucleocytoplasmic proteins and the localization of identified O-Man residues mirror that of the O-GlcNAc glycoproteome found in other eukaryotic cells, indicating that the two different types of O-glycosylations serve the same important biological functions. The discovery opens for exploration of the enzymatic machinery that is predicted to regulate the nucleocytoplasmic O-Man glycosylations. It is likely that manipulation of this type of O-Man glycosylation will have wide applications for yeast bioprocessing.

All eukaryotic cells except yeast harbor a simple type of protein O-glycosylation designated O-GlcNAcylation [dynamic cycling of N-Acetylglucosamine (GlcNAc) on serine (Ser) and threonine (Thr) residues] in the cytosol and nucleus (1). O-GlcNAcylation is an essential process involving addition and removal of a single GlcNAc at Ser and Thr residues of nuclear, cytoplasmic, and mitochondrial proteins (2–5). O-GlcNAcylation is a widespread modification found on, for example, nucleoporins, transcription factors, kinases, and cytoskeletal and chromatin proteins; it is involved in a plethora of biological processes and believed to play causal roles in diabetes, cancer, cardiovascular, and Alzheimer’s disease (6–8). Sites of O-GlcNAcylation are often found at or in close proximity to protein phosphorylation sites, and the intricate interplay between both modifications is known to modulate many important processes in cells (8, 9). The transfer of GlcNAc to proteins is carried out by the O-GlcNAc transferase (OGT) using uridine diphosphate α-d-GlcNAc (UDP-GlcNAc) as a donor substrate, and a second hydrolytic enzyme, O-GlcNAcase (OGA), is available to remove the GlcNAc monosaccharide in a regulated dynamic process (4, 5). Discovered over 30 years ago, the OGT/OGA enzyme pair was initially identified in mammals, but subsequent work has demonstrated O-GlcNAcylation in bacteria, filamentous fungi, plants, and metazoans. The only eukaryotic cell type without identifiable orthologous OGT/OGA genes is yeast, although OGA orthologs have not been identified in plants. Yeast uses primarily Ser and Thr for phosphorylation, with tyrosine (Tyr) phosphorylation being used to an extremely low extent (10). This is in contrast to other eukaryotic cells that phosphorylate all three residues for extensive and vital signaling. It has been a longstanding puzzle why in yeast the coregulatory functions of O-GlcNAcylation are apparently not required or whether another type of protein O-glycosylation, such as O-linked mannose (O-Man), takes on this role (6).

All eukaryotes except nematodes and plants have a well-characterized O-Man glycosylation machinery for proteins trafficking the secretory pathway, and the enzymes involved in this process are multitransmembrane-spanning dolichol phosphate β-d-Man (Dol-P-Man):protein O-mannosyltransferases (PMTs) using the membrane-associated Dol-P-Man donor substrate and transferring a single Man residue to selected Ser and Thr residues of proteins. These enzymes are located in the ER, with their catalytic domains oriented into the lumen (11, 12). Higher eukaryotic cells have two PMT isoenzymes (POMT1 and POMT2), and the initial O-Man glycans are elongated, branched, and capped by sialic acids by a series of glycosyltransferases. Deficiencies in many of the enzymes involved in protein O-mannosylation in humans underlie a group of congenital muscular dystrophies (13). Yeast, in contrast, have a family of at least six PMTs, and the initial Man monosaccharide is extended only by additional mannose residues through the actions of Golgi resident mannosyltransferases of the KTR/MNT and MNN family that use GDP α-d-Man (GDP-Man) as a donor substrate (11). The O-Man glycosylation and the action of multiple PMTs are essential for yeast (14), and O-Man plays major roles in maintaining yeast cell wall integrity (15).

Our knowledge of the proteins undergoing O-Man glycosylation in yeast and the specific sites of glycosylation is limited (11, 12), but recently we developed a glycoproteomic strategy to probe the O-Man glycoproteome of human cells using genetic engineering to simplify the O-glycan structures, the so-called “SimpleCell” strategy, in combination with Concanavalin A (ConA) lectin chromatography for enrichment of glycopeptides and mass spectrometric sequencing (16). This resulted in identification of a large number of O-Man glycoproteins and O-Man glycosites, demonstrating, for example, that cadherins and protocadherins are major carriers of O-Man glycans. In the present study, we modified this strategy to probe the yeast O-Man glycoproteome from total cell lysates. In striking contrast to our studies with human cell lines, we in addition to proteins entering the secretory pathway also identified a large number of O-Man proteins annotated as classical nuclear, cytosolic, or mitochondrial proteins that are not expected to be exposed to the known O-Man glycosylation machinery in the secretory pathway. The nucleocytoplasmic O-Man glycosites were located on proteins and in positions resembling that of the O-GlcNAcylation process in higher eukaryotes. Here, we describe this O-Man glycoproteome and suggest that the nucleocytoplasmic O-Man modifications in yeast represent the missing equivalent to the O-GlcNAcylation process of higher eukaryotes, and future studies can now address the biosynthetic machinery and biological functions.

