Báo cáo khoa học: 7,8-Diaminoperlargonic acid aminotransferase from Mycobacterium tuberculosis, a potential therapeutic target Characterization and inhibition studies pptx

Keywords 7,8-diaminopelargonic acid aminotransferase; amiclenomycin; biotin biosynthesis; Mycobacterium tuberculosis; S-adenosyl- L -methionine Correspondence O.. Mycobacterium tuberculo

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,

boıˆte 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

pKa 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: KmAdoMet¼ 0.78 ± 0.20 mm,

KmKAPA¼ 3.8 ± 1.0 lm, kcat¼ 1.0 ± 0.2 min)1, KiKAPA ¼ 14 ± 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: Ki¼

12 ± 2 lm, kinact¼ 0.35 ± 0.05 min)1, and Ki¼ 20 ± 2 lm, kinact¼ 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 1 specifically 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.

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 tuberculosisDAPA 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-MTHis6bioA, contained the same ribosome-binding site and a sequence coding for His6inserted 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 His6-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

NH 2

O

KAPA

NH 2

H 2 N COOH

NH 2

OH

DAPA

COOH

NH 2

NH 2

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.

biotin-free Luria–Bertani agar plate (containing

0.45 UÆmL)1avidin), 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 His6-tagged M

tuberculo-sis DAPA AT was low in E coli JM105⁄

pUC18-MTHis6bioA: 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 lacIqstrain

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-MTHis6bioA and 0.12 mUÆmg)1 in E coli BL21

CodonPlus(DE3)RP⁄ pUC18-MTHis6bioA This slightly

increased production compared with that in E coli

JM105⁄ pUC18-MTHis6bioA was attributed to the

overproduction of the rare tRNAs However, when

produced in E coli BL21 CodonPlus(DE3)RP⁄

pUC18-MTHis6bioA, DAPA AT was predominantly in an

insoluble form Unfortunately, attempts to solubilize

the precipitated proteins in 8 m urea and renature the

DAPA AT were unsuccessful The His6-tagged DAPA

AT was thus purified from the soluble crude extract of

E coliBL21 CodonPlus(DE3)RP⁄ pUC18-MTHis6bioA

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 precipitasolu-tion 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 His6 -tagged enzyme, which shows that the six N-terminal histidine residues of His6-tagged DAPA AT do not perturb the catalytic activity

Biochemical characterization

As shown in Fig 2, wild-type and His6-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 His6 -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, A414⁄ A280, was 0.219 for the pure enzyme The specific activity of His6-tagged M tuberculosis DAPA AT was measured at different pH values, from 6.8 to 9.1, in the presence of 20 lm KAPA 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 (pKa1):

a¼ amax=ð1 þ 10pKa1 pH

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 pKa values

Fig 2 Analysis by SDS ⁄ PAGE of the purification of His6 -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 m M imidazole; lane

10, purified wild-type M tuberculosis DAPA AT; lane 11, purified

His6-tagged M tuberculosis DAPA AT; lane 12, molecular mass

standards (66 kDa, 45 kDa, 36 kDa, 29 kDa, 24 kDa, 20 kDa).

obtained were 7.6 and 8.0 for E coli and M

tuberculo-sisDAPA AT, respectively It should be noted that the

data points between pH 6.5 and pH 7.1 for M

tuber-culosisDAPA 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 tuberculosisenzyme, 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 His6-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., KiKAPA, KmKAPA,

KmAdoMet, and Vm 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 Km⁄ Vm as a function of KAPA con-centration gave: KiKAPA ¼ 14 ± 2 lm, KmAdoMet¼

0.1

1.0

10.0

100.0

6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0

-1)

pH

Fig 4 Activity versus pH profile for the His6-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) h E coli DAPA AT; d His6-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

Wavelength (nm)

Fig 3 UV-visible spectrum of His6-tagged M tuberculosis DAPA

AT The absorption spectrum of purified His6-tagged M

tuber-culosis DAPA AT (0.44 mgÆmL)1) in 50 m M Tris ⁄ HCl buffer

(pH 8.0) ⁄ 10 m M 2-mercaptoethanol was recorded against a blank

containing the same buffer.

