Local changes in the catalytic site of mammalian histidine decarboxylase can affect its global conformation and stability Carlos Rodrı ´ guez-Caso 1 , Daniel Rodrı ´ guez-Agudo 1 , Aurelio A. Moya-Garcı ´ a 1 , Ignacio Fajardo 1 , Miguel A ´ ngel Medina 1 , Vinod Subramaniam 2, * and Francisca Sa ´ nchez-Jime ´ nez 1 1 Department of Molecular Biology and Biochemistry, Faculty of Sciences, Ma ´ laga, Spain; 2 Max Planck Institute for Biophysical Chemistry, Goettingen, Germany Mature, active mammalian histidine decarboxylase is a di- meric enzyme of carboxy-truncated monomers ( 53 kDa). By using a biocomputational approach, we have generated a three-dimensional model of a recombinant 1/512 fragment of the rat enzyme, which shows kinetic constants similar to those of the mature enzyme purified from rodent tissues. This model, together with previous spectroscopic data, al- lowed us to postulate that the occupation of the catalytic center by the natural substrate, or by substrate-analogs, would induce remarkable changes in the conformation of the intact holoenzyme. To investigate the proposed con- formational changes during catalysis, we have carried out electrophoretic, chromatographic and spectroscopic analy- ses of purified recombinant rat 1/512 histidine decarboxylase in the presence of the natural substrate or substrate-analogs. Our results suggest that local changes in the catalytic site indeed affect the global conformation and stability of the dimeric protein. These results provide insights for new alternatives to inhibit histamine production efficiently in vivo. Keywords: histidine decarboxylase; histamine; a-fluoro- methylhistidine; L -histidine methyl ester; pyridoxal phos- phate-dependent enzymes. Mammalian histidine decarboxylase (HDC), the enzyme responsible for the biosynthesis of histamine, is a pyridoxal 5¢-phosphate (PLP)-dependent enzyme that belongs to the evolutionary group II of L -amino acid decarboxylases [1–3]. Histamine is involved in several physiological responses (immune responses, gastric acid secretion, neurotransmis- sion, cell proliferation, etc.) and is also implicated in widely spread human pathologies (inflammation-related diseases, neurological disorders, cancer and invasion) [4–8]. In spite of the importance of these pathologies, HDC has not been fully characterized, and important questions about the regulation of the enzyme expression, sorting, processing, structural characterization and turnover remain unan- swered [9–15]. Mature HDC purified from mammalian tissues has been reported to be a dimer. Although the exact sequence of each monomer is not known, it is generally believed that the 74 kDa precursor is processed to a carboxy-truncated form of 53–58 kDa [16,17]. The N-terminus of the polypeptide (residues 1–480) exhibits a moderately high degree of identity with the porcine DOPA decarboxylase (DDC), another dimeric group II L -amino acid decarboxylase for which an X-ray structure has been solved [18]. Recently, we have characterized the catalytic mechanism of a recombin- ant carboxy-truncated form of the rat enzyme (fragment 1–512, also named HDC 1/512) [19], which shows kinetic constants similar to those of the mature enzyme purified from rodent tissues [16,17]. Mammalian HDC and DDC appear to share several catalytic features [19,20]. First, the PLP-enzyme internal Schiff base consists mainly of an enolimine tautomeric form (free holoenzyme). Second, Michaelis complex formation leads to a polarized ketoenamine form of the Schiff base. Third, after transaldimination, the coenzyme–substrate Schiff base exists mainly as an unprotonated aldimine. Finally, decarboxylation occurs and the free holoenzyme is recovered after protonation and a reverse transaldimination that releases the amine product. In spite of these shared features, the following observations [19] suggest that key structural differences must exist between the mammalian HDC catalytic site and those of the other group II L -amino acid decarboxylases: (a) HDC is the least efficient enzyme of its group; (b) mammalian HDC activity is less sensitive to the presence of thiol reducing compounds in the medium than other homologous and nonhomologous PLP-dependent decarboxylases; and (c) transaldimination from the internal to the external aldimine involves a higher degree of rotation in the torsion angle (v) than those observed for other homologous enzymes. The last observation is in agreement with the remarkably low catalytic efficiency of mammalian Correspondence to F. Sa ´ nchez Jime ´ nez, Departamento de Biologı ´ a Molecular y Bioquı ´ mica, Facultad de Ciencias, Universidad de Ma ´ laga, 29071 Ma ´ laga, Spain. Fax: + 34 95 2132000, Tel.: + 34 95 2131674, E-mail: kika@uma.es Abbreviations:DDC,aromatic L -amino acid decarboxylase or DOPA decarboxylase (EC 4.1.1.28); DOPA, L -3,4-dihydroxyphenylalanine; a-FMH, alpha-fluoromethylhistidine; a-FMHA, alpha-fluoromethyl- histamine; HDC, histidine decarboxylase (EC 4.1.1.22); HisOMe, L -histidine methyl ester; PLP, pyridoxal 5¢-phosphate. Enzymes:aromatic L -amino acid decarboxylase or DOPA decarb- oxylase (EC 4.1.1.28); histidine decarboxylase (EC 4.1.1.22). *Present address: Advanced Science and Technology Laboratory, AstraZeneca R & D Charnwood, Loughborough, UK. (Received 7 August 2003, revised 9 September 2003, accepted 12 September 2003) Eur. J. Biochem. 