Investigating Glyoxylate-Mediated Transamination Using Dipeptide

Sep 20, 2018 - Glyoxylate-mediated transamination (GT) is a classic, potentially general, and N-terminus-specific protein modification method useful f...
0 downloads 0 Views 5MB Size
Download PDF
Article Cite This: Bioconjugate Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/bc

Investigating Glyoxylate-Mediated Transamination Using Dipeptide Arrays and Proteomic Peptide Mixtures Xiaohong Tan*,† and Chuan-Fa Liu*,‡ †

Chemistry Department and Center for Nucleic Acids Science and Technology, Carnegie Mellon University, Pittsburgh, Pennsylvania 15213, United States ‡ School of Biological Sciences, Nanyang Technological University, 637551 Singapore

Downloaded via UNIV OF TEXAS SW MEDICAL CTR on October 9, 2018 at 11:51:12 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Glyoxylate-mediated transamination (GT) is a classic, potentially general, and N-terminus-specific protein modification method useful for the preparation of bioconjugates. However, there is a lack of information on whether and how readily a particular N-terminal amino acid (in the context of a peptide chain) can be converted to the 2-oxoacyl moiety under GT conditions. Here, we conducted a systematic investigation of GT using membrane-bound dipeptide arrays that include all the 400 possible dipeptide combinations of the 20 genetically encoded amino acids. This colorimetric method offers a convenient way to assess the GT reaction tendency of N-terminal residues by the naked eye. It also provides interesting information about the effect of the second residues on GT, which has not been reported previously. In addition, we also designed a proteomics approach to study GT in solution using tryptic peptide mixtures, which not only confirmed many of our findings in peptide array assays but also revealed potential side reaction products. Taken together, our studies will make the future use of GT for protein modification in a much more predictable way.

■

can also proceed in the presence of urea7 or guanidine hydrochloride which increases the solubility of proteins. Therefore, in principle, any protein with a free N-ter amine can be transaminated, and the resultant carbonyl from the 2oxoacyl moiety can be used in a subsequent bioconjugation reaction through the formation of a hydrazone, oxime, or thiazolidine linkage. Although a number of proteins have been reported to be transaminated by glyoxylate,6−12 there is still no sufficient information available that can guide one to reliably predict whether and how readily an N-ter residue can undergo GT. This has prompted us to conduct a systematic study of GT. In this report, first we used a membrane-bound dipeptide library to assess the propensities of each of the 20 N-ter amino acid residues to undergo GT under the effects of the next amino acid residue. This library consisted of all the 400 possible dipeptide combinations of the 20 genetically coded amino acids, each represented by one spot on the membrane. After

INTRODUCTION Protein chemical modification is recognized as a critically important tool for protein functional study as well as for many other applications.1−3 Most of these methods functionalize specific amino acids such as Cys or Lys residues, resulting in product mixtures that present the new functional groups in multiple locations on the protein surface, which may adversely affect protein function. Therefore, it is highly desirable to develop alternative methods capable of modifying a protein once at a single site. The classical reaction of glyoxylatemediated transamination (GT) represents a potentially general method for protein N-terminal (N-ter) modification. Typically, in the presence of glyoxylate and copper(II) or nickel(II) ions and in a weakly acidic aqueous buffer, the N-ter residue of a peptide is converted to a 2-oxoacyl moiety (Scheme 1). As this reaction requires a free amine alpha to a carboxyamide (or carboxyl group for amino acids) and the N-ter α-amine is a much weaker base than the ε-amine of lysine, it is expected that transamination is specific only to the N-ter amine of a polypeptide chain.4,5 In addition, with the catalysis of metalions, this reaction is fast. For example, transamination of cytochrome c could be completed within 15 min.6 Moreover, it © XXXX American Chemical Society

Received: July 3, 2018 Revised: September 8, 2018 Published: September 20, 2018 A

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry Scheme 1. Reaction Mechanism of a Glyoxylate-Mediated Transamination

being subjected to the standard treatment conditions with glyoxylate and copper(II) ion, the reactivity of each of the dipeptide spots toward transamination was directly visualized in a colorimetric assay with a dye-modified hydrazide. Second, we used a proteomic method to study GT on an E. coli tryptic peptide mixture. After transamination, the successfully modified peptides were conjugated with an aminooxyfunctionalized biotinylated peptide containing a TEV cleavage site through the formation of an oxime linkage. After avidin affinity purification and TEV cleavage, LC-MS/MS analysis and a protein database search were conducted to identify the sequences of E. coli tryptic peptides which were able to undergo transamination.