0.58 ± 0.1 mm and Vm¼ 22 ± 6 mUÆmg)1 (Fig 5B).

Thus the catalytic constant is kcat¼ 1.0 ± 0.2 min)1

To estimate KmKAPA, 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 KmAdoMet¼ 0.96 ± 0.1 mm and Vm¼ 21 ± 7 mUÆmg)1 and KmKAPA¼ 3.8 ± 1.0 lm Because two different values for KmAdoMet 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 Km values for the latter are 3–4 times higher, and that the kcatfor the M tuberculosis enzyme is eight times lower than that of the E coli enzyme Furthermore, the inhibition constant, KiKAPA, 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 His6-tagged DAPA AT

by amiclenomycin and 4-(4c-aminocyclohexa-2,5-dien-1r-yl)propanol (compound 1)

When the His6-tagged M tuberculosis DAPA AT was preincubated, at pH 8.0, in the presence of amicleno-mycin or compound 1 at 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, kobs, varied hyperbolically with the inactivator concentration Thus, the simple two-step model for irreversible inactivation may apply, and the following kinetic parameters, Ki and kinact, were derived from a Kitz–Wilson plot (Fig 6B), for amiclenomycin and compound 1, respectively: Ki¼

12 ± 2 lm, kinact¼ 0.35 ± 0.05 min)1, and Ki¼

20 ± 2 lm, kinact¼ 0.56 ± 0.05 min)1 The inactiva-tion was irreversible, as a sample of His6-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 mm PLP The partition ratio for the inactivation by amiclenomycin was meas-ured by incubating the His6-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

3 min)

[AdoMet] (mM)

A

0.0

10.0

20.0

30.0

40.0

Km

3 min)

[Racemic-KAPA] (µM)

B

Fig 5 Hanes–Woolf plot of the inhibition of His6-tagged M

tuber-culosis DAPA AT by KAPA (A) The activity was measured at

var-ious AdoMet and KAPA concentrations n 20 l M KAPA; h 50 l M

KAPA; d 70 l M KAPA; s 100 l M KAPA; r 140 l M KAPA (B)

Re-plot of the ordinate intercepts against KAPA concentrations Data

were fitted to straight lines using linear regression analysis.

amount of amiclenomycin and by measuring the

resid-ual enzymatic activity Plotting the fraction of residresid-ual

activity against the molar ratio of amiclenomycin over

that of the enzyme active sites (Fig 7) gave a straight

line that intersected the abscissa at 1.1 molar ratio

The same experiment was run for the inactivation by

compound 1, and the partition ratio was found to be 1.0 Thus, these suicide substrates inactivate M tuber-culosis DAPA AT almost every turnover, which make these inactivators particularly efficient Figure 8 shows the mechanism by which amiclenomycin and analogues inactivate E coli DAPA AT [25] There is no doubt that the inactivation of M tuberculosis DAPA AT observed here follows the same reaction pathway This mechanism is reminiscent of that proposed by Rando [31] for the inactivation of c-aminobutyric acid transa-minase by gabaculine and more recently by others for the inactivation of c-aminobutyric acid transaminase [32], d-amino acid aminotransferase [33], and alanine racemase [34] by cycloserine It is quite interesting to note that all these PLP-dependent enzymes are inhib-ited by a similar mechanism, ultimately yielding an aromatic ring that does not dissociate from the active site We investigated if the M tuberculosis DAPA AT was inhibited by gabaculine and the antituberculous drug cycloserine, which seems to be poorly specific In fact, none of these compounds inhibited DAPA AT even at high concentration (1.3 mm), showing that the DAPA AT is quite specific

As shown above, the nature of the main chain amino acid in amiclenomycin versus alcohol in com-pound 1 has only a very moderate effect on the inacti-vation parameters This finding is quite encouraging for the design of new inhibitors because the synthesis

of amiclenomycin takes longer than that of

com-10

100

Time (min)

A

0

2

4

6

8

10

-0.10 -0.05 0.00 0.05 0.10 0.15 0.20

1/[Amiclenomycin] (µ M-1)

B

Fig 6 Kinetics of inactivation of His6-tagged M tuberculosis

DAPA-AT by amiclenomycin (A) The enzyme was preincubated in

the presence of various concentrations of amiclenomycin: d no

inhibitor; s 5.7 l M ; n 11.4 l M ; n 17.1 l M ; r 25.7 l M At different

time points, the residual activity was measured and plotted on a

semi-log plot against time The data were fitted to simple

exponen-tial decay The slopes of the lines gave an estimate of the observed

inactivation constants, kobs (B) Double-reciprocal plot of the

observed rate of inactivation, kobs, against the inhibitor

concentra-tion The data were fitted to straight lines using a linear regression

analysis.

