Proc. Int. Symp. Biomol. Struct. Interactions, Suppl. J. Biosci., Vol. 8, Nos 1 & 2, August 1985, pp. 223–238. © Printed in India.

A guest-host approach to oligodepsipeptide structure M. GOODMAN, Υ. V. VENKATACHALAPATHI, S. MAMMI and R. KATAKAI* Department of Chemistry, B-014, University of California, San Diego, La Jolla, California, 92093, USA *Department of Chemistry, College of Technology, Gunma University, Tenjin-cho, Kiryu-shi 376, Japan Abstract. In this paper we investigate the effect of main chain isosteric replacement of specific amino acid residues by α-hydroxy acids. As part of a long term program specifically protected heptaglutamates were prepared and their circular dichroism and nuclear magnetic resonance spectra in various solvents were examined. From these experiments conformational preferences were deduced. We have also prepared oligo-(γ-methyl-glutamates) replacing the amino acids at specific positions along the chain with S-lactic acid and have elucidated the effect of these main chain isosteric replacements on oligopeptide structure. Analogues of collagen also have been prepared with glycolic acid replacing specific glycine residues. We synthesized the model hexamers Ac-Ala-Gly-Pro-Ala-Gly-Pro-NHMe, Ac-AlaGlc-Pro-Ala-Gly-Pro-NHMe, and Ac-Ala-Gly-Pro-Ala-Glc-Pro-NHMe in order to study their structural characteristics under various conditions. Preliminary nuclear magnetic resonance and circular dichroism results are presented. Keywords. Oligodepsipeptides; nuclear magnetic resonance; circular dichroism; hydrogen bonding.

Introduction As a part of our programme on the study of biologically interesting peptides and polypeptides, we have concerned ourselves with the replacement of specific amino acids with α-hydroxy acid residues to assess the effect of removal of key hydrogen bond donor groups on secondary structure. We have prepared polydepsipeptides in which Slactic acid residues were incorporated in poly-alanine (Becktel et al.,1980; Goodman et al., 1980), poly-γ-methyl-L-glutamate (Katakai and Goodman, 1982), and a L-lysine-γmethyl-L-glutamate copolymer (Katakai and Goodman, 1982). The conformational effects of these specific isosteric replacements were determined and interpreted on a thermodynamic basis. In this paper we examine similar replacements on monodisperse oligo-γ-methyl-L-glutamates and sequence oligopeptides related to collagen structure. Experimental section Nuclear magnetic resonance (NMR) spectra were obtained on a 360 MHz spectrometer built in-house from a Varian instrument equipped with an Oxford magnet and Abbreviations used: NMR, Nuclear magnetic resonance; CD, circular dichroism; TFE, trifluoroethanol; UV, ultra-violet

223

224

Goodman et al.

Nicolet 1280 computer. NMR chemical shifts are in ppm from TMS as internal standard. Circular dichroism (CD) measurements were carried out on a Cary 61 spectropolarimeter which was modified by replacing the original Pockel cell with a 50 KHz photoelastic modulator (Hinds International FS-5/PEM-80), used in conjunction with a lock in amplifier (EG and G Princeton Applied Research no. 128) to detect and integrate the modulation. System automation, multiple scan signal averaging, and base line subtraction were accomplished by a DEC 11/02 computer interfaced directly to both the Cary 61 and the amplifier. The system software and custom hardware interfaces were designed by Allen Microcomputer Services Inc. CD spectra were obtained with a 0.02 cm cell, by signal-averaging 20 scans.

Results and discussion NMR study of oligodepsipeptides in CDCl3 In previous NMR studies of oligopeptides such as Boc-(Met)7-OMe the guest-host method (Ribeiro et al., 1978) and incorporation of α-deuterated methionyl residues at specific sites along the chain (Naider et al., 1979) were used to assign the amide NH and α-CH resonances to individual amino acids. The identical assignment patterns were found in the same solvents for Z-[Glu(OEt)]n-OEt extensive homonuclear decoupling experiments combined with guest-host analogs of co-oligopeptides were used (Ribeiro et al., 1980). In Boc-[Glu(OMe)]n-OPOE where the oligoglutamates are bound to polyoxyethylene chain the amide proton resonances were unequivocally assigned to individual amino acid residues from the studies of α-deuterated glutamic residues at specific sites along the chain. In all these studies similar patterns and positions of amide resonances were shown to be present in methionyl as well as in γmethyl-glutamyl peptides. Preliminary results are presented for a series of oligodepsipeptides of the general form:

