An Oxyferrous Heme/Protein-based Radical Intermediate Is Catalytically Competent in the Catalase Reaction of Mycobacterium tuberculosis Catalase-Peroxidase (KatG)

Article (PDF Available)inJournal of Biological Chemistry 284(11):7017-29 · February 2009with67 Reads
DOI: 10.1074/jbc.M808106200 · Source: PubMed
Abstract
A mechanism accounting for the robust catalase activity in catalase-peroxidases (KatG) presents a new challenge in heme protein enzymology. In Mycobacterium tuberculosis, KatG is the sole catalase and is also responsible for peroxidative activation of isoniazid, an anti-tuberculosis pro-drug. Here, optical stopped-flow spectrophotometry, rapid freeze-quench EPR spectroscopy both at the X-band and at the D-band, and mutagenesis are used to identify catalase reaction intermediates in M. tuberculosis KatG. In the presence of millimolar H2O2 at neutral pH, oxyferrous heme is formed within milliseconds from ferric (resting) KatG, whereas at pH 8.5, low spin ferric heme is formed. Using rapid freeze-quench EPR at X-band under both of these conditions, a narrow doublet radical signal with an 11 G principal hyperfine splitting was detected within the first milliseconds of turnover. The radical and the unique heme intermediates persist in wild-type KatG only during the time course of turnover of excess H2O2 (1000-fold or more). Mutation of Met255, Tyr229, or Trp107, which have covalently linked side chains in a unique distal side adduct (MYW) in wild-type KatG, abolishes this radical and the catalase activity. The D-band EPR spectrum of the radical exhibits a rhombic g tensor with dual gx values (2.00550 and 2.00606) and unique gy (2.00344) and gz values (2.00186) similar to but not typical of native tyrosyl radicals. Density functional theory calculations based on a model of an MYW adduct radical built from x-ray coordinates predict experimentally observed hyperfine interactions and a shift in g values away from the native tyrosyl radical. A catalytic role for an MYW adduct radical in the catalase mechanism of KatG is proposed.
An Oxyferrous Heme/Protein-based Radical Intermediate Is
Catalytically Competent in the Catalase Reaction of
Mycobacterium tuberculosis Catalase-Peroxidase (KatG)
*
S
Received for publication, October 22, 2008, and in revised form, December 23, 2008 Published, JBC Papers in Press, January 12, 2009, DOI 10.1074/jbc.M808106200
Javier Suarez
‡§
, Kalina Ranguelova
‡1
, Andrzej A. Jarzecki
‡§
, Julia Manzerova
, Vladimir Krymov
, Xiangbo Zhao
,
Shengwei Yu
, Leonid Metlitsky
, Gary J. Gerfen
, and Richard S. Magliozzo
‡§2
From the
Department of Chemistry, Brooklyn College, Brooklyn, New York 11210, the
§
Departments of Chemistry and
Biochemistry, Graduate Center, City University of New York, New York, New York 10016, and the
Department of Physiology and
Biophysics, Albert Einstein College of Medicine, Bronx, New York 10461
A mechanism accounting for the robust catalase activity in cata-
lase-peroxidases (KatG) presents a new challenge in heme protein
enzymology. In Mycobacterium tuberculosis, KatG is the sole cata-
lase and is also responsible for peroxidative activation of isoniazid,
an anti-tuberculosis pro-drug. Here, optical stopped-flow spectro-
photometry, rapid freeze-quench EPR spectroscopy both at the
X-band and at the D-band, and mutagenesis are used to identify
catalase reaction intermediates in M. tuberculosis KatG. In the
presence of millimolar H
2
O
2
at neutral pH, oxyferrous heme is
formed within milliseconds from ferric (resting) KatG, whereas at
pH 8.5, low spin ferric heme is formed. Using rapid freeze-quench
EPR at X-band under both of these conditions, a narrow doublet
radical signal with an 11 G principal hyperfine splitting was
detected within the first milliseconds of turnover. The radical and
the unique heme intermediates persist in wild-type KatG only dur-
ing the time course of turnover of excess H
2
O
2
(1000-fold or more).
Mutation of Met
255
, Tyr
229
,orTrp
107
, which have covalently linked
side chains in a unique distal side adduct (MYW) in wild-type
KatG, abolishes this radical and the catalase activity. The D-band
EPR spectrum of the radical exhibits a rhombic g tensor with dual
g
x
values (2.00550 and 2.00606) and unique g
y
(2.00344) and g
z
values (2.00186) similar to but not typical of native tyrosyl radicals.
Density functional theory calculations based on a model of an
MYW adduct radical built from x-ray coordinates predict experi-
mentally observed hyperfine interactions and a shift in g values
away from the native tyrosyl radical. A catalytic role for an MYW
adduct radical in the catalase mechanism of KatG is proposed.
Heme enzymes such as catalases and peroxidases are ubiqui-
tous in aerobic organisms and are principally responsible for
eliminating the potential damaging effects of hydrogen perox-
ide. The structure and function of the monofunctional enzymes
have received abundant attention for many years, whereas the
dual function enzyme, catalase-peroxidase (KatG),
3
found in
bacteria and fungi, is a relative newcomer and is less well char-
acterized. In the pathogen Mycobacterium tuberculosis, which
still infects and kills millions of people each year, KatG is the
only catalase and is required for virulence (1, 2). The peroxidase
activity of KatG is considered to be responsible for activation of
the anti-tuberculosis therapeutic agent, isoniazid (3, 4), and
mutations in this enzyme are the principal origin of widespread
resistance to this pro-drug (5, 6).
