|Home | About | Journals | Submit | Contact Us | Français|
Polyhydroxybutyrate (PHB) synthases catalyze the conversion of 3-hydroxybutyryl coenzyme A (HBCoA) to PHB of molecular weight 1.5 million. The class III synthase from Allochromatium vinosum is a tetramer of PhaEPhaC (each 40 KDa). The polymerization involves covalent catalysis using C149 of PhaC with one PHB chain per PhaEC dimer. Two mechanisms for elongation have been proposed. The first involves an active site composed of two monomers where the growing hydroxybutyrate (HB) chain alternates between C149 on each monomer. The second involves C149 and covalent and non-covalent (HB)nCoA intermediates. Two approaches were investigated to distinguish between these models. The first involved the wt-PhaEC primed with sTCoA (a CoA ester of (HB)3 in which the terminal HO is replaced with an H) which uniformly loads the enzyme. The primed synthase was reacted with [1-14C] HBCoA by a rapid chemical quench method and analyzed for covalent and non-covalent intermediates. Radiolabel was found only with the protein. The second approach used C149S-PhaEC which catalyzes polymer formation at 1/2200 the rate of wt PhaEC (1.79 min−1 vs 3900 min−1). C149S-PhaEC was incubated with [1-14C] HBCoA and chemically quenched on the min time scale to reveal non-covalently bound [1-14C] (HB)2CoA and (HB)3CoA as well as covalently labeled protein. Synthesized (HB)nCoA (n = 2, 3) were shown to acylate PhaEC with rate constants of 1 to 2 min−1 and these species are converted into polymer. The (HB)n-CoA function as kinetically and chemically competent intermediates. These results support the mechanism involving covalently and non-covalently bound intermediates.
Polyhydroxybutryate (PHB) synthases catalyze the conversion of 3-(R)-hydroxybutryl coenzyme A (HBCoA) to PHBs. These enzymes are representative of a large number of polymerases found in nature which use common, water soluble metabolites as substrates, require no template, and undergo a phase transition during the polymerization process to form insoluble inclusions or granules.(1–6) PHBs are produced by most bacteria when they find themselves in a nutrient limited environment and have a readily available carbon source. Up to 95% of the cell dry weight of the organism can be converted into polyoxoesters. When the environment is no longer nutrient limited, the polymers are degraded by depolymerases, releasing energy and monomers for biosynthesis.(1, 7–8) PHBs and polyhydroxyvalerates in the appropriate ratio are of general interest as they have properties of thermoplastics and are biodegradable. They can potentially provide an alternative to the petroleum based, non-biodegradable plastics, if they can be produced in an economically competitive fashion. Many of the proteins involved in biosynthesis and degradation of PHBs have been identified.(8–14) Our efforts have focused on understanding the mechanism of initiation, elongation and granule formation, and termination.(7–8, 15–21) The prototypes for these studies have been the class I synthase from Ralstonia eutropha (22) and the class III synthase from Allochromatium vinosum (18, 23–24) In the present paper we describe experiments that establish that both covalent and non-covalent intermediates are involved in the polymerization process in the class III synthases.
The class III synthase from A. vinosum (PhaEC), isolated from E. coli, is composed of 1:1 complex of two proteins: PhaC (39 kDa) and PhaE (40 kDa).(23) PhaC catalyzes the polymerization reaction, while PhaE is not homologous to any protein of known function. However, the presence of PhaE is essential for an active and well behaved synthase PhaC.(18) The class III synthase is a tetramer of PhaEC with a small amount of dimer. Both oligomeric states are active. However, it should be noted, that the synthase has never been isolated from its host organism.
