A Proton Relaxation Rate Study of the Copper Analog of Escherichia coli Alkaline Phosphatase*

SUMMARY Proton relaxation rates for the binary complex of the Cu*f analog of Escherichia coli alkaline phosphatase and water have been measured. Titration of apoenzyme into copper salts shows approximately two tight copper-binding sites, Ki = 10m6 M, and 13 weak sites per enzyme dimer molecule. Enhancement of l/TIP a factor of 2 upon binding enzyme observed; the time the metal in substrate binding. the the frequency TIP* and from the of T,: Tz x Cu2+-water proton distance r = the number ligand q, = r = results

Baltimore, Maryland 21205 SUMMARY Proton relaxation rates for the binary complex of the Cu*f analog of Escherichia coli alkaline phosphatase and water have been measured.
Titration of apoenzyme into copper salts shows approximately two tight copper-binding sites, Ki = 10m6 M, and 13 weak sites per enzyme dimer molecule. Enhancement of l/TIP by a factor of 2 upon binding to enzyme is observed; this result for the first time suggests the participation of the metal in substrate binding. From the temperature dependence of TIP* and the fact that T1, > TQ it is concluded that TIP is not exchange-limited.
From the frequency dependence of TIP* and from the ratio of T,: Tz we estimate a correlation time, TV, of 4.6 x 10m9 s-l which corresponds to the Cu2+-water proton distance r = 3.6 A when the number of ligand sites, q, = 1 or r = 3.2 when q = W. The addition of phosphate has little effect on TIP" for the copper enzyme.
These results suggest that at least 1 water molecule or hydroxyl ion has access to the inner coordination sphere of the Cu*+ ion and that inorganic phosphate does not displace this ligand.
Escher&&a coli alkaline phosphatase is a dimeric metalloeneyme which catalyzes the nonspecific hydrolysis of organic phosphomonoesters.
The native enzyme, molecular weight = 80,000 to 86,000 (1, 2), contains 4 zinc atoms (3, 4) which can be replaced by numerous other divalent transition metal ions (4,5). Only the zinc enzyme possesses complete catalytic activity; however, the cobalt and copper enzymes are partially active (6,7). All of the characterized metallo analogues bind phosphate; the Cu*+-, Mn*+-, and Co*+-enzymes, in particular, form stable phosphoryl intermediates (8). The use of paramagnetic probes to study protein metal-binding sites, as well as the interaction of small molecules with metalrequiring proteins, by nuclear magnetic resonance is well established (9). Because water can be considered a substrate of alkaline phosphatase, an investigation of the interaction of Hz0 and a paramagnetic analogue of the enzyme is valuable in elucidating the role of metal. Cottam and Thompson (10) have recently reported some proton relaxation rate data on water protons in the presence of Mn*+-alkaline phosphatase.
We have chosen to study the copper-enzyme, because copper is also a relatively efficient paramagnetic relaxer, and affords a partially active ( = 5% of the zinc-enzyme activity) analogue. Electron paramagnetic resonance data provide evidence that there are two specific tight copper-binding sites per enzyme dimer, as well as weak sites, and that the tightly bound copper atoms each bind at least three nitrogen ligands, probably histidyl side chains (11,12). Moreover, EPR and Sephadex-binding studies suggest that the two tightly binding copper atoms compete for the catalytic zinc sites (11).
In the following section 2'1 and T2 are the observed spin-lat,tice and spin-spin relaxation times; TIP and TQ are paramagnetic contributions to T1 and T2; TI,* and TQ* are the values for the enzyme-bound metal; 7c is the correlation time; 7,, TV, and 7m are the rotational, electron spin, and chemical exchange correlation times, respectively; Q is the number of ligands in the coordination sphere of the metal; and cl is the observed enhancement of T1. More detailed definitions and discussions of these terms can be found in the references (9, 13).
Alkaline phosphatase was isolated from E. coti C 90 (a generous gift from Dr. B. Bachmann of the Genetic Stock Center, Yale University) and purified by a modification of the original Mallamy and Horecker procedure (14), using spheroplast formation and, in the final purification and concentration step, batch addition of triethylaminoethylcellulose (Bio-Rad).
Only enzyme with specific activity 250 pmoles mg-* min-1 which corresponds to the maximal reported activity was used. The enzyme was found to be homogeneous by ultracentrifugation.
Apoenzyme was prepared by incubation of the purified native enzyme with Chelex (Bio-Rad) as described by Csopak (15). Copper-alkaline phosphatase was prepared by direct titration of spectroscopic grade CuS04 (Fisher "Spec-Pure"), CuC&, or Cu(N03)2 ("Baker analyzed" reagent) into the apoenzyme. Zinc removal and copper incorporation were monitored by atomic absorption spectrometry.
Water T's were measured using the pulsed NMR method of Carr and Purcell (16) with a nuclear magnetic resonance pulsed spectrometer operating at 8.1, 24.3, or 40.0 MHz. T1:T2 ratios were measured on a Varian NMR spectrometer, equipped with Fourier Transform and operating at 220 MHz; Ts were det,ermined by the null method and T's were determined by line width measurements. Fig. 1 shows the titration of 0.21 mM apoalkaline phosphatase in 10 mM Tris-Cl buffer, pH 7.0, at 4", with CuS04, CuC12, and Cu(NO&. l/Tl,* varies linearly with copper concentration and cl = eb = 2.2, until [Cu]/[enzyme] = 2; l/TIP* versus [CUSO~] then becomes nonlinear with a decreasing slope. l/TIP* at 25" for a solution of 0.40 mM CuS04 at the same enzyme concentration is 0.26 s-l; thus, the paramagnetic contribution to the spin-lattice relaxation rate shows a negative temperature coefficient.
