Bio-inspired approaches to accelerating metal ion-promoted reactions: enzyme-like rates for metal ion mediated phosphoryl and acyl transfer processes

Introduction

Over 300 Zn(II)-containing enzymes are known to be involved in catalyzing a variety of important biological processes involving acyl or phosphoryl transfer processes such as hydrolysis (peptidases like the amino- and carboxypeptidases and class B β-lactamases), hydration (carbonic anhydrases), and phosphoryl transfer to various acceptors (nucleases and phosphatases) [1–7]. Some of these contain one Zn ion in the active site (e.g., carboxypeptidase, thermolysin, carbonic anhydrase) while others contain two or more (e.g., class B β-lactamase, phospholipase C, phosphotriesterase). The common motifs for the Zn(II) forms in the active sites contain a mononuclear Zn(II):(⁻OH) or dinuclear Zn(II):(⁻OH):Zn(II) as schematized in 1 or 2, where the metal associated hydroxide acts as a base or nucleophile and the metal ions are bound by three or more additional enzyme-based ligands. Not surprisingly, the apparent simplicity of these motifs has prompted numerous researchers to create reductionist biomimetic catalysts for the decomposition of carboxylate amides and esters, phosphate diester RNA and DNA models, and phosphate triesters (for compendia of references pertaining to such models see refs. [8–10]). However, as a somewhat general rule, the majority of the models are not very good at catalyzing the hydrolytic reactions of the test substrate and in fact most are not better than hydroxide with several being far worse, although there are some notable exceptions [11–18]. It might be expected that a hydroxide coordinated to an electropositive metal ion is less basic and nucleophilic than free hydroxide, thus accounting for some of the instances where the M^x+:(⁻OH) is less active than hydroxide. However, in order to achieve a larger activity than hydroxide alone it is generally accepted that a cooperative interaction is required where the metal ion exerts additional roles such as a Lewis acid to activate the substrate and also to deliver its coordinated hydroxide as a nucleophile or base in a transient substrate:M^x+:(⁻OH) complex [19]. Nevertheless, despite the intense effort that has been directed at this problem, the lack of examples of really active man-made catalysts suggests that other effects, not yet appreciated, might be operative.

Our initial foray into this area was inspired by works suggesting that “enzyme active sites are non-aqueous, and the effective dielectric constants resemble those in organic solvents rather than that in water” [20, 21]. The idea of using reduced polarity solvents to enhance some biologically important reactions is not new. For example, it is known that the decarboxylation of pyruvate promoted by 3,4-dimethylthiazolium ion (a model for thiamine pyrophosphate catalyzed reactions in enzymes) occurs 10⁴–10⁵ faster in ethanol than in water [22]. In addition, Stockbridge and Wolfenden recently demonstrated that the hydrolyses of bis-neopentyl phosphate (3) and mono-neopentyl phosphate (4) in wet cyclohexane are accelerated by 10⁹ and 10¹² times respectively relative to their hydrolyses in water [23]. The large accelerating effect of the low polarity solvent in these examples was attributed to desolvation of the ground state and better solvation of a charge-dispersed transition state. These examples suggest that similar polarity and solvation effects might be important contributors to the enhanced reactivity in the active sites of enzymes. However, with the exception of Yatsimirski’s studies [24–27], this concept has not received extensive investigation in the cases of metal ion catalyzed solvolytic reactions of anionic phosphate and neutral carboxylate substrates although these are just the sort of reactions where one expects to see profound medium effects on binding and reactivity. Among the organic solvents, the light alcohols, methanol and ethanol, have structure and properties closest to water, and their dielectric constants (D_r = 31.5 and 24.3 respectively) are substantially lower than that of water (D_r = 78) [28]. Importantly, these two solvents also retain H-bonding characteristics and so are capable of solvating both anionic and cationic entities relatively well.

One of our first examples comparing solvent effects involved the La³⁺(⁻OR)-promoted solvolysis of paraoxon (5) in water and methanol as in eq. (1) [25]. In aqueous solution the second order rate constants for the La³⁺- and hydroxide-promoted reactions are similar at 7 × 10⁻³ dm³ mol⁻¹ s⁻¹ and 6 × 10⁻³ dm³ mol⁻¹ s⁻¹ respectively. However, in methanol the metal ion is far more reactive than methoxide alone, the second order rate constants for the catalyzed methanolysis being 23.2 and 1 × 10⁻² dm³ mol⁻¹ s⁻¹ respectively. It subsequently became clear to us that, in general, the M^x+:(⁻OR) forms in the light alcohols are far more reactive than methoxide or ethoxide alone and, as we will see, in some cases the ratios of k^{(Mx+:(−OR))}/k^(−OR) are as large as 10⁸.