We have established that the polymerization reaction catalyzed by PhaEC involves covalent catalysis using a catalytic dyad: histidine (H331) and cysteine (C149).(18) H331 activates C149 for nucleophilic attack on HBCoA to form an acylated thiol ester intermediate. Ester formation requires an aspartate, D302, which is proposed to function as a general base catalyst in the activation of the hydroxyl group of HBCoA for nucleophilic attack on the acylated enzyme.(17, 19) We have previously proposed two mechanisms for PHB formation. Both involve chain elongation through acylated C149. The mechanism in Scheme 1A proposes that subsequent to acylation of C149 (HB-PhaC), a second HBCoA reacts with the HB-C149 to generate a non-covalently bound (HB)2-CoA that then rapidly re-acylates C149. This mechanism is a variant of that proposed for some type III polyketide synthases.(25) The mechanism in Scheme 1B requires that the active site is at the interface of two PhaC monomers. In this model, the growing (HB)n chain is always covalently attached to C149 and switches from one monomer to the other upon addition of each subsequent HB unit. This mechanism is a variant of that established for fatty acid synthases where a cysteine from a ketosynthase and a pantetheinylated thiol attached to the acyl carrier protein are involved in covalent catalysis.(26)
Efforts to distinguish between the two mechanisms are the focus of this paper. A number of peculiarities associated with PhaEC have made a distinction between the two mechanisms more difficult than might appear at first glance. We have established with both wild-type (wt) class I and III synthases that the rate of elongation is much faster than the rate of initiation and that uniform loading of the protein essential for studying the elongation process specifically cannot be achieved.(20, 22) Even when the substrate to enzyme (S/E) ratio is 5, a small amount of protein is covalently linked to a huge HB-polymer, while most of the protein remains unmodified. To overcome this problem and ensure uniform loading of the synthase, we previously designed “artificial” primers,(21) such as the saturated trimer-CoA (sTCoA), an analog of (HB)3CoA where the terminal hydroxyl group is substituted with a hydrogen. Our previous studies showed that incubation of wt-PhaEC with [3H]-sTCoA resulted in the labeling of 0.5 sT per PhaEC based on CoA release.(16) When the acylated protein, however, was subjected to rapid denaturation, trypsin digestion and HPLC analysis, three peptides with the same sequence were isolated with trimeric, tetrameric and pentameric HB units attached to C149.(18, 21) PhaEC was caught in the act of elongation based on the presence of a small amount of HBCoA present in the sTCoA solution due to the fortuitous breakdown of the sTCoA on storage. Thus, our first strategy to distinguish between mechanisms in Scheme 1 was to use wt-PhaEC acylated with sTCoA followed by rapid mixing with [1-14C] HBCoA and rapid chemical quench (RCQ) methods to look for sT-(HB)nCoA intermediates and PhaC-(HB)n. Only PhaC-(HB)n was detected suggesting that if noncovalent sT-(HB)nCoA intermediates are generated they rapidly re-acylate C149.
The second strategy was to examine a variety of PhaEC mutants to find one with reduced rate of re-acylation. The active site serine mutant (C149S) is shown to form PHB with a rate constant of 1.79 min−1, 1/2200 the rate of the wt-PhaEC.(27) Studies are presented here in which C149S-PhaEC incubated with [1-14C]-HBCoA resulted in detection of (HB)nCoA (n = 2, 3) and PhaC-(HB)n. Experiments with synthetically prepared (HB)nCoA (n = 2, 3) revealed that these noncovalently detected intermediates function in a chemically and kinetically competent fashion. The results together support the mechanism proposed in Scheme 1A involving covalent and non-covalent intermediates.
Racemic (R,S)-[1-14C]HBCoA was purchased from American Radio Labeled Chemicals Inc. and was diluted with (R)-HBCoA synthesized according to Yuan et al.(27) Compounds sTCoA and [3H]-sTCoA were synthesized according to Jia et al.(17) ESI-MS and MALDI-TOF mass spectra were performed by MIT proteomics core facility and MIT biopolymers lab, respectively. RCQ experiments were carried out on RQF-3 from Kintek. Unless otherwise specified, HPLC was performed using an adsorbosphere nucleoside-nucleotide column (Alltec, 7μ, 4.6 × 250 mm) on a Waters HPLC instrument equipped with a 515 pump and 2996 photodiode array detector. A typical elution protocol used 20 mM KPi, pH 4.7 (solvent A) and methanol (solvent B) at a flow rate of 1.0 mL/min and a linear gradient from 10–70% in 0–35 min, 70–95% in 35–45 min, and 95–95% at 45–50 min. Typical recoveries based on scintillation counting were 80–90% of the loaded material.
Recombinant wt- and C149S-PhaEC were over-expressed from plasmids pET-UM4 and pET-UM22, respectively and purified to homogeneity as previously reported.(16, 18) Synthases were assayed by the discontinuous method using 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB) to monitor CoA release as previously described.(16) The specific activity for wt- and C149S-PhaEC was 120 and 0.05 μmol/min/mg at 37 °C, respectively.
