Molecular Characterization of the Arabidopsis thaliana Flavoprotein AtHAL3a Reveals the General Reaction Mechanism of 4′-Phosphopantothenoylcysteine Decarboxylases*

The Arabidopsis thaliana flavoprotein AtHAL3a, which is linked to plant growth and salt and osmotic tolerance, catalyzes the decarboxylation of 4′-phosphopantothenoylcysteine to 4′-phosphopantetheine, a key step in coenzyme A biosynthesis. AtHAL3a is similar in sequence and structure to the LanD enzymes EpiD and MrsD, which catalyze the oxidative decarboxylation of peptidylcysteines. Therefore, we hypothesized that the decarboxylation of 4′-phosphopantothenoylcysteine also occurs via an oxidatively decarboxylated intermediate containing an aminoenethiol group. A set of AtHAL3a mutants were analyzed to detect such an intermediate. By exchanging Lys34, we found that AtHAL3a is not only able to decarboxylate 4′-phosphopantothenoylcysteine but also pantothenoylcysteine to pantothenoylcysteamine. Exchanging residues within the substrate binding clamp of AtHAL3a (for example of Gly179) enabled the detection of the proposed aminoenethiol intermediate when pantothenoylcysteine was used as substrate. This intermediate was characterized by its high absorbance at 260 and 280 nm, and the removal of two hydrogen atoms and one molecule of CO2 was confirmed by ultrahigh resolution mass spectrometry. Using the mutant AtHAL3a C175S enzyme, the product pantothenoylcysteamine was not detectable; however, oxidatively decarboxylated pantothenoylcysteine could be identified. This result indicates that reduction of the aminoenethiol intermediate depends on a redox-active cysteine residue in AtHAL3a.

The biosynthesis of coenzyme A from pantothenate includes the decarboxylation of (R)-4Ј-phospho-N-pantothenoylcysteine (PPC) 1 to 4Ј-phosphopantetheine (PP), a reaction that intro-duces the reactive cysteamine residue of coenzyme A. In eubacteria, the decarboxylation of PPC is catalyzed by the NH 2terminal CoaC domain of the Dfp proteins (1,2). Dfp is a bifunctional enzyme, and the COOH-terminal CoaB domain catalyzes the synthesis of PPC from 4Ј-phosphopantothenate and L-cysteine using cytidine 5Ј-triphosphate as the activating nucleotide (3). The coenzyme A biosynthetic pathway in plants is not fully understood. However, it was recently shown that the Arabidopsis thaliana trimeric flavoprotein AtHAL3a, which is linked to plant growth and salt and osmotic tolerance (4), catalyzes the same reaction as the CoaC domain of Dfp, the decarboxylation of PPC (5).
Dfp and AtHAL3a belong to a new family of flavoproteins that was named HFCD (homo-oligomeric flavin containing Cys decarboxylases (2,6)). Other members of this flavoprotein family include the LanD flavoenzymes EpiD and MrsD, which catalyze the oxidative decarboxylation of peptidylcysteines to peptidyl-␤-aminoethenethiols (7)(8)(9)(10)(11), a reaction involved in the biosynthesis of lantibiotics containing unsaturated thioether bridges such as epidermin and mersacidin. Lantibiotics are a group of ribosomally synthesized and posttranslationally modified antibiotic peptides containing the thioether amino acid lanthionine and other unusual amino acid residues (12,13). The structure of the EpiD peptidyl-␤-aminoethenethiol reaction products has been elucidated by NMR spectroscopy and mass spectrometry (9,14). EpiD, AtHAL3a, and Dfp all bind the cofactor FMN (4,10,15), whereas MrsD is a FAD-dependent enzyme (11). The HFCD proteins share the flavin binding motif and conserved active-site residues and are trimeric or dodecameric enzymes catalyzing the decarboxylation of cysteine residues (2,6).
The crystal structure of an active-site mutant of EpiD with a bound substrate peptide gave the first insight into the decarboxylation mechanism used by the HFCD proteins (6). Surprisingly, it looks like C ␣ -C ␤ dehydrogenation is not the initial reaction but rather the oxidation of the thiol group by the flavin cofactor. The spontaneous decarboxylation of the thioaldehyde group-containing intermediate then leads to the peptidylaminoethenethiolate reaction product (6).
