Nickel(II), Copper(II) and Zinc(II) Complexes of 9-[2- (Phosphonomethoxy)ethyl]-8-azaadenine (9,8aPMEA), the 8-Aza Derivative of the Antiviral Nucleotide Analogue 9-[2-(Phosphonomethoxy)ethyl] adenine (PMEA). Quantification of Four Isomeric Species in Aqueous Solution

The acidity constants of the twofold protonated acyclic nucleotide analogue 9-[2-(phosphonomethoxy)- ethyl]-8-azaadenine, H2(9,8aPMEA)±, as well as the stability constants of the M(H;9,8aPMEA)+ and M(9,8aPMEA) complexes with the metal ions M2+ =Ni2+, Cu2+ or Zn2+, have been determined by potentiometric pH titrations in aqueous solution at I=0.1 M (NaNO3) and 25℃. The result for the release of the first proton from H2(9,8aPMEA)+ (pKa= 2.73), which originates from the (N1)H+ site, was confirmed by UV-spectrophotometric measurements. Application of previously determined straight-line plots of log KMM(R-PO3) versus PKH3(R-HPO3)' for simple phosph(on)ate ligands, R- PO-, where R represents a residue without an affinity for metal ions, proves that the primary binding site of 9,8aPMEA2- is the phosphonate group for all three metal ions studied. By stability constant comparisons with related ligands it is shown, in agreement with conclusions reached earlier for the Cu(PMEA) system [PMEA2-=dianion of 9-[2- (phosphonomethoxy)ethyl]adenine], that in total four different isomers are in equilibrium with each other, i.e. (i) an open isomer with a sole phosphonate coordination, M(PA)op, where PA2-=PMEA2-or 9,8aPMEA2-, (ii) an isomer with a 5-membered chelate involving the ether oxygen, M(PA)cl/o, (iii) an isomer which contains 5- and 7-membered chelates formed by coordination of the phosphonate group, the ether oxygen and the N3 site of the adenine residue, M(PA)cl/O/N3, and finally (iv) a macrochelated isomer involving N7, M(PA)cl/]N7. The Cu2+ systems of PMEA2- and 9,8aPMEA2- behave quite alike; the formation degrees for Cu(PA)op, CuM(PA)cl/O, Cu(PA)cl/O/N3 and Cu(PA)cl/N3 are approximately 16, 32, 45 and 7%, respectively, which shows that Cu(PA)cl/N7 is a minority species. In the Ni2+ and Zn2+ systems the open isomer is the dominating one followed by M(PA)cl/O, but there are indications that the other two isomers also occur to some extent.

PMEA and its relatives affect the viral reproduction cycle at the stage of DNA synthesis, i.e., they serve in their diphosphorylated form as substrates for polymerases and lead after their incorporation to the termination of the growing nucleic acid chain [1]. Since polymerases depend on the presence of metal ions [4], we have studied over the past few years the metal ion-binding properties of PMEA in detail [2,5,6], and suggested also a mechanism [7] which explains why diphosphorylated PMEA is initially an excellent substrate for nucleic acid polymerases [8,9].
The stability determining binding site of PMEA 2is the phosphonate group; however, biologically important metal ions like Mg2+, Ca2+, Mn 2+ and Zn 2+ are able to interact also with the ether oxygen atom and this gives rise to the following intramolecular equilibrium (1) [2,5,6]: () This proposed metal ion-ether oxygen interaction is crucial for the suggested polymerase mechanism [7] which agrees with the observation that deletion of this ether oxygen or a change in its position in the aliphatic chain leads to compounds which are biologically inactive [8][9][10].