The switch from water to alcohol solvent provides at least four important effects that greatly accelerate the M^x+:(⁻OR) catalyzed phosphoryl and acyl transfer reactions that are the subject of the present account. These are: a) an increased electrostatic ion-ion and ion-dipole association of the metal ion and oppositely charged or polarized substrates; b) stronger binding between M^x+ and ligands, stabilizing the transiently formed or fully formed complexes in solution; c) an enhanced solubility of metal ions at ‘pH’ values above the ‘pK_a’ of the M^x+(HOR)_n ←→ M^x+(⁻OR)(HOR)_m + H⁺ acid dissociation process; and, d) a medium effect accelerating the reactions where charge is dispersed in transition states of a transforming substrate:M^x+(⁻OR) complex. As will become evident from the examples below, these factors and others where significant assistance of the departure of leaving groups (known as leaving group assistance or LGA) is observed due to their interaction with the electropositive metal, give highly active systems that facilitate the normally very slow reactions of phosphoryl and acyl transfer.

The cleavage of a series of 2-hydroxypropyl aryl phosphates (RNA models) promoted by a dinuclear Zn(II) catalyst in methanol and ethanol

The kinetics of the methoxide-promoted cyclization reactions of a series of 2-hydroxylpropyl aryl phosphates (6a–g) where the aryl groups have a series of decreasing electron withdrawing substituents were studied in methanol. In a typical example (eq. (2)) with the p-nitro derivative 6a, the methoxide promoted deprotonation of the 2-hydroxy group leads to cyclization with expulsion of p-nitrophenol (or phenoxide depending on the solution ‘pH’) and the corresponding cyclic phosphate diester (7). The k₂^OMe for this process is 2.6 × 10⁻³ dm³ mol⁻¹ s⁻¹ at 25 °C [29, 30].

A subsequent study of the kinetics of the cyclization of 6a–g promoted by the dinuclear Zn(II) complex 8 revealed that the general mechanism involves saturation binding of the substrate to the complex followed by unimolecular decomposition of a 8:6 complex to give the corresponding phenoxide products and the cyclic ester 7 [31] as in eq. (3) where K_M refers to the dissociation constant of the 8:6 complex and k_cat refers to the unimolecular rate constant for formation of product. Given in Fig. 1 is a Brønsted plot of the lg(k₂) or lg(k_cat/K_M) vs. pssKa of the parent phenol for the catalyzed cyclization of 6a–g which exhibits two important features (the super and subscript ‘s’ refers to a pK_a that is determined in methanol and referenced to that solvent [32]). There is a plateau (slope = 0) where the limiting second order rate constant for the reaction of substrates with good leaving groups is ∼275 000 dm³ mol⁻¹ s⁻¹. With poorer substrates having a higher pssKa for the parent phenol, the gradient of the plot is −1.1, indicating a large dependence of the rate of cyclization on the goodness of the leaving group. The fact that the overall plot is concave downward indicates that there is a change in rate-limiting step as a function of parent phenol pssKa. This is interpreted as a transition from a step dependent on the rate of binding for substrates with good leaving groups, changing to a chemical step involving the actual cleavage reaction when the leaving group gets poorer.

Fig. 1:

A Brønsted plot of the lg(k₂) = lg(k_cat/K_M) vs. pssKa (of the parent phenol) for the cyclization of 6a–g at 25 °C.

When the solvent is changed to ethanol the cleavage of 6a–g mediated by 8 (this time having a coordinated ethoxide) is also rapid but there are some additional aspects imposed by the lower polarity solvent [33]. The kinetic behavior still adheres to the Michaelis-Menten process in eq. (3), but the lower dielectric constant of ethanol makes the binding of 8 and 6 at least 300 times stronger in ethanol so that the k_cat term can be evaluated for all members of 6. Figure 2 shows the Brønsted plot of lg(k_cat) vs. pssKa (in ethanol), where the maximum rate constant for good substrates has a limiting value of ∼150 s⁻¹ and is independent of the nature of the leaving group. The Fig. 2 plot is still concave downward indicative of a change in rate limiting step and the slope changes from 0 for good substrates to −1.12 for poor substrates. The gradient of 0 with good substrates is interpreted as arising from a conformational change in the 8:6 complex which more favorably positions the substrate for the catalyzed chemical process which then becomes rate-limiting for poorer substrates.

