Kinetic and Spectroscopic Studies of the ATP:Corrinoid Adenosyltransferase PduO from Lactobacillus reuteri: Substrate Specificity and Insights into the mechanism of Co(II)corrinoid Reduction

Abstract

The PduO-type ATP:corrinoid adenosyltransferase from Lactobacillus reuteri (LrPduO) catalyzes the formation of the essential Co–C bond of adenosylcobalamin (coenzyme B₁₂) by transferring the adenosyl group from co-substrate ATP to a transient Co¹⁺corrinoid species generated in the enzyme active site. While PduO-type enzymes have previously been believed to be capable of adenosylating only Co¹⁺cobalamin (Co¹⁺Cbl⁻), our kinetic data obtained in this study provide in vitro evidence that LrPduO can in fact also utilize the incomplete corrinoid Co¹⁺cobinamide (Co¹⁺Cbi) as an alternative substrate. To explore the mechanism by which LrPduO overcomes the thermodynamically challenging reduction of its Co²⁺corrinoid substrates, we have examined how the enzyme active site alters the geometric and electronic properties of Co²⁺Cbl and Co²⁺Cbi⁺ by using electronic absorption, magnetic circular dichroism, and electron paramagnetic resonance spectroscopic techniques. Our data reveal that upon binding to LrPduO that was pre-incubated with ATP, both Co²⁺corrinoids undergo a partial (~40–50%) conversion to distinct paramagnetic Co²⁺species. The spectroscopic signatures of these species are consistent with essentially four-coordinate, square-planar Co²⁺ complexes, based on a comparison with the results obtained in our previous studies of related enzymes. Consequently, it appears that the general strategy employed by adenosyltransferases for effecting Co²⁺→Co¹⁺ reduction involves the formation of an “activated” Co²⁺corrinoid intermediate that lacks any significant axial bonding interactions, so as to stabilize the redox-active, Co 3d_z²-based molecular orbital.

Coenzyme B₁₂, which is also known as adenosylcobalamin (AdoCbl¹), provides one of very few known examples of a bio-organometallic species (1). AdoCbl contains a cobalt(III) ion that is equatorially ligated by four nitrogens of a tetrapyrrole macrocycle, termed the corrin ring, and axially coordinated by the 5′-carbon atom of an adenosyl (Ado) moiety on the “upper” face and a nitrogen atom from 5,6-dimethylimidazole (DMB) on the “lower” face (Figure 1) (2). The DMB is part of a nucleotide loop that is tethered to the corrin ring at C¹⁷; in adeonsylcobinamide (AdoCbi⁺), this loop is absent and the lower axial coordination site is occupied by a water molecule to complete a distorted octahedral ligand environment of the Co³⁺ ion (3).

Figure 1.

Schematic representation of the adenosylcobalamin cofactor (AdoCbl). In adenosylcobinamide (AdoCbi⁺), the nucleotide loop including the DMB base is absent and a water molecule occupies the lower axial position.

AdoCbl serves as a cofactor for numerous distinct enzymes that catalyze radical-induced rearrangement reactions (4–10), such as methylmalonyl-CoA mutase, glutamate mutase, diol dehydratase, and ethanolamine ammonia lyase, or ribonucleotide reduction, as exemplified by the ribonucleotide tri-phosphate reductase (11). Common to these enzymes is that the first step in their respective catalytic cycles involves homolytic cleavage of the cofactor’s Co–C bond to yield Co²⁺cobalamin (Co²⁺Cbl) and an organic radical centered on the 5′-carbon of the Ado moiety. The Ado• radical generated in this process then abstracts a hydrogen atom from substrate to initiate a protein-mediated rearrangement reaction that eventually leads to product formation and regeneration of the AdoCbl cofactor (12–14).

Eukaryotes lack the biosynthetic machinery needed to synthesize AdoCbl de novo. Instead, those eukaryotes that utilize AdoCbl in their metabolism (which does not include plants) convert exogenous cobalamins, such as vitamin B₁₂ (in which the upper axial coordination site is occupied by a CN⁻ moiety rather than Ado), to AdoCbl by using a class of enzymes termed adenosyltransferases (15, 16). The importance of these enzymes is highlighted by the fact that the malfunctioning of the human adenosyltransferase (hATR) can cause methylmalonic aciduria, an autosomal disorder that is often fatal in infants (17). Many prokaryotes also require AdoCbl; e.g., for the utilization of propanediol and ethanolamine to produce substrates for ATP production and as sources of carbon and nitrogen, respectively (7, 8). Some of these organisms are capable of synthesizing AdoCbl de novo in a process involving over 25 enzymes, which catalyze the stepwise conversion of 5-aminolevulinic acid to the complete cofactor via an intermediate named cobyric acid, a cobalamin precursor that possesses all of the necessary functionality except for the nucleotide loop and the axial Ado ligand (18).