In a total volume of 350 μL, 40 μM wt-PhaEC and 0.52 mM sTCoA in 20 mM KPi (pH 7.5) and 50 mM NaCl (Buffer A) were incubated for 1 min at 37 °C. The sT-wt-PhaEC was immediately loaded into one syringe of the RCQ apparatus and mixed with 2 mM [1-14C] HBCoA (specific activity (SA) = 2000 cpm/nmol) at an S/E ratio of 1 or 50 in the second syringe. At different time points (2.5–820 ms) the reaction was quenched with 2% HClO4. At the end of the time course, each aliquot was centrifuged and the supernatant was removed and placed on ice. The precipitated protein was washed with H2O (75 μL × 3), redissolved in 100 μL 10% SDS and analyzed by liquid scintillation counting. The pooled washes were combined with the supernatant and the pH of the solution was adjusted to ~ 6.0 by careful titration with 0.5 M NaOH at 4 °C. The solution was then filtered using a Whatman microsyringe filter. The filtration unit was washed with H2O (75 μL× 3). The filtrate and washes were combined, concentrated, and analyzed by HPLC. Fractions (1 mL) were collected and each fraction was analyzed by scintillation counting.
Using the rapid chemical quench method, 40 μM wt-PhaEC in Buffer A was mixed at 37 °C with an equal volume of 0.52 mM [3H]-sTCoA (S/E = 13, SA = 2400 cpm/nmol) in the same buffer. The reaction was quenched with 2% HClO4 at 50 ms, 100 ms, 300 ms, 500 ms and 1 s. At the end of the experiment, each aliquot was centrifuged and the supernatant was removed. The precipitated protein was washed with H2O (75 μL × 3), redissovled in 100 μL 10% SDS and analyzed by liquid scintillation counting.
The reaction mixture of 300 μL contained 100 μM C149S-PhaEC and 1.3 mM [3H]-sTCoA (S/E=13, SA=1278 cpm/nmol) in Buffer A and was incubated at 37 °C. At 10, 30, 45, 60, 90 and 120 s, a 40 μL aliquot was withdrawn and quenched with 100 μL 10% trichloroacetic acid (TCA). The quenched reaction mixtures were centrifuged followed by the removal of the supernatant. The precipitated protein was washed with H2O (75 μL × 3), redissolved in 100 μL 10% SDS and analyzed by scintillation counting.
Assays were carried out in a final volume of 200 μL in Buffer A containing 20 μM C149S-PhaEC and 50 μM to 6 mM HBCoA. The mixture was incubated at 37 °C and 20 μL aliquots were removed at various times and quenched with 50 μL 10% TCA. Each sample was centrifuged to remove the precipitated protein, and then 65 μL of the quenched reaction mixture was added to 70–250 μL of 0.25 mM DTNB in 0.5 M KPi (pH 7.8) and A412 was measured. Each substrate concentration was run in triplicate. The kinetic parameters were determined by fitting the data to Eq. 1.
In a final volume of 10 μL, 20 μM C149S-PhaEC was incubated with 100 μM to 6 mM [1-14C]-HB-CoA (SA = 2 × 104 cpm/nmol) in Buffer A for 60 min at 37 °C. Experiments with S/E ratios of 5, 50, 100, and 300 were carried out. The reactions were stopped by addition of an equal volume of Laemmli buffer with no reducing reagent. Furthermore, the sample was not boiled. From each quenched reaction mixture, 15 μL (12 μg protein) was loaded onto a 10% SDS-PAGE gel of 1.0 mm thickness. The gel was stained with Coomassie for 10 min, destained in fast destain solution for 30 min, transferred to slow destain solution for 15 min, rinsed in H2O for 1 min, and dried immediately. The dried gels were exposed to the low-energy screen (Molecular Dynamics) for 48 h and then scanned using the Storm Imaging System and analyzed using ImageQuant TL software (Amersham Biosciences).
In a final volume of 155 μL (at S/E = 50) or 455 μL (at S/E = 300), 20 μM C149S-PhaEC in Buffer A was incubated with 1 (S/E = 50) or 6 mM (S/E = 300) [1-14C] HBCoA (SA=2000 cpm/nmol) at 37°C. Aliquots of 50 μL were withdrawn at different time points and quenched with 50 μL 2% HClO4. At the end of the time course, the reaction mixture was centrifuged and the supernatant was removed and placed on ice. The precipitated protein was washed with H2O (90 μL × 2), redissolved in 100 μL 10% SDS and analyzed by liquid scintillation counting. The pooled washes were combined with the supernatant and the pH was adjusted to ~ 6.0 by careful titration with 0.5 M NaOH at 4 °C. The neutralized solution was filtered using a Whatman microsyringe filter. The filtration unit was washed with H2O (80 μL × 3). The filtrate and washes were combined and analyzed by HPLC. In the case of S/E = 50, 1 mL fractions were collected. In the case of S/E = 300, fractions were manually collected at 0–20.0, 20.0–20.2, 20.2–20.4, 20.4–21.2, 21.2–22.0, and 22.0–50.0 min (flow rate, 1 mL/min). Each fraction was analyzed by scintillation counting.
