An engineered high affinity Fbs1 carbohydrate binding protein for selective capture of N-glycans and N-glycopeptides

Abstract

A method for selective and comprehensive enrichment of N-linked glycopeptides was developed to facilitate detection of micro-heterogeneity of N-glycosylation. The method takes advantage of the inherent properties of Fbs1, which functions within the ubiquitin-mediated degradation system to recognize the common core pentasaccharide motif (Man3GlcNAc2) of N-linked glycoproteins. We show that Fbs1 is able to bind diverse types of N-linked glycomolecules; however, wild-type Fbs1 preferentially binds high-mannose-containing glycans. We identified Fbs1 variants through mutagenesis and plasmid display selection, which possess higher affinity and improved recovery of complex N-glycomolecules. In particular, we demonstrate that the Fbs1 GYR variant may be employed for substantially unbiased enrichment of N-linked glycopeptides from human serum. Most importantly, this highly efficient N-glycopeptide enrichment method enables the simultaneous determination of N-glycan composition and N-glycosites with a deeper coverage (compared to lectin enrichment) and improves large-scale N-glycoproteomics studies due to greatly reduced sample complexity.

Figure 1. Fbs1 binds to diverse types of N-glycomolecules.

(a) Fbs1 binding to RNase B is N-glycan dependent. RNase B or deglycosylated RNase B was subjected to an Fbs1 pulldown assay and analysed by SDS–PAGE. Lane 1, RNase B input control (CTL). Lane 2, essentially no RNase B deglycosylated by PNGase F is pulled down by Fbs1. Lane 3, RNase B with N-glycans is efficiently pulled down by Fbs1 beads. A representative SDS–PAGE gel is shown from two experiments. (b) Fbs1 binds to the N-glycosylated heavy chain of human IgG. Denatured and reduced human IgG were subjected to an Fbs1 pulldown assay and analysed by SDS–PAGE. Heavy chains of human IgGs are typically N-glycosylated. Lane 1 is a control showing the IgG light chain and heavy chain. Lane 2 is Fbs1 beads only. Some SNAP-Fbs1 protein leaches from the prototype Fbs1 beads (denoted by an asterisk). Lanes 3 and 4 show that only the glycosylated heavy chain is bound by Fbs1 beads. A representative SDS–PAGE gel is shown from two experiments. (c) Fbs1 binding affinity (Kd value) to sialylglycopeptide (SGP), M3N2 and M3N2F was measured by isothermal titration calorimetry. Structures of SGP, M3N2 and M3N2F are shown in the left panel. M3N2F is M3N2 with α1-6 fucosylation at the reducing end GlcNAc. The left panel summarizes the Kd values of SGP (n=4), M3N2 (n=5) and M3N2F (n=5) interacting with wt Fbs1. There is no significant difference between the Kd values of M3N2 and M3N2F (P value 0.85>0.05, mean±s.e.m., t-test, two-tailed). (d) wt Fbs1 shows binding bias to different N-glycopeptides. SGP was labelled with TMR fluorophore to facilitate detection. N-glycans of SGP-TMR (1) were then trimmed with exoglycosidases to produce asialo-SGP-TMR (2), SGP-TMR without sialic acids and galactose (3) and SGP-TMR without sialic acids, galactose and GlcNAc (4). Binding of the trimmed glycopeptides to Fbs1 beads was analysed. The relative binding affinity to wt Fbs1 is reported as the recovery percentage (TMR fluorescence on beads/input TMR fluorescence). For simplicity, TMR is only indicated in N-glycopeptide structure 1. Results represent the mean±s.e.m. of three replicates.

Figure 2. High-salt conditions increase complex N-glycomolecule binding to wt Fbs1.

