Angeletti, Pomezia Roma, Italy; 3 Dipartimento di Scienze Biochimiche ƠRossi FanelliÕ, Universita'di Roma ƠLa SapienzaÕ, Roma, Italy Ligand binding by the aryl hydrocarbon receptor AhR,
Trang 1A model for recognition of polychlorinated dibenzo- p -dioxins
by the aryl hydrocarbon receptor
M Procopio1, A Lahm2, A Tramontano3, L Bonati1and D Pitea1
1 Dipartimento di Scienze dellÕAmbiente e del Territorio, UniversitaÁ degli Studi di Milano-Bicocca, Milano, Italy;
2 Istituto di Ricerche di Biologia Molecolare P Angeletti, Pomezia (Roma), Italy;
3 Dipartimento di Scienze Biochimiche ƠRossi FanelliÕ, Universita'di Roma ƠLa SapienzaÕ, Roma, Italy
Ligand binding by the aryl hydrocarbon receptor (AhR), a
member of the bHLH-PAS family of transcriptional
reg-ulatory proteins, has been mapped to a region within the
second ƠPASÕ domain, a conserved sequence motif ®rst
discovered in the Per-ARNT-Sim family of proteins In
addition to the bacterial photoactive yellow protein (PYP),
which had been proposed as a structural prototype for the
three dimensional fold of PAS domains, two crystal
structures of the PAS domain have recently been
deter-mined: the human potassium channel HERG and the
heme binding domain of the bacterial O2 sensing FixL
protein The three structures reveal a highly conserved
structural framework in evolutionary rather distant PAS
domains, provide a more general view of how these domains can recognize their ligands and suggest a structure±function relationship that we exploited to build a three-dimensional model of the ligand binding domain (LBD) of the mouse aryl hydrocarbon receptor (mAhR) The model allowed us to putatively identify the residues responsible for the recognition of polychlorinated dibenzo-p-dioxins (PCDDs) by AhR receptors and to formulate an hypothesis on the signal transduction mechanism Keywords: aryl hydrocarbon receptor; polychlorinated dibenzo-p-dioxins; structure prediction; protein modelling; molecular recognition
Studies on the biological mechanism of action of
polychlo-rinated dibenzo-p-dioxins (PCDDs) indicate that their
biological effects are mediated by binding to a speci®c
cytoplasmic protein, the aryl hydrocarbon receptor (AhR)
[1]
Ligand-induced activation of AhR initiates a process
whereby the receptor is transformed into a nuclear
transcription factor by forming a complex with the protein
ARNT (Ah receptor nuclear translocator) Speci®c
recog-nition of XRE DNA sequences (xenobiotic responsive
elements) by the ligand-activated AhR/ARNT heterodimer
then induces transcription of genes encoding xenobiotic
metabolizing enzymes [2] Understanding the PCDD±AhR
binding process at a molecular level is therefore a key step
for gaining insight into the biological mechanism of action
of these compounds
The structure±activity relationship (SAR) of the PCDD±
AhR interaction has been studied with the aim of
correlat-ing physico-chemical properties of the ligands and their biological activities [3±5] In particular, we analysed a series
of PCDDs with varying binding af®nities [4,5] on the basis
of their molecular electrostatic potential (MEP) and molec-ular polarizability and concluded that the requirements for high af®nities are the concentration of negative MEP values
at the extremes of the ligand's long axis and a depleted charge above and below the aromatic rings This led to the hypothesis that there are favorable interactions with a receptor nucleophilic site in the central part of the ligand and with electrophilic sites at both sides of the principal molecular axis A necessary next step to understand the PCDD±AhR interaction and to identify the amino-acid residues directly interacting with PCDDs is the construction
of a three-dimensional model for the AhR ligand binding domain (LBD)
AhR and ARNT belong to the Per-ARNT-Sim (PAS) family of proteins [6,7], whose members act as transcrip-tional activators, sensor modules of two-component regu-latory systems or as ion channels [8] PAS domains are found predominantly in proteins that are involved, directly
or indirectly, in signal transduction Their known functions are in some cases to mediate protein±protein interactions and, in other cases, such as for AhR, ligand and/or cofactor binding [8]
In AhR, two PAS domains are present in a 270-residue region encompassing two imperfect repeats of 110 amino acids (PAS-A and PAS-B) separated by a sequence of 50 amino acids A minimal LBD was mapped in the mouse AhR (mAhR) between amino acids 230 and 397, the region that encompasses the PAS-B repeat [9] While deletion of the PAS-A repeat (amino acids 121±182) reduced ligand binding only to 30%, deletion of the PAS-B repeat (amino acids 259±374) completely abolished binding, as did deletion
Correspondence to A Tramontano, Department of Biochemical
Sciences ƠRossi FanelliÕ, University of Rome ƠLa SapienzaP.le Aldo
Moro, 500185 Rome Fax: + 39 06 91093482,
Tel.: + 39 06 91093207, E-mail: Anna.Tramontano@uniroma1.it
Abbreviations: AhR, aryl hydrocarbon receptor; PYP, photoactive
yellow protein; LBD, ligand binding domain; mAhR, mouse aryl
hydrocarbon receptor; PCDD, polychlorinated dibenzo-p-dioxin;
ARNT, Ah receptor nuclear translocator; XRE, xenobiotic responsive
elements; SAR, structure±activity relationship; MEP, molecular
electrostatic potential; PAS, Per-ARNT-Sim protein family; TCDD,
tetrachlorodibenzo-p-dioxin.
