Outer-membrane transport of aromatic hydrocarbons
as a first step in biodegradation
Elizabeth M. Hearn, Dimki R. Patel, and Bert van den Berg*
Program in Molecular Medicine, University of Massachusetts Medical School, 373 Plantation Street, Worcester, MA 01605
Edited by Tom A. Rapoport, Harvard Medical School, Boston, MA, and approved April 7, 2008 (received for review February 7, 2008)
Bacterial biodegradation of hydrocarbons, an important process
for environmental remediation, requires the passage of hydropho-
bic substrates across the cell membrane. Here, we report crystal
structures of two outer membrane proteins, Pseudomonas putida
TodX and Ralstonia pickettii TbuX, which have been implicated in
hydrocarbon transport and are part of a subfamily of the FadL fatty
acid transporter family. The structures of TodX and TbuX show
significant differences with those previously determined for Esch-
erichia coli FadL, which may provide an explanation for the sub-
strate-specific transport of TodX and TbuX observed with in vivo
transport assays. The TodX and TbuX structures revealed 14-
stranded ␤-barrels with an N-terminal hatch domain blocking the
barrel interior. A hydrophobic channel with bound detergent
molecules extends from the extracellular surface and is contiguous
with a passageway through the hatch domain, lined by both
hydrophobic and polar or charged residues. The TodX and TbuX
structures support a mechanism for transport of hydrophobic
substrates from the extracellular environment to the periplasm via
a channel through the hatch domain.
membrane protein 兩 x-ray structure
T he monoaromatic hydrocarbons benzene, toluene, ethyl-
benzene, and xylene, referred to as BTEX, pose a signif-
icant human health concern. Human exposure to BTEX can
cause liver and kidney damage, nervous disorders, and repro-
ductive and developmental effects (1– 4). Moreover, benzene
is classified as a human carcinogen (5). Despite their negative
impact on human health, BTEX compounds continue to be
produced for use in the petroleum industry, in plastic and
polymer manufacturing, and as solvents in paints and house-
hold cleaners (1– 4). In 2005, industrial disposal practices in the
United States resulted in ⬎1.5 million lb of BTEX released to
surface waters and land (6). BTEX-contaminated water and
soil environments are remediated primarily through biodeg-
radation by natural microbial populations (7, 8). In particular,
Gram-negative bacteria are known for their biodegradative
abilities (8).
In Gram-negative bacteria, five aerobic pathways for BTEX
biodegradation have been characterized in great detail, both
chemically and genetically (9): the toluene o-monooxygenation
(tom) (10), toluene m-monooxygenation (tbu) (11), toluene
p-monooxygenation (tmo) (12), xylene monooxygenase (xyl)
(13), and toluene dioxygenation (tod) pathways (14). Although
much is known about the intracellular degradation of BTEX,
virtually nothing is known about how such compounds are taken
up by the cell, a prerequisite for their biodegradation. In addition
to the enzymatic genes, the BTEX degradation pathways include
genes encoding outer membrane proteins, which may be re-
quired for hydrocarbon uptake across the outer membrane
(OM). Because the Gram-negative bacterial OM is surrounded
by a hydrophilic lipopolysaccharide layer (15), membrane pro-
teins are required to facilitate transport of hydrophobic com-
pounds across the OM. The involvement of OM proteins in
BTEX uptake was demonstrated in vivo by the enhanced toluene
metabolism upon expression of the Pseudomonas putida F1
TodX (16) and Ralstonia pickettii PKO1 TbuX (17) proteins.
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0801264105
These proteins and the P. putida XylN (18) and Pseudomonas
mendocina TmoX proteins (19) show 40–80% identity to each
other and low (15–25%) sequence identity to the Escherichia coli
long-chain fatty acid transporter FadL [supporting information
(SI) Fig. S1]. The TodX-type proteins form a subfamily of the
FadL transporter family (20), lending support to their proposed
function in hydrocarbon transport.
The E. coli FadL protein is the only well characterized OM
protein known to facilitate the uptake of hydrophobic com-
pounds (21). Recently, two FadL crystal structures were solved
(21), revealing a monomeric 14-stranded ␤-barrel whose in-
terior is blocked by an N-terminal hatch domain. An inward-
pointing kink formed by strand S3 in the ␤-barrel wall con-
BIOCHEMISTRY
strains the N terminus located within the interior of the barrel.
From the FadL structures, a transport model was proposed
that involves the formation of a channel through the hatch
domain. However, no such channel was observed in either
FadL structure. Here, we present the crystal structures of P.
putida TodX and R. pickettii TbuX, the first structures of any
transporters involved in xenobiotics uptake. Both structures
show a continuous passageway leading from the extracellular
surface through the hatch domain, all the way to the periplas-
mic surface. We discuss the implications of these structures on
the mechanism of hydrocarbon transport of TodX family
proteins.
