Cell-Wide Survey of Amide-Bonded Lysine Modifications by Using Deacetylase CobB

Background Lysine post-translational modifications are important regulators of protein function. Proteomic and biochemical approaches have resulted in identification of several lysine modifications, including acetylation, crotonylation, and succinylation. Here, we developed an approach for surveying amide-bonded lysine modifications in the proteome of human tissues/cells based on the observation that many lysine modifications are amide-bonded and that the Salmonella enterica deacetylase, CobB, is an amidase. Results After the proteome of human tissues/cells was denatured and the non-covalently bonded metabolites were removed by acetone washes, and the amide-bonded modifiers were released by CobB and analyzed using liquid- and/or gas chromatography/mass spectrometry metabolomic analysis. This protocol, which required 3–4 days for completion, was used to qualitatively identify more than 40 documented and unreported lysine modifications from the human proteome and to quantitatively analyze dynamic changes in targeted amide-bonded lysine modifications. Conclusions We developed a method that was capable of monitoring and quantifying amide-bonded lysine modifications in cells of different origins.


Background
Many amino acids are modified post translationally to regulate functions of proteins. Adding phosphor groups to serine, threonine and tyrosine constitutes protein phosphorylation and one of the major mechanisms of cellular signal transduction [1,2]. Other amino acid residues such as histidine, proline and cysteine can also be phosphorylated [3,4], hydroxylated [5] or acylated [6], respectively, to convey various biological functions. Lysine is the most heavily modified residue in proteins. More than 90 different lysine modifications have been identified on lysine [7]. Many lysine modifications are physiologically significant and have been studied extensively. For example, lysine acetylation of histones and other nuclear proteins is critical for chromatin remodeling and regulation of gene tran-scription [8,9], whereas acetylation of metabolism-related enzymes is important for regulation of metabolism [10,11]. Lysine methylation marks histones and other proteins for regulation of transcription and protein activity [12]. Although lysine methylation is an important modification, the majority of lysine modifications are amide-bonded. This was recently confirmed after an array of metabolites, including propionate, butyrate, malonate, crotonate, and succinate and their respective coenzyme A derivatives, was shown to contain amide-bonded lysine residues [13][14][15][16].
Cell-wide survey of known lysine modifications, such as acetylation [17] or methylation [18], has been documented. However, a more comprehensive understanding of the importance of lysine modifications will require elucidation of the cell-wide dynamics and types of lysine modifications. Thus, protocols for systematically and quantitatively surveying different types of lysine modifications are necessary. Because most lysine modifications are formed via amide bonds, cell-wide surveys of lysine modifications can be conducted once a nonspecific amidase is identified. Accordingly, in this study, we developed an approach for surveying amide-bonded lysine modifications in the proteome of human tissues/cells.

Results
While investigating the activities of Salmonella enterica deacetylase CobB [10], we unexpectedly observed that CobB possesses nonspecific amidase activity. Recombinant CobB purified from Escherichia coli efficiently cleaved all tested amide-bonded lysine modifications, including propionylation, succinylation, crotonylation, and acetylation, on synthetic peptides (Fig. 1). This suggested that CobB could be used as a general amidase to release amide-bonded lysine modifiers and thereby identify novel lysine modifications in cells. Thus, we developed a CobB-based protocol to survey amide-bonded lysine modifications in human cells. Briefly, proteins in cell lysates were precipitated and washed extensively with acetone to remove any small molecules that were noncovalently bound to the proteins, the amide-bonded modifications to the proteome were released by CobB treatment, and the released modifiers were analyzed by liquid chromatography/mass spectrometry (LC/MS) or gas chromatography/mass spectrometry (GC/MS).
Metabolites, the levels of which were significantly higher in CobB-treated samples than in untreated control samples, were considered possible lysine modifiers (Fig. 2) because only the ε-amine of lysine and the N-terminus amine of a protein could form amide bonds; however, we confirmed that the amide bonds formed with the Nterminal amine were not cleaved by CobB. Using this protocol, we successfully identified more than 40 new lysine modifications, including lysine aminoacylations, which facilitate sensing and signal transduction of intracellular amino acids in human liver cancer tissues and HEK293T cells [19].
A successful survey using the above described protocol will generate the following results: 1) significantly higher (> 2-fold, p value < 0.05) GC or LC peak areas should be obtained in CobB-treated samples than in matched untreated samples ( Fig. 3-1); 2) the MS/MS fingerprint spectra ( Fig. 3-2) should match some of the spectra in the NIST mass spectral library ( Fig. 3-3); 3) the retention time in GC or LC and the MS/MS spectrum of identified metabolites should match those of authentic standard metabolites ( Fig. 3-4). Insignificant increases (< 2 fold, p value < 0.05) in GC or LC peak areas are often related to insufficient removal of non-covalently bonded metabolites or CobB Fig. 1 CobB is an amidase. The abilities of CobB to cleave synthetic propionylated, succinylated, crotonylated and acetylated peptides were tested. The M/Z values of synthetic (left each group) and cleaved (right each group) peptides were determined by mass spectrometry and marked inactivation; a lack of GC or LC peaks with significant increases in area may be related to the first reason, whereas a lack of significant increases in GC or LC peak areas may be related to the second reason. The protocol can be improved accordingly.

