MAXM. MARSH
1896
samples. The maximum variation in Soret absorption was 77,, and there was no correlation between the small differences in absorption and the differences in activity. As a final complication, we should note that there was a difference in the distribution of visible light absorption between the three peaks of Lots 5457 and 5458, as observed in the sedimentation dia-
[CONTRIBUTION FROM
Vol. 84
grams. In Lot 5458 absorption was associated with the leading peak only, whereas all three species absorbed light in Lot 5457. Acknowledgments.-This work was supported by research grants from the National Institute of Arthritis and Metabolic Diseases, National Institutes of Health, and from the National Science Foundation.
ANALYTICAL RESEARCH DEPARTMENT, ELI LILLYAND
CO
, INDIANAPOLIS 6, IND.]
Spectropolarimetric Studies on Proteins. Bovine Plasma Albumin and Insulin BY MAX11,MARSH RECEIVED MAY18, 1961 Optical rotatory dispersion curves from which the contributions of the sulfur-containing amino-acids were subtracted have been obtained for insulin and bovine plasma albumin. Drude equation and Moffitt equation constants derived from these curves have been calculated. Similar data on these proteins after cleavage of disulfide bonds provide a comparison of effects of the cleavage itself on optical rotatory properties of the proteins; application of two different methods of cleavage provides a comparison of effects of specific procedures, as well. Measurements on all materials were made in two different solvent systems. T h e selected solvents represent different extremes in terms of the conformational behavior of protein molecules. The presence of zinc in the insulin molecule is shown t o have a n appreciable effect on its optical rotatory properties.
Considerable emphasis is presently being placed on the measurement and interpretation of optical rotatory dispersion curves of proteins as an aid in the evaluation of the secondary and tertiary structural features of their configurations. In particular, the work of Moffitt,IF3, D ~ t y , ~Yangzs4.5, -~ B l o ~ t , ' - and ~ others involving optical rotatory dispersion measurements has shown interesting correlations with the apparent helical nature of polypeptides and proteins in solution. A complicating feature of the analysis of optical rotatory dispersion curves is the necessity for separation of the contributions t o the optical rotatory dispersion of the ordered secondary structure itself from those of the amino acid constituents of the protein. A working concept of these contributions is one which consists of a consideration of the anomalous character of the dispersion curve as being due to the ordered structure while that of the intrinsic rotatory contributions of the amino acid residues is thought of as being simple dispersion. A difficulty in utilizing this concept was pointed out by Turner, Bottle and Haurowitz,'O who demonstrated the effect of oxidation of cystine bridges in albumin qn the optical rotation values a t the D-line (5890 A . ) . In this case the anomalous rotatory behavior of certain of the amino acid constituents themselves namely cystine, can be interpreted as contributing (1) W. Moffitt, J . Chem. P h y s . , 25, 467 (1956). (2) W. M o m t t and J. T. Yang, Proc. Nall. Acad. Sci., 42, 596 (1956). (3) W. Moffitt. ibid., 42, 736 (1956). (4) P. D o t y and 5. T. Yang. J . A m . Chem. Soc., 7 8 , 498 (1956). ( 5 ) J. T. Yang and P. Doty, ibid., 79, 761 (1957). (6) P. D o t y and R. D. Lundberg. Proc. Null. Acad. Sci., 43, 213 (1957). (7) E. R. Blout and R. H. Karlson, J. A m . Chem. Soc., 80, 1259 (1958). (8) E. R. Blout and L. Stryer, Proc. Araf[. Acad. Sci., 46, 1591 (1959). (9) G . D. Fasman, hl. Idelson and E. R. Blout, J . A m . Chem. Soc., 83, 709 (1961). (10) J. E. Turner, R. T. Bottle and F. Haurowitz, ibid., 80, 4117 (1958).
