Single-Site Oxidation, Cysteine 108 to Cysteine Sulfinic Acid, in d-Amino Acid Oxidase from Trigonopsis variabilis and Its Structural and Functional Consequences

Materials.

NADH was from Seppim S.A. (Sees, France), and l-lactate dehydrogenase was from Roche (Mannheim, Germany). All other chemicals, including proteomics grade modified trypsin, were obtained from Sigma-Aldrich (St.Louis, MO). Nanoflow liquid chromatography (NanoLC) eluents were prepared in deionized ultrapure water generated by the Nanopure Water system (Barnstead, Newton, WA). A biocatalyst preparation of TvDAO from strain Trigonopsis variabilis ATCC 10679 was kindly provided by Sandoz GmbH (Kundl, Austria). Its specific enzyme activity was 40 U/mg, determined by a reported coupled enzyme assay using d-alanine as the substrate (34). If not otherwise stated, all buffers contained 1 mM dithiothreitol (DTT).

Separation of native TvDAO and its modified protein forms.

The enzyme was stored in aliquots at −25°C for later processing. An aliquot was thawed, and precipitated protein was removed from it by centrifugation (Sorval centrifuge, rotor T865) at 80,000 × g (30 min; 4°C). The resulting supernatant was subjected to gel filtration (HiPrep 26/10 column [2.6 cm by 10 cm; Amersham Biosciences, Uppsala, Sweden], and BioLogic DuoFlow system [Bio-Rad, Hercules, CA]), and the buffer was exchanged for 5 mM Tris-HCl, pH 7.5, containing 1 mM DTT. Pooled eluate was concentrated to the original volume by ultrafiltration (Vivaspin ultrafiltration concentrator tubes; Vivascience AG, Hanover, Germany)at4,400 × g using molecular mass cutoff membranes of 10 kDa. The enzyme was loaded on an anion-exchange MonoQ HR 5/5 column (0.5 cm by 5 cm; Amersham Biosciences) using a BioLogic DuoFlow system (Bio-Rad). The column had been equilibrated with 1 mM DTT-10 mM Tris-HCl, pH 7.5 (buffer A). The bound protein was eluted with six steps of isocratic flow of 0.1 to 0.6 M KCl in buffer A (within 96 min; flow rate, 0.5 ml/min), followed by isocratic flow of 1 M KCl in buffer A (5 min; flow rate, 0.75 ml/min). UV detection at 280 nm was used. The collected fractions were pooled according to eluted protein peaks.

The fractions were checked by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), nondenaturing anionic PAGE, and isoelectric focusing using Coomassie blue for the staining of protein bands. Enzyme preparations which were not used immediately were aliquoted and stored at −25°C. Protein concentrations were determined by the Bradford method (protein microassay from Bio-Rad, Hercules, CA) using bovine serum albumin as a standard.

Size exclusion chromatography.

Size exclusion chromatography was performed at 10°C using a Superdex 200 HR 10/30 column and an Äkta FPLC system (both from Amersham Biosciences). Intact TvDAO or its modified form (500 μg; 100 μl) was loaded on the column equilibrated with 10 mM Tris-HCl (pH 7.5; 0.2 M KCl) containing 1 mM DTT. Elution was carried out isocratically with a flow rate of 0.5 ml/min and detection at 280 nm. Molecular mass determinations were referenced against appropriate protein standards (Bio-Rad).

Spectral measurements and enzyme assays.

Protein absorbance spectra were collected with a Beckmann DU800 UV-VIS spectrophotometer (Beckmann-Coulter, Inc., Fullerton, CA) at 20 ± 0.1°C using a 10-mm-path-length quartz cell. The enzyme concentration was 10 μM in 10 mM Tris-HCl, pH 7.5, containing 1 mM DTT.

Activity of TvDAO was measured with an assay based on the decolorization of 2,6-dichloroindophenol (DCIP) and a coupled enzymatic assay (TvDAO/l-lactate dehydrogenase) using conditions reported recently (34). d-Methionine (20 mM) or d-alanine (50 mM) was employed as the substrate. Enzymatic oxygen consumption rates were recorded with a fiber optic fluorescence microsensor, as reported elsewhere (34).

