FREDKARUSH
2714
Vol. 72
[ CONTRIBUTXON FROM THE SLOAN-KETTERING INSTITUTE FOR CANCER RESEARCH]
The Competitive Interaction of Organic Anions with Bovine Serum Albumin' BY FREDKARUSH? The accumulating evidence furnished by studies Chemicals Division of E. I. du Pont de Nemours and Company. It was the same sample employed in our earlier on the binding properties of serum albumins investigation3 where its analysis was reported. The dye, points to the conclusion that the intrinsic associa- #-( 2-hydroxy-5-methylphenylazo) -benzoic acid, prepared tion constants of the binding sites, for any partic- by coupling diazotized #-aminobenzoic acid to p-cresol, ular albumin, are not identical but may be spread was the same material described before.' Dialysis Method.-The binding of the dye was deterover a wide range of values, Such variations mined by equilibrium dialysis with 15-ml. volumes inside suggest that there exist considerable structural and outside, in the manner previously indicated.' All differences among the several combining regions the experiments were conducted at 25.0 * 0.1' in 0.05 M of the protein molecule. We have previously phosphate buffer, #H 7.0 with an initial protein conceninside the bag, of 3.00 X 10-6 M . This is based ~ h o w nfor , ~example, ~~ that the binding by bovine tration, on a molecular weight of 69,000 for bovine serum albumin.' albumin of two structurally dissimilar organic The effect of bound dodecyl sulfate on the binding of the anions, a long-chain alkyl sulfate and a car- dye was investigated by conducting a series of experiments boxylic azo dye, required in each instance the in which various predetermined amounts of the detergent were included in the albumin solutions. These quantities assumption of heterogeneity for its quantitative were selected so that the average number of detergent interpretation. molecules bound per protein molecule would be integral The heterogeneity of the binding sites of al- yalues. The range of these values was from 0 to 8, inclubumin raises an interesting problem which is sive. For the values 1to 6 , a 1% excess over the amount of detergent which was to be bound was used and for the particularly significant for the hypothesis of two higher values the necessary total concentrations were configurational adaptability which we have ad- estimated from the binding curve of dodecyl sulfate.8 In vanced to account for the distinctive binding each case controls were included in which dye was omitted. properties of serum albumin^.^ This is the prob- It was also established that the dialysis times used, ten'to twelve hours, were sufficient t o permit the detergent dislem of the relative heterogeneity of the binding tribution t o reach equilibrium. Duplicate tubes were sites with respect to the various complexing always used in the dialysis experiments. anions. Do the sites which bind one anion most Analytical Methods.-Dye concentrations were deterstrongly also bind other anions most effectively? mined by spectrophotometric measurement in the manner previously The analysis for dodecyl sulfate T h a t the binding sites are not completely specific was carried described.' out by a sensitive colorimetric method recently in their selection is clear from such studies as developed.8 This method permits detection of the deterthose of Klotz and c o - w ~ r k e r swho ~ ~ ~showed by gent at a minimum concentration of about 1 X 1 0 9 M. spectral means that various organic acids could In the presence of the dye a small correction must be made displace methyl orange and azosulfathiazole from due t o its interfering spectral absorption. bovine albumin. These studies were not deResults signed, however, to deal with the problem of The experimental results are given in Table I. relative heterogeneity. These include a correction for the casing adsorpIn the attempt to elucidate this problem we tion of the dye; no correction is necessary for the have investigated the competitive binding by detergent since i t is not adsorbed. Table I bovine serum albumin of sodium dodecyl sulfate also shows the values of r and r l c where r is the and the azo dye, p-(2-hydroxy-5-methyIphenyl- average number of moles of dye bound per mole azo)-benzoic acid. These anions were selected of protein a t the free dye equilibrium concenbecause they had previously been studied3i4 tration c. These quantities have been plotted and because the analysis of each anion in the in Fig. 1 for the moles of detergent bound per presence of the other is feasible. mole of protein ranging from 0 to 8. The curve for detergent absent is taken from earlier data.4 Experimental Materials.-The bovine serum albumin used in the Curves have been drawn through the points study was crystallized protein obtained from Armour and according to the best visual fit and extrapolated Company. Corrections for moisture and ash were made linearly to the r / c axis. I n each experiment as previously described.' The sodium dodecyl sulfate was analyses for dodecyl sulfate were carried out with a specially purified sample generously supplied by the Fine the outside solutions of the control, in which no (1) Presented in part before the Division of Biological Chemistry dye was used, and the samples containing the at the Atlantic City meeting of the American Chemical Society, largest amount of dye. These determinations September, 1949. demonstrated that the integral values for the (2) Investigation conducted during tenure of a Fellowship in Candetergent bound were correct in all cases to cer Research of the American Cancer Society, recommended by the Committee an Growth of the National Research Council; present within about 1%. Also, they showed that not address: The Neurological Institute, New York 32, N . Y . more than 1% of the bound detergent was dis(3) F. Karush and M. Sonenberg, THIS J O U R N A L , 71, 1369 (1949). placed by the bound dye. (4) F.Karush, i b i d . , 71, 2705 (1950). (5) I. M. Klotz, ibid., 68, 2299 (1946). (6) I. M. Klotz, H. Triwush and F. M. Walker, i b i d . , (1948).