The existence of oxidatively decarboxylated reaction products in the case of the LanD enzymes is a direct hint that the homologous PPC decarboxylases also use oxidation of the thiol group to enable decarboxylation of cysteine residues. Therefore, it was concluded that a "peptidyl"-aminoenethiol is an intermediate of the PPC decarboxylases (2,5) and that this compound is finally reduced to 4Ј-phosphopantetheine (see Fig.  7). Mechanistic studies on the Dfp enzyme such as detection of substrate/product-induced charge-transfer complexes also favor a mechanism via oxidation of the thiol group and the presence of a thioaldehyde group-containing intermediate (16). However, the existence of this oxidatively decarboxylated intermediate has not been proven, and up to now, the intermediate itself has not been purified or characterized.
The four characterized HFCD proteins EpiD, MrsD, Dfp (CoaC) and AtHAL3a are similar in sequence and the reaction they catalyzed, but they use different substrates, with peptidylcysteines on one hand and 4Ј-phosphopantothenoylcysteine on the other. Crystal structure analysis of EpiD with bound substrate peptide and a theoretical model for the binding of PPC to AtHAL3a showed that substrate binding involves an NH 2 -terminal substrate binding helix and a COOH-terminal substrate binding clamp (5,6). The substrate binding clamp of EpiD forms an antiparallel ␤-sheet with the residues SSG (Ser-152 to Gly-154) in the turn region. The binding clamp and substrate peptide together form a three-stranded ␤-sheet (6). The substrate binding clamp of the PPC decarboxylases AtHAL3a and Dfp is four residues shorter than that of EpiD and contains a conserved ACGD motif. The conserved Cys residue of this motif (Dfp, Cys 158 ; AtHAL3a, Cys 175 ) aligns with Ser 153 of EpiD, and modeling of the AtHAL3a-PPC complex suggests that the conserved Cys residue is in the vicinity of the substrate cysteinyl moiety and might participate in catalysis (5).
For Dfp the substrate binding structural elements have been partially characterized and a PPC decarboxylase signature defined. The conserved lysine residue of the NH 2 -terminal G(G/ S)IAXYK motif of the Dfp proteins is probably important for the binding of the phosphate group of PPC, whereas the exchange of the residue Cys 158 led to loss of PPC decarboxylase activity (1).
To elucidate the reaction mechanism of PPC decarboxylases, two central questions have to be answered. First of all, it has to be proven that decarboxylation of PPC involves an oxidatively decarboxylated substrate. Second, elucidation of the mechanism by which this compound is reduced by FMNH 2 to complete the reaction cycle of the flavoprotein must occur. To address these questions, we chose the following approach. It should be possible to trap the reaction at the oxidatively decarboxylated intermediate and to prevent complete reduction by changing the geometry between the enzyme-bound intermediate and reduced flavin coenzyme. This could be achieved by changing residues involved in the binding of PPC, such as residues of the NH 2 -terminal binding helix or the substrate recognition clamp and/or by changing the substrate. Using site-directed mutagenesis, it should also be possible to identify amino acid residues involved in reoxidation of FMNH 2 and to trap the intermediate in this way. We decided to carry out these experiments with AtHAL3a, which encodes only the PPC decarboxylase but not the PPC synthetase activity.
Here we present data that AtHAL3a not only decarboxylates PPC but also pantothenoylcysteine (PC) and that the conserved Lys residue of the NH 2 -terminal binding helix is important for discriminating between phosphorylated and non-phosphorylated substrate. We then show the purification and characterization of oxidatively decarboxylated PC obtained by incubation with mutant AtHAL3a enzymes. Furthermore, we present data indicating that the conserved Cys 175 residue is important for reduction of the oxidized intermediate.

Plasmid Construction
In General-PCR amplifications were performed with Vent-DNA polymerase (New England BioLabs). The entire sequences of the AtHAL3a-coding regions of the constructed plasmids were verified. The oligonucleotides used were purchased from MWG Biotech.