Results and Discussion

Probing the O-Man Glycoproteome of Saccharomyces cerevisiae and Schizosaccharomyces pombe.

Because yeast is known to produce heterogeneous elongated polymannose structures, we initially used the KRE2ΔKTR1ΔKTR3Δ mutant strain (Mut), lacking the α1,2-mannosyltransferase (KRE2) involved in the second biosynthetic step of O-Man glycans and additional gene family members (KTR1 and KTR3), to enable analysis of the O-Man glycoproteome with more simplified O-Man glycan structures (Fig. 1A) resembling our SimpleCell strategy (16).

Fig. 1.

Identification of a novel type of nucleocytoplasmic protein O-mannosylation. (A) Graphic depiction of the glycoproteomic approach for identification of yeast O-mannosylation. WT or KRE2ΔKTR1ΔKTR3Δ Mut yeast total cell lysates treated with trypsin and PNGase F are enriched by ConA LWAC and glycopeptides identified by MS. (B) Pie chart showing the cellular localization of identified proteins with O-Man glycosites from total cell lysates of WT and Mut S. cerevisiae and a cytoplasmic preparation (S3) from WT S. cerevisiae. The composite results (Total) illustrate the total nonredundant number of O-Man glycoproteins identified. The identified O-Man glycoproteins assigned to the secretory pathway (Extracellular) will be described elsewhere. (C) GC-MS analysis of mannitol and sorbitol standards (gray trace). GC-MS profile of yeast reducing-end sugars released by reductive β-elimination (black trace). (D) Analysis of O-Man anomericity using jack bean α-mannosidase. Bar chart shows total (nuclear, cytoplasmic, and mitochondrial) number of identified O-Man glycoproteins for WT and Mut S. cerevisiae strains before (–) and after (+) α-mannosidase treatment.

We used total cell lysates obtained by vortexing with glass beads in Rapigest detergent for trypsin digestion. N-linked glycans were removed by PNGase F digestion, and the digests were subjected to ConA lectin weak affinity chromatography (LWAC) for enrichment of O-Man glycopeptides. Enriched O-Man glycopeptides were further fractionation by isoelectric focusing (IEF) and analyzed by nanoflow liquid chromatography-mass spectrometry (nLC-MS/MS) (Fig. 1A). Through this approach using the KRE2ΔKTR1ΔKTR3Δ Mut, we identified a considerable number of O-glycoproteins and O-glycosites with one or more hexoses attached using both higher energy collision dissociation (HCD) and electron transfer dissociation (ETD) fragmentation modes. A summary of the identified glycoproteins from S. cerevisiae is presented in Fig. 1B. In total we mapped 291 unique O-Man glycoproteins and ∼1,000 O-Man glycosites in Mut total cell lysates.

Discovery of a Nucleocytoplasmic O-Man Glycoproteome.

Analysis of the identified proteins revealed that a large proportion (n = 83, 29%) were known or predicted nuclear and/or cytosolic proteins without recognizable signal peptides. This finding was in striking contrast to our previous analysis of the O-Man glycoproteome of human cells, where all identified glycoproteins were known or predicted to traffic the secretory pathway (16). We therefore also inspected all identified glycoproteins with signal peptides for the specific localization of O-glycosites with respect to known or predicted membrane orientation and found sites predicted to be located in the cytosolic compartment (Dataset S1). Thus, more than a third of the O-Man glycoproteins identified in Mut S. cerevisiae were classical nuclear, cytosolic, or mitochondrial proteins that are not exposed to the known O-Man glycosylation machinery in the secretory pathway (Fig. 1B).

To exclude the possibility that the unexpected finding was related to the KRE2ΔKTR1ΔKTR3Δ Mut, we further explored this by analyzing total cell lysates of wild-type (WT) S. cerevisiae, which resulted in identification of 261 O-Man glycoproteins of which a similar fraction of 91 (35%) glycoproteins were from nuclear, cytoplasmic, or mitochondrial compartments. Also, we sought to enrich for cytoplasmic proteins by analyzing a crude cytoplasmic fraction (S3) of fractionated WT yeast, and this resulted in identification of a number of unique nucleocytoplasmic glycoproteins (Fig. 1B and Dataset S1). In total, we identified 162 unique glycoproteins from Mut and WT S. cerevisiae strains with O-glycosites that were either only found in cytosolic, nuclear, or mitochondrial compartments (n = 160) or where the sites identified were located in the cytosolic part of transmembrane proteins (n = 2).