For stability measurements, purified preparations of TvDAO (20 μM) dissolved in 10 mM Tris-HCl, pH 7.5, were incubated at 50 ± 1°C on an orbital shaker operated at an agitation rate of 300 rpm. Samples (350 μl) were taken at certain incubation times, and part of the sample (50 μl) was used to determine the residual enzyme activity with the DCIP-based assay. The remainder volume of the sample was used for light-scattering measurements, which served as particular probes of protein aggregation. Light scattering was recorded with an F4500 spectrofluorometer (Hitachi, Tokyo, Japan) under the conditions described above, except that excitation and emission wavelengths were both set to 500 nm.

The number of free sulfhydryl groups in purified fractions of TvDAO were determined using colorimetric titration with Ellman's reagent, 5,5′-dithio-bis (2-nitrobenzoic acid) (DTNB) (12). Briefly, exchange of buffer (80 mM sodium phosphate, pH 8.0, containing 1.2 mM EDTA) and exhaustive removal of DTT were achieved by repeated steps of gel filtration with a HiPrep 26/10 column (Amersham Biosciences), followed by concentration through ultrafiltration. The protein solution (6 μM; 400 μl) was mixed with 50 μl DTNB solution (10 mM; 100 mM sodium phosphate buffer, pH 7.8), corresponding to a molar ratio of DTNB to protein subunit of 200, and incubated at 25 ± 1°C for up to 240 min. The time course of absorbance at 412 nm was monitored continuously, and a molar absorption coefficient of 13,600 M⁻¹ cm⁻¹ for the 2-nitro-5-thiobenzoate anion was assumed for the calculation of the concentration of protein sulfhydryl groups reacted. Appropriate controls were recorded in all cases, and the reported values are corrected for minor blank readings. The standard error of determination was ± 20% (n = 6), and the results obtained with different concentrations of DTNB were fully consistent. Thiol group titration in unfolded TvDAO was carried out as just described after denaturation of the protein in the presence of SDS (15 mg/ml) for 15 min.

MS.

Analysis by mass spectrometry (MS) was performed using an Agilent 1100 series ion trap (IT) mass spectrophotometer (Agilent Technologies, Palo Alto, CA). The instrument, equipped with an orthogonal electrospray or nanospray source, was coupled to a microinfusion pump (KD Scientific, St. Louis, MO) or on line to an Agilent 1100 series nanoLC system (Agilent Technologies), respectively. The mass spectrometer was tuned using a calibration mixture provided by the manufacturer. The spectral data were acquired and monitored using the Agilent ChemStation software package (Agilent Technologies).

(i) ESI-MS.

For mass analysis, the TvDAO samples were prepared in 60% (vol/vol) acetonitrile, 1% formic acid solution and then directly infused into an electrospray ionization (ESI) source using a microsyringe (100 μl) coupled to the microinfusion pump (KD Scientific) at a flow rate of 5 μl/min. The spectra were obtained in the automatic acquisition mode scanning from 200 to 2,200 m/z. The tip was held at 3.5 kV in a positive ion detection mode, and nebulization was assisted by nitrogen gas at a flow rate of 8 liters/min. Data processing was done using the deconvolution module of the ChemStation software.

(ii) In-solution tryptic digestion.

A suitable volume of protein solution was diluted in Tris buffer (100 mM, 6 M urea, pH 7.8) and reduced with 5 mM DTT for 1 h at room temperature. Iodoacetamide (IAM) was added to the mixture in a fourfold excess over DTT and allowed to react for 1 h at room temperature in the dark. Subsequently, unreacted IAM was consumed by addition of the equimolar amount of reducing reagent (DTT) during the next hour, and then the reaction mixture was diluted with purified water (Nanopure Water System) to yield a final concentration of urea of 0.6 M. Sequencing grade modified trypsin was prepared by adding 20 μl of 1 mM HCl to a vial containing 20 μg of trypsin, and this solution was added to the pretreated sample at a substrate-to-enzyme ratio of 50:1 (wt/wt). The mixture was incubated overnight at 37°C, and the reaction was stopped by addition of acetic acid. The digest was used without any additional processing steps before analysis.