(Received 10 August 2001, revised 20 September 2001, accepted 16
October 2001)
Trang 2of the complete PAS region [9] In the same study it was
already shown that modi®cations outside the PAS domain
had no effect on ligand binding
A structural prototype for the three-dimensional fold of
the PAS domain superfamily has been proposed to be the
structure of the photoactive yellow protein (PYP), a
bacterial light-sensing protein [10] However, the crystal
structures of two other PAS domains [11,12] have been
recently determined and their analysis allowed us to build a
three-dimensional model of the mAhR LBD and to
investigate its ligand binding site at the molecular level
R E S U L T S A N D D I S C U S S I O N
Structure prediction
Application of a recursive PSI-BLAST [13] search (default
parameters) against the nonredundant protein sequence
database revealed a high number of matches between the
mAhR LBD and many other PAS proteins, including
hypoxia-inducible factor 1a, several histidine kinases, light
receptors, regulatory proteins, clock proteins (such as the
period clock protein PER), sensor proteins (oxygen/redox
sensors) and ion channels The crystal structures of the PAS
domains of two of these proteins were recently solved: the
human potassium channel HERG [11] and the heme
binding domain of the bacterial O2sensing FixL protein
[12] Both structures were detected after four (HERG) or
eight (FixL) PSI-BLAST iteration cycles, as was the PYP
protein (iteration 6) Although E values were initially rather
high (> 0.1) for all three structures, E values for HERG
and FixL became highly signi®cant (< 10)4) as the search
progressed A search including only the database of known
protein structures neither found any of these structures, nor
highlighted any other statistically signi®cant homologies
The structures of the HERG and FixL PAS domains are
shown in Fig 1 together with the PYP structure Given the
low level of sequence identity [15], the high structural conservation is quite unexpected: all the structures are formed by a ®ve-stranded antiparallel b sheet with a helices
on one side Although all three domains belong to proteins involved in signalling processes and are expected to transmit
a signal through protein±protein interactions, they have developed quite different mechanisms to perform their function While the HERG PAS domain does not bind a ligand [11], both the FixL PAS domain and PYP are activated by ligands; in FixL, oxygen binding at the heme binding PAS domain controls the activity of a histidine kinase domain [12]; and in PYP, a local conformational change occurs once the p-hydroxycinnamoyl cromophore is bound [10]
The largest conformational difference between the FixL and the HERG and PYP structures occurs in the so-called helical connector, the long central helix, which shows a translational displacement of 7 AÊ that allows the accom-modation of the heme cofactor (Fig 1C) [12] The hydro-phobic core of the three domains is generally well conserved, but two buried residues in FixL differ signi®cantly in size from the structurally equivalent residues in PYP and HERG, again favoring the heme binding These are glycines
224 and 251 that substitute Phe96 and Val120 in PYP, and Phe98 and Leu127 in HERG [16]
For both FixL and PYP, structures are known for the inactive and active signaling states In the case of PYP, conformational changes occur in the neighborhood of the p-hydroxycinnamoyl cromophore and are transmitted to the surface of the protein primarily through the cromophore and Arg52 [10] In FixL, the heme propionate groups are suggested to relay the spin transition signal by transducing the increased planarity of the puckered porphyrin ring into backbone and side-chain conformational changes within a loop (residues 211±215) immediately following the helical connector [12] The suggested signal transducing regions of PYP and FixL are thus located at the opposite ends of the
Fig 1 Schematic representation of the HERG (A), PYP (B), and FixL (C) PAS domains displaying the high degree of structural similarity The largest shift amongst the conserved secondary element position occurs in FixL due to the presence of the large heme cofactor Secondary structure elements are colored blue (strands) and red (helices), cofactor ligands green (A), (B) and (C) were generated using RIBBONS [28] Coordinate sets used correspond to entries 1BV5(FixL) [12], 2PYP(PYP) [10] and 1BYW(HERG) [11] of the PDB protein data bank [14].