Purification of His-AtHAL3a Proteins-500 ml of isopropyl-1-thio-␤-D-galactopyranoside-induced E. coli BL21 (DE3) pET28a(ϩ) AtHAL3a cells were harvested and disrupted by sonication in 10 ml of 20 mM Tris-HCl, pH 8.0. 3 ml of the cleared lysate, obtained by two centrifugation steps (each 20 min at 30,000 ϫ g at 4°C), was diluted with 3 ml of column buffer (20 mM Tris-HCl, pH 8.0, 10 mM imidazole, 300 mM NaCl) and applied to an equilibrated Ni-NTA column containing ϳ0.5 ml of Ni-NTA-agarose (Qiagen). The column was then washed with 10 ml of column buffer. His-AtHAL3a and mutant His-AtHAL3a proteins were eluted with column buffer containing 250 mM instead of 10 mM imidazole, and the yellow peak fractions (ϳ400 l) were collected. If not otherwise stated, immediately after elution from the column, DTT was added to a final concentration of 5 mM. For gel filtration, a 25-l aliquot of the Ni-NTA eluate was subjected to a Superdex 200 PC 3. sulfate-polyacrylamide (10%) gel electrophoresis (18) under reducing conditions.
AtHAL3a Assays-Approximately 50 -100 g of PPC as a calcium salt (2) were incubated with 0.5-1.0 g of His-AtHAL3a for 20 min at 37°C in a total volume of 0.70 ml of 50 mM Tris/HCl, pH 8.0, 5 mM DTT. The decarboxylation of pantothenoylcysteine used as barium salt (2) was assayed in the same way; however, for each assay 475 g of substrate were used in a total volume of 1 ml. The purity of both substrates was in the range of 60 -80%, and it was verified by mass spectrometry that no 4Ј-phosphopantetheine or pantothenoylcysteamine was present in the synthesized substrates. However, PPC contained minor amounts of the unphosphorylated pantothenoylcysteine. The His-AtHAL3a mutants were adjusted to the same absorbance at 455 nm to enable a direct comparison of their activities. The reaction mixtures were kept at -80°C and then were successively separated by reversed phase chromatography (RPC) with a RPC C 2 /C 18 SC 2.1/10 column on FIG. 2. Gel filtration of mutant His-AtHAL3a proteins under reducing conditions. Immediately after elution from the Ni-NTA-agarose column, 5 mM DTT was added to the protein solutions. The IMAC-enriched His-AtHAL3a and mutant His-AtHAL3a proteins were then separated by gel filtration, and the elution was followed by absorbance at 280 nm (upper line), 378 nm (not shown), and 450 nm (lower line). The Superdex 200 PC 3.2/30 column used was calibrated with standard proteins to correlate the elution volume with molecular weight information as recently described (2). All the His-AtHAL3a proteins eluted at about 1.382 ml. This elution volume corresponds to an apparent molecular mass of about 110 kDa (5) and is in accordance with the trimeric structure of His-AtHAL3a-FMN (21).
FIG. 3. Gel filtration of mutant His-AtHAL3a proteins under oxidizing conditions. IMAC-enriched His-AtHAL3a protein solutions containing no DTT were subjected to gel filtration after different times of incubation at 4°C (His-AtHAL3a wt, 120 min; His-AtHAL3a N142D, 30 min; His-AtHAL3a C175S, 20 h; His-AtHAL3a G179A, 20 h). His-AtHAL3a wt containing 5 mM DTT was used in the control experiment. In the absence of DTT, His-AtHAL3a wt, N142D, and G179A oxidize and form higher multimers, whereas the elution volume of His-AtHAL3a C175S is not changed.
a SMART system (Amersham Biosciences). Compounds were eluted with a linear gradient of 0 -50% acetonitrile, 0.1% trifluoroacetic acid in 5.8 ml with a flow rate of 200 l/min. The absorbance was measured simultaneously at 214, 260, and 280 nm to enable identification of oxidatively decarboxylated intermediates (7,8). The fractions obtained were analyzed by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry (ESI-FTICR-MS) as described below.
