Engineering of a small protein scaffold to recognize sulfotyrosine with high specificity

Abstract

Protein tyrosine O-sulfation is an essential posttranslational modification required for effective biological processes such as hemostasis, inflammatory response, and visual phototransduction. Due to its unstable nature under mass spectrometry conditions and residing on low abundance cell surface proteins, sulfated tyrosine (sulfotyrosine) residues are difficult to detect or analyze. Enrichment of sulfotyrosine-containing proteins (sulfoproteins) from complex biological samples are typically required before analysis. In this work, we seek to engineer the phosphotyrosine binding pocket of a Src Homology 2 (SH2) domain to act as an anti-sulfotyrosine antibody mimic. Using tailored selection schemes, several SH2 mutants are identified with high affinity and specificity to sulfotyrosine. Further molecular docking simulations highlight potential mechanisms supporting observed characteristics of these SH2 mutants. Utilities of the evolved SH2 mutants were demonstrated by the detection and enrichment of sulfoproteins.

Protein tyrosine O-sulfation (PTS) is one of the most common type of tyrosine posttranslational modifications (PTMs).¹ It is present exclusively on transmembrane and secreted proteins.^{2, 3} Two tyrosyl-protein sulfotransferase (TPST) enzymes within the trans Golgi, TPST-1 and TPST-2, catalyze the sulfation of tyrosine residues in the presence of the universal sulfate donor 3’-phosphoadenosine 5’-phosphosulfate (PAPS; Figure 1).^{4, 5} Sulfated tyrosine (sTyr) residues are an important recognition element facilitating many essential protein-protein interactions in a variety of biological processes involved in hemostasis,⁶ visual phototransduction,^{7, 8} reproduction⁹ and inflammatory response.¹⁰ In addition, protein tyrosine O-sulfation has been implicated in the development of infectious diseases,^11–13 autoimmune diseases,^14–16 and cancers.^{17, 18}

Figure 1. Protein tyrosine O-sulfation.

This type of protein posttranslational modification entails the transfer of a sulfate group from 3’-phosphoadenosine 5’-phosphosulfate (PAPS) to a tyrosine residue to form sulfotyrosine (sTyr) under the catalysis of tyrosylprotein sulfotransferases (TPSTs).

Up to 1% of tyrosine residues in animals are predicted to be sulfated,^{19, 20} yet biomolecular probes for the elucidation, enrichment, and characterization of these tyrosine sulfated proteins (sulfoproteins) have been inadequate. Antibodies of high affinity and selectivity have been essential in the analysis of PTMs. For example, anti-phosphotyrosine (pTyr) antibodies provide means of selective isolation of pTyr-containing proteins/peptides from complex cellular samples for further downstream analysis.^21–23 In addition, high affinity anti-pTyr antibodies can function as potential diagnostic tools and therapeutics for various human diseases.²⁴ Similarly, anti-sTyr antibodies are commercially available from multiple vendors and mostly based on two clones, Sulfo-1C-A and 7C5.^{25, 26} Both antibodies were identified using in vitro phage display as previous attempts at immunization to identify anti-sTyr antibodies proved unsuccessful likely due to the essential nature of PTS in vivo.²⁵ While these antibodies are specific for sTyr over pTyr, only moderate affinity was observed.^25–27 Due to the challenging nature of evolving anti-sTyr antibodies,²⁶ our group looked to the Src Homology 2 (SH2) domain, a small modular domain of ~100 amino acids, to engineer an anti-sTyr antibody mimic.²⁷ The significantly smaller size and reduced cost of production of an engineered sTyr-recognizing SH2 domain would provide a useful alternative to currently available antibodies. Indeed, evolution of the pTyr-binding pocket resulted in several SH2 domain variants identified with high affinity for sTyr.²⁷ Unfortunately, the retention of high affinity for pTyr limited the application of these evolved SH2 domain for targeted enrichment and identification of sulfoproteins in complex biological samples. Here we report to further engineer an SH2 domain to display either improved affinity or specificity for sTyr. Furthermore, we seek to identify potential interactions which may facilitate more specific recognition of sTyr through molecular docking simulations.

RESULTS AND DISCUSSION

Library construction.

As a family of over 100 different members, SH2 domains are composed of a beta-sheet sandwiched between two alpha helices which recognize pTyr within varying acidic motifs.^{28, 29} With no known consensus sequence, PTS has also been shown to predominantly occur within acidic sequences further supporting the application and engineering of the SH2 domain as an anti-sTyr antibody mimic.³⁰ The conserved pTyr-binding site is composed of several residues forming a shallow basic pocket capable of recognizing the pTyr’s phosphate group (Figure 2).²⁸ In order to switch the specificity of SH2 domain from pTyr to sTyr recognition, we used, SH2-tm (T40V, C45A and K63L),^{27, 31} as the starting point and randomized four residues (R35, S37, E38 and T39) that interact with the phosphate group of pTyr (Figure 2).²⁸ These four residues were randomized to all 20 amino acids using the NNK degenerate codon to produce a theoretical library diversity of 1.05 × 10⁶. The coverage of the constructed SH2 library was verified using DNA sequencing (Figure S1).

