|Home | About | Journals | Submit | Contact Us | Français|
Aldh1l1, also known as 10-formyltetrahydrofolate dehydrogenase (FDH), contains the carboxy-terminal domain (Ct-FDH), which is a structural and functional homolog of aldehyde dehydrogenases (ALDHs). This domain is capable of catalyzing the NADP+-dependent oxidation of short chain aldehydes to their corresponding acids, and similar to most ALDHs it has two conserved catalytic residues, Cys707 and Glu673. Previously, we demonstrated that in the Ct-FDH mechanism these residues define the conformation of the bound coenzyme and the affinity of its interaction with the protein. Specifically, the replacement of Cys707 with an alanine resulted in the enzyme lacking the ability to differentiate between the oxidized and reduced coenzyme. We suggested that this was due to the loss of a covalent bond between the cysteine and the C4N atom of nicotinamide ring of NADP+ formed during Ct-FDH catalysis. To obtain further insight into the functional significance of the covalent bond between Cys707 and the coenzyme, and the overall role of the two catalytic residues in the coenzyme binding and positioning, we have now solved crystal structures of Ct-FDH in the complex with thio-NADP+ and the complexes of the C707S mutant with NADP+ and NADPH. This study has allowed us to trap the coenzyme in the contracted conformation, which provided a snapshot of the conformational processing of the coenzyme during the transition from oxidized to reduced form. Overall, the results of this study further support the previously proposed mechanism by which Cys707 helps to differentiate between the oxidized and reduced coenzyme during ALDH catalysis.
Aldehyde dehydrogenases are a family of NAD(P)+-utilizing enzymes that catalyze conversion of a variety of aldehydes to their corresponding carboxylic acids. In humans, this family is represented by 19 distinct genes, which encode protein products that form homodimers or homotetramers . ALDHs have a very similar structure, consisting of catalytic, nucleotide binding and oligomerization domains [2-8]. While structural boundaries between these domains can be clearly seen, the functional separation between them is somewhat arbitrary. For example, the nicotinamide ring of bound NAD(P)+ protrudes into the catalytic domain to come in close proximity to catalytic residues. Of note, the nicotinamide moiety forms only a few contacts with an ALDH molecule and most of the interactions involve two ribose rings and the adenine moiety forming contacts with the protein within a classical Rossmann fold conserved throughout the family [3, 5, 6, 8, 9].
In addition to the structural conservation, there is also a functional conservation with regard to ALDH mechanism. Thus, ALDHs have two highly conserved catalytic residues, a cysteine and a glutamate . Numerous structural, site-directed mutagenesis and enzymatic studies of ALDHs have established the role of the cysteine as the active site nucleophile (reviewed in ). The proposed mechanism for ALDH catalysis includes two steps, (i) acylation and (ii) deacylation [3, 5, 10-13]. The cysteine forms a thiohemiacetal intermediate with the substrate during the acylation step, which requires deprotonation of its sulfhydryl group (Reviewed in ). The glutamate has been proposed to facilitate the cysteine deprotonation, either directly or indirectly, and has also been suggested as a residue involved in the deacylation step of the reaction by activating a water molecule in the active site [3, 5, 8, 10, 14, 15]. The dual role of the glutamate residue in combination with the conserved geometry of the active site created the requirement for coenzyme isomerization during catalysis . In the first stage of the reaction the nicotinamide ring of the cofactor must be sufficiently close to the catalytic cysteine to receive the hydride. The corresponding stretched conformation of NAD(P)+ was named “extended”. Once the hydride transfer is accomplished, the nicotinamide ring vacates the catalytic pocket to allow access of a water molecule that hydrolyzes the thioester intermediate, releasing the reaction product. This conformation of NAD(P)H was named “contracted” because the isomerization brings the nicotinamide moiety closer to the adenine part. In some crystal structures of ALDHs the nicotinamide ring of the reduced coenzyme is not seen due to disorder [8, 16]. This conformation of NADP(H) is a functional analogue of the contracted conformation, as it also allows space for the deacylation stage to proceed.
