31P NMR of Phosphate and Phosphonate Complexes of Metalloalkaline Phosphatases”

of phosphate phosphonate of Escherichia alkaline phosphatase One equivalent of P,, bound to to a ‘lP resonance ppm downfield for P,, assignable the noncovalent phosphate in of eq dimer to a resonance the position expected for free P,. At 5.1, resonance ppm downfield for free P,, is assignable to the covalent complex, E-P. The large downfield shift the is highly bond At phosphatase mol of a phosphoryl group mol of with a single ‘lP resonance ppm downfield from the resonance for P,. Formation of the apophosphoryl removal of the from the

with the following exception.
For the NMR studies, the Co(R) and Mn(II) enzymes were prepared by the addition of slightly less than 2 eq of Me(R) per apoenzyme dimer. Titration of apoalkaline phosphatase with Me(R) ions as followed by several spectroscopic techniques shows that the first two metal ions added were tightly bound at the sites occupied by the catalytically active Zn(I1) ions of the native enzyme (6)(7)(8)(9)(10)(11). When using paramagnetic metal ions to reconstitute apoalkaline phosphatase, addition of more than 2 eq of Me(R) per enzyme dimer gives rise to an ESR signal significantly different from that generated by the first 2 eq of metal ion (8,9). Concentrations of Me(R) above 2 eq per enzyme dimer also lead to broadening of the NMR lines of nuclei carried on active site ligands which is similar to that produced by free metal ions in solution (10,11). The same phenomenon was observed in the present work with both phosphate and phosphonate as ligands. Addition of more than 2 eq of Co(R) to the enzyme resulted in broadening of the a1P resonances for both compounds similar to that observed with hydrated Co(R) (see under "Results").
All NMR samples and equipment were prepared metal-free following procedures described in the preceding paper (5,11 1D). One of these is -2 ppm upfield from the position of the resonance for enzyme-bound phosphate observed at pH 8 and -2 ppm downfield from the resonance position of orthophosphate at pH 5.1, which suggests that the noncovalently bound phosphate has become protonated.
The second resonance is over 8 ppm downfield from the resonance position for free inorganic phosphate at pH 5.1 (Fig. 1D). The latter resonance must represent that of the covalently bound phosphate complex, E-P, since previous evidence shows that E-P becomes the predominant species at low pH (5,6). Effect of Co(ZZ) and Mn(ZZ) on s1P Resonance of Enzymebound Phosphate-Addition of 1 eq of phosphate to Co(I1) alkaline phosphatase (pH 8.0) containing 2 g at Co(II)/mol of enzyme dimer results in complete disappearance of the 31P resonance (Fig. 1E). Under the present system of data collection, and at millimolar concentrations, the linewidth would have to be at least 300 Hz to be undetectable. The addition of a 2nd eq of P, to the Co(R) enzyme results in the appearance of a highly resolved resonance at the position expected for free inorganic phosphate at this pH (Fig. 1F). This resonance is not detectably broadened, showing that the 2nd eq of P, is not in rapid exchange with the 1st eq, and that there is negligible free Co(I1) present in solution. Addition of any more than 2 eq of Co(I1) to the enzyme broadens this line in a manner similar to that observed on the addition of Co(I1) to a solution of inorganic phosphate. Exactly the same phenomena are observed when Mn(I1) is used instead of Co(I1).
31P NMR of Phosphate Complexes of Cd(ZZ) and Apoalkaline Phosphatase-Since Cd(R) alkaline phosphatase forms a stable and well characterized phosphoryl enzyme near neutral pH (5), the 31P NMR characteristics of the phosphoryl enzyme were further explored by preparation of the phosphorylated Cd(I1) enzyme. Addition of 2 eq of P, to Cd(I1) alkaline phosphatase at pH 6.5 results in the appearance of two SIP resonances; one at the position of the resonance for inorganic phosphate, and the other of similar amplitude 8 ppm downfield from the resonance position for inorganic phosphate (Fig. W). Dialysis of the Cd(I1) enzyme at pH 6.5 against metal-free buffer results in removal of the resonance at the position of inorganic phosphate and retention of the downfield resonance ( Fig. 2B). As demonstrated by S2P-labeling (5), this resonance must represent the phosphorus nucleus of the Cd(I1) phosphoryl enzyme.
The Cd(I1) ion can be completely removed from the dialyzed Cd(R) phosphoryl enzyme by dialysis against 5 x 1O-s M l,lO-phenanthroline to form the apophosphoryl enzyme (5). The s1P resonance of the apophosphoryl enzyme ( Fig. 2C) shifts -2.3 ppm (88 Hz) upfield from the resonance position of the Cd(R) phosphoryl enzyme (Fig. 2B), but it is still -5 ppm downfield from the resonance for free inorganic phosphate at this pH.
results in a broadened resonance such that a value of V,, cannot be obtained on inspection (Fig. 3A). Computer simulation of the undecoupled spectrum using a T, of 0.024 s, and assuming an A,X spin system,2 permits assignment of a Vr, for the apophosphoryl enzyme of 13 Hz (Table I). The broadening of the proton-decoupled phosphoryl resonance in the presence of Cd(I1) (T, = 0.016 s) compared to that of the apophosphoryl enzyme (T, = 0.024 s) (Fig. 2, B and C) is not due to cadmium-phosphorus spin-spin coupling. There are two isotopes of Cd with nuclear spin of i/z (l%d and lllCd), each the exchange rate of the exchangeable protons. For comparison of linewidths under different conditions see Table I. The assumed equivalence of the @ protons which permit analysis in terms of an A,X spin system may not be valid for the enzyme phosphoserine.