Figure 2. The pTyr-binding pocket of the SH2 domain.

(A) Crystal structure (PDB: 1SPS) of pTyr-binding pocket of SH2-wt bound to pTyr. Residues in red were selected for randomization. (B) Sequence alignment of SH2-wt and SH2-tm mutant. Bolded residues were mutated in SH2-tm. Underlined residues were selected for randomization.

Monovalent display of SH2 variants fused to pIII viral coat proteins on the surface of M13KO7 phage was used in our previous work for SH2 domain engineering.²⁷ Unfortunately, we were unable to identify any SH2 mutants that could differentiate sTyr-containing peptide (sulfopeptide) from pTyr-containing peptide (phosphopeptide).²⁷ In an effort to overcome such problem in the present work, we explored multivalent display of SH2 mutants using the hyperphage system.³² In comparison to the traditional phage display that only displays a single copy of protein of interest on up to 1% of phage, hyperphage display offers advantages by displaying five copies of protein of interest on 100% of phage.³³ Indeed, an increased display of SH2 by hyperphage was observed by using an immunoassay that qualitatively identified SH2-tm-pIII fusion proteins displayed on the surface of phage variants (Figure S2). To generate phage library, genes encoding SH2 mutants were fused to a full length pIII gene via a cleavable linker, which allows the re-generation of functional pIII protein for phage infection after biopanning.³²

Biopanning.

Hyperphages displaying the SH2 mutant library were panned against biotinylated sulfopeptides or phosphopeptides, respectively (Figure 3B). A total of three selection schemes were employed as outlined in Figure 3A. Three rounds of panning were carried out for each selection scheme. Selection scheme I was restricted to positive selections only against the sulfopeptide with non-specific elution of bound phage. As such, selection scheme I was likely to only identify SH2 mutants with improved affinity for the sulfopeptide. The remaining two selection schemes employed techniques to identify SH2 mutants that are more specific for the sulfopeptide over the phosphopeptide. Selection scheme II included one round of negative selection against the phosphopeptide to eliminate SH2 mutants with retained affinity for pTyr. While selection scheme III only used positive selections against the sulfopeptide, a competitive elution method using phenyl sulfate was used to release bound phage. We hypothesized that competitive elution of SH2 mutants displayed on hyperphage by phenyl sulfate would more selectively release SH2 variants with high specificity for the sulfopeptide. Given the avidity effect of the hyperphage system, traditional monovalent phage display was used in the final round of panning in order to identify SH2 mutants with the most improved affinity for sulfopeptides.

Figure 3. Biopanning.

(A) Three different biopanning schemes that were used in this work. The major differences among three schemes are bolded. Multivalent display used hyperphage and monovalent display used M13KO7. Non-specific elution used trypsin to cleave away bound SH2 and release phage. Specific elution used phenyl sulfate to compete with bound SH2-pIII phage and release phage complex. (B) Structures of sTyr, pTyr, and peptides for selection and characterization.

Characterization of hits.

Ten SH2 mutants from each selection scheme were qualitatively analysed for their binding affinities using phage ELISA (Figure S3). Sixteen out 30 screened SH2 variants (80%, 20%, and 60% of SH2 mutants from selection scheme I, II, and III, respectively) showed moderate to high affinity for sulfopeptides. Among all examined hits, SH2-3.1, obtained from selection scheme III, was the only mutant with apparent specificity for the sulfopeptide over the phosphopeptide. Interestingly, the inclusion of a negative selection (scheme II; Figure 3A) proved unsuccessful for conferring specificity for the sulfopeptide as it appeared to facilitate the removal of all SH2 mutants with any affinity for sulfopeptide, including those with potentially low affinity for phosphopeptide but with high affinity for sulfopeptide.

According to DNA sequencing results (Table S1), an alanine mutation at position 37 was commonly observed among SH2 mutants with improved affinity relative to the starting template SH2-tm. Previous mutation studies of SH2 support improved sulfopeptide affinity of SH2 with the mutation S37A.³⁴ In contrast, those hits selected with no improvement or reduced affinity favored valine at this position. Selection of basic residues (i.e., arginine or lysine) were favored for SH2 mutants with improved affinity for the sulfopeptide. All selected SH2 mutants retained essential arginine at position 35. Four SH2 mutants, SH2-1.5 (S37A and T39G), SH2-1.8 (S37A and T39R; with highest affinity to sulfopeptides), SH2-3.1 (S37H, E38P, and T39F; with highest specificity to sulfopeptides), and SH2-3.10 (S37A and T39K) were selected for further quantitative binding affinity analysis, molecular docking, and downstream applications.

