Characterization of the catalase-peroxidase KatG from Burkholderia pseudomallei by mass spectrometry by

The electron density maps of the catalase-peroxidase from Burkholderia pseudomallei (BpKatG) presented two unusual covalent modifications. A covalent structure linked the active site Trp111 with Tyr238 and Tyr238 with Met264, and the heme was modified, likely by a perhydroxy group added to the vinyl group on ring I. Mass spectrometry analysis of tryptic digests of BpKatG revealed a cluster of ions at m/z 6585, consistent with the fusion of three peptides through Trp111, Tyr238, and Met264, and a cluster at m/z approximately 4525, consistent with the fusion of two peptides linked through Trp111 and Tyr238. MS/MS analysis of the major ions at m/z 4524 and 4540 confirmed the expected sequence and suggested that the multiple ions in the cluster were the result of multiple oxidation events and transfer of CH3-S to the tyrosine. Neither cluster of ions at m/z 4525 or 6585 was present in the spectrum of a tryptic digest of the W111F variant of BpKatG. The spectrum of the tryptic digest of native BpKatG also contained a major ion for a peptide in which Met264 had been converted to homoserine, consistent with the covalent bond between Tyr238 and Met264 being susceptible to hydrolysis, including the loss of the CH3-S from the methionine. Analysis of the tryptic digests of hydroperoxidase I (KatG) from Escherichia coli provided direct evidence for the covalent linkage between Trp105 and Tyr226 and indirect evidence for a covalent linkage between Tyr226 and Met252. Tryptic peptide analysis and N-terminal sequencing revealed that the N-terminal residue of BpKatG is Ser22.


INTRODUCTION
The heme-containing catalase-peroxidases are bifunctional enzymes that degrade hydrogen peroxide either as a catalase (2H 2 O 2 ® 2H 2 O + O 2 ) or as a peroxidase (H 2 O 2 + 2AH ® 2H 2 O + 2A . ). The catalatic reaction, with a more rapid turnover rate, dominates over the peroxidatic reaction, and the in vivo peroxidatic substrate remains unidentified, suggesting that the main role of the enzyme is the removal of H 2 O 2 , preventing the formation of highly reactive and damaging breakdown products of H 2 O 2 . However, the enzyme has a close sequence resemblance to plant peroxidases (1,2), and it remains a possibility that the peroxidatic reaction has a metabolic significance outside of degrading H 2 O 2 . Indeed, it is clear that the catalatic function evolved as an adaptation of the peroxidatic function because the simple change of a tryptophan to a phenylalanine in the distal heme pocket reduces catalatic activity by 1000 fold (of E. coli HPI) and increases peroxidatic activity by 3 fold (3,4,5). Furthermore, the core structure of both the N-and C-terminal domains of the catalase-peroxidases from Haloarcula marismortui and Burkholderia pseudomallei closely resemble the structure of plant peroxidases (6,7). Finally, the conversion of isoniazid into its active antitubercular form by KatG of Mycobacterium tuberculosis is clearly a result of the peroxidatic reaction using INH as a substrate that must mimic the actual in vivo substrate.
The structures of the catalase-peroxidases from H. marismortui and B. pseudomallei have been reported (6,7) and have revealed several features that are, so far, unique to this class of enzyme. Present in both structures is an unusual adduct or covalent linkage among the side chains of a tryptophan, a tyrosine and a methionine (Fig 1). The likely mechanistic significance 5 The oligonucleotides CTGTTCATCAAAATGGCATGG (AAA encoding Lys in place of Arg 108 ), CGCATGGCATTTCACAGCGCG (TTT encoding Phe in place of Trp 111 ), and TTCGCGCGCCTGGCGATGAAC (CTG encoding Leu in place of Met 264 ) were purchased from Invitrogen. They were used to mutagenize a 600 bp fragment from pBG306 generated by KpnI-ClaI restriction following the Kunkel procedure (11), which was subsequently reincorporated into pBG306 to generate the mutagenized katG gene. Sequence confirmation of all sequences was by the Sanger method (12) on double-stranded plasmid DNA generated in JM109. Subsequent expression and purification were carried out as described previously (3,8).
The catalase and peroxidase specific activities of the variants compared to the native BpKatG are summarized in Table 1.