Regardless of whether an organism synthesizes AdoCbl de novo or salvages incomplete corrinoids from its environment, both processes require two one-electron reduction steps to convert Co³⁺ to Co¹⁺ and the formation of the essential Co–C bond by transferring the Ado group from a molecule of ATP to the transiently formed Co¹⁺corrinoid species (19, 20). While previous studies revealed that the Co^3+/2+corrinoid reduction potential is sufficiently positive to ensure that the Co³⁺corrinoid be converted to the Co²⁺ state in the reducing environment of the cytoplasm (21), the Co^2+/1+corrinoid potential is lower than the midpoint potentials of putative in vivo reducing agents. For example, the Co^2+/1+ reduction potential of aqueous Co²⁺Cbl (E° = −610 mV) is well below that of the semiquinone/reduced flavin couple of flavodoxin A (FldA, E° = −440 mV) (22–24). Despite this apparent thermodynamic dilemma, it has been shown that in the presence of adenosyltransferases, the Co²⁺→Co¹⁺ reduction can be effected under physiologically relevant conditions (25–27).

Previously, our spectroscopic and computational studies of two representative members of the family of adenosyltransferases, CobA from Salmonella enterica and hATR, have provided definitive clues as to how these enzymes overcome the thermodynamically challenging task of reducing Co²⁺corrinoids to their Co¹⁺ oxidation state (28, 29). Specifically, we have shown that upon binding of Co²⁺cobinamide (Co²⁺Cbi⁺) to CobA complexed with co-substrate ATP, the axially bound water molecule (partially) dissociates so as to generate an effectively four-coordinate square-planar Co²⁺ species. As a result of this (partial) ligand dissociation, the redox-active Co 3d_z2 orbital that is oriented toward the axial coordination sites is significantly stabilized in energy, thereby raising the Co^2+/1+ reduction midpoint potential by an estimated 250 mV (28). Remarkably, the same spectroscopic signatures characterizing this “activated” four-coordinate Co²⁺ species features have been found to develop upon binding of Co²⁺Cbl to hATR in the presence of ATP (29), even though hATR and CobA are structurally and evolutionary unrelated (7, 30, 31).

Recently, St. Maurice et al. have purified an hATR-type adenosyltransferase from Lactobacillus reuteri, termed LrPduO, and reported its X-ray crystal structure with ATP bound to the active site (32). In vivo and in vitro experiments have confirmed that LrPduO can convert Co¹⁺Cbl⁻ to AdoCbl and revealed that it can substitute for CobA in Salmonella enterica, implying that LrPduO possesses the ability to adenosylate both Co²⁺Cbl and Co²⁺Cbi⁺ (16, 32). In the present study, we have performed a detailed kinetic and spectroscopic characterization of LrPduO to obtain molecular-level insight into certain steps of the catalytic mechanism employed by this enzyme. Our kinetic parameters obtained for the reaction of Co¹⁺Cbi with LrPduO/ATP confirm that cobinamide can serve as an alternative substrate for this adenosyltransferase, while our spectroscopic data indicate that LrPduO utilizes the same general strategy for promoting Co²⁺→Co¹⁺ reduction as CobA and hATR; namely, to generate an effectively four-coordinate Co²⁺corrinoid species regardless of whether Co²⁺Cbl or Co²⁺Cbi⁺ is used as the substrate. Nonetheless, a comparison of the results obtained for LrPduO and those reported previously for hATR and CobA reveals small but notable differences with regards to the substrate specificity and degree of Co²⁺corrinoid activation.

MATERIALS AND METHODS

Cofactors and Chemicals

Aquacobalamin (H₂OCbl⁺), dicyanocobinamide ((CN)₂Cbi), and sodium borohydride (NaBH₄) were purchased from Sigma and used as obtained. Diaquacobinamide ((H₂O)₂Cbi²⁺) was prepared by reducing (CN)₂Cbi with NaBH₄, loading the reaction mixture on a C18 SepPack column, washing with doubly distilled H₂O, and eluting the product with methanol, as described in a previous report (28). Co²⁺Cbl and Co²⁺Cbi⁺ were prepared by adding NaBH₄ to degassed aqueous solutions of H₂OCbl⁺ and (H₂O)₂Cbi²⁺, respectively, in the presence of 60% (v/v) glycerol, and the progress of the conversion was monitored spectrophotometrically.