Quantitative binding affinity analyses using fluorescence polarization assays were used to determine dissociation constants of wild-type SH2 (SH2-wt), SH2-tm, a previously evolved SH2-60.1 mutant,²⁷ and SH2 mutants identified from this work including SH2-1.5, SH2-1.8, SH2-3.1 and SH2-3.10 (Figure 4A & Figure S4). In addition to fluorescein-labeled sulfopeptide and phosphopeptide (Figure 3B), a non-modified peptide was included to assess the specificity for modified tyrosine residue. All selected and purified (Figure S5) SH2 mutants displayed improved affinity for sulfopeptides as compared to SH2-wt and SH2-tm (Figure 4A). Mutant SH2-1.5 had a 17-fold improvement in affinity for sulfopeptides as compared to SH2-wt. SH2-1.5 retained greater affinity for phosphopeptides over sulfopeptides. Mutant SH2-1.8 had the greatest affinity for sulfopeptides exhibiting a 1.5-fold improvement in observed affinity for sulfopeptides over the previously reported SH2-60.1 mutant. SH2-1.8 also displayed a comparable affinity for both sulfopeptides and phosphopeptides. The retained high affinity for phosphopeptides is not surprising since SH2-1.5 and SH2-1.8 were identified in selection scheme I that only contained positive selections (Figure 3A). On the other hand, the change in specificity observed for SH2-3.1 was confirmed with a greater than 15-fold improvement in affinity for sulfopeptides over phosphopeptides (Figure 4A). Remarkably, the observed dissociation constant of SH2-3.1 for sulfopeptides also showed over a 40-fold improvement as compared to SH2-wt. As a second comparison, the affinity of SH2-3.1 to its sulfopeptide substrate is about 2.5-fold higher than that of SH2-wt to its phosphopeptide substrate. Surprisingly, mutant SH2-3.10, identified in selection scheme III, also displayed a change in specificity with a 1.9-fold improvement in affinity for sulfopeptides over phosphopeptides (Figure 4A). A greater than 90-fold improvement in affinity for sulfopeptides as compared to SH2-wt is observed for SH2-3.10, however, significant affinity is retained for phosphopeptides (Figure 4A). All identified SH2 mutants appeared more specific than SH2-60.1 towards peptides containing modified tyrosine over the peptide containing non-modified tyrosine (Figure 4A). To better understand the function of each selected residue within mutant SH2-3.1, four additional SH2 variants were purified (Figure S6) and characterized: SH2-S37H, SH2-E38P, SH2-T39F and SH2-S37H/E38P (Figure 4B & Figure S4). Based on observed dissociation constants, the S37H mutation appears to modulate sulfopeptide specificity of SH2-3.1. Mutants SH2-S37H and SH2-S37H/E38P exhibited 1.4-fold and 6.6-fold improvement in affinity for sulfopeptides over phosphopeptides, respectively (Figure 4B). Mutations E38P and T39F are nonetheless essential for further improving the affinity for sulfopeptides. SH2-E38P displayed a 17-fold improvement in affinity for sulfopeptides as compared to SH2-wt. SH2-T39F had comparable affinities for both sulfopeptides and phosphopeptides with a greater than 35-fold improvement in affinity for sulfopeptides as compared to SH2-wt. It should be noted here that the absolute K_d values of SH2-wt, SH2-tm and SH2-60.1 in this work are different from our previous work. However, the relative binding strengths among different SH2 variants in this work are consistent from multiple measurements.

Figure 4. Quantitative binding analysis of selected SH2 mutants.

(A) Dissociation constants of evolved SH2 variants to FITC-labeled peptides; (B) Dissociation constants of SH2-3.1 back mutants to FITC-labeled peptides. Data fitting and plot are shown in Figure S4. R² for all determined K_d above 0.97. Reported data are the average measurement of three samples with standard deviations.

Modeling.

Comparative modeling³⁵ and molecular peptide docking³⁶ simulations of the identified mutants bound to either sulfopeptide or phosphopeptide revealed the effects of mutations within the binding pocket (Figure 5). SH2 mutants, SH2-1.5, SH2-1.8 and SH2-3.1, were modeled using an available SH2-tm crystal structure (PDB:4F5B). The top model for each SH2 mutant was docked with either a sulfopeptide or a phosphopeptide. An expanded binding pocket is afforded to SH2-1.5 with mutations S37A and T39G (Figure 5A–C). No significant changes in the conformation of sTyr or pTyr are observed as compared to bound pTyr in SH2-wt (PDB:1SPS). The sTyr residue within the binding pocket of SH2-1.8 adopted a different conformation than that of pTyr as observed in SH2-wt (PDB:1SPS) when bound to pTyr (Figure 5F). Similar to SH2-1.5, the S37A mutation allowed for an expanded binding pocket for SH2-1.8. The T39R mutation within SH2-1.8 is seen enclosing the bound sTyr in Figure 5D. Flexibility of side chain confirmations of T39R, observed for phosphopeptide docking simulations within the SH2-1.8 binding pocket, underline potential function in both high affinity interaction with bound modified tyrosine and as binding pocket ‘lid’ upon peptide binding. The mutations in SH2-3.1 binding pocket, S37H, E38P and T39F, reduced the binding pocket size preventing deep binding of sTyr and forcing a closer interaction with R15 (Figure 5G). The imidazole ring of mutation S37H significantly obstructs the binding pocket displaying close interaction with bound sTyr (Figure 5G). These docking results compliment observed binding affinities of SH2-3.1 back mutants, SH2-S37H, and SH2-S37H/E38P. The reduced binding pocket size of SH2-3.1 may explain the improved specificity for sTyr over pTyr. The electron density cloud of the phosphate of pTyr at pH 7.4 is significantly larger than that of the sulfate of sTyr.³⁷ The reduced electron density of sTyr can be better accommodated within the smaller binding pocket of SH2-3.1, allowing for essential interactions with R15, R35, and H37 to be maintained. Confirmation of these results through structure determination in the future may further expose the mechanisms of interaction conferring improved binding properties of SH2 mutants for sTyr.

