OUP user menu

Characterization of a bifunctional catalase–peroxidase of Burkholderia cenocepacia

Panagoula Charalabous, Janet M. Risk, Rosalind Jenkins, Andrew J. Birss, C. Anthony Hart, John W. Smalley
DOI: http://dx.doi.org/10.1111/j.1574-695X.2007.00224.x 37-44 First published online: 1 June 2007


Isolates of Burkholderia cenocepacia express a putative haem-binding protein (molecular mass 97 kDa) that displays intrinsic peroxidase activity. Its role has been re-evaluated, and we now show that it is a bifunctional catalase–peroxidase, with activity against tetramethylbenzidine (TMB), o-dianisidine, pyrogallol, and 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulphonic) acid (ABTS). Both peroxidase and catalase activities are optimal at pH 5.5–6.0. The gene encoding this enzyme was cloned and expressed in Escherichia coli. We have named it katG because of its similarity to other katGs, including that from Burkholderia pseudomallei. It is substantially similar to a previously described catalase–peroxidase of B. cenocepacia (katA). MS analysis indicated that the initial katG translation product may be post-translationally modified in B. cenocepacia to give rise to the mature 97-kDa catalase–peroxidase.

  • catalase–peroxidase
  • Burkholderia cenocepacia
  • cystic fibrosis


Burkholderia cenocepacia (genomovar IIIa) isolates are associated with life-threatening lung infections that may progress to septicaemia (‘cepacia syndrome’) in cystic fibrosis (CF) patients (Hart & Winstanley, 2002). These bacteria were shown to express a 97-kDa putative haem-binding protein (Smalley et al., 2001), but more recent studies (Smalley et al., 2003) have shown that this protein does not bind haem dose-dependently, and is peroxidase-positive even without prior exposure to exogenous iron(III) protoporphyrin IX. These newer findings suggest the possibility that it is not a true haem-binding protein. In the present study, we have re-evaluated the properties of the 97-kDa protein and demonstrated it to be a bifunctional catalase–peroxidase. The gene encoding this protein was cloned and expressed in Esherichia coli.

Materials and methods

Bacterial strains and growth conditions

Burkholderia cenocepacia clonal isolates BC7, C5424, C6433 and J2315 (expressing the 97-kDa putative haem-binding protein) were maintained by subculture on horse blood agar. For peroxidase and catalase studies, the cells were subcultured three times on Columbia or M9 Minimal Salts Medium agar (Sigma Chemical Company) before growth in bulk on these solid media as lawn growths for 3 days. Cells were harvested into and washed twice in 0.14 M NaCl/0.1 M Tris-HCl (pH 7.5) to remove any contaminating growth medium constituents. For molecular genetic studies, E. coli strains were grown in Luria–Bertani broth or agar, supplemented with 100 µg ampicillin mL−1 where appropriate.

Sodium dodecyl sulfate polyacrylamide gel eletrophoresis (SDS-PAGE) and staining for peroxidase and catalase activity

Cell samples of strain J2315 were solubilized in nonreducing sample buffer (37°C for 1 h), electrophoresed on 10% acrylamide gels, and stained with 3,3′,5,5′-tetramethylbenzidine (TMB)–H2O2 (Smalley et al., 2001). Gels were counterstained with Coomassie Blue (Smalley et al., 2003) to precisely identify the positions of peroxidase bands. The chromogenic peroxidase substrates o-dianisidine (3,3′-dimethoxybenzidine), pyrogallol (1,2,3-trihydroxybenzene), 4-chloronaphthol, guaiacol (2-methoxyphenol) and 2,2′-azino-bis(3-ethylbenzthiazoline-6-sulphonic) acid (ABTS) (all at 6.3 mM) were also tested using the same procedure as for TMB. Catalase activity was detected in samples as above using the K3Fe(CN)6–FeCl3–H2O2 method of Katsuwon & Anderson (1992), in which catalase-positive bands appear as clear zones against a dark brown background.