TO, 2935
(7) E. J. Cohn, W. L. Hughes, Jr., and J. H. Weare, ibid., 69, 1753 (1947). (8) F. Karush and M. Sonenberg, Anal. Chcm., 12, 176 (1950).
June, 1950
COMPETITIVEINTERACTION OF ORGANIC ANIONSWITH BOVINESERUM ALBUMIN 2715
r for detergent = 8; initial detergent concn. 26.0 TABLE I 0.766 9.83 3.43 COMPETITIVE BINDINGOF DODECYLSULFATE AND DYE .560 6.89 2.37 BY BOVINE SERUM ALBUMIN AT 25' IN 0.05 M PHOSPHATE 4.92 1.70 .396 BUFFER,PH 7.0; INITIAL PROTEIN CONCN. 3.00 X M 1.04 .224 2.96 Initial dye Concn. of free concn. m./L dye(c) m./l. y/c 1.98 0.67 .169 x 105 x 106 I x 10-4 r for detergent = 1; initial detergent concn. 3.03 X
17.80 11.80 7.88 5.94 3.95 17.73 11.80 7.86 5.90 3.96
3.10 1.73 1.00 0.72 0.45 3.01 1.71 1.01 0.71 0.44
3.66 2.67 1.89 1.45 0.984 3.70 2.68 1.88 1.445 0.997
A!f
11.8 15.4 18.9 20.15 21.9 12.30 15.7 18.7 28.4 22.6
X M 2.24 2.36 2.33 2.16 2.52
Discussion Theoretical Considerations.-Before we attempt an interpretation of the data in terms of relative heterogeneity we wish t o consider theoretically the problem of competitive binding and derive some useful relations. We shall designate the component whose binding is being measured (the dye) by A and the competitive inhibitor (the detergent) by B. We assume that
11.1 r for detergent = 2; initial detergent concn. 6.06 X 9.72 3.48 3.38 17.80 12.94 1.94 2.51 11.80 15.82 1.13 1.79 7.88 1.41 18.1 5.94 0.78 0,964 20.1 3.95 0.48 12.66 1.96 2.48 11.76 1.78 15.7 7.85 1.13 18.2 0.77 1.40 5.90 20.1 0.48 0.966 3.95 r for detergent = 3; initial detergent concn. 9.09 X 7.80 3.93 3.06 17.82 2.29 10.22 11.78 2.24 12.6 1.32 1.67 7.90 1.313 14.4 0.91 5.95 15.8 0.57 0.901 3.96 c
for detergent = 4; 17.82 11.78 7.90 5.95 3.96
r for detergent -5;
14.75 9.85 6.90 4.92 2.98
Af
initial detergent concn. 12.12X lO-'M 4.42 2.70 6.11 2.64 1.99 7.54 1.57 1.48 9.43 1.160 10.35 1.12 0.800 11.3 0.71 initial detergent concn. 15.15X10-6 M 1.97 4.89 4.03 2.51 1.44 5.74 1.097 6.70 1.64 1.13 0.810 7.16 0.64 0.524 8.2
24
-. 20 I
2 X
< 16 *.
12
8
r for detergent=6; initial detergent concn. 18.18X10-5M
14.75 9.85 8.90 4.92 2.98
4.51 2.90 1.95 1.36 0.79
1.62 1.16 0.870 .644 .413
3.59 4.00 4.46 4.74 5.23
r for detergent = 7; initial detergent concn. 22.2 X 10-6 M 9.83 3.20 0.935 2.92 .684 3.11 6.89 2.20 4.92 1.54 .511 3.32 2.96 0.94 .298 3.17 1.98 0.63 .I98 3.14
0
2
1
3
4
r. Fig. 1.-The effect of bound dodecyl sulfate on the binding by bovine serum albumin of an anionic azo dye at 25' in 0.05 M phosphate buffer, PH 7.0. The average number of detergent anions bound per molecule of protein i s indicated at;he right of the &es.