In a final volume of 150 μL containing 40 μM C149S-PhaEC and 393 μM to 3.14 mM (HB)2CoA or 686 μM to 2.74 mM (HB)3CoA in Buffer A, aliquots (30 μL) were removed and quenched with 20 μL 10% TCA at various times. Each sample was centrifuged to remove the precipitated protein. Forty-five μL of the quenched reaction mixture was added to 55 μL of 0.25 mM DTNB in 0.5 M KPi (pH 7.8) and A412 was measured. Values for kacyl and Ki were determined by fitting the data to Eq. 2,
where [E]0 is total enzyme concentration, [Ea] is active enzyme concentration, [I] is concentration of (HB)nCoA (n = 2,3) and t is time.
In a final volume of 250 μL, 20 μM C149S-PhaEC was incubated with 1 mM [1-14C] HBCoA (S/E=50, SA=2000 cpm/nmol) in Buffer A at 37°C for 60 min. The reaction mixture was then divided into 200 μL (I) and 50 μL (II) aliquots. Aliquot I was immediately transferred to a YM-30 microcentrifuge tube and centrifuged at ×14,000 g at 4°C for 10 min. The protein was washed with Buffer A (100 μL × 2). The protein was then transferred to an Eppendorf tube using 100 μL Buffer A and analyzed by A280nm and liquid scintillation counting. The filtrate was analyzed by HPLC and 1 mL fractions were collected and analyzed by scintillation counting. Fractions eluting from 29 to 31 min and 32 to 43 min were combined and concentrated to ~300 μL using a Speedvac. The pH of the concentrated sample was adjusted to 2 using 2M HCl before it was extracted three times with Et2O (200 μL × 3). The organic layers were combined and evaporated to dryness and the residue was redissolved in 20 μL MeOH. The aqueous and Et2O extracts were analyzed by liquid scintillation counting before the material from the Et2O extract was analyzed by MALDI-TOF using α-cyano-4-hydroxycinnamic acid as the matrix.
Aliquot II was adjusted to pH 2 by addition of 0.5 M HCl before being transferred to a YM-30 microcentrifuge tube and centrifuged at ×14,000 g at 4°C for 10 min. The protein was washed with Buffer A (100 μL × 2). The protein was then transferred to an Eppendorf tube using 30 μL Buffer A. Fifteen microliter of the protein was analyzed by A280nm and liquid scintillation counting. The rest of the protein (15 μL) was immediately analyzed by ESI-MS performed on the QSTAR Elite quadrupole-TOF mass spectrometer. The sample was diluted to a concentration of 2–3 pmol/μL in water containing 0.1% formic acid and loaded onto the HPLC autosampler. A protein microtrap (Michrom BioResources) was used to bind the protein sample, which was subsequently desalted with water containing 0.1% formic acid and then eluted with water acetonitrile mixture (v/v, 1:1) containing 0.1% formic acid for MS analysis.
Our previous studies have shown that with wt synthases the rate of elongation is faster than the rate of initiation.(20, 22) Therefore, our first strategy to distinguish between the two mechanisms depicted in Scheme 1 was to acylate wt-PhaEC with sTCoA. By acylation with sTCoA, the wt-PhaEC is uniformly loaded (0.5 eq./PhaEC) and thus the elongation rate can be examined. Wild type-PhaEC was primed with sTCoA at an S/E ratio of 13. Since hydrolysis of the sT-wt-PhaEC occurs on the minute time scale (0.07 min−1), this mixture was loaded as rapidly as possible onto one syringe of a RCQ apparatus without removing excess sTCoA. The primed synthase was then rapidly mixed with 1 mM [1-14C]-HBCoA (S/E = 1 or 50) and then quenched with perchloric acid from 2.5 to 820 ms. The precipitated protein was removed and analyzed by scintillation counting and the supernatant was analyzed by HPLC and scintillation counting. The results of the experiment from analysis of the protein fractions are shown in Figure 1. The PhaEC is acylated with a rate constant of 2.4 s− 1 at S/E = 1 and of 52.2 s−1 at S/E = 50. The rate constant of 52.2 s−1 is very similar to the turnover number of 65 s−1. Analysis of fractions from the supernatant revealed decreasing amounts of (R)-[1-14C]-HBCoA and no evidence for any sT-(HB)nCoA products (lower detection limit is ~3 pmol). At the end of the reaction (S)-[1-14C]-HBCoA remained, which is consistent with previously established studies that (S)-HBCoA is neither a substrate, nor an inhibitor of wt-PhaEC. All of the radioactivity was associated with PhaEC.
