OYE flavoprotein reductases initiate the condensation of TNT-derived intermediates to secondary diarylamines and nitrite

Article abstract of DOI:10.1021/es071449w

Polynitroaromatic explosives such as 2,4,6-trinitrophenol (picric acid) and 2,4,6-trinitrotoluene (TNT) are toxic and recalcitrant environmental pollutants. They persist in the environment due to the highly inactivated π system of their aromatic rings, wh

Full text of DOI:10.1021/es071449w

Environ. Sci. Technol. 2008, 42, 734–739
acid) and 2,4,6-trinitrotoluene (TNT) were regarded as
OYE Flavoprotein Reductases Initiate
the Condensation of TNT-Derived
Intermediates to Secondary
Diarylamines and Nitrite
extremely recalcitrant. Nevertheless, a convincing catabolic
route was described in gram-positive Rhodococcus and
Mycobacterium strains, whereby the aromatic rings of these
trinitroarenes are reduced to Meisenheimer hydride com-
plexes (4–6). These findings clearly demonstrated that, while
mono and dinitroaromatics are preferentially attacked by
bacterial oxygenase enzymes, the highly inactivated aromatic
π system of trinitroarenes can only be activated by reductases
such as those found in the aforementioned microorganisms.
Low concentrations of azoxy- and azo condensates from
intermediary hydroxylamine and nitroso derivatives have also
been identified and in a few cases partially denitrated
intermediates or end-products were reported as well, though
their proposed structures were not always unequivocally
confirmed (3).
In contrast to picric acid, which was shown to be
completely mineralized through H--trinitrophenol forma-
tion, further reduction by a hydride transferase and subse-
quent ring cleavage (7), TNT biotransformation by the above
bacteria resulted in the formation of reduced aromatic ring
systems, partially reduced nitro groups and concomitant
nitrite release as end-products. However, the Meisenheimer
dihydride complex formed by Rhodococcus erythropolis strain
HL PM-1 has been described as having a very short half-life
of only a few hours, whereas investigating its structure in a
nuclear magnetic resonance spectrometer, this reactive
compound was assumed to undergo rapid decomposition,
presumably by hydrolytic denitration in deuterated water
(6). Also important in this context and from an environmental
point of view, is the immobilization of biologically generated
reactive products into nonextractable polymeric soil struc-
tures. Thus, they could escape structural identification as
potential environmental pollutants (8, 9).
In the meantime, the description of TNT-reducing bacteria
has become legion (10–12). OYE flavoproteins have been
identified in gram-negative bacteria such as Enterobacter
cloacae from which pentaerythritol tetranitrate (PETN)
reductase was purified and crystallized (13, 14). This enzyme
not only can reduce nitro groups of nitrate esters but also
of TNT. Additionally, it simultaneously reduces the aromatic
ring leading to nitrite release for growth (13). Similarly, the
so-called xenobiotic reductase B from Pseudomonas fluo-
rescens I-C (15, 16) catalyzes analogous reactions, thereby
confirming earlier observations (17, 18) of the reductive
potential of bacteria of the genus Pseudomonas. Moreover,
the OYE-like N-ethyl maleimide reductase from Escherichia
coli demonstrates similar capacities (2) and has been shown
to be responsible for allowing growth of this bacteria with
TNT as sole nitrogen source (19, 20). However, in most of the
abovementioned cases, the structures of the denitrated
intermediate(s) and end products, information critical for
determining the importance of these reactions for the
bioremediation of TNT, still await unequivocal elucidation.
Here, we demonstrate that stabilization of the TNT-
derived Meisenheimer dihydride complex originating either
from in vitro biological enzyme reactions with a recombinant
flavoprotein reductase of Pseudomonas putida JLR11, or
generated chemically, occurs through the condensation with
biologically or chemically prepared reactive hydroxylamines
to form stable secondary diarylamines.
R O L F - M I C H A E L W I T T I C H , * , †
A L Í H A Ï D O U R , † , ‡ P I E T E R V A N
D I L L E W I J N , A N D J U A N - L U I S R A M O S †
Departamento de Protección Ambiental, Estación
Experimental del Zaidín, Consejo Superior de Investigaciones
Científicas, Calle Profesor Albareda 1, E-18008 Granada, Spain, and
Unidad de Resonancia Magnética Nuclear, Centro de
Instrumentación Científica, Universidad de Granada, Paseo
Juan Osorio s/n, E-18071 Granada, Spain
Received June 15, 2007. Revised manuscript received November
20, 2007. Accepted November 21, 2007.
Polynitroaromatic explosives such as 2,4,6-trinitrophenol
(picric acid) and 2,4,6-trinitrotoluene (TNT) are toxic and
recalcitrant environmental pollutants. They persist in the
environment due to the highly inactivated π system of their
aromatic rings, which are inaccessible to dioxygenases that
normallyinitiatethebacterialaerobiccatabolismof(nitro-)aromatic
compounds. Aside from reductive transformation of nitro
side groups to hydroxylamines, trinitroarenes are prone to
aromatic ring reductions by some flavin reductases to yield
Meisenheimer mono and dihydride complexes. Here we show
that the simultaneous accumulation of Meisenheimer
complexes and aromatic hydroxylamines derived from TNT
gives rise to the condensation of both types of reactive
intermediates to secondary diarylamines and nitrite as the end-
products of this environmentally relevant reaction sequence.
