100
I R V I N G M. KLOTZ AND JEAN M. URQUHART
T H E B I N D I S G O F ORGANIC ION3 BY PROTEIXS. B l T F E R EFFECTS's2
INTRODUCTION
The ability of crystallized proteins, particularly serum albumin, to 1)iiicl numerous organic and inorganic ions has been demonstrated repeatedly by sesreral types of invest'igation ( 5 , 7 , 9 , 12, 18). From this \vorlc it has become increasingly evident that protein molecules, and those of serum albumin in particular, do not exist in simple uncombined form in common electrolyte solutions, but rather as complexes with these small ions. These complexes may differ appreciably in physicochemical propert,ies from t'hose to be anticipated for the simple protein molecule. Electrophoretic memurements are particularly sensit'ive to associations betiyeen the protein and small ions in solut'ion. It has been realized, in fact, for a long time that protein mobilit'ies depend on the nature of the buffer as well as on the ionic strength and pH of the solution (1, 3, G , 17, 19). Davis and Cohn (6), furthermore, have attempt'ed to analyze the measurements of mobility in buffer solutions of varying ionic strength in terms of equations designed to account for electrostatic interactions between the different species. However, they have found that the mobilities calculated by extrapolation to infinite dilution are not independent of the buffer; hence they have suggested tJhat ion-protein association may be a contributing factlor to the observed mobility. It has seemed appropriate, therefore, to investigate the extent of complex formation between prot'eins and several common buffer ions both for its importance in the interpretation of electrophoretic mobilities and for its general significance in studies of protein complexes. Several alternative procedures are available by \yhich one might obtain this information. Since the primary interest in this Laboratory is in competition between the buffer and certain reference anions, the relative degrees of binding of different buffers have been evaluated by an examination of their relative competith-e abilities in the clisplacement of dye ions from their protein compleses. EXPERIMENTAL
The extent of binding of the dye anions (methyl orange and azosulfatliiazole, respectively) \vas measured by t'he differential dialysis technique described 1 Presented nt tlie T\veiity-secoiicl National Colloid Syrnposiuiii, \\hicli \VAS held under tllc nuspices of t h e Division of Colloid Cheniistr,y of the Aniericnn Chemical Society a t C.Linl)iidgr, llassnchusetts. Julie 23-25. 1048. 2 This inves~igntionu w s u p p o i ~ t c dIiy grniitn froin tlie Rockefeller Foundat ion and froin
the Office of Naval Research.
101
BINDING O F O R G h N I C IOKS BY P R O T E I N S
previously (13). The experiments lvere carried out. with mechanical shaliing for an 18.h . period. in an ice bath at 0.0"C.
ELECTROLYTE
I
.
IPH
NOLARITT
IONIC STRENGTH
5.00
0.1000 0.0555
0.1000
5.00
0.0500 0.0250
0.1000
S n C 2 H0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . HC.H,OL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
l
Pot:issiuiii acid plithnlate . . . . . . . . . . . . . . . . . . . . . . . . . SnOH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
NnC1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
'
5.01*
0.099G
0 . OODG
SaCl. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
5.72t
0.1086
0.1086
Citric acid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SaOH.,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
(
5.73
0.0233 0.0565
0 .009g
Nnh-08. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
'i
5.73$
0.1000
0.1000
6.63
0.099G
0.0990
6.53
0.0349 0.027G
0.1323
6.9
0 . 0139 0.0110
0.0527
0.06G3 0.0400
0.0400
NnCl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
~
1
SnrHPOJ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IiH2POr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i Xa?HPOr. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . RH1PO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
'1
S'eronnl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NnOH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1
S.80
0.100
0.100
Glycine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 NaOH.,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i
9.0s
0.4000 0.1000
0.1000
10.5
0.0345 0.0036
0.1000
S R H C O ~...,. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Na2c.03
..........................................
HCl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
i
I i
* p H oht:iiiietl by :idding -4.65 cc . of 0.0025 hydrochloi~ic:icitl to 500 c c . of solutioii . t pH obtniiietl by ndtling 2.50 cc . of 0.0025 -1' liytlrochlorio iicitl to 500 cc . of solution . 2 pH obtniiietl by :itltling -4.03cc . of 0.0026 .V hytlrocliloric :wit1 to 500 cc . of solutioti .
