Splice variant PRKC-ζ(-PrC) is a novel biomarker of human prostate cancer.

BACKGROUND: Previously, using gene-knockdown techniques together with genome expression array analysis, we showed the gene protein Kinase C (PKC)-zeta (PRKCZ) to mediate the malignant phenotype of human prostate cancer. However, according to NCBI, the gene has undergone several major iterations. Therefore, to understand the relationship between its structure and biological activities, we have analysed its expressed sequence in prostate cancer cell lines and tissues. METHODS: Transcriptome-walking and targeted PCR were used to sequence the mRNA transcribed from PRKCZ. Hydropathy analysis was employed to analyse the hypothetical protein sequence subsequently translated and to identify an appropriate epitope to generate a specific monoclonal antibody. RESULTS: A novel sequence was identified within the 3'-terminal domain of human PRKCZ that, in prostate cancer cell lines and tissues, is expressed during transcription and thereafter translated into protein (designated PKC-ζ(-PrC)) independent of conventional PKC-ζ(-a). The monoclonal antibody detected expression of this 96 kD protein only within malignant prostatic epithelium. INTERPRETATION: Transcription and translation of this gene sequence, including previous intronic sequences, generates a novel specific biomarker of human prostate cancer. The presence of catalytic domains characteristic of classic PKC-ß and atypical PKC-ι within PKC-ζ(-PrC) provides a potential mechanism for this PRKCZ variant to modulate the malignant prostatic phenotype out-with normal cell-regulatory control.

Deltoid contracture mimicking shoulder dislocation in a 7-year-old boy.

Contracture of the deltoid muscle, a relatively uncommon disorder in children, can be caused by repeated intramuscular injection, trauma, or congenital disease. The typical clinical manifestations of deltoid contracture (i.e., a palpable fibrous cord within the deltoid muscle, abduction contracture of the shoulder, winged scapula, and skin dimpling over the fibrous bands), however, may be atypical or even lacking, thus, leading to misdiagnosis. The procedure going from misdiagnosis to recognition of the correct diagnosis is reviewed in a 7-year-old boy with deltoid contracture.

Role of the conserved phenylalanine 181 of NADPH-cytochrome P450 oxidoreductase in FMN binding and catalytic activity.

NADPH-cytochrome P450 oxidoreductase (P450 reductase, EC 1.6.2.4) is an essential component of the P450 monooxygenase complex and binds FMN, FAD, and NADPH cofactors. Residues Tyr140 and Tyr178 are known to be involved in FMN binding. A third aromatic side chain, Phe181, is also located in the proximity of the FMN ring and is highly conserved in FMN-binding proteins, suggesting an important functional role. This role has been investigated by site-directed mutagenesis. Substitution of Phe181 with leucine or glutamine decreased the cytochrome c reductase activity of the enzyme by approximately 50%. Ferricyanide reductase activity was unaffected, indicating that the FAD domain was unperturbed. The mutant FMN domains were expressed in Escherichia coli, and the redox potentials and binding energies of their complexes with FMN were determined. The affinity for FMN was decreased approximately 50-fold in the Leu181 and Gln181 mutants. Comparison of the binding energies of the wild-type and mutant enzymes in the three redox states of FMN suggests that Phe181 stabilizes the FMN-apoprotein complex. The amide 1H and 15N resonances of the Phe181Leu FMN domain were assigned; comparison of their chemical shifts with those of the wild-type domain indicated that the effect of the substitution on FMN affinity results from perturbation of two loops which form part of the FMN binding site. The results indicate that Phe181 cooperates with Tyr140 and Tyr178 to play a major role in the binding and stability of FMN.

Stopped-flow kinetic studies of flavin reduction in human cytochrome P450 reductase and its component domains.