In a preliminary study, we also explored the O-Man glycoproteome of the fission yeast S. pombe (WT) using the same experimental approach as above (Fig. 1A) with total cell lysates. We identified 178 O-Man glycoproteins, of which 87 (49%) are known to have cytosolic, nuclear, or mitochondrial localization (Fig. S1 and Dataset S2). The identified O-Man proteins classified as nucleocytoplasmic greatly expanded the total nucleocytoplasmic glycoproteome, as there was little overlap among the datasets from S. cerevisiae and S. pombe, with only a few identified in both.

Fig. S1.

Pie chart showing the distribution of S. pombe O-Man glycoproteins with glycosylations in nucleocytoplasmic, extracellular (including ER and Golgi lumen), and “not assignable” subcellular compartments.

In the following, we focus on the deepest nucleocytoplasmic O-Man glycoproteome data obtained from S. cerevisiae (Dataset S1). The O-Man glycoproteome subset comprising glycoproteins and domains known to be exposed to the secretory pathway will be presented in a separate publication.

The Nucleocytoplasmic O-Glycans Are Based on Man-α-O-Ser/Thr.

Our O-glycoproteomics strategy largely hinges on the α-Mannose specificity of the lectin ConA used for the LWAC as well as the MS identification of the mass increment for hexose residues. Thus, the approach does not reveal the absolute stereochemistry of the identified modifications. We therefore performed a series of experiments to verify the anomeric and epimeric configuration of the identified peptide-linked hexose residue. First, we released O-linked glycans by reductive β-elimination and analyzed the O-glycan profiles by MALDI-TOF mass spectrometry and found Hex_2–5 oligosaccharides to be the dominating species (Fig. S2). This glycoprofiling does not enable us to selectively quantify the composition of hexose structures on the nucleocytoplasmic glycoproteins, but as shown in Dataset S1, we identified nucleocytoplasmic proteins with both a single and a disaccharide hexose structure, showing that at least some elongation occurs. Mammalian nucleocytoplasmic O-GlcNAc is not elongated by other glycans, but in plants this type of glycosylation is elongated (17). The glycans released by reductive β-elimination were subsequently subjected to acid hydrolysis and trimethylsilyl derivatization before GC-MS analysis. This approach enables the reducing-end hexose to be differentiated from internal or terminal hexose residues through its distinct retention time in GC-MS, and for the β-eliminated yeast O-glycans, the reducing-end hexose displayed a similar retention time and fragmentation pattern as the mannitol standard, thus confirming that the peptide-linked hexose is a mannose residue (Fig. 1C and Fig. S3). Having resolved the epimeric configuration of the glycan, we then turned our attention to the anomeric configuration of the mannose–peptide linkage. We performed jack bean α-mannosidase digestion on the ConA-enriched and IEF-fractionated samples from both WT and KRE2ΔKTR1ΔKTR3Δ yeast and found that essentially all mannose residues were hydrolyzed by the α-mannosidase treatment (Fig. 1D and Dataset S1). Approximately 25% of the identified O-Man glycopeptides were readily identified as the corresponding peptides without the Man residues after digestion (Dataset S1). This suggests that the anomericity of the mannose linkage is in an α-configuration. Thus, we conclude that all identified O-Man glycosites involved α-linked Man residues hitherto known only on proteins trafficking the secretory pathway.

Fig. S2.

Yeast O-linked glycans from a crude preparation of cytoplasmic proteins (S3) obtained by ultracentrifugation (described in Materials and Methods). The total pool of O-linked glycans from the S3 fraction were released by reductive β-elimination, permethylated, and detected as sodiated ions by MALDI-TOF. Hexose (mannose) residues are depicted as green circles.

Fig. S3.

Identification of epimeric configuration of hexose monosaccharides released from yeast fraction S3. (A) GC-MS of sorbitol, mannitol, and mannose standards (Upper) together with hexoses from yeast S3 fraction released by reductive β-elimination (Lower). Automatic area (AA) integration was performed to estimate the relative amounts of mannose (84%) and mannitol (16%) from the yeast S3 sample. All samples have been spiked with myo-inositol (internal standard), depolymerized by acid treatment, and TMS-derivatized before injection. Retention times are normalized relative to the myo-inositol standards. (B) Mass spectra from peaks at ∼30 min in A showing the TMS-hexitol fragmentation pattern for sorbitol, mannitol, and yeast S3 (mannitol).

Nucleocytoplasmic O-Man Proximity to Phosphorylation Sites.

