Avidin. II. Composition H. Fraenkel-Conrat: Prom the Western Regional
and Mode of Action of Avidin A N. S. Snell and E. D. Ducay Research Laboratory,2
Received October 16, 1951
In the preceding paper (1) ‘the occurrence of avidin in egg white in the form of a complex with either nucleic acid (avidin NA) or with an acidic protein (avidin XA) has been described, as well as methods of separating the common albumin-like moiety, avidin A, which carries the biotin-binding activity. This paper contains analytical data for avidin A, and attempts to elucidate the mode of fixation of biotin by the protein. METHODS Characterization of Carbohydrate Components The orcinol method (2) was used for calorimetric approximation of the hexose content of glycoproteins; the Elson and Morgan method (3) was used for the hexosamines, after 4 hr. of refluxing with N hydrochloric acid. Many fruitless attempts were made to separate, identify, and determine the carbohydrate constituents of avidin (or other glycoproteins, e.g. ovomucoid) by paper chromatography of sulfuric acid hydrolyzates, neutralized with baryta, by the techniques of Werner and Odin (4). While pure sugars, or mixtures of these gave clearcut spots3 of reproducible RF values by ascending chromatography with a variety of solvents, the same sugars, in the presence of ad hoc mixtures of amino acids or in hydrolyzates of glycoproteins (5-257c carbohydrate) showed faint, diffuse and uncharacteristic spots and bands of low Rp. No good separation could be detected upon elution and calorimetric analyses of various areas of the paper. In contrast to Werner and Odin, we found sugar-amino acid interaction an insurmountable obstacle in the analysis of such solutions. It was thereupon decided to separate the sugars from amino acids and peptides 1 Present address: Virus Laboratory, University of California, Berkeley, Calif. 2 Bureau of Agricultural and Industrial Chemistry, Agricultural Research Administration, U. S. Department of Agriculture. 3 For the detection of sugars, both the method of Partridge (5), and that of McCready et al. (6) were used with equal success. 97
by means of ion-exchange resins. When a 2 N sulfuric acid hydrolyzate was passed through a column of t,he basic resin, Pcrmutit S (washed with ITCI, X:1011 and water), developing with N sulfuric acid until arid appealed in the elu:~tr, N()~l()O~O of the sugar and glucosamine and only 10m20y0 of the nitrogen of thr hytlrolyz:tt,c (largely arginine) were in the eluate. EIowever, chromatography of such solutions st.ill did not permit, separations of glucosamine or galactose from mannosr, as the) can readily be obtained with pure sugar solutions. The use of Zeo-Rex (corresponding to Zeocarh 215 of the English literature (7)) was then attempted. This acidic resin removes all amino acids, but unfortunately also glucosamine, from neutral solution. However, elution with dilute saline permits the recovery of glucosamine (7). When protein samples (7-12 my.) wcrc hydrolyzed with 1 ml. of 2 N sulfuric acid (12 hr. at 100” in a sealed tube), then freed from sulfate by means of baryta and the neutral solution concentrated and passed through a Zeo-Rex column, at least 90% of the hrxose of the hydrolyzate appeared in the filtrate. Paper chromat,ographic analysis of such material (descending or ascending chromatography with butanolLacetic acid or ppridine-amyl alcohol as solvents) yielded clearly resolved spots of reproducible li)~ values. Corresponding areas of the paper could be rlut,ed with w:tt,er and analyzed for reducing sugars (8).
