Structural studies and organic ligand-binding properties of bovine plasma albumin.

Abstract Cyanogen bromide cleavage of bovine plasma albumin in 75% formic acid gave two fragments, N and C, in yields of 80%. After reduction of the cross-linking disulfide bonds of Fragment N, two peptides containing 88 and 98 amino acid residues were isolated. The single sulfhydryl group of albumin was found to be located in the 88 amino acid residue peptide which occupies the amino-terminal position of the molecule. After reduction of Fragment C, three peptides containing 211, 148, and 34 amino acid residues were isolated. These five peptides together account for the albumin molecule. Limited tryptic hydrolysis of defatted bovine plasma albumin at pH 8.8 and 0° gave a fragment of molecular weight of about 40,000 in a yield of 21%. The fragment is derived from the carboxyl-terminal two-thirds of the albumin molecule as shown by a comparison of its three cyanogen bromide peptides with those from albumin. These data were used for the alignment of the five cyanogen bromide peptides of albumin. Binding studies showed that the tryptic fragment and defatted albumin both had one primary site for octanoate and l-tryptophan. d-Tryptophan competitively displaced either of these two ligands from the fragment and from albumin, thus suggesting that all three ligands bind at the same site. The binding constants of the fragment for the three ligands were all about one-third of those for albumin. The ratio of the binding constants of l-and d-tryptophan for the fragment was nearly the same as that for albumin, and this was also the case for octanoate and l-tryptophan. These similarities strongly suggest that the binding site in the fragment is the one present in albumin.

Cyanogen bromide cleavage of bovine plasma albumin in 75% formic acid gave two fragments, N and C, in yields of 80%.
After reduction of the cross-linking disulfide bonds of Fragment N, two peptides containing 88 and 98 amino acid residues were isolated.
The single sulfhydryl group of albumin was found to be located in the 88 amino acid residue peptide which occupies the amino-terminal position of the molecule.
After reduction of Fragment C, three peptides containing 211, 148, and 34 amino acid residues were isolated.
These five peptides together account for the albumin molecule.
Limited tryptic hydrolysis of defatted bovine plasma albumin at pH 8.8 and 0" gave a fragment of molecular weight of about 40,000 in a yield of 21%.
The fragment is derived from the carboxyl-terminal two-thirds of the albumin molecule as shown by a comparison of its three cyanogen bromide peptides with those from albumin.
These data were used for the alignment of the five cyanogen bromide peptides of albumin.
Binding studies showed that the tryptic fragment and defatted albumin both had one primary site for octanoate and L-tryptophan. D-Tryptophan competitively displaced either of these two ligands from the fragment and from albumin, thus suggesting that all three ligands bind at the same site. The binding constants of the fragment for the three ligands were all about one-third of those for albumin. The ratio of the binding constants of L-and D-tryptophan for the fragment was nearly the same as that for albumin, and this was also the case for octanoate and L-tryptophan.
These similarities strongly suggest that the binding site in the fragment is the one present in albumin.
One of the interesting properties of bovine or human plasma albumin is its propensity to bind a wide variety of organic and inorganic ligands.
Because of this property, albumin is believed to function as a transport protein, and possibly serves as a mechanism in regulating the blood level of certain drugs and metabolites (2). The binding of diverse organic ligands by * This research was supported in part by a United States Public Health Service Grant AI-08445. A preliminary account of this work has been published (1). albumin is in contrast to the specific interactions of enzymes and substrates or of antigens and antibodies.
The binding affinities of albumin for different ligands are often as high as those of the specific interactions.
The present studies were made with the intention to understand better what molecular parameters of albumin are required for this binding property.
Our approach has consisted of stepwise degradation of albumin jnto fragments and an examination of the binding properties of such fragments.
This has the advantage of providing structural information about this most abundant plasma protein for which only fragmentary data are presently available (3,4). Both bovine and human plasma albumins are readily available and studies carried out with either one will be equally useful for the understanding of the other because of probable similarities in structure.
Bovine albumin was chosen since our first degradation procedure was cyanogen bromide cleavage at methionyl bonds, and bovine albumin has one less methionine than does the human albumin (4). Our second degradation procedure follows the findings of other workers that limited tryptic hydrolysis of bovine albumin gave intermediates of large molecular sizes (5, 6).
EXPERIMENTAL PROCEDURE 2MateriaZs-Crystalline bovine plasma albumin samples (lot B-70411 and D-71209) were from Armour.
Cyanogen bromide was from Aldrich.
Formic acid, 97 to 100% from Matheson Coleman and Bell (Division of the Matheson Company, East Rutherford, New Jersey) was purified by distillation before use. Dithiothreitol was from Calbiochem. Urea and guanidine hydrochloride were of ultrapure quality from Mann. Carboxypeptidase A was from Sigma (lot C-33B80).
Trypsin, crystallized twice was from Worthington (TRL 6254). 14C-Iodoacetamide with specific activity of 3.87 mCi per mmole from New England Nuclear was diluted wit,h cold carrier to a specific activity of 0.43 mCi per mmole before use. On reaction with excess cysteine followed by acid hydrolysis, 99% of the radioactivity was found in the carboxymethylcysteine peak separated on the amino acid analyzer.