(iii) NanoLC/ESI-MS/MS.

The peptide mixture (1 μl) was trapped on a precolumn (ZORBAX 300SB-C18 column; 0.3 mm; Agilent Technologies) with a total loading time of 5 min and a sample solution flow rate of 200 nl/min. The loaded mixture was then flushed to a 0.75-μm (inside diameter) nanoLC separation column (ZORBAX 300SB-C18; 50 mm; Agilent Technologies), where the mobile-phase flow rate was 250 nl/min. Ultrapure water was used as mobile phase A, with addition of 0.1% high-performance liquid chromatography grade trifluoroacetic acid (vol/vol), and high-performance liquid chromatography grade acetonitrile-trifluoroacetic acid (99.9:0.1 [vol/vol]) was used as mobile phase B. Peptides were eluted from the stationary phase with a 60-min linear gradient ranging from 2 to 60% mobile phase B at a constant flow (250 nl/min). The column effluent was directed to an etched fused-silica ESI tip (360-50-8-D-20; New Objective, Inc., Cambridge, MA), and ions formed by electrospray were focused on the entrance of the IT mass spectrometer. The voltage at the spray tip and the heated MS inlet capillary temperature were optimized at 1.3 kV and 235°C, respectively. The IT operated in an automated data-dependent acquisition mode under the control of the Agilent ChemStation software system. The system was run in positive ion mode with a maximal ion accumulation time of 100 ms; the collision gas was helium. In all MS/MS experiments, a full MS scan between 200 and 2,200 m/z was followed by two maximal-resolution scans and six full MS/MS scans for the two most abundant ions from each MS scan.

(iv) Data processing.

Protein sequences were obtained from the National Center for Biotechnology Information (NCBI) database. The proteolytic digest pattern derived from the purified TvDAO was compared with a virtual digest of the protein derived using Peptide Tools software (Agilent Technologies). The acquired spectra were analyzed with Mascot (Matrix Science, London, United Kingdom) and Spectrum Mill, provided by Agilent Technologies, employing peptide mass fingerprint, MS/MS search, and de novo sequencing. Peptides containing cysteine residues were manually interpreted for confirmation.

Separation and identification of multiple forms of TvDAO.

Figure 1 shows the protein elution profile during MonoQ anion-exchange chromatography of a representative TvDAO preparation that is used, after immobilization, in the conversion of cephalosporin C. Two of six distinct and baseline-separated protein peaks were found to contain significant yet different specific enzyme activities in the range of 40 to 220 U/mg (measured by the coupled enzyme assay using d-alanine as the substrate). The other peaks did not contain activity and were therefore not further analyzed. The pattern of multiple protein forms and their separation, as depicted in Fig. 1, was fully reproducible with different batches of the enzyme and did not change substantially regardless of the workup procedure used to obtain the technical-grade biocatalyst (data not shown). The recovery of the procedure, in respect to both enzyme activity and protein, was greater than 95%. Upon rechromatography on the MonoQ column, the isolated protein forms 1 (F1) and 2 (F2) eluted exactly as they did during fractionation of the protein mixture, as indicated by the superimposition of the respective UV traces in Fig. 1. It was proven that upon incubation at 4°C and 25°C in buffer (1 mM DTT-10 mM Tris-HCl, pH 7.5), isolated F1 and F2 were not interconverted into one another, clearly indicating that they represent stable protein forms of TvDAO.

Molecular size determination of isolated forms of TvDAO.

The two separated, as well as rechromatographed, enzyme forms F1 and F2 migrated in SDS-PAGE as single protein bands with apparent molecular masses of about 39 kDa (data not shown). The molecular masses of F1 and F2 were calculated from ESI-MS spectra to be 39,214.1 ± 10 Da and 39,246.3 ± 10 Da, respectively. There are two different sequences of TvDAO reported in the NCBI database: entries Q99042 (or CAA90322) and AAR98816.