and Chemical Methods
As in previous investigations (9), the micro-Kjeldahl and manometric van Slyke methods were used for N and amino N determinations, respectively; colorimetric methods were used for the determination of trypt,ophan and cystine; while the other amino acids were determined microbiologically (9). The determination of total basic groups in proteins in unbuffered solution was performed as follows: To duplicate or triplicate samples of protein (2-3 mg. or 1 ml. solution) were added water to give a final volume of 4 ml., 1 ml. of 0.02 iV HCl, and 1, 2 and at times 3 ml. of an 0.1% solution of Orange G. After 24 hr. of equilibration, the unbound dye was determined calorimetrically in t,he supernatant as usual (10). For determination of the total acid groups in unbuffered solution, similar protein samples were brought to 1, 2, and 3 ml. with water, tieated with 0.2,0.4, and 0.6 ml. of 0.1 N NaOH and 1,2, and 3 ml. of 0.2% safranine solution, respectively. Equilibration, etc., was done in t)he usual manner. The amounts of the dyes hound, and thus the numher of acid and basic groups, were for several test proteins in good accord with values obtained by the old method (lo), as well as with those calculated from the literat,ure, reproducibilit)?; being slightly better in the new analyses without buffer. The chemical modification reactions were carried out as in previous related studies (11-16). Many of these experiments were performed before the sensitivity of avidin to trace metals (1) was fully recognized. Thus dialysis against running tap water was not avoided. It is therefore possible that some of the partial decreases in activity observed were due to such causes, unrelated to the reactions performed. It appears noteworthy that avidin NA was rendered water-soluble by acetylation of its amino groups. The nucleic acid was not removed, which is not surprising since it is not dialyzahle (17), but it appears probable t,hat it was no longer complex-bound.
Composition of Avidin A The low nitrogen content of avidin A (Table I) suggested that it might be a glycoprotein. The presence of about 9% of carbohydrate, half of it, hexosamine, was indicated by calorimetric analyses and subsecluently demonstrat.ed by ion-exchange separation and paper chromatographic identification of the sugars. The hexose was identified as mannose. When the two complex forms of avidin were similarly analyzed, avidin XA was found to contain appreciably more hexose than nvidin A or avidin NA (Table I) ; upon chromatographic analysis there TABLE Typical Type of preparation
A PU‘A X.4
N 104 g:
13.8 14.2 11.8
oj Intact AE;ini;
Avidin Avidin Avidin
Avidin Basic ~ ~ kwyP% Equlv./ ~ 1O”g. -ipp-
&%?,, 104 g.
6.8 7.9” 8.7
I j Hexosamine
% 1.10 1.01 0.87
7.9 7.2 6.2
~ 1 1
9.7 7.6 7.1
(1On air-dry basis, containing 8P10y0 moisture. b Orcinol method, thus only of comparative value. c Twelve and one-tenth for 2.8-3.1 !lI ammonium sulfate fractions up to I .45yo I’, compared to 1 .O’% for most avidin X,4 preparations.
appeared two distinct spots in two solvents, corresponding in their positions to mannose and galactose (about 3.4oj, of each). Avidin A was generally free from galactose. Thus the galactose must be attributed to the complexing X-protein. The amino acid make-up of avidin A is very similar, as would be expected, to that of 90% pure avidin NA (9) (Table II). Cystine and cysteine are probably absent from the pure protein. Tyrosine and histidine contents are comparatively low. The relatively greater amounts of aspartic acid and threonine, as compared to glutamic acid and serine, are somewhat unusual (Table II). The amino acids found amount to only 85yo of the weight of the glycoprotein. After addition of 9% sugar, there remains about 20% of undetermined composition; part of this is known to be alanine. The determination of the number of acid and basic groups of avidin
has posed certain problems. While avidin has always been described as a very basic protein, analyses by the dye method (10) indicated an excess of acid over basic groups in many preparations of avidin NA (9). This paradoxical behavior is similar to that of tobacco mosaic virus under t,he same conditions (18). Thus, the virus nucleoprotein has been TABLE Amino Component Alanine Arginine Aspartic acid Cystine Glutamic acid Glycine Histidine Hydroxyproline Isoleucine Leucine Lysine Methionine Phenylalanine Proline Serine Threonine Tryptophan Tyrosine Valine
II of Avidin
Avidin A %“- *
Not determined 6.7 (22)* 10.6 0.170 (0.4) Not determined 5.1 0.87 (3.3) Sot determined 5.1 4.9 6.6 (26) 1.5 (5.9) 6.2 1.4 (7.0) 4.5 12.6 6.0 0.93 (3.0) 4.5
I’l.eparations Avidin NA (2) %“* *
Present 6.5 9.7 0.5 6.6 1.6 0.96 (4.0) 5.5 4.9 6.2 1.4 (6.0) 5.9 1.6 (8.9) 4.5 10.5 5.4 0.88 (3.1) 4.2
n Grams/100 g. dry protein, or nucleoprotein cont,aining 10% nucleic acid, rcspectively. * Figures in parentheses indicate equivalents/mole protein, assuming a molecul:tr weight of 58,000 and 64,000 for avidin A and avidin NA, respectively. c Avidin XA was used for this analysis; other amino acid analyses performrtl and histidine, indicated no with avidin XA, particularly tyrosinr, t r~q~lophm, significant difierences with avidin A.
