阿木雜紀: 2月 2016

Title

Identification of 2-oxohistidine interacting proteins using E. coli proteome chips.

Date: 20160206

Version: 1

Running Title

Identification of 2-oxohistidine interacting proteins

Abbreviations

The abbreviations used are: PTM, post-translational modification, MCO, metal-catalyzed oxidation, RAGE, receptors for advanced glycation end-products, Aβ, amyloid beta, AD, Alzheimer’s disease, GO, Gene Ontology, KEGG, Kyoto Encyclopedia of Genes and Genomes, BSA, bovine serum albumin, TBST, tris-buffered saline with tween 20, K_d, dissociation constant, AG peptide, AGAQVAHGNEVAG, SE peptide, SEAGVNHGSAGQA, IA peptide, IAVENVHAQGLA, Oxo-AG peptide, 2-oxohistidine residue in AG peptide, Oxo-SE peptide, 2-oxohistidine residue in SE peptide, Oxo-IA peptide, 2-oxohistidine residue in IA peptide.

Summary

Cellular proteins are constantly damaged by reactive oxygen species generated by cellular respiration. Due to its metal-chelating property, histidine residues are easily oxidized in the presence of Cu/Fe ions and H₂O₂ via metal-catalyzed oxidation, usually converted to 2-oxohistidine. We hypothesize that cells may have evolved antioxidant defenses against the generation of 2-oxohistidine residues on proteins, and therefore there would be cellular proteins which specifically interact with this oxidized side chain. Using two chemically synthesized peptide probes containing 2-oxohistidine, high-throughput interactome screening was conducted using the E. coli K12 proteome microarray containing >4200 proteins. Ten interacting proteins were successfully validated using fluorescence polarization assay through a third peptide probe of different sequence, as well as binding constant measurements. We discovered 9 out of 10 identified proteins seem to be involved in redox-related cellular functions. We also built the functional interaction network to reveal their interacting proteins. The network showed our interacting proteins were enriched in oxido-reduction process, ion binding, and carbon metabolism. A consensus motif was identified among these 10 bacterial interacting proteins based on bioinformatic analysis, which also appeared to be present on human S100A1 protein. The preferential binding of S100A1 with 2-oxohistidine over histidine was successfully validated using all three peptide probes, suggesting that the capacity to recognize 2-oxohistidine modification may be evolutionarily conserved from bacteria to humans. Besides, we found our consensus motif among our identified proteins, including bacteria and human, were all alpha-helix form and faced the outside of proteins which mean the motif has a chance to interact with the other proteins. The combination of chemically engineered peptide probes with proteome microarrays proves to be an efficient discovery platform for protein interactomes of unusual post-translational modifications, sensitive enough to detect even the insertion of a single oxygen atom in this case.

Introduction

The complexity of the proteome arises in a large part due to the hundreds of post-translational modifications (PTMs) already discover. Many PTMs are enzyme-catalyzed, such as phosphorylation, glycosylation, or ubiquitination (1, 2), but there are also numerous non-enzymatic PTMs caused by chemical reactions between reactive molecules and protein side chains, such as glycation, nitrosylation, and oxidation by reactive oxygen species (ROS) (3, 4). As protein side chains are enzymatically modified, there are generally specialized factors in the cell to recognize such changes. For instance, 14-3-3 family protein can recognize protein phosphorylation motifs (5) and various lectins can recognize protein glycosylation (6). However, recognition factors may also exist for non-enzymatic PTMs, such as receptor for advanced glycation end-products (RAGE) (7). In this study we seek to uncover cellular binding factors for 2-oxohistidine, the oxidized product of histidine, which is an important but less understanding non-enzymatic PTM.

The generation of ROS is an unavoidable consequence of cellular respiration, which leads to the oxidation of proteins, lipids, and nucleic acids (4, 8). ROS play regulatory roles in cellular signaling pathways under low levels (9), but high levels of ROS are cytotoxic and lead to the accumulation of damaged cellular components (10, 11). The reactions of proteins with ROS may lead to almost 100 side chain modifications (12, 13). Histidine is highly susceptible to ROS damage, because it has strong metal chelation affinities and often constitutes the binding site for metal ions (14, 15). The presence of H₂O₂ and redox-active metals (Cu and Fe) can lead to metal-catalyzed oxidation (MCO, also called Fenton-type chemistry), which converts histidine side chains to 2-oxohistidine (16, 17).

The conversion of histidine to 2-oxohistidine alters its charge state, hydrogen bonding property, and metal chelation affinity, and hence may have seriously impact on protein structure and function. The net reaction is oxygen insertion (+16 Da), which makes it an irreversible PTM. It is unclear if cells simply tolerate such damages on histidines or employ active mechanisms to recognize them and use them as redox sensors or as damage markers for promoting protein degradation. The only known biological function of 2-oxohistidine is to serve as a redox sensor on bacterial transcription factor PerR (18), while other studies have used 2-oxohistidine as a stable marker of protein damage during oxidative stress (12, 19).

Judging by the potential biological significance of 2-oxohistidine modification, we hypothesized that there may be cellular factors to recognize it. Previous research on 2-oxohistidine had been impeded by the difficulty in generating this side chain with reasonable yields. Recently, we managed to greatly improve the yield of 2-oxohistidine conversion by optimizing MCO reaction conditions using the copper/ascorbate system (20), allowing us to synthesize and purify peptide probes containing 100% 2-oxohistidine for this study.

Here, we used 2-oxohistidine-containing peptides to mimic the oxidative conversion of histidine residues on native proteins. Then, we utilized the E. coli K12 proteome chip to identify 2-oxohistidine-interacting proteins via high-throughput screening, and the interactors turned out to be largely involved redox-related metabolism. From the bacterial interactors we predicted a consensus binding motif, which could be validated across different species and correctly predicted S100A1 as a human binding factor for 2-oxohistidine. Thus, recognition of 2-oxohistidine appears to be an evolutionarily conserved capacity from bacteria to human.

Experimental Procedures

Fabrication of E. coli K12 proteome chip

The high throughput protein expression, protein purification, and protein printing were modified from the previous study (21). Briefly, we expressed and purified E.coli K12 protein in 96-well plate format and subsequently printed the proteome microarray. All purified proteins were spotted in duplicate on each aldehyde slide (BaiO, China) by SmartArrayer 136 (CapitalBio, China) at 4°C. After printing proteins, the proteome microarray chips were kept at 4°C for protein immobilization on the slides for 12 h. In the end, the chips were stored at -80°C before probing with samples.

Peptide oxidation

Solutions containing 1 mM peptide, 5 mM Cu2+ and 200 mM sodium ascorbate were exposed to air with gentle shaking at 37 °C for 24 hrs (AG and SE peptide) or 6 hrs (IA peptide). The oxidation reaction was quenched with 20 mM EDTA and analyzed by reverse-phase HPLC (10-30% acetonitrile and 0.1% TFA in water, C18 column from Dr. Maisch, Ammerbuch, Germany) to determine the reaction yield. For LC-MS/MS analysis of crude reaction mixtures and HPLC fractions, 10 μL samples was acidified with 2 μL 10% TFA and desalted with ZipTip (Millipore, Billerica, MA) following manufacturer’s protocols. Oxidized peptides were purified by semi-preparative HPLC (C18 column, Dr. Maisch). LC-MS/MS experiments were conducted under previously reported conditions (20).

Peptide labeling

Oxidized and non-oxidized peptides were dissolved in 50 mM sodium borate buffer at pH 7.5 and analyzed by HPLC to determine peptide concentration by 210 nm absorbance. DyLight-conjugated NHS esters were dissolved in anhydrous DMF to 10 mg/mL and added to peptide solutions for 1 hr incubation at room temperature, at the following fluorophore/peptide ratios: DyLight 650:AG =3:1, DyLight 650:SE = 5:1, DyLight 650:oxo-IA = 1.5:1; DyLight 550:oxo-AG = 5:1, DyLight 550:oxo-SE = 7:1, DyLight 550:IA = 3:1. Labeled peptides were analyzed and purified by HPLC as described above. Labeled products were verified by LC-MS/MS, and quantified by absorbance measurements based on fluorophore properties.

