Polypeptide binding specificities of Saccharomyces cerevisiae oligosaccharyltransferase accessory proteins Ost3p and Ost6p

Abstract

Asparagine-linked glycosylation is a common and vital co- and post-translocational modification of diverse secretory and membrane proteins in eukaryotes that is catalyzed by the multiprotein complex oligosaccharyltransferase (OTase). Two isoforms of OTase are present in Saccharomyces cerevisiae, defined by the presence of either of the homologous proteins Ost3p or Ost6p, which possess different protein substrate specificities at the level of individual glycosylation sites. Here we present in vitro characterization of the polypeptide binding activity of these two subunits of the yeast enzyme, and show that the peptide-binding grooves in these proteins can transiently bind stretches of polypeptide with amino acid characteristics complementary to the characteristics of the grooves. We show that Ost6p, which has a peptide-binding groove with a strongly hydrophobic base lined by neutral and basic residues, binds peptides enriched in hydrophobic and acidic amino acids. Further, by introducing basic residues in place of the wild type neutral residues lining the peptide-binding groove of Ost3p, we engineer binding of a hydrophobic and acidic peptide. Our data supports a model of Ost3/6p function in which they transiently bind stretches of nascent polypeptide substrate to inhibit protein folding, thereby increasing glycosylation efficiency at nearby asparagine residues.

Introduction

Asparagine (N)-linked glycosylation is an abundant co- and post-translocational modification of secretory and membrane proteins in eukaryotes, and also occurs in archaea and some bacteria.^1–3 N-glycosylation is catalyzed in the lumen of the endoplasmic reticulum (ER) by the multiprotein complex oligosaccharyltransferase (OTase).⁴ OTase physically associates with ribosomes bound at the translocon⁵ and transfers an oligosaccharide (typically Glucose₃ Mannose₉ N-acetylglucosamine₂) from a dolichol-pyrophosphate donor to selected Asn side-chains in nascent polypeptide substrates. Glycosylation efficiency of Asn residues is greatly enhanced if they are present in glycosylation “sequons” (Asn-Xaa-Thr/Ser; Xaa≠Pro),⁴ although Asn not in sequons can in rare cases also be glycosylated.⁶ The presence of N-glycans on proteins assists productive folding in the ER intrinsically by virtue of their bulky hydrophilic nature, and also indirectly by specifically recruiting molecular chaperones.⁷ N-glycans can be modified during transport through the Golgi, resulting in highly diverse structures that can differ between organisms and are subject to regulation at the level of the cell, protein, or glycosylation site. The precise glycan structures present on a mature protein can influence protein function in a variety of ways and are important in processes including infection, immunity, cancer, and development.^8–11

Diversification and regulation of OTase activity seems to have occurred by various mechanisms throughout evolution. Organisms including some N-glycosylating bacteria and divergent eukaryotes have multiple single subunit OTase enzymes homologous to the catalytic Stt3p protein.^4,12 The various enzymes from a single organism can have different glycan and protein substrate specificities, such that gene duplication and divergence may have increased the catalytic range of efficient glycosylation.^13–17 Higher eukaryotes have evolved multiprotein complex OTases, in which subunits are involved in selection of mature glycan substrate and regulation of activity.^18–20 Duplication and divergence of multiprotein OTase subunits has also occurred, with genes encoding Stt3p and Ost3/6p proteins present in two copies in some organisms.⁴