to the over-all anomalous rotatory characteristic of this protein. Since the studies mentioned above were carried out in a solvent (S8yo formic acid) in which helix formation is not believed to be favored, the results might be misleading in the sense that the implication is that the oxidation of cystine bridges does not alter the helical content of the protein. Of course, one expects the helical content to be virtually zero either before or after oxidation of albumin in S8% HCOOH; however, a more rigid structure such as that found in insulin might exhibit significant changes in secondary order on cleavage of -S-S- bonds even in formic acid. The problem also arises regarding results of such oxidative procedures in solvents wherein the formation of a helix is favored. It is also possible that the manner in which the disulfide bonds are cleaved could have some effect on the optical rotatory dispersion, particularly if the reaction conditions could permit interaction of some groups other than -S-S-; if the hypothesis of Turner, et al., is supported, however, the manner of cleavage should not cause significant alteration of rotatory properties if the disul?de bond contributions and those of their conversion products are first subtracted (assuming no other changes in the molecule). In order to get a more complete picture of the optical rotatory changes which take place, the proteins in this study-bovine plasma albumin and insulin-were examined spectropolarimetrically; thus the continuous optical rotatory dispersion curves from 300 to 589 mp were obtained, rather than [Q]D values alone. Likewise to make some interpretation of the effect of the manner of cleavage on the rotatory changes seen, the -S-S- bonds of these proteins were cleaved reductively with sulfite as well as oxidatively with performic acid in separate experiments. The oxidation of -S-S- bonds with performic acid
May 20, 1962
SPECTROPOLARIMETRY OF BOVINEPLASMA ALBUMLV
is an irreversible procedure, but the sulfite cleavage is more or less reversible (depending on the environment of the bridge).’l For this reason, the possibility of re-formed protein or newly-formed bridges in the isolated reaction product could not be excluded. Amperometric titration, however, of the parent proteins and cleavage products after they had been isolated, lyophilized, and stored several months under refrigeration indicated disulfide contents of only 0.051 mole per mole of sulfite-cleaved insulin and no measurable residual disulfide content of sulfite-cleaved bovine plasma albumin. The same procedure12 yielded values of 3.10 moles of disulfide per mole of insulin and 17.1 moles per mole of bovine plasma albumin while their performic acid-oxidized counterparts both exhibited zero measurable disulfide contents. As a basis for the comparison of optical rotatory changes which have occurred, the residual optical rotatory dispersion curves of the native proteins and their oxidized and sulfite-cleaved counterparts have been obtained. These values were arrived a t by selecting specific rotation values from the observed dispersion curves a t each of twelve wave lengths and subtracting the apparent rotatory contribution of the S-containing amino acid a t that same wave length. The result was a 12-point 0.r.d. curve for that protein minus the contribution for any S-containing amino acid. I t may be somewhat naive to presume that the 0.r.d. contribution of the amino acid residue in the protein is the same as it is in the free amino acid state, but Wiirz and Haurowitz,13have shown this to be approximately true for cystine and its conversion to cysteic acid or S-sulfocysteine in a variety of proteins under varying conditions when rotation measurements were made a t the D-line only. Experimental T o help answer the question of solvent effects on the optical rotatory properties of the materials examined, two different extremes of conditions with respect to the hydrogen bonding relationships of solvent and solute were examined. On the one hand, 887, formic acid, in which hydrogen bonding of solvent molecules to solute would be expected t o be favored was used; on the other, a solvent composed mainly of 2-chloroethanol-known to favor a more ordered configuration and, presumably, intramolecular hydrogen-bonding of polypeptide solutesl*-was employed. It was found necessary to modify the latter solvent b y addition of small airiounts of trifluoroacetic acid and water in order to obtain sufficient solubility of some of the materials examined. The final composition of the solveut was 9Oc/‘, C1CH2CH20H, 57;)CF3COOH and 5%> HzO. I n this case, a compromise in conditions had to be accepted since greater order in the protcin configuration appears to be associated with decreased solubility-at least in solvents which appear to favor intratnolecular hydrogen bonding. It is probable that some interaction between solvent and solute occurs in these solvent systems. In any event, the comparison of parent protein and its oxidized and reduced forms is made under exactly the same solvent conditions for all three materials. The effect of such alteration of optical rotatory properties as may be attributed t o the reaction of the protein with the solvent should thus be minimized. A. Materials. 1. Proteins.-(a) Zinc-insulin [I(Zn)], lot no. 708850 (Eli Lilly and Co.), was used for nearly all studies. A few comparisons with other manufactured lots ._
(11) R. Cecil and J. R. hlcPhee, Biochem. J.,60, 496 (1955). (12) K. E. Benesch, H. A. Lardy and R. Benesch, J . Bioi. Chem.,
ale, 663 (1955).
(1.7) H. Wurz and F. Haurowitz, J . A m . Chem. Soc., 83,280 (1961).
(14) R. E. Weber and C. Tanford, ibid., 81,3255 (1959).