Investigation of the properties of KatG is of interest because
the correlation between unusual features of its structure and
particular mechanistic steps is not fully defined nor under-
stood. The post-translational modification of residues Met
255
,
Tyr
229
, and Trp
107
, the side chains of which form an adduct on
the distal side of the heme (Fig. 1), is the most intriguing struc-
tural feature and is required for catalase but not peroxidase
activity (7–10). KatG catalytic function has features in common
with classical mono-functional enzymes; for example, an oxof-
erryl-porphyrin
-cation radical species is formed from resting
enzyme in turnover with alkyl peroxides (11–13), and this inter-
mediate is taken to be common to both peroxidase and catalase
reaction paths. Its formation has not been observed in the pres-
ence of hydrogen peroxide however, except in mutants such as
W107F and Y229F that lack catalase activity (7, 14). Interest-
ingly, a poor rate of reaction with hydrogen peroxide of KatG
Compound (Cmpd) I prepared using alkyl peroxide was
recently presented as evidence that the catalase mechanism in
KatG diverges from the typical monofunctional enzyme path-
way exemplified by bovine liver catalase (15). A reaction
between hydrogen peroxide and Cmpd I would be expected to
constitute the second phase of the dismutation in which H
2
O
2
acts as a two-electron reductant. A mechanism explaining the
robust catalase activity exhibited by KatGs is therefore of spe-
cial interest because novel mechanistic features are likely to be
uncovered; Cmpd I is apparently not a catalytically competent
intermediate, and high activity is absent from other homolo-
* This work was supported, in wholeor in part, by National Institutes of Health
Grants AI060014 (NIAID) (to R. S. M.) and GM075920 (to G. J. G.). The costs
of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement”in
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
S
The on-line version of this article (available at http://www.jbc.org) contains
supplemental Figs. S1–S5.
1
Present address: Laboratory of Pharmacology, National Institute of Environ-
mental Health Sciences, National Institutes of Health, P. O. Box 12233, MD
F0-01, 111 TW Alexander Dr., Research Triangle Park, NC 27709.
2
To whom correspondence should be addressed: Dept. of Chemistry, Brook-
lyn College, 2900 Bedford Ave., Brooklyn, NY 11210. E-mail: rmaglioz@
brooklyn.cuny.edu.
3
The abbreviations used are: KatG, catalase-peroxidase; KatG[W107F],
W107F mutant of KatG; KatG[Y229F], Y229F mutant of KatG; KatG[M255A],
M255A mutant of KatG; Cmpd, Compound; PAA, peroxyacetic acid; MYW,
Met-Tyr-Trp adduct; WT, wild type; PDB, Protein Data Bank; DFT, density
functional theory.
THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 284, NO. 11, pp. 7017–7029, March 13, 2009
© 2009 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.
MARCH 13, 2009 VOLUME 284 NUMBER 11 JOURNAL OF BIOLOGICAL CHEMISTRY 7017
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
http://www.jbc.org/cgi/content/full/M808106200/DC1
Supplemental Material can be found at:
gous class I peroxidases. The near absence of catalase activity in
KatG mutants in which replacements are made in the residues
of the Met-Tyr-Trp (MYW) adduct strongly suggests a specific
mechanistic role is filled by modification of the distal side
residues.
KatG dismutates hydrogen peroxide in a nonscrambling
mechanism such that both oxygen atoms of dioxygen derive
from the same molecule of hydrogen peroxide (16, 17). Unlike
classical catalases, however, which do not accumulate interme-
diates during H
2
O
2
turnover because of very rapid rates of both
the peroxide reduction and the peroxide oxidation steps, KatG
forms a species characteristic of oxyferrous heme (peroxidase
Cmpd III) in the presence of high concentrations of peroxide
(11, 15, 18). This intermediate is usually stable in peroxidases
and is considered to be catalytically inert, but it must be highly
unstable in KatG because catalase turnover occurs while the
heme is in this form (In cytochrome c peroxidase oxyferrous
heme is unstable because of internal redox chemistry involving
Trp
191
, but cytochrome c peroxidase does not exhibit catalase
activity (19).) and the catalase activity of the W321F mutant of
M. tuberculosis KatG is only moderately reduced (20). In the
companion paper (69), formation of oxyenzyme from the car-
bonyl enzyme shows that in WT KatG, oxyferrous heme is too
stable to be a viable catalase intermediate unless some other
unique feature produced in the oxyferrous enzyme formed in
the presence of hydrogen peroxide alters its properties. A pro-
tein-based radical proposed here to be formed on the distal side
amino acid adduct, which co-exists with oxyferrous heme when
peroxide turnover occurs, is suggested to be a catalytically com-
petent intermediate in catalase turnover in WT KatG.
Our focus here is on identifying intermediates formed in M.
tuberculosis KatG during catalase turnover. Stopped-flow opti-
cal kinetics techniques are applied to follow the changes in
heme oxidation state in parallel with rapid freeze-quench-EPR
experiments used to monitor the production and decay of pro-
tein-based radicals. A heretofore unknown narrow doublet rad-
ical signal is characterized by X-band and D-band EPR spec-
troscopy. DFT calculations provide additional insights into the
structure of the radical. A reaction scheme originally suggested
in other work (15) is described, incorporating the new radical
proposed to be localized on the MYW adduct into the cata-
lase reaction pathway. A detailed analysis of the catalase
reaction mechanism in M. tuberculosis KatG, the only cata-
lase found in M. tuberculosis , contributes to our understand-
ing of the biology of this pathogen, which resides in host
macrophages upon infection and is subjected to extensive
oxidative stress in that environment.
MATERIALS AND METHODS
M. tuberculosis KatG was prepared from an overexpression
system in Escherichia coli strain UM262, as published previ-
ously (21), and was used usually within a few days of prepara-
tion. Mutagenesis was performed using the QuickChange II
site-directed mutagenesis kit from Stratagene (La Jolla, CA).