Figure 1. Demonstration of the reactivity and suitability of DabsylAla(β)-hydrazine as a colorimetric reagent. A) Two peptide spots with the H-Thr-Ala-β-Ala peptide without periodate oxidation as the control; B) two peptide spots with the H-Thr-Ala-β-Ala peptide with periodate oxidation as positive results. After the periodate treatment, H-Thr-Ala-β-Ala on cellulose is converted to oxoacetyl-Ala-β-Ala which can conjugate with Dabsyl-β-Ala-hydrazine to generate a visible spot.

bound dipeptide libraries were incubated in the reaction solution, 3 M sodium acetate buffer (pH 5.5) containing 5 mM copper sulfate and 0.1 M glyoxylic acid, for 1 and 3 h, respectively. In a separate experiment, the third library was treated for 1 h with the same acetate buffer containing 0.1 M glyoxylic acid but no copper(II) ion. After the transamination reaction, all of the above peptide membranes were washed extensively and treated with the red dye reagent for 22 h at 37 °C and washed extensively. The results were shown in Figure 2A, B, and C. The peptide array results are summarized as in the following observations. First, when the membrane was treated by GT solution without Cu+, there was no transamination, and the membrane showed no obviously colored spots (Figure 2A), which served as a negative control for our experiments. This is consistent with early findings that metal ions such as copper(II) and nickel(II) are required when using glyoxylate as the deamination reagent,6 as shown in the proposed mechanism (Scheme 1). The few very light spots on the membrane might be due to nonspecific hydrophobic binding between the dye molecule and the membrane-bound dipeptides. Second, when simply comparing the intensities of the colored spots on the membranes treated for 1 and 3 h (Figure 2B and C), the darkest spots were those that displayed dipeptides with these N-ter residues: Gly, Ala, Leu, Met, Ser, and Lys. These amino acids are either small or contain long, unhinderd side chains, which result in little steric hindrance to the transamination reaction on the α-amine. The next group is from these rows: Phe-X, Tyr-X, Val-X, Thr-X and Ile-X, Asn-X, Trp-X and perhaps Glu-X that displayed moderate color spots. It was not surprising that, being more hindered, an N-ter Thr would react more slowly than Ser. The aromatic residues PheX and Tyr-X also displayed reasonable reactivity in the GT. Val-X and Ile-X are expected to be less reactive than the first group, and the color on the spots was less intense. Furthermore, light to very light to invisible color was seen with the remaining rows of Asp-X, Cys-X, Arg-X, Gln-X, His-X,

■

RESULTS AND DISCUSSION The Design and Synthesis of Peptide Arrays. The arrangement of the library was in a matrix system containing 400 spots, each representing a dipeptide. If the N-ter amino acid residue is converted to the oxoacyl moiety during GT, the ketone or aldehyde (when the N-ter residue is Gly) group should react easily with Dabsyl-β-Ala-hydrazide to form a hydrazone, and the reactivity of each dipeptide should be shown by the intensity and the size of the red dot formed on the membrane. The β-Ala residue in this dye reagent would serve as a good spacer and reduce any steric hindrance that the bulky Dabsyl group might impose on the reacting hydrazide. Since the R1 group in the 2-oxoacyl moiety may have an influence on the yield of hydrazone formation, having this βAla spacer in this dye reagent would also help minimize any influence of the R1 group. To check whether Dabsyl-β-Alahydrazide is a suitable reagent to detect membrane-bound reactive carbonyls, we first synthesized a test peptide, H-ThrAla-β-Ala-cellulose on a small Whatman filtration paper using the spot synthesis technique developed by Ronald Frank.13 After oxidation by periodate, the N-ter Thr was converted to oxoacetyl which was then subjected to reaction with the dye reagent at 1 mM in 0.1 M acetate buffer containing 50% DMSO at pH 5.7 for 2 h.14 As a result, the two treated spots showed up as dark red dots, whereas the other two untreated spots showed no color at all (Figure 1, only black and white pictures are presented in this report). The color stayed on the two spots after extensive washing with DMF and acetic acid solution, indicating stable covalent binding of the dye to the membrane. This result indicates that Dabsyl-β-Ala-hydrazide is a good reagent to visualize the presence of the oxoacyl moiety. GT Study with Peptide Arrays. Three identical dipeptide libraries, each containing 400 dipeptides, were prepared for transamination study on solid phase. Two of the membraneB

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry

Figure 2. Using dipeptide membrane libraries to reveal GT reactivity trends. The intensities of the spots indicate the extent of the reaction of the 2oxoacyl-dipeptides with the Dabsyl-β-Ala-hydrazine. A is the negative control. B, C, and D indicate GT treatment for 1, 3, or 10 h, respectively. The letters on the left edge of each membrane array image denote the N-ter amino acid residues, while the letters on top indicate the amino acids at the second position.