Protein Production and Purification

Recombinant PduO protein from Lactobacillus reuteri was obtained by overexpressing a pTEV3 plasmid transformed into Escherichia coli strain BL21(DE3). pTEV3 encodes the LrPduO protein with an N-terminal (His)₆ tag (32, 33). This (His)₆ tag was subsequently clipped off using rTEV protease, and the tag-free enzyme was purified using a HisTrap FF column (Amersham Biosciences), as described previously (32). Purified LrPduO was stored at −80 °C in Tris-HCl buffer (0.1 M, pH 8 at 4 °C) containing NaCl (0.5 M). Protein samples used for kinetic analyses additionally contained 10% (v/v) glycerol.

Kinetic Studies

Activity assays were performed as described previously (32) without modifications (final LrPduO concentration was 1 μg/mL). AdoCbi⁺ formation was monitored spectrophotometrically at 388 nm, which corresponds to the peak position of the dominant absorption feature of Co¹⁺Cbi. The values of V_max and K_m were determined from plots of the initial velocity versus substrate concentration. All kinetic parameters were determined in duplicates; their values and uncertainties as reported in this paper correspond to the average of the two measurements and their standard deviations, respectively.

Sample Preparation

Solutions of LrPduO (in 50 mM Tris-HCl buffer, pH 8 containing 0.5 M NaCl) in the absence or presence of 3 mM MgATP were purged separately with Ar gas for 30 min at 4 °C before they were combined with the degassed free Co²⁺corrinoid solutions in a 0.9:1.0 Co²⁺corrinoid:LrPduO molar ratio in a sealed anaerobic vial (final enzyme concentration of 0.4 mM). The reaction mixtures, which contained 60% (v/v) glycerol, were then transferred anaerobically into the appropriate sample cells (previously purged with Ar gas) and immediately frozen in liquid N₂.

Spectroscopy

Low-temperature electronic absorption (Abs) and magnetic circular dichroism (MCD) spectra were collected on a Jasco J-715 spectropolarimeter in conjunction with an Oxford Instruments SM-4000 8T magnetocryostat. All MCD spectra presented in this paper were obtained by taking the difference between spectra collected with the magnetic field oriented parallel and antiparallel to the light propagation axis to remove contributions from the natural CD and glass strain.

X-band EPR spectra were obtained by using a Bruker ESP 300E spectrometer in conjunction with an Oxford ESR 900 continuous flow liquid helium cryostat and an Oxford ITC4 temperature controller. The microwave frequency was measured with a Varian EIP model 625A CW frequency counter. All spectra were collected using a modulation amplitude of 10 G, a modulation frequency of 100 kHz, and a time constant of 41 ms. EPR spectral simulations were performed using the WEPR program developed by Dr. Frank Neese (34).

RESULTS AND ANALYSIS

Kinetic Studies

To explore whether LrPduO possesses the ability to adenosylate incompletely assembled corrinoids in addition to Co¹⁺Cbl⁻ (32), initial velocity kinetic measurements were performed using chemically reduced Co¹⁺Cbi as the substrate. In these experiments, the concentration of Co¹⁺Cbi was varied while that of the co-substrate ATP was kept at saturation. The apparent K_m for Co¹⁺Cbi (K_m = 0.096 μM) was found to be similar to the one reported for Co¹⁺Cbl⁻ (K_m = 0.13 μM). Likewise, the rates at which LrPduO can adenosylate Co¹⁺Cbl⁻ and Co¹⁺Cbi are nearly indistinguishable (k_cat = 2.0 and 2.4 × 10⁻² s⁻¹, respectively, see Table 1), indicating that, at least in vitro, the nucleotide loop is not required for the binding and adenosylation of the corrinoid substrate by LrPduO.

Table 1.

Kinetic parameters for the LrPduO catalyzed conversion of Co¹⁺Cbl⁻ and Co¹⁺Cbi to AdoCbl and AdoCbi⁺, respectively^a

Substrate	kcat (s−1)	Km (μM)	kcat/Km (M−1 s−1)
Co1+Cbl−	2.4 ± 0.1 × 10−2	13 ± 1 × 10−2	1.8 ± 0.2 × 105
Co1+Cbi	2.0 ± 0.2 × 10−2	9.6 ± 1.4 × 10−2	2.1 ± 0.4 × 105

^aData for Co¹⁺Cbl^- were taken from ref (32).

Co²⁺Cbl ↔ LrPduO Interactions

(A) Co²⁺Cbl + LrPduO

The low-temperature Abs spectra of Co²⁺Cbl in the absence and presence of LrPduO are nearly identical with respect to band positions and intensities (cf Figures 2A and 2B). Although the addition of LrPduO causes a minor blue-shift of the prominent feature in the visible spectral region (the α-band (35)) of the Co²⁺Cbl Abs spectrum, from 21010 to 21050 cm⁻¹ (Table 2), it has no discernible effect on the position of the dominant Abs feature in the near-UV region at ~32150 cm⁻¹ (note that the shoulder near 18520 cm⁻¹ in the Co²⁺Cbl + LrPduO Abs spectrum is due to the presence of a small fraction of re-oxidized corrinoid in that sample).

