Active acetyl-CoA synthase from Clostridium thermoaceticum obtained by cloning and heterologous expression of acsAB in Escherichia coli

Abstract

Acetyl-CoA synthase from Clostridium thermoaceticum (ACS_Ct) is an α₂β₂ tetramer containing two novel Ni-X-Fe₄S₄ active sites (the A and C clusters) and a standard Fe₄S₄ cluster (the B cluster). The acsA and acsB genes encoding the enzyme were cloned into Escherichia coli strain JM109 and overexpressed at 37^oC under anaerobic conditions with Ni supplementation. The isolated recombinant His-tagged protein (AcsAB) exhibited characteristics essentially indistinguishable from those of ACS_Ct, from which Ni had been removed from the A cluster. AcsAB migrated through nondenaturing electrophoretic gels as a single band and contained a 1:1 molar ratio of subunits and 1.0–1.6 Ni/αβ and 14–22 Fe/αβ. AcsAB exhibited 100–250 units/mg CO oxidation activity but no CO/acetyl–CoA exchange activity. Electronic absorption spectra of thionin-oxidized and CO-reduced AcsAB were similar to those of ACS_Ct, with features typical of redox-active Fe₄S₄ clusters. Partially oxidized and CO-reduced AcsAB exhibited EPR signals with g values and low spin intensities indistinguishable from those of the B_red state of the B cluster and the C_red1 and C_red2 states of the C cluster of ACS_Ct. Upon overnight exposure to NiCl₂, the resulting recombinant enzyme (ACS_Ec) developed 0.06–0.25 units/mg exchange activity. The highest of these values is typical of fully active ACS_Ct. When reduced with CO, ACS_Ec exhibited an EPR signal indistinguishable from the NiFeC signal of Ni-replete ACS_Ct. Variability of activities and signal intensities were observed among different preparations. Issues involving the assembly of these metal centers in E. coli are discussed.

Certain archaea and bacteria grow chemoautotrophically using the Wood/Ljungdahl pathway (1). The central enzyme of this pathway is acetyl-CoA synthase (ACS), also called carbon monoxide dehydrogenase (CODH). The enzyme from Clostridium thermoaceticum (ACS_Ct) catalyzes two reactions, including the reversible reduction of CO₂ to CO, and the synthesis of acetyl-CoA from CO, CoA, and a methyl group bound to a corrinoid-iron-sulfur protein (2). This corrinoid protein is methylated by methyltetrahydrofolate in a reaction catalyzed by a methyltransferase.

ACS_Ct is an α₂β₂ tetramer containing three types of metal-sulfur clusters called A, B, and C (2, 3). The B and C clusters are located in the β subunit, whereas the A cluster is located in α. The B cluster is a standard [Fe₄S₄]^2+/1+ cluster involved in electron transfer (4), and it exhibits an EPR signal with g_av = 1.94 when reduced to the B_red state (5). The C cluster is the active site for CO₂/CO redox catalysis (6–9) and likely consists of an Ni complex weakly coupled through an unidentified bridge to an Fe₄S₄ cluster (10, 11). The one- and three-electron reduced C_red1 and C_red2 states exhibit EPR signals with g_av = 1.82 and 1.86, respectively. The A cluster is the active site for acetyl-CoA synthesis and is also composed of an Ni center bridged to an Fe₄S₄ cluster (12–15). The Ni ion of the A cluster is labile and can be removed easily by incubating ACS_Ct in the chelator 1,10-phenanthroline. The Ni ion is readily reinserted by incubating phenanthroline-treated enzyme with NiCl₂ (13–15). When the A cluster is reduced by one electron and bound with CO, it yields a state (A_red-CO) that exhibits the so-called NiFeC EPR signal (g = 2.08, 2.07, 2.03). For unknown reasons, only a fraction of the A, B, and C clusters are redox and catalytically active (16). This heterogeneity is reflected in the low spin intensities of the NiFeC, g_av = 1.94, 1.82, and 1.86 signals (≈0.2, 0.6, 0.3, and 0.3 spins/αβ, respectively).