1897
were made to assure consistency of optical rotatory dispersion data from lot to lot. The zinc content of lot 708850 was 0.56y0 on a n anhydrous basis; moisture content was 3%. Cleavage experiments were conducted with this material. ( b ) Zinc-free insulin [I(Zn-free)], lot no. 683300 (Eli Lilly and Co.), was used as a reference material only. Moisture content was 4.7$& and Zn content was 0.0061%. ( c ) Oxidized insulin [OI] was prepared by performic acid oxidation according to Sanger.16 Zinc content of the lyophilized product was 0.48Yc. ( d ) Sulfite-cleaved insulin [SO3-I]: the method of Bailey and Colel6 was used to prepare this material. The lyophilized product contained only 0.0016% Zn. ( e ) Bovine plasma albumin [BPA]: Armour lot no. U68712 and T68204 were used interchangeably after satisfactory comparison of the optical rotatory dispersion curves of the original samples. Oxidized [OBPA] and sulfitecleaved [SOa-BPA] albumin preparations were made using essentially the same techniques as were employed for insulin; purification of OBPA involved repeated precipitation with 5 M S a c 1 rather than acetone, however. 2. Amino A c i d s . - ( a ) L-Cystine, Merck reagent; ( b ) cysteic acid, Eastman Kodak Co. (E93-18 spec. list); ( c ) S-sulfocystrine, gift of Prof. F. Haurowitz, synthesized (as Na Salt) by method of Clarke.” 3. Solvents.-(a) Formic acid 88-90%, Baker and Adamson, reagent; ( b ) 2-chloroethanol, Eastman Kodak Co., redistilled; (c) trifluoroacetic acid, Eastman Kodak
c o.
B. Apparatus.-Optical rotatory dispersion data were obtained using a model 260 Rudolph recording spectropolarimeter a t the “medium” sensitivity range. The reproducibility of the observed rotations is approximately f 0 . 0 1 ” on this position. Samples of the various amino acids and proteins were measured in their respective solvents at either 0.5 or 1.0% concentration in 10-cm. or 1-cm. cells. The shorter path lengths were used in the shorter wave length regions of rotation when the observed rotatory values were sufficiently high to avoid gross error factors. The light absorption of the proteins and of the solvents prevented accurate measurements below 300 mM in most cases. As the studies described here progressed, a great many solvents and solvent mixtures were examined and subsequently discarded; the criterion for acceptance was the fact t h a t the parent proteins, their oxidation products, their sulfitecleavage products, and the individual S-containing amino acids involved had to be soluble therein. Calculations.-As a method for evaluating the optical rotatory changes which have occurred under the different solvent conditions and independent of the S a m i n o acid contributions, both single-term DrudeI8 equations and hfoffitt3 equations of the “reduced” 0.r.d. curves have been evaluated. The one-term Drude equations [ a ] = k / ( X 2 XC2) were evaluated graphically for k and A, (intercept and slope, respectively) from plots of [a] vs. X 2 [a]. The more complex Moffitt equation
(where IZ = refractive index, I10 = average residue wt., [wz’] = effcctive residue rotation) was treated empirically as suggested by I)oty.’$ The evaluation was made first in the conventioiial inantier using XO = 2100 for all the data. This was done through a inodification of the graphical approach used by hloffitt and Yang2; in the present study, however, some of the tedious aspects of the graphical evaluation were avoided by using a computer program. Thus it was possible automatically to convert the twelve pairs of [ a ] 11s. X data to values for 1 / ( X 2 - X ~ 2 ) and [nz’] (X2-Xo2) and t o compute the slope and intercept for the best straight line through the data points. \Then the slope and intercept terms, bo Xo4 and no Xo2, were evaluated, values for a,, and bu-constants believed to be related to secondary structural characteristics and/or solvent-solute interaction effectswere obtained.
x.
(15) F. Sanger, Biochem.J.,44, 126 (1949). (16) J. L. Bailey and R. D. Cole, J . B i d . Chem., 234, 1733 (1959). (17) H.T.Clarke, ibid., 97, 235 (1932). (18) P. Drude, “ T h e Theory of Optics,” Leipzig, 1900; cf. T. hf. L u w r y , “Optical Rotatory Power,” Longman, Greens, London, 1937. (19) P. Doty, “Proceedings of t h e Fourth International Congress of Biochemistry. Vol. VIII, Proteins,” Pergamon Press, New York, N. Y.,1960, pp. 6-22.