Oligodepsipeptide structure

225

Table 1 shows the chemical shifts and assignments of the amide resonances in CDCl3 of these oligodepsipeptides. To accomplish the assignments of the amide resonances we used a guest-host approach based on comparison with analogous oligopeptides and with smaller analogues. In all these compounds the resonance of Glu1-NH is unequivocally assigned as that occurring at high field (~ 5·6 ppm) since this proton is a part of urethane grouping. The guideline in the assignments was the inductive effect of the ester bond on the amide protons of the adjacent residues. The experimental result, which is common to all the compounds studied, is that the resonances of the amide protons immediately preceding and following the lactic acid are shifted respectively to higher and lower field when compared to the corresponding amide resonance in oligopeptide analogues. In the series Boc-Glu(OMe)-Lac-[Glu(OMe)]n-OMe where n = 1–5, the effect of the lactic acid can be followed systematically. The position of the urethane NH is clearly affected by the presence of the ester bond: it is always upfield shifted when compared to the analogous peptide. The assignment of the other resonance in the tridepsipeptide Boc-Glu(OMe)-Lac-[Glu(OMe)]-OMe (I) is straightforward, appearing at 7·24 ppm. The same residue in a glutamate tripeptide resonates at 7·14 ppm. The comparatively small downfield shift arises from the fact that this residue is not only adjacent to the lactic acid, but is also a C-terminal residue. In the analogous tetrapeptide, the Glu3-NH resonates at 7·36 ppm and the terminal residue amide proton appears at 7·21 ppm (Saltman, R. P., Goodman, Μ. and Ribeiro, Α. Α., unpublished results). In the tetradepsipeptide Boc-Glu(OMe)-Lac[Glu(OMe)]2-OMe (II) the two amide proton resonances occur at 7·44 and 6·96 ppm. The resonance at 7·44 ppm was therefore assigned to the residue Glu3 by comparison of the corresponding peptide and depsipeptide. In the pentapeptide the amide resonances of residues 3, 4 and 5 occur at 7·63, 7·35 and 7·15 ppm. In the depsipeptide these resonances occur at 7·69, 7·07 and 6·99 ppm. The resonance at 7·69 ppm was assigned to the Glu3-NH whereas the other two resonances belong to C-terminal residues. The inductive effect of lactic acid is distance-dependent. The effect is probably very small on amide protons further away than one residue. Thus, the behaviour of such residues in the molecule is similar to a normal peptide. With this in mind, we assigned the resonances at 7·07 and 6·99 ppm to the Glu4 and to the Cterminal residue, respectively. In the glutamate hexapeptide derivative, the Glu3-NH occurs at 7·88 ppm while in the corresponding depsipeptide Boc-Glu(OMe)-Lac-[Glu(OMe)]4-OMe (IV) it appears at 8·01 ppm. Lastly, in the glutamate heptapeptide, the Glu3-NH is at 7·95 ppm whereas in the depsipeptide Boc-Glu(OMe)-Lac-[Glu(OMe)]5-OMe (V) the corresponding Glu3 amide proton is at 8·08 ppm. In these two depsipeptides the resonances of amide protons not adjacent to the lactic acid were assigned by comparison with analogous homo-oligopeptides, as described above for the pentapeptide. Using this approach, all the amide resonances were assigned to individual residues. In the Boc-[Glu(OMe)]2-Lac-[Glu(OMe)]n-OMe series a second glutamic acid residue is present at the N-terminal side of the lactic acid. The resonances for this residue are also downfield shifted with respect to the amide of the second residue in the analogous homo-oligoglutamate peptides. This can be seen in Boc-[Glu(OMe)]2-Lac-

226

Goodman et al.

[[

Oligodepsipeptide structure

227

228

Goodman et al.