shown to give a very much lower value for basic groups, and a slightly higher one for acid groups, than calculated from the amino acid composition (19). In contrast, the virus protein, from which the nucleic acid had been removed, gave values in excellent agreement with the calculated (18, 19).4 It appears that in the case of both the virus nucleo4 Kausche and Hahn actually believed (18) that there was a great discrepancy between the acid groups as determined )Jy the dye method, and those calculated
protein and avidin NA, part of the basic groups of the protein are bound so firmly to the nucleic acid that they cannot bind the acidic dye at pH 2.2; on the other hand, in alkaline solution some acid groups from the nucleic acid seem to become available to the basic dye, particularly in the case of avidin NA. In support of this interpretation, avidin fractions, enriched in nucleic acid by ammonium sulfate refractionation (2.8-3.1 M) showed exceptionally high acid group values. Avidin XA also contained more acid than basic groups, which may be similarly explained on the basis of the marked acidity of the complexing protein. In contrast, avidin A contained a definite excess of basic over acid groups, in conformity with its high isoelectric point (Table I). Since electrophoretic data indicated a selective affinity of avidin A for phosphate ions,6 it was attempted to rule out interference by such an effect in the analysis for acid and basic groups, by modifying the technique to exclude phosphate buffers. There were repeated indications that the new technique yielded slightly higher values for basic and lower for acid groups than the old, as would be expected if firmly bound phosphate were present in the test solutions; but the differences were not sufficiently marked or consistent to permit a definite conclusion. It was attempted to calculate the number of peptide chains of avidin A from analytical data by the same three methods previously employed: (a) difference between van Slyke amino N and lysine, (6) difference between total basic groups and arginine plus histidine plus lysine, and (c) difference between total acid groups and glutamic plus aspartic acid, plus tyrosine, minus amide N (9). The resultant three values, however, diverged greatly. No single reasonable assumption of error, such as the possible destruction of basic amino acids during the hydrolysis of the glycoprotein, or the presence of free amino groups from the gluc*osamine, appeared to suffice in resolving this divergence. It appears t,hat, our present, as yet incomplete, knowledge of the composition of this glyroprotein does not permit any significant structural conclusions.6 from t,he amino acid composition of tobacco mosaic virus. This discrepancy was due, however, to an oversight, since they neglected to substract the amide N from the aspartic plus glutamic acid analyses. When this is done, the calculated value for the acid groups is 10.8 (per lo4 g. protein), while the values found were 10.1, and 11.0, and thus in good agreement. 6 Ward, W., in preparation for press. 6 The number of peptide chains of avidin has recently been investigated by the I]NP method (Fracnkel-Conrnt and I’ortrr, IRiorhim. cl Bioph?/s. :Irln in press).
