7,8-Diaminoperlargonic acid aminotransferase from
Mycobacterium tuberculosis, a potential therapeutic
target
Characterization and inhibition studies
Ste
´phane Mann and Olivier Ploux
Synthe
`se Structure et Fonction de Mole
´cules Bioactives, Universite
´Pierre et Marie Curie-Paris 6, UMR 7613, Paris, France
Tuberculosis remains one of the major infectious dis-
eases in the world, with a newly infected human every
second, one-third of the total population already infec-
ted, and 2 million deaths per year, according to the
WHO [1]. Furthermore, strains of Mycobacterium
tuberculosis, the pathogen, that are resistant to one or
several antibiotics used in therapy have been identified
and might thus compromise efforts to eradicate the
disease. New therapeutic targets and drugs, as well as
new vaccines and public health efforts are thus
urgently needed to decrease the incidence of tuberculo-
sis worldwide.
Keywords
7,8-diaminopelargonic acid
aminotransferase; amiclenomycin; biotin
biosynthesis; Mycobacterium tuberculosis;
S-adenosyl-L-methionine
Correspondence
O. Ploux, Synthe
`se Structure et Fonction de
Mole
´cules Bioactives, UMR7613 CNRS-
UPMC, Universite
´Pierre et Marie Curie,
boı
ˆte 182, 4, place Jussieu, F-75252 Paris
cedex 05, France
Fax: +33 1 44 27 71 50
Tel: +33 1 44 27 55 11
E-mail: olivier.ploux@upmc.fr
URL: http://www.upmc.fr/umr7613/
(Received 1 June 2006, revised 13 July
2006, accepted 23 August 2006)
doi:10.1111/j.1742-4658.2006.05479.x
Diaminopelargonic acid aminotransferase (DAPA AT), which is involved
in biotin biosynthesis, catalyzes the transamination of 8-amino-7-oxonona-
noic acid (KAPA) using S-adenosyl-l-methionine (AdoMet) as amino
donor. Mycobacterium tuberculosis DAPA AT, a potential therapeutic tar-
get, has been overproduced in Escherichia coli and purified to homogeneity
using a single efficient step on a nickel-affinity column. The enzyme shows
an electronic absorption spectrum typical of pyridoxal 5¢-phosphate-
dependent enzymes and behaves as a homotetramer in solution. The pH
profile of the activity at saturation shows a single ionization group with a
pK
a
of 8.0, which was attributed to the active-site lysine residue. The
enzyme shows a Ping Pong Bi Bi kinetic mechanism with strong substrate
inhibition with the following parameters: K
mAdoMet
¼0.78 ± 0.20 mm,
K
mKAPA
¼3.8 ± 1.0 lm,k
cat
¼1.0 ± 0.2 min
)1
,K
iKAPA
¼14 ± 2 lm.
Amiclenomycin and a new analogue, 4-(4c-aminocyclohexa-2,5-dien-1r-
yl)propanol (referred to as compound 1), were shown to be suicide sub-
strates of this enzyme, with the following inactivation parameters: K
i
¼
12±2lm,k
inact
¼0.35 ± 0.05 min
)1
, and K
i
¼20±2lm,k
inact
¼
0.56 ± 0.05 min
)1
, for amiclenomycin and compound 1, respectively. The
inactivation was irreversible, and the partition ratios were 1.0 and 1.1 for
amiclenomycin and compound 1, respectively, which make these inactiva-
tors particularly efficient. compound 1(100 lgÆmL
)1
) completely inhibited
the growth of an E. coli C268bioA mutant strain transformed with a
plasmid expressing the M. tuberculosis bioA gene, coding for DAPA AT.
Reversal of the antibiotic effect was observed on the addition of biotin or
DAPA. Thus, compound 1specifically targets DAPA AT in vivo.
Abbreviations
AdoMet, S-adenosyl-L-methionine; DAPA, 7,8-diaminopelargonic acid (7,8-diaminononanoic acid); DAPA AT, 7,8-diaminopelargonic acid
aminotranferase; KAPA, 8-amino-7-oxononanoic acid; PLP, pyridoxal 5¢-phosphate.
4778 FEBS Journal 273 (2006) 4778–4789 ª2006 The Authors Journal compilation ª2006 FEBS
The biosynthesis of biotin (vitamin H), a cofactor
for carboxylases, decarboxylases and transcarboxylas-
es, has been identified as an interesting target for anti-
biotics and herbicides. Indeed, this metabolic pathway
is specific to micro-organisms and higher plants [2].
