10-Formyltetrahydrofolate Dehydrogenase Requires a 4′-Phosphopantetheine Prosthetic Group for Catalysis*

10-Formyltetrahydrofolate dehydrogenase (FDH) consists of two independent catalytic domains, N- and C-terminal, connected by a 100-amino acid residue linker (intermediate domain). Our previous studies on structural organization and enzymatic properties of rat FDH suggest that the overall enzyme reaction, i.e. NADP+-dependent conversion of 10-formyltetrahydrofolate to tetrahydrofolate and CO2, consists of two steps: (i) hydrolytic cleavage of the formyl group in the N-terminal catalytic domain, followed by (ii) NADP+-dependent oxidation of the formyl group to CO2 in the C-terminal aldehyde dehydrogenase domain. In this mechanism, it was not clear how the formyl group is transferred between the two catalytic domains after the first step. This study demonstrates that the intermediate domain functions similarly to an acyl carrier protein. A 4′-phosphopantetheine swinging arm bound through a phosphoester bond to Ser354 of the intermediate domain transfers the formyl group between the catalytic domains of FDH. Thus, our study defines the intermediate domain of FDH as a novel carrier protein and provides the previously lacking component of the FDH catalytic mechanism.

The subunit of 10-formyltetrahydrofolate dehydrogenase (FDH 3 ; ALDH1L1; EC 1.5.1.6) represents a single polypeptide consisting of three domains, with two of these domains possessing their own catalytic activities (1)(2)(3)(4). The N-terminal domain of FDH (residues 1-310) functions as 10-formyltetrahydrofolate (10-fTHF) hydrolase, converting 10-fTHF to tetrahydrofolate (THF) and formate ( Fig. 1) (2). The C-terminal domain (residues 420 -902) is an aldehyde dehydrogenase-homologous enzyme capable of NADP ϩ -dependent oxidation of short chain aldehydes to the corresponding acids (3). The two catalytic domains are separated by a short (ϳ100 residues) linker domain (intermediate domain). Although the physiological significance of the catalytic functions of the separate domains of FDH is not clear, the well established function of FDH is the conversion of 10-fTHF to THF in an NADP ϩ -dependent dehydrogenase reaction (5)(6)(7). It is believed that this reaction regulates intracellular 10-fTHF/THF pools (8), controls de novo purine biosynthesis (9,10), and affects the methylation potential of the cell (11). This reaction is observed only when the two catalytic domains are linked in one polypeptide by the intermediate domain (2). Studies of FDH structure and catalytic mechanism (12)(13)(14)(15)(16)(17)(18) suggest that this reaction proceeds in two steps: (i) hydrolytic removal of the formyl group from 10-fTHF and (ii) NADP ϩ -dependent oxidation of formyl to CO 2 . The first step takes place in the N-terminal hydrolase domain, whereas the second step occurs in the C-terminal aldehyde dehydrogenase domain (2,3,17,18).
The connection between the two steps is the transfer of the formyl group between the two catalytic domains. However, the nature of such a transfer is not obvious. We have demonstrated previously that the two catalytic domains of FDH do not form close contacts in the absence of the intermediate domain (2). It has been suggested that the intermediate domain keeps the N-and C-terminal domains of FDH in close proximity to each other and in the correct orientation to create an interface between the two domains and to allow the formyl group transfer between the two catalytic centers. In support of this, the engineering of FDH with limited flexibility within the intermediate domain, which presumably permanently uncouples the two catalytic domains, resulted in an enzyme with no 10-fTHF dehydrogenase activity (15). The crystal structures of the two domains have demonstrated, however, that the catalytic centers of FDH, the hydrolase and aldehyde dehydrogenase, are buried within their corresponding domains, so residues of one catalytic center cannot contact residues of the other catalytic center (17,18). This implies that the direct transfer of the formyl group between the two centers is improbable and suggests a role for the intermediate domain in the transfer of the substrate between the catalytic centers.
