Glutamate mutase is one of a group of adenosylcobalamin-dependent enzymes that use free radicals to catalyse unusual and chemically difficult rearrangements involving 1,2-migrations of hydrogen atoms. A key mechanistic feature of these enzymes is the transfer of the migrating hydrogen atom between substrate, coenzyme and product. The present review summarizes recent experiments from my laboratory that have used rapid chemical quench techniques to identify intermediates in the reaction and probe the mechanism of hydrogen transfer through a variety of pre-steady-state kinetic isotope effect measurements.
Approximately a dozen AdoCbl (adenosylcobalamin; coenzyme B12)-dependent enzymes are known that catalyse unusual 1,2-isomerization reactions in which a hydrogen atom on one carbon is interchanged with either an -OH or -NH2 group, or a carbon-containing fragment on an adjacent carbon [1–6]. The first step in the mechanism of all these enzymes is homolysis of the unique cobalt–carbon bond of AdoCbl to generate a 5′-deoxyadenosyl radical. This radical then abstracts a hydrogen atom from the substrate to generate 5′-deoxyadenosine and a substrate radical, which is thereby activated to undergo subsequent rearrangement (Figure 1).
Structure of AdoCbl and the minimal mechanistic scheme describing the rearrangements catalysed by AdoCbl-dependent isomerases
AdoCbl-dependent enzymes constitute one group in a much larger class of enzymes that use carbon-based radicals to effect a variety of chemically difficult reactions involving the cleavage of carbon–carbon, carbon–oxygen and carbon–nitrogen bonds [7–11]. These include enzymes that use S-adenosylmethionine to generate adenosyl radical and methionine through a one-electron reduction of S-adenosylmethionine mediated by an iron–sulfur cluster [12,13], and enzymes that use a glycyl radical on the backbone of the protein to initiate catalysis .
The unusual reactions catalysed by AdoCbl and other radical enzymes have intrigued bio-organic, biophysical and bioinorganic chemists, and over the last decade, our understanding of both the mechanism and structure has advanced significantly. Mechanistic studies in my laboratory have focused on glutamate mutase [2,15], the first AdoCbl enzyme to be discovered , and more recently on the glycyl radical enzyme benzylsuccinate synthase [17–19]. The isomerization of L-glutamate to L-threo-3-methylasparate catalysed by glutamate mutase (Figure 2) provides a model system to investigate how enzymes harness the intrinsic reactivity of free radicals towards catalysis, and exemplifies many mechanistic features common to other AdoCbl, glycyl radical and S-adenosylmethionine radical enzymes. This brief review focuses on recent pre-steady-state kinetic experiments designed to test the mechanism of glutamate mutase.
Isomerization of L-glutamate to L-threo-3- methylasparate catalysed by glutamate mutase
The mechanism of glutamate mutase
A mechanism for the isomerization of glutamate to methylaspartate, which is fully reversible, catalysed by glutamate mutase, is presented in Figure 3. The catalytic cycle is initiated by homolysis of the AdoCbl cobalt–carbon bond to generate a 5′-deoxyadenosyl radical (II). The adenosyl radical then abstracts the migrating hydrogen from the substrate to give a glutamyl radical and 5′-deoxyadenosine (III). Next, the adenosyl radical undergoes rearrangement to form the methylaspartyl radical (V); this occurs through a fragmentation and recombination mechanism with glycyl radical and acrylate as intermediates (IV). The methylaspartyl radical then re-abstracts a hydrogen from 5′-deoxyadenosine to give the product and regenerate the adenosyl radical (VI). Finally, the adenosyl radical recombines with the cobalt to regenerate the coenzyme (VII), thereby completing the catalytic cycle.
The mechanism for the reversible rearrangement of L-glutamate to L-threo-3-methylaspartate catalysed by glutamate mutase
Homolysis of AdoCbl
Homolysis of AdoCbl to produce cob(II)alamin and adenosyl radical is accompanied by characteristic changes in the electronic spectrum of the coenzyme that can be monitored by stopped-flow spectrometry. Early in our investigation of glutamate mutase, we studied the kinetics of AdoCbl homolysis in response to substrate binding . We found that, surprisingly, homolysis of AdoCbl proceeded more slowly when the substrate (glutamate or methylaspartate) was deuterated. If adenosyl radical is a discrete intermediate, isotopic substitution is not expected to affect the rate of substrate-triggered homolysis, because this step does not formally involve hydrogen abstraction from the substrate.