Catalase, peroxidase, protein and spectral determination.
Catalase activity was determined by the method of Rrrth and Jensen (13) in a Gilson oxygraph equipped with a Clark electrode. One unit of catalase is defined as the amount that decomposes 1 µmol of H 2 O 2 in 1 min in a 60 mM H 2 O 2 solution at pH 7.0 at 37 o C. Protein was estimated according to the methods outlined by Layne (14). Peroxidase activity was determined by the method of Smith et al (15). One unit of peroxidase is defined as the amount is defined as the amount that decomposes 1 µmol of ABTS (3-ethylbenzothiazolinesulfonic acid) in one minute at 20 o C. Absorption spectra were obtained using a Milton Roy MR3000 spectrophotometer. Samples were dissolved in 50 mM potassium phosphate, pH 7.0. The N-terminal sequence was determined by the Proteomics facility at the Institut de Biotecnologia i Biomedicina (UAB).

Mass spectrometry analysis
For mass spectrometry, protein was dialysed into 5 mM ammonium acetate. The intact proteins were analysed by electrospray ionization in an orthogonal time-of-flight mass spectrometer (16,17). The declustering voltage was varied in order to assess the stability of the protein-heme complex. Digests of the proteins were prepared using TPCK-treated trypsin, and these were analysed on the Manitoba/Sciex prototype MALDI QqTOF instrument (18). Initial analysis was done with equal volumes (0.5 mL) of digested protein and 2,5-dihydroxybenzoic acid (DHB, 160 mg/mL in water:acetonitrile 3:1, 2% formic acid) spotted onto a custom target. and eluted with a linear gradient of 1 -80% acetonitrile (0.1% TFA). Column effluent (4 mL/min) was collected at 1 min intervals by hand. Under the conditions used, the vast majority of tryptic fragments were eluted in 40 min. These were spotted onto the target as above.

MS characterization of the Trp-Tyr-Met adduct in BpKatG
The existence of a covalent structure linking the side chains of Trp 111 , Tyr 238 and Met 264 in BpKatG (Fig.1) was originally deduced from the electron density maps derived from crystals of both HmCPx (6) and BpKatG (7). In order to confirm the existence of such an unusual structure, the peptide mixture generated by trypsin digestion of BpKatG was analyzed by mass 7 spectrometry. Each of the key residues in the structure is located on a separate tryptic peptide fragment, and the absence of these fragments combined with the presence of larger fragments of appropriate mass would confirm the presence of the adduct (Fig. 2). Some of the peptides identified by MALDI mass spectrometry from both BpKatG and its W111F variant are listed in Table 2. Significantly, the expected fragment at m/z 1179 (containing Trp 111 ) is completely absent from the BpKatG spectrum, but the equivalent ion in the spectrum of the W111F variant,  (Table 4). Assigning X as homoserine, which would arise from hydrolysis of the Tyr-Met covalent bond (Fig 4) and which has a mass 30 Da less than that of Met, explains the mass differential between the ions at m/z 2062 and m/z 2092. In addition, this represents further indirect evidence for the covalent link between Tyr 238 and Met 264 .