Effect of pH on the peroxidase and catalase activities

Replicate gel tracks of cell samples (strain J2315; 25 µg of protein) were incubated, after electrophoresis, in 0.14 M NaCl buffered at pH 9.0, 8.5, 8.0 or 7.5 with 0.1 M Tris-HCl, or at pH 7.0, 6.5, 6.0 or 5.5, buffered with 0.1 M sodium acetate/acetic acid, for 1 h. The gel strips were developed in K3Fe(CN)6–FeCl3–H2O2 or TMB–H2O2 to reveal catalase and peroxidase bands, respectively.

pH activity profiles for whole cell catalase and peroxidase

Peroxidase activity was measured by monitoring ΔOD645 nm after incubation of cell suspensions (J2315; 250 µg protein mL–1) in the above buffers at 20°C with 6.3 mM TMB plus 10 mM H2O2 (Josephy et al., 1982). Catalase activity was measured at 20°C in the above suspensions by monitoring ΔOD240 nm (Beers & Sizer, 1952), using 8 mM H2O2 as substrate. Catalase and peroxidase activities were expressed as either pmol H2O2 degraded or pmol TMB oxidized per minute.

Protein separation and MS

Protein bands from Coomassie Blue-stained one-dimensional gels were excised, and destained and dehydrated by incubation with 50% acetonitrile/50 mM ammonium bicarbonate for 1 h at room temperature, followed by vacuum drying in a SpeedVac (Eppendorf). The gel pieces were rehydrated in 50 mM ammonium bicarbonate containing 40 ng µL−1 modified trypsin (Promega), and incubated for 16 h at 37°C. Peptides were extracted from the gel by incubation with two changes of 60% acetonitrile/1% trifluoroacetic acid (TFA), and the resulting supernatants were again dried in a SpeedVac. The extracts were desalted using C18 ZipTips according to the manufacturer's instructions (Millipore), and were reconstituted in a final volume of 30 µL of 5% acetonitrile/0.1% TFA. Aliquots of 0.5 µL were spotted onto a matrix-assisted laser-desorption ionization (MALDI) target plate together with an equal volume of 5 mg mL−1α-cyano-4-hydroxycinnamic acid (LaserBiolabs, France) in 50% acetonitrile/0.1% TFA. Peptide mass fingerprints were acquired either on a Voyager DE Pro MALDI (Applied Biosystems, CA) or a M@LDI (Micromass, Manchester, UK) instrument in positive ion reflector mode. Data were submitted for screening via the mascot search engine (Matrix Science, London). The mass tolerance was set to 100 p.p.m., and one missed cleavage and no modifications were allowed. For liquid chromatography (LC)-MS/MS analysis, aliquots of 5 µL were delivered into a QSTAR Pulsar i hybrid mass spectrometer (Applied Biosystems) by automated in-line LC [Integrated LCPackings System, 5-mm C18 nano-precolumn and 75 µm × 15 cm C18 PepMap column (Dionex, CA)] via a nano-electrospray source head and a 10-µm inner diameter PicoTip (New Objective, MA). A gradient from 5% acetonitrile/0.05% TFA (v/v) to 48% acetonitrile/0.05% TFA (v/v) in 60 min was applied at a flow rate of 300 nL min−1, and MS and MS/MS spectra were acquired automatically in positive ion mode using information-dependent acquisition (Analyst, Applied Biosystems). Database searching was carried out using mascot with mass tolerances set to 1.2 Da for MS and 0.6 Da for MS/MS, and with deamidation as a variable modification.

Cloning of the catalase–peroxidase (katG) gene from B. cenocepacia strain J2315

Genomic DNA was amplified by PCR using the primers CACCATGTCGAACGAAGGGCAGT and CGATGTACCACCGCTTTT. PCR was performed with an initial denaturation step at 94°C for 5 min, followed by 35 cycles each of 1 min at 94°C, 1 min at 58°C, 3.5 min at 72°C, and a final extension step at 72°C for 5 min. PCR products were electrophoresed on a 1% (w/v) agarose gel, and visualized using ethidium bromide to confirm a single product of the correct size before cloning into pET-100/D-TOPO (Invitrogen) to produce pKatG and transformation of One Shot® TOP 10 chemically competent E. coli cells (Invitrogen). Successful insertion of the gene in the correct orientation was confirmed by restriction analysis.

