Iso-coenzyme A*

Iso-coenzyme A is an isomer of coenzyme A in which the monophosphate is attached to the 2′-carbon of the ribose ring. Although iso-CoA was first reported in 1959 (Moffatt, J. G., and Khorana, H. G. (1959) J. Am. Chem. Soc. 81, 1265–1265) to be a by-product of the chemical synthesis of CoA, relatively little attention has been focused on iso-CoA or on acyl-iso-CoA compounds in the literature. We now report structural characterizations of iso-CoA, acetyl-iso-CoA, acetoacetyl-iso-CoA, and β-hydroxybutyryl-iso-CoA using mass spectrometry (MS), tandem MS, and homonuclear and heteronuclear NMR analyses. Although the 2′-phosphate isomer of malonyl-CoA was recently identified in commercial samples, previous characterizations of iso-CoA itself have been based on chromatographic analyses, which ultimately rest on comparisons with the degradation products of CoA and NADPH or have been based on assumptions regarding enzyme specificity. We describe a high performance liquid chromatography methodology for separating the isomers of several CoA-containing compounds. We also report here the first examples of iso-CoA-containing compounds acting as substrates in enzymatic acyl transfer reactions. Finally, we describe a simple synthesis of iso-CoA from CoA, which utilizes β-cyclodextrin to produce iso-CoA with high regioselectivity, and we demonstrate a plausible mechanism that accounts for the existence of iso-CoA isomers in commercial preparations of CoA-containing compounds. We anticipate that these results will provide methodology and impetus for investigating iso-CoA compounds as potential pseudo-substrates or inhibitors of the >350 known CoA-utilizing enzymes.

Iso-coenzyme A is an isomer of coenzyme A in which the monophosphate is attached to the 2Ј-carbon of the ribose ring. Iso-CoA was first reported in 1959 by Moffatt and Khorana (1,2) to be a by-product of the chemical synthesis of CoA. The final step of this synthesis entailed acid-catalyzed hydrolysis of cyclic coenzyme A (Scheme 1) to produce a 50:50 mixture of two products that were separated using epicholorohydrin triethanolamine (ECTEOLA) cellulose ion exchange chromatography (1,2). Direct structural analysis was not readily available in 1959; consequently, the authors deduced the structure of iso-CoA by establishing that, in contrast to CoA, iso-CoA was not a substrate for the enzyme phosphotransacetylase (1,2). Moffatt and Khorana also used paper chromatography of phosphodiesterase hydrolysates to show that enzymatic hydrolysis of iso-CoA produces adenosine-2Ј,5Ј-diphosphate, exclusively (1,2). Since the original work of Moffatt and Khorana was done, four additional chemical syntheses of CoA have been reported (3)(4)(5)(6)(7)(8), all of which produce cyclic CoA as the penultimate product; cyclic CoA is then hydrolyzed either chemically to produce a mixture of CoA and iso-CoA (1, 2, 4 -6, 8) or enzymatically with ribonuclease T2 to produce CoA regioselectively (3)(4)(5)(6)(7).
Recently, Minkler et al. (9) observed the 2Ј-phosphate isomer of malonyl-CoA in commercial samples. These authors reported that HPLC 1 -purified samples of each isomer exhibited identical UV, HPLC-MS, and HPLC-MS/MS properties, and the identity of the 2Ј-phosphate isomer was convincingly demonstrated by comparing the one-dimensional NMR spectra to that of 2Ј,5Ј-ADP. In related work, Retey and co-workers have synthesized several dethia-CoA analogs (10 -15), in which the sulfur of CoA is replaced by a methylene, as mixtures of the 2Ј-and 3Ј-phosphate isomers; two of these analogs were purified to homogeneity and characterized by one-dimensional 1 H-NMR, 31 P-NMR, and, in one case, fast atom bombardment MS (13,14). Dethia-CoA analogs were also synthesized and characterized by Stewart and co-workers (16 -18); unfortunately, they were unable to separate the 2Ј-and 3Ј-phosphate isomers, and the NMR data could only be obtained on isomeric mixtures (17). However, to the best of our knowledge no full structural characterization has ever been reported in the literature for iso-CoA, acetyl-iso-CoA, acetoacetyliso-CoA, or ␤-hydroxybutyryl-iso-CoA.