Several mechanisms for channeling of a substrate within an FDH molecule could be envisioned. If the formyl group moves from the hydrolase catalytic center into the aldehyde dehydrogenase catalytic center by passive diffusion, the role of the intermediate domain could be in maintaining an orientation of the two centers that cages formyl and minimizes its loss to the outer solution. Another potential mechanism by which the interme-diate domain can facilitate the transfer of a reaction intermediate between the catalytic centers is formation of a molecular tunnel, as has been proposed for a number of multidomain enzymes (reviewed in Refs. 19 and 20). The fact that such tunnels are mostly hydrophobic in nature (20) while the intermediate domain of FDH is noticeably hydrophilic (15) argues against such a mechanism. A third alternative for assisted intermediate transfer involves the employment of a prosthetic group, which can function as a swinging arm (21). The canonical examples of such a transfer are pyruvate carboxylase (22), the pyruvate dehydrogenase complex (23), and acyl carrier proteins (ACPs) (24), which use biotin, lipoate, and 4Ј-phosphopantetheine (4Ј-PP), respectively, as prosthetic groups.
The NMR structure of the intermediate domain of human FDH (Protein Data Bank code 2CQ8) demonstrates an ACPlike fold (21). In this study, we evaluated the sequence and structure similarity between the intermediate domain of rat FDH and carrier proteins with a 4Ј-PP prosthetic group (see Fig.  2). We also provide evidence that the intermediate domain of rat FDH is covalently modified by 4Ј-PP and that this prosthetic group transfers formyl from the hydrolase catalytic center to the aldehyde dehydrogenase catalytic center during the FDHcatalyzed 10-fTHF dehydrogenase reaction.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Rat FDH was expressed in either Escherichia coli or insect cells using a baculovirus expression system as we described previously (18,25). It was purified by affinity chromatography on immobilized 5-fTHF-Sepharose, followed by size-exclusion chromatography on Sephacryl S-300. If necessary, additional purification was done by fast protein liquid chromatography using a Mono Q column as described (25). The S354A mutant of FDH was expressed in insect cells and purified by the same procedure as used for the wild-type enzyme. The intermediate domain of rat FDH was expressed in E. coli and purified from inclusion bodies by sizeexclusion chromatography on Sephacryl S-300 (15).
Assay of Enzyme Activity-The hydrolase, aldehyde dehydrogenase, and 10-fTHF dehydrogenase activities of FDH were evaluated spectrophotometrically as described previously (3). Briefly, the hydrolase and 10-fTHF dehydrogenase activities were measured using 10-formyldideazafolate as a substrate, with the assays for dehydrogenase activity containing additionally 100 M NADP ϩ . The monitored increase in absorbance at 295 nm (the characteristic spectrum maximum for the reaction product, dideazafolate) was used to calculate the specific activity. All assays were carried out at 30°C in a 1-cm quartz cuvette using a Shimadzu 2401PC double-beam spectrophotometer. Aldehyde dehydrogenase activity was evaluated by assaying propanal oxidation in the presence of FDH and NADP ϩ . The reduction of NADP ϩ to NADPH measured at 340 nm (the characteristic spectrum maximum for the reduced form of the coenzyme) provided the reaction rate for the aldehyde oxidation.
Site-directed Mutagenesis-The S354A mutant of rat FDH was generated using the QuikChange kit (Stratagene) as we described previously (13,14).
Reactivation of FDH Expressed in E. coli-Purified recombinant rat FDH expressed in E. coli (50 l of a 2 mg/ml solution in 20 mM Tris-HCl (pH 7.5)) was incubated with 2 l of 5 mM coenzyme A solution in the presence and absence of 50 or 100 l of lysate from Sf9 insect cells. The cell lysate was obtained by mechanical disruption of a suspension (1.0 ml) of 10 ϫ 10 6 cells in 20 mM Tris-HCl (pH 7.5) containing 10 mM MgCl 2 , 10 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride using a Dounce homogenizer for 2 min. 200 l of glycerol was then added to the homogenate, and insoluble content was removed by centrifugation at 13,000 ϫ g for 5 min.
Synthesis of a Fluorescent Reporter-A fluorescent reporter was synthesized according to a published procedure (26). 100 l of a solution of fluorescein maleimide (7 mg in 0.7 ml of Me 2 SO) was mixed with 2 ml of 0.1 M of MES (pH 6.0) containing 3.8 mg of coenzyme A and kept on ice for 30 min, followed by incubation for 10 min at room temperature. Complete conversion of coenzyme A into the fluorescent reporter was verified by TLC (5:2:4 (v/v) butanol/acetic acid/water). Non-reacted fluorescein maleimide was quenched by the addition of 40 l of 0.5 M dithiothreitol in water. The solution was used directly for protein labeling. It was stable upon freezing for at least 1 month.