The 160 identified O-Man nucleocytoplasmic glycoproteins identified in S. cerevisiae included transcription factors, nucleoporins, histones, and kinases, which are all classical proteins undergoing O-GlcNAcylation as well as phosphorylation in higher eukaryotes (Dataset S1). For the identified nucleocytoplasmic yeast glycoproteins, 80% are known to be phosphorylated (18–21), and as much as 19% (221 in total) of the identified O-Man glycosites were found to be identical to previously identified phosphorylation sites (Dataset S3), a relatively high number considering that neither the current yeast phosphoproteome nor the O-Man glycoproteome presented here is complete.

An illustrative example was the glycogen phosphorylase (GP) enzyme, which mobilizes cellular energy by breaking down glycogen into glucose-1-phosphate. The mechanism of GP activation is conserved among eukaryotes and involves, in addition to allosteric elements, a single phosphorylation at Thr31 in yeast and Ser15 in mammals, both located in the N-terminal regulatory domain (Fig. 2A) (22, 23). We identified an O-Man glycan on the regulatory Thr31 of yeast GP, demonstrting that nucleocytoplasmic O-Man in yeast may compete directly with phosphorylation for functionally important sites. Clearly O-Man glycosylation of Thr31 in GP will block phosphorylation, and it is expected that this will severely affect activation of the enzyme. Thus, the nucleocytoplasmic O-Man glycosylation may therefore function as an important control switch for utilization of reserve carbohydrates in yeast.

Fig. 2.

Graphic illustration of cross-talk between O-Man glycosylation and phosphorylation with analogy to O-GlcNAcylation in other eukaryotes. (A) Sequence alignment (grayscale proportional to conservation) of human and yeast GP with expansion of the N-terminal domain showing the co-occupancy at the regulatory Thr31 residue. (B) Alignment of human and yeast histone H2B proteins demonstrating C-terminal O-GlcNAc and O-Man modifications, respectively. Human H2B lysine 120 is indicated in yellow. (C) Alignment of mouse and yeast plasma membrane ATPase proteins. Left expansion shows part of the ATP binding motif (yellow) with overlapping O-GlcNAc/O-Man occupancy at Thr558. Right expansion shows the C-terminal regulatory domain modified by O-Man and phosphorylations.

We also compared the proximity of all O-Man glycosites to known phosphorylation sites, which further demonstrated that the O-Man glycosites assigned to nucleocytoplasmic proteins in general tended to have phosphorylation sites closer than the O-Man glycosites assigned to proteins and protein domains exposed to the secretory pathway (Fig. S4). Thus, the nucleocytoplasmic O-Man glycoproteins clearly resemble the mammalian O-GlcNAcylated proteins.

Fig. S4.

Distance to the closest phosphorylation site from different glycosylation site types. Unambiguously assigned site data for mammalian O-GlcNAc (802 proteins and 2,601 sites) and S. cerevisiae O-Man data (92 cytosolic proteins, 202 sites, and 151 extracellular proteins, 855 sites) as well as phosphorylation data from refs. 18–21 were used to find the distance to the closest phosphorylation site within the same compartment (determined using the transmembrane region prediction from TMHMM) up to a distance of 200 amino acids. A higher proportion of cytosolic proteins bearing O-Man have nearby phosphorylation sites than the set of extracellular O-Man proteins. The cytosolic O-Man has a higher proportion of sites closer to phosphorylation sites, similar to O-GlcNAc, suggesting a similar relationship between O-Man glycosylation and phosphorylation as found for O-GlcNAc.

We further analyzed the amino acid sequences surrounding the nucleocytoplasmic O-Man glycosites, but this approach did not reveal any apparent sequence motifs for glycosylation (Fig. S5) similar to what has been reported for O-GlcNAcylation (24). However, a higher proportion of flanking hydrophobic residues was observed for nucleocytoplasmic O-Man glycosites compared with O-Man glycosites found on extracellular proteins.

Fig. S5.

WebLogo plots comparing chemical properties of neighboring amino acids for (A) mammalian O-GlcNAc (2,371 sites), (B) yeast nucleocytoplasmic O-Man (202 sites), and (C) yeast extracellular O-Man sites (855 sites). Amino acids are encoded such that “S” encodes for small and nonpolar residues (S,T,G,A), “A” encodes for acidic residues (D,E), “B” encodes for basic residues (K,R,H), “P” encodes for polar residues (N,Q,Y), and “H” encodes for hydrophobic residues (C,F,I,L,M,P,V,W). The O-GlcNAc neighborhood is substantially similar to the previously published logo plot (24).

Analysis of the subset of proteins for which mammalian orthologs could be reliably established (79/160) revealed that 24% of these have also been reported to be O-GlcNAcylated in rodents or humans (Table S1) (24–27). This analysis is likely biased by the limited depth of the O-glycoproteome data available currently, and we expect the overlap to increase with deeper characterization of the two glycoproteomes. Thus, the identified O-Man glycosylation of nucleocytoplasmic proteins is likely to represent the equivalent to O-GlcNAcylation found in higher eukaryotes, and the discovery suggests that yeast possesses a hitherto unidentified O-glycosylation machinery that operates in cytoplasmic, nuclear, and mitochondrial compartments.