Results and Discussion
Overall Structures. Because the yield of protein expressed within
the OM was low, P. putida TodX was purified from inclusion
bodies and refolded in detergent. X-ray crystal structures of
refolded TodX were obtained in two space groups: orthorhom-
bic (I222; 3.2-Å resolution, one molecule in the asymmetric unit)
and triclinic [P1; 2.6-Å resolution, two molecules in the asym-
metric unit (Table S1)]. Neither molecule in the triclinic struc-
ture shows visible electron density for the N-terminal five
residues, whereas the orthorhombic structure has a complete N
terminus. Both structures are missing electron density for resi-
dues 405–414 of extracellular loop L7 and smaller regions of
loops L1 (residues 72–77), L3 (residues 206–209), and L6
(residues 359–364). The orthorhombic and triclinic forms of
TodX are very similar (Fig. 1A), with an average rms deviation
for the C␣ atoms of 0.9 Å.
R. pickettii TbuX was purified from the membrane fraction
and crystallized in its native form. TbuX crystals were obtained
in the orthorhombic (F222) space group at a resolution of 2.8 Å,
Author contributions: B.v.d.B. designed research; E.M.H. and D.R.P. performed research;
E.M.H. and B.v.d.B. analyzed data; and E.M.H. and B.v.d.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The atomic coordinates have been deposited in the Protein Data Bank,
www.pdb.org [PDB ID codes 3BRZ (TodX I222), 3BS0 (TodX P1), and 3BRY (TbuX)].
*To whom correspondence should be addressed. E-mail: bert.vandenberg@umassmed.edu.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0801264105/DCSupplemental.
© 2008 by The National Academy of Sciences of the USA
PNAS 兩 June 24, 2008 兩 vol. 105 兩 no. 25 兩 8601– 8606
Fig. 1. Structural features of TodX and TbuX. (A) Comparison of triclinic TodX (Left), TbuX (Center) and monoclinic FadL (Right), shown as ribbon
representations. The extracellular loops are shown in purple (L3), light blue (L4), and dark blue (L5). Although the electron density for some of the extracellular
loops in the TbuX structure was weak and could not be modeled accurately, the conformations of the loops are similar to TodX. The region in ␤-strand S3
associated with the kink is colored pink. Detergent molecules are indicated in red (C8E4) and blue (LDAO). The N and C termini are indicated, and the hatch domain
is colored yellow. The approximate position of the OM is indicated by horizontal lines with the extracellular side (E) at the top and the periplasm (P) at the bottom.
(B) Ribbon representations of the N-terminal hatch domains, colored by B-factor, of triclinic and orthorhombic TodX (superimposed, Left), TbuX (Center), and
monoclinic and hexagonal FadL (superimposed, Right). All figures were made with PyMOL (www.pymol.org).

with two molecules in the asymmetric unit (Table S1). Electron
density is weak or absent in parts of extracellular loops L1
(residues 61–66), L3 (residues 206–212), L4 (residues 258–261),
and in the majority of loops L6 (residues 362–376) and L7
(residues 406–418), which could not be modeled. Despite the
incompleteness of the extracellular loops, the overall features of
the native TbuX structure are similar to those of refolded TodX
(average C␣ rms deviation of 4 Å; Fig. 1 A), indicating that the
TodX structure is not likely to be an artifact of refolding.
TodX and TbuX have a 14-stranded ␤-barrel similar to that
observed in the fatty acid transporter FadL (21) (Fig. 1 A). Like
FadL, TodX and TbuX have an N-terminal hatch domain
consisting of three short helices that fold within the barrel
interior. The N-terminal 10–12 residues in the TodX and TbuX
structures have significantly higher B-factors relative to the rest
of the hatch domain (Fig. 1B), suggesting that the N terminus is
flexible. Moreover, the first five residues in the triclinic TodX
structure are not visible in the electron-density maps, presum-
ably because of high mobility. The mobility of the N terminus is
reminiscent of the situation in FadL, where two otherwise
identical crystal structures showed different conformations for
the N-terminal seven residues (21).
Substrate Binding by Aromatic Hydrocarbon Transporters. Within the
extracellular domains of both TodX and TbuX, several tubes of
electron density are visible, which were assigned as C8E4 detergent
molecules used during the later stages of the purification process
(Methods). Two C8E4 molecules, designated as upper and lower,
were observed at slightly different positions in both the orthor-
hombic and triclinic TodX structures. A single C8E4 molecule was
identified in TbuX, in approximately the same position as the lower
C8E4 molecule in TodX (Fig. 1). Inspection of the TodX and TbuX
structures reveals a continuous hydrophobic channel extending
from the extracellular surface of the protein to the N terminus (Fig.