FREDKARUSH
2716
there are n moles of binding sites per mole of protein characterized by n values of the intrinsic association constant K, all of which may be different. We neglect any electrostatic interaction of the bound ions and assume the absence of any other interaction. If ai is the fraction of the i t h sites occupied a t a particular concentration by the appropriately indicated component, then i t follows readily from the law of mass action that
VOl. 72
geneity in the binding of A and B. We define identical heterogeneity to mean that Kf/KB is the same for all i, which ratio is indicated by d. Since A(A) is determined in the absence of inhibitor and A(B) is found under the condition CA + 0, then at of (9) and a? of (10) are given by
Under the condition that r.4 = that
rB
= r, i t follows
and
This can be put in the form
which on substituting dKp for Kf gives (1
-af)c,d~y= r ]
In the absence of inhibitor (4)
and
Since the K B ’ s are independent then the corresponding coefficients must be equal
If inhibitor is present, then
By virtue of the fact that we have assumed E a ?
CaB, i t follows that
=
i
CB/CAd
= 1 and a? = a?.
i
Under these circumstances, i. e., when 1
+ cBK? = 1/(1 - a?)
which on substitution in (6) yields
CA
=0
(7)
A(A) = A(B) (for all
’
This expression on replacing a: with a: gives the equation for self-competition and becomes equivalent to (4). I n this case the meaning of rA/CA is easily seen if one imagines the experiment in which is added to the protein solution, containing a particular concentration of free dye, an additional known amount of chemically identical dye, but experimentally distinguishable by virtue of the inclusion of a radioactive isotope. The value of rA/CA given by (8) would be the limiting value of r/c for the radioactive dye as its concentration approached zero. It is convenient to define two experimentally determinable quantities A(A) and A(B) which give parallel measures of self-competition and non-self-competition, respectively.
A(B) =
(:)o -
(z)B,o
= CaYKf
In view of equations (9) and (10) we have therefore the experimental criterion for identical heterogeneity that
(Fa:
E
‘B)
(10)
We now inquire as to the relation between A(A) and A(B) when there is identical hetero-
rA e
rg)
(15)
In the case that the binding of either constituent is homogeneous simple relations are obtained. If the binding of A is homogeneous, which means that the Kf’s are identical, i t follows from (10) that A(B) = K* E a ?
KAyg
(16)
i
For the homogeneous binding of B, a: = On substitution in (lo), this gives A(B) = 9 n
YB/n.
Kf i
Thus, in both instances, i t would be observed that successive unit increases in r B would be accompanied by equal successive increments of A(B). It is worth noting that if A(B) does not change in this way with r B , i t can be inferred that the binding of both constituents is heterogeneous. Relative Heterogeneity-In order to evaluate our data in terms of the criteria developed above, we have listed in Table I1 the competition differentials A(A) and A(B) and their increments. For the case of self-competition we have used the dye binding data previously ~ b t a i n e d . ~The quantities (rA/CA)B,o are, 0: course, found by the extrapolation of the dye binding curves obtained
June, 1950
COMPETITIVE INTERACTION OF ORGANIC ANIONSWITH BOVINESERUM ALBUMIN 2717
these sites for complexing with detergent relative to any other sites (including Group 2) which also bind detergent is greater than their superiority relative to the Group 2 sites in the binding of the dye. This permits the explanation of the fact that the effect on dye binding of only seven bound detergent anions is equivalent to the effect of more than 11 bound dye anions. To put i t another TABLE I1 way, i t accounts for the fact, already pointed out, THECOMPETITIVE EFFECT O F DODECYL SULFATE ON T H E that the A(B) increments exceed the A(A) increBINDINGOF A DYEAT 25' IN 0.05 M PHOSPHATE BUFFER, ments in the r range of 3 to 7. PH 7.0. A COMPARISON OF SELF-COMPETITION AND DEThese results, on the whole, serve to support the TERGENT COMPETITION IN TERMS OF THE COMPETITION idea that the binding properties of serum albuDIFFERESTIALS,A(A) A N D A(B), AND THEIRINCREMENTS mins are associated with the configurational adaptSelf competition Detergent competition ability of a number of regions of the protein4 A(B), The fact that the same sites (Group 1) bind most YA/CS A(A) A ( A L 1 (YA/CA)B,O A(B) A(B),-~ I x 10-4 x 1 0 - 4 x 1 0 - 4 x io-' x 10-4 x 10-4 strongly these two structurally different anions, 0 32.0 0 ... 