A member of ALDH family, Aldh1l1 (FDH, 10-formyltetrahydrofolate dehydrogenase) is a folate-metabolizing enzyme, which is a product of a natural fusion of three unrelated genes . One of these genes, an ancient aldehyde dehydrogenase, encodes for the carboxyl-terminal portion (residues 400-902, identified thereafter as Ct-FDH) of Aldh1l1. This domain has high sequence similarity (up to 50%) with ALDH1 and ALDH2, and can catalyze both NADP+-dependent ALDH or NADP+-independent esterase reactions using short-chain aldehydes as substrates [17-19]. Like many other ALDHs, Ct-FDH possesses two key catalytic residues, a cysteine (Cys707) and a glutamate (Glu673); replacement of either resulted in the catalytically inactive enzyme [8, 20]. The ALDH reaction catalyzed by this domain is an integral part of the entire enzyme mechanism, which is the NADP+-dependent conversion of 10-formyltetrahydrofolate to tetrahydrofolate and CO2 . Of note, higher animals have the second gene encoding for a similar enzyme, Aldh1L2, which is a mitochondrial protein (in contrast to Aldh1l1, which is a cytosolic enzyme) [21, 22].
Crystal structures of apo Ct-FDH and the holo enzyme in complexes with either NADP+ or NADPH have demonstrated a high similarity of the protein molecule to other ALDHs . Ct-FDH (as well as full-length Aldh1l1 ) forms a homotetramer, with each monomer displaying catalytic, coenzyme binding and oligomerization domains. Previous studies have demonstrated an unusual feature of the complex of Ct-FDH with the oxidized coenzyme: a transient covalent bond between the sulfur atom of the catalytic cysteine and the C4N atom of the nicotinamide (Fig. 1) [8, 23]. Formation of such a bond has been also predicted for mitochondrial ALDH2 using in silico simulations  and was experimentally shown for other ALDHs . It has been further demonstrated that the ability to form such a covalent bond is important for the discrimination between oxidized and reduced coenzyme bound to Ct-FDH . In particular, replacement of the catalytic cysteine of Ct-FDH with an alanine resulted in an enzyme that bound both NADP+ and NADPH in the extended conformation. This study also suggested that the conserved catalytic glutamate controls the binding and discharging of the coenzyme, presumably through long-range communications with helix G that interacts with its adenine moiety. In the present work we aimed to clarify the role of the cysteine and the functional significance of its covalent bond with NADP+ in proper positioning of the coenzyme. We further wanted to elucidate if the nicotinamide ring of the coenzyme, which is known to only loosely bind to ALDHs, plays a significant part in defining the overall conformation of NADP(H). To this end, we solved crystal structures of the C707S mutant of Ct-FDH in complex with NADP+ or NADPH and a structure of wild-type Ct-FDH in complex with thio-NADP+, an NADP+ analogue with an altered nicotinamide group. Here we report the structural analysis of these proteins with regard to coenzyme binding and effects of Cys707 and Glu673 on the conformation of bound dinucleotide, and we expand the model for the conformational processing of the coenzyme during the transition from oxidized to reduced form.
Site-directed mutagenesis was carried out using a QuickChange site-directed mutagenesis kit (Agilent Technologies) and confirmed by DNA sequencing of the mutant constructs. Wild type and mutant Ct-FDH were expressed in E. coli as constructs with 5xHis tag at the amino-terminus and purified on Ni-NTA resin (GE-Healthcare) as we previously described .
Crystals were grown by the vapor diffusion method in hanging drops over wells containing 1.5-1.6 M ammonium sulfate and 0.1 M MES-NaOH, pH 6.4 or 0.1 M HEPES-NaOH, pH 7.2, as described . Crystals of binary complexes with NADP+, NADPH or thio-NADP were produced by overnight soaking of protein crystals in the presence of 5 mM of the corresponding coenzyme. Prior to mounting, crystals were passed through the mother liquor solution supplemented with the corresponding coenzyme and containing 27.5% glycerol and flash-cooled in situ at 100 K using an X-Stream Cryostream (Rigaku MSC).
Data sets were collected on an RAXIS IV++ image plate detector mounted on a RU-H3R rotating anode X-ray generator operating at 50 kV and 100 mA (Medical University of South Carolina). All the crystals belonged to space group C2 with cell dimensions a=258.6, b=194.8, c=97.4, and β=108.8. The data were processed using HKL2000 . The statistics are shown in Table 1.
