Comparison between the enzymatic activity, structure and substrate binding of mouse and human lecithin retinol acyltransferase.

Lecithin retinol acyltransferase (LRAT) is involved in the visual cycle where it catalyzes the formation of all-trans retinyl ester. The mouse animal model has been widely used to study LRAT. Primary sequence alignment shows 80% identity and 90% similarity between human and mouse LRAT. However, human LRAT has a proline at position 173 (hLRAT (P173)) while an arginine can be found at this position for the mouse protein (mLRAT (R173)). Moreover, residue 173 is important for the human protein since a substitution mutation of this residue to a leucine (P173L-hLRAT) caused night blindness in a patient. The present study was thus undertaken to determine whether mouse and human LRAT have a similar enzymatic activity, structure and substrate binding affinity using a truncated form of LRAT (tLRAT). The enzymatic activity and binding affinity to the substrate, all-trans retinol, of mtLRAT (R173) were found to be 2.7- and 3.9-fold lower, respectively, than that of htLRAT (P173). Moreover, the enzymatic activity of P173L-htLRAT is 6.3-fold lower compared to that of htLRAT (P173). Furthermore, a significant difference was observed between the intrinsic fluorescence emission as well as between the circular dichroism spectra of mtLRAT (R173) and htLRAT (P173). In addition, mtLRAT proteins are less thermostable than htLRAT proteins, which suggests that structural differences exist between the mouse and human proteins. Altogether, these data strongly suggest that the much lower catalytic activity of mtLRAT (R173) compared to that of htLRAT (P173) mostly results from differences between their structure, predominantly revealed by their dissimilar thermal stability, as well as their efficiency to bind all-trans retinol. Therefore, conclusions regarding the behavior of human LRAT based on measurements performed with mouse LRAT must be made with caution. Also, the much lower enzymatic activity of P173L-htLRAT compared to that of htLRAT (P173) might explain the night blindness of a patient carrying this mutation.

Systematic analysis of the expression, solubility and purification of a passenger protein in fusion with different tags.

Purification of recombinant proteins is often achieved using a purification tag which can be located either at the N- or C-terminus of a passenger protein of interest. Many purification tags exist and their advantages and limitations are well documented. However, designing fusion proteins can be a challenging task to get a fully expressed, soluble and highly purified passenger protein. Besides, there is a lack of systematic studies on the use of a single tag versus combined tags and on the effect of the position of the tags in the construct. In the present study, 9 different fusion proteins were expressed in Escherichia coli using some of the most commonly used purification tags: maltose-binding protein (MBP), glutathione S-transferase (GST) and polyHis tag. The expression and purification of N-terminus single-tagged fusion proteins (MBP, GST and polyHis) and fusion proteins with combined tags at different positions have been tested. Both the identity of the tag(s) and its position were found to have a strong effect on the expression, solubility and purification yields of the fusion proteins. Consequently, the different fusion proteins assayed have shown varying expression, solubility and purification yields, which were also dependent on the passenger protein. Therefore, there is a compelling need to design various fusion proteins with different single or combined tags to identify optimized constructions allowing to achieve high levels of expression, solubility and purification of the passenger protein.

Novel approaches to probe the binding of recoverin to membranes.

Recoverin is a protein involved in the phototransduction cascade by regulating the activity of rhodopsin kinase through a calcium-dependent binding process at the surface of rod outer segment disk membranes. We have investigated the interaction of recoverin with zwitterionic phosphatidylcholine bilayers, the major lipid component of the rod outer segment disk membranes, using both 31P and 19F solid-state nuclear magnetic resonance (NMR) and infrared spectroscopy. In particular, several novel approaches have been used, such as the centerband-only detection of exchange (CODEX) technique to investigate lipid lateral diffusion and 19F NMR to probe the environment of the recoverin myristoyl group. The results reveal that the lipid bilayer organization is not disturbed by recoverin. Non-myristoylated recoverin induces a small increase in lipid hydration that appears to be correlated with an increased lipid lateral diffusion. The thermal stability of recoverin remains similar in the absence or presence of lipids and Ca2+. Fluorine atoms have been strategically introduced at positions 4 or 12 on the myristoyl moiety of recoverin to, respectively, probe its behavior in the interfacial and more hydrophobic regions of the membrane. 19F NMR results allow the observation of the calcium-myristoyl switch, the myristoyl group experiencing two different environments in the absence of Ca2+ and the immobilization of the recoverin myristoyl moiety in phosphatidylcholine membranes in the presence of Ca2+.