Sequencing of katG

The katG gene was sequenced from pKatG plasmid DNA using multiple overlapping IRD-700-labelled forward and reverse primers (MWG-Biotech Ltd, UK; details on request) and employing the SequiThermEXCEL II DNA Sequencing Kit-LC (Epicentre Technologies) in an LI-COR 4200S autosequencer. The sequencing data were viewed using the li-cor base imagir software in conjunction with the sequencher program (http://www.genecodes.com).

Expression analysis

BL21 Star One Shot E. coli cells were transformed with the pKatG plasmid, and induction of expression was undertaken by growth at 37°C in the presence of 0.1 M isopropyl β-d-thioglucopyranoside (IPTG; Sigma). Pelleted cells were electrophoresed as above, and assessed for catalase and peroxidase activity by in-gel staining.


As previously reported (Smalley et al., 2001), a major peroxidase-positive protein of 97 kDa was seen for B. cenocepacia strain J2315 (Fig. 1, tracks b and h). The 97-kDa enzyme also showed peroxidase specificity towards o-dianisidine, ABTS and pyrogallol (Fig. 1, tracks c, e and g), but guaicol (methoxyphenol) and 2-chloronaphthol were not peroxidized (Fig. 1, tracks d and f). Using in-gel K3Fe(CN)6–FeCl3–H2O2 staining, the 97-kDa peroxidase protein was also shown to display catalase activity (Fig. 1, track a), a phenomenon also demonstrated by the clonal isolates BC7, C5424 and C6433 (data not shown). The catalase activity of the 97-kDa enzyme of strain J2315 was not inhibited by pre-exposure of the cells for 1 h to the specific monofunctional catalase inhibitor 3-amino-1,2,4-triazole (20 mM). Peroxidase activity against TMB, ABTS and o-dianisidine was also unaffected by the inhibitors isonicotinic acid and niacinamide (all 20 mM) (Lefebre & Valvano, 2001) (data not shown). In general, the catalase activity visualized with the in-gel assay was stronger than the peroxidase staining.

Figure 1

Catalase and peroxidase specificity of the 97-kDa protein of Burkholderia cenocepacia strain J2315 as shown after SDS-PAGE under nonreducing conditions. Catalase activity was demonstrated using the potassium ferricyanide–ferric chloride–H2O2 method (a), and peroxidase activity was assessed against (b) TMB, (c) o-dianisidine, (d) 2-methoxyphenol, (e) ABTS, (f) 4-chloronaphthol, and (g) pyrogallol. (h) Coomassie Blue counterstaining for protein following the peroxidase reaction. See ‘Materials and methods’ for details.

Both catalase and peroxidase activities were observed in SDS-PAGE gels over the pH range 5.5–8.5, whereas little or no activity was seen at pH 9.0 (Fig. 2). Both enzyme activities were generally low at alkaline pH values, and highest over the acid pH range, and scanning densitometry confirmed maximal catalase and peroxidase activities at pH 6.0 and 5.5, respectively. Peroxidase activity of suspensions of whole cells of strain J2315 against TMB was not detected at neutral or alkaline pH values, but maximal activity was seen at pH 6.0 (Fig. 3a). In contrast, low levels of catalase activity of whole cells were observed at alkaline pH, rising in the acid pH range to a maximum at pH 6.0 (Fig. 3b).

Figure 2

Effect of pH on the activity of the bifunctional catalase–peroxidase as shown after SDS-PAGE under nonreducing conditions. Peroxidase activity (a) was revealed using TMB–H2O2, and catalase was assayed using the K3Fe(CN)6–FeCl3–H2O2 staining method (b). Gel loadings were 25 µg of protein per track.