Inability to detect small molecule intermediates in the supernatant prevents us from differentiating between the two mechanisms shown in Scheme 1. If the model in Scheme 1A is valid, then failure to detect these intermediates could simply indicate that the rate of re-acylation of wt-PhaEC by sT-(HB)nCoA is much faster than its rate of dissociation. Therefore, it is apparent that a mutant enzyme with decreased rate of re-acylation has to be employed in order to detect the noncovalently bound intermediates should they be generated.
Our previous studies on a variety of mutants of PhaEC showed that HBCoA was a substrate for C149S-PhaEC(16) with a turnover number of 1/2200 that of the wt-PhaEC. Furthermore, studies of alcoholysis versus thiol ester exchange of thiol esters demonstrate that the rate constant for the former at pH 8 is 50- to 100-fold slower than the latter.(28) Thus, we hypothesized that the rate of re-acylation of (HB)nCoA analogs with the serine mutant would be greatly reduced compared with wt enzyme and could potentially facilitate the detection of proposed intermediates.
As an indicator of differences in acylation rates, we investigated the rate of acylating wt- and C149S-PhaEC by [3H]-sTCoA at an S/E =13 at 37 °C using RCQ and hand quench methods. Analysis of the radioactivity associated with the precipitated protein ([3H]-sT-wt-PhaEC) revealed a rate constant for acylation (kacyl) of 5.12 ± 2.01 s−1. This rate constant is lower than the turnover number of the wt-PhaEC (65 s−1) and likely reflects the high Km for sTCoA and the fact that under the experimental conditions examined, that the PhaEC was not saturated with [3H]-sTCoA due to its limited availability.
A similar experiment was carried out with C149S-PhaEC under identical conditions (S/E = 13). The analysis of the precipitated protein gave kacyl of 2.1 ± 0.3 min−1 (0.03 s−1). The kacyl of the serine mutant is thus ~170 fold slower than the wt. Therefore, as predicted from model reactions, we might be able to use the serine mutant to trap intermediates due to the decreased rate of re-acylation.
Our previous studies on the wt-PhaEC showed biphasic kinetics of CoA release from HBCoA. Analysis of the initial rapid phase gave a Km of 0.130 mM for HBCoA and a kcat of 3900 min−1.(27) With C149S-PhaEC, HBCoA showed a monophasic release of CoA. The data fitted to Eq. 1 gave a Km for HBCoA of 1.39 mM and a kcat of 1.79 min−1 (Figure S1, Supporting Information (SI)). From these kinetic parameters, the catalytic efficiency of HBCoA for wt-enzyme (kcat/Km = 3.00 × 104 mM−1·min−1) is 20,000-fold higher than C149S mutant (kcat/Km = 1.29 mM−1·min−1) and kcat is ~2,200 times faster than with the serine mutant.
We have previously monitored chain elongation intermediates generated from reaction of [1-14C]-HBCoA with wt-PhaEC at many S/E ratios using SDS-PAGE gel analysis and autoradiography.(20) We have repeated these studies for comparison with the C149S mutant. The results with wt-PhaEC at S/E ratios of 5 and 50 are shown in Figure 2B, lanes 6 and 7. They reveal four regions of radioactivity. The label in region I is proposed to be PhaEC acylated with (HB)n (n = 3 to 10). The label in region II is proposed to be associated with (HB)n and n = 40 to 100, determined by a comparison of the migratory properties with D302A-PhaEC acylated with HBCoA at a S/E ratio of 100.(19) The label in region III is proposed to be predominately tetrameric (320 KDa) with a small amount of dimeric (160 KDa) PhaEC acylated with (HB)n. Region IV is PHB of large molecular weight attached to PhaC that does not enter the gel. (20) It has to be noted that the gel shown in Figure 2 was run in the absence of mercaptoethanol and without boiling samples before the protein was loaded onto the gel.
