The structure of arylamine N-acetyltransferase from Mycobacterium smegmatis--an enzyme which inactivates the anti-tubercular drug, isoniazid.

Abstract

Arylamine N-acetyltransferases which acetylate and inactivate isoniazid, an anti-tubercular drug, are found in mycobacteria including Mycobacterium smegmatis and Mycobacterium tuberculosis. We have solved the structure of arylamine N-acetyltransferase from M. smegmatis at a resolution of 1.7 A as a model for the highly homologous NAT from M. tuberculosis. The fold closely resembles that of NAT from Salmonella typhimurium, with a common catalytic triad and domain structure that is similar to certain cysteine proteases. The detailed geometry of the catalytic triad is typical of enzymes which use primary alcohols or thiols as activated nucleophiles. Thermal mobility and structural variations identify parts of NAT which might undergo conformational changes during catalysis. Sequence conservation among eubacterial NATs is restricted to structural residues of the protein core, as well as the active site and a hinge that connects the first two domains of the NAT structure. The structure of M. smegmatis NAT provides a template for modelling the structure of the M. tuberculosis enzyme and for structure-based ligand design as an approach to designing anti-TB drugs.

Full text

Page 1

The Structure of Arylamine N-acetyltransferase from
Mycobacterium smegmatis—An Enzyme which
Inactivates the Anti-tubercular Drug, Isoniazid
James Sandy1, Adeel Mushtaq1, Akane Kawamura1, John Sinclair1
Edith Sim1and Martin Noble2*
1Department of Pharmacology
University of Oxford
Mansfield Road
Oxford OX1 3QT, UK
2Laboratory of Molecular
Biophysics, South Parks Road
Oxford OX1 3QU, UK
Arylamine N-acetyltransferases which acetylate and inactivate isoniazid,
an anti-tubercular drug, are found in mycobacteria including Mycobacterium
smegmatis and Mycobacterium tuberculosis. We have solved the structure of
arylamine N-acetyltransferase from M. smegmatis at a resolution of 1.7 A˚
as a model for the highly homologous NAT from M. tuberculosis. The
fold closely resembles that of NAT from Salmonella typhimurium, with a
common catalytic triad and domain structure that is similar to certain
cysteine proteases. The detailed geometry of the catalytic triad is typical
of enzymes which use primary alcohols or thiols as activated nucleo-
philes. Thermal mobility and structural variations identify parts of NAT
which might undergo conformational changes during catalysis. Sequence
conservation among eubacterial NATs is restricted to structural residues
of the protein core, as well as the active site and a hinge that connects
the first two domains of the NAT structure. The structure of M. smegmatis
NAT provides a template for modelling the structure of the M. tuberculosis
enzyme and for structure-based ligand design as an approach to
designing anti-TB drugs.
q 2002 Published by Elsevier Science Ltd
Keywords: mycobacteria; TB; catalytic triad; X-ray crystallography;
arylamine N-acetyltransferase
*Corresponding author
Introduction
Arylamine N-acetyltransferase activity was first
identified as the enzymic activity in humans
responsible for the inactivation of isoniazid, the
major anti-tubercular agent.1Arylamine N-acetyl-
transferases are now known to constitute a major
family of enzymes which acetylate a range of aryl-
amine, arylhydroxylamines and arylhydrazines
using acetyl-CoA as the acetyl donor.2,3Other
members of the family, including the terminal
protein in the cluster of enzymes responsible for
rifamycin synthesis,4are likely to be involved in
amide bond formation using a different acyl
donor. It is therefore possible that individual
members of the NAT family provide a range of
activities in the different organisms in which they
are found. Mycobacteria are distinguished by dis-
tinctive metabolism resulting in the synthesis of
mycolic acids, long chain fatty acids that form a
layer of the tough mycobacterial cell wall. A
further distinctive feature of mycobacteria is their
unique sensitivity to isoniazid.5Isoniazid inhibits
mycolic acid synthesis. However, isoniazid is a
pro-drug and must be activated by oxidation. The
gene product of katG, which has both catalase and
peroxidase activity associated with it, activates
isoniazid.6,7The activated drug is thought to exert
its effect, at least partially, through inhibition of an
enoyl-acyl carrier protein (ACP) reductase (InhA).8
This is an enzyme of the multicomponent fatty
acid synthase type II (FAS II) complex, involved in
mycolic acid synthesis. Another component of
FAS II, a b-ketoacyl ACP synthase (KasA) has
also been identified as a target.9If isoniazid is
acetylated by NAT, the product acetylisoniazid
cannot be oxidized to its active form by katG. The
gene for NAT is present in the mycobacterial
speciesMycobacterium
bacilli Calmette-Gue ´rin (BCG), Mycobacterium avium
and Mycobacterium smegmatis.11When the gene
forM. tuberculosisNAT
M. smegmatis the result is that resistance to isoniazid
tuberculosis,10
M.bovis
isover-expressedin
0022-2836/02/$ - see front matter q 2002 Published by Elsevier Science Ltd
E-mail address of the corresponding author:
martin@biop.ox.ac.uk
Abbreviations used: NAT, N-acetyltransferase;
TB, tuberculosis.
doi: 10.1016/S0022-2836(02)00141-9 available online at http://www.idealibrary.com on
B
w
J. Mol. Biol. (2002) 318, 1071–1083