Analyses for the dyes were made with the Beckman spect.rophotomet.er. Thc wave length of maximum absorption (9) \IW used for. this purpose . Buffers v.ere prepni.ed from reagent'-gtade materials. except for veronal. \vhich was U.S.P. grade. and glycine. \vhich ]\.as a recrystallized Eastman product . Tlic specific composit.ion. ionic strength. and pH of each IxifTer are listed in table 1
102
IRVING 31. ICLOTZ .iKD JEAN hl. URQUH.4RT
The methyl orange and azosnlfathinzole were very pure samples described previously (9). The bovine serum nlliumin \Y:LS n crystallized sample from Armour and Company. I t s water content was determined by drying a separate portion in an oven at 110°C. Except when specifically noted to the conti-ary, binding measurements were carried out with a 0.2 per cent solution of the a l h m i n . P-Lactoglobalin, fotir times recrystallized, was generously supplied by Dr. T. L. RlcMeekin. Its concentration in solution was determined by its absorption of ultraviolet light, calibration being made by comparison with a sample which had been dried to constant \\.eight a t 110°C. Here again, R 0.2 per cent protein solution was used for the binding measurements.
L
.-
GLY. - - - - --- 7-0
Ftios.
.5
r-0
-
-
AC.
PHTH.
/
\
BICARB.
'1
CIT
0
CAAB
0
CHLORIDE
VER.
0
\ \ NAOH
-
RESULTS llND DISCUSSION
E$'ect of changes i r i p H Before one can consider specific buffer effects, it is necessary to exaniiiie changes which might be brought about by gross variations in hydrogen-ion concentration. For this purpose the degrees of binding of methyl orange a t 1 X 114 concentration of free dye by n 0.2 per cent albumin solution are plotted in figure 1. It is apparent first that the binding decreases to zero a t pH's near 12. This behavior confirms that observed earlier (10) by a spectrophotometric technique, where it \vas found that the changes in dye spectrum caused by albumin were not procluced a t pIi 12. The absence of complex formation in these basic solutions supports the Tie\\. that quaternary nitrogen :ttoms of th4 lysine (and perhaps arginine) residues in the protei11 are the foci of attachment of the anions. At pH's belo\\. 9 (figure 1) there is clearly no consistent pultern, the degree of binding being primarily z: function of tlic nnture of the buffer. -it first glance one might be inclined to conclutlr tlint there is :I gradual decretzse in extent, of
binding as t'he pH drops from 9 to 3 , such :is might, be indicated by the lower broken line in figure 1. Such a trend seems nnlikely, however, in view of the following obseivntions. The degree of binding \vas measrired in two separate sohitions of socliiitn chloride, each a t 0.1 ionic strength but#a t pH's 5.i and 6.6, respectively. This part'icular interval of pH \\w chosen since i t corresponds to that in \vhich n very sribstanhial rise in Iinding is ohser.iwl if one compares cit>rate and phosphate buffers in figure I. Xevertheless, in tJlie chloride solutions there is no significant difference in the binding of met>hylorange 11y albumin (figure 2). Evidently then the dm\ving of the lower broken line in figure 1, 11,san indication of the effect)of pH, is not warranted.
2 -
1
It has Leer1 :issunled, therefore, for t81iepurposes of par1 of the following discussion that the pH ixhavior is that, indicated by t'he iiorixo,zia.l broken line in figure 1. Tlie height a t \\,hiell this line is dra\vn implies that glycinate does not interfere a t all \\.it11 tlie lincling of methyl orange by albumin. While it is difficult to demonstrate t'liis vunclrisively by the dialysis technique, it is pertinent, t o mention that n fen. ine:tsuremeiits in 2.5 molal glycine in phosphate buffer at p H 7 slio\vecl increased binding of methyl orange by albumin, rather than t,he decrp;i.st: uiie \\.uiild cspect from cffec,ts of cunipciition. In terms of t,he variables in figlire 1 , the ordinate \viirild be npprosirnately 2. Tlinl glyciriatc sliould be hound less by albumin than is acetat.e also seems i,ensonahle from stiwcturiil considerations, since the a-amino group in the former tl it mure hydrophilic; hence the protein-binding constant should anion ~ v o ~ i lIiiake be decreased. These considerations lend one, therefore, to expect a minimum of int,erferenc.e \ v i t h methyl orange-alhiniin complexes in buffer solutions containing only glyc,inat,e.
Spcczjic
a)iioii cflccfs
(1) l l e t h y l orange as reference anion
In several systems it has been possible to examine the effects of lwious liuffer anions a t a fixed pH so that assumptions in connection with variations in pH need not enter. d n example is illustrated in figure 3. I t is quite apparent that phthalate anions are more active competing agents than acetate ions and hence that the former nre more strongly bound by albumin. Such behavior \voultl be expected, since the molecular \wight of plithnlate is snbstantially greater than that of ncetnte.
..
LOG t DYE JFREE FIG.3 . Compai~ieonof hiticling of methyl orange by bovine albumin in buffers a t pH 5.0. 0 , acetate; 0, phthalate.