VOl. 84
MAXM. MARSH
1898
TABLE I OPTICALROTATORY DISPERSIOSDATA( [ a ]) FAa
Sample
-
BPA OBPA SOa-BPA I (Zn) 1 (Zn-free)
5890
SO3-I L-Cystine L-Cysteic acid S-Sulfocysteine
82’ 68 62 -80 70 -50 - 58 -259 13 -84
Sample
FA
-
0-1
+
-
‘I
Solvent F A
-255’ -225 -232 -245
-232 -174 -195 -785 44 -232
+ =
28’ 16 13
-441
- 38 -19
- 24 -256
+ 64 -81
3800 BPA OBPA SOB-BPA I (Zn) I (Zn-free) 0-1 S03-I L-Cystine L-Cysteic acid S-Sulfocysteine
2Ca
FA
-
-
94’
82 84
94 85 G3
-
72 -316 14 92
+
-
5500
-
-
-304’ -290
+118
58 -2G9
-237
-276 -292 -277 -209 -232 -912
+
47 24
-
22 -300
+ 70 -100
FA
-139’ 82 98 -1G7 -1145 - 85 - 92 -744
-
FA
- 34’ - 18 - 16 - 51
3000
2C
2c
88% HCOOH; solvent 2C
=
48’ 24
- 20 - 71 - ti3 - 33 - 30 -386 t 78 -126
+
3400
-181’
-208 -182 -10G -118 -868 +138 -276
2C
-
-118’ -106 -110 -117 -113 - 83 - 93 -409 17 -123
2C
-108 -133
5000
F .4
-159’ -136 -144 -155 -145 -107 -121 -513 i 24 -102
4600
IN
A.
2c
-
($8’
34
-189’ -159
40 9G 85 - 44 - 46 -486 90 -150
-1G8 -180 -167 -125 -140 -589 f 29 -184
+
4250
2C
FA
2c
-366’ -330 -342 -349 -33G -251 -279 -1120 75 -326
-245’ -140 -189 -266 -234 -132 -162 -1026 4-164 -318
-451’ -390 -426 -427 -414 -310 -337 -1277 +IO1 -391
-343’ -220 -280 -370 -320 -190 -220 -1198
4000 2c
-
-
85’ 44 53
-113 99
-
54 5G
-556
+ 98 -178 3100
3200
FA
+
FA
FA
2c
- 4990
- 4200
FA
?C
-2190
-1120 - 62 - 74 -138 -121 - 69 - 72 -652
-191 -200 -212 -201 -147 -166 -692 37 -2ou
+
+lo4 -209 3000 F A 2C
-488 -457 -450 -300 -400 -1410
-5900 -550 -520 -500 -540 -410 -400 -1bOO
+I21
+2d0
+148
-324 -1520
+ZOO
-280 -342 --id0 -380 -230 -270 -1320
-3138
-399
-420
-470
-470
-460
-5400 -310
-435 -510 -470 -290
+278
90% ClCH2CHZOH, 5:c CF3COOH, 5 5 , H2O.
tion of equations representing only one type of secondary The computer program further afforded the opportunity structure yields inconsistencies is not surprising. t o verify easily the most effective linearization of the data in Since the comparison of optical rotatory constants in terms of XO. T h e Moffitt equations for a particular set .of 0.r.d. data were calculated for Xo values from 1800 to 2900 A . these experiments is relative to the constants of the parent at 100-A. intervals; the interaction standard deviation of protein, it is not necessary to assign quantitative significance the x and y parameters ( x = 1 / X 2 - An2) and y = [m’](X2 - to them. Thus, although bo values near -630 are indicative Xo2) was also automatically deduced and expressed as S 2 z . v ; of essentially helical structures in synthetic polypeptides and the best linear fit of the data was taken as t h a t value of XO in some proteins, the numerical values obtained here do not necessarily indicate a “per cent. helix” based on the fraction for which S2,,, was a minimum. It was found that for many obtained by dividing the bo for the particular protein by of the “reduced” 0.r.d. curves as well as for some of the un-630. The interpretation of results depends only on the corrected X O for the most linear plots of the parameters \:as not 2100 A . U’hen the equations containing Xo # 2100 order of magnitude and direction of change of the constants A . were solved for their constants, values of Ax,,and B X O ( L O , bo, XO, A x 0 and BXO.I t must be recognized that comparisons of different bo, Xo m t i B i o ternis is made only with rewere obtained (to distinguish them from no and bo since the gard to the similarity or lack oi similarity of behavior of identity of the latter with XO = 2100 A. has already been these terms as a result of the transformation o f the protein established). As timy be noted from thc data listed iii tlirougli the two different cleavage paths. Table I, the selection of wave lengths has been weighted to favor the ultraviolet region. In general, the precision of Results and Conclusions these ineasurerneiits is greater than the longer wave length Table I1 and I11 report constants for the hlofitt values; they more clearly reflect the approach to the region d anonlalous rotatory behavior. The procedure is much equation evaluations, both uncorrected and with the more analogous to the graphical