and Pro-X. No-reactivity for His-X dipeptides was not unexpected as previous studies in solution reactions had shown that an N-ter His could not undergo transamination.5 It is assumed that the copper ion binds to histidine via the N-ter amino group and the N-3 of the imidazole side chain to form a six-membered chelate ring.15 It was also known that an N-ter Pro could not be transaminated due to its secondary amine in nature. Third, when the effect of the second amino acid residue on transamination was assessed, two residues, Pro and His, were found to have a clearly negative impact, as these two columns displayed much lighter color than did the other columns. The X-Pro column in particular had almost no colored spots (Figure 2B, C), indicating that having a Pro as the second amino acid residue strongly inhibited the reaction. A possible explanation is that the presence of Pro at the second position would disallow the formation of the copper complex intermediate. As seen from Scheme 1, enolization of the amide bond, presumably a prerequisite step for the formation of a stable copper complex intermediate, would not be possible for an X-Pro peptide bond. Although an alternative copper complex involving the carbonyl oxygen of the keto form amide

is also conceivable, the tertiary amide nature of an X-Pro peptide bond would still impose considerable steric hindrance and disfavor its formation. A clear albeit less drastic inhibitory effect was also observed for the X-His dipeptides (Figure 2B, C). This inhibition effect might be caused by a competing chelate complex that engages the side chain imidazole ring of the His residue and the N-ter amine (Scheme 2), making it less available for transamination. This kind of complex structures is known to exist in nature, as seen in the naturally occurring copper peptide GHK-Cu found in human plasma which is a copper complex of the tripeptide Gly-His-Lys.16 Because GT of membrane-bound peptides may be slower than GT of peptides in solution, a new membrane containing Scheme 2. Proposed Copper Chelating Mechanism for Peptides with a His Residue at the Second Position

C

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry Scheme 3. Protocol for Using a Proteomic Method To Study GT

various N-terminal residues. For this, we developed a proteomic method to study GT with peptide mixtures generated by tryptic digestion of bacterial cellular proteins. Such a proteomics study not only can validate the peptide array experiment but also may help reveal any side reactions that might accompany GT. GT Study with a Proteomic Method. As shown in Scheme 3, the whole protein extraction of E. coli cells was treated with trypsin to generate a peptide mixture. After reduction and thiol-alkylation, we conducted GT on the tryptic peptides. Then samples were loaded on a C18 column with a 1 h wash by buffer A (0.045% TFA) to remove the large excess of glycoxylate, followed by a quick elution using buffer B (90% acetonitrile, 0.05% TFA) to collect the retained peptides. After that, a biotin-hydroxylamine peptide, containing a TEV protease cleavage site, reacted with the N-ter 2-oxoacyl of successfully transaminated peptides through oxime formation.19,20 The peptide mixture was loaded into a C18 column, following a RP-HPLC gradient (0%−80% buffer B in 40 min) to remove the excess of biotin-hydroxylamine peptide. Finally, peptides were enriched by avidin affinity selection, and all successfully modified peptides were eluted by TEV protease cleavage. The eluted peptides would contain a unique N-ter modification by an oxime-linked tripeptide designated SYK (Scheme 3), which brings a molecular-weight increase of 449.191 Da compared to the unmodified peptides prior to transamination. The sample was analyzed by LC-LTQ mass spectrometry (LC-MS/MS), and the mass data was searched in an E. coli tryptic peptide database by GPM (http://www. thegpm.org/), with a constant N-ter modification (+449.191) and other potential modifications (see the Experimental Procedures for details). After the database search, 629 peptides were identified, which have a constant N-ter modification (+449.191) (see Table S1 for details). These peptides are derived from 139 E. coli proteins. The distribution pattern of different N-ter residues in the identified oxime products is shown in Figure 3 (blue bars). Since the experimentally found frequencies solely are not adequate to compare the behaviors of different N-ter residues, we also calculated, using a PERL script, the theoretical frequency (Figure 3, red bars) for each of the 20 amino acids at the N-ter positions of all predicted tryptic peptides from these 139 proteins. The ratio of the experimentally found frequency over the calculated frequency should be a better parameter for the relative reactivity of a particular N-ter amino acid (Figure 3, insert). Several observations are noteworthy from Figure 3. First, the over-

selected peptides was prepared, which contained the dipeptide rows that did not have good color in the previous experiments. So, this membrane was incubated in GT solution for 10 h before reacting with the dye reagent (Figure 2 D). As expected, after this extended reaction, most of the less colored rows in the previous experiment showed up in much darker red color. This was especially true for the rows of Val-X, Ile-X, Asn-X, Asp-X, Thr-X, and Glu-X. It should be mentioned that this membrane was first treated with iodoacetamide to block the thiol group of Cys residue before the 10-h GT. For this row of modified Cys*-X, obvious red spots were observed, indicating that an unmodified N-ter Cys undergoes GT very slowly (Figure 2 B, C), probably because the thiol group in its side chain and the N-ter amine might chelate with the metal ion. In addition, the rows of His-X, Gln-X, and Arg-X were still negative. The reason for the negative results of Gln-X and ArgX is not clear at this stage. It is likely that these peptides may very well undergo GT, but the ketone group in the products might react with side-chain functional groups of Gln and Arg to form a stable ring structure, preventing reaction with the dye reagent. It was reported in an early study by Francis et al. that the transaminated product of N-ter Gln obtained from a biomimetic transamination reaction is resistant to subsequent conjugation.17 Also, an N-ter Gln can undergo pyruglutamate formation, a reaction with a half-life of several hours,18 which reduces its availability for GT. As for the effect of the second residue, it was confirmed that a second Pro or His residue was detrimental to the transamination reaction as the two columns had the least colored spots on this membrane. It appears that spots with Ser/Thr at the second position are slightly darker than the spots in other columns for the membranes receiving 1 and 3 h treatment for transamination (Figure 2B and C). However, no such difference was observed when transamination was performed for 10 h (Figure 2D). This suggests that the reaction might be slightly faster for the peptides on the spots of the Ser/Thr columns. The reason for this is not clear. As a plausible explanation, the presence of the hydroxyl in Ser/ Thr might help solvate the membrane-bound peptides in the aqueous buffer, making the peptides more solvent-exposed for transamination. The above membrane-bound peptide array results reveal interesting trends on the reactivity of the dipeptides toward GT. These results confirm that GT is a relatively general Nterminal protein modification reaction as most of the membrane-bound dipeptides were able to get modified. Next, we wanted to see how GT would work in solution on real biological samples containing mixtures of peptides with D

