J. CHEM. SOC. FARADAY TRANS., 1994, 90(5), 727-732 Enthalpies of Interaction between Dimethyldioctadecylammonium Bromide Vesicles in Aqueous Solution and either Dipicolinate or Sulfate Anions Michael J. Blandamer, Barbara Briggs, Michael D. Butt, Paul )rA. Cullis and Matthew Waters Department of Physical Chemistry, University of Leicester, Leicester, UK LEI 7RH Jan B. F. N. Engberts Department of Organic & Molecular Inorganic Chemistry, University of Groningen , Nijenborgh 4, 9747 AG Groningen, The Netherlands Dick Hoekstra Department of Physiological Chemistry, University of Groningen , Bloemsingel 10,9712 KZ Groningen, The Netherlands Injection of small aliquots of dipicolinate anions (sodium salt) into an aqueous solution containing dimethyl- dioctadecylammonium bromide (DOAB) vesicles is endothermic at 50 “C, becoming first more and then less endothermic.The injection process is effectively athermal for solutions containing more than equimolar amounts of DOAB and dipicolinate anions. A similar pattern is observed when small aliquots of sodium sulfate(aq) are injected into DOAB(aq). The overall patterns of enthalpy changes are attributed to the vesick dianion interaction which is exothermic and head-group dehydration with bromide ion displacement which is endothermic. Nevertheless, a complexity emerges if the solutions include a buffer which turns out to play a less than passive role. This conclusion is supported by differential scanning microcalorimetry for DOAB(aq) in the presence and absence of HEPES buffer.In aqueous solution, dimethyldioctadecylammonium bromide (DOAB) forms whose structures resemble closed bilayer aggregates formed from phospholipids. Fendler and co-workers’ estimated that the hydrodynamic radius of each vesicle is 80 nm. Cuccovia et aL6 estimated a radius of 65 nm. Vesicles can fuse to form larger vesicles. Fusion is an extremely important physiological process being a key step in, for example, cell fertilisation. Fusion is often initiated by adding a fusogenic agent. For e~ample,~ Ca’+ ions induce fusion between di-n-didodecylphosphate vesicles and phos- phatidylserine liposomes. Fendler and co-workers’ report that SO:-ions induce the fusion of DOAB vesicles. A mechanism was advanced’ in which fusion of two vesicles involves dehydration of head groups and the formation of a trans complex through an SO:-ion which interacts with the head groups on the surfaces of the vesicles.Following estab- lishment of this link, the bilayers coalesce in the area of contact, the two inner aqueous compartments of the vesicles join together and the ellipsoid-shaped assembly reforms to a larger vesicle. In view of the importance of the initial vesicle-SOi-interaction, we decided to study this process using a titration calorimeter with the aim of measuring the associated enthalpy changes. The expectation was that the titration calorimeter would record a series of exothermic pulses when SO:-(aq) aliquots were injected into DOAB(aq). In the event, the recorded trace showed a series of endo- thermic peaks for the first few injections followed by a series of exothermic peaks.However, when the data were corrected for the dilution of Na,SO,(aq) the trace showed a series of endothermic peaks, the intensity of the peaks decreasing with the injection of each new aliquot of SOz-(aq). The dianion of dipicolinic acid7 (DPA, pyridine-2,6-dicar- boxylic acid) was used as a fusogenic agent for didodecyl- dimethylammonium bromide (DDAB) vesicles. This fusogenic agent is more hydrophobic than SO:-anions. In terms of the mechanism proposed by Fendler’ for vesicle fusion, the DPA dianions have structures which would allow these ions to form a bridge between vesicles as the first step in fusion.Again the expectation was, therefore, that addition of DPA to DOAB(aq) would be exothermic. We report that this is not the case. However, these experiments revealed two further complications. The first was not unexpected in the light of our experience in using a differential scanning micro- calorimeter to record the gel-to-liquid transitions.’ There we found that our calorimetric methods could detect variations in vesicle properties associated with various methods of prep- aration. However, adopting standard protocols for vesicle preparation led to entirely reproducible results. The same experience emerged with respect to titration calorimetry involving vesicle solutions. The second complication emerged from the use of buffered solutions in experiments using DPA.Thus both DPA(aq) and DOAB(aq) were buffered at pH 6. The aim was to remove any complications in the recorded heat from dilution of a weak acid. Therefore, in order to facilitate direct comparison we used similar buffered solutions in the experiments using SO:-(aq) as the fusogenic agent. However, we then observed that the results for these experi- ments changed when the buffer was omitted. Nevertheless, addition of SO:-(aq) to DOAB(aq) remained surprisingly endothermic. We discuss the reasons why this pattern is observed rather than the intuitively anticipated exothermic trend. Experimental Materials Following experiments using a number of protocols for the preparation of solutions, satisfactory results were obtained when aqueous solutions containing DOAB vesicles were pre- pared by either of two methods.In the ‘hot-water’ solid DOAB was dissolved in hot water and held at 55°C for 30 min. In the ‘sonicated’ method, DOAB was dissolved in hot water and sonicated for 30 min. The solution was heated to 70°C and held at this temperature for 5 min. The solution was allowed to stand for 1 h before being used to fill the sample cell in the titration calorimeter. In two series of experiments the solutions were prepared using HEPES buffer (Sigma). Aqueous solutions prepared with HEPES buffer at pH 6 contain in associated and disso- ciated forms, H+(aq), Na+(aq) total = 5 x mol dm-3, CH3CO; (as) total = 5 x mol dm-3, and HEPES [N'-(2-hydroxyethyl)piperazine-N-ethanesulfonate, XSO; , total = 5 x mol dmW3].In other words, these constitu- ents are for the most part amphipathic. These concentrations exceed those of DOAB monomers and of the added anions in the titration experiments. Consequently, the vesicular systems are saturated with the HEPES-buffered systems. We return to this point below. Indeed, for reasons which will become clear below we identify the HEPES anion by the symbol XSO,. Calorimetry An Omega titration micr~calorimeter~ (MicroCal Inc., USA) recorded the heat, q, associated with the injection of a dilute solution of a fusogenic agent into DOAB(aq). The sample cell (reservoir) contained 1.4115 cm3 of the DOAB solution at 323 K. A syringe driven by a stepping motor introduced ali- quots of fusogenic solution at predetermined time intervals.In each experiment a sequence of 25 injections was recorded with a time delay of 3 min between each injection. This time delay was sufficient for the calorimeter to bring sample and reference (containing water) cells to the same temperature. If the injection of one aliquot of solution from the syringe was exothermic, the recorded trace showed the rate of heating of the reference cell (recorded in cal s-l) during the time inter- val required for sample and reference cells to attain the same temperature. If the injection was endothermic, the trace showed the rate of heating of the sample cell during the cor- responding time interval. In other words, the raw data com- prised a plot which has a number of extrema showing rates of heating following the injections as a function of time.The experiment was repeated except that the sample cell con- tained only water (or the buffer solution) but no DOAB. The pulsed sequence of injections was repeated and hence the trace recorded the contribution to the injection pattern described above which could be attributed to dilution of the fusogenic agent. The latter trace was subtracted from the first trace to yield the information required in the next stage of the analysis. Using the Omega software (MicroCal Inc.), the pulses were integrated to obtain a plot of heat, q, as a func- tion of injection number. N. A differential scanning microcalorimeter (MicroCal, Inc., USA) was used to record the dependence of relative isobaric heat capacities on temperature for aqueous solutions contain- ing DOAB(aq) in solutions prepared both in the presence and absence of buffer. In each case, the DOAB solutions were placed in the calorimeter and the temperature scanned from 15 to 90 "C.Each solution was cooled to 15 "C and held at that temperature for at least 3 h. A new scan was then recorded over the range 15-90 "C. Results The calorimetric investigation was based on a temperature of 323 K for two reasons. First, with respect to using SO:-(aq) as the fusogenic agent, we found that the results recorded for solutions at 298 K were irreproducible, a consequence of pre- cipitation in the reservoir following injection of SO:-(aq).This conclusion followed visual inspection of the recovered solutions. Secondly this temperature (323 K) is above the temperature' at which the hydrocarbon chains in the DOAB vesicles undergo a gel-liquid-crystal transition, 318 K. In fact, vesicle fusion requires that the hydrocarbon chains are in the liquid-crystalline form. A typical trace is shown in Fig. 1 for a J. CHEM. SOC. FARADAY TRANS., 1994, VOI,. 