Fig. 2:

A Brønsted plot of lg(k_cat) vs. pssKa for the cleavage of 6a–g mediated by 8 in ethanol at 25 °C. Adapted with permission from C. T. Liu, A. A. Neverov, R. S. Brown. J. Am. Chem. Soc. 130, 16711 (2008) [33]. Copyright 2012, American Chemical Society.

While the reactions are very fast for substrates with good leaving groups in the above two examples, they become linearly slower for substrates having leaving groups with high pssKa values. In the following section we deal with a metal ion promoted method for accelerating the departure of leaving groups from such substrates.

Assisting the departure of leaving groups by their association with metal ions (LGA)

There are at least four main modes by which metallo-enzymes are proposed to act in cleaving substrates that include phosphate diesters and amides. These are: 1) Lewis acid activation of the substrate via M^x+:O=P binding; 2) delivery of metal-bound hydroxide or alkoxide that serves as a nucleophile or a base; 3) electrostatic stabilization of the increasingly anionic substrate and nucleophile/base through binding to its (+)-charged active site thereby lowering the transition state energy of the reaction; and 4) assisting the departure of the leaving group by coordinating it to the metal ion (LGA). The latter role may not be particularly important when the substrate has a good LG such as an aryloxy group with a strong electron withdrawing substituent, but it should be important for ‘poor’ substrates having leaving groups such as amides and alkoxides with high parent pK_a values that are increasingly reluctant to leave as anions. As important as the latter LGA effect might be it is exceedingly difficult to demonstrate in small molecules probably due to the difficulty in positioning a metal ion close enough to the scissile group that it can bind prior to, or during, LG departure.

Complex 8 in methanol also promotes the cleavage of a series of O-methyl O-aryl phosphate diesters 9a–h by a mechanism similar to what was shown in eq. (3) where there is a pre-equilibrium binding of 8 and 9 followed by a rate-liming cleavage of the 8:9 complex [34]. Given in Fig. 3 are three Brønsted plots: 1) lower left, a plot of the lg of the pseudo-first order rate constant (k_cat) vs. phenol pssKa for methanolysis of 9 in the presence of 1 mol dm⁻³ methoxide in methanol (◊), gradient −0.57; 2) center, a plot of lg(k_cat) vs. phenol pssKa for the cleavage of 9 mediated by 8 (■), gradient −0.59; and 3) upper right, a plot of lg(k_cat) vs. phenol pssKa for the cleavage of 9 bearing an ortho-NO₂ or C(=O)OCH₃ mediated by 8 (○), gradient −0.34.

Fig. 3:

Three Brønsted plots: 1) lower left, a plot of the log of the pseudo-first order rate constant for methoxide attack on 9 vs. phenol pssKa for methanolysis of 9 in the presence of 1 mol dm⁻³ methoxide (◊); 2) center, a plot of lg(k_cat) vs. phenol pssKa for the cleavage of 9 mediated by 8 (■); and upper right, a plot of lg(k_cat) vs. phenol pssKa for the cleavage of 9 bearing an ortho-NO₂ or C(=O)OCH₃ mediated by 8 (○). Lines through the data are linear least squares fits; see text for gradients.

Figure 3 shows that the methoxide and 8-mediated processes have indistinguishable gradients of −0.57 and −0.59. In simple terms, the gradients of the Brønsted line are a measure of the degree of cleavage of the departing OAr group in the transition state and, by inference, are a reflection of the amount of negative charge building up on the departing group in the TS [34–36]. Williams has suggested that the mechanism for oxyanion promoted cleavage of phosphate diesters is concerted with concurrent (but not necessarily synchronous) nucleophile/nucleofuge bond formation and breaking [37]. The similarity of the Brønsted slopes would suggest a comparable degree of bond cleavage and charge development on the departing aryloxide in both the methoxide and 8-promoted reactions. However, for 8-mediated cleavage of the substrates 9 having an ortho-NO₂ or C(=O)OCH₃ group the gradient is −0.34, suggesting that there is considerably less negative charge building up on the departing aryloxy group. (Note: It has been correctly pointed out by Professor Charles Perrin, University of California, San Diego, that the O-nitro and O-carbomethoxy phenols may lie on a different Brønsted line due to their long known [38, 39] propensity for intramolecular H-bonding which gives them different solution properties such as reduced solubility and greater volatility relative to their meta and para counterparts.) We note that these same phenols do adhere to the Brønsted plot for the methoxide-promoted transesterifcation of 9 so that the shallower slope in the case of the 8-promoted reaction more likely stems from an interaction of the phenoxide departing group with the complex in the transition state of the cleavage reaction. Control experiments do establish that the ortho-NO₂ and O−C(=O)OCH₃ substituted phenols, as their phenoxides, actually form stable complexes with 8 in solution [34].