Table S1.

Orthology mapping of yeast O-Man proteins to mammalian O-GlcNAc proteins

Yeast	Mouse	Rat	Human
60S ribosomal protein L36-B		60S ribosomal protein L36
Glyceraldehyde-3-phosphate Dehydrogenase 3	Glyceraldehyde-3-phosphate dehydrogenase	Glyceraldehyde-3-phosphate dehydrogenase
Pyruvate kinase 1	Pyruvate kinase PKM
Tubulin beta chain	Tubulin beta-2A chain
Tubulin beta chain	Tubulin beta-3 chain
Tubulin beta chain	Tubulin beta-4A chain
Tubulin beta chain	Tubulin beta-4B chain
Tubulin beta chain	Tubulin beta-2B chain
Tubulin beta chain	Tubulin beta-6 chain
Nucleoporin NSP1		Nuclear pore glycoprotein p62	Nuclear pore complex protein Nup214
General transcriptional corepressor CYC8	Lysine-specific demethylase 6A
Actin-binding protein	Src substrate cortactin
Actin-binding protein	Drebrin-like protein
Eukaryotic translation initiation factor 2 subunit alpha	Eukaryotic translation initiation factor 2 subunit 1
Malate dehydrogenase, cytoplasmic	Malate dehydrogenase, mitochondrial
Probable DNA-binding protein SNT1	Nuclear receptor corepressor 2
Probable DNA-binding protein SNT1	Septin-9
Actin cytoskeleton-regulatory complex protein SLA1	RalBP1-associated Eps domain-containing protein 1
Actin cytoskeleton-regulatory complex protein END3	RalBP1-associated Eps domain-containing protein 1
Eukaryotic initiation factor 4F subunit p150			Eukaryotic translation initiation factor 4 gamma 1
E3 ubiquitin-protein ligase RSP5	E3 ubiquitin-protein ligase NEDD4
GTPase-activating protein BEM2/IPL2	Breakpoint cluster region protein	Nuclear pore glycoprotein p62
ADP ribosylation factor GTPase-activating protein effector protein 2	Stromal membrane-associated protein 2
PAB1-binding protein 1	Ataxin-2
PAB1-binding protein 1	Ataxin-2–like protein		Ataxin-2–like protein
Reticulon-like protein 1	Reticulon-1
Reticulon-like protein 1	Reticulon-3
Reticulon-like protein 1	Reticulon-4
Reticulon-like protein 2	Reticulon-1
Reticulon-like protein 2	Reticulon-3
Reticulon-like protein 2	Reticulon-4

Identified yeast O-Man proteins (left column) and mapped mammalian O-GlcNAc modified orthologs shown for mouse, rat and human (right columns). Orthology mapping data were retrieved from the Saccharomyces Genome Database (39), for pairwise mappings between Yeast/Mouse, Yeast/Rat and Yeast/Human, and assembled into the table by mapping associated UniProt protein identifiers.

O-Man Glycosites and Comparison with Mammalian O-GlcNAcylation Sites.

We selected examples of orthologous proteins with known O-GlcNAcylation sites and identified O-Man glycosites for more detailed analysis. Although the existence of O-GlcNAc on histones was questioned in a recent study (28), O-GlcNAcylation is believed to be one of several posttranslational modifications (PTMs) that constitute the histone code and regulate histone interactions with DNA and effector proteins, and this PTM has been found on histones H2A, H2B, and H4 (29). It has further been demonstrated that O-GlcNAcylation of human histone H2B in response to glucose levels modulates the transcriptional response by promoting monoubiquitination of lysine residue 120 (30). A glucose-dependent response is also observed in yeast where the orthologous histone H2B.1 undergoes monoubiquitination at the conserved lysine residue, although the preceding step promoting this monoubiquitination has not been identified (31). We found O-Man glycosites on the highly conserved yeast histone ortholog H2B.1 positioned virtually in the same region as O-GlcNAcylation on the human histone H2B (Fig. 2B). We found O-Man residues on the peptide TKYSSST¹²⁹ but could not assign the exact positions of the glycosites; however, an O-GlcNAc site has been identified in the highly conserved TSS sequon. Although we in this first limited study did not identify O-Man glycosites on other histones, the results suggest that O-Man glycosylation is part of the histone code in yeast, similar to O-GlcNAcylation in higher eukaryotes.