This paradox is resolved by recognizing that for a reversible reaction, the individual steps in the mechanism form a series of kinetically coupled equilibria, so perturbations in one step propagate to adjacent steps. In this case, homolysis of AdoCbl is energetically unfavourable, so that adenosyl radical and cob(II)alamin are formed only at very low concentrations in a rapidly established but unfavourable equilibrium. The subsequent (favourable) reaction with substrate depletes adenosyl radical and thus displaces the equilibrium towards homolysis. When the substrate is deuterated, a kinetic isotope effect causes it to react more slowly with the adenosyl radical; therefore the equilibrium is established more slowly and gives rise to an apparent isotope effect on AdoCbl homolysis. Importantly, this observation implies that adenosyl radical must be a high-energy intermediate that does not accumulate on the enzyme.
Rapid chemical-quench studies
Apart from cob(II)alamin, the other reaction intermediates are spectrally indistinguishable; therefore in delving into the mechanism of glutamate mutase, we have made extensive use of rapid chemical-quench methods to establish the identity of proposed intermediates in the reaction and investigate their kinetics of formation . This technique uses apparatus that rapidly mixes two reactant solutions (e.g. enzyme and substrate) and after a short time interval (milliseconds to seconds) mixes a third quenching solution to stop the reaction; often a strong acid is used to denature the protein. These experiments should carry the following health warning: ‘This technique uses much more enzyme than spectroscopic methods and has the added disadvantages that kinetic data are obtained discontinuously; furthermore, analysis of each data point is time consuming and labour-intensive’. Nevertheless, with careful experimental design, including appropriate internal standards, accurate and insightful data can be obtained on enzyme intermediates. One advantage is that the identity of intermediates can be established by chemical analysis; in particular, MS of isolated intermediates has proved especially valuable for measuring pre-steady-state deuterium isotope effects.
Identification of reaction intermediates
The mechanism in Figure 3 predicts that 5′-deoxyadenosine and acrylate are formed during turnover as stable (if transient) intermediates. Using rapid quench methods, we demonstrated that both intermediates are, indeed, formed during the reaction at rates high enough for them to be kinetically competent [22,23]. The fact that 5′-deoxyadenosine was an intermediate was uncontroversial; however, the mechanism by which the substrate and product radicals rearranged had been the subject of considerable speculation and debate because chemical model studies had failed to replicate this aspect of the mechanism. By using 14C-labelled glutamate as substrate and isolating [14C]acrylate and the quenched glycyl radical as [14C]glycine, we were able to unambiguously establish the mechanism .
Pre-steady-state measurements of isotope effects
The isomerizations catalysed by AdoCbl-dependent enzymes involve two hydrogen transfer steps: one from substrate to coenzyme and the other from coenzyme to product; therefore steady-state kinetic isotope effect measurements, which encompass both steps, are hard to interpret. For glutamate mutase, rapid quench methods have allowed us to measure deuterium and tritium isotope effects for hydrogen transfer between either glutamate or methylaspartate and the coenzyme. These have provided insights into the free-energy profile of the overall reaction and the transition states for the individual hydrogen transfer steps.
Tritium partitioning experiments
The partitioning of hydrogen isotopes from an enzyme-bound intermediate between products and substrates provides an elegant way of investigating the relative heights of energetic barriers in an enzyme-catalysed reaction. Using tritium-labelled AdoCbl in the 5′-(exchangeable) position, we examined how tritium partitioned between substrate and product when the reaction was run either in the forward direction (glutamate to methylaspartate) or in the reverse direction (methylaspartate to glutamate) . As illustrated in Figure 4, once substrate radical and tritiated 5′-deoxyadenosine are generated, the reaction may either proceed forward to product or reverse back to substrate, thereby carrying tritium either into the product or into the substrate. The distribution of tritium depends on the relative heights of the energy barriers leading to product and substrate. It is important to ensure that the transfer of tritium from the coenzyme to either the substrate or product is effectively irreversible; otherwise, if tritium makes multiple passages through the enzyme, the partitioning information is lost. The rapid quench technique allows the reaction to be stopped after very short time intervals (shorter than a single turnover) so that only the initial tritium transfer event is captured.
