of two additional components that appear to be aldol condensation products. These are designated aldol 1 and aldol 2 in Table 11. They became the major components in samples prepared from one batch of diethyl aminomalonate, perhaps because of the presence of some impurity that catalyzed their formation. However, their content was almost indetectable in the solutions for which data are reported here.
+
DISCUSSION Our goal in this study was to establish the state of protonation and pK, values for quinonoid species derived from PLP in enzymes or in nonenzymic systems. The four enzymic quinonoid species (Figure 1) have similar spectra with vibrational subbands from 1020 to 1110 cm-' apart. They all appear to be in pH-independent equilibria with Schiff base
QUINONOID SPECIES FROM PYRIDOXAL PHOSPHATE
forms having the state of protonation of the ring shown in Scheme I. The fact that the quinonoid species from 0methyl-PLP and L-aspartate has a spectrum very similar to those given by PLP (Table I) suggests strongly (Chen et al., 1987) that the chelated proton in quinonoid I11 of Scheme I is on the phenolic oxygen rather than on the imine nitrogen. Previously Matsumoto and Matsushima (1972, 1974) had suggested that migration of the proton from the imine nitrogen of the Schiff base to the phenolic -0- accompanies formation of the quinonoid structure. Replacement of the -OH by -OA12+ could provide the observed stabilization of the quinonoid forms in methanol by aluminum salts. Ledbetter et al. (1979) observed that laser-induced formation of a quinonoid band was favored in an A13+-phenolate complex of Schiff bases of pyridoxal plus amines but not when chelation of the metal occurred. The characteristic shape of the quinonoid adsorption bands is reminiscent of those of cyanine dyes and various other elongated chromophores (Malhotra & Whiting, 1960). It is probable that the prominent shoulder on the high-energy side of the highest peak represents vibrational structure. Such spectra can sometimes be approximated as the sum of a series of Gaussian bands whose relative heights follow a Poisson distribution (Gal et al., 1973). We have shown previously (Metzler et al., 1987; Chen et al., 1987) that the 491-nm absorption band of the erythro-3-hydroxyaspartatecomplex of cytosolic aspartate aminotransferase can be represented as the sum of a series of three nearly Gaussian lognormal curves of constant width and skewness and placed at intervals of about 1200 cm-l. This spacing of the subbands may be related to the 1238-cm-’ frequency seen in the coherent anti-Stokes Raman spectrum of the aspartate aminotransferaseserythro-3-hydroxyaspartatecomplex (Benecky et al., 1985). Our present results show that the use of Gaussian (skewness = 1.O) or nearly Gaussian curves will give an extremely good fit for all of the quinonoid bands, whether of enzyme complexes, diethyl aminomalonate derivatives, or aluminum complexes of alanine ethyl ester and pyridoxal. The narrow Gaussian bands have been combined with lognormal curves for the envelope of broader absorption bands to give complete descriptions of the spectra (Figures 1, 3B, 4B, 6, and 7; Table 1). An important question is whether or not the nonenzymic quinonoid species are chemically the same as those present in enzymes. The similarities are obvious: a narrow structured absorption band of high intensity and long wavelength. However, there are differences. The spacing between the component vibrational bands is somewhat greater (1330-1430 cm-l) for the nonenzymic spectra. The 1330-cm-’ spacing seen for the methanolic ethyl alaninate-pyridoxal-AI3+ system corresponds closely to the intense 1332-cm-’ frequency of the coherent anti-Stokes Raman scattering on a similar phenylalanine ethyl ester system (Benecky et al., 1985). The wavelengths of maximum absorption are from 480 to 500 nm for enzyme spectra but about 460 nm for the intense bands from diethyl aminomalonate. In addition, the diethyl aminomalonate systems contain a weaker band system on the high-energy (low-wavelength) side that appears to be absent in enzymes. One of the strongest arguments in favor of the quinonoid structure is the high intensity of the absorption bands. The molar areas of typical Schiff bases of pyridoxal phosphate, whether free or in enzymes, are about 350-400 Mm/mol, while that of an enzymic quinonoid species has been estimated as 698 Mm/mol (Metzler Lk Metzler, 1987). The areas of