270, 4376–4387 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03834.x HDC, as this angle defines the degree of conjugation between the pyridine ring and the imine bond (maximum at v ¼ 0), which is important for catalytic efficiency of PLP- dependent enzymes [21]. In fact, the catalytic steps after the first transaldimination are rate-limiting in the entire catalytic reaction of mammalian HDC [19]. On the other hand, the crystal structure of DDC has shown that the catalytic sites of this enzyme are located in the interface between monomers [18]. Assuming that the catalytic sites of HDC are also in the interface between monomers, we postulated the hypothesis that the observed high degree of rotation of the torsion angle v induced by the presence of the substrate could produce a remarkable conformational change in the whole holoenzyme. In this work, we have used biocomputational methods to locate the catalytic center and predict the shape of the dimeric protein. The model reinforced our hypothesis proposed above. To further address this hypothesis, we have characterized the protein conformational changes duringcatalysis,byusingastrategysimilartothatused previously for the catalytic mechanism characterization [19]. Substrate analogs capable of blocking the HDC catalytic site at different catalytic steps are known, allowing the detection and analysis of the respective conformational states of the protein during the reaction. Thus, electro- phoretic, chromatographic and spectroscopic analyses carried out with purified preparations of the recombinant HDC 1/512 version in the presence and absence of substrate or substrate analogs have allowed us to characterize the conformational changes of the holoenzyme whenever a PLP substrate- or PLP product-like adduct is inside the catalytic center. We have also evaluated the stability of these conformational states against several agents that disrupt structure, i.e. detergent, thiol reductants, and temperature. Materials and methods Biocomputational analyses An initial model of the target protein (residues 5–479) was generated from the rat HDC sequence (Swiss-Prot accession number P16453) using the automated comparative protein modeling server SWISS - MODEL [22–24] in First Approach mode. The two pig DDC structures obtained from the Protein Data Bank (PDB ID 1JS3 and 1JS6) were used as templates. The docking program GRAMM [25] was used to build the structure of the dimeric rat HDC from the coordinate file provided by SWISS - MODEL . A low resolution docking was performed with a grid step of 6.8 A ˚ and 20° rotation increments. This type of docking is designed to overcome the problems of conformational flexibility and induced fit movements inherent to the formation of a protein–protein complex. The lowest energy docking solution was selected as representative of the dimer structure. Secondary structure of the model was calculated with the DSSP program [26]. Energy minimization calculations were performed with the program XPLOR [27] in an SGI Altix 3000 under GNU/ L INUX R ED H AT 7.2. Three-dimensional visualization and analysis were per- formed with RASMOL [28] and SWISS - PDBVIEWER [22]. Molecular graphics were generated using MOLSCRIPT [29], RASTER 3 D [30] and PYMOL [31]. Recombinant HDC purification procedures and enzyme activity assay The DNA encoding for residues 1–512 of rat HDC [32] was subcloned in the pET-11a vector (Novagen, USA). The recombinant plasmid transformed into the Escherichia coli BL21(DE3)pLysS strain. Transformed cultures were induced to express the HDC 1/512, which was purified by applying three chromatographic steps (Phenyl-Sepharose CL-4B, DEAE interchange, and hydroxyapatite). The final preparations were dissolved in 50 m M potassium phosphate, 0.1 m M PLP, pH 7.0. Purity of the HDC 1/512 construct was checked by Coomassie blue staining and Western blotting, and was higher than 95% in the final preparations. HDC activity was assayed by following 14 CO 2 release from L -His-[U- 14 C] (American Radiolabeled Chemicals, USA). Analogs and histidine were provided by Sigma-Aldrich (Spain). All of these procedures (overexpression, purifica- tion and analysis) are described extensively elsewhere [19]. When required, the enzyme was concentrated in different Amicon (USA) ultrafiltration systems (cut-off between 10 and 30 kDa) depending on the initial volume. To avoid interference by free PLP, the final preparation was subjected to size-exclusion gel chromatography by using a Sephadex G25 column or, alternatively, a Sephacryl HiPrep S-200 column (Pharmacia Biotech, Sweden) in 50 m M potassium phosphate buffer (pH 7) immediately before starting spectroscopic analyses. Chromatographic analysis Purified HDC preparations were incubated at room tem- perature in either the presence or absence of 1 m M alpha- fluoromethylhistidine (a-FMH) for 1 h. After incubation, protein was subjected to size-exclusion gel chromatography on a HiPrep Sephacryl S-200 high-resolution column, pre- equilibrated with 50 m M potassium phosphate and coupled to a FPLC system (Pharmacia). Absorbance at 280 nm was monitored continuously to detect protein peaks. M r swere calculated after calibration of the column with the following molecular-mass standards (all from Sigma): alcohol dehy- drogenase (M r 14 2000), bovine serum albumin (M r 65 000), chymotrypsinogen A (M r 25 000) and cyto- chrome c (M r 12 400). Electrophoretic and Western blotting analysis Aliquots of the purified enzyme (1–3 lg) were incubated at room temperature for 1 h in the presence of either 1 m M histidine-analogs or 10 m M histidine, or in their absence (untreated enzyme). All solutions were adjusted to pH 7. Ten millimolar histidine (more than 20-fold the previously reported K m value) was chosen to maximize the percentage of enzyme taking part in the enzyme–substrate complex. When indicated, 2-mercaptoethanol was added after 55 min of incubation to yield a final concentration of 80 m M , followed immediately by loading buffers. For conventional denaturing SDS/PAGE experiments, the protocol described by Laemmli was followed [33]. Samples were boiled and Ó FEBS 2003 Conformational changes of histidine decarboxylase (Eur. J. Biochem. 