Two antibiotics isolated from Streptomyces species,
actithiazic acid [3] and amiclenomycin [4–8], have been
found to be active against mycobacteria, and target
enzymes of the biotin biosynthesis pathway. Further-
more, bioA, the gene coding for 7,8-diaminopelargonic
acid aminotransferase (DAPA AT; EC 2.6.1.62), which
is involved in biotin biosynthesis, has been implicated
in long-term survival of mycobacteria [9]. It thus seems
that biotin biosynthesis, and in particular the transami-
nation step catalyzed by DAPA AT, are valid targets
for antibiotic directed against mycobacteria. Obvi-
ously, mycobacteria could reverse the effect of such
antibiotics by taking up external biotin. However, such
a transporter remains elusive in the annotated genes of
M. tuberculosis [10,11], and reversal of the amicleno-
mycin antibacterial effect is observed at biotin concen-
trations above 0.01 lgÆmL
)1
[4], a concentration at
least 10 times higher than that found in human plasma
[12]. Interestingly, the recently described bioA mutant
of Mycobacterium smegmatis survived poorly in rich
medium, suggesting that the observed phenotype was
not reversed by the presence of external biotin [9].
DAPA AT is a pyridoxal 5¢-phosphate (PLP)
enzyme that catalyzes the transamination of 8-amino-
7-oxononanoic acid (KAPA) to yield 7,8-diaminonona-
noic acid (DAPA) [13,14] (Fig. 1). In Escherichia coli,
the amino donor in this reaction is S-adenosyl-l-
methionine (AdoMet) [15]. The enzyme from E. coli
has been well characterized [13–17], and its 3D struc-
ture determined [18,19]. We have reported the total
synthesis of natural amiclenomycin [20] and some of
its analogues [21] and have deciphered the mode of
action of this antibiotic at the molecular level [22–25].
It irreversibly inactivates E. coli DAPA AT by forming
an aromatic adduct with the bound PLP. Interestingly,
modification of the structure of amiclenomycin gave
some active compounds, encouraging the design of
new inhibitors that might be useful in antibiotic devel-
opment [25].
In an effort to contribute to the discovery of new
therapeutic targets in M. tuberculosis, it is our inten-
tion to fully characterize M. tuberculosis DAPA AT
and screen likely molecules for their inhibiting proper-
ties. We report here the cloning and heterologous
expression of the M. tuberculosis bioA gene. M. tuber-
culosis DAPA AT was purified to homogeneity and
characterized. We also provide evidence that amicleno-
mycin and a new analogue irreversibly inactivate
M. tuberculosis DAPA AT.
Results and Discussion
Cloning, expression and purification of
M. tuberculosis DAPA AT
We used PCR-based technology to construct two
M. tuberculosis bioA genes which were cloned into a
pUC18 vector, downstream of the lac promoter. The
first construct, pUC18-MTbioA, contained a ribosome-
binding site consensus sequence 7 bp ahead of the first
ATG codon, while the second, pUC18-MTHis
6
bioA,
contained the same ribosome-binding site and a
sequence coding for His
6
inserted between the first and
second codon of the bioA gene. The first construct
would therefore produce a M. tuberculosis DAPA AT
with the wild-type sequence (referred to as wild-type
DAPA AT in this work), and the latter would give an
N-terminal His
6
-tagged DAPA AT, for convenient
purification. The sequence of the recombinant genes
was verified by DNA sequencing. The functionality of
the recombinant enzymes was demonstrated in vivo by
transforming E. coli C268 bioA
–
cells with both con-
structs. Transformed cells were able to grow on a
COOH
NH2
O
KAPA
NH2
H2NCOOH
NH2
OH
DAPA
COOH
NH2
NH2
DAPA
aminotransferase
AdoMet
S-Adenosyl-
(2-oxo-4-thiobutyrate)
Biotin
Amiclenomycin Compound 1
Fig. 1. The reaction catalyzed by DAPA AT
and the chemical structure of amicleno-
mycin and compound 1.
S. Mann and O. Ploux M. tuberculosis DAPA aminotransferase
FEBS Journal 273 (2006) 4778–4789 ª2006 The Authors Journal compilation ª2006 FEBS 4779
biotin-free Luria–Bertani agar plate (containing
0.45 UÆmL
)1
avidin), thus reversing the bio
–
phenotype
by complementation, which proved that the heterolo-
gous expression was efficient and that both recombin-
ant M. tuberculosis DAPA ATs were functional.