In order to determine if the serine mutant behaves similarly to the wt-PhaEC, the same experiment was carried out with [1-14C]-HBCoA at S/E ratio of 5, 50, 100 and 300. It has to be noted that we used high enzyme concentration and specific activity of [1-14C]-HBCoA in the mutant experiment because wt-PhaEC is 2200-fold more active than C149S-PhaEC. The results of SDS-PAGE analysis with Coomassie staining and the autoradiography are shown in Figure 2A and B, respectively. As with wt-PhaEC, most of the mutant remained unmodified based on Coomassie staining (lanes 2–5 in Figure 2A). Autoradiography revealed multiple forms of protein with (HB)n bound, similar to the observations with the wt enzyme. The tight radiolabel band in region I migrates between PhaC and PhaE in both wt and mutant synthases. The radiolabel associated with the mutant in region I is broader than that with wt and is likely the result of a 3 × increased protein loading and higher SA of the [1-14C]-HBCoA with the mutant used in the experiment. Species II is assigned to wt-PhaC-(HB)n where n = 40 to 100. However, with the mutant enzyme, species II has a distinct broad and dark feature that migrates slower than PhaE with a MW in the range of 44 to 46 KDa. This feature is proposed to be a C149S-PhaC-(HB)n where n = 73, 75 based on the ESI mass spectrometric analysis presented subsequently. Comparison of the autoradiogram of region II of the wt-PhaEC and the serine mutant, suggests that the mutant forms shorter PHB chains and appears less polydisperse than the wt enzyme. Similar to the wt-PhaEC, species III in C149S-PhaEC migrates as a tetramer (320 KDa) with longer (HB)n chains bound. We propose that the subunits remain associated during PAGE as the samples were not boiled prior to loading. The faster migrating band in region III, which has been attributed to dimeric PhaEC loaded with PHB, is not detected likely due to a sensitivity issue. Species IV does not enter the gel and corresponds to a small amount of protein associated with high MW PHB polymer. Thus even with the serine mutant, the loading of the substrate is nonuniform and the behavior appears similar to wt enzyme. Therefore, studies with this mutant are likely to be representative of the mechanism with wt-PhaEC.
To look for (HB)nCoA intermediates, C149S-PhaEC (20 μM) and [1-14C]HBCoA (6 mM) were incubated at 37 °C and the reaction was monitored by HPLC by A260nm and scintillation counting (Figure 3). A decrease in amount of (R)-HBCoA (retention time (Rt), 14.6 min) and an increase of CoA (Rt, 11.4 min) were observed (data not shown). In addition, two new species with Rt of 20.8 and 25.1 min were observed and analysis by scintillation counting revealed both species were radiolabeled (Figure 3).
The new species were isolated, concentrated and characterized by MALDI-TOF mass spectrometry. The results are shown in Figure 4. The calculated mass of (HB)2CoA in the negative mode is 938.18 Da. The observed masses of the species with a retention time of 20.8 min are 938.02 and 961.97 Da. These masses are consistent with (HB)2CoA (M–H)− and (M–H+Na)−, respectively. Since MALDI-TOF mass spectroscopy uses soft ionization, decomposition of (HB)2CoA is not expected. However, a small peak with a mass of 833.98 Da was observed, which corresponds to crotonyl-CoA with a calculated mass of 834.13 Da for (M–H)−. This species could be produced by elimination of HB from (HB)2CoA (Scheme 2). Moreover, as shown in Figure 4, there is a peak with A260nm with a Rt of 20.3 min just prior to elution of (HB)2CoA, which might be associated with crotonyl-CoA.
The calculated mass of (HB)3CoA is 1024.22 Da. The compound with a Rt of 25.1 min has observed masses of 1024.12 and 1047.89 Da, corresponding to the (M–H)− and (M–H+Na)−, respectively of (HB)3CoA. Further confirmation of the identity of the species with Rt of 20.8 and 25.1 comes from their respective co-elution with chemically synthesized (HB)2CoA and (HB)3CoA as described in Scheme S1 in Supporting Information.
We have previously reported that (HB)2CoA and (HB)3CoA function in vitro as primers of class I PhaC from R. eutropha.(21) However, their detailed syntheses have not been reported. To confirm the identity of the new CoA analogs detected during the turnover of C149S PhaEC and to help establish the kinetic competence of their formation, (HB)2CoA and (HB)3CoA were synthesized. The detailed protocol is described in Scheme S1 (SI) as is their characterization.
Both (HB)2CoA and (HB)3CoA are mechanism based substrates/inhibitors that can acylate C149S-PhaEC (Scheme 3). To determine the dissociation constant (Ki) for (HB)nCoA binding to C149S-PhaEC and kacyl, (HB)nCoA (n = 2, 3) was incubated at various concentrations with C149S-PhaEC and aliquots removed as a function of time and quenched with 10% TCA to measure CoA release. The results of the analysis using Eq 2 are shown in Figure 5A and B. The Ki for (HB)2CoA is 2.02 mM with a kacyl of 1.1 min−1. Similar experiments with (HB)3CoA gave a Ki of 1.54 mM and a kacyl of 2.35 min−1. The rate constant for PHB production by C149S-PhaEC is 1.79 min−1 and thus both of these oligomers are capable of being involved in polymer formation in a kinetically competent fashion.