270) 4377 stored at )20 °C until their use. However, for semidena- turing SDS/PAGE experiments, samples were not boiled and the loading buffers lacked 2-mercaptoethanol. Two different loading buffers were used. Loading buffer A (pH 7.4) lacked SDS, while loading buffer B (pH 6.8) contained 0.7% SDS. Immediately after addition of the respective loading buffer, electrophoresis was performed at 4 °C in a 7% acrylamide SDS-free PAGE gel (semidena- turing condition A) or in a 7% acrylamide 0.1% SDS/ PAGE gel (semidenaturing conditions B), both lacking stacking gel. The running buffer always contained 0.1% SDS. To avoid problems with any minor contaminant band in the purified preparations, electrophoretic results were visualized by Western blotting, following a protocol described elsewhere [15]. The anti-HDC K9503 serum was generously supplied by L. Persson (Lund University, Sweden). Recombinant Strep-tag unstained standards 161–0362 (Bio-Rad, USA) were used as references for electrophoretic migrations. In each experiment, treatments were carried out in parallel on aliquots of the same purified enzyme preparations. The results of semidenaturing gels shown here are representative of at least four different independent experiments. Spectroscopic analysis Absorption spectra were obtained using a HP8452A diode array spectrophotometer (Hewlett-Packard, USA). The acquisition time for each absorption spectrum was 2 s. Fluorescence spectra were obtained in a QuantaMaster SE spectrofluorimeter (Photon Technology International Inc., USA). Integration time was 0.1 sÆnm )1 ; three spectra were averaged. CD spectroscopy was carried out with a Jasco J-715 spectropolarimeter at a scan speed of 50 nmÆmin )1 ; 10 spectra were averaged. Analogs were used at the specified final concentrations. Unless otherwise indicated, all spectroscopic measurements were carried out at room temperature. Fluorescence (300–400 nm, excita- tion at 275 nm) was not detectable from solutions of 10 m M histidine or 1 m M analogsinanenzyme-free buffer. CD signals (195–250 nm) were not detectable from solutions of 1 m M a-FMH, 14 l M PLP or both together in an enzyme-free buffer. Results and discussion Protein modeling of rat HDC dimer predicts the location of its catalytic site We have previously observed that transaldimination from the internal to the external aldimine of HDC involves a higher degree of rotation in the torsion angle (v)thanin other homologous enzymes [19]. This observation led us to postulate that the interaction of HDC with its substrate could induce significant conformational changes, at least in the catalytic center environment. In addition, this experimental observation also indicated that some differ- ences must exist in the relative position of the cofactor– substrate adduct when compared with homologous enzymes. The moderately high degree of identity between mam- malian HDC and DDC (> 50%), the published structural models [2,3] and the reported crystal structure of porcine DDC [18], in combination make it possible to apply comparative modeling methods to overcome the lack of reported crystal structures for mammalian HDC. Figure 1 shows the lowest energy predicted 3D structure of the rat HDC monomer and its quaternary structure, obtained after docking and energy minimization calculations carried out as described in the Materials and methods section. One thousand dimer structures were generated without any manual fitting at all. At least the first 200 were examined and found to be very similar. They are conformed as twofold axial symmetry dimers similar to those described for the crystal structures of pig DDC [18]. Nevertheless, it predicts a more occluded catalytic center when compared with the DDC crystal structure. This is not surprising, as HDC has a more restrictive catalytic center, and it could be suspected from experimental data previously shown [19]. Figure 2 shows the alignment of rat HDC and pig DDC primary sequences, as well as the distribution of a-helices and b-sheets in both the crystal structure of pig DDC and that estimated from the rat HDC 3D model. This figure stresses that the pattern of secondary structures in both enzymes is very similar, in spite of their differences in primary structure, as expected. The complete distribution of consensus secondary structure estimated from the model is as follows: 39% of a-helices, 9% of b-sheets, 12% of turns, 21% of random coil and 19% of other structures. These estimations are similar to those obtained from rat HDC primary sequence, and they are consistent with estimations from far-UV CD spectra (controls at 20 °CinFig.9,and not shown here to avoid redundancy). In spite of the lack of overall sequence identity, a common PLP-binding motif consisting of clusters of conserved residues is present in decarboxylases belonging to groups I, II and III [2]. The PLP-binding site of Morganella morganii AM-15 HDC was experimentally located in its K233 residue [34]. This lysine residue is extremely conserved, and corresponds to K303 of pig and rat DDC, also previously shown to play this role [18,35], and to K308 in rat HDC. A histidine residue, corres- ponding to H197 in rat HDC, also is strictly conserved in group II of mammalian L -amino acid decarboxylases, in which it seems to be stacked in front of the cofactor pyridine ring [18]. Thus, our model combined with the previously reported structural and mechanistic character- ization of these enzymes [19,20,36] allowed us to locate the HDC catalytic center at the interface between the monomers (Fig. 1), as is the case of DDC [18]. One of the monomers (monomer A) would contain the major part of the catalytic site pocket, including K308 and H197 in rat HDC. Figure 3 shows the most important residues close to H197 and K308, as predicted by our 3D model. The predicted catalytic center contains a number of polar residues (D276, N305, H197 and K308). All of these residues are strictly conserved in mammalian DDC (see Fig. 2), where they take similar positions to those predicted in HDC. In mammalian DDC, D271 and N300 (counterparts of D276 and N305 in rat HDC) are predicted to interact with the imidazole ring and the phosphate group of PLP, respectively [18,36,37]. It is noteworthy that in spite of the high flexibility of the 4378 C. Rodrı ´ guez-Caso et al. (Eur. J. Biochem. 