Bioinorganic Chemisoy and Applications
With certain metal ions like Cu PMEA2-may also undergo an adenine interaction. This adenine interaction occurs for a minority species via N7 [11], i.e., the phosphonate-coordinated metal ion forms a macrochelate as indicated in equilibrium (2), phospbonate-ribose-base 2/ phosp__honate-r --- base-e (2) and which is well known to occur in the complexes of AMP, where a phosphate group is the primary binding site [12,13]. The majority species, however, results with Cu 2+ from an interaction with N3 [2,11,14] in such a way that a M(PMEA) species, which exists as a fivemembered chelate (eq. (1)), forms in addition a sevenmembered chelate involving N3; this .species is designated as M(PMEA)cvom3 and consequently, the macrochelated (eq. (2)) and ether oxygen-bound isomers (eq. (1)) are abbreviated as M(PMEA)cvN7 and M(PMEA)vo, respectively, and.the open isomer seen in equilibria (1) and (2) as M(PMEA)op. The indicated situation regarding Cu(PMEA) is most fascinating because for the first time a quantitative evaluation of a system in which four isomers occur in equilibrium was possible 11 ].
The relative affinities of N3 versus N7 of an adenine residue are of general interest since N7 is exposed to the solvent in the major groove of DNA whereas N3 is located in the minor groove [15]. Therefore it was desirable to confirm the observations summarized above for M(PMEA) systems with another acyclic nucleoside phosphonate. We selected 9-[2-(phosphonomethoxy)ethyl]-8-azaadenine (9,8aPMEA) [16], which also exhibits some antiviral activity [17] and which is shown in its dianionic form together with PMEA 2in Figure 1, and studied its metal ion-binding properties with Ni2+, Cu 2+ and Zn2+. We selected these metal ions since they are known [18] to have a relatively pronounced affinity toward N donors. To complete the picture, the previously obtained equilibrium data [5,11] for the Ni 2+ and Zn 2+ complexes of PMEA 2were now also evaluated regarding the equilibrium scheme (3) Vol. 2, Nos. [3][4]2004 Nickel(ll), Copper(ll) and Zinc(ll)Complexes of 9-[2-Phosphonomethox)O -8-azaadenine (9,8aPMEA) where PA 2-= PMEA 2or 9,8aPMEA2-. The presented results prove that at least with Cu 2+ all four isomers occur in solution with both ligands, whereas with Ni 2+ and Zn 2+ the proof of their occurrence is more difficult since the differences in complex stability between the various species are small.  [1] and of 9-[2-(phosphonomethoxy)ethyl]-8-azaadenine (= 9,8aPMEA2-), together with the structure of PME-R2-, where R is a non-interacting residue, and which represents the metal ioncoordinating properties of the ether-phosphonate chain occurring in PMEA 2and 9,8aPMEA2-. A further ligand to be considered in this study is 9-(4-phosphonobutyl)adenine, which is abbreviated as dPMEA 2-(= 3-deoxa-PMEA -) to indicate that its structure corresponds to that of PMEA 2except that the ether O atom is replaced by a CH2 group.

Materials
Twofold protonated 9-[2-(phosphonomethoxy)ethyl]-8-azaadenine, i.e. Hz (9,8aPMEA) , was synthesized by alkylation of 8-azaadenine with a synthon carrying the structural constituents of the required side chain [16]; in fact, the same lot of compound was used as previously [19]. The aqueous stock solutions of the ligand were freshly prepared just before the experiments by dissolving the substance in deionized, ultrapure (MILLI-Q185 PLUS; from Millipore S.A., 67120 Molsheim, France) CO2-free water, adjusted to pH about 8.5 by adding 2 equivalents of 0.1 M NaOH.
The disodium salt of 1,N,N',, potassium hydrogen phtha|ate, HNO3, NaOH (Yitrisol), and the nitrate salts of Na+, Ni2+, Cu 2+ and Zn + (all pro analysi) were from Merck AG, Darmstadt, FRG. All solutions for the potentiometric pH titrations were prepared with ultrapure CO2-free water. The buffer solutions (pH 4.00, 7.00, 9.00 based on the NBS scale; now NIST) used for calibration of the pH-measuring instruments were from Metrohm AG, Herisau, Switzerland.
The exact concentrations of the stock solutions of the divalent metal ions were determined by potentiometric pH titrations via their EDTA complexes. The exact concentration of the ligand solutions was in each experiment newly determined by the evaluation of the corresponding titration pairs, i.e. the difference in NaOH consumption between solutions with and without ligand (see Section 2.3).

