Article Cite This: Anal. Chem. 2018, 90, 13533−13540
pubs.acs.org/ac
A Ubiquitous but Overlooked Side Reaction in Dimethyl Labeling of Peptides Rijing Liao,*,† Yuan Gao,‡ Ming Chen,⊥ Lulu Li,† and Xuye Hu§ †
Downloaded via UNIV OF RHODE ISLAND on December 2, 2018 at 13:35:15 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Shanghai Institute of Precision Medicine, The Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200125, China ⊥ State Key Laboratory of Bioorganic and Natural Products Chemistry, Shanghai Institute of Organic Chemistry, Chinese Academy of Sciences, 345 Lingling Road, Shanghai 200032, China § Shanghai Clinical Center, Chinese Academy of Sciences, Shanghai 200031, China S Supporting Information *
ABSTRACT: Reductive dimethylation using formaldehyde and NaBH3CN to label peptides or proteins on their N-termini and lysine residues is one of the most widely used labeling methods in the quantitative proteomics field. In this study, we characterized a ubiquitous side reaction in dimethylation labeling, causing mass increments of 26 Da on the N-termini of peptides. It can occur extensively on most peptides, which significantly compromises data quality in terms of sensitivity, dynamic range, and peptide- and protein-identification rates. Nevertheless, this side reaction was so-far overlooked, largely because the current database search algorithms limited the detection of unknown modifications. In order to illustrate the chemical nature of this side reaction, 1D and 2D nuclear magnetic resonance (NMR) was performed to elucidate the exact structure of the modification formed through this side reaction, revealing that the side reaction produced an N-methyl-4imidazolidinone moiety between the first two residues of the undesirably labeled peptides. On the basis of the mechanism proposed for the side reaction, we optimized the reaction conditions for dimethyl-labeling. Compared with the current typical labeling method, our approach can dramatically suppress the side reactions at both the standard protein and proteome levels. As a result, with our optimal labeling method, peptide- and protein-identification rates were significantly increased compared with those from the traditional labeling method.
T
can be distinguished on the basis of their isotopic tags, which are isotopically distinct in MS or MS/MS spectra. The benefit is that these isotopic tags are chemically identical; thus, the cognate peptides from different samples will coelute in chromatographic separation and have the same electrosprayionization efficiency in MS analysis, which improves the accuracy of quantification when compared with label-free approaches. Among reported stable-isotope-labeling methods, reductive dimethylation1−6 of peptides with formaldehyde in the presence of sodium cyanoborohydride (NaBH3CN) is, in terms of procedure, straightforward, fast, and undemanding. Furthermore, this method is quite budget friendly in comparison with other isotope-labeling methods such as stable-isotope-labeling by amino acids in cell culture (SILAC)7 and isobaric tags for relative and absolute quantification (iTRAQ).8 During dimethyl labeling, the
he advent of mass spectrometry (MS) in combination with liquid chromatography (LC) has enabled proteomics to identify and quantify thousands of proteins from complex biological samples. Nevertheless, the accurate quantification of proteins across multiple biological samples is not without difficulty. The signal intensities or extracted-ioncurrent (XIC) peak areas of peptides are commonly used when comparing their abundances across multiple samples. Unfortunately, the MS signal intensity of a given peptide is not only determined by its abundance but also prone to the impacts of several other factors (e.g., elution time and matrix effects). Thus, the variations between separate LC fractionation and LC-MS analysis are inevitable and will compromise the accuracy of protein relative quantification between different samples. To minimize the variation, stable-isotope-labeling strategies have been developed to enable processing and analyzing multiple samples under the same chromatography and mass-spectrometry conditions. Samples to be compared are differentially labeled with isotopic tags (a set of isotopic variants) and then mixed before further processing and LCMS/MS analysis. Peptides and proteins from different samples © 2018 American Chemical Society