Pairs of complementary primers (synthesized and purified by
Operon Biotechnologies, Inc.) were designed to introduce the
required mutations. The oligonucleotide pairs (mutated
codons are in boldface) were as follows: Y390F, 5-
1153
CGGG-
TGGATCCGATCTTTGAGCGGATCACGCGTC
1186
-3 and
5-
1186
GACGCGTGATCCGCTCAAAGATCGGATCCAC-
CCG
1153
-3 (4); M255A, 5-
748
GAGACGTTTCGGCGCGCGG-
CCATGAACGACGTC
780
-3 and 5-
780
GACGTCGTTCAT-
GGCCGCGCGCCGAAACGTCTC
748
-3. Mutagenesis was
performed according to the manufacturer’s protocol, and the
reaction products were transformed into the E. coli XL1-Blue
strain for selection purposes. The presence of the mutated
codons in the katG gene was confirmed by DNA sequencing
(Gene Wiz, Inc.), and the mutated plasmid was electroporated
into E. coli strain UM262 for protein overexpression. The
mutants Y229F and W107F were prepared as reported previ-
ously (7, 22).
The catalase activity of purified KatG enzymes was evaluated
optically based on the disappearance of H
2
O
2
,atpH6or7,in20
m
M potassium phosphate buffer or in 20 mM sodium borate
buffer, pH 8.5, according to published procedures (23). No
direct comparison is presented for the rate of catalase turnover
measured optically and the intensity changes in RFQ-EPR
measurements (described below) under identical conditions
because of the large difference in sensitivity of the two tech-
niques; high enzyme concentration is needed for detection of
the radical in EPR spectra, whereas such conditions produce
bubbling in solutions that strongly interfere with absorbance
measurements.
RFQ-EPR methodology has been published elsewhere (22,
24, 25); briefly, enzyme and peroxide solutions are mixed at
room temperature in a 1:1 ratio, and the mixture is sprayed into
a funnel attached to an EPR tube submerged in an isopentane
bath held at approximately 130 °C. The frozen powder is
packed into precision bore EPR tubes, and the packed sample,
which retains isopentane, is frozen in liquid nitrogen. For quan-
titative EPR, signal intensity is based on a copper standard, and
the total intensity for RFQ samples is multiplied by 2 to correct
for isopentane dilution. The foaming of solutions because of the
rapid evolution of dioxygen leads to further dilution by an addi-
tional variable amount around 2-fold, but no further correction
of EPR signal intensity was made. In some cases, manually
mixed samples were prepared in borate buffer, pH 8.5, by mix-
ing protein solution (100
M) with an 8000-fold molar excess
of H
2
O
2
, loading the mixture into precision bore EPR tubes, and
immediately immersing the tube in liquid nitrogen. EPR spec-
tra were recorded on a Bruker E500 ElexSys X-band spectrom-
eter with samples held either in a finger Dewar for spectra at 77
K or in an Oxford liquid helium cryostat for lower tempera-
tures. Signal averaging was employed to improve the signal-to-
noise ratio when needed. Simulation of X-band EPR spectra
was performed using SimFonia software (Bruker). Optical
stopped-flow spectroscopy was performed using a HiTech
16-DX instrument as described elsewhere (4, 7, 11) in the buff-
ers listed above.
Cross-link Formation—To rule out the formation of dity-
rosine cross-links during catalase turnover, 10
M WT KatG
was incubated with 1000 8000-fold molar excess of H
2
O
2
for
15 min at room temperature. SDS-PAGE of KatG from these
samples was carried out using a PhastGel system (Amersham
Biosciences).
Catalase Activity in M. tuberculosis KatG
7018 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 NUMBER 11 MARCH 13, 2009
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
DFT Calculations—DFT calculations for the analysis of
MYW adduct models were carried out using the Gaussian 03
program package (26). Molecular structures were optimized at
the B3LYP/6–31G* level of theory in combination with the
pruned 99-point Euler-Maclaurin radial grid and 590-point
angular grid for all atoms. In addition to a single molecule
approximation (gas phase model), a mimic of the protein envi-
ronment of the MYW adduct was explored by employing the
polarizable continuum model (27) with various solvents span-
ning permittivity constants from 1 (vacuum) to about 80
(water). An initial geometry of the MYW adduct was built based
on the coordinates for the covalently linked residues Met
255
,
Tyr
229
, and Trp
107
taken from the WT M. tuberculosis KatG
structure (PDB entry 2CCA), from which we carried out a full
optimization. All atoms of the side chains up to the C-
carbons
(substituted by methyl groups) were optimized in the model.
The relative position of the three C-
carbons and the two
dihedral angles that freeze the rotations of the aromatic rings of
the tyrosyl and tryptophanyl moieties with respect to their
respective C-
atoms were kept constant retaining the values
found in the x-ray structure. This set of constraints maintained
the relative position of the MYW residues with respect to the
backbone atoms (C-
carbons), whereas their relative position
with respect to one another could be fully optimized.
The natural spin population analysis of the radical models
was carried out as implemented in the NBO 5.0 program pack-
age (28). Hyperfine coupling values for hydrogen atoms in the
radical models were computed and compared with experimen-
tal data. Reliability of the chosen level of theory (B3LYP/6
31G*) to predict the spin distribution and hyperfine parameters
was examined by comparison of the calculated electron spin
densities in free tyrosyl and neutral tryptophanyl radicals to
experimental values (29, 30), which demonstrated a good agree-
ment. An increased size of basis sets (6–311G**) did not affect
computed spin distribution or hyperfine parameters signifi-
cantly. The g value calculations for the radicals were performed
using the ADF program package (31).
High Field-EPR D-band (130 GHz) spectra were obtained
using a spectrometer assembled at the Albert Einstein College
of Medicine, Bronx, NY, that uses a quadrature detection
bridge (100-milliwatt pulsed output) and probe supplied by HF
EPR Instruments, Inc. (V. Krymov, New York). Cylindrical res-
onators operating in the TE
011
mode provide typical
/2 pulse
widths of 30–50 ns at maximum bridge power. The magnetic
field is generated by a specially designed 7 tesla superconduct-
ing magnet with a 0.5 tesla superconducting sweep/active
shielding coil (Magnex Scientific). Frozen RFQ samples were
held in quartz capillary tubes with inner and outer diameters of
0.5 and 0.6 mm, respectively, and an active volume of 0.2
l.
