Interactions between metal cations and the ionophore lasalocid. Part 13. Structure of 1:1 and 2:1 lasalocid anionndash;divalent cation complexes in methanol

摘要

J. CHEM. SOC. PERKIN TRANS. 2 1995 1939 Interactions between metal cations and the ionophore lasalocid. Part 13.' Structure of 1:1 and 2 :1 lasalocid anion-divalent cation complexes in methanol Mostafa Mimouni; Patrice Malfreyt," Rachid Lyazghi," Marc Palma: Yves Pascal,b Giirard Dauphin and Jean Juillard a URA CNRS 434, UniversitP Blaise Pascal, 63177 Aubi2re Cedex, France ' URA CNRS 485, UniversitP Blaise Pascal, 63177 AubiGre Cedex, France +The successive formation of complexes MA and MA, of alkaline-earth cations M2 + with lasalocid HA is observable in methanol. Knowing the corresponding formation constants, it was possible to access NMR parameters specific to these two types of species for the four alkaline-earth cations. 13C and 'H chemical shifts are reported for MA' and MA,; 'H-IH coupling constants for MA+ only. Experiments at low temperature and experiments using the paramagnetic cation Mn2 + determined the coordination sites.All these data show an appreciable variability of coordination in the series of alkaline-earth cations. Computations using the semi-empirical quantum methods AM 1 and PM3 and Monte-Carlo simulations in methanol according to BOSS, mainly on MgA' and BaA' fully confirmed these findings. 'The natural ionophore lasalocid (Fig. 1) is known to be able to by acting on the respective concentrations of the components. 1ransport both divalent and monovalent cations across NMR parameters corresponding to these two species could thus be obtained in CD,OD. This was done for all the alkaline-earth cations.Previous work on 13C and 'H NMR of the free acid, anion and potassium salt of lasalocid in methanol 'facilitated both acquisition and interpretation of these data. Using a paramagnetic cation, Mn2 , yielded the coordination sites + involved in these two species. All these experiments were expected to supply answers to the two questions: What are the structures of the two successive lasalocid complexes? How do Fig. 1 Lasalocid formula with carbon and oxygen numbering scheme they vary as a function of the cation? Additional data were obtained by modelling interactions and structures usingmembranes. With divalent cations, the formation at the water- quantum semi-empirical methods and Monte-Carlo simulations membrane interface of a neutral complex able to migrate taking as a base our recent study on molecular modelling ofthrough the membrane is obviously a two step kinetic process lasalocid free acid and anion.' involving successive formation of a 1:1 and then a 2 :1complex. In solvent systems, such as the water-chloroform biphasic system, only the complete reaction leading to the formation, Experimental from the ionophore HA and the cation M2+, of the neutral salt MA,, is observed in equilibrium studies. Both formation Chemicals constants and structures of these species were recently Lasalocid was obtained as previously; likewise its tetraethyl- i ivestigated in the water-chlorofonn system for the whole ammonium salt.4 Specification of alkaline-earth and manganese Thealkaline-earth series.Energies associated respectively with perchlorate samples used has already been ~tated.~.~ each of the two steps could not be derived from such work and preparation of the neutral salts MA, was described in a recent how each of the two anions is implicated in the coordination of paper.' Solvents CH30H and CD,OD were as previously tie cation cannot be stated with any certainty. specified. q9 Studying equilibria in a more polar solvent affords the energies associated with the successive formations of the two NMR experiments species MA' and MA, and also provides some insight into These were conducted as previously,' using the same apparatus their structure, i.e. the coordination sites of the cation and (Bruker MSL 300) and methods. the conformation of the ligands.Standard or apparent thermo- dynamic functions for the reactions (I) and (2) were previously Computations Semi-empirical quantum calculations were carried out using AM1 11,12 andPM3"-13 asstatedinthelastpaper." Monte-Carlo statistical mechanics simulations were carried out with the BOSS14 program. Atomic charges were first calculated using AM1 and then introduced in the parameter file. azquired in the solvent methanol for alkaline-earth cations Except when otherwise stated, a first minimization was done in (AGO,, AH" and AC; and AVO4) for some transition a continuum with a relative permittivity (dielectric constant), metal cations AH" and ASo7) and for some heavy D= 32.66. The solute molecule was then placed with 378 metal cations (AGO, AHo and AS" *).methanol molecules l5 in a cubic cell (ca. 26.7 x 26.7 x 40 A3) Knowledge of the equilibria then allows the predominant with periodic boundary conditions. Metropolis ' and preferen- formation of either MA+ or MA, in methanol to be favoured tial sampling were used in the isothermal isobaric ensemble 1940 at 25 "C and 1 atm. For these simulations equilibration was carried out for 3 x lo6 configurations, volume being prevented from moving during the first 3 x lo5 configurations, followed by averaging for 3 x lo6 configurations. Intermolecular interaction energies were obtained using the AMBER/OPLS force field. ' This force-field is constituted of intramolecular energy terms identical to those of the AMBER force-field l9 and inter- and intra-molecular energy terms between non-bonded atoms.These last interactions are represented by a Coulombic term and a Lennard-Jones 6, 12 term. For two molecules A and B in which Aij = (amp;Ajj); Cij = (CiiC..)