(a) The presence of 2 M NaCl increases SGP-TMR binding to wt Fbs1 in an N-glycan-dependent manner. PNGase F (+) indicates SGP-TMR was pretreated with PNGase F to cleave the glycan from the fluorophore-labelled peptide (sequence KVANKT). SGP-TMR with or without PNGase F treatment was incubated with Fbs1 beads in low-salt (LS) conditions or high-salt (HS) conditions. SGP-TMR binding to Fbs1 beads was measured, and affinity to Fbs1 is indicated by percentage of recovery (amount of bound SGP-TMR/amount of input SGP-TMR). Results represent the mean±s.e.m. of three replicates. (b) HS conditions increase Fbs1 binding to sialylated fetuin relative to RNase B, which contains high-mannose N-glycans. A mixture of denatured fetuin and RNase B was subjected to an Fbs1 bead pulldown assay. Lane 1 indicates the input ratio of fetuin to RNase B. Lanes 2 and 3 show the amounts of fetuin and RNase B pulled down by Fbs1 beads in LS and HS conditions. Asterisk denotes a small amount of SNAP-Fbs1 that leaches from the Fbs1 beads. N-glycan structures present within fetuin and RNase B are illustrated. A representative SDS–PAGE gel is shown from two experiments. (c) Reciprocal pulldown of SNAP-Fbs1 by denatured fetuin or RNase B beads in LS or HS conditions. A representative SDS–PAGE gel is shown from two experiments. (d) HS conditions have no effect on Fbs1 binding to asialo-SGP-TMR. SGP-TMR was trimmed with α2-3,6,8 Neuraminidase to produce asialo-SGP-TMR (structures shown in Fig. 1d, glycopeptide 1 and 2). SGP-TMR and asialo-SGP-TMR were incubated with Fbs1 beads in LS buffer or HS buffer. SGP-TMR or asialo-SGP-TMR relative affinity to Fbs1 is indicated by the recovery percentage. Results represent the mean±s.e.m. of three replicates.

Figure 3. Isolation of Fbs1 mutants with improved binding affinity for complex N-glycans.

(a) Schematic illustration of p50-Fbs1 plasmid display. (1) p50-Fbs1 fusion protein is expressed in the E. coli cytosol (driven by the lac promoter (P_Lac)). The encoding plasmid contains p50 binding sites (bs). (2) p50-Fbs1 protein stably binds to the encoding plasmid via the tight interaction between p50 and its binding sites. (3) Upon gentle cell lysis, the p50-Fbs1-encoding plasmid complex is selected by the interaction between Fbs1 and complex N-glycan of fetuin (immobilized on a solid support, Affi-Gel 15). After stringent washing, Fbs1 variants with high affinity to complex N-glycans are enriched. The plasmid DNA is extracted and transformed into E. coli cells for the next selection cycle. (b) The table lists the mutants of human Fbs1 that were obtained from plasmid display screening and alanine scanning. Top two rows: the amino acid residues within mouse Fbs1 that correlate to human Fbs1. The amino acid (a. a.) position is numbered according to the full-length sequence of each respective protein. PPG, PPS, PPR and YR mutants were obtained by plasmid display screen. S155A was obtained by alanine scanning, and S155G was a further mutation based on S155A. (c) Relative affinity of Fbs1 variants to complex N-glycans as determined by a fetuin bead pulldown assay. E. coli cell lysate containing the same amount of wt Fbs1 or variant Fbs1 fusion protein was subjected to a fetuin bead pulldown assay. The bound Fbs1 protein was analysed by SDS–PAGE (upper panels) and quantified by ImageJ. The amount of bound variant Fbs1 protein was standardized to that of bound wt Fbs1 and the fold change was calculated (bottom panels, bar graphs). The relative affinity to fetuin of an Fbs1 mutant is indicated by the fold change. A representative SDS–PAGE gel is shown from two experiments. (d) Combination of the mutations in c results in Fbs1 variants with even higher affinity to complex N-glycans. All assays are the same as in c. A representative SDS–PAGE gel is shown from two experiments.

Figure 4. Fbs1 GYR and PPRYR variants display reduced binding bias between high-mannose and complex N-glycans.