Received: August 7, 2018 Accepted: October 29, 2018 Published: October 29, 2018 13533
DOI: 10.1021/acs.analchem.8b03570 Anal. Chem. 2018, 90, 13533−13540
Article
Analytical Chemistry
incubated at −20 °C overnight. The precipitate was collected by centrifugation and washed twice with ice-cold acetone. The pellet was dissolved in 1 M urea and 50 mM ammonium bicarbonate (pH 8.0). This was followed by trypsin digestion at 37 °C overnight. After a desalting step using Sep-PAK C18 (Waters), the tryptic peptides were dried with a SpeedVac and stored at −80 °C until use. Dimethylation Labeling. Two labeling methods were used in this study to compare their labeling performance. All labeling reactions and quenching steps should be carried out in a fume hood to protect the researchers. In the traditional method, peptides were dissolved in 100 μL of 100 mM triethylammonium bicarbonate (TEAB, pH 8.5). Formaldehyde solution (4 μL; 4%, v/v) was added to the peptide solution, which was followed by brief mixing and centrifugation. Immediately, 4 μL of 0.6 M NaBH3CN was added to the sample, which was followed by brief mixing and centrifugation. After incubation at 30 °C for 1 h, the labeling reaction was quenched by adding 16 μL of 1% (v/v) ammonia solution. In our optimized method, peptides were dissolved in 100 μL of 100 mM NaH2PO4 (pH 6.0). Immediately before use, 4% (v/v) formaldehyde was mixed with 1.2 M NaBH3CN in a ratio of 1:1 (v/v); then, 8 μL of the mixture was added to the peptides solution; this was followed by brief mixing and centrifugation. The labeling reaction was incubated at 30 °C for 1 h and then quenched by adding 16 μL of 1% (v/v) ammonia solution. Elucidation of Byproduct Structure by NMR. Peptide LVNE (10 mg) was dissolved in 10 mL of 100 mM triethylammonium bicarbonate (TEAB, pH 8.5). Formaldehyde solution (400 μL; 4%, v/v) was added to the peptide solution, followed by 400 μL of 0.6 M NaBH3CN. After incubation at 30 °C for 1 h, the labeling reaction was quenched by adding 1.6 mL of 1% (v/v) ammonia solution. After desalting using a Waters Sep-pak C18 column, the labeling products were further purified by semipreparative reversedphase HPLC (Agilent series 1200 system). About 1 mg of the byproduct with a mass increment of 26 Da was obtained. The byproduct was dissolved in 0.5 mL of CD3OD and then subjected to a Bruker 600 MHz NMR for the acquisition of the 1 H NMR, 13C NMR, COSY, HSQC, and HMBC data. Nano-RPLC-MS/MS. Samples were loaded onto a trap column (75 μm i.d., 2 cm, C18, 3 μm, 100 Å, Dionex) with an online desalting wash and then separated with a reversedphased column (75 μm i.d.,10.2 cm, C18, 3 μm, 120 Å, New Objective) by an EASY-nLC 1000 (Thermo Scientific) at a flow rate of 300 nL/min. The synthetic-peptide samples were separated using a 30 min LC gradient (10−50% B from 0 to 30 min; solvent A, 0.1% formic acid in H2O; solvent B, 0.1% formic acid in acetonitrile), BSA tryptic digests were separated using a 1 h LC gradient (6−24% B over 40 min, 24−40% B from 40 to 47 min, 40−90% B from 47 to 50 min, and continuing at 90% B up to 60 min), and the HeLa samples were separated with a 3 h LC gradient (5−22% B over 115 min, 22−40% B from 115 to 163 min, 40−90% B from 163 to 167 min, and continuing at 90% B up to 180 min). The nano-ESI voltage was set to 2.0 kV, and the capillary temperature was 275 °C. The MS data were acquired on an Orbitrap Fusion mass spectrometer (Thermo Scientific) in data-dependent mode. Each full-scan MS (m/z 350−2000, resolution of 60k) was followed with several HCD MS/MS scans (normalized collision energy of 30, m/z 150−2000, resolution of 15k) for the most intense precursor ions within a
primary amine-groups of peptides, including the N-terminal amines and the ε-amines of lysine residues, can react with formaldehyde to form the imine (also called Schiff-base) intermediates. In the presence of reducing agent such as NaBH3CN, the imine intermediates are further reduced to give the N-methyl group. However, the imine intermediates are also reactive with the side-chains of many amino acid residues (e.g., arginine, histidine, and tyrosine), resulting in intra- or interpeptide cross-links.9 Undoubtedly, these side reactions will compromise the quality of analysis in terms of sensitivity, dynamic range, and peptide- and protein-identification rates. Side reactions reduce the yields of