The plasma membrane H⁺-ATPase (PMA1) is O-GlcNAcylated in the cytosolic region in mammals (24), and we identified O-Man glycosites on the yeast PMA1 ortholog that were conserved and overlapping with the known O-GlcNAcylation sites as well as phosphorylation sites (Fig. 2C). In particular, we found O-Man at Thr558, a conserved residue within the ATP binding motif of PMA1 (32), which is also O-GlcNAcylated in mammals (24). The consequence of glycosylation at Thr558 is still unclear, but given that Thr558 is in immediate proximity to the ATP molecule, it is reasonable to predict that the O-glycan will have an impact on ATP binding (Fig. S6). PMA1 was further identified with O-Man modifications on the C-terminal regulatory domain where one site, Ser899, has also been identified as being phosphorylated (Fig. 2C). The C-terminal region is a critical region that undergoes glucose-dependent phosphorylation, leading to rapid PMA1 activation (33, 34). Upon glucose depletion, PMA1 is reversibly deactivated within minutes, suggesting a negative regulation driven by PTMs, although this mechanism is not fully understood.

Fig. S6.

O-Man modification at Thr558 of PMA1. (A) Sequence alignment of mammalian orthologs to yeast PMA1 demonstrating the conservation of residues surrounding part of the ATP binding motif (yellow). Black arrow indicates the modified Thr residue, which is observed with O-Man in yeast and O-GlcNAc in mammals. (B) Crystal structure of the orthologous plant plasma membrane ATPase 2 (Protein Data Bank ID code 3B8C) with the nonhydrolysable ATP analog AMPPCP (38). Expansion shows the corresponding Thr558 residue and the proximity to the ATP analog. The catalytic Asp378 residue is also shown. Cartoon representation was generated by the PyMOL molecular graphics system.

In conclusion, we describe the existence of a hitherto unknown nucleocytoplasmic type of protein O-mannosylation in S. cerevisiae and S. pombe. We previously used the same strategy to characterize the human O-Man glycoproteome and found no evidence of O-mannosylation of nucleocytoplasmic proteins (16). Because yeast is also the only known eukaryote to lack the OGT/OGA enzymes responsible for O-GlcNAcylation, it is likely that nucleocytoplasmic O-mannosylation is unique to yeast. Although other fungi including molds like Aspergillus niger have OGT/OGA homologous genes, further studies are needed to determine if these also have nucleocytoplasmic O-Man glycoproteins and ultimately if the two nuclecytoplasmic types of glycosylation can coexist. The O-Man glycosites on nucleocytoplasmic proteins share the characteristic feature of overlap with phosphorylation sites and are positioned similarly to O-GlcNAc sites on evolutionary conserved proteins and sites. We therefore propose that O-mannosylation of nucleocytoplasmic proteins in yeast serves the equivalent functions as the O-GlcNAcylation process in higher eukaryotes. The nutrient-sensing role of O-GlcNAcylation is likely to be mirrored by O-mannosylation in yeast, with the difference being that GDP-Man, and not UDP-GlcNAc, will likely serve the role of donor substrate and key nutrient sensor in yeast. The biosynthesis of GDP-Man would thus link glucose and nucleotide metabolism in a network node capable of integrating nutrient signals that are ultimately relayed as nucleocytoplasmic O-Man glycosylation. The apparent shift to nutrient sensing by UDP-GlcNAc in higher eukaryotes from GDP-Man may indicate a need to evolve a more complex nutrient sensor capable of incorporating glucose, nucleotide, amino acid, and fatty acid metabolism by flux through the hexosamine biosynthetic pathway. Yeast is the only eukaryotic cell type without identifiable orthologous OGT/OGA genes involved in O-GlcNAcylation, and our findings here provide an explanation, as the enzymes required for transfer and removal of α-Mannose residues are likely to have entirely different structures and be encoded by nonhomologous genes. Identifying the enzymes responsible for the nucleocytoplasmic O-Man will provide unique opportunities to regulate a broad spectrum of cellular processes in yeast with huge potential for yeast-based bioproduction.

Materials and Methods

Yeast Strains and Culture Conditions.