The production of soluble His
6
-tagged M. tuberculo-
sis DAPA AT was low in E. coli JM105 ⁄pUC18-
MTHis
6
bioA: the crude extract had a specific activity
of 0.04 mUÆmg
)1
. Induction by isopropyl b-d-thiogal-
actopyranoside (0.1–0.5 mm) in this lacI
q
strain moder-
ately increased the production of soluble enzyme by a
factor of 2. As we noted the presence in the M. tuber-
culosis bioA sequence of codons rarely used in E. coli
(eight CCC Pro, one AGG Arg and three CGA Arg
codons), we attempted to produce the enzyme in
E. coli Rosetta(DE3) ⁄pLysS or E. coli BL21 Codon-
Plus(DE3)RP. In these hosts, the expression was con-
stitutive, as the lac repressor is not overproduced,
and the specific activity of the soluble extract was
0.10 mUÆmg
)1
in E. coli Rosetta(DE3) ⁄pLysS ⁄pUC18-
MTHis
6
bioA and 0.12 mUÆmg
)1
in E. coli BL21
CodonPlus(DE3)RP ⁄pUC18-MTHis
6
bioA. This slightly
increased production compared with that in E. coli
JM105 ⁄pUC18-MTHis
6
bioA was attributed to the
overproduction of the rare tRNAs. However, when
produced in E. coli BL21 CodonPlus(DE3)RP ⁄pUC18-
MTHis
6
bioA, DAPA AT was predominantly in an
insoluble form. Unfortunately, attempts to solubilize
the precipitated proteins in 8 murea and renature the
DAPA AT were unsuccessful. The His
6
-tagged DAPA
AT was thus purified from the soluble crude extract of
E. coli BL21 CodonPlus(DE3)RP ⁄pUC18-MTHis
6
bioA
using a single purification step, nickel affinity chroma-
tography. Homogeneous enzyme was thus obtained,
as judged by SDS ⁄PAGE analysis (Fig. 2). Three
milligrams of pure protein with a specific activity of
8.8 ± 0.3 mUÆmg
)1
, was obtained from 1 L of culture,
making this purification scheme quite efficient, with a
73-fold purification. Concentration of the protein solu-
tion was achieved by ammonium sulfate precipitation
followed by solubilization and dialysis rather than by
ultrafiltration which caused precipitation. The pure
enzyme was kept at )80 C without significant loss of
activity.
Wild-type M. tuberculosis DAPA AT was similarly
produced in E. coli BL21 CodonPlus(DE3)RP ⁄pUC18-
MTbioA. However, purification of the enzyme from
the soluble fraction required a two-step purification
protocol using Q-Sepharose and Mono Q columns.
The specific activity of the pure enzyme was 9.4 ±
0.3 mUÆmg
)1
, a value very similar to that for the His
6
-
tagged enzyme, which shows that the six N-terminal
histidine residues of His
6
-tagged DAPA AT do not
perturb the catalytic activity.
Biochemical characterization
As shown in Fig. 2, wild-type and His
6
-tagged
M. tuberculosis DAPA AT showed a single band when
separated by SDS ⁄PAGE, with an approximate
molecular mass of 45 kDa, in agreement with the bioA
DNA sequence. The two enzymes were separately
chromatographed on a calibrated Superdex HR S200
column, in native conditions, at pH 8.0. Both recom-
binant proteins were eluted as a single species with an
estimated molecular mass of 189 kDa. Therefore,
M. tuberculosis DAPA AT behaved as a homotetramer
in solution.
The electronic absorption spectrum of pure His
6
-
tagged M. tuberculosis DAPA AT, at pH 8.0, exhibited
characteristic bands at 332 nm and 414 nm, typical of
the internal aldimine of PLP-dependent enzymes [26],
which we attributed to the internal aldimine between
the bound PLP and the enzyme (Fig. 3). The absorb-
ance ratio, A
414
⁄A
280
, was 0.219 for the pure enzyme.
The specific activity of His
6
-tagged M. tuberculosis
DAPA AT was measured at different pH values, from
6.8 to 9.1, in the presence of 20 lmKAPA and 1 mm
AdoMet. Figure 4 shows the data on a log-log plot
together with the pH profile for E. coli DAPA AT,
measured in the same conditions, for comparison. The
data were fitted to Eqn (1) assuming one ionisable
group on the enzyme (pK
a1
):
a¼amax=ð1þ10pKa1 pHÞð1Þ
As Fig. 4 shows, the maximum specific activity for the
M. tuberculosis enzyme is 10 times lower than that
measured for the E. coli enzyme. The pK
a
values
Fig. 2. Analysis by SDS ⁄PAGE of the purification of His
6
-tagged
M. tuberculosis DAPA AT on a Ni-affinity column. Lane 1, crude
extract; lane 2, unretained fraction; lane 3, molecular mass stand-
ards (from top to bottom: 66 kDa, 45 kDa, 36 kDa, 29 kDa,
24 kDa); lanes 4–9, fractions eluted with 400 mMimidazole; lane
10, purified wild-type M. tuberculosis DAPA AT; lane 11, purified
His
6
-tagged M. tuberculosis DAPA AT; lane 12, molecular mass
standards (66 kDa, 45 kDa, 36 kDa, 29 kDa, 24 kDa, 20 kDa).