As a consequence, overall mass balances of aerobic
biotransformations of TNT become possible for the first time.
In our study, the process of TNT activation was enzymatically
initiated by the xenobiotic reductase B (XenB)-like flavin
reductase of Pseudomonas putida JLR11 and then completed
chemically by autodimerization. The structures of the formed
end products were unequivocally elucidated by NMR.
Introduction
The increasing number of publications on the old yellow
enzyme (OYE) flavoprotein reductases of yeast and bacteria
strongly suggests that their activity with regard to the
reduction of cyto- and hepatotoxic nitroaromatic compounds
is well established, and of considerable interest for applica-
tions in industrial and medical biotransformations, and whole
cell bioremediation technologies (1, 2). Catabolic pathways
for aerobic bacterial mineralization of mono and dinitroaro-
matic compounds, generally initiated by electrophilic di-
oxygenase systems, have long been known (3). However,
trinitrated compounds such as 2,4,6-trinitrophenol (picric
Materials and Methods
* Corresponding author phone: +34-958-181 600 X138; fax: +34-
958-129 600; e-mail: rolf.wittich@eez.csic.es.
† Consejo Superior de Investigaciones Científicas.
‡ Universidad de Granada.
Chemicals. Analytical standards of amino- and hydroxyl-
aminodinitrotoluenes as well as azoxy and azo adducts were
from AccuStandard (New Haven, CT). All other nitroaromatics
9
734 ENVIRONMENTAL SCIENCE & TECHNOLOGY
/
VOL. 42, NO. 3, 2008
10.1021/es071449w CCC: $40.75
2008 American Chemical Society
Published on Web 01/03/2008

were from Aldrich Chemie, Steinheim, Germany, and all other
chemicals (p.a. grade) from Fluka, Buchs, Switzerland. TNT
was obtained from Unión Española de Explosivos (Madrid,
Spain).
20SXB FT-IR spectrometer. Mass spectral data were obtained
in the negative ionization mode with a Varian model 1200L
quadrupole spectrometer coupled to a Varian Prostar HPLC
system. Routine HPLC analyses were performed on a Hewlett-
Packard model 1050 chromatograph equipped with a PDA
at 25 °C on a Waters Nova-Pak C-8 column (5 µm, 3.9 × 150
mm). Solvent A of eluent system I consisted of water
containing 5 mM tetrabutylammonium phosphate (pH 7.0,
Fluka) as the ion-pairing agent and solvent B was gradient-
grade acetonitrile. The compounds mentioned below (See
Construction of an Expression Vector for Recombinant
Protein Production. Pseudomonas putida JLR11 (no. PSC
44, EEZ, Pseudomonas Reference Culture Collection, Gran-
ada) was grown routinely at 30 °C in M9 minimal medium
supplemented with 0.5% (wt/vol) glucose or 10 mM sodium
benzoate as a carbon source (21). Chromosomal DNA from
P. putida JLR11 was obtained using an AquaPure genomic
DNA isolation kit (Bio-Rad). To obtain the xenB homologue
gene of P. putida JLR11, genomic DNA was used as a template
for amplification by PCR using 5′-TTTGGATCCATAAAAG-
CACTGGCCCAC-3′ as a forward primer and 5′-AAAAAGCT-
TCAACCGCGGATAATCGATG-3′ as a reverse primer. The
amplification product was then digested with the restriction
enzymes BamHI and HindIII prior to ligation into a
pET28b(+) vector (Novagen) which likewise had been
digested with BamHI and HindIII. The resulting plasmid
contained the coding sequence in frame with a DNA sequence
encoding a His6-tag at its 3′ end which was verified by DNA
sequencing. The nucleotide sequence of the xenobiotic
reductase B of P. putida JLR11 is 100% identical to the
sequence of the same name (Genebank ID AAN66545) in P.
putida KT2440.
Purification and Assay of XenB-like Reductase. For
protein-His6 purification, the pET28b+ derivative plasmid
was transformed into E. coli BL21 pLysS (Novagen). The cells
were grown in several one-liter batches at 30 °C in 2 × YT
culture medium (21) with 50 µg mL-1 kanamycin to an A660
between 0.5 and 0.7 and then induced with 1 mM IPTG. Cells
were harvested after overnight induction at 18 °C, resus-
pended in 500 mM NaCl, 10% (vol/vol) glycerol, 25 mM
sodium phosphate buffer pH 7.5, 0.1 mM EDTA, and 1 µM
FMN and protease inhibitor cocktail (complete, Roche) and
disrupted by treatment with 20 µg mL-1 of lysozyme and
French press. Following centrifugation at 20000g for 30 min,
the protein was found predominantly (more than 80%) in
the soluble fraction. XenB-His6 was purified by nickel affinity
chromatography and eluted with a continuous imidazole
gradient. Peak fractions were pooled and the protein
concentration determined using the Bio-Rad Protein Assay
kit. The molecular mass was determined by SDS-PAGE using
defined protein standards (Bio-Rad). Results of the purifica-
tion are shown in Supporting Information (SI) Figure 1).