The issue of whether or not enzymes that act upon CoAcontaining substrates discriminate between the 2Ј-and 3Јphosphate isomers is of obvious relevance to the design of new classes of inhibitors and pseudosubstrates. As mentioned above, in the early work of Moffatt and Khorana the enzyme phosphotransacetylase was reported to be unreactive toward iso-CoA (1,2). Similarly, Retey and co-workers have reported that FAD-dependent isobutanoyl-CoA dehydrogenase is reactive with isobutanoyl-dethia-CoA but not with the iso-CoA analog (13) and that methylmalonyl-CoA pyruvate carboxylase also distinguishes between the isomers, reacting exclusively with propionyl-dethia-CoA (10). In contrast, Thorpe et al. (18) reported that substrate mixtures containing both dethia-iso-CoA and dethia-CoA derivatives are capable of reducing enzyme-bound FAD. Likewise, Rossier (19) reported that choline acetyltransferase is inhibited by both seleno-CoA and selenoiso-CoA, and Wagner et al. (14) have reported that N-myristoyltransferase is competitively inhibited by both carbadethia-CoA and carbadethia-iso-CoA, although in both cases the iso-CoA analogs are less potent. In the cases of citrate synthase and carnitine palmitoyltransferase (17), it is difficult to draw a clear conclusion from the data because the experiments were performed with mixtures containing both the 2Ј-and 3Ј-isomers. Clearly this issue needs to be reexamined for many CoA-utilizing enzymes using purified and characterized iso-CoA substrate analogs.
Herein we report structural characterizations of iso-CoA, acetyl-iso-CoA, acetoacetyl-iso-CoA, and ␤-hydroxybutyryliso-CoA. Historically, characterizations of the structure of iso-CoA have been based on chromatographic analyses (1, 2) that rest on comparisons with the degradation products of CoA and NADPH (20 -22) or have been based on assumptions regarding enzyme specificity (5,6,8). We describe HPLC methodology to separate the isomers of several CoA-containing compounds and the characterization of iso-CoA structures using MS, MS/MS, and NMR analyses. We also report here the first examples of iso-CoA-containing compounds acting as substrates in enzymatic acyl transfer reactions. Finally, we describe a simple synthesis of iso-CoA from CoA that utilizes ␤-cyclodextrin to produce iso-CoA with high regioselectivity, and we demonstrate a plausible mechanism that accounts for the existence of iso-CoA isomers in commercial preparations of CoA-containing compounds.
HPLC Mass Spectrometry-Commercial CoA and acetyl-CoA standards were analyzed with a Micromass Quattro LC triple quadrupole SCHEME 1. Final reaction in the total synthesis of CoA. Chemical hydrolysis of cyclic CoA (I) produces two products, CoA (II) and iso-CoA (III). Carbons are labeled according to d'Ordine et al. (29). tandem mass spectrometer equipped with an ESI source connected to an Hewlett-Packard series 1100 HPLC system. Resolution of the isomers was achieved with an Agilent Hypersil AA-ODS 2.3 ϫ 200-mm column attached to a Phenomenex SecurityGuard TM (C18, 4-mm length ϫ 3-mm inner diameter), using an isocratic mobile phase of 96% 200 mM ammonium acetate (pH 6.0) and 4% acetonitrile at a flow rate of 0.2 ml/min. ESI mass spectra were obtained in the positive ion mode with nitrogen used as the nebulizer and desolvation gas at flow rates of ϳ80 and 580 liter/h, respectively. The cone voltage was set to 50 V and the capillary voltage to 3. NMR Analyses-NMR spectra of iso-CoA were obtained in D 2 O in a 5-mm sample tube on a Bruker DRX 500 spectrometer at 500.13 mHz for 1 H and 202.46 mHz for 31 P using an inverse triple resonance probe maintained at 298 K. The COSY experiment was performed using the standard Bruker program cosyprqf with presaturation during the 2-s relaxation delay on the HOD signal, 2048 data points in the F2 dimension, and 128 increments in F1. The data matrix was processed to give a matrix of 1024 ϫ 1024 points, and a sine bell apodization function was applied before Fourier transformation.