Trang 3long central helix, highlighting the importance of this region
and the ¯anking loops as the critical regulatory region of the
PAS domain family [16], with the remainder of the PAS fold
serving as a structural scaffold
When sequence similarity between the target and
potential template sequences is low, as in our case, the
correctness of the alignment plays a crucial role both in the
selection of the correct template and in the expected ®nal
quality of the model Other information, such as predicted
and observed secondary structures of the target and
template proteins, and sequence and structure conservation
in their families, should therefore be used to re®ne the
alignment
The consensus secondary structure for residues 230±397
of the mAhR LBD as predicted by theJPREDserver [17,18] is
reported in Fig 2 together with the observed secondary
structure of the three template candidates The ®nal
sequence alignment used for modelling is reported in the
same ®gure This sequence alignment differs somehow from
PAS domain alignment recently proposed [8], as it was produced manually taking into account the predicted secondary structure of AhR LBD, the observed secondary structure and FSSP structural alignment for FixL, HERG and PYP, and the conservation pattern in a multiple alignment of AhR sequences For clarity, in Fig 2 we only show some of these latter sequences and, for comparison, a subset of sequences from the related ARNT proteins The AhR sequences in Fig 2 were selected for their differences in the response to PCDDs: the human Ah receptor that has an af®nity for 2,3,7,8-TCDD sixfold lower than mAhR [19]; the AhR-1 ortholog of Caenorhabditis elegans (AhR-1C.E.), neither photoaf®nity labeled by a dioxin analog, nor activated by b-naphto¯avone in a yeast system [20]; the rainbow trout AhRa that binds TCDD [21] and the Microgadus Tomcod AhR also activated by TCDD [22]
All alignments were manipulated using the interactive display programSEAVIEW[23]
Fig 2 Alignment of Ah receptors and their predicted secondary structure against the three structural templates aligned according to FSSP a Helices and b strands are represented as white and black bars, respectively Secondary structure assignment for FixL, PYP and HERG is derived from the PDB entries Colouring scheme for residues: red: acidic; blue: basic; purple: polar; yellow: Cys; brown: aromatic; green: hydrophobic; orange: Ser,Thr; grey: Pro; white: Gly.
Trang 4Because of the closer functional homology (noncovalent
interaction with a ligand) we used FixL as a template for
modelling This choice was also motivated by the
observa-tion that the helical connector in FixL is translated away
from the b sheet with respect to HERG and PYP (Fig 1)
thus allowing binding of the heme-ligand, a situation
expected to be present in a similar fashion also in AhR
The sequence corresponding to mAhR residues 275±380
was therefore inscribed onto the structural template
pro-vided by FixL according to the alignment shown in Fig 2,
and, subsequently, the necessary insertions and deletions were modeled (Fig 3A) AhR residues 381±397 were not modeled because the corresponding helix in FixL is pointing away from the barrel and should not be involved in ligand binding AhR residues 286±288 could be modeled using the corresponding loop of equal length from the HERG structure, while all other insertions and deletions were constructed using a fragment database search procedure [24] The one-residue insertion at position 314 correlates well with the presence of a hydrophilic residue at the spatially close position 282 replacing the buried hydrophobic Ile present in FixL
Fig 3 Comparison between the mAhR model (A,C) and the parental FixL PAS domain structure (B,D) In (A) and (B) residues that in¯uence the size
of the ligand pocket are highlighted The arrow indicates the shift of the helical connector (orange) in AhR with respect to FixL Loop regions where insertions or deletions had to be accomodated in the AhR model are coloured magenta In (C) and (D) a close-up of the AhR and FixL ligand binding pockets is shown The key elements in the proposed signal transduction mechanism for FixL, a change in side-chain conformation of Arg206, Thr210 and Asp212, are conserved in AhR with Arg333, Thr337 and Glu339 equivalently positioned, ready to sense and transduce the presence of the PCDD ligand Additional AhR residues involved in ligand recognition and discrimination are Arg282 and Gln377 close to the polar end of the ligand inside the pocket TCDD and heme cofactor atoms are colored green (carbon), red (oxygen), blue (nitrogen), yellow (iron) and magenta (chlorine).