The enzyme reaction contained in a total volume of 1 mL:
0.85 mL of 50 mM sodium potassium phosphate buffer of
pH 7.0 saturated with TNT (overnight with an excess of
mortar-ground TNT followed by filtration before starting the
experiment, final concentration 0.49 mM at 25 °C), 0.02 mL
of isopropanol as the substrate for alcohol dehydrogenase,
also saturated with TNT and giving a total concentration of
0.9 mM of TNT in the assay; 0.01 mL of NADPH (10 mM),
and 0.1 mL of secondary alcohol dehydrogenase (5 U) from
Thermoanaerobium brockii (Sigma) for recycling of formed
NADP+. The reaction was initiated by the addition of 0.02
mL (2.25 µM) of purified recombinant xenobiotic reductase
B of the bacterial strain Pseudomonas putida JLR11 and
proceeded at 25 °C in the autosampler of the HPLC system,
which served to monitor the course of the enzymatic reaction
(concentration of NADPH and NADP), the depletion of TNT
(1), and the transient formation of the monohydride complex
of TNT (2), the subsequent dihydride complex (3), and the
two isomeric secondary diarylamines 9 and 10 over time
(see Figure 1).
also Figure 1) were separated at a flow rate of 0.85 mL min-1
.
The gradient program was as follows: 0-13% B in 3 min,
then to 25% in 17 min, and to 62% in 23 min; reequilibration
was for 9 min. Retention times (min) and maxima of selected
UV–vis spectra (nm) were as follows: nitrite, 3.3; nitrate, 4.8;
the five isoforms of the Meisenheimer dihydride complex
(3), 3a 9.1 (231, 422), 3b 9.9 (236, 469), 3c 12.5 (265, 438), 3d
14.4 (263, 502), 3e 15.9 (262, 496); TNT (1), 16.3; 2-hydroxy-
lamino-4,6-dinitrotoluene (6), 13.8; 4-hydroxyl-amino-2,6-
dinitrotoluene (5), 14.8; 2-amino-4,6-dinitrotoluene, 15.1;
4-amino-2,6-dinitrotoluene, 15.7; Meisenheimer monohy-
dride complex of TNT (2) (253, 476), 21.3; minor (3–5% of
(2)) Meisenheimer monohydride complex of TNT (hydrogen
enters geminal with regard to the methyl group) 25.2 (208,
248, 470); the 3 new compounds (7), 29.6 (200, small shoulders
at 230 and 292); (9), 30.7 (201, 276, shoulder at 320); and (10),
31.5 (204, shoulder at 232, 268, shoulder at 320); 4,4′,6,6′-
tetranitro-2,2′-azoxytoluene, 33.3; 2,2′,6,6′-tetranitro-4,4′-
azoxytoluene, 33.9. (Bio-) transformation reactions were
routinely monitored at 210, 230, and 450 nm, and UV–vis
spectra recorded within the range from 200 to 600 nm were
stored on a PC connected to the instrument. Purifications of
compounds 7, 9, and 10 were carried out on a semipreparative
Nova-Pak HR C-18 column (6 µm, 7.8 × 300 mm, Waters)
with acetonitrile-water (40:60%, v:v, system II) at ambient
temperature (22 °C). Retention times were as follows: 38.0
min for 7, 47.2 min for 10, and 55.0 min for 9.
Reduction Products from TNT and Synthesis of the
Dimers. Standard procedures (syntheses of compounds 3,
5, and 6, see Figure 1): Due to the low solubility of TNT in
salts-buffered water (113 mg, 0.49 mmol at 25 °C) (22), an
excess of the mortar-ground compound was dispersed in 2
L of water on a rotary shaker (120 rpm) overnight at 30 °C
before being kept at 25 °C. To the resulting saturated solution
obtained after filtration, small quantities of KBH4 were added
slowly under stirring. The formation of 2 (reddish-brown)
and 3 (orange-yellow) was monitored by ion-pair reversed-
phase HPLC (C-18) after every increment of KBH4 until only
3 was present (monitored as its five isoforms; it has to be
mentioned that 3 is being formed from 2 much faster than
2 from 1, meaning that the presence of 2 is only transient
and 3always present in aqueous solution). We have previously
shown that compound 3 remains stable in aqueous solution
for several weeks while very slowly degrading to 2-hydroxyl-
amino-6-nitrotoluene 4 (19). Since the synthesis of 3 was
performed stoichiometrically from 1 and the UV spectra of
each of the isoforms differ only slightly, the sum of the
isoforms was used to calculate the concentration of 3. On
the other hand, in a 100 mL Erlenmeyer flask flushed with
nitrogen, 1 (681 mg, 3.0 mmol in 20 mL of ethanol) and
ammonium chloride (1.0 g, 18.7 mmol in 10 mL of water)
were combined under constant stirring. Zinc powder (800
mg, 12.3 mmol) was added stepwise and the solution was
kept stirring under temperature control (50 °C). The formation
of compounds 5 and 6 was monitored by HPLC over the
entire time of reaction as had been done for the production
of compound 3. Calculation of the concentrations and
stoichiometry was achieved using the respective reference
compound supplied by AccuStandard (New Haven, CT).