Figure 2.

Abs (gray traces, right axis) and 7 T MCD (black traces, left axis) spectra collected at 4.5 K of (A) free Co²⁺Cbl, (B) Co²⁺Cbl in the presence of LrPduO, and (C) Co²⁺Cbl in the presence of the LrPduO/ATP complex. The peak position of the dominant Abs feature (the α-band) is indicated by the vertical solid line.

Table 2.

Changes in the α-band positions of Co²⁺Cbl and Co²⁺Cbi⁺ caused by the addition of LrPduO in the absence and presence of co-substrate ATP, as determined from the 4.5 K Abs spectra in Figures 2 and 4^a

Species	Free Corrinoid	+ LrPduO	+ LrPduO/ATP
Co2+Cbl	21010 cm−1 (476.0 nm)	21050 cm−1 (475.0 nm)	21480 cm−1 (465.5 nm)
Co2+Cbi+	21250 cm−1 (470.5 nm)	21250 cm−1 (470.5 nm)	21480 cm−1 (465.5 nm)

^aNote that a blue-shift of the α-band reflects a weakening of the axial ligand–Co²⁺corrinoid bonding interaction.

A complementary and considerably more sensitive probe of the interaction between Co²⁺Cbl and LrPduO is provided by low-temperature MCD spectroscopy. Because the low-energy region of the Co²⁺corrinoid MCD spectra is dominated by ligand-field (LF) transitions, this region is particularly valuable for monitoring changes in the Co²⁺ coordination environment (35). Hence, the fact that the MCD spectra of Co²⁺Cbl in the absence and presence of LrPduO are virtually identical (cf Figures 2A and 2B) rules out any major structural perturbations to the Co²⁺ center upon cofactor binding to the enzyme active site.

Consistent with our MCD results, the EPR spectra of Co²⁺Cbl in the absence and presence of LrPduO are nearly superimposable (cf Figures 3A and 3B). However, closer examination of the EPR spectrum of Co²⁺Cbl + LrPduO reveals the presence of a broad feature near 2600 G that has no counterpart in the spectrum of the free cofactor. This feature is characteristic of base-off Co²⁺Cbl (36, 37), a species in which a water molecule binds to the lower axial coordination site of the Co²⁺ center that was originally occupied by the DMB moiety in base-on Co²⁺Cbl (Figure 1). A quantitative analysis of the Co²⁺Cbl + LrPduO EPR spectrum reveals that this sample contains ~30% base-off Co²⁺Cbl. This finding correlates nicely with the slight blue-shift of the α-band that is observed in the Co²⁺Cbl Abs spectrum upon the addition of LrPduO (Figures 2A and 2B, Table 2). The g and ⁵⁹Co hyperfine values A(Co) obtained from simulations of the EPR spectra presented in Figure 3 are summarized in Table 3.

Figure 3.

EPR spectra collected at 20 K of (A) free Co²⁺Cbl, (B) Co²⁺Cbl in the presence of LrPduO, and (C) Co²⁺Cbl in the presence of the LrPduO/ATP complex. Spectrum D was obtained by subtracting 60% of spectrum B from spectrum C. Spectral simulations (thin lines) were performed using the parameters provided in Table 3.

Table 3.

EPR g Values and ⁵⁹Co Hyperfine Values A(Co) (in MHz) from the spectral simulations presented in Figures 3 and 5

Species	Spectruma	g1	g2	g3	A1(Co)	A2(Co)	A3(Co)
Free Co2+Cbl b	3A	2.00	2.23	2.28	305	30	40
Co2+Cbl + LrPduO c	3B	2.00 2.00	2.23 2.34	2.28 2.37	305 400	30 220	40 220
Co2+Cbl + LrPduO/ATP	3D	1.99	2.70	2.72	770	755	595
Free Co2+Cbi+ b	5A	2.00	2.34	2.34	405	220	220
Co2+Cbi+ + LrPduO	5B	2.00	2.34	2.37	400	220	220
Co2+Cbi+ + LrPduO/ATP	5D	1.99	2.65	2.74	810	665	565

^aThe labels refer to the spectra in Figures 3 and 5.

^bFrom ref (28).

^cThis spectrum has contributions from ~70% base-on and ~30% base-off Co²⁺Cbl.