Labeling FDH with the Fluorescent Reporter-Purified FDH (50 l of a 2.0 mg/ml solution in 20 mM Tris-HCl (pH 7.8)) was incubated with 50 or 100 l of lysate from Sf9 cells and 10 l of the fluorescent reporter mixture for 3 h at 37°C. Following incubation, 10 l of this solution was mixed with 10 l of SDS-PAGE loading buffer (4% SDS, 1% 2-mercaptoethanol, 4% glycerol, and 0.01% bromochlorophenol blue in 50 mM Tris-HCl (pH 6.8)) and subjected to SDS-PAGE. Immediately following electrophoresis, the gel was imaged under UV light irradiation and then stained with Coomassie Blue.
Labeling FDH with Fluorescein Maleimide-1 mg of FDH in 1 ml of phosphate buffer (pH 7.0) was incubated with 25 l of a stock solution of fluorescein maleimide (Pierce) in Me 2 SO (20 mg/ml) at room temperature for 3 h. Non-reacted fluorescein maleimide was quenched by the addition of 25 l of 100 mM dithiothreitol and removed by gel filtration on a PD-10 column (GE Healthcare).
Treatment with Alkaline Phosphatase-Purified wild-type FDH or its S354A mutant (ϳ1 mg) was incubated with 20 units The N-terminal domain of FDH produces hydrolase activity (converts 10-fTHF to THF and formate); the C-terminal domain produces aldehyde dehydrogenase activity (NADP ϩ -dependent conversion of short chain aldehydes to the corresponding acids); and full-length FDH carries out NADP ϩ -dependent conversion of 10-fTHF to THF and CO 2 (the other two reactions also remain).
of calf intestinal alkaline phosphatase (CIAP; New England Biolabs) in 50 mM Tris-HCl (pH 7.9) containing 100 mM NaCl, 10 mM MgCl 2 , and 1 mM dithiothreitol at 37°C for 24 h. The protein component was then separated by filtration on a Centriprep cartridge with molecular mass cutoff of 3000 Da (Millipore Corp.). The treated FDH was evaluated for all three catalytic activities as described above. The low molecular mass filtrate was concentrated and subjected to HPLC on a C 18 column.
HPLC-The concentrated low molecular mass filtrate was diluted with 20 mM ammonium acetate buffer (pH 6.8) at a 1:1 ratio, and 20 l was loaded onto a Microsorb-MV C 18 column (4.6 ϫ 250 mm, 5-m diameter particles; Rainin) equilibrated with the same buffer. Elution was performed with a linear gradient of acetonitrile (0 -60%) in the above buffer over 60 min. Chromatography was carried out on a Waters 515 dual-pump solvent delivery system at a flow rate of 1 ml/min; peaks were monitored at 224 nm using a Waters 2487 dual-wavelength absorbance detector. Fractions of 1 ml were collected.
Matrix-assisted Laser Desorption Ionization (MALDI) Mass Spectrometry-Fractions corresponding to peaks observed upon HPLC separation were analyzed by MALDI mass spectrometry. A MALDI time-of-flight Altoflex Smartbeam instrument (Bruker Daltonics, Billerica, MA) was used in linear mode to detect average ion masses. Samples were prepared by mixing 1 part of a sample solution (0.5 l) with 1 part of matrix (5 mg/ml 2,5-dihydroxybenzoic acid in 50% ethanol) on the sample plate and allowed to air dry prior to analysis. Experiments were performed at the Biomolecular Mass Spectrometry Facility of the Medical University of South Carolina.

FDH Treated with Alkaline Phosphatase Lacks 10-fTHF
Dehydrogenase Activity-4Ј-PP is bound to ACP by a phosphodiester bond, which can be hydrolyzed by the action of a phosphodiesterase, resulting in the removal of the 4Ј-PP group (27). To study whether the FDH intermediate domain has the same type of modification as ACPs, we treated fully active rat FDH expressed in insect cells with CIAP. Alkaline phosphatases in general have been shown to possess phosphodiesterase activity (28 -30). Therefore, we suggested that CIAP is also capable of such an activity and will hydrolyze the phosphodiester bond between FDH and 4Ј-PP. Indeed, we observed that the treatment with alkaline phosphatase completely deprived FDH of 10-fTHF dehydrogenase activity, whereas the treated enzyme was still capable of 10-fTHF hydrolase activity and aldehyde dehydrogenase activity (Table 1).