Figure 3

pH activity profile of catalase and peroxidase activities of suspensions of whole cells of Burkholderia cenocepacia strain J2315 grown on M9 Minimal Salts Medium agar. Peroxidase was measured using TMB as substrate. Catalase activity was assayed by UV absorbance. The reactions were carried out at 20°C.

Masses of tryptic peptides of the 97-kDa catalase–peroxidase were obtained by MALDI–time of flight MS, and matches were obtained to the KatG catalase–peroxidases of B. pseudomallei (Donald et al., 2003) and Mycobacterium tuberculosis (Sonnenberg & Belisle, 1997), and to an archeal catalase–peroxidase of Haloarcula marismortui (Cannac-Caffrey et al., 1998). These genomic data were aligned with the sequence of the J2315 strain (http://www.sanger.ac.uk/Projects/B_cenocepacia/) and used to design primers to a putative ORF of 2211 bases (768 amino acids) on chromosome 2, which was cloned and resequenced (accession number DQ112341). This ORF was subsequently identified in the published B. cenocepacia genome (Sanger Institute), and was calculated to have a size of 80.5 kDa. A second ORF was also identified at 3 612 912–3 615 098 on chromosome 1 that possessed 73% identity and 81% similarity to the first catalase–peroxidase gene at the amino acid level and may represent another catalase–peroxidase protein.

Paired amino acid alignments (blastp) revealed a high degree of homology between the B. cenocepacia catalase–peroxidase and the minor catalase–peroxidase described by Lefebre et al., (2005), those of other selected bacterial species, including the cell surface catalase–peroxidase of B. pseudomallei (KatG), and the 77-kDa iron(III) protoporphyrin IX monomer-binding protein (accession number DQ114424; Smalley et al., 2005) (Table 1). Multiple sequence alignment analysis using clustalw revealed striking similarities between the B. cenocepacia catalase–peroxidase and the other selected enzymes in both the C-terminal and N-terminal regions. These included the conserved amino acid triad Arg104-Trp107-His108, and the second haem ligand (His270) of M. tuberculosis KatG (Zamocky et al., 2001). Because of the similarities of the catalase–peroxidase to these well-characterized KatG proteins, the gene encoding the B. cenocepacia enzyme was named katG.

View this table:
Table 1

Analysis of the Burkholderia cenocepacia catalase–peroxidase KatG (GenBank accession number DQ112341) amino acid homology with bacterial and archaeal catalase–peroxidases using paired alignment comparisons performed in blastp

OrganismAccession no.Reference% identity% similarity
Haloarcula marismortuigbY16851Cannac-Caffrey et al., (1998)56.869.4
Halobacterium salinarumgbAF069761Long & Salin (2001)56.468.9
Mycobacterium bovisspP46817Heym et al., (1995)60.071.8
Mycobacterium fortuitumgbY07865Menendez et al., (1997)58.165.9
Mycobacterium intracellularespQ04657Morris et al., (1992)58.669.3
Mycobacterium tuberculosisspQ08129Heym et al., (1993)59.971.6
Escherichia colispP13029Triggs-Raine et al., (1988)59.971.8
E. coli (0157: H7)gbX89017Brunder et al., (1996)55.266.2
Legionella pneumophilagbAF078110Bandyopadhyay & Steinman (1998)57.169.3
Salmonella typhimuriumspP17750Loewen & Stauffer (1990)60.671.9
Streptomyces reticuligbY14317Zou et al., (1999)64.074.0
Yersinia pestisgbAF135170Garcia et al., (1999)55.667.0
Burkholderia pseudomalleigbAAK72466Loprasert et al., (2002)70.779.0
Burkholderia cenocepaciagbDQ114424Smalley et al. (2005)71.079.0
Burkholderia cenocepaciagbAF317697Lefebre et al., (2005)94.094.0
  • Accession numbers are from GenBank (gb) or Swiss-Prot (sp).