E. coli K12 proteome chip assays with 2-oxohistidine peptides

The chips were first blocked with 3% bovine serum albumin (BSA) (Sigma-Aldrich, US) for 5 min. Ten μM of DyLight^TM 550-conjugated 2-oxohistidine peptide and DyLight^TM 650-conjugated non-oxohistidine peptide were probed together onto the chip with LifterSlips^TM (Thermo Scientific, US) at room temperature for 45 min. Finally, the chips were washed by Tris-buffered saline-Tween 20 (TBST) in an orbital shaker three times and 5 min each time. The chip was dried by centrifugation and then scanned with a LuxScan^TM microarray scanner (CapitalBio, China). Signal intensities, foreground median subtract background median, were acquired and analyzed using GenePix Pro 6.0 software. Then, we used quantile normalization to normalize the signal intensity from both 2-oxohistidine containing probes and non-oxohistidine containing probes. To identify positive 2-oxohistidine interacting proteins, four cutoffs were set. 1) The signal from experimental groups was greater than 1.5 standard deviations away from the mean for experimental groups. 2) To get the large signal difference between experimental groups and negative controls, the delta, defined as signal difference between experimental group and control group, was greater than 1.5 standard deviations away from the mean for all deltas. 3) To exclude the non-specific binding to 2-oxohistidine residue, the signal from the negative control was less than 1.5 standard deviations away from the mean for control group. 4) To remove the irreproducible hits among triplicate chip assays, the student’s t-test p-values between experimental groups and negative controls were less 0.05.

Heat Map

The R programming language (22) was used to display heat map. The data was presented by signal intensity of foreground subtract background. The gplots package (23) was used for classifying 2-oxohistidine containing peptides and non-oxohistidine containing peptides in hierarchy.

Functional interaction analysis

The identified proteins were used for functional interaction analyses by using EcID (24) and Cytoscape (25). Briefly, the files of EcID entities and EcID pairs were downloaded from EcID database. Before mapping identified proteins to their EcID entities and EcID pairs, we removed the pairs which based on the prediction mode, such as phylogenetic profiles, gene neighborhood, mirror tree, insilicon 2 hybrid, or context mirror. After mapping, we used Cytoscape to generate the functional interaction network, and visualized the identified proteins and their interacting proteins. Later on, we used AmiGO 2 (26) and KOBAS 2.0 (27) to generated gene ontology (GO) (28) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (29) results, respectively.

Fluorescence polarization assay

After blocking the 96-well black plate (Thermo Scientific, US) with 1% BSA at room temperature for 1 h, the identified proteins was added to the plate. The concentrations of 10 identified proteins (ThrS, YqjG, YajL, HemE, IlvA, PrpD, Zwf, Eda, Gor, and PqqL) were 12.0, 25.7, 10.7, 15.6, 3.4, 18.6, 19.5, 11.8, 26.1, and 5.9 μM, respectively. And the concentrations of BSA, as a negative control, were as same as the protein they compared to. Ten nM of DyLight^TM 550-conjugated 2-oxohistidine peptide was incubated with protein or BSA in a Micromixer MX4 (FINEPCR, South Korea) at room temperature for 1 h. After incubation, the degree of polarization of each well was detected by a Synergy 2 (BioTek, US), using an excitation wavelength of 540 nm and an emission wavelength of 590 nm with a dichroic mirror of 570 nm.

Measurement of dissociation constant (K_d)

Identified proteins and S100A1 (Abnova, Taiwan) were printed on aldehyde chips in a multiple-well format. After printing, the chips were immobilized at 4 °C for 12 h and then stored at -80 °C. The printed chips were blocked at room temperature for 5 min with 3% BSA. Two folds serial-diluted DyLight^TM 550-conjugated 2-oxohistidine peptides, DyLight^TM 650-conjugated non-oxohistidine peptides, and quenched fluorescent dyes were probed onto the wells of the chip individually with Multi-Well Microarray Hybridization Cassettes (Arrayit, US), and incubated at room temperature for 45 min. The fluorescent dyes, DyLight^TM 550 and DyLight^TM 650, were already quenched by 5M Tris-HCl (Bionovas, Canada). To check whether calcium affects interaction between S100A1 and 2-oxohistidine, 1 mM CaCl₂ was added in the assay buffer. After washes with TBST, the chips were dried by centrifugation and then scanned with a microarray scanner. The K_d value was calculated by double-reciprocal plot analysis which y is one divided by fluorescence intensity, and x is one divided by peptide concentration. Set the regression line formula in the form of y = ax, which “a” is the slope of regression line. The K_d value will be “a” times concentration of identified protein.

Motif Search with GLAM2

All identified proteins were converted to FASTA format and analyzed by Gapped Local Alignment of Motifs (GLAM2) (30) for surveying consensus motif. The parameters of GLAM2 were set as default. The resultant motif was then searched in entire E. coli K12 proteome and human proteome by GLAM2SCAN (30).

Protein 3D structure and secondary structure prediction

All protein 3D structures were provided by their provider (31-38) and RCSB PDB (39). The colors in protein 3D structures were visualized by RasMol software (40). We used the EcoGene 3.0 (41) which contains the QUARK prediction method (42) to predict the secondary structure of those proteins which do not have protein 3D structures.

Results

Many researches revealed that the 2-oxohistidine residue had been discovered in several peptides or proteins (16, 43-51). We used the E. coli K12 proteome chip to identify proteins which can bind specifically to 2-oxohistidine residue. To accomplish our purpose, we fabricated the E. coli K12 proteome chips, generated the 2-oxohistidine containing peptides, and probed these peptides with E. coli K12 proteome chips. After identified the positive hits, we used fluorescence polarization assays to validate the interactions and measured the binding affinity by dose-response measurements. Then, we surveyed the consensus motif among these identified proteins and applied to human proteome to look for the possible human 2-oxohistidine interacting proteins. Finally, we used the functional interaction network to find out the possible interacting proteins and used GO and KEGG to figure out possible process and pathway (Fig. 1).

Oxidation of peptide histidine residue

Histidine residues are placed in the middle of 12-mer or 13-mer peptides to eliminate possible charge effects at N-terminus and C-terminus, creating a context similar to proteins. Easily oxidized amino acids, such as methionine, cysteine, tyrosine, tryptophan phenylalanine, lysine, and arginine, are avoided. Three peptides containing a single histidine residue and random selections of other residues, namely AGAQVAHGNEVAG (AG), SEAGVNHGSAGQA (SE), and IAVENVHGGLA (IA), were used for chip assays. We carried out MCO reaction using the copper/ascorbate/air system shown in Figure 2. The HPLC yield of mono-histidine peptides AG and SE were around 10%, and for IA peptide around 20% (Fig. 2).

E. coli K12 proteome chip assays

To investigate 2-oxohistidine interacting proteins, AGAQVAH*GNEVAG (Oxo-AG peptide) and SEAGVNH*GSAGQA (Oxo-SE peptide) were conjugated to DyLight^TM 550 fluorophore molecular probes. Non-oxidized AG and SE peptides were conjugated to DyLight^TM 650 as negative controls. In the chip assay, 2-oxohistidine containing peptide and its negative control were probed with E. coli K12 proteome chip in triplicate (Fig. 3). The examples of 2-oxohistidine interacting proteins compared with non-oxohistidine containing peptide profiling were shown in Figure 4.

To identify the specific hits to 2-oxohistidine peptides, we set several cutoffs. First, we chose the hits had strong intensity in experimental groups. Second, we wanted the hits had high signal in experimental groups and low signal in negative controls. Thus, we chose the hits had large difference between experimental groups and negative controls. Third, although we chose the hits had large difference between two groups, there still were some strong signals in negative controls. To exclude this kind of non-specific binding to 2-oxohistidine residue, we removed the hits which greater than 1.5 standard deviationa away from the mean for negative controls. Fourth, in order to have reproducibility results among triplicate chip assays, we excluded the hits which had large variances as we described in the section of experimental procedures. Under such criteria, 38 and 20 protein hits were found to bind oxo-SE peptide and oxo-AG peptide, respectively (supplementary Table S1-S2). To avoid the non-specific binding due to the different peptide sequences, we chose the hits shared by both 2-oxohistidine peptides among those proteins. Only 10 proteins (ThrS, YqjG, YajL, HemE, IlvA, PrpD, Zwf, Eda, Gor, and PqqL) were identified by both 2-oxohistidine containing peptides (Table 1).