Citrate, on the other hand (figure A), is about equal to acetate, at equivalent ionic strength, in its ability to form a complex with bovine albumin. Such behavior also is not surprising, for despite its larger molecular weight, the large number of polar groups and high average charge (2-3) on citrate ions at pH 5.7 would tend to decrease the extent of binding by the protein. On the other hand, the curve for the binding of methyl orange in the presence of chloride a t pH 5.7 (figure 4) might not have been predicted. It is true that evidence for the binding of chloride by bovine albumin is available from the osmotic experiments of Scatchard, Batchelder, and Brown (1G). Severt helesr, measurements of the stability of human albumin (4) indicate that acetate has a greater protective action than chloride against heat denaturation, a result which might he interpreted to mean that chloride is bound less strongly. Perhaps even more surprising than the behavior of chloride ions is the strong displacing ability of nitrate (figure 4). This anion is sitbstnntially in the same class as phthalate in its ability to form a comples with albumin, despite the much
BISDIKG O F OHG.\SIC' IOXS B Y P R O T E I N S
105
smuller molecular \veiglit of the inorganic iinion. In this connection it is perhaps of interest' to note tlint' the order of binding by albumin, Nos- > C1- >
CH7COO-, i q knilar to that of certain physiological effect5 of these anions (8). In fact, the ability t o forin complexes 11 itli albumin parallels the Hofmeister series in ninny reqpects, a b has heen pointed out by Scntchaiad (E), even though thr
106
I R V I N G AI. Iy tmviiic ;ilbumin in various buffers. 0 , glycinate, pH 9.1;O , bivarlwiiatv5 pll S.6; 0 , vvrnnal, pH 8.0; 0 , carbonate, pH 10.5.
6. Xost important perhaps is the observation that veronal buffers produce a very pronounced decrease in the binding of methyl orange by albumin. Thus the Verona1 anions in themselves niust be bound very strongly by the protein. Since this buffer is used so commonly in biochemical studies, it is essential to keep in mind that it exerts substantial effects on the properties of protein (and presumably enzyme) molecules. (2) Aeosulfathiazole as reference anion I n order to examine the possible influence of the charge of the reference anion on the buffer competitions, a few experiments have been carried out with azosulfathiazole, an anion which is divalent up t o about pH 7. The pertinent data are summarized in figure 7. Acetate and chloride ions have preserved their same yelathe positions. On the other hand, it is of interest to note that citrate acts as a much more potent competitor in the binding of divalent azosulfathiazolc
BINDING O F ORGllNIC IONS BY PROTEINS
:lo7
than of monovalent methyl orange. This difference is probably due to the increased electrostatic repulsion between bound multivalent citrate and divalont :izosulfathiazole, as compared to monol-dent methyl orange.
E#ffectof ioiiic slrength Several attempts ha\ e been made to obtain information on the extent of binding in buffer-free solutions as a basis for coinparison with solutions of varying ionic strength. Very discordant results have been obtained, however, by the dialysis method. 111a typical experiment with azosulfathiazole and albumin, for example, no significant binding was detected (table 2), yet it is quite obvious
from the change in spectrum (figure 8) under equivalent conditions that o t ~ 90 r per cent of tJhedye is in the protein comples. I t is iiecessary, of course. t o innlic suit'able corrections for the Donnaii effect in the dialysis experiments I11 i)iilf'erfree solutions. Reliable estimates of this correction could not be rni.ried out, hon-ever., because of inconsistent values of the pH readings in sirccessive experiments. Similar difficxltieu in salt-free solr1t)ionsof protein hn7.c 1)cenencountered by others (2, 10) and seem t o be tliie t o traces of acid 01' allcdi dissolved from t'h.e surface of the glass app:watus. Since the veronsl system \\xs inT-estigntccl nt, 0.0-4 p , \\-hereas all of t'lia other buffers were examined near 0.1 p, it seeinec! tlcsii,nble to make certain that no proiiounced ionic strength effects occur in this region. F o r this purpose, therefore, the binding of methyl orange by n1l)iiniin in phosphate buffer a t pH 6.8 was examined a t 0.05 ionic st,rength also. The results obtained we compared \vit:h
108
IRVING M. KLOTZ h K D J E S N M. URQUHART
the previous ones near 0.1 p in figure 9. It is apparent that there is n o significant difference between the two. A rough theoretical calculation of the Donnan effect at p H G.8 and 0.04 p , based on a valence of about 10 for the albumin obTABLE 2 Apparent absence of binding of azosulfathathiazole in biiffer-frec solidion pH 5.1-5.6; 0°C.; 0.189 per cent albuniin
---. -.
COKCENTRATIOY I N T U B E \ \ I T A P R O T E I N
-.
- - .- .
CONCENTRATION 111 CONTROL T U B E
M
dl
0.50 X 1.16 1.30 2.61 3.1B 5.18 6.24 12.35
0.49 X 10-5 1.1s 1.32 2.63 3.24 5.12 6.27 12.50
-
.?I
g.25
z W
n -I Q
Y
t .lo
> 5000
5500
4500
WAVELENGTH I N A .