[Glu(OMe)]3-OMe (VI) and Boc-[Glu(OMe)]2-Lac-[Glu(OMe)]4-OMe (VII), where the Glu2 NH appears at 8·10 and 8·08 ppm, respectively. It is possible unambiguously to assign the amide resonances in the series Boc[Glu(OMe)]3-Lac-[Glu(OMe)]n-OMe using the same approach. There are two low field resonances which can be assigned to Glu2 and Glu3 amide protons. As the chain length is increased, the resonances of the amide protons following the lactic acid are assigned as above for the Boc-Glu(OMe)-Lac-[Glu(OMe)]n-OMe series. In the case of Boc-[Glu(OMe)]3-Lac-[Glu(OMe)]3-OMe(X) there is an ambiguity between Glu3 and Glu5, since the two resonances almost overlap. The ambiguity was resolved by comparison of the temperature coefficients of the amide protons along the series. With similar reasoning it was possible to assign the resonances in Boc-[Glu(OMe)]4Lac-[Glu(OMe)]4-OMe (XI) and Boc-[Glu(OMe)]2-Lac-[Glu(OMe)]2-Lac[Glu(OMe)]2-OMe (XII). Temperature coefficients of various amide protons for the series of depsipeptides in chloroform are also presented in table 1. In all of these depsipeptides, the urethane amide proton has a temperature coefficient between 4·8 × 10–3 and 6·1 × 10–3 ppm/°C which is indicative of weak solvent shielding, possibly by intramolecular C7 hydrogen bonding. In fact, similar types of temperature coefficients were seen in Boc[Glu(OMe)]7-OPOE and Z-[Glu(OEt)]7-OEt where such intramolecular hydrogen bonds were proposed (Ribeiro et al., 1978; Saltman, R. P., Goodman, Μ. and Ribeiro, Α. Α., unpublished results). From their temperature dependence in CDCl3 we can divide the amide protons into three categories: the N-terminal, internal, and C-terminal region. For residues 1, 2, and 3, the temperature coefficient is greater than 4·5 × 10–3 ppm/°C. This has been interpreted in terms of a hydrogen bond, present at room temperature, which is disrupted at higher temperatures (Pysh and Toniolo, 1977). Analogous results in the peptide series have been interpreted as indicative of the presence of intramolecular hydrogen bonds (Saltman, R. P., Goodman, Μ. and Ribeiro, Α. Α., unpublished results). We therefore propose that the depsipeptide analogues similarly fold in chloroform, possibly forming C 7 hydrogen bonds. The second category of temperature dependence includes most of the internal residues that follow the lactic acid residue. For these amides, the value of the temperature coefficient is between 2·0 × 10–3 ppm/°C and 4·5 × 10–3 ppm/°C, which has been interpreted in the literature as indicative of non-hydrogen bonded amides. Inspection of the temperature coefficient of the C-terminal residues reveals two types of behaviour. An intermediate value indicates solvated amides. A very low value, such as 0·7 × l0–3 ppm/°C in Boc-[Glut(OMe)]4-Lac-[Glu(OMe)]4-OMe has been interpreted in terms of hydrogen bonds strong enough not to be disrupted by temperature. We believe that these strong hydrogen bonds at the C-terminus arise from intermolecular association. The different behaviour of the various compounds studied thus reflects their different tendencies to aggregate. We have carried out preliminary experiments on the effect of addition of a strongly hydrogen bonding solvent such as DMSO on the structure of depsipeptides in chloroform. Figures 1–3 show the solvent dependence of the chemical shifts of amide protons in 3 heptadepsipeptides in CDCl3/DMSO-d6 mixtures. It is clear from these plots that in all cases the urethane amide proton is easily accessible to the strongly

[

Oligodepsipeptide structure

229

Figure 1. Solvent dependence of the amide NH chemical shifts in Boc-[Glu(OMe)]-Lac[Glu(OMe)]5-OMe in CDCl3 /DMSO-d6 mixtures. The concentration is 4·2 mM.

hydrogen bonding solvent. This result is consistent with the temperature studies in chloroform if we assume that this amide is involved in a weak intramolecular hydrogen bonding which is affected by temperature, and disrupted by a strong hydrogen bond acceptor such as DMSO. The main feature of the plot for compound (V) is the uniqueness of the behaviour of the Glu3 residue. The NH for this residue does not shift at all, while the other NH resonances exhibit a linear dependence with positive slope. This suggests intramolecular hydrogen bonding for the Glu3 amide in chloroform. Such a hydrogen bond deshields the proton, so that the addition of DMSO does not cause more deshielding. The behaviour of the other two heptadepsipeptides (VII) and (X) toward addition of DMSO is complex. Three types of solvent dependencies are observed for the amide protons: (i) large initial effect and plateau at higher concentrations of DMSO for Glu2NH of both compound (VII) and compound (X), (ii) large linear dependence for residues following the lactic acid residue in the chain, but not adjacent to it, and (iii) small linear dependence for Glu3-NH and Glu5-NH in compound (X). The Glu4-

230

Goodman et al.

