Biochemical Journal

Research article

The structure of CYP101D2 unveils a potential path for substrate entry into the active site

Wen Yang , Stephen G. Bell , Hui Wang , Weihong Zhou , Mark Bartlam , Luet-Lok Wong , Zihe Rao


The cytochrome P450 CYP101D2 from Novosphingobium aromaticivorans DSM12444 is closely related to CYP101D1 from the same bacterium and to P450cam (CYP101A1) from Pseudomonas putida. All three are capable of oxidizing camphor stereoselectively to 5-exo-hydroxycamphor. The crystal structure of CYP101D2 revealed that the likely ferredoxin-binding site on the proximal face is largely positively charged, similar to that of CYP101D1. However, both the native and camphor-soaked forms of CYP101D2 had open conformations with an access channel. In the active site of the camphor-soaked form, the camphor carbonyl interacted with the haem-iron-bound water. Two other potential camphor-binding sites were also identified from electron densities in the camphor-soaked structure: one located in the access channel, flanked by the B/C and F/G loops and the I helix, and the other in a cavity on the surface of the enzyme near the F helix side of the F/G loop. The observed open structures may be conformers of the CYP101D2 enzyme that enable the substrate to enter the buried active site via a conformational selection mechanism. The second and third binding sites may be intermediate locations of substrate entry and translocation into the active site, and provide insight into a multi-step substrate-binding mechanism.

  • access channel
  • crystal structure
  • cytochrome P450
  • multi-step binding mechanism
  • Novosphingobium aromaticivorans DSM12444
  • open conformation


The CYP (cytochrome P450) superfamily of haem-dependent mono-oxygenases is widely distributed and catalyses C–H bond oxidation in a broad range of organic molecules [1,2]. Through this chemically remarkable reaction and other activities, CYP enzymes are involved in a variety of biochemical processes, including biosynthesis, biodegradation and xenobiotic metabolism [3,4]. Despite low sequence identities, structurally characterized CYP enzymes have a similar overall fold, but considerable structural diversity within this fold leads to different substrate specificity [5]. The structures of the substrate-free and substrate-bound forms of CYP enzymes often have similar closed or open conformations with little or no change upon substrate binding. CYP enzyme access channels have been classified via a systematic analysis of structurally characterized enzymes [6]. Substrate entry and binding are mainly controlled by residues within the B, B′, F and G helices, and the B/C and F/G loops [68]. However, the details of how substrates are recognized and enter the active site are often poorly defined. Molecular dynamic simulations of camphor binding to CYP101A1 from Pseudomonas putida suggest that protein backbone conformational changes and aromatic side chain rotations occur to allow camphor entry into the active site, and experiments have shown that salt-bridge interactions modulate substrate access [7,9,10]. On the basis of theoretical and NMR studies, a second camphor-binding site, on the protein surface of CYP101A1, and a two-step mechanism for camphor binding have been proposed [11]. Multi-step binding mechanisms and multiple substrate occupancy have been investigated in the mammalian enzymes CYP3A4 [12] and CYP1A2 [13].

Crystal structures of different substrate-bound forms of CYP119 from Sulfolobus acidocalderius show significant conformational dissimilarities [1416], and NMR studies indicate that multiple conformations of the substrate-free enzyme may exist [17,18]. This may allow substrate binding to occur via a process of conformation selection, as suggested from kinetic analysis of substrate binding to P450 EryK (CYP113A1) from Saccharopolyspora erythraea [19]. Crystal structures of many forms of CYP101A1 have been determined in order to better understand P450 catalysis. These include the substrate and product complexes [20,21], the ferrous–carbon monoxide- and ferrous–oxygen-bound forms [2224] and mutant enzyme structures [25,26]. In the majority of these structures, a closed conformation is observed and structural information on substrate entry into the active site is not available. Open conformations of CYP101A1 have been induced by binding adamantyl substrates linked to dansyl fluorophores or ruthenium complexes that force open the enzyme structure [2729]. Recently, substrate-free structures of CYP101A1 have been reported in an open conformation similar to those observed with the linker substrates [30].