the desired products, thereby reducing their chances of detection. Moreover, the byproducts produced by side reactions increase the sample complexity, which further reduces the probability of lowabundance peptides being selected for MS/MS fragmentation in data-dependent-acquisition (DDA) mode. However, side reactions in labeling are often overlooked by proteomic researchers, and consequently, the chemical natures of their products remain unclear. This is largely due to the current database search algorithms, which cannot identify the side products without knowing the exact changes in mass and the positions of the modifications caused by the side reactions. In this work, we characterized a side reaction that could occur significantly in reductive-dimethylation labeling of peptides with formaldehyde and NaBH3CN. We first observed this side reaction on a synthetic peptide, and further analysis of complex biological samples revealed that it can occur extensively on most peptides, resulting in a 26 Da mass increment at the N-termini of peptides. The exact chemical structure of the modification moiety caused by the side reaction was elucidated by 1D and 2D nuclear-magneticresonance (NMR) techniques, including 1H NMR, 13C NMR, COSY, HSQC, and HMBC, revealing that an N-methyl-4imidazolidone moiety was formed between the first and second residues of the peptides in the side reaction. The illustrated structure suggests that the side reaction forms an imine intermediate, which is further converted to the imidazolidone via intrapeptide cross-linking with the adjacent amide group. On the basis of this proposed mechanism, we optimized the labeling conditions to suppress the side reactions in dimethyl labeling. Using the optimized labeling conditions, we can suppress this side reaction and increase the peptideidentification rates. We demonstrated that about 15% more peptides and 10% more proteins were identified in the same the HeLa tryptic digests with the optimal labeling method than with the traditional labeling method.
■
MATERIALS AND METHODS Chemicals and Reagents. Chemicals were purchased from Sigma-Aldrich unless stated elsewhere. The double deionized water was produced by a Milli-Q gradient A10 system (Millipore). The standard peptides were purchased from GL Biochem Ltd. Bovine-serum-albumin (BSA) standard protein was purchased from Beyotime Ltd. Mammalian-Cell Culture and Protein Preparation. HeLa 229 cells were cultured as previously described.10 The cells were washed once in PBS buffer and then sonicated in 6 M urea/2 M thiourea on ice until the lysate appeared clear. The proteins were reduced with 10 mM dithiothreitol (DTT) at 37 °C and then alkylated with 25 mM iodoacetamide at room temperature for 30 min in the dark. The protein solution was mixed with 5 times its volume of ice-cold acetone and 13534
DOI: 10.1021/acs.analchem.8b03570 Anal. Chem. 2018, 90, 13533−13540
Article
Analytical Chemistry
at the N-terminal amino group (Figure 1b). However, the minor peak at16.02 min was an unknown product 26 Da heavier than the unmodified peptide. Comparing the MS/MS spectra from the correctly dimethyl-labeled peptide and the side product, both spectra had the same y-ion series but distinct b-ion series (Figure 1b,c), suggesting that the side product was an adduct of DRVYIHPF with the modification appended at the peptide N-terminus. In particular, the y7 fragment ions (m/z 931.51) detected in the both MS/MS spectra further supported the hypothesis that the modification was attached at the first residue. In order to assess the frequency of this side reaction, we then applied the dimethyl-labeling reaction to BSA tryptic digests following the same procedure. The labeling products were subjected to LC-MS/MS for analysis, and the raw data was processed by Proteome Discoverer software (Materials and Methods). The database-search results showed that this side reaction occurred on most BSA tryptic peptides. In the BSA digests, 42 peptides were identified as correctly dimethyl labeled, among which 31 peptides were also detected as being undesirably labeled. Moreover, there were four peptides detected only in the undesirably labeled form. Table 1 shows the 31 peptides identified with both the correctly dimethylated and the undesirably modified forms. The XCorr score of each undesirably labeled peptide was comparable to that of the corresponding correctly