We used heat map to visualize the intensity of these 10 identified proteins among 2-oxohistidine and non-oxohistidine containing probing results (Fig. 5). The heat map shows that our 10 identified proteins clearly classified the 2-oxohistidine peptides from non-oxohistidine peptides.

Functional interaction analysis

We exploited EcID to find our 2-oxohistidine interacting proteins’ partners that indirectly interacted to 2-oxohistidine. The EcID database (Escherichia coli Interaction Database) (24) provided a framework for the integration of several interactional source, including EcoCyc (metabolic pathways, protein complexes and regulatory information), KEGG (metabolic pathways), MINT and IntAct (protein interactions), high-throughput experiment (protein complexes), and iHOP (text mining).

In this study, we only selected interactions from experimental mode which was proved by many databases and the results would be more reliable and confident. We chose the interacting proteins that had at least interacted 3 out of the 10 identified 2-oxohistidine interacting proteins. As shown in Figure 6, four 2-oxohistidine interacting proteins (thrS, zwf, eda, and ilvA) were ‘‘hubs’’ that connected many interacting proteins in the network. From this functional interaction analysis, 26 interacting proteins were found to have interactions with at least 3 out of the 10 identified proteins. We further analyzed this functional interaction network, including 10 identified proteins and 26 interacting proteins, by using AmiGO 2 (26) and KOBAS 2.0 (27) to provide the GO (28) and KEGG (29) results, respectively (supplementary Table S3-S5). Interestingly, fifteen out of the 36 proteins (~40%) were in the oxidation-reduction process, which shows significant enrichment (p < 0.05). Table 2 summarizes the related GO terms and KEGG pathways. Oxidation-reduction process is a metabolic process that involved in the transfer of electrons between chemical species (52). This result suggested that our identified 2-oxohistidine interacting proteins and their interacting proteins from the network may involve in the oxidation-reduction process. In the molecular function, ion binding and cofactor binding were enriched in our network. This result suggested that our proteins may interact with metal ion which can lead to MCO reaction. Besides, oxoacid metabolic process and carbon metabolism were also discovered. These kinds of metabolism usually accompany with energy metabolism that the reducing power and ROS may also carry out in the process (53). Changes to the oxidation state of a molecule were frequently carried out as a secondary metabolite were synthesized or modified (54). Therefore, the biosynthesis of secondary metabolites was also enriched in our interaction network. These data showed that identified proteins and their binding proteins may involve in the redox process or the oxygen sensitive environment to responsible for such kinds of oxidation change or be a protector or sensor to the oxidative stress.

Fluorescence polarization assays

Although there were positive results in the chip assays, we still could not exclude the possible bias of this kind of heterogeneous approach. Fluorescence polarization assay is a kind of homogeneous binding detection methods to mimic the interaction between two compounds in the cellular environment (55-59). Fluorescence polarization assays, investigation of the binding between two molecules were used to validate the 10 identified proteins in this study. Once the protein bound to fluorescent 2-oxohistidine containing peptides, a high degree of polarization was expected. As shown in Figure 7, all the 10 identified proteins had higher polarization than the negative control, BSA. Besides, the polarization distribution of two oxidative peptides was similar to each other. It indicated that interaction between proteins and 2-oxohistidine was not affected by different peptide sequences. The result confirmed that 10 identified proteins can bind to 2-oxohistidine in both AG and SE peptides.

Measurement of binding affinity

Dissociation constant (K_d) described the propensity of a ligand-protein complex to dissociate reversibly into its components. We measured the K_d of these identified proteins to oxidative peptides, normal peptides, and quenched fluorescent dyes by dose-response measurements. Fluorescent 2-oxohistidine containing peptides with different concentrations probed onto the slide, where the identified proteins were immobilized (supplementary Fig. S1A). Using double-reciprocal plot analysis, we calculated the K_d values for all identified proteins (supplementary Fig. S1B). The same procedures were done in normal peptides and fluorescent dyes, too. The result showed our 10 identified proteins had a strong affinity to 2-oxohistidine from 10^-8 to 10^-10 M, especially the hemE protein which had the highest K_d (~10^-10 M) in both 2-oxohistidine containing peptides (Table 3). We also found our proteins slightly preferred oxo-SE peptide than oxo-AG peptide, but the difference of K_d was not greater than one order of magnitude. On top of that, the K_d values from oxidative peptides were significant difference to the normal peptides, and quenched fluorescent dyes (p < 0.05). To check the interaction between 2-oxohistidine and identified proteins again in order to be certain. We used a third peptide, IAVENVH*QGLA (Oxo-IA peptide) and its negative control (IAVENVHQGLA, IA peptide), which had different peptide sequence and we also swapped their fluorescent dyes to each other to avoid the influence of fluorescent dyes. The result also showed the statistically significant difference to its negative controls (p < 0.05). This indicated that our 10 identified proteins had a strong binding affinity to 2-oxohistidine, and were not affected by different peptide sequences and different fluorescent dyes.

Motif Searching in E. coli proteome and human proteome

Based on fluorescence polarization and binding affinity results, we performed the GLAM2 (Gapped Local Alignment of Motifs) (30) to survey whether a consensus motif among these identified proteins. In this study, we found the consensus motif among these identified proteins is [SD][QV][AEDT]A[YIL][CE][AK][ARL][MV][AHK]?[KET][LV] [AYLF]E (Fig. 8). In addition, we used this motif to query entire E. coli K12 proteome by GLAM2SCAN (30). The result showed top ten ranking proteins containing this motif were identical to our identified proteins (Table 4). This indicates that motif was significantly unique in the entire E. coli K12 proteome (p < 0.05). We also applied this motif to entire human proteome, and found the ranked top one protein is S100 Calcium Binding Protein A1 (S100A1), which is a member of the S100 family (supplementary Table S6).

After motif screening in E. coli and human proteome, we further investigated the secondary structure of the motif in our identified proteins and S100A1 by using protein 3D structures (Fig. 9). However, there were 3 proteins (hemE, zwf, and pqqL) were not available. For these three proteins, we used the QUARK prediction method to predict their secondary structures. By proteins 3D structure analysis or QUARK prediction, the result showed that this motif was usually an alpha-helix in these proteins except for yajL, which contains 36% beta-sheet and 64% alpha-helix in the motif (Table 5). Besides, we found these kinds of alpha-helix formed motifs generally faced the outside of the proteins which mean they had chance to interact with outside molecules. Our finding suggested that 2-oxohistidine recognized motif was an alpha-helical structure and conversed between E. coli and human.

K_d measurement between human S100A1 protein and the oxidative peptides

To validate the interaction of human S100A1 protein we found by GLAM2SCAN on entire human proteome, we calculated the K_d values according to dose-response measurements for all oxidative peptides, including oxo-AG peptide, oxo-SE peptide and oxo-IA peptide. The result showed that S100A1 protein had a strong affinity to all 2-oxohistidine containing peptides and significant difference to the other unoxidized peptides and fluorescent dyes (p < 0.05) (Table 6). The binding affinity of S100A1 to 2-oxohistidine were 10-fold to 100-fold higher than the negative controls, indicating that S100A1 actually had an ability to bind to the 2-oxohistidine. Since we knew S100A1 is calcium binding protein, we wondered whether calcium would affect the interaction or not. The result showed calcium was not involved in the interaction of S100A1 to 2-oxohistidine peptides or the other groups (p > 0.1). This suggested the E. coli K12 proteome chip was able to be a feasible platform for motif screening in cross-species studies.

Discussion

Enzymatic and non-enzymatic PTMs are comparable in their diversity and chemical complexity, but past research efforts have mostly focused on the former, leaving a huge gap in our understanding of biological phenomena associated with non-enzymatic PTMs. Even though non-enzymatic PTMs are not generated by enzyme actions, there may still be specific enzymes to chemically reverse such modifications, or specific receptors to detect such modification. For example, the chemical oxidation of methionine to methionine sulfoxide can be reduced back to methionine by specific reductases MsrA and MsrB (60); RAGE can recognize protein glycation and lead to inflammatory responses (7). However, there are still many non-enzymatic PTMs for which the biological functions are little known.