The parameters for the two-dimensional heteronuclear multiple quantum correlation ( 1 H-31 P HMQC) experiment were the hmqcqf standard Bruker program, 1.5-s recycle delay, 1024 data points in F2, 128 increments in F1, an 8-Hz coupling constant, GARP (globally optimized alternating phase rectangular pulse) 31 P decoupling during acquisition, and shifted sine-squared apodization before Fourier transform. The chemical shift assignments for the HPLC-purified CoA and iso-CoA are labeled above their respective peaks in the proton spectra. The few unlabeled peaks were either due to the iso-CoA disulfide formed during purification and analysis in the absence of reducing agents or to an unknown impurity. Coupling constants were obtained through successive decoupling experiments, and the unresolved coupling constants were then obtained by simulating the spectra using the gNMR (23) and MestRe-C (24) software packages.
HPLC-HPLC analyses of CoA, acetyl-CoA, acetoacetyl-CoA, ␤-hydroxybutyryl-CoA, and the isomers were performed on a Waters LC Module I Plus HPLC system with auto-injection of samples using the Millennium chromatography manager software (Waters, Milford, MA). A Waters NovaPak C18 reverse phase column of 4-m particle size and 150 ϫ 3.9-mm inner diameter was used with a Phenomenex Security-Guard TM containing a C18 guard cartridge (4-mm length ϫ 3-mm inner diameter). Isocratic analyses were performed at a wavelength of 261 nm with a mobile phase containing 96% 200 mM ammonium acetate (pH 6.0) and 4% acetonitrile at a flow rate of 1.0 ml/min. Semi-Prep HPLC purification was performed on an Allsphere ODS-2 semi-prep column, 250 ϫ 10-mm inner diameter, with a Phenomenex SecurityGuard TM containing a C18 guard cartridge (4-mm length ϫ 3-mm inner diameter) using isocratic elution with a mobile phase of 97.5% 200 mM ammonium acetate (pH 6.0) and 2.5% acetonitrile at 7 ml/min.
Two peaks for each CoA-containing sample were observed with four different reversed phase HPLC columns and three different HPLC systems using both UV-visible and MS detection. HPLC-purified samples of CoA were stable in neutral solution, and iso-CoA was not generated in deionized water at pH 3 or at pH 11 (data not shown).
Enzymatic Analyses of Iso-CoA and Acyl-iso-CoA Compounds-The enzymes ␤-ketothiolase (E.C. 2.3.1.16), acetoacetyl-CoA reductase (E.C. 1.1.1.36), and poly-(␤)-hydroxybutyric acid (PHB) synthase (currently not classified), all of which utilize CoA-containing substrates, were examined for the ability to react with the respective iso-CoA-containing analogs. Enzymatic reactions were carried out in a solution of 16 M acetyl-CoA isomers, 1 mM NADPH, and 5 mM TCEP in 150 mM 4-(2-hydroxyethyl)piperazine-1-propanesulfonic acid with a final pH of 7.8 at 37°C. The first reaction was initiated by addition of ␤-ketothiolase and acetoacetyl-CoA reductase, and the second reaction was initiated with PHB synthase. Enzyme reactions were quenched with 0.27 M ice-cold perchloric acid followed by centrifugation prior to HPLC analysis as described above.
CoA Disulfide (CoASSCoA)-Ten milligrams (13 mM) of HPLC-purified CoA (II) were dissolved in 1 ml of 20 mM ammonium carbonate pH 8.5, reacted overnight at 37°C, and then lyophilized to produce CoA disulfide with 100% yield by HPLC. The reaction progress was monitored via HPLC using the HPLC procedure described above, except that the mobile phase contained 6% acetonitrile.
Cyclic CoA Disulfide-The CoA disulfide lyophile was dissolved in 1 ml of 20 mM MES (pH 6.0), 10 mg (52 mM) of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC) was added, and the pH was readjusted. Reaction progress was monitored by HPLC after rapidly reducing an aliquot of the disulfide to the free thiol with TCEP at pH 8.0 for 2 min. When the reaction was complete, the entire mixture was lyophilized. The lyophile was dissolved in 200 mM ammonium acetate (pH 6.0) buffer, and the entire solution was applied to a 500-mg Supelco C8 solid phase extraction cartridge that was pre-equilibrated with the same buffer. Residual EDC and the urea by-product were eluted from the cartridge with the equilibration buffer, and the cyclic CoA disulfide was eluted in ϳ5 ml of 90% 200 mM ammonium acetate (pH 6.0) and 10% acetonitrile. The cyclic CoA-containing fractions were pooled and lyophilized.