Trang 5The postulated signaling loop of FixL (amino acids 211±
215) following the helical connector had to be substituted by
a shorther fragment in mAhR (Fig 2, amino acids 335±338)
that could only be achieved in our model by manually
shifting the helix towards the b sheet
A last insertion occurred at Gly368, in a loop region
unlikely to be involved in interactions with the ligand
The backbone geometry of the resulting model was
regularized withWHAT IF [25], the side-chain rotamers of
substituted residues optimized using SCWRL [26] and the
model analyzed without any further modi®cations
A model for recognition of PCDDs by Ah receptors
The most noticeable conformational difference between the
mAhR model and the FixL template is the relative position
of the helical connector that moves closer to the b sheet, thus
reducing the size of the binding cavity entrance (Fig 3) The
helix position, intermediate between that observed in HERG
and in FixL, correlates well with the functional role of the
hydrophobic core in the three proteins, while HERG lacks
any binding activity; the modeled mAhR binds PCDDs and
FixL has to accomodate the larger heme cofactor
The AhR residues at positions important for heme
binding in FixL support our model Gly224 and Gly251 in
the hydrophobic core of FixL correspond to Leu347 and
Ala375 in mAhR thus reducing the size of the cavity This is
also consistent with site-directed mutagenesis results that
identi®ed Ala375 as critical for the ligand binding activity
[19] Interestingly, there is also a good correlation between
the size of the side-chain at this position and the size of
the ligand While the latter decreases from FixL to AhR
to HERG, the side-chain volume increases (from Gly251
to Ala375 to Leu127) Moreover, human AhR and
AhR-1C.E., both with reduced af®nity for PCDDs, have
bigger side-chains at this position (Val and Leu, respectively)
partially ®lling the binding cavity The residue coordinating
the ferric heme ion in FixL, His200, is substituted by Cys327
in all AhR receptors, except for AhR-1C.E where
methi-onine is present
At the entrance of the FixL ligand cavity, Arg220, that
binds a heme propionate group, is replaced by Thr in all
AhR (Thr343 in mAhR), except for human AhR and
AhR-1C.E that have isoleucine and leucine, respectively
While the CG2 methyl group of Thr could mediate
hydro-phobic interactions with the ligand, both isoleucine and
leucine will partially block the entrance and reduce af®nity
None of these residues, characteristic of the AhR
proteins, are conserved in the homologous ARNT proteins
(Fig 2) which have no ligand binding activity
Additional information about the PCDD±AhR binding
can be deduced by analyzing the proposed mechanisms for
signal transduction of FixL [12,16,27] According to Perutz
et al [27], the pathway starts at Ile215, Leu236, Ile238,
which form a hydrophobic triad around the heme ligand
The movements of these residues are transmitted to, and
ampli®ed by, a loop that includes Pro213, and then
transmitted to other atoms including the heme
propio-nates A different key residue has been proposed by Gong
et al [12] who indicated the interaction between heme
propionates and His (or Arg) 214 as the starting event of
the protein conformational change It has also been
observed that, going from the unbound to the bound
state, Arg206 affects the position of Asp212, which in turn undergoes the largest conformational change of all the sidechains [12]
Interestingly, although the conformation of the signaling loop had to be altered in the mAhR model, Arg206 and Thr210 of FixL are in equivalent structural positions as Arg333 and Thr337 in mAhR and Asp212 of FixL is replaced by the very similar Glu339 These three residues are conserved in all Ah receptors and are not present in the other PAS proteins analyzed Therefore, by analogy with the FixL mechanism, it is conceivable that, once PCDD is bound, Arg333 in mAhR is involved in the interaction with one of the chlorine atoms and breaks the hydrogen bond with Glu339 that changes conformation