(B) Co²⁺Cbl + LrPduO/ATP

While the spectral changes in response to Co²⁺Cbl binding to substrate-free LrPduO are quite modest (see above), a rather significant blue-shift of the α-band is observed in the Abs spectrum of Co²⁺Cbl upon the addition of the LrPduO complexed with the co-substrate ATP (cf Figures 2B and 2C, Table 2). This shift occurs in parallel with a dramatic change to the MCD spectrum, most notably the appearance of new features at ~12340 and 20580 cm⁻¹ (Figure 2C). Because the MCD feature at 12340 cm⁻¹ has no discernible counterpart in the Abs spectrum, it can be attributed to a magnetic-dipole allowed Co²⁺ d→d transition (38, 39). An analogous low-energy MCD feature has been observed for Co²⁺Cbl bound to the hATR/ATP complex, where it was shown to reflect the presence of an effectively four-coordinate Co²⁺Cbl species (29). Our MCD data thus provide compelling evidence that a four-coordinate Co²⁺Cbl species is also formed when Co²⁺Cbl binds to the LrPduO/ATP complex. This conversion is not complete, however, as the negative features at 15250 and 16880 cm⁻¹ and the positive features at 18000 and 18850 cm⁻¹ in the MCD spectrum of Co²⁺Cbl + LrPduO/ATP (Figure 2C) are characteristic of base-on Co²⁺Cbl (see Figure 2A). This fraction of five-coordinate Co²⁺Cbl corresponds to either unbound corrinoid substrate or to an enzyme-bound species that has resisted conversion to the four-coordinate form.

In agreement with our MCD data presented in Figure 2, the EPR spectrum of Co²⁺Cbl + LrPduO/ATP (Figure 3C) reveals the presence of at least two distinct paramagnetic species, one exhibiting an EPR spectrum characteristic of base-on Co²⁺Cbl (see Figure 3A) and the other giving rise to the appearance of a series of widely spread resonances in the low-field region. To better resolve the features associated with the latter species, the suitably scaled (×0.6) Co²⁺Cbl + LrPduO EPR spectrum (Figure 3B) was subtracted from the composite spectrum in Figure 3C to obtain the trace shown in Figure 3D. As expected on the basis of our MCD data analysis (vide supra), the resulting spectrum is very similar to that reported for the four-coordinate Co²⁺Cbl species in samples of Co²⁺Cbl + hATR/ATP (29). In particular, the g_2,3 and A(Co) values obtained from a fit of the EPR spectrum in Figure 3D are much larger than those of any five-coordinate Co²⁺Cbl species (Table 3), and are instead characteristic of an essentially square-planar Co²⁺ complex that lacks any significant axial bonding interactions (28, 40, 41).

Co²⁺Cbi⁺ ↔ LrPduO Interactions

(A) Co²⁺Cbi⁺ + LrPduO

The low-temperature Abs and MCD spectra of Co²⁺Cbi⁺ in the absence and presence of LrPduO are very similar to each other (cf Figures 4A and 4B). However, while the addition of LrPduO to the Co²⁺Cbi⁺ solution has no effect on the peak position of the α-band (Table 2), it causes a small but notable broadening of several features in the 16000 – 21000 cm⁻¹ region of the MCD spectrum. Similarly, the EPR spectrum of Co²⁺Cbi⁺ is slightly perturbed when LrPduO is present, especially between 2800 and 3100 G (cf Figures 5A and 5B). Collectively, these data suggest that Co²⁺Cbi⁺ binds to LrPduO even in the absence of co-substrate ATP, but retains a five-coordinate Co²⁺ center with an axially bound water molecule. Interestingly, our simulation of the Co²⁺Cbi⁺+ LrPduO EPR spectrum reveals that g₂ ≠ g₃ for this species, indicating that the Co²⁺ ion is in a ligand environment of rhombic symmetry, while for free Co²⁺Cbi⁺ an axially symmetric spectrum with g₂ = g₃ is observed (see Table 3).

Figure 4.

Abs (gray traces, right axis) and 7 T MCD (black traces, left axis) spectra collected at 4.5 K of (A) free Co²⁺Cbi⁺, (B) Co²⁺Cbi⁺ in the presence of LrPduO, and (C) Co²⁺Cbi⁺ in the presence of the LrPduO/ATP complex. The peak position of the dominant Abs feature (the α-band) is indicated by the vertical solid line.

Figure 5.

EPR spectra collected at 20 K of (A) free Co²⁺Cbi⁺, (B) Co²⁺Cbi⁺ in the presence of LrPduO, and (C) Co²⁺Cbi⁺ in the presence of the LrPduO/ATP complex. Spectrum D was obtained by subtracting 50% of spectrum B from spectrum C. Spectral simulations (thin lines) were performed using the parameters provided in Table 3.