270) Ó FEBS 2003 fragments (see Fig. 2) containing most of these residues, our prediction located them very close in space and at similar positions to those described for DDC [18]. Therefore, it seems to be likely that they play a similar role in mammalian HDC, delimiting what could be termed as the PLP interaction region (PLP-IR, see Fig. 3). After formation of the holoenzyme, the substrate (histi- dine) should enter the catalytic site from the bottom part of Fig. 3 through a space delimited by the PLP-IR and a region in which our model predicts the location of several residues of both monomers able to establish hydrophobic interactions; for instance, Y84 (Fig. 3) and the fragment PAL 85–87 from monomer A (the latter not shown in Fig. 3 for clarity), and F331, I364 and L356 from monomer B. These predictions are in agreement with previous bio- physical and kinetic studies from our laboratory indicating that, in the internal aldimine form, the catalytic site of HDC is enriched in hydrophobic residues, leading to an enolimine tautomeric form of PLP [19]. A hydrophobic channel for the substrate has also been proposed for DDC [38,39]. However, the specific hydrophobic residues of monomer B contributing to this region of DDC (I101 and F103, as deduced from data in reference [18]) are different, as expected from the structural differences between their respective substrates. It is also noteworthy that some of the closest hydrophobic residues of monomer B (for instance, F331) are part of or close to the Ôflexible loopÕ described for mammalian DDC, which could not be solved from the crystal structure (residues 328–339 in Fig. 2). In pig DDC, a conformational change of this loop in response to substrate binding has been demonstrated [18,37]. A similar role of its counterpart in mammalian HDC in relation to the conformational change described in the present work could be suspected. Finally, from this model we have predicted that the occupation of the catalytic center by the polar substrate or an analog through a hydrophobic channel could induce drastic conformational changes of the holoenzyme that would probably affect, at least locally, interactions at the monomer interface and vice versa. This reinforces our starting hypothesis. Fig. 1. Three-dimensional model of rat HDC structure. The model was generated from res- idues 2–475 of the primary sequence, as des- cribed in Material and methods section. (A) and (B) Surface representations of the dimeric form, one monomer in white and the other in red. (C) Surface representation of one mono- mer in white. The predicted interface between monomers is shown in red. The active site residues, K308 and H197, are shown in blue. W and Y residues are depicted in green. In (A) and(C),thewhitemonomerisshowninthe same position. (B) is left-twisted around the z-axis with respect to (A) to show the localization of K308 and H197 within the monomer interface. A double-headed arrow in (A) indicates the maximum distance determining the Stokes’ radius predicted for the dimer. Ó FEBS 2003 Conformational changes of histidine decarboxylase (Eur. J. Biochem. 270) 4379 Active HDC1/512 is a dimer and the presence of a substrate analog in its active site diminishes its Stokes’ radius HDC1/512 is a recombinant carboxy-truncated form of the rat enzyme that we have previously used to study structure/ function relationships of the mature HDC [10,13,14,19,40]. We have recently shown that HDC1/512 has kinetic constants similar to those of the mature enzyme purified from rodent tissues [19]. Figure 4 shows the results of size- exclusion gel chromatography of purified HDC1/512. A major peak (M r 107 000) was observed for the untreated enzyme. Some inactive higher molecular weight HDC aggregates were detected by Western blots (data not shown). In fact, the purified enzyme preparations slowly tend to form inactive aggregates when incubated at room tempera- ture or higher (C. Rodrı ´ guez-Caso, D. Rodrı ´ guez-Agudo, A. A. Moya-Garcı ´ a, M. A ´ . Medina, V. Subramanian & F. Sa ´ nchez-Jime ´ nez, unpublished observations). Enzymatic activity was only detectable in fractions corresponding to the major peak. These results indicated that the quaternary Fig. 2. Alignment of rat HDC and pig DDC sequences. The 5–479 fragment of rat HDC (Swiss-Prot accession number P16453) and pig- DDC sequence from the Protein Data Bank (PDB ID 1JS6) were aligned using the ProdModII method in Swiss-Model server (http:// www.expasy.org/swissmod/SWISS-MODEL.html). The secondary structure of the crystallized pig DDC and that predicted from the rat HDC 3D model are shown below the aligned sequences: h, helix; s, sheet; ?, unpredicted conformation in crystallized pig-DDC. Fig. 3. Structural neighborhood of the PLP-binding site. The most relevant residues closer than 7 A ˚ to H197 or K308, as predicted by our 3D apoenzyme model, are depicted. The line delimits the PLP inter- action region (PLP-IR). The putative entrance for the substrate between the PLP-IR and the hydrophobic region is marked with a star. Fig. 4. Size-exclusion gel chromatography of the free-holoenzyme and the a-FMH-treated enzyme. Purified enzyme was incubated for 1 h in the presence or absence of 1 m M a-FMH and submitted to size- exclusion gel chromatography, using a FPLC system as described in the Material and methods. No monomer was observed. Fractions 1 to 38 represented void volume. For the free-holoenzyme extracts, enzyme activity was coincident with the major peak. Arrows indicate the peaks of the M r standards: 1, alcohol dehydrogenase (M r 142 000); 2, bovine serum albumin (M r 65 000); 3, chymotrypsinogen A (M r 25 000); 4, cytochrome c (M r 12 400). 4380 C. Rodrı ´ guez-Caso et al. (Eur. J. Biochem. 