The CYP101A1 homologues CYP101D2 and CYP101D1 from Novosphingobium aromaticivorans DSM12444 both bind camphor and oxidize it stereoselectively to 5-exo-hydroxycamphor with high activity and coupling of product formation to NADH consumption. Both enzymes utilize a class I electron-transfer system, consisting of a NADH-dependent FAD-containing ferredoxin reductase ArR and a [2Fe–2S] ferredoxin Arx, which are analogous to PdR (putidaredoxin reductase) and Pdx (putidaredoxin) of the CYP101A1 system [31]. We have previously solved the crystal structures of ArR, Arx and CYP101D1, a complete physiological class I electron-transfer system [31]. The structures and the kinetic data indicate that electrostatic interactions are important in protein–protein recognition. In the present paper we report the structure of CYP101D2 in the native and camphor-soaked forms. Both structures have an open conformation with an access channel. The camphor-soaked structure contains three regions of electron density that could be modelled as camphor molecules. One putative camphor molecule was located in the access channel close to the F/G and B/C loops, whereas the second was found at the enzyme surface on the F helix side of the F/G loop. The third camphor molecule was bound within the active site where its carbonyl oxygen interacted with the water ligand of the haem iron.


Cloning, expression and purification

An N-terminal His6 tag was incorporated into CYP101D2 by transferring the encoding gene into the expression vector pET28a(+) (Novagen) using the NdeI and HindIII restriction enzymes [32]. The recombinant plasmid was transformed into Escherichia coli BL21(DE3). A single colony was cultured at 37 °C in LB (Luria–Bertani) medium containing 50 μg·ml−1 kanamycin, to an D600 of 0.6−0.8, and then induced with 0.5 mM IPTG (isopropyl β-D-thiogalactopyranoside) for 18 h at 20 °C. The cells were harvested by centrifugation (6000 g for 15 min), resuspended in 1× PBS, pH 7.4, and then disrupted by sonication at 4 °C. The supernatant obtained by centrifugation at 27000 g for 30 min was loaded on to a Ni2+-chelating affinity column pre-equilibrated with 1× PBS. After washing with 200 ml of 1× PBS, pH 7.4, containing 20 mM imidazole, the target protein was eluted from the column with 1× PBS, pH 7.4, containing 300 mM imidazole. The protein was concentrated and then applied to a Superdex-200 (GE Healthcare) column equilibrated with 1× PBS. The red-coloured fractions were collected and concentrated in buffer A [20 mM Tris/HCl, pH 8.0, and 1 mM DTT (dithiothreitol)]. The protein was further purified using a Resource Q anion-exchange column (GE Healthcare), eluting with a linear gradient of 0−1 M NaCl in buffer A. The purity of CYP101D2 was estimated to be >95% by SDS/PAGE analysis (Supplementary Figure S1 at

Crystallization and substrate-soaking experiments

The purified CYP101D2 protein was concentrated to 50 mg·ml−1 in crystallization buffer (20 mM Tris/HCl, pH 8.0, and 150 mM KCl). Crystallization was performed by the hanging-drop vapour-diffusion method at 18 °C. The protein solution (1 μl) was mixed with an equal volume of reservoir solution and equilibrated with 200 μl of reservoir solution. Crystal screening was carried out with Hampton Research Crystal Screen kits. Initially, small crystals were obtained using 0.1 M Hepes, pH 7.5, 2% (v/v) PEG [poly(ethylene glycol)] 400 and 2.0 M ammonium sulfate (condition number 39 of Crystal Screen I) and further optimization was performed by adjusting the buffer, pH and concentration of precipitant. Good quality crystals were obtained after approx. 1 week from 0.1 M Tris/HCl, pH 8.3, 2.1 M ammonium sulfate and 4% (v/v) PEG 400. The camphor substrate was soaked into crystals of native CYP101D2 by adding solid camphor to the drops containing the crystals. Camphor saturated the crystal solution after several days and the soaking time was varied from 1 week to 1 month.

Data collection and structure determination

Immediately prior to data collection all crystals were cryoprotected by the addition of 20% (v/v) glycerol. X-ray diffraction data were collected in-house at −173 °C on a mar345 image plate using Cu Kα radiation (λ=1.5418 Å; 1 Å=0.1 nm) from a Rigaku MicroMax-007 rotating-anode X-ray generator operating at 40 kV and 20 mA. All diffraction data were indexed, integrated and scaled with the HKL2000 package [33].

The structure of native CYP101D2 was solved by the MR (molecular replacement) method using the program Phaser [34] in the CCP4 suite [35] with CYP101D1 as a search model (PDB code 3LXH). The initial model was then rebuilt using Coot [36] and refined by Refmac5 [37]. The CYP101D2 camphor-soaked structure was solved by the MR method using the native CYP101D2 as a search model. The programs Coot and Refmac5 were used for manual adjustment and refinement of the model.