(B) Co²⁺Cbi⁺ + LrPduO/ATP

As in the case of Co²⁺Cbl + LrPduO/ATP (Figure 2), a significant blue-shift of the α-band (Table 2) and the appearance of prominent MCD features at 12310 and 20530 cm⁻¹ are observed when Co²⁺Cbi⁺ is incubated with LrPduO complexed with the co-substrate ATP (cf Figures 4A and 4C). Again, the conversion to a four-coordinate Co²⁺corrinoid species is incomplete, based on the observation of weak positive features in the 16000 – 20000 cm⁻¹ region of the Co²⁺Cbl + LrPduO/ATP MCD spectrum that are characteristic of five-coordinate Co²⁺Cbi⁺ (Figure 4A).

As anticipated on the basis of our MCD data in Figure 4, the EPR spectrum of Co²⁺Cbi⁺ + LrPduO/ATP (Figure 5C) has contributions from two distinct paramagnetic species, an effectively four-coordinate Co²⁺Cbi⁺ species (responsible for the widely spread resonances in the low-field region) and five-coordinate Co²⁺Cbi⁺ possessing an axially bound water molecule (see Figure 5B). To obtain a pure spectrum of the four-coordinate Co²⁺Cbi⁺ fraction, the suitably scaled (×0.5) Co²⁺Cbi⁺ + LrPduO EPR spectrum (Figure 5B) was subtracted from the composite spectrum in Figure 5C. Although the resulting spectrum (Figure 5D) is qualitatively very similar to that exhibited by the four-coordinate Co²⁺Cbl fraction bound to the LrPduO/ATP complex (Figure 3D), the g_2,3 and A(Co) values are in fact somewhat different (Table 3). This result suggests that even though the DMB moiety no longer serves as an axial ligand in the four-coordinate Co²⁺Cbl + LrPduO/ATP fraction, it nonetheless modulates the geometric and electronic structures of the Co²⁺ center to a small but noticeable degree, presumably via hydrogen-bonding to amino-acid residues near the enzyme active site.

DISCUSSION

LrPduO can adenosylate Co¹⁺Cbl⁻ and Co¹⁺Cbi

Thus far, only the CobA-type adenosyltrasferases were believed to participate in the adenosylation of complete as well as incomplete corrinoids (42). However, the kinetic data obtained in this study provide evidence that in vitro a PduO-type adenosyltransferase can utilize incomplete corrinoids as alternative substrates, in support of a previous proposal that was based on in vivo evidence (32) Interestingly, LrPduO adenosylates Co¹⁺Cbl⁻ and Co¹⁺Cbi with approximately the same catalytic efficiency, as judged on the basis of the similar k_cat/K_m values (Table 1). Additionally, under saturating ATP conditions the apparent K_m of LrPduO for Co¹⁺Cbi is nearly identical to that reported for Co¹⁺Cbl⁻ and at least 10-fold lower than for any previously reported adenosyltransferase (32). These results suggest that the nucleotide tail (Figure 1) is relatively unimportant for corrinoid binding to the enzyme active site and for product formation. It is important to note, though, that chemically reduced Co¹⁺corrinoids were used in our activity assays. Thus our kinetic data do not rule out the possibility that the nucleotide tail may be required for allowing the physiological electron donor to accomplish the Co²⁺ → Co¹⁺corrinoid reduction. However, such a requirement is quite unlikely, based on the close resemblance of our spectroscopic data obtained for Co²⁺Cbl and Co²⁺Cbi⁺ bound to the LrPduO/ATP complex (see below) and their obvious similarities with those reported for the CobA adenosyltransferase (28).

Interaction between Co²⁺corrinoids and LrPduO in the absence of co-subtrate ATP

The fact that the Abs and MCD spectra of Co²⁺Cbl and Co²⁺Cbi⁺ change only slightly when LrPduO is added (Figures 2–5) indicates that the geometric and electronic structures of these species are largely preserved in the presence of the enzyme. Nevertheless, small but noticeable changes are observed in each case, signaling that Co²⁺corrinoid binding to the enzyme active site does occur even in the absence of co-substrate ATP. Specifically, when a solution of Co²⁺Cbl is incubated with LrPduO, a significant fraction (~30%) of the cofactor converts to the base-off form, in which the Co²⁺ center is axially ligated by a water molecule instead of the DMB moiety (Figure 1) (35). Alternatively, the addition of LrPduO to a solution of Co²⁺Cbi⁺ causes a small rhombic splitting of the EPR signal (Table 3), possibly reflecting a perturbation of the equatorial ligand environment of the Co²⁺ ion; e.g., via hydrogen-bonding of active-site residues to the peripheral side chains of the corrin macrocycle.