In C. thermoaceticum, the enzymes involved in the Wood/Ljungdahl pathway are encoded by the acs genes (17–19). Genes acsA and acsB encode the β and α subunits of ACS_Ct, respectively. Genes acsC and acsD encode the corrinoid-iron-sulfur protein subunits, and acsE encodes methyltransferase. The acs genes contain promoter-like sequences and translational signals that are recognized by Escherichia coli. Roberts et al. have cloned and expressed these genes in E. coli and found that only AcsE was catalytically active (17, 19). AcsCD was inactive but could be activated by incubation with cobalamin, iron, and sulfide ions (20). AcsAB was inactive, but a method of activation has not been reported. Roberts et al. reported that the recombinant α and β subunits migrated independently on nondenaturing electrophoretic gels, indicating the absence of the native tetrameric structure (17). They suggested that AcsAB was inactive because the bacteria were grown aerobically without metal ion supplementation. Morton et al. (19) later suggested that recombinant subunits might lack competent NiFeS centers. Genes encoding ACS homologs in methanogenic archaea have been sequenced, cloned, and/or overexpressed in E. coli (21–28), but to our knowledge no active recombinant enzyme from a methanogen has been reported.

The purple nonsulfur photosynthetic Rhodospirillum rubrum contains a monomeric carbon monoxide dehydrogenase homologous to the β subunit of ACS_Ct (29). CODH_Rr exclusively catalyzes the reversible oxidation of CO to CO₂ (30) and contains only the B and C clusters. The cooS gene encoding this enzyme has been cloned and sequenced (31). Using conjugative mating followed by homologous recombination, the gene was introduced into a strain of R. rubrum lacking functional CODH_Rr. This elegant method afforded functional enzyme and allowed the construction and study of numerous functional site-directed mutants (32, 33). It has also allowed consideration of issues involving the assembly of the metal-sulfur clusters. Roberts and coworkers have identified three genes (cooCTJ), deletions in which result in Ni-deficient CODH_Rr (31). This suggests that CooC, CooT, and CooJ play roles in inserting Ni into CooS during maturation. CooC possesses a nucleotide-binding “P-loop” region and is homologous to GTPases required to insert Ni into urease and NiFe hydrogenases (34–37).

Given the tremendous utility of recombinant metalloenzyme systems, we wanted to generate active enzyme by overexpressing cloned acsAB. However, because of the reported inability of E. coli to serve as a host for this, our strategy was to clone and manipulate acsA and acsB in E. coli and then introduce them into a host containing accessory proteins required for ACS metal cluster assembly. In this paper, we describe the surprising result that the purified recombinant AcsAB from E. coli is a tetramer that contains assembled B and C clusters and exhibits CO oxidation activity levels typical of native ACS_Ct. Furthermore, the recombinant protein can be activated by treatment with NiCl₂ to afford CO/acetyl-CoA exchange activity and the NiFeC EPR signal, indicative of assembled A cluster. Thus, the activated recombinant protein is essentially indistinguishable from native ACS_Ct. Implications of these results for probing the mechanism of cluster assembly and the origin of heterogeneity are discussed.

Materials and Methods

Materials.

Restriction enzymes, T4 DNA ligase, alkaline phosphatase, isopropylthio-β-D-galactoside and anti-rabbit IgG (Fc) AP conjugate were from Promega. Taq polymerase was from Perkin–Elmer. Ni-nitrilotriacetic acid (Ni-NTA) agarose, E. coli strain M15(pREP4), QiaPrep spin miniprep kit, QIAquick PCR purification kit, and vector pQE-60 were purchased from Qiagen. This vector was used to install a His-tag at the C terminus of the α subunit of cloned acsAB. Anti-ACS_Ct rabbit serum was prepared by Alpha Diagnostics (San Antonio, TX).

Cloning of acsAB.

C. thermoaceticum cells were grown as described (38) under CO₂/H₂ in a Coy Laboratory Instruments chamber. Genomic DNA was isolated using the DNA isolation kit from Puregene. PCR primers 5′-GGCGAGATCTGAATTCATCCTCAACCAC-3′ and 5′-GGCGAGATCTCATAATGGGATCCATGGT-3′, designed to hybridize to the ends of acsA and acsB, were synthesized by Genosys (The Woodlands, TX). Genes acsAB were amplified from genomic DNA using these primers, a Stratagene Gradient 96 Robocycler, and a PCR optimizer kit (Invitrogen). The resulting 4.5-kilobase PCR product was purified with the QIAquick PCR purification kit and digested with BglII. Vector pQE-60 was digested with BglII and incubated with alkaline phosphatase to prevent self-ligation. The digested vector and PCR product were ligated and transformed into M15(pREP4) competent cells. Plasmid pT316 was isolated from kanamycin- and ampicillin-resistant colonies.