Page 2

is increased,12and, as would be expected, knocking
out the nat gene increases the sensitivity of
M. smegmatis to isoniazid.13It is unclear whether
NAT from mycobacteria also has an endogenous
role, although knocking out the nat gene from
M. smegmatis increases the duration of the lag
phase of bacterial cell growth.13Therefore, on the
basis of its participation in isoniazid metabolism,
and the effect of knocking out the nat gene in
M. smegmatis, NAT from M. tuberculosis is a strong
candidate as a target for anti-tubercular therapy.
In order to apply an approach of rational drug
design, the structure of NAT from a mycobac-
terium is required. The crystal structure of NAT
from Salmonella typhimurium has been obtained at
2.8 A˚resolution.14This was the first member of
the NAT family for which a crystal structure has
been determined. NAT from S. typhimurium is
only 32% identical to NAT from M. tuberculosis at
the amino acid level. Therefore, it has been impor-
tant to determine the structure of a mycobacterial
NAT. The NAT from M. smegmatis is 60% iden-
ticalto theNATfrom
both enzymes are able to acetylate isoniazid.15
M. smegmatis NAT, in contrast to M. tuberculosis
NAT, when expressed as a recombinant protein in
M. tuberculosis,and
Escherichia coli, is highly soluble, and crystallises
readily. We have therefore been able to determine
the crystal structure of M. smegmatis NAT at a
resolution of 1.7 A˚.
Results
Purification and characterisation
The NAT from M. smegmatis has been generated
as a recombinant soluble protein with an N-termi-
nal hexahistidine tag in E. coli. The protein has
been purified to homogeneity on a nickel affinity
resin and the tag has been removed with thrombin.
Thrombin cleavage leaves three additional non-
authentic residues (a glutamic acid, a serine and a
histidine) at the amino terminus. The pure protein
is active and catalyses the acetylation of isoniazid
(Km, 87((^18) mM; Vmax, 115(^33) nmol min21mg21
protein) as well as the arylamines 4-aminoveratrole
(Km 1.42(^0.12) mM; Vmax 6.5(^0.6) mmol min21
mg21protein) and 4-anisidine (Km, 245(^35) mM;
Vmax,114(^28) nmol min21mg21
Km value for isoniazid is very similar to that
reported for recombinant NAT from M. tuberculosis,
protein).The
Figure 1. Sequence alignment of prokaryotic NATs. Conserved active site residues of Cys, His and Asp are high-
lighted in yellow. Other conserved residues are highlighted in black, residues with conservative substitution are
shown highlighted in grey. Domains I (1–88), II (89–188) and III (188–280) are indicated as filled bars above the
sequence alignment (coloured black, red and blue, respectively). Residues identified as being likely to be involved in
substrate recognition are indicated in green. Amino acids are represented by the single letter code. Numbers refer to
amino acid positions of the aligned sequences with reference to the sequence from M. smegmatis. Sequences were
aligned using the ClustalW service (www.ebi.ac.uk/ClustalW).
1072
Structure of Mycobacterial NATs