VOL. 2 7 , N O . 1 3 , 1988
4931
the six Gaussian subbands in the diethyl aminomalonate-PLP spectra are estimated as over 1000 Mm/mol. Both the greater intensity of the quinonoid bands from diethyl aminomalonate and the greater width than for those from ethyl alaninate may be a result of the presence of the second ester group. This may also account for the fact that the wavelength of maximum absorption is at a higher energy for the PLP-diethyl aminomalonate complex (468 nm in methanol) than for the A13+ complex of pyridoxal and alanine ethyl ester (486 nm in methanol). The opposite orientation of the two-electron-withdrawing ester grouping in the diethyl aminomalonate derivatives may cause this shift to higher energy. In terms of energy, however, the difference is small. The energy of excitation is only 11 kJ/mol less for photons of 480-nm wavelength than for photons of a 460-nm wavelength. Some differences in positions of quinonoid bands in enzymes or nonenzymic systems may result from “solvent effects”. Thus, the difference in peak position for the PLP-diethyl aminomalonate quinonoid between water and methanol is 11 nm (Table I). The alternative possibility, suggested by Sala et al. (1987), is that the intense 460-nm band with diethyl aminomalonate represents not a quinonoid species but a Schiff base. The strongest argument in favor of a Schiff base structure is that a precipitated quinonoid compound, when dissolved in deuteriated dimethyl sulfoxide, gave the proton N M R spectrum expected for a Schiff base. Thus, the C4-H resonance was found at 9.06 ppm. By using PLP and N-methyl-PLP rather than pyridoxal, we were able to run NMR spectra on solutions of a variety of compositions and pH values in DzO. We also found the N M R spectra to resemble closely those of Schiff bases. For example, at pH 6.2, a solution of 2 mM PLP and 4 mM diethyl aminomalonate appeared to contain only two major components: PLP (C4-H at 10.4 ppm) and a component with the C4-H resonance at 9.18 ppm. However, we suggest that this latter resonance may arise from a rapidly equilibrating mixture of Schiff base plus quinonoid form. We find it difficult to predict the NMR spectrum of the quinonoid species. It could be very close to that of the Schiff base. We thought that the N M R resonance of the N-methyl group of the quinonoid form derived from N-methyl-PLP might undergo a strong upfield shift from that in the free aldehyde. However, this was not observed. A possible explanation is that the quinonoid character of the intermediate is seen mostly in the photoexcited state and that in the ground state a more localized carbanionic structure, as proposed by Martell (1 982), predominates. If we accept the structure of the PLP-diethyl aminomalonate complexes as quinonoid IV of Scheme 11, we must ask what causes the weaker bands in the 360-nm region. We suggest that these may represent tautomer V, in which the chelated proton has migrated back to the imine nitrogen. Similarly, tautomerism between VI and VI1 could occur in the higher pH form shown in Figure 5 . During the experiment depicted in Figure 2 the ratio of the areas of the higher energy transition (Gaussian subbands 4-6) to the areas of the first three subbands was constant through the first 122 min of decay. If we assume that both tautomers IV and V (Scheme 11) have the same molar area, the ratio of V (bands 4-6)/IV (bands 1-3) = 0.27. An alternative possibility is that the second set of subbands represents not a tautomer but an additional electronic transition. However, the @-methyl-PLP-containing quinonoid complex (Figure 1B) has no absorption that could be attributed to a second electronic transition. Tautomerism is also im-
4932
BIOCHEMISTRY
METZLER ET AL.