270) Ó FEBS 2003 structure of the active recombinant purified enzyme used in this work is, indeed, a dimer, as also deduced for the native enzyme purified from natural sources [17]. Figure 5 shows a scheme of the HDC reaction and the specific steps interfered by the substrate analogs histidine methyl ester (HisOMe) and a-FMH, deduced from previous reports in the literature [19,41]. HisOMe, a reversible competitive inhibitor, blocks the reaction after formation of an external aldimine tautomeric form very similar to that of the PLP–histidine adduct (Fig. 3 and [19]). By using the substrate analog a-FMH, the reaction can proceed (inclu- ding the decarboxylation step) to form a-fluoromethylhis- tamine (a-FMHA; Fig. 5 and [41]). In the case of fetal rat HDC, this reaction has been reported to proceed much slower than with the natural substrate histidine [42]. Nevertheless, after decarboxylation and elimination of the fluoride, a reverse transaldimination can occur, so that an enamine form of the product can either leave the catalytic site or react again with the internal aldimine to form a PLP- adduct covalently attached to the catalytic center. It has been proposed that the occurrence frequency of these two possibilities depends on how long the enamine remains in the active site and its rate of attack on the aldimine bond, and can be modified by slight differences in the position of the residues [41]. As covalent binding was proposed to occur, at least partially, between PLP–a-FMHA derivatives and the enzyme, this analog is considered as a suicide inhibitor of PLP-dependent HDC [43]. Complexes III and VII in Fig. 5 would correspond to the major final forms stabilized at short-term in the catalytic site after the reactions with HisOMe and a-FMH, respectively. Based on the shape of the predicted dimeric HDC (Fig. 1), a conformational change affecting the monomer interface would change the Stokes’ radius of the protein, as the major diameter of the dimer is predicted to be given by the distance between the carboxy termini of both monomers (from the lower left to the upper right of Fig. 1B), which in turn is dependent on the dimerization surface conformation. Gel filtration is a validated method to distinguish changes in the Stokes’ radius of an oligomeric enzyme. Among the different compounds tested (the natural substrate and substrate-analogs), a-FMH is the only one that can accumulate a stable PLP-adduct cova- lently bound to the enzyme (Fig. 5), thus being able to withstand a gel filtration procedure. An apparent reduction of the M r was indeed observed in the a-FMH-treated samples (Fig. 4), suggesting a treatment-induced change of the dimeric structure to a conformation with diminished Stokes’ radius. Analog-treated HDC shows altered electrophoretic mobility under semidenaturing conditions It is well known that quaternary structure of proteins is frequently established, at least partially, through hydropho- bic interactions that can be weakened by SDS and other detergents. Thus, electrophoresis of the samples carried out in the presence of SDS as the only denaturing agent could reveal: (a) reinforcements of monomer associations, as only the strongest associations could survive the denaturing agent; and (b) any change in the volume of a single polypeptide (or a polypeptide association). We analyzed HisOMe- and a-FMH-treated samples under the semide- naturing conditions described in the Material and methods section. Figure 6 shows that treatments with analogs change the relative electrophoretic mobility of untreated HDC under semidenaturing conditions, supporting our hypothe- sis on global conformational changes of HDC induced by the presence of analogs in the active center. Furthermore, these findings also seem to indicate that the analogs can Fig. 5. Scheme of the HDC reactions with the natural substrate histidine and the histidine- analogs HisOMe and a-FMH. This scheme was built from the major forms for each step mentioned in the text deduced from the pre- vious information ([18,19] and the present results). The absorption spectrum maxima described for the tautomeric forms mentioned in the text are indicated in brackets. The pro- posed major forms reached with HisOMe and a-FMH are shown inside dashed boxes. T, transaldimination steps. Ó FEBS 2003 Conformational changes of histidine decarboxylase (Eur. J. Biochem. 270) 4381 induce changes of the enzyme to conformational states more resistant to denaturation by detergents, especially in the case of conformations adopted during the external aldimine state (HisOMe-treated samples). Absorption spectra of HDC during reaction with histidine and analogs reveal details of the catalytic mechanism Taking into account the extremely low reaction rate reported for mammalian HDC [19], and especially in the presence of a-FMH [42], it is expected that different steps of the reaction can be distinguished as a function of time in a conventional UV-visible spectrophotometer. Before Fig. 6. Western blots of the free-holoenzyme and analog-treated samples under different semidenaturing and denaturing conditions. Aliquots of the same purified preparation were incubated for 1 h in the absence (control, C) or presence of 1 m M histidine analogs (histidine methyl ester, HisOMe; a-fluoromethylhistidine, a-FMH) and submitted to the semidenaturing conditions A (A, loading buffer A) and B (B) and (C) described in the Materials and methods section. In (C), samples had been treated for 5 min with 2-mercaptoethanol. (D) corresponds to a conventional denaturing SDS/PAGE electrophoresis. In this case, molecular mass standards are shown as M r · 10 )3 . In (A–C), bands are designed according to their relative electrophoretic mobilities as F (the fastest mobility), S (the slowest mobility) and I (intermediate mobilities). Fig. 7. Changes with time of the absorption spectra of HDC in the presence of the natural substrate histidine or histidine-analogs. Con- centrated, neutralized stocks (6.36 lL) of the natural substrate histi- dine (A), HisOMe (B) or a-FMH (C) were added to 70 lLofa 13–14 l M solution of purified and gel filtered protein to reach final concentrations of 10 m M histidine or 1 m M analogs. 4382 C. Rodrı ´ guez-Caso et al. (Eur. J. Biochem. 