Double-strand sequencing by the Gene Technologies Laboratory (Department of Biology, Texas A&M University) revealed that pT316 contained acsAB, albeit with three mutations relative to that reported by Morton et al. (19). One mutation was silent, one yielded Ser instead of Pro at position 259 in AcsA, and the third yielded Asp instead of Glu at position 527 of AcsB. A second cloning (from genomic DNA) and sequencing of the two affected regions indicated Pro-259 in AcsA and Asp-527 in AcsB. This suggests that native acsA and acsB encode Pro-259 and Asp-527, respectively. The QuikChange site-directed mutagenesis kit (Stratagene) was subsequently used to change Ser-259 to Pro-259. The resulting plasmid, pTM02, was subsequently transformed and expressed in E. coli strain JM109 [e14-(McrA-) recA1 endA1 gyrA96 thi-1 hsdR17(r_K⁻ m_K⁺) supE44 relA1 Δ(lac-proAB) F′ traD36 proAB lacI^qZΔM15] (Stratagene).

Expression of AcsAB.

E. coli JM109 (pTM02) cells were grown in 25 liters of Begg's medium (1% tryptone/0.5% yeast extract/0.4% glucose/100 mM potassium phosphate, pH 6.5/15 mM sodium formate/1 μM sodium molybdate/1 μM sodium selenite/5 μM NiCl₂) (39) under stirred (100 rpm) anaerobic (N₂) conditions using a custom-made (Chemglass, Inc.) water-jacketed all-glass Bioreactor maintained at 37°C with a refrigerated recirculator (Polysciences). When the optical density of cells reached 0.65–0.75 at 600 nm (1-cm path length), isopropylthio-β-D-galactoside and NiCl₂ were added to final concentrations of 0.1 and 0.5 mM, respectively. Cells were harvested anaerobically and frozen in liquid N₂. The presence of AcsAB, and thus the expression of acsAB, was indicated by Western blotting (40).

Purification of AcsAB.

Using standard anaerobic methods (38), 70 g of frozen cells, 70 mg of lysozyme, and a trace of DNase were suspended in 350 ml of buffer A (50 mM NaP_i, pH 8.0/0.3 M NaCl/10 mM β-mercaptoethanol). After 1 h, the suspension was sonicated (Branson Sonifier 450) at room temperature for 5 min using 1-s pulses (duty cycle constant 60% and output control 7), and then centrifuged at 23,500 × g with a GSA rotor (Sorvall) for 1.5 h. The supernatant was loaded onto an Ni-NTA agarose column (1.6 × 12 cm) equilibrated in buffer A. The column was washed with 100 ml of 10 mM imidazole in buffer A and then eluted with 100 mM imidazole in buffer A. The brown-colored eluent was diluted 4× with buffer B (50 mM Tris-Cl, pH 8.0/10 mM DTT/2 mM sodium dithionite) and loaded on a DEAE column (2.6 × 15 cm) equilibrated in buffer B. The column was washed with 150 ml of 0.2 M NaCl in buffer B followed by 300 ml of a 0.2–0.4 M gradient of NaCl in buffer B. Fractions exhibiting CO oxidation activity were combined, concentrated, and frozen in liquid N₂.

One batch of AcsAB was isolated as described (38) without the use of the Ni-NTA agarose column and further purified by FPLC, as described (41) using a POROS 50D column (PerSeptive Biosystems, Framingham, MA), equilibrated in buffer C (50 mM Tris, pH 8.0). After loading, the column was washed with 125 ml of 0.1 M NaCl in buffer C and eluted with 125 ml of a 0.1–0.4 M gradient of NaCl in buffer C.

Characterization and Activation of AcsAB.