Potentiometric pH Titrations
The pH titration curves for the determination of the equilibrium constants in H20 were recorded with a Metrohm E536 potentiograph connected to a Metrohm E665 dosimat and a Metrohm 6.0222.100 combined macro glass electrode. The pH calibration of the instrument was done with the mentioned buffer solutions at pH 4.00, 7.00 and 9.00. The titer of the NaOH used was determined with potassium hydrogen phthalate. The direct pH meter readings were used in the calculmions of the acidity constants; i.e. these constants determined at I 0.1 M (NaNO3) and 25 C are so-called practical, mixed or Bronsted constants [20]. They may be converted into the corresponding concentration constants by subtracting 0.02 from the listed pK, values; this conversion term contains both the junction potential of the glass electrode and the hydrogen ion activity [20,21]. It should be emphasized that the ionic product of water (Kw) and the mentioned conversion term do not enter into our calculation procedures because we always evaluated the differences in NaOH consumption between a pair of solutions, i.e. with and without ligand. The stability constants determined are, as usual, concentration constants.
The individual results for the stability constants showed no dependence on pH or on the excess of metal ion concentration used. The results are in each case the averages of at least 5 independent pairs of titration curves.

Spectrophotometric Measurements
The acidity constant that describes the release of the proton from the (Nl)H site of the adenine residue in H H2(9,8aPMEA)i, PKHz(9,8aPMEA) (eq (4)), was also determined by spectrophotometry. The UV-Vis spectra of 9,8aPMEA (1.2 mM) were recorded in aqueous solution (25 C; I--0.1 M, NaCI) and l-cm quartz cells with a Varian Cary 3C spectrophotometer connected to an IBM-compatible desk computer (OS/2 system) and an EPSON Stylus 1500 printer. The pH of the solutions was adjusted by dotting with relatively concentrated HC1 and measured with a Metrohm 713 pH meter using a Metrohm 6.204.100 glass electrode.

BioinotNanic Chemistry and Applications
The spectra were recorded within the range of 205 to 330 nm; for further details see Figures 2 and 3 in Section 3.1.

RESULTS AND DISCUSSION
Derivatives of purines are well known to undergo self-association via rt-stacking [24]. Therefore, all potentiometric pH titrations (25 C; I 0.1 M, NaNO3), the results of which are summarized below, were carried out with a ligand concentration of 0.4 mM. Under these conditions self-stacking is negligibly small as has been shown for PMEA [5]. Hence, it is ascertained that the results given below reflect the properties of monomeric species.
3.1. Acidity Constants of H2 (9,8aPMEA) From the structure of 9,8aPMEA 2-(see Figure l) it is evident that this species can accept three protons, two at the phosphonate group and one at the N1 site of the 8-azaadenine residue [25,26]. Further protonations at an adenine residue are possible at N7 and N3, but these protons are released with pKa < 0 [27]; similarly, release of the first proton from the -P(O)(OH)2 group of H3(PMEA) + occurs with pKa 1.2 [26,28] and the same may be surmised for H3 (9,8aPMEA) . Hence, in the present study, for which all potentiometric pH titrations were carried out at pH > 2.8, only the following two deprotonation reactions, in which 9,8aPMEA 2and related species like PMEA 2- (Figure 1) are abbreviated as PA 2-(this also holds for other equations further below), need to be considered: Indeed, all the experimental data from the potentiometric pH titrations in aqueous solution could be excellently fitted by taking into account equilibria (4) and (5). The acidity constants obtained in the present study for H2(9,8aPMEA) are given in Table together with some related data [29][30][31].
From a quick comparison of the acidity constants in Table it is immediately evident that the first proton released from H2(9,8aPMEA) according to equilibrium (4) is from the (N1)H site and the second one according to equilibrium (5) from the -P(O)2(OH)-group. This site attribution is confirmed by the spectrophotometric measurements seen in Figure 2; the change in absorption of the H2(9,8aPMEA)/ Table 1 Negative Logarithms of the Acidity Constants of H2(9,8aPMEA) and H2(PMEA) (eqs (4) and (5) [30].