Figure 5. Molecular modeling of binding interaction between SH2 and bound peptide using RosettaCommons.

(A) SH2-1.5 bound to phosphopeptide; (B) SH2-1.5 bound to sulfopeptide; (C) SH2-1.5 bound to sulfopeptide superimposed with SH2-wt bound to phosphopeptide; (D) SH2-1.8 bound to phosphopeptide; (E) SH2-1.8 bound to sulfopeptide; (F) SH2-1.8 bound to sulfopeptide superimposed with SH2-wt bound to phosphopeptide; (G) SH2-3.1 bound to phosphopeptide; (H) SH2-3.1 bound to sulfopeptide; (I) SH2-3.1 bound to sulfopeptide superimposed with SH2-wt bound to phosphopeptide. SH2-wt, PDB:1SPS.

Detection of sulfoproteins.

To demonstrate their utilities, engineered SH2-1.8 and SH2-3.1 mutants were first applied to immunoassays (Far-Western blot) for the identification of sulfoproteins. A naturally tyrosine-sulfated protein, complement 4 (C4), was employed. It contains a posttranslationally sulfated tyrosine residue at position 1422. The peptide fragment (1418–1428; EDYEYDELPAK) of C4, which contains a different amino acid sequence from the peptide used for selection, was fused to a glutathione S-transferase protein to yield C4-Tyr-GST. The installation of sTyr at position Y1422 was achieved through unnatural amino acid mutagenesis,³⁸ which afforded C4-sTyr-GST. Both C4-Tyr-GST and C4-sTyr-GST were purified through affinity chromatography using GSH-Sepharose resin (Figure S7). As shown in Figure 6A, C4-sTyr-GST could be readily detected by both SH2-1.8 and SH2-3.1 at all concentrations examined. As a control, C4-Tyr-GST was not detected when 0.1 μg/mL or 0.05 μg/mL SH2 mutants were used (Figure 6A). As another control, chemically phosphorylated BSA (BSA-P) was not detected when 0.1 μg/mL or 0.05 μg/mL of SH2-3.1 was used. It should be noted that BSA was chemically phosphorylated and contains multiple pTyr residues.

Figure 6. Identification and enrichment of sulfoproteins.

(A) Dot blot analysis of effective concentrations and substrate specificity of SH2 domain mutants. 1 μg of designated protein detected using various concentrations of SH2-1.8 or SH2-3.1; (B) Dot blot analysis of limit of detection. Serial dilutions of C4-sTyr-GST were detected by 1 μg/mL of SH2-1.8, SH2-3.1, or 1/1000 dilution of anti-sTyr monoclonal antibody; (C) Enrichment experiments. Lanes 2 to 5, enrichment of sulfoprotein, C4-sTyr-GST, in the presence of 100-fold BSA using engineered SH2 mutants. Lanes 6 to 9, enrichment of non-sulfated protein, C4-Tyr-GST, in the presence of 100-fold BSA using engineered SH2 mutants. Lane 1, molecular weight marker; Lane 2 & 6, pre-enrichment mixture; Lane 3 & 7, enrichment with SH2-1.8; Lane 4 & 8, enrichment with SH2-3.1; Lane 5 & 9, enrichment without SH2 mutants; (D) Selective enrichment from 1:1 BSA-P:C4-sTyr-GST mixture. Non-treated denotes no phosphatase treatment prior to enrichment. Treated denotes 16 hrs of phosphatase treatment prior to enrichment. Lane 1, protein ladder; Lane 2 & 5, pre-enrichment mixture; Lane 3 & 6, enrichment using SH2-1.8; Lane 4 & 7, enrichment using SH2-3.1.

To further demonstrate the excellent affinity of SH2 mutants for sTyr, serial dilutions of C4-sTyr-GST detected by either SH2-1.8, SH2-3.1, or anti-sTyr monoclonal antibody were used to identify qualitative limit of detections for each variant, respectively (Figure 6B). All immunoassays were performed with the same amount of immobilized protein and substrate exposure time of two minutes and thirty seconds. As shown in Figure 6B, SH2-1.8 displayed the best affinity, followed by SH2-3.1. In contrast, the anti-sTyr antibody was only able to detect C4-sTyr-GST at the highest concentration tested. This is consistent with our previous observation that the dissociation constant of the anti-sTyr monoclonal antibody was larger than 1,000 nM.²⁷

Enrichment of sulfoproteins.