Instrumentation and Analyses. Nuclear magnetic reso-
nance (NMR) spectra were measured on a Varian 500 MHz
spectrometer at room temperature in deuterated acetonitrile.
Infrared (IR) spectra were recorded on a NICOLET model
9 735
/ ENVIRONMENTAL SCIENCE & TECHNOLOGY
VOL. 42, NO. 3, 2008

FIGURE 1. Release of nitrite by rearomatization of 2,4,6-trinitrotoluene-derived Meisenheimer dihydride complexes through
condensation with TNT-derived hydroxylamines. Enzymatic reductions as well as chemical reductions furnish identical products
which undergo the shown condensation reactions.
Upon completion of the reduction of TNT to both
hydroxylaminodinitrotoluenes, the solution containing com-
pounds 5 (78%) and 6 (22%) was cooled to 25 °C, diluted
10-fold with distilled water, neutralized to pH of 7.0 and a
Results and Discussion
We describe here the condensation (dimerization) of the
Meisenheimer dihydride complex (compound 3 in Figure
1) with hydroxylaminodinitrotoluenes (compounds 5 and
stoichiometric amount of compound 3 added, to simulate
6) derived from the chemical or enzymatic reduction of
the conditions in the biological system described above. The
TNT, to form the corresponding secondary diarylamines
(compounds 9 and 10) with the concomitant release of
reaction mixture was stirred in the presence of air and started
to turn turbid after several minutes, indicative for the
significant, nearly stoichiometric amounts of nitrite (Fig-
formation of the dimers, which precipitated nearly completely
ures 2, 3). In both cases identical condensation products
within 4 h due to their low solubility in water. Samples were
were formed, which confirms that this process of nitrite
analyzed every 55 min by HPLC. After 2 more hours of
incubation, the crude products were collected by filtration
and the filtrate extracted with ethylacetate, dried over
anhydrous sodium sulfate and the solvent evaporated in
vacuum. The final yield was 323 mg of a red-brownish solid,
which was dissolved in acetonitrile. The mixture of the newly
formed compounds containing minute amount of residual
TNT was analyzed by LC-MS and separated by semiprepara-
tive HPLC, using a mixture of acetonitrile-water (40:60, v:v).
Fractions from eight separations containing compounds 7,
9, and 10 were pooled, saturated with sodium chloride to
achieve transfer into the organic phase, and the aqueous
phase discarded. The organic phases were dried over
anhydrous magnesium sulfate and the solvent evaporated
in vacuum, yielding in all cases light-yellow semicrystalline
compounds which were further analyzed by NMR and FT-IR
spectrometry.
elimination is basically a chemical SN reaction and thus
not enzyme-catalyzed. Products obtained from both the
enzymatic and chemical dimerization were analyzed by
LC-MS, 1H NMR, 13C NMR, and FT-IR. However, we could
not find any evidence for the occurrence of a carbon-
carbon coupling reaction suggested previously to furnish
isomeric aminodimethyltetranitrobiphenyls (16, 20). In
contrast to several reports on C-C coupling reactions
starting from Meisenheimer complexes, those reported to
yield (rearomatized) diaryl amines are rare. These poly-
functional amines are of interest since they represent
building blocks for many pharmaceuticals, natural prod-
ucts, and other chemicals. Apart from the relevance of
above-mentioned microbially induced processes for en-
vironmental biogeochemistry, the herein reported struc-
tures may also be of potential interest as highly energetic
and insensitive explosives and/or propellant formulations
(23).
9 ENVIRONMENTAL SCIENCE & TECHNOLOGY
/ VOL. 42, NO. 3, 2008
736

isopropanol which is the substrate of the secondary alcohol
dehydrogenase used to recycle NADPH (2). Under these
conditions, additional peaks close to the known azoxy adducts
became detectable although much less hydroxylamine or
amino derivatives of TNT appeared than expected from
literature. Further analyses of the precipitated material
revealed that its two distinct peaks found by HPLC cor-
responded to the aforementioned newly formed peaks (see
SI Figure 2). Since nitrite formation was nearly half of that
expected stoichiometrically with regard to the depletion of
TNT, we therefore hypothesized that the expected hydroxyl-
amine derivatives of TNT were intercepted by a condensation
reaction, whose products form part of the observed
precipitate.