Among non-enzymatic PTMs, 2-oxohistidine is particularly interesting because of its minimal size, involving the insertion of just one oxygen atom. It probably represents the smallest atom-scale alteration associated with a known PTM, and we investigated if cells have evolved the ability to monitor such a small change on the surface of proteins. Because histidine often plays critical roles in protein function, both structurally and catalytically, we hypothesized there would be cellular factors that specifically recognize 2-oxohistidine side chains, and this hypothesis was tested with specially synthesized peptide probes, and E.coli proteome chips.

Using three peptide probes with homogeneous 2-oxohistidine modification, we were able to identify 10 proteins that show preferential binding for 2-oxohistidine-containing peptides over non-oxidized control peptides (Table 1). Since these three probes have very different flanking sequences, it is very likely that we have identified proteins which specifically recognize side-chain differences between 2-oxohistidine and histidine, and we will refer to them as 2-oxohistidine recognition factors. Before this study, the recognition factors of 2-oxohistidine had never been proposed or identified.

In theory, the recognition of 2-oxohistidine could play several different biological roles. First, it may act as a redox sensor, similar to S-nitrosylation (61). Secondly, it may identify oxidatively damaged proteins and mark it for degradation. Third, it may trigger cellular stress responses and antioxidant pathways. Although there is no known involvement of 2-oxohistidine in different E. coli physiological pathways, several of the recognition factors in E. coli appear to be related to redox pathways and antioxidant pathways.

Among the 10 putative recognition factors identified via proteome array, 9 seem to be involved in redox-related cellular functions. Gor is a glutathione reductase, involved in the generation of glutathione, which maintains the reducing environment of the cell (62). YqjG is glutathionyl hydroquinone reductase, which utilizes glutathione to reduce a wide range of organic molecules (38). HemE is an uroporphyrinogen decarboxylase involved in the biothesis of the heme group, which is an important cofactor for antioxidant enzymes like catalase and peroxidase (63). Zwf is a glucose-6-phosphate dehydrogenase, which helps supply NADPH through the pentose phosphate pathway (64), and NADPH is a cofactor used as a reducing agent by many metabolic enzymes (65, 66). PqqL in E. coli is a putative zinc metalloprotease, but functionally it may be similar to pqqF in Klebsiella pneumoniae, which has a supportive role in pyrroloquinoline quinone biosynthesis (67). Pyroloquinoline quinone is a redox cofactor that provides reducing power for the cell, and also a ROS scavenger (68).

YajL is an anti-oxidative-stress chaperone, which promotes disulfide formation to help maintain order in the thiol proteome (69). Interestingly, the human homolog of yajL, DJ-1, is also an anti-oxidative stress protein, and its mutations are known to cause familial Parkinsonism (70). On the other hand, ilvA and thrS are both involved in threonine metabolism, and known to be regulated by oxygen levels in the cell. IlvA, a threonine dehydratase, converts threonine to 2-oxobutanoate, and its promoter is activated by oxygen (71). ThrS is a threonyl-tRNA synthetase, and potentially also an oxygen sensor in the cell through Cys182 oxidation (72). Eda, Entner-Doudoroff aldolase (also called KDPG aldolase), is involved in the Entner-Doudoroff pathway that generates pyruvate and NADPH by consuming glucose. Eda is a multi-functional aldolase which also catalyze the addition of pyruvate to electrophilic aldehydes to detoxify harmful byproducts generated by oxidative stress (73).

PrpD, a 2-methylcitrate dehydratase, does not appear to be directly involved in redox functions, but it converts propionyl-CoA into pyruvate through the methylcitrate cycle (74), and pyruvate can be utilized by the aforementioned eda to detoxify oxidized organic molecules with aldehydes. Therefore, all 10 putative recognition factors for 2-oxohistidine identified here appear to be involved in supplying reducing power to the cell or in oxygen-sensitive regulation of carbon metabolism. This strongly implies that recognition of 2-oxohistidine in E. coli may play certain roles in redox sensing and metabolic regulation, but further experiments are required to elucidate its actual function.

Using motif analysis by GLAM2 and GLAM2SCAN, we identified putative 2-oxohistidine binding motif from these 10 recognition factors, which turned out to be: [SD][QV][AEDT]A[YIL][CE][AK][ARL][MV][AHK]?[KET][LV][AYLF]E. We further validated this binding motif by searching for the highest-scoring match in the human proteome, which turned out to be DVDAVDKVMKELDE on S100A1 protein, and we verified that S100A1 indeed exhibited 2-oxohistidine binding affinity. S100A1 is a calcium binding protein highly expressed in the brain and heart, and its calcium binding affinity is greatly enhanced by the oxidative nitrosylation of Cys86 (75). It is believed to regulate calcium and nitric oxide signaling in neuronal cells, affecting neurotransmitter release as well as inflammation (76). Interestingly, since S100A1 is also secreted extracellularly (77), it may bind to oxidized amyloid beta (Aβ) with 2-oxohisitidine side chains, which are released from extracellular senile plaques which trap metals and generate ROS (45, 78, 79). Since Aβ is known to cause calcium misregulation (80), oxidative stress (81), and inflammatory response (82) in the brain, the interaction between S100A1 and oxidized Aβ through 2-oxohistidine recognition may play a role in Alzheimer’s disease (AD) pathogenesis, which warrants future investigation.

Our preliminary evidence suggests that both bacteria and humans have cellular factors which can recognize 2-oxohistidine side chains, and a conserved binding motif has been putatively identified. Through the course of evolution, the recognition of 2-oxohistidine may carry important cellular functions related to redox signaling. We have also shown that E. coli K12 proteome microarray is capable of being exploited as a motif library for screening small molecule binding, and that even a single-atom modification on the molecule may be recognized. We expect a wide application of this approach for studying the interaction of other post-translational modifications, such as phosphorylation, methylation, acetylation, amidation, thiolation, sulfation, nitrosylation, as well as many non-enzymatic PTMs. With regard to 2-oxohistidine, future work is required to elucidate how single-oxygen insertion can be recognized on the protein surface, and how recognizing this modification regulates biological functions.

References

1. Wold, F. (1981) In vivo chemical modification of proteins (post-translational modification). Annu Rev Biochem 50, 783-814

2. Walsh, C. T., Garneau-Tsodikova, S., and Gatto, G. J., Jr. (2005) Protein posttranslational modifications: the chemistry of proteome diversifications. Angew Chem Int Ed Engl 44, 7342-7372

3. Harding, J. J. (1985) Nonenzymatic covalent posttranslational modification of proteins in vivo. Adv Protein Chem 37, 247-334

4. Davies, M. J. (2005) The oxidative environment and protein damage. Biochim Biophys Acta 1703, 93-109

5. Morrison, D. K. (2009) The 14-3-3 proteins: integrators of diverse signaling cues that impact cell fate and cancer development. Trends Cell Biol 19, 16-23

6. Kilpatrick, D. C. (2002) Animal lectins: a historical introduction and overview. Biochim Biophys Acta 1572, 187-197

7. Sparvero, L. J., Asafu-Adjei, D., Kang, R., Tang, D., Amin, N., Im, J., Rutledge, R., Lin, B., Amoscato, A. A., Zeh, H. J., and Lotze, M. T. (2009) RAGE (Receptor for Advanced Glycation Endproducts), RAGE ligands, and their role in cancer and inflammation. J Transl Med 7, 17

8. Muller, F. L., Lustgarten, M. S., Jang, Y., Richardson, A., and Van Remmen, H. (2007) Trends in oxidative aging theories. Free Radic Biol Med 43, 477-503

9. Ray, P. D., Huang, B. W., and Tsuji, Y. (2012) Reactive oxygen species (ROS) homeostasis and redox regulation in cellular signaling. Cell Signal 24, 981-990

10. Martin, K. R., and Barrett, J. C. (2002) Reactive oxygen species as double-edged swords in cellular processes: low-dose cell signaling versus high-dose toxicity. Hum Exp Toxicol 21, 71-75

11. Jang, Y. Y., and Sharkis, S. J. (2007) A low level of reactive oxygen species selects for primitive hematopoietic stem cells that may reside in the low-oxygenic niche. Blood 110, 3056-3063