method than would appear S-containing amino acid contribution subtracted ; to be the case if equal wave length intervals had been utithe former contains au and bo evaluations in the conlized. ventional manner while the latter shows results of The observation that MoRitt equation data may someselection of Xo to give the best linear fit. The values times to be more appropriatFly fitted to a linear plot with values of ho other than 2120 A . has been made by othersz0~21; of < l x o and Bxo in Table 111 coqespond to a0 and this fact has a significant bearing on the utilization of the bo bo when the A,, selected is 2100 A. Table IV conterm for expressing degree of order among helical molecules. tains constants derived for one-term Drude equaTJnder these circumstances, bo( XO = 2100 A , ) is not a con- tions of the curves of the same proteins. stant since the LMoffitt plot is not truly linear. On the There are certain general conclusions which can other hand, whil? RXOvalues derived from equations in be drawn from a comparison of these values: I . which XO # 2100 A . are constants, they cannot be simply related to each other unless the coefficients being compared The differences between the optical rotatory diswere obtained using the same Xo; thus for a given set of X persioii curves of bovine plasma albumin and its i’s. [fa’] values, the magnitude of Bxo is some function of oxidatively and reductively cleaved counterparts Xu. I n qualitative terms, the results suggest that some chrodo not appear to be completely accounted for by mophoric contributions are made by the secondary order or spatial asymmetry o f the molecule itself; changes in this the differences in S-containing amino acid contrisecondary order are seen as changes in the frequency terms bution-even in a solvent favoring the random cono f linear Moflitt and Drude equations. It is fairly obvious figuration of the peptide secondary structure. The froin recent discussions by Schellnian and SchellmannZ2that Bxo values in Table I11 show excellent agreement the optical rotator!- dispersion of proteins such as albumin and insulin can be expected to be a composite of the confor B P h and OBPA, but are obtained from two tribution of several structural types; the fact that evaluadifferent values ofoXo. Since both XO’Sare quite dif(20) A. R. Downie, A. Elliott, W. B. H a n b y a n d B K Llalcul~n, Proc. Royal Soc. ( L o a d o n ) , A242, 325 (1957). (21) P. J. Urnes, K . Imahori and P. Dot>’, Proc. S n t l . d c n d . .Cci., 47, 163.5 (1961) (22) J. A . Schellman and C. G. Schellmaun, J . I ’ O ~ ~ J Sci., U C ~49, 132
(19613.
ferent from 2100 A , , the bo’s are probably not too useful for comparison purposes, either. The Drude equation constants also indicate some differences among the preparations not accounted for by subtracting the S-containing amino acid contributions.
1899
SPECTROPOLARIMETRY OF BOVINEPLASMAALBUMIN
May 20, 1962
TABLE I1 MOFFITTEQUATION EVALUATIOX (A, = 2100 A,) c
Difference curves @-amino acids subtractedBovine plasma albumin Insulin A. Formic acid A. Formic acid a0 bo ao
BPiZ -392 -419 OBPA S03-BPA.2 -372 B.
- 66 -100 -108
Cbloroethanol
ao
-13 -67 +32
BPA OBPA S03-BPA
bo
-12 -85 -88 -81
I (Zn) -288 I (Zn-free) -229 -329 0-1 so3-I -263 B.
bo
-430 -277 -416
I
Chloroethanol a0 bo
I (Zn) - 29 I (Zn-free) - 4 0-1
-149 f 14
soa-I
Uncorrected curves Bovine plasma albumin Insulin A. Formic acid 4. Formic acid ao bo no
-492 -417 -422
BPA OBPA SOB-BPA B.
-54 -92 -98
Chloroethanol
-110 - 39 - 20
BPA OBPX SO1-BPX
B. bo
ao
-296 -264 -155 -246
I (Zn) I (Ztl-free) 0-1 s03-1
-396 -288 -404
bo
-500 -4245 -320 -365
+l5 -58 -69 -62
Chloroethanol ao
I (Znj -226 I (Zn-free) -201 0-1 -93 so3-I - 87
bo
-266 -234 -175 -2230
TABLE I11 MOFFITTEQUATIOX EVALVATION Difference curves (S-amino acids subtracted)Bovine plasma albumin Insulin A. Formic acid A. Formic acid i o , A . AXO BAO io,&. Aho BPA 2500 -272 +36 I (Zn) 2600 -175 0-BPA 2600 -266 +36 I IZn-free) 2600 -152 SOaBPA 2700 -213 +27 0-1 2600 -209 SOrI 2600 -1GY
,-----
B.
2-Chloroethanol h,A. A h -27 BPA 2200 2400 -75 0-BPA SOsBPA +32 2100
-