Figure 3 also reveals an important feature relating to the shallower gradient of the plot for 8-promoted cleavage of the phosphate dieters having the ortho-NO₂ or C(=O)OCH₃ groups, relative to the plot with the diesters that do not. The dinuclear catalyst accelerates the cleavage of substrates with poorer leaving groups to a greater extent than it does the reaction of those with better leaving groups. This is an important consequence of LGA since, as the gradient of the metal ion promoted Brønsted plot approaches zero, the catalytic rate constant becomes independent of the goodness of the leaving group, as least as it is measured by its parent pK_a. Such an occurrence is a definite advantage in biological systems where the leaving groups are generally far poorer than aryloxides.

Cases where there is really good LGA in the cleavage of phosphate mono and diesters

We consider here a specially constructed pair of phosphate mono- and diesters in which a Cu(II) ion is closely positioned to a departing aryloxy group by way of its binding to a covalently attached phenanthroline group as in 10a,b [40, 41]. The X-ray diffraction of a (2-(2phenoxy)-1,10-phenanthroline)₂Cu(II)₂ complex with a bridging μ-acetate, [Cu₂L₂(μ-MeCO₂)][PF₆] [42] indicates it exists as a phenolate-bridged dinuclear complex with phenoxy-O--Cu(II) distances of 1.889 Å and two equivalent ligands containing 2-phenoxy and phenanthroline rings that are only 7° from coplanarity. The close positioning of the phenoxy oxygen and Cu(II) in this system suggests that the metal ion would be ideally located in the corresponding phosphate mono- and diester to assist in the solvolytic cleavage of the P–OAr bond in 10a,b.

Our own studies [40] indicate that the binding of the 2-(2-phenoxy)-1,10-phenanthroline to Cu(II) is very strong. In Scheme 1 is presented a thermodynamic cycle from which we can evaluate the various acid dissociation and equilibrium binding constants as follows. The pssKa for the acid dissociation constant of 11 in methanol was determined to be 16.16 from a UV/vis spectroscopic titration of 11 as a function of increasing [methoxide]. The K_dis dissociation constant of (11Cu)²⁺ was determined to be 2.8 × 10⁻⁷ mol dm⁻³ from the titration study of a solution containing ligand plus equimolar Cu(II). The potentiometric titration data were analyzed using Hyperquad 2000 [43] with inputs for the pssKa of 11 and 11-H⁺ of 16.16 and 4.73 [40] thereby yielding a pssKa for the (11Cu)²⁺ ↔ (11Cu)¹⁺ + H⁺ equilibrium of 0.49, and a K_dis of 2.3 × 10⁻²⁴ mol dm⁻³ for the metal dissociation from 11^-, ((11Cu)⁺ ↔ 11^- + Cu(II)).

Scheme 1:

Thermodynamic cycle depicting various acid dissociation and equilibrium Cu(II) binding constants for 11 and 11.