The stereochemical quality of the structures was examined with the program PROCHECK [38]. The data collection and refinement statistics are summarized in Table 1.

View this table:
Table 1 X-ray data collection and structure refinement statistics for native (DEG-bound) and camphor-soaked CYP101D2

Values in parentheses are for the highest-resolution shell. Rmergei|Ii-〈I〉|/Σ〈I〉, where Ii is an individual intensity measurement and 〈I〉 is the average intensity for all the reflections. Rwork/Rfree=Σ‖Fo|−|Fc‖/Σ|Fo|, where Fo and Fc are the observed and calculated structure factors respectively.


Structure of CYP101D2

The crystal structures of native and camphor-soaked CYP101D2 were solved at 2.4 Å and 2.2 Å resolution respectively. The final models of both structures contained a single CYP101D2 molecule. The native structure contained a single DEG (diethylene glycol; C4H10O3) molecule and was traced from residues 10−413 (out of 416) with the exception of residues 92 and 93 in the B′ helix and residues 189–191 in the F/G loop. The camphor-soaked structure was also traced from residues 10–413, and the entire B′ helix and F/G loop could be modelled.

CYP101D2 had the characteristic trigonal prism-shaped, mixed α/β structure commonly found for CYP enzymes (Figure 1A). The organization of the secondary-structure elements of native and camphor-soaked CYP101D2 were virtually identical [RMSD (root mean square deviation) of 0.49 Å for 399 Cα atoms], and both were similar to CYP101D1 (PDB code 3LXH) and CYP101A1 (PDB code 2CPP) (Supplementary Figure S2 and Supplementary Table S1 at However, both structures had an open conformation resulting from shifts of the F helix (residues 173–188), G helix (192–217), H helix (225–232) and the N-terminus of the I helix (241–261). These helices were pushed away from the haem, thus opening an access channel from the haem to the enzyme surface (Figures 1B–1D and Supplementary Figure S2C). These arrangements resemble the open conformations found in the substrate-free (PDB code 3L61) [30] and linker-substrate-bound CYP101A1 structures {Ru-F8bp-Ad [ruthenium tris(bipyridine)-4,4′-octafluorobiphenyladamantane], PDB code 1K2O [28]; D-4-Ad (dansylbutyladamantane), PDB code 1RF9; and D-8-Ad (dansyloctyladamantane), PDB code 1RE9 [29]} (Figure 1D and Supplementary Figure S2).

Figure 1 Overall structure of CYP101D2

(A) The secondary structure of camphor-soaked CYP101D2. The B′, F and G helices are coloured red, orange and salmon respectively. The haem moiety is shown in yellow. (B) An overlay of the camphor-soaked form of CYP101D2 and CYP101D1 (grey) highlighting the movement of the F, G and H helices and the N terminus of the I helix. The B′, F, G, H and I helices of CYP101D2 are coloured red, orange, salmon, blue and magenta respectively. (C) The distal surface of camphor-soaked CYP101D2. The B′, F and G helices and F/G loop (in green) open up to form an access channel (black circle). (D) Overlay of camphor-soaked CYP101D2 and an open CYP101A1 conformer with an adamantyl-linked ruthenium complex (Ru-F8bp-Ad) bound (PDB code 1K2O). The secondary structure of CYP101A1 is shown in grey and the Ru-F8bp-Ad molecule is shown in green. The open form of CYP101A1 (PDB code 3L61) has a similar structure to that of Ru-F8bp-Ad-bound CYP101A1 and CYP101D2. The positions of the F, G, H and I helices of CYP101D2 closely resemble those in both these structures; the F/G loop and the residues at the N-terminus of the G helix in the open forms of CYP101A1 are bent further back.

The structure of native CYP101D2 contained electron density in the active site that was modelled as DEG, which is present in the PEG 400 from the crystallization buffer. The carbon atoms of the DEG molecule were in van der Waals contact with Leu250, Gly254, Thr258 and Val301 (Supplementary Figure S3 at, whereas the O2 and O3 atoms were hydrogen-bonded to Wat-568 (3.1 Å), the sixth ligand to the haem iron (Fe–O, 2.2 Å), and Glu303 (3.3 Å) respectively.