IPTG induction of the E. coli BL21Star cells carrying the 80.5-kDa catalase–peroxidase gene resulted in a protein product that electrophoresed on 10% gels as a single band with an apparent molecular mass of c. 80 kDa (Fig. 4). The recombinant protein stained positively for both peroxidase and catalase, showing the gene product to be functionally active (Fig. 4). To confirm the identity of the recombinant enzyme expressed in E. coli as the product of katG, MS/MS analysis was performed after SDS-PAGE and trypsin digestion. On 7% acrylamide gels, it was found that the recombinant protein was separated into two bands with calculated molecular masses of 79 and 83 kDa, denoted R1 and R2, respectively (Fig. 5), both of which were positive for peroxidase (data not shown). The observation of the band R2 is in keeping with the expected size of an initial translation product based on the vector system employed, which results in the addition of 36 amino acids to the N-terminus of the expressed protein. We speculate that the lower molecular mass band R1 arises as a result of proteolytic cleavage of the initial translation product. In addition to proteins R1 and R2, a very faint Coomassie Blue-stained band of c. 97 kDa (denoted R3) was observed that was not expressed by E. coli cells carrying the empty plasmid. This band was peroxidase-positive, as revealed by TMB staining for a longer time period (data not shown). MS/MS analysis of these three proteins showed that they matched B. cenocepacia KatG (Table 2). Taken together, these data confirmed that the product of katG was the bifunctional catalase–peroxidase. The presence of a higher molecular mass form of the enzyme suggests that the initial translation product may be post-translationally modified to give the mature 97-kDa catalase–peroxidase.

Figure 4

SDS-PAGE on 10% polyacrylamide gels of BL21Star Escherichia coli cells expressing the catalase–peroxidase, after growth in the presence of 0.1 M IPTG. C, catalase; Per, peroxidase; Prot, protein staining; M, molecular mass markers. The in-gel catalase staining was performed at pH 6.0 in 0.5 M sodium acetate.

Figure 5

SDS-PAGE of the recombinant Burkholderia cenocepacia KatG catalase–peroxidase enzyme on a 7% polyacrylamide gel. Recombinant protein bands R1, R2 and R3 gave mass matches to KatG after trypsinization and MS/MS analysis (see Table 2). The gel was stained with Coomassie Blue.

View this table:
Table 2

Masses of tryptic peptides derived from the recombinant protein

mascot score=875520–532681.87581361.6575R.IQGEFNSTQPGGK.K
mascot score=871383–389449.2540896.4279R.FDPVYEK.I
mascot score=964520–532681.84081361.6575R.IQGEFNSTQPGGK.K
639–6551004.98402008.9530K.HGVFTDQPETLTVDFFR.N (D)
  • Analysis was performed with one missed cleavage allowed and deamidation as the only variable modification.

  • A mass tolerance of 1.2 Da was allowed for MS, and a mass tolerance of 0.6 Da was allowed for MS/MS analysis.

  • R1, R2 and R3 refer to the arrowed bands on Fig. 5.

  • m/z, mass/charge; Mr (calc.), calculated relative molecular masses of the sequenced tryptic peptides; D, deamidation.


We have re-evaluated the role of the 97-kDa putative haem-binding protein of B. cenocepacia (Smalley et al., 2001). This protein shows peroxidase specificity towards TMB, o-dianisidine, pyrogallol, and ABTS, but not against 4-chloronaphthol and 2-methoxyphenol (guaicol). It also has catalase activity, but is not inhibited by the specific monofunctional catalase inhibitor 3-amino-1,2,4,-triazole. This protein does not show dose-dependent binding of iron(III) protoporphyrin IX in either the monomeric or µ-oxo oligomeric form (Smalley et al., 2003), and does not bind to haem–agarose (J.W. Smalley et al., unpublished results). Collectively, these data show that this component is not a true haem-binding protein, but a bifunctional catalase–peroxidase, in contrast to the 77-kDa and 149-kDa iron(III) protoporphyrin IX-binding, outer-membrane components, which do not possess intrinsic catalase activity, and which are only peroxidase-positive after exposure to, and binding of, iron(III) protoporphyrin IX monomers (Smalley et al., 2003).