32.0 0 ... suggests that these sites have a high degree of 1 26.8 5.2 5.2 26.2 5.8 5.8 structural adaptability, i. e., they can assume 2 21.6 10.4 5.2 24.6 7.6 1.8 structures which are complementary to a wide 3 16.9 15.1 4.7 19.0 13.0 5.4 range of configurations. Thus, i t would follow 4 13.2 18.8 3.7 13.5 18.5 4.5 that i t is this property which confers upon the al5 10.3 21.7 2.9 9.5 22.5 4.0 bumins the distinctive ability to form complexes 6 8.1 23.9 2.2 5.8 26.2 3.7 with a wide variety of anions. From this point of 7 6.5 25.5 1.6 3.3 28.7 2.5 view, the Group 2 sites could be interpreted as 8 5.4 26.6 1.1 2.4 29.6 0.9 being more restricted in their range of adaptabil9 4.6 27.4 0.8 ity with, consequently, smaller binding constants 10 4.0 28.0 .6 and therefore more selective in the anionic struc11 3.5 28.5 .5 tures with which they associate. 12 3.1 28.9 .4 Since the binding of dye and detergent anions 13 2.7 29.3 .4 probably involves a displacement of buffer an14 2.4 29.6 .3 i o n ~ it, ~is unclear what net contribution, if any, is made to the competition differentials by the From Table I1 several interesting inferences can electrostatic effect of these bound anions. In any be made about the relative heterogeneity in the case, our conclusions would not be affected by the binding of the dye and detergent. I n the first inclusion of an electrostatic correction. Thus, for place, the inequality in the increments of A(B) example, the maximum effect on the dye binding verifies the conclusion previously reached that the of 7 bound detergent anions was calculated on the binding of both dye and detergent is heterogene- assumption that these decrease the net charge on ous. Secondly, the lack of agreement of A(A) and the protein molecule by 7. The calculation was A(B) demonstrates that this heterogeneity is not carried out in a manner similar to that employed identical for these two anions. Thirdly, i t will be by Scatchard and Black.'* If the reduction in the noted that for low values of r , A(A) is larger than number of available sites is taken into account, a the corresponding values of A(B). For r = 6 and value of 14.1 X lo4is obtained for (IA/CA)B,O comlarger, however, A(B) exceeds A(A). This means, pared to the experimental value of 3.3 x 10'. in general terms, that for small r self-competition This indicates that there is heterogeneity in the is more effective than detergent competition. At dye binding sites and that the detergent binds high r values, the reverse is true. preferentially to the Group I sites. Otherwise the For a more detailed interpretation of this re- experimental value would not be less than 14.1 x versal let us recall that the dye binding sites can 104. be divided into two groups: Group 1 contains beWith the knowledge now available of the relatween four and five sites and is characterized by a tive heterogeneity of the binding sites of bovine high binding constant and Group 2 with about serum albumin i t is possible to account for some seventeen sites and a relatively low binding con- interesting but unexplained observations of Klotz, ~ t a n t . We ~ also note that the A(B) increments Triwush and Walker.6 They attempted to.deterexceed those of A(A) for Y ranging from 3 to 7. mine the binding constant of bovine albumin for The picture emerges, then, that there is a t least sodium dodecyl sulfate by its competitive disone site, and perhaps two, which binds the deter- placement of methyl orange. This was measured gent strongly. This site (or sites) is not a member by the spectral shift associated with the binding of Group 1, but may belong to Group 2. Aside of the dye. On the assumption that the binding, from this site, the Group 1 sites are not only most G. Longsworth and C. F. Jacobsen, J . Pkys. Colloid Chcm., effective for the dye but also bind the dodecyl sul- 68,(9)126L.(1949). fate most strongly. In fact, the superiority of (10) G. S c a t c h a r d and E. S. Black, THISJOURRAL, 68, 88 (1949). with the several values of r ~ .For low values of r B significant error could be introduced if the linear extrapolation were not valid. For T B = 1, in particular, there is considerable uncertainty about the extrapolation and we shall therefore, in our main discussion, not include reference to the corresponding value of A(B).