The reduction by NADPH of the FAD and FMN redox centers in human cytochrome P450 reductase and its component domains has been studied by rapid-mixing, stopped-flow spectroscopy. Reduction of the isolated FAD-domain occurs in three kinetically resolvable steps. The first represents the rapid formation (>500 s(-)(1)) of a charge-transfer species between oxidized FAD and NADPH. This is followed by an isomerization ( approximately 200 s(-)(1)) to a second charge-transfer species, characterized by a more intense absorption in the long-wavelength region. The third step represents hydride transfer from NADPH to FAD and is accompanied by a change in the tryptophan fluorescence of the FAD-domain. Flavin reduction is reversible, and the observed rate of hydride transfer displays a complex dependence on NADPH concentration. Two-electron-reduced FAD-domain is active in electron transfer reactions with the isolated FMN domain through the formation of a weakly associating electron transfer complex. Reduction of the CPR by NADPH occurs without direct spectral evidence for the formation of charge-transfer species, although the presence of such species is inferred indirectly. Transfer of the first hydride ion leads to the accumulation of a blue di-semiquinoid species of the reductase, indicating rapid transfer of one electron to the FMN domain. The di-semiquinoid species decays on transfer of the second hydride ion. A third phase is seen following prolonged incubation with NADPH and is assigned to a series of equilibration reactions between different redox species of the enzyme as the system relaxes to its thermodynamically most stable state. As with the isolated FAD-domain, the first hydride transfer in the reductase shows a complex dependence on NADPH concentration. At high NADPH concentration, the observed rate of hydride transfer is slow (approximately 20 s(-1)), and this attenuated rate is attributed to the reversible formation of an less active complex resulting from the binding of a second molecule of NADPH. The kinetic data are discussed with reference to the potentiometric studies on the enzyme and its component domains presented in the preceding paper in this issue [Munro, A., Noble, M., Robledo, L., Daff, S., and Chapman, S. (2001) Biochemistry 40, 1956-1963].

Structural consequences of site-directed mutagenesis in flexible protein domains: NMR characterization of the L(55,56)S mutant of RhoGDI.

The guanine dissociation inhibitor RhoGDI consists of a folded C-terminal domain and a highly flexible N-terminal region, both of which are essential for biological activity, that is, inhibition of GDP dissociation from Rho GTPases, and regulation of their partitioning between membrane and cytosol. It was shown previously that the double mutation L55S/L56S in the flexible region of RhoGDI drastically decreases its affinity for Rac1. In the present work we study the effect of this double mutation on the conformational and dynamic properties of RhoGDI, and describe the weak interaction of the mutant with Rac1 using chemical shift mapping. We show that the helical content of the region 45-56 of RhoGDI is greatly reduced upon mutation, thus increasing the entropic penalty for the immobilization of the helix, and contributing to the loss of binding. In contrast to wild-type RhoGDI, no interaction with Rac1 could be identified for amino-acid residues of the flexible domain of the mutant RhoGDI and only very weak binding was observed for the folded domain of the mutant. The origins of the effect of the L55S/L56S mutation on the binding constant (decreased by at least three orders of magnitude relative to wild-type) are discussed with particular reference to the flexibility of this part of the protein.

Two-iron rubredoxin of Pseudomonas oleovorans: production, stability and characterization of the individual iron-binding domains by optical, CD and NMR spectroscopies.