possible for this species. The A13+complex of pyridoxal and alanine ethyl ester in methanol (Table I) also has no proton that could participate in such tautomerism. It does have a weak structured band but a considerably higher energy relative to that of the major band. The band at 327 nm in the 1.25-min spectrum of Figure 6 cannot be PMP, even though the latter compound appears later at this same position. It could very reasonably represent a higher energy transition belonging to the quinonoid species. A corresponding band appears to be present in the spectrum of the erythro-3-hydroxyaspartatecomplex of aspartate aminotransferase (Figure 1A). The reason for the 10-nm difference in the position of the lowest energy band for the A13+-pyridoxal-alanine ethyl ester complex alone and in the presence of excess alanine ethyl ester (Table I) is not obvious. Perhaps a molecular complex is formed at the higher concentration ratio. We found nearly the same values for the positions and widths of the Gaussian and lognormal curves used to resolve both spectra of Figure 7 as well as one at an intermediate pH. We assumed that each of the two most intense bands was accompanied by a weaker band appropriately spaced to represent vibrational structure (Table I). Two of the resolved spectra are shown in Figure 7. It is clear that our representation is reasonable. The spacing between the major bands is 2270 cm-I. Although this is greater than the 8 10-cm-' difference between the positions of the first subbands of forms H(SB + Q)2-and (SB Q)3-of eq 3, it is still reasonable to propose that the spectral changes observed with pH in Figure 7 are related to that seen in Figure 5. It is also possible that quinonoid bands with maxima around 400 nm in enzymes may represent dissociated structures such as VI of Scheme 11. We are unable yet to say whether there is a significant amount of Schiff base in equilibrium with the neutral quinonoid forms within the complex H(SB Q)2-of eq 3 or within (SB Q)*- of eq 1. However, KQ is probably quite high. The pK, of diethyl aminomalonate is 13.3, about 11 units lower than that of ethyl acetate (Pearson & Dillon, 1953). The pK, of the cy C H of the most protonated form of the Schiff base of alanine with 3-hydroxypyridine-4-aldehyde has been estimated by Dixon and Bruice (1973) as approximately 9. We may expect that the pK, for the corresponding Schiff base of diethyl aminomalonate will be several units lower than this as a result of addition of the second electron-withdrawing substituent. The same should be true for Schiff base form H2SB- of eq 3. For this to be true KO must be quite high (>loo) or the pK, of the protonated imine group of H2SBmust be unexpectedly low.
+
+
+
ACKNOWLEDGMENTS We thank Dr. Victor Chen for the N-methyl- and 0methyl-PLP, Rukmani Viswanath for the data for the graph in Figure lD, Dr. R. David Scott for help with the NMR spectroscopy, and the referees for valuable suggestions. REFERENCES Abbott, E. H., & Bobrick, M. A. (1973) Biochemistry 12, 846-85 1. Abbott, E. H., & Martell, A. E. (1973) J . Am. Chem. SOC. 95, 5014-5019. Benecky, M. J., Copeland, R. A., Rava, R. V., Feldhaus, R., Scott, R. D., Metzler, C. M., Metzler, D. E., & Spiro, T. G. (1985) J . Biol. Chem. 260, 11671-11678. Braunstein, A. E., & Shemyakin, M. M. (1953) Biokhimia (MOSCOW) 18, 393-41 1.