M. tuberculosis DAPA aminotransferase S. Mann and O. Ploux
4780 FEBS Journal 273 (2006) 4778–4789 ª2006 The Authors Journal compilation ª2006 FEBS
obtained were 7.6 and 8.0 for E. coli and M. tuberculo-
sis DAPA AT, respectively. It should be noted that the
data points between pH 6.5 and pH 7.1 for M. tuber-
culosis DAPA AT do not fit well to the simple ioniza-
tion model described by Eqn (1). Further pH studies
are necessary to clarify this. Indeed, interpretation of
simple pH effects on activity are not straightforward
[27], but because the substrates were almost at satur-
ating concentrations, one can reasonably attribute the
ionization observed to the active-site base that cata-
lyses the proton transfer. In E. coli DAPA AT, the
active-site base has been proposed to be Lys274 on the
basis of structural data. This lysine residue is con-
served in all DAPA AT sequences known so far [19].
There is no doubt that the corresponding lysine in the
M. tuberculosis enzyme, Lys283, plays the same role.
Several potential amino donors were tested at high
concentration (5 mm) on our enzyme: l-Asp, l-Glu,
l-Met, l-Lys, and d,l-homocysteine. None of them
was a substrate for the transamination reaction, i.e.,
no activity was detected when AdoMet was replaced
by these amines. Consequently, AdoMet was consid-
ered to be the natural amino donor and used for
further kinetic studies. Of all the DAPA ATs
characterized [13,14,28], the enzyme from Bacillus
subtilis is the only one that does not use AdoMet as
the amino donor. It uses lysine as the amino donor
instead [29]. Thus, there might be two different classes
of DAPA AT that differ with regard to the second
substrate.
Determination of the kinetic parameters of the
His
6
-tagged DAPA AT
The double-reciprocal plot of initial velocities against
KAPA concentration for several concentrations of the
second substrate, AdoMet, is typical of a Ping Pong
Bi Bi mechanism, with strong substrate inhibition by
KAPA (Fig. S1) [30]. The plot is very similar to those
already published by Stoner & Eisenberg [14] and us
[24], for the E. coli enzyme, i.e., at low KAPA concen-
tration the lines appear parallel, whereas at higher
KAPA concentration they bend up as they approach the
ordinate axis. Such plots are problematic for determin-
ing the four kinetic parameters, i.e., K
iKAPA
,K
mKAPA
,
K
mAdoMet
, and V
m
. We thus used a different strategy
to obtain an estimation of the kinetic parameters (a
full description is available in the Supplementary
material). When KAPA concentrations above 10 lm
were used, this substrate appeared as a simple compet-
itive inhibitor of the reaction, as shown in the Hanes–
Woolf plot of the data (Fig. 5A). The parallel lines are
characteristic of competitive inhibition by KAPA, i.e.,
KAPA forms a dead end complex with the enzyme–
PLP form, in competition with AdoMet. Replotting
the apparent K
m
⁄V
m
as a function of KAPA con-
centration gave: K
iKAPA
¼14 ± 2 lm,K
mAdoMet
¼
0.1
1.0
10.0
100.0
6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0
Specific activity (mU.mg-1)
pH
Fig. 4. Activity versus pH profile for the His
6
-tagged M. tuberculo-
sis DAPA AT and for E. coli DAPA AT. The specific activities of
both enzymes were determined in the presence of saturating con-
centrations of substrates at various pH values. See Experimental
procedures for details. The specific activity was plotted against the
pH on a log-log plot. The data points were fitted to Eqn
(1). hE. coli DAPA AT; dHis
6
-tagged M. tuberculosis DAPA AT.
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
300 350 400 450 500 550
Absorbance
Wavelength (nm)
Fig. 3. UV-visible spectrum of His
6
-tagged M. tuberculosis DAPA
AT. The absorption spectrum of purified His
6
-tagged M. tuber-
culosis DAPA AT (0.44 mgÆmL
)1
)in50mMTris ⁄HCl buffer
(pH 8.0) ⁄10 mM2-mercaptoethanol was recorded against a blank
containing the same buffer.