Should there be a significant chemical shift difference between the B protons, the s1P spectrum corresponds to the X of an ARX system, in which case only the sum of the proton-phosphorus couplings can be determined from the linewidth of the undecoupled spectrum.
The observed linewidth, -33 Hz, is not in conflict with conclusions drawn with respect to the structural features of the phosphoryl enzyme (see "Discussion"). The 31P-(1H)3 spectra of phosphoserine and the apophosphoryl enzyme yield T, vaiues of 0.53 and 0.024 s, respectively, from the measured linewidth (see Table  I  The results are shown in Fig. 4. The slP resonance of the phosphoseryl group in apoalkaline phosphatase at pH 8 is -2.5 ppm further downfield than that of isolated phosphoserine at the same pH (Fig.  4). The 31P resonance of free phosphoserine moves upfield by -4 ppm as the monoester is protonated (20). The pK, for this transition is -5.7, the normal second pK, for phosphoserine.
In marked contrast, the usual pK, of phosphoserine is not observed in the 31P resonance of the phosphoseryl group of the apoenzyme as it is titrated from pH 8 to pH 2 (Fig. 3). The 31P resonance of the apoenzyme shifts only slightly between pH 8 and pH 4. The resonance shifts upfield to the resonance position of the model phosphoserine when the pH changes from 4 to 3.5 (Fig. 4) Fig. 7, and was determined by two methods. The first used a constant concentration of Co(I1) enzyme and varied the phosphonate concentration.
The second method utilized constant concentrations of phosphonate and apoenzyme to quently, in the absence of an independent determination of T, from studies at other frequencies, we have assumed that the dipolar mechanism is the sole contribution to T, Obs. Thus, T, Obs = T,,, and the 7c is taken from a theoretical curve for q ("'P-(H)) versus T, constructed in a similar manner to that described previously for TJ ( 13Cv lHl) (19). Using the value of rC of 5 x 10M9 s and the measured T,, a value for r31 p.~H of 2.5 A can be calculated, which agrees well with that predicted for 31P-O-C,-'H distances in phosphate esters.
which fractional equivalents of Co(I1) or Mn(I1) were added. Data from the second method are also plotted as a function of moles of Me(B) added/m01 of enzyme dimer. The first 2 eq of Me(B) bound to the apoenzyme show much less relaxation enhancement of the 31P nucleus of the phosphonate than do the Co(I1) or Mn(I1) ions added above the 2-eq point. The relaxation observed from the first 2 eq of Me(B) is that characteristic of the metal ion at the active site. Calculation of Me(ZZ)-a'P Distances in Phosphonate Complexes of Mn(ZZ) and Co(ZZ) Alkaline Phosphatase-The generalized Bloembergen-Solomon equations relating the metalinduced relaxation times, T,, and TZM, to the nuclear-electron distance, r, are given by Equations 1 and 2, where y, is the nuclear gyromagnetic ratio; wS is the electron precession frequency; wI is the nuclear precession frequency; A is the isotropic indirect hyperfine interaction; T, is the correlation time for the anisotropic dipolar interaction for which l/re = (l/rJ + (l/r?), where rr is the rotational correlation time and re is the correlation time for the isotropic hyperfine interaction. l/r1 = ( ~/TJ + (l/7,,,), where rS is the electron spin relaxation time and T,,, is the residence time of a nucleus in the coordination sphere (or environment) of the metal ion. The other parameters have their usual meanings (22, 23).
Applications of the above equations to the relaxation of 'H of H,O or other ligands by Mn(I1) ions attached to macromolecules have generally assumed that the isotropic indirect hyperfine or scalar interaction is negligible (22). For Mn(II) relaxing 'H this assumption has generally appeared to be justified, and T,, has been observed to be controlled primarily by TV (as predicted if the first (dipolar) term of Equation 2  coordinated ligands than the free metal ion (small 7?).' For Co(I1) this assumption is not justified, since the hyperfine interaction is greater. When the nucleus undergoing paramagnetic relaxation is 81P the neglect of terms in A is not justified for either metal ion, since the scalar interaction is expected to be much greater for $'P than for 'H. In fact, the A terms may dominate the relaxation, and the metal ion bound to the macromolecule may be less effective in relaxing ligand nuclei 'For Mn(II) attached to rapidly rotating molecules, T, e 7,. since the ra for Mn (II) is generally near 1Om8 s. However, for a ligand attached to a macromolecule containing Mn(II), T, may be sufficiently increased (-1O-8 s) for rb to contribute to TV. The change in relaxation of ligand protons observed on comparing relaxation by free and bound Mn(I1) is generally due to the change in r,, even though r6 may decrease somewhat on binding of Mn(I1) to the macromolecule.