90 I I -0 50 tirne/rnin Fig. 1 Calorimetric titration of DPA(aq) (20 x mol dm--3) in 5.64 x dm3 aliquots into 1.4115 cm3 of DOAB (aq) (pH 6; 1 x mol dm-3) at 50°C calorimetric titration in which DPA(aq) was injected into DOAB(aq). The reservoir contained DOAB (1 x lov3 mol dm-3) in an aqueous solution at 50°C having pH 6 (recorded at 25 "C) and prepared using the sonication method.The recorded pulses change from endo- to exo-thermic after eight injections. Between each pulse a small length of 'baseline' was recorded showing that the processes responsible for each pulse are effectively complete within 1 min. In other words, no information concerning kinetics of physical/chemical processes was forthcoming. In considering traces of the form shown in Fig. 1, it is often informative to re-express the quantities in molar terms. So, for example, in this experiment there are 1.42 x mol of DOAB in the sample cell. Each aliquot of fusogenic agent adds 0.113 x rnol of this agent into the reservoir. After eight injections (cf. the crossover in Fig. l), 0.904 x mol have been added.DPA and DOAB are roughly equimolar around the 13th injection. At the 25th injection there are 2.72 x mol of DPA in the sample cell. The results illustrated typically by the trace in Fig. 1 were obtained following several preliminary experiments. If the concentration of solution in the syringe was too high, the reaction was over following the first injection. If this concen- tration was too low, the extrema were small, being lost in the noise. Comparison between different titration plots must take into account the compositions of the solutions involved. However, the recorded trace in Fig. 1 includes a contribution from the dilution of the DPA solution into the buffer solu- tion. As noted above, this latter contribution is obtained by repeating the experiment with no DOAB present in the reservoir.In this case, the pulses showed an exothermic dilu- tion. The recorded trace was subtracted from the trace in Fig. 1. The calculated pulses were integrated (OMEGA software, MicroCal) to yield the calorimetric titration curve shown in Fig. 2. At low injection numbers the enthalpy of reaction is endo- thermic becoming more and then less endothermic with increase in injection number. After 25 injections the amounts of DPA and of DOAB in the reservoir are 2.82 x mol and 1.412 x mol respectively, a ratio of ca. 2 :1. The titration curves approach the athermal conditions when slightly more DPA than DOAB is present in the reservoir. The actual molar ratio was difficult to identify precisely but seems in the region of 1.6 :1 DPA :DOAB.The reason for the uncertainty can be traced to the small enthalpies. For a solution more concentrated in DOAB prepared using the sonication method, the endothermic extremum was more J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1.E c I 3 n-z 1.0 W-m 0.6 3 i,d x 0 1 I I I I I 0 10 20 injection number Fig. 2 Calorimetric titration showing dependence of enthalpy of reaction expressed per mol of DPA as a function of injection number; system described in Fig. 1; plot corrected for dilution of DPA intense, Fig. 3. The endothermic maximum was ca. 2.5 kcal (mol DPA)-l. With decrease in concentration of DOAB, the intensity of the endothermic maximum decreased although the scatter on the plots increased, Fig.4. Here the first injec- tion was marginally exothermic. Using freshly prepared solu- tions the patterns shown in Fig. 1-4 were reproducible, although it was difficult to ascertain whether the first injec- tion in the system described in Fig. 4 was marginally endo- or exo-thermic. A similar pattern emerged when SO:-(as sodium salt) was injected into the buffered DOAB solution prepared using the sonication method. The titration plot for a typical experiment is given in Fig. 5 which records the endothermic injection of 5 x lov6dm3 aliquots of SO:-(aq) (22.6 x mol dm-3) into DOAB (1 x mol dm-3). Again it is useful to re- express these details in molar quantities.The sample cell con- tains 1.412 x moles DOAB. Each aliquot of injected solution contained 0.113 x mol SO:-ions. In Fig. 5, the change from endo- to exo-thermic occurs at the sixth injection