Page 3

heterologously expressed in M. smegmatis (Km,
100 mM15). The values are also similar to those
described for recombinant M. smegmatis NAT in
E. coli cell lysates12although the value for iso-
niazid, whilst of the same order of magnitude, is
higher than that reported previously (25 mM), and
may be explained by the different assay methods
employed. The Km values for anisidine and
4-aminoveratrole are of the same order of magni-
tude as the values reported for S. typhimurium
NAT witha C-terminal
although the relative values for the two arylamines
are reversed: anisidine (1.1 mM) and 4-amino-
veratrole (0.6 mM).
hexahistidinetag,16
Comparison with S. typhimurium NAT
Molecular replacement was used to determine
the structure of M. smegmatis NAT in all of the
crystal forms described in Table 1. The best crystals
diffracted to a resolution of 1.7 A˚. In all crystal
forms, the structure was found to be very similar
throughout, with the exception of the confor-
mationally mobile N terminus and some side-
chain conformations. The analysis of the structure
presented here derives from the 1.7 A˚tetragonal
crystal form, but applies equally to all available
crystal forms. Structure building and refinement
was guided in part by sequence alignments
of NAT from M. smegmatis with NATs from
M. tuberculosis, S. typhimurium and E. coli. The
NAT sequences from S. typhimurium and E. coli are
74% identical to each other, while the NAT
sequences from M. smegmatis and M. tuberculosis
are 60% identical to each other (Figure 1).
Theoverall sequence
S. typhimurium and M. smegmatis NAT is 34%, with
52% identity in the first two domains.11The overall
fold of the two molecules is very similar. The
root-mean-squared coordinate difference (rmsd)
betweenthetwostructures
255 equivalent residues used in the structural
alignment. As well as genuine structural dif-
ferences due to insertions and deletions in the
identity between
is1.5 A˚over
M. smegmatis sequence and changes in the relative
disposition of NAT domains (discussed below),
this high value may reflect the relatively low resol-
ution of the S. typhimurium structure.14
M. smegmatis NAT is 275 amino acid residues
long, compared with 281 of S. typhimurium. Apart
from a deletion of eight amino acid residues at the
C terminus of M. smegmatis NAT, the alignment of
M. smegmatis and S. typhimurium NATs shows the
mycobacterial enzyme has insertions after residue
47 (P), after residue 99 (DD) and after residue 115
(GA). There are three single amino acid deletions
in M. smegmatis NAT between residues 195 and
256 with respect to S. typhimurium NAT. Number-
ing corresponds to the sequence of S. typhimurium
NAT.All oftheseinsertions
are remote from the catalytic site and so are
unlikely toinfluence
enzyme activity.
and deletions
substratebindingand
Dimerisation
Inall crystalforms observedforboth
S. typhimurium and M. smegmatis NAT (see Table
1), a similar association of monomeric NAT
molecules to form dimers is observed (Figure 2).
In some cases, this association occurs across
crystallographic 2-fold axes, while in others the
dyad symmetry isnon-crystallographic.
dimeric association involves residues of the edge
strand of domain 3 (strand b14 according to the
nomenclature of Sinclair et al.14). Despite sharing a
common mode of association, the dimeric species
do not perfectly superimpose, due to a different
twist of the b-sheets which form the dimer inter-
face. The catalytic cysteine residues of the dimeric
related molecules are separated by 55 A˚, and so
direct cross-talk between active sites by this route
would not be expected.
In solution, ultracentrifugation and gel filtration
analyses demonstratethat
S. typhimurium NAT are formed on storage of the
purified recombinant protein for two weeks, and
that higher-order active oligomers are formed
The
active dimers of
Figure 2. Dimerisation observed
in crystals of prokaryotic NAT. Two
representative monomers from the
asymmetric unit of the 1.7 A˚reso-
lutionstructure
NAT are shown, to indicate their
mode of dimerisation. Molecules
are drawn in a cartoon represen-
tation and are colour ramped from
N to C terminus starting with blue,
through yellow, to red. Terminii of
the two molecules are labelled.
ofM. smegmatis
Structure of Mycobacterial NATs
1073

Page 4

Table 1. Statistics of datasets collected and structure refinement
Crystallization conditionsa
csII, 19 csI, 14mdI, 24csI, 24
Space group
Unit cell (A˚, 8)
P212121
P41212I4122P6222
a ¼ 101:3;b ¼ 105:8;c ¼ 141:3;
a ¼ b ¼ g ¼ 90:0
a ¼ b ¼ 100:7;c ¼ 130:4;
a ¼ b ¼ g ¼ 90:0
a ¼ b ¼ 121:5;c ¼ 191:7;
a ¼ b ¼ g ¼ 90:0
a ¼ b ¼ 144:8;c ¼ 127:8;
a ¼ b ¼ 90:0;g ¼ 120:0
Data statistics (highest resolution shell)
Resolution (A˚)
Number of observations
Unique reflections
Completeness (%)
I/s(I)
Rmergeb
1.7 (1.76–1.70)
624,263 (34,821)
163,227 (14,137)
98.1 (87.4)
3.1 (1.3)
0.074 (0.568)
1.9 (2.07–1.90)
283,364 (24,174)
57,212 (5402)
100.0 (99.3)
8.3 (1.5)
0.063 (0.471)
2.3 (2.39–2.30)
282,264 (32,406)
32,406 (2749)
91.2 (83.5)
10.6 (1.9)
0.049 (0.431)
2.4 (2.48–2.40)
101,254 (6642)
30,456 (2767)
97.5 (91.7)
8.5 (1.8)
0.096 (0.592)
Refinement
Monomers in asymmetric unit
Rconvc(Rfreed)
rms deviation from bond ideality (A˚)
rms deviation from angle ideality (8)
Number of water molecules added
Mean main-chain temperature factor (A˚2)
4211
0.209 (0.241)
0.022
1.967
755
38.1
aAll conditions were discovered from commercially available crystal screens. csI and csII denote the crystal screens from Hampton Research Inc., while mdI denotes crystal screen I from Mol-
ecular Dimensions Ltd.
bRmerge¼P
cRconv¼P
dRfreeis equivalent to Rconvfor a 5% subset of reflections not used in the refinement.
h
P
jllIh;jl 2 lkIhlll=P
hllFohl 2 lFchll=P
h
P
jlIh;jl; where Ih,jis the jth observation of reflection h.
hlFohl; where Fohand Fchare the observed and calculated structure factor amplitudes, respectively, for the reflection h.