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry

oxime products of N-ter Glu (9.24%) were observed than that of N-ter Asp (0.16%). We reasoned that N-ter Asp would be able to undergo transamination smoothly, but a side reaction has probably happened after the transamination reaction. As we know, decarboxylation of a carboxylate alpha to a carbonyl, as present in the transamination product of an N-ter Asp, is a classic reaction in organic chemistry, and transaminationinduced Asp decarboxylation was indeed observed in previous studies.21,22 After decarboxylation, the final transamination product of N-ter Asp is the same as that of N-ter Ala. Therefore, we searched the database again by considering this side reaction, Asp decarboxylation (−43.990), in addition to the standard N-ter modification (+449.191). As shown in Figure 4, 934 peptides are now found, about 30% of which are

Figure 3. Experimental distribution (blue bars) of the N-ter amino acids in all the peptides found in the proteomics study versus the theoretical distribution (red bars) of the same amino acids at the Nterminal positions of the predicted tryptic peptides of the identified E. coli proteins (assuming complete tryptic digestion). The inserted figure displays the quotient values of the experimental distribution over the theoretical distribution of individual N-ter residues.

representation of Ala is particularly striking, suggesting it is the most preferred amino acid for GT. Gly is also well represented. This is not surprising since Ala and Gly are the smallest amino acids, which present little steric hindrance to the transamination reaction on the α-amine. Interestingly, as the residue containing the smallest side chain, N-ter Gly gives less oxime products than N-ter Ala. Second, N-ter Glu and Leu, which contain long, unhindered side chains and should undergo transamination smoothly, are found in 9.24% and 6.21%, respectively, of the peptides. Third, the fewest products were observed from the group comprising Pro, His, and Trp. It should also be mentioned that because we used tryptic peptides for our transamination study, there ought to be no or very few N-ter Lys and Arg peptides identified from this proteomics study, since trypsin cleaves a peptide bond after Lys or Arg. This turns out to be true, as only four N-Lys peptides (and no N-Arg peptides) were found (Table S1) which were produced due to cleavage of a dibasic peptide bond [Arg-Lys (peptide No. 536) or Lys-Lys (peptide No. 539)] and relative resistance of the following Lys-Asp and Lys-Glu bonds toward trypsin digestion in these two peptides. Surprisingly, one prolinyl-peptide was identified in the proteomics study, which is a fragment of the E. coli protein 30S ribosomal protein S5 (Table S1). This peptide was not generated by tryptic digestion but rather via spontaneous cleavage of the Asn-Pro bond from its parent sequence. However, it is not clear how this peptide got picked up in the transamination−oxime conjugation experiments since an N-ter Pro is not able to undergo transamination. Another surprise is with Gln which was largely unreactive on membrane but showed up in a number of enriched tryptic peptides in the proteomics study. A possible reason is that GT of a Glnpeptide and the subsequent conjugation might exhibit differences in kinetics and product stability between solid phase and solution reactions. Nevertheless, one should also note that, when compared to the probability of an N-ter Gln in the predicted tryptic peptides, the number of Gln-peptides found in the proteomics experiment was relatively not so high (Figure 3). This method could be useful to reveal the information on side products. For example, as we learned from Figure 2D, after 10 h GT incubation, N-ter Asp displayed very similar product spots as N-ter Glu. However, as seen in the proteomics study (Figure 3), a much greater number of

Figure 4. Oxime product distribution of the five most reactive Nterminal amino acid residues from all identified transaminated peptides, which was obtained after database search with consideration of Asp decarboxylation. The percentage value in brackets indicates the calculated probability of these N-terminal amino acids in the predicted tryptic peptides.

the N-ter Asp peptides (Figure 4 and Table S2), indicating that decarboxylation of N-ter Asp is a common side reaction under the GT reaction conditions used in our study. The overrepresentation of N-ter Asp peptides also suggests that the byproduct so-formed, which is identical to the transamination product of N-ter Ala, is particularly amenable to oxime conjugation.