Sample purity was quantified by scanning Coomassie blue-stained SDS/PAGE gels with an AlphaImager 2000 (Alpha Innotech, San Leandro, CA) densitometer. Native gel electrophoresis was performed anaerobically as described (42). CO oxidation and CO/acetyl-CoA exchange activities were performed as described (38, 43). The Ni and Fe contents of purified AcsAB were determined using atomic absorption spectrometry (12). EPR spectra of AcsAB were determined using a Bruker ESP300 spectrometer and an Oxford Instruments ER910A cryostat. Signals were quantified as described (44) using a 1.00 mM CuEDTA standard. UV-visible absorption spectra were determined using a Beckman DU 640B spectrophotometer. Samples were freed from reductant by passage through a column of Sephadex G-25 (1 × 17 cm) equilibrated in buffer D (100 mM Mops, pH 7.5). A sample was transferred to a quartz cuvette that was subsequently sealed with a septum and removed from the glove box. Thionin solution was added until a weak band at 600 nm appeared because of oxidized thionin, ensuring a fully oxidized sample. Another sample was reduced by exposure to 1 atm (1 atm = 101.3 kPa) CO. Protein concentrations were determined as described (45) using BSA as a standard. The A cluster of reductant-free AcsAB (≈3–8 mg/ml⁻¹) in buffer D was activated by incubation with 200 μM (final concentration) NiCl₂ overnight under an Ar or CO atmosphere at room temperature.

Results

Genes acsA and acsB from C. thermoaceticum were cloned and overexpressed in E. coli. The recombinant protein AcsAB contained a His-tag, which allowed rapid purification. Samples were >95% pure according to densitometric analysis of SDS/PAGE gels. AcsAB consisted of two subunits that migrated, in SDS/PAGE gels, at the same rates as the α and β subunits of ACS_Ct (Fig. 1). Based on relative staining densities, the α and β recombinant subunits were present in a 1:1 molar ratio. Moreover, AcsAB had the same mobility as ACS_Ct on a nondenaturing electrophoresis gel (data not shown). These results suggest that AcsAB possesses the same α₂β₂ tetrameric structure as ACS_Ct.

Figure 1.

SDS/PAGE gel (12%) of ACS_Ct (A) and AcsAB (B).

The three batches of AcsAB whose metal contents were determined contained an average of 1.3 ± 0.3 Ni/αβ and 17 ± 4 Fe/αβ (Table 1). To address the possibility that the measured Ni arose from ions that leached from the Ni-NTA column and bound onto the His-tag of AcsAB during purification, the Ni content of the batch purified without this column was also determined. That sample (75% pure) contained 1.7 Ni/αβ (assuming no Ni in the contaminating proteins), indicating that the Ni ions measured in the Ni-NTA purified samples were associated with AcsAB. A control sample of ACS_Ct contained 1.8 Ni/αβ and 12 Fe/αβ. ACS_Ct has been reported to contain 2.0 ± 0.2 Ni/αβ (n = 10) and 12 ± 2 Fe/αβ (n = 11).¶ Thus, our metal analyses suggest that AcsAB contains a full complement of FeS clusters, relative to ACS_Ct (i.e., ≥12 Fe/αβ) and 1.0–1.6 Ni/αβ.

Table 1.

Characteristics of purified AcsAB before and after Ni activation

Batch	1	2	3
Mass of cell paste (g)	70	65	80
Ni content before activation (Ni/αβ)	1.5	1.6	1.0
Fe content before activation (Fe/αβ)	16	22	14
CO oxidation activity before/after activation (units/mg)	160/140	200(220)/250	100(90)/100
CO exchange activity before/after activation (units/mg)	0/0.06	0/0.15	0/0.25
NiFeC Signal intensity before/after activation (spin/αβ)	0/0.15	0/0.38	0/0.15
gav = 1.94 Signal intensity before/after activation (spin/αβ)	1.12/0.75	0.95(0.55)/0.93	0.83(0.66)/0.82
gav = 1.82 Signal intensity before activation (spin/αβ)	0.29	0.31	0.25
gav = 1.86 Signal intensity before/after activation (spin/αβ)	0.21/0.24	0.13/0.18	0.08/0.09

Batch 1 was activated with Ni under 1 atm Ar, while batches 2 and 3 were activated under 1 atm CO. Values in parentheses for batches 2 and 3 are from samples incubated overnight in the absence of reductants and under 1 atm Ar. 