Brown, F. C., Hodgins, W. R., & Roswell, J. A. (1969) J. Biol. Chem. 244, 2809-28 15. Chen, V. J. (198 1) Ph.D. Dissertation, Iowa State University. Chen, V. J., Metzler, D. E., & Jenkins, W. T. (1987) J . Biol. Chem. 262, 14422-14427. Ching, W., & Kallen, R. G. (1979) Biochemistry 18, 821-830. Davis, L., & Metzler, D. E. (1972) Enzymes (3rd Ed.), 62-74. Dixon, J. E., & Bruice, T. C. (1973) Biochemistry 12, 4762-4766. Drewe, W. F., & Dunn, M. F. (1986) Biochemistry 25, 2494-2501. Dunn, M. F., Robustell, B., & Roy, M. (1987) in Biochemistry of Vitamin B6 Catalysis, Proceedings of the 7th International Congress on Chemical and Biological Aspects of Vitamin B6 Catalysis (Korpelo, T., & Christen, P., Eds.) Birkhaeuser, Basel; private communication. Gal, M., Kelly, G. R., & Kurucsev, T. (1973) J. Chem. SOC., Faraday Trans. 2 69, 395-402. Iwata, C. (1968) Biochem. Prep. 12, 117-121. Jenkins, W. T. (1961) J . Biol. Chem. 234, 1121-1 125. June, D. S., Suelter, C. H., & Dye, J. L. (1 98 1) Biochemistry 20, 27 14-27 19. Karube, Y., & Matsushima, Y . (1976) J. Am. Chem. SOC.98, 3725-3726. Karube, Y., & Matsushima, Y . (1977) J. Am. Chem. SOC.99, 73 56-7 358. Kurucsev, T. (1974) J . Chem. Educ. 55, 128-129. Ledbetter, J. W., Askins, H. W., & Hartmann, R. S.(1979) J . A m . Chem. SOC.101, 4284-4289. Malhotra, S. S., & Whiting, M. C. (1960) J . Chem. SOC., 3812-3822. Martell, A. E. (1982) Adv. Enzymol. Relat. Areas Mol. Biol. 53, 163-199. Martell, A. E., & Taylor, P. (1984) Znorg. Chem. 23, 27 34-27 35. Martin, R. B. (1987) Znorg. Chem. 26, 2197-2198. Matsumoto, S.,& Matsushima, Y . (1972) J. Am. Chem. SOC. 94, 721 1-7213. Matsumoto, S., & Matsushima, Y . (1974) J. Am. Chem. SOC. 96, 5228-5232. Matthew, M., & Neuberger, A. (1963) Biochem. J . 87, 60 1-6 12. Metzler, C. M., & Metzler, D. E. (1987) Anal. Biochem. (in press). Metzler, C. M., Cahill, A., & Metzler, D. E. (1980) J . Am. Chem. SOC.102, 6075-6082. Metzler, D. E., Ikawa, M., & Snell, E. E. (1954a) J . Am. Chem. SOC.76, 648-652. Metzler, D. E., Longenecker, J. B., & Snell, E. E. (1954a) J . A m . Chem. SOC.76, 639-642. Metzler, D. E., Harris, C. M., Johnson, R. J., Siano, D. B., & Thompson, J. S. (1973) Biochemistry 12, 5377-5392. Morino, Y., & Snell, E. E. (1967) J . Biol. Chem. 242, 2800-2809. Nagano, K., & Metzler, D. E. (1967) J. Am. Chem. SOC.89, 2891-2900. Pearson, R. G., & Dillon, R. L. (1953) J . Am. Chem. SOC. 75, 2439-2443. Pocker, A., & Fischer, E. H . (1969) Biochemistry 8, 5 18 1-5188. Sala, L. F., Martell, A. E., & Motekaitis, R. J. (1987) Znorg. Chim. Acta 135, 123-127.
Biochemistry 1988,27, 4933-4938 Schirch, L., & Slotter, R. A. (1966) Biochemistry 5, 3 175-3 18 1. Scott, R. D., Chang, Y.-C., Graves, D. J., & Metzler, D. E. (1985) Biochemistry 24, 7668-768 1.
4933
Thanassi, J. W. (1975) Bioorg. Chem. 4, 132-135. Thanassi, J., & Fruton, J. S.(1962) Biochemistry I , 975-982. Ulevitch, R. J., & Kallen, R. G. (1977) Biochemistry 16, 5350-5 354.