*;Aii = 4 eiioii12;JJ Cii = 4eiioii6, oii and eii being the usual Lennard-Jones parameters. The non-bonded contribution to the intramolecular energy is evaluated with the same expression for all pairs of atoms separated by more than three bonds. Considering solute, solvent molecule and solvent box dimensions, the cut-off was set at 11 A for both solute-solvent and solvent-solvent interactions. It was very large for intramolecular interactions in the solute.It was accepted that all C-C and C-0 bonds could vary during the simulation. For each interaction the allowed change of each of the dihedral angles of lasalocid was set at less than 5" to limit the rejection rate of the generated structures; this rate was thus contained at about 40. All these computations were performed using a DECa3000/ 400s computer. Results and discussion 13Cand 'H NMR data on alkaline-earth lasalocid complexes Two types of solutions were prepared. Type 1 resulted from dissolving the tetramethylammonium lasalocid salt and the metal perchlorate in CD,OD analytical concentrations being respectively cz = 0.1 mol dmp3 and c;l; = 0.2 mol drn-,. Assuming like formation constants in the two solvents CH,OH and CD,OD, calculations using constants previously obtained show that in these conditions the MA+ species is the predominant one (percentages from 96 for Mg2 + to 99.9 for Ba2+).This was also obtained by dissolving the lasalocid metal salt MA, in CD,OD and the metal chloride or perchlorate MX, in a 1 :2 ratio (type 1' solution). Type 2 solutions were obtained by dissolving the neutral lasalocid alkaline-earth salts at a concentration of 0.1 mol dm-, in CD,OD. Calculations using constant values previously reported for the formation of both MA+ and MA, gave the proportion in solution of the three species MA,, MA+ and A-; percentages of A engaged in MA, lay between 76 for Ba2+ and 84 for Mg2+ in these concentration conditions. All the 13C and 'H spectra obtained with these two types of solutions were well resolved, denoting rapid exchange between various species or between the two ligands involved in the same species.and 'H resonance frequencies were independently and unequivocally assigned for each solution from the 'H spectrum, the 13C broad-band and J-mod spectra, the 'H-'H (COSY 45) and 13C-'H correlation contour plots as previously described.' 'H and 13C spectra corresponding to a 100species were obtained by correcting for the presence of other species using 6 = in which 6 is the actual experimental chemical shift, di the chemical shift of species i and ri the fraction of the total amount of the ionophore engaged in the species i. The MA' spectra were readily obtained from solutions of type 1 or 1 ', which then gave the MA, spectra from J. CHEM.SOC. PERKIN TRANS. 2 1995 (b1 19 56 15 . I 13 1; lII111IIIIII1II1l1,, 200 150 100 50 PPm Fig. 2 13C spectra ('H decoupled) in methanol of lasalocid-calcium 1: 1 complex, CaA' alone (a) and in presence of manganese chloride (b, c) (ratio Mn/Ca = r). (a) r = 0 analytical concentrations c*-(CaA,) = 0.026, c*(CaCI,) = 0.14 mol dm-3 which results in cCaA+= 99.4 and cA-= 0.6. (b)r = lo-*. (c) r = 4 x lop2. solutions of type 2 with corrections for the presence of MA+ and A -.'These corrected values of 'H and 3C chemical shifts are reported in Tables 1 and 2. 'H-'H coupling constants were also obtained for the MA' species using procedures already reported. 'Those which could be accessed are given in Table 3.No attempt was made to determine coupling constants for the MA, species since they would have been mean values for the two ligating anions. Identification of the coordination sites was achieved using a paramagnetic cation Mn2 '. Both longitudinal and transverse relaxation times, T, and T,, of a given carbon nucleus are affected by the presence in its neighbourhood of a paramagnetic cation. It is generally accepted that such interactions are mainly dipolar, enhancement of the relaxation times being related to the through-space carbon-cation distance. On the 3C broad- band spectra, T, enhancement corresponds to signal broadening. The experiments conducted here were essentially qualitative; they were intended to provide information on the oxygens involved in the coordination. Starting from a solution in which the calcium species investigated, CaA+ or CaA,, is strongly preponderant, small amounts of MnCl, or MnA, were added.Formation constants of manganese or calcium lasalocid complexes are of the same order of sizes of the two cations Mn2+ and Ca2+ are similar. Thus gradual substitution of Ca2+ by Mn2+ could be expected. Effects observed in these experiments are presented in Figs. 2 and 3. No attempt was made to determine good values of T, or, better, of T, and by correlation to calculate metal-oxygen distances using Solomon and Bloembergen equations,20 as done by Hanna et J. CHEM. SOC. PERKIN TRANS. 