Mode of Action of Avidin The remarkable activity of firm fixation and inactivation of two moles of biotin per mole avidin, rapid even at extreme dilutions, suggests the occurrence of sites of very specific configuration in this protein. It was thought that chemical modification of avidin might yield indications as to the nature of these sites. The approach was at first similar to that which had been used with other biologically active proteins (11-16). However, avidin was found remarkably resistant to almost all reagents employed (Table III). No clear-cut dependence of its activity upon the integrity of amino, phenolic, imidazole, or disulfide groups could be established. A possible role for some carboxyl, indole, amide, or guanidyl groups was not as clearly excluded. The retention of most of the activity after iodination under a variety of conditions favoring both subst’itution and oxidation was particularly surprising. Since biotin or oxybiotin are available for fixation even in the form of the esters or corresponding alcohols (20, 21), t,he nonionic nature of the linkage is evident, and it is understandable that reactions such as acetylation of the amino groups did not affect the activity of avidin. However, the great variety of unessential groups suggests that the activity is localized in a single specific group of the protein. This group must be expected to show some structural relationship to the cyclic ureido group of biotin which, apart from the shape of the molecule, appears the sole prerequisite for combinability with avidin (20-23). Another cue in regard to the nature of the site of activit,y comes from pH-stability studies. Avidin is comparatively resistant to many denaturing agents (heat, urea, surface forces, organic solvents). Treat,ment with N acid and alkali for 4 hr. at 23” caused only slight loss of activity. But exposure for prolonged periods of time at 36” to 0.01 or 0.02 N hydrochloric acid caused progressive inactivation; the effect of 0.02 N sodium hydroxide was even slower. The slow rat,e and temperature dependence of this inactivation at’ high and low pH levels (pH 2.2 and 10.5) suggest a reaction involving primary bonds, most probably hydrolytic in nature, rather than a rupture of only secondary or ionic bonds, i.e. denaturation. Acid treatment slowly rendered part) of the avidin insoluble; denaturation may thus be a secondary effect of the exposure to acid. The protein also became digestible by pepsin. In contrast to free avidin, the avidin-biotin complex was completely resistant to 0.02 N acid or alkali for prolonged periods of time at elevated temperature. No biotin was released, no precipit,ate formed, and pepsin was
Derivatives of Avidin Protein
Type of preparation
Number el quiv./lO4
Residual activity ( ~.
NAG NAa XAa A NA’ NA NAa NAa NAa NA’ 9 NA’” A NAa NAa NAa NAa A NA’= NAa NAO NA NA NA A
snhydr. (lZO),* pH 7 Do. (600) pH 7 Do. (120) pII 7 Do. pH 7 Methanol-HCl, -5”, 3 days Do. 24”, 2 days Iodine (l), pH 5.2, 24 hr. Do. (2), pH 5,2, 9 M urea, 24 hr. Do. (35), pH 7.6, 10 min. Do. (63)@pH 7.6, 2 hr. Do. (35)” pH 7.6, 20 min. Diazobenzene sulf. acid (4) Do. (pH 7.6) (5) 1 Formaldehyde, pH 11, 15 min. Do. f &nine, pH 6.1 Do. + proline, pH 6.1 Do. + acetamide pH 6.1 Do. + Do. pH 6.1 Glucose, pH 7.6, 53”, 4 days Thioglycol (lo), pH 5.0, 17 hr. Same, followed by iodoacetamide (PH 8.5) OH-radicals (Fe + HzOz) Hydroxylamine (3.5$&j, 53”, 2 days Do. (15y0), Do. Do. (6701, Do.
Amino Amino Bmino Amino Carboxyl Carboxyl Phenol Phenol Phenol
4.0 5.0 4.4
Amine Amino Amino Disulfide Disulfide
0.35 0.32 1.8 0.8 1.5
70 64 120 180 80 150
6.8 4.1 3.4 4.5 3.2
62 32 50 66 47
% 1010 6tic 4oc 108 74 (m)d 36# 85 99 128f 79 110 62 2ii 60 37 68 1ld.i 75 62 72 100 30 70 0 0
a Older preparations, assaying only 50-70y0 of the activity of the pure preparations, were used for these reactions. * Unspecified figures in parentheses represent equivalenta per 1W g. protein. ’ All acetyl derivatives are water-soluble, yet those of avidin NA retain the nucleic acid. d Activity of the fraction soluble in 0.4 M saline. e Percentage of activity, as compared to control solution treated at 24” with methanol, and aqueous HCl (2 days each). f Treatment with excess sulfite to release labilely bound (imidszole N-bound) iodine (12) decreased the iodine content to 0.52 equivalents, and did not affect the activity. 0 About five equivalents of iodine were used up. * About 2.5 equivalents of iodine were used up. ’ A single assay. ’ Mostly insoluble.