Engineered SH2 mutants were used to enrich C4-sTyr-GST in the presence of a large excess (100-fold) of BSA (Figure 6C). The experiment procedure was based on reported protocols for the enrichment of phosphopeptides.^{39, 40} Both SH2-1.8 and SH2-3.1 were able to enrich C4-sTyr-GST with minimal BSA contamination (Lanes 2–5; Figure 6C). The identity of enriched sulfoprotein at ~30 kDa was further verified as C4-sTyr-GST by mass spectrometry (Figure S8). Greater than 20-fold improvement in the ratio of C4-sTyr-GST to BSA before and after enrichment using SH2-1.8 and SH2-3.1 was exhibited following semi-quantitative analysis of triplicate enrichments using ImageJ (Figure S9). A trace amount of BSA after enrichment was likely due to non-specific binding of BSA to Ni-Sepharose resin since similar amount of BSA was detected in the absence of any SH2 mutants (Lanes 5 & 9; Figure 6C). As a control, enrichment experiments were carried out with C4-Tyr-GST under the same conditions. No enrichment of non-sulfated C4-Tyr-GST was observed. This further demonstrated the selectivity of engineered SH2 mutants to sTyr over Tyr.

Engineered SH2 mutants were used to selectively isolate C4-sTyr-GST from a mixture containing chemically phosphorylated BSA-P. As shown in Figure 6D, both SH2-1.8 and SH2-3.1 were able to enrich C4-sTyr-GST with SH2-1.8 being more efficient (Lanes 2–4; Figure 6C). Since the molecular weight of chemically phosphorylated BSA is non-homogeneous due to non-uniform modification, a smear was observed for BSA-P. Nevertheless, no significant contamination of BSA-P was observed after enrichment. As a control, an enrichment was carried out in which the initial mixture was first treated with a phosphatase to reduce the presence of phosphorylated tyrosine in BSA-P. Similar enrichment results were observed as that of non-treated enrichment. This further demonstrated that the engineered SH2-1.8 and SH2-3.1 mutants were capable of significant and specific enrichment of sulfoprotein in the presence of phosphoproteins.

Taking a step further, engineered SH2 mutants were used to isolate a sulfoprotein from HEK293T cell lysate. A natural sulfoprotein, C-X-C chemokine receptor type 4 (CXCR4), was employed. CXCR4 contains three sulfation sites at position Y7, Y12, and Y21. To avoid complications in isolating membrane proteins in this proof of concept study, the peptide fragment (1–38; MEGISIYTSDNYTEEMGSGDYDSMKEPCFREENANFNK) of CXCR4, which contains a different amino acid sequence from the peptide used for selection (Figure 3B), was fused to an enhanced green fluorescent protein (EGFP). sTyr was installed only at position Y21 using noncanonical amino acid mutagenesis in HEK293T cells,⁴¹ which afforded CXCR4-sTyr-EGFP. Both SH2-1.8 and SH2-3.1 were able to enrich CXCR4-sTyr-EGFP from the soluble HEK293T cell lysate (Lanes 4 and 5; Figure 7). As a control, no enrichment of CXCR4-sTyr-EGFP was observed when no SH2 was used for enrichment (Lane 6; Figure 7). This further demonstrated the application of engineered SH2-1.8 and SH2-3.1 for the enrichment of sulfoprotein from complex biological samples.

Figure 7. Enrichment of sulfoprotein from a complex biological sample.

Immunoblotting shows enrichment of sulfoprotein, CXCR4-sTyr-EGFP, from HEK293T cell lysate using engineered SH2 mutants. Lane 1, molecular weight marker; Lane 2, total HEK293T lysate; Lane 3, total soluble HEK293T lysate; Lane 4, enrichment with SH2-1.8; Lane 5, enrichment with SH2-3.1; Lane 5, enrichment without any SH2 mutants. Equal amounts (7 μg) of protein were added to each lane as determined using a Bradford assay.

CONCLUSION

In summary, the combination of hyperphage display with a specific elution method allowed for the identification of SH2-3.1 with high affinity and desired specificity for sTyr. The hyperphage system enhances substrate peptide binding by enforcing oligovalent display of SH2 mutants on every phage particle. A more specific elution with phenyl sulfate increases the stringency of selection for the identification of SH2 domain mutants selective for sTyr over pTyr. Engineered SH2 mutants in immunoassay and enrichment applications displayed promising results as an anti-sTyr antibody mimic. Both SH2-1.8 and SH2-3.1 were capable of specifically enriching a sulfoprotein in the presence of nonmodified or phosphorylated proteins and within complex biological samples. With significantly higher affinity, engineered SH2 mutants performed superior to a commercially available anti-sTyr monoclonal antibody. Application of these engineered SH2 mutants as suitable anti-sTyr antibody alternatives for the analysis of sulfoproteome is underway. Additional applications as biomolecular tools for the specific analysis of protein tyrosine O-sulfation are yet another possibility for these engineered SH2 mutants. Further, engineering of the residues involved in sequence specificity of SH2 variants may allow for the development of targeted SH2 mutants for the study of specific tyrosine O-sulfation sites.