The ligand with the highest af®nity for the AhR is 2,3,7,8-TCDD and our model can be used to investigate its mode of binding, under the assumption that the molecular plane of TCDD is in a similar position as that of the heme group in FixL We highlight in Figs 3C,D, the residues predicted to mediate key ligand interactions in the proposed binding cavity The size of Ala375 is important for ligand accom-modation, Cys327 could interact with the electrophilic central region of TCDD [4], Thr343 possibly stabilizes the complex by hydrophobic interactions, Arg333, at the entrance of the cavity, may guide TCDD towards its binding site by long-range electrostatic interactions and, by interacting with chlorine atoms of TCDD, may promote a signal transduction mechanism through Glu339, similar to that of FixL Two additional residues, Arg282 and Phe345, are shown in the Fig 3 While Arg282, replaced by Gln in some Ah receptors and pointing to the TCDD chlorinated side, may contribute to the binding by electrostatic interac-tions or hydrogen bond, Phe345, lining one side of the ligand pocket, could interact with the aromatic ringsystem
of TCDD Ultimately, Gln377, characteristic of all Ah receptors and not present in other PAS proteins, could form hydrogen bonds with chlorine atoms in the predicted binding cavity for TCDD
Most of the proposed interactions ®t well with the electrostatic characteristics we highlighted in previous QSAR studies on ligand properties [4,5] The requirements
of a nucleophilic site in the central part of the ligand and of electrophilic sites at the sides of the principal molecular axis are both explained by our model
C O N C L U S I O N S
Given the limitations in today's modelling and prediction techniques, the model presented here has to be considered only an approximate and probably incomplete picture of the ligand binding domain of AhR and of its interactions with PCDDs It should also be emphasized that the LBD is part
of a much larger protein and some features of the Ah receptor system might not be explainable in terms of the isolated domain
PYP has been previously proposed as an appropriate structural template for AhR, but our analysis of the recently determined structures of the FixL and the HERG PAS domains strongly suggests that a model based on FixL is more likely to be correct On one hand, the availability of the three structures indicates that the position of the helical connector can differ On the other, the closer functional homology between FixL and AhR, the secondary structure
Trang 6prediction and the size of the ligand all point to FixL as a
more suitable candidate
Our model, although based on low sequence similarity, is
capable of explaining all known experimental and
theoret-ical data and therefore we believe it to be accurate enough to
serve as a framework for further experiments such as site
directed mutagenesis of residues proposed to mediate the
AhR±PCDD interaction and docking calculations to more
accurately de®ne the orientation of the ligand in the binding
cavity
A C K N O W L E D G E M E N T S
The ®nancial support by the Italian CNR (grant no 98.03245.ST74)
and the Fondazione Lombardia per l'Ambiente is gratefully
acknowl-edged.
R E F E R E N C E S
1 Landers, J.P & Bunce, N.J (1991) The Ah receptor and the
mechanism of dioxin toxicity Biochem J 276, 273±287.
2 Sogawa, K & Fujii-Kuriyama, Y (1997) Ah receptor, a novel
ligand-activated transcription factor J Biochem 122, 1075±1079.
3 Waller, C.L & Mckinney, J.D (1995) Three-dimensional
quan-titative structure-activity relationships of dioxins and dioxin-like
compounds: model validation and Ah receptor characterization.
Chem Res Toxicol 8, 847±858.
4 Bonati, L., Fraschini, E., Lasagni, M., Modoni, E.P & Pitea, D.
(1995) A hypothesis on the mechanism of PCDD biological
activity based on molecular electrostatic potential modeling.
Part 2 J Mol Struct (THEOCHEM) 340, 83±95.
5 Fraschini, E., Bonati, L & Pitea, D (1996) Molecular
polariz-ability as a tool for understanding the binding properties of
polychlorinated dibenzo-p-dioxins: de®nition of a reliable
com-putational procedure J Phys Chem 100, 10564±10569.
6 Hahn, M.E., (1991) The aryl hydrocarbon receptor: a comparative
perspective Comp Biochem Physiol Part C 121, 23±53.
7 Gu, Y.Z., Hogenesch, J.B & Brad®eld, C.A (2000) The PAS
superfamily: sensors of environmental and developmental signals.
Annu Rev Pharmacol Toxicol 40, 519±561.
8 Taylor, B.L & Zhulin, I.B (1999) PAS domains: internal sensors
of oxygen, redox potential, and light Microbiol Mol Biol Rev.