■

CONCLUSION We have conducted a systematic investigation of the glyoxylic acid-mediated transamination reaction using membrane-bound dipeptide libraries and E. coli tryptic peptide mixtures. Our study revealed some interesting trends of GT for the different amino acid residues at the N-ter or second position. Based on the results obtained from the peptide arrays and the proteomic method, we present a set of guidelines for using GT-enabled conjugation in Table 1. The nature of the side chains of N-ter amino acid residues greatly influences GT and the stability of resultant 2-oxoacyl products, which then determines whether or not a particular N-ter residue is suitable for GT-enabled bioconjugation. As shown in Table 1, 14 N-ter amino acids are well or reasonably suited for this purpose. The first group contains 8 residues (Ala, Asp, Gly, Glu, Leu, Met, Ser, and Lys) which appear to be the most preferred N-ter amino acids for GT-enabled conjugation. The second group includes 6 residues (Ile, Val, Thr, Phe, Tyr, and Asn) which are also suitable albeit with slower kinetics. The last group contains 6 E

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry

interest. In addition, our data suggest that, as an N-terminusdirected reaction, GT is potentially useful for special proteomics studies. For example, the Wells lab has previously developed a so-called N-terminomics technology based on the use of subtiligase, an artificially engineered peptide ligase that can add a peptide tag to the N-terminal amine of a protein. Using this technology, they have successfully identified a large number of cellular protein substrates cleaved by caspases24 and characterized endogenous and exogenous proteolysis in human plasma and serum.25 Taken together, our study provides useful guidelines for the future use of glyoxylate-based transamination for protein modification, bioconjugation, and proteomics applications.

Table 1. Suitability for GT-Enabled Conjugation of N-ter and the Second Residue Predicted from This Study N-ter residues

suitability

A, D, G, E, L, M, S, K I, V, T, F, Y, N C, Q, R, W, H, P the second residues

excellent good to be avoided suitability

H, P

to be avoided

residues (Cys, Gln, Arg, Trp, His, and Pro), and these residues either cannot be transaminated or the GT is very slow. Moreover, we also report the effect on the second residue toward transamination, i.e., Pro and His should be avoided at position 2 since they would strongly inhibit GT reaction of the N-ter residues. However, it should be pointed out that the rules in Table 1 are derived mostly from peptide membrane array studies. Since we did not characterize the conjugation products on the spots, the nature of some of these products may be different from what is expected. So some caution may need to be applied when using these rules. For example, the membrane-peptide array study showed that an N-ter Ser displayed very good reactivity toward GT and subsequent conjugation with the dye reagent, but in the proteomic study, this residue did not show up at a representation rate expected from its high abundance in proteins. One possible explanation for this discrepancy is that the GT products underwent some side reactions that altered the molecular weights of the final conjugation products, leading to missed identification of N-ter Ser peptides in the proteomic study. As shown previously from a biomimetic transamination study using pyridoxal-5-phosphate, β-elimination of the Ser transamination product was indeed a significant side reaction.21 However, when we screened the database with the modification of Ser β-elimination (−18), no such modified N-ter Ser was observed (data not shown), indicating that Ser β-elimination alone is not an observable side reaction in the GT conditions reported in this study. Based on the N-ter Thr and Ser data in peptide arrays, we can say that N-ter Ser can readily undergo transamination under GT conditions but with unknown side reactions. We were unable to determine the nature of these side reactions. There may also be other unknown side reactions involving some other amino acid residues. For example, Phe was fairly reactive in the membrane assay but under-represented from the proteomics experiment. Nevertheless, the two methods used herein prove complementary in studying GT. In recent years, the Francis lab used pyridoxal-5-phosphate to perform a transaminate reaction on proteins,23 and they also investigated the effects of the identity of N-ter residues on the reactivity using a tetrapeptide library.17 They reported that Nter residues such as His, Trp, Lys, Pro, Cys, Ser, and Gln should be avoided in the pyridoxal-5-phosphate based transamination reaction, and pyridoxal-5-phosphate-based transamination also has the preference for positive charges near the N-ter positions.17 Herein, to our best knowledge, our report is the first one to study glyoxylate-mediated transamination in a systematic way. Our data reveal interesting trends of all amino acids (as the N-ter residue) toward GT, as well as how the neighboring residues (at the second position) affect GT, increasing our predictive power for this reaction. An initial set of guidelines is provided, that may be used to increase the applicability of this reaction to specific proteins of