0 20 40 60 80 100

0.0 0.2 0.4 0.6 0.8 1.0 1.2

Molar ratio

Fig 7 Titration of His6-tagged M tuberculosis DAPA AT by ami-clenomycin The enzyme and amiclenomycin, at various molar ratios, were incubated at pH 8.0 for 45 min at 37 C The residual activity was then determined and plotted against the molar ratio The data were fitted to a straight line using linear regression analysis.

pound 1 and other analogues [20,21] Furthermore,

varying this part of the molecule might afford new

interesting properties such as bioavailability

In vivo antibiotic effect of compound 1

When E coli C268⁄ pUC18-MTHis6bioAwas grown on

solid rich medium devoid of biotin

(avidin-supplemen-ted Luria–Bertani agar medium), growth was inhibi(avidin-supplemen-ted

by increasing the concentration of compound 1 The

minimal concentration that completely inhibited growth

was found to be 100 lgÆmL)1 When avidin was not

added, the biotin present in the Luria–Bertani medium

( 0.2 lm) was sufficient to reverse the growth

inhibi-tion Furthermore, adding 100 lm DAPA to the

med-ium in the presence of avidin also reversed the growth

inhibition Taken together, these data indicate that,

firstly, compound 1 is able to cross the cell wall, and

that, secondly, M tuberculosis DAPA AT is the only

in vivotarget of this inhibitor The in vivo effect of

com-pound 1 was similarly tested on M smegmatis CIP

56.5, a wild-type strain The minimal concentration that

completely inhibited growth, measured in Luria–Bertani

medium without biotin, was 10 lgÆmL)1 Biotin present

in the medium was sufficient to reverse the effect These

experiments could not be repeated with amiclenomycin,

because we did not have sufficient amounts of this

molecule However, Okami et al [4–6] reported a

mini-mum inhibitory concentration for amyclenomycin of

3–6 lgÆmL)1, on mycobacteria This value is lower than

that measured for compound 1 for the E coli strain but

similar to that obtained for M smegmatis In our case,

the target, DAPA AT, is overproduced in the E coli

strain, thus increasing the minimum inhibitory

concen-tration The in vivo effect of compound 1 on M

tuber-culosiscells needs to be studied

In conclusion, we have purified and characterized

DAPA AT from M tuberculosis The purification

scheme is simple enough to provide sufficient pure enzyme for structural studies This work is underway

in our laboratory We have also provided evidence that amiclenomycin and compound 1 are suicide substrates

of M tuberculosis DAPA AT The fact that com-pound 1 is active in vivo and that it specifically targets DAPA AT makes this molecule an interesting lead to new antibiotics

Experimental procedures

Materials and equipment

The M tuberculosis H37Rv bioA (Rv1568) gene, cloned into

a plasmid, was a gift from P Alzari (Institut Pasteur, Paris, France) M smegmatis CIP 56.5 was obtained from the Institut Pasteur Collection E coli strains JM105 and BL21

USA) and E coli BL21 CodonPlus(DE3)RP was from Stratagene (La Jolla, CA, USA) E coli C268 (DbioA his