labeled species, suggesting that the +26 Da modified peptides were not likely to be false positives or artifacts of the search. Interestingly, each +26 Da modified peptide showed longer retention time than the corresponding dimethylated peptide, indicating that this +26 Da modification caused a larger increase in peptide hydrophobicity than the dimethylation did. The observation that this +26 Da modification could extensively occur on most BSA tryptic peptides motivated us to study it in detail. The fact that this side reaction can happen to the peptides starting with glycine or aliphatic amino acids (Table 1) indicates that side chains of amino acids are not involved in the side reaction. The 26.0162 Da mass increment of the side product, corresponding to a nominal C2H2 moiety, suggested an increase of 1 degree of unsaturation in the side products. In other words, either one ring or one double bond was introduced to the side products by the side reaction, but given the fact that the labeling reactions were carried out in the presence of NaBH3CN, which is a reducing reagent that is strong enough to reduce double bonds, a ring structure was more likely to be the one formed in the side products. One possible structure of the side product was N-methyl-4imidazolidinone (Supporting Information Figure S1a). A previous study conducted by Metz et al.9 reported that formaldehyde, in the absence of NaBH3CN, could cross-link the N-terminus amine group with the adjacent amide group via a methylene bridge, forming a 4-imidazolidinone moiety with a mass increase of 12 Da. In the presence of NaBH3CN, the 4imidazolidinone would be further converted to N-methyl-4imidazolidinone. The structure of N-methyl-4-imidazolidinone exactly fit the mass increase observed in the side product, and it could also explain the longer retention time in the side products than in the dimethylated peptides. However, there were still two hurdles to confidently assign the modification moiety as N-methyl-4-imidazolidinone. First, the reaction conditions reported to yield N-methyl-4imidazolidinone were quite different from the traditional dimethyl labeling protocol. In the previous report,9 peptides
cycle time of 3 s. The maximum ion-injection time (MIT) for MS1 and MS/MS were 50 and 100 ms, and the auto gain control target for MS1 and MS/MS were 2 × 105 and 2 × 105, respectively. The dynamic-exclusion time was set as 60 s. Data Analysis. Raw data were viewed in Xcalibur (Thermo Scientific). Proteome Discoverer 2.2 (Thermo Scientific) with the SEQUST engine was used for database searches and quantification. The raw data of the BSA and HeLa samples were searched against the bovine Uniprot and human UniProtKB databases, respectively. Enzyme specificity was set to trypsin, allowing for up to two missed cleavages. The mass tolerance was set to 10 ppm for the precursor ions and 0.02 Da for the fragment ions. Carbamidomethylation (+57.0215 Da) on cysteine and dimethylation (light: +28.0313 Da, heavy: +30.0380 Da) on N-terminal amine and lysine were set as static modifications. In order to assess the levels of side products in labeling samples, database searches were carried out with the same settings as above except using a fixed +26.0156 Da modification to replace the N-terminal dimethylation. Decoy database searches were also performed in parallel, and peptides with a less than 1% falsediscovery rate (FDR) were accepted.
■
RESULTS AND DISCUSSION We first noted this side reaction on a synthetic peptide, DRVYIHPF, which was dimethyl-labeled with formaldehyde and NaBH3CN following the protocol described previously.3 We analyzed the labeling products with LC-MS, and the basepeak chromatogram showed two peaks eluting at 13.02 and 16.02 min (m/z 537.7903 and 536.7827, respectively; Figure 1a). Confirmed by the MS/MS spectra, the major peak at 13.02 min was the desired product with correct dimethylation
Figure 1. LC-MS/MS analysis of the dimethyl-labeling products of the synthetic peptide DRVYIHPF. (a) Base-peak chromatogram of the labeling products. (b,c) MS/MS spectra of the correctly dimethylated peptide (b) or the side product (c). 13535
DOI: 10.1021/acs.analchem.8b03570 Anal. Chem. 2018, 90, 13533−13540
Article
Analytical Chemistry