The following S. cerevisiae strains were used: WT (BY4741, MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0), the KTR/MNT triple mutant (BY4741 except kre2Δ::His3-GFP, ktr1Δ::SAT, ktr3Δ::KanMX4; a gift of Howard Bussey, McGill University, Montreal), and BY5457 loxp::pep4Δ::loxp [WT BY5457 (MATα ura3-52::p1785 leu2-3,112 his4-519 trp1 pho3-1 pho5-1 can^r) with the additional loxp::pep4Δ::loxp modification; a gift from Rosa Laura Lopez Marques, University of Copenhagen, Copenhagen). WT cells (BY4741) were grown in rich medium containing 2% (wt/vol) glucose, 2% (wt/vol) peptone, and 1% (wt/vol) yeast extract (YPD) under aerobic conditions at 30 °C with constant shaking at 170 rpm in a rotary shaker incubator (I Series 26, New Brunswick Scientific Co., Inc.). Cell growth was determined by measuring the optical density at a wavelength of 600 nm (OD₆₀₀). S. cerevisiae cultures were grown to midlog phase (OD₆₀₀ 0.8–1.1) and harvested by centrifugation at 3,000 × g for 5 min. Growth conditions and harvesting of BY5457 cells are described below. The S. pombe WT (no-marker h–) strain, a derivative of the WT heterothallic strains 972h– and 975h+, was used. The fission yeast cells were propagated at 30 °C in YES media (5 g/L yeast extract, 30 g/L glucose, 225 mg/L adenine, 225 mg/L leucine, and 225 mg/L uracil) and harvested in the late exponential phase by centrifugation (3,000 × g for 2 min) and lysed using glass beads as described below.

LWAC Isolation of O-Man Glycopeptides.

Cell extracts were prepared from a total of 100 OD₆₀₀ units of packed yeast cells with acid-washed 5-mm glass beads in ice-cold 0.1% Rapigest (Waters Corp.) in 50 mM NH₄HCO₃ by vortexing in reciprocal shaker (Hybaid RiboLyser) for 4 × 25 s with 1-min intervals at 4 °C. The bottom of the tube was punctured, and the lysate was collected. Unbroken cells and larger cell debris were removed by a low-speed centrifugation step at 1,500 × g for 5 min at 4 °C. The cleared lysates were heated at 80 °C for 10 min, followed by reduction in 5 mM dithiothreitol at 60 °C for 30 min and alkylation in 10 mM iodoacetamide at room temperature (RT) in darkness for 30 min. Samples were digested with 25 µg trypsin (Roche) overnight (ON), heat-inactivated by incubation at 95 °C for 20 min, and treated with 8 U PNGase F ON at 37 °C. An additional 4 U PNGase F was added and incubated at 37 °C for 4 h. The digests were acidified with 12 µL TFA, incubated at 37 °C for 20 min, cleared by centrifugation at 10,000 × g for 10 min, and purified by Sep-Pak C18 (Waters) columns. The LWAC protocol for isolation of O-Man glycopeptides was as previously described (16). Sep-Pak–purified peptides were concentrated by evaporation, and the reduced solution was diluted with an equal volume of 2 × ConA buffer A (40 mM Tris·HCl, pH 7.4, 300 mM NaCl, 2 mM CaCl₂/MgCl₂/MnCl₂/ZnCl₂, 1 M urea) before loading in a 2.8-m-long ConA lectin agarose column. The column was washed with 10 column volumes (CVs) of ConA buffer A at 100 µL/min before elution with five CVs of ConA buffer B (20 mM Tris·HCl, pH 7.4, 150 mM NaCl, 1 mM CaCl₂/MgCl₂/MnCl₂/ZnCl₂, 0.5 M methyl-α-d-glucopyranoside/methyl-α-d-mannopyranoside) at 50 µL/min. Fractions containing glycopeptides were purified by in-house packed Stage tips (Empore disk-C18, 3M) and further fractionated by IEF as previously described (35). For each IEF fraction, 50% was analyzed by nLC-MS/MS as described below. The remaining 50% of each IEF fraction was digested at 37 °C ON with 30 U/mL jack bean α-mannosidase (Prozyme) in 100 mM sodium acetate, 2 mM Zn²⁺, pH 5 before analysis by nLC-MS/MS.

nLC-MS/MS and Data Analysis.

Mass spectrometric analyses were performed essentially as previously described (16). Samples were analyzed on a setup composed of an EASY-nLC 1000 (Thermo Fisher Scientific) interfaced via a nanoSpray Flex ion source to an LTQ-Orbitrap Velos Pro hybrid spectrometer (Thermo Fisher Scientific). The EASY-nLC 1000 was equipped with a polar end-capped C18-silica column 21 cm in length, 75 μm in inner diameter, and 1.9 μm in particle size. A data-dependent mass spectral acquisition routine, HCD, and subsequent ETD scan were used for all runs. Briefly, a precursor MS1 scan (m/z 355–1,700) of intact peptides was acquired in the Orbitrap at a resolution setting of 30,000, followed by Orbitrap HCD-MS2 and ETD-MS2 of the five most abundant multiply-charged precursors in the MS1 spectrum; a minimum MS1 signal threshold of 50,000 ions was used for triggering data-dependent fragmentation events; and MS2 spectra were acquired at a resolution of 15,000. Data processing was carried out using Proteome Discoverer 1.4 software (Thermo Fisher Scientific) as previously described (16) with minor modifications, as outlined below. Raw data files (.raw) were processed using the Sequest HT node and searched against the canonical S. cerevisiae proteome (7,225 entries) downloaded from the Uniprot database (October 2013) or the canonical S. pombe proteome (5,092 entries; September 2015). Spectral assignments at the medium confidence level (P > 0.01) and below were resubmitted to a second Sequest HT node using semispecific tryptic cleavage. Final results were filtered for high-confidence (P < 0.01) identifications only. Spectra matched to peptides with nucleocytoplasmic O-Man modifications were inspected manually to verify the accuracy of the assignments. The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium (36) via the PRoteomics IDEntifications (PRIDE) partner repository with the dataset identifier PXD002924. Signal peptides were retrieved manually from Uniprot or through the SignalP tool (server version 4.1), and transmembrane domains were predicted with Trans Membrane Hidden Markov Model (TMHMM) (server version 2.0); both tools are available at www.cbs.dtu.dk/services/.

