Membrane orientation of the FMO antenna protein from Chlorobaculum tepidum as determined by mass spectrometry-based footprinting

Abstract

The high excitation energy-transfer efficiency demanded in photosynthetic organisms relies on the optimal pigment-protein binding orientation in the individual protein complexes and also on the overall architecture of the photosystem. In green sulfur bacteria, the membrane-attached Fenna-Matthews-Olson (FMO) antenna protein functions as a “wire” to connect the large peripheral chlorosome antenna complex with the reaction center (RC), which is embedded in the cytoplasmic membrane (CM). Energy collected by the chlorosome is funneled through the FMO to the RC. Although there has been considerable effort to understand the relationships between structure and function of the individual isolated complexes, the specific architecture for in vivo interactions of the FMO protein, the CM, and the chlorosome, ensuring highly efficient energy transfer, is still not established experimentally. Here, we describe a mass spectrometry-based method that probes solvent-exposed surfaces of the FMO by labeling solvent-exposed aspartic and glutamic acid residues. The locations and extents of labeling of FMO on the native membrane in comparison with it alone and on a chlorosome-depleted membrane reveal the orientation. The large differences in the modification of certain peptides show that the Bchl a #3 side of the FMO trimer interacts with the CM, which is consistent with recent theoretical predictions. Moreover, the results also provide direct experimental evidence to confirm the overall architecture of the photosystem from Chlorobaculum tepidum (C. tepidum) and give information on the packing of the FMO protein in its native environment.

Photosynthesis is a fundamental biological process that harvests solar energy to power life on Earth (1). A diverse family of pigment-protein complexes and elegant architectures accomplish the necessary light-harvesting and energy-storage processes (2–5). In photosynthetic green sulfur bacteria, light absorbed by a large antenna complex known as a chlorosome (6–8) is transferred through a protein called Fenna-Matthews-Olson (FMO) (9) to the reaction center, which is embedded in the cytoplasmic membrane (CM). Together, they form a funnel-like architecture to facilitate energy transfer. The specific orientation of the critical linker, the FMO protein, however is unknown (Fig. 1A).

Fig. 1.

Photosystem from C. tepidum and structure of FMO. (A) Model architecture of photosystem from C. tepidum. The 2 possible orientations of FMO on the CM are presented. Bchl a #3 is shown as a star. (B) Top view of the FMO trimer with the Bchl a #3 side shown. All of the pigments are omitted except Bchl a #3, which is colored cyan. The side chains of all of the D/E residues are highlighted as red sticks. In each FMO monomer, there are 21 D and 20 E residues plus a C-terminal carboxyl group. (C) Side view of the FMO trimer shown as cartoon, ribbon and mesh for clarity. Positions of Bchl a #3 (cyan) and Bchl a #1 (red) are labeled in the monomer shown by cartoon. All of the phytol tails of pigments are omitted for clarity.

The structure of the FMO protein was the first (bacterio)chlorophyll binding protein to be determined by X-ray crystallography. Structures of this protein from 2 species, Prosthecochloris aestuarii 2K (10, 11) and Chlorobaculum tepidum (12) are now available, and they show strong structural and spectral similarities. The FMO protein consists of 3 identical subunits of mass 40 kDa related by a 3-fold axis of symmetry. The 3 monomers form a disc with a C3 symmetry axis perpendicular to the disc plane (Fig. 1B). There are 7 BChl a molecules in a monomer, although an eighth pigment has been resolved in newly solved structures (13, 14). Each pigment experiences a different local environment (Fig. 1C), and their site energies are fine-tuned by specific interactions with the protein. Bchl a #3 and Bchl a #1, for example, are on the opposite sides of the FMO protein from the side view of the FMO trimer (Fig. 1C).

