Ternary Copper(II) Complexes in Solution Formed With 8-Aza Derivatives of the Antiviral Nucleotide Analogue 9-[2-(Phosphonomethoxy)Ethyl]Adenine (PMEA).

The stability constants of the mixed-ligand complexes formed between Cu(Arm)(2+), where Arm= 2,2'-bipyridine (Bpy) or 1,10-phenanthroline (Phen), and the dianions of 9-[2-(phosphonomethoxy)ethyl]-8-azaadenine (9,8aPMEA) and 8-[2-(phosphonomethoxy)ethyl]-8-azaadenine (8,8aPMEA) (both also abbreviated as PA(2-)) were determined by potentiometric pH titrations in aqueous solution (25 ( degrees )C; I = 0.1 M, NaNO(3)). All four ternary Cu(Arm)(PA) complexes are considerably more stable than corresponding Cu(Arm)(R-PO(3)) species, where R-PO(3) (2-) represents a phosph(on)ate ligand with a group R that is unable to participate in any kind of interaction within the complexes. The increased stability is attributed to intramolecular stack formation in the Cu(Arm)(PA) complexes and also to the formation of 5-membered chelates involving the ether oxygen present in the -CH(2)-O-CH(2)-PO(3) (2-) residue of the azaPMEAs. A quantitative analysis of the intramolecular equilibria involving three structurally different Cu(Arm)(PA) species is carried out. For example, about 5% of the Cu(Bpy)(8,8aPMEA) system exist with the metal ion solely coordinated to the phosphonate group, 14% as a 5-membered chelate involving the -CH(2)-O-CH(2)-PO(3) (2-)residue, and 81% with an intramolecular stack between the 8-azapurine moiety and the aromatic rings of Bpy. The results for the other systems are similar though with Phen a formation degree of about 90% for the intramolecular stack is reached. The existence of the stacked species is also proven by spectrophotometric measurements. In addition, the Cu(Arm)(PA) complexes may be protonated, leading to Cu(Arm)(H;PA)(+) species for which it is concluded that the proton is located at the phosphonate group and that the complexes are mainly formed by a stacking adduct between Cu(Arm)(2+) and H(PA)(-). Conclusions regarding the biological properties of these azaPMEAs are shortly indicated.

Considering the broad biological activity of PMEA, it is not surprising that many derivatives have been synthesized and studied, [2] and it is now clear that in order to be antivirally active, PMEA and its derivatives must be phosphorylated in the cell to the diphosphate (PMEApp4-) and this is then recognized by DNA polymerases as a substrate and incorporated into the growing nucleic acid chain which is terminated thereafter. [3'4] Knowing that polymerases depend on the presence of metal ions [15'16] and that the nucleoside 5'-triphosphates must be present as complexes (mostly Mg2+) [7] we have been studying complexes of PMEA [8"24] and a mechanism of action has been proposed recently [25'261 in which the correct location of two metal ions at the triphosphate chain for achieving a desired reaction is emphasized. [26] In other words, the correct orientation of the nucleoside 5'-triphosphate or its analogue in the active-site cavity of the enzyme is crucial. [26] One of the ways in which the correct anchoring process of a substrate in the active site of an t'z7] enzyme can be achieved, is via stacking interactions of the nucleobase residue, e.g. with an indole moiety of a tryptophan unit. [28] For this reason we became interested in the PMEA relatives, 9-[2-(phosphonomethoxy)ethyl]-8-azaadenine (9,8aPMEA) and 8-[2-(phosphonomethoxy)ethyl]-8azaadenine (8,8aPMEA) (see Fig. 1), and the question was: Do the stacking properties of PMEA, 9,8aPMEA and 8,8aPMEA differ? To this end we measured the stabilities of the mixed iigand Cu(Arm)(PA) complexes where Arm 2,2'-bipyridine (Bpy) or 1,10-phenanthroline (Phen) and 2 2 PA-9,8aPMEAor 8,8aPMEA2-. These Cu(Arm)(PA) complexes can fold such that the

Potentiometric pH Titrations
The apparatus for the potentiometric pH titrations, the calibration procedure, the computers, and the calculatio_n methods used now are the_same as in [24]. The stability constants hmt(H.9 8aP4ZA) and Kt (9 8aP4ZA complexes were determined under the corresponding conditions. Furthermore, the above conditions are similar to those described in [24].

Spectrophotometric Measurements
The UV-Vis spectra of the Cu2+/Phen/9,8aPMEA or 8,8aPMEA systems were recorded in aqueous solution and 1-cm 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 NaOH and measured with a Metrohm 713 pH meter using a Metrohm 6.204.100 glass electrode. Further details are given in the legend of Figure 4 in Section 3.5.