Two isoforms of OTase are present in yeast, defined by the presence of either of the homologous proteins Ost3p or Ost6p, ²¹ that differ in their protein substrate-specific activities at the level of individual glycosylation sites.²² Ost3p and Ost6p have an ER lumenal thioredoxin-like domain followed by four transmembrane helices,²³ although other topologies have been proposed.²⁴ The crystal structure of the lumenal domain of Ost6p (Ost6L), shows a putative peptide-binding groove adjacent to its CxxC thioredoxin motif in the oxidized form of the enzyme [Fig. 1(A, B), PDB ID code 3G7Y].²³ However, this groove is not present in the reduced form, as a loop forming one face of the groove is flexibly disordered. This structural change correlates with peptide-binding activity, as a specific peptide transiently binds to Ost6L but only when its CxxC motif is oxidized.²³ A model has been proposed²³ in which Ost3p and Ost6p increase glycosylation efficiency of specific Asn in substrate proteins by transiently binding stretches of nascent polypeptide, allowing efficient glycoyslation of nearby Asn residues. A key prediction of this model is that the peptide-binding grooves of Ost3p and Ost6p will bind different stretches of polypeptide determined by amino acid sequences. Previous studies using a crude mixture of yeast peptides applied to Ost6L covalently linked to CNBr-sepharose identified a single peptide that specifically bound to the peptide-binding groove of Ost6L.²³ Here, we extended this observation and performed in vitro analyses to identify specific peptides that bind to the peptide-binding grooves of Ost3p and Ost6p from yeast. We based our analysis on the previously reported ability of Ost6L to specifically bind peptides in the oxidized but not the reduced state.²³ We tested the peptide-binding activity of wild type and variant Ost3L and Ost6L proteins using peptides from Gas1p, a yeast N-glycoprotein. Our results show that Ost3L and Ost6L bind stretches of polypeptide with complementary characteristics to their peptide-binding grooves. Further, we identified the features of these peptides and the peptide-binding grooves of Ost3/6p that determine this binding, consistent with the role of Ost3/6p in determining the in vivo protein substrate specific activity of OTase isoforms.

Figure 1.

Peptide-binding groove of S. cerevisiae Ost6p and Ost3p. Surface representation of (A) top and (B) side views of the ER lumenal domain of oxidized Ost6p (Ost6L; PDB code 3G7Y) with residues lining the peptide-binding groove coloured by hydropathy (black, hydrophobic to white hydrophilic); blue, basic; and yellow, cysteine. (C) Sequence alignment of sections of S. cerevisiae Ost6p and Ost3p, with surface-exposed residues in the Ost6p peptide-binding groove and residues of Ost3p mutated in MBP-Ost3_Q103K,Q106K variant bolded. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Results

Ost3p and Ost6p are accessory proteins of the multiprotein complex OTase and are involved in determining the specificity and activity of OTase at the level of individual glycosylation sites.²² A model of Ost3/6p function has been suggested²³ wherein stretches of nascent polypeptide transiently bind to the peptide-binding groove of Ost3p or Ost6p, enabling efficient glycosylation of nearby sequons by the active site of OTase. Previous genetic and MS analysis in yeast showed that Gas1p is a physiological substrate of Ost6p, as Ost6p is required for efficient glycosylation of N₂₅₃ in Gas1p in vivo.²² To further investigate its interactions with Ost3p and Ost6p, we expressed and purified yeast Gas1p in E. coli, to obtain unglycosylated polypeptide, similar to nascent polypeptide exiting the translocon into the ER, which is the physiological substrate of OTase. Gas1p was efficiently expressed and purified from E. coli as a periplasmically targeted MBP-fusion protein. To determine which stretches of polypeptide interacted with Ost3/6L, we digested MBP-Gas1p with trypsin, inactivated residual trypsin activity, and applied peptides to resin with bound MBP-Ost3L or MBP-Ost6L, either oxidized or reduced. Of the 35 MBP-Gas1p tryptic peptides robustly detected by MALDI-TOF-MS, three peptides were identified that were retained by resin with oxidized, but not reduced, MBP-Ost6L, indicating that they bound at the peptide-binding groove of Ost6p [Fig. 2(A, E, F, G); Supporting Information Table 1; 1057.6¹⁺, LVIWINGDK (MBP_33-41], P = 0.003; 1189.7¹⁺, AGLTFLVDLIK (MBP_216-226), P = 0.001; and 2287.2¹⁺, ALNDADIYVIADLAAPATSINR (Gas1p_106-127), P = 0.003). In contrast, no peptides were identified that showed significant retention to resin with MBP-Ost3L oxidized versus reduced [Fig. 2(B)]. Comparison of the biophysical characteristics of peptides that bound or did not bind to the peptide-binding groove of Ost6L showed that peptides with high GRAVY and aliphatic indices bound to Ost6L [Fig. 3(A, B)], while there was no difference in length or pI. Binding of hydrophobic peptides by the peptide-binding groove of Ost6L correlates with the hydrophobic amino acids Val103, Met45, Val88, Leu100, and Val95 forming the base of this groove in the crystal structure of Ost6L [Fig. 1(A, B)]. The peptides that bound to Ost6L all contained acidic residues, and this correlated with the presence of two basic residues Lys96 and Lys99 lining the Ost6L peptide-binding groove [Fig. 1(A, B)]. To test if these lysine residues contributed to specific peptide binding by Ost6L, we generated variant MBP-Ost6_K96Q,K99Q and repeated peptide binding experiments. These showed that the MBP-Ost6_K96Q,K99Q double mutant variant abolished binding of the peptides that bound to MBP-Ost6L [Fig. 2(C, E–G)]. Sequence alignment with Ost6L showed that the residues forming the base of the Ost3L peptide-binding groove were also hydrophobic [Fig. 1(C)], but that the residues lining the groove were neutral. We tested if we could engineer the peptide-binding activity of Ost3L by introducing basic lysine residues corresponding to their location in Ost6L [Fig. 1(C)]. Indeed, this MBP-Ost3_Q103K,Q106K variant showed significant binding to the 1057.6¹⁺ peptide (P = 0.001) that also bound to wild-type MBP-Ost6L [Fig. 2(D, H)].