2 1995 Table 1 'H Chemical shifts for MA+ and MA, complexes of lasalocid HA and alkaline-earth cations M2+ in methanol at room temperature Proton A-a MgA + MgA, CaA + CaA, SrA + SrA, BaA + BaA , 5 7.03 7.27 7.08 7.09 7.07 7.08 7.05 7.08 7.05 6 6.57 6.79 6.66 6.61 6.6 1 6.6 1 6.60 6.62 6.58 8A 3.36 3.40 3.29 3.31 3.35 4.00 3.62 4.17 3.76 8B 2.94 3.09 3.02 2.99 2.95 2.32 2.71 2.24 2.56 9A 1.73 1.76 1.76 1.76 1.72 1.83 1.78 1.77 I .79 9B 1.55 1.62 1.68 1.61 1.56 1.65 1.60 1.69 1.66 10 1.75 1.76 1.72 1.76 1.72 1.83 1.78 1.87 1.79 11 3.99 4.03 4.01 4.02 4.04 4.64 4.28 4.62 4.35 12 3.04 2.96 3.04 3.13 3.08 3.16 3.08 3.13 3.09 14 2.83 2.87 2.84 2.99 2.95 3.16 2.97 3.09 2.94 15 3.87 3.82 3.87 3.92 3.90 3.83 3.86 3.97 3.96 16 2.14 2.28 2.20 2.26 2.25 2.46 2.30 2.49 2.32 17 A 1.88 2.03 1.89 1.96 1.92 2.07 1.98 2.08 1.98 17 B 1.71 1.52 1.72 1.67 1.72 1.65 1.62 1.69 1.66 19 3.60 3.66 3.61 3.75 3.69 3.87 3.73 3.88 3.76 20 A 1.73 1.76 1.72 1.76 1.72 1.83 1.78 1.84 1.83 20 B 21 A 21 B 1.53 1.a 1SO 1.62{1.87 1.54{1.68 1.61{1.76 1.56 1.72 1.56 1.65{1.83 1.60{1.67 1.77{1.84 1.66{1.79 23 3.82 3.86 3.84 3.92 3.98 4.4 1 4.14 4.26 4.07 24 25 A 25 B 1.22{1.38 1.26{1.42 1.27{1.39 1.31{1.44 I .28{1.43 1.34{1.50 1.30{1.42 1.37{1.53 1.31{1.50 26 27 A 27 B 0.97{1.67 0.96 1.87 1.42 0.98 1.72 1.54 0.94 1.76 1.52 0.93 1.78 1.56 1.07 2.07 1SO 0.98 1.96 1.58 1.09 2.08 I .58 1.03 1.83 1.56 28 0.91 0.93 0.94 0.99 0.97 0.96 0.94 0.97 0.94 29 1.06 1.14 1.08 1.11 1.09 1.16 1.10 1.17 1.12 30 A 1.97 1.95 1.97 1.96 2.00 2.07 2.02 2.08 1.98 30 B 1.57 1.44 1.54 1.61 1.56 1.65 1.60 1.69 1.56 31 0.93 0.93 0.94 0.98 0.97 0.96 0.93 0.94 0.87 32 0.99 1.04 1.oo 1.03 1.02 1.14 1.06 1.13 1.08 33 0.94 1.03 0.97 0.99 0.98 0.99 1.03 0.99 0.97 34 2.19 2.26 2.23 2.20 2.22 2.22 2.22 2.22 2.21 From ref.9. that interactions between the two nuclei M and C are mainly dipolar, which cannot be ascertained. Structure of MA 'complexes It was thus possible from this data to obtain information on the structure of the two successive complexes of the divalent cations with the lasalocid anion in methanol.As suspected from thermodynamic data 3,73 actual structures are significantly dependent on the cation involved. Structural aspects from NMR experiments. The relevant NMR data are clearly illustrated in Fig. 4in which variations of 13C chemical shifts in methanol from the free anion to its MA' complexes are considered. For a given carbon atom this variation is mainly related to interactions of the metal ion with a neighhouring oxygen atom, to conformational changes in its neighbourhood and, to a small extent, to local changes in the interactions with the solvent molecules. It is not very easy to attribute these chemical shift variations to particular causes.Nevertheless, it can be suggested from the strong positive l,,,,l,,,'t,,,,l,,,, variation of C-13 chemical shift that coordination of 0-5 to 200 150 100 50 Ba2+ and Sr2 'is very strong, to Ca2 + weaker and to Mg2' PPm very weak. From shift variations of C- I5 and C-18, involvement Fig. 3 I3C spectra ('H decoupled) in methanol of lasalocid-calcium of 0-6 in the coordination of the cation is clear for Ba2 'and 2 1complex salt CaA, alone (a) and in presence of MnA, (b)(ratio Sr2+, still appreciable for Ca+ and somewhat weak for Mg2+.Nln/Ca =r). (a) r =0, analytical concentration c*(CaA,) = This conclusion could also be reached from Degani and9.4 x 10 'mol dm which results in cCaA2=76.2, cA =cCaA-= Friedman's data; 23 analysis of the circular dichroism spectra 11.9.(b)r =5 x10 3. of MA' at 290 mm, a wavelength for which the contribution of the ketone chromophore is large, showed a perturbation of the and Lallemand et u/.,,* respectively for MnA, and CuA, C=O group increasing with the size of the cation. From the in chloroform solution. This distance determination assumes 2o effects observed on C-22, involvement of 0-8 is suggested for J. CHEM. SOC. PERKIN TRANS. 