High Resolution Mass Spectrometry-ESI-FTICR-MS (19) measurements were carried out with a passively shielded 4.7-Tesla APEX TM II-ESI/matrix-assisted laser desorption ionization-FTICR mass spectrometer (Bruker Daltonik). The mass spectrometry software XMASS version 5.0.10 (Bruker Daltonik) was used for mass calculation, data acquisition, and processing. Mass calculation was performed with the standard elemental mass compilation of Audi and Wapstra (20). The measured mass range was m/z 200 -2000, and broadband excitation took place from m/z 100 to 2000. In general, 524,288 data points were acquired. For positive ion investigations, 50 l of methanol and 1% acetic acid were added to the sample solutions (ϳ40 -70-l fractions from RPC). Electrospray ionization was performed in the positive mode with a grounded capillary sprayer needle mounted 60°off-axis (Analytica of Branford). No nebulizer gas was necessary to support the spray process. In general, the spectrum of a blank run (solvent and 1% acetic FIG. 4. PPC decarboxylase activity of mutant AtHAL3a proteins. PPC was incubated with purified wild-type and mutant His-AtHAL3a proteins. The reaction mixtures were separated by RPC, and the elution was followed by absorbance at 214 nm. The reaction product PP eluted before PPC, as has been shown by electrospray ionization mass spectrometry (2). The mutant His-AtHAL3a proteins were analyzed under comparable conditions together with wild-type His-AtHAL3a in two sets of experiments (A, control, AtHAL3a wt, AtHAL3a V30I, AtHAL3a I33L, AtHAL3a I33V, and AtHAL3a K34N; B, AtHAL3a wt, AtHAL3a K34Q, AtHAL3a K34R, AtHAL3a R95Q, AtHAL3a N142D, AtHAL3a M145L, AtHAL3a A174S, AtHAL3a A174V, AtHAL3a C175S, AtHAL3a D177N, AtHAL3a G179A, and AtHAL3a G181A). The mutant His-AtHAL3a proteins N142D, M145L, and C175S showed no activity (only with high enzyme concentrations very low residual activities might be detectable), whereas AtHAL3a D177N and G179A very significantly reduced PPC decarboxylase activity. The mutant proteins A174S and G181A showed significantly reduced activity. An additional reaction product (labeled with an asterisk) was clearly detectable for the Lys 34 mutants K34N and K34Q. However, minor amounts of this reaction product were also detectable for all active AtHAL3a mutants (including the R95Q mutant, data not shown) and wild-type enzyme. Further experiments showed that the substance labeled with an asterisk is pantothenoylcysteamine, which is derived from pantothenoylcysteine by decarboxylation (compare Fig. 5 and Fig. 7).

RESULTS AND DISCUSSION
Purification of Mutant His-AtHAL3a Proteins-AtHAL3a and mutant AtHAL3a genes were expressed as His-tag fusion proteins purified from the corresponding E. coli clones by immobilized metal affinity chromatography (IMAC) and additionally characterized by gel filtration (Figs. 1 and 2). All the mutant proteins have the same elution volume as wild-type His-AtHAL3a, and the ratio of the absorbance values at 280 and 450 nm was not significantly altered by the introduced mutations. We conclude that the introduced mutations do not affect the trimeric structure and coenzyme binding of AtHAL3a, indicating that the overall three-dimensional structure is not significantly altered. In the absence of DTT, trimeric His-AtHAL3a proteins oxidize and form higher multimers (Fig. 3).

Mutations within the NH 2 -terminal Binding Helix of AtHAL3a-Recent molecular characterization of CoaC activity
showed that mutations within the NH 2 -terminal G(G/S)IAXYK motif of the Dfp proteins, especially the exchange of the residue Lys 20 for Gln or Asn, decreased PPC decarboxylase activity (1). It was suggested that the conserved Lys residue is involved in binding of the phosphate group of PPC. This assumption is supported by the model for binding of PPC to AtHAL3a (5). Therefore, we were interested in characterizing the Lys 34 residue within the motif GSVAAIK in the plant AtHAL3a protein.
For the K34Q mutant, we could show a small decrease of PPC decarboxylase activity (Fig. 4). However, we also observed that exchange of the residue Lys 34 for Asn or Gln led to an additional reaction product. This compound is only present in very low amounts if wild-type enzyme is used and, therefore, has not been detected before. Small amounts of this additional compound are also observed when the mutants I33V and K34R are used. Because we were aware that the synthetic PPC used is contaminated with pantothenoylcysteine, we decided to investigate if, in contrast to Dfp, AtHAL3a and mutant AtHAL3a proteins are able to decarboxylate PC (see below). We have previously shown that AtHAL3a is not as specific as Dfp (5).