■

EXPERIMENTAL PROCEDURES Chemicals, Reagents, and Materials. All chemicals and solvents were of analytical grade and were used as received from commercial sources unless otherwise noted. All Fmocprotected amino acids, Rink-amide-MBHA-resin, DCC (dicyclohexylcarbodiimide), 2-chlorotrityl chloride resin, B O P [ b e n z o t r i a z o l e - 1 - y l o x y - t r i s ( d i m et h y l a m in o ) phosphonium-hexafluorophosphate], DMAP (4-dimethylamino-pyridine), and DMSO (dimethyl sulfoxide) were all obtained from GL Biochem (Shanghai, China). DMF (N,N′dimethylformamide), DCM (dichloromethane), piperidine, TFA (trifluoroacetic acid), TIS (triisopropylsilane), DIEA (diisopropylethylamine), sodium acetate, methanol, biotinhydrazine, hydrazine hydrate, triethylamine, and dioxane were obtained from Merck. 1,3-Diaminopropane, iodoacetamide, Dabsyl-Cl [4-(dimethylamino)azobenzene-4′-sulfonyl-chloride], and copper(II) sulfate pentahydrate were purchased from Aldrich. E. coli strain MG1655 is a gift from Dr. Sze in the School of Biological Sciences, Nanyang Technological University. HPLC Assay. Reverse-phase high performance liquid chromatography (RP-HPLC) was performed on an Agilent system (Agilent Tech., USA) with a Vydac C18 reverse-phase column (5 μm, 250 mm × 4.6 mm). The analysis was carried out using the mixture of two solvents, A and B, as the mobile phase. Solution A was H2O (deionized) containing 0.045% TFA, and solution B was acetonitrile/H2O (90/10) containing 0.04% TFA. The mobile phase flow rate was 1 mL/min, and the separation temperature was 25 °C. UV detection was carried out at 220, 254, and 280 nm. Mass Spectrometry. ESI-MS data were obtained at the NTU SBS Mass Spectrometry Core Facility using the Finnigan LCQ Deca XP MAX. Preparation of Dipeptide Library. An 8 × 8 cm sheet of cellulose paper (Whatman #1 chromatography paper) was treated with the symmetrical anhydride of Fmoc-β-alanine (6 mmol), DCC (3 mmol), and DMAP (0.4 mmol) in 25 mL of DMF at room temperature for 3 days.26 Washed by DCM, DMF, and methanol, the paper was treated with 20% piperidine in DMF for 1 h to remove the Fmoc group, and the piperidine was washed 5−8 times until the wash was bromophenol blue negative. The paper was washed over methanol and dried. Then it was used for the “spot-synthesis” of the library containing 400 dipeptides according to standard Fmoc machine protocols using a MultiPep synthesizer from Intavis Bioanalytical Instruments AG. A capping step was included after the first coupling step by immerging the membranes in 10% Ac2O solution in DMF. When the synthesis was completed, the side chain-protecting groups F

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry

General Procedure for Transamination of Tryptic Peptide Mixture. E. coli strain MG1655 was grown in LB medium at 37 °C for overnight. Cells were harvested by centrifugation, washed twice by PBS, and sonicated into 3 mL of PBS containing 1% triton X-100 on ice with 12 short bursts of 10 s followed by intervals of 30 s for cooling. After centrifugation (21000 × g for 15 min at 4 °C), proteins in the supernatant solution were precipitated by 4 volumes of cold acetone, rinsed twice by cold acetone, and resuspended in 3 mL of 50 mM ammonium bicarbonate buffer, pH 8. Around 3 mg of proteins was digested with 60 μg of trypsin at 37 °C for 16 h (20 μg of fresh trypsin was added after 5 h), prior to adding 2 mM TCEP at 37 °C for 30 min. After reduction, proteins were treated by 6 mM iodoacetamide for 1 h in the dark. After that, the transamination reaction was carried out with addition of 10% pyridine, 0.2 M glyoxylic acid, and 10 mM CuSO4 . Undissolved peptides were collected by centrifugation and resuspended into 3 M guanidine hydrochloride containing 10% pyridine, 0.2 M glyoxylic acid, and 10 mM CuSO4. Both peptide samples were incubated at 37 °C for 3 h prior to combination together for RP-HPLC desalting with a Vydac semipreparation C18 reverse-phase column, in which glyoxylic acid was removed by 1 h wash using buffer A. The collected peptide solution, eluted by quickly raising the HPLC eluent to 100% buffer B, was lyophilized into power and resuspended into 0.1 M phosphate buffer containing 3 M guanidine hydrochloride, pH 6.0. Aminooxy-functionalized biotin-peptide (2 mM) was added, and the solution was incubated at 37 °C for 24 h prior to injection into the HPLC system using the same C18 column. The fractions corresponding to the peak containing an excess of the aminooxy biotinpeptide were discarded; all remaining fractions were collected into 10 portions under a gradient of 0−80% B in 40 min. All peptides were lyophilized to powders which were dissolved into 0.2 mL of 100 mM Bicine, pH 8, 100 mM NaCl, and 1 M guanidine hydrochloride. Peptides were incubated with streptavidin beads (Pierce Biotechnology 100 μL) for 3 h at room temperature for enrichment of the N-terminusbiotinylated peptides. The streptavidin beads were washed extensively with 50 mM Bicine (pH 8) × 1, 4 M NaCl × 2, 50 mM Bicine (pH 8) × 3 and finally 100 mM Bicine (pH 8), DTT (1 mM) × 1. The captured peptides were released from streptavidin beads by overnight treatment with 5 μL of TEV protease, respectively, into the last wash buffer. General Procedures for LC-MS/MS Analysis. The released peptides were desalted by Ziptip-C18 and submitted into an LC-LTQ system. Peptides were separated with a 90 min linear gradient of 0−80% buffer B (90% acetonitrile and 0.1% methanoic acid) and eluted directly into the LTQ spectrometer. MS/MS spectra were collected automatically during the LC-MS runs. Each scan was set to acquire a full MS scan followed by four MS/MS scans of the four most intense ions from the preceding MS scan. Database Searching. After data acquisition, MS/MS spectra were then extracted and searched against in an E. coli tryptic peptide database by GPM (http://www.thegpm.org/). Complete modifications include +449.1910 Da at all N-termini of peptides and carbamidomethyl of Cys (+57.021). Potential modifications included oxidation of Met (+15.995), deamidation of Asn (+0.984), and decarboxylation of Asp (−43.990). The parent tolerance is 20 ppm, and fragment tolerance is 0.5 Da.