Given that the FMO protein plays a critical role in the energy transfer pathway, significant effort has been made to understand its electronic structure. Quantum effects (15–20), which were recently discovered in this complex, may function to improve the energy-transfer efficiency. A defined energy-transfer pathway was also elucidated by both 2D electronic spectroscopy (21) and novel theoretical calculations (20). The pigment with the lowest site energy, the assignment of which was historically controversial (22–24), is predicted to be Bchl a #3 on the basis of coupling with the dipole of adjacent alpha helices (19). This energy-sink pigment is expected to be close to the CM to ensure efficient energy transfer from the FMO protein to the reaction center (RC) (20). Thus, this side of the FMO trimer (Bchl a #3 side) should be in close contact with the RC in the CM.

The opposite orientation, however, was predicted from the structure and properties of the isolated protein. Hydrophobicity analysis of the FMO protein favors an interaction of the Bchl a #1 side of the protein with the CM (12), in accord with another suggestion based on the existence of an eighth pigment (13). In this latter model, the extra pigment forms an energy transfer bridge between the FMO and the RC.

The experimental evidence for the orientation of the FMO comes from linear dichroism (25) and 3D reconstitution data based on STEM images (26). Both suggest that the FMO disc sits flat on the CM with its C3 symmetry axis perpendicular to the plane of the membrane. However, the specific orientation of the disc (i.e., which side interacts with the CM), cannot be determined by using these methods.

Moreover, the overall architecture of the photosystem, including the relative orientation and the extent of the interaction between the individual antenna complexes, is also poorly understood. The interaction between the flat surface of the FMO trimer and the RC, shown by the STEM image (26), is not as strong as proposed on the basis of protein hydrophobicity, suggesting that the FMO is probably partially buried in the CM (12). On the chlorosome side, the detailed interaction between the FMO and the CsmA protein is not clear, although surface plasmon resonance (27) and cross-linking data (28) suggest that FMO protein directly interacts with the CsmA protein and is probably partially buried in the CsmA layer (28). In short, a comprehensive interaction map of the various components, chlorosome, FMO, and RC, at the molecular level is still needed.

We report here a method that combines carboxyl group modification with mass spectrometry to afford surface mapping or footprinting (29, 30) of the protein, revealing the interaction of proteins associated with membranes. We chose the reagent, glycine ethyl ester (GEE), which labels any solvent accessible carboxyl groups on glutamic acid (E), aspartic acid (D), and the C terminus by zero-length cross-linking (28, 31–33). Although the use of labeling reagents for mapping and cross-linking is not new and liquid chromatography/tandem mass spectrometry (LC/MS/MS) is commonly used for complex proteomics, the combination is shown to be useful for protein footprinting particularly with highly accurate mass measurements.

Three states of the protein were investigated for comparison: the isolated FMO protein, the protein attached to the CM but with chlorosomes removed, and the protein in the native membrane (i.e., with chlorosomes attached). After labeling the FMO protein in the 3 states, the labeled sites were located by LC-MS/MS analysis of peptides produced by in-gel trypsin digestion of the protein following its isolation (Fig. 2). The modification levels of various peptides from the 3 samples, upon quantitative analysis on the basis of extracted ion current (EIC) chromatograms (34), show that the Bchl a #1 side of the FMO protein in the native membrane was modified to a lesser extent compared to that after the chlorosome was removed. When the FMO protein is attached to the CM, with or without chlorosomes, the modification levels of the Bchl a #3 side of the FMO protein were never as high as those for the free protein. Thus, it is the Bchl a #3 side of the protein that is in contact with the CM.

Fig. 2.

Schematic experimental design and procedure. The solvent-exposed surfaces of isolated FMO protein, FMO from chlorosome-depleted membrane, and FMO from native membrane were probed by a small molecule, which can be covalently attached to certain residues. The labeling sites were determined by MS after protein purification and enzyme digestion. The modification levels of different peptides were compared to determine the interaction interface.