Regioselective Synthesis of Iso-CoA (III)-The cyclic CoA disulfide lyophile was dissolved in 160 ml of 0.05 mM bicarbonate buffer (I ϭ 0.01), pH 9.5, containing 3 M KCl and 15 mM ␤-cyclodextrin, and the pH was readjusted. The solution was stirred at 30°C with the pH maintained at 9.5 for ϳ5 days or until the reaction was deemed complete by HPLC, at which point the pH was adjusted to 3.0 to quench the reaction. This provided a final solution of iso-CoA with 83% regioselectivity by HPLC. Acetone precipitation and filtration served to remove some of the KCl and ␤-cyclodextrin from the iso-CoA-containing solution and permitted rotary evaporation followed by vacuum evaporation to dryness. The final solid was then dissolved in 200 mM ammonium acetate (pH 6.0), the pH was readjusted, and the solution was desalted by passing through a 10-g Supelco C8 solid phase extraction cartridge. The desalted isomeric mixture of 83% iso-CoA and 17% CoA was eluted as described previously, and the desalted mixture was lyophilized. Final purification of iso-CoA was performed by dissolving the desalted lyophile in a minimal quantity of 5 mM TCEP in 4-(2-hydroxyethyl)piperazine-1-propanesulfonic acid (pH 8.0) to reduce the CoA disulfide to the free thiol, passing the mixture through a 10-g solid phase extraction cartridge and eluting with 200 mM ammonium acetate (pH 6.0), followed by a final purification using semi-prep HPLC, as described previously.
Acid Catalyzed Synthesis of Iso-CoA (III)-HPLC-purified CoA (II) (1.6 mg) was dissolved in 0.5 ml of 0.5 M HCl, and the reaction was monitored by HPLC at room temperature until equilibrium was reached, generating a mixture of 60% CoA and 40% iso-CoA. The identity of iso-CoA produced in this manner was confirmed by NMR analysis after HPLC purification and lyophilization as described previously.  Fig. 1 shows an HPLC-MS chromatogram of a commercial CoA sample that exhibits two peaks with identical (M ϩ H) ϩ ions at m/z 768, suggesting that commercial CoA samples contain two compounds with identical masses. Therefore, HPLC-MS/MS experiments using MRM were carried out to garner further structural information. It is evident from the right portion of panel A in Fig. 1 that the CoA samples exhibit two peaks for the MRM transitions m/z 7683159, m/z 7683261, and m/z 7683428. Thus, these HPLC-MS and HPLC-MS/MS results suggest the presence of isomeric compounds with identical fragmentation patterns. To confirm these analyses, each peak in the commercial CoA sample was purified by preparative HPLC, and high resolution exact mass and MS/MS analyses were performed individually on each isomer. As shown in the Structural Characterization of Iso-CoA by NMR-One-dimensional 1 H-NMR spectra for the two HPLC-purified isomers of CoA are shown in panels A and B of Fig. 2. The spectrum in Fig. 2A for the earlier eluting isomer is very similar to several previously published CoA spectra (25)(26)(27)(28)(29)(30), and the chemical shift assignments were made according to d'Ordine et al. (29). In contrast, the spectrum shown in Fig. 2B for the later eluting isomer is very different in the 4 -5 ppm region. Therefore, two-dimensional 1 H-1 H COSY experiments (Fig. 3) were obtained for this isomer in order to assign the chemical shifts. The chemical shift assignments and coupling constants are listed in Table I. It is evident that for the later eluting isomer the 2Ј-proton is shifted downfield by ϳ0.20 ppm, whereas the 3Ј-and 4Ј-protons are shifted up field by 0.15 and 0.20 ppm, respectively, and the two 5Ј-protons have significantly different coupling patterns. Because it seemed likely that the position of the phosphate on the ribose ring gives rise to these differences, 1 H-31 P HMQC experiments were carried out to explicitly assign the position of each phosphate. As shown in Fig. 4, the 1 H-31 P HMQC spectra for the later eluting compound unequivocally establishes that the monophosphate is attached at the 2Ј-carbon, thus designating this compound as iso-CoA. The HPLC experiments presented in Fig. 5 established that commercial samples of CoA, acetyl-CoA, acetoacetyl-CoA, and ␤-hydroxybutyryl-CoA, designated as 93, 93, 90, and 99% pure, respectively, contain ϳ10, 15, 18, and 3% of the 2Ј-monophosphate isomers, respectively.