63, 479±506.
9 Fukunaga, B.N., Probst, M.R., Reisz-Porszasz, S & Hankinson,
O (1995) Identi®cation of functional domains of the aryl
hydro-carbon receptor J Biol Chem 270, 29270±29278.
10 Pellequer, J.L., Wager-Smith, K.A., Kay, S.A & Getzo, E.D.
(1998) Photoactive yellow protein: a structural prototype for the
three-dimensional fold of the PAS domain superfamily Proc Natl
Acad Sci USA 95, 5884±5890.
11 Cabral, J.H.M., Lee, A., Cohen, S.L., Chait, B.T., Li, M &
Mackinnon, R (1998) Crystal structure and functional analysis of
the HERG potassium channel N terminus: a eukaryotic PAS
domain Cell 95, 649±655.
12 Gong, W., Hao, B., Mansy, S.S., Gonzalez, G., Gilles-Gonzalez, M.A & Chan, M.K (1998) Structure of a biological oxygen sensor: a new mechanism for heme-driven signal transduction Proc Natl Acad Sci USA 95, 15177±15182.
13 Altschul, S.F., Madden, T.L., SchaÈer, A.A., Zhang, J., Zhang, Z., Miller, W & Lipman, D.J (1997) Gapped BLAST and PSI-BLAST: a new generation of protein database search programs Nucleic Acids Res 25, 3389±3402.
14 Berman, H.M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T.N., Weissig, H., Shindyalov, I.N & Bourne, P.E (2000) The protein data bank Nucleic Acids Res 28, 235±242.
15 Holm, L & Sander, C (1996) Mapping the protein universe Science 273, 595±602.
16 Pellequer, J.L., Brudler, R & Getzo, E.D (1999) Biological sensors: more than one way to sense oxygen Current Biol 9, R416±R418.
17 Cu, J.A., Clamp, M.E., Siddiqui, A.S., Finlay, M & Barton, G.J (1998) JPred: a consensus secondary structure prediction server Bioinformatics 14, 892±893.
18 Cu, J.A & Barton, G.J (1999) Evaluation and improvement of multiple sequence methods for protein secondary structure prediction Proteins 34, 508±519.
19 Ema, M., Ohe, N., Suzuki, M., Mimura, J., Sogawa, K., Ikawa, S.
& Fujii-Kuriyama, Y (1994) Dioxin binding activities of poly-morphic forms of mouse and human arylhydrocarbon receptors.
J Biol Chem 269, 27337±27343.
20 Powell-Coman, J.A., Brad®eld, C.A & Wood, W.B (1998) Caenorhabditis elegans orthologs of the aryl hydrocarbon recep-tor and its heterodimerization partner the aryl hydrocarbon receptor nuclear translocator Proc Natl Acad Sci USA 95, 2844±2849.
21 Abnet, C.C., Tanguay, R.L., Hahn, M.E & W.Peterson, R.E (1999) Two forms of aryl hydrocarbon receptor type 2 in rainbow trout (Oncorhynchus mykiss) Evidence for dierential expression and enhancer speci®city J Biol Chem 274, 15159±15166.
22 Roy, N.K & Wirgin, I (1997) Characterization of the aromatic hydrocarbon receptor gene and its expression in Atlantic tomcod Arch Biochem Biophys 344, 373±386.
23 Galtier, N., Gouy, M & Gautier, C (1996) SEAVIEW and PHYLO_WIN: two graphic tools for sequence alignment and molecular phylogeny Comput Appl Biosc 12, 543±548.
24 Molecular Simulation Inc (1998) INSIGHT II Molecular Simu-lation Inc San Diego, CA, USA.
25 Vriend, G (1990) WHAT IF: a molecular modeling and drug design program J Mol Graph 8, 52±56.
26 Bower, M.J., Cohen, F.E & Dunbrack, R.L Jr (1997) Prediction
of protein side-chain rotamers from a backbone-dependent rot-amer library: a new homology modeling tool J Mol Biol 267, 1268±1282.
27 Perutz, M.F., Paoli, M & Lesk, A.M (1999) Fix L, a haemo-globin that acts as an oxygen sensor: signalling mechanism and structural basis of its homology with PAS domains Chem Biol 6, R291±R297.
28 Carson, M (1991) RIBBONS 2.0 J Appl Cryst 24, 958±961.