when 0.678 x mol of SO:-have been injected. At the 25th injection the molar ratio SO:-: DOAB was 2 :1. The corresponding integrated plot, corrected for the dilution of SOi-(aq), shows that it is not possible to define precisely t 0 10 20 injection number Fig. 3 Calorimetric titration of DPA(aq) (20 x mol dmP3) in 10 x dm’ aliquots into 1.4115 cm3 of DOAB(aq) (pH 6; 2 x mol drn-j) at 50°C; plot corrected for dilution of DPA 729 0.3C c Iz o.20 c)-E W0.10 m +. $0 -0.10 inject ion number Fig.4 Calorimetric titration of DPA(aq) (9 x mol dm-3) into DOAB(aq) (pH 6; 0.4 x rnol dm-’) at 50°C;plot corrected for dilution of DPA(aq) the molar ratio at the point where the injection becomes athermal, Fig. 6. Nevertheless, the endothermicity is clear-cut, being more dramatic than in the case of the DPA injections. Consequently, the scatter on the integrated titration curves was less. An interesting indication of the reproducibility is given in Fig. 7. This plot shows the dependence of the enth- alpy of reaction as a function of injection number for three independent titrations using freshly prepared solutions. This plot yields an averaged quantity by noting that at the 25th injection the summed enthalpy of reaction is ca. 22 kcal (mol SO:-)-’.Hence, on average, each injection was endothermic to the extent of 0.8 kcal (mol SO:-)-’.The results summarised in Fig. 5-7 refer to titration calori- metric experiments in which both sulfate and DOAB solu- tions were buffered. We wanted to draw direct comparison with the experiments where DPA(aq) was injected into DOAB(aq), both solutions being buffered to pH 6. However, when the buffer was not used, the intensity of the endo- thermic extrema for SO:-titrations increased dramatically. In Fig. 8 we show the titration plot for a typical run, the DOAB(aq) being prepared using the sonication method. Again the switch from endothermic to exothermic injection is apparent but, as before, the exothermicity is a consequence of dilution of SO:-(aq).A clear indication of the effect of the 15 -10 v)-m 05 z \ P 0 -5 I I I 0 50 100 ti me/m i n Fig. 5 Titration of Na,SO,(aq) (22.6 x lo-’ mol dm-3) aliquots of 5.0 x dm-3 into a solution (1.4115 an3) of DOAB(aq) (1 x lo-’ mol dm-3) at 323 K; dependence of rate of heating follow- ing each injection as a function of time; buffered solution J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 01 LI I I I I I 0 10 20 injection number Fig. 6 Integrated titration curve for the titration of SO:-into DOAB(aq); for details see caption to Fig. 5 buffer is afforded by comparison of the integrated plots in Fig. 7 and 9, both runs reporting the enthalpy changes for solutions containing DOAB (1 x mol dm-3).In both cases the injections were continued until the molar ratio SO:-: DOAB was 2 :1. There is a slight difference between the titration calorimet- ric results for DOAB solutions prepared using sonication and hot-water methods. We attribute this difference to differences 7 20 n IClt z-0E v-m 5 10 i5w I I I I I 1 I 0 10 20 injection number Fig. 7 Summed integrated enthalpies of reaction for three indepen- dent titrations; for details see caption to Fig. 5 I I I 16.67 41.67 time/s Fig. 8 Titration of Na,SO,(aq) (22.6 x mol dm-j) aliquots of 10 x dm3 into DOAB (2 x mol dm-3); unbuffered solu- tions I I I I I I 0 10 20 injection number Fig. 9 Integrated titration curves for the titration of SO:-(aq) into DOAB(aq); for details see caption to Fig. 8.The results for dilution of SO:-(aq) have been removed ;unbuffered solutions. in size and distribution of the vesicles prepared by the two methods described in the Experimental. The results in Fig. 10 show that, on average, each injection is endothermic to the extent of 1.8 kcal (mol SO:-)-', roughly twice that observed in the presence of buffer. The A. I1 I I I I 1 0 10 20 injection number -I I t. -P A. I I I 1 I I 0 10 20 injection number Fig. 10 Summed integrated enthalpies of reaction for independent determinations using SO:-(aq) and DOAB(aq) in unbuffered solu- tions; DOAB(aq) prepared using (a) sonication and (b) hot-water methods J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 I\ Inl 20 40 60 TrC Fig. 11 Dependence on temperature of the relative isobaric heat capacity for DOAB(aq), 6C,, monitored using a differential scanning microcalorimeter. The plots have been displaced for clarity on the dC, axis. The curves characterise solutions of DOAB prepared according to four different protocols: see Experimental; (a) hot-water method; (b) sonication