A minigene encoding the C-terminal domain of the 2Fe rubredoxin of Pseudomonas oleovorans was created from the parental alk G gene contained in the expression plasmid pKK223-3. The vector directed the high-level production of the C-terminal domain of this rubredoxin; a simple procedure was used to purify the recombinant domain in the 1Fe form. The 1Fe form of the C-terminal domain was readily converted into the apoprotein and cadmium forms after precipitation with trichloroacetic acid and resolubilization in the presence or absence of cadmium chloride respectively. In steady-state assays, the recombinant 1Fe C-terminal domain is redox-active and able to transfer electrons from reduced rubredoxin reductase to cytochrome c. The absorption spectrum and dichroic features of the CD spectrum for the iron- and cadmium-substituted C-terminal domain are similar to those reported for the iron- and cadmium-substituted Desulfovibrio gigas rubredoxin [Henehen, Pountney, Zerbe and Vasak (1993) Protein Sci. 2, 1756-1764]. Difference absorption spectroscopy of the cadmium-substituted C-terminal domain revealed the presence of four Gaussian-resolved maxima at 202, 225, 240 and 276 nm; from Jørgensen's electronegativity theory, the 240 nm band is attributable to a CysS-Cd(II) charge-transfer excitation. Attempts to express the N-terminal domain of the 2Fe rubredoxin directly from a minigene were unsuccessful. However, the N-terminal domain was isolated through cleavage of an engineered 2Fe rubredoxin in which a factor Xa proteolysis site had been introduced into the putative interdomain linker. The N-terminal domain is characterized by absorption spectra typical of the 1Fe rubredoxins. The domain is folded as determined by CD and NMR spectroscopies and is redox-active. However, the N-terminal domain is less stable than the isolated C-terminal domain, a finding consistent with the known properties of the full-length 2Fe and cadmium-substituted Ps. oleovorans rubredoxin.

Structure-activity relationships in flexible protein domains: regulation of rho GTPases by RhoGDI and D4 GDI.

The guanine dissociation inhibitors RhoGDI and D4GDI inhibit guanosine 5'-diphosphate dissociation from Rho GTPases, keeping these small GTPases in an inactive state. The GDIs are made up of two domains: a flexible N-terminal domain of about 70 amino acid residues and a folded 134-residue C-terminal domain. Here, we characterize the conformation of the N-terminal regions of both RhoGDI and D4GDI using a series of NMR experiments which include (15)N relaxation and amide solvent accessibility measurements. In each protein, two regions with tendencies to form helices are identified: residues 36 to 58 and 9 to 20 in RhoGDI, and residues 36 to 57 and 20 to 25 in D4GDI. To examine the functional roles of the N-terminal domain of RhoGDI, in vitro and in vivo functional assays have been carried out with N-terminally truncated proteins. These studies show that the first 30 amino acid residues are not required for inhibition of GDP dissociation but appear to be important for GTP hydrolysis, whilst removal of the first 41 residues completely abolish the ability of RhoGDI to inhibit GDP dissociation. The combination of structural and functional studies allows us to explain why RhoGDI and D4GDI are able to interact in similar ways with the guanosine 5'-diphosphate-bound GTPase, but differ in their ability to regulate GTP-bound forms; these functional differences are attributed to the conformational differences of the N-terminal domains of the guanosine 5'-diphosphate dissociation inhibitors. Therefore, the two transient helices, appear to be associated with different biological effects of RhoGDI, providing a clear example of structure-activity relationships in a flexible protein domain.

DNA-binding mechanism of the Escherichia coli Ada O(6)-alkylguanine-DNA alkyltransferase.

The C-terminal domain of the Escherichia coli Ada protein (Ada-C) aids in the maintenance of genomic integrity by efficiently repairing pre-mutagenic O:(6)-alkylguanine lesions in DNA. Structural and thermodynamic studies were carried out to obtain a model of the DNA-binding process. Nuclear magnetic resonance (NMR) studies map the DNA-binding site to helix 5, and a loop region (residues 151-160) which form the recognition helix and the 'wing' of a helix-turn-wing motif, respectively. The NMR data also suggest the absence of a large conformational change in the protein upon binding to DNA. Hence, an O:(6)-methylguanine (O:(6)meG) lesion would be inaccessible to active site nucleophile Cys146 if the modified base remained stacked within the DNA duplex. The experimentally determined DNA-binding face of Ada-C was used in combination with homology modelling, based on the catabolite activator protein, and the accepted base-flipping mechanism, to construct a model of how Ada-C binds to DNA in a productive manner. To complement the structural studies, thermodynamic data were obtained which demonstrate that binding to unmethylated DNA was entropically driven, whilst the demethylation reaction provoked an exothermic heat change. Methylation of Cys146 leads to a loss of structural integrity of the DNA-binding subdomain.