MATERIALS AND GENERAL METHODS

Standard molecular biology techniques were used throughout. Biotinylated and fluorescein isothiocyanate peptide probes were purchased from CHI Scientific. Hyperphage were purchased from Progen Biotechnik GmbH. Neutraviden 96-well plates and T4 ligase were purchased from Thermo Scientific. KOD polymerase and anti-sulfotyrosine monoclonal antibody were purchased from EMD Millipore. Restriction enzymes, Taq DNA ligase and T5 exonuclease were purchased from New England Biolab. Trypsin, alkaline phosphatase and BSA-P were purchased from Sigma Aldrich. Potassium phenyl sulfate was purchased from Tokyo Chemical Industry. HRP conjugated anti-M13 antibody was purchased from GE Health Care Life Science. Anti-FLAG mouse antibody was purchased from Invitrogen. Anti-mouse goat antibody was purchased from BioRad. E. coli TOP10F’ were used for all cloning and phage propagation experiments. E. coli BL21(DE3) were used for all protein expression experiments. All solutions were prepared using deionized water passed through the Barnstead Nanopure ultrapure water filtration system. Standard antibiotic concentrations used were 100 μg/mL ampicillin, 50 μg/mL kanamycin and 5 μg/mL tetracycline.

SH2 Library Construction.

SH2-tm was subcloned into the pSEX vector via BamHI and NcoI cloning cut sites using Gibson Assembly to produce pSEX-SH2-tm. The pSEX plasmid allowed for expression and incorporation of SH2 domain as part of the full length pIII viral coat proteins into phage. Using pSEX-SH2-tm as DNA template, two fragments were produced using PCR while introducing randomization by means of NNK degenerate codons at positions R35, S37, E38, and T39. Following overlapping PCR, SH2-lib was cloned into the pSEX vector via BamHI and NcoI cloning sites using Gibson Assembly to produce pSEX-SH2-lib. A N-terminal FLAG-tag (DYKDDDDK) was included for the analysis of displayed levels of SH2-pIII fusion on surface of phage. Sequences of primers are included Table S2.

Plasmid construction.

SH2 variants were subcloned into the pET30b vector via NdeI and XhoI cloning sites to produce plasmids pET30b-SH2-wt, pET30b-SH2-tm, pET30b-SH2-1.5, pET30b-SH2-1.8, pET30b-SH2-3.1, pET30b-SH2-3.10, pET30b-SH2-60.1, pET30b-SH2-S37H, pET30b-SH2-E38P, pET30b-SH2-T39F, and pET30b-SH2-S37H/E38P. The pET30b plasmid allowed for the expression of C-terminal 6xHis-tag SH2 variants for subsequent purification. Sequences of primers are included in Table 2 of Supplementary Information.

The compliment 4 peptide sequence, EDYEYDELPAK, was cloned to the N-terminus of GST and inserted via NdeI and Blp1 cloning sites to produce pLei-C4-Tyr-GST. Sited direct mutagenesis was used to mutate desired Tyr1422 codon of peptide motif to an amber nonsense codon. The pLei plasmid allowed for amber suppression using the plasmid pBK-sTyrRS-5A, containing an optimized aminoacyl-tRNA synthetase for sTyr incorporation⁴² and an amber tRNA. Sequences of primers are included in Table S2.

The gene encoding the N-terminal fragment of CXCR4 fused to EGFP was inserted between HindIII and Pme1 cloning sites to produce pcDNA3.1-CXCR4-EGFP. Sited direct mutagenesis was used to mutate the desired Tyr21 codon into an amber nonsense codon. The amber suppression machinery is encoded on the plasmid pcDNA3.1-sTyrRS,⁴¹ containing an optimized aminoacyl-tRNA synthetase and an amber tRNA for sTyr incorporation in mammalian cells. Sequences of primers are included in Table S2.

Phage Display and Panning.

Transformed E. coli TOP10F’ harboring necessary F’ episome for phage infection with pSEX-SH2-lib were infected with either M13KO7 phage or hyperphage for generation of SH2-lib phage library. Briefly, following transformation the culture was grown overnight with shaking at 30 °C in media containing ampicillin, tetracycline, and 100 mM glucose. In the subsequent day, a subculture in media containing ampicillin, tetracycline, and 100 mM glucose with 1% of overnight culture was grown at 37 °C with shaking until OD₆₀₀ reached 0.5. Cells were infected with phage at a M.O.I. of 20 for 30 mins with no shaking at 37 °C followed by 30 mins with shaking at 37 °C. Infected cells were pelleted at 3,220 g for ten mins at 4 °C. The supernatant was removed, and the cells were resuspended in eight times volume of the initial subculture of media containing ampicillin and kanamycin. Phage propagation was carried out overnight at 30 °C with shaking. To isolate phage, E. coli TOP10F’ were collected by centrifuging at 16,000 g for ten mins. Five times concentrated polyethylene glycol/NaCl was added to the supernatant to precipitate the phage for one hour. Precipitated phages were pelleted at 16,000 g for ten mins. The supernatant was removed, and the final phage pelleted was concentrated at 5,000 g for three mins. Phage pellet was resuspended in 1% the volume of propagation culture in phage blocking buffer (PBB, 0.2% BSA, 0.05% Tween-20, PBS pH 7.4). Any remaining cells were removed by centrifuging at 21,000 g for 5 mins. All phage isolation steps were carried out at 4 °C. Phage concentration was determined using NanoDrop at OD₂₆₈ = 5×10¹² pfu/mL.