2 1995 Table 2 13C Chemical shifts for MA+ and MA, complexes of lasalocid HA and alkaline-earth cations Mz' in methanol at room temperature Carbon A-" MgA' MgA, CaA+ CaA, SrA' SrA, BaA' BaA, 1 175.9 177.2 177.2 180.0 178.7 177.1 177.6 177.1 177.0 2 119.3 117.6 117.8 116.8 117.7 117.6 118.2 117.6 118.5 3 161.1 160.7 160.8 161.4 161.3 161.7 161.3 161.7 161.4 4 123.4 124.8 123.8 123.8 123.8 124.0 123.7 124.0 123.8 5 132.6 135.5 133.4 133.6 133.3 133.3 133.0 133.3 132.9 6 121.3 122.5 121.7 121.6 121.6 121.7 121.4 121.7 121.4 7 144.8 144.7 145.1 145.1 145.1 144.6 144.7 144.6 144.7 8 34.1 34.9 34.0 33.8 34.0 33.1 33.7 33.1 33.7 9 38.1 38.1 37.8 37.7 37.9 37.7 37.9 38.1 38.3 10 35.6 35.6 35.4 35.4 35.4 34.3 35.0 34.4 35.0 11 75.5 75.8 75.4 76.3 75.4 72.5 74.2 72.1 74.0 12 49.6 50.3 49.7 50.4 50.2 50.1 50.0 49.9 50.0 13 217.8 218.6 217.7 221.3 219.3 225.5 22 1.6 225.1 222.0 14 57.6 55.8 57.2 56.5 56.9 56.1 56.8 56.3 57.2 15 85.5 85.1 85.7 86.0 86.2 86.2 85.9 87.3 86.3 16 37.8 35.7 37.5 36.9 37.1 35.2 36.5 35.5 36.5 17 41.O 38.9 40.7 39.7 40.2 38.5 39.6 38.1 39.5 18 86.8 88.1 88.0 88.2 87.8 89.3 88.3 90.2 88.8 19 73.5 71.2 73.2 72.4 72.7 71.4 72.2 71.6 72.5 20 22.1 20.6 22.0 21.9 21.9 21.2 21.5 20.9 21.4 21 30.2 30.5 30.3 30.3 30.4 30.0 30.2 30.5 30.4 22 72.1 72.3 72.1 73.0 72.7 73.7 72.9 73.8 73.3 23 77.8 77.4 77.7 78.0 77.6 76.7 77.4 77.7 77.9 24 14.8 14.3 14.8 14.4 14.5 13.6 14.2 13.6 14.2 25 32.2 32.7 32.2 31.9 32.1 31.9 32.0 31.9 32.1 26 6.8 6.7 6.8 6.7 6.8 6.6 6.8 6.6 6.9 27 30.8 31.0 30.7 31.1 30.5 30.5 30.5 30.5 30.6 28 8.8 9.6 8.8 9.0 9.1 9.8 9.3 9.8 9.3 29 16.9 15.8 16.2 16.3 16.6 15.5 16.3 15.5 16.1 30 19.7 17.2 19.5 19.4 19.6 17.3 18.6 17.0 18.2 31 12.7 12.2 12.8 13.0 13.0 12.0 13.2 12.1 12.8 32 14.2 13.4 14.2 13.7 13.9 13.6 13.2 13.6 13.1 33 13.0 12.8 12.9 13.0 13.1 12.4 12.7 12.4 13.4 34 16.2 15.8 16.6 16.0 16.2 16.2 16.1 16.2 16.2 a From ref.9. Table 3 Vicinal proton coupling constants for MA' complexes of lasalocid HA with cations M2 ' Computational Experimental (in methanol) Gaseous state Methanol * MgA' CaA' SrA+ BaA+ A-" I,(AM 1) IM,,(PM~) IMg(BOSS) XM,(BOSS) I,,(BOSS) 8A-9A 2.4 4.4 3 2.3 4.0 2.3 4.4 2.0 2.2 2.3 8A-9B 12.6 11.2 12.0 11.3 11.2 12.3 12.7 11.7 12.8 12.4 8B-9A 10.2 11.7 11.0 12.0 11.2 12.3 12.7 11.7 10.7 12.3 8B-9B 6.0 5.2 4.5 5.O 5.6 5.6 4.4 6.7 6.4 5.3 10-33 7.0 6.0 6.0 6.8 6.5 7.2 7.2 7.2 7.2 7.2 10-1 1 1.2 1.5 1.5 2.2 1.8 3.1 2.0 3.2 1.7 3.7 11-12 10.2 9.7 10.5 10.2 10.0 9.9 10.1 9.9 10.3 10.2 12- 32 7.2 7.2 7.2 7.5 7.0 7.0 7.0 7.0 7.0 7.0 14-30A 10.8 10.4 11.3 11.3 10.0 11.9 12.1 11.9 10.8 12.2 14-30B 2.4 3.9 -3 4.0 2.3 2.5 2.3 2.4 4.6 14-15 1.2 3.3 3 I .5 4.0 1.8 2.6 1.6 1.6 3.4 15-16 10.8 7.4 10.5 10.5 10.0 7.4 7.9 5.7 10.7 10.3 16-17A 7.0 5.2 7.0 6.0 8.0 7.0 9.0 6.0 6.9 6.7 16-17B 4 8.3 9.2 10.5 11.0 10.4 9.3 10.8 9.4 11.5 16-29 6.3 6.4 6.0 6.5 6.5 7.2 7.2 7.2 7.2 7.2 19-20 5.4 7.5 9.0 11.3 10.0 11.2 11.2 10.8 5.4 10.9 19-20B 2.4 3.0 3.5 3 3.2 3.7 3.2 4.8 2.3 4.8 23-24 7.2 7.0 7.2 7.5 7.0 6.4 6.4 6.4 6.4 6.5 25-26 7.0 7.6 7.2 7.5 7.0 7.5 7.5 7.5 7.5 7.5 27-28 7.5 7.2 7.6 7.0 6.8 7.5 7.5 7.5 7.5 7.5 30-3 1 7.5 6.5 7.6 7.8 6.8 7.5 7.5 7.5 7.5 7.5 " From ref.9. * Mean values using a Boltzman distribution. Ba2+,Sr2 and, to a lesser extent for Ca2 +,not clearly for effects across the benzene ring suggest a stronger coordination + +Mg2+.For 0-1 or 0-2, appreciable variations observed on C-4, of Mg2 than other cations. Nevertheless, the strong effects C-5 and C-6, probably related to transmission of electronic observed on C-1 and C-2 for Cat+ are also noteworthy. J. CHEM. SOC. PERKIN TRANS. 2 1995 ll 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 Ill IIIIIIr I I I I I I I 1'1'1 I I I I I 1' 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 Fig.4 Variation in the 13Cchemical shifts of all the carbon atom of lasalocid from its free anion A-to its 1 :1 complexes MAf in methanol. d 13C-n(A in MA+) -6 I3C-n (A in A-) from C-1 to C-34. Appreciable conformational changes appear on the area C-9, C-12 for BaA+ and SrA+, not for CaA' or MgA' and on the C-13, C-15 area for MgA', BaA' and SrA' but weakly for CaA' (as shown by C-14 and C-30 chemical shift variations). Concerning the two heterocycles, conformational changes are suggested for the four cations though they must be weaker for CaA+, maximal variations being observed for the furan ring (on C- 16, C- 17) in BaA and for the pyran ring (on C- 19 and C-20) + in MgA '.This last observation is supported by data on 'H-'H coupling constants.On the whole, from the I3C chemical shift, the smallest conformational changes from A- to MA+ would seem to be observed for CaA+. 'H-'H coupling constants' vdues are in close agreement with this for the aliphatic part of the molecule from C-8 to C-15 but not for the rings for which a marked deformation of both the furan and the pyran rings are observed in CaA '. Examination of 19-H, 20-H coupling ccnstants suggest an inversion of the pyran ring for MgA', the sti-ucture observed in A-only being found in BaA'. These two constants show intermediate values for CaA' and SrA+ which suggest either twist forms or resonance between the two chair forms. NMR experiments involving Mn2 yield information on the + coordination sites of the cation in MnA 'the structure of which is expected to be not very different from that of CaA' in methanol. Adding MnCl, to a solution of type 1 in CD30D +containing mainly CaA (concentration ratio cG/cX = 0.04) resulted, as shown in Fig.2, in the disappearance of C-I, C-2, C-7 and C-22 signals and broadening of C-3 and C- 13 signals, which suggests a closed