Site-directed Mutagenesis of Ser 354 -Comparison of the sequences of the intermediate domain of FDH and ACPs indicates that Ser 354 is equivalent to the serine residue of ACPs that is modified by 4Ј-PP ( Fig. 2A). To study the role of this residue in FDH function, we expressed the S354A mutant enzyme. The mutant enzyme was expressed in insect cells using a baculovirus expression system and purified to near homogeneity by the same procedure as used for the wild-type enzyme (16). Assay of the activities of the mutant enzyme revealed that it possesses hydrolase and aldehyde dehydrogenase activities similar to those of wild-type FDH (Table 1). In contrast, the mutant did not produce a noticeable level of 10-fTHF dehydrogenase activity ( Table 1).
Reactivation of FDH Expressed in E. coli-We observed previously that rat FDH expressed in E. coli lacked the dehydrogenase activity toward 10-fTHF, whereas the two other activities, hydrolase and aldehyde dehydrogenase, residing in specific domains, remained. 4 These results were initially explained by the lack of ability of E. coli to fold FDH into a functional enzyme. However, if FDH requires a modification of its intermediate domain, it is possible that the enzyme expressed in bacteria is inactive because of the lack of specific enzymes in E. coli capable of modifying FDH with 4Ј-PP. This was found to be the case: incubation of FDH expressed in E. coli with lysates of insect cells in the presence of CoA (CoA is a source of 4Ј-PP for ACP (24)) restored 10-fTHF dehydrogenase activity (Table 1). Similar results were obtained in experiments with lysates from human A549 cells (data not shown). To confirm that activation of FDH proceeds through a modification of Ser 354 , we expressed S354A mutant FDH in E. coli and subjected it to the same procedure. These experiments revealed that this mutant was not reactivated into a functional 10-fTHF dehydrogenase ( Table 1).
Fluorescent Labeling of FDH-Further evidence of the covalent modification of FDH at Ser 354 in the presence of CoA came from our experiments with a fluorescent reporter synthesized according to a published procedure (26). In this reporter, the fluorescent label fluorescein maleimide was attached to a 4Ј-PP part of CoA through the sulfhydryl group (Fig. 3A). This reporter allows the fluorescent labeling of proteins undergoing 4Ј-PP modification (26), as illustrated in Fig. 3B. We incubated purified recombinant FDH or its intermediate domain expressed in E. coli (both presumably unmodified at Ser 354 ) with the reporter and lysates from Sf9 insect cells and subjected the mixture to SDS-PAGE. These experiments clearly demonstrated incorporation of the fluorescent label into both FDH and its intermediate domain (Fig. 3, C and F). In contrast, the S354A mutant was not labeled under the same conditions (Fig.  3D), indicating that this serine is a modification site. We further used insect cell-expressed FDH, possessing 10-fTHF dehydrogenase activity, as a target for modification by this reporter. We failed to observe modification of FDH with fluorescently 4 S. A. Krupenko, unpublished data. labeled 4Ј-PP (Fig. 3E), presumably because the enzyme is already fully modified with 4Ј-PP. We also used fluorescein maleimide to modify sulfhydryl groups of insect cell-expressed FDH. Because this enzyme is a functional 10-fTHF dehydrogenase, we expected that it bears a 4Ј-PP prosthetic group with a sulfhydryl group accessible for modification. The labeled protein, separated from non-reacted fluorescein maleimide, was then treated with CIAP to release components bound to the protein through a phosphodiester bond. The released low molecular mass components were separated from the protein by passage through a 3000-Da cutoff filtration cartridge. In a control experiment, labeled protein untreated with CIAP was subjected to the same procedure. We observed strong fluorescence of the filtrate in the case of phosphatase-treated FDH, whereas no fluorescence was seen in the control untreated sample. In contrast, S354A mutant FDH did not produce labeled low molecular mass components in similar experiments. This further confirmed that a sulfhydryl-containing cofactor is bound to Ser 354 of FDH through a phosphodiester bond.