Membrane fluidity is a driving force for recoverin myristoyl immobilization in zwitterionic lipids.

Recoverin is the only protein for which the phenomenon of calcium-myristoyl switch has been demonstrated without ambiguity. It is located in rod disk membranes where the highest content in polyunsaturated lipid acyl chains can be found. However, although essential to better understand the inactivation of the phototransduction process, the role of membrane fluidity on recoverin recruitment is unclear. We have therefore investigated the immobilization of the recoverin myristoyl moiety in the presence of phosphocholine bilayers using 2H solid-state NMR spectroscopy. Several lipids with different acyl chains were selected to investigate model membranes characterized by different fluidity. Immobilization of the recoverin myristoyl moiety was successfully observed but only in the presence of calcium and in specific lipid disordered states, showing that an optimal fluidity is required for recoverin immobilization.

Discriminating Lipid- from Protein-Calcium Binding To Understand the Interaction between Recoverin and Phosphatidylglycerol Model Membranes.

Recoverin is a protein involved in the phototransduction cascade by regulating the activity of rhodopsin kinase through a calcium-dependent binding process at the surface of rod outer segment disk membranes. Understanding how calcium modulates these interactions and how it interacts with anionic lipid membranes is necessary to gain insights into the function of recoverin. In this work, infrared spectroscopy allowed us to show that the availability of calcium to recoverin is modulated by the presence of complexes involving phosphatidylglycerol (PG), which in turn regulates its interactions with this negatively charged lipid. Calcium can indeed be sequestered into strongly bound complexes with PG and is thus sparingly available to recoverin. The thermal stability of recoverin then decreases, which results in weakened interactions with PG. By contrast, when calcium is fully available to recoverin, the protein is thermally stable, indicating that it binds two calcium ions, which results in favorable interactions with negatively charged lipids. Consequently, the protein induces an increase in the chain-melting phase transition temperature of PG, which is indicative of an enhanced lipid chain packing resulting from the peripheral location of the protein. The secondary structure of recoverin is not affected by its interactions with anionic membrane lipids. Similar results have been obtained with saturated and unsaturated anionic lipids. This work shows that the recruitment of recoverin at the surface of anionic lipid membranes is dependent on the availability of calcium.

Phosphatidylserine Allows Observation of the Calcium-Myristoyl Switch of Recoverin and Its Preferential Binding.

Recoverin undergoes a calcium-myristoyl switch during visual phototransduction. Indeed, calcium binding by recoverin results in the extrusion of its myristoyl group, which allows its membrane binding. However, the contribution of particular lipids and of specific amino acids of recoverin in its membrane binding has not yet been demonstrated. In the present work, the affinity of recoverin for the negatively charged phosphatidylserine has been clearly shown to be governed by a cluster of positively charged residues located in its N-terminal segment. Moreover, the calcium-myristoyl switch of recoverin was only observed upon binding onto monolayers of phosphatidylserine and not in the case of other anionic phospholipids. Fluorescence microscopy experiments with mixed lipid monolayers allowed confirmation of the specific binding of myristoylated recoverin to phosphatidylserine, whereas the extent of penetration of recoverin in phosphatidylserine monolayers was estimated by ellipsometry. A model has thus been proposed for the membrane binding of myristoylated recoverin in the presence of calcium.

Lipid Selectivity, Orientation, and Extent of Membrane Binding of Nonacylated RP2.