Pantothenoylcysteine Is Decarboxylated by AtHAL3a-To analyze if AtHAL3a not only decarboxylates PPC but also PC, synthetic PC was incubated with wild-type and mutant His-AtHAL3a proteins. The reaction mixtures were then separated by RPC, and the masses of the eluted compounds were determined. AtHAL3a enzymes, which were active in decarboxylation of PPC, were also active in the decarboxylation of PC (Fig.  5). The mass difference between PC and the observed reaction product was determined to be 43.990 Da (not shown but see Fig. 6), which is in excellent agreement with the monoisotopic mass of CO 2 (the calculated monoisotopic mass is 43.9898 Da). The reaction product pantothenoylcysteamine eluted with a retention time of about 24.0 -24.2 min. This retention time is in agreement with that of the observed additional reaction product of the AtHAL3a K34N/K34Q reaction with synthetic PPC (see above; minor shifts in the overall retention times are explained by changes in the column material with increasing numbers of applications). We conclude that, in a mixture of PPC and PC, wt AtHAL3a preferentially decarboxylates PPC and that this discrimination between both substrates depends on Lys 34 . This is a direct hint that Lys 34 of AtHAL3a binds the phosphate group of PPC. If no PPC is present, there should be no difference between AtHAL3a and AtHAL3a K34N/K34Q in the reaction with PC, because this substrate lacks the phosphate group (the data presented in Fig. 5 support this view). The elucidation of the kinetic parameters for AtHAL3a and AtHAL3a K34N/K34Q for both substrates is impeded by the lack of a suitable enzyme assay. Moreover, the determination of decarboxylase activities for different ratios of PPC and PC is only possible by separating the reaction products (as it is shown in Fig. 4).
The Conserved Asn 142 and Met 145 Residues of the PXM-NXXMW Motif-The PXMNXXMW motif is conserved within the HFCD proteins, and crystal structure analysis of EpiD revealed the importance of this motif (residues 114 -121 of EpiD) for binding of the substrate and the flavin cofactor (6). The conserved residue Asn 117 of EpiD contacts the C ␤ hydrogen atoms and the carboxylate group of the substrate cysteinyl moiety. It appears that this Asn residue (EpiD, Asn 117 ; MrsD, Asn 125 ; Dfp, Asn 125 ; AtHAL3a, Asn 142 ) and the conserved His residue of the HCFD proteins (EpiD, His 67 ; MrsD, His 75 ; Dfp, His 75 ; AtHAL3a, His 90 ) are essential for the decarboxylase reaction. Met 120 of EpiD is located above the pyrimidine system of the FMN cofactor. For the eubacterial enzyme Dfp, the importance of the conserved Asn and Met residues for the PPC decarboxylase activity has already been published (1,2). To verify these data and the proposed model of PPC binding to AtHAL3a, the mutant proteins His-AtHAL3a N142D and His-AtHAL3a M145L were purified and characterized. As expected, both enzymes were inactive in decarboxylating PPC and PC (Figs. 4 and 5).
Mutations within the Substrate Recognition Clamp of AtHAL3a-Recently, crystal structure analysis of EpiD H67N with bound peptide DSYTC revealed that the pentapeptide is embraced by a 20-amino acid substrate recognition clamp comprising residues Pro 143 to Met 162 . Residues of the NH 2 -terminal binding helix, the residue Asn 117 , and the NHI motif (containing the above mentioned conserved His residue) are also important for substrate binding (6). We were interested in changing the way that PC and PPC bind to AtHAL3a to detect the oxidatively decarboxylated intermediate. Therefore, we not only exchanged Asn 142 and residues of the NH 2 -terminal binding helix (see above) but also residues within the proposed PPC binding clamp of AtHAL3a. Characterization of the substrate recognition clamp of HFCD proteins by site-directed mutagenesis has not been published apart from the preliminary characterization of the conserved Cys residue of the PPC decarboxylases (1, 5). We mutated residues Ala 174 , Cys 175 (see below), Gly 179 , and Gly 181 of the AtHAL3a substrate recognition clamp Pro 168 -Ile-Lys-Lys-Arg-Leu-Ala 174 -Cys 175 -Gly-Asp-Ile-Gly 179 -Pro-Gly 181 -Arg-Met 183 and elucidated the activity of the altered enzymes with both PPC and PC (Figs. 4 and 5). All of these residues are conserved in the eubacterial Dfp proteins (1). The mutant protein AtHAL3a D177N was included in the studies, because both AtHAL3a and E. coli Dfp have the ACGD motif, which is not present in the LanD enzymes EpiD and MrsD. Activity of AtHAL3a A174S, G179A, and G181A was significantly reduced with both substrates compared with wildtype AtHAL3a. D177N showed very low activity. However, activity of AtHAL3a A174V was comparable with wild-type activity, indicating that a hydrophobic side chain in position 174 is important for hydrophobic interactions with the substrate (compare the published model of PPC binding to AtHAL3a (5)). As described below, further characterization of the AtHAL3a G179A reaction with pantothenoylcysteine then led to the identification of the proposed intermediate.