without effect. The latter fact, or rather the complete resistance of the avidin-biotin complex against all proteolytic enzymes (24, 25), naturally represents one of the prime causes for the syndrome of egg white
injury.’ The simplest model for biotin, 2-imidazolidone when added in great. excess, had a similar, though less complete prot,eeting a&on; urea was inactive (Table IV). These findings appear to indicate that t,he site of the protein most susceptible to hydrolytic destruction is also the one adapted to the fixation of cyclic ureido compounds, and stabilized by that fixation. A group crosslinking two suitably diacylimide group, e.g. an -NHplaced aspartic or glutamic side chains (formula I) would fulfill all these requirements. It represents a complementary structure to the ureido Protein / \ CO-NH-CO NH-CO-NH
CH I CH-R
AH2 ‘ls/ I
group with which it could combine by a triad of adjacent hydrogen bonds supplying the necessary specificity and composite bond strength. This type of bond would be expected to be inert in the chemical reactions to which avidin was subjected without inactivation. It would be the group in a protein most susceptible to hydrolysis, and it would be stabilized by its firm combination with biotin. Model experiments performed with diacetylimide (CH&O-NHCOCHB) showed that it hydrolyzed slowly to acetic acid and acetamide under similar conditions of acid and alkaline pH and temperature as cause slow inactivation and insolubilization of avidin. The acetic acid formed could be titrated in the presence of added protein. In 0.02 N HCl at 36” the reaction reached completion in about 12 days at 36”. Similar treatment of avidin seemed to liberate in one experiment a significant number of titratable carboxyl groups, while slightly gentle1 treatment (7 days in 0.0144.02 N acid), though sufficient to cause inactivation, yielded small or no increases in titration. It must be noted, however, that the hydrolysis of two hypothetical diacylimide sites per 7 The stabilization Gyijrgy (24).
has been observed
avidin molecule would be almost within the range of experimental error of t’he technique employed. IJpon isolation of the samples, and analysis of acid groups (lo), the warm acid-inactivated avidin was found regularly higher in acid groups (by 4-5/male) [see Ref. (I), footnote I I], t)han bhe control sample that had been kept in acid solution at 3” and retained it,s activity. These results, while not sufficient,ly consistent to TSBLE
lQ?‘ect of Acid and Alkali Reaction
Avidin N.4 (0.5 mg./ml.) + biotin (2.4 equiv./mole) + 2.imidazolidone (7000 cquiv./ mole) + urea (7000 equiv./molc) .1vidin NA (0.5 mg./ml.) + biotin (2.4 equiv./molc) + 2.imidazolidone (7000 equiv./ mole) + urea (7000 equiv./molc) + sodium barbital (7000 cquiv./ mole) Avidin XA (2 mg./ml.) + hiotin (2.1 equiv./molej (purest prep.) + hiot)in (2.3cquiv./ mole)
on Avidin Activity bath O.OlN
retained after incu(36”) in presence of HCl
3 3 3
20”. b 94c 57
6‘1 95 83
3 1 I I
I.16 3 no 93
1 1 2 2 4
* 9lY 98
” Upon addition of pepsin t,o such samples, about 60% of the protein \v:ts rendered dialyzable wit,hin 24 hr. at 36”. * Precipitates formed during incubation, but a-erc dissolved by pepsin. CThe addition of pepsin had no effect on such samples. All the protein and all the biotin remained undialyzable.
be regarded as proof, definitely suggest the new-formation of carboxyl groups upon inactivation of avidin with warm acid, under conditions which do not appreciably affect other proteins. The fact that 50-100 mg. of protein is needed for each of t’hese experiments has been a serious obstacle preventing the necessary repetitions and refinements. In a search for reactions characteristic for cyclic diacylimide groups, the introduction of amines (-CO-NH-CO*NH”% -CONHR + NH2-CO-) was investigated. Hydroxylamine was used because of the ease of det,ectability of protein-bound hydroxylamides (13). Tests with
N. R. SNELL
AND E. D. DUCAY
diacetylimide showed that hydroxylamine (fivefold excess) was acylated at an appreciable rate at room temperature, but] the cyclic model, succinimide, reacted slowly even at 53”. Avidin was found to react, and to become inactivated and partly insolubilized by treat,ment with high concentrations of hydroxylamine at elevated temperature (Table III), but t,he concentrations necessary to achieve this were such that transamidation of ordinary amide groups and other reactions could not be excluded. Thus, inactivation of avidin by hydroxylamine can be regarded as no more than weak substantiating evidence for the occurrence of diacylimide groups in this protein. While the final proof for the proposed st,ructure (formula I) of the biotin-combining site in avidin is still missing, and other possibly related structures may be envisaged and have been tested, all observed results can be explained on the basis of some such structure, and none are contradictory to it. 8UMMrlRY
Avidin A is a glycoprotein conbaining mannose and hexosamine. Data caoncerning its amino acid composition and its amino, acidic, and basic groups are given. Avidin XA, while similar in most analytical respects, contains galactose also. A variety of cahemical modification reactions did not appreciably af?ect the biotin-binding activit,y of avidin, which suggests that chemically reactive amino, phenolic, imidazole, carboxyl, or disulfide groups do not occur in the active site. The susceptibility of avidin activity to prolonged exposure to dilute acsid or alkali was investigated, as well as the protection afforded by hiotin and imidazolidone. A hypothetical structural component of the biotin-binding site is proposed, which would explain its mode of function as well as its resistance to all but hydrolytic agents. REFERENCES 9.