Multiple amino acid alignment analysis of the translated B. cenocepacia catalase–peroxidase gene revealed a strong homology with other bacterial catalase–peroxidases, and supported the above biochemical observations. It possessed the Arg88-Trp91-His92 triad, which is conserved among all known catalase–peroxidases (Zamocky et al., 2001), and displayed the greatest cross-species amino acid homology (70.7% identity and 79% similarity) to that of B. pseudomallei KatG, a homodimer of subunit size 81.6 kDa (Carpena et al., 2002, 2003), which plays a role in protecting against hydrogen peroxide (Loprasert et al., 2003). For this reason, the B. cenocepacia catalase–peroxidase gene was named katG. A catalase–peroxidase gene katA (accession number AF317697), which is similar to the gene identified in this article, has recently been described in B. cenocepacia strain C5424 by Lefebre et al., (2005).

MS/MS analysis clearly demonstrated that the recombinant catalase–peroxidase was the product of katG. The detection of a higher molecular mass enzyme matching KatG shows that expression of katG in E. coli is also accompanied by some post-translational processing, and suggests that this step may be more efficient in B. cenocepacia, giving rise to the mature 97-kDa catalase–peroxidase. At present, however, the nature of any post-translational modifications is unclear. The electrophoretic mobility of the native 97-kDa enzyme from B. cenocepacia does not change upon reduction with dithiothreitol, and nor does it react with phosphoprotein stains or periodic acid–Schiff reagent (data not shown).

Recent B. cenocepacia J2315 sequence database releases indicate that the katG gene (BCAM2107) may actually be extended by 20 amino acids at the N-terminus to give a 756 amino acid, 82.6-kDa protein. This, together with other sequence differences that we have noted between our data, the B. cenocepacia J2315 database sequence, and the katA gene of Lefebre et al., (2005), may mean that the ORF is not yet correctly identified. Lefebre et al., (2005) also demonstrated a second catalase–peroxidase gene in B. cenocepacia J2315. We confirm the presence of this second gene (BCAL3299), and observe that it is 73% identical and 81% similar to our B. cenocepacia katG gene at the amino acid level.

Burkholderia cenocepacia katA mutants are sensitive to H2O2, and katA also appears to contribute to the normal functioning of the tricarboxylic acid cycle (Lefebre et al., 2005), but the extent to which the catalase–peroxidase described herein contributes to growth and survival in vivo is not clear. Although the pH of the liquid surface layer of the lung in health is c. 6.9 (Jayaraman et al., 2001), endobronchial pH values of c. 6.5 have been recorded. In addition, respiratory mucins, to which B. cenocepacia binds specifically (Sajjan et al., 1992), are highly sulphated, especially those produced by CF patients (Chace et al., 1983, 1985; Cheng et al., 1989), and this also contributes to the acidity of the secretions. The pH is also reduced as a result of defective transmembrane conductance regulator function (Coakley & Boucher, 2001) and the mucopurulent secretions formed in the CF lung during infection also have an acid pH (Bodem et al., 1983; Yoon et al., 2006). In view of the above and the acid pH optima of the catalase–peroxidase, it is likely that bacterial cells expressing this enzyme would be advantaged in enduring attack by macrophage-derived H2O2 in the slightly acidic conditions prevailing in the lung during chronic infection and inflammation. Although members of the ‘B. cepacia complex’ display catalase and peroxidase activities (Lefebre & Valvano, 2001), we have generally found that these activities in other species of the complex are very low compared to B. cenocepacia strains (Charalabous, 2004). Bacterial catalase–peroxidases display wide substrate specificities (Marcinkeviciene et al., 1995; Zamocky et al., 2001), but it is not known which compounds represent natural peroxidase substrates for the B. cenocepacia enzyme or whether it plays any role in attacking and degrading other host (macro)molecules for defensive or nutritive purposes.


We would like to thank the Cystic Fibrosis Trust (Grant number CF RS 22) and the MRC Proteomics Initiative for financial support.


  • Editor: Kai Man Kam


View Abstract