Discussion
With the increasing discovery of the importance of PTMs, there is a growing need for discovery of novel PTMs. This protocol represents a general method for identifying and quantitatively analyzing amide-bonded modifications in cells. By using this method, it is not difficult to find dozens of new PTMs. Although we successfully used the method to identify amide-bonded lysine modifications in human kidney and liver cancer cells [19], this approach could also be applied to cell types of other human tissues or origins, as long as the proteome of these cells (bacteria, plants, and other mammalian cells) is readily available. Moreover, a welldesigned and well-performed analysis will allow quantification of the dynamic changes in cellular amide-bonded lysine modifications. Of note, the principle of our survey can be expanded to detect other PTMs of distinct chemical natures. For example, if a proper cleaving enzyme is identified, modifications such as acylation in cysteine, ubiquitinoylation or SUMOylation in glycine and glycosylation on various amino acids can be screened cell wide.
Lysine modifications are identified using distinct approaches (Table 1). For example, investigations on biochemical reactions have led to the identification of lysine acetylation [20] and methylation [21], and proteomic analysis has resulted in the identification of a  number of lysine modifications, including propionylation [13], butyrylation [13], and crotonylation [16]. Our current protocol utilized the amidase activity of CobB in combination with metabolomic analysis to identify multiple lysine modifications. To the best of our knowledge, this is the first protocol to allow cell-wide identification and quantification of amide-bonded lysine modifications. We confirmed that CobB did not cleave the N-terminus amine-derived amide bonds and peptide bonds in the protein backbone, suggesting that modifiers obtained after CobB cleavage were putative lysine modifiers [19]. However, the protocol had two limitations. First, there was a slight possibility that the CobB-cleaved modifiers were not lysine modifiers because we did not test whether CobB could cleave other chemical bonds. Second, although we identified an array of new lysine modifications using this protocol [19], whether this approach may enable complete identification of amide-bonded lysine modifications was unclear because the ability of CobB to cleave all types of amide-bonded modifications could not be exhaustively tested. Nevertheless, our protocol represents a useful tool for surveying potential novel lysine modifications.

Conclusions
This method was developed because of the lack of method for systematically and quantitatively surveying different types of lysine modifications. We propose a method that can be accurate, facile, and reproducible to identify and quantify new amide-bonded lysine modifications in cells of different origins.

Reagent Setup Homogenization Buffer
Homogenization buffer was composed of 10 mM KCl, 1.5 mM MgCl 2 , 10 mM Tris, and 5 mM nicotinamide. The solution was prepared by dissolving 0.373 g KCl, 0.071 g MgCl 2 , 0.606 g Tris, and 0.305 g nicotinamide in 400 mL deionized water, adjusting the pH to 7.5 with HCl, and adding deionized water to a final volume of 500 mL. This solution can be stored for up to 1 month at room temperature before use.

Oximation Mix (for GC/MS)
This solution was composed of 250 mM methoxyamine hydrochloride in pyridine. This solution was prepared by dissolving 20 mg methoxyamine hydrochloride in 1 mL pyridine. This solution must be freshly prepared on the day of the experiment.