method; (c) hot-water method using HEPES buffered solution; (d)sonication method using HEPES buffered solu- tions. significance of the presence of buffer to the properties of DOAB solutions is confirmed in Fig. 11 which records four typical scans obtained using the differential scanning micro- calorimeter .For DOAB(aq) prepared in the absence of buffer using both the hot-water and sonicated methods, two extrema are observed in the DSC scans near 36 and 45°C where [DOAB] = 2 x mol dm-3. When prepared in the pres- ence of buffer only one extremum is observed in these traces near 47°C. In other words, addition of HEPES buffer raises the gel-to-liquid crystal transition temperature ;i.e. stabilises the gel state of the vesicle. Discussion As mentioned in the Introduction, we anticipated that the interaction between vesicle and fusogenic agents would be exothermic. Consequently, several features of the injection plots in Fig. 1, 5 and 8 are of immediate interest. Following each injection, the endothermic pulses are sharply resolved, a short length of 'baseline' being recorded between each pulse.This pattern signals that within a few tens of seconds the solution has recovered an equilibrium state. We cannot comment more precisely on the rate because the time depen- dence is determined by the calorimeter response time. Never- theless, we can be confident that there is no slow (chemical) reaction or process involving either DPA or SO:-ions and DOAB vesicles. The endothermicity decreased with each injection of an aliquot of Na2S04(aq). In one possible model for the process, added SO:-ions displace bromide ions from the double layer. If this was the sole explanation, one might expect a series of pulses of similar magnitude for each aliquot until all the bromide ions had been displaced.The resultant pattern would be a plot with a well defined step from constant endo- thermicity to athermal conditions. This pattern is not observed but, continuing with the essential features of this model, bound SO:-ions clearly modify the energetics of dis- placement of Br- ions by the next aliquot of SO:-ions. Returning to the possible step shape discussed above, the expectation was that, on the basis of charge numbers, each SO:-ion would displace two bromide ions. In these terms it is surprising that the injection becomes athermal at near equimolar amounts of SO:-and Br-. We also note that two processes associated with the head groups on the outside and on the inside of vesicles are not detected. Taken in conjunc- tion with comments concerning the baseline between pulses, we conclude that DOAB vesicles above the transition tem- perature are very leaky allowing SO:-and Br- to pass freely across the vesicle bilayer.The observed pattern can be accounted for in terms of an equilibrium involving SO:-ions which are either bound to vesicles or 'free' in the aqueous solution. The binding reac- tion might therefore involve displacement of bound Br -ions from the vesicles. This process would be characterised by a binding constant and an enthalpy of binding. Moreover, the dependence of 4 on amount of added SO:-(aq) would form a sigmoidal dependence centred around 4 for a solution where the ratio [Na2S04] : [DOAB] is unity. The fact that this pattern is not observed rules out an equilibrium-based model. Rather, the process involves direct replacement of Br- by SO:-, Scheme 1. liquidchains I I I Scheme 1 However, as shown in Scheme 1, the process would be exo- thermic.The dominant endothermic process stems from an accompanying dehydration of the cationic head groups which is also known to be a precursor to fusion." Thus injection becomes athermal when all sites on the vesicle surface are occupied by SO:-ions. Interaction between N'Me, head groups and SO:-ions is exothermic but the endothermicity is the consequence of displacing tightly bound water into the bulk solution and replacement by bound SO:-ions. When small amounts of SO:-ions are added, the boundary layer comprises mainly bromide ions with relatively small numbers of SO:-ions. In other words, the mean separation of SO:-on the surface of the vesicles is large.Nevertheless, the SO:--Br -repulsion means that displacement of further bromide ions and water by added SO:-(aq) is less endo- thermic. This process continues until all bromide ions have been displaced. In other words, there appear to be two processes, one endothermic and one exothermic, which depend on the extent to which bromide ions have been displaced. Each contrib- uting process