Analysis of the crystal-packing interactions of CYP101D2 revealed that the F and G helices and F/G loop had extensive interactions with neighbouring molecules (Supplementary Figure S4 at This is also the case for the open CYP101A1 structure (PDB code 3L61) and these crystal-packing interactions may help to maintain the open conformation observed in these crystal structures.

Structure of camphor-soaked CYP101D2

Co-crystallization experiments with CYP101D2 and different camphor concentrations were performed for >500 conditions. CYP101D2 always crystallized in an open conformation, similar to that described above for the native structure, rather than a closed conformation as might be expected. The open camphor-soaked CYP101D2 structure showed three regions of difference in the electron-density map, compared with the native form, consistent with the presence of bound camphor molecules. One camphor molecule was located in the active site, the second in the access channel and the third in a groove on the enzyme surface (Figure 2A). The electron densities of the second and third camphor molecules were weaker than that of the active-site camphor, presumably due to greater disorder and lower occupancy as the substrate became more exposed to the external solvent.

Figure 2 Active site, access channel and surface cavity of camphor-soaked CYP101D2

(A) The position of the three potential camphor electron densities relative to the active site and access channel. The electron density (FoFc contoured at 3σ) of the two potential camphor-binding sites that are located in the access channel and on the enzyme surface are shown in orange. The electron density (2FoFc contoured at 1σ) of the camphor molecule bound in the active site is shown in blue. The camphor and haem molecules are shown in green and yellow respectively. (B) The structure around the camphor-recognition site. The electron density (FoFc contoured at 3σ) of the camphor is coloured orange. The helices are shown in grey and the neighbouring residues are shown in green. Asp257 forms a salt bridge with Arg186 in the camphor-soaked structure, whereas in the native structure this salt bridge is broken and the Arg186 side chain moves away from Asp257 towards the bulk solvent and forms a new salt bridge with Glu156 (residues Arg186 and Glu156 in the native form are coloured in cyan). (C) The access channel of camphor-soaked CYP101D2. The electron density (FoFc contoured at 3σ) of the camphor molecule located in the access channel is shown in orange. The residues involved in camphor binding are shown in green. In the B′ helix the Tyr96 residue moves 5 Å when compared with the native CYP101D2 structure to accommodate the camphor in the access channel. The rotation of the side chain of Phe87 and the poor density map (2FoFc contoured at 1σ), coloured blue, in the camphor-soaked form are also shown (residues Tyr96 and Phe87 in the native form are coloured cyan). (D) The active site of camphor-soaked CYP101D2. The residues involved in substrate binding are shown in green and the water molecule (Wat-589) that occupies the sixth haem iron co-ordination site and which hydrogen-bonds to the camphor carbonyl is shown as a red sphere. The electron density (2FoFc contoured at 1σ) of the camphor molecule is shown in blue.

The camphor closest to the enzyme surface is located 15–16 Å from the haem iron, in a cavity on the F helix side of the F/G loop that was defined by the residues Glu156, Pro159, Val160, Ser178, Ala181, Arg182, Thr185, Arg186, Leu256, Asp257 and Val260 (Figures 2A and 2B, Supplementary Figures S5 and S6, and Supplementary Table S2 at The electron-density-difference map revealed a positive peak (5.7σ), the size and shape of which was consistent with a camphor molecule; the contacts with surrounding residues were all within 3.5–4 Å and only solid camphor was added to the soaking drop. After refinement of the camphor-soaked CYP101D2 structure, the average B-factor of this camphor was ~67.6 Å2. In the native structure only a single water molecule could be identified in this cavity with a hydrogen bond with the carbonyl oxygen of Ser178 (3.2 Å). No camphor was detected in this cavity in the crystal structures of CYP101A1 or CYP101D1, which were instead filled with several water molecules [20,31]. However, a camphor molecule was computationally docked in this cavity in CYP101A1, and a binding constant of 43 μM was calculated for a second camphor molecule using NMR techniques (~16 Å from the haem iron, but of unknown location, although the distance is consistent with the position of this surface cavity) [11].