For positive selection panning, biotinylated sulfopeptide in peptide dilution buffer (PDB, 0.1% NaN₃, 0.2% BSA PBS pH 7.4) was incubated within a neutravidin well for one hour with rocking. Unbound peptide was removed by washing three times with 200 μL of phage washing buffer (PWB, 0.05% Tween-20, PBS pH 7.4). A total of 5,000 times the diversity (5.24×10⁹ pfu) of SH2-lib phages were incubated for one hour with rocking. Unbound/non-specific phage were removed by washing ten times with 200 μL of PWB. All panning steps were carried out at room temperature. Non-specific elution of the phage was achieved by incubating bound phage with 100 μL of 10 μg/mL trypsin for 30 mins at 37 °C. Specific elution was carried out by incubating bound phage with 100 μL of 200 mM phenyl sulfate for 10 mins at 37 °C followed by the addition of 10 μL of 10 μg/mL trypsin for 20 mins at 37 °C. Released phages were used for infection of 10 mL E. coli TOP10F’ at OD₆₀₀ 0.4 as previously outlined. Amplified phage pool was used for subsequent panning.

For negative selection panning, biotinylated phosphopeptide in PDB was incubated within a neutravidin well for one hour with rocking. Unbound peptide was removed by washing three times with 200 μL of PWB. A total of 5,000 times the diversity (5.24×10⁹ pfu) of SH2-lib phages were incubated for one hour with rocking. In tandem, biotinylated sulfopeptide in PDB was incubated within a neutravidin well for one hour with rocking. Unbound peptide was removed by washing three times with 200 μL of PWB. Unbound phages were incubated within the sulfopeptide containing well for one hour with rocking. Unbound/non-specific phage were removed by washing ten times with 200 μL of PWB. All panning steps were carried out at room temperature. Non-specific elution of the phage required incubation of bound phage with 100 μL of 10 μg/mL trypsin for 30 mins at 37 °C. Released phages were used for infection of 10 mL E. coli TOP10F’ at OD₆₀₀ 0.4 as previously outlined. Amplified phage pool was used for subsequent panning.

Phage ELISA.

Enriched phage library or individual phage clones were displayed on M13KO7. Biotinylated sulfopeptide or phosphopeptide in PDB were incubated within a neutravidin well for one hour with rocking. As a control, an empty well was incubated with only peptide dilution buffer. Unbound peptide was removed by washing three times with 200 μL of PWB. Phages (100 μL at 1×10¹⁰ pfu/mL) were incubated for one hour with rocking. Unbound phage were removed by washing five times with 200 μL of PWB. Anti-M13KO7 antibody (100 μL, 1/3000 dilution) in PBB was incubated for 30 mins with rocking. Unbound antibody was removed by washing five times with 200 μL of PWB. Tetramethyl benzamidine (100 μL; ThermoScientific) was used as substrate and incubated for 15 mins prior to quenching of the reaction with 50 μL H₂SO₄. OD₄₅₀ was measured using a Synergy H1 plate reader (BioTek Instruments, Inc.). Duplicate or triplicate measurements were obtained for control, sulfopeptide and phosphopeptide wells.

Protein Expression and Purification.

E coli BL21(DE3) transformed with desired pET30b-SH2, pLei-C4-Tyr-GST, or pLei-C4TAG-GST/pBk-sTyrRS-5A plasmids were cultured overnight with shaking at 37 °C. On the following day, a subculture (1%) of the overnight culture was grown at 37 °C with shaking until OD₆₀₀ reached 0.6. Protein expression was induced with IPTG (0.25 mM) for 18.5 hrs with shaking at room temperature after which the culture was pelleted at 5,000 g for 10 mins. For incorporation of sTyr, 1 mM sTyr was supplemented to the media. Isolated cells were lysed using sonication. Cellular debris was removed by centrifuging 21,000 g for 30 mins. Ni-Sepharose resin was used for purification of the 6xHis-tagged SH2 variants as per the manufacturer’s protocol. GSH-Sepharose resin was used for purification of C4-Tyr-GST and C4-sTyr-GST as per the manufacturer’s protocol. SDS-PAGE analysis was used to confirm purification. Purified proteins were desalted and buffer exchanged into protein storage buffer (20 mM Tris-HCl pH 7.5 150 mM NaCl, 10% glycerol) using an EconPac Desalting column (BioRad) as per the manufacturer’s protocol. Purified protein was flash frozen using liquid nitrogen and stored at −80 °C.

Immunoblotting Assay.

For Western and Far-Western blot assays, protein mixtures were first separated using SDS-PAGE gel electrophoresis. Separated proteins from SDS-PAGE gel were transferred to nitrocellulose membrane (0.45 μm, BioRad). For dot blot assays, 5 μL of protein mixture was spotted onto a nitrocellulose filter and allowed to dry for 1 hr at room temperature. Nitrocellulose membrane was blocked for 1 hr at room temperature. Blocking solution was discarded and the membrane was washed two times with PBS with 0.05% Tween-20 (PBST). Optimized far-Western immunoassays incubated 1 μg/mL Src-SH2 in PBST with 1 % BSA (PBSTB) prior to primary antibody. Primary antibody (1:3000) in PBSTB was added and incubated for 1 hr at room temperature. Solution was discarded and the membrane was washed two times. Secondary antibody (1:3000 goat anti-mouse HRP conjugated monoclonal antibody) was added and incubated for 1 hr at room temperature. Solution was discarded and the membrane was washed two times. Substrate was incubated with membrane for desired visualization.