(a) Comparison of N-glycoprotein pulldown by wt Fbs1, Fbs1 GYR and PPRYR variant proteins. A mixture of denatured RNase B and fetuin was subjected to an Fbs1 pulldown assay with wt, GYR and PPRYR Fbs1 beads in low salt (50 mM ammonium acetate, pH7.5) and high salt (2M ammonium acetate, pH7.5). All three Fbs1 bead types were conjugated with the same amount of the respective Fbs1 protein (Supplementary Fig. 1). Left panel is the SDS–PAGE gel showing the bound (Lanes 1–6) and input ratio (Lane 7) of RNase B and fetuin. An asterisk denotes the SNAP-Fbs1 protein leaching from the Fbs1 beads. Right panel shows the recovery percentage (bound protein amount/input protein amount) of each substrate glycoprotein using the different conditions. A representative SDS–PAGE gel is shown from two experiments. (b) Fbs1 GYR variant binding to a diverse set of N-glycopeptides is substantially unbiased. The experiment in Fig. 1d was repeated using Fbs1 GYR beads. The data shown in Fig. 1d are presented in this figure to facilitate the comparison between wt Fbs1 and Fbs1 GYR. N-glycans of SGP-TMR (1) were trimmed with different combinations of exoglycosidases to produce asialo-SGP-TMR (2), SGP-TMR without sialic acids and galactose (3) and SGP-TMR without sialic acids, galactose and GlcNAc (4). Identities of the trimmed SGP-TMR derivatives were confirmed by LC-MS. The trimmed glycopeptides were then added to binding assays with wt Fbs1 or Fbs1 GYR beads in 50 mM ammonium acetate pH7.5. The relative binding affinity to wt Fbs1 or Fbs1 GYR is reported as the recovery percentage (TMR fluorescence on beads/input TMR fluorescence). For simplicity, TMR is not shown in the N-glycopeptide structures (1–4). Results represent the mean±s.e.m. of three replicates.

Figure 5. The Fbs1 GYR variant substantially improves N-glycopeptide enrichment.

The total ion chromatogram (TIC) (upper panel) from an LC-MS analysis shows wt Fbs1 or Fbs1 GYR-mediated binding and enrichment of N-glycopeptides from a complex peptide mixture. The complex peptide mixture is a tryptic digest of RNase B spiked with SGP and SGP-TMR. RNaseB contains non-glycosylated peptides and two major high-mannose N-glycopeptides labelled in the enlargement as: M5N2-NLTK and M6N2-NLTK. M5N2 or M6N2 indicates 5 or 6 mannose residues and 2 GlcNAc residues, respectively. NLTK is the peptide sequence of the N-glycopeptide produced by trypsin treatment of RNase B. The enrichment was performed in low salt (50 mM ammonium acetate, pH7.5). The black line indicates the chromatogram of a 50% input mixture. The orange and blue lines indicate the chromatograms of Fbs1 GYR and wt Fbs1 enrichment samples. The major N-glycopeptide peaks (M5N2-NLTK, M6N2-NLTK, SGP and SGP-TMR) are indicated. N-glycopeptides were quantified from the extracted ion chromatogram of the LC-MS analysis. The ions with the correct monoisotopic m/z values, that is, M5N2-NLTK:1691.98^1- (theoretical, 1691.72^1-), M6N2-NLTK: 1854.07^1-(theoretical,1853.72^1-) and SGP: 1433.47^2- (theoretical, 1433.10^2-) were extracted, integrated and quantified. The amount of SGP-TMR was determined by fluorescence measurement of the LC elution. Recovery of each N-glycopeptide (enriched peptide amount/input peptide amount) is shown as a bar graph (lower panel). A representative TIC profile is shown from three experiments.

Figure 6. N-glycopeptide enrichment from human serum tryptic peptides by Fbs1 GYR.