Figure 2. Solvent dependence of the amide NH chemical shifts in Boc-[Glu(OMe)] 2-Lac[Glu(OMe)] 4-OMe in CDCl3/DMSO-d6 mixtures. The concentration is 3·7 mM.

NH in compound (VII) is similarly affected. Clearly, additional experiments are necessary to interpret the behaviour and to deduce a complete picture of the effect of solvent on the structure of these depsipeptides. CD study of oligoglutamates and oligodepsipeptide analogues We are investigating these depsipeptide series using CD to assess overall secondary structure. The studies described here have been conducted in trifluoroethanol (TFE), at an approximate concentration of 10–3 mol/1. Chloroform does not allow sufficient penetration into the ultra-violet (UV) region to follow the characteristic transitions of the amide and ester bonds. We were therefore forced to use a transparent solvent such as TFE, bearing in mind that the results found in this solvent likely involve different conformations from those present in chloroform. In fact, the secondary structures proposed for Boc-[Glu(OMe)]n-OPOE in the two solvents, based on NMR

Oligodepsipeptide structure

231

Figure 3. Solvent dependence of the amide NH chemical shifts in Boc-[Glu(OMe)]3-Lac[Glu(OMe)]3-OMe in CDCl3/DMSO-d6 mixtures. The concentration is 3·7 mM.

studies and computer simulations are very different. A series of C7 hydrogen bonds was proposed for the heptapeptide in chloroform. For the same compound the α-helix is the most probable ordered structure present in TFE (Saltman, R. P., Goodman, Μ. And Ribeiro, Α. Α., unpublished results). In the Boc-[Glu(OMe)]-Lac-[Glu(OMe)]n-OMe series (figure 4) where n varies from 1 to 5, there is a consistent increase of the intensity of the negative band below 200 nm, and a red shift of its position. The spectrum of Boc-[Glu(OMe)]-Lac[Glu(OMe)]-OMe is dominated by a positive band centered around 208 nm, which probably arises from the coupling of the n–π* transition of the ester bond and the π–π* transition of the amide. With the addition of a glutamyl residue (compound II), a negative band appears at about 195 nm which represents the contribution of the coupling between the π–π* transitions between different amides. The trend proceeds through the series. At the hexamer level the contribution from the positive band is

232

Goodman et al.

Figure 4. CD spectra of Boc-[Glu(OMe)]-Lac-[Glu(OMe)]n-OMe in TFE. The concentrations are: n = 1, 1·3 mM; n = 2, 1·4 mM; n = 3, 1·0 mM; n = 4, 1·0 mM; n = 5, 1·0 mM.

negligible. In the heptamer the negative band has shifted to 200 nm, and a strong shoulder appears centered around 225 nm. In the Boc-[Glu(OMe)]3-Lac-[Glu(OMe)]n-OMe series (figure 5) the pentamer n = 1 shows the negative band centered around 192 nm and the positive band centered around 210 nm. When n = 2, both undergo a slight red shift, but the relative intensity changes drastically. When n = 3, there is a definite shift of the negative band from 194 nm to 199 nm without appreciable change in its intensity. It is interesting to follow the effect of the substitution of lactic acid for glutamic acid maintaining the total chain length constant and moving the hydroxy acid residue along the chain. In the case of the pentamers we notice a strong dependence of the spectrum on the position of the substitution. The spectrum of Boc-[Glu(OMe)]-Lac[Glu(OMe)]3-OMe is very similar to that of Boc-[Glu(OMe)]5-OMe, suggesting that the amide-ester interaction is not very effective. On the other hand, the importance of this interaction is clear in the spectrum of Boc-[Glu(OMe)]3-Lac-[Glu(OMe)]-OMe. We do not believe that any kind of ordered structure is present in these short compounds. Therefore, the differences found in the CD spectra, simply reflect the effect of the ester bond on the chiroptical properties of a randomly coiled oligopeptide.

Oligodepsipeptide structure

233

Figure 5. CD spectra of Boc-[Glu(OMe)]3- Lac- [Glu(OMe)]n-OMe in TFE. The concentrations are: n = 1, 1·0 mM; n = 2, 1·1 mM, n = 3, 1·1 mM.