The precipitate after being dissolved in acetonitrile
produced two mayor peaks in HPLC-DAD analysis using the
ion-pair eluent system I. Each peak gave a pseudo molecular
mass of 376 [M-H]- when analyzed by HPLC-MS using the
solvent system II in the negative ionization mode, (SI Figure 3
A) matching an empirical formula of C14H11N5O8. Since little
evidence for possible solid structures of the condensed
products or for their synthesis pathways were found in the
literature (16, 20), we decided to elucidate the corresponding
structures by establishing a chemical model reaction by de
novo synthesis. We started with the chemically synthesized
Meisenheimer dihydride complex of TNT (3) dissolved in
water (the absence of TNT or the monohydride complex had
previously been confirmed by HPLC-DAD and 1H NMR) and
pure (AccuStandard) 4-hydroxylamino-2,6-dinitrotoluene or
2-hydroxylamino-4,6-dinitrotoluene dissolved in acetonitrile
(5 or 6, respectively, Figure 1). The condensation of 3 and
5 and of 3 and 6 and the concomitantly released nitrite were
followed by HPLC-DAD analysis over time. The formation of
10 from 3and 6is shown in Figure 2 and clearly demonstrated
the stoichiometry of the reaction sequence depicted in Figure
1. The condensation of 3 and 5 to generate 9 gave a similar
image (not shown). The peaks shared by chemical and
biological reactions exhibited identical retention times and
UV–vis spectra. Also LC-MS [M-H]- analysis furnished
identical pseudo molecular masses of m/z)376 (C14H11N5O8)
for each of the two new peaks (SI Figure 3C and D).
FIGURE 2. Chemical autocondensation of 2-hydroxylamino-4,6-
dinitrotoluene (6) with the Meisenheimer dihydride complex of
TNT represented by the overall sum of the five isoforms (3) over
time in aqueous solution as monitored by HPLC. Samples were
diluted with acetonitrile (1:1, vol) prior to analysis in order to
redissolve the partially precipitated secondary diarylamine (10).
The scaled up synthesis aimed for the isolation and in-
depth identification of the new compounds unexpectedly
furnished three instead of only the two main peaks found
previously in the optimized enzymatic and in the chemical
reaction using the authentic standard compounds. The
solution containing the three precipitated condensation
products was filtered, the precipitate dissolved in acetonitrile,
purified by preparative RP-HPLC, and subsequently analyzed
in more detail. The identical molecular mass and the very
similar IR spectra of compounds 9 and 10, respectively,
suggested that compound 10 is an isomer of compound 9,
which was confirmed by its 1H NMR spectrum. Either
compound exhibited two equivalent protons as a singlet,
two coupled allylic protons, but only in compound 10, two
separated methyl protons, indicative for its nonsymmetric
structure. The third new peak in the HPLC chromatogram,
representing compound 7 (SI Figure 3B), was only observed
in the preparative chemical reaction. It eluted first in RP-
HPLC and provided a pseudomolecular mass of 392 with
FIGURE 3. Time course by HPLC analysis of the enzymatic
transformation of 1 by recombinant xenobiotic reductase from
Pseudomonas putida JLR11. Within 3 h the target compound 1 is
transformed to the predominant compounds 3, 9, and 10. The
reaction does not appear to be fully stoichiometric because the
secondary diarylamines and azoxy toluenes (always less than 11%
of corresponding 9 and 10, not shown) started to precipitate after
about 20-30 min of the reaction. Dissolving the reaction mixture
with acetonitrile (1:1 by vol) resulted in their almost stoichiometric
recovery, of at least 89% with regard to eliminated nitrite. Very
small amounts of 5 and 6, as well as the corresponding amines
were also detected (not shown).
