Ground state destabilization by anionic nucleophiles contributes to the activity of phosphoryl transfer enzymes

Abstract

Enzymes stabilize transition states of reactions while limiting binding to ground states, as is generally required for any catalyst. Alkaline Phosphatase (AP) and other nonspecific phosphatases are some of Nature's most impressive catalysts, achieving preferential transition state over ground state stabilization of more than 10²²-fold while utilizing interactions with only the five atoms attached to the transferred phosphorus. We tested a model that AP achieves a portion of this preference by destabilizing ground state binding via charge repulsion between the anionic active site nucleophile, Ser102, and the negatively charged phosphate monoester substrate. Removal of the Ser102 alkoxide by mutation to glycine or alanine increases the observed Pi affinity by orders of magnitude at pH 8.0. To allow precise and quantitative comparisons, the ionic form of bound P(i) was determined from pH dependencies of the binding of Pi and tungstate, a P(i) analog lacking titratable protons over the pH range of 5-11, and from the ³¹P chemical shift of bound P(i). The results show that the Pi trianion binds with an exceptionally strong femtomolar affinity in the absence of Ser102, show that its binding is destabilized by ≥10⁸-fold by the Ser102 alkoxide, and provide direct evidence for ground state destabilization. Comparisons of X-ray crystal structures of AP with and without Ser102 reveal the same active site and P(i) binding geometry upon removal of Ser102, suggesting that the destabilization does not result from a major structural rearrangement upon mutation of Ser102. Analogous Pi binding measurements with a protein tyrosine phosphatase suggest the generality of this ground state destabilization mechanism. Our results have uncovered an important contribution of anionic nucleophiles to phosphoryl transfer catalysis via ground state electrostatic destabilization and an enormous capacity of the AP active site for specific and strong recognition of the phosphoryl group in the transition state.

Figure 1. Active site models for the AP ground and transition state.

AP ground state (A) and transition state (B) models based on previously solved X-ray crystal structures (PDB Codes 3TGO and 1B8J, respectively). Proposed active site contacts are illustrated with dashed lines. The proposed electrostatic destabilization from Ser102 in the ground state model is illustrated by the red hash marks.

Figure 2. Free energy reaction profiles illustrating preferential E•S ground state destabilization.

(A) Hypothetical uncatalyzed reaction profile. (B) Hypothetical enzyme that stabilizes the ground and transition states equally () so that the resulting reaction barrier is equal to the uncatalyzed reaction barrier under saturating conditions. This enzyme is not a catalyst as stabilization of the transition state without parallel stabilization of the ground state is required for catalysis. (C) Hypothetical enzyme that makes additional, specific transition state stabilizing interactions, so that the reaction barrier between E.S and E.P is lower than that for the uncatalyzed barrier. This enzyme is a catalyst. (D) Hypothetical enzyme that makes specific ground state destabilizing interactions, , to further enhance the catalytic properties of the enzyme relative to that in panel (C). This destabilization is discussed in the text.

Figure 3. Structural comparisons of noncovalently bound P_i in AP and Ser102 mutants.

(A) Overlay of WT AP with vanadate transition state analog covalently bound to Ser102 (black, 1PDB code 1B8J) and P_i noncovalently bound (gray, PDB code 3TGO). (B) Overlay of WT AP (grey) and S102G AP (blue, PDB code 1ELZ), both with bound P_i. (C) Overlay of WT AP (grey) and R166S AP (red, PDB code 3CMR). Mutation of Arg166 to Ser results in rotation and 1.0 Å translation of the bound P_i. (D) Overlay of WT AP (grey) and S102G/R166S AP (purple). Removal of the Arg166 side chain (R166S AP) results in a rearrangement of the bound P_i with Ser102 present (A→C) but not with Ser102 mutated (B→D) (see Text S5). While it is likely, based on the results herein and previously , that P_i is bound as the trianion in all cases and Ser102 is protonated when present, the X-ray data lack the resolution needed to identify protons.