(thr-1 leuB6(Am) glnV44(AS) bioA109 LAM- rfbC1 thi-1; CGSC#7257) [36] was generously given by the E coli Gen-etic Stock Center (Yale, NH, USA) Plasmid pUC18 was from Promega Synthetic oligonucleotides were products of Proligo (Paris, France) and were used without any further purification Chemicals were purchased from Sigma-Aldrich (St Louis, MO, USA) and were of the highest purity avail-able Racemic-KAPA was obtained as already described [37] (7S,8R)-DAPA was a gift from J Crouzet (Sanofi-Aventis, Vitry sur Seine, France) Amiclenomycin was pre-pared as already described [20] Dethiobiotin synthase (EC 6.3.3.3) was expressed and purified as previously described [24] Restriction endonucleases, Taq polymerase, T4 DNA ligase and molecular biology kits were from either Promega

or Roche (Meylan, France) Culture medium components were purchased from Difco Laboratories (Detroit, MI, USA) Chromatographic equipment (GradiFrac, FPLC) and column phases were from Amersham Biosciences

Enz-Lys

NH 2

N

O O

HO 3 P

N H R

H H

H

Enz-Lys

NH3

N

O O

HO3P

N H R

H

H

Enz-Lys

NH2

N

O O

HO3P

N H R

H

H

H

Enz-Lys

NH2

N

O O

HO3P

N H R

H H

External aldimine Quinonoid PMP adduct Aromatic adduct

Fig 8 Proposed inactivation mechanism of DAPA AT by amiclenomycin and analogues The conserved active-site Lys residue is probably responsible for the transamination reaction, and the aromatization step is promoted by an as yet unknown base R represents the various main chains of the analogues.

(Orsay, France) UV-visible spectra were obtained on a

Uvikon-930 Kontron (Munchen, Germany)

spectrophoto-meter or a Lambda-40 Perkin–Elmer (Norwalk, CT, USA)

apparatus Sonication was performed on a VibraCell

carried out on a Bio-Rad (Hercules, CA, USA) Protean II

system, using the conditions described by the manufacturer

DNA electrophoresis was performed on a Mupid

per-formed in a Sorval RF5plus centrifuge (Kendro,

Brucker spectrometer (Rheinstetten, Germany) CI mass

spectra were obtained with a Nermag R 30–10 apparatus

(Quad Service, Poissy, France)

Synthesis of compound 1

The aminocyclohexadiene 1,

4-(4c-aminocyclohexa-2,5-dien-1r-yl)propanol, was prepared from its N-allyloxycarbonyl

precursor 2, i.e.,

allyl[4c-(3-oxopropyl)-cyclohexa-2,5-dien-1r-yl]-carbamate, the synthesis of which has already been

published [20] Phenylsilane (150 lL, 1.2 mmol) and a

dichloro-methane (1.5 mL) were added under argon to a solution of

precursor 2 (167 mg, 0.7 mmol) in dry dichloromethane

(1.5 mL) The mixture was stirred for 50 min at room

tem-perature, concentrated under vacuum and chromatographed

on a silica gel (flash silica, Merck 230, 0.04–0.063 mm)

Eluted fractions were combined and acidified to pH 4 using

1 m HCl After concentration under vacuum the

hydrochlo-ride salt of compound 1 was obtained as an yellow oil

(80 mg, 60%)

1

Cloning of M tuberculosis bioA gene

were obtained using PCR amplification of the bioA gene

cloned into plasmid pDEST17 (P Alzari, Institut Pasteur)

Amplifications were achieved using the Taq DNA

polym-erase (Promega) under the conditions recommended by the

sulfoxide The following sets of primers were used to obtain

the wild-type recombinant gene: 5¢-CGCGCGAATTCAG

GAGGAATTTAAAATGGCTGCGGCGACTGGCGGG-3¢

containing an EcoRI restriction site and a ribosome-binding

site, and 5¢-GCAAGCTTTCATGGCAGTGAGCCTACG AGCCG-3¢ containing a HindIII restriction site For the

5¢-CGCGCGAATTCAGGAGGAATTTAAAATGCACCAC CACCACCACCACGCTGCGGCGACTGGCG-3¢ contain-ing an EcoRI restriction site, a ribosome-bindcontain-ing site and

GCAGTGAGCCTACGAGCCG-3¢ containing a HindIII restriction site The DNA fragments were purified (PCR Preps, Promega), digested with EcoRI and HindIII, purified

on agarose gel, and ligated into pUC18 previously cut by the same restriction enzymes After transformation in

plas-mids were extracted and purified (Wizzard Plus Minipreps, Promega) for DNA sequencing (ECSG, Evry, France)

thus obtained and used to transform various E coli strains

Transformation and phenotype determination

no plasmid DNA was run at the same time Biotin auxotro-phy was determined by plating the cells on Luria–Bertani