MS and MS/MS spectra, more evidence was required to determine which structure was the right one. To date, nuclear magnetic resonance (NMR) is still the most powerful tool that can provide detailed and decisive information for the structural elucidation of small molecules. In order to elucidate the exact chemical structure of the side product, we decided to determine the structure by high-resolution NMR. Identification of the Chemical Structure of the Side Product by NMR. In the dimethyl labeling of BSA tryptic digests, peptide LVNELTEFAK was shown to produce a relatively high percentage of the side product. Moreover, its first and second amino acid residues have aliphatic side chains, preventing any other side reactions involving the side chains. Therefore, we decided to use it as a model peptide to elucidate the chemical structure of the side product formed in dimethyl labeling. In order to reduce the complexity of the NMR spectra, a shortened peptide, LVNE, was synthesized and used instead of the LVNELTEFAK. The tetrapeptide was labeled with formaldehyde and NaBH3CN on a large scale, following the same protocol we used for the dimethyl labeling of BSA digests. After multiple steps of isolation, about 1 mg of the byproduct was purified and reconstituted in deuterated methanol for the acquisition of the 1H NMR, 13C NMR, COSY, HSQC, and HMBC data (see the Materials and Methods for details). The 1H NMR, 13C NMR (Table 2), and HSQC spectra (Supporting Information Figure S2a−c) indicated that the
Table 1. BSA Tryptic Peptides Identified in Both the Dimethylated Form and the +26 Da Modified Form through the Traditional Dimethyl-Labeling Method Xcorr
peptide sequences
dimethyl labeled
+26 Da modified
ratio between the two speciesa
HPEYAVSVLLR HLVDEPQNLIK CCTESLVNR SHCIAEVEK DLGEEHFK NECFLSHK ETYGDMADCCEK GACLLPK LGEYGFQNALIVR EACFAVEGPK LFTFHADICTLPDTEK QTALVELLK LVNELTEFAK NECFLSHKDDSPDLPK DAIPENLPPLTADFAEDK TCVADESHAGCEK DDSPDLPK TVMENFVAFVDK YICDNQDTISSK HPYFYAPELLYYANK SLHTLFGDELCK DAFLGSFLYEYSR KQTALVELLK EYEATLEECCAK ECCHGDLLECADDR GLVLIAFSQYLQQCPFDEHVK DDPHACYSTVFDK LKPDPNTLCDEFK CCAADDKEACFAVEGPK RHPEYAVSVLLR DAIPENLPPLTADFAEDKDVCK
2.76 3.98 2.18 2.60 2.38 2.27 4.16 2.26 3.92 3.12 4.28 3.23 3.26 3.73 3.25 3.76 2.59 4.72 3.44 3.88 4.53 3.96 3.58 4.53 5.14 4.86 4.20 5.08 5.19 5.97 5.41
3.77 4.59 2.58 3.00 2.53 2.37 4.21 2.25 3.85 3.02 4.18 3.09 3.11 3.58 3.07 3.53 2.31 4.37 3.08 3.47 3.90 3.29 2.73 3.60 3.98 3.34 2.51 3.32 3.38 4.07 2.92
0.20 0.12 1.78 3.55 0.18 1.94 0.15 0.04 0.01 0.04 0.69 0.05 0.23 0.15 1.44 1.04 0.37 0.02 0.23 0.61 0.30 0.03 0.32 0.09 0.67 0.01 0.25 0.03 0.41 0.03 0.15
Table 2. 1H (600 MHz) and 13C (150 MHz) NMR Data for the Adduct of LVNE in CD3OD Leu
a
Ratio = (intensity of the +26 Da modified peptide)/(intensity of the dimethyl-peptide).
Val
were first incubated with formaldehyde alone for several hours before the addition of NaBH3CN, allowing the formation of 4imidazolidinone within the first two amino acid residues. In the dimethyl-labeling reaction, however, NaBH3CN was added immediately after formaldehyde. More importantly, the structure of the methylene bridge between the two residues was inconsistent with the MS/MS result (Figure 1c), which suggested that the modification should located at the first amino acid residues of the peptides. In view of the disagreement between 4-imidazolidinone and the MS/MS spectra, we propose another possible structure, N-methylaziridine (Supporting Information Figure S1b), for the side product. This structure also introduces a nominal C2H2 moiety to the products. As the N-methylaziridine moiety is more hydrophobic than the dimethylated amine group, it could also account for the fact that the side products had longer retention times than the corresponding dimethylated products. Furthermore, the structure of N-methylaziridine was consistent with the result of the MS/MS spectra, which showed the modification located at the first amino acid residue. Nevertheless, in spite of the hints provided by the retention time and
Asn
Glu
a
C no.a
δC, type
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
20.93, CH3 24.47, CH 21.37, CH3 36.59, CH2 64.19, CH 38.17, CH3 65.37, CH2 169.18, C 60.96, CH 27.46, CH 18.08, CH3 17.17, CH3 168.74, C 50.20, CH 36.27, CH2 172.95, C 171.34, C 51.62, CH 26.57, CH2 29.62, CH2 174.08, C 173.02, C
δH 1.03, 2.07, 1.03, 1.85, 3.89, 3.02, 4.58,
d m d m; 1.69, m br s d; 4.99, d
4.35, 2.19, 1.03, 0.93,
d m d d
4.78, dd 2.80, dd; 2.65, dd
4.45, dd 2.21, m; 1.95, m 2.43, t
The C numbers are labeled in Figure 2.