The present in-depth study of TNT transformation was
motivated by the observation of the formation of a relatively
large amount of an unexpected brownish-yellow precipitate
at the end of an in vitro reaction with a nitro reductase from
the TNT-converting Pseudomonas putida strain JLR11 which
shares 88% identity with the xenobiotic reductase B (XenB)
of Pseudomonas fluorescens (15). However, monitoring of
this enzyme reaction over several hours by HPLC using
conditions described previously (16), only showed the
production of small quantities of Meisenheimer dihydride
complexes and very minute amounts of azoxy derivatives;
the latter only appearing after several hours. Hydroxyl-amine
derivatives of TNT were hardly detectable. We optimized the
conditions of the reaction by increasing the substrate
concentration by about 9-fold from the initial 0.1 mM by
using TNT-saturated buffer solution as well as TNT-saturated
1
LC-MS, equivalent to a molecular mass of m/z ) 393. H
NMR of compound 7 revealed four aromatic protons as a
singlet and six aliphatic protons corresponding to the methyl
groups, similar to compound 9. However, the FT-IR spectrum
of compound 7 showed a very broad band at 3304 cm-1
whereas the IR spectra of the other two compounds 9 and
10 exhibited significantly smaller bands at 3382 cm-1
,
altogether indicative of a secondary hydroxylamine and two
secondary amines, respectively, and confirming mass spectral
data: the differentiation between these structures was
9 737
/ ENVIRONMENTAL SCIENCE & TECHNOLOGY
VOL. 42, NO. 3, 2008

deduced from results obtained by LC-MS and IR spectro-
metry. Results from analyses by 13C NMR, in which all carbon
signals were unequivocally assigned by HSQC and HMBC
correlations, allowed the definitive assignment of substituents
of all three compounds, which were finally identified as the
symmetric N,N-bis(3,5-dinitrotolyl) hydroxylamine (7), N-(2-
methyl-3,5-dinitrophenyl)-4-methyl-3,5-dinitroaniline (10),
and the symmetric N,N-bis(3,5-dinitrotolyl) amine (9). The
individual detailed analytical data of compounds 7, 9, and
10 were as follows:
(7): MS (HPLC-MS, low resolution): 392 [M-H]-. 1H NMR
(500 MHz, CD3CN, 25 °C, TMS): δ ) 2.47 (s, 6H; CH3), 7.93
(s, 4H; ArH), 8.82 (s, 1H, ) NOH); 13C NMR (125 MHz, CD3CN,
25 °C, TMS): δ ) 14.21 (C-7,7′), 118.74 (C-2,2′,6,6′), 122.25
(C-1,1′), 146.64 (C-4,4′), 152.58 (C-3,3′,5,5′). FT-IR (ν cm-1):
3304 ()NOH very br.), 3095 (Ar-H), 2919, 2851 (-CH3), 2261,
2115 ()N-OH def.), 1623 (-NH def.), 1537, 1349 (NO2 str.);
UV λACN nm, 200 (log ꢀ 4.87).
(9): MS (HPLC-MS, low resolution): 376 [M-H]-. 1H NMR
(500 MHz, CD3CN, 25 °C, TMS): δ ) 2.42 (s, 6H; 2 × CH3),
7.77 (s, 4H; ArH 2,2’6,6′), NH > 9.0 (var.); 13C NMR (125 MHz,
CD3CN, 25 °C, TMS): δ ) 13.95 (C-7,7′), 116.85 (C-2,2′,6,6′),
119.13 (C-1,1′), 141.57 (C-4,4′), 152.83 (C-3,3′,5,5′). FT-IR (ν
cm-1): 3376 (-NH br.), 3089 (Ar-H), 2922, 2853 (-CH3), 1624
(-NH def.), 1537, 1347 (NO2 str.); UV λACN nm, 201 (log ꢀ
4.67), 275 (log ꢀ 4.49).
(10): MS (HPLC-MS, low resolution): 376 [M-H]-. 1HNMR
(500 MHz, CD3CN, 25 °C, TMS): δ ) 2.426 (s, 3H; CH3), 2.432
(s, 3H; CH3), 7.33 (s, )NH), 7.66 (s, 2H; ArH2 and ArH6), 8.26
(d, 1H; ArH6′, J ) 2.23 Hz), 8.32 (d, 1H; ArH4′, J ) 2.23 Hz);
13C NMR (125 MHz, CD3CN, 25 °C, TMS): δ ) 14.23 (C-7),
14.67 (C-7′), 114.41 (C-4′), 116.75 (C-2,6), 118.74 (C-6′), 118.79
(C-1,1′), 132.64 (C-2′), 143.54 (C-4), 143.88 (C-3′), 147.49 (C-
5′), 153.19 (C-3,5). FT-IR (ν cm-1): 3382 (-NH br.), 3091
(Ar-H), 2923, 2852 (-CH3), 1632 (-NH def.), 1535, 1347 (NO2
str.); UV λACN nm, 204 (log ꢀ 4.58), 268 (log ꢀ 4.34).
We consider symmetrical compound 7 as the more stable
intermediate in the formation of compound 9 since we did
not find any clear evidence for an analogous hydroxylamine
precursor (8) of compound 10. The conversion of chemically
generated diaryl hydroxylamines to the corresponding
secondary amines in the presence of oxygen has already been
reported to proceed via nitroxyl radicals as intermediates,
possibly through an unusual disproportioning (24), but we
failed to identify further oxidized products and tentatively
consider the mechanism an autoreduction similar to that of
aromatic hydroxylamines to the corresponding amines. As
already stated above, in the enzymatic reaction with XenB,
mainly products 9 and 10 could be identified (Figure 3) but
with the help of the spectroscopic data (HPLC) of 7 we were
able to identify small amounts of this compound within small
overlapping peaks of yet unidentified minor compounds.
The fact that only very small amounts of secondary hy-
droxylamine could be found in the enzymatic reaction can
be explained by the relatively slower and, therefore, rate-
limiting velocity of the biological reaction. In contrast to the
chemical reduction of TNT with Zn in the presence of NH4Cl,
which generates predominantly (around 80%) 4-hydroxyl-
amino-2,6-dinitrotoluene, the enzymatic reduction of TNT
predominantly furnishes 2-hydroxylamino-4,6-dinitrotolu-
ene, which is reflected by the significantly higher concentra-
tion of the corresponding adduct 10 (data not shown). This
suggests that the condensation reaction to compound 10 is
probably faster compared to that yielding compound 9
through compound 7, or that the reaction exhibits certain
selectivity. On the other hand, HPLC-purified compound 7
dissolved in acetonitrile appears to be slightly unstable since
it was partially (around 35%) transformed into compound 9
after about two months in the dark at room temperature.