The titration of CuS04 into the reconstituted Zn2-enzyme is also shown in Fig. 1; 110 enhancement within experimental error is observed.
This result suggests that copper added to apoenzyme binds at the catalytic zinc sites. This possibility is substantiated by EPR data (Taylor and Coleman (11)) which show that addition of two coppers to apoenzyme results in a spectrum with high g values and nitrogen splitting, but that addition of copper to Znz-enzyme does not produce this effect. Moreover, it is unlikely that these observations represent a conformational change in the apoenzyme when it is reconstituted with zinc, since neither ultraviolet optical rotary dispersion nor circular dichroism studies of alkaline phosphatase (17) show any change in the enzyme spectrum when zinc is added. Fig. 2 shows a Scatchard plot (18) for the binding of CuSO4 to apoalkaline phosphatase at 4". As can be seen there are between two and three specific tight binding sites for copper, Ki = 1.0 PM, and 13 weak sites, Ki = 50 PM. The data were obtained from the titration of apoenzyme into a solution that is 0.5 InM CuSO4 in 10 mM Tris-Cl, pH 7.0. Formation of the metalloenzyme was followed by the enhancement parameter, E, obtained at 24.3 MHz; eb = 2.2 was obtained from Fig. 1.
The data in Table I illustrate the frequency dependence of l/!/'~,* for 0.125 mM copper-alkaline phosphatase in the same buffer at three copper concentrations and three frequencies, lITI,* is seen to be frequency-dependent for a given concen- Apoensyme in 10 mM Tris-Cl btier, pH 7.0, was titrated into 0.50 mrd CuSOc in the same buffer; complex formation was followed by the enhancement parameter, e, assuming eb = 2.2 from Fig. 1. tration. Fig. 3 shows a plot of TIP* versus d. The correlation time, TV, as determined from (slope/intercept)1'2, is 4.6 x 10Pg s. Table II summarizes the TI and Tz data at 220 MHz for 0.125 mM apoenzyme and 0.2 mM CuSO4 in the above buffer. I/T@* is greater than l/Tl,* by a factor of 28. No chemical shift in the enzyme sample is detected upon broadening of the Hz0 proton signal by copper. Table III shows 7c calculated from the T1 : Tz ratio and from the plot of I/T,* versus w. As the value derived from the TI : Tz ratio represents a lower limit for TV, 4.6 x 10V9 is probably the better estimate.
This corresponds to r = 3.6 A when q = 1.
Addition of a a-fold excess of inorganic phosphate to 0.125 mM  apoenzyme, 0.250 mM CuSO4 at pH 7.0 results in no change of l/TIP* within experimental error. The data for the titration of apoenzyme with copper indicate the existence of two copper sites per dimer of alkaline phosphatase. These titrations have been carried out numerous times at a variety of enzyme concentrations and temperatures; a constant E = 2 is always observed until [Cu2+]/enzyme = 2 after which E decreases. The Scatchard binding plot, based on data for the titration of copper with apoenzyme, indicates between two and three tight copper sites per dimer; the inherent error in this analysis, however, is greater.
Together, the data suggest that there are two tight copper sites, in agreement wit,h atomic absorption data reported by Csopak (7), which also shows two copper sites, and in contrast to reports which show four sites for zinc, manganese, and cobalt (4). TIP* has a negative temperature coefficient.
In addition, T,, > Tzp. These results clearly indicate that TIP is not exchange-limited and that, therefore, f. TIP* = T1,. Thus, if 7c is known, f.T,,* can be used to calculate the interatomic distance, r, between the water protons and the copper atom.
We have estimated 7c by two methods (Table III). First, we have determined 7c graphically by a plot of TIP* versus ~2 (Fig.  2) (19). The square root of the slope/intercept affords a value of 4.5 x 10eg s for TV. (The fact that the plot is linear is good evidence that 7c is not dominated by rs (9) .) We have also calculated a lower limit for rc from the T, : Tt ratio (20) ; this method gives a value of 4.6 x 10eg s for TV. The two values for the correlation time are in close agreement and correspond to r = 3.6 A when q = 1 or r = 3.2 A when q = 35. We include the possibility of q = $$ to take account of the fact that hydroxyl ion may be the active species in catalysis by CL?+ or Zn*+ at high pH, as for example in the case of carbonic anhydrase (21). It should be possible to distinguish the two possibilities by a determination of q using 170 NMR of water. We have recently initiated appropriate experiments toward this end. The interatomic distance between the protons of an axial water ligand and metal ion in copper proline dihydrate, a compound which approximates the environment of copper in a protein, has been determined by x-ray crystallographic data to be 3.2 A (22). Thus, Hz0 or OH-would appear to be binding in the first coordination sphere of the copper atom of Cu*f-alkaline phosphatase. The enhancement, ti = 2, observed for l/TIP upon binding corroborates the idea that innersphere relaxation is being observed.
This result is in marked contrast to that observed for manganese-alkaline phosphatase by Cottam and Thompson Together these results suggest a tentative model for copperalkaline phosphatase, with the following features: (a) at least 1 water molecule or hydroxyl ion can exchange on and off of the metal site of the enzyme; (6) this water molecule or hydroxyl ion actually enters the first coordination sphere of the copper ion; and (c) inorganic phosphate binds at a site other than the copper site or it binds at the copper site without displacing the Hz0 or OH-ligand.
It should be noted that these results do not exclude additional, slowly exchanging water molecules, which would not be detected by these methods.
It is significant that in the case of the partially active copper analogue of E. coli alkaline phosphatase, metal is implicated in H20 binding, whereas t,his is apparently not true for the inactive manganese enzyme.
These findings suggest the need for further investigation into possible correlations between HzO-metal distances for the various phosphatase analogs and relative activities.
These studies are now in progress in this laboratory.