For preparation of YI-half-cystinyl-BPA', it was diluted with cold n-cystine to give 7 x lo5 cpm per pmole. The radiochemical purity of the Issue of November 25, 1970 T. P. King and M. Spencer 6135 diluted cystine was established by radioactivity determination of the cystine peak separated on an amino acid analyzer.
L-and n-Tryptophan obtained from Calbiochem were purified by recrystallization from water; they had identical ultraviolet spectra, but they had opposite optical rotations, [ar], of -30.8 and +30.8" (C, 1, in water), respectively.
lJ4C-L-Tqptophan with a specific activity of 9.6 mCi per mmole was also from Calbiochem (lot 850014). For binding studies, it was diluted with cold L-tryptophan to give 1.5 x lo5 cpm per pmole.
The radiochemical purity was established by chromatography on Sephadex G-25; tryptophan was eluted after the total column volume, and there was coincidence of the radioactivity distribution with the tryptophan peak. The stereochemical purity of I%-n-tryptophan was not checked. The manufacturer stated that it was prepared by enzymatic resolution and that it was free of D isomer as tested by cocrystallization in the presence of excess D isomer.
Sodium 1-14C-octanoate with a specific activity of 17 mCi per mmole was from Amersham-Searle Corporation, Arlington Heights, Illinois (lot CPA 93). For binding studies, it was diluted with cold sodium octanoate (acid purified by vacuum distillation) to give 1.5 X lo5 cpm per nmole. To check its radiochemical purity, binding studies were also carried out after further dilution with carrier to give 0.5 X lo5 cpm per pmole.
Methods-Amino acid analyses were made on a Beckman-Spinco model 120-B analyzer after hydrolysis of the samples in 6 N HCl at 110" for 22 and 72 hours (7). For good separation of the homoserine lactone and ammonia peaks, a short column (15 cm) was used; for resolution of homoserine and glutamic acid peaks, analyses on the long column (55 cm length) were started at 48", and the temperature was changed to the normal operating value of 55" at the position of threonine peak. For 22-hour hydrolysates, the following decomposition factors were used for carboxymethylcysteine, threonine, cystine, and serine: 3%, 5%,, 7%, and lo%, respectively (7). Tryptophan was determined on the basis of the molar extinction coefficients of the peptides at 280 nm.
For quantitative NHz-terminal end group analysis, the cyanate procedure was used, and only the hydantoin Fractions A and B of the isolation scheme of Stark (8) were examined. For qualitative analyses, the Edman procedure described by Fambrough and Bonner (9) and the dansylation (10) procedure were used. The phenylthiohydantoins were identified by chromatography on Eastman Kodak silica sheets with fluorescent indicator using two solvent systems; heptane-1 , a-dichloroethaneformic acid (99q/) in volume ratiosuf 6:3t? or 3:6: 1 (only the upper phase used). The spots were visualized under ultraviolet light, and also after spraying first with starch solution, followed by a solution of NaIs and NaNa (11). The dansyl amino acids were identified by chromatography on polyamide sheets (12), with the exception of dansyl-arginine which was identified on paper electrophoresis (10) and verified by Sakaguchi reaction.
For COOK-terminal end group analyses, the peptide in 0.1 M ammonium bicarbonate was digested with carboxypeptidase A at a substrate to enzyme molar ratio of 80 for 2 hours at 25". After lyophilization to remove buffer salt, an aliquot of the digest dissolved in pH 2.2 citrate buffer was examined on the amino acid analyzer.
Disc electrophoreses at alkaline and acid pH values were performed in Tris-urea gel (13) and in acetate-urea gel (14), respectively.
The sample was applied to the gel as a solution in 8 M urea with the appropriate buffer, and the sample gel was omitted.
For staining with Amido schwarz dye (15) about 30 pg of sample were used. For staining with Coomassie blue dyes (16)) about 8 pg of sample were used. Paper electrophoresis was carried out on Whatman No. 3 paper, using a pH 6 buffer containing 34 ml of pyridine and 3.5 ml of acetic acid per liter of water; the electrophoretic conditions were about 20 volts per cm for 3 hours.
Radioactivity determinations of samples (50-to 200-/J aliquots) were made after dilution with 4 ml of ethanol and 15 ml of scintillation fluid. The scintillation fluid was prepared by diluting 53 ml of Liquifluor (New England Nuclear) with 1 liter of toluene.
Alternatively, the radioactivity determinations of samples (0.2.ml aliquots) were made after dilution with 5 ml of Bray's solution (17). The scintillation counter which was used had an efficiency of 60% for the 14C-isotope.
For calculation of the molar extinction coefficients of peptides at 280 nm, the extinction coefficients of tryptophan, tyrosine, and cystine were taken to be 5690, 1280, and 120, respectively (18). The molar extinction coefficient at 280 nm for BPA in Tris-HCl buffer (pH 7.96) was found to be 4.2 X lo4 (.t5%), employing amino acid analysis and the composition given in Table I in order to establish the concentration of the BPA solution.