The frozen sample tubes were loaded into the probe under liq-
uid nitrogen and inserted into the low temperature continuous
flow cryostat (Oxford Instruments, Spectrostat model 86) and
were maintained at7Ktoanaccuracy of approximately 0.3 K
using an ITC503 temperature controller. The magnetic field
was calibrated to an accuracy of 2 G using a sample of Mn(II)-
doped MgO (32). Field-swept two-pulse (Hahn) echo detected
spectra were obtained with 180° phase cycling of the first pulse
to provide suppression of base-line artifacts. Specific experi-
mental parameters are given in the figure legends.
HF EPR spectra were simulated using software described
previously (33, 34). The hyperfine interactions were treated to
first order, and transition probabilities were taken as unity. To
fit the 130 GHz echo-detected spectra, a weighted sum of two
simulations having different g
x
values was required. Parameters
for the simulations are given in the legend to Fig. 5. Uncertainty
in the reported g values is 0.00003 and for the hyperfine cou-
pling value was 0.5 G (33, 35).
Preparation of RFQ Samples for High Field EPR—The RFQ
apparatus and its calibration for preparation of the high fre-
quency EPR samples will be described in detail elsewhere.
Briefly, it consists of a System 1000 syringe driving ram (Update
Instruments, Madison, WI) and associated reactor hoses to
produce a sample frozen at 35 ms and the following compo-
nents designed and manufactured in-house: a Wiskind-type
mixing chamber, an ejection nozzle, a freezing unit comprising
copper-beryllium alloy wheels, and a sample collection basin.
4
Two syringes of equal volume were mounted on the ram unit;
one was filled with the KatG solution (150
M) and the other
with 600 m
M H
2
O
2
. A 25-hole spraying nozzle was attached to
the outlet of the aging hose. The reaction mixture was ejected
through the nozzle onto the rapidly rotating (1500 rpm) cop-
per-beryllium alloy wheels cooled with liquid nitrogen. The fro-
zen solution was continuously scraped off the wheels by stain-
less steel blades and was collected as a fine powder under liquid
nitrogen. This powder was then divided and packed into either
standard size precision bore quartz EPR tubes (4 mm outer
diameter) for measurements at X-band or into capillaries for
D-band measurements. The 4-mm tubes were attached to a
glass funnel using heat-shrink tubing, and this assembly was
immersed in isopentane kept at 140 °C; the sample powder
was transferred into the funnel with cold metal spatulas and
packed using a stainless steel packing rod with a Teflon tip. A
packing factor of 0.8 was determined separately by comparing
the volume of frozen and then melted contents of a tube. A
home-built platform immersed in liquid nitrogen was used to
pack D-band capillary tubes.
RESULTS AND DISCUSSION
RFQ-EPR and Optical Stopped-flow Experiments—The high
catalase activity of M. tuberculosis KatG is conveniently meas-
ured in the presence of millimolar H
2
O
2
without enzyme deg-
radation or apparent inhibition. Among the accumulated
mechanistic details for this unusual behavior in a class I perox-
idase are the findings of a nonscrambling mechanism shown in
isotope-ratio experiments (16), the finding that superoxide is
not a product during turnover (16), and the demonstration that
mutation of Met
255
, Tyr
229
,orTrp
107
, the side chains of which
form the distal side adduct in KatG shown in Fig. 1, reduces the
rate thousands-fold (7–10, 14, 36–38). A recent comprehen-
sive kinetic study of heme intermediates formed during catalase
turnover (15) and suggestions from that report provided the
impetus for the application of RFQ-EPR spectroscopy to probe
for protein-based radicals in the catalase reaction pathway in
4
J. Manzerova, V. Krymov, and G. Gerfen, unpublished results.
Catalase Activity in M. tuberculosis KatG
MARCH 13, 2009 VOLUME 284 NUMBER 11 JOURNAL OF BIOLOGICAL CHEMISTRY 7019
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
WT KatG and in several mutants. Optical stopped-flow exper-
iments and low temperature EPR were also applied to be able to
monitor the kinetics of changes in heme oxidation or spin state.
RFQ-EPR sampling allows mixing of ferric KatG with H
2
O
2
followed by freeze-quenching of reactions in the millisecond
time range. Fig. 2 (top) shows spectra recorded at 77 K in the g
2 region as a function of incubation time after mixing KatG with
a 1000-fold molar excess of H
2
O
2
at pH 7 and 8.5. A narrow
doublet signal at g 2.0034 with a principal hyperfine splitting
of 11 G appears at the earliest time point accessible here and
was the only signal in the g 2 region in spectra recorded at 77
K. The intensity of the signal was maximal in the sample frozen
after 20 ms (0.1 spins/heme) and decayed thereafter until it
was no longer detectable after 200 ms (Fig. 2, top) for the
reaction run at pH 7. The actual spin concentration is likely
greater than 0.1 spin/heme because the considerable foaming
under these conditions dilutes the solutions; no correction is
made for this dilution. At pH 8.5, the narrow doublet signal
persisted 8 times longer (up to 1.7 s) (Fig. 2, bottom). When
using an 8000-fold molar excess, the narrow doublet signal per-
sists long enough that an EPR sample could be prepared by
manually mixing ferric enzyme with peroxide and rapidly
transferring and freezing the solution by immersion of the EPR
tube in liquid nitrogen. This method for sample preparation
was used below for examining radical formation in mutant
enzymes.
The catalase specific activity of M. tuberculosis KatG (4000
units/mg at pH 7 and 439 units/mg at pH 8.5) allows estima-
tion that a 1000-fold excess of peroxide should be completely
consumed within 0.2 s at pH 6 or 7 and after 1.6 s at pH 8.5.