The second camphor molecule in CYP101D2 was located above the active site, in the access channel and approx. 12 Å from the haem iron (Figures 2A and 2C). The peak intensity in the difference map was similar to that located in the cavity on the enzyme surface (5.8σ). Again, this electron density could be assigned to a camphor molecule despite being relatively poor, which may be the result of partial occupancy and multiple binding conformations due to weak interactions with the residues lining the access channel. Following refinement of the camphor-soaked CYP101D2 structure, the average B-factor of this second camphor was ~67.6 Å2. Residues Phe87, Tyr96, Met98, Met184, Thr185, Leu199, Leu253 and Ile401 were found to be in contact with this camphor molecule (Figure 2C, Supplementary Figure S6 and Supplementary Table S3 at, and the camphor carbonyl oxygen atom could interact with water molecules or the backbone carbonyls of Met184 and Thr185. Further analysis on the possible orientations of these two camphor molecules is provided in Supplementary Figure S6.

The third camphor molecule was unambiguously modelled into the difference-electron-density map with a strong positive peak of 8.9σ (the average B-factor of this camphor was ~40.4 Å2). The camphor was bound within the active site, and the centre of the molecule was positioned approx. 7 Å from the haem iron (Figure 2D). Residues Ile86, Met98, Thr101, Leu250, Leu253, Gly254, Thr258, Val301, Ile401 and Val402 were in van der Waals contact with the camphor carbon atoms (Figure 2D, Supplementary Figure S7 at, Table 2 and Supplementary Table S1). The camphor carbonyl oxygen formed a hydrogen bond (2.8 Å) with Wat-589, the sixth ligand of the haem iron. The closest camphor carbon to the iron atom was C10 (4.9 Å) and the camphor carbonyl oxygen was 4.4 Å away. Although the selectivity of camphor oxidation by CYP101D2 is ≥99% for 5-exo-hydroxycamphor, the C5 carbon of camphor was 7.5 Å from the haem iron.

View this table:
Table 2 Protein–camphor interactions at the active site of CYP101D2, CYP101D1 and CYP101A1

CYP101D1 values are taken from [31], CYP101A1 values are taken from [20].

The access channel and active site

The access channel was 20 Å deep and flanked by residues Tyr29, Phe87, Pro89, Tyr96, Met184, Thr185, Arg186, Pro187, Met195, Leu199, Asn203, Asp257, Val401 and Val402 (Supplementary Figure S3A and Supplementary Table S3). The side chains of Met184, Thr185 and Leu199 were pushed 5.5–6.0 Å away from the active site compared with the equivalent residues in the closed CYP101D1 and CYP101A1 structures, where they form a cap over the substrate (Supplementary Figure S8 at In both CYP101D2 structures, the side chain of the B′ loop residue, Tyr96 pointed away from the haem towards the solvent (Figure 2C, Supplementary Figures S3A and S8), whereas in both native and camphor-bound CYP101D1 and CYP101A1 Tyr96 points inwards at the haem, and the side chain forms a hydrogen bond with the camphor carbonyl. Notably, the Tyr96 side chain had moved by ~5 Å towards the entrance of the access channel in the camphor-soaked CYP101D2 structure, presumably to accommodate the camphor molecule that was located in the access channel, but that had not yet moved inside the active site (Figure 2C). The side chain of Phe87 had also shifted significantly and was flipped by ~90 ° between the native and camphor-soaked CYP101D2 structures (Figure 2C). The electron density of Phe87 in the camphor-soaked structure was less well defined compared with the native structure. The lower resolution of the Phe87 side chain and the second camphor molecule in this structure may be due to different camphor occupancy and binding orientations resulting in different Phe87 orientations. Even though the concentration of camphor and the soaking times were varied, electron-density maps obtained for the camphor molecule and Phe87 side chain were always similar.

In the camphor-soaked CYP101D2 structure, Asp257 formed a salt bridge with Arg186, but in the native structure this salt bridge was broken and the Arg186 side chain had moved towards the bulk solvent and formed a new salt bridge with Glu156 (Figure 2B and Supplementary Figure S9 at The backbone carbonyl oxygen of Asp257 in both CYP101D2 structures interacts with the side-chain nitrogen of Asn261 (Figure 3A). This aspartate residue–asparagine residue interaction and the resultant rotation of the aspartate residue's carbonyl backbone is also found in CYP101D1 and in the ferrous deoxy form of CYP101A1 [24]. The I helix groove of CYP101D2 had shifted slightly towards the N-terminus compared with its position in closed CYP101A1 and CYP101D1 structures, mirroring the shift of this groove in the open conformations of CYP101A1 (Supplementary Figure S10 at [29]. These movements in CYP101D2 resulted in a shift of the side chain of Thr258 to a similar orientation to that observed for Thr252 in the ferrous–oxy form of CYP101A1, but different from those seen in the closed structures of CYP101A1 and CYP101D1.