Expression and purification of wild-type and His6-tagged recombinant M tuberculosis DAPA AT

Wild-type DAPA AT

(800 mL batches in Erlenmeyer flasks), supplemented with

The cells were collected by centrifugation (4000 g, 15 min)

Crude extract

The cells were thawed on ice and suspended in 50 mm

PLP, and disrupted by sonication for 70 s (seven 10-s pulses with intermittent 1-min cooling periods) The cellular debris were removed by centrifugation (10 000 g, 20 min)

Q-Sepharose

The supernatant was loaded on a Q-Sepharose column

2-mer-captoethanol) After the column had been washed with

50 mL buffer A, the proteins were eluted using a linear

gradient (0–0.4 m NaCl in buffer A, 400 mL) Active

frac-tions were detected using the coupled enzymatic assay (see

below) and pooled

FPLC Mono Q

FPLC column using a linear salt gradient (0–450 mm NaCl

in buffer A, 80 mL) Active fractions were detected using

the coupled enzymatic assay (see below) and pooled The

purified enzyme solution was desalted and concentrated by

repetitive ultrafiltrations (Centriprep 30; Millipore, Bedford,

His6-tagged DAPA AT

An overnight preculture (50 mL Luria–Bertani medium,

used to inoculate 5 L Luria–Bertani medium (1-L batches

col-lected by centrifugation (4000 g, 15 min) and kept at

)20 C until use The following steps were all run at 4 C

The cell paste was resuspended in 40 mL buffer B: 50 mm

was sonicated on ice for 70 s (seven 10-s pulses with

(10 000 g, 20 min), the supernatant was supplemented with

PLP (0.1 mm final concentration) and directly loaded on a

nickel affinity column (chelating Sepharose; 1.6 cm internal

diameter, 5 cm long, 10 mL) prepared as recommended by

the manufacturer and equilibrated with buffer B After

loading, the column was washed with 100 mL buffer B,

and the proteins were eluted with 100 mL buffer B

contain-ing 100 mm imidazole and then 100 mL buffer B containcontain-ing

200 mm imidazole The column was run at a flow rate of

pres-ence of protein in the fractions was detected using the

Bradford assay, and the purity of individual fractions was

AT were pooled, and the protein was precipitated by the

addition of ammonium sulfate at 70% saturation The

pre-cipitated protein was recovered by centrifugation (10 min at

pH 8.0, 10 mm 2-mercaptoethanol and dialyzed overnight

against 1 L of the same buffer supplemented with 0.1 mm

Determination of the oligomerization state

-tagged DAPA AT was estimated by gel filtration on an

FPLC apparatus equipped with a calibrated Superdex 200

flow rate, detection set at 280 nm) Commercial standards

dex-tran, l-tryptophane, b-amylase, BSA, chymotrypsin, alco-hol dehydrogenase, cytochrome c, carbonic anhydrase, standards from Sigma) were separated on the column, and

logarithm of their molecular mass [39] The linear plot thus obtained was used to estimate the native molecular mass of

used for the determination)

Protein assay

Protein concentrations were determined using the colori-metric assay described by Bradford [40] and as supplied by Bio-Rad

DAPA AT assays Coupled assay

The assay (100-lL final volume) consisted of, unless other-wise stated: 100 mm 4-(2-hydroxyethyl)-1-piperazinepro-panesulfonic acid (EPPS) buffer, pH 8.6, 10 mm ATP,

(800 ng), 0.1 mm PLP, 20 lm KAPA, 1 mm AdoMet and DAPA AT (185 ng) The mixture was preincubated for

add-ing the DAPA AT The reaction was stopped by addadd-ing

formed was quantified by the standard disc bioassay proce-dure, using E coli C268 as described [24] One unit is defined as the amount of enzyme producing 1 lmol product per min in the conditions described above

Direct assay

The assay (100 lL final volume) consisted of, unless other-wise stated: 100 mm EPPS buffer, pH 8.6, 0.1 mm PLP,