2). On the extracellular surface of the protein, the space between
the two antiparallel ␣-helices of loop L3 forms a hydrophobic cleft,
which contains the upper C8E4 molecule (Fig. 2B). This detergent
molecule is in close proximity (⬍4.5 Å) to residues Ser-171,
Val-173, Thr-177, Leu-182, Val-187, Ala-191, Val-194, and Ala-199
from L3 (Fig. 2). The lower C8E4 molecule is close to residue
Leu-134 in loop L2; residues Leu-162, Leu-164, Leu-166, Leu-168,
Gln-172, and Phe-202 in loop L3; residues Ile-270, Val-272, and
Phe-275 in loop L4; residues Val-313, Val-324, and Leu-326 in loop
L5; and residues Leu-367 and Ile-370 in loop L6 (Fig. 2 A).
Together, these residues form the lining of a pronounced hydro-
phobic channel that extends all the way to the N terminus in the
interior of the ␤-barrel. The hydrophobic nature of the channel is
conserved among the homologues implicated in hydrocarbon trans-
port (Fig. S1). We propose that the hydrophobic channel forms a
conduit for the initial binding and transport of aromatic hydrocar-
bon substrates by members of the TodX subfamily. Interestingly,
inspection of the structures of FadL with those of TodX/TbuX
reveals that the detergent-binding sites are located at similar
positions in the structures, with the upper C8E4 molecule in TodX
at a position analogous to the C8E4 molecule in the low-affinity
binding site of FadL, and the lower TodX C8E4 molecule at a
position analogous to that occupied by an LDAO molecule in the
high-affinity binding site of FadL (ref. 21; Fig. 1).
FadL Family Members Are Substrate-Specific. Because transport
assays involving volatile substrates like BTEX are technically chal-
lenging, we tested the ability of TbuX and TodX to transport
long-chain fatty acids, using an in vivo assay with radiolabeled oleic
acid (22). Although E. coli cells expressing FadL showed a high level
of oleic acid uptake, cells expressing TodX and TbuX were unable
to transport oleic acid, despite the observation that the (outer)
membrane levels of TodX and TbuX are higher than those of FadL
(Fig. 3). These data demonstrate that members of the FadL family
are substrate-specific transporters. Substrate specificity provides an
explanation for the fact that many bacteria have more than one

8602 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0801264105 Hearn et al.

Fig. 2. Detergent binding to TodX. (A) Stereo diagram view showing the locations of bound C8E4 detergent molecules (green, with oxygens in red), designated
as ‘‘upper’’ and ‘‘lower.’’ 2Fo⫺Fc density, contoured at 1.5␴, is shown as an orange mesh. Loop L3 is colored yellow, and the segment of loop L5 that interacts
with L3 is colored magenta. Residues that are located within 4.5-Å distance of the C8E4 molecules are shown as stick models (carbons gray, oxygens red, and
nitrogens blue). (B) Surface view from the extracellular side, rotated ⬇45° relative to A, showing the upper C8E4 molecule bound within the hydrophobic cleft
formed by loop L3. Polar residues are colored salmon, and hydrophobic residues are colored gray.
FadL homolog (for example, Pseudomonas aeruginosa and Vibrio
cholerae each have three). The substrate specificity of FadL homo-
logues is different from that of multidrug efflux pumps, which
display a broad specificity for various hydrophobic compounds (23).
Which structural features underlie the different substrate
specificities of FadL proteins? The answer to this question is not
yet clear, but one of the most obvious possibilities would be
differences within the substrate-binding sites. In FadL, two
positively charged residues (Arg-157 and Lys-317) that likely
interact with the fatty acid carboxylate group (21) are present in
the otherwise completely hydrophobic high-affinity binding
pocket. Sequence and structural alignments of FadL with TodX
and TbuX show that the residues corresponding to Arg-157 and
Lys-317 in FadL are hydrophobic residues in the aromatic
hydrocarbon transporters (Leu-166 and Leu-326 in TodX; Fig.
S1), which is in accord with the uncharged nature of BTEX
substrates. Thus, as a first step in elucidating the basis for
substrate specificity in FadL family members, we mutated both
Fig. 3. TodX and TbuX are substrate specific. Upper shows activity for oleic
acid uptake, normalized to the cell density (OD600), as measured in whole cells
expressing the various proteins. Scale bars represent the average of at least
three independent measurements, and error bars represent the standard
deviation. Representative Western immunoblots indicating expression levels
of the various proteins within the OM (see Methods) are shown in Bottom.
Hearn et al.
Leu-166 and -326 in TodX to arginine and lysine, respectively
(L166R/L326K). Subsequent oleate transport assays showed
very little, if any, transport for the TodX double mutant (Fig. 3),
indicating that these residues alone are not responsible for
substrate specificity. Future more extensive mutagenesis exper-
BIOCHEMISTRY
iments will be required to uncover the structural features
governing the substrate specificity of FadL family members.