12. Shacter, E. (2000) Quantification and significance of protein oxidation in biological samples. Drug Metab Rev 32, 307-326

13. Xu, G., and Chance, M. R. (2007) Hydroxyl radical-mediated modification of proteins as probes for structural proteomics. Chem Rev 107, 3514-3543

14. Tainer, J. A., Roberts, V. A., and Getzoff, E. D. (1991) Metal-binding sites in proteins. Curr Opin Biotechnol 2, 582-591

15. Regan, L. (1993) The design of metal-binding sites in proteins. Annu Rev Biophys Biomol Struct 22, 257-287

16. Uchida, K., and Kawakishi, S. (1994) Identification of oxidized histidine generated at the active site of Cu,Zn-superoxide dismutase exposed to H2O2. Selective generation of 2-oxo-histidine at the histidine 118. J Biol Chem 269, 2405-2410

17. Lewisch, S. A., and Levine, R. L. (1995) Determination of 2-oxohistidine by amino acid analysis. Anal Biochem 231, 440-446

18. Traore, D. A., El Ghazouani, A., Jacquamet, L., Borel, F., Ferrer, J. L., Lascoux, D., Ravanat, J. L., Jaquinod, M., Blondin, G., Caux-Thang, C., Duarte, V., and Latour, J. M. (2009) Structural and functional characterization of 2-oxo-histidine in oxidized PerR protein. Nat Chem Biol 5, 53-59

19. Davies, M. J., Fu, S., Wang, H., and Dean, R. T. (1999) Stable markers of oxidant damage to proteins and their application in the study of human disease. Free Radic Biol Med 27, 1151-1163

20. Huang, C. F., Liu, Y. H., and Tai, H. C. (2015) Synthesis of peptides containing 2-oxohistidine residues and their characterization by liquid chromatography-tandem mass spectrometry. J Pept Sci 21, 114-119

21. Chen, C. S., Korobkova, E., Chen, H., Zhu, J., Jian, X., Tao, S. C., He, C., and Zhu, H. (2008) A proteome chip approach reveals new DNA damage recognition activities in Escherichia coli. Nat Methods 5, 69-74

22. Team, R. C. (2015) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing

23. Warnes, G. R., Bolker, B., Bonebakker, L., Gentleman, R., Huber, W., Liaw, A., Lumley, T., Maechler, M., Magnusson, A., and Moeller, S. (2009) gplots: Various R programming tools for plotting data. R package version 2

24. Andres Leon, E., Ezkurdia, I., Garcia, B., Valencia, A., and Juan, D. (2009) EcID. A database for the inference of functional interactions in E. coli. Nucleic Acids Res 37, D629-635

25. Shannon, P., Markiel, A., Ozier, O., Baliga, N. S., Wang, J. T., Ramage, D., Amin, N., Schwikowski, B., and Ideker, T. (2003) Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 13, 2498-2504

26. Carbon, S., Ireland, A., Mungall, C. J., Shu, S., Marshall, B., Lewis, S., Ami, G. O. H., and Web Presence Working, G. (2009) AmiGO: online access to ontology and annotation data. Bioinformatics 25, 288-289

27. Xie, C., Mao, X., Huang, J., Ding, Y., Wu, J., Dong, S., Kong, L., Gao, G., Li, C. Y., and Wei, L. (2011) KOBAS 2.0: a web server for annotation and identification of enriched pathways and diseases. Nucleic Acids Res 39, W316-322

28. Ashburner, M., Ball, C. A., Blake, J. A., Botstein, D., Butler, H., Cherry, J. M., Davis, A. P., Dolinski, K., Dwight, S. S., Eppig, J. T., Harris, M. A., Hill, D. P., Issel-Tarver, L., Kasarskis, A., Lewis, S., Matese, J. C., Richardson, J. E., Ringwald, M., Rubin, G. M., and Sherlock, G. (2000) Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 25, 25-29

29. Kanehisa, M., and Goto, S. (2000) KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 28, 27-30

30. Frith, M. C., Saunders, N. F., Kobe, B., and Bailey, T. L. (2008) Discovering sequence motifs with arbitrary insertions and deletions. PLoS Comput Biol 4, e1000071

31. Rajashankar, K. R., Kniewel, R., Solorzano, V., Lima, C.D., Burley, S.K., New York SGX Research Center for Structural Genomics (2004) Crystal Structure of 2-methylcitrate dehydratase.

32. Mittl, P. R., Berry, A., Scrutton, N. S., Perham, R. N., and Schulz, G. E. (1994) Anatomy of an engineered NAD-binding site. Protein Sci 3, 1504-1514

33. Gallagher, D. T., Gilliland, G. L., Xiao, G., Zondlo, J., Fisher, K. E., Chinchilla, D., and Eisenstein, E. (1998) Structure and control of pyridoxal phosphate dependent allosteric threonine deaminase. Structure 6, 465-475

34. Dock-Bregeon, A. C., Rees, B., Torres-Larios, A., Bey, G., Caillet, J., and Moras, D. (2004) Achieving error-free translation; the mechanism of proofreading of threonyl-tRNA synthetase at atomic resolution. Mol Cell 16, 375-386

35. Wright, N. T., Varney, K. M., Ellis, K. C., Markowitz, J., Gitti, R. K., Zimmer, D. B., and Weber, D. J. (2005) The three-dimensional solution structure of Ca(2+)-bound S100A1 as determined by NMR spectroscopy. J Mol Biol 353, 410-426

36. Wilson, M. A., Ringe, D., and Petsko, G. A. (2005) The atomic resolution crystal structure of the YajL (ThiJ) protein from Escherichia coli: a close prokaryotic homologue of the Parkinsonism-associated protein DJ-1. J Mol Biol 353, 678-691

37. Fullerton, S. W., Griffiths, J. S., Merkel, A. B., Cheriyan, M., Wymer, N. J., Hutchins, M. J., Fierke, C. A., Toone, E. J., and Naismith, J. H. (2006) Mechanism of the Class I KDPG aldolase. Bioorg Med Chem 14, 3002-3010

38. Green, A. R., Hayes, R. P., Xun, L., and Kang, C. (2012) Structural understanding of the glutathione-dependent reduction mechanism of glutathionyl-hydroquinone reductases. J Biol Chem 287, 35838-35848

39. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N., and Bourne, P. E. (2000) The Protein Data Bank. Nucleic Acids Res 28, 235-242

40. Sayle, R. A., and Milner-White, E. J. (1995) RASMOL: biomolecular graphics for all. Trends Biochem Sci 20, 374

41. Zhou, J., and Rudd, K. E. (2013) EcoGene 3.0. Nucleic Acids Res 41, D613-624

42. Xu, D., and Zhang, Y. (2012) Ab initio protein structure assembly using continuous structure fragments and optimized knowledge-based force field. Proteins 80, 1715-1735

43. Uchida, K., and Kawakishi, S. (1993) 2-Oxo-histidine as a novel biological marker for oxidatively modified proteins. FEBS Lett 332, 208-210

44. Lewisch, S. A., and Levine, R. L. (1999) Determination of 2-oxohistidine by amino acid analysis. Methods Enzymol 300, 120-124

45. Atwood, C. S., Huang, X., Khatri, A., Scarpa, R. C., Kim, Y. S., Moir, R. D., Tanzi, R. E., Roher, A. E., and Bush, A. I. (2000) Copper catalyzed oxidation of Alzheimer Abeta. Cell Mol Biol (Noisy-le-grand) 46, 777-783

46. Schoneich, C. (2000) Mechanisms of metal-catalyzed oxidation of histidine to 2-oxo-histidine in peptides and proteins. J Pharm Biomed Anal 21, 1093-1097

47. Gunther, M. R., Peters, J. A., and Sivaneri, M. K. (2002) Histidinyl radical formation in the self-peroxidation reaction of bovine copper-zinc superoxide dismutase. J Biol Chem 277, 9160-9166

48. Hovorka, S. W., Biesiada, H., Williams, T. D., Huhmer, A., and Schoneich, C. (2002) High sensitivity of Zn2+ insulin to metal-catalyzed oxidation: detection of 2-oxo-histidine by tandem mass spectrometry. Pharm Res 19, 530-537