Mass Spectrometric Structural Characterization of CoA and
Iso-CoA-containing Compounds Are Substrates for ␤-Ketothiolase, Acetoacetyl-CoA Reductase, and PHB Synthase-The reaction time courses shown in Fig. 6 for CoA, acetyl-CoA, and ␤-hydroxybutyryl-CoA and their respective iso-CoA isomers demonstrate that the three enzymes, ␤-ketothiolase, acetoacetyl-CoA reductase, and PHB synthase, can catalyze reactions with either isomer. These data represent the first examples of iso-CoA-containing compounds acting as substrates in enzymatic acyl transfer reactions. In the coupled two-enzyme system presented in Fig. 6A, ␤-ketothiolase condenses two molecules of acetyl-CoA to acetoacetyl-CoA, which is immediately reduced to ␤-hydroxybutyryl-CoA by the second enzyme, acetoacetyl-CoA reductase. In Fig. 6B a third enzyme, PHB synthase, has been added that converts the ␤-hydroxybutyryl-CoA to PHB. It is evident from the time courses in both panels that in all cases both the CoA and iso-CoA isomers are processed with equal facility. In other experiments we found that acetoacetyl-CoA and its iso-CoA isomer react with equal facility in the back reaction to produce the acetyl-CoA product (data not shown).
Regioselective Synthesis of Iso-CoA-Because iso-CoA and acyl-iso-CoA compounds may be useful as potential inhibitors or pseudosubstrates, a synthesis capable of producing high yields of the 2Ј-phosphate isomer is desirable; to this end we report a simple synthesis of iso-CoA from CoA. As shown in Scheme 2, CoA was initially converted to the dimeric disulfide, and the water-soluble coupling agent EDC was then used to form the 2Ј,3Ј-cyclic CoA intermediate in a reaction analogous to the synthesis of 2Ј,3Ј-cyclic-NADP ϩ (31). We note that the first steps of the synthesis are carried out using the CoA dimer to avoid possible quenching of the EDC; the dimer is reduced  back to CoA in step 3. Finally, base-catalyzed hydrolysis is carried out in the presence of ␤-cyclodextrin in high salt at pH 9.5 to produce iso-CoA with a 83% yield.
Acid-catalyzed Phosphate Migration-We investigated the acid-catalyzed migration of the CoA monophosphate group in an attempt to furnish a plausible mechanism for the existence of the 2Ј-phosphate isomers in commercial preparations of CoAcontaining compounds. In these experiments HPLC-purified CoA was incubated in the presence of 0.5 M HCl, and the relative amounts of CoA and iso-CoA were determined by HPLC as a function of time. As shown in Fig. 7, a final equilibrium mixture of 60% CoA and 40% iso-CoA was attained within 60 h. We note that ϳ10% iso-CoA is present after only 4 h; thus, the iso-CoA content of commercial preparations may indeed be produced in this fashion during the processing of the product. DISCUSSION The vital role of coenzyme A in metabolism is underscored by the sheer quantity of classified enzymes (33) 2 that react with CoA-containing molecules; we count Ͼ9% of the total known enzymes to be of this type. The 2Ј-phosphate isomer of CoA, first named iso-CoA in the 1959 report of Moffatt and Khorana (1, 2), has traditionally been considered an undesirable synthetic by-product or has been simply ignored. Indeed, a review of the literature reveals that only eight of the more than 350 CoA-utilizing enzymes 2 have been examined for the ability to discriminate between the 2Ј-and 3Ј-phosphate isomers (1,2,10,13,14,(17)(18)(19)34), and in two of these cases the enzymes have been shown to accept purified 2Ј-phosphate isomers as substrates or inhibitors (14,19). Moreover, the majority of the compounds examined in these eight cases are synthetic analogs of CoA, such as the dethia or seleno derivatives, and not iso-CoA itself.