Structural Differences Between FadL and Aromatic Hydrocarbon
Transporters. One of the most striking differences between TodX/
TbuX and FadL is the conformation of the extracellular loops L3
and L4 (Fig. 1). In TodX (and TbuX), loop L3 consists of two
antiparallel ␣-helices that lie flat on the top of the barrel (Figs. 1 A
and 2 A), forming a hydrophobic cleft that is occupied by a
detergent molecule. In E. coli FadL, loop L3 also consists of two
antiparallel ␣-helices, but these protrude into the extracellular
environment (Fig. 1 A). Between L3 and L4 in FadL, a hydrophobic
groove is present that serves as an initial low-affinity-binding site for
fatty acids (21). Thus, while the first interaction between FadL and
substrate occurs in a hydrophobic groove between loops L3 and L4,
in TodX and TbuX the hydrophobic cleft is formed solely by loop
L3. The reason for the differences in the arrangement of the L3 and
L4 loops between FadL and TodX/TbuX is unclear but is likely
functionally significant for several reasons. First, very similar con-
formations of L3 and L4 are observed in all three aromatic
hydrocarbon transporter structures. Second, the tip of loop L3
(residues Gly-321–Val-324) forms extensive interactions with loop
L5 (residues Asn-181–Gly-184; Fig. 2 A), suggesting that the con-
formation of L3 is not likely to be transient. Finally, the structures
of several E. coli FadL mutant proteins all show L3 and L4 loops
that are very similar in conformation to that of wild-type FadL (data
not shown), indicating that crystal-packing effects are not likely to
be responsible for the structural differences of L3 and L4 between
FadL and TodX/TbuX.
A second structural difference between TodX/TbuX and
FadL that is likely to be functionally important is the confor-
mation of ␤-strand S3. E. coli FadL has an unusual inward-
pointing kink in strand S3 of the barrel wall (ref. 21; Fig. 1 A).
The kink makes a number of interactions with the N terminus,
thereby likely constraining the mobility and conformational
space accessible to the N terminus (21). In TodX and TbuX, the
kink is much less pronounced (Figs. 1 A and 4). This is most
evident in TodX, where the interstrand hydrogen-bonding pat-
tern between S3 and the neighboring strands S2 and S4 is largely
intact (Fig. 1 A). The S3 kink region in TodX and TbuX may be
flexible, as suggested by the relatively high B-factors for the kink
residues (100–104 in TodX and 102–106 in TbuX) relative to the
neighboring residues in S3 (Fig. S2). The absence of a pro-
nounced kink in TodX/TbuX suggests there may be fewer
PNAS 兩 June 24, 2008 兩 vol. 105 兩 no. 25 兩 8603

Fig. 4. TodX and TbuX have a channel through the hatch domain. (A) Stereo
diagrams viewed from the extracellular side, showing the conformational differ-
ences in the hatch domains (most notably the N terminus and helix 1) and the S3
kink region for TodX (P1 crystal form, red), TbuX (blue), and both crystal forms of
FadL (white). The superposition (showing the barrels of only FadL and TodX for
clarity) was made in Coot (34). The position of the hatch channel is indicated by
the asterisk. (B) Stereo cutaway surface view from the side, showing the TodX
hatch channel as a continuous black tube. The upper and lower C8E4 detergent
molecules are shown as orange stick models (with oxygens in red). The residues
at the narrowest constriction of the channel (Tyr-9, Gln-83, and Phe-100) are
shown in green as stick models (oxygens, red and nitrogens, blue). The hatch
domain is shown in pink and the signature NPA sequence in yellow.
interactions of this segment with the N terminus; as a result, we
speculate that the N terminus is less constrained in the aromatic
hydrocarbon transporters compared with the N terminus in E.
coli FadL. This is borne out by the fact that the N-terminal ⬇10
residues in TodX and TbuX have high B-factors or do not show
electron density at all (Fig. 1B).
Mechanism of Substrate Transport. Using the two crystal struc-
tures of E. coli FadL (21), a transport model was proposed in
which hydrophobic substrates are captured from the extracel-
lular medium by a low-affinity-binding groove, followed by
diffusion into an adjacent high-affinity-binding site. Subse-
quently, spontaneous conformational changes of the N termi-
nus, postulated from the observation of different conforma-
tions of the N terminus in both FadL structures, result in
substrate release from the high-affinity-binding site. As the
final and most crucial step of the transport model, further
hypothetical conformational changes in the hatch and possibly
the S3 kink were proposed that would open a channel for
substrate diffusion through the hatch domain into the
periplasm (21).
The hatch domains, consisting of three short helices connected by
short loops, have very similar folds in FadL and TodX/TbuX (Fig.
4A). However, the positions of the hatch domains within the barrel
are clearly different between FadL and TodX/TbuX, with shifts
8604 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0801264105
occurring for several of the helices and loops, most notably for the
N terminus and helix 1 (residues 11–15; Fig. 4A). These shifts,
together with the presence of a much less prominent kink in strand
S3, result in the presence of a channel through the hatch in TodX
and TbuX. The presence of the hatch channel generates a contin-
uous substrate passageway that leads from the extracellular sub-
strate-binding site, via the hydrophobic channel (Fig. 2), to the
periplasm (Fig. 4B). The hatch channel is located at identical
positions in TodX and TbuX, on the opposite side from the
absolutely conserved NPA sequence (Fig. S2). The channel is lined
by (TodX) residues 5, 8–12, 14 (N terminus, hatch helix 1), 24–27
(hatch helix 2), 45, 47, 51 (strand S1), 83 (strand S2), 100, 102–105
(S3 kink), 132, 134, 136 (strand S4), 388, 389, 391 (strand S13), and
426, 428, 430, 432 (strand S14; C terminus; Fig. S3). The narrowest
point (⬇4.5–5-Å diameter; atom-center to atom-center distance) of
the channel is close to residues Gln-83, Phe-100, and Tyr-9 (Fig.