Derivatization Mix (for GC/MS)
This solution was composed of 20% (v/v) MTBSTFA in pyridine. This solution was prepared made by dissolving 200 μL MTBSTFA in 800 μL pyridine. This solution must be freshly prepared on the day of the experiment.

Derivatization Mix (for LC/MS)
This solution was composed of 3 M HCl in 1-butanol. This solution was prepared by dissolving 250 μL HCl in 750 μL 1-butanol. This solution must be freshly prepared on the day of the experiment.

High-Salt Buffer
This solution was composed of 20 mM Tris, 25% (v/v) glycerol, 1.5 mM MgCl 2 , 0.2 mM EDTA, and 1.2 M KCl. The solution was prepared by dissolving 1.211 g Tris, 12.5 mL glycerol, 0.071 g MgCl 2 , 0.029 g EDTA, and 44.730 g KCl in 400 mL deionized water; adjusting the pH to 7.4 with HCl; and adding deionized water to a final volume of 500 mL. This solution can be stored for up to 1 month at room temperature before use. Add protease inhibitors described below to a 1× concentration just before use for cell lysis.

Coomassie Brilliant Blue
This solution was prepared with 0.25% (w/v) Coomassie brilliant blue G-250, 45% (v/v) methanol, and 10% (v/v) acetic acid. G-250 brilliant blue powder was solubilized in methanol, acetic acid was added, and deionized water was added to an appropriate volume. The solution was filtered through a 0.45-μm filter-top bottle before immediate use and could be stored at room temperature for future use.

Acrylamide:Bisacrylamide Solution (for Making SDS Gels)
This solution consisted of 250 g acrylamide dissolved in 417 mL deionized water. The solution was prepared by stirring the acrylamide solution overnight before adding 1.67 g bisacrylamide. The solution was filtered through a 0.45-μm filter-top bottle before use and could be stored at room temperature (20-25°C) in a dark container.

Electrophoresis Running Buffer
The buffer was prepared by dissolving 3.02 g Tris base, 18.8 g glycine, and 1 g SDS in 1 L deionized water. This buffer could be stored at room temperature until use.

PBS
PBS was composed of 137 mM NaCl, 8 mM Na 2 HPO 4 , 2.7 mM KCl, and 1.47 mM KH 2 PO 4 , pH 7.1, in deionized water. The solution was prepared by dissolving 8 g NaCl, 1.14 g Na 2 HPO 4 , 200 mg KCl, and 200 mg KH 2 PO 4 in 800 mL deionized water; adjusting the pH to 7.1 with HCl; and bringing the volume up to 1 L with additional deionized water. The solution was autoclaved in a glass bottle to sterilize. This buffer can be stored at room temperature for at least several months.

Binding Buffer (for CobB Purification)
Binding buffer was composed of 100 mM sodium phosphate, 150 mM NaCl, and 20 mM imidazole. The solution was prepared by dissolving 8.197 g sodium phosphate, 4.383 g NaCl, and 0.681 g imidazole in 400 mL deionized water; adjusting the pH to 7.2 with HCl; and adding deionized water to a final volume of 500 mL. This solution must be freshly prepared on the day of the experiment.

Elution Buffer (for CobB Purification)
Elution buffer was composed of 100 mM sodium phosphate, 150 mM NaCl, and 300 mM imidazole. The solution was prepared by dissolving 8.197 g sodium phosphate, 4.383 g NaCl, and 10.212 g imidazole in 400 mL deionized water; adjusting the pH to 7.2 with HCl; and adding deionized water to a final volume of 500 mL. This solution must be freshly prepared on the day of the experiment.

Desalination Buffer (for CobB Purification)
Desalination buffer was composed of 50 mM Tris and 150 mM NaCl. The solution was prepared by dissolving 3.029 g Tris base and 4.383 g NaCl in 400 mL deionized water, adjusting the pH to 7.5 with HCl, and adding deionized water to a final volume of 500 mL. This solution can be stored for up to 1 month at room temperature before use.