Results

Three samples were prepared and subjected to chemical modification: isolated FMO protein as a control (Fig. 3A), the chlorosome-depleted membrane (Fig. 3B), and the native membrane (Fig. 3C). In the native membrane, the strong chlorosome absorptions at 746 nm, 457 nm, and 336 nm obscure the spectral features of the other components (Fig. 3C), and the Q_y absorption band of the FMO is just a shoulder. After chlorosome depletion by the chaotropic reagent NaI, the FMO protein is still attached to the CM and shows the characteristic absorption. Peaks from the RC (671 nm) and from carotenoids in the 400–500 nm region are recognizable in the absorption of the chlorosome-depleted membranes (Fig. 3B). The isolated FMO protein showed identical absorption spectra before and after GEE modification (Fig. 3A), indicating no significant conformational change after protein modification.

Fig. 3.

Characteristic absorption spectra. (A) Absorption spectra of purified FMO protein (red) and FMO protein after GEE modification (black). (B) Chlorosome-depleted membrane. (C) Native membrane from C. tepidum.

The modification of the D/E residues on the FMO protein by GEE was done under physiological conditions by using the zero-length cross-linker, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC). Once the samples were modified, they were submitted to SDS/PAGE separation. Separate samples of the FMO protein, visualized as appropriate gel bands (Fig. S1) were in-gel digested with trypsin, and the peptides were loaded to LC-MS/MS to identify and quantify the modification sites.

In the LC-MS analysis (an example output is shown in Fig. 4), peptides and their modifications were identified by both accurate mass measurement (accuracy < 10 ppm) and tandem mass spectrometry (MS/MS). The accurate mass is needed to add specificity for analysis of the complex mixtures that arise by protein digestion. The modification of one D/E site by GEE shifts the peptide mass +85.0528 Da (C₄H₇NO) (Fig. 4D). The added ester group also undergoes hydrolysis under either the basic conditions during protein purification or under the acidic conditions used for LC/MS, or both. After hydrolysis, the mass shift for a given one-site modified peptide changes to +57.0215 Da (C₂H₃NO) (Fig. 4C). The hydrolysis conserves a carboxyl group, producing little conformational stress, and maintaining the stabilization of the protein. The +57.0215 Da peptide generally eluted at approximately the same time as or a little earlier than the unmodified peptide, whereas the +85.0528-Da peptide always eluted approximately 1–3 min later than the unmodified peptide, consistent with its increased hydrophobicity. This chromatographic pattern helps in the identification of the modified peptides (Fig. S4).

Fig. 4.

Typical chromatograms from LC/MS showing a peptide and the peptide with D/E modifications. (A) LC chromatogram of the trypsin-digested FMO protein. (B) Extracted ion current (EIC) chromatogram of the unmodified peptide (m/z = 698.3719 Da). (C) EIC of the peptide with D/E modification (+57.0215 Da). (D) EIC of the peptide with D/E modification (+85.0528 Da). The D/E modification sites were determined by the product ions (Fig. S6). The retention times of the peptides and the areas of EIC chromatograms were labeled, and the mass accuracy was better than 10 ppm.

Quantitative MS analysis was accomplished by obtaining extracted ion current chromatograms of the trypsin-digested peptides. The modification level of a given peptide was computed to be the ratio of EIC of D/E modified peptides by the sum of the EICs of both modified and unmodified peptides. For example, the modification level of peptide 67–79 from the modified free protein, shown in Fig. 4, is equal to (1.94 × 10⁷ + 1.24 × 10⁷)/(1.94 × 10⁷ + 1.24 × 10⁷ + 3.11 × 10⁸) × 100% = 9%. The same peptides from the treated FMO protein in chlorosome-depleted membranes and in native membranes were analyzed separately. This peptide has 2 possible modification sites. Remarkably, the modification of both sites was identified, and they were well-separated by the HPLC method we adopted. An EIC signal for the modification of both sites of the same peptide (mass increments of +114.0430 Da, +170.1056 Da, or +142.0743 Da) could not be detected at a signal-to-noise ratio of 2:1. In general, the EIC signal corresponding to multiple modifications of any peptide containing 2 or more carboxyl side chains was at or below the noise level, and they were not considered in the analysis (Fig. S2).