Mass Spectrometry Analysis of FDH Modification-We analyzed the molecular mass of the modifying group of FDH by MALDI time-of-flight mass spectrometry. The fact that treatment with CIAP completely deprived the enzyme of 10-fTHF dehydrogenase activity indicates that the essential non-protein cofactor was removed from the protein core by this treatment. We separated the cofactor from the protein moiety by concentrating the mixture on a filter with a 3000-Da molecular mass cutoff and subjected the low molecular mass content, after passage through the filter, to HPLC on a C 18 column. The chromatography produced two major peaks with retention times of 17.5 and 49 min, respectively. The MALDI time-of-flight anal- ysis revealed that the main component in the first peak has a molecular mass of 550.60 Da, whereas the main component in the second peak has a molecular mass of 360.08 Da (Fig. 4). The molecular mass of the latter corresponds well to the calculated molecular mass of 4Ј-PP, 358.33 Da. As for the first peak, we suggest that its main component represents an oxidized form (dimerized via a disulfide bond) of the two pantetheine residues, produced by removal of phosphate from 4Ј-PP by CIAP. The calculated molecular mass of such a dimer is 552 Da. As a control, both FDH expressed in E. coli and the S354A mutant expressed in insect cells were subjected to the same procedure. None of the proteins yielded corresponding peaks after HPLC (data not shown).

DISCUSSION
The amino acid sequence of FDH suggests that the enzyme is a natural fusion of at least two unrelated genes. Although the relationships of the N-terminal domain with other enzymes that use 10-fTHF as a substrate, as well as the nature of the C-terminal domain as an aldehyde dehydrogenase family member, were apparent, the origin of the linker between the two domains remained unclear. Indeed, this domain does not have an extensive sequence similarity to known proteins. However, recent analysis of the sequence of rat FDH using a data base of protein domains (www.expasy.org/prosite/) suggested that the intermediate domain is a member of the group of carrier proteins with a 4Ј-PP swinging arm.
These carrier proteins are components of multienzyme complexes involved in fatty acid, polyketide, and non-ribosomal peptide biosynthesis (21,31). Their specific feature is a 4Ј-PP component covalently attached to a conserved serine residue through a phosphodiester bond (24). Another enzyme with a 4Ј-PP modification is ␣-aminoadipate reductase, a component of lysine synthesis specific to fungi (32). The 4Ј-PP arm is a crucial functional component of the carrier proteins: it holds a growing chain of fatty acids, polyketides, or peptides during the reaction. In mammals, there is only one known metabolic pathway (fatty acid biosynthesis) that requires this type of carrier proteins. Structures of several carrier proteins have been solved (reviewed in Refs. 21, 24, and 31). They demonstrate a similar fold, a distorted four-helix bundle with the 4Ј-PP-bearing conserved serine located in the loop. The NMR structure of the intermediate domain of human FDH has recently become available (Protein Data Bank code 2CQ8). The model of the rat enzyme, which was used in our studies, was built using the human enzyme as a template. The overall fold of the intermediate domain is similar to that of other carrier proteins (33)(34)(35), with the best superposition (root mean square deviation between C-␣ atoms of 4.1 Å) with the peptidyl carrier protein domain of non-ribosomal tyrocidine synthase III from Bacillus brevis (Fig. 2B) (35). Most of the structural deviations among the carrier proteins are due to shifted positions of the four ␣-helixes within each structure, as well as alterations in the length and conformation of the loops connecting the ␣-helixes. But the position of the conserved serine residue that bears the 4Ј-PP arm is very close in all structures (Fig. 2B). Interestingly, despite the overall structural similarity, the amino acid sequence conservation within this group is not so profound: the multiple sequence alignment of the four carrier proteins from different pathways (non-ribosomal peptide synthase, polyketide synthase, ACP, and the intermediate domain) revealed only two strictly conserved residues, other than the serine modified by 4Ј-PP.