270) Ó FEBS 2003 addition of any substrate, we observed the same PLP absorption profile previously reported for the free holo- enzyme (Fig. 7, untreated samples in all panels, and [19]): a major enolimine form (maximum at  335 nm, complex I in Fig. 5) and a minor ketoenamine form (maximum at  420 nm) of the internal aldimine. However, a few seconds after substrate or analog addition, a new peak arose at 390 nm (Fig. 7, all panels), which must corres- pond to accumulation of enzyme molecules at the external aldimine stage, as reported previously ([19], see also complex III in Fig. 5). The peak was observed not only with histidine but also with both analogs, corresponding to their reported action mechanisms. After 1 min, this 390-nm peak could still be observed in all cases. As mentioned before, this external aldimine complex (com- plex III) is the final product of the reaction with HisOMe. In fact, after 5 min, the spectra of the HisOMe-treated samples had stabilized with the 390-nm peak as the major and final one. In the reaction in the presence of an excess of the natural substrate histidine, a shoulder around 430 nm is also observed (Fig. 7A), which must correspond to Michaelis complexes (complex II in Fig. 5) and/or to ketoenamine forms of external aldimines (that is to say, to other stages of the reaction), many of them having typical absorption maxima around 430 nm [44]. On the other hand, when using a-FMH, accumulation of a PLP-derivative with an absorption maximum at  345 nm can be clearly observed after the reaction had passed through a maximum concentration of the external aldimine complex (Fig. 7C). From the first minutes on, this peak became the major one clearly observed in the a-FMH-treated enzyme, suggesting that it corresponds to a major molecular form of the PLP–a-FMHA derivative that was rather stable for at least the first hour of treatment (complex VII in Fig. 5). Nevertheless, as the absorption at wavelengths higher than 400 nm was even increasing with the treatment period, other noncovalently bound PLP adducts cannot be ruled out. Bhattacharjee and Snell [41], working with the bacterial M. morganii HDC enzyme, proposed that the covalently bound adduct can be converted slowly into other PLP adducts with absorption maxima higher than 400 nm (not shown in Fig. 5), which can be removed from the catalytic center by dialysis and boiling. The absorption maximum observed at 345 nm, as well as the shape of the final spectra, are extremely similar to that reported [41] for the product of Fig. 8. Fluorescence emission (excitation at 274 nm) of the free-HDC holoenzyme and the enzyme treated with the natural substrate or analogs. Ten microliters of 50 m M potassium phosphate pH 7 (control condition), or 10 lL of concentrated neutralized stocks of the natural substrate histidine (A), HisOMe (B) or a-FMH(CandD)wereaddedto90lLofa6.5–7 l M solution of purified and gel filtered protein to reach final concentrations of 10 m M histidine or 1 m M analogs. Stability of the control spectra were assessed by three different determinations. Ó FEBS 2003 Conformational changes of histidine decarboxylase (Eur. J. Biochem. 270) 4383 the enamine reaction with the internal aldimine, that is, a PLP-a-FMHA derivative covalently bound to the enzyme of Gram negative microorganisms (also Fig. 5, complex VII). As far as we know, this is the first time that the spectra of the PLP-adducts during the reaction of a mammalian enzyme with a-FMH are recorded. Previous studies of the reaction were carried out on partially purified extracts of the rat enzyme, working with radiolabeled a-FMH [42]. These authors deduced that one-third of the decarboxylated a-FMH products seemed to be covalently attached to the enzyme. Our data are also consistent with this proposal. As the inhibitor is in excess with respect to the enzyme, successive decarboxylations would accumulate the covalent adduct in the assay period, thus becoming the major form detected in the spectra. Summarizing, data shown in Fig. 7 reinforce our previ- ous findings on the tautomeric forms of the internal aldimine in the free holoenzyme [19] and the reaction carried out with the suicide analog a-FMH [40]. The conformational changes induced in HDC by histidine and histidine-analogs involve alterations in the environment of the aromatic amino acid residues of the protein Conformational changes of proteins can modify the intrin- sic fluorescence from their aromatic residues. Both HisOMe and a-FMH block the HDC catalytic center after the first transaldimination step (after external aldimine formation), leading to catalytic sites occupied by different PLP-adducts. HDC 1/512 contains 11 W and 13 Y residues. From them, six W and seven Y residues are predicted to be in the monomer interface (Fig. 1). These represent most of the W (86%) and Y (64%) residues predicted to be exposed to the monomer surface (seven W and 11 Y). Thus, it seemed likely that a conformational change in the monomer interaction surface could be reflected in the fluorescence measurements of the free holoenzyme and the enzyme after addition of histidine and histidine analogs. An increase in W fluores- cence intensity is most commonly associated with reduced exposure of the W residues to the solvent, i.e. a transition from a predominantly solvent-exposed to a more hydro- phobic situation brought about by a conformational change. Although an increase in interaromatic energy transfer is also possible, this is not the most likely reason for a fluorescence increase. Indeed, when there are several aromatic residues in close proximity, interactions quenching the fluorescence tend to occur [45]. Figure 8 shows the fluorescence emission spectra (from 300 to 400 nm, excitation at 274 nm) of HDC holoenzyme obtained before and after substrate or analog addition. In all cases, increases of fluorescence were observed to occur within the first minute after the compound addition, suggesting that, indeed, all of them are able to induce structural changes that shield aromatic residues from solvent interactions. It is Fig. 10. Thermal denaturation profile at 222 nm of the free HDC holoenzyme and the analog-treated samples. Aliquots of a 3-l M solution of purified and gel filtered enzyme were treated with or without the histidine-analogs at 1 m M final concentration. CD spectra were recorded after stabilization of the samples at the different assayed temperatures. Stabilization times were 2–5 min. Fig. 9. Thermal denaturation of the free HDC holoenzyme and the a-FMH-treated enzyme. Aliquots of a 3-l M solution of purified and gel filtered enzyme solution were treated (B) or not (A, free holoenzyme) with 1 m M a-FMH. CD spectra were recorded after stabilization of the samples at the different assayed temperatures. Stabilization times were 2–5 min. 4384 C. Rodrı ´ guez-Caso et al. (Eur. J. Biochem. 