2718
EDWARD L. ENGELHARDT, FRANK S. CROSSLEY AND
both of the dye and detergent, is homogeneous i t was calculated from the experimental results that more detergent was bound than was added. The explanation is, undoubtedly, that there is heterogeneity in the binding of both anions and that the same sites are most effective in both instances. Under these circumstances the detergent is a more effective competitor than the homogeneity assumption would predict. I n fact, if, as is probably the case, i t is the Group 1 sites which are involved, the experimental observations are readily accounted for. In view of the heterogeneity in the binding of anions, the interpretation of the binding of a substance by its displacing effect is subject to considerable limitation. Depending on the relative heterogeneity, the binding of the competitor may be underestimated or even overlooked entirely. When compared to data obtained by other methods, e. g., electrophoretic mobility, considerable disagreement in the results may appear. Acknowledgment.-I am indebted to Dr. C. P. Rhoads for making available to me, during the conduct of this investigation, the excellent laboratory facilities of the Sloan-Kettering Institute for Cancer Research. This work was supported by the Office of Naval Research.
[CONTRIBUTION FROM
THE
JAMES
M. SPRAGUE
Vol. 72
summary The effect of bound sodium dodecyl sulfate on the binding by bovine serum albumin of an anionic azo dye has been investigated a t 25' in 0.05 M phosphate buffer, p H 7.0. The amount of bound detergent was varied from 1 mole per mole of protein to 8 moles per mole protein. The data are analyzed in terms of a comparison of self-competition versus detergent competition. For this purpose there are introduced the quantities A(A) and A(B) which are named competition differentials and whose values can be derived from the experimental data. It is shown that the criterion for identical heterogeneity is that A(A) = A(B) for ?A = ?E$. A comparison of A(A) and A(B) over the whole range of Y A and r B studied leads to the conclusion that those sites (Group 1) which bind the dye most strongly also, for the most part, bind the detergent most effectively. I n terms of configurational adaptability the Group 1 sites are interpreted as being able t o assume structures complementary to a wide range of configurations whereas the other sites are more restricted in this respect. The latter bind less strongly and are more selective. NEWYORK,N. Y.
RECEIVED NOVEMBER 7, 1949
DEPARTMENT OF ORGANIC CHEMISTRY, MEDICAL RESEARCH DIVISION,SHARPAND DOHME,
INC.]
The Reductive Alkylation of Arylalkanolamines BY EDWARD L. ENGELHARDT, FRANK S. CROSSLEY AND JAMES M. SPRAGUE Cope and Hancock' have described a con- nature of the anhydro compounds formed from venient method for the preparation of N-alkyl norephedrine (I, R1 = phenyl) has not been derivatives of alkanolamines by the catalytic studied, the steric effect of the phenyl group reduction of mixtures of the alkanolamine and an might be expected to favor the alkylidene-aminoaldehyde or ketone. As a part of a pharma- alcohol (IIb). cological study of N-alkyl derivaR&C=O + RI-CH-CH-CHI tives of phenethanolamine we have RI-CHOH-CH-CHs I I NH I extended the method of Cope and 0 NHz Hancock to the alkylation of 1I IIa \e' phenyl-2-amino-1-propanol and cer/\ tain nuclear substituted derivatives. Rz Ra With aliphatic aminoalcohols, If CoDe and Hancock have shown that YI H2 the' structure of the anhydro coma R1-CHOH-CH-CH~ pounds (11) that are thought to be R1-cHoH- -CH-CHs I (catalytic) I intermediates in the reaction, deNHCH&RI N=CRA pends upon steric factors in the carIIb bonyl compound. Aldehydes and The only failures encountered in the norunhindered ketones give oxazolidines (IIa) while highly branched ketones give alkylidene-amino- ephedrine series (Table I) occurred when acetalalcohols (IIb). The alkylaminoalkanol results dehyde or acetophenone was employed as the either from reduction of the alkylidene inter- carbonyl component. The former was polymediate (Ira) or from hydrogenolysis of the oxazol- merized, the latter apparently failed to condense idine (IIb). Although in the present work the with the aminoalcohol. 3,4-Dihydroxynorephedrine gave good yields of the expected products (1) Cope and Hancock, THIS JOURNAL, 64, 1503 (1942); 66, 1453 when ketones were employed as the carbonyl (1944); Hancock and Cope, i b i d . , 66, 1738 (1944).
+
L