FIG.S. hbsorption spcrtra of axosulfathiaxole i n buffer-free solutions. -4,in wat,er, pH 5.2; B, in 0.2 per rent bovine albumin, pH 5.0.
tained from a graph of the data of Scatchard, Batchelder, and Bronn (IG), indicates a correction of only slightly more than 1 per cent. Thus no significant errors are introduced from this source. Experiment and theoretical calculation thus indicate that the low binding in 0.04 p veronal solutions is not due to an appreciable Donnan effect but is actually an espression of the relatively strong affinity of this buffer for the protein.
BINDING O F ORGAXIC IONS BY PROTEIXS
109
Eflect of prolein concentration In all past calculations of binding, it has been assumed iniplicitly that the concentration of protein has no significant effect on the binding constant. It has seemed appi'opriate to examine this assumption experimentally, particularly because of its possible impoi*tancein connection with the interpretation of electrophoretic measurements a t different protein concentrations. The results otitained in a phosphate buffer of 0.1 ionic strength at two protein concentrations, 0.2 and 1 per cent, respectively, are summarized in figure 10. It is appni'ent that there is :I I-ery substantial decrease in complex formation \vith methyl orange when the
albumin concentration is increased by a factor of 5. Evidently the activity coefficients of the species PA, - 1, where P represents the protein and X the methyl orange anion, must be decreased more than those of PA, \\.hen the albumin concentration is increased.
E-fect of fype of protein As has been pointed out previously (9) not all proteins give indications of forming complexes with anions. In particular, earlier ivork has demonstrated that bovine serum yglobulin binds 'every few anions (11) under conditions where bovine albumin forms complexes extensively. I n the present investigation these studies have been extended to @-lactoglobulin. The binding of methyl orange by this protein, compared to bovine albumin, is illustrated in figure 11. Again the albumin stands out, h i t not by so large a margin. In fact, if the comparison
110
IRVING M. IiLOTZ .4ND JE4N M. URQUHART
LOG C DYE 1 FREE
BINDING OF O R G d S I C I O N S BY PROTEINS
111
iwre iniLde on a grain basis instead of it mole basis, one should use tlic dotted line (figuit 11) for the lactoglobulin, and it becomes apparent that this protein will bind methyl orange t'o about' one-t'hirtl t,lie extent, obtained with a l h n i i n under similar conditions. There is as yet no clear indication of the properties of the protein molcculc u.hich favor binding of anions. While it is t'riie t,hut both Iiovine albumin ant1 @-lactoglol~ulincontain :L large iirimber of amino acid residues \\.ith qiinternary nitrogen atoms, the same is true of bovine yglobulin, and yet it shoivs no appreciable binding properties. It' i\.ould appear thatJ the relative position of the qiiat'ernarg nitrogen in the protein structure is a ci.itical factor. The nat,ure of t8histopological factor remains to be \vorlalloci on protein available for binding anions then in the absence of elect8rost8nticint'eract'ions bet8ween successively t)ound mi ons :
It is evident that in the limit of zero concentration of dye ion, equation 1 reduces t,o
For the purposes of the present discussioii it is assumed that there is no competition between glycinate ions and methyl orange or, in other \\-ords,that li, = 0 for glycinate. I n addition it is postulated that rL and lit undergo no change in the pH region 5-0 under consideration. Implicitly this assumes that in this p H region the protein uncleigoes no modification irhicli \\.oultl affect the binding properties. On this basis it is possible to e\.aluate the product ( t ) / ; J by an extrapolation to zero (A) of the binding data in glycinate buffer in the graph of ,*,/(A) us. (A). For any other buffer solution the limiting value of ?.%/(A)may be obtained in n similar fashion, m d the ratio of limits may be used t o evaliiate Ihalbumin. Changes in pH, a t constant' ionic strength, do not seem to affect the binding of methyl orange b y albumin until about p H 0. I n more basic solutions binding i \ l l nltri.nntive view, suggested by Professoi, Scatchard, is that OH- ions, being i n high concenti~atiou:it high pH, act as strong competitive anions.
114
ROBERT
.\.
-4LBERl'Y
drops rapidly and reaches zero a t p H 12. The degree of hinding at fixed pH falls with increase in protein concentration. h comparison of the binding affinities of bovine albumin, p-lnctoglobulin, and bovine y-globulin indicates a decrease in t'lie order listed. Rl.:FF;RESC'I,-secoiidh-iltioiial Colloid S.yinposiuin, whicli WHY Iieltl under tlic nuspic,es of tlic Division of Colloid C,lic,inistr!. of the A 4 ~ n e i ~ i Chemical c~ii Society at' ('niiil Iritlgc , .R Inssa cl iiise ti s . ,Jriiic 23-25, 1 9-l-R .