Preparation of Half-cyst&$ Bovine Plasma Albumin-To a solution of BPA (3.12 g, 44 Mmole) in 100 ml of 0.10 M Tris and 0.06 M HCl (pH 7.96) was added a solution of n-cystine (67.4 mg, 280 pmole) in 142 ml of the same buffer. Because of the limited solubility of L-cystine, it was dissolved first in 1.5 ml of 1.08 N NaOH, then immediately diluted with 140 ml of buffer. After 17 hours at 25", the sulfhydryl titer had decreased from the initial value of about 0.65 mole per mole of BPA to less than 0.01 mole, as determined by the Elhnan procedure ( It was next applied to a Sephadex G-75 column (2.8 X 195 cm) and eluted with the pH 2.86 ammonium formate buffer at a flow rate of 24 ml per hour. Fractions of 12-ml volume were collected.
Cuts containing Fragments N and C ( Fig. 1) were concentrated by ultrafiltration, then rechromatographed on the same column.
Lyophilization of the fragments was avoided since this process and the storage in the lyophilized form led to aggregated fragments.
For the reduction experiments to be deseribed below, lyophilized fragments were used. The reaction time required for cyanogen bromide cleavage in formic acid could be decreased to 3 hours by increasing the concentration of cyanogen bromide 12-fold. This gave the fragments in yields and purities identical with those obtained after the longer reaction time.
Cleavage in 0.1 N HCl was carried out in the same manner as that in formic acid.
Reduction and Carboxymethylation of Fragment N-To After 3 hours at 25", solid iodoacetamide (300 mg, 1600 pmole) was added. After another 25 min, the mixture was desalted on a Sephadex G-25 column (2 x 66 cm) equilibrated with 2 N acetic acid. The front peak was collected and lyophilized.
For separation of the reduced peptides on Sephadex G-100 (Fig. 3), loads larger than 1 pmole of Fragment C led to poor resolution.
For carboxymethylation, 0.8 pmole of Fragment C was reduced in 1 ml of 2 M Tris and 0.5 M HCl containing guanidine hydrochloride (1.65 g) and dithiothreitol (31 mg) for 3 hours, then a solution of iodoacetic acid (130 mg) in 1 ml of 2 M Tris was added. After 30 min the mixture was freed of excess reagents by dialysis against 0.1 M ammonium bicarbonate and used directly for separation on a Sephadex G-150 column (Fig. 4). Lyophilization of the reduction mixture was avoided, since one of the peptides (t,he arginyl peptide) redissolved with difficulty.
Limited Tryptic Hydrolysis of Half-cystinyl Bovine Plasma Albumin-A solution of half-cystinyl-BPA (150 mg) in 3.0 ml of 0.025 M Tris and 0.015 M HCl was cooled to O", and its pH was adjusted to 8.8 by the addition of 0.21 ml of 0.1 N NaOH. A 150.~1 aliquot of a trypsin solution (10 mg per ml in 1 X lwa N HCI) was added. The pH of the digestion mixture was monitored with a pH meter and was kept at pH 8.80 zt 0.05 by periodic addition of 20-/J aliquots of 0.1 N NaOH.
After reaching the desired extent of hydrolysis, the reaction mixture was acidified to pH 3.1 with 25 ~1 of 6 N HCl, and 6.3 ml of a buffer consisting of 1.74 M formic acid and 0.26 M ammonium formate (pH 2.86) was added.
The mixture was next separated on a column of Sephadex G-150 (200 X 0.9 cm). The column was eluted at 25" with a buffer of 0.174 M formic acid and 0.026 M ammonium formate at a flow rate of 12 ml per hour, and fractions of 3-ml volume were collected.
The desired Fraction T-2 ( Fig. 6) was concentrated by ultrafiltration and rechromatographed on the same column.
The concentrate of rechromatographed Fraction T-2 was neutralized to pH 7.4 by addition of 2 M Tris and was dialyzed against 0.10 M Tris and 0.08 M HCI (pH 7.61). It was next separated on a column (0.9 x 22 cm) of DEAE-cellulose (Whatman DE-32 grade).
The column was eluted at 25" with a linear gradient formed with 200 ml each of 0.10 M Tris and 0.08 M HCl, and 0.10 M Tris, 0.08 M HCl, and 0.10 M NaCl at a flow rate of 36 ml per hour, and 3-ml fractions were collected.
The desired fractions, T-22 and T-23 (Fig. 7), were concentrated and dialyzed against the starting buffer. They were each rechromatographed to reduce their cross contamination with a recovery yield of 50 to 60%. For cyanogen bromide cleavage of Fragment T-23, the concentrate was dialyzed against water and then lyophilized.
The procedures used for cyanogen bromide cleavage and for reduction and carboxamidomethylation were the same as those used for half-cystinyl-BPA.
Equilibrium Dialysis-The binding of organic ligands by half-cystinyl-BPA and it fragments were determined by equilibrium dialysis in a simple apparatus containing 14 cells. The two cylindrical compartments (width, 0.094 inch and diameter, 0.7 inch) of each cell were separated by a sheet of Visking $-$ dialysis tubing.
The total volume of each compartment was 0.5 ml.