If there are no sites missed in the trypsin digestion, the FMO protein will be digested into 25 peptides containing at least 6 amino acids. Five peptides (132–143, 216–222, 287–303, 325–331, and 340–347) do not have D/E residues, and they all were observed. Of the 20 remaining peptides, 5 (169–181, 182–199, 225–238, 348–354, and 355–366) have D/E residues, but the D/E residues are located in the connection region used to form a trimer, and they are buried under the trimer surface. The signals from these unmodified peptides were identified, but no modifications could be observed. When the FMO trimers were denatured, however, these sites could be easily modified, consistent with tight binding of the 3 monomers. Two long tryptic peptides (1–29 and 97–126) were not detected in the LC/MS experiment. The loss of large peptides is a common problem in in-gel trypsin digestion. Signals for peptides 82–93, 152–168, and 269–285 were barely above the noise level and only occasionally could be found; they also are not considered further.

All other tryptic peptides were detected in the 3 samples; they are classified and listed in 3 groups in Fig. 5. Peptides 36–52, 67–79, and 304–314 (group A) of the modified FMO protein purified from chlorosome-depleted membranes and from the native membranes were not modified as extensively as those from the GEE-modified free FMO protein (compare the red bar with the green and blue bars in Fig. 5A), indicating that the corresponding regions of the protein are clearly protected. The modification levels of these peptides, from FMO either in the chlorosome-depleted membrane or in the native membrane, are approximately identical whether or not the chlorosome is removed from the membrane. This indicates that protection comes from the membrane and not from the chlorosomes.

Fig. 5.

Modification level of certain peptides and their location in the protein 3D structure. (A) Identified cytoplasmic membrane-protected peptides of FMO (group A). (B) Location of peptides in group A (highlighted in red). (C) Identified chlorosome-protected peptides of FMO (group B). (D) Location of peptides in group B (highlighted with purple). (E) Peptides that are identified and showed an approximately similar modification level. (F) Location of peptides 248–259 and 260–268 (highlighted in orange) and location of peptide 332–339 (highlighted in blue). In B, D, and F, the side view of the FMO monomer is presented for clarity. Only Bchl a #3 (cyan) and Bchl a #1 (red) are labeled to show the orientation.

In contrast, several peptides, 53–62, 144–151, 203–215, 239–247 (group B), showed a statistically significant increase in modification after the chlorosome was removed from the CM (compare the green bars with the blue bars in Fig. 5C). In the native environment, the protein regions corresponding to these peptides are likely to be covered by the bulky chlorosomes, but they become available for labeling when the chlorosomes are not present, regardless of the presence or absence of the membrane.

Higher modification levels of peptides (green bars in Fig. 5C) are seen when the chlorosome was removed and the membrane remains than when the protein is free (red bars in Fig. 5C). This is likely due to the relatively higher concentration of labeling reagents that were used to treat the chlorosome-depleted membrane. We chose to add relatively more labeling reagent to the chlorosome-depleted membrane samples (refer to Methods section) to compensate for other sites in the CM that would consume the labeling reagents. The variability in reagent load makes it difficult to do a direct comparison between the 3 samples. Nevertheless, even under high reagent loading, peptides in group A from the FMO associated with the chlorosome-depleted membrane are not modified to the same level as peptides from the free FMO protein. Protection by the membrane is apparent from the results.

There are also several peptides (Fig. 5E) that show no apparent trend as those seen for peptides in groups A and B. The modification levels of peptides 248–259 and 260–268 in the isolated FMO protein, for example, are the same as or slightly higher than those from FMO associated with the chlorosome-depleted membrane and with the native membrane. Both peptides are located in the middle and on the side of the FMO protein (highlighted in orange in Fig. 5F). They show slight protection when the membrane is present. Peptide 332–339 is modified to a somewhat higher level in FMO taken from the chlorosome-depleted membrane and from the native membrane than from the free protein. The D/E residues in this peptide are located in a flexible loop at the bottom region of the FMO disc and stick out of the protein body (blue colored in Fig. 5F). The local environment of these residues in the 3 samples is expected to be similar; this peptide may be an indicator of the amount of labeling reagents that can approach the FMO protein held in the chlorosome-depleted membrane and in the native membrane.