form of the molecule through main coordination of Mn2 'in MnA' by 0-1or 0-2,O-5 and 0-8. Computational studies. Modelling the structure of MA ' complexes was first attempted in a vacuum using quantum semi-empirical methods. Firstly, a doubly charged 'sparkle' of radius 0.7 A was introduced in some of the geometries previously derived for the lasalocid anion, its initial location being somewhat subjective.Structure optimization was carried out using either AM1 or PM3," two versions of a quantum semi-empirical method derived from MNDO. According to the starting situation and the program used various geometries were obtained. Their formation energies, of the order of 645 kcal mol-', lie, taking into account a systematic difference related to parametrization differences between PM3 and AMI, in a 2 kcal mol-' margin. However their structures differ appreciably. Nevertheless, some general features con- cerning the coordination of the cation modelled by this sparkle are observed: 0-1 and 0-2 are both coordinating sites, standing at a distance of the order of 3 A from the cation.Coordination to 0-5 is significant, distances from the sparkle to this oxygen being with high regularity of the order of 3.5 A. The terminal oxygen 0-8 is frequently at a distance of 4 8, from the cation, though in some cases the O(8)H O(1) hydrogen bond is retained. Distances to the sparkle from 0-4, 0-6 and 0-7 range from 4.1 to 4.8 A, those of the oxygens nearest to the cation being variable. As an example, results for a structure I, obtained using anion geometry I," and AM1 program optimization are reported in Tables 3 and 4. Such a structure corresponds closely to coordination features previously described for MnA'; the cation is mainly bound to 0-1 and 0-2, 0-5 and 0-8. Corresponding 'H-'H coupling constants were calculated using the Durette and Horton formula24 as previously.Substituting the Mg2 + cation (which is parametrized in PM3) for the 2' sparkle affords very spectacular results. Starting from closed geometry I, here described in Tables 3 and 4, after optimization using PM3, gave I,,-also described in these Tables and presented in Fig. 5. A very strong coordination of Mg2+ by the carboxylate, which results in a marked shortening of the M 0-1 and M I .85 A) is observed. This corresponds to a strong opening of the structure, the nearest potentially coordinating oxygens being 0-5 and then 0-4 and 0-8 but at distances greater than 4.5 A. Calculation of the 'H-'H coupling constants, as shown in Table 3, gave values consistent with the experimental ones for MgA' except for the rings.Other computed MgA+ conform- ations also present the same general trends: strong coordination of the magnesium cation to the carboxylate and breaking of head-tail hydrogen bonds of the ligand. The above computations are for the gaseous state. To compare experimental data Monte-Carlo simulations were carried out in methanol using the program BOSS as described in the experimental part. Given interpretations proposed for thermodynamic data and the present results, it was considered important to make the size of the cation variable. All alkaline- earth cations were parametrized, according to Aqwist ,'in the program used but, to save computer time, investigations were restricted here to the smallest and largest of these cations, the structural features observed thus being assumed to be representative of two extreme situations.The initial geometry of BaA' was derived from X-ray data for a BaA2-2H,O crystal.26 The second anion ligand and the J. CHEM. SOC. PERKIN TRANS. 2 1995 +Table 4 Parameters for various structures of MA+ species obtained using 2+ sparkle, MgZ or Ca2 + cation and resulting, in the gaseous state from quantum semi-empirical calculations (AM 1 or PM3) and in methanol from Monte-Carlo simulations (BOSS): metal cation-lasalocid oxygen and C( ltC(22) distances, hydrogen bond length (all distances in A), conformation of the rings, steric angle 0 between benzene and carboxylate planes, mean number of methanol molecules in the first and second solvation shells of cation n, and n, with corresponding oxygen-cation mean distances d, and d,.Iz(AM1) I,,(PM3) M * 0-1 3.3 1.8 M * * 0-2 2.8 1.8 M * * 0-4 4.8 5.1 M * -* 0-5 3.5 4.6 M * * 0-6 4.3 6.0 M 0-7 4.3 5.8 M * * 0-8 3.5 5.3 08-H * 01 2.3 6.1 08-H * -02 4.2 6.4 03-H 01 2.0 1.8 C( 1) * * C(22) 5.4 7.7 THF a Twist THP Chair Chair 6 (deg) -39 -4 n1, n2 dl7 d2 (4 a Between twist and envelope. water molecules were withdrawn and the protons were located optimally; meanwhile care was taken to maintain the O(8) -H 0(1) hydrogen bond. A chloride ion was also added to the cell, but at a sizeable distance to ensure system electroneutrality and also to avoid interactions between this ion and the ionophore-barium complex.Computations for the BaA+ complex gave a globular geometry I,, depicted in Fig. 6, in which the cation is coordinated to oxygens of the ligand. Corresponding relevant parameters are given in Tables 3 and 4. In this structure, the