Protease Inhibitor Cocktail (50×)
This solution contained 50 mM PMSF, 0.05 mg/mL aprotinnin, 0.05 mg/mL leupeptin, and 0.05 mg/mL pepstatin in dimethyl sulfoxide. This solution can be stored for up to 12 weeks at − 20°C and diluted 50-fold (v/v) before use. ▲CRITICAL 1 mM DTT and PMSF should be added to the diluted protease inhibitor cocktail solution before use.
! CAUTION Protease Inhibitor Cocktail is extremely toxic; handle with care.

Experimental Design
In addition to the cleavage of amide-bonded lysine modifiers by CobB, the key to the success of this protocol was related to obtaining precleaned proteomes. To achieve this goal, the proteomes of human cells were denatured using 85% acetone, which allowed further removal of non-covalently bonded metabolites while preserving covalently bonded lysine modifications and inactivated enzymes that may remove the modification. The denatured proteome was subject to extensive washing with 85% acetone to remove non-covalently bonded metabolites. The "precleaned" proteome was then subjected to CobB treatment to release the amidebonded lysine modifiers, and the cleaved modifiers were obtained by collecting the supernatants from the centrifuged denatured CobB reaction mixture. The cleaved modifiers in the supernatant were directly assayed using LC/MS, GC/MS, or both after the modifiers were oximated by methoxyamine hydrochloride. Notably, CobB-untreated samples should be run in parallel with CobB-treated samples as controls. ▲CRITICAL The purity of the synthetic peptides should be greater than 98% to ensure accurate quantification.
2. Mix known concentrations of synthetic peptide into samples to be analyzed.
3. Run samples with known synthetic peptides and samples without synthetic peptides through steps 1-54. 4. Compare the areas of GC or LC peaks of the targeted modifiers in these samples. 5. Calculate the levels of modifications in these samples by comparing with the internal peptide calibrator. 6. Alternatively, obtain a working curve by running various levels of synthetic peptides through steps 1-54, and compare areas of GC or LC peaks of samples with the working curve.

Procedure
This protocol can be used to survey lysine modifications (steps 1-59, Fig. 2) or to quantify targeted lysine modifications (Table 2).
Obtaining the Human Proteome •TIMING: 1-5 H 1. The human proteome was obtained from either hepatocellular carcinoma (HCC) tissue or HEK293T cultured human embryonic kidney cells. HCC tissues were obtained with consent from the patient and processed (start from step 2) within 2 h of the patient's operation. HEK293T cells were processed (start from step 13) within 15 min of harvesting.
▲CRITICAL The following steps (2-22) must be performed on ice.
2. Dice 2 g HCC tissue into pieces of approximately 0.1 mm using a razor. 3. Resuspend the tissue particles in 10 mL ice-cold (4°C) homogenization buffer supplemented with 1× protease inhibitor cocktail. 4. Homogenize tissue particles in a Dounce homogenizer with at least 40 strokes. 5. Pass the homogenized suspension through a 100-mmdiameter cell filter (pore size 125 μM) to remove debris. 6. Centrifuge the filtered solution at 1000×g for 10 min at 4°C, and collect the supernatant. 7. Resuspend the pellet from step 6 in 10 mL ice-cold (4°C) homogenization buffer supplemented with 1× protease inhibitor cocktail by vortexing. 8. Recentrifuged the samples at 1000×g for 10 min at 4°C, and collect the supernatant. 9. Resuspend the pellet from step 8 in 2 mL high-salt buffer with vortexing, and place the suspension on a rotary shaker for 30 min at 4°C to extract nucleic proteins. 10. Centrifuge at 1000×g for 10 min at 4°C, and collect the supernatant. 11. Repeat steps 9-10, and collect the supernatant. 12. Combine the supernatants from steps 6, 8, 10, and 11, and go to step 23. 13. Harvest two dishes (15 cm) of confluent HEK293T cells by scraping, and wash the cells three times using 10 mL PBS with centrifugation at 1000×g. 14. Resuspend the cells in 10 mL ice-cold (4°C) homogenization buffer supplemented with 1× protease inhibitor cocktail. 15

26.
Add DTT (final concentration: 1 mM) to the cytosolic (from step 23) and mitochondrial (from step 25) proteomes to break disulfide bonds within proteins. Allow the reaction to proceed for 10 min on ice. 27. Add prechilled (− 80°C) acetone to the proteome to reach a final concentration of 85% (v/v) to precipitate proteins. Place the sample at − 80°C for 30 min to allow complete precipitation of proteins.
! CAUTION Acetone is highly flammable and toxic; handle with caution.
▲CRITICAL Do not exceed 800×g during centrifugation to avoid tight packing of the protein pellet, which can cause inefficient washing in the following steps.