byproduct had five methyl groups, five methylenes (CH2), six methines (CH), and six carbonyl carbons (δC 168.7−174.9). The presence of six methines ruled out the possibility that the byproduct possessed a N-methylaziridine moiety, because the LVNE peptide containing a N-methylaziridine moiety should have 5 methines. The resonances at δH 3.02 and δC 38.2 were attributed to the N-methyl group. The proton of the N-methyl 13536
DOI: 10.1021/acs.analchem.8b03570 Anal. Chem. 2018, 90, 13533−13540
Article
Analytical Chemistry
accelerate the reduction of imine and increase the yield of the desired methylated peptides. In most published protocols, dimethyl labeling is performed in triethyl ammonium bicarbonate (TEAB) buffer, and the addition of formaldehyde occurs before that of NaBH3CN.2−5 Theoretically, adding formaldehyde first might increase the side reactions to some extent because the imine intermediates will form after the addition of formaldehyde, and in the absence of NaBH3CN, the imine will be converted to the unwanted side products only. Therefore, we premixed formaldehyde and NaBH3CN immediately before their addition to the labeling solution, allowing the simultaneous addition of both formaldehyde and NaBH3CN. Our experiments indicated that adding the labeling reagents simultaneously did reduce the peptide spectra matches (PSMs) of the undesirably labeled species when compared with those from the traditional method (from 107 PSMs to 95 PSMs on average, Figure 3). Thus, in the following experiments, we always premixed formaldehyde and NaBH3CN immediately before their use and added them simultaneously to start the labeling reaction.
showed long-range HMBC (Supporting Information Figure S2e) correlations to the methine carbon at δC 64.2 and the methylene carbon at δC 65.4, allowing for the assignment of the methine and methylene as the α-C of Leu and the methylene of 4-imidazolidine, respectively. The assignment of methylene in 4-imidazolidine was further supported by the HMBC correlations from H-7 (δH 4.58/4.99) to C-6 (δC 38.2) and the long-range 1H−13C correlation from H-9 (δH 4.35) to C-7 (δC 65.4). Taken together, the chemical structure of the modification moiety in the byproduct was established as an Nmethyl-4-imidazolidinone moiety between the Leu and Val residues (Figure 2). The key 1H−1H COSY and HMBC correlations in the adduct are also shown in Figure 2.
Figure 2. Key COSY and HMBC correlations of the adduct of the LVNE peptide produced by the side reaction.
Optimization of the Dimethyl-Labeling Conditions. On the basis of the structure elucidation, this ubiquitous side reaction of dimethyl labeling was established as the intermolecular cross-linking reaction between the N-terminal amine group and the adjacent amide group. A mechanism for this side reaction was proposed and is shown in Scheme 1. In Scheme 1. Primary Reaction and Side Reaction Involved in Dimethyl Labeling of Peptides with Formaldehyde and NaBH3CN
Figure 3. PSM numbers of the dimethyl-labeled and the undesirably labeled peptides in BSA tryptic digests labeled under different conditions. All numbers are the arithmetic means of three technical replicates.