Interaction between Co²⁺corrinoids and the LrPduO/ATP complex

While the spectroscopic properties of Co²⁺Cbl and Co²⁺Cbi⁺ are only weakly affected by the addition of LrPduO, rather drastic spectral changes are observed when the enzyme was pre-incubated with the co-substrate ATP (Figures 2–5). In each case, binding of the Co²⁺corrinoid substrate to the LrPduO/ATP complex leads to a substantial blue-shift of the dominant Abs feature in the visible spectral region, termed the α-band (Table 2), which corresponds to the lowest-energy corrin-centered π → π* transition (43–45). Previous studies revealed that the corrin π-based donor orbital involved in this transition also possesses some Co d_z² character, the extent of which is governed by the σ-donor strength of the axial ligand. As a result, the α-band of Co²⁺corrinoids shifts to higher energy with decreasing electron donating ability of the axial ligand (e.g., from Co²⁺Cbl to Co²⁺Cbi⁺, see Table 2) or a weekending of the axial bonding interaction (35, 46).

The large blue-shift of the α-band in response to Co²⁺Cbl and Co²⁺Cbi⁺ binding to the LrPduO/ATP complex is accompanied by drastic changes to the corresponding MCD spectra, most notably the appearance of a prominent low-energy feature at ~12340 cm⁻¹ (Figures 2 and 4), as well as to their EPR data, in particular a large increase in the g₂, g₃, and A(Co) values (Table 3). Collectively, these spectral perturbations are consistent with a sizeable stabilization of the singly occupied Co 3d_z²-based molecular orbital (28) owing to a significant weakening of the axial ligand–Co²⁺corrinoid bonding interaction, so as to generate essentially square-planar Co²⁺ species in the enzyme active site. Although these “activated” forms of Co²⁺Cbl and Co²⁺Cbi⁺ bound to the LrPduO/ATP complex are best described as possessing an effectively four-coordinate Co²⁺ center that lacks any significant axial bonding interactions, the corresponding MCD and EPR spectra (Figures 2–5) exhibit small but notable differences. Specifically, when the alternative substrate Co²⁺Cbi⁺ is used instead of Co²⁺Cbl, the prominent MCD feature in the near-IR spectral region displays a minor red-shift (by ~30 cm⁻¹, Table 4) and the EPR g and A(Co) values become more rhombic (Table 3). These observations suggest that even though the DMB moiety is displaced from the Co²⁺ ion in the Co²⁺Cbl + LrPduO/ATP complex, it may still play a role in organizing the enzyme active site for maximum catalytic activity (e.g., by promoting Co²⁺ → Co¹⁺ reduction).

Table 4.

Peak position of the prominent MCD feature in the near-IR spectral region observed for Co²⁺corrinoids bound to several alkyltransferase/ATP complexes

Co2+Cbl + LrPduO/ATP	Co2+Cbi+ + LrPduO/ATP	Co2+Cbi+ + CobA/ATPa	Co2+Cbl + hATR/ATPb
12340 cm−1 (810.0 nm)	12310 cm−1 (812.5 nm)	12340 cm−1 (810 nm)	12580 cm−1 (795 nm)

^aFrom ref (28).

^bFrom ref (29).

The spectroscopic data obtained in this work nicely complement the information that has recently been obtained from structural and kinetic studies of LrPduO. First, Mera et al. prepared numerous variants of LrPduO that were characterized by using kinetic and X-ray crystallographic techniques to develop a better understanding of the function of several conserved residues in the enzyme active site (47). Importantly, these studies revealed that substrate inhibition occurs at subsaturating concentrations of ATP, in strong support of an ordered binding scheme according to which the enzyme must bind co-substrate ATP prior to the corrinoid substrate. Second, in the course of this investigation, St. Maurice et al. succeeded in obtaining an X-ray crystal structure (at 1.9 Å resolution) of Co²⁺Cbl bound to the LrPduO/ATP complex (48). As expected on the basis of our spectroscopic data reported here, this structure provides visual evidence for the formation of a base-off, four-coordinate Co²⁺Cbl intermediate in the catalytic cycle of LrPduO. This structure also offers clues as to how the enzyme imposes such a unique coordination environment on the Co²⁺corrioind substrate. Specifically, it shows that the N-terminus of LrPduO that is disordered in the substrate-free enzyme (32) becomes ordered upon ATP-binding, thereby inducing a distinct active-site conformation that is poised for Co²⁺corrinoid binding (48). Moreover, a conserved Phe residue moves to within 3.8 Å of the Co²⁺ ion so as to promote dissociation of the axially bound 22 solvent ligand that would normally be present in base-off Co²⁺Cbl. Although not obvious from the LrPduO crystal structures, our MCD and EPR data indicate that the formation of such a four-coordinate Co²⁺Cbl intermediate leads to a significant stabilization of the redox-active Co 3d_z2 orbital that is oriented toward the axial coordination sites, thereby raising the Co^2+/1+ reduction midpoint potential by an estimated 250 mV (28).