Two things become apparent from the data presented in Scheme 1. First, the presence of the Cu(II) in (11Cu)²⁺, decreases the pssKa for H⁺ dissociation of the associated phenol from 16.16 to 0.49, so the formation of the phenoxide is strongly stabilized by association with the electropositive metal ion. Correspondingly the K_dis values for removal of Cu(II) from 11 and 11^- suggest that the anionic ligand binds the metal ion by ∼10¹⁶ times better than does the neutral ligand which in terms of ΔG is 97 kJ mol⁻¹ dm³. Part of this strong interaction is realized in the transition states for solvent mediated cleavage of 10a,b in methanol [40], ethanol and water [41]. In Fig. 4 is presented pssH vs. lg(k_obs) profiles for the decomposition of these in methanol at 25 °C [40]. At low pssH, the top plot for 10a is nearly coincident with that of 10b because the phosphate monoester is protonated and has the same charge of −1 as the diester. The latter decomposes spontaneously between pssH 2.5 and 10 with an average k_obs of 2.4 × 10⁻³ s⁻¹. However, as the pssH increases the monoester deprotonates (with a pssKa of 7.8) to form the dianion which decomposes in the plateau region with a k_obs of 15 s⁻¹, fully 6000 times faster than the diester. The acceleration stems from the fact that the monoester decomposes by a highly dissociative process driven by the dianionic nature of the TS. This is revealed by its activation parameters of ΔH^≠ = 90 kJ mol⁻¹ dm³ and ΔS^≠ = 75 J mol⁻¹ dm³ K⁻¹. By comparison, the diester’s activation parameters for spontaneous decomposition are ΔH^‡ = 90 kJ mol⁻¹ dm³ and ΔS^‡ = 9.7 J mol⁻¹ dm³ K⁻¹ [40].

Fig. 4:

Two pssH vs. lg(k_obs) profiles for the decomposition of 10a (top plot) and 10b (bottom plot) in methanol at 25 °C. Adapted with permission from C. T. Liu, A. A. Neverov, C. I. Maxwell, R. S. Brown. J. Am. Chem. Soc.132, 3561 (2010) [40]. Copyright 2010. American Chemical Society.

Presented in eq. (4) is a pathway with for methanolystic cleavage of 10a,b by way of solvent attack on the P with displacement of the Cu(II)-bound 2-(2′-phenoxy)-1,10-phentholine group. For the fully deprotonated monoester 10a, the net charge of the Cu(II) complex is zero, with strong charge dispersal in the TS. The diester complex, 10b, has a net charge of +1 and also decomposes via a charge dispersed TS. Importantly, as the P–OAr bond is cleaving the increasingly (-)-charged phenoxy group becomes strongly associated with the Cu(II) ion in both cases. In general, the LGA solves the problem of facilitating a difficult bond cleavage by offsetting the endothermic latter process with an exothermic one where the departing group has an increasing affinity with a metal ion. This is so effective in the present cases that the nucleophile required to effect displacement is the weakest one in solution, namely solvent [40, 41].

The effects of varying solvent on the LGA provided by the Cu(II) for cleavage of 10a,b in methanol, ethanol and water have been determined. For the monoester, there is only a small difference in the cleavage rates in methanol and ethanol (15 s⁻¹ and 7.7 s⁻¹): the same is true for the diester in these two solvents (2.4 × 10⁻³ s⁻¹ and 3.6 × 10⁻³ s⁻¹). However, in water, there is a retardation in rate constant to 0.115 s⁻¹ for the monoester and 5.6 × 10⁻⁶ s⁻¹ for the diester. The aqueous reactions are still fast the cleavage of such inert compounds, but the comparisons do point out that the larger effects of LGA are really manifested in the lower polarity solvents. The rate enhancing effects of the Cu(II)-promoted leaving group assistance in all three solvents is substantial, estimated at 10¹²−10¹⁵ for the monoester, and 10¹²⁻¹⁴ for the diester relative to their background reactions [41].

LGA in the solvolysis of amides: M^(II)-promoted solvolysis of N,N-bis(2-picolyl) benzamides

In Scheme 2 is a generalized mechanism for the metal ion promoted solvolysis of amides that portrays the various modes of catalysis exerted by the metal ion. Once the tetrahedral intermediate is formed, the forward reaction must be promoted by one or more ways. These may involve: 1) general acid protonation of the departing amide anion by a metal ion-bound HOS or less effectively; 2) by a general acid catalysis by solvent HOS; or 3) translocation of the metal ion to coordinate to the departing amide anion as shown below.

Scheme 2:

Generalized mechanism for metal ion promoted solvolysis of amides with metal ion promoted LGA.