These time intervals correspond very well to the interval during
which the radical is present, suggesting that it represents a
kinetically competent intermediate of the catalase reaction
pathway. The RFQ-EPR conditions are not amenable to optical
experiments to directly follow the rate of disappearance of
hydrogen peroxide because of active bubbling; therefore, no
visual comparison in rates is presented for the results using
these two techniques.
Stopped-flow optical experiments were performed to record
changes in heme species at pH 6 and 8.5. Previous work had
shown that small excesses of H
2
O
2
do not lead to accumulation
of new species (11). Upon mixing of enzyme with 1000-fold
molar excess of H
2
O
2
at pH 6, a species different from resting
enzyme appears at the earliest time point (5 ms) (Fig. 3, A and
B). This species lacks the CT bands of the resting (ferric)
enzyme at 500 and 630640 nm and has the features of oxyfer-
rous enzyme, including a Soret maximum at 413 nm and with
and
bands at 545 and 579 nm. The spectrum of oxyferrous
heme in M. tuberculosis KatG (11, 15) was confirmed in the
companion paper (69). The reaction pathway to oxyenzyme can
occur through reduction of Fe(IV) by peroxide, superoxide
addition to ferric heme, or dioxygen binding to ferrous iron.
The finding that this intermediate is observed only when very
large excesses of hydrogen peroxide are present argues for the
first pathway, which had been confirmed for M. tuberculosis
KatG[Y229F] (7).
In similar experiments at pH 8.5, a different heme interme-
diate was found at the earliest time point. (The optical spectrum
of the resting enzyme does not change as a function of pH from
pH 6 to 8.5.) The new species has a Soret maximum at 418 nm,
FIGURE 1. The distal side structure of M. tuberculosis KatG (from the coor-
dinates of PDB code 2CCA) showing the covalent Met-Tyr-Trp adduct.
FIGURE 2. RFQ-EPR of M. tuberculosis KatG reacted with H
2
O
2
. Top, sam-
ples were frozen at the indicated time points after mixing ferric KatG with
1000-fold excess H
2
O
2
at pH 7.2 or 8.5. Spectra (average of nine scans) were
recorded using the following conditions: T 77 K, microwave frequency
9.442 GHz; microwave power 0.1 milliwatt; modulation amplitude 1G.
Bottom, intensity (spins/heme) of the narrow doublet EPR signal as a function
of time in RFQ-EPR samples frozen after mixing KatG with 1000-fold excess of
H
2
O
2
at pH 8.5. The curve is a fit of the data to a single exponential function
using Sigma Plot 9.0.
Catalase Activity in M. tuberculosis KatG
7020 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 NUMBER 11 MARCH 13, 2009
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
a broad shoulder at 521 nm, and no other prominent features in
the visible region (Fig. 3C). The spectrum of this intermediate
decayed sharply after 1.5 s (Fig. 3D), and upon reaction of the
enzyme with 2000-, 4000-, and 8,000-fold molar excess of
hydrogen peroxide persisted for longer periods proportional to
the amount of peroxide added (data not shown). At pH 7, the
optical spectrum formed in the first milliseconds was a mixture
of the low and high pH species (data not shown). The time
interval during which the oxyenzyme intermediate was
observed at pH 6 (and also at pH 7) exceeded the time interval
during which the narrow doublet EPR signal was detected (Fig.
2) and beyond the predicted time interval for complete con-
sumption of peroxide assuming a linear reaction rate through-
out. These observations indicate interesting features of the
reaction pathway that are addressed under “Conclusions.”
Low temperature EPR spectra of RFQ-EPR samples prepared
at pH 7 did not contain a signal in either the g 6 or the g 2
regions that could be assigned to a paramagnetic intermediate
(Fig. 4). This observation is consistent with the optical evidence
for oxyferrous heme, which is nonparamagnetic. For the RFQ-
EPR samples at pH 8.5, the resting enzyme signals around g 6
were also absent, and a signal consistent with a low spin ferric
heme species was detected in the g 2 region during the cata-
lase turnover interval. The new signal was broad, but two par-
tially resolved features at g 2.24 and 2.15 of what could be a
rhombic signal were seen. The g values of the two features are
similar to those reported for the deprotonated ferric-peroxo
complex of myoglobin (39) or the peroxo-horseradish peroxi-
dase complex (40). A third feature below g 2 could not be
resolved. These results indicate a change in the steady-state
heme species from the oxyferrous intermediate to a proposed
ferric-peroxo species at alkaline pH. Note that the actual pH of
frozen RFQ-EPR samples in phosphate buffer will be below 7
because of a decrease in pH with temperature for this buffer.
This artifact does not have any impact on the narrow doublet
radical signal shown in Fig. 2, but it does remove the signal
assigned to a ferric-peroxo species at alkaline pH in Tris-male-
ate buffer (Fig. 4), the pH of which is more resistant to changes
with temperature.
The optical spectrum of the pH 8.5 samples does not resem-
ble that assigned to the ferric peroxo form of horseradish per-
oxidase or heme oxygenase (39, 40), whereas the EPR data are
clearly consistent with a low spin ferric peroxy heme; the inter-
mediate formed at pH 8.5 at room temperature had been con-
firmed elsewhere to be a ferric species because in the presence
of cyanide, a typical six coordinate low spin ferric complex was
directly formed from it (41). A complete understanding of the
identity of all species under these conditions awaits further
study.
Characterization and Identification of the Radical in WT
KatG—Turnover of KatG with H
2
O
2
is expected to generate
Cmpd I (oxoferryl porphyrin
-cation radical), which in the
experiments above is apparently rapidly reduced by endoge-
nous electron transfer producing a radical on a protein site. The
rate of this step is too rapid for finding Cmpd I in either the
optical or the EPR data gathered here. In results published ear-
lier, turnover of ferric M. tuberculosis KatG with alkyl peroxides
such as PAA generated Cmpd I and also tyrosyl and tryptopha-
nyl radicals. RFQ-EPR spectra of such samples contain wide
doublet, singlet, and also a narrow doublet EPR signal (11, 22,
24) under certain conditions. Wide doublet and singlet signals
were found when the enzyme was treated with PAA alone, and
a narrow doublet signal was found in the presence of isoniazid.