S. Mann and O. Ploux M. tuberculosis DAPA aminotransferase
FEBS Journal 273 (2006) 4778–4789 ª2006 The Authors Journal compilation ª2006 FEBS 4781
0.58 ± 0.1 mmand V
m
¼22 ± 6 mUÆmg
)1
(Fig. 5B).
Thus the catalytic constant is k
cat
¼1.0 ± 0.2 min
)1
.
To estimate K
mKAPA
, we used the constant ratio
method [14,30]. Initial velocities were measured using a
constant molar ratio of the two substrates, AdoMet
and KAPA. The double-reciprocal plot in these condi-
tions gave straight lines, a characteristic of the Ping
Pong mechanism, but in our case they did not intersect
at the same point on the ordinate axis, because inhibi-
tion by KAPA was not negligible (Fig. S2). The secon-
dary plots (Fig. S3 and Fig. S4) allowed the estimation
of K
mAdoMet
¼0.96 ± 0.1 mmand V
m
¼21 ± 7
mUÆmg
)1
and K
mKAPA
¼3.8 ± 1.0 lm. Because two
different values for K
mAdoMet
were obtained by our
analyses, the mean of these values (0.78 ± 0.20 mm)
was considered to be the best estimate.
Comparison of the kinetic parameters of the E. coli
and M. tuberculosis enzymes shows that the K
m
values
for the latter are 3–4 times higher, and that the k
cat
for
the M. tuberculosis enzyme is eight times lower than
that of the E. coli enzyme. Furthermore, the inhibition
constant, K
iKAPA
, for the M. tuberculosis enzyme is
half that measured for the E. coli enzyme. Overall,
M. tuberculosis DAPA AT is much less efficient than
the E. coli enzyme. This result is quite surprising as
the two enzymes share strong sequence identity (50%)
and all the active-site residues are conserved. Deter-
mination of the 3D structure of M. tuberculosis DAPA
AT will certainly shed light on this issue.
Inactivation and titration of His
6
-tagged DAPA AT
by amiclenomycin and 4-(4c-aminocyclohexa-
2,5-dien-1r-yl)propanol (compound 1)
When the His
6
-tagged M. tuberculosis DAPA AT was
preincubated, at pH 8.0, in the presence of amicleno-
mycin or compound 1at various concentrations,
inactivation occurred. The remaining activity was
measured under standard conditions. In these condi-
tions, the inhibitor was diluted in the assay mixture
(30-fold dilution), thus stopping the inactivation pro-
cess. Figure 6A shows the remaining activity against
time on a semi-log plot for the inactivation by ami-
clenomycin. Because M. tuberculosis DAPA AT is a
rather slow enzyme, its concentration was sometimes
comparable to the inactivator concentration in these
experiments. Nevertheless, the data fitted well to a
pseudo-first-order kinetic process, and the observed
inactivation rates, k
obs
, varied hyperbolically with the
inactivator concentration. Thus, the simple two-step
model for irreversible inactivation may apply, and
the following kinetic parameters, K
i
and k
inact
, were
derived from a Kitz–Wilson plot (Fig. 6B), for
amiclenomycin and compound 1, respectively: K
i
¼
12±2lm,k
inact
¼0.35 ± 0.05 min
)1
, and K
i
¼
20±2lm,k
inact
¼0.56 ± 0.05 min
)1
. The inactiva-
tion was irreversible, as a sample of His
6
-tagged
M. tuberculosis DAPA AT inactivated at 90% by ami-
clenomycin did not recover its activity after prolonged
dialysis in the presence of 0.1 mmPLP. The partition
ratio for the inactivation by amiclenomycin was meas-
ured by incubating the His
6
-tagged M. tuberculosis
DAPA AT (5.7 lm) for 45 min with a substoichiometric
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
01234567
[AdoMet]/v (103 min)
[AdoMet] (mM)
A
0.0
10.0
20.0
30.0
40.0
0 50 100 150
Km app/Vm (103 min)
[Racemic-KAPA] (µM)
B
Fig. 5. Hanes–Woolf plot of the inhibition of His
6
-tagged M. tuber-
culosis DAPA AT by KAPA. (A) The activity was measured at var-
ious AdoMet and KAPA concentrations. n20 lMKAPA; h50 lM
KAPA; d70 lMKAPA; s100 lMKAPA; r140 lMKAPA. (B) Re-
plot of the ordinate intercepts against KAPA concentrations. Data
were fitted to straight lines using linear regression analysis.
M. tuberculosis DAPA aminotransferase S. Mann and O. Ploux
4782 FEBS Journal 273 (2006) 4778–4789 ª2006 The Authors Journal compilation ª2006 FEBS