Structures were obtained by molecular replacement using Molrep  with the structure of apo Ct-FDH (PDB code 2O2P)  as the search model. Coenzyme molecules were identified by examining the FO-FC and 2FO-FC electron density maps. Models were refined by alternating rounds of manual building using O  and Coot  and restrained refinement in REFMAC5 . Water molecules were introduced with Coot  and verified manually. During refinement, 5% of the data were reserved to calculate the free R-factor. Table 1 shows the refinement statistics. Stereochemistry of the models was verified by PROCHECK . The coordinates and structure factors have been deposited in the Protein Data Bank with codes 4GNZ, 4GO0 and 4GO2 for the C707S in complex with NADP+, the C707S in complex with NADPH and wild type Ct-FDH in complex with thio-NADP+, respectively. Figures were prepared using PyMol (www.pymol.org).
The ALDH activity of Ct-FDH and the C707S mutant was measured by following the increase in absorbance at 340 nm as a result of formation of NADPH. The standard assay contained 22 μg/ml the enzyme, 120 μM NADP+, and 5 mM propanal in 50 mM HEPES-NaOH buffer, pH 9.0. The reaction was initiated by the addition of propanal. All measurements were done at 25 °C using a Shimadzu 2401PC double beam spectrophotometer. The activity of wild type Ct-FDH with thio-NADP+ (Sigma) was measured by monitoring the increase in absorbance at 398 nm as a result of formation of thio-NADPH.
Wild type Ct-FDH or the mutated proteins were incubated with increasing concentrations of NADP+ (up to 4.2×10−5 M), NADPH (up to 5×10−5 M) or thio-NADP (up to 4.2×10−5 M) for 1 hour at room temperature in 50 mM Tris-HCl buffer, pH 7.5 containing 100 mM NaCl and 0.5 mM DTT. Emission fluorescence spectra of proteins were recorded on a Hitachi F-2500 fluorescence spectrophotometer by scanning from 300 to 400 nm with excitation at 295 nm. Changes in NADPH fluorescence at 450 nm were recorded upon excitation at 330 nm. Kd values were determined by non-linear fitting of experimental data to equations described in our previous publication .
All crystals obtained in this study belong to the C2 space group with unit cell dimensions identical to those of the wild type Ct-FDH crystals . Protein subunits share the same fold and show no appreciable structural rearrangements compared to these previously published models.
We determined two structures of the C707S mutant of Ct-FDH, the complex with NADP+ at 2.3 Å and the complex with NADPH at 3.4 Å resolution. These structures show no significant conformational deviations from wild type Ct-FDH, and their active sites are virtually identical (Fig. 2). Similarly to the structure of the wild-type enzyme with bound NADP+  and to the structure of the corresponding mutant of human mitochondrial ALDH , in the C707S-NADP+ complex the coenzyme is bound in the extended conformation, with the nicotinamide ring displacing the side chain of the catalytic Glu673 to attain close proximity to Ser707 (Fig. 2A). However, the serine residue in place of the catalytic cysteine does not form a covalent bond with the C4N atom of the nicotinamide ring.
A remarkable feature of the C707S-NADPH complex is the extended conformation of the reduced coenzyme (Fig. 2B). In contrast, in the wild type Ct-FDH the nicotinamide ring of the reduced coenzyme is disordered, presumably due to the lack of contacts with the protein . We have previously demonstrated the identical arrangement of the reduced coenzyme (extended conformation) bound to another Ct-FDH, the C707A enzyme . In that enzyme, however, the extended conformation of the reduced coenzyme could be possible because the smaller side-chain of alanine allows additional space in the catalytic pocket for the nicotinamide ring. Here our data demonstrated that despite a higher structural and chemical similarity between a serine and a cysteine versus an alanine and a cysteine, NADPH still is in the extended conformation when bound to the C707S enzyme. Of note, the average distance between the hydroxyl oxygen of Ser707 and the C4N atom of NADPH in this structure is 2.9±0.2 Å, which provides enough space to accommodate the nicotinamide ring. Thus, the sulfur atom of Cys707 (likely due to the ability to form the transient covalent bond with the C4N atom of nicotinamide) seems to be crucial for discriminating between the oxidative states of the coenzyme. This sensory function either allows the oxidized nicotinamide ring to enter the catalytic pocket or repels the reduced ring.