This drug is a peroxidase substrate but is also known to quench
the wide doublet radical in M. tuberculosis KatG (22, 24, 42).
The narrow doublet radical had been suggested to be localized
on free Tyr
229
, i.e. on the tyrosine not incorporated into the
FIGURE 3. Optical stopped-flow spectra of M. tuberculosis KatG reacted
with H
2
O
2
. A, spectrum (solid line) recorded after 5 ms of incubation, an inter-
mediate spectrum (dashed line) recorded 3.5 s after mixing, and a spectrum
(dotted line) recorded 10 s after mixing KatG with 1000-fold excess of H
2
O
2
at
pH 6. B, time course of the reaction followed at 407 and 580 nm (pH 6). C, spec-
trum (solid line) recorded after 1.3 ms of incubation, an intermediate spec-
trum (dashed line) recorded 1.3 s after mixing, and a spectrum (dotted line)
recorded 2 s after mixing KatG with 1000-fold excess of H
2
O
2
at pH 8.5. D, time
course of the reaction followed at 407 and 520 nm. These wavelengths cor-
respond to the Soret maximum at the start of the time course and the new
maximum in the visible region of the spectrum of the steady-state species
formed at pH 8.5.
FIGURE 4. RFQ-EPR spectra of M. tuberculosis KatG reacted with 1000-fold
molar excess of H
2
O
2
frozen at the indicated time points. Spectra (average
of 9 scans) were recorded under the following conditions: T 4 K; microwave
frequency 9.3879 GHz; microwave power 1 milliwatt; modulation ampli-
tude 4 G. At pH 8.5, a ferric heme iron intermediate is present (g 2.24,
2.15, ?); At pH 7, the only signal present is the narrow doublet signal at g
2.0034.
Catalase Activity in M. tuberculosis KatG
MARCH 13, 2009 VOLUME 284 NUMBER 11 JOURNAL OF BIOLOGICAL CHEMISTRY 7021
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
MYW adduct in freshly isolated enzyme, and the wide doublet
was also considered to represent a radical on this residue having
a different ring orientation in the absence of bound isoniazid
(22). Here we consider the following new interpretation of the
earlier findings, the narrow and wide doublet signals may be
localized on different sites and the narrow doublet only became
evident when the radical giving the wide doublet was quenched
by reaction with isoniazid. If this is the case, these two different
radicals could be formed in tandem after Cmpd I is produced. A
feature consistent with a contribution from the narrow doublet
signal can be found superimposed on the wide doublet
described in earlier reports (7, 22) (supplemental Fig. S1). The
important observation for the current results is that a radical
species giving a narrow doublet signal is suggested to be pro-
duced when ferric KatG is treated with alkyl peroxide or hydro-
gen peroxide, consistent with its formation from endogenous
electron transfer initiated by Cmpd I. Further study is needed to
confirm this idea.
Interestingly, turnover of WT KatG with excess H
2
O
2
does
not produce the wide doublet radical signal nor singlet signals
seen when Cmpd I is generated by an alkyl peroxide (in the
absence of H
2
O
2
). Also important is the observation that the
narrow doublet radical does not propagate to other protein
sites; for example, the initial wide doublet produced during
reaction with PAA evolves to other species over a period of
seconds (22, 24). Thus, the species giving the narrow doublet in
the presence of H
2
O
2
must decay in a reaction with a redox
partner that does produce the same secondary protein radicals.
These results strengthen the suggestion that the narrow dou-
blet represents an obligatory radical intermediate of the cata-
lase reaction.
EPR of Distal Side Mutants—Mutations in residues of the
distal side MYW adduct in KatG are known to disrupt catalase
activity but not peroxidase activity (7, 9, 14). EPR samples
were prepared using KatG[Y229F], KatG[W107F], and
KatG[M255A] reacted with 8000-fold excesses of H
2
O
2
. The
narrow doublet EPR signal was not formed in any of these reac-
tions (at pH 8.5) (Fig. 5). Instead, a very low yield of a singlet
EPR signal was observed, as was found in similar experiments
with KatG[W107F] and Y229F mutants using PAA (22). These
observations are fully consistent with the suggestion that the
radical species giving the narrow doublet signal in WT KatG
plays a kinetically competent role for catalase turnover. The
absence of catalase activity in these mutants could then be due
to the loss of the radical or loss of a structural feature particular
to the radical form of the WT enzyme. It is important to note
that KatG[Y229F], W107F, and M255A mutants form oxyen-
zyme in the presence of excess hydrogen peroxide just as the
WT enzyme does, which means that it is not loss of this heme
intermediate that is responsible for the loss in catalase function.
Furthermore, greatly enhanced stability of the oxyenzyme con-
tributes to the lost catalase function in the KatG[Y229F] and
W107F mutants (69).
High Field RFQ-EPR—More complete characterization of
the structure of the radical giving rise to the narrow doublet
EPR signal in WT KatG was pursued using high field RFQ-EPR
spectroscopy. Here, a 4000-fold molar excess of peroxide was
used to increase the yield of radical at an early time point and
thereby improve the signal-to-noise in D-band experimental
data. A rhombic signal with g values similar to those of tyrosyl
radicals was found for samples frozen 35 ms after mixing rest-
ing KatG with H
2
O
2
(Fig. 6A). The principal hyperfine splitting
estimated to be 11 G through simulation of the spectrum (see
below) demonstrated that this signal represents the same spe-
cies as that found at X-band. Also, a portion of the identical
sample used for the D-band measurements was loaded into a
conventional EPR tube and examined at X-band; it exhibited a
narrow doublet radical intensity equal to 20% of the heme
concentration, approximately twice the yield estimated in the
X-band data at 20 ms. The slow relaxation characteristics of
the two-pulse echo-detected D-band EPR signal suggest that the
radical site is isolated from other paramagnetic species. A sim-
ilar conclusion was drawn based on the power saturation of the
ND signal in X-band spectra (data not shown).