(a) Left panel: The total ion chromatogram (TIC) from an LC-MS analysis of HSA-depleted human serum tryptic peptides without Fbs1 enrichment (pre-enrichment) or the peptides from Fbs1 GYR enrichment. Right panel: TIC from an LC-MS analysis of peptides enriched by Fbs1 GYR treated with active PNGase F (blue curve) or heat-inactivated PNGase F (95 °C, 10 min) (orange curve). A representative TIC profile is shown from three experiments. (b) N-glycan profiling of human serum (-HSA) tryptic peptides without Fbs1 enrichment (pre-enrichment, dark curve) or the peptides from Fbs1 GYR Enrichment (red curve) by 2-AB labelling and UPLC. Glycan structures are assigned to major peaks according to the glucose unit of each peak (Supplementary Fig. 8). The reduced peaks in the Fbs1 GYR enrichment sample are labelled with an ‘*' The red curve is offset for better comparison with the black curve. A representative glycan profile is shown from two experiments. (c) N-glycosite identification using PNGase F deglycosylation in ¹⁸O water. HSA-depleted human serum tryptic peptides without Fbs1 enrichment (pre-enrichment) or enriched by Fbs1 GYR (Fbs1 GYR enrichment) were deglycosylated by PNGase F in ¹⁸O water. Upon N-glycan removal, asparagine (N) in the N-glycosite is deamidated to aspartic acid resulting in a peptide tagged with an additional 2.988 daltons. Peptide spectra with N[+2.988] within the canonical N-glycosylation motif N-X-T/S can be confidently assigned as N-glycopeptides. Detailed MS spectrum information is in Supplementary Data 1 and 2. The MS data were combined from two mass spectrometric experiments. (d) Comparison of unique N-glycosites and N-glycoproteins identified without (pre-enrichment) or with Fbs1 GYR enrichment. Left panel: 183 and 477 unique N-glycosites were identified in pre-enrichment and Fbs1 GYR enrichment, respectively. One hundred and seventy-two (94%) of the 183 N-glycosites in the pre-enrichment were also identified in the Fbs1 GYR enrichment sample. Right panel: 89 and 230 N-glycoproteins were identified in pre-enrichment and Fbs1 GYR enrichment, respectively. Eighty-three (93%) of the 89 N-glycoprotein in the pre-enrichment were also identified in the Fbs1 GYR enrichment sample. Detailed N-glycosite and N-glycoprotein information is in Supplementary Data 1 and 2.

Figure 7. Fbs1 GYR enrichment greatly improves intact N-glycopeptide identification.

(a) Intact N-glycopeptide identification using Fbs1 GYR enrichment, lectin enrichment or no enrichment (pre-enrichment, pre-). MS data were obtained from three mass spectrometric experiments (n=3). The average number of MS spectra of intact N-glycopeptides (orange) and non-N-glycopeptides (blue) are shown in the bar graph. The number of N-glycopeptide spectrum from Fbs1 GYR enrichment was significantly higher relative to lectin enrichment (***P<0.001, mean±s.e.m., t-test, two-tailed). Detailed intact N-glycopeptide MS data is in Supplementary Data 3–5. (b) Comparison of unique intact N-glycopeptides and unique N-glycosites identified from pre-enrichment, lectin enrichment and Fbs1 GYR enrichment. The numbers were generated from combining three intact N-glycopeptide MS data sets (Supplementary Data 3–5). Peptide miscleavage was manually corrected. (c) Comparison of N-glycosylation micro-heterogeneity identified from pre-enrichment, lectin enrichment and Fbs1 GYR enrichment. The overall determination of N-glycosylation micro-heterogeneity is indexed by glycan type per glycosite (number of unique intact N-glycopeptides/number of unique N-glycosites) and spectral count per glycosite (number of spectral counts/number of unique N-glycosites). (d) Micro-heterogeneity of N-glycosylation in human Complement C3 protein (as an example) illustrated by intact N-glycopeptide identification without enrichment (pre-enrichment) or using Fbs1 GYR enrichment and lectin enrichment. The orange row indicates N-glycoprotein identification. Beneath protein identification, N-glycosites are listed (in green rows). Beneath N-glycosites, N-glycan compositions attached to this N-glycosite are listed (in red). N@ indicates the asparagine with N-glycan modification. The light blue columns indicate the spectral counts of each N-glycoform. A full list of N-glycosylation micro-heterogeneity illustrated by intact N-glycopeptide identification is shown in Supplementary Data 6.