H(9,8aPMEA)pair occurs in this range of wavelengths where protonation/deprotonation reactions of related aromatic moieties are commonly seen [32]. A further reason for the spectrophotometric measurements was that the formation degree of the H2(9,8aPMEA) species that could be reached in the potentiometric pH titrations was relatively low (see Section 2.3). This means that it was desirable to determine the acidity constant for equilibrium (4) also by another independent method. Therefore we measured the absorption spectra of 9,8aPMEA as a function of pH; a representative set of spectra is shown in Figure 2. The evaluation of the same experiment by a curvefitting procedure, but involving more data, is given in Figure 3. Since NaNO3 absorbs in part of the wavelength range needed for the evaluation of 9,8aPMEA data, I was now adjusted to 0.1 M with NaCI. The H 2.73 + 0.08, and this value is final result from two independent series of measurements is PKH2 (9,SaPMEA) in excellent agreement with the constant given in Table and Figure 2. The final result H (PKH2(9,SaPMEA) 2.73 + 0.08 (3Or)) is the average of two independent experimental series. Nickel(ll), Copper(il) and Zinc(ll)Complexes of 9-[2-.Phosphonomethoxy) -8-azaadenine (9, 8aPMEA) The most obvious conclusions from the data in Table are that replacement of (C8)H by a nitrogen atom reduces the pK, of the (N1)H + site by about A pK, 1.5, i.e., this site becomes considerably more acidic as follows from a comparison of entries and 2 with 3 and 5. In contrast, entries 2-4 demonstrate that the nucleobase residue hardly affects the release of the proton from the -P(O)z(OH)-group. However, elimination of the ether oxygen from the R-CH2CH2 -O-CH2 -POchain enhances the basicity of the -POgroup remarkably (cf. entries 2-6).
It should be noted that in formulas like M(H;PA) + the H + and PA 2are separated by a semicolon to facilitate reading, yet they appear within the same parentheses to indicate that the proton is at the ligand without defining its location.
Indeed, together with equilibria (4) and (5), equilibria (6) and (7) are sufficient to obtain excellent fitting of the titration data (see Section 2.3), provided the evaluation is not carried into the pH range where formation of hydroxo species occurs, which was evident from the titrations without ligand. Of course, equilibria (6) and (7) The results are listed in column 4 of Table 2 together with the constants for the corresponding M(PMEA) complexes and some further related data. The stability constants given in footnote "e" for the M(H;9,8aPMEA) + complexes need to be considered as estimates since the formation degree of these species was low (see Section 2.3). The stability constants of the M(9,8aPMEA) complexes show the trend expected for divalent 3d metal ions, i.e., they vary within the series Ni 2+ < Cu 2+ > Zn2+, and this holds for the constants due to the M(H;9,8aPMEA) + species as well. The analysis of potentiometric pH titrations only yields the amount and distribution of the species of a net charged type; i.e., further information is required to locate the binding sites of the proton and the metal ion in the M(H;9,8aPMEA) species. At first one may ask where the proton is located because binding of a metal ion to a protonated ligand commonly leads to an acidification of the ligand-bound proton [34,35]. Hence, the acidity constants according to equilibriiam (8) are needed; these values are calculated with the data listed in Tables and 2 by application of equation (9) (Table 1). This comparison shows that the proton in M(H;9,8aPMEA) is bound to the phosphonate group, hence, one may tentatively assume that the metal ion is coordinated preferentially to the nucleobase, since a monoprotonated phosphonate group is only a weak binding site. Indeed, this suggestion agrees with evidence obtained previously for other related M(H;PA) + species [5,14,36].