Figure 3 Environment of the I helix, haem moiety and proximal face of CYP101D2

(A) Overlay of the residues in the I helix between CYP101D2 (green) and CYP101A1 (salmon). The conformation of the carbonyl oxygen of Asp257 is flipped by ~90 ° towards Asn261 due to a hydrogen bond with the side-chain nitrogen of Asn261 (2.8 Å). Two water molecules hydrogen bond with the Asp257 side chain and are shown as red spheres. (B) The haem environment in the native form of CYP101D2. The A-ring propionate group forms a salt bridge with Arg305. The O1 atom forms hydrogen bonds with two water molecules, Wat-565 and Wat-436, and Wat-565 forms another hydrogen bond with Wat-566. The O2 atom of the A-ring propionate group is hydrogen-bonded to the side-chains of Tyr75 and Glu303. The Oϵ1 atom of Glu303 is hydrogen-bonded to the DEG molecule and the Oϵ2 atom is hydrogen-bonded to Wat-566 and the A-ring propionate group. The haem and the relevant residues are shown in yellow and green respectively. The water molecules are shown as red spheres. The electron density (2FoFc contoured at 1σ) of the DEG molecule is shown in blue. (C) The haem environment in the camphor-soaked structure of CYP101D2. Compared with the native structure, the A-ring propionate group is rotated by approx. 90 °, and both the oxygen atoms of Glu303 are hydrogen-bonded to the propionate group. The propionate group also forms a salt bridge with Arg305 and hydrogen bonds with two water molecules, Wat-498 and Wat-464. Glu303 is hydrogen bonded to Tyr75. The haem and residues are shown in yellow and green respectively. The water molecules are shown as red spheres and the electron density (2FoFc contoured at 1σ) of the camphor molecule is shown in blue. (D) The proximal face of CYP101D2. Negatively and positively charged surface areas are coloured red and blue respectively. The positively charged proximal face is similar to that of CYP101D1 and more positively charged than that of CYP101A1 (see Supplementary Figure S12 at The basic residues of CYP101D2 (Arg112, Arg125, Arg289, Arg349, Arg362 and Arg370) and Ser76 which aligns with Arg77 in CYP101D1 are highlighted.

A hydrogen-bonded network of four water molecules was located in the vicinity of the haem propionate groups of the native CYP101D2 structure (Figures 3B and 3C). In the camphor-soaked structure there are fewer water molecules in this region and the Glu303 side chain and the A-ring propionate group of the haem occupied different positions compared with the native structure (Figure 3C). Several water molecules were also located in the access channel of both structures of CYP101D2 (Supplementary Figure S11 at

Proximal face

The haem-proximal surface of CYP101D2 had a preponderance of positively charged residues, including Arg112, Arg125, Arg289, Arg349, Arg362 and Arg370, that are conserved spatially and in sequence in CYP101D1 (Figure 3D and Supplementary Figure S12 at By comparison, the proximal face of CYP101A1 is less positively charged. The ferredoxin gene Arx is genomically associated with the CYP101D2 gene of N. aromaticivorans, and together ArR and Arx support the mono-oxygenase activity of CYP101D2, CYP101D1 (a kcat of 39 and 41 s−1 respectively) and three other N. aromaticivorans CYP enzymes [31]. These basic residues in CYP101D2 and CYP101D1 probably interact with acidic surface residues on Arx in the competent electron-transfer complex (Supplementary Figure S12). Electrostatic interactions are important in Arx–CYP101D2 electron transfer, and the charge distributions are consistent with the low cross-reactivity with the CYP101A1 and CYP101D2 systems when the physiological ferredoxins are exchanged [31,32,39].


CYP101D2 has a primary sequence identity of 62% with CYP101D1 and of 44% with CYP101A1 (Supplementary Figure S13 at and Supplementary Table S1). However, owing to crystal packing interactions, the native and camphor-soaked structures of CYP101D2 had an open conformation with a clearly defined access channel, in contrast with the closed structures observed with CYP101D1 [31]. The helices of the native structure of CYP101D2 overlaid closely with those of the open forms of CYP101A1. Unexpectedly, the carbonyl oxygen of the active-site camphor was hydrogen-bonded to the haem-iron-bound water rather than the Tyr96 side chain, as observed in the closed structures of camphor-bound CYP101A1 and CYP101D1 (Table 2). Soaking of crystals of the native open form of CYP101D2 in a camphor solution or co-crystallization of camphor with CYP101D2 did not result in a closed form in which the camphor in the active site was completely shielded from the solvent. Camphor-binding titrations with CYP101D2 showed an approx. 40% shift to the high-spin form and a Kd of 3.1 μM (Supplementary Figure S14 at, similar to that observed previously for CYP101D1 [31].