We used a coenzyme analogue, thio-NADP+ (Fig. 3), to further probe the role of the two catalytic residues, Cys707 and Glu673, in the coenzyme binding. The structure of wild-type Ct-FDH in complex with thio-NADP+, determined at 2.3 Å resolution, maintains the same protein fold, but the coenzyme is bound in two different conformations with approximately equal occupancies. While one of these conformers is mostly disordered, the other forms a “crooked” structure with the nicotinamide part showing minimal contacts with the protein (Fig. 4A). In this new conformation the nicotinamide ring is shifted 4.6 Å outward from the catalytic center, and it closely resembles the canonical contracted conformation of NADH in the complex with ALDH2  (Fig. 4B). Furthermore, as was previously shown in the structure of the Ct-FDH/NADP+ complex, binding of the nicotinamide ring in the extended conformation of the coenzyme is associated with a movement of the side chain of Glu673 out of the active site (Fig. 4C) . In this new position, Glu673 forms tight contacts with several other residues, water molecules and the carboxamide group of the coenzyme (Fig. 4C). In contrast, in the structure of the Ct-FDH/thio-NADP+ complex the side chain of Glu673 remains in contact with Cys707 as was shown previously for the ligand-free structure , blocking the entrance to the catalytic center (Fig. 4A). It is likely that the bulky sulfur atom of thio-NADP+ substituting for the carboxamide oxygen of NADP+ cannot be accommodated in the extended conformation, and its repulsion from the carboxyl group of the glutamate results in the nicotinamide moiety being expelled from the active site. Of note, thio-NADP+ appears in the extended conformation in the structures of DHFR  and transhydrogenase , the fact suggesting that this conformation is possible. Thus, we suggest that thio-NADP+ bound to Ct-FDH represents the conformation of the natural coenzyme on its way out after the completion of the hydride transfer.
We have previously demonstrated that C707A Ct-FDH is catalytically inactive . A similar effect was observed when the catalytic cysteine of other aldehyde dehydrogenases was replaced with alanine . However, the replacement of the cysteine with a serine in mitochondrial rat liver aldehyde dehydrogenases (Aldh2) resulted in an enzyme that retained residual catalytic activity . In the present study, analysis of the C707S Ct-FDH revealed that it was catalytically inactive. We have also determined Kds for binding both oxidized and reduced coenzymes by the mutant enzyme. The K for NADP+ d was determined based on quenching of tryptophan fluorescence of the C707S enzyme upon binding of the coenzyme as we previously described . The decrease in fluorescence was likely due to quenching of fluorescence of two tryptophans near the coenzyme binding site, Trp573 located about 4 Å from the pyrophosphate group of the coenzyme and Trp582 located about 8 Å from the nicotinamide ring. The replacement of Cys707 with a serine resulted in a significant decrease in the affinity for coenzyme with Kd of 11.0 ± 1.5 μM (Kd for the wild-type enzyme is 0.9 ± 0.1 μM). The above method cannot be used to measure the Kd for NADPH binding due to the fluorescence resonance energy transfer from the tryptophan residues to NADPH . Therefore, changes in NADPH fluorescence at 450 nm were monitored after excitation at 330 nm (binding to Ct-FDH leads to enhanced NADPH fluorescence) . These experiments demonstrated that the wild-type and C707S enzymes have similar Kd values for NADPH (data not shown). This is in contrast to C707A enzyme, which binds NADPH more tightly than the wild-type enzyme does . We suggest the difference is due to the larger side chain of the serine as compared to the alanine residue.