can be described by the following empirical equation, where i = 1 denotes an exothermic process and i = 2 an endothermic one. AHi = AH: exp(-mi n/N)[N -n)/N] (1) Here AH: represents the enthalpy change for the displace- ment of the first bromide ion by SO:-where AH(obs) = AHl + AH2.Here N represents the number of sites on the surface of the bilayer and n is the number of sites occupied by added ions (cf:injection number). The exponential term through the parameters a1 and a2 produces a characteristic shape to the recorded plot whereas the term linear in (N -n) assures that the plot passes through zero at n = N.We show in Fig. 12 a typical plot which emerges from eqn. (1) for a set of selected parameters. In a process of trial and error we attempted to reproduce as closely as possible the pattern of the recorded J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 closely related to the added fusogenic ions, have already par- tially dehydrated the head groups.I I I I 0 100 200 injection number Fig. 12 Dependence of enthalpy change as a function of injection number, AH, as predicted using eqn. (1) for (a) endothermic process with AHo = -1.5 kcal mol-' and a = 0.4, (b) exothermic process with AHo = 2.0 kcal mol-' and o! = 0.6, and (c) resultant enthalpy change titration plots. Therefore, we reached the conclusion that the endothermic peaks (cf: Fig. 5) in the titration calorimetric experiments following addition of SOt-(aq) to DOAB(aq) are not linked directly to fusogenic processes. In all probability, the fusion process continues in the background to the changes recorded here. It is interesting to note that a reversal in sign of enthalpies of interaction was observed by Jones et al. in their study of interactions between alkylsulfates and the macromolecules, ribonuclease.' 'J' When DOAB(aq) is prepared in the presence of buffer, the gel state is, according to the DSC experiments, stabilised.The presence of a double extremum in the absence of buffer is attributed to melting of patches of vesicles in close proximity (intervesicular) and to the melting of patches unaffected by intervesicular interactions. When buffer is added, the amphi- pathic anions (e.g.XSO, and, marginally, CH,CO,) displace bromide ions from near the vesicle surface. These bound XSO, ions stabilise and insulate the vesicles by weakening intervesicular interactions. Consequently, a single extremum at higher temperatures is recorded in the DSC, Fig. 11. In the titration experiment added SO;-ions displace the bound XSO, ions, Scheme 2.Hence the overall process is less endo- thermic. The associated dehydration of the polar N'Me, head groups is not as dramatic because XSO, ions, being 1 Scheme 2 Overall, the two sets of titration results differ because SOt-(aq) ions are added to DOAB vesicles having quite dif- ferent compositions in their associated double layers. The titration plots describing injection of DPA(aq) into DOAB(aq) are more complicated than observed when the fusogenic agent SO:-was used. However, the production of an endothermic maximum points again to two contributions, one exothermic and one endothermic, eqn. (1) and Fig. 12. However, as the experiments show with buffered solutions using SOi-(aq) injections, the buffer is not passive.We thank the University of Leicester for a travel grant to M.J.B. We thank SERC for a grant to M.D.B. This research was supported under the Molecular Recognition Initiative at the University of Leicester. References 1 J. H. Fendler, Acc. Chem. Res., 1980,13, 7. 2 J. H. Fendler, Chem. Rev., 1987,87,877. 3 T. Kunitake, Angew. Chem., Int. Ed., 1992,31,709. 4 A. M. Carmona-Ribeiro, Chem. SOC.Reu., 1992,21,209. 5 D. Yogev, B. C. R. Guillaume and J. H. Fendler, Langmuir, 1991, 7,623. 6 I. M. Cuccovia, E. Feitosa, H. Chaimovich, L. Sepulveda and W. Reed, J. Phys. Chem., 1990,94,3722. 7 T. A. A. Fonteyn, D. Hoekstra and J. B. F. N. Engberts, J. Am. Chem. SOC.,1990,112,8870. 8 M. J. Blandamer, B. Briggs, P. M. Cullis, J. A. Green, M. Waters, G. Soldi, J. B. F. N. Engberts and D. Hoekstra, J. Chem. SOC.,Faraday Trans., 1992, in the press. 9 T. Wiseman, S. Williston, J. F. Brandts and L-N. Lin, Anal Biochem., 1989,179, 131. 10 L. A. M. Rupert, J. F. L. Breemen, D. Hoekstra and J. B. F. N. Engberts, J. Phys. Chem., 1988,92,4416. 11 M. N. Jones, Chem. SOC.Rev., 1992,21,85. 12 M. I. Paz Andrade, E. Boitard, M. A. Saghal, P. Manley, M. N. Jones and H.A. Skinner, J. Chem. SOC., Faruday Trans. I, 1981, 77,2939. Paper 3/0555OG; Received 15th September, 1993