Retinitis pigmentosa 2 (RP2) is an ubiquitary protein of 350 residues. The N-terminus of RP2 contains putative sites of myristoylation and palmitoylation. The dually acylated protein is predominantly localized to the plasma membrane. However, clinically occurring substitution mutations of RP2 in photoreceptors lead to the expression of a nonacylated protein, which was shown to be misrouted to intracellular organelles using different cell lines. However, the parameters responsible for the modulation of the membrane binding of nonacylated RP2 (naRP2) are still largely unknown. The maximal insertion pressure of naRP2 has thus been determined after its injection into the subphase underneath monolayers of phospholipids, which are typical of photoreceptor membranes. These data demonstrated that naRP2 shows a preferential binding to saturated phospholipid monolayers. Moreover, polarization modulation infrared reflection absorption spectroscopy has allowed comparison of the secondary structure of this protein in solution and upon binding to phospholipid monolayers. In addition, simulations of these spectra have allowed to determine that the ß-helix of naRP2 has an orientation of 60° with respect to the normal, which remains unchanged regardless of the type of phospholipid. Finally, ellipsometric measurements of naRP2 demonstrated that its particular affinity for saturated phospholipids can be explained by its larger extent of insertion in this phospholipid monolayer compared to that in polyunsaturated phospholipid monolayers.

Enzymatic activity of Lecithin:retinol acyltransferase: a thermostable and highly active enzyme with a likely mode of interfacial activation.

Lecithin:retinol acyltransferase (LRAT) plays a major role in the vertebrate visual cycle. Indeed, it is responsible for the esterification of all-trans retinol into all-trans retinyl esters, which can then be stored in microsomes or further metabolized to produce the chromophore of rhodopsin. In the present study, a detailed characterization of the enzymatic properties of truncated LRAT (tLRAT) has been achieved using in vitro assay conditions. A much larger tLRAT activity has been obtained compared to previous reports and to an enzyme with a similar activity. In addition, tLRAT is able to hydrolyze phospholipids bearing different chain lengths with a preference for micellar aggregated substrates. It therefore presents an interfacial activation property, which is typical of classical phospholipases. Furthermore, given that stability is a very important quality of an enzyme, the influence of different parameters on the activity and stability of tLRAT has thus been studied in detail. For example, storage buffer has a strong effect on tLRAT activity and high enzyme stability has been observed at room temperature. The thermostability of tLRAT has also been investigated using circular dichroism and infrared spectroscopy. A decrease in the activity of tLRAT was observed beyond 70°C, accompanied by a modification of its secondary structure, i.e. a decrease of its α-helical content and the appearance of unordered structures and aggregated ß-sheets. Nevertheless, residual activity could still be observed after heating tLRAT up to 100°C. The results of this study highly improved our understanding of this enzyme.

The thermal stability of recoverin depends on calcium binding and its myristoyl moiety as revealed by infrared spectroscopy.

To evaluate the structural stability of recoverin, a member of the neuronal calcium sensor family, the effect of temperature, myristoylation, and calcium:protein molar ratio on its secondary structure has been studied by transmission infrared spectroscopy. On the basis of the data, the protein predominantly adopts α-helical structures (∼50-55%) with turns, unordered structures, and ß-sheets at 25 °C. The data show no significant impact of the presence of calcium and myristoylation on secondary structure. It is found that, in the absence of calcium, recoverin denatures and self-aggregates while being heated, with the formation of intermolecular antiparallel ß-sheets. The nonmyristoylated protein (Rec-nMyr) exhibits a lower temperature threshold of aggregation and a higher intermolecular ß-sheet content at 65 °C than the myristoylated protein (Rec-Myr). The former thus appears to be less thermally stable than the latter. In the presence of excess calcium ions (calcium:protein ratio of 10), the protein is thermally stable up to 65 °C with no significant conformational change, the presence of the myristoyl chain having no effect on the thermal stability of recoverin under these conditions. A decrease in the thermal stability of recoverin is observed as the calcium:protein molar ratio decreases, with Rec-nMyr being less stable than Rec-Myr. The data overall suggest that a minimal number of coordinated calcium ions is necessary to fully stabilize the structure of recoverin and that, when bound to the membrane, i.e., when the myristoyl chain protrudes from the interior pocket, recoverin should be more stable than in a Ca-free solution, i.e., when the myristoyl chain is sequestered in the interior.

Beneficial 'unintended effects' of a cereal cystatin in transgenic lines of potato, Solanum tuberosum.