49. Schoneich, C., and Williams, T. D. (2002) Cu(II)-catalyzed oxidation of beta-amyloid peptide targets His13 and His14 over His6: Detection of 2-Oxo-histidine by HPLC-MS/MS. Chem Res Toxicol 15, 717-722

50. Schiewe, A. J., Margol, L., Soreghan, B. A., Thomas, S. N., and Yang, A. J. (2004) Rapid characterization of amyloid-beta side-chain oxidation by tandem mass spectrometry and the scoring algorithm for spectral analysis. Pharm Res 21, 1094-1102

51. Inoue, K., Garner, C., Ackermann, B. L., Oe, T., and Blair, I. A. (2006) Liquid chromatography/tandem mass spectrometry characterization of oxidized amyloid beta peptides as potential biomarkers of Alzheimer's disease. Rapid Commun Mass Spectrom 20, 911-918

52. Boyer, R. F. (2005) Concepts in Biochemistry, 3 edition Ed., Wiley

53. Jensen, P. R., and Michelsen, O. (1992) Carbon and energy metabolism of atp mutants of Escherichia coli. J Bacteriol 174, 7635-7641

54. Dewick, P. M. (2009) Secondary Metabolism: The Building Blocks and Construction Mechanisms. Medicinal Natural Products, pp. 7-38, John Wiley & Sons, Ltd

55. Lundblad, J. R., Laurance, M., and Goodman, R. H. (1996) Fluorescence polarization analysis of protein-DNA and protein-protein interactions. Mol Endocrinol 10, 607-612

56. Jameson, D. M., and Seifried, S. E. (1999) Quantification of protein-protein interactions using fluorescence polarization. Methods 19, 222-233

57. Allen, M., Reeves, J., and Mellor, G. (2000) High throughput fluorescence polarization: a homogeneous alternative to radioligand binding for cell surface receptors. J Biomol Screen 5, 63-69

58. Parker, G. J., Law, T. L., Lenoch, F. J., and Bolger, R. E. (2000) Development of high throughput screening assays using fluorescence polarization: nuclear receptor-ligand-binding and kinase/phosphatase assays. J Biomol Screen 5, 77-88

59. Moerke, N. J. (2009) Fluorescence Polarization (FP) Assays for Monitoring Peptide-Protein or Nucleic Acid-Protein Binding. Curr Protoc Chem Biol 1, 1-15

60. Kim, H. Y., and Gladyshev, V. N. (2007) Methionine sulfoxide reductases: selenoprotein forms and roles in antioxidant protein repair in mammals. Biochem J 407, 321-329

61. Martinez-Ruiz, A., Araujo, I. M., Izquierdo-Alvarez, A., Hernansanz-Agustin, P., Lamas, S., and Serrador, J. M. (2013) Specificity in S-nitrosylation: a short-range mechanism for NO signaling? Antioxid Redox Signal 19, 1220-1235

62. Mittl, P. R., and Schulz, G. E. (1994) Structure of glutathione reductase from Escherichia coli at 1.86 A resolution: comparison with the enzyme from human erythrocytes. Protein Sci 3, 799-809

63. Nishimura, K., Nakayashiki, T., and Inokuchi, H. (1993) Cloning and sequencing of the hemE gene encoding uroporphyrinogen III decarboxylase (UPD) from Escherichia coli K-12. Gene 133, 109-113

64. Henard, C. A., Bourret, T. J., Song, M., and Vazquez-Torres, A. (2010) Control of redox balance by the stringent response regulatory protein promotes antioxidant defenses of Salmonella. J Biol Chem 285, 36785-36793

65. Lim, S. J., Jung, Y. M., Shin, H. D., and Lee, Y. H. (2002) Amplification of the NADPH-related genes zwf and gnd for the oddball biosynthesis of PHB in an E. coli transformant harboring a cloned phbCAB operon. J Biosci Bioeng 93, 543-549

66. Shi, F., Li, K., Huan, X., and Wang, X. (2013) Expression of NAD(H) kinase and glucose-6-phosphate dehydrogenase improve NADPH supply and L-isoleucine biosynthesis in Corynebacterium glutamicum ssp. lactofermentum. Appl Biochem Biotechnol 171, 504-521

67. Xiong, X., Yang, L., Han, X., Wang, J., and Zhang, W. (2010) [Knockout and function analysis of pqqL gene in Escherichia coli]. Wei Sheng Wu Xue Bao 50, 1380-1384

68. Misra, H. S., Khairnar, N. P., Barik, A., Indira Priyadarsini, K., Mohan, H., and Apte, S. K. (2004) Pyrroloquinoline-quinone: a reactive oxygen species scavenger in bacteria. FEBS Lett 578, 26-30

69. Le, H. T., Gautier, V., Kthiri, F., Malki, A., Messaoudi, N., Mihoub, M., Landoulsi, A., An, Y. J., Cha, S. S., and Richarme, G. (2012) YajL, prokaryotic homolog of parkinsonism-associated protein DJ-1, functions as a covalent chaperone for thiol proteome. J Biol Chem 287, 5861-5870

70. Bonifati, V., Rizzu, P., van Baren, M. J., Schaap, O., Breedveld, G. J., Krieger, E., Dekker, M. C., Squitieri, F., Ibanez, P., Joosse, M., van Dongen, J. W., Vanacore, N., van Swieten, J. C., Brice, A., Meco, G., van Duijn, C. M., Oostra, B. A., and Heutink, P. (2003) Mutations in the DJ-1 gene associated with autosomal recessive early-onset parkinsonism. Science 299, 256-259

71. Lopes, J. M., and Lawther, R. P. (1989) Physical identification of an internal promoter, ilvAp, in the distal portion of the ilvGMEDA operon. Gene 76, 255-269

72. Wu, J., Fan, Y., and Ling, J. (2014) Mechanism of oxidant-induced mistranslation by threonyl-tRNA synthetase. Nucleic Acids Res 42, 6523-6531

73. Murray, E. L., and Conway, T. (2005) Multiple regulators control expression of the Entner-Doudoroff aldolase (Eda) of Escherichia coli. J Bacteriol 187, 991-1000

74. Brock, M., Maerker, C., Schutz, A., Volker, U., and Buckel, W. (2002) Oxidation of propionate to pyruvate in Escherichia coli. Involvement of methylcitrate dehydratase and aconitase. Eur J Biochem 269, 6184-6194

75. Lenarcic Zivkovic, M., Zareba-Koziol, M., Zhukova, L., Poznanski, J., Zhukov, I., and Wyslouch-Cieszynska, A. (2012) Post-translational S-nitrosylation is an endogenous factor fine tuning the properties of human S100A1 protein. J Biol Chem 287, 40457-40470

76. Wright, N. T., Cannon, B. R., Zimmer, D. B., and Weber, D. J. (2009) S100A1: Structure, Function, and Therapeutic Potential. Curr Chem Biol 3, 138-145

77. Perrin, R. J., Craig-Schapiro, R., Malone, J. P., Shah, A. R., Gilmore, P., Davis, A. E., Roe, C. M., Peskind, E. R., Li, G., Galasko, D. R., Clark, C. M., Quinn, J. F., Kaye, J. A., Morris, J. C., Holtzman, D. M., Townsend, R. R., and Fagan, A. M. (2011) Identification and validation of novel cerebrospinal fluid biomarkers for staging early Alzheimer's disease. PLoS One 6, e16032

78. Curtain, C. C., Ali, F., Volitakis, I., Cherny, R. A., Norton, R. S., Beyreuther, K., Barrow, C. J., Masters, C. L., Bush, A. I., and Barnham, K. J. (2001) Alzheimer's disease amyloid-beta binds copper and zinc to generate an allosterically ordered membrane-penetrating structure containing superoxide dismutase-like subunits. J Biol Chem 276, 20466-20473

79. Schoneich, C., and Williams, T. D. (2003) CU(II)-catalyzed oxidation of Alzheimer's disease beta-amyloid peptide and related sequences: remarkably different selectivities of neurotoxic betaAP1-40 and non-toxic betaAP40-1. Cell Mol Biol (Noisy-le-grand) 49, 753-761