The electronic absorption spectrum of thionin-oxidized AcsAB (Fig. 2) revealed a broad absorption shoulder in the 400-nm region, caused by the presence of [Fe₄S₄]²⁺ clusters. Spectral intensity at 420 nm declined when the protein was exposed to CO, reflecting the reduction of these clusters to the 1+ core oxidation state. These spectra were similar to those of ACS_Ct (51). The extinction coefficient difference (Δɛ₄₂₀ = ɛ₄₂₀ oxidized − ɛ₄₂₀ reduced) quantified to 14.1 mM⁻¹cm⁻¹ for AcsAB compared to 14.7 mM⁻¹cm⁻¹ for ACS_Ct (51). These results suggest that AcsAB and ACS_Ct contain the same number and types of FeS clusters.

Figure 2.

Electronic absorption spectra of AcsAB. Solid line, AcsAB exposed to 1 atm CO. Dotted line, AcsAB oxidized with a slight excess of thionin. The slight feature in the 600-nm region is attributable to absorption from oxidized thionin. (Inset) Oxidized minus-reduced difference spectrum.

EPR spectra of three batches of reductant-free partially oxidized AcsAB consisted of two signals (Fig. 3A), one with g_av = 1.82 (g = 2.01, 1.80, 1.64) and the other with g_av = 1.94 (g = 2.04, 1.94, 1.90). At 20 mW and 10 K, the g_av = 1.94 and 1.82 signals showed little evidence of saturation (Fig. 3B). At temperatures higher than ≈16 K, the g_av = 1.82 signal relaxation broadened, whereas the g_av = 1.94 signal did not. Quantification of the g_av = 1.82 signals yielded an average of 0.28 spin/αβ. In the same redox state, ACS_Ct exhibits two signals, a g_av = 1.94 signal arising from B_red and a g_av = 1.82 signal from C_red1. Because the signals in AcsAB have indistinguishable g values and saturation/relaxation properties as those in ACS_Ct, we conclude that they arise from clusters with the same structures as the B and C clusters of ACS_Ct and refer to them by the same names. Moreover, the average quantified spin intensity of the C_red1 state of the C cluster in AcsAB (Table 1) is similar to that observed in ACS_Ct, indicating a similar proportion of functional C clusters in ACS_Ct and AcsAB.

Figure 3.

10 K X-band EPR spectra of AcsAB (batch 2, 6.8 mg/ml). (A) Partially oxidized with thionin, 0.05 mW microwave power; (B) same as A, but at 20 mW microwave power. Other spectrometer conditions: microwave frequency, 9.42 GHz; modulation frequency, 100 kHz; modulation amplitude, 11.8 G; sweep time, 164 s; time constant, 328 s.

In batches 1 and 2, CO-reduced AcsAB exhibited two EPR signals, one with g_av = 1.94 and the other with g_av = 1.86 (g = 1.97, 1.87, 1.75) (Fig. 4A and B). Essentially indistinguishable signals are observed in spectra of CO-reduced ACS_Ct. These signals in ACS_Ct arise from B_red and C_red2, and we conclude that the corresponding signals in AcsAB arise from analogous states. The B_red and C_red2 signals of batch 3 differ slightly from those of batches 1 and 2 (Fig. 4 C and D). The reason for this is unknown, but the differences probably reflect minor perturbations in protein or cluster structure. The g_av = 1.86 and g_av = 1.94 signals from the three batches quantified to an average of 0.14 spin/αβ and 0.97 spin/αβ, respectively (Table 1). The g_av = 1.94 signal is ≈30% more intense than in ACS_Ct, possibly suggesting a shift in spin state from S = 3/2 to S = 1/2 for the B_red state in AcsAB. The intensity of the C_red2 signal from AcsAB is about half of that reported for ACS_Ct (in ACS_Ct, both g_av = 1.82 and 1.86 signals typically quantify to ≈0.3 spin/αβ). We may have underestimated the g_av = 1.86 signal intensity in these spectra, as the dominating g_av = 1.94 signal is superimposed. Alternatively, the redox potential for the C_int/C_red2 couple may be slightly more negative in AcsAB than in ACS_Ct, resulting in a greater proportion of C clusters in the EPR-silent C_int state in AcsAB (52). In any event, the presence of the g_av = 1.86 signal provides conclusive evidence that the C cluster is present in AcsAB and has the same basic structure and properties as that in ACS_Ct.

Figure 4.

10 K X-band EPR spectra of AcsAB. (A) CO-reduced (batch 2, 6.8 mg/ml), 0.05 mW microwave power; (B) Same as A, except 20 mW microwave power; (C) CO-reduced (batch 3, 4.6 mg/ml), 0.05 mW microwave power; (D) same as C, except 20 mW microwave power. Other conditions were as in Fig. 3.