270) Ó FEBS 2003 noteworthy that these conformational changes take place within the time period in which most of the enzyme is passing through the external aldimine state, irrespective of the substrate or analog added (Fig. 7). At least with both histidine and HisOMe, a shift to the blue in the emission maximum could be clearly observed, supporting a transition to a more hydrophobic environment. On the other hand, an increased, red-shifted fluorescence is observed in the a-FMH-treated enzyme after 30 min of the reaction (Fig. 8), when most of the enzyme is forming the covalently bound adduct with the PLP-a-FMHA derivative (Fig. 7). These observations suggest that the conformational change involves alterations in the environ- ment of the aromatic amino acid residues of the protein. Resistance of the different conformational states to thermal denaturation Results obtained with semidenaturing electrophoresis seem to indicate that histidine analogs can induce changes in the enzyme to conformational states more resistant to dena- turation by detergents. To test whether these analog- induced conformational states are also more resistant to thermal denaturation, we carried out CD analysis of HDC under several treatments and at different temperatures. Changes in the secondary structure of the enzyme during temperature-induced unfolding can be deduced by using this approach. Figure 9 shows far-UV CD spectra (190–250 nm) of the free-holoenzyme (Fig. 9A) and the a-FMH-treated enzyme (Fig. 9B) incubated at different temperatures after treatment. Unfolding of the protein can be deduced from changes in the spectra observed with increasing temperatures (Fig. 9), as well as from the changes observed in the ellipticity at 222 nm (Fig. 10). Nevertheless, these temperature-induced changes were more evident for the free holoenzyme than for the a-FMH-treated enzyme sample, indicating that the enzyme that had a covalently bound PLP-adduct was more resistant to temperature- induced denaturation. Scarce 2D structural information can be obtained from similar experiments made with HisOMe, due to the basal CD absorption of this compound. However, the observed increasing trend in the ellipticity at 222 nm of the HisOMe-treated enzyme preparations (Fig. 10) suggests that the reversible inhibitor was not able to protect the enzyme against thermal denaturation. The increased resistance of the a-FMH-treated protein to thermal denaturation would indicate that the covalent binding of the adduct to the catalytic center could fix some secondary structure in the enzyme, suggesting several interaction points for the adduct within the catalytic center in addition to those established by the internal aldimine alone. From the shape of the a-FMH-treated protein spectra, stabilization of some a-helix by the cofactor adduct could be suspected (Fig. 9). It is noteworthy that most of the catalytic site is predicted to adopt a-helix and random coil secondary structures (predictions not shown, also derived from Figs 1 and 2). Concluding remarks Since the initial suggestions by Pauling in 1948 and the later formulation of the induced-fit hypothesis by Koshland, it is well established that the interactions between an enzyme and its substrate induce conformational changes at the active site. In most cases, these conformational changes are only local and relatively small. However, for some enzymes these changes may be remarkable. This seems to be the case of mammalian HDC. We had previously observed that transaldimination from the internal to the external aldimine of HDC involves a higher degree of rotation in the torsion angle (v)than those observed in other homologous enzymes [19]. This observation allowed us to postulate the hypothesis that the interaction of HDC with its substrate could induce signi- ficant conformational changes, at least, in the catalytic center environment. By using biocomputational tools, we have located the catalytic center of this enzyme in the interface between the monomers (Fig. 1). Furthermore, we have predicted that the occupation of the catalytic center by the substrate or an analog could induce global conforma- tional changes in the intact holoenzyme, reinforcing our starting hypothesis. Evidence has also been obtained suggesting differences between these conformations before and after decarboxylation, revealed by using HisOMe and a-FMH, respectively. As the catalytic site of HDC is located at the dimer interface, small changes in the interaction surface between monomers could affect the exposure of the catalytic pocket content to the medium. During reaction, PLP-substrate and PLP-product adducts are not covalently bound to the enzyme. Thus, it would make sense that conformational changes occur to keep the reaction intermediates within the catalytic site, at least for several seconds, which is the time reported to complete the decarboxylation reaction of a single histidine molecule [16,19,46]. It is worthwhile mentioning that some conformational changes have also been suggested for homologous enzymes during the decarboxylation reaction. For