There is a little appreciated mode of catalysis of metal ion promoted solvolysis of amides and related species that involves initial Lewis acid coordination to the amide N as in 12. Unfortunately, such coordination is only rarely observed in small molecule chemistry because the more basic coordination site is the oxygen of the C=O unit. However, N-coordination can be forced by providing metal binding ligands on the amidic N that position the metal ion at the latter’s lone pair. Interestingly, a report by Houghton and Puttner in 1970 described the observation that N,N-bis(2-picolyl) amides, when placed in a methanol solution containing Cu²⁺ (therein forming species such as 14) rapidly cleaved to form the corresponding methyl ester and Cu(II):N,N-bis(2-picolyl)amine [44]. The implications and additional demonstrations of this discovery remained relatively dormant until much later when physical studies of such M(II):amides [45–47] and the demonstration of their synthetic utility [48] appeared. These, as well as reports on the methanolysis of metal complexes of secondary amides [49, 50] prompted us to investigate the mechanism of metal ion promoted alcoholysis under pssH controlled conditions [51–53].

A comprehensive investigation of the Cu(II)-promoted methanolysis of a series of substituted benzoyl N,N-bis(2-picolyl) amides (15a–g) demonstrated that the pssH vs. lg(k_obs) profiles for the seven members of the series determined in the presence of a slight excess of Cu(II) (as the triflate) were bowed downward suggestive of the process shown in eq. (5). The kinetically active forms contain a Cu(II):(⁻OCH₃) formed by ionization of a Cu(II) bound methanol having pssKa values of ≤ 6.5 [52, 53]. The Hammett plot of the lg(k_max) vs. σ_x value of the substituents on 15 has a gradient of 0.80 ± 0.05 [52] showing that there is some negative charge building up on the aryl group in the rate limiting TS. Since the plot is linear over the entire group, there is no evidence for a change in mechanism or rate limiting step in passing from substrates having electron donors to those having withdrawing substituents. Additional evidence gathered included activation ΔH^‡ values of 80.2 and 89.4 kJ mol⁻¹ dm³ and ΔS^‡ of (−22.6 ± 2.5) and (−8.4 ± 4) J mol⁻¹ dm³ K⁻¹ for 15b and 15g respectively, suggesting that there is little restriction in the degrees of freedom in the TS relative to the ground state of 15:Cu^(II):(⁻OCH₃)(HOCH₃). As a final note, the solvent kinetic deuterium isotope effects on the k_max values were k_H/k_D = 1.12 and 1.20 for 15b and 15g. All these data are consistent with the hypothesis that the reaction proceeds by way of a rate-limiting intramolecular delivery of a Cu(II)-coordinated methoxide to the C=O of the benzoyl group with subsequent fast breakdown of a presumed tetrahedral intermediate to product.

Additional mechanistic information is provided by DFT calculations [52] on the reaction pathway with the computed free energy profiles starting from the initial state of the 15:Cu^(II):(⁻OCH₃)(HOCH₃) complexes for 15b, 15e, and 15g proceeding to products as shown in Fig. 5. Starting with the trigonal bipyramidal GS structure, the Cu(II):(⁻OCH₃) intramolecularly attacks the benzoyl C=O in all three substrates. There is a larger degree of Cu(II):N amidic interaction during nucleophilic attack, as indicated by a shortening of that bond (from 2.73–2.76 Å to 2.03–2.05 Å) and an increase in nitrogen pyramidalization (χ_N decreases from ∼150–154° to 132.6–132.9°. In all cases, the TS_Nu structure leads to a single tetrahedral intermediate, INT, which occupies a shallow minimum on the free energy surface. Breakdown of INT is virtually barrierless, such that it is considered too unstable to be a vibrationally equilibrated intermediate and proceeds by way of concurrent fracture of the Cu(II)-methoxide and C–N bonds and shortening of the Cu(II)-trigonal N bond distance by 0.02–0.03 Å to 1.99 Å with simultaneous opening of the (Py)N-Cu(II)-N(Py) angle resulting in planarization of the Cu(II) in the product structure (P). The (Pyr-CH₂)₂N⁻---Cu(II) departs as the coordinated anion, but having departed or nearly so, is assumed to be rapidly protonated by the medium or buffer components therein.

Fig. 5:

DFT-computed reaction pathway for the cleavage of the Cu(II):(⁻OCH₃)(HOCH₃) complexes of 15b (― ― ―); 15e (―·―·―); and 15g (······) in methanol. All free energy values are to scale and are reported in kJ mol⁻¹ dm³ at 298 K relative to the GS structure. Adapted with permission from M. A. R. Raycroft, C. I. Maxwell, R. A. A. Oldham, A. Saffouri Andrea, A. A. Neverov, R. S. Brown. Inorg. Chem.51, 10325 (2012) [52]. Copyright 2012. American Chemical Society.