S., ANI) Duc*.\Y,
92, 80 (1952). 2. FRIEDMANN, Il., Riochem. Rio;hys.
J. 44, 117 (1949). IhSON, L. A.. AND L~~ORGAK,W. T. J., ~iochem. .I. 27, 1824 (1933). WERNER, I., AND ODIN, L., Lrpsala [email protected]
Piirh. 64.69 (1949). PARTRIDGE, S. M., Nature 164,443 (1949). MCCREADY, R. M.. WALTER, E. D., AND MACLAY, W. D., Pood ‘I’echnol. 19 (1950). 7. PARTRIDGE, S. M., Biochem. J. 42, 238 (1948); ibid. 46,459 (1949). S. SOMOGYI, M., J. Riol. Chem. 160, fil (1945).
3. It. 5. 6.
9. LEWIS, J. C., SNELL, N. S., HIRSUMASN, D., AND FRAENKEL-CONRAT, H., J. Biol. Chem. 188,23 (1950). 10. FRAENKEL-CONRAT, H., ASD COOPER,M., .J. Biol. Chem. 164, 239 (1944). 11. FRAENKEL-CONRAT, II., REAS, R. S., AND LINEWEAVER, IT., J. BioZ. Chem.
177, 385 (1949). 12. FRAENKEL-CONHAT, II., A&. Hi&em. 27, 109 (1950); ibid. 28,kQ (1950). 13. FRAENKEIXOSRAT, H., .~NDF~SEY, R. fi., Arch. Biochem. 29, 101 (1950). 11. FRAESKEL-CONRAT, kr., ANI) l'~hesKEr,-(~on'~.4~, ,I., Biochim. ~1 Biophys. Ada 6, 98 (1950). 15. FR.4ENmL-CoNRAT, .I., ASI> I~‘~AE~K~I,-C~~RA,I~, II., Hiochim. rl Hiqh/S. Ada 5, 59 (1950). 16. ~RA~NKEI.-CUNR.~T,II., BEAS, It. S.,l)ucau, I(:. D., A.UL)OI.COY~,II. S., .I&. Biochem. Biophys. 37, 393 (1952). 17. FRAENKEI,-COXRAT, H., WARD, W., Smr,r., S., AXI) DUCAY, 11:. I)., .J. Am. Chem. Sot. 72, 3826 (1950). IS. KAUSCHE, G. A., ASD HAHN, F., Z. Naturjorsch. 3b, 437 (1948). 19. KNIGHT, C. A., .I. Biol. Chem. 171, 297 (1947). 20. VIGSEAUD, V. DU, DmTnmR,K., HOFiMaNN, K., AND R,~EI,VILLE,D. u., hoc. Sot. Esptl. Biol. Med. 60, 374 (1942). 21. WINNICK, T., HOFMANN, K., PII,GRIM, F. J., AND AXELROD, A. E., J. Bid. Chem. 181, 405 (1945). 22. DITTYER, K., ASU VIC~NEAUI),V. DU, J. Biol. Chem. 189,63 (1947). 23. WRIUI-IT, L. D., SKEGGS,II. It., .ZSD CRESSON,l3. L., Proc. Sot. Ezpll. Biol. Med. 84, 150 (19-l7). 24. GY~RGY, I'., ROSE, C. S., AXD TOMARELLI, R., J. Bid. Chem. 144, 169 (19$2). 25. GY~RGY, P., ANU ROSE, C. S., Proc. Sot. Ezptl. Biol. Med. 63, 55 (1943).