Purification and Characterization of Oxidatively Decarboxylated Pantothenoylcysteine-From the characterization of the flavoenzyme EpiD it was known that oxidatively decarboxylated peptidylcysteines are characterized by their absorbance properties. The aminoenethiol group NHOCHACHOSH is a strong chromophore, and the UV FIG. 7. Reaction mechanism of the decarboxylation of pantothenoylcysteine catalyzed by AtHAL3a. Recently, it has been suggested that C ␣ -C ␤ dehydrogenation is not the initial reaction of the LanD proteins EpiD and MrsD but, rather, oxidation of the thiol group by FMN. In this mechanism a thioaldehyde group-containing intermediate is formed, and its spontaneous decarboxylation (in analogy to the decarboxylation of ␤-keto acids) leads to the peptidylaminoenethiolate/peptidylaminoenethiol reaction product (6). However, in case of the PPC decarboxylases it is not the aminoenethiolate compound but the cysteamine derivative that is the final product. The results presented in this paper clearly demonstrate that the decarboxylation of pantothenoylcysteine follows the oxidative decarboxylation mechanism of the LanD proteins, since oxidatively decarboxylated PC could be identified by mass spectrometry and UV-visible spectroscopy. The enethiol/enethiolate intermediate is then reduced by FMNH 2 completing the reaction cycle. The reaction of AtHAL3a C175S with PC does not lead to pantothenoylcysteamine (decarboxylated PC). However, pantothenoylaminoethenethiol (oxidatively decarboxylated PC) was detectable, and therefore, it is reasonable to assume that Cys 175 is involved in the oxidation of FMNH 2 and/or reduction of the aminoenethiol intermediate. Furthermore, this Cys residue is conserved in the eubacterial Dfp proteins, and it has been shown that Dfp C158A and Dfp C158S are inactive in the decarboxylation of PPC to PP (1,5). However, the exact catalytic role of the conserved Cys residue remains open for further investigations. We assume that the decarboxylation mechanism is not dependent of the used substrate (PC or PPC). However, the oxidatively decarboxylated intermediate was present in assignable amounts for PC but not for PPC. This probably reflects the fact that PPC is more tightly bound by the enzyme, and therefore, changes within the substrate binding clamp do not influence the second half reaction of the flavoenzyme to the same extent. spectra of the EpiD reaction products were pH-dependent, with an absorption maximum of 259 nm under acidic conditions (pH Յ 4.2) for the enethiol form and 283 nm at pH 7.0 for the enethiolate form; the isosbestic point was determined to be 271 nm. It was shown that the molar extinction coefficient (⑀) of SFNSYVONHOCHACHOSH at 259 nm is at least ⑀ Ϸ 6,800 M Ϫ1 cm Ϫ1 and that the pK a value of the enethiol group is about pH 6.0 (7). Therefore, the reversed phase separation of the AtHAL3a reaction mixtures was not only followed by absorbance at 214 nm but also by absorbance at 260 and 280 nm. Using PPC as substrate, we did not unambiguously detect compounds with an increased absorbance at 260 nm, even if mutant AtHAL3a proteins were used in the decarboxylation assay (data not shown). However, the combination of pantothenoylcysteine as a substrate and mutant His-AtHAL3a enzymes such as R95Q, A174S, A174V, C175S, D177N, and G179A enables the identification of a compound that showed strong absorption at 260 nm (Fig. 5B). The residues Ala 174 , Cys 175 , Asp 177 , and Gly 179 are within the proposed substrate recognition clamp of AtHAL3a and could directly contact the substrate. Gly 179 could also be important for the maintenance of the correct secondary/tertiary structure of the clamp. We believe that changing the structure of the recognition clamp results in a different active-site architecture and that the different geometry of the substrate/ intermediate FMN/FMNH 2 pair prevents complete reduction of the oxidatively decarboxylated intermediate by FMNH 2 . It is also possible that structural changes within the recognition clamp enables reoxidation of FMNH 2 by oxygen. Modeling of PPC binding to AtHAL3a shows that also residue Arg-95 could contact the substrate (5).