were removed with 82.5% TFA, 5% phenol, 2.5% ethanedithiol, and 5% H2O for 2 h. The paper was washed with DCM, DMF, and methanol and air-dried. Preparation of Dabsyl-Ala(β)-hydrazide. The 2-chlorotrityl resin (1.6 mmol) was soaked in dry DCM for 20 min. Hydrazine hydrate (16 mmol) was incubated with the resin in dry DCM for 1 h. After DCM wash to remove excess hydrazine, Fmoc-β-Ala-OH was coupled to the resin by the standard Fmoc chemistry. After removing the Fmoc group and washing, the resin was soaked in DMF/DCM (1:1), and 4 equiv of Et3N and 4 equiv of Dabsyl-Cl were added. The mixture was incubated overnight with shaking. The product was cleaved from the resin by 50% TFA in DCM (v/v) for 2 h and purified by HPLC using a C18 column and checked by ESI-MS (MH+ found 391.4; MW calc. 390.5). Transamination Reaction of the Dipeptide Library. The membrane displaying the dipeptide library was incubated in a plastic tray containing 50 mL of 3 M sodium acetate solution (5 mM copper sulfate, 0.1 M glyoxylic acid, pH 5.5) for 1 or 3 h at RT. Another dipeptide membrane was incubated in 20 mL of 5 mM iodoacetamide in 0.1 M sodium acetate (pH 5.6) for 5 min at RT and washed carefully by 25 mM sodium acetate buffer (pH 5.5). Dried with methanol, the membrane was incubated in the above transamination solution for 10 h. The control one was incubated in the same transamination reaction solution for 1 h without the addition of copper(II) ion. All reactions were stopped by the addition of 10 mM EDTA, and then membranes of dipeptide libraries were washed by 25 mM sodium acetate buffer, pH 5.5. Verification of the Colorimetric Detection Method Based on the Use of Dabsyl-β-Ala-hydrazine. A membrane containing H-Thr-Ala-β-Ala-cellulose spots was treated with 5 mL of 1 mM sodium periodate in 5 mM sodium acetate buffer (pH 5.6) for 15 min in the dark at RT. Then the membrane was washed carefully by 5 mM sodium acetate (pH 5.6) and followed by incubation with the dye reagent solution (1 mM Dabsyl-β-Ala-hydrazine and 50% DMSO in 0.1 M acetate buffer, pH 5.7) for 2 h at 37 °C. The control group was directly treated by the dye reagent without periodate oxidation. Both were washed carefully by DMF and acetic acid solution (pH 2.4) until the background was clean. The color development was monitored visually and recorded by photography. Reaction of Transaminated Dipeptide Libraries with Dabsyl-β-Ala-hydrazine. Dabsyl-β-Ala-hydrazide was dissolved in 50% DMSO/0.1 M sodium acetate (pH 5.7) at a concentration of 1 mM. The transaminated dipeptide libraries were treated with this Dabsyl-β-Ala-hydrazide solution (50 mL) for 22 h at 37 °C. Then membranes were washed carefully by DMF and acetic acid solution (pH 2.4) until the background was clean. The color development was monitored visually and recorded by photography. Synthesis of Aminooxy-Functionalized Biotin-Peptide. Fmoc-Lys-(Mtt)-OH was coupled on Rink-AmideMBHA-resin, and the Mtt group was removed by 3% TFA. After wash, (Boc-aminooxy) acetic acid was coupled on the side-chain of lysine, and the remaining residues were coupled in turn using standard Fmoc chemistry. The sequence of this peptide is Biotin-Ahx-Kε-Kε-TENLYFQ-SYK(ONH2)-NH2 (MH+ found 1960.7; MW calc. 1959.7). Kε indicates that the amide bond is formed on the side-chain of the Lys residue by using Boc-Lys(Fmoc)-OH. G