Evaluation of the Stabilities of the M(9,8aPMEA) Complexes
For the M(9,8aPMEA) complexes the question arises: Does the 8-azaadenine residue also participate in metal ion binding next to the phosphonate group? Should such an additional interaction with the nucleobase residue occur then it has to be reflected in an increased complex stability [37]. Hence, it is necessary to define the stability of a pure -PO-/M 2+ interaction. This can be done by applying the previously defined M H plots for simple [5] straight-line correlations which are based on log KM(R_PO3) versus PKH(R_PO3) phosphate monoesters [38] and phosphonates [5]; these ligands are abbreviated as R-PO-, where R represents a noncoordinating residue. The parameters for the corresponding straight-line equations, which are defined by equation (l l), have been tabulated [2a,5,39,40], i.e., the slopes m and the intercepts b with the y-axis. Hence, with a known pKa value for the deprotonation of a -P(O)2(OH)group an expected stability constant can be calculated for any phosph(on)ate-metal ion complex.
M H according to equation (11) are shown in Figure 4 for The plots of log KM(R_PO3) versus PKH(R_PO 3) the :1 complexes of Cu 2+ and Zn2+, as examples, with the data points (empty circles) of the eight simple ligand systems used [5] for the determination of the straight baselines. The two solid circles refer to the corresponding M(9,8aPMEA) complexes and the crossed ones to the M(PMEA) species. For further comparison also the data points for the related M(PME-R) (solid squares) and M(dPMEA) (empty squares) systems are shown.
All the latter mentioned data points are clearly positioned above their reference lines thus proving that beyond the -PO]-/M 2+ binding additional interactions occur. The smallest stability increase is observed for the M(dPMEA) complexes, where dPMEA 2-= 3'-deoxa-PMEA 2-(i.e., the ether O is replaced by CH2) 9-(4-phosphonobutyl)adenine ( Figure 1); in these instances macrochelates according to equilibrium (2) involving N7 of the adenine residue are formed 11]. For the M(PME-R) complexes the stability increase is more pronounced and clearly attributable to equilibrium (1) since no other additional binding site but the ether O atom is available (Figure t) [5,30]. However, the stability increase observed for the Cu(9,8aPMEA), Cu(PMEA) and Zn(9,8aPMEA) species is much larger than the one for the M(dPMEA) and M(PME-R) complexes, thus indicating that an accumulation of extra interactions occurs as it is depicted in the .equilibrium scheme (3). No meaning should be attributed to the apparent equality of the stability increase seen in Figure 4 for the Zn(PMEA) and Zn(PME-R) complexes because the stability constant for Zn(PMEA) is only an estimate carrying a large error limit (see Table 2, entry c in column 4).

Extent of the Total Amount of Chelates Formed in the M(PA) Systems
Before considering the situation in the M(PMEA) and M(9,8aPMEA) complexes according to the equilibrium scheme (3) in more detail (see Section 3.5), it is appropriate to evaluate first the total amount of closed species, M(PA)evtot, for all four PA 2ligands considered (Figure 1) because evidently the sum of all the closed species, independent of their structure, is responsible for the observed stability increase. Stability enhancements like those seen in Figure 4 can be quantified by the differences between the experimentally (exptl) measured stability constants and those calculated (calcd) according to equation (11); this difference is defined in equation (12) Table 2 the values for the terms of equation (12)   The least-squares lines (eq. (11)) are drawn through the corresponding 8 data sets (O) taken from ref. [38] for the phosphate monoesters and from ref. [5] for the phosphonates. The points due to the equilibrium constants for the M2+/pA 2systems are based on the values listed in Tables (column 4) and 2 (columns 4 or 6). The vertical broken lines emphasize the stability differences from the reference lines; they equal log AM/pA as defined in eq. (12) for the M(PA) complexes. All the plotted equilibrium constants refer to aqueous solutions at 25 C and I 0.1 M (NaNO3).