Kinetic competence implies that not only is the rate of utilization of (HB)nCoA greater than or equal to the rate of polymer formation, but their rate of formation must also be faster than the overall rate of the reaction. To examine this rate, C149S-PhaEC (20 μM) was incubated with [1-14C]HBCoA (6 mM) and the reaction progress was monitored by HPLC for (HB)2CoA and by scintillation counting for (HB)n-C149S-PhaCE subsequent to protein isolation by filtration (Figure 3 and Figure 6). Intermediate (HB)2CoA was formed at a rate constant of 0.30 min−1 and (HB)n-C149S-PhaC at 1.61 min−1. The reaction rate is thus 2.21 min−1 and is similar to that detected for CoA release from HBCoA (1.79 min−1). The relative rates of re-acylation vs (HB)nCoA dissociation appear to be about 2.7:1. Thus the kinetics reveal that (HB)2CoA can function as an intermediate in polymer formation. Given the relative rates of acylation by C149 vs S149, it seems reasonable that our inability to detect this intermediate with wt synthase is due to its enhanced re-acylation as we proposed.
A similar analysis for the (HB)3CoA could not be carried out due to the low levels of material. The apparent slow rate of formation of (HB)3CoA and the inability to detect (HB)nCoA (n≥4) may be related to the slow rate constants for dissociation of longer HB chains and hence the ratio of re-acylation to dissociation increases.
In an experiment designed to monitor formation and disappearance of (HB)2CoA and (HB)3CoA, a reaction of C149S-PhaEC with [14C]-HBCoA at an S/E = 50 was monitored by A260nm and scintillation counting over 60 min. As depicted in Figure 7A monitoring the changes in A260nm, (HB)2CoA production increased during the first 30 min and then decreased, while (HB)3CoA only decreased. Examination of a similar reaction at 60 min, revealed that while ~50% of the radioactivity was associated with (HB)2CoA, ~ 25% eluted between 29 and 43 min with limited A260nm. Our studies on the stability of sT-wt-PhaEC indicated that the acid of the sT was formed with a rate constant of 0.07 min−1. Thus, given the long incubation times, we postulated that the radioactivity eluted between 29 and 43 min was likely associated with oligomers of hydroxybutyric acids (HB)nCO2H.
Knowing the distribution of n might be informative about the optimal length of the primer and the size of the (HB)n associated with region I of the SDS gel (Figure 2). If the products that eluted between 29 and 43 min (Figure 7B) are (HB)nCO2H, then acidification of the solution to pH 2, should allow them to be extracted into diethylether. Fractions 29–31 min and 32–43 min were pooled separately and examined by this procedure. For 29–31 min, 90% of the radioactivity was extracted and analysis of this material by MALDI-TOF in the positive mode revealed masses consistent with (HB)nCO2− as Na/K salt where n = 4 or 5 (Figure S2, SI). Extraction of fractions 32–43 min resulted in 73% of the radioactivity recovered in the ether layer and MALDI-TOF analysis revealed oligomeric acids with n = 6, 7 and 8 as Na/K salt (Figure S3, SI). As described in the methods section, 4 nmol PhaEC was incubated with 200 nmol of HBCoA and 50 nmol of radiolabeled material was eluted between 29 and 43 min by HPLC analysis. The sizes of the (HB)n acids detected and the previously measured rate of hydrolysis of the sT-PhaEC suggests that during the 60-min experiment it is reasonable that acids might be detected. Moreover, the results suggest that the acids might be indicative of the acylated PhaC in region I of the gel (Figure 2) and indicative of a primed enzyme.
In an effort to establish the range of n for (HB)n attached to the protein, the reaction between C149S-PhaEC and [1-14C]-HBCoA at 60-min time point was acidified to pH 2 to stop the reaction and minimize hydrolysis of the acylated enzyme. The protein was separated from small molecules using YM-30 microcentrifuge tube and was immediately analyzed by ESI-MS. The results are shown in Figure 8. Tentative assigments of mass to structure is summarized in Table 1. The species at 39760 and 40440 Da correspond to C149S-PhaC and PhaE (a mutant with two mutations relative to wt), respectively. The other species with masses of 46080 and 46211 have been assigned to C149S-PhaC-(HB)n where n is 73 or 75. We propose that the C149S-PhaC-(HB)73/75 might be the dark band observed in region II in SDS-PAGE analysis (Figure 2). The masses attributed to C149S-PhaC-(HB)n (n=247 and 334) might migrate in the region of species III in the corresponding autoradiography.