anion ligand is closed through hydrogen bonding 0-8 to both 0-1 and 0-2 roughly equivalently. It thus wraps the cation, all the oxygens of the crown being involved to some extent in its coordination (Fig. 5) in agreement with experimental features for the '3C chemical shifts. Calculated 'H-'H coupling constants agree acceptably with experimental ones. For the magnesium complex, various computations were also carried out using the BOSS program.A Mg2+ cation was substituted for the sparkle in the previous form I, (AM 1). After optimization in a continuum the resulting geometry was then placed in the methanol molecule box. Equilibration resulted in geometry TM,(BOSS) depicted in Fig. 7. Significant Mg 0 mean distances are also reported in Table 4.Comparison of the H-'H mean coupling constants to the accessible experimental ones revealed some structural inadequacies for both this geometry and geometry IM,,(PM~). Discrepancies mainly concern the THF ring C(15)-C(16) bond and the THP ring C(19)-C(20) bond. Observed coupling constants J,9-20 clearly result from an inversion, compared to other lasalocid species, of the THP ring; this inversion for example does not occur, as shown in Table 3, for the BaA' complex.Accordingly, starting from the 12(AM1) geometry, an inversion of the THP cycle was carried out and dihedral angles were thus fitted to values calculated from known 'H-'H coupling constants. This was done using an appropriate procedure in the SYBYL package.29 The resulting geometry was thus directly placed in methanol. Equilibration resulted in geometry XM,(BOSS) the parameters of which are also presented in Tables 3 and 4. As expected, an improved agreement was found between experimental and calculated H-'H coupling constants for MgA+ in methanol. Mg2+ strongly bound to the carboxylate group. In comparison coordination to the other oxygens of the ligand is rather weak. A small opening of the structure is observed.These features previously shown by calculations IM,(BOSS) XM,(BOSS) I,,(BOSS) 3.3 k 0.2 1.9 k 0.1 5.7 2 0.6 1.8 k 0.1 1.9 _+ 0.1 3.9 amp; 0.5 4.2 k 0.2 4.8 k 0.1 4.1 k 0.5 4.5 k 0.3 6.3 2 0.2 3.3 k 0.9 5.7 _+ 0.3 7.7 _+ 0.3 3.8 2 0.7 4.5 k 0.2 7.3 +_ 0.3 3.5 k 0.3 3.0 +_ 0.2 4.4 2 0.2 3.4 amp; 0.7 3.1 5.2 2.4 3.5 3.9 2.5 2.I 2.0 2.4 5.9 6.5 4.8 Twist Twist Twist Chair Chair Chair -39 -24 -19 3.0, 3.0 4.0, 3.1 5.0,4.2 1.93,4.2 2.08,4.0 2.75,4.8 Fig. 5 Geometry I,,(PM3) for the MgA+ complex in a vacuum using semi-empirical quantum calculation with program PM3 as described. 0carbon, 0hydrogen @oxygen metal cation. using PM3 are thus confirmed by Monte-Carlo simulations in methanol. Concerning the method, it must be stressed that conform- ation I,, and I,, were first optimized, using the AMBER/OPLS J.CHEM. SOC. PERKIN TRANS. 2 1995 A Fig. 6 Geometry I,,(BOSS) for the BaA+ complex in methanol using Monte-Carlo simulation in methanol. Top: a view showing the coordination of Ba2+ by the oxygens of the ligand. Bottom: a view showing the coordination of Ba2+ to both the lasalocid anion and the five nearest methanol molecules. force-field in a dielectric continuum (E = 32.66). These conformations are thus expected to correspond at least to lacal enthalpy minima. Monte-Carlo computations using the 'preferential sampling option' mainly act on the first solvation shell solvent molecule distribution and on the cation location, little on the solute, the initial optimized geometry of which is barely modified.On 3 x lo6 configurations, enthalpies and their standard deviations for systems (MA' + C1-+ 378 MeOH) were found to be respectively -3738 L-4 kcal mol-' for conformation IBa, -3881 k 4 for conformation I,, and -3845 k 3 for conformation X,,. The low values of the standard deviations show that the configurations retained during the sampling are narrowly scattered, which suggests that thzse systems can be considered as equilibrated. Calculations of the enthalpies of association of MA' in methanol from these data and analogous data for separate M2 + and A- ions yield to unrealistic values, compared with the experimental ones; not surprising considering Jorgensen's own statement ' that searching for enthalpy differences of less than 10 kcal mol-' is currently impractical.Differences observed between enthalpies of systems I,, and X,, have also to be noted. The X,, conformation is more compatible with structural data but its eni.halpy appears in this computation as less favourable. Would its entropy be more favourable? Monte-Carlo simulations using BOSS also showed that the desolvation of the alkali-metal cations is only partial in the MA' complexes. Fig. 8 shows the radial distribution function between the cation involved and the methanol molecule oxygen atcms. They are compared to analogous radial distribution fur: ctions, obtained in comparable conditions, for the cation alone.By integration of the first two peaks of these radial disx-ibution functions, mean numbers of near neighbouring Fig. 7 Skeleton schemes of geometry IMg (top) and