Gentamicin nucleotidyltransferase. Stereochemical inversion at phosphorus in enzymatic 2'-deoxyadenylyl transfer to tobramycin.

Gentamicin nucleotidyltransferase-catalyzed reaction of (Sp)-[alpha-17O]dATP with tobramycin produced 2"-(2'-deoxyadenosine 5'-[17O]phosphoryl)tobramycin. The configuration at phosphorus in this product was shown to be Rp by chemical degradation to chiral [17O, 18O]dAMP using a stereochemically defined procedure, and determination of the configuration at phosphorus in this product. Periodate-base treatment of 2"-(2'-deoxyadenosine 5'-[17O]phosphoryl)tobramycin followed by NaBH4 reduction produced (2-glyceryl)-[17O]dAMP, which upon snake venom phosphodiesterase-catalyzed hydrolysis in H(2)18O produced [17O,18O] dAMP. The configuration at phosphorus in this product was shown to be S by enzymatic phosphorylation to [17O,18O]dATP, adenylylcyclase (Bordetella pertussis)-catalyzed cyclization to 3',5'-cyclic [17O,18O]dAMP, and 31P NMR analysis of the ethyl esters. Since snake venom phosphodiesterase-catalyzed hydrolyses proceed with retention of configuration at phosphorus, (Sp)-[17O,18O]dAMP must have been produced from (Rp)-(2-glyceryl)-[17O]dAMP; and since the chemical degradation to the latter compound did not involve cleavage of any bonds to phosphorus, the initial enzymatic product must have been (Rp)-2"-(2'-deoxyadenosine 5'-[17O]phosphoryl)tobramycin. Therefore, nucleotidyl transfer catalyzed by gentamicin nucleotidyl-transferase proceeds with inversion of configuration at phosphorus, and the reaction mechanism involves an uneven number of phosphotransfer steps. Inasmuch as this is an uncomplicated two-substrate group transfer reaction, the mechanism probably involves direct nucleotidyl transfer from the nucleoside triphosphate to the aminoglycoside. The B. pertussis adenylylcyclase reaction was shown to proceed with inversion at phosphorus, as has been established for other adenylylcyclases.

The B. pertussis adenylylcyclase reaction was shown to proceed with inversion at phosphorus, as has been established for other adenylylcyclases.  (1) (1-4). The resulting nucleotidylaminoglycosides are inactive as drugs for combating Gram-negative bacteria. The enzyme can be purified by affinity chromatography and is known to follow a sequential kinetic pathway (5, 6).
One means of obtaining information about the number of nucleotidyl transfer steps in the mechanism, and therefore about the possible involvement of a covalent nucleotidylenzyme intermediate, is to determine the stereochemical consequences of nucleotidyl transfer using a nucleotidyl donor substrate having a center of chirality at P'. If the configuration at this center is inverted in the product, it can be concluded that the mechanism involves an uneven number of phosphotransfer steps, whereas if the configuration is retained, an even number of steps is involved (7,8). In uncomplicated two-substrate, two-product nucleotidyltransferase reactions, inversion strongly implies a mechanism involving a one-step, direct group transfer between bound substrates; whereas retention implies a two-step, double displacement mechanism mediated by a covalent nucleotidyl-enzyme as a compulsory intermediate (7). We here report stereochemical inversion at phosphorus in the gentamicin nucleotidyltransferase-catalyzed reaction of (S,)-[-'70]dATP with tobramycin, consistent with a single displacement mechanism. EXPERIMENTAL PROCEDURES'