Placing the hydroxyacid moiety in different positions along the chain has different effects that may be due to the different relative strength of coupling between the transitions of the ester and the amide groups as compared to the coupling of transitions between two amides. If the chain length is further increased to six residues, placement of a lactic acid residue at position 2 or 4 results in rather similar spectra, even though the negative band is much stronger in the former compound. In both of the spectra the ester-amide interaction is overwhelmed by the amide-amide π–π* coupling. The same type of analysis is valid for the heptamer series where the compounds examined thus far contain lactic acid at position 2 or 4. In neither compound the n–π* transition of the ester is efficiently coupled with the π–π* transition of the amide, and the π–π* coupling of different amides is prevailing. These compounds are interesting also from another point of view. The parent heptapeptide is the shortest oligoglutamate showing some degree of α-helical structure in TFE (Goodman et al., 1969). The removal of a hydrogen bond donor can disrupt the ordered structure to an extent that depends on the position of the substitution. In an α-helical array of the peptide chain, the first residue involved in a hydrogen bond as a donor is the fourth from the N-terminus. Consequently, if the substitution occurs at the second residue, the effect should be minimal., while a large effect should be found for substitution at position 4. The [

234

Goodman et al.

substantial difference in the spectra of Boc-[Glu(OMe)]-Lac-[Glu(OMe)]5-OMe and of Boc-[Glu(OMe)]3-Lac-[Glu(OMe)]3-OMe is consistent with this reasoning: the spectrum of the former is similar to the parent peptide while the spectrum of the latter is not. Another interesting comparison is between Boc-[Glu(OMe)]3-Lac-[Glu(OMe)]3OMe and Boc-[Glu(OMe)]4-Lac-[Glu(OMe)]4-OMe (figure 6). These data suggest a definite increase of α-helical structure in the nonamer with respect to the heptamer. The presence of the lactic acid substantially reduces the α-helical content in the heptamer, but not in the nonamer.

Figure 6. CD spectra of Boc-[Glu(OMe)]3-Lac[Glu(OMe)]3-OMe (1) and Boc[Glu(OMe)]4-Lac-[Glu(OMe)]4-OMe (2) in TFE. The concentrations are 1·1 and 1·0 mM, respectively.

We have also investigated the effect of a double substitution in a single chain. In figure 7, the spectra of Boc- [Glu(OMe)]2-Lac- [Glu(OMe)]2-Lac- [Glu(OMe)]2-OMe and Boc-[Glu(OMe)]3-Lac[Glu(OMe)]3-Lac-[Glu(OMe)]3-OMe are compared. It is apparent that the effect of the hydroxy acid moiety is large in the former compound. We believe that a higher degree of ordered structures is present in the undecamer than in the octamer.

Oligodepsipeptide structure

235

Figure 7. CD spectra of Boc-[Glu(OMe)]2-Lac-[Glu(OMe)]2-Lac-[Glu(OMe)]2-OMe (1) and Boc-[Glu(OMe)]3-Lac-[Glu(OMe)]3-Lac.[Glu(OMe)]3-OMe (2) in TFE. The concentrations are 0·7 and 0·6 mM, respectively.

NMR and CD study of collagen model compounds We are also employing the guest–host technique to understand the importance of hydrogen bonds to the stability of structures present in biological systems. During these studies we have synthesized model hexamers of collagen-like composition. The structures of these compounds are as follows: Ac-Ala-Gly-Pro-Ala-Gly-Pro-NMHe Ac-Ala-Glc-Pro-Ala-Gly-Pro-NHMe Ac-Ala-Gly-Pro-Ala-Glc-Pro-NHMe where Glc represents glycolic acid, the hydroxy analogue of glycine. We present here initial results of our spectroscopic studies on these compounds. NMR. The amide region of the proton spectra for these compounds is presented in figure 8. The assignments were accomplished by 2D-COSY spectroscopy. The ambiguity between the two alanines was resolved by comparison of the acetylated peptide or depsipeptide with the corresponding benzyloxycarbonyl or t-butyloxycarbonyl protected derivatives. The ambiguity between the two glycines in

236

Goodman et al.