The primary structures of the TvDAOs described above differ at four positions: 87 (Ser in Q99042 and Asn in AAR98816), 102 (Val/Gly), 142 (Arg/Ala), and 344 (Val/Ile). The molecular masses of full-length TvDAOs, predicted from the genes including the initiator methionine without the bound cofactor, are 39,301.1 Da (Q99042) and 39,214.0 Da (AAR98816). According to full-mass scan spectra of both protein forms and molecular mass information, F1 is identified as the translated gene product of sequence AAR98816, and F2 appears to be a modified form of TvDAO. The mass difference of 32 Da (measured value, 32.2 Da) between F1 and F2 could occur if F2 had incorporated two additional oxygen atoms that are not present in F1 (Table 1). In nondenaturing anionic PAGE, F2 showed an unmistakably higher mobility (data not shown). Its pI value of ≈5.5 revealed in isoelectric focusing was slightly but significantly higher than that of F1 (ΔpI ≈ 0.12) (data not shown). This result, combined with the stronger retention of F2 in anion-exchange chromatography, suggests that the overall negative charge of F2 is probably greater than that of F1.

Structural characterization of isolated forms of TvDAO.

The primary structure of TvDAO is predicted from the coding gene to contain six cysteines. Colorimetric titration with DTNB shows that all six thiol side chains are reduced in F1 and accessible for chemical derivatization in the folded state of the protein. In contrast, F2 has five reactive SH groups, suggesting that one cysteine may have been modified, most probably in an irreversible fashion. Tryptic peptide fingerprinting for F1 and F2 was carried out using ESI-MS, and the results are summarized in Table 1. The data reveal the identity of amino acids Asn87, Gly102, and Val344 and thus reinforce the above conclusion that the primary structure of F1 corresponds to the amino acid sequence predicted from gene AAR98816.

The peptides containing cysteine residues were identified unambiguously: Cys108, peptide 12; Cys145 and Cys159, peptide 16; Cys193, peptide 20; Cys257, peptide 27; and Cys298, peptide 32.

Tandem MS was used to clearly resolve conflicting published information regarding the N-terminal sequence of TvDAO (positions 4 to 22; peptide 2) (Fig. 2), which plays an important role in the function of the enzyme by providing contacts with the FAD cofactor. The N-terminal sequence of F1, I⁴VVIGAG¹⁰V¹¹AGLT¹⁵T¹⁶A¹⁷L¹⁸Q¹⁹LLR²², corresponds to the sequences in the NCBI database but is different from another published sequence, I⁴VVIGAX¹⁰G¹¹AGLA¹⁵G¹⁶X¹⁷G¹⁸L¹⁹LLR²² (22), as indicated by boldface letters. The overall sequence coverages obtained by MS sequencing were 95% and 90% for F1 and F2, respectively.

Single-site oxidation of Cys108 distinguishes F2 from F1.

The MS analysis shown in Table 1 suggested that peptides 12 derived from F1 and F2 might be different, eventually causing the observed mass difference of 32.2 Da for the intact protein forms. Tandem MS was therefore used to characterize the molecular details of the peptides with the sequence L¹⁰⁰EGAMSAIC¹⁰⁸QANPWFK¹¹⁵, focusing in particular on the modification of Cys108 and using manual interpretation of the spectra. NanoLC-ESI-MS analysis of F1 detected a mass peak with an m/z value of 912.4. The calculated monoisotopic mass of the carboxamidomethyl adduct is 57.021, and the corresponding m/z value for the [M + 2H]²⁺ ion of peptide 12 with a carboxamidomethylated (Cam) Cys108 residue is 912.4. The mass of F1 peptide 12 was shifted upward by m/z 28.5 relative to the expected mass of the unmodified peptide.

To confirm the modification of the cysteine residue at position 108, we performed MS/MS sequencing of the precursor ion at m/z 912.4. The MS/MS spectrum of the ESI-produced [M + 57 + 2H]²⁺ ion consisted of typical fragment ions. A nearly complete series of b ions (b₂ to b₁₀, b₁₂, and b₁₃) and y ions (y₂ to y₉ and y₁₁) demonstrated that Cys108 had gained 57 atomic mass units (Fig. 3a), fully consistent with the notion that the original thiol group of F1 had been alkylated with IAM during sample preparation for mass spectrometric analysis. In the mass spectrum of the tryptic digest of F1, we did not observe mass peaks at m/z 883.9 and 1766.8 that would correspond to, respectively, the doubly and singly charged ions of peptide 12 containing reduced Cys108. Also lacking were mass peaks of peptide 12 bearing oxidized Cys108 (i.e., Cys108-SOH, Cys108-SO₂H, and Cys108-SO₃H). We conclude, therefore, that the side chain of Cys108 is reduced in native F1 and fully carboxamidomethylated upon alkylation with IAM.