BACKGROUND: Studies reported unintended pleiotropic effects for a number of pesticidal proteins ectopically expressed in transgenic crops, but the nature and significance of such effects in planta remain poorly understood. Here we assessed the effects of corn cystatin II (CCII), a potent inhibitor of C1A cysteine (Cys) proteases considered for insect and pathogen control, on the leaf proteome and pathogen resistance status of potato lines constitutively expressing this protein. RESULTS: The leaf proteome of lines accumulating CCII at different levels was resolved by 2-dimensional gel electrophoresis and compared with the leaf proteome of a control (parental) line. Out of ca. 700 proteins monitored on 2-D gels, 23 were significantly up- or downregulated in CCII-expressing leaves, including 14 proteins detected de novo or up-regulated by more than five-fold compared to the control. Most up-regulated proteins were abiotic or biotic stress-responsive proteins, including different secretory peroxidases, wound inducible protease inhibitors and pathogenesis-related proteins. Accordingly, infection of leaf tissues by the fungal necrotroph Botryris cinerea was prevented in CCII-expressing plants, despite a null impact of CCII on growth of this pathogen and the absence of extracellular Cys protease targets for the inhibitor. CONCLUSIONS: These data point to the onset of pleiotropic effects altering the leaf proteome in transgenic plants expressing recombinant protease inhibitors. They also show the potential of these proteins as ectopic modulators of stress responses in planta, useful to engineer biotic or abiotic stress tolerance in crop plants of economic significance.

Binding of a truncated form of lecithin:retinol acyltransferase and its N- and C-terminal peptides to lipid monolayers.

Lecithin:retinol acyltransferase (LRAT) is a 230 amino acid membrane-associated protein which catalyzes the esterification of all-trans-retinol into all-trans-retinyl ester. A truncated form of LRAT (tLRAT), which contains the residues required for catalysis but which is lacking the N- and C-terminal hydrophobic segments, was produced to study its membrane binding properties. Measurements of the maximum insertion pressure of tLRAT, which is higher than the estimated lateral pressure of membranes, and the positive synergy factor a argue in favor of a strong binding of tLRAT to phospholipid monolayers. Moreover, the binding, secondary structure and orientation of the peptides corresponding to its N- and C-terminal hydrophobic segments of LRAT have been studied by circular dichroism and polarization-modulation infrared reflection absorption spectroscopy in monolayers. The results show that these peptides spontaneously bind to lipid monolayers and adopt an α-helical secondary structure. On the basis of these data, a new membrane topology model of LRAT is proposed where its N- and C-terminal segments allow to anchor this protein to the lipid bilayer.

Influence of the physical state of phospholipid monolayers on protein binding.

Langmuir monolayers were used to characterize the influence of the physical state of phospholipid monolayers on the binding of protein Retinis Pigmentosa 2 (RP2). The binding parameters of RP2 (maximum insertion pressure (MIP), synergy and ΔΠ(0)) in monolayers were thus analyzed in the presence of phospholipids bearing increasing fatty acyl chain lengths at temperatures where their liquid-expanded (LE), liquid-condensed (LC), or solid-condensed (SC) states can be individually observed. The data show that a larger value of synergy is observed in the LC/SC states than in the LE state, independent of the fatty acyl chain length of phospholipids. Moreover, both the MIP and the ΔΠ(0) increase with the fatty acyl chain length when phospholipids are in the LC/SC state, whereas those binding parameters remain almost unchanged when phospholipids are in the LE state. This effect of the phospholipid physical state on the binding of RP2 was further demonstrated by measurements performed in the presence of a phospholipid monolayer showing a phase transition from the LE to the LC state at room temperature. The data collected are showing that very similar values of MIP but very different values of synergy and ΔΠ(0) are obtained in the LE (below the phase transition) and LC (above the phase transition) states. In addition, the binding parameters of RP2 in the LE (below the phase transition) as well as in the LC (above the phase transition) states were found to be indistinguishable from those where single LC and LE states are respectively observed. The preference of RP2 for binding phospholipids in the LC state was then confirmed by the observation of a large modification of the shape of the LC domains in the phase transition. Therefore, protein binding parameters can be strongly influenced by the physical state of phospholipid monolayers. Moreover, measurements performed with the α/ß domain of RP2 strongly suggest that the ß helix of RP2 plays a major role in the preferential binding of this protein to phospholipids in the LC state.