Binding experiments were carried out at 24 f lo in a buffer of 0.05 M Tris, 0.03 M HCl, and 0.10 M NaCl (pH 7.95). Into one compartment of each cell were introduced 300 ~1 of a protein solution (about 0.6 x low4 M), and into the other compartment were introduced 300 ~1 of a ligand solution.
The protein solution was thoroughly dialyzed against the desired buffer before use. Duplicate analyses at several different ligand concentrations ranging from 0.3 X 10U4 M to 70 X 1w4 M were made. After equilibrating for about 16 hours, a 200~~1 aliquot was withdrawn from each compartment for determination of the concentration of free and bound ligands by radioactivity. The concentration of bound ligand was calculated by taking the difference of the concentrations in the two compartments, and the concentration of free ligand was that in the compartment without protein.
The average number of moles of l&and bound per mole of protein, denoted as ii, was calculated by dividing the concentration of bound ligand by the protein concentration. The protein concentration was taken to be equal to that initially introduced into the compartment.
At the end of an experiment, about 95% of the volume of solution initially introduced into each compartment was recoverable, and the total recovery of ligand from each cell ranged from 85 to 98%. The low recovery was probably due to absorption to the cell, since this occurred when the free ligand concentration was less than 0.1 X lo-' M. Binding of n-tryptophan was also determined spectrophotometrically.
After equilibration, a 200+1 aliquot from each compartment was diluted with 0.8 ml of 10% trichloracetic acid. The solution from the protein-containing compartment was centrifuged to remove the precipitates before measuring its absorbance.
The binding data given in Figs as described below. The binding of ligands by a protein with i classes of n equivalent and independent sites is described by the following equation: (1) where i; represents the average number of moles of ligand bound per mole of protein, K; is the apparent binding constant for the site of class i, and (A) is the free ligand concentration. In this study, the data were treated on the assumption that there are only two classes of binding sites, since our main interest is concerned with the tight binding primary class.
In the case where another ligand (B) competitively binds only to one of the two classes of ligand A binding sites, Equation 1 becomes : In this work, the free ligand concentration of B was not determined but was approximated to be equal to the total concentration used. The error from this approximation is small since the molar ratio of ligand B to protein was greater than 100.

RESULTS
Most albumin preparations are mixtures in which about two-thirds of the molecules have a free cysteinyl residue (23), but the remaining one-third has this cysteine in the form of a mixed disulfide with L-cysteine or glutathione (24). To avoid oxidation of the cysteine residue during cyanogen bromide cleavage, the albumin preparations used in this work were converted to the mixed disulfide form by reaction with L-cystine or to the S-carboxamidomethylated derivative with iodoacetamide. These two derivatives will be referred to as half-cystinyl-BPA and X-carboxamidomethylated-BPA, respectively. Isolation of Cyanogen Bromide Fragments of Bovine Plasma Albumin-Five peptides should result on cyanogen bromide cleavage of BPA, since albumin contains 4 methioninyl residues. However, the presence of cross-linking disulfide bonds between the peptides decreases the number of products to two. The two products designated as Fragments N and C were isolated from the cleavage mixture by chromatography on Sephadex G-75. The upper graph of Fig. 1 was obtained with half-cystinyl-BPA after cleavage with cyanogen bromide in 0.1 N HCl (25), and the lower graph was obtained with 14C-S-carboxamidomethylated-BPA after similar cleavage in 7501, aqueous formic acid. The elution positions of Fragments N and C indicated their molecular weights to be about 20,000 and 40,000, respectively. There was significantly more of the materials eluted in front of Fragment C when the cleavage was done in HCl rather than in formic acid. This difference is due to the different solvents used for cleavage and is not related to the nature of the protecting group. When half-cystinyl-BPA was cleaved in formic acid, the result was identical with that obtained with S-carboxamidomethylated-BPA.
The radioactivity distribution of the i4C-labeled protecting group showed that the single cysteine residue of albumin is present in Fragment N.
The yields of the two fragments after rechromatography were both 80% from cleavage of half-cystinyl-BPA in formic acid. Their homogeneity was studied by disc electrophoreses in Trisurea gel and in acetate-urea gel. In both buffer systems the fragments migrated as a single band together with a very faint  Table I. The sum of their compositions agrees well with the composition of BPA. The weight recoveries of Fragments N and C as amino acid residues were 98 and 94%, respectively.
The low weight recovery for Fragment C was probably due to bound buffer ions, since its nitrogen recovery was nearly quantitative.
The sample of Fragment C analyzed was prepared by lyophilization from 0.1 M ammonium bicarbonate. The molar extinction coefficients for Fragments N and C at 280 nm in ammonium formate buffer of pH 2.86 were found to be 17,100 f 1,000 and 20,500 f 1,200, respectively, and the values calculated from their amino acid compositions are 16,600 and 19,900.