We have proposed two mechanisms for PHB polymerization catalyzed by PhaEC (Scheme 1A and 1B) based on available information. The mechanism in Scheme 1B is less appealing as it requires that the active site is at the interface of two monomers and that each monomer contributes a single cysteine involved in covalent catalysis. No structures are available of the class I or III synthases. However, all synthases possess a lipase box and threading models suggest that the synthase is an α,β hydrolase superfamily member with the active site nucleophile at the elbow of a strand turn helix.(16, 29) Lipases are in general monomers and their active sites are deeply buried, requiring a moderately long tunnel for access of their triacylglycerol substrates. Given this information, it is thus difficult to see how two deeply buried active sites, one from each PhaC monomer, could generate the required active site at their interface. Moreover, earlier attempts failed to trap the second covalent intermediate using potential chain terminators such as 3-methoxy- and 3-fluoro-butyryl-CoA.(15) Addition of the terminator to the pre-acylated class I enzyme (sT-PhaC) did not release any detectable CoA and thus failed to provide support for a second covalent intermediate.
The mechanism shown in Scheme 1A is based on some type III polyketide synthases(25) and requires only one active site nucleophile. According to this model, it predicts two equivalents of sT per dimer of PhaEC, in contrast with the experimental observations. However, similar phenomena of half site reactivity have been observed with the homodimers of chalcone synthase (30) and prenyltransferases (31) where multiple reaction steps occur within a single reactive site.
Furthermore, the mechanism in Scheme 1A implicating non-covalent intermediates is consistent with our recent studies using an HBCoA substrate analog in which the CoA was replaced by N-acetylcysteamine (NAC, HBNAC).(32) This substrate has a turnover number of 1/100 that of HBCoA. The PHB polymers generated from HBNAC are considerably smaller than those generated from HBCoA (75 KDa vs 1.5 MDa) with much higher polydispersity. Furthermore, the C-terminus of the polymer ends with NAC rather than an acid. One interpretation of these observations is that a non-covalent intermediate dissociated from the active site presumably due to the high Kd of PHB-NAC.
The mechanism proposed in Scheme 1A, which predicts non-covalently bound (HB)nCoA intermediates, would likely be difficult to demonstrate with wt-PhaEC, as the enzyme is known to produce polymers of 1.5 million Da with low polydispersity. If intermediates dissociated during the polymerization process, they then would compete with monomer for re-binding and the polydispersity of the PHB would be expected to be high as observed with HBNAC. Thus our inability to detect sT-(HB)nCoA analogs by the RCQ approach is not surprising when sT-PhaEC is reacted with HBCoA. C149S-PhaEC proved to be an ideal candidate for examining the mechanism of polymerization based on the differences in reactivity of oxygen vs thiol nucleophiles, the slow rate of PHB formation with this mutant, and its similaries to wt-PhaEC in the elongation process (Figure 2). When C149S-PhaEC was reacted with [1-14C]-HBCoA, non-covalent intermediates (HB)2CoA and (HB)3CoA along with covalently modified synthase were observed for the first time. These species were demonstrated to be chemically and kinetically competent, providing strong support for the proposed chain elongation mechanism shown in Scheme 1A.
Finally isolation of (HB)nCO2H where n is 4 to 8 after a long incubation of C149S-PhaEC with HBCoA, might be providing us with insight about the priming process catalyzed by PhaEC. These acids are proposed to originate from the hydrolysis of the modified C149S-PhaEC corresponding to the species I [C149S-PhaC-(HB)n (n=3–10)] in the autoradiogram (Figure 2B). Thus our working model is that the polymer exit tunnel(20) may accommodate up to 8 HB units. Unfortunately efforts to detect the acylated C149S-PhaC-(HB)n (n=3–10) by ESI-MS were unsuccessful. However, detection of species tentatively assigned to C149S-PhaC-(HB)73/75 suggests that as the HB chain gets longer, the rate of hydrolysis of the covalent linkage to PhaC is dramatically reduced. This proposal is consistent with the low polydispersity of the PHB.
Using the active site C149S-PhaEC has allowed detection for the first time of covalently and non-covalently bound HB intermediates. The detailed mechanism of chain extension and termination requires a structure to think about the complexity of this reaction. Efforts with mechanism-based inhibitors to obtain such a structure are in progress.
The authors thank Rachael Buckley for purifying C149S-PhaEC.
§This work was supported by NIH Grant GM49171 to J.S. P.L. was supported by NIH Kirschstein–NRSA postdoctoral fellowship F32GM082067.
Supporting Information Available
Synthesis of (HB)2CoA and (HB)3CoA, kinetic analysis of HBCoA with C149S-PhaEC, and MALDI-TOF mass spectra of (HB)nCO2H. This material is available free of charge via the Internet at http://pubs.acs.org.