Lecithin retinol acyltransferase and its S175R mutant have a similar secondary structure content and maximum insertion pressure but different enzyme activities.

Recent work on Lecithin:retinol acyltransferase (LRAT) allowed to gather a large amount of information on its secondary structure, enzymatic properties and membrane binding. A truncated form of LRAT (tLRAT) as well as its S175R mutant leading to retinis pigmentosa, a severe form of retinal dystrophy, were studied to understand the role of this mutation on the dysfunction of this protein. Consistently with previous reports, the S175R-tLRAT mutant was shown to lack enzyme activity. However, very similar secondary structures probed by circular dichroism have been obtained with the S175R-tLRAT mutant and tLRAT. Moreover, similar values of maximum insertion pressure of the S175R-tLRAT mutant and tLRAT have been obtained using Langmuir monolayers, thus suggesting that the S175R mutation has no effect on the membrane binding properties of tLRAT. These findings leave open the possibility that the loss of enzymatic activity associated with the S175R mutant is related to loss of an essential nucleophile near the active site, or alternatively, to steric obstruction of the active site that impedes substrate binding.

Structural information and membrane binding of truncated RGS9-1 Anchor Protein and its C-terminal hydrophobic segment.

Visual phototransduction takes place in photoreceptor cells. Light absorption by rhodopsin leads to the activation of transducin as a result of the exchange of its GDP for GTP. The GTP-bound ⍺-subunit of transducin then activates phosphodiesterase (PDE), which in turn hydrolyzes cGMP leading to photoreceptor hyperpolarization. Photoreceptors return to the dark state upon inactivation of these proteins. In particular, PDE is inactivated by the protein complex R9AP/RGS9-1/Gß5. R9AP (RGS9-1 anchor protein) is responsible for the membrane anchoring of this protein complex to photoreceptor outer segment disk membranes most likely by the combined involvement of its C-terminal hydrophobic domain as well as other types of interactions. This study thus aimed to gather information on the structure and membrane binding of the C-terminal hydrophobic segment of R9AP as well as of truncated R9AP (without its C-terminal domain, R9AP∆TM). Circular dichroism and infrared spectroscopic measurements revealed that the secondary structure of R9AP∆TM mainly includes ⍺-helical structural elements. Moreover, intrinsic fluorescence measurements of native R9AP∆TM and individual mutants lacking one tryptophan demonstrated that W79 is more buried than W173 but that they are both located in a hydrophobic environment. This method also revealed that membrane binding of R9AP∆TM does not involve regions near its tryptophan residues, while infrared spectroscopy validated its binding to lipid vesicles. Additional fluorescence measurements showed that the C-terminal segment of R9AP is membrane embedded. Maximum insertion pressure and synergy data using Langmuir monolayers suggest that interactions with specific phospholipids could be involved in the membrane binding of R9AP∆TM.

Membrane binding properties of the C-terminal segment of retinol dehydrogenase 8.

Light absorption by rhodopsin leads to the release of all-trans retinal (ATRal) in the lipid phase of photoreceptor disc membranes. Retinol dehydrogenase 8 (RDH8) then reduces ATRal into all-trans retinol, which is the first step of the visual cycle. The membrane binding of RDH8 has been postulated to be mediated by one or more palmitoylated cysteines located in its C-terminus. Different peptide variants of the C-terminus of RDH8 were thus used to obtain information on the mechanism of membrane binding of this enzyme. Steady-state and time-resolved fluorescence measurements were performed using short and long C-terminal segments of bovine RDH8, comprising one or two tryptophan residues. The data demonstrate that the amphipathic alpha helical structure of the first portion of the C-terminus of RDH8 strongly contributes to its membrane binding, which is also favored by palmitoylation of at least one of the cysteines located in the last portion of the C-terminus.

Optimized transformation, overexpression and purification of S100A10.

As a member of the S100 protein family, S100A10 has already been purified. However, its purity, or even yield, have often not been reported in the literature. To facilitate future biophysical experiments with S100A10, we aimed to obtain it at a purity of at least 95% in a reasonably large amount. Here, we report optimized conditions for the transformation, overexpression and purification of the protein. We obtained a purity of 97% and performed stability studies by circular dichroism. Our data confirmed that the S100A10 obtained is suitable for experiments to be performed at room temperature up to several days.