Fragment N was found to contain 2 NH,-terminal residues, aspartic acid and alanine (Table II). It contained 2 COOHterminal residues of homoserine as indicated by its composition, although carboxypeptidase-A digestion released only 0.45 residue of it together with traces of other amino acids (less than 0.05 residue each). These data indicate that Fragment N contains two cyanogen bromide peptides. Fragment C was found to contain 3 NHz-terminal residues, glutamic acid and proline, by the cyanat,e procedure (Table II) and by Edman degradation,  The composition found is in close agreement with those in the literature (26,27 b Reduced and carboxymethylated Fragment C was used, since it gave a better yield of the end groups than Fragment C did. In addition to the residues reported, about 0.5 mole of carboxymethylcysteine was found. This is believed to be an artifact, since no such end group was detected on dansylation. and arginine by the procedure of dansylation. The yield of dansyl arginine was qualitatively estimated to be about 30%, but affirmative evidence for its presence is given by the work of Brown et al. (28), who isolated a tryptic peptide of BPA con-taining the sequence of Met-Arg. Fragment C contains 2 homoserine residues, and carboxypeptidase A digestion released 0.60 residue of homoserine together with 1.39 residues of alanine, 0.86 residue of leucine, 0.63 residue of threonine, and 0.58 residue of valine. These data together would indicate that Fragment C contains three cyanogen bromide peptides. As aspartic acid and alanine are, respectively, the known NH2-and COOH-terminal residues of BPA (4), it follows then that Fragments N and C contain, respectively, the NH%-terminal and the COOH-terminal portions of BPA. The end group results of Table II indicated that the NHZ-terminal residues of Fragments N or C were not found in equimolar quantities when they were obtained from cleavage in HCI, but that they were nearly equimolar when the cleavage was done in formic acid. This finding suggested incompletely cleaved methionyl bonds in the fragments, a finding useful for establishing the relative positions of the two peptides of Fragment N to be described later.
During the course of isolation, Fragment N was found to have the interesting property of aggregation in the pH range of 4.1 to 5.3, as detected by chromatography on a Sephadex G-100 column with acetate buffers. At pH 4.6, about 50yo of it was aggregated in the form of a polymer, and 30% of it was monomeric. The polymeric Fragment N could not be dissociated This finding establishes that, in intact albumin, the rate of 12 ml per hour, and fractions of 4-ml volume were collected.
alanyl peptide occupies a position penultimate to that of the The fractions were analyzed by ninhydrin color after alkaline NH*-terminal aspartyl peptide. hydrolysis of 100~~1 aliquots (29). The upper graph was obtained with 1 pmole of Fragment N from cleavage of half-cystinyl-BPA The aspartyl and the alanyl peptides were purified by rein 0.1 N HCI, and the Zovrer graph was obtained with 0.6 pmole of chromatography.
Their yields were about 65% when Fragment Fragment N from cleavage of XX'-carboxamidomethylated-N, obtained from cleavage in formic acid, was used as the starting BPA in formic acid. material.
The amino acid composition of the purified peptides  Vol. 245,No. 22 are given in Table III. The molecular weights of the aspartyl and the alanyl peptides were taken to be 10,300 and 12,100, respectively, for the calculation of their amino acid compositions, since only with these molecular weight values was the sum of the compositions of the peptides in close agreement with that of Fragment N. The molar extinction coefficients at 280 nm of About 40 mg (0.83 pmole) of Fragment C was used.
the aspartyl and the alanyl peptides in 0.1 M ammonium bicarbonate mere found to be 3,500 i 500 and 12,800 i 1,300, respectively. The corresponding values calculated from their amino acid compositions were 2,560 and 13,370.
To determine the location of the cysteinyl residue of BPA, the 14C-X-carboxamidomethylated Fragment N was next reduced and carboxymethylated, and the two resulting peptides were separated as shown in the lower graph of Fig. 2. About 80% of the radioactivity in Fragment N was present in the aspartyl peptide fraction, and about 6% was in the alanyl peptide fraction. On rechromatography of the alanyl peptide fraction, all of its radioactivity was removed, thus indicating that the single cysteinyl residue of albmnin is located in the aspartyl peptide region.
Arginyl, ProlyE, Glutamy Peplides oj l+agment C-These three peptides designated according to their N&terminal residues, were isolated in the form of the carboxamidomethylated derivatives, as well as in the form of the carboxymethylated derivatives after reduction of the disulfide bonds of Fragment C. In the experiments to be described below, only Fragment C, isolated from cleavage of albumin in formic acid, was used.
The separation of the carboxamidomethylated peptides on Sephadex G-100 in 2 N acetic acid is given in Fig. 3. Cuts 2, 5, and 4 contained the desired peptides. The N&-terminal residue of Cut b was shown qualitatively to be arginine using  T. P. King and M. Spencer the dansyl technique, and those of Cuts 3 and .4 were found to be proline and glutamic acid in yields of about 70% using the cyanate method. The yields of the arginyl, the prolyl, and the glutamyl peptides were 75, 65, and 75%, respectively. After rechromatography the yields of the arginyl and the prolyl peptides were 52 and 42%. Their homogeneity was checked on disc electrophoresis in acetate-urea gel, each showing one heavy band together with a series of faint slower moving bands suggestive of aggregates of increasing sizes. On paper electrophoresis at pH 6 of the glutamyl peptide, only a single ninhydrinpositive spot migrating toward the anode was observed. The amino acid compositions of these peptides are given in Table IV, and the molecular weights of the arginyl, the prolyl, and the glutamyl peptides were taken to be 26,500, 18,300, and 4,130, respectively, on the basis of their elution positions shown in Fig. 3. The sum of the compositions of the peptides accounts satisfactorily for the composition of Fragment C. The weight recovery of amino acid residues was about 94% for the arginyl and the prolyl peptides, but their nitrogen recovery was nearly quantitative. The low weight recovery of these two peptides might, be due to tightly bound acetic acid, since these samples were obtained on lyophilization from 2 N acetic acid. The molar extinction coefficients at 280 nm in 2 N acetic acid for the arginyl and the prolyl peptides were found to be 13,800 ZIZ 1,400 and 4,660 % 500, respectively. The calculated values from their amino acid compositions are 14,650 and 3,840.