RESULTS AND DISCUSSION
In preliminary experiments, we investigated the possibilities for determining the stereochemical course of nucleotidyl transfer by using either (S,)or (I?,)-ATPaS' as the nucleo-Portions of this paper (including "Experimental Procedures" and Table I  15995 tidy1 donor. We readily found (S,)-ATPaS to be an excellent substrate, whereas the (R,)-epimer did not react. However, we were unable to establish the configuration at phosphorus in the product, since it proved not to be a substrate for snake venom phosphodiesterase or nuclease SI, which have been shown to exhibit high selectivities for the R, and S, configurations, respectively, in thiophosphodiester substrates (15,16,18). AMP-tobramycin also proved to be refractory to nuclease action; and neither acetylation nor succinylation of the amino groups in AMP-tobramycin, which destroyed the positive charges, altered this refractoriness. Therefore, [a-170,180] ATP was not considered to be a practical alternative substrate since the configuration of the product at phoshorus could not readily be established by nuclease degradation to [170,180] AMP. We chose (S,)-[(Y-'~O]~ATP as the substrate for this stereochemical study because dAMP-tobramycin (see Structure 1) has a single periodate cleavage site (C4"-C3"), and the cleavage product can be degraded by alkaline elimination to a substrate for snake venom phosphodiesterase without affecting the configuration at phosphorus.
Synthesis of (Sp)-[a-170]dATP-The procedure outlined in Scheme I was followed for preparing this substrate. Michelson   -workers (20). We have used this procedure, which is outlined in Scheme 111. Enzymatic phosphorylation of ['70,'80]dAMP by phosphoenolpyruvate in the presence of adenylate kinase, pyruvate kinase, and a catalytic amount of dATP produced a randomly pyrophosphorylated mixture of [a-'70,180]dATP stereoisomers. Adenylylcyclase-catalyzed cyclization produced the corresponding three stereoisomers of 3',5'-cyclic dAMP, one of which shown in Scheme I11 contained "0 but no I7O. Species containing 170 were produced but are not shown because they were NMR-silent in the analysis. Alkylation by diazoethane produced a mixture of axial and equatorial ethylated 3',5'-cyclic dAMP which was subjected to 31P NMR analysis for bridging and nonbridging doubly bonded), doubly bonded "0, singly bonded "0, and no l8O. The relative signal intensities reveal the configuration at phosphorus.
The configurations shown in Scheme I11 for illustrative purposes are those actually established by the analysis of (Sp)- ['70,'80]dAMP resulting from the degradation of ['70]dAMPtobramycin according to Scheme 11. The 31P NMR integration data are given in Table 11. (Sp)-['70,'80]dAMP in Scheme I11 would arise from the degradation of (Rp)-['70]dAMP-tobramycin in Scheme 11, and this in turn would be produced as the result of inversion of configuration at phosphorus in the nucleotidyltransferase-catalyzed reaction of (Sp)-[a-170] dATP. As shown by Table 11, the 31P NMR data are consistent with these configurations, that is, with inversion and not with retention as the stereochemical course of nucleotidyl transfer. Therefore, gentamicin nucleotidyltransferase catalyzes the reaction of (Sp)-[a-'70]dATP (Scheme I) with the C2"-OH of tobramycin to form (Rp)-['70]dAMP-tobramycin (Scheme 11), i.e. with inversion of configuration.
Stereochemical Course of the B. pertussis Adenylylcyclase Reaction-The adenylylcyclase step in Scheme I11 is shown as proceeding with inversion of configuration, as has been shown for adenylylcyclases from two species other than B. pertussis (21,22). In the present work, a crude preparation of the enzyme from B. pertussis was used for the configurational analysis of (Sp)-['70,'80]dAMP; therefore, our conclusion is based on the assumption that the adenylylcyclase of B. pertussis also catalyzes this reaction with inversion. We verified this by carrying out the reaction with (Sp)-[a-'80]ATP as the substrate and the B. pertussis extract as the source of enzyme.
We analyzed the configuration of the resulting 3',5'-cyclic ["OIAMP at phosphorus by the procedure outlined in Scheme I11 for ['70,'s0]dAMP and obtained the results given in Table  111. This particular analysis was complicated by a 20% loss of "0 during the enzymatic cyclization; however, after taking this loss into account in calculating the theoretical signal intensities for inversion and retention of configuration, the data clearly showed that the reaction proceeded with inversion of configuration.
Mechanism of Nucleotidyl Group Transfer-Inasmuch as the gentamicin nucleotidyltransferase reaction proceeds with inversion of configuration at P' of dATP, the reaction mechanism must involve an uneven number of displacement steps. The simplest interpretation, and the one we favor, is that the mechanism involves the sequential binding of a nucleoside triphosphate and an aminoglycoside at adjacent binding sites of the enzyme to form a ternary complex, in which the C2"-

TABLE I11 31P NMR intensities for the ethyl esters of 3',5'-cyclic ['"O]AMP
The individual entries quote the percent contribution of each species to the total signal for the equatorial or axial triester. The theoretical values are calculated from the isotopic composition of (Sp)-[a-'80]ATP assuming inversion to (Rp)-3',5'-cyclic ["OIAMP or retention to (Sp)-3',5'-cyclic ['"OIAMP and corrected for 20% loss of "0 during the cyclization reaction. Owing to experimental error, 10% or less epimerization at phosphorus would not be apparent in the data. OH of the aminoglycoside is positioned near P' of the nucleotide. Direct in-line displacement of pyrophosphate from P' by the C2"-OH produces the nucleotidylaminoglycoside in a single step accompanied by inversion of configuration at phosphorus. This mechanistic pathway is consistent with both our stereochemical result and the fact that the kinetics for the reaction is sequential. If in the sequential kinetic pathway the conversion of the substrate ternary complex to the product ternary complex had involved an enzyme-mediated double displacement with a n intervening covalent nucleotidyl-enzyme, the nucleotidyl transfer would have taken place with overall retention of configuration at phosphorus.
The present work extends the list of two-substrate, twoproduct phosphotransferases that follow sequential kinetics and proceed with inversion of configuration at phosphorus.
Thus far, all two-substrate, two-product phosphotransferase reactions that proceed with overall retention a t phosphorus follow ping-pong kinetic pathways ( 7 ) . This pattern for Bi Bi reactions is extensively documented and has been rationalized on the basis that the role of covalent intermediates in the simplest Bi Bi phosphotransferases is to protect the phosphoester or phosphoanhydride bond energy during the changeover of acceptors by dissociation and association at a common binding site in ping-pong pathways ( 7 ) . Such a role for the enzyme is unnecessary in the sequential pathways since the donor and acceptor substrates occupy adjacent sites in the ternary complexes and group transfer is a direct, single step process. Ter Ter and Ter Bi reactions such as ATP-dependent synthetases will not necessarily follow this rule, however, since additional roles (such as group translocation between subsites) for covalent intermediates can be expected to be important within quaternary complexes. Stereochemical analysis of such reactions may uncover these alternative roles for covalent intermediates.