4B). The constriction site observed in the TodX and TbuX struc-
tures may be large enough to allow passage of substrate molecules;
manual fitting of benzene or toluene within the constriction is
possible without having any interatom distances ⬍2.5 Å (atom-
center to atom-center). In marked contrast with the exclusively
hydrophobic character of the initial substrate passageway (Fig. 2),
the hatch channel in TodX/TbuX is lined by both hydrophobic and
polar and charged residues (Fig. S4), some of which are not highly
conserved among xenobiotic transporters (Fig. S1). We speculate
that the more polar character of the hatch channel provides
substrate-binding sites of lower affinity compared with those in the
hydrophobic channel, which prevent the substrate from stalling in
the channel during diffusion toward the periplasm. Thus, the TodX
and TbuX structures support the transport model derived from the
E. coli FadL structures, with spontaneous conformational changes
within the hatch domain and the S3 kink opening a channel for
substrate diffusion through the hatch. This hypothesis is supported
by the fact that removal of the N-terminal 12 residues from the E.
coli FadL structures results in a continuous channel through the
FadL hatch domain (Fig. S5).
The presence of a continuous channel through TodX and TbuX
might be expected to generate a channel for conductance of small
ions in planar lipid bilayer experiments, analogous to what has been
observed for many other OM channels. Preliminary experiments
with both TbuX and FadL, however, did not show any conductance
(data not shown). We speculate that the absence of measurable ion
conductance in FadL family members may be due to the hydro-
phobic character of the initial substrate passageway (Fig. 2), which
may preclude the passage of ions.
Based on the new crystal structures of TodX and TbuX, we
propose the following general model for transport of BTEX
hydrocarbons across the OM, representing the essential first step in
the biodegradation of these compounds: (i) substrate capture from
the extracellular environment by the hydrophobic cleft in extracel-
lular loop L3; (ii) movement of the substrate through the hydro-
phobic channel to a position close to the N terminus; (iii) confor-
mational changes in the S3 kink and the hatch domain, resulting in
the opening of a continuous channel through the hatch; and (iv)
diffusion through the hatch channel and substrate release into the
periplasm. Upon emergence from the hatch channel, the hydro-
carbons must diffuse through the periplasm and cross the inner
membrane to be accessible to the cytosolic mono- or dioxygenase
enzymes. Because BTEX hydrocarbons are relatively water-soluble
(the aqueous solubility of toluene is ⬇6 mM; ref. 24), they are likely
to enter and diffuse through the periplasm without the aid of
periplasmic-binding proteins. Crossing the inner membrane is likely
to occur by spontaneous diffusion, and, like OM transport, is driven
by biodegradation occurring in the cytoplasm, providing a sink.
The putative substrate diffusion channel observed in TodX and
TbuX is unique among OM proteins for several reasons. First, the
distance from the initial substrate-binding site on the extracellular
surface (Fig. 2) to the periplasmic exit of the hatch channel is ⬎50
Hearn et al.
Å. Along much of this length, the channel diameter does not exceed
⬇10 Å. This long and narrow channel is very different from those
present in other OM proteins, such as porins (OmpC, OmpF) and
substrate-specific channels like LamB, which typically have short
constrictions (often called ‘‘eyelets’’) with lengths of ⬇5–10 Å (15).
Second, the TodX/TbuX channels consist of an initial, very hydro-
phobic part (Fig. 2) and a more polar part contributed by the hatch
channel (Fig. S4). All other known OM channels have entirely polar
passageways, in accordance with the polar nature of their
substrates.