Figure 2.

Peptide binding to S. cerevisiae (A) Ost6L, (B) Ost3L, (C) Ost6_K96Q,K99Q variant, and (D) Ost3_Q103K,Q106K variant. All plots show y-axis as relative retention of peptides by oxidized versus reduced protein, and x-axis as wash fraction (1–9). Positive relative retention values represent peptides that are retained more by oxidized than reduced proteins, indicating binding at the peptide-binding groove (see text). Retention of individual peptides with m/z of (E) 1057.6¹⁺, (F) 1189.7¹⁺ and (G) 2287.2¹⁺ to Ost6p (▪) and Ost6_K96Q,K99Q variant (□); and of (H) 1057.6¹⁺ to Ost3p (▪) and Ost3_Q103K,Q106K variant (□). Data are average of at least three independent replicates. Error bars in (E–H) show standard error. Retention of peptides is statistically significant in (E–G), Ost6p (▪); and (H), Ost3_Q103K,Q106K variant (□) (see text).

Figure 3.

Characteristics of peptides bound (b) or not bound (n) by the peptide-binding groove of Ost6p. (A) Grand average of hydropathy, (B) aliphatic index, (C) length, (D) pI. Values show mean. Error bars show standard error.

Discussion

We have performed in vitro peptide binding experiments to identify stretches of polypeptide that interact with the peptide-binding grooves of yeast OTase accessory proteins Ost3p and Ost6p. The crystal structure of Ost6L²³ shows a groove with a hydrophobic base lined by neutral and basic amino acids [Fig. 1(A, B)]. Fitting with the characteristics of this peptide-binding groove, Ost6p bound peptides rich in hydrophobic amino acids and with acidic residues. Consistent with the peptides described here (Supporting Information Table 1 and above), the peptide previously shown to bind to the peptide-binding groove of Ost6L (Cwp1p_21-31; DSEEFGLVSIR) also shares these characteristics.²³ We did not detect significant binding of peptides to wild type Ost3L. However, the variant Ost3_Q103K,Q106K, which had two neutral Gln residues lining the Ost3p peptide-binding groove replaced by the basic Lys residues present at the corresponding positions in Ost6p [Fig. 1(C)], showed increased binding of one peptide, 1057.6¹⁺, which also bound to wild type Ost6L. The amino acid residues forming the base of the peptide-binding groove which differ in Ost3p (Leu91 and Phe102) and Ost6p (Val88 and Val95) [Fig. 1(C)] are all hydrophobic, although these different residues may affect the strength and specificity of hydrophobic interactions of peptides with Ost6L and Ost3L. These and other sequence variations may contribute to the differences in peptide-binding activity that remain between Ost6L and the Ost3_Q103K,Q106K variant. Together, these data suggest that the peptide-binding grooves of Ost3p and Ost6p, and their homologues in other organisms, can transiently and noncovalently bind stretches of polypeptide with characteristics complementary to the amino acid residues in their grooves. The different peptide-binding specificities we observe for Ost3L and Ost6L is consistent with their role in determining the in vivo protein substrate specific activity of OTase isoforms previously observed in genetic experiments.^{21–23,25,26} This is consistent with the proposed model of Ost3/6p activity,²³ in that different stretches of nascent polypeptide would be bound by OTase isoforms containing Ost3p or Ost6p, which would enhance glycosylation efficiency of nearby sequons in an OTase isoform-dependent manner.