Figure 4. The pH dependence of P_i and tungstate binding for AP with and without Ser10

(A) The pH-dependent binding of P_i to R166S AP (open circles) and S102G/R166S AP (closed circles). See Methods for assay details. Weighted, nonlinear least-squares fits of Equations S1 and S2 (see Text S9) derived from the binding models in parts (C) and (D) for R166S and S102G/R166S AP, respectively, are shown as solid lines. For R166S AP,  = 7.6±0.1 and  = 110±20 µM. For S102G/R166S AP, was fixed at 6.5 based on the tungstate binding data in part (B) and was fixed at 10^−6.1 M based on the ³¹P NMR data in Figure 6. Fits yielded  = 93±8 nM, and  = 210±20 fM. (B) The pH-dependent binding of tungstate to R166S (open squares) and S102G/R166S AP (closed squares). A weighted, nonlinear least-squares fit of derived from a two-state tungstate binding model () yielded a fit of the R166S AP data with  = 170±25 µM and  = 7.6±0.1. The corresponding fit to the observed tungstate affinity of S102G/R166S AP yielded a  = 890±90 µM and  = 6.5±0.1. The tungstate affinity of S102G/R166S AP is weaker than R166S AP, indicating that Ser102 plays a favorable role in tungstate binding, possibly by allowing formation of octahedral tungstate as observed in other proteins that bind tungstate ,. At pH values ≥8 where the P_i affinity is strongest, the observed competition of ³²P_i binding is likely influenced by competition from contaminating, unlabeled P_i in the tungstate stock rather than tungstate (see Materials and Methods). The dashed portion of the S102G/R166S AP tungstate fit line illustrates where the observed affinity can be accounted for by 0.5 ppm P_i contamination. Omitting the pH 9 and 10 points in the fit did not significantly change the fitted or values. (C) Binding model used to fit the pH-dependent P_i affinity of R166S (and WT) AP. (D) Binding model used to fit the pH-dependent P_i affinity of S102G/R166S AP.

Figure 5. Thermodynamic cycle for binding AP with Ser102 protonated.

(A) The value of for R166S AP is from a fit of the model in Figure 4C to the pH-dependent P_i binding affinity in Figure 4A. The Ser102 pK _a is an upper limit , and thus, the dissociation constant between protonated Ser102 and () is also an upper limit (≤69 pM). The same cycle was used to establish an upper limit for the dissociation constant between WT AP with Ser102 protonated and , ≤290 fM, from the following values: ≤5.5,  = 11.7, and  = 0.46 µM (Table 2). (B) Relationship derived from the thermodynamic cycle of part (A).

Figure 6. The pH-dependence of the ³¹P chemical shift of P_i bound to S102G/R166S AP.

(A) ³¹P NMR spectra of P_i bound to S102G/R166S AP at various pH values. See Materials and Methods for conditions. (B) The chemical shift of P_i-bound to S102G/R166S AP versus the solution pH. A nonlinear least-squares fit to an equation (see Materials and Methods) derived from the binding model in (C) yields δ_upfield and δ_downfield values of −0.74 and 1.94 ppm, respectively, and a value for of 10^−6.1 M as defined in (D).

Figure 7. Models for AP binding P_i dianion and trianion and summary of AP binding energetics.

(A) The dissociation constant limit for dianionic phosphate monoester binding to WT AP was determined in ref. . This limit is also expected to hold for R166S AP where an interacting residue is removed. (B) Removing Ser102 strengthens binding of a dianion by ≥10³-fold, as was estimated to be ∼10 nM (Table 2). (C) Trianion binding (; Table 2) was estimated to be ∼1 fM and is 10⁷-fold stronger than dianion binding. (D) The AP rate enhancement is 10²⁷-fold (for methyl phosphate dianion hydrolysis: k _cat/K _M = 1.2×10⁶ M⁻¹ s⁻¹ and k _uncat∼4×10⁻²² M⁻¹ s⁻¹ [75]), corresponding to a theoretical dissociation constant for transition state binding of 10⁻¹² fM (derived in Figure S1 of ref. [16]). This theoretical affinity reflects binding of the enzyme to the transition state while accompanied by replacement of water by the active site Ser102 nucleophile. The energetics of these two processes cannot be separated and the formal dissociation constant reflects contributions from both binding interactions and positioning of the Ser102 nucleophile and substrate.