MS characterization of the Trp-Tyr-Met adduct in HPI
The presence of the covalent adduct in catalase-peroxidases from two such disparate sources as the archaebacterium H. marismortui and the Gram negative bacterium B.
pseudomallei suggested that it may be a feature common to all catalase-peroxidases. This was 9 explored in an analysis of the tryptic digest of HPI (KatG) from E. coli and its W105F variant (Table 5). All three of the fragments containing respectively Trp 105 (m/z 1149), Met 252 (m/z 2532) and Tyr 226 (m/z 3206) are present, their identities confirmed by MS/MS analysis. While this suggests that the adduct may not be present in HPI, a cluster of ions, also separated by approximately 16 Da, is evident near m/z 4350 (Fig 6A) in the digest of native HPI but not in the digest of the W105F variant ( Fig 6B). Analysis of the predominant ion by MS/MS reveals a fragmentation pattern consistent with the presence of the Trp-Tyr covalent structure (Fig 6C).
The location of the modifications, either from the addition of CH 3 -S-(+47 less one proton for +46) or oxidation (+48 for three oxygen atoms less 2 protons for +46), could be localized to the hybrid fragment bounded by ions y25 and y16 ( Fig. 6C and Table 6

Identification of the N-terminus of BpKatG
The N-terminal 34 residues predicted by the DNA sequence were not evident in the electron density maps of BpKatG (7), raising the question of whether they were present and disordered or absent as a result of N-terminal processing. The tryptic fragments corresponding to residues 1 to 9 and 19 to 26, including possible partial digest fragments, are absent from the spectrum, whereas the fragment corresponding to residues 27 to 40 is present and its sequence corroborated by MS/MS analysis (Table 2). From these data, it can be concluded that the N- The core structure of the individual N-and C-terminal domains of catalase-peroxidases is very similar to the core structure of plant peroxidases, suggesting that the enzyme is a peroxidase that has adopted an efficient catalatic activity during evolution. A small number of clues for how this adaptation took place are provided by the structure of the active site. The such as cytochrome c peroxidase and ascorbate peroxidase do not exhibit significant catalatic activity. Furthermore, the two other key active site residues in peroxidases, the arginine and histidine equivalent to Arg108 and His112 of BpKatG, are spatially oriented much the same as in BpKatG. Indeed, the root mean square deviation of the C α of 133 residues in conserved α-helical segments, including the three active site residues, is just 0.97 C comparing BpKatG and cytochrome c peroxidase (7). The most significant differences between the catalase-peroxidases and the peroxidases, aside from sheer size, lie in the unusual post-translational modifications in the catalase-peroxidases, the Trp-Tyr-Met adduct and the modified heme, and it is reasonable to consider how these features might make the catalase reaction possible.
The covalently linked residues would form a very rigid structure that would fix the position of the indole nitrogen of the essential Trp relative to the heme iron and imidazole ring of the essential His. Such precise positioning with no possibility of movement may be necessary to generate optimal interaction distances with the H 2 O 2 for the reduction of compound I. Indeed, mutation of Met 264 , which would prevent formation of at least part, and possibly all, of the covalent structure, significantly reduces catalatic activity, with little effect on peroxidatic activity ( Table 1) and mutation of the equivalent of Tyr 238 in Synechocystis KatG has a similar effect (20). The covalent linkages may also affect the electronic environment of the indole, enhancing its ability to bind H 2 O 2 for compound I reduction. In addition the adduct creates an obvious 1 2 route for delocalization of the radical from the heme of compound I, a process recently demonstrated in M. tuberculosis KatG (21). From the standpoint of the peroxidatic reaction, electron tunneling from a peroxidatic substrate on the surface (7) to the heme for reduction of compound I or compound II may also be facilitated by the adduct. It is tempting to speculate about the reaction mechanism responsible for the Trp-Tyr-Met covalent structure, and both free radical and ionic mechanisms can be presented that may be initiated by oxidation of the most reactive group, Met 264 . However, structural analysis of variants lacking the three involved residues, and other nearby residues, are required to determine what residues are necessary and to see if partial adducts can be formed will provide a much firmer basis for such speculation.   y18 and all the b-series are 30-31 Da larger than expected from the sequence shown, whereas the y6 to y15 ions agree well with the expected sizes. The fragment sizes are summarized in Table 3. The ion at m/z 1792 was not identified.   Table 4.  Table 6.