Interestingly, simulation of the HF-RFQ-EPR data demon-
strated that two different g
x
values, equal to 2.00550 and
2.00606, and a single isotropic hyperfine splitting of 11 G
were required for simulating the 1:2:1 triplet feature at g
x
. The
apparent splitting at g
x
is not because of the presence of two
equivalent I
1
2
hyperfine coupling interactions, which could
also produce a triplet feature, because the X-band spectrum
lacks the multiplicity expected if more than a single 11 G split-
ting were present. The principal g values (2.00606/2.00550,
2.00344, and 2.00186) obtained by simulation rule out peroxyl
(43), glycyl (44, 45), tryptophanyl (30, 46, 47), and thiyl radicals
(35, 48, 49).
The presence of multiple g
x
values has been documented in
HF-EPR spectra of protein-based tyrosyl radicals in which a
heterogeneous environment is present at the phenolic oxygen
(50–52). The simulation shown in Fig. 6 was composed of
nearly equal proportions of two rhombic signals differing only
in the value of g
x
. The proximity of the two g
x
values and the
similarity of the rest of the spectral features suggest that both
signals likely arise from the same species in slightly different
FIGURE 5. Manual freeze quench samples of KatG distal side mutants
treated with 8000-fold molar excess of H
2
O
2
at pH 8.5. Spectra (average of
nine scans) were recorded under the following conditions: T 77 K, micro-
wave frequency 9.442 GHz; microwave power 0.1 milliwatt; modulation
amplitude 1G.
Catalase Activity in M. tuberculosis KatG
7022 JOURNAL OF BIOLOGICAL CHEMISTRY VOLUME 284 NUMBER 11 MARCH 13, 2009
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
environments. The actual g
y
and g
z
values are quite invariant for
known tyrosyl radicals and are in general equal to or larger than
2.0042 and 2.0020, which contrasts with the values here (35,
53–56). The anisotropy of the g matrix for the new radical sig-
nal (g
x
g
z
0.00420 and 0.00364, g
y
g
z
0.00158), how-
ever, is near the range typical of tyrosyl radicals (g
x
g
z
0.007
to 0.004, g
y
g
z
0.002) (57). In addition, the rhombicity of the
g matrix is similar to that expected for a tyrosyl radical, even
though the isotropic g value is slightly below that observed pre-
viously in unmodified tyrosyl radicals. Thus, an electronically
modified tyrosyl-like radical is reasonably predicted by these
results.
Structural Information from the X-band Spectrum—Simula-
tion of the narrow doublet spectrum could be confidently
approached adopting the experimentally determined g values
to obtain additional hyperfine coupling information. Simula-
tions using a single isotropic hyperfine coupling value of 11 G
to achieve the principal splitting produced a symmetrical line
shape unlike that in the X-band spectrum. More satisfactory
simulation was achieved by adding in a second weaker isotropic
hyperfine coupling. Values of 10.5 and 3.2 G gave the most
satisfactory fit to the data (Fig. 6B). Inclusion of additional
hyperfine interactions with a
iso
values exceeding 2.5 G
decreased the quality of fitting to the experimental spectrum.
These observations limit the number and magnitude of inter-
actions that can contribute to the narrow doublet line shape in
addition to the main splitting. Note that the X-band experi-
mental spectrum was recorded at very low power and modula-
tion amplitude to ensure against microwave saturation and line
shape distortions, and this spectrum is assumed to optimally
indicate all the hyperfine features resolvable at this frequency.
The impact on the spectra caused by the two g
x
values evaluated
at D-band is negligible at X-band.
The two hyperfine interactions used in simulating the
X-band spectrum are tentatively assigned to the
-methylene
hydrogens of a tyrosyl-like radical. This assignment is consist-
ent with the high field-EPR results demonstrating the isotropic
nature of the 11 G hyperfine interaction, which is a feature of
such couplings in tyrosyl radicals. Additional hyperfine split-
tings because of relatively strong anisotropic dipolar interac-
tions for the 3 and 5 ring hydrogens would also be expected
in the X-band spectrum if the species were an unmodified (native)
tyrosyl radical, with a
iso
values around 6–7 G, and with the 2
and 6 hydrogens exhibiting weaker couplings of around 3 G
(29, 53, 58). However, these interactions could not be included
in the simulations without severely degrading the quality of the
fits (data not shown). Couplings in the 6 G range are decidedly
absent from the narrow doublet spectrum as they would intro-
duce greater multiplicity in the line. In fact, the upper limit to
additional hyperfine coupling interactions was 2.5 G as stated
above. These observations suggest that couplings to ring hydro-
gens are less than 2.5 G or that the ring hydrogens are absent
from the radical structure.
Further insights can be gained from the isotropic hyperfine
coupling values assigned to a pair of
-methylene hydrogens if
the spin density in the ring of the tyrosyl-like radical could be
estimated. Then a prediction can be made about the couplings
expected for the ring hydrogens. A user-friendly approach
based on the McConnell equation designed for easy analysis of
EPR spectra of tyrosyl radicals in proteins (59), which correlates
hyperfine couplings with C-1 (C-
) spin density and phenol
ring plane orientation, shows that the experimental couplings
of 10.5 and 3.2 G (0.5) would arise for
-methylene hydro-
gens when a small ring orientation angle,
, occurs (Fig. 7) in a
radical in which the spin density on C-1 is 0.2. In this geom-
etry, the weakly coupled hydrogen lies close to the plane of the
ring, and the strongly coupled hydrogen lies close to the p-
singly occupied orbital direction on C-1 of the phenolic ring.