80. Kuchibhotla, K. V., Goldman, S. T., Lattarulo, C. R., Wu, H. Y., Hyman, B. T., and Bacskai, B. J. (2008) Abeta plaques lead to aberrant regulation of calcium homeostasis in vivo resulting in structural and functional disruption of neuronal networks. Neuron 59, 214-225

81. Garcia-Alloza, M., Dodwell, S. A., Meyer-Luehmann, M., Hyman, B. T., and Bacskai, B. J. (2006) Plaque-derived oxidative stress mediates distorted neurite trajectories in the Alzheimer mouse model. J Neuropathol Exp Neurol 65, 1082-1089

82. Du Yan, S., Zhu, H., Fu, J., Yan, S. F., Roher, A., Tourtellotte, W. W., Rajavashisth, T., Chen, X., Godman, G. C., Stern, D., and Schmidt, A. M. (1997) Amyloid-beta peptide-receptor for advanced glycation endproduct interaction elicits neuronal expression of macrophage-colony stimulating factor: a proinflammatory pathway in Alzheimer disease. Proc Natl Acad Sci U S A 94, 5296-5301

Figure Legends

Figure 1. Overall strategy for the identification of 2-oxohistidine interacting proteins using E. coli K12 proteome chip. We expressed and purified ~4,300 E. coli proteins in high-throughput to fabricate the E. coli K12 proteome chip. We used an improved condition to obtain 2-oxohistidine peptides in high purity. 2-Oxohistidine peptides were then probed to E. coli K12 proteome chip and identified the preferential binding proteins. We also built their functional interaction network to investigate their biology. Fluorescence polarization assays were used to validate the identified proteins. We conducted dose-response fluorescence assays to measure the K_d of these proteins. Furthermore, we used GLAM2 to search consensus motif among these identified proteins and also applied this motif to entire E. coli K12 proteome and human proteome by GLAM2SCAN.

Figure 2. Summary scheme for the synthesis of 2-oxohistidine-containing peptides. The process was synthesized by using metal-catalyzed oxidation, and the histidine side chain on peptides was converted to 2-oxohistidine.

Figure 3. Schematic of E. coli K12 proteome chip assays with 2-oxohistidine peptide probes. To detect the 2-oxohistidine interacting proteins, E. coli K12 proteome chips were probed with 2-oxohistidine-containing peptides and un-oxidized control peptides labeled with different fluorophores. Each protein was printed in duplicate on the chips.

Figure 4. Representative images of the E. coli K12 proteome chips probed with 2-oxohistidine containing peptide (Oxo-SE peptide) and non-oxohistidine containing peptide (SE peptide). The representative positive hits (yqjG and thrS) and non-specific binding protein (yeiG) on the chip were enlarged from sample images of oxo-SE peptide and SE peptide, respectively. The contrast and brightness of images had been equally adjusted using the same parameters.

Figure 5. The heat map of 10 identified proteins. The heat map showed the classification of 10 identified proteins in oxo-AG, oxo-SE, AG and SE chip assay probing result. Each peptide probes had triplicate results. The R programming language and gplots package were used to display heat map.

Figure 6. The functional interaction network of the 10 identified proteins and 26 interacting proteins. The interaction pairs for 10 identified proteins were downloaded from EcID database, and functional interaction network was visualized by Cytoscape. We only showed the interacting proteins that interact with at least 3 out of 10 identified proteins, and 26 interacting proteins were identified. Four out of 10 identified proteins, eda, ilvA, zwf, and thrS, had many interactions and considered to be hubs. Square shapes represented the 10 identified proteins, and round shapes represented the 26 interacting proteins. The node color showed the number of interactions, the red is greater than 10 interactions, the green is greater than 5 interactions, and the others are yellow which smaller than 5 interactions. The thicker edge lines symbolized that more databases showed the interaction between 2 proteins.

Figure 7. Validation of the interactions between 2-oxohistidine peptides and identified proteins using fluorescence polarization assays. In fluorescence polarization assays, the polarizations of the tested proteins were compared with same concentration of BSA, as a negative control. A. The fluorescence polarization assays for oxo-AG peptide and identified proteins. B. The fluorescence polarization assays for oxo-SE peptide and identified proteins. The black bar is identified proteins and the gray bar is BSA. The asterisks mean the polarizations of the identified proteins were significant difference to the BSA control (p < 0.05).

Figure 8. Consensus motif among the 10 validated proteins. A motif [SD][QV][AEDT]A [YIL][CE][AK][ARL][MV][AHK]?[KET][LV][AYLF]E was identified by GLAM2. The table showed the protein sequences of 10 validated proteins aligned with consensus motif.

Figure 9. Protein 3D structure of E. coli identified proteins and human S100A1. Only 7 E. coli identified proteins (thrS, yqjG, yajL, ilvA, prpD, eda, gor) and human S100A1 had protein 3D structures. The protein 3D structures were provided by their provider and RCSB PDB, and visualized by RasMol software. Beta-sheets are shown in yellow bands; alpha-helices are shown as pink bands and random coil as white lines. The blue bands are the consensus motif we found by GLAM2. Only yqjG, yajL, prpD, gor and S100A1 were provided by homodimer structure. The other is the monomer structure.

Tables

Table 1. 2-Oxohistidine interacting proteins identified by E. coli K12 proteome chips. There were 38 and 20 proteins are identified by oxo-AG peptide and oxo-SE peptide chip assays, respectively. To avoid the non-specific binding due to the different peptide sequences, we only chose the hits were shared by both 2-oxohistidine containing peptides (oxo-AG peptide and oxo-SE peptide).

Accession ID	Protein Symbol	Protein Name	Protein Function
EG11001	thrS	Threonyl-tRNA synthetase	An enzyme involved in protein synthesis which is regulated by aerobic and anaerobic metabolisms
EG12746	yqjG	Glutathionyl-hydroquinone reductase	Reduction of organic small molecules
EG13272	yajL	Anti-oxidative stress chaperone	A covalent chaperone for thiol-containing proteome, also promoting disulfide formation
EG11543	hemE	Uroporphyrinogen decarboxylase	Involved in the synthesis of heme group, which is a critical cofactor for antioxidant enzymes
EG10493	ilvA	Threonine dehydratase	A metabolic enzyme that converts threonine to 2-oxobutanoate, regulated by an oxygen-responsive promoter
EG13603	prpD	2-Methylcitrate dehydratase	A metabolic enzyme in the methylcitrate cycle that converts propionyl-CoA to pyruvate
EG11221	zwf	Glucose-6-phosphate dehydrogenase	A metabolic enzyme in the pentose-phosphate pathway that supplies reducing power to cells generating NADPH
EG10256	eda	KDPG aldolase	An enzyme in the Entner-Doudoroff pathway, also a multi-function aldolase to detoxify aldehydes generated by oxidative stress
EG10412	gor	Glutathione reductase	An enzyme that generates glutathione to maintain a reducing environment in the cell
EG11744	pqqL	Putative periplasmic M16 family zinc metalloendopeptidase	An enzyme involed in pyrroloquinoline quinone biosynthesis, which is a redox cofactor that supplies reducing power

Table 2. Summary for functional analysis of 36 proteins from functional interaction network. The 36 proteins, including 10 identified proteins and 26 interacting proteins, were used to do the functional analysis. The GO and KEGG results were generated by AmiGO 2 and KOBAS 2.0, respectively. We summarizedthe related GO terms and KEGG pathways in this table. The entirely detailed information of GO and KEGG results were shown on supplementary Table S3-S5.

GO Term (Biological process)	ID	Protein involved numbers	p-value
Oxoacid metabolic process	GO:0043436	21	5.29E-08
Oxidation-reduction process	GO:0043436	15	5.03E-03
GO Term (Molecular function)	ID	Protein involved numbers	p-value
Ion binding	GO:0043167	27	3.43E-05
Cofactor binding	GO:0048037	15	2.38E-06
KEGG	ID	Protein involved numbers	p-value
Carbon metabolism	eco01200	11	2.89E-03
Biosynthesis of secondary metabolites	eco01110	19	1.25E-02

Table 3. K_d values for 2-oxohistidine peptides binding to identified proteins. All K_d values were determined by dose-response measurements. Different concentration of fluorescent oxidative peptides, normal peptides, and fluorescent dye were probed onto the chip which 10 identified proteins immobilized. Based on the dose-response, we could use double-reciprocal plot to calculate the K_d values. We also used the oxo-IA peptide, which was different peptide sequence and labeled different fluorescent dye, and its negative control (IA peptide) to confirm the binding between identified proteins and 2-oxohistidine.