Discussion

Membrane Orientation of FMO.

When peptides in group A are mapped onto the known FMO 3D structure (pdb code: 1M50), we see that all are located on the Bchl a #3 side of the protein (Fig. 5B). In contrast, peptides in group B are all located on the Bchl a #1 side of the protein (Fig. 5D). Considering both peptide sets, we are able to pinpoint the interactions between both the FMO protein and the chlorosome and the FMO and the CM. That is, the Bchl a #3 side of the protein interacts with the CM, whereas the Bchl a #1 side interacts with the chlorosome. This orientation of the FMO trimers on the CM indicates that the newly resolved eighth Bchl a in the FMO protein (13, 14) is located close to the chlorosome baseplate, suggesting that this pigment functions as a linker to facilitate the energy transfer from the baseplate protein to the core pigments of the FMO protein.

This orientation confirms recent theoretical predictions that Bchl a #3 functions as a trap and transfers excitation to the RC (15, 19–21). Nevertheless, the question of how this pigment connects to the energy-acceptor pigment in the RC is still unanswered. A detailed analysis of the docking of FMO and RC and the energy transfer process from the FMO to the RC requires an atomic-resolution structure of the whole complex.

Packing of the Chlorosome, FMO, and CM Layers.

Having answered the principal question, what is the orientation of the FMO protein in the CM, we turn to some additional questions about the architecture of the system. One concern is the packing between the chlorosome, FMO, and the CM layers; this packing must affect the energy transfer efficiency and other cellular processes. The labeling results, obtained using small probes to map solvent-exposed surfaces, also allows some tentative conclusions to be made about the packing of the 3 layers.

First, for peptides in group A, the presence of the CM doesn't preclude modification (green and blue bars in Fig. 5A); rather the extent of modification is decreased to 30–40% compared to the free FMO protein. This outcome indicates that the interaction between the FMO protein and the CM is not sufficiently strong to lock the protein in the CM, a conclusion that is consistent with the STEM results (26). Further, peptide 332–339 (Fig. 5E) of FMO in both the chlorosome-depleted membrane and the native membrane was modified to a similar extent as that from the free protein. If the Bchl a #3 side of the FMO was tightly locked with or significantly immersed in the CM, such a high level modification of this peptide on the CM would not be expected.

Second, the comparable extents of some peptide modification from the chlorosome-protected FMO and the free FMO (compare blue and red bars in Fig. 5C) indicate that the packing between the FMO and chlorosome layers has permeability to solvent water carrying the mapping reagent. It is likely that the FMO protein is available for labeling because it is not buried in the CsmA layer; indeed, 15–20 aa from the C terminus of CsmA stick out of the chlorosome envelope (35). In addition, the newly resolved eighth Bchl a and the possible linker function, as described above, may diminish the packing of the 2 layers.

All previously reported models (2, 36, 37) of this membrane system hold that the FMO proteins are located beneath the chlorosomes and function mainly to transfer excitation energy from the chlorosomes to the RCs. Nevertheless, some FMO proteins may not be covered by the chlorosomes. If so, moderate modification of chlorosome-protected peptides from FMO in the native membrane would be expected and is observed. Those uncovered proteins that increase the extent of modification are unlikely to be free because we did pellet both the purified chlorosome-depleted membrane and native membrane to remove any unbound FMO proteins.

When the FMO protein was oriented through gel squeezing, linear dichroism experiments found that the 3C symmetry axis of a fraction of FMO trimers was not perfectly perpendicular to the membrane surface (25). Such a tilt might explain why there is still low-level modification of peptides when FMO is interacting with the chlorosome (blue bars in Fig. 5C). This does not affect our overall conclusion on the orientation of the FMO proteins in the membrane.