In order to increase the reducing-reaction rate for Schiff-base intermediates, we doubled the concentration of NaBH3CN (final concentration of 44.4 mM) in TEAB buffer, which was expected to promote the conversion of imine intermediates to methylated products while competitively suppressing the unwanted side reactions. As shown in the Figure 3, increasing the NaBH3CN concentration in the labeling solution further reduced the undesirably labeled PSMs (from 95 PSMs to 84 PSMs on average, Figure 3). In organic synthesis, NaBH3CN is a mild reducing agent often used in acidic reaction solutions rather than basic solutions.11,12 We therefore tried to labeled the BSA digests in a slightly acidic phosphate buffer (100 mM NaH2PO4, pH 6.0). The results indicated that this could dramatically reduce the side products when compared with the use of TEAB buffer (from 84 PSMs to 19 PSMs on average, Figure 3). Further decreasing the pH value of the phosphate buffer from pH 6.0 to 4.0 could slightly reduce the side products. However, the PSMs of the correctly labeled peptides were also decreased (from 432 PSMs to 421 PSMs on average, Figure 3). This was likely because the lower pH value also reduced the reaction efficiency of the desired dimethyl labeling while also reducing the side reactions. As our previous experiments showed that doubling the concentration of NaBH3CN could reduce the
the side reaction, the N-terminal amine groups of the peptides first react with formaldehyde and form hydroxymethyl groups. The hydroxymethyl groups undergo condensation to give the imine groups (Schiff bases), which were also the key intermediates for the desirable dimethylation reactions. In the presence of reducing agents (e.g., NaBH3CN), most imine intermediates are reduced quickly to the desirable dimethylated products, but a small proportion of imine intermediates undergo the intramolecular nucleophilic addition to form the unwanted side products. Thus, the key factor affecting the ratio between the demethylated products and the byproducts is the reactivity of the reductant used in the labeling reaction. Obviously, enhancing the reducibility of NaBH3CN would 13537
DOI: 10.1021/acs.analchem.8b03570 Anal. Chem. 2018, 90, 13533−13540
Article
Analytical Chemistry
Figure 4. (a) Numbers of undesirably labeled PSMs and (b) numbers of dimethyl-labeled PSMs, unique peptides, and grouped proteins identified from HeLa digests labeled under the traditional or optimized conditions. All numbers are the arithmetic means of three technical replicates.
reduction in peptide and protein identification. Side reactions lower the yields of correctly labeled species, thus reducing the detectability of peptides, especially low-abundance peptides. More importantly, the byproducts produced by side reactions substantially increase the sample complexity and thus compromise the probabilities of low-abundance peptides being selected for MS/MS sequencing. Given the large dynamic range of protein abundances in the whole proteome, which typically spans 6 to 10 orders of magnitude,18,19 the byproducts from high-abundance proteins would still be more abundant than the low-abundance peptides and therefore compete with the low-abundance peptides for MS/MS sequencing in DDA mode. Obviously, reducing byproducts can enhance the concentrations of correctly labeled species. Moreover, it reduces the competition between byproducts and low-abundance peptides for MS/MS selection. As a result, identification, especially the identification of low-abundance peptides, will be remarkably increased in the optimally labeled samples. As shown in our results, although the increases in correctly labeled PSMs were observed within each peptideintensity bin with optimal labeling, the PSMs in the lowintensity bins were increased more significantly than those in the high-intensity bins (Supporting Information, Figure S3a,b). By further comparing the PSM numbers of the HeLa samples labeled with the optimal and traditional methods, we found a significant discrepancy between the reduced number of undesirably labeled PSMs and the increased number of correctly labeled PSMs: the optimal method reduced about 870 undesirably labeled PSMs but increased more than 2800 correctly labeled PSMs (Figure 4). Given the fact that the average MS/MS acquisition time for the undesirably labeled or dimethyl-labeled species were close, we speculated that the optimal labeling conditions had reduced other byproducts in addition to those with N-methyl-4-imidazolidinone for dimethyl labeling. It has been reported that formaldehyde, in the absence of NaBH3CN, can induce many intrapeptide crosslinking reactions, including the imidazolidinone we characterized in present study and other cross-linking between primary amine groups (N-terminal amines and the ε-amines of lysine) and the reactive side chains of arginine, tyrosine, and histidine.9,20 These side-chain cross-linking products, unlike the N-methyl-4-imidazolidinone, cannot be detected by a typical database search because of the difficulty of interpreting their MS/MS spectra. However, they would also reduce the probability of the correctly labeled species being detected. Because all these side-chain cross-linking reactions occur through a similar mechanism as that of imidazolidinone, the optimal labeling conditions that suppress the formation of
PSM numbers of byproducts while increasing the PSM numbers of correctly labeled species in TEAB buffer, we doubled the NaBH3CN concentration in the pH 6.0 phosphate buffer, allowing further reduction of the PSMs of byproducts while increasing the PSMs of correctly labeled peptides (Figure 3). It is worth noting that the labeling efficiency under all the labeling conditions was high, because only negligible PSMs (