Cytoplasmic proteins (S3 fraction) were prepared as follows: Cells (BY5457 loxp::pep4Δ::loxp) were grown as above in YPAD medium (YPD supplemented with 0.01% Adenine hemisulfate wt/vol). Cells were pelleted by centrifugation at 800 × g for 15 min at 4 °C and suspended in a 1:5 ratio in 50 mL lysis buffer (50 mM Tris·HCl, pH 7.5, 150 mM NaCl) supplemented with one tablet of cOmplete protease inhibitor (Roche) and lysed by French press at 4 °C at 40 Kpsi (S1 fraction). The S1 fraction was cleared by centrifugation at 5,000 × g at 4 °C for 15 min. The supernatant (S2) was subjected to ultracentrifugation at 45,000 × g at 4 °C for 30 min, followed by 100,000 × g at 4 °C for 3 h, resulting in the cytoplasmic S3 (supernatant) fraction. The S3 fraction was trypsin-digested, ConA-enriched, and purified by Stage tips as described above. For each LWAC elution fraction, 5% of the total sample was injected and analyzed by HCD fragmentation only.

MALDI and GC-MS Analysis.

The S3 fraction (13 mL) was dialyzed against 6 L 50 mM NH₄HCO₃, pH 7.8, using 8,000 Da MWCO membranes, and concentrated by evaporation. Following reduction and alkylation (described above), the cytosolic proteins were trypsin-digested ON and purified by Sep-Pak C18 (Waters) columns. The tryptic peptides were dried, resolubilized in 100 mM NaOH and 1 M NaBH₄, and incubated ON at 50 °C. The β-elimination reaction was terminated by the addition of 8 µL glacial acetic acid, and the released O-glycan alditols were separated from proteins by a second Sep-Pak C18 (Waters) purification. Reduced O-glycans were desalted by Dowex AG 50W ×8 cation exchange resin (Bio-Rad) followed by repeated (×5) addition of 500 µL 1% (vol/vol) acetic acid in methanol and evaporation over a stream of N₂ gas. For MALDI analysis, released oligosaccharides were permethylated essentially as previously described (37). Briefly, released oligosaccharides were dried in glass vials to which 18 mg NaOH powder, 150 µL dimethyl sulfoxide with 0.1% H₂O (vol/vol), and 30 µL methyl iodide were added; the mixture was incubated at RT for 1 h, and the reaction was terminated by addition of 150 µL ice-cold H₂O followed by 200 µL chloroform. The organic phase was washed five times with 1 mL H₂O and finally dried with a stream of N₂ gas. Permethylated oligosaccharides were reconstituted in 30 µL 50% methanol in H₂O (vol/vol), and 1 µL was cocrystalized with an equal amount of matrix [10 mg/mL 2,5-dihydroxybenzoic acid and 2.5 mM sodium acetate in 70% acetonitrile in H₂O (vol/vol)]. MALDI-TOF (Autoflex Speed, Bruker Daltonics) was operated in the reflector mode using positive polarity and 2,000 laser shots per spot. For GC-MS, reduced O-glycans were spiked with 1 µg myo-inositol (internal standard) and depolymerized by incubation in 0.5 N methanolic HCL (Supelco) at 80 °C for 16 h. Monosaccharides were per-O–trimethylsilylated with Tri-Sil reagent (Thermo Scientific) at 80 °C for 30 min. GC-MS analysis was performed using a TRACE GC Ultra gas chromatograph coupled to a PolarisQ ion trap mass spectrometer. Samples were injected (splitless mode) at 40 °C (1 min), and the oven temperature was ramped to 150 °C (25 °C/min) followed by an increase to 200 °C (1 °C/min) before a final ramp to 260 °C (10 °C/min), where it was held for 5 min. Monosaccharides were identified by comparison of retention times and mass spectra to hexose and hexitol standards. All retention times were relative to the myo-inositol internal standard.