Crystal structures of human Delta4-3-ketosteroid 5beta-reductase (AKR1D1) reveal the presence of an alternative binding site responsible for substrate inhibition.

The 5beta-reductases (AKR1D1-3) are unique enzymes able to catalyze efficiently and in a stereospecific manner the 5beta-reduction of the C4-C5 double bond found in Delta4-3-ketosteroids, including steroid hormones and bile acids precursors such as 7alpha-hydroxy-4-cholesten-3-one and 7alpha,12alpha-dihydroxy-4-cholesten-3-one. In order to elucidate the binding mode and substrate specificity in detail, biochemical and structural studies on human 5beta-reductase (h5beta-red; AKR1D1) have been recently undertaken. The crystal structure of a h5beta-red binary complex provides a complete picture of the NADPH-enzyme interactions involving the flexible loop B, which contributes to the maintenance of the cofactor in its binding site by acting as a "safety belt". Structural comparison with binary complexes of AKR1C enzymes, specifically the human type 3 3alpha-hydroxysteroid dehydrogenase (AKR1C2) and the mouse 17alpha-hydroxysteroid dehydrogenase (AKR1C21), also revealed particularities in loop B positioning that make the steroid-binding cavity of h5beta-red substantially larger than those of the two other enzymes. Kinetic characterization of the purified recombinant h5beta-red has shown that this enzyme exerts a strong activity toward progesterone (Prog) and androstenedione (Delta4) but is rapidly inhibited by these substrates once their concentrations reach 2-times their K(m) value. A crystal structure of the h5beta-red in ternary complex with NADPH and Delta4 has revealed that the large steroid-binding site of this enzyme also contains a subsite in which the Delta4 molecule is found. When bound in this subsite, Delta4 completely impedes the passage of another substrate molecule toward the catalytic site. The importance of this alternative binding site for the inhibition of h5beta-red was finally proven by site-directed mutagenesis, which demonstrated that the replacement of one of the residues delineating this site (Val(309)) by a phenylalanine completely abolishes the substrate inhibition. The results of this report provide structural insights into the substrate inhibition of h5beta-red by C19- and C21-steroids.

The crystal structure of human Delta4-3-ketosteroid 5beta-reductase defines the functional role of the residues of the catalytic tetrad in the steroid double bond reduction mechanism.

The 5beta-reductases (AKR1D1-3) are unique enzymes able to catalyze efficiently and in a stereospecific manner the 5beta-reduction of the C4-C5 double bond found into Delta4-3-ketosteroids, including steroid hormones and bile acids. Multiple-sequence alignments and mutagenic studies have already identified one of the residues presumably located at their active site, Glu 120, as the major molecular determinant for the unique activity displayed by 5beta-reductases. To define the exact role played by this glutamate in the catalytic activity of these enzymes, biochemical and structural studies on human 5beta-reductase (h5beta-red) have been undertaken. The crystal structure of h5beta-red in a ternary complex with NADP (+) and 5beta-dihydroprogesterone (5beta-DHP), the product of the 5beta-reduction of progesterone (Prog), revealed that Glu 120 does not interact directly with the other catalytic residues, as previously hypothesized, thus suggesting that this residue is not directly involved in catalysis but could instead be important for the proper positioning of the steroid substrate in the catalytic site. On the basis of our structural results, we thus propose a realistic scheme for the catalytic mechanism of the C4-C5 double bond reduction. We also propose that bile acid precursors such as 7alpha-hydroxy-4-cholesten-3-one and 7alpha,12alpha-dihydroxy-4-cholesten-3-one, when bound to the active site of h5beta-red, can establish supplementary contacts with Tyr 26 and Tyr 132, two residues delineating the steroid-binding cavity. These additional contacts very likely account for the higher activity of h5beta-red toward the bile acid intermediates versus steroid hormones. Finally, in light of the structural data now available, we attempt to interpret the likely consequences of mutations already identified in the gene encoding the h5beta-red enzyme which lead to a reduction of its enzymatic activity and which can progress to severe liver function failure.

Mouse 17alpha-hydroxysteroid dehydrogenase (AKR1C21) binds steroids differently from other aldo-keto reductases: identification and characterization of amino acid residues critical for substrate binding.