In the results presented here, CoA and iso-CoA were purified by HPLC, and high resolution exact mass MS and MS/MS analyses unequivocally established that the compounds are constitutional isomers with extremely similar fragmentation patterns. Direct structural identification of the two HPLCpurified isomers was then performed using 1 H-NMR, 1 H-1 H COSY, and 1 H-31 P HMQC, and the results unequivocally established that the monophosphate of the iso-CoA isomer is attached to the 2Ј-carbon of the ribose ring. Structural identification of acetyl-iso-CoA, acetoacetyl-iso-CoA, and ␤-hydroxybutyryl-iso-CoA was also carried out using HPLC-MS and HPLC-MS/MS analyses.
We report here the first example of iso-CoA-containing compounds acting as substrates in acyl transfer reactions. Three enzymes, ␤-ketothiolase, acetoacetyl-CoA reductase, and PHB synthase, successfully react with the 2Ј-isomers of their natural substrates, and the reaction time courses indicate that all three enzymes react with either isomer with equal facility. These results suggest that other enzymes should be examined for their reactivities toward the 2Ј-phosphate isomers of their CoA-containing substrates. We note that ϳ40 enzymes have been evaluated for their abilities to interact with dephospho-CoA analogs. In the case of phosphotransacetylase, Iyer and Ferry (35) have identified a salt bridge at the binding site of this enzyme that they have suggested imparts a 350-fold preference for CoA over dephospho-CoA; a salt bridge between an arginine and the 3Ј-phosphate moiety of CoA has also been reported for choline acetyltransferase (36,37). This type of salt bridge formation may be a general feature of enzymes that react with CoA-containing substrates. If so, it remains to be determined whether this binding interaction would be distorted by the presence of a phosphate at the 2Ј-position rather than the 3Ј-position of the ribose ring.
The traditional method for resolving isomers of CoA-containing compounds, ion exchange chromatography (1,2,14,34), is often lengthy, does not achieve base-line resolution, and thus may have hampered investigations into isomers of CoA (19). We report here an efficient HPLC methodology that achieves base line resolution of CoA and acyl-CoA isomers. This technique, which improves upon earlier work by Norwood et al. (38), is the first HPLC method to separate a series of iso-CoA compounds, and we were able to separate four iso-CoA isomers within 30 min with base-line resolution using convenient isocratic elution. We note that inspection of published HPLC elution profiles of CoA-containing compounds reveal several chromatograms that appear to contain iso-CoA compounds that The 1 H-31 P HMQC for the HPLCpurified isomer unequivocally demonstrates that the monophosphate, located at Ϫ0.9 ppm, clearly corresponds to the 2Ј-proton of the ribose ring; thus, this isomer is designated as iso-CoA. Additionally, because the two phosphates of the pyrophosphate linkage couple with the protons on the 5Ј and 1Љ carbons, no other phosphate rearrangements are evident.
FIG. 6. The first example of enzymatic acyl transfer reactions occurring with 2-isomers of CoA-containing compounds. Enzymatic reaction time courses for the enzymes ␤-ketothiolase, acetoacetyl-CoA-reductase, and PHB synthase indicate that these enzymes catalyze identical reactions exhibiting equivalent kinetic rates with both 2Ј-and 3Ј-isomers of acetyl-CoA, CoA, and ␤-hydroxybutyryl-CoA. Panel A, ␤-ketothiolase and acetoacetyl-CoA reductase are added to a solution containing acetyl-CoA and its isomer, NADPH, and TCEP, buffered at pH 7.8. This coupled-enzyme reaction initiates the consumption of two molecules of acetyl-CoA or acetyl-iso-CoA, thus generating one molecule of CoA or iso-CoA and one molecule of ␤-hydroxybutyryl-CoA or the isomer. Panel B, addition of PHB synthase to the previous coupledenzyme system (panel A) results in the conversion of ␤-hydroxybutyryl-CoA and its isomer to PHB; this irreversible polymerization stimulates the consumption of further acetyl-CoA isomers with CoA or iso-CoA liberation during both processes. Acetoacetyl-CoA, the product of the ␤-ketothiolase-mediated condensation of 2 acetyl-CoA molecules, is below the limit of detection by HPLC. Top sections (panels A and B): f, acetyl-CoA (AcCoA); ‚, acetyliso-CoA (Iso-AcCoA). Middle sections: f, CoA; E Iso-CoA. Bottom sections: ࡗ, ␤-hydroxybutyryl-CoA (HBCoA); ▫, ␤-hydroxybutyryl-iso-CoA (Iso-HBCoA). SCHEME 2. Regioselective synthesis of Iso-CoA from CoA. This is the first regioselective chemical synthesis of iso-CoA reported and utilizes ␤-cyclodextrin in high salt to confer iso-CoA regioselectively in high yields.
were not identified as such by the authors (39 -43). Furthermore, iso-CoA isoforms may have been overlooked by previous authors because the isomers will co-elute under stronger HPLC elution conditions.