Tritium partitioning experiments to probe the free-energy profile of glutamate mutase
Our experiments found that tritium partitions in an almost 1:1 ratio between substrate and product . This indicates that the energetic barriers for transfer of tritium from 5′-deoxyadenosine to glutamyl radical and methylaspartyl radicals are of equal height. Furthermore, the partitioning ratio is essentially independent of the direction in which the reaction is proceeding, implying that the interconversion of glutamyl and methylaspartyl radicals is fast relative to tritium transfer. (If the interconversion of radicals was very slow, most of the tritium would partition back to the substrate, regardless of the direction of the reaction; a phenomenon we have observed with a mutant glutamate mutase enzyme .) The wild-type glutamate mutase appears to have optimized the free-energy profile so that no single step is significantly rate-determining.
Secondary isotope effect measurements
More recently, we have used rapid quench pre-steady-state techniques to measure the secondary tritium kinetic isotope effects associated with homolysis of AdoCbl and abstraction of hydrogen from the substrate [26–28]. Secondary isotope effects are sensitive to changes in bond stiffness during reactions and as such can provide valuable insights into transition states . These measurements have to be made in the pre-steady-state regime, otherwise the isotope effect information is lost during turnover, making these very challenging experiments. To accomplish this, we used 14C,5′-3H-labelled AdoCbl and examined how the 3H/14C ratio changes as 5′-deoxyadenosine is formed during a single turnover on the enzyme (Figure 5). The apparent isotope effect extrapolated to t=0 represents the secondary kinetic isotope effect, whereas the apparent isotope effect extrapolated to t=∞ represents the secondary equilibrium isotope effect.
Secondary tritium isotope effects uncover coupled motion and hydrogen tunnelling in glutamate mutase
The secondary equilibrium tritium isotope effect reports on both breaking of the AdoCbl cobalt–carbon bond and transfer of hydrogen from substrate to 5′-deoxyadenosine, whereas the secondary kinetic isotope effect is dominated by the latter step, which primary isotope effect measurements indicate is relatively slow . Surprisingly, a large inverse equilibrium isotope effect of 0.72±0.04 was found for the overall reaction, indicating that the 5′-C–H bonds become significantly stiffer in going from AdoCbl to 5′-deoxyadenosine, even though the 5′-carbon remains formally sp3 hybridized . The kinetic isotope effect for the formation of 5′-deoxyadenosine was 0.76±0.02, which is usually interpreted as a ‘late’ transition state for the reaction .
However, when we repeated these measurements using glutamate deuterated in the position of hydrogen abstraction, so that 5′-deutero-deoxyadenosine was formed, we noticed something very strange. The effect of introducing a primary deuterium kinetic isotope effect on the hydrogen transfer step was to deflate the secondary kinetic isotope effect to a value close to unity, 1.05±0.08, whereas the equilibrium secondary isotope effect remained unchanged . Normally, isotope effects obey the ‘Rule of the Geometric Mean’, which basically states that there are no isotope effects on isotope effects .
A detailed discussion of the reason for the rule, and why it breaks down, is beyond the scope of this review; however, the breakdown of the rule may be interpreted as follows. Reduction in the secondary kinetic isotope effect is consistent with concerted motion in the transition state of the 5′-hydrogen atoms adjacent to the hydrogen that is transferred between substrate and coenzyme (illustrated in Figure 5), in a reaction that involves a large degree of quantum tunnelling. Note that theoretical studies indicate that it is necessary to have both coupled motion and quantum tunnelling to satisfactorily account for the dependence of the secondary isotope effects on the primary isotope effect . Furthermore, coupled motion also has the effect of reducing the primary kinetic isotope effect, which is often much larger than the semi-classical limit of ∼7 when quantum tunnelling is involved, so that even if hydrogen tunnelling is occurring, the deuterium isotope effect may be well within the semi-classical range.