The separation of these three peptides as carboxymethylated derivatives was carried out on a column of Sephadex G-150 in 0.1 M ammonium bicarbonate (top graph, Fig. 4). The arginyl peptide was isolated in 80% yield from Cut 1. Its amino acid composition is identical with that shown in Table IV. It was also present in Cut Z? as indicated by the similar patterns of these two cuts on disc electrophoresis (top of Fig. 4). The arginyl peptide in Cut 1 was aggregated because, when it was rechromatographed on the same column using 0.1 M ammonium bicarbonate containing 0.04% sodium dodecyl sulfate as a solvent, its elution position was significantly retarded. It was eluted just in front of the proline peptide (the lower graph of Fig. 4) in accord with their molecular weight differences. Cut 3 of Fig. 4 has not been characterized. It was electrophoretically different from Cuts 2 and 4 (top of Fig. 4). Cut 4 contained the prolyl peptide in 50% yield. Its amino acid composition was the same as that given in Table IV, with the exception that its glycine content was consistently found to be only about 0.4 residue. Disc electrophoresis of Cut 4 in Tris-urea gel showed the presence   Fig. 4), while disc electrophoresis of the carboxamidomethylated prolyl peptide in acetate-urea gel showed only one band. These observations may be of significance in relation to the known chromatographic heterogeneity of albumin. The glutamyl peptide was present in Cut 6 in 75% yield.
Its composition was identical with that given in Table IV, and on paper electrophoresis at pH 6 it migrated more rapidly than the carboxamidomethylated peptide because of its two additional negative charges. On carboxypeptidase A digestion of the glutamyl peptide, 1.6 moles of alanine and 0.9 mole of leucine were released, thus showing it to be the COOH-terminal peptide of albumin. Isolation of Tryptic Fragments of Bovine Plasma Albumin-Both defatted and nondefatted half-cystinyl-BPA were tested as starting materials. This is because most albumin preparations contain bound fatty acids (21)) and albumin with bound organic ligands is known to be stabilized against denaturation (30) and proteolytic digestion (31). The rate of tryptic hydrolysis of defatted and nondefatted half-cystinyl-BPA at pH 8.8 is shown in Fig. 5. At 25" the nondefatted half-cystinyl-BPA was hydrolyzed too rapidly for convenient isolation of the fragments. Therefore, the temperature was lowered to decrease the rate of hydrolysis.
At 0" the defatted and the nondefatted samples both showed initial rapid cleavages followed by slow cleavages. But the nondefatted sample was hydrolyzed more slowly than the defatted one, probably as a result of ligand-stabilized conformation or conformations. The digests after acidification were separated on a column (1 CNBr-T-23, Fragment T-23 after cyanogen bromide cleavage. b RCAM-CNBr-T-23, Fragment T-23 after cyanogen bromide cleavage followed with reduction and carboxamidomethylation. The two cuts were obtained as shown in Fig. 9. In addition to the amino acids reported, about 4.5 residues of carboxymethylcysteine were found, and these are believed to be an artifact. of Sephadex G-150. The top chromatogram in Fig. 6 was from a 51-min digest of nondefatted half-cystinyl-BPA, and the lower chromatogram was from a 29-min digest of defatted '*C-halfcystinyl-BPA.
Both digests showed four major ultraviolet absorbing zones in different proportions.
Cut T-l consisted mainly of undigested albumin as indicated by disc electrophoresis and end group analysis (Table VI).
There were 25% and 4% of undigested albumin for the nondefatted and defatted samples, respectively.
Cut T-d contained two large fragments both representing the COOH-terminal two-thirds of the BPA molecule, as will be shown by the experiments described below.
The yield of T-d from the defatted sample was greater than that from the nondefatted one. The presence of 14C-half-cystinyl group in Cut T-S indicated that this cut contained fragments originating from the NHz-terminal region of BPA, as we have already established that the single sulfhydryl group of BPA is located within the first 88 ammo acid residues of the mole- cule. The materials in Cut T-S have molecular weights of about 20,000 since trypsin was eluted in this region.
The two fragments in Cut T-d were separated by chromatography on DEAF-cellulose, designated as T-61 and T-23 in the upper graph of Fig. 7. Both fragments were rechromatographed, but only the rechromatography of T-23 is shown in the lower graph of Fig. 7. After rechromatography, the two fragments were still slightly cross contaminated with each other as shown by disc electrophoresis at pH 8.9. The disc electrophoretic mobilities of Fragments T-22 and T-23 were, respectively, slightly slower and faster than that of half-cystinyl-BPA.