This study has demonstrated that the intermediate domain of FDH is a functional analog of carrier proteins with the 4Ј-PP arm. This prosthetic group must be attached to the conserved serine residue (Ser 354 ) in order for FDH to be catalytically active. Elimination of this modification (in our experiments via replacement of the serine with alanine) resulted in FDH lacking 10-fTHF dehydrogenase activity. Furthermore, the treatment of active FDH with alkaline phosphatase deprived the enzyme of 10-fTHF dehydrogenase activity. Such a treatment also yielded a compound with a molecular mass that corresponds to the molecular mass of 4Ј-PP. To complement these experiments, incubation of catalytically inactive FDH expressed in E. coli with lysates of mammalian or insect cells and CoA as a source of 4Ј-PP restored the enzyme activity. Incorporation of 4Ј-PP into FDH as well as into the intermediate domain itself was further confirmed in our in vitro experiments with a fluorescent reporter. Thus, it becomes apparent that the intermediate domain uses a 4Ј-PP arm to transfer a formyl group between the two catalytic centers (Fig. 5).
Atomic level details concerning the interface between FDH domains and the nature of their interaction with the 4Ј-PP swinging arm await the crystal structure of full-length FDH. But because the 4Ј-PP arm extends to ϳ20 Å in length (21), both the hydrolase and aldehyde dehydrogenase catalytic centers of FDH are well within reach, as can be concluded from their  respective crystal structures (17,18,36). The catalytic center of the hydrolase domain is relatively open (17,36), which perhaps facilitates easy access of the 4Ј-PP arm to the formyl group of the substrate. Moreover, two potential mechanisms could further facilitate access of the 4Ј-PP arm to the formyl group in the hydrolase catalytic center. In one of the mechanisms, a shift of the flexible loop bearing the catalytic residue Asp 142 could expose formyl to the domain surface (17). Another mechanism may involve larger scale rearrangements within the hydrolase domain making the catalytic center cleft more open (36). In contrast, the aldehyde dehydrogenase active site is less accessible, with its key catalytic cysteine located at the end of a 12-Å deep substrate entrance tunnel (18). However, the length of the 4Ј-PP arm is sufficient to bring the formyl group in close vicinity of this residue. Interestingly, the substrate entrance tunnel of the aldehyde dehydrogenase domain of FDH shares a characteristic with aldehyde dehydrogenases specific for larger aldehyde substrates, viz. a wide tunnel made up of amino acids with small side chains (37,38). Apparently, such a wide tunnel is necessary to allow access of the 4Ј-PP arm to the active site.
Our experiments have answered the previously unclear question of how the two functional domains of FDH communicate to merge the hydrolase and aldehyde dehydrogenase catalytic engines into one mechanism (Fig. 5). Notably, this study has identified the second metabolic pathway in higher organisms (in addition to fatty acid biosynthesis) that uses the 4Ј-PP prosthetic group. Interestingly, the two pathways, fatty acid biosynthesis and folate metabolism, are not closely related. Our finding also indicates that FDH requires post-translational modification to become a functional enzyme. A recently reported human 4Ј-PP transferase with broad specificity (39) could be a candidate to carry out this modification. (It appears that there is only one 4Ј-PP transferase gene in the human genome (40).) Interestingly, FDH expressed in E. coli appeared to be unmodified, although E. coli possesses several 4Ј-PP transferases (41). These results might indicate that these transferases are not efficient toward large mammalian proteins. Alternatively, the lack of a typical consensus motif for 4Ј-PP attachment can make FDH an inefficient substrate. Indeed, the corresponding sequence of rat FDH (Fig. 2B) indicates mismatches with the consensus motif (41,42), including two residues immediately up and downstream of the conserved serine. A third possibility to be considered is the activity of E. coli phosphodiesterase that removes the 4Ј-PP moiety from proteins (27).
Post-translational addition of the 4Ј-PP arm to ACP occurs after the protein is fully folded (41). Folded but unmodified FDH may occur if the modification process becomes deficient. Alternatively, hydrolytic enzymes, similar to a recently characterized ACP phosphodiesterase (27), may also alter the extent of phosphopantetheinylation of FDH in the cell. Whatever the mechanism, FDH without its 4Ј-PP prosthetic group could function as a folate-binding protein. Interestingly, the folate binding function was originally proposed for FDH (43). It is not clear at present whether FDH exists in the cell in an unmodified form. If it does, the distribution between catalytically functional and nonfunctional FDHs and the cellular conditions that affect this distribution would be of interest. Future studies should allow clarification of this matter.