Besides the FadL family, the only other OM proteins in which the
barrel lumen is occluded by hatch domains are the TonB-dependent
receptors (TBDRs), which transport large (⬎800 Da) hydrophilic
molecules such as vitamin B12 and iron–siderophore complexes
(25). TBDRs have much larger barrels (22 ␤-strands) than FadL
family members (14 ␤-strands) and consequently much larger hatch
domains comprising ⬇150–200 residues. TBDRs bind their sub-
strates with high (nanomolar) affinities, require exogenous energy
input for transport provided by the proton motive force, and require
the inner membrane protein TonB. TonB interacts with the hatch
domain of TBDRs, likely resulting in local unfolding of the hatch,
generating a transient wide passageway through the hatch for
transport of the large substrates (26). This elaborate mechanism
may have evolved because a permanently open pore of a size
TodX and TbuX proteins, membrane protein purification was performed as
described (21). TodX protein expressed from plasmid pTodX(ib) was purified from
inclusion bodies and refolded as follows. Cells were harvested, resuspended in
TSB buffer [20 mM Tris, 300 mM NaCl, 10% (vol/vol) glycerol, pH 8] and ruptured
by passage at 15,000 –20,000 psi through a microfluidizer (Avestin). Inclusion
bodies and cell membranes were collected by centrifugation at 17,000 ⫻ g,
washed in 10 mM Tris (pH 8), 50 mM NaCl, and resuspended in 10 mM Tris (pH 8),
50 mM NaCl containing 8 M urea. The suspension was stirred for 2 h at room
temperature. After centrifugation (46,000 ⫻ g), the supernatant containing
urea-denatured protein was slowly diluted 10-fold in 1% (wt/vol) N-lauroylsar-
kosine (sarkosyl) in TSB. Protein folding was achieved by stirring the sarkosyl
solution at 4°C for 18 h. Folded TodX protein was purified by nickel affinity and
gel filtration chromatography as described for FadL (21). Success of the sarkosyl
folding was confirmed by observing the heat-modifiability on SDS/PAGE gels (32);
before boiling, TodX migrates on SDS/PAGE gels with an apparent molecular
mass of ⬇28 kDa, and after boiling, the migration shifts to ⬇46 kDa (data not
shown).
Crystallization. Crystallization trials were set up by using the hanging-drop
technique with commercially available crystallization screens (The Classics and
MB Class II screens, Qiagen). For TodX, the best diffracting crystals (native and
selenomethionine) were obtained at 22°C in 25% (wt/vol) PEG4000, 0.2 M am-
monium sulfate, 0.1 M sodium acetate (pH 4.6). TbuX crystallized at 22°C in 20%
(wt/vol) PEG4000, 0.1 M magnesium acetate, 0.05 M sodium acetate (pH 4.5).
Crystals were flash-frozen (100 K) in reservoir solution containing C8E4 and 20%
(vol/vol) glycerol by plunging in liquid nitrogen.

required for passage of TBDR substrates would likely be harmful
for the cell. By contrast, the substrates of FadL family members are
small, removing the need for formation of a large channel through
the hatch. Moreover, substrates likely bind to FadL proteins with
relatively low affinities (27), allowing small and spontaneous con-
formational changes to suffice for transport via passive diffusion.
The current structures of TodX and TbuX support and extend
the general transport model for hydrophobic molecules that was
proposed previously, based on structures of E. coli FadL (21). The
structural data provide a solid framework to test the proposed
transport mechanism by site-directed mutagenesis experiments, in
combination with in vivo and in vitro substrate-binding and trans-
port assays. Such experiments will also shed light on the structural
determinants of substrate specificity of FadL family members. A
detailed understanding of how FadL family members transport
their hydrophobic substrates not only is of fundamental importance
but could also lead to the design of novel hydrophobic drugs and
more efficient biodegrading bacterial strains.
Methods
Vector Construction. Signal sequence cleavage sites were predicted to occur
between residues 20 and 21 for P. putida F1 TodX and residues 23 and 24 for R.
pickettii PKO1 by using the SignalP program (28). For expression and OM local-
ization of TodX and TbuX, the mature todX and tbuX genes (lacking the endog-
enous signal sequences and the first two residues, and modified with a C-terminal
hexahistidine tag) were amplified by PCR from P. putida F1 (American Type
Culture Collection 700007) genomic DNA and from R. pickettii PKO1. The mature
gene fragments were cloned in-frame with the E. coli FadL signal sequence (plus
the first two residues of the mature FadL sequence) into the pBAD22 vector (29).
The resulting plasmids were designated pTodX(native) and pTbuX(native). For
expression of TodX in inclusion bodies, plasmid pTodX(ib) was constructed by
cloning the mature TodX gene sequence (corresponding to residues 22– 453),
with a C-terminal hexahistidine tag, downstream of the arabinose-inducible
promoter in pBAD22 (29).
Protein Expression and Purification. Proteins were overexpressed in E. coli
C43(DE3) (30) by induction with 0.2% (wt/vol) arabinose in 2⫻YT medium for 3 h
at 37°C. Selenomethionine-substituted proteins were produced in E. coli
C43(DE3) by inhibition of the methionine biosynthesis pathway (31). For native
BIOCHEMISTRY
Data Collection and Structure Determination. Datasets for the TodX and TbuX
crystals were collected on beamline X6A at the NSLS (Brookhaven National
Laboratory, Upton, NY) and processed using HKL2000 (33). For selenome-
thionine-substituted TodX, diffraction data were collected at the selenium
peak wavelength. Selenium sites were identified using SOLVE (34) and
refined using SHARP (35). This structure was used to solve the structure of
the native TodX crystal by molecular replacement using PHASER (36). For all
structures, model building was done manually using Coot (37), and
refinement was done with CNS (38).