instance, it has been proposed that the fragment 328–339, which contains residues proven to be important for the activity of the enzyme and which cannot be properly resolved by X-ray diffraction studies, could be a flexible part of the molecule that changes its conformation during catalysis [18,37,38]. More recently, Hayashi et al. [47] have reported an important conformational change in aspartate aminotrans- ferase after substrate binding, which promotes the catalytic reaction, as it favors maximum imine–pyridine conjugation. Aspartate aminotransferase is also a dimeric PLP-depend- ent enzyme with a similar fold and some catalytic properties in common with both DDC and HDC [19,21,48]. The exact nature of the mammalian HDC conformational change is still unknown; nevertheless, a process similar to that occurring in the transaminase could lead to a more severe rotation in angle v up to negative values [19], so that these conformational changes are related to the catalytic effi- ciency of the enzyme. Our results indicate that mammalian HDC adopts, at least two well-differentiated conformations during the catalytic reaction. The one corresponds to the fully active internal aldimine of the enzyme, and the second takes place during the presence of a PLP-adduct (PLP-substrate or PLP-product) in the catalytic site. These conformational changes are part of the HDC reaction with its natural substrate. The latter conformation represents inactive forms Ó FEBS 2003 Conformational changes of histidine decarboxylase (Eur. J. Biochem. 270) 4385 [...]... conformation, would not necessarily have to involve the entrance of any inhibitor into the catalytic site, as the conformational change seems to affect the whole enzyme structure In addition, the present results and experimental approaches could also be interesting for groups working in other enzymes with a similar folding model (i.e L-amino acid decarboxylases, amino transferases) 10 11 12 13 Acknowledgements... aromatic 1-aromatic acid decarboxylase: spectroscopic and kinetic analysis of the coenzyme and reaction intermediates Biochemistry 32, 812–818 Hayashi, H., Mizugushi, H & Kagamiyama, H (1998) The iminepyridine torsion of the pyridoxal 5¢-phosphate Schiff base of aspartate aminotransferase lowers its pKa in the unliganded enzyme and is crucial for the successive increase in the pKa during catalysis Biochemistry... (Eur J Biochem 270) of the enzyme, which occur during the rate-limiting steps of mammalian HDC catalysis [19] Therefore, we suggest that the full molecular characterization of these conformational changes could be useful to look for new strategies to inhibit specifically and efficiently histamine production in vivo This alternative strategy, blocking the enzyme in the second type of conformation, would... Multiple forms of rat stomach histidine decarboxylase may reflect post-translational activation of the enzyme Regul Peptides 81, 41–48 Fleming, J.W & Wang, T.C (2000) Amino and carboxyterminal PEST domain mediate gastrin stabilization of rat 1 -histidine decarboxylase isoforms Mol Cell Biol 20, 4932–4947 ´ Olmo, M.T., Urdiales, J.L., Pegg, A.E., Medina, M.A & Sanchez´ Jimenez, F (2000) In vitro study of proteolytic... degradation of rat histidine decarboxylase Eur J Biochem 267, 1527–1531 ´ ´ Rodrı´ guez-Agudo, D., Sanchez-Jimenez, F & Medina, M.A (2000) Rat histidine decarboxylase is a substrate for m-calpain in vitro Biochem Biophys Res Commun 271, 777–781 ´ ´ Fajardo, I., Urdiales, J.L., Medina, M.A & Sanchez-Jimenez, F (2001) Effects of phorbol ester and dexamethasone treatment on histidine decarboxylase and ornithine decarboxylase. .. ¨ (University of Malaga) and Dr J V Fleming (University of Massachusetts Medical Center) for their valuable comments, and to Drs J.A Ranea and A Valencia (National Centre of Biotechnology, Madrid, Spain) for advice during 3D structure prediction of rat HDC Thanks are due to the Department of Architecture of Computers ´ (University of Malaga) for allowing us to get access to its computing facilities... Characterization and expression of the complementary DNA encoding rat histidine decarboxylase Proc Natl Acad Sci USA 87, 733– 737 33 Laemmli, U.K (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4 Nature 227, 680– 685 34 Hayashi, H., Tanase, S & Snell, E.E (1986) Pyridoxal 5¢-phosphate-dependent histidine decarboxylase J Biol Chem 261, 11003–11009 35 Nishino, J., Hayashi,... Mutation of residues in the coenzyme binding pocket of Dopa decarboxylase Eur J Biochem 268, 2975–2981 ´ ´ 40 Rodrı´ guez-Caso, C., Rodrı´ guez-Agudo, D., Sanchez-Jimenez, F & Medina, M.A (2003) Green tea epigallocatechin-3-gallate is an inhibitor of mammalian histidine decarboxylase Cell Mol Life Sci 60, 1760–1763 41 Bhattacharjee, M.K & Snell, E.E (1990) Pyridoxal 5¢-phosphatedependent histidine decarboxylase: ... decarboxylase: mechanism of inactivation by alpha-fluoromethylhistidine J Biol Chem 265, 6664–6668 42 Kubota, H., Hayashi, H., Watanabe, T., Taguchi, Y & Wada, H (1984) Mechanism of inactivation of mammalian 1 -histidine decarboxylase by (S)-a-fluoromethylhistidine Biochem Pharmacol 33, 983–990 43 Colucci, R., Fleming, J.V., Xavier, R & Wang, T.C (2001) 1 -Histidine decarboxylase decreases its own transcription... reaction of the Lys303 mutant aromatic 1-amino acid decarboxylase unravels the role of the residue in catalysis J Biochem 121, 604–611 36 Poupon, A & Mornon, J.P (1998) Populations of hydrophobic amino acids within protein globular domains: identification of conserved ÔtopohydrophobicÕ positions Proteins 33, 329–342 37 Ishii, S., Hayashi, H., Okamoto, A & Kagamiyama, H (1998) Aromatic 1-amino acid decarboxylase: . Local changes in the catalytic site of mammalian histidine decarboxylase can affect its global conformation and stability Carlos Rodrı ´ guez-Caso 1 ,. histidine and histidine- analogs involve alterations in the environment of the aromatic amino acid residues of the protein Conformational changes of proteins can