Page 5

upon longer storage. The purified M. smegmatis
NAT protein also forms oligomers which are active
(Table 2). These oligomers, which migrate with an s
value around 5.3 S, are likely to represent compact,
non-spherical tetramers. This is supported by gel
filtration studies in which the Stokes’ radius of the
large oligomer of S. typhimurium NAT has been
determined as 4.8 nm, corresponding to a globular
protein with a molecular mass of 165 kDa, con-
sistent with a non-spherical tetrameric assembly.
Neither S. typhimurium nor M. smegmatis NAT
oligomers are disulphide linked, since a band
corresponding to an apparent molecular mass
of 31 kDa is observed on SDS-PAGE without
reduction of the protein.
In solution, mammalian NATs do not appear
toformdimersas
centrifugation17,18
and gel filtration studies.17,19
However, none of these studies has been carried
out with recombinant protein. There has been a
report of dimer formation using partially purified
pigeon NAT in the presence of acetyl-CoA.20
Analysis of sequence conservation (Figure 1)
shows that strand b14 is not conserved in sequence
across prokaryotic NATS.
determined byultra-
Rotations of the “lid domain”
A substantial part of the structural difference
between NAT structures from the different species
results from a different relative orientation of the
third domain with respect to the first two highly
conserved domains. Optimal superposition of
M. smegmatis NAT on S. typhimurium NAT over
the whole length of the two structures results in
an rms difference of 1.6 A˚for 264 equivalent Ca
positions, whereas superposition of the core of the
first two domains (residues 10–170 of M. smegmatis
NAT) results in an rms difference of 1.2 A˚for 156
Capositions. Figure 3 shows the relative position
of the third domains after superposition of the
structures on the basis of the other two domains.
The transformationrelating
domains of the two species of NAT corresponds
approximately to a rotation of 78, hinged by the
short helix at residues H206-T209.
theC-terminal
Temperature factors/mobility
The high resolution of the M. smegmatis NAT
structure has allowed accurate determination of
Table 2. Oligomerisation of pure recombinant procaryotic NATs determined by density gradient ultracentrifugation
ProteinDays storagea
s20,w(S)b
Percentage total activityc
Oligomeric state
(A) S. typhimurium NAT2
17
17
17
30
30
30
40
40
40
3.23 ^ 0.18
3.55 ^ 0.11
4.41 ^ 0.11
5.32 ^ 0.11
3.53 ^ 0.11
4.38 ^ 0.11
5.32 ^ 0.11
3.53 ^ 0.11
4.41 ^ 0.11
5.32 ^ 0.11
.95
12 ^ 3
80 ^ 5
8 ^ 2
8 ^ 5
55 ^ 14
37 ^ 15
,5
60 ^ 12
35 ^ 15
Monomer
Monomer
Dimer
Tetramer
Monomer
Dimer
Tetramer
Monomer
Dimer
Tetramer
(B) M. smegmatis NAT40
40
40
3.53 ^ 0.11
4.38 ^ 0.11
5.32 ^ 0.11
,5
48 ^ 14
50 ^ 10
Monomer
Dimer
Tetramer
aPure recombinant protein was stored at 4 8C prior to ultracentrifugation and the total activity was not substantially diminished
over the storage period, as described.16
bThe sedimentation coefficient is expressed as the average ^1 SD of at least three independent determinations. Activity was deter-
mined with 100 mM anisidine as substrate in each fraction following ultracentrifugation.
cThe activity is expressed as a percentage of the total recovered which was greater than 90% of the activity applied to the gradient.
Figure
comparison between S. typhimurium
andM. smegmatisis
tures of S. typhimurium (red) and
M. smegmatis(green)
superimposed on the basis of their
first twodomains.
mational difference
approximately to a rotation of the
third domains around an axis as
indicated by the arrow. Domains 1
(helical bundle), 2 (b-barrel), and 3
(mixed a/b) are labelled D1–D3.
3.Domainorientation
NAT.Struc-
NATare
The
corresponds
confor-
Structure of Mycobacterial NATs
1075