The binding orientation of the camphor molecule in the active site is not consistent with the observed product profile of almost exclusive formation of 5-exo-hydroxycamphor. In addition, it is inhibitory towards the first electron-transfer step. The enzyme has to undergo a change to a closed conformation with a concomitant shift in the camphor orientation, e.g. breaking the hydrogen bond with the axial water and rotation to form a hydrogen bond with the Tyr96 side chain to orient the camphor C5 over the haem iron. This suggests that crystal-packing interactions may have effectively locked CYP101D2 in the open conformation, resisting the conversion into the closed structure expected upon substrate binding and resulting in the observed camphor-binding orientation. In solution, conformational dynamics of the enzyme may result in rapid transition to the closed form, perhaps even during camphor entry into the active site, initially leading to different transient binding orientations of the camphor molecule before the enzyme–substrate complex reaches its equilibrium structure.

Multi-step binding mechanisms have been proposed for CYP3A4, CYP1A2, CYP101A1 and recently for CYP107L1 (P450 PikC) [1113,40,41]. In such mechanisms the substrate binds to a recognition site on the protein surface before entering the active site. The product could also leave via the same route, although alternative routes, such as via the F and G helices or the E/F loop, have been suggested [6]. Binding of multiple camphor molecules to CYP101A1 has been reported from NMR and UV studies [11,4244]. The observation of a putative camphor molecule in a surface cavity, coupled to the presence of a second camphor in the access channel of camphor-soaked CYP101D2, may be further evidence of a multi-step binding mechanism with initial substrate recognition occurring at the enzyme surface [11]. Modelling and NMR relaxation studies led to the proposal that a camphor entry/recognition site (Kd of 43 μM) may be located in this cavity in CYP101A1, and it is of note that these experiments showed that this surface-bound camphor molecule was in a highly fluxional state [11]. The B-factors associated with the camphor molecules in the access channel and the surface cavity are similar to those observed for the solvent, whereas the camphor molecule in the active site has a B-factor similar to that of the protein. The less-specific interactions and mobile nature of the camphor molecules located in the CYP101D2 surface cavity and access channel compared with the camphor in the active site suggest that these may be snapshots of intermediate binding positions in the substrate entrance pathway into the active site (or of product release).

Many structurally characterized CYP enzymes have closed conformations with buried active sites [20,31]. However, clearly defined access channels are not uncommon [45,46]. Other P450 enzymes show open conformations in the absence of a bound substrate, but adopt a more closed conformation upon substrate binding, e.g. P450 EryK [19]. Double-jump substrate-binding studies of P450 EryK suggest a conformational selection model for substrate recognition and binding [19]. Crystal structures of CYP119 reveal that the enzyme undergoes major rearrangements at the active site to accommodate the binding of different substrates, and NMR studies of this enzyme tagged with the unnatural amino acid 13C-p-methoxyphenylalanine showed that it can exist in different conformations [15,17,18]. These conformers are thought to be important in substrate recognition and binding, and the equilibrium between them can be perturbed using different substrates. The open structure of CYP101D2 may be a conformer that facilitates substrate entry and product exit from the active site. The movement of Tyr96 between the open and closed conformers may assist in the recognition and binding of camphor. We note, however, that CYP101D2 is probably held in this open conformation by crystal packing contacts and the camphor concentration of ~8 mM (1.2 g·l−1) is higher than those in turnover reactions (1 mM). Therefore we cannot rule out adventitious binding at one or both of the locations outside the active site. The substrate might enter (or the product exit) the active site that is opened up by normal thermal fluctuations of the structure via the channel without forming an intermediate state. On the other hand, no adventitious binding of camphor was observed to the surface cavity of the closed conformations of CYP101D1 under a similar camphor-soaking regime [31].