We have also evaluated the aldehyde dehydrogenase activity of wild-type Ct-FDH using thio-NADP+ as the coenzyme. The enzyme did not exhibit any catalytic activity with this derivative, a finding that is in agreement with our structural data demonstrating that the nicotinamide moiety is positioned too far from the catalytic cysteine to allow the hydride transfer; the closest distance between the C4N atom of the nicotinamide of thio-NADP+ and the sulfur of the catalytic cysteine is 4.5 Å compared to 1.6 Å in the case of NADP+ (Fig. 1) . The Kd value (17.0 ± 0.8 μM) for the binding of thio-NADP+ evaluated by quenching of tryptophan fluorescence was significantly increased compared to the Kd value for wild-type enzyme.
Nicotinamide dinucleotide molecules, NAD+/NADH and NADP+/NADPH, transfer a hydride ion in oxido-reductive reactions and are among the most common coenzymes found in the cell, being utilized by several hundreds of different enzymes [36, 37]. While the precise mode of the accommodation of these coenzymes in their respective binding pockets varies among enzymes and is defined by structural arrangements of catalytic and nucleotide binding centers, the enzyme mechanism, and by specific residues involved in the coenzyme binding and catalysis , there are several common features of dinucleotide binding. For example, the most common structural motif involved in the binding of NAD(P)+ molecules is the Rossmann fold (reviewed in [36, 39]), which is found in many different classes of dehydrogenases including the ALDH family [2, 3]. Furthermore, while the nicotinamide adenine dinucleotide is a flexible molecule that can adopt different conformations depending on its environment , the coenzyme is bound to most proteins preferentially in the extended conformation . An interesting feature of the coenzyme binding by ALDHs, though, is two distinct conformations of the dinucleotide molecule, extended and contracted [2, 3, 5, 8, 9, 41]. It is generally accepted that the former represents the hydride transfer position of the oxidized coenzyme in the acylation step of the catalysis while the latter corresponds to the reduced coenzyme in the second, deacylation, step when the nicotinamide ring has already left the catalytic pocket.
Crystal structures of the holo Ct-FDH in complex with NADP+ or NADPH demonstrated two conformations of the dinucleotide molecule as well, an extended one for the oxidized coenzyme and one which was interpreted as the contracted conformation for the reduced coenzyme . Of note, in that presumably contracted conformation, there was no electron density seen for the nicotinamide ring of NADPH, suggesting the flexibility of positioning of the reduced ring in this structure. Apparently, the nicotinamide ring in the oxidized coenzyme forms additional contacts with the enzyme that fix it in the extended conformation, whereas the lack of such contacts for the ring in the reduced coenzyme allows it to move into various positions. In the present study, however, the contracted conformation of the coenzyme, with a well-ordered nicotinamide ring, has been observed in the complex of Ct-FDH with thio-NADP+.
Our recent studies of Ct-FDH shed light on the mechanism by which ALDHs could retain the oxidized coenzyme in the extended conformation and also discriminate against the binding of the reduced coenzyme . This novel mechanism of the interaction between ALDHs and NAD(P)+ involves a transient covalent bond formed between the catalytic nucleophile (the sulfhydryl group of the cysteine residue) and the C4N atom of the nicotinamide ring . Since this finding was reported, covalent bonds between the catalytic cysteine and the nicotinamide ring have been observed in crystal structures of several other aldehyde dehydrogenases [24, 25], the fact implicating this phenomenon as an integral part of the catalytic mechanism. We initially suggested that such a covalent bond facilitates stronger binding of the coenzyme and the proper orientation of the nicotinamide ring within the catalytic center . The follow up study from our laboratory has further demonstrated that the covalent bond between the cysteine and nicotinamide ring plays an additional role as the sensor recognizing the oxidative state of the coenzyme . This hypothesis has received further support in the present study.