Chemical Modification of Prothrombin Fragment 1: Documentation of Sequential, Two-Stage Loss of Protein Function+ Dean J. Welsch, Carol H. Pletcher,t and Gary L. Nelsestuen* Department of Biochemistry, College of Biological Sciences, The Uniuersity of Minnesota. St. Paul, Minnesota 551 08 Received January 11, 1988; Reuised Manuscript Receiued March 3, 1988
of prothrombin fragment 1 (amino acids 1-156 of prothrombin) were derivatized by acetylation, amidination, and reductive methylation. Conditions that caused complete acetylation of protein amino groups produced a fragment 1 derivative which no longer displayed a metal ion dependent intrinsic fluorescence change and had lost its membrane binding capability as well. However, when derivatized in the presence of calcium ions, extensive acetylation yielded a product that underwent protein fluorescence quenching at metal ion concentrations similar to those observed for the native protein. This derivative bound to membranes in a calcium-dependent manner with only a small reduction in affinity. Several results showed the existence of a partially functional protein that was characterized by a high degree of calcium-dependent protein fluorescence quenching but which had a requirement for 10-fold higher calcium concentration. This derivative was produced by partial acetylation (>3 equiv) of metal-free protein. This partially acetylated protein had greatly diminished membrane binding. The calcium-protected amino group, therefore, was among the most reactive acetylation sites in the metal-free protein. The second site, responsible for abolishing all metal ion induced fluorescence change, was resistant to acetylation and became derivatized a t the last stages of amino group acetylation. The second site did not function as a substitute for the first site. That is, both sites were shown to be essential for full protein function so that calcium actually protected both sites from acetylation. The second site, but not the first site, could undergo deacetylation under alkaline conditions and mild heat (pH 10,50 "C)or with hydroxylamine and mild heat (0.2 M, 50 "C). Thus, the fully acetylated protein could be returned to its intermediate state of function (high calcium requirement, low membrane binding) by these treatments. Amidination and reductive methylation of fragment 1 produced derivatives which were similar to the partially acetylated protein. These derivatives underwent protein fluorescence quenching which required 10-fold more calcium than native protein and had greatly reduced membrane affinity. This indicated that these subtle changes abolished the function of the first site but did not alter the function of the second site. ABSTRACT: The amino groups
E o t h r o m b i n , a plasma glycoprotein, is one of a group of proteins which requires vitamin K for the posttranslational carboxylation of glutamyl residues. The resulting calcium binding protein, which contains 10 y-carboxyglutamyl (Gla)' residues in its amino-terminal domain, is capable of calciumdependent membrane binding [reviewed in Nelsestuen (1984)l. Prothrombin fragment 1 (amino acids 1-1 56 of prothrombin) contains all of the Gla residues of prothrombin and has been used as a model peptide for calcium and membrane binding studies of vitamin K dependent proteins [Gitel et al., 1973; reviewed in Nelsestuen (1984)l. Recent X-ray crystallographic analyses of fragment I provide the potential for a detailed understanding of calcium ion binding to a Gla-containing protein (Tulinsky & Park, 1986). Relatively little is currently known about the precise metal ion binding sites in fragment
1. Loss of as few as two Gla residues results in considerable loss of protein function (Malhotra et al., 1985; Borowski et al., 1985; Wright et al., 1986). This has greatly restricted the ability for modification of Gla residues as an approach to understanding the metal binding functions of fragment 1. The present investigations were initiated to expand our understanding of the fragment 1 structure and to identify specific groups important to the various functions of fragment 1. These approaches may also offer methods of producing chemically modified proteins which allow insertion of spectroscopic probes at precise sites in the protein. The results provided direct chemical evidence for involvement of amino groups in the calcium and membrane binding activities of prothrombin fragment 1. Two essential sites were found that were sensitive to acetylation. Both were protected from ace-
'This work was supported in part by Grant HL-15728from the National Institutes of Health. Portions of this work have been presented previously in abstract form (Pletcher, 1981;Welsch & Nelsestuen, 1987). *Present address: Cargill Research, P.O. Box 9300,Minneapolis, M N 55440.
Abbreviations: Gla, y-carboxyglutamic acid; fragment 1, amino acids 1-156 of the amino terminus of bovine prothrombin; EDTA, ethylenediaminetetraacetic acid; TNBS, trinitrobenzenesulfonic acid; Tris, tris(hydroxymethy1)aminomethane; TSP,(trimethylsily1)propionate sodium salt.
0006-2960/88/0427-4933$01.50/0
'
0 1988 American Chemical Society