? TROUBLESHOOTING
29. Resuspend the pellets in 5 mL of 85% (v/v) acetone on a vortex mixer for 2 min.
■PAUSE POINT The dried pellet can be stored at − 80°C until further processing.
Release Amide-Bonded Modifiers from the Denatured Proteome with CobB. •TIMING: 4-5 H 33. Dissolve 0.5-2 mg protein from step 32 in 1 mL deionized water, and sonicate for 5 min at room temperature to increase solubility.

?TROUBLESHOOTING
34. Determine the protein concentration of the suspension using Bradford reagent. 35. Adjust the protein concentration of the suspension to 0.2 mg/mL. 36. To 1 mL proteome suspension, add recombinant CobB (purity > 98%; for reaction system, see Table 3; for preparation, see Table 4) to reach a proteome:CobB ratio of 100:1. 37. Allow the reaction to proceed at 37°C for 4 h, and stop the reaction by adding 4 volumes of stop solution supplemented with 100 μM adonitol, which serves as an internal control for metabolite quantification in subsequent processes. 38. Store the stopped reaction at − 80°C for more than 12 h to allow complete precipitation of all proteins in the reaction mix. 39. Centrifuge at 12,000×g for 10 min at 4°C, collect the supernatant, and discard the pellet. 40. Vacuum dry the supernatant (which contains CobBcleaved 2′-O-metabolite-ADP-ribose derivatives) in an Eppendorf vacuum drier at room temperature. The anionic ion-pair reagent HFBA was added to the mobile phase to improve analyte interactions with the stationary phase. An aqueous solution of HFBA (approximately 0.5 M) was diluted in water (mobile phase A) and methanol (mobile phase B) to a final concentration of 0.5 mM.
2. Gradient system (pair of two alternating columns) The gradient system used two identical columns (Agilent Zorbax SB-C18; 2.1 mm × 50 mm, 1.8 μm; Waldbronn, Germany) connected to a 10-port two-position switching valve. Using the column switching valve and a second binary pump, the two columns were applied alternately. When one column was used for the analytical gradient, the other column was cleaned and re-equilibrated. From the prepared samples, 5 μL was injected into the HPLC system. The analytical gradient (pump 1; flow rate: 0.5 mL/min) increased linearly from 10% B to 50% B within 11 min. After isocratic elution for 0.5 min, the gradient returned to starting conditions until 11.6 min, and isocratic flow was performed for 10% B from 11.6 to 12.5 min at 0.5 mL/min. This short return of pump 1 to starting conditions was necessary for flushing the tubing from the pump to switching valve before the freshly equilibrated second column was switched in for the next injection. Parallel to the analytical gradient, the other column was cleaned and equilibrated by pump 2; within 1.0 min, the gradient was increased from 10% B to 95% B. The isocratic flow of 95% B was held for 4.5 min. From 5.5 to 6.5 min, initial conditions of the analytical gradient (10% B) were achieved and retained at isocratic flow until 12.5 min for flushing and equilibrating the column. Total time from injection to injection was 13.3 min (including the autosampler operation time of 0.8 min).   50. Transfer 100 μL derivatized metabolite solution into GC vials for GC/MS analysis (for settings, see Table 6).
▲CRITICAL Analyze derivatized metabolites as soon as possible. The stability of derivatized metabolites at − 80°C can range from a few days to a few months.
?TROUBLESHOOTING.  those with 3-fold or higher increases represent possible positive hits.

?TROUBLESHOOTING
54. Obtain MS/MS spectra of possible positive hits and search against NIST mass spectral library to identify corresponding metabolites. ▲CRITICAL Empirically, no more than a 0.5% retention time shift in GC or LC is tolerated.

Troubleshooting
Step