Based on our data, we propose the following mechanism for discharging the coenzyme from its binding site (schematically depicted in Fig. 5). This mechanism presents two catalytic residues, Cys707 and Glu673, as the key determinants defining the mode of interaction between the enzyme and the coenzyme, and can be described as a two-phase process: (i) the displacement of the reduced nicotinamide ring from the catalytic center, and (ii) dissociation of the coenzyme from the enzyme. In the first phase, upon binding the oxidized coenzyme, the sulfhydryl group of the cysteine forms a transient covalent bond with the C4N atom. This bond, which was also demonstrated for the hydrolytic ALDH from Pseudomonas aeruginosa, can easily dissociate via an elimination reaction . The fact that we observed two positions of the sulfur atom of the cysteine with the 50% probability might suggest that the formation and dissociation of this bond is a constant process. We hypothesize that the ability to form such a temporary bond serves as a sensor of the oxidized state of the coenzyme (it cannot be formed with the reduced form, NADPH). This step is schematically depicted in Fig. 5 (panels A and B) as the balance between two stages, with and without the covalent bond. In the second step of this phase, following the hydride transfer and the reduction of the coenzyme, the sulfhydryl group comes into steric hindrance with the nicotinamide ring (Fig. 5, panel C). In the third step, this hindrance results in the displacement of the ring from the catalytic center thus changing the coenzyme conformation from extended to contracted (Fig. 5, panel D).
In our crystal structures, the latter stage is characterized by the lack of well-defined electron density for the reduced nicotinamide ring, but well-defined density for the adenosine diphophosphoribose. We interpreted this conformation of NADPH as a disordered nicotinamide ring and the rest of the coenzyme rigidly bound to helix G, which itself is well-ordered at this stage (Fig. 6, panel A). Based on our previous studies , we hypothesize that such an arrangement (an ordered helix G) is caused by the distant effect of the catalytic glutamate, which at this point is displaced from the catalytic pocket by the nicotinamide ring (Fig. 6, panel A). The second phase of the coenzyme dissociation involves the rotation of the glutamate back into the nicotinamide-binding pocket within the catalytic center, which may initiate remote conformational alterations in helix G (as shown previously for the E673Q enzyme ) thus causing its disorder (Fig. 6, panel B). The disordered helix is losing contact with the adenine moiety, which allows coenzyme dissociation from the enzyme (there would be no strong contacts between the enzyme and the coenzyme with such an arrangement). It should be noted that Glu673 is also likely to participate in the first stage of coenzyme discharging, through energetically unfavorable interactions with carboxamide of nicotinamide, thus cooperating with Cys707 in dissociation of the nicotinamide from the catalytic center. In this regard, the glutamate could discriminate between different orientations of the planes of the carboxamide group relative to the nicotinamide ring in the oxidized and reduced coenzyme, a mechanism described for other enzymes [32, 33].
It is not clear at present whether the proposed mechanism of discrimination between the reduce and oxidize states of the coenzyme is unique for Ct-FDH, limited to certain members of ALDH family, or whether it is a more common phenomenon for this class of enzymes. In this regard, while the catalytic cysteine is conserved in all active aldehyde dehydrogenases, the glutamate is less conserved . It is also not clear whether the mechanism is extended to other classes of oxido-reductases. Of note, the ALDH reaction is essentially unidirectional , whereas for example alcohol dehydrogenases catalyze a reversible reaction . Thus, the specific feature of the ALDH catalysis could be the necessity to discharge the reduced coenzyme without the requirement to bind and utilize it for the reverse reaction. Toward this end, our study underscores the role of the cysteine and glutamate in discriminating between the reduced and oxidized coenzyme, a feature that perhaps would not be required for enzymes catalyzing the reaction in both directions. Curiously, it appears that one of the ALDHs, ALDH18A1, catalyzes the reversal of ALDH reaction. ALDH18A1, an NADPH-dependent γ-glutamyl phosphate reductase, is a part of the bifunctional pyrroline-5-carboxylate synthase, the key enzyme in the biosynthesis of proline and ornithine [45-48]. It is attractive to suggest that the reductase does not have the catalytic glutamate analogous to Glu673 of Ct-FDH and binds NADPH in the extended conformation. However, while the crystal structure of the apo enzyme from Thermotoga maritima has been solved , the structure of its complex with the coenzyme awaits future studies.
The authors would like to thank Dr. John Hempel, University of Pittsburg for helpful discussions. This work was supported by The National Institutes of Health grant DK54388. Kyle C. Strickland was supported by a Ruth L. Kirschstein National Research Service Award for Individual Predoctoral MD/PhD Fellows F30DK083215. The X-ray crystallography facility used for this work is supported by the Medical University of South Carolina’s Research Resource Facilities program.
Conflict of interest statement None.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.