The switch between the open and the closed forms of CYP101A1 involves breaking and formation of inter-residue salt bridges and hydrogen bonds, and the rearrangement of hydrophobic interactions [29]. The presence or otherwise of these interactions, together with adventitious binding of molecules, such as DEG or deliberate introduction of longer substrates such as D-4-Ad, as well as crystal packing effects, may control whether crystals with open or closed conformation are obtained. Salt-bridge interactions between the F helix or F/G loop and the I helix are less extensive in CYP101D2 (and CYP101D1) compared with CYP101A1 (Supplementary Table S2). A network of interactions between Lys178, Asp182, Arg186 and Asp251 in CYP101A1 has been shown to play a key role in the control of the diffusion step of camphor binding and provide a link from the I helix to the bulk solvent at the enzyme surface (Supplementary Table S2) [10,47]. These interactions are broken in the open conformations of CYP101A1 [29,30]. The Asp257−Arg186 salt bridge is conserved in camphor-soaked CYP101D2, but Ser178 and Arg182 replace Lys178 and Asp182. Ser178 is located in the access channel, whereas Arg182 points towards the enzyme surface (Figure 2B and Supplementary Figure S5B). The Asp257−Arg186 salt bridge separates the camphor molecules in the surface recognition site and in the access channel site. The breaking of this salt bridge and the movement of Arg186 to interact with Glu156, which lines the surface recognition site, could open up a short route of substrate entry or product exit between these two sites. This blocking of the proposed surface recognition site and the active-site camphor by the equivalent salt bridge in CYP101A1 has been noted previously [11], and in the substrate-free open CYP101A1 structure, Arg186 interacts with Glu156 [30]. The structure and function of mutants with substitutions to break salt bridges will be of interest [10,47,48].

The proximal faces of CYP101D2 and CYP101D1 are very similar, and more positively charged than CYP101A1, whereas the corresponding interaction region on the Arx ferredoxin is more negatively charged compared with Pdx [31], consistent with electrostatic interactions being important in Arx–CYP101D2 and Arx–CYP101D1 binding and recognition. However, the precise role of each of the acidic residues on Arx and the basic residues on the CYP enzymes remains to be elucidated. The ArR/Arx system also supports the mono-oxygenase activity of at least three other CYP enzymes from N. aromaticivorans (CYP101B1, CYP101C1 and CYP111A2). The crystal structures of the other P450 enzymes supported by Arx will be required for a fuller analysis, and this is the subject of ongoing investigations.


CYP101D2 and CYP101D1 are analogues of CYP101A1 that can catalyse the fast, efficient and stereoselective oxidation of camphor to 5-exo-hydroxycamphor. Although the structure of CYP101D1 is similar to that of CYP101A1, the camphor-soaked structure of CYP101D2 has an open conformation and reveals potential camphor-binding sites at the active site, in the access channel and at the enzyme surface. This open structure may be a conformer involved in substrate recognition and binding. The camphor-binding sites at the enzyme surface and in the access channel may be intermediate locations of the substrate in its progression into the active site of CYP101D2 via multi-step and conformational selection mechanisms.


Wen Yang, Stephen Bell and Hui Wang performed the protein production, purification and biochemical assays. Wen Yang and Hui Wang performed the crystallization and soaking experiments. Wen Yang, Weihong Zhou and Mark Bartlam conducted the structure determination and refinement. Stephen Bell, Wen Yang, Mark Bartlam, Luet Wong and Zihe Rao designed the research, analysed the data and wrote the manuscript. All authors helped with editing the manuscript before submission.


This work was supported by the Ministry of Science and Technology of China Project 973 [grant number 2007CB914301 (to M.B.)]; the Tianjin Municipal Science and Technology Commission [grant number 08SYSYTC00200]; and by the Higher Education Funding Council for England.


We are grateful to Dr Zhiyong Lou (Tsinghua University, Beijing, China) for help in data collection and processing.


  • The co-ordinates for the crystal structures of native CYP101D2 and camphor-soaked CYP101D2 have been deposited in the PDB under the accession codes 3NV5 and 3NV6 respectively.

Abbreviations: CYP, cytochrome P450; D-4-Ad, dansylbutyladamantane; D-8-Ad, dansyloctyladamantane; DEG, diethylene glycol; MR, molecular replacement; PdR, putidaredoxin reductase; Pdx, putidaredoxin; PEG, poly(ethylene) glycol; RMSD, root mean square deviation; Ru-F8bp-Ad, ruthenium tris(bipyridine)-4,4′-octafluorobiphenyladamantane


View Abstract