20 lm KAPA, 1 mm AdoMet and DAPA AT (0.65 lg) The assay was run as described above for the coupled assay The DAPA formed was quantified by the standard disc bioassay procedure using E coli MEC1 [13] in a modi-fied Vogel–Bonner minimal medium [24], prepared without casamino acids A range of authentic DAPA samples from 0.7 to 20 pmol was used as standards

Although the direct assay is more sensitive and simpler (no coupling enzyme) than the coupled assay, we found the former less reliable Therefore, the coupled assay was used throughout this work except when high sensitivity was

required (inactivation studies and alternate amino donor

experiments, see below)

pH profile

activity was measured using the coupled assay (as described

above) but at various pH values, using the following buffer

solutions: 100 mm Hepes from pH 6.8 to pH 8.2, 100 mm

EPPS from pH 7.6 to pH 8.6, and 100 mm TAPS

from pH 7.7 to pH 9.1 The E coli DAPA AT was purified

as previously described [18] and assayed under the same

conditions Control experiments were run to ensure that the

dethiobiotin synthase-catalyzed step was never rate

deter-mining in these conditions The activity versus pH profile

data were fitted to Eqn (1), using a nonlinear regression

analysis supported by Kaleidagraph software (Synergy

Soft-ware, Reading, PA, USA)

Amino donor

The direct enzymatic assay was carried out as described

above, except that AdoMet was replaced by various amines

-tagged M tuberculosis DAPA AT was used (1.3 lg)

Determination of the kinetic parameters

of DAPA AT

DAPA AT using the coupled enzymatic assay, as described

above, and by varying AdoMet and KAPA concentrations

The inhibition type and the inhibition constant displayed by

method, i.e., the rate of the reaction was measured while

keeping the substrate concentrations at a constant ratio

[AdoMet]⁄ [KAPA]ratio was set at 125, 250 and 375, and

[AdoMet] was varied from 0.25 mm to 4 mm It was

experi-mentally difficult to obtain good quality data using lower or

higher molar ratios, because increasing the KAPA

concen-tration led to strong inhibition and increasing the AdoMet

concentration above 4 mm also led to a slight inhibition

See the Supplementary material for detailed analysis

Kinetics of DAPA-AT inactivation by

amiclenomycin and compound 1

2-mercaptoethanol and amiclenomycin or compound 1 at

various concentrations (for amiclenomycin: 5.7, 11.4, 17.1,

25.7 lm, and 9.8, 22.0, 24.6, 34.4, 30.0, 40.0, 50.0 lm for

was measured at different time points by withdrawing

3.5-lL samples of the preincubation mixture and adding them

to the coupled assay mixture The dethiobiotin formed was then quantified by the standard disc bioassay, as described above Plotting the logarithm of the residual activity against time gave straight lines, the slope of which gave the

against inhibitor concentration in a double-reciprocal plot

experi-ments using d-cycloserine or l-cycloserine (1.3 mm) or gabaculine (1.3 mm) were run under the same conditions

Titration of DAPA AT by amiclenomycin

a molecular mass of 45 kDa per monomer) was incubated in

2-merca-ptoethanol and amiclenomycin or compound 1 (from 1.7 to

withdrawn (3.5 lL, corresponding to 0.89 lg DAPA AT), and the residual enzymatic activity was measured using the coupled assay, as described above Plotting the residual activity against the molar ratio of inactivator over that of DAPA AT active sites gave a straight line that extrapolated

to 1.1 for amiclenomycin and to 1.0 for compound 1

Irreversibility of the inactivation by amiclenomycin

final volume of 17.5 lL A 3.5-lL sample was withdrawn, and the residual activity was measured using the direct en-zymatic assay, as described above At the same time, the remaining mixture was diluted to a final volume of 100 lL

The resulting solution was dialyzed against 0.25 L 50 mm

enzyme solution was then measured, taking into account the dilution factor due to the dialysis step A blank contain-ing all components but without the inhibitor was carried out in parallel and treated in the same way

Effect of compound 1 on E coli C268/pUC18-MTHis6bioA and M smegmatis growth