The GEE probe molecule, (diameter: approximately 2.5 Å; length: approximately 7 Å) may be appropriate to respond to differences caused by tight binding, but it may be too small to be sensitive to weaker binding and the resulting larger distances between the interacting bodies. A series of probes with different sizes and shapes are now being sought to probe the packing between the 3 layers in a way that is more sensitive to distance.

Packing of FMO on the CM.

Bryant and coworkers (36, 37) estimate that approximately 150–200 FMO trimers and 25–40 RCs are present per chlorosome. A range of 4–8 FMO trimers are associated with each RC. Although the stoichiometry of the purified FMO-RC complex is still uncertain, it appears that each RC only has 1 or 2 FMO binding sites according to the STEM images (26). Therefore, there may be lateral energy transfer from FMO to FMO on the native CM, and this would require tight packing of the FMO proteins to increase the energy-transfer efficiency.

The STEM images of the FMO-RC complex (26), however, indicate that the 2 possible FMO binding sites are not closely associated with each other. Biophysical studies of the purified FMO protein also are in accord with the proposal that there is little energy coupling between the 3 monomers of a FMO trimer even though they are tightly packed (23, 24). Another concern is that ferredoxin should be able to move freely to accept the electron delivered from the RC. There should be enough space or channels between the chlorosomes and CM to permit its diffusion, although in a proposed model in which all of the RCs are on the edge of the chlorosome (36), such channels are not required. Considering all these concerns, efficient lateral energy transfer does not appear to be likely.

Owing to the complexity of the system, it is difficult to predict the labeling pattern of the side of the FMO protein on the CM. Nevertheless, peptides 248–259 and 260–268 were modified to a similar extent when the FMO protein is on the CM compared to when it is free. This suggests there is no tight binding between them. A higher resolution experimental approach (possibly AFM) is required to determine the distribution of FMO on the CM.

To conclude, the green sulfur photosynthetic bacteria contain a remarkably efficient and complex architecture to harvest sunlight and transfer the energy to the RC, where electron transfer quenches the excitation. Specific protein-membrane and protein-protein interactions play a crucial role in accomplishing the high efficiency transfer. We have established the orientation of the FMO protein in its native setting. Furthermore, from a semiquantitative consideration of the labeling results, we are able to conclude the packing of the FMO layer is permeable to solvent water carrying the mapping reagent. These conclusions arise from results taken by an efficient protein footprinting method. Indeed, the reagent used in this research works remarkably well under physiological conditions. Given that D/E residues are common in most soluble proteins, we believe this method can be extended to study a wide variety of protein-protein, protein-membrane, and protein-ligand interactions.

Materials and Methods

Cells of the thermophilic green sulfur bacterium Chlorobaculum tepidum strain TLS were grown anaerobically at 45 °C, 150 uE light intensity for 2 days. The cells were harvested by centrifugation at 10,000 × g for 15 min.

Native Membrane Preparation.

After the harvested cells were washed with 50 mM phosphate buffer (pH = 7.6), they were broken by sonication; the cell debris was pelleted by low-speed centrifugation, and the supernatant liquid was ultracentrifuged at 150,000 × g for 2 h. The pellet, containing the native membrane, was collected for later analysis.

FMO Protein Purification.

The FMO protein from C. tepidum was isolated according to a modification of the method described by Li et al. (12). The main difference is that the starting material was the native membrane instead of the broken cells. After Na₂CO₃ incubation and ultracentrifugation, the supernatant containing the FMO protein was dialyzed against 100 times volume of 20 mM Tris-HCl buffer (pH = 8.0) to remove residual CO₃²⁻. The solution was then purified by using ion exchange and gel filtration chromatography until OD267/OD371 < 0.6. The FMO protein was concentrated by Centricon 100 MWCO and stored for further use.

Chlorosome-Depleted Membrane Preparation.