CO-reduced AcsAB did not exhibit the characteristic NiFeC EPR signal of CO-reduced ACS_Ct. To address the possibility that AcsAB contained a partially assembled A cluster (an assembled Fe₄S₄ cluster without Ni), the protein was incubated overnight in NiCl₂. Ni-treated AcsAB, to be called ACS_Ec, exhibited an EPR signal (Fig. 5B) with g values, saturation/relaxation properties, and spin intensities (Table 1) indistinguishable from the NiFeC signal of ACS_Ct. Thus, we conclude that ACS_Ec contains an A cluster with redox and spectroscopic properties similar to that in ACS_Ct.

Figure 5.

130 K X-band EPR spectra of CO-reduced ACS_Ec (batch 2, 5.7 mg/ml). (A) before addition of NiCl₂ and (B) after addition of NiCl₂. Other spectrometer conditions were as in Fig. 3 except that microwave power was 80 mW.

Purified ACS_Ec (which exhibited the NiFeC signal when reduced with CO) was loaded on the Ni-NTA agarose column, eluted with 100 mM imidazole, and then subjected to Sephadex G-25 chromatography to remove excess imidazole and Ni. The resulting material did not exhibit the NiFeC signal, suggesting that the Ni-NTA purification step can strip Ni from activated A cluster.

The catalytic properties of ACS_Ec and AcsAB were examined. The three batches of AcsAB and ACS_Ec exhibited 100 to 250 units/mg of CO oxidation activity (Table 1). The activity of batch 2 is within the range of values observed in our laboratory for fully active ACS_Ct (280 ± 70 units/mg; n = 44)^¶. The higher activity of batch 2 AcsAB and ACS_Ec is also consistent with the more intense C cluster signals observed for this batch. Enzyme allowed to incubate overnight in the absence of any reducing agents resulted in a lower intensity of the B_red signal and CO oxidation activity, suggesting that AcsAB may not be quite as stable as ACS_Ct. None of the AcsAB batches exhibited CO/acetyl-CoA exchange activity, whereas all Ni-treated ACS_Ec samples did (Table 1). The activities of batches 2 and 3 were within the range observed for fully functional ACS_Ct (0.17 ± 0.10 units/mg; n = 39). This result demonstrates that ACS_Ec can contain fully assembled and functional A cluster.

Discussion

Our results demonstrate that the acsAB genes from C. thermoaceticum can be cloned and heterologously expressed in E. coli to yield a recombinant protein able to catalyze the oxidation of CO. AcsAB has an α₂β₂ tetrameric quaternary structure, fully assembled B and C clusters, and at least the Fe₄S₄ component of the A cluster. Because the purification method used seems able to remove the labile Ni from the A cluster, it is unclear whether AcsAB in E. coli has a fully assembled A cluster. In any event, an Ni ion can be inserted in vitro, affording a fully functional A cluster and enzyme capable of catalyzing CO/acetyl-CoA exchange. The measured redox and spectroscopic properties of these clusters are nearly indistinguishable from those of native ACS_Ct. In brief, our results indicate that Ni-treated recombinant enzyme (ACS_Ec) is essentially equivalent to native ACS_Ct, exhibiting both catalytic activities and the full complement of metal clusters that define this enzyme.

These results and conclusion contrast sharply with those of Roberts et al., who reported that expression of acsAB in E. coli afforded nonfunctional α and β subunits that migrated independently (17). We obtain functional recombinant enzyme, but of lesser yields, regardless of the strain of E. coli used, the growth conditions (anaerobic or bubbled with air,‖ 15–45°C), or the presence of supplemented Ni in the media. Furthermore, recombinant protein seems relatively stable in its tetrameric form.

The results of this study provide a convenient means for biosynthesizing site-directed mutants of this enzyme. These mutants could be used to identify ligands to the metal centers as well as reveal their role in establishing the catalytic, redox, and/or spectroscopic properties that distinguish these centers. This achievement also provides a means to study metal cluster assembly.