Fluorescence Polarization Assays.

Purified SH2 variant was buffer exchanged into fluorescence polarization buffer (FPP, 20 mM potassium phosphate pH 7.35, 100 mM NaCl, 2 mM DTT, 0.1% BGG) without bovine γ-globulin (BGG) using an AmiconUltra 10,000 MW centrifuge filter (EMD Millipore) as per the manufacturer’s protocol. Protein concentration was determined using a Bradford assay (Biorad). Fluorescence polarization assay followed a similar procedure as reported.⁴³ Briefly, 10 nM FITC non-modified, sulfopeptide or phosphopeptide were incubated with serial dilutions of SH2 variant for 20 mins at room temperature in the dark. A Synergy H1 plate reader (BioTek Instruments, Inc.) equipped with standard filter cube (λ_Ex = 485 nm, BP = 20 nm; λ_Em = 528 nm, BP = 20 nm) was used to measure fluorescence polarization intensities. Serial diluted protein and FITC probe only controls were included in each assay. Parallel and perpendicular intensities of serial diluted protein only were accounted for each FITC probe tested. Corrected intensities were converted initially to fluorescence polarization values before determining fractional occupancy of SH2 binding pocket as reported.⁴⁴ Fluorescence polarization assays for each FITC probe were repeated in triplicate. Obtained data was plot as fractional occupancy vs SH2 concentration and fit to the Hill Equation using MatLab to determine dissociation constant.

Comparative Modeling and Peptide Docking.

Previously reported methods^{35, 36} for Rosetta were applied for comparative modeling and peptide docking using identified SH2 mutants. Additional information on commands, flags and scripts used are included in the Supplementary Material.

Transfection and Lysis of Mammalian Cells.

HEK293T cells were maintained in DMEM containing 10 % FBS at 37 °C 5 % CO₂. At 70 % confluency, media was replaced with fresh media. Cells were transfected with plasmids pcDNA3.1-sTyrRS and pcDNA3.1-CXCR4-EGFP using Lipofectamine2000 (ThermoFisherScientific) as per the manufacturer’s protocol. Media was supplemented with 5 mM sTyr. Transfected cells were incubated for 48 hrs prior to cell lysis using RIPA buffer (25 mM Tris pH 7.6, 1 % NP-40, 150 mM NaCl) supplemented with Halt Protease Inhibitor (ThermoFisherScientific). To ensure complete cell lysis, cells were further sonicated for 30 seconds at 50 % amplitude. Cell lysate was centrifuged for 15 mins at 14,000 g. Soluble fraction was applied to designed sulfoprotein enrichment protocol. All cell lysis steps were carried out at 4 °C.

Sulfoprotein Enrichment.

Sulfoprotein enrichment was based on reported protocols for the enrichment of phosphopeptides.^{39, 40} Sulfoprotein enrichment from BSA/C4-sTyr-GST mixture was conducted by using 20 μL of Ni-sepharose 6 Fast Flow affinity resin. Resin was washed with 5 column volume of 20 mM imidazole wash buffer (PBS pH 7.4, 0.1 % NP-40). The SH2 variant of interest (50 μg) was incubated for 10 mins. For sulfoprotein enrichment from BSA-P/C4-sTyr-GST mixtures, 5 μL of resin was used with only 10 μg of immobilized SH2. For alkaline phosphatase treated mixture, 0.5 units/μg total protein were incubated with mixture for 16 hrs at 22 °C. For sulfoprotein enrichment of CXCR4-sTyr-EGFP, 30 μL of resin was used with 1 mg of immobilized SH2. Bound SH2 was washed with 3 column volume wash buffer. Desired sulfoprotein mixture was incubated for 2 hrs. Bound protein was washed using two 25 column volume wash buffer and nine 25 column volume wash buffer containing 50 mM imidazole. Bound sulfoprotein was eluted using four column volume elution buffer (PBS pH 7.4, 0.1% NP-40, 200 mM phenyl sulfate). All steps were carried out at 4 °C.

Mass Spectrometric Analysis.

Mass spectrometric analysis was carried out by the Proteomics and Metabolomics Facility at the Centre of Biotechnology at the University of Nebraska-Lincoln. Excised gel pieces corresponding to expected C4-sTyr-GST following enrichment were subjected to trypsin digest. Tryptic digest was analyzed by nano LC-MS/MS experiments, running a 1 h gradient on a 0.075 mm x 250 mm C18 Waters CSH column on a Q-Exactive HF mass spectrometer. All MS/MS data were analyzed using Mascot version 2.6.1. Mascot was set up to search for common contaminants using the cRAP_20150130.fasta database and user-supplied protein sequences. Mascot was searched with a fragment ion mass tolerance of 0.060 Da and a parent ion tolerance of 10.0 PPM. Deamidation of asparagine and glutamine, oxidation of methionine, carbamidomethyl of cysteine and phosphorylation and sulfation of serine, threonine and tyrosine were specified in Mascot as variable modifications.