RESULTS AND DISCUSSION
All potentiometric pH titrations (25 C; I 0.1 M, NaNO), the results of which are summarized below, were carried out with a ligand concentration of 0.4 mM and at CuZ+/Arm concentrations equal to or below 4.4 mM. Under these conditions self-stacking of the ligands is negligibly small as has been shown [81 for PMEA; the same applies to the self-association of Cu(Phen)r+. [3] This means, the self-association is negligible for any of the reactants under the present experimental conditions and the results given below certainly refer to monomeric species.

Definition of the Equilibrium Constants
The ligands 9,8aPMEA 2and 8,8aPMEA2-, abbreviated as PA 2- ( provided the evaluation of the data is restricted to the pH range below the beginning of the formation of hydroxo complexes which was evident from the titrations of M 2+ without ligand. It should be noted that in formulas like M(H;PA) + the H + and the 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.
Equilibria (3a) and (4a) are connected via equilibrium (5a), and the corresponding acidity constant (eq. (5b)) may be calculated with equation (6) The equilibrium constants according to equations (3), (4), and (5) are listed in columns 2, 3, and 4 of Table 1, respectively. The acidity constants of the ligands (see footnote "a" in Table 1) and the stability constants of the binary Cu(H;PA) + and Cu(PA) complexes will be discussed in a different context. [33] Here we concentrate on the properties of the ternary complexes. 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 Cu(Arm)(H;PA) + species. A comparison of the acidity constants of H2(9,8aPMEA) +, P/H2(9,8aPMEA) 2.73 an.d P/H(9,8aPMEA)= 6.85, with pKCu(Arm)(H;9,8aPMEA) 3.7 (Table 1) of the Cu(Arm)(H;9,8aPMEA)* complexes reveals that the proton in these complexes must be located at the phosphonate group, since metal ion coordination must give rise to an acidification, [36,371 which amounts to A pK a pKIH(9,8aPMEA)-pKCu(Arm)(H;9,8a+PMEA) (6. An evaluation following exactly the route described in [2] leads to the conclusion that for all four Cu(Arm)(H;PA) + systems the stacked species in equilibrium (7)  --log/C(PA) According to the general rule for complex stabilities, K > K?, equilibrium (8a) is expected to be on the left side with negative values for A log KCu/Arm/PA, in agreement with statistical considerations, [38'39] i.e., A log Kcu/statist -0.5. [39] From te values listed in column 5 of  (8) is significantly displaced to the right side. More important, however, is a comparison with the results obtained for the dianion of phosphonomethoxyethane (PME:-), CH3CH2-O.CH2-PO-, which reflects the properties of the side chain of9,8aPMEA 2and 8,8aPMEA'-- (Fig. 1). Indeed, these results, A log KCu/Bt)v/PME 0.13 + 0.04 and A log Kcu/Phen/PME 0.17 + 0.05,[311are considerably smaller than the valuers of A log KCu/Arm/P A for the Cu(Arm)(PA) complexes, which means that the adenine residue contributes to the stability of the Cu(Arm)(PA) species. This comparison thus provides the first clear hint for the occurrence of an intramolecular stacking interaction in these latter mentioned complexes. Another way to evaluate the increased stability of ternary Cu + complexes, independently of the properties of the binary Cu(9,8aPMEA) and Cu(8,8aPMEA) species, rests on the previously established [2'311 straight-line correlations for log KUu(rrmm) RPO versus p/ po) plots (eqs (10) c ) ( 3) and (11) (10) and (11) are shown in Figure 2, where the stability constants log Ko u(Arm) versus the acidity constants pK of the 9,8aPMEA and Cu( m)(P 8,8aPMEA species are also pAlrotte, together with the corresponding dat'a l] of the Cu(Arm)(PME) systems. All these data points are above their reference lines, proving an increased complex stability which must mean [41 that aside from 2+ the Cu(Arm) -phosphonate coordination further interactions take place. The vertical distances in Figure 2 between the data points due to Cu(Arm)(9,8aPMEA), Cu(Arm)(8,8aPMEA) and Cu(Arm)(PME) and the reference lines are a measure for the extent of the intramolecular interactions in these complexes and they can be defined according to equation (12)  which only a -PO3-/Cu(Arm) interaction occurs. The first term on the right hand side in equation (12) is the experimentally determined stability constant (eft. (4)), whereas a value for log can be calculated with the acidity constant P/'"PA' and the straight-line equations (1--0utV(l]ai"As indicated above, such a calculated value har{tifies the stability of the open isomer. The ligand PME 2offers to metal ions the phosphonate group for coordination, but an interaction with the ether oxygen is also possible as has repeatedly been proven. [822'41'421 This gives rise to 5-membered chelates and therefore equilibrium (13a) needs to be considered" Evidence for an enhanced stability of the ternary Cu(Arm)(9,8aPMEA) Cu(Arm)(8,SaPMEA) ((C),O), and Cu(Arm)(PME) (A,,) complexes based on the relationship between log /,,_u(Ann u(Ann)(r-PO3) or log "Cu(Ann)(PA)and PK(R-PO3) or pKIH(pA) in aqueous solution at I 0.1 M (NaNO3) and 25 C. The plotted data for 9,8aPMEA and 8,8aPMEA are from Table and those for PME from [31]. The ASu(Ann) two reference lines represent the log Cu(Ann)(r.po3)versus P/(-PO3) relationship for ternary Cu(Arm)(R-PO3) complexes (eqs (10) and (11)); R-POsymbolizes phosphonates or phosphate monoesters in which the group R is unable to undergo any kind of hydrophobic, stacking or other type of interactions, i.e. ligands like D-ribose 5-monophosphate, methanephosphonate or ethanephosphonate. [31 The broken line holds for Arm Bpy and the solid line for Arm Phen. Both straight lines represent the situation for ternary complexes without an intramolecular ligand-ligand interaction. The vertical dotted lines emphasize the stability differences from the reference lines; they equal log ACu/Arm/PA as defined in equation (12). and now, knowing K, the percentage of the closed isomer, Cu(Arm)(PME)c, in equilibrium (13a) can be obtained with equation (15)" % Cu(Arm)(PME)c 100. K/(I + KI) (15)  For the error limits see footnote 'b' of Table 1. These values are from column 3 in Table 1.