Intrinsic kinetic isotope effect measurements
Diagnosing hydrogen tunnelling in enzymes requires measuring the intrinsic kinetic isotope effect ; however, very often, even using pre-steady-state techniques, this is not possible because other chemical steps obscure the isotopically sensitive step and the isotope effect is suppressed (i.e. measured as smaller than its true value). Therefore we devised an experiment to measure the intrinsic kinetic isotope effect for hydrogen atom transfer from methylaspartate to 5′-deoxyadenosine catalysed by glutamate mutase [33,34]. The experiment takes advantage of the fact that for hydrogen transfer from methyl groups the intrinsic deuterium isotope effect can be measured by specifically labelling the methyl carbon with one or two deuterium atoms and analysing the isotopic composition of the reaction products . The isotope effect can be measured, even when the isotopically sensitive step is not rate-determining, because it is manifested though intramolecular competition between protium and deuterium atoms, which remain chemically equivalent even in the enzyme active site owing to the rapid rotation of the methyl group.
The principle of the experiment is shown in Figure 6. Glutamate mutase is reacted with methylaspartate specifically monodeuterated in the methyl group. During the reaction, 5′-deoxyadenosyl radical, generated by homolysis of AdoCbl, is confronted with the choice of abstracting either protium or deuterium from the methyl group of the same substrate molecule. Hydrogen or deuterium abstraction generates methylaspartyl radical that rapidly rearranges to the much more stable glutamyl radical, so that at sufficiently short times (<100 ms) the reaction is effectively irreversible and the intrinsic isotope effect is manifested.
Scheme illustrating how the intrinsic deuterium isotope effect for hydrogen transfer from methylaspartate to 5′-deoxyadenosine can be measured by intramolecular competition between hydrogen and deuterium on the methyl group of methylaspartate
The deuterium content of the 5′-deoxyadenosine was determined by ultra-high-resolution Fourier transform ion cyclotron MS. The 5′-deoxyadenosine isolated from the reaction contains a mixture of isotopomers, with contributions from natural abundance 13C and 15N isotopes in addition to any deuterium from the substrate. Normally, the presence of 13C complicates the analysis of deuterium kinetic isotope effects because the peaks due to deuterated and 13C-containing molecules overlap. However, the resolution of this type of instrument is sufficiently high that peaks due to 13C-, 2H- and 15N-containing molecules are resolved, removing this complication.
The intrinsic deuterium isotope effect measured by our experiment is quite modest: 4.1. This contrasts with much larger isotope effects (in the range of 30–50) measured for other B12 enzymes such as methylmalonyl-CoA mutase [36–39] and B12 model systems (measured by different techniques) that have been attributed to extensive hydrogen tunnelling. This argues strongly for a role for glutamate mutase in modulating the transition state for hydrogen transfer, and thereby changing the isotope effect. However, as mentioned above, extensive coupling of the motions of the primary and secondary hydrogen atoms will result in the primary isotope effect remaining quite small even if extensive hydrogen tunnelling is occurring. Further isotope effect measurements are in progress in our laboratory to better explain the contribution hydrogen tunnelling makes to catalysis in glutamate mutase.
Enzyme Mechanisms: Fast Reaction and Computational Approaches: Biochemical Society Focused Meeting held at Manchester Interdisciplinary Biocentre, U.K., 9–10 October 2008. Organized and Edited by Nigel Scrutton and Andrew Munro (Manchester, U.K.).
I acknowledge the contribution of the talented graduate students and postdoctoral scientists who have contributed to the work described here. Among them are Daniel Holloway, Hao-Ping Chen, Hung-Wei Chih, Prashanti Madhavapeddi, Ipsita Roymoulik, Marja Huhta, Mou-Chi Cheng, Li Xia, Anjali Patwardhan, Hyang-Yeol Lee and Miri Yoon. I have also benefited from productive collaborations with Professor Dave Ballou, Professor Roseanne Sension and Professor Kristina Håkansson at the University of Michigan.
This work was partially supported by the National Institutes of Health [grant number GM 59227].