The dimensions of the peptide-binding groove in Ost6p are ∼15x10x10 Å (Fig. 1, PDB code 3G7Y²³), which corresponds well to the approximate dimensions of an extended stretch of polypeptide ∼5 amino acids long, or an alpha helix of ∼10 amino acids. Inspection of the sequences of the peptides we detected that bind to Ost6p (Supporting Information Table 1 and above) shows that the hydrophobic amino acid residues are not evenly distributed but are instead clustered in local stretches of ∼5 residues, in agreement with the size of the Ost6p peptide-binding groove. As alpha helices can form in nascent polypeptides in the translocon,²⁷ and the groove of Ost6p could accommodate a helix, amphiphatic alpha helices with one hydrophobic side could also bind at the groove in vivo.

As well as displaying transient noncovalent peptide binding activity, Ost6L is an active oxidoreductase in vitro, and mutation of its CxxC active site motif affects glycosylation of specific sites in vivo.²³ Ost3p or Ost6p may bind nascent polypeptide either noncovalently or through a mixed disulfide.²³ In this study, we have not investigated this mixed disulfide bond formation. However, noncovalent interaction of cysteine-containing polypeptide with the peptide-binding groove is likely to precede formation of a mixed disulfide with the exposed oxidized half cystine of Ost3/6p. As structural disulfide bonds are typically internal to protein domains, cysteine residues are generally located within or adjacent to stretches of hydrophobic amino acids, and so will likely be bound to the peptide-binding grooves of Ost3/6p.

In the proposed model of Ost3/6p function, transient binding of stretches of nascent polypeptide increases glycosylation efficiency of nearby sequons by increasing the time they are accessible to the OTase active site.²³ The transient noncovalent binding of hydrophobic stretches of polypeptide by Ost3/6p that we observe here would also efficiently slow protein folding of nascent polypeptide in vivo, as these hydrophobic stretches will tend to be positioned internal to folded domains. N-glycosylation by OTase requires unfolded polypeptide substrate,^28,29 and so inhibition of substrate protein folding by Ost3/6p binding sequences internal to mature protein structures is an efficient strategy to enable optimal presentation of protein substrate to the active site of OTase.

Mutations in the genes encoding the human homologues of Ost3p and Ost6p, IAP (MAGT1) and N33 (TUSC3), cause a congenital disorder of glycosylation in which the primary symptom is nonsyndromic mental retardation.^30,31 Several reports also implicate loss of N33 activity in cancer.^32–34 The work we present here provides a basis for understanding the function of Ost3/6p proteins in human development and disease, and in biotechnology.

Materials and Methods

Plasmid constructs

The DNA sequence corresponding to the respective ER lumenal domains of Ost3p and Ost6p (from the signal peptidase cleavage sites to the beginning of the first transmembrane helix) as predicted in the UniProt Knowledgebase: residues 23-185 of Ost3p (P48439, YOR085W); residues 25-188 of Ost6p (Q03723, YML019W), was amplified using PCR primers incorporating EcoRI and SalI restriction sites and cloned into the pMAL-c2x derivative of the pMAL E.coli expression vector, giving pMBP-OST3L and pMBP-OST6L, respectively. Variants with amino acid replacements (pMBP-OST6L_K96Q,K99Q and pMBP-OST3L_Q103K,Q106K) were constructed as described by Imai et al.³⁵ For Gas1p expression in E. coli, the DNA sequence corresponding to mature Gas1p (from the signal peptidase cleavage site to the GPI-anchor site) as predicted in the UniProt Knowledgebase: residues 23-528 of Gas1p (P22146, YMR307W), was cloned into a derivative of the pMAL-p2x vector, using PCR primers incorporating BamHI and HindIII restriction sites, giving pMBP-GAS1.