These constants were calculated with eqs (10)or (11) and the H pK(pA values given in footnote 'a' of Table 1. See eqs (13b) and (14).
From the results given in the lower part of Table 2 it is evident that the closed isomer of Cu(Arm)(PME) is an important species with a formation degree of about 75%. Naturally, the formation of the corresponding isomer involving the ether oxygen is also to be expected (see Fig. 1) for the Cu(Arm)(9,8aPMEA) and for Cu(Arm)(8,8aPMEA) systems and we designate it as Cu(Arm)(PA)vo. However, the log ACu/Arm/PA values listed in column 4 of Table 2 are by about 0.7 to log unit larger for the latter mentioned complexes than for the Cu(Arm)(PME) species and this must mean that in the systems with 9,8aPMEA and 8,8aPMEA, next to Cu(Arm)(PA)op and Cu(Arm)(PA)l/o, a third isomer must occur which involves the adenine residue. The ligands 9,8aPMEA zand 8,8aPMEA 2offer only two such possibilities" The phosphonate-coordinated Cu(Arm) 2+ forms (i) a macrochelate with one of the nitrogens of the adenine residue, or (ii) an intramolecular stack between the aromatic ring systems of Bpy/Phen and the adenine moiety. That the first possibility is not of relevance has been discussed in detail for 3'-deoxa-PMEA, [21 and the same arguments also apply here, whereas for the second possibility involving intramolecular stacks, many examples exist. [2,28,30,31,39'40'43] Hence, the additional enhanced complex stability may be attributed indeed to intramolecular stack formation. Application of space-filling molecular models reveals that the adenine residue of the 9,8aPMEA or 8,8aPMEA ligands, which are equatorially chelated to Cu(Arm) 2+ via the phosphohate group and the ether oxygen, cannot stack well with the aromatic rings of the also equatorially coordinated Arm; a substantial and strain-free overlap of the aromatic systems is only possible if the ether oxygen is not equatorially coordinated to Cue+. This latter situation is depicted in Figure 3 for 9,8aPMEA. However, from the molecular models it is also evident that an apical ether oxygen coordination and simultaneous stack formation would be compatible with each other in the Cu(Arm)(PA) species with PA 2-9,8aPMEA or 8,8aPMEA. Hence, there are various intramolecularly stacked Cu(Arm)(PA) species possible including those with somewhat different orientations of the aromatic rings toward each other. As there is at present no way to distinguish these various isomers and conformers from each other, we treat all the stacked species together and designate them as Cu(Arm)(PA)s t. The sum of the above reasonings then gives rise to the equilibrium scheme (16), where the pure phosphonate-coordinated isomer is designated as Cu(Arm)(PA)op. It is evident that the upper branch of this equilibrium scheme reflects equilibrium (13a) while the lower branch reflects the stacking interaction (Fig. 3).
With these definitions the experimentally accessible equilibrium constant (4b) can be reformulated as equation (20): [20a,31] [ In those instances where the stacked species do not form, the above equations reduce to the twoisomer problem treated in equations (13) and (14). It is evident that/I Kl/tot according to equation (21a) can be calculated via the values log ACu/Arm/PA as defined by equation (12) and listed in the upper part of column 4 in Table 2.  The results are the same for both calculation methods yet the error limits are understandably larger for the second method (data not shown).
The resulting K values are given in the fourth column of the upper part of Table 3 Table 3.
Considering the equilibrium scheme (16) and the corresponding results summarized in Table  3 several conclusions are evident: (i) All three structurally different species are formed in appreciable amounts in the Cu(Phen)(9,8aPMEA) and Cu(Phen)(8,SaPMEA) systems. (ii) The stacked species (Fig. 3) clearly dominate, reaching formation degrees of about 80 to 90%. (iii) Consequently, the formation degree of the five-rnembered chelates involving the ether oxygen is suppressed, roughly speaking to about 10%, compared with the approximately 75% present in the Cu(Arrn)(PME) systems (cf Table 2). Derivatives of the Antiviral Nucleotide Analogue 9-[2-(Phosphonomethoxy)ethyl)Adenine(PMEA) A further aspect that warrants emphasis is the fact that the values for Ki/st are lower by a factor of about one half for the complexes containing Bpy compared with those containing Phen (Table 3,  this is the result of the smaller aromatic-ring system of 2,2'-bipyridine, compared to that of 1,10phenanthroline, which gives rise to a less pronounced overlap with the adenine residue.