Comparison to other adenosyltransferases

The involvement of unprecedented, essentially four-coordinate Co²⁺corrinoid intermediates in the catalytic cycles of adenosyltransferases has originally been proposed on the basis of spectroscopic and computational studies of CobA from Salmonella enterica and the human enzyme hATR (28, 29). As expected by taking into consideration that hATR is a PduO-type adenosyltransferase, Co²⁺Cbl interacts similarly with this enzyme and LrPduO. Specifically, while binding of Co²⁺Cbl to hATR in the absence of ATP leads to a partial (~40%) conversion to the base-off form, a significant blue-shift of the α-band and the appearance of the MCD and EPR spectroscopic signatures characteristic of an approximately four-coordinate Co²⁺corrinoid species are observed when hATR was pre-incubated with co-substrate ATP (29). However, the prominent low-energy feature in the MCD spectrum of Co²⁺Cbl + hATR/ATP is blue-shifted by 240 cm⁻¹ relative to that of Co²⁺Cbl + LrPduO/ATP (Table 4), signifying a slightly stronger residual axial bonding interaction in the former species. Moreover, a comparison of the corresponding EPR spectra reveals that the conversion from the five- to the four-coordinate Co²⁺Cbl species is more complete in the case of hATR (~90%) than for LrPduO (~40%). Collectively, these results suggest that a stronger interaction could exist between the dissociated DMB moiety of the “activated” Co²⁺Cbl species and the enzyme active site in hATR than in LrPduO. It is tempting to speculate, then, that due to this increased interaction with the nucleotide loop, hATR may exhibit an increased substrate specificity and thus be unable to adenosylate incompletely assembled corrinoids, such as Co²⁺Cbi⁺. Although to our knowledge the range of possible corrinoid substrates for hATR has not yet been established, it is interesting to note that the PduO-type adenosyltransferase from S. enterica appears to be specific for AdoCbl synthesis, despite its high sequence similarity with LrPduO (49).

In the case of CobA, for which the natural substrate is a Co²⁺Cbl precursor that lacks the nucleotide loop and thus more closely resembles Co²⁺Cbi⁺ (16, 50), MCD and EPR spectroscopic studies revealed that in the Co²⁺Cbi⁺ + CobA/ATP complex, a predominant fraction (~70%) of four-coordinate Co²⁺corrinoid species is formed (28). When the alternative substrate Co²⁺Cbl is added to the CobA/ATP complex, however, the conversion to a similar four-coordinate Co²⁺ species is largely suppressed (to ~10%). While hATR and CobA are thus relatively specific for their respective substrates, the results obtained in this study reveal that a sizeable a fraction of a four-coordinate Co²⁺corrinoid species is formed regardless of whether Co²⁺Cbl or Co²⁺Cbi⁺ binds to the LrPduO/ATP complex (40% and 50%, respectively). Although this low substrate-specificity of LrPduO is consistent with our kinetic data reported in the Results and Analysis section (Table 1), this finding is rather unexpected, because among the three types of adenosyltransferases found in Salmonella enterica, CobA has been presumed to be the only enzyme capable of adenosylating incomplete corrionid species (7, 8, 16, 28). Yet, the nearly identical positions of the prominent near-IR features in the MCD spectra of Co²⁺Cbi⁺ + LrPduO/ATP and Co²⁺Cbi⁺ + CobA/ATP (Table 4) suggest that Co²⁺Cbi⁺ does in fact interact similarly with the LrPduO/ATP and CobA/ATP complexes. Despite these similarities, a more complete conversion to a four-coordinate Co²⁺Cbi⁺ species occurs in the latter enzyme (50% and 70% for LrPduO and CobA, respectively).

Although small differences exist among the spectroscopic signatures of the formally four-coordinate Co²⁺corrinoid species that have now been identified in the active sites of various adenosyltransferases (28, 29), the general strategy of these enzymes evidently involves the formation of an essentially square-planar Co²⁺corrinoid intermediate, so as to reduce the thermodynamic barrier for the reduction to the Co¹⁺ state through stabilization of the redox-active, Co 3d_z²-based molecular orbital that is oriented along the axial coordination sites of the Co²⁺ ion. Remarkably, for each adenosyltransferase studied to date (28, 29), this “activated” Co²⁺corrinoid intermediate is formed in high yield only in the presence of co-substrate ATP, which provides a mechanism for protecting the enzyme active site from being attacked by the transiently formed Co¹⁺corrinoid “supernucleophile”. Collectively, the insights gained in this study and the recent structural and kinetic characterizations of LrPduO provide an excellent basis for further investigations into the fascinating bio-organometallic reaction catalyzed by PduO-type adenosyltransferases.