Overall rotational diffusion and internal mobility in domain II of protein G from Streptococcus determined from 15N relaxation data.

The backbone dynamics and overall tumbling of protein G have been investigated using 15N relaxation. Comparison of measured R2/R1 relaxation rate ratios with known three-dimensional coordinates of the protein show that the rotational diffusion tensor is significantly asymmetric, exhibiting a prolate axial symmetry. Extensive Monte Carlo simulations have been used to estimate the uncertainty due to experimental error in the relaxation rates to be D(parallel)/D(perpendicular) = 1.68 +/- 0.08, while the dispersion in the NMR ensemble leads to a variation of D(parallel)/D(perpendicular) = 1.65 +/- 0.03. Incorporation of this tensorial description into a Lipari-Szabo type analysis of internal motion has allowed us to accurately describe the local dynamics of the molecule. This analysis differs from an earlier study where the overall rotational diffusion was described by a spherical top. In this previous analysis, exchange parameters were fitted to many of the residues in the alpha helix. This was interpreted as reflecting a small motion of the alpha helix with respect to the beta sheet. We propose that the differential relaxation properties of this helix compared to the beta sheet are due to the near-orthogonality of the NH vectors in the two structural motifs with respect to the unique axis of the diffusion tensor. Our analysis shows that when anisotropic rotational diffusion is taken into account NH vectors in these structural motifs appear to be equally rigid. This study underlines the importance of a correct description of the rotational diffusion tensor if internal motion is to be accurately investigated.

Structural and biochemical characterization of recombinant wild type and a C30A mutant of trimethylamine dehydrogenase from methylophilus methylotrophus (sp. W(3)A(1)).

Trimethylamine dehydrogenase (TMADH) is an iron-sulfur flavoprotein that catalyzes the oxidative demethylation of trimethylamine to form dimethylamine and formaldehyde. It contains a unique flavin, in the form of a 6-S-cysteinyl FMN, which is bent by approximately 25 degrees along the N5-N10 axis of the flavin isoalloxazine ring. This unusual conformation is thought to modulate the properties of the flavin to facilitate catalysis, and has been postulated to be the result of covalent linkage to Cys-30 at the flavin C6 atom. We report here the crystal structures of recombinant wild-type and the C30A mutant TMADH enzymes, both determined at 2.2 A resolution. Combined crystallographic and NMR studies reveal the presence of inorganic phosphate in the FMN binding site in the deflavo fraction of both recombinant wild-type and C30A proteins. The presence of tightly bound inorganic phosphate in the recombinant enzymes explains the inability to reconstitute the deflavo forms of the recombinant wild-type and C30A enzymes that are generated in vivo. The active site structure and flavin conformation in C30A TMADH are identical to those in recombinant and native TMADH, thus revealing that, contrary to expectation, the 6-S-cysteinyl FMN link is not responsible for the 25 degrees butterfly bending along the N5-N10 axis of the flavin in TMADH. Computational quantum chemistry studies strongly support the proposed role of the butterfly bend in modulating the redox properties of the flavin. Solution studies reveal major differences in the kinetic behavior of the wild-type and C30A proteins. Computational studies reveal a hitherto, unrecognized, contribution made by the S(gamma) atom of Cys-30 to substrate binding, and a role for Cys-30 in the optimal geometrical alignment of substrate with the 6-S-cysteinyl FMN in the enzyme active site.

Mapping the binding site for the GTP-binding protein Rac-1 on its inhibitor RhoGDI-1.