In the mass spectrum of F2, a mass peak with m/z 900.0, which corresponds to the [M + 32 + 2H]²⁺ ion of peptide 12, was observed. The calculated m/z value for the doubly charged ion of peptide 12 plus two oxygens is 899.9. In contrast to the Cys108-containing peptide in F1, the series of b ions (b₂ and b₃ and b₅ to b₁₃) and y ions (y₂ and y₃ and y₅ to y₁₃) derived from the parent ion at m/z 900.0 demonstrated that the thiol residue at position 108 had gained 32 mass units, as shown in Fig. 3b. Therefore, this finding strongly suggests that Cys108 is oxidatively modified and present as cysteine sulfinic acid (Cys108-SO₂H) in F2. Considering the currently available analytical tools (5, 7, 11), the results in Fig. 3b probably present the most direct determination of the site-specific modification of a protein cysteine, showing unambiguously the oxidation status of its thiol side chain.

An important conclusion that has general relevance for the analysis of oxidized protein cysteines with electrospray MS can be derived from Fig. 3b. The results reveal that the observed fragment ions of the Cys108-SO₂H-containing peptide did not display loss upon collision-induced dissociation (32). As was already seen for alkylated Cys108 in Fig. 3a, the b₇ ion was most abundant, while the b₉ and y₇ ions, which result from the cleavage of the peptide bond at the C-terminal side of Cys108-SO₂H, had an abundance similar to that of the product ions emanating from cleavages at other sites within peptide 12. These findings are at variance with recently reported fragmentation patterns of synthetic peptides containing different oxidation levels of a cysteine side chain (17, 35), suggesting that gas phase collision-induced fragmentation of the F2-derived peptide bearing Cys108-SO₂H may be affected by the particular sequence context.

Under the conditions of data acquisition used, we did not detect additional peptide fragments obtained from F2 in which Cys108 was reduced, alkylated, or oxidized to a different extent than that reported. The different modifications of sulfur-containing residues, especially Met104, and aromatic residues were analyzed manually by checking all MS/MS spectra that were acquired in data-dependent mode and not identified by Spectrum Mill. No detectable ions of modified peptides were found. Therefore, the results of the detailed mass spectrometric analysis of intact protein forms and their tryptic digests suggest that at a sequence coverage of >95%, microheterogeneity at only a single side chain, that of Cys108, distinguishes F2 from F1.

Structural and functional consequences of modification of Cys108.

A structural model of the TvDAO subunit was generated with the Swiss-PDBViewer using the X-ray structure of pig kidney DAO (PDB code 1kif) (16) as a template. From this model, Cys108 is proposed to be located within a flexible loop extending from the protein remote from the active center (>20 Å) (data not shown). F1 and F2 eluted from a Superdex 200 HR size exclusion column as single 80-kDa protein peaks that contained all the enzyme activity, suggesting that both forms of TvDAO are functional dimers.

Absorption spectra of F1 and F2 are shown in Fig. 4. The spectrum of F1 exhibits two strong bands in the visible region, centered at 365 nm and 455 nm, which are characteristic of enzyme-bound FAD in the oxidized state. The value of 6.5 for the A₂₈₀/A₄₅₅ ratio is consistent with values reported in the literature (28). The spectrum of F2 is significantly altered compared with that of F1. It exhibits bands that are similar in position but far less pronounced than the corresponding bands in F2. Upon stepwise addition of sodium sulfite, the FAD bands in F1 and F2 were strongly quenched, reflecting the formation of a covalent N(5) adduct between the cofactor and sulfite, a feature characteristic of oxygen-reactive flavoenzymes (29). Figure 4 (inset) shows the dependence of the extent of band quenching in F1 and F2 on the molar excess of sulfite added. The results reveal that the reactivities of enzyme-bound FAD toward sulfite are identical in F1 and F2 and that at high levels of sulfite, the quenching is complete in both TvDAO forms. Therefore, this implies that within the limits of detection (±5%), FAD cofactors bound to F1 and F2 are present exclusively in their oxidized states. The observed changes in the absorption spectrum of F2 in comparison with that of F1 may thus be ascribed to long-range effects of the oxidative modification of Cys108 on the microenvironment of the FAD cofactor, which are manifested in particular functional properties of F2 (see below).