We also report here the first regioselective chemical synthesis of iso-CoA from CoA. Previously reported methods that produce iso-CoA compounds as a by-product of cyclic CoA hydrolysis yield mixtures containing both the 2Ј-and 3Ј-phosphate isomers in ratios of ϳ40 and 60%, respectively (1,2,5,6,8). In our simple synthesis, the cyclic CoA moiety is efficiently produced from CoA, and then base-catalyzed hydrolysis in the presence of ␤-cyclodextrin in high salt produces iso-CoA regioselectively in high yields. Our results are in agreement with those of Komiyama and coworkers (44,45), who reported previously that ␤-cyclodextrinbased hydrolyses of adenosine and guanosine 2Ј,3Ј-cyclic phosphate compounds give reversed regioselectivity (i.e. 3Ј-cleavage) as compared with enzymatic hydrolysis by ribonuclease. These authors theorize that the adenine moiety forms an inclusion complex with ␤-cyclodextrin, which generates steric constraints that direct hydrolysis toward the 3Ј-bond, producing the 2Ј-phosphate species regioselectively.
We propose acid catalysis as a plausible mechanism to account for the existence of the 2Ј-phosphate isomers in commercial preparations of CoA-containing compounds. In 1954, Kaplan and co-workers published preliminary reports about the acid-catalyzed synthesis of iso-CoA from CoA (21,46); however, these authors did not prove iso-CoA formation but only demonstrated that exposing CoA to strong acid decreased specific 3Ј-nucleotidase activity. Nevertheless, the acid-catalyzed migration of the 2Ј-and 3Ј-phosphate groups of NADPH (47) and nucleotides (48 -51) is well established in the literature. In our hands, treatment of CoA with 0.5 M HCl produced an equilibrium mixture of 40% iso-CoA and 60% CoA within 60 h, and 10% of the isomer could be generated within 4 h. These results indicate that acid catalysis is a viable mechanism for rapidly producing iso-CoA from CoA, especially considering that strong acids are often used in several published CoA isolation procedures (52)(53)(54)(55)(56).
The biological significance of iso-CoA compounds is currently unknown. The original report of Moffatt and Khorana (1,2) stating that phosphotransacetylase does not react with iso-CoA probably has led to the generally held contention that the 2Ј-isomers are unnatural compounds. In contrast, Kurooka et al. (57) suggested that iso-CoA might be formed from crude cell extracts of Proteus mirabilis in the presence of 3Ј-dephospho-CoA. Similarly, Michaelson (3) reported that an unnamed, par-tially purified enzyme from calf brain selectively generated iso-CoA from cyclic CoA in vitro. This enzyme was probably 2Ј,3Ј-cyclic 3Ј-phosphodiesterase, which selectively produces 2Јphosphate nucleosides from their 2Ј,3Ј-cyclic species, such as NADP ϩ generation from cyclic-NADP ϩ (31). Although this enzyme constitutes ϳ4% of all myelin protein, to date the actual in vivo substrate for this enzyme's esterase activity has not been established (32,58,59). These reports suggest that iso-CoA could possibly be generated enzymatically in vivo from either dephospho-CoA or 2Ј,3Ј-cyclic CoA; however, no definitive evidence for this notion yet exists.
Although the biological significance of iso-CoA is presently unknown, the potential for this class of isomeric compounds is clearly significant. Iso-CoA isoforms have potential as pseudosubstrates and inhibitors and as probes for investigating CoA binding to the extensive class of CoA-utilizing enzymes. We anticipate that our investigation, in addition to generating the first reports of enzymes performing acyl transfer reactions with iso-CoA isomers, will provide the necessary methodology for future studies into iso-CoA and acyl-iso-CoA compounds.