X,, (bottom) of MgA' in methanol resulting both from Monte-Carlo simulations using BOSS methanol molecules n, in the first and n2in the second solvation shells of the cation were calculated. They are also reported in Table 4 along with corresponding cation-methanol oxygen mean distances, respectively d, and d,. Corresponding com- puted values for free cations in methanol are n, = 8.3, d, = 2.8, n2 = 8.2, d, = 5.0 for Ba2' and n, = 6.0, d, = 2.0, n, = 6.1, d, = 4.1 for Mg2+. Concerning this last cation it is interesting to observe the close agreement between these values and those resulting from X-ray diffraction studies and molecular dynamic simulation of a 0.6 mol dm-3 solution of MgCI, in methanol by Tamura et a/.28The height of the peak corresponding to the second solvation shell differs appreciably in the two simulations; this could be due to differences in solvent model, expression of interaction energies and concentrations; however, integration of the two peaks results in analogous values of methanol molecule number in this second shell: between 6 and 7.Concerning MgA+ and BaA' complexes it can be observed in Fig. 8 that the radial distribution function tends to I as expected when r increases, consistent with good equilibration of the system. Concluding on the MA' structure in methanol.On the basis of the comparison of standard thermodynamic functions associated with MA complex formation for lasalocid and + salicylic acid in methanol, it was previously suggested that3 going from magnesium to barium, the cation shifted from a location opposite the carboxylate group to one in the centre of the pseudo-crown. Also, examination of the electronic absorption spectrum of the salicylate chromophore revealed that the salicylate moiety in lasalocid MA' complexes became less involved as the size of the cation increased. Both are 2.5 5.0 7.5 10.0 0 2.5 5.0 7.5 10.0 0 2.5 5.0 7.5 10.0 2.5 5.0 7.5 10.0 0 2.5 5.0 7.5 10.0 M.... Omethanol distance J 8, Fig. 8 Radial distribution functions g(r) as a function of the metal cation-methanol oxygen (M2'-O) distance r in A.(a) Ba2+ alone in methanol. (b)Ba2+ in the BaA' complex in methanol (geometry I,, BOSS). (c) Mgz+ alone in methanol. (d)Mg2+ in the MgA+ complex in methanol (geometry IMgBOSS). (e) As (d)(geometry XMgBOSS). supported here. From both NMR data and computational modelling, Ba2 + is coordinated to 0-5, 0-6, 0-7 and 0-8 and weakly to 0-2. Conversely, Mg2+ is strongly bound to the carboxylate and weakly to 0-8 and 0-4. Coordinating systems of other cations are intermediate as clearly shown by the I3C chemical shift patterns in Fig. 4. The structure of SrA+ is very near that of BaA'. Coordination to the carboxylate clearly occurs with Ca2 + concurrently with a weakening of the cation coordination to 0-5, 0-6 and 0-8.Experiments with the para- magnetic cation Mn2 + show main coordination to the carboxy- late, 0-5 and 0-8. Thus the wide variability of the coordination of the cation by lasalocid anion is confirmed. This variation is clearly a function of the size of the cation which determines continuous changes in both conformation of the ligand and involvement of the coordination sites. One of the aspects of this concerns the conformation of the pyran cycle. Continuous variations with the size of the cation from BaA' to MgA+ result in a complete inversion of the cycle from BaA+ to MgA'. Owing to its better fit to the experimental 'H-'H coupling constants, geometry X,, should correspond at best to the actual structure of MgAf in methanol.However, the modelling does not reveal why, whether because of steric hindrance or interactions, such an inversion of the pyran ring is favoured. Results in Table 4 and Fig. 6 stress the importance of the residual solvation of the cation in the MA+ complexes. Some of the molecules of the solvent, by completing the coordination shell of the cation, contribute to the stability of these complexes. In the case of MgA+ the strong remaining solvation of the cation, probably also contributes to appreciable opening of the structure. Structure of 2:l complexes MA,. Owing to budgetary constraints on both programs and computers, computational simulations involving two ligating molecules such as in MA, J. CHEM. SOC. PERKIN TRANS. 2 1995 complexes could not be carried out.Structural information concerning these species are thus derived only from NMR experiments. At ambient temperature 13C and 'H spectra of MA, complex salts are well resolved for all species studied, which denotes either a symmetric role of the two A- ligands or their rapid exchange. Complementary experiments were carried out at low temperature (240 to 220 K) on calcium type 2 solutions, in which mainly CaA, was present. With respect to reference signals, such as those of the methyl carbons, an appreciable broadening of signals corresponding to C- 11, C- 13 and C- 10, C- 17 and C- 18 and also C-20 was observed. This can be taken as a sign of the dissymetric role played by the two A-ligands in CaA,, their exchange being slower at low temperature.Moreover, these results suggest analogous co- ordination of the cation by the carboxylate and 0-8 and possibly the 0-6 and 0-7 of the two ligands and specific involvement of 0-4 of one ligand and 0-5 of the other one. Analogous conclusions can be drawn from experiments with Mn2+. Adding small amounts of either MnCl, or MnA, to a solution of type 2 containing mainly CaA, had identical effects: C-1, C-2, C-3, C-7; C-8, (2-9, C-10; C-11; C-13; C-22 signals disappeared or were markedly broadened, which again suggests the involvement of 0-1 or 0-2,O-4,0-5 and 0-8 borne by one or the other of the two anions ligands. 