The mouse 17alpha-hydroxysteroid dehydrogenase (m17alpha-HSD) is the unique known member of the aldo-keto reductase (AKR) superfamily able to catalyze efficiently and in a stereospecific manner the conversion of androstenedione (Delta4) into epi-testosterone (epi-T), the 17alpha-epimer of testosterone. Structural and mutagenic studies had already identified one of the residues delineating the steroid-binding cavity, A24, as the major molecular determinant for the stereospecificity of m17alpha-HSD. We report here a ternary complex crystal structure (m17alpha-HSD:NADP(+):epi-T) determined at 1.85 A resolution that confirms this and reveals a unique steroid-binding mode for an AKR enzyme. Indeed, in addition to the interactions found in all other AKRs (van der Waals contacts stabilizing the core of the steroid and the hydrogen bonds established at the catalytic site by the Y55 and H117 residues with the oxygen atom of the ketone group to be reduced), m17alpha-HSD establishes with the other extremity of the steroid nucleus an additional interaction involving K31. By combining direct mutagenesis and kinetic studies, we found that the elimination of this hydrogen bond did not affect the affinity of the enzyme for its steroid substrate but led to a slight but significant increase of its catalytic efficiency (k(cat)/K(m)), suggesting a role for K31 in the release of the steroidal product at the end of the reaction. This previously unobserved steroid-binding mode for an AKR is similar to that adopted by other steroid-binding proteins, the hydroxysteroid dehydrogenases of the short-chain dehydrogenases/reductases (SDR) family and the steroid hormone nuclear receptors. Mutagenesis and structural studies made on the human type 3 3alpha-HSD, a closely related enzyme that shares 73% amino acids identity with the m17alpha-HSD, also revealed that the residue at position 24 of these two enzymes directly affects the binding and/or the release of NADPH, in addition to its role in their 17alpha/17beta stereospecificity.

Crystal structures of mouse 17alpha-hydroxysteroid dehydrogenase (apoenzyme and enzyme-NADP(H) binary complex): identification of molecular determinants responsible for the unique 17alpha-reductive activity of this enzyme.

Very recently, the mouse 17alpha-hydroxysteroid dehydrogenase (m17alpha-HSD), a member of the aldo-keto reductase (AKR) superfamily, has been characterized and identified as the unique enzyme able to catalyze efficiently and in a stereospecific manner the conversion of androstenedione (Delta4) into epitestosterone (epi-T), the 17alpha-epimer of testosterone. Indeed, the other AKR enzymes that significantly reduce keto groups situated at position C17 of the steroid nucleus, the human type 3 3alpha-HSD (h3alpha-HSD3), the human and mouse type 5 17beta-HSD, and the rabbit 20alpha-HSD, produce only 17beta-hydroxy derivatives, although they possess more than 70% amino acid identity with m17alpha-HSD. Structural comparisons of these highly homologous enzymes thus offer an excellent opportunity of identifying the molecular determinants responsible for their 17alpha/17beta-stereospecificity. Here, we report the crystal structure of the m17alpha-HSD enzyme in its apo-form (1.9 A resolution) as well as those of two different forms of this enzyme in binary complex with NADP(H) (2.9 A and 1.35 A resolution). Interestingly, one of these binary complex structures could represent a conformational intermediate between the apoenzyme and the active binary complex. These structures provide a complete picture of the NADP(H)-enzyme interactions involving the flexible loop B, which can adopt two different conformations upon cofactor binding. Structural comparison with binary complexes of other AKR1C enzymes has also revealed particularities of the interaction between m17alpha-HSD and NADP(H), which explain why it has been possible to crystallize this enzyme in its apo form. Close inspection of the m17alpha-HSD steroid-binding cavity formed upon cofactor binding leads us to hypothesize that the residue at position 24 is of paramount importance for the stereospecificity of the reduction reaction. Mutagenic studies have showed that the m17alpha-HSD(A24Y) mutant exhibited a completely reversed stereospecificity, producing testosterone only from Delta4, whereas the h3alpha-HSD3(Y24A) mutant acquires the capacity to metabolize Delta4 into epi-T.