The amino acid compositions of the two fragments were indistinguishable within the experimental error (Table V), and they were very similar to that of Fragment C obtained from cyanogen bromide cleavage of BPA.
The two tryptic fragments differed in their NHz-terminal end groups (Table VI) ; T-23 has valine as its end group, and T-22 has two end groups, leucine and tyrosine.
In practice, it was found more convenient to terminate the digestion at 29 min, since this obviates the need of a careful separation of the undigested material.
By carrying the digestion of the defatted halfcystinyl-BPA to 50 min, the yield of Fragment T-23 decreased to 17%, while that of Fragment T-22 was about 12%. Thus, at all stages of digestion, the yield of Fragment T-23 was higher than that of Fragment T-22, while the yield of Fragment T-22 remained approximately constant. This dependence of the yields of the two fragments on the length of di,gestion suggests that they both are unstable intermediates, and that Fragment T-22 is probably formed from Fragment T-23.
Cyarwgen Bromide Cleavage of Fragment T-,%-Cleavage of the 2 methioninyl residues of Fragment T-23 yielded a single product in 80% yield (Fig. 8). The product had an amino acid composition identical with that of the intact fragment with the difference of replacement of 2 methionine residues by 2 homoserine residues. The product had two new NHz-terminal end groups of proline and glutamic acid in addition to the expected end group of valine (Table VI).
These results indicated that the product was a complex of three peptides cross-linked by disulfide bonds.
This was about 0.6 X 10-+ M, and the initial ligand concentration ranged from 0.3 to 30 X 10-* M. 0, the data for octanoate binding; X, the data for competitive binding of octanoate in the presence of 50 X 10--' M of n-tryptophan.
The curves are calculated with the constants given in Table VIII. Comnetitive bindina of octanoate bv halfcystinyl-BPA was also deteimined in the iresence of 90 X"W4 M of n-tryptophan, and the results were in agreement with the calculated curve.
which is also a complex of three peptides cross-linked by disulfide bonds with NHz-terminal end groups of arginine, proline, and glutamic acid.
The three peptides of the cleaved Fragment T-23 were separated after reduction and carboxamidomethylation. This is depicted in the upper chromatogram of Fig. 9. Cuts 1, 2, and S contained the desired peptides.
Cuts 1 and ,9 were rechromatographed to reduce their cross contamination as shown in the lower graph of Fig. 9. The yield of Cut S was 75%, while the yields of the rechromatographed Cuts 1 and ,S were 31 and 48%, respectively.
The NHS-terminal end groups of Cuts 1 and d were found to be valine and proline, respectively (Table VI). Cut S must have glutamic acid as its NHS-terminal end group as deduced from the end group results of the cyanogen bromide cleaved Fragment T-23 and Cuts 1 and 2. The ammo acid compositions of the three cuts are given in Table VII. Cuts 2 and Shad compositions identical with those of the prolyl and the glutamyl peptides isolated from Fragment C, while Cut 1 had a composition very similar to that of the arginyl peptide of Fragment C. These findings, therefore, provide strong support that FIG. 12. n-Tryptophan binding by Fragment T-23 (upper graph) and by half-cystinyl-BPA (lower graph). The experimental conditions were the same as those in Fig. 6 with the exception that the initial ligand concentration ranged from 0.2 to 5X lo+M. l , the data for n-tryptophan binding; X, the data for competitive binding of n-tryptophan in the presence of 50 X 10-* M n-tryptophan.
The curves are calculated with the constants given in Table VIII. Fragments T-23 and C are derived from the same region of the BPA molecule.
The alignment of the three cyanogen bromide peptides in the intact Fragment T-23 must be in the order of valyl peptide, prolyl peptide, and glutamyl peptide, since this is required by the NHn-terminal residue of the valyl peptide and by the absence of a homoserine residue in the glutamyl peptide.
This information has also permitted the alignment of the three peptides of Fragment C in the intact BPA as shown in Fig. 10.

Ligand-binding
Properties of Fragment T-93--The tryptic Fragment T-23 was found to bind octanoate anion, L-, and n-tryptophan, as does intact BPA (32)(33)(34). The binding of octanoate and of Ltryptophan was determined by equilibrium dialysis, while the binding of n-tryptophan was determined indirectly by competitive binding with L-tryptophan and with octanoate.
The data for the fragment, as well as those for defatted half-cystinyl-BPA are shown in Figs. 11 and 12. The data were analyzed according to the equations given under "Experimental Procedure," with the assumption that there is a maximum of two classes of binding sites present in the protein.
The number of sites in each class and their apparent binding constants which would best fit the data are listed in Table VIII. These values were used to calculate the curves shown in the figures.
Binding studies were also carried out with Fragment C. The

DISCUSSION
On cyanogen bromide cleavage of BPA, two fragments, N and C, were formed.
After reduction of the crosslinking disulfide bonds of the fragments, five peptides were isolated in over-all yields of 45 to 60%. These peptides with NHz-terminal residues of aspartic acid, alanine, arginine, proline, and glutamic acid contain 88, 98, 211, 148, and 34 amino acid residues, respectively.