	Oxidative Peptides			Normal Peptides			Fluorescent Dyes
Name	DyLight 550 oxo-AG	DyLight 550 oxo-SE	DyLight 650 oxo-IA	DyLight 650 AG	DyLight 650 SE	DyLight 550 IA	DyLight^TM 550	DyLight^TM 650
thrS	1.2E-8 ± 9.6E-10^a	6.7E-9 ± 1.9E-9^a	3.9E-8 ± 4.7E-9a	1.4E-7 ± 3.5E-8	2.5E-7 ± 8.3E-8	1.0E-7 ± 6.8E-8	8.5E-8 ± 1.1E-8	1.1E-7 ± 1.6E-8
yqjG	1.1E-8 ± 9.2E-10^a	3.4E-9 ± 2.4E-10^a	1.4E-8 ± 1.7E-9^a	4.9E-7 ± 2.1E-7	1.0E-7 ± 4.5E-9	1.2E-7 ± 7.0E-8	2.9E-7 ± 9.7E-8	1.4E-7 ± 2.7E-8
yajL	5.6E-8 ± 2.8E-8^a	1.4E-8 ± 7.5E-9^a	1.0E-7 ± 3.1E-8^a	3.7E-7 ± 1.9E-7	8.6E-7 ± 2.8E-7	2.2E-7 ± 1.2E-7	2.5E-7 ± 1.3E-7	4.4E-7 ± 2.9E-7
hemE	8.7E-10 ± 6.1E-11^a	6.9E-10 ± 2.3E-11^a	5.6E-9 ± 5.3E-10^a	1.9E-7 ± 3.4E-8	1.2E-8 ± 7.7E-10	1.2E-7 ± 3.5E-8	1.4E-7 ± 5.4E-8	4.8E-8 ± 5.7E-9
ilvA	1.6E-8 ± 4.1E-9^a	2.8E-8 ± 3.6E-8^a	5.9E-7 ± 1.7E-7^a	2.9E-7 ± 4.1E-8	2.9E-7 ± 5.9E-8	1.5E-6 ± 6.8E-7	8.5E-8 ± 5.4E-8	1.3E-6 ± 5.3E-7
prpD	1.6E-7 ± 1.7E-7^a	9.0E-8 ± 1.8E-7^a	1.3E-7 ± 2.1E-8^a	7.3E-7 ± 1.0E-7	6.2E-7 ± 4.2E-7	7.9E-7 ± 1.7E-7	1.3E-6 ± 8.7E-7	2.4E-6 ± 5.7E-7
zwf	1.4E-8 ± 1.5E-9^a	2.3E-8 ± 2.7E-8^a	7.2E-8 ± 1.7E-8^a	1.2E-6 ± 5.3E-7	5.8E-7 ± 2.2E-7	3.7E-7 ± 2.7E-7	5.6E-7 ± 3.6E-7	6.4E-7 ± 2.7E-7
eda	3.9E-9 ± 2.7E-10^a	2.0E-9 ± 2.6E-10^a	9.2E-8 ± 1.6E-8^a	3.3E-7 ± 1.1E-7	2.6E-7 ± 9.4E-8	2.6E-7 ± 1.1E-7	2.2E-7 ± 2.4E-7	4.6E-7 ± 2.1E-7
gor	2.4E-8 ± 3.5E-9^a	1.2E-8 ± 1.8E-9^a	8.3E-8 ± 3.4E-8^a	1.1E-6 ± 5.5E-7	8.6E-7 ± 5.0E-7	2.2E-7 ± 8.2E-8	9.0E-7 ± 5.0E-7	6.1E-7 ± 3.7E-7
pqqL	5.4E-9 ± 4.7E-10^a	1.8E-9 ± 2.3E-10^a	6.7E-8 ± 1.1E8^a	3.0E-7 ± 2.0E-7	3.0E-7 ± 1.1E-7	3.7E-7 ± 2.7E-7	1.5E-7 ± 9.2E-8	6.3E-7 ± 3.8E-8
^a Significant difference to its normal peptide control and fluorescent dye control ( p < 0.05).

Table 4. Top 10 protein list of [SD][QV][AEDT]A[YIL][CE][AK][ARL][MV][AHK]? [KET][LV][AYLF]E enriched in entire E. coli K12. The motif was searched in entire E. coli K12 proteome by GLAM2SCAN.

Rank	Name	EcoGene Accession	START	SITE	END	SCORE
1	ilvA	EG10493	266	DSDAICAAMKDLFE	279	29.2
2	thrS	EG11001	116	DVEALEKRMHELAE	129	28.1
3	yqjG	EG12746	202	SQEAYDEAVAKVFE	215	26
4	pqqL	EG11744	357	MQDAANALMAELAT	370	24.3
5	prpD	EG13603	293	SQTAVEAAM.TLYE	305	23
6	eda	EG10256	53	AVDAIRAIAKEVPE	66	22.6
7	gor	EG10412	87	SRTAYIDRIHTSYE	100	22.5
8	zwf	EG11221	54	DKAAYTKVVREALE	67	21.8
9	yajL	EG13272	101	IVAAICAAPATVLV	114	21.1
10	hemE	EG11543	174	DPQALHALLDKLAK	187	20

Table 5. Secondary structure of Motifs from 10 E. coli K12 identified proteins and human S100A1 proteins. The secondary structure of motifs for each protein was provided by their provider and RCSB PDB with protein 3D structures. However, the hemE, zwf and pqqL do not have the protein 3D structure in RCSB PDB. We used the EcoGene 3.0 which contains the QUARK prediction method to predict the secondary structure of motifs.

Name	Secondary structure of Motif	Source	PDB ID	QUARK ID	Reference
thrS	Alpha-helix	RCSB PDB	1TJE		(34, 39)
yqjG	Alpha-helix	RCSB PDB	4G0L		(38, 39)
yajL	Beta-sheet+Alpha-helix	RCSB PDB	2AB0		(36, 39)
hemE	Alpha-helix	EcoGene 3.0		E11780	(41, 42)
ilvA	Alpha-helix	RCSB PDB	1TDJ		(33, 39)
prpD	Alpha-helix	RCSB PDB	1SZQ		(31, 39)
zwf	Alpha-helix	EcoGene 3.0		E14278	(41, 42)
eda	Alpha-helix	RCSB PDB	1WAU		(37, 39)
gor	Alpha-helix	RCSB PDB	1GEU		(32, 39)
pqqL	Alpha-helix	EcoGene 3.0		E12551	(41, 42)
S100A1	Alpha-helix	RCSB PDB	1ZFS		(35, 39)

Table 6. K_d values for 2-oxohistidine peptides binding to S100A1.

	Oxidative Peptides			Normal Peptides			Fluorescent Dyes
S100A1	DyLight 550 oxo-AG	DyLight 550 oxo-SE	DyLight 650 oxo-IA	DyLight 650 AG	DyLight 650 SE	DyLight 550 IA	DyLight^TM 550	DyLight^TM 650
w/o calcium^b	5.3E-9 ± 4.9E-9^a	5.2E-9 ± 2.1E-9^a	1.2E-8 ± 4.2E-9^a	3.9E-8 ± 4.0E-8	8.2E-8 ± 5.7E-8	3.1E-8 ± 2.0E-8	3.4E-7 ± 2.4E-7	3.6E-7 ± 3.6E-7
w/ calcium	7.3E-9 ± 4.8E-9^a	2.9E-9 ± 6.4E-10^a	2.8E-8 ± 9.8E-9^a	1.3E-7 ± 9.1E-8	8.5E-8 ± 5.5E-8	6.6E-8 ± 3.7E-8	5.6E-8 ± 3.8E-8	1.2E-7 ± 8.4E-8
^a Significant difference to its normal peptide control and fluorescent dye control ( p < 0.05). ^b No significant difference to with calcium group (p > 0.1).