Self-reduction represents a common feature of aromatic
hydroxylamines.
RP-HPLC analyses of aqueous solutions of the chemically
synthesized Meisenheimer dihydride complex of TNT over
time only allowed the identification of its isoforms and of
the adducts described here, but never resulted in the
identification of hydroxylamines or other rearomatized
species except after several weeks that of compound 4 on
which we reported recently (19). On the other hand, upon
further reduction with KBH4, the colorless product obtained
from the orange-yellow dihydride complex, yielded pre-
dominantly 4-amino-2,6-dinitrotoluene upon slow rearo-
matization without any detectable release of nitrite (not
shown).
In Figure 2 the self-condensation reaction of compounds
3 and 6 to yield compound 10 under stoichiometric release
of nitrite is shown over time. Similar patterns are obtained
when 3 condenses with 5 to yield 9 (data not shown). Finally,
the prepared amounts of purified compounds 9 and 10
allowed the full quantification for the mass balance of the
enzymatic reaction initiated by the XenB homologue (Figure
3). Results presented in this figure clearly show that also
here the release of nitrite is nearly stoichiometrically related
to the end-products formed. Thus, by knowing the concen-
trations of compounds 9 and 10, the concentrations of
compounds 5 and 6 intercepted by the condensation can be
calculated and the final mass balance of the overall reaction,
which is only initiated enzymatically by this OYE flavin
reductase, correctly computed.
Our earlier observations (18, 19) together with the
aforementioned results now allow us to clearly explain why
some microorganisms can utilize TNT as a sole source of
nitrogen. The results also clearly confirm that nitrite release
from TNT degradation is not enzyme-catalyzed but a
secondary chemical reaction. Furthermore, we have found
that nitrodiarylhydroxylamines and nitrodiarylamines can
be produced in aqueous solvents from Meisenheimer
complexes of TNT and hydroxylaminodinitrotoluenes with-
out the need of aryl halides as starting material and
(transition) metal catalysts. In addition, the identification of
the above compounds may explain why only low concentra-
tions of compounds 5 and 6 (if at all) and the corresponding
amines were detected in biological TNT degradation experi-
ments; they were intercepted (scavenged) by the dimerization
reaction with the reactive dihydride complex of TNT.
Currently, we are characterizing a number of bacterial OYE
reductase gene products with the help of the quantitative
and qualitative data obtained in the present work.
Acknowledgments
This work was supported by grant contract MADOX QLRT-
2001-00345 from the European Union. We thank Dr. Fernan-
do Lafont from the University of Córdoba for molecular mass
determinations by LC-MS and the Unión Española de
Explosivos for providing TNT.
Supporting Information Available
A photograph of an SDS-PAGE gel showing protein purifica-
tion steps of the XenB protein, termed Figure S1. Figure S2
shows a representative separation by HPLC with diode array
detection of metabolites derived from TNT reduction and
the formed condensation products. Figure S3A-D shows ESI-
MS spectra of newly described condensation products from
TNT-derived Meisenheimer dihydride complexes and hy-
droxylaminodinitrotoluenes. This information is available
free of charge via the Internet at http://pubs.acs.org.
Literature Cited
(1) Williams, R. E.; Bruce, N. C. ‘New uses for an old enzyme’–the
Old Yellow Enzyme family of flavoenzymes. Microbiology 2002,
148, 1607–1614.
9 ENVIRONMENTAL SCIENCE & TECHNOLOGY
/ VOL. 42, NO. 3, 2008
738

(2) Williams, R. E.; Rathbone, D. A.; Scrutton, N. S.; Bruce, N. C.
Biotransformation of explosives by the Old Yellow Enzyme family
of flavoproteins. Appl. Environ. Microbiol. 2004, 70, 3566–3574.
(3) Marvin-Sikkema, F. D.; de Bont, J. A. M. Degradation of
nitroaromatic compounds by microorgnisms. Appl. Microbiol.
Biotechnol. 1994, 42, 499–507.
structural basis of reactivity of pentaerythritol tetranitrate
reductase with NADPH, 2-cyclohexenone, nitroesters, and
nitroaromatic explosives. J. Biol. Chem. 2002, 277, 21906–21912.
(15) Blehert, D. S.; Fox, B. G.; Chambliss, G. H. Cloning and sequence
analysis of two Pseudomonas flavoprotein xenobiotic reductases.
J. Bacteriol. 1999, 181, 6254–6263.
(4) Lenke, H.; Knackmuss, H.-J. Initial hydrogenation during
catabolism of picric acid by Rhodococus erythropolis HL 24–2.