Using standard enzyme assays reported elsewhere (34), the activity of F2 was found to be approximately one-fourth that of F1, a significant difference. Initial rate measurements were carried out under conditions in which the natural cosubstrate dioxygen (220 μM) or the artificial electron acceptor DCIP (150 μM) was employed at a constant concentration and the substrate concentration was varied (d-alanine, 2 to 50 mM). Kinetic parameters were obtained from nonlinear least-squares fits of a rectangular hyperbola to the data. Within the limits of experimental error (±5%), the apparent Michaelis constants (K_m) of F1 and F2 for d-alanine were 16 mM and showed no dependence on the cosubstrate used. The apparent turnover number (^appk_cat) of F1 for reaction with dioxygen (^appk_cat = 62 ± 2 s⁻¹) was four times that of F2 (^appk_cat = 14 ± 2 s⁻¹), in good agreement with the results of activity measurements using DCIP. Kinetic parameters of F1 and F2 for reduction of dioxygen were obtained from initial rate measurements in which the depletion of O₂ was monitored directly (34). For both enzymes, the dependence of the O₂ consumption rate showed a linear dependence on the O₂ concentration in the range 50 to 500 μM (Fig. 5). The slopes of the linear plots were used to obtain second-order rate constants (k_cat/K_m) of 0.149 μM⁻¹ s⁻¹ and 0.040 μM⁻¹ s⁻¹ for F1 and F2, respectively. The results reveal that the catalytic efficiency of enzyme-bound FADH₂ for reduction of O₂ is decreased approximately fourfold as a result of the site-specific oxidation of Cys108, which fully explains the observed lower activity of F2. The data in Fig. 5 also show that the decrease in k_cat/K_m for F2, compared with F1, is partly due to an increase in the apparent K_m for O₂. The cofactor absorbance spectrum of F2, which is clearly different from that of F1, arguably reflects this significant decrease in reactivity.

Oxidative modification of TvDAO alters the thermal denaturation pathway.

Under temperature conditions (50°C) that promote rapid inactivation of TvDAO, we compared the aggregation propensities of F1 and F2 using light scattering as a reporter of the aggregation event. Figure 6 shows time courses of aggregation of F1 and F2. The results reveal that the conversion of Cys108 into cysteine sulfinic acid causes a complete suppression of protein aggregation under the thermal-stress conditions. For F1, there is a clear kinetic correlation between loss of activity and aggregation. In the case of F2, inactivation and aggregation are partly uncoupled at 50°C so that a time-dependent decrease in activity, yet at a significant two- to threefold-higher rate than with F1, is seen even in the absence of detectable aggregation. Although the detailed comparison of inactivation pathways for F1 and F2 is beyond the scope of this paper, the results indicate clearly that Cys108 will be an important target for the stabilization of TvDAO under conditions in which the effect of protein aggregation is the main contributor to the inactivation rate. They provide a novel basis for the design of engineered enzyme variants which couple high activity (F1; Cys108 reduced) and suppressed aggregation (F2; Cys108 oxidized) to show improved performance in biocatalytic processes and analytical applications.

In summary, we have identified Cys108 as the prominent site of oxidative posttranslational modification of TvDAO and performed a detailed structure-function relationship analysis for the “chemical point mutation” Cys108→cysteine sulfinic acid. The evidence, aside from the obvious applied component associated with it, appears to be relevant in the context of TvDAO lifespan in the yeast. The enzyme is located in the peroxisomal compartment, and site-specific oxidation could mark it for turnover through one of the cellular protein degradation pathways (8, 30).

Finally, the reported analytical protocols could be useful for other researchers wishing to explore, on a defined structural basis, the stability and stabilization of oxidation-sensitive biocatalysts.