Nickel(ll), Copper(ll) and Zinc(ll)Complexes oJ'9-[2-Phosphonomethoxy)
-8-azaadenine (9,8aPMEA) tt 0 All values for log AM/PA are positive with the single exception of the one for the Zn(dPMEA) complex where log AZn/dPMEA is zero within the error limits (Table 2, entry 4c in column 6). The 'total' of the dimensionless intramolecular equilibrium constant, Kvtot, is defined by equation (13) (see also below eq. (21) and values for Kl/tot can be calculated following known procedures [5,12,37,39,40], i.e., via equation (14): Knowledge of Kl/to allows then according to equation (15) % M(PA)cl/tot 100"Kt/tot](1 +Kl/tot) to obtain the percentage of the sum of all the closed isomers (cl/tot) present in equilibrium, i.e., their total formation degree. The corresponding results for the four PA 2ligands of Figure and their Ni2+, Cu e+ and Zn 2+ complexes are summarized in columns 6-8 of Table 2.
The most easily understood result of the evaluation is the one given under entry 3 in Table 2 because the PME-R 2ligand can only form the two isomeric complexes seen in equilibrium (1) and % M(PA)l/tot % M(PME-R)vo (Table 2, column 8). Similarly simple is the situation with dPMEA 2because in this case an additional metal ion interaction, next to the one with the -POgroup, must occur with the adenine residue and it was previously concluded [11] that this is the N7 site; hence, here equilibrium (2) applies. Consequently, for the M(dPMEA) complexes it holds Kl/tot K/N7, as defined by equation (17) It is evident that the situation for the complexes formed with PMEA 2and 9,8aPMEA > is more complicated, since more possibilities for the formation of closed isomers exist, and that these possibilities materialize at least in part is evident from the observed rather large stability increases, log AM/PA (Table 2, column 6), and also from the high formation degrees calculated for % M(PA)mot. Furthermore, it is rcvealing to see that the values given in column 8 of Table 2 for % M(PMEA)cvtot and % M(9,8aPMEA)I/,ot (entries and 2) are for a given metal ion very similar or even identical within their error limits. 3.5. Formation Degrees of the Four Isomers Existing in Equilibrium for the M(PMEA) and M (9,8aPMEA) Species Up to now the Cu2+/PMEA system is the one most thoroughly studied. Indeed, it had originally been proven [14] that three isomers are important for the Cu(PMEA) system [7]: (/) An 'open' isomer, Cu(PMEA)op, in which the metal ion is solely coordinated to the phosphonate group; (ii) an isomer which involves the ether oxygen (see Figure 1) as shown in equilibrium (1), designated as Cu(PMEA)cvo; and (iii) an isomer in which not only a 5-membered chelate but in addition a 7-membered one involving N3 exists, i.e. Cu(PMEA)cvom3. More recently [11] evidence was provided that there is a fourth isomer, a minority species, in which the phosphonate-coordinated Cu 2+ interacts with N7 of the adenine residue forming a macrochelate, Cu(PMEA)dmT, as indicated in equilibrium (2). In this context it is important to emphasize that for steric reasons no macrochelate involving only N3 can be formed by PMEA 2and Cu 2+ [2a]. If one tries to form such a species with molecular models, one automatically forces the ether oxygen into the coordination sphere of the metal ion, giving rise to the already mentioned Cu(PMEA)vom3 isomer [2a]. If one summarizes all these results then the simple equilibrium (7a) must be replaced for the Cu(PMEA) system by the rather complicated equilibrium scheme (3) already introduced in Section 1. Of course, exactly the same reasonings also apply to the PMEA 2complexes formed with Ni 2+ and Zn 2+ as well as for the M(9,8aPMEA) species.
For these systems a quantitative evaluation toward the formation degree of the various isomers needs now to be carried out.