BACKGROUND: Members of the Rho family of small GTP-binding proteins, such as Rho, Rac and Cdc42, have a role in a wide range of cell responses. These proteins function as molecular switches by virtue of a conformational change between the GTP-bound (active) and GDP-bound (inactive) forms. In addition, most members of the Rho and Rac subfamilies cycle between the cytosol and membrane. The cytosolic guanine nucleotide dissociation inhibitors, RhoGDIs, regulate both the GDP/GTP exchange cycle and the membrane association/dissociation cycle. RESULTS: We have used NMR spectroscopy and site-directed mutagenesis to identify the regions of human RhoGDI-1 that are involved in binding Rac-1. The results emphasise the importance of the flexible regions of both proteins in the interaction. At least one specific region (residues 46-57) of the flexible N-terminal domain of RhoGDI, which has a tendency to form an amphipathic helix in the free protein, makes a major contribution to the binding energy of the complex. In addition, the primary site of Rac-1 binding on the folded domain of RhoGDI involves the beta4-beta5 and beta6-beta7 loops, with a slight movement of the 3(10) helix accompanying the interaction. This binding site is on the same face of the protein as the binding site for the isoprenyl group of post-translationally modified Rac-1, but is distinct from this site. CONCLUSIONS: Isoprenylated Rac-1 appears to interact with three distinct sites on RhoGDI. The isoprenyl group attached to the C terminus of Rac-1 binds in a pocket in the folded domain of RhoGDI. This is distinct from the major site on this domain occupied by Rac-1 itself, which involves two loops at the opposite end to the isoprenyl-binding site. It is probable that the flexible C-terminal region of Rac-1 extends from the site at which Rac-1 contacts the folded domain of RhoGDI to allow the isoprenyl group to bind in the pocket at the other end of the RhoGDI molecule. Finally, the flexible N terminus of RhoGDI-1, and particularly residues 48-58, makes a specific interaction with Rac-1 which contributes substantially to the binding affinity.

The role of tyrosine-9 and the C-terminal helix in the catalytic mechanism of Alpha-class glutathione S-transferases.

Glutathione S-transferases (GSTs) play a key role in the metabolism of drugs and xenobiotics. To investigate the catalytic mechanism, substrate binding and catalysis by the wild-type and two mutants of GST A1-1 have been studied. Substitution of the 'essential' Tyr(9) by phenylalanine leads to a marked decrease in the k(cat) for 1-chloro-2,4-dinitrobenzene (CDNB), but has no affect on k(cat) for ethacrynic acid. Similarly, removal of the C-terminal helix by truncation of the enzyme at residue 209 leads to a decrease in k(cat) for CDNB, but an increase in k(cat) for ethacrynic acid. The binding of a GSH analogue increases the affinity of the wild-type enzyme for CDNB, and increases the rate of the enzyme-catalysed conjugation of this substrate with the small thiols 2-mercaptoethanol and dithiothreitol. This suggests that GSH binding produces a conformational change which is transmitted to the binding site for the hydrophobic substrate, where it alters both the affinity for the substrate and the catalytic-centre activity ('turnover number') for conjugation, perhaps by increasing the proportion of the substrate bound productively. Neither of these two effects of GSH analogues are seen in the C-terminally truncated enzyme, indicating a role for the C-terminal helix in the GSH-induced conformational change.

Protein-ligand interactions measured by 15N-filtered diffusion experiments.

NMR diffusion coefficient measurements have been shown to be sensitive to the conformational and oligomeric states of proteins. Recently, heteronuclear-filtered diffusion experiments have been proposed [Dingley et al. (1997) J. Biomol. NMR, 10, 1-8]. Several new heteronuclear-filtered diffusion pulse sequences are proposed which are shown to have superior sensitivity to those previously proposed. One of these new heteronuclear-filtered diffusion experiments has been used to study the binding of an SH3 domain to a peptide. Using this system, we show that it is possible to measure binding constants from diffusion coefficient measurements.

Determinants of specificity for aflatoxin B1-8,9-epoxide in alpha-class glutathione S-transferases.