0-4, which is not involved in coordination in the MnA' complex, thus appears to be in MnA,.C-15, C-18 and C-19, C-23 signals are not appreciably affected, which suggests that 0-6 and 0-7 are not strongly involved in the coordination of Mn2 + in MnA,. The two anion ligands thus have different conformations. At ambient temperature what are observed in spectra are mean parameters for the two anion ligands involved in the complexation of the cation. Structural information on these two ligating anions can nevertheless be obtained assuming that adding the second molecule does not perturb the first one. In these conditions, assuming for example that '3C chemical shifts ofthe first molecule (A'-)' are those of A- in MAf, the '3C chemical shifts can be calculated for the second molecule (A-)". If this is done, it appears that except in the case of Mg2+, (A-)" exhibits only very small differences from free anion A- in methanol. Some perturbations nevertheless occurring in C- 1, C-2, C-31 and C-32 suggest that the main involvement is that of the oxygens of the carboxylate. However, examination of data resulting from experiments involving Mn2 + and experiments concerning CaA, at low temperature suggest for these two cations an involvement of the 0-4 of the second molecule in their coordination.In addition, it must be mentioned that contrary to what was observed in chloroform' for SrA, and BaA, no significant shift of H-28, expected to result from the cycle current of the benzene ring of the other molecule, occurs here. The structure of the MA, species in methanol must then be a more open one than in chloroform.The second molecule (A")- acts as a rather mobile lip of a bowl formed by the first molecule (A')- and hosting the cation. References 1 Part 12, R. Lyazghi, Y. Pointud, G. Dauphin and J. Juillard, J. Chem. SOC.,Perkin Trans. 2, 1993, 1681. 2 J. Juillard, C. Tissier and G. Jeminet, J. Chem. Soc., Faraday Trans. 1, 1988,84,951. 3 Y. Pointud, E. Passelaigue and J. Juillard, J. Chem. SOC.,Faraday Trans. I, 1988,84, 1713. 4 J. Woznicka, C. Lhermet, N. Morel-Desrosiers, J.-P. Morel and J. Juillard, J. Chem. Soc., Faraday Trans. 1, 1989,85, 1709. 5 P. Laubry, C. Tissier, G. Mousset and J. Juillard, J. Chem. SOC., Faraday Trans. I, 1988,84,969. 6 P. Laubry, G. Mousset, P. Martinet, M. Tissier, C.Tissier and J. Juillard, J. Chem. SOC.,Faraday Trans. I, 1988,84, 3175. 7 Y. Pointud and J. Juillard, J. Chem. Soc., Faraday Trans. 1,1990,86, 3395. J. CHEM. SOC. PERKIN TRANS. z 1995 8 M. Mimouni, Y. Pointud and J. Juillard, Bull. SOC.Chim. Fr., 1994, 131, 58. 9 R. Lyazghi, A. Cuer, G. Dauphin and J. Juillard, J. Chem. SOC., Perkin Trans. 2, 1992, 35. 10 P. Malfreyt, Y. Pascal and J. Juillard, J.Chem. SOC., Perkin Trans.2, 1994,2031. 11 Program MOPAC v. 6.0, QCPE no 455 (Dept. Chem., Indiana University), according to refs. 12 and 13. 12 M. J. S. Dewar, E. G. Zoebisch, E. F. Healy and J. J. P. Stewart, J. Am. Chem. SOC., 1985,107,3902. 13 J. J. P. Stewart, J. Comput. Chem., 1989, 10, 209. 14 W. L. Jorgensen, BOSS v. 3.4, Yale University, New Haven, CT, 1991. 15 W. L. Jorgensen, J. Phys. Chem., 1986,90, 1276. 16 N. Metropolis, A. W. Rosenbluth, M. N. Rosenbluth, A. Teller and E. J. Teller, J. Chem. Phys., 1953, 21, 1087. 17 W. L. Jorgensen, J. Phys. Chem., 1983,87,5304. 18 W. L. Jorgensen and J. Tirado-Rives, J. Am. Chem. Soc., 1988,110, 1657. 19 S. J. Weiner, P. A. Kollman, D. A. Case, V. C. Singh, C. Ghio, G. Alagena, S. Profeta and P. Wernier, J.Am. Chem. SOC., 1984,106, 765. 20 I. Solomon and W. Bloembergen, J. Chem. Phys., 1956,25,261. 21 D. A. Hanna, C. Yeh, J. Shaw and C. W. Everett Jr., Biochemistry, 1983,22,5616. 22 J. Y. Lallemand, R. Rao and T. Prange, Noun J. Chim., 1980, 4, 315. 23 H. Degani and H. L. Friedman, Biochemistry, 1974, 13, 5022. 24 P. L. Durette and D. Horton, Org. Magn. Reson., 1971,3,417. 25 J. Aqwist, J. Phys. Chem., 1990,94, 8021. 26 I. H. Suh, K. Aoli and H. Yamazaki, Acta Crystallogr., Sect. C, 1989, 45,415. 27 W. J. Jorgensen, B. Bigot and J. Chandrasekha, J. Am. Chem. SOC., 1982,104,4584. 28 Y. Tamura, E. Spohr and K. Heinziger, Ber. Bunsenges. Phys. Chem., 1992,96, 147. 29 Tripos Associates, 1699 S. Hanley Road, suite 303, St. Louis, MO, 63144, Sybyl 6.03, 1993. Paper 4/07870E Received 29th December 1994 Accepted 7th June 1995 0Copyright 1995 by the Royal Society of Chemistry