Protein expression and purification

E. coli TOP10 cells containing pMAL-OST3L, pMAL-OST6L, pMAL-GAS1 or variant expression plasmids were grown in LB glucose medium containing 100 μg/mL of ampicillin at 37°C until the OD_600nm reached 0.5. Cultures were supplemented with 1 mM IPTG to induce protein expression. After 1 h induction, the cells were harvested by centrifugation and resuspended in 10 mL of ice-cold 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1× Complete Mini EDTA free protease inhibitor (Roche Diagnostics) and DNase. Cells were lysed using a French Press and the lysates were passed through 0.22 μm filters before they were applied to pre-washed gravity flow columns containing amylose-agarose resin (New England Biolabs), which were washed with column buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA). MBP-Ost3L, MBP-Ost6L, and variants were retained on the amylose-agarose resin for use in peptide binding analyses. All purification steps were performed at 4°C. MBP-Gas1p was eluted with 0.1% SDS. The identity and final purity of MBP-Gas1p, MBP-Ost3p, MBP-Ost6p, and variants were confirmed by SDS-PAGE (data not shown).

Preparation of MBP-Gas1p peptides

Purified MBP-Gas1p was reduced/alkylated by treatment with 1% SDS, 10 mM DTT and 50 mM Tris-HCl pH 8, heated at 95°C for 10 min, followed by addition of acrylamide to 25 mM and incubation at 20°C for 16 h. Protein was precipitated and washed as previously described.³⁶ The pellet was air dried, resuspended in column buffer, and digested with trypsin at 37°C for 16 h. After the digest, residual trypsin activity was eliminated by addition of PMSF to 1 mM and incubation at 20°C for 30 min.

Peptide binding analysis

Peptide binding to resin containing bound MBP-Ost3L, MBP-Ost6L, or variants was performed essentially as previously described.²³ Peptides were applied to resin, gravity flow washed with column buffer and 600 μL fractions collected. All steps were performed at 4°C. MBP-Ost3/6/variant proteins pre-reduced by incubation with DTT were used as negative controls. Peptides in each fraction were desalted with C-18 ZipTips (Millipore), eluted in 2 μL (2 mg/mL α-Cyano-4-hydroxycinnamic acid in 70% Acetonitrile) on to the MALDI target plate and analyzed by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-Tof/Tof-MS). Binding to MBP-Ost3L, MBP-Ost6L, or variants were analyzed by three biological replicates.

MALDI-Tof/Tof-MS analysis

Peptides in load and wash fractions were measured using a Bruker Daltonics autoflex III smartbeam MALDI-Tof/Tof instrument in positive ion mode, averaging 2000 laser shots with an m/z range of 700–4000 using external calibration. Data were analyzed as described²³ with some variations. The relative enrichment of each peptide in each wash fraction was determined by the abundance of the peptide measured by the signal to noise ratio as a fraction of the total abundance of all detected peptides in that fraction, normalized to the relative abundance of that peptide in the load. Interactions with the peptide-binding groove of Ost3L, Ost6L, or variants were determined for each peptide by the difference in the peptide's relative abundance in each wash fraction in oxidized and reduced versions of the same protein. Retention of peptides by oxidized versus reduced protein was tested by comparison of the relative abundance in reduced and oxidized conditions of the peptide in each wash fraction using two-sided paired student's t-test with P < 0.005. This analysis makes use of the only change in the crystal structure of Ost6L between oxidized and reduced states being in the loop forming one side of the peptide-binding groove (PDB ID codes 3G7Y and 3G9B).²³ This means that peptides bind at the peptide-binding groove of Ost3/6p if they are retained more strongly by resin containing oxidized Ost3/6L protein than by reduced Ost3/6L protein.²³ Peptide characteristics were determined using ProtParam (http://expasy.org/tools/protparam.html).

Glossary

Abbreviations:

ER

endoplasmic reticulum

MALDI-Tof/Tof-MS

Matrix assisted laser desorption/ionization tandem time-of-flight mass spectrometry

N-glycosylation

asparagine-linked glycosylation

Ost3/6L

ER lumenal domain of Ost3/6p

OTase

oligosaccharyltransferase.