We have used homology modelling, based on the crystal structure of the human glutathione S-transferase (GST) A1-1, to obtain the three-dimensional structures of rat GSTA3 and rat GSTA5 subunits bound to S-aflatoxinyl-glutathione. The resulting models highlight two residues, at positions 208 and 108, that could be important for determining, either directly or indirectly, substrate specificity for aflatoxin-exo-8,9-epoxide among the Alpha-class GSTs. Residues at these positions were mutated in human GSTA1-1 (Met-208, Leu-108), rat GSTA3-3 (Glu-208, His-108) and rat GSTA5-5 (Asp-208, Tyr-108): in the active rat GSTA5-5 to those in the inactive GSTA1-1; and in the inactive human GSTA1-1 and rat GSTA3-3 to those in the active rat GSTA5-5. These studies show clearly that, in all three GSTs, an aspartate residue at position 208 is a prerequisite for high activity in aflatoxin-exo-8,9-epoxide conjugation, although this alone is not sufficient; other residues in the vicinity, particularly residues 103-112, are important, perhaps for the optimal orientation of the aflatoxin-exo-8,9-epoxide in the active site for catalysis to occur.

Crystal structure of the FMN-binding domain of human cytochrome P450 reductase at 1.93 A resolution.

The crystal structure of the FMN-binding domain of human NADPH-cytochrome P450 reductase (P450R-FMN), a key component in the cytochrome P450 monooxygenase system, has been determined to 1.93 A resolution and shown to be very similar both to the global fold in solution (Barsukov I et al., 1997, J Biomol NMR 10:63-75) and to the corresponding domain in the 2.6 A crystal structure of intact rat P450R (Wang M et al., 1997, Proc Nat Acad Sci USA 94:8411-8416). The crystal structure of P450R-FMN reported here confirms the overall similarity of its alpha-beta-alpha architecture to that of the bacterial flavodoxins, but reveals differences in the position, number, and length of the helices relative to the central beta-sheet. The marked similarity between P450R-FMN and flavodoxins in the interactions between the FMN and the protein, indicate a striking evolutionary conservation of the FMN binding site. The P450R-FMN molecule has an unusual surface charge distribution, leading to a very strong dipole, which may be involved in docking cytochrome P450 into place for electron transfer near the FMN. Several acidic residues near the FMN are identified by mutagenesis experiments to be important for electron transfer to P4502D6 and to cytochrome c, a clear indication of the part of the molecular surface that is likely to be involved in substrate binding. Somewhat different parts are found to be involved in binding cytochrome P450 and cytochrome c.

NMR detection of slow conformational dynamics in an endonuclease toxin.

The cytotoxic activity of the secreted bacterial toxin colicin E9 is due to a non-specific DNase housed in the C-terminus of the protein. Double-resonance and triple-resonance NMR studies of the 134-amino acid 15N- and 13C/15N-labelled DNase domain are presented. Extensive conformational heterogeneity was evident from the presence of far more resonances than expected based on the amino acid sequence of the DNase, and from the appearance of chemical exchange cross-peaks in TOCSY and NOESY spectra. EXSY spectra were recorded to confirm that slow chemical exchange was occurring. Unambiguous sequence-specific resonance assignments are presented for one region of the protein, Pro65-Asn72, which exists in two slowly exchanging conformers based on the identification of chemical exchange cross-peaks in 3D 1H-1H-15N EXSY-HSQC, NOESY-HSQC and TOCSY-HSQC spectra, together with C alpha and C beta chemical shifts measured in triple-resonance spectra and sequential NH NOEs. The rates of conformational exchange for backbone amide resonances in this stretch of amino acids, and for the indole NH of either Trp22 or Trp58, were determined from the intensity variation of the appropriate diagonal and chemical exchange cross-peaks recorded in 3D 1H-1H-15N NOESY-HSQC spectra. The data fitted a model in which this region of the DNase has two conformers, NA and NB, which interchange at 15 degrees C with a forward rate constant of 1.61 +/- 0.5 s-1 and a backward rate constant of 1.05 +/- 0.5 s-1. Demonstration of this conformational equilibrium has led to a reappraisal of a previously proposed kinetic scheme describing the interaction of E9 DNase with immunity proteins [Wallis et al. (1995) Biochemistry, 34, 13743-13750 and 13751-13759]. The revised scheme is consistent with the specific inhibitor protein for the E9 DNase, Im9, associating with both the NA and NB conformers of the DNase and with binding only to the NB conformer detected because the rate of dissociation of the complex of Im9 and the NA conformer, NAI. is extremely rapid. In this model stoichiometric amounts of Im9 convert, the E9 DNase is converted wholly into the NBI form. The possibility that cis-trans isomerisation of peptide bonds preceding proline residues is the cause of the conformational heterogeneity is discussed. E9 DNase contains 10 prolines, with two bracketing the stretch of amino acids that have allowed the NA [symbol: see text] NB interconversion to be identified, Pro65 and Pro73. The model assumes that one or both of these can exist in either the cis or trans form with strong Im9 binding possible to only one form.