Because there is some error in evaluating the hyperfine cou-
plings by simulation of the X-band spectrum, a small range of
angles is considered possible here. For ring orientations in a
range with 10
30°, the spin density on carbon-1 of the
FIGURE 6. A, echo-detected pseudo-modulated D-band (130 GHz) RFQ-EPR
spectrum of M. tuberculosis KatG frozen 35 ms after mixing resting enzyme
with 4000-fold molar excess of H
2
O
2
at pH 8.5. Pulse widths, 40 and 80 ns;
interpulse delay, 120 ns; temperature, 7 K; repetition rate, 10 Hz; averages per
point, 30; 16 scans, total scan time 73 min. The overlaid simulation (dotted line)
was generated as described in the text using the following parameters. Spe-
cies 1 (59% of total simulation): g
x
2.00606, g
y
2.00344, g
z
2.00186, A
x
11.9 G; A
y
10.3 G; A
z
10.3 G. Species 2 (41% of total simulation) with same
parameters as Species 1 except g
x
2.00550. Note that the smaller (2G)
hyperfine coupling required to fit the X-band simulation is unresolved
because of inhomogeneous broadening in the D-band spectra. B, RFQ
X-band EPR spectrum (solid line) of WT KatG treated with 1000-fold excess
H
2
O
2
at pH 8.5 frozen 20 ms after mixing. Experimental conditions, T 77 K,
microwave frequency 9.442 GHz; microwave power 0.1 milliwatt; mod-
ulation amplitude 1 G; average of nine scans. Dotted line, simulation using
the following parameters: g
1,2,3
2.00606, 2.00344, 2.00186; A
iso
(1) 10.5 G,
A
iso
(2) 3.2 G; line broadening 3.6 G (average.). a.u., arbitrary units.
Catalase Activity in M. tuberculosis KatG
MARCH 13, 2009 VOLUME 284 NUMBER 11 JOURNAL OF BIOLOGICAL CHEMISTRY 7023
by Richard Magliozzo on April 4, 2009 www.jbc.orgDownloaded from
phenolic ring would remain close to 0.2. An alternative geom-
etry in which
is close to 66° also predicts hyperfine cou-
plings near 10.5 and 3.2 G, but in this case the C-1 spin density
must be large (0.5). For the narrow doublet signal, this high
spin density rules out assignment to a native tyrosyl radical
because the EPR spectrum lacks typical hyperfine splittings that
would be expected for 3,5 hydrogens and for 2,6 hydrogens
(29). Furthermore, a spin density greater than 0.4 has not
been observed in any tyrosyl radical reported to date nor do
calculations predict densities significantly greater than that
value (60). Therefore, the analysis points to the conclusion that
the narrow doublet signal arises from a tyrosyl-like radical with
low spin density at C-1 that may lack 3,5 hydrogens, or a
tyrosyl radical with high spin density that cannot contain ring
hydrogens. In either case, the radical giving the narrow doublet
EPR signal must be localized on an electronically and/or struc-
turally modified tyrosine residue according to both HF and
X-band EPR analyses.
An MYW Adduct Radical?—The only structurally unusual
tyrosine residue in KatG is Tyr
229
linked within the distal side
MYW adduct. Beyond this and upon inspection of the M. tuber-
culosis KatG crystal structure, one tyrosine residue that may
also have unusual electronic properties is Tyr
390
. This residue
exhibits a
-cation interaction with Arg
249
(PDB code 2CCA)
and also happens to have a small
angle possibly consistent
with the structural predictions above. The finding that
KatG[Y390F] exhibits the same narrow doublet signal as WT
KatG (supplemental Fig. S2) rules out assignment to this resi-
due. There are no other tyrosines in the KatG structure with
small
angles. These observations and the loss of the narrow
doublet EPR signal in mutants that lack the distal side adduct
(KatG [Y229F], KatG [W107F], and KatG[M255A]) argues very
strongly that this structure is either the site of the radical or is
needed to produce the radical on another modified (electroni-
cally unusual) tyrosine in KatG.
A range of electronic environments for known tyrosyl radi-
cals in proteins is usually attributed to hydrogen bonding
effects at the phenolic oxygen. For example, in cases where
there is relatively good hydrogen bonding, a g
x
value close to
2.006 is observed, whereas in the absence of hydrogen bonding,
this value can be as large as 2.009 (53, 55, 59, 61). Calculations
illustrate that the origin of the increased g
x
value in the absence
of hydrogen bonding is associated with increased unpaired spin
density on oxygen, which coincides with decreased unpaired
spin density on C-1 of the ring (60, 62). The difference in spin
density at C-1, however, is on the order of only 0.03 for the
limiting cases in which an isolated tyrosyl radical (no hydrogen
bond) is compared with one with a strong hydrogen bonding
partner (imidazole) (60). For the narrow doublet here, whereas
the g
x
value could be considered consistent with a good hydro-
gen bond at the phenolic oxygen of a tyrosyl-like radical (for
example, the tyrosyl radical in PSII (63)), the predicted spin
density at C-1 is much smaller than the typical values whether
there is a polarizing interaction at the oxygen or not. Further-
more, the direction for change would be toward increased spin
density in the ring with better hydrogen bonding, rather than an
unusually low spin density suggested by the results above.
Therefore, a native tyrosyl radical is again ruled out as the spe-
cies giving the narrow doublet.
The atypical g values argue for some electronic effect other
than hydrogen bonding that alters the spin density at C-1 in the
tyrosyl-like radical. Knowledge that M. tuberculosis KatG
treated with excess alkyl peroxide forms dimers because of
di-tyrosine cross-links (64)
5
opens the possibility that modified
tyrosines could harbor a radical produced during catalase turn-
over. However, cross-linking of KatG does not occur during
turnover of the enzyme with excess hydrogen peroxide, ruling
out dityrosine as a locus for the narrow doublet radical.
To assist in assignment of the new KatG radical, DFT calcu-
lations were performed that focused on the electronic features
of MYW adduct </