For TbuX, an initial model was obtained from a dataset collected from a
selenomethionine-substituted TbuX crystal that diffracted to 3.2 Å at the sele-
nium peak wavelength. A partial model was built and used to solve the structure
of a native TbuX crystal (space group F222) by molecular replacement using
PHASER (36). Model building was done manually using Coot (37), and refinement
was done with CNS (38).
TodX and TbuX Activity Assays Toward Fatty Acid. Plasmids pBAD22, pFadL (21),
pTodX(native), pTbuX(native), and pTsx (39) were transformed into E. coli LS6164
(⌬fadR ⌬fadL) cells (40). The TodX L166R/L326K mutant was constructed using the
QuikChange Site-directed Mutagenesis kit (Stratagene). Transport assays with
[3H-9,10]-oleic acid (specific activity 40 Ci mmol⫺1; Sigma) were performed in E.
coli LS6164 (⌬fadR ⌬fadL) cells as described by Kumar and Black (22), except cells
were induced with 0.001% (wt/vol) arabinose for 1 h at 37°C, and chloramphen-
icol was omitted from the buffers. To determine the presence of FadL, TodX,
TbuX, or Tsx proteins in the OM, cells were incubated with BugBuster (Novagen)
while shaking for 20 min at 25°C and centrifuged for 10 min at 10,000 ⫻ g to
separate solubilized proteins and inclusion bodies. The soluble and membrane
proteins were subjected to western immunoblotting using an anti-His antibody
(Qiagen).
ACKNOWLEDGMENTS. We thank V. Stojanoff, M. Allaire, J. Jakoncic, and F.
Yokaichiya (Beamline X6A, National Synchrotron Light Source) for beamtime
and beamline support. We thank J. Kukor (Rutgers, The State University of
New Jersey, Newark) for providing R. pickettii PKO1, and L. Movileanu and A.
Wolfe (Syracuse University, Syracuse, NY) for planar lipid bilayer experiments.
We are grateful to P. Black and C. Petteys (Ordway Research Institute, Albany,
NY; and Albany Medical College, Albany, NY) for strain D10 and for technical
advice on the fatty acid transport assays. This work was supported by the
National Institutes of Health (Training Grant GM079820, to E.M.H.) and by a
National Institutes of Health R01 Research Project Grant (GM074824, to
B.v.d.B.).

1. Agency for Toxic Substances and Disease
Benzene (US Department of Health and
Atlanta).
2. Agency for Toxic Substances and Disease
Toluene (US Department of Health and
Atlanta).
Registry (2007) Toxicological Profile for
Human Services, Public Health Service,
Registry (2000) Toxicological Profile for
Human Services, Public Health Service,
3. Agency for Toxic Substances and Disease Registry (2007) Toxicological Profile for
Ethylbenzene (US Department of Health and Human Services, Public Health Service,
Atlanta).
4. Agency for Toxic Substances and Disease Registry (2007) Toxicological Profile for
Xylene (US Department of Health and Human Services, Public Health Service,
Atlanta).

Hearn et al. PNAS 兩 June 24, 2008 兩 vol. 105 兩 no. 25 兩 8605
 5. National Toxicology Program (2005) in Report on Carcinogens (US Department of
Health and Human Services, Public Health Service, Atlanta, GA), 11th Ed, pp 26 –27.
6. Environmental Protection Agency (2007) Toxics Release Inventory (US Environmental
Protection Agency, Washington, DC).
7. Head IM, Jones DM, Röling WFM (2006) Marine microorganisms make a meal of oil. Nat
Rev Microbiol 4:163–182.
8. Leahy JG, Colwell RR (1990) Microbial degradation of hydrocarbons in the environ-
ment. Microbiol Rev 54:305–315.
9. Ellis LB, Hou BK, Kang W, Wackett LP (2003) The University of Minnesota biocatalysis/
biodegradation database: Post-genomic data mining. Nucleic Acids Res 31:262–265.
10. Newman LM, Wackett LP (1995) Purification and characterization of toluene 2-mono-
oxygenase from Burkholderia cepacia G4. Biochemistry 34:14066 –14076.
11. Olsen RH, Kukor JJ, Kaphammer B (1994) A novel toluene-3-monooxygenase pathway
cloned from Pseudomonas pickettii PKO1. J Bacteriol 176:3749 –3756.
12. Yen, K-M, et al. (1991) Cloning and characterization of a Pseudomonas mendocina KR1
gene cluster encoding toluene-4-monooxygenase. J Bacteriol 173:5315–5327.
13. Williams PA, Murray K (1974) Metabolism of benzoate and the methylbenzoates by
Pseudomonas putida (arvilla) mt-2: Evidence for the existence of a TOL plasmid. J
Bacteriol 120:416 – 423.
14. Zylstra GJ, Gibson DT (1989) Toluene degradation by Pseudomonas putida F1: nucle-
otide sequence of the todC1C2BADE genes and their expression in Escherichia coli.