The putative oxidatively decarboxylated PC (pantothenoylaminoethenethiol) was purified in larger amounts by incubation of PC with His-AtHAL3a G179A, because incubation with this enzyme led to the largest amounts of the intermediate. After separation of the reaction mixture by RPC, the reaction products were analyzed by ESI-FTICR-MS (Fig. 6). The MS data prove that PC is decarboxylated to pantothenoylcysteamine and that the compound with increased absorbance at 260 nm (and 280 nm) is pantothenoylaminoethenethiol. Interestingly, a second compound with the mass of pantothenoylaminoethenethiol could also be detected, and we propose that the two compounds are cis and trans isomers of the aminoenethiol intermediate.
The Conserved Cys 175 Residue of the Substrate Binding Clamp-Recently, it has been shown that the conserved Cys residue of the substrate recognition clamp of Dfp proteins, which is missing in the LanD enzymes EpiD, MrsD, and MutD (6), is essential for activity (1,5). We could prove this for AtHAL3a, because His-AtHAL3a C175S is inactive in decarboxylation of PPC to PP and of PC to pantothenoylcysteamine (Figs. 4 and 5). However, we detected pantothenoylaminoethenethiol when this mutant His-AtHAL3a enzyme was incubated with PC (Fig. 5B). Therefore, the C175S mutant differs from the R95Q, A174S, A174V, D177N, and G179A mutants, which were able to reduce the oxidized intermediate but had a very low decarboxylase activity (Figs. 4 and 5). These mutants are probably substrate binding mutants. We suggest that Cys 175 is directly involved in the reaction mechanism by reoxidation of the FMNH 2 cofactor and/or by reduction of the oxidized intermediate (Fig. 7). Oxidation of the thiol side chain of the substrates would still be possible with the C175S mutant (if Cys 175 has the proposed function), and consequently, spontaneous decarboxylation occurs. However, the reduced coenzyme will not be reoxidized by the substrate so that the native reaction cycle is interrupted, and only oxidatively decarboxylated but no decarboxylated compounds are detectable. Partial reoxidation of FMNH 2 by oxygen can explain the fact that oxidatively decarboxylated intermediates are present in greater molar amounts than the enzyme used. Because C 175 itself is part of the substrate recognition clamp, the previously mentioned structural changes of the clamp, for example the G179A exchange, not only change the binding of the reaction intermediate but also the geometry between reduced coenzyme, Cys 175 , and the intermediate.
The elucidation of the mechanism of the second half reaction of the PPC decarboxylases will require further detailed biochemical and structural studies. However, it is tempting to speculate that the aminoenethiol/enethiolate group of the intermediate, NHOCHACHOSH, and the side chain of Cys 175 , CH 2 OSH, form the mixed disulfide NHOCH 2 OCH 2 OSOSOCH 2 during one stage of the second half reaction of the PPC decarboxylases. In this model, we suggest that disulfide formation concurs with reduction of the double bond, although the mechanism of this reaction is not clear. The disulfide is finally reduced by FMNH 2 , generating the cysteamine residue of the substrate and restoring the thiol side chain of Cys 175 .
Interestingly, in contrast to wild-type AtHAL3a and other mutants, trimeric AtHAL3a C175S cannot form dimers (of trimers) under oxidizing conditions (Fig. 3). In the absence of a substrate, the recognition clamp has no defined structure (6), and Cys 175 could form intermolecular disulfide bridges.
Conclusions-In this paper, we have shown that the plant flavoprotein AtHAL3a not only decarboxylates PPC but also PC and that the discrimination between both substrates involves the conserved Lys 34 residue. We were able to show that pantothenoylaminoethenethiol is the intermediate in decarboxylation of PC. We conclude that the decarboxylation of PC and PPC follows the mechanism decarboxylation by initial oxidation (Fig. 7). The role of the flavin cofactor of the PPC decarboxylases AtHAL3a and Dfp is therefore the same as for the LanD enzymes EpiD and MrsD. However, it appears that the difference in structure of the substrate binding clamp of PPC decarboxylases compared with LanD enzymes (for example the presence of the ACGD motif in the PPC decarboxylases) not only causes the difference in substrate specificity but also a different overall reaction. PPC decarboxylases are able to reduce the aminoenethiol group to a cysteamine residue, whereas LanD decarboxylases are not able to reoxidize their flavin cofactor in this way. Starting with the characterization of the peptidylcysteine decarboxylase EpiD a decade ago, we now understand how the decarboxylation of Cys residues occurs.