Appl. Environ. Microbiol. 1992, 58, 2933–2937.
(5) Vorbeck, C.; Lenke, H.; Fischer, P.; Knackmuss, H.-J. Identifica-
tion of a hydride-Meisenheimer complex as a metabolite of
2,4,6-trinitrotoluene by a Mycobacterium strain. J. Bacteriol.
1994, 176, 932–934.
(6) Vorbeck, C.; Lenke, H.; Fischer, P.; Spain, J. C.; Knackmuss,
H.-J. Initial reductive reaction in aerobic microbial metabolism
of 2,4,6-trinitrotoluene. Appl. Environ. Microbiol. 1998, 64, 246–
252.
(16) Pak, J. W.; Knoke, K. L.; Noguera, D. R.; Fox, B. G.; Chambliss,
G. H. Transformation of 2,4,6-trinitrotoluene by purified xe-
nobiotic reductase B from Pseudomonas fluorescens I-C. Appl.
Environ. Microbiol. 2000, 66, 4742–4750.
(17) Haïdour, A.; Ramos, J. L. Identification of products resulting
from the biological reduction of 2,4,6-trinitrotoluene, 2,4-
dinitrotoluene, and 2,6-dinitrotoluene by Pseudomonas sp.
Environ. Sci. Technol. 1996, 30, 2365–2370.
(18) Esteve-Núñez, A.; Ramos, J. L. Metabolism of 2,4,6-trinitro-
toluene by Pseudomonas sp. JLR11. Environ. Sci. Technol. 1998,
32, 3802–3808.
(7) Hofmann, K. W.; Knackmuss, H.-J.; Heiss, G. Nitrite elimination
and hydrolytic ring cleavage in 2,4,6-trinitrophenol (picric acid)
degradation. Appl. Environ. Microbiol. 2004, 70, 2854–2860.
(8) Heiss, G.; Knackmuss, H.-J. Bioelimination of nitroaromatic
compounds: immobilization versus mineralization. Curr. Opin.
Microbiol. 2002, 5, 282–287.
(9) Weiss, M.; Geyer, R.; Russow, R.; Richnow, H. H.; Kästner, M.
Fate and metabolism of [15N] 2,4,6-trinitrotoluene in soil.
Environ. Toxicol. Chem. 2004, 23, 1852–1860.
(19) González-Pérez, M. M.; van Dillewijn, P.; Wittich, R.-M.; Ramos,
J. L. Escherichia coli has multiple enzymes that attack TNT and
release nitrogen for growth. Environ. Microbiol. 2007, 9, 1535–
1540.
(20) Stenuit, B.; Eyers, L.; Rozenberg, R.; Habib-Jiwan, J.-L.; Agathos,
S. N. Aerobic growth on 2,4,6-trinitrotoluene (TNT) as sole
nitrogen source by Escherichia coli and evidence of TNT
denitration with whole cells and cell-free extracts. Appl. Environ.
Microbiol. 2006, 72, 7945–7948.
(10) Esteve-Núñez, A.; Caballero, A.; Ramos, J. L. Biological degrada-
tion of 2,4,6-trinitrotoluene. Microbiol. Mol. Biol. Rev. 2001, 65,
335–352.
(21) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A
Laboratory Manual, 2nd Ed.; Cold Spring Harbor Laboratory:
Cold Spring Harbor, NY, 1989.
(11) Stenuit, B.; Eyers, L.; El Fantroussi, S.; Agathos, S. N. Promising
strategies for the mineralisation of 2,4,6-trinitrotoluene. Rev.
Environ. Sci. Bio/Technol. 2005, 4, 39–60.
(22) Prak, D. J. L.; O’Sullivan, D. W. Solubility of 2,4-dinitrotoluene
and 2,4,6-trinitrotoluene in seawater. J. Chem. Eng. Data 2006,
51, 448–450.
(12) Symons, Z. C.; Bruce, N. C. Bacterial pathways for degradation
of nitroaromatics. Nat. Prod. Rep. 2006, 23, 845–850.
(13) French, C. E.; Nicklin, S.; Bruce, N. C. Aerobic degradation of
2,4,6-trinitrotoluene by Enterobacter cloacae PB2 and by pen-
taerythritol tetranitrate reductase. Appl. Environ. Microbiol.
1998, 64, 2864–2868.
(23) Agrawal, J. K. Some new high energy materials and their
formulations for specialized applications. Propellants Explos.
Pyrotech. 2005, 30, 316–328.
(24) Yost, Y.; Gutmann, H. R. Carbonyl compounds and secondary
amines from diarylhydroxylamines via nitroxides. J. Org. Chem.
1973, 38, 165–167.
(14) Khan, H.; Harris, R. J.; Barna, T.; Craig, D. H.; Bruce, N. C.;
Munro, A. W.; Moody, P. C. E.; Scrutton, N. S. Kinetic and
ES071449W
9 739
/ ENVIRONMENTAL SCIENCE & TECHNOLOGY
VOL. 42, NO. 3, 2008