Figure 8. Partial 1H NMR spectra of Ac-Ala-Gly-Pro-Ala-Gly-Pro-NHMe (1), Ac-AlaGlc-Pro-Ala-Gly-Pro-NHMe (2), and Ac-Ala-Gly-Pro-Ala-Glc-Pro-NHMe (3) in DMSO-d6. c = cis and t = trans. The concentrations are 5·1, 11·9, and 16·2 mM, respectively.

the hexapeptide was resolved by comparison with the two hexadepsipeptides. We note that the Ala1 resonance is less affected by the cis-trans isomerism than the resonance of Ala4. The same behaviour is found for the two glycines. This result is consistent with the fact that Ala4 and Gly5 are in the centre of the sequence. This leads to a cumulative effect of the prolines at either side and possibly an increased rigidity. Alternatively, this result could reflect the effectiveness of cis-trans isomerism only on the residues that follow proline. The N-methylamide NH always shows a doubling of the peaks. An important feature revealed by the spectra is the deshielding effect of the ester bond. In both depsipeptides we observe that the alanine adjacent to the glycolic acid resonates at lower fields than the other alanine. Moreover, the position of this resonance is downfield shifted in the depsipeptide when compared with the same resonance in the parent peptide. This behaviour is different from the behaviour found in the guest-host compounds of oligo-γ-methylglutamates and lactic acid described [

Oligodepsipeptide structure

237

above. These differences arise likely from the different nature of the peptides and the solvents employed. CD. The analysis of the CD data in TFE is based on the assumption that the conformation is the same in all the three cases. The spectra of the three compounds under study are presented in figure 9. We see that the spectrum of one of the depsipeptides is superimposable on the spectrum of the peptide. In the spectrum of the other depsipeptide, it is apparent that a positive band is present at about 230 nm. In the latter compound, the glycolic acid is in position 5 and there is only one residue following it. The coupling of the n–π* transition of the ester and the π– π* transition of the amide is very likely responsible for the positive band. We think that changing the position of the glycolic acid does not have an effect on the conformation assumed by these compounds in TFE. The effect we see simply arises from the different contributions of the amide and of the ester chromophores in analogy with the oligo-γmethylglutamates-lactic acid series. Studies are in progress in our laboratories to examine the effect of the position of the ester group on the conformation of the

Figure 9. CD spectra of Ac-Ala-Gly-Pro-Ala-Gly-Pro-NHMe, and Ac-Ala-Glc-Pro-AlaGly-Pro-NHMe (1), and Ac-Ala-Gly-Pro-Ala-Glc-Pro-NHMe (2) in TFE. The concentrations are 1·0, 1·1, and 0·8 mM, respectively.

238

Goodman et al.

depsipeptide. This information will help us interpret the spectra of polydepsipeptide model compounds related to collagen. The studies presented above indicate the subtleties of replacement of specific amino acid residues with α-hydroxy acids. We are probing these effects in order to elucidate the conformational characteristics of peptides modified in a guest-host sense. Acknowledgements This work is supported by NSF grant CHE 80-23002 and by NIH grant GM 18694. We wish to thank Dr. Piero Andrea Temussi for helpful discussions and Joseph Taulane of our laboratories for building the state of the art circular dichroism instrument from the basic Cary 61. References Becktel, W. J., Mathias, L. J. and Goodman, Μ. (1980) Macromolecules, 14, 203. Goodman, Μ., Verdini, A. S., Toniolo, C, Phillips, W. D. and Bovey, A. (1969) Proc. Natl. Acad. Sci. USA, 64, 444. Goodman, Μ., Becktel, W. J., Katakai, R. and Wouters, G. (1980) Makromol. Chem. Suppl., 4, 100. Katakai, R. and Goodman, Μ. (1982) Macromolecules, 15, 25. Naider, F., Sipzner, R., Steinfeld, A. S., Becker, J. M., Ribeiro, A. A. and Goodman, Μ. (1979) in Proceedings of the 6th American Peptide Symposium (eds E. Gross and J. Meienhofer) (Rockford: Pierce Chemical Co.) p. 185. Pysh, E. S. and Toniolo, C. (1977) J. Am. Chem. Soc., 99, 6211. Ribeiro, Α. Α., Goodman, Μ. and Naider, F. (1978) J. Am. Chem. Soc., 100, 3903. Ribeiro, Α. Α., Saltman, R. P. and Goodman, M. (1980) Biopolymers, 19, 1771.