Proteins containing novel metal centers, such as urease, NiFe-hydrogenase, and nitrogenase, have been shown to require specific accessory proteins for assembly and insertion of their metal centers (53–58). The previous report of nonfunctional ACS (17) suggested that such proteins were required for the biosynthesis of this metalloenzyme but were absent in E. coli. This was easily rationalized, as E. coli does not naturally contain ACS, and the metal centers in this enzyme appear to be unique in biology. Therefore, our success was surprising, as it suggests that E. coli contains accessory proteins that assemble and insert the B and C clusters into recombinant AcsAB, as well as the FeS component of the A cluster. The B cluster and the Fe₄S₄ components of the C and A clusters may have been assembled and inserted using IscS and IscU, enzymes in E. coli that catalyze the formation of FeS clusters (59).

It is intriguing to consider how Ni was inserted into the C cluster in vivo. A series of gene clusters in E. coli is responsible for assembling the active sites of the organism's three NiFe hydrogenases, including hya, hyb, hyc, and hyp (55, 56, 60–63). When hypB is deleted, all three hydrogenases lack Ni in their active sites, suggesting that HypB participates in inserting Ni into these sites (36, 37). HypB is a homolog of CooC, a protein involved in inserting Ni into the C cluster of CODH_Rr (31). We conjecture that in E. coli, HypB served as a surrogate of CooC to insert Ni into AcsAB, thereby forming the C cluster. Given that our batches of AcsAB had somewhat variable CO oxidation activity and C cluster EPR signal intensities, this surrogate may not be as effective as CooC in assembling the C cluster.

Whether Ni was inserted into the labile Ni site of the A cluster by accessory proteins in E. coli remains unknown. E. coli may lack an accessory protein capable of serving as a surrogate for the unidentified accessory protein in C. thermoaceticum that is used to insert Ni during A cluster assembly. Alternatively, the Ni in an assembled A cluster may have been lost during purification. Fortunately, Ni could be inserted into this site simply by incubating the enzyme overnight in NiCl₂.

Finally, these results impact our understanding of the redox and spin-state heterogeneity that plagues this enzyme. The evidence for heterogeneity is extensive (5, 10, 13–16, 46, 50) and will only be summarized here. Approximately 40% of A clusters contain labile Ni ions, function catalytically, and can be reduced to the A_red-CO state (exhibiting the NiFeC signal). The remaining 60% contain nonlabile Ni ions, are nonfunctional, and cannot be reduced. This situation is reflected in the low spin intensity of the NiFeC signal. The C cluster is also heterogeneous. Approximately 40% of C clusters in the C_ox state can be reduced by one, two, and three electrons, forming C_red1, C_int, and C_red2, respectively. The remaining 60% can be reduced by one electron to the C_S=3/2 state. Again, the low spin intensities of the g_av = 1.82 and 1.86 signals reflect this situation. The B cluster appears to exhibit only spin-state heterogeneity, with about 60% of B_red having an S = ½ ground state (affording the g_av = 1.94 signal) and the remainder having S = 3/2 (affording EPR features between g = 4–6). Obvious reasons for this heterogeneity, including errors in determining protein, metal, or spin concentrations or damage incurred during purification, have been considered and discounted (38). Fraser and Lindahl recently suggested that heterogeneity might arise from events occurring during the growth of C. thermoaceticum (16). However, the results presented here indicate that the same type of heterogeneity present in ACS_Ct plagues recombinant AcsAB (and ACS_Ec). Thus, it would seem that the factors causing heterogeneity are independent of the host organism and of the accessory proteins used to assemble the clusters and are intrinsic to the enzyme. Such intrinsic heterogeneity could arise if the two αβ dimers of the α₂β₂ tetramer were symmetry-inequivalent.

With active recombinant ACS from C. thermoaceticum now available, we plan to prepare site-directed mutants to help elucidate details of the catalytic mechanism of this enzyme and the roles of its metal clusters. We also plan to probe the mechanism of cluster assembly and uncover the origin of the heterogeneity.

Abbreviations

ACS_Ct

acetyl-CoA synthase from C. thermoaceticum, also called carbon monoxide dehydrogenase or CODH_Ct

AcsAB

isolated His-tagged protein produced by expressing C. thermoaceticum acsA and acsB genes in E. coli

ACS_Ec

AcsAB after activating with NiCl₂

CODH_Rr

carbon monoxide dehydrogenase from R. rubrum

Ni-NTA

Ni-nitrilotriacetic acid