Practical aspects of the measurement of the diffusion of proteins in aqueous solution.

NMR diffusion measurements have been shown to be a useful tool for investigating the conformation and oligomeric state of proteins. Four main problems associated with making diffusion measurements of proteins in aqueous solution are identified and solutions proposed. The resulting experiment is demonstrated for an aqueous solution of hen egg white lysozyme.

NMR structural studies of glutathione S-transferase.

The use of nuclear magnetic resonance (NMR) spectroscopy for the structure determination of small proteins is now widely recognized; what is less frequently reported is the application of NMR techniques for high-resolution studies of large proteins (M(r) larger than 30 kD). We demonstrate here how an integrated approach, using heteronuclear NMR and X-ray crystallography, can provide useful and biologically important information for large protein systems. The dynamic features of the human Al-1 glutathione S-transferase and the role of the C-terminal region are being probed by NMR; in the X-ray crystal structure, the electron densities for this region of the protein are uninterpretable.

Determinants of the substrate specificity of human cytochrome P-450 CYP2D6: design and construction of a mutant with testosterone hydroxylase activity.

Cytochrome P-450 CYP2D6, human debrisoquine hydroxylase, metabolizes more than 30 prescribed drugs, the vast majority of which are small molecules containing a basic nitrogen atom. In contrast, the similar mouse protein Cyp2d-9 was first characterized as a testosterone 16alpha-hydroxylase. No common substrates have been reported for the two enzymes. Here we investigate the structural basis of this difference in substrate specificity. We have earlier used a combination of NMR data and homology modelling to generate a three-dimensional model of CYP2D6 [Modi, Paine, Sutcliffe, Lian, Primrose, Wolf, C.R. and Roberts (1996) Biochemistry 35, 4541-4550]. We have now generated a homology model of Cyp2d-9 and compared the two models to identify specific amino acid residues that we believe form the substrate-binding site in each protein and therefore influence catalytic selectivity. Although there are many similarities in active site structure, the most notable difference is a phenylalanine residue (Phe-483) in CYP2D6, which in the model is located such that the bulky phenyl ring is positioned across the channel mouth, thus limiting the size of substrate that can access the active site. In Cyp2d-9, the corresponding position is occupied by an isoleucine residue, which imposes fewer steric restraints on the size of substrate that can access the active site. To investigate whether the amino acid residue at this position does indeed influence the catalytic selectivity of these enzymes, site-directed mutagenesis was used to change Phe-483 in CYP2D6 to isoleucine and also to tryptophan. CYP2D6, Cyp2d-9 and both mutant CYP2D6 proteins were co-expressed with NADPH cytochrome P-450 reductase as a functional mono-oxygenase system in Escherichia coli and their relative catalytic activities towards bufuralol and testosterone were determined. All four proteins exhibited catalytic activity towards bufuralol but only Cyp2d-9 catalysed the formation of 16alpha-hydroxytesterone. Uniquely, the CYP2D6F483I mutant acquired the ability to metabolize testosterone to a novel product, which was identified by MS and proton NMR spectroscopy as 15alpha-hydroxytestosterone. NMR spin relaxation experiments were used to measure distances between the haem iron and protons of testosterone bound to the CYP2D6F483I mutant. These experiments demonstrate that very minor modifications to the active site structure of CYP2D6 can have a profound influence on the substrate specificity of the enzyme.