The metal ion catalysis of the alcoholysis of bis(2-picolyl) amides is not limited to methanol and Cu(II), but also occurs in ethanol and in the presence of Ni(I) and Zn(II). Zn(II) and Ni(II) are very effective, but either do not bind as tightly as Cu(II), or do not exhibit saturation behaviour in their k_obs vs. pssH plots which generally precludes comparison of their kinetic data with those of the Cu(II) complexes. The overall kinetic behaviour is best with Cu(II) as it binds tightly to the amide, and shows saturation behaviour in the plots of k_obs as a function of pssH [53]. A fit of the k_obs vs. [H⁺] data obtained with the Cu(II) complex of 15b to eq. (6), derived for the process in eq. (5), gives a kinetic pssKa of 5.79 ± 0.07 and k_max of (5.7 ± 0.4) × 10⁻³ s⁻¹.

All of the metal ions in each alcohol are believed to react by mechanisms comparable to that depicted in eq. (5). The metal ions appear to commonly exert a trifunctional catalytic role, acting as a Lewis acid to activate the substrate by reducing the amidic resonance along with providing a closely positioned intramolecular nucleophile. Its subsequent role, to provide LGA for the rapid departure of the coordinated bis(2-picolyl)amide anion is essential, since in this sort of two step process the catalyst needs to promote both the nucleophilic addition and breakdown steps. The overall effectiveness of catalysis can be quantified simply by comparing the rate constants for methoxide attack on 15b in the absence and presence of coordinated Cu(II). A methanol solution containing 150 mmol dm⁻³ of 15b and 0.3 mol dm⁻³ of KOCH₃ shows no sign of product formation after 250 days. Assuming one could detect 1 mmol dm⁻³ of product, an upper limit for the rate constant for ⁻OCH₃ attack on 15b of 1 × 10⁻⁹ mol⁻¹ dm³ s⁻¹ can be computed. In comparison, one can easily compute [53] a rate constant of 5.4 × 10⁸ mol⁻¹ dm³ s⁻¹ for methoxide attack on Cu(II):15b:(HOMe)₂, which is at least 5 × 10¹⁷ larger than the reaction in the absence of metal ion.

An alternative and more visually appealing way to compare the catalysis provided by the metal ion is to determine how tightly the metal ion binds to a transition state containing 15b and methoxide [53]. In Scheme 3 is a thermodynamic cycle depicting the various rate and equilibrium constants that are necessary to derive the ΔG^o values to determine the free energy of stabilization due to binding of a Cu(II) to a TS comprising 15b + methoxide (ΔΔG^‡_stab). Since the binding constant (K_b) and second order rate constants for methoxide attack on 15b and Cu(II):15b are known or can be calculated, the ΔΔG^‡_stab at standard state can be computed by inserting these values into eq. (7)

Scheme 3:

A thermodynamic cycle describing the free energies for various equilibrium and kinetic steps for methoxide attack on substrate 15b, equilibrium binding of the metal ion to15b and methoxide attack on the 15b:Cu(II) complex (metal ion charges omitted for clarity). Products include the dipicolyl amine (DPA) and methyl benzoate.

Shown in graphical form in Fig. 6 is the free energy plot for the various species involved in the Cu(II) catalyzed and methoxide promoted methanolysis of 15b. Of note is the fact that there is a 124.3 kJ mol⁻¹ dm³ lowering of the TS energy for methoxide attack on 15b by associating it with Cu(II). The graphical presentation also emphasizes the fact that the role of Cu(II) is to assemble the methoxide and substrate in a proximity that permits the reaction to occur with a TS energy close to that of the free species.

Fig. 6:

A free energy diagram at standard state depicting the ΔG^o values for the binding of Cu(II) to 15b, and the TS energies for methoxide attack on 15b and its attack on Cu(II):15b. (Adapted with permission from M. A. R. Raycroft, C. I. Maxwell, R. A. A. Oldham, A. Saffouri Andrea, A. A. Neverov, R. S. Brown. Inorg. Chem.51, 10325 (2012) [52]. Copyright 2012. American Chemical Society.