The four equilibrium constants seen in scheme (3) are defined by the already mentioned equations (16) and (17) together with the also necessary equations (18) With these definitions the measured overall stability constant (eq. (7b)) can be redefined as given in equations (20a) The connection between the overall intramolecular equilibrium constant Kl/toD already introduced in Section 3.4, and the accessible stability enhancement (eq. (12)) is given by equations (21a) Values for gl/to were already calculated with equations (12) and (14) in Section 3.4; they are listed in column 7 of Table 2 (entries and 2). The relation between Kvtot and the other three intramolecular equilibrium constants follows from equations (2 b) and (2 c). Based on the reasonable assumption [7] that the stability of the M(PA)cvo isomer, where PA z-= PMEAor 9,8aPMEA2-, is well represented by that of the 5-membered M(PME-R)cvo species (Figure 1) and the stability of the M(PA)vN7 isomer by that of the M(dPMEA)clm7 macrochelate, values for Kvo, which define the position of equilibrium (1), and KvN7, which refer to equilibrium (2), are also known (see the second to the last paragraph in Section 3.4). Hence, the only unknown constant in equation (21 e) is Kvom3 (eq. (19)) and thus values for this constant can be obtained, and consequently, the formation degrees for all four isomers appearing in scheme (3) can now be calculated. The corresponding results are summarized in Table 3 for the M(PMEA) and M(9,8aPMEA) systems; as far as the error limits are concerned it needs to be emphasized that three times the standard errors (3o) are given. From Table 3 it is evident that Cu(PMEA) and Cu(9,8aPMEA) (entries b and 2b) have practically identical properties: The Cu(PA)cvOm3 species with the 5-and 7-membered chelate rings dominate with formation degrees of about 45% followed by Cu(PA)cvo with about 30%. As far as Cu(PMEA)cvOm3 is concerned, the result with 41 + 12% is within the error limit identical with the previously obtained 49 + 10% where the formation of the fourth isomer, Cu(PMEA)vN7, had not been taken into account [5,7]. This demonstrates immediately that the Cu(PA)cvN7 isomer must be a minority species; indeed, the present calculations show that the formation degrees of Cu(PMEA)cvN7 and Cu(9,8aPMEA)cvN7 amount only to about 7% (see also ref. [11]). it is interesting to see that for the Ni(PMEA) and Zn(PMEA) systems about 50% each exist as the open isomer and the remaining half of the species is present as chelates (Table 3, entries a and c). In the case of Ni(PMEA) all three chelated isomers occur with comparable concentrations though the formation degrees of Ni(PMEA)cvo and Ni(PMEA)vo/N3 appear to be slightly favored. With Zn(PMEA) the Zn(PMEA)vo isomer seems to be the dominating species, the formation degrees of the other chelates being zero within the error Nickel(ll), Copper(II) and Zinc(ll)Complexes of 9-[2-Phosphonomethoxy) -8-azaadenine(9, 8aPMEA) V V limits; here it should be recalled that the overall stability constant for Zn(PMEA) is an estimate only (Table  2, entry e) [5]. For Zn(9,SaPMEA) ( Table 3; entry 2c) the results are more clear-cut since in this case the overall stability constant of the complex could actually be measured (see Section 2.3): Again the Zn(9,8aPMEA)cvo chelate dominates. However, in this case it may be helpful to rewrite the results for Zn(9,8aPMEA)cvo, Zn(9,8aPMEA)cm7 and Zn(9,SaPMEA)cvom3 with one standard deviation (lcy) only, that is 32 + 4, 10 +/-7, and 24 + 9%, respectively. This view confirms that Zn(9,8aPMEA)cvo dominates but that Zn(9,8aPMEA)cvom7 most likely also exists, whereas Zn(9,SaPMEA)cmv is definitely also for this system a minority species. The great similarity between the Zn(PMEA) and Zn(9,8aPMEA) systems is evident, despite all shortcomings, from a comparison of the values in entries c and 2c of Table 3. This is also true for the Ni(PMEA) and Ni(9,SaPMEA) systems for which the values seen in entries a and 2a of Table 3 overlap within their error limits.

CONCLUSIONS
The presented results prove that systems in which four different isomers occur in equilibrium in solution can be treated in a quantitative way. They prove further that both N3 and N7 of an adenine residue may bind to metal ions provided primary binding sites promoting a favorable steric orientation are available. With regard to nucleic acids this result is of relevance; in fact, that the more basic N7 [27] is suited for such purposes is by now general knowledge 12,39] whereas this property of N3 has only been recognized more recently 14,27b, 35,36a,4 ].