J Biol Chem 264:14940 –14946.
15. Nikaido H (2003) Molecular basis of outer membrane permeability revisited. Microbiol
Mol Biol Rev 67:593– 656.
16. Wang Y, et al. (1995) Identification of a membrane protein and truncated LysR-type
regulator associated with the toluene degradation pathway in Pseudomonas putida
F1. Mol Gen Genet 246:570 –579.
17. Kahng HY, Byrne AM, Olsen RH, Kukor JJ (2000) Characterization and role of tbuX in
utilization of toluene by Ralstonia pickettii PKO1. J Bacteriol 182:1232–1242.
18. Kasai Y, Inoue J, Harayama S (2001) The TOL plasmid pWW0 xylN gene product from
Pseudomonas putida is involved in m-xylene uptake. J Bacteriol 183:6662– 6666.
19. Ramos-González MI, et al. (2002) Cross-regulation between a novel two-component
signal transduction system for catabolism of toluene in Pseudomonas mendocina and
the TodST system from Pseudomonas putida. J Bacteriol 184:7062–7067.
20. Saier MH, Jr, Tran CV, Barabote RD (2006) TCDB: The transport classification database
for membrane transport protein analyses and information. Nucleic Acids Res 34:D181–
D184.
21. van den Berg B, Black PN, Clemons WM, Jr, Rapoport TA (2004) Crystal structure of the
long-chain fatty acid transporter FadL. Science 304:1506 –1509.
22. Kumar GB, Black PN (1991) Linker mutagenesis of a bacterial fatty acid transport
protein: identification of domains with functional importance. J Biol Chem 266:1348 –
1353.
8606 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0801264105
23. Higgins CF (2007) Multiple molecular mechanisms for multidrug resistance transport-
ers. Nature 446:749 –757.
24. Eastcott L, Shiu WY, Mackay D (1988) Environmentally relevant physical-chemical
properties of hydrocarbons: a review of data and development of simple correlations.
Oil Chem Pollut 4:191–216.
25. Ferguson AD, Deisenhofer J (2002) TonB-dependent receptors–structural perspectives.
Biochim Biophys Acta 1565:318 –322.
26. Shultis DD, Purdy MD, Banchs CN, Wiener MC (2006) Outer membrane active transport:
Structure of the BtuB:TonB complex. Science 312:1396 –1399.
27. Black PN (1990) Characterization of FadL-specific fatty acid binding in Escherichia coli.
Biochim Biophys Acta 1046:97–105.
28. Bendtsen JD, Nielsen H, von Heijne G, Brunak S (2004) Improved prediction of signal
peptides: SignalP 3.0. J Mol Biol 340:783–795.
29. Guzman LM, Belin D, Carson MJ, Beckwith J (1995) Tight regulation, modulation and
high-level expression by vectors containing the arabinose PBAD promoter. J Bacteriol
177:4121– 4130.
30. Miroux B, Walker JE (1996) Over-production of proteins in Escherichia coli: mutant
hosts that allow synthesis of some membrane proteins and globular proteins at high
levels. J Mol Biol 260:289 –298.
31. van Duyne GD, Standaert RF, Karplus PA, Schreiber SL, Clardy J (1993) Atomic structures
of the human immunophilin FKBP-12 complexes with FK506 and rapamycin. J Mol Biol
74:105–124.
32. Black PN (1991) Primary sequence of the Escherichia coli fadL gene encoding an outer
membrane protein for long-chain fatty acid transport. J Bacteriol 173:435– 442.
33. Otwinowski Z, Minor W (1997) Processing of X-ray diffraction data collected in oscil-
lation mode. Methods Enzymol 276:307–3261997.
34. Terwilleger TC, Berendzen J (1997) Automated MAD and MIR structure solution. Acta
Crystallogr D55:849 – 851.
35. de La Fortelle E, Bricogne G (1997) Maximum-likelihood heavy-atom parameter re-
finement and phasing program for multiple isomorphous replacement and multi-
wavelength anomalous dispersion MAD methods. Methods Enzymol 276:472– 494.
36. McCoy AJ, Grosse-Kunstleve RW, Storoni LC, Read RJ (2005) Likelihood-enhanced fast
translation functions. Acta Crystallogr D61:458 – 464.
37. Emsley P, Cowtan K (2004) Coot: Model-building tools for molecular graphics. Acta
Crystallogr D60:2126 –2132.
38. Brunger AT, et al. (1998) Crystallography and NMR system: A new software suite for
macromolecular structure determination. Acta Crystallogr D54:905–921.
39. Ye J, van den Berg B (2004) Crystal structure of the bacterial nucleoside transporter Tsx.
EMBO J 23:3187–3195.
40. Ginsburgh CL, Black PN, Nunn WD (1984) Transport of long-chain fatty acids in
Escherichia coli: Identification of a membrane protein associated with the fadL gene.
J Biol Chem 259:8437– 8443.
Hearn et al.