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX

Article

Bioconjugate Chemistry

■

(16) Lau, S. J., and Sarkar, B. (1981) Biochem. J. 199, 649−656. (17) Scheck, R. A., Dedeo, M. T., Iavarone, A. T., and Francis, M. B. (2008) J. Am. Chem. Soc. 130, 11762−11770. (18) Ribo, M., Bosch, M., Torrent, G., Benito, A., Beaumelle, B., and Vilanova, M. (2004) Eur. J. Biochem. 271, 1163−71. (19) Calzadilla, M., Malpica, A., and Cordova, T. (1999) J. Phys. Org. Chem. 12, 708−712. (20) Rosenberg, S., Silver, S. M., Sayer, J. M., and Jencks, W. P. (1974) J. Am. Chem. Soc. 96, 7986−7998. (21) Lee, S. H., Goto, T., and Oe, T. (2008) Chem. Res. Toxicol. 21, 2237−2244. (22) Silva, A. M. N., Borralho, A. C., Pinho, S. A., Domingues, M. R. M., and Domingues, P. (2011) Rapid Commun. Mass Spectrom. 25, 1413−1421. (23) Gilmore, J. M., Scheck, R. A., Esser-Kahn, A. P., Joshi, N. S., and Francis, M. B. (2006) Angew. Chem., Int. Ed. 45, 5307−5311. (24) Wiita, A. P., Seaman, J. E., and Wells, J. A. (2014) Methods in Enzymology pp 327−358, Academic Press, DOI: 10.1016/B978-0-12417158-9.00013-3. (25) Wildes, D., and Wells, J. A. (2010) Proc. Natl. Acad. Sci. U. S. A. 107, 4561−4566. (26) Tam, J. P., Liu, C. F., Shao, J., and Rao, C. (1995) Int. J. Pept. Protein Res. 45, 209−216.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.bioconjchem.8b00475. Identified oxime peptides and their corresponding E. coli proteins through the proteomic method (Tables S1 and S2) (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: cfl[email protected]. ORCID

Xiaohong Tan: 0000-0001-7272-4292 Chuan-Fa Liu: 0000-0001-7433-2081 Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We appreciate the late Professor B. F. Dixon (University of Cambridge) for his helpful discussions and advice. We appreciate Professor Jinming Li (Southern Medical University, China) for his help in calculating the theoretical frequency for the N-ter residues of all identified E. coli proteins. We also thank the Ministry of Education (MoE) of Singapore (MOE 2016-T3-1-003) for financial support, as well as Nanyang Technological University. We are grateful to the David Scaife Family Charitable Foundation for a postdoctoral fellowship to X.T.

■

REFERENCES

(1) Francis, M. B. (2008) Chemical Biology pp 593−634, Wiley-VCH Verlag GmbH, DOI: 10.1002/9783527619375.ch10c. (2) Boutureira, O., and Bernardes, G. J. L. (2015) Chem. Rev. 115, 2174−2195. (3) Spicer, C. D., and Davis, B. G. (2014) Nat. Commun. 5, 4740. (4) Dixon, H. B., and Weitkamp, L. R. (1962) Biochem. J. 84, 462− 468. (5) Papanikos, A., Rademann, J., and Meldal, M. (2001) J. Am. Chem. Soc. 123, 2176−2181. (6) Dixon, H. B., and Moret, V. (1965) Biochem. J. 94, 463−9. (7) Suenaga, M., Ohmae, H., Okutani, N., Kurokawa, T., Asano, T., Yamada, T., Nishimura, O., and Fujino, M. (1999) J. Chem. Soc., Perkin Trans. 1 1, 3727−3733. (8) Metzler, D. E., Olivard, J., and Snell, E. E. (1954) J. Am. Chem. Soc. 76, 644−648. (9) Bunzli, H. F., and Bosshard, H. R. (1971) Hoppe Seylers Z. Physiol. Chem. 352, 1180−1182. (10) Fields, R., and Dixon, H. B. (1971) Biochem. J. 121, 587−589. (11) Van Heyningen, S., Tipton, K. F., and Dixon, H. B. (1968) Biochem. J. 108, 508−509. (12) Nishimura, O., Suenaga, M., Ohmae, H., Tsuji, S., and Fujino, M. (1998) Chem. Commun. 10, 1135−1136. (13) Frank, R. (2002) The SPOT-synthesis technique. J. Immunol. Methods 267, 13−26. (14) Shao, J., and Tam, J. P. (1995) J. Am. Chem. Soc. 117, 3893− 3899. (15) Sugiura, Y. (1977) Eur. J. Biochem. 78, 431−435. H

DOI: 10.1021/acs.bioconjchem.8b00475 Bioconjugate Chem. XXXX, XXX, XXX−XXX