A method modified from Feick et al. (38) was used to purify the chlorosome-depleted membrane from C. tepidum. Solutions of 10% sucrose and 2 M NaI (Mallinckrodt) were added to the membrane suspension. The mixture was sonicated for 10 min in a water bath sonicator. Deriphat 160c detergent (0.05%) (Henkel) was added to prevent the membrane aggregation. A subsequent centrifugation at 80,000 × g for 60 min resulted in a floating pellet, which is mainly chlorosome, and a supernatant. The supernatant was ultracentrifuged at 180,000 × g for 150 min, and the supernatant was harvested as the chlorosome-depleted membrane. Another round of centrifugation and ultracentrifugation yielded clean chlorosome-depleted membranes.

Carboxyl Group Modification.

The modification reaction was carried out for 2 h at 4 °C, dark, phosphate buffer at pH = 7.6, with 0.3 M GEE (Sigma) and 50 mM EDC (Pierce). The reaction was quenched by adding the same volume of 1 M sodium acetate, and the samples were immediately loaded onto SDS/PAGE gel to isolate the FMO protein. Before the modification reaction, both the native membrane and the chlorosome-depleted membrane were ultracentrifuged again to pellet the membranes and to ensure that there was no free FMO protein in the sample. The pellets were resuspended in 50 mM phosphate buffer (pH = 7.6), and the modification reaction carried out. The OD 809 of the chlorosome-depleted membrane was approximately 0.8. The OD 747 of the chlorosome peak in the native membrane was approximately 70. The isolated FMO protein was originally in Tris buffer; buffer exchange was done by diluting the concentrated protein stock into phosphate buffer, then concentrated, and diluted again several times. The OD 809 of the FMO protein was approximately 4.

LC-MS/MS.

The FMO band was cut from the SDS/PAGE gel, and the in-gel protein was trypsin digested following the manufacturer's instructions by using proteomic grade trypsin from Sigma. The LC-MS/MS running method was adapted from Sperry et al. (39). The peptide solution was loaded onto a reverse-phase C₁₈ column (0.075 mm × 150 mm) custom-packed with silica media (5 μm, 120 Å; Michrom Bioresources) The peptides were separated over 70 min using a NanoLC-1D (Eksigent) with the LC gradient from 2 to 60% acetonitrile with 0.1% formic acid for 60 min and then from 60% to 80% acetonitrile with 0.1% formic acid for 10 min at 260 nL/min followed by a 12 min reequilibration step by deionized water with 0.1% formic acid. The solution was sprayed directly from the column into the LTQ-Orbitrap mass spectrometer (Thermo-Scientific) using a PicoView Nanospray Source (PV550; New Objective) with an spray voltage of 1.8 kV, no sheath gas and capillary voltage of 27 V. Mass spectra of the tryptic peptides (m/z range: 350–2000) were acquired at mass resolving power of 60,000 (at m/z = 400) with an Orbitrap mass spectrometer while product-ion scans (MS/MS) of the 6 most abundant ions were performed in the ion trap part of the instrument at 35% of the normalized collision energy. An isolation width of 2 Da and an activation time of 30 ms were used.

Peptides and the D/E modifications were identified from the peptide accurate masses and product-ion sequencing by searching against the bacteria entries in the NCBI database using Mascot (Matrix Science). The EIC chromatograms were used to give quantitative information about chemical modification level on each peptide, as described in the Results section. Peak areas were obtained by integration of the various peaks by using Qual Browser (Xcalibur: Thermo-Scientific).

In the ESI MS analysis, some peptides show different charge-state distributions; furthermore, a small fraction of certain peptides undergo other modifications (e.g., methionine oxidation and N/Q deamidation). Larger peptides from a few missed cleavages can also be found. These complications can affect in a small way the calculation of modification levels, but they do not change the trends (Figs. S3 and S4). In the analysis, we always chose the dominant fractions, considered only the unmodified peptide, the +57.0215 peptide, and the +85.0528 peptide, except for peptide 36–52 and peptide 144–151. For peptide 36–52, the dominant material is a C propionamide (+C₃H₇NO) formed by reaction with free acrylamide during SDS/PAGE, whereas that for peptide 144–151 is an M-oxidized species (Fig. S5).