Spectrophotometric Confirmation of Stacking in the Cu(Phen)(PA) Systems
The results of Section 3.4 provide very clear though indirect evidence, obtained via stability constant comparisons, for stack formation in the ternary complexes. Direct evidence for the formation of the stacks can be obtained via either H NMR shift experiments [28'4] or spectrophotometric measurements. [28,,46,471 The first-mentioned method is excluded because of the line broadening effects of Cu2+. However, the second method, which is based on the experience that the formation of stacked adducts is connected with the observation of charge-transfer bands, [28''46'47] should be suitable.
Indeed, the spectrophotometric measurements carried out for the Cu(Phen)(9,SaPMEA) and Cu(Phen)(8,SaPMEA) systems, which are summarized in Figure 4, confirm the formation of stacks in the ternary species. The difference spectra reveal the occurrence of new absorption bands at approximately_ 335 and 350 nm in accordance with previous observations of various Cu(Bpy) 2+ [46] and Cu(Phen) 2+ [47] systems with nucleotides, p '48] The spectrophotometric measurements seen in Figure 4 are not very precise. It was difficult to adjust a pH of 4.5 in the solutions without obtaining a precipitation of hydroxo complexes and the values of the difference-absorptions (AA) measured are very small. Therefore, a quantitative evaluation of these data is not appropriate; important for the present context is the simple fact that the data confirm stacking.  [Cu(Phen) 2+ is nearly completely formed under the given conditions]. NaNO3 was added to all four solutions to maintain I 0.1 M. The pH was always adjusted to 4.50 + 0.02 by dotting with relatively concentrated NaOH (at higher pH hydroxo-complex formation occurs). See also Section 2.3.

CONCLUSIONS
The present study reveals several interesting properties of PMEA and its derivatives: (i) It is relatively astonishing to observe that the binding of the side chain either at N9 or at N8 (see Fig. 1) does not affect the stacking properties in the ternary complexes studied; i.e., the extent of stacking is within the error limits identical in Cu(Arm)(9,8aPMEA) and Cu(Arm)(8,8aPMEA) ( Table 3, column 6 in the lower part). This means most likely that the side chain is long enough to provide the flexibility needed for a favorable orientation of the aromatic rings in the stacks in both types of complexes. (ii) The extent of stacking of the two azaPMEAs in their Cu(Arm) 2+ complexes is within the error limits identical with that of PMEA [] and 3'-deoxa-PMEA [21 in the corresponding ternary complexes. This means, the deletion of the ether oxygen from the side chain as well as the presence of an 'extra' nitrogen atom in the adenine residue do not affect the stacking properties of the purine system. (iii) Furthermore, even the stacking properties of the Cu(Arm)(AMP) species [441 are within the error limits identical with those of the mentioned ternary complexes. In other words, as far as the stacking properties are concerned, all of the mentioned PMEA derivatives resemble closely the parent nucleotide, adenosine 5'-monophosphate (AMp2-).
A further point of interest is that 9,8aPMEA shows in vitro some antiviral activity whereas the 8,8a isomer does not. [49,5] Why? Based on the presented results one may suggest that both nucleotide analogues could possibly bind via stacking to the active site of the enzyme in question, but that the orientation within the active site is different preventing thus the desired I26] biological action.