This sort of LGA that is attributable to Zn(II), Ni(II) or Cu(II) is also realized in other acyl transfer processes involving more easily prepared N,N-bis((1H-benzimidazol-2-yl)methyl)benzamides such as 16 [53], ureas such as 17 [54] and carbamates like 18a–e [55] where the overall mechanism is essentially the same as for the M(II) complexes of the amides.

Conclusions, caveats and speculative implications for metallo-enzyme catalysis of phosphoryl and acyl transfer processes

The above examples deal with two relatively new and little studied ways for accelerating metal ion catalyzed phosphoryl and acyl transfer reactions. These are: medium effects tailored to reactions of anionic and neutral phosphates and carboxylate esters and amides, and leaving group assistance (LGA). The former builds on the idea that the active sites of enzymes have a reduced dielectric constant more like that of an organic solvent than water [20]. This is an interesting and controversial concept given that dielectric constant is a macroscopic property of an isotropic medium while the enzyme active site resembles the anisotropic interior of a molecular bottle, the insides of which are adorned with functional groups having specific interactions with the transforming substrate [21]. Nevertheless, at the present level of investigation, the optimum requirement for high activity in the sort of systems described above is a reduced polarity, light alcohol medium that enhances the electrostatic ion-ion and ion-dipole interactions of substrate and catalytic metal ion and also provides H-bonding capabilities that are necessary for optimal rate accelerations. The examples provided demonstrate that the effect of methanol and ethanol is large enough that accelerations range from 10¹¹ to 10¹³ for a dinuclear Zn(II) promoting the cleavage of phosphate diesters relative to the background alcoholysis reaction at pssH near neutrality where the catalyst is active. In the case of specially constructed Cu(II) complexes 10a,b, the accelerations afforded by LGA for phosphate P–O cleavage are estimated to be 10¹⁴–10¹⁵ for the monoester and 10¹⁴ for the diester relative to the computed background reactions. It is also notable that these accelerating effects are not limited to methanol and ethanol, but also occur in water albeit with about a 100–1000-fold reduction in rate.

The generality of metal ion catalyzed LGA is difficult to demonstrate in small molecule examples due to the fact that metal ions preferentially coordinate to the C=O group unless there is some modification of the amide structure to position the metal ion close to the N lone pair. Nevertheless, metal ion promoted rate accelerations for appropriately configured amides can amount to 10¹⁷ relative to the nucleophilic attack by the lyoxide of the solvent. The present amide cleavage system using LGA exhibits multi-functional roles for the M(II) interaction with the substrate and transition states leading to and from a tetrahedral intermediate that are rarely observed in metal ion-catalyzed solvolyses where the metal ion is not positioned to enable it to coordinate with the amidic N. These are: 1) an unusual N-coordinated Lewis acid activation of the substrate; 2) intramolecular delivery of a M(II)-coordinated lyoxide; and 3) assistance of the departure of the LG as a metal-coordinated amide anion.

Much effort and speculation has been expended to explain how enzymes achieve their remarkable rate accelerations. Zhang and Houk have discussed the great bulk of the hypotheses and analyzed [56] a large number of enzymatic systems, concluding that the best catalytic systems have transition state binding energies far greater than can be achieved by non-covalent interactions between the substrate, cofactors and enzyme. That efficiency stems from additional covalent effects such as heavy atom- and H-bond formation, as well as strong interactions of cofactors like metal ions that develop between the enzyme (catalyst) and transition state, thereby altering the mechanism from what is seen in the absence of catalyst. It is an interesting observation that this sort of covalent bonding is manifested in the way the Cu(II) binds the transition states for the two-step solvolysis reaction of amides described herein and how the dinuclear Zn(II) catalyst binds the transition state for phosphate diester cleavage [31, 33, 34]. Here it seems that the reduced dielectric medium plays an important part in encouraging the strong binding to the TS. Admittedly, extrapolation of the large solvolytic rate effects observed with light alcohol solvents and small, specially designed complexes and substrates to the actual biological systems cannot be made without serious criticism. However, such large rate accelerations might lead one to suggest that coupling of the two effects contributed to the sorts of rate accelerations observed in enzymes that promote solvolytic acyl and phosphoryl transfer reactions. One additionally might speculate that some of the limitations imposed in small molecules may not appear in enzyme-promoted reactions where the tertiary structure of the enzyme, and its conformational mobility, controls the placement of the metal ion relative to the transforming substrate and can reposition the metal ion to optimize interactions.