The five peptides together account for the molecule of BPA with 580 amino acid residues. The molecular weight calculated from this composition is 65,500. This value is slightly less than the value of about 67,000 determined by physical techniques (4), but the difference is well within the experimental error of the methods used.
It was possible to position three of the five peptides in the BPA molecule from the results of end group analyses and by the isolation of an incompletely cleaved Fragment N. The positions of these three peptides, the aspartyl, the alanyl, and the glutamyl peptides, are shown in Fig. 10. The alignment of the remaining two peptides was made possible from studies of a tryptic fragment of BPA.
This tryptic fragment, T-23, containing 2 methionyl residues, was derived from the COOH-terminal two-thirds of BPA, just as Fragment C was. The data on the relative positions of the three cyanogen bromide peptides of Fragment T-23 provided the necessary information to align the two remaining arginyl and prolyl peptides in BPA (Fig. 10). This unique arrangement of the five cyanogen bromide peptides of BPA is consistent only with the interpretation that BPA is a molecule of a single polypeptide chain.
BPA has a single cysteine residue and 34 half-cystine residues forming 17 disulfide bonds. The present work has located the single cysteine residue in the NHz-terminal region of the molecule, Residues 1 to 88. Almost half of the sequence in this region is already known.
The structure of a nonapeptide containing the cysteine residue of BPA was determined by Witter and Tuppy (35), and an identical structure was found for that from human albumin.
The sequences of Residues 1 to 24 of bovine, rat, and human albumins were elucidated by Bradshaw and Peters (36), and extensive similarities were observed.
Of the 17 disulfide bonds of BPA, 5 are located in the Fragment N region and 12 in the Fragment C region.
On the basis of the findings of other workers, the disulfide bonds in the Fragment C region may be divided further into three separate sets as given in Fig. 10. The presence of two disulfide bonds in the COOH-terminal one-fifth of BPA molecule follows from the work of Peters and Hawn (3) who isolated a 77-residue peptic fragment corresponding to this region of the molecule, and from our own observation that the glutamyl peptide is released from Fragment C only after reduction.
The presence of two sets of five disulfide bonds in the central region of BPA can be deduced from the work of Pederson and Foster (37). These workers found that subtilisin cleaved a detergent-complexed BPA into two half-molecules of 270 and 320 residues having eight and nine disulfide bonds, respectively. This distribution of the disulfide bonds in four separate regions of BPA is in accord with its physical property of conformational flexibility in acid and in alkaline solutions (38,39). It is also in accord with proposed models of BPA consisting of compact regions linked by flexible parts as indicated by the low angle x-ray diffraction studies (40), as well as by the isolation of large fragments on limited proteolysis of albumin with pepsin (41), trypsin (5,6), and chymotrypsin (42,43).
The homogeneity of BPA is still an unresolved issue, since a number of workers have reported the heterogeneity of BPA by chromatographies on hydroxylapatite (44), DEAE-cellulose (45), and DEAE-Sephadex (46) as well as by solubility studies (47,48). The data on the compositions of the cyanogen bromide peptides of BPA do not help to settle the issue of whether or not heterogeneity of albumin has a structural origin because of the large sizes of two of the five peptides.
However, the present work does offer an orderly approach to the study of this problem.
A comparison of the organic ligand-binding properties of the tryptic Fragment T-23 and defatted half-cystinyl-BPA showed that they have similar properties.
The fragment and halfcystinyl-BPA both had one primary binding site for the three ligands tested. The primary binding constant of the fragment for each ligand was about one-third of that of half-cystinyl-BPA. The primary binding constants in Table VIII are estimated to have an error of about 20% with the exception of the value for octanoate binding by half-cystinyl-BPA. This value may have an error of as much as 500/, due to the scattering of data in the region of 7 less than 1 (lower graph of Fig. 11). n-Tryptophan competitively displaced L-tryptophan and octanoate from the primary site of the fragment, as well as from that of halfcystinyl-BPA.
This suggests that all three ligands bind at the same site. The ratio of the primary binding constants of L-and n-tryptophan is 30 for the fragment and 22 for half-cystinyl-BPA, and the ratio of the primary binding constants of octanoate and n-tryptophan is 6 for the fragment and 7.6 for half-cystinyl-BPA. These ratios for the fragment and for half-cystinyl-BPA are nearly identical within the experimental error. Therefore, these similarities suggest strongly that the primary binding sites of the fragment and half-cystinyl-BPA share common steric and structural features, thus implying that the primary binding site of the fragment is the same one present in halfcystinyl-BPA.
A notable difference between the fragment and half-cystinyl-BPA is the contribution of the secondary sites to the over-all binding process. The secondary sites of the fragment made only a small contribution to the over-all binding of octanoate, and were apparently absent for the binding of n-tryptophan. This is not the case for half-cystinyl-BPA, since the secondary sites contributed significantly to the over-all binding of octanoate or n-tryptophan.
The number of secondary sites and their binding constants listed in Table VIII are only approximate due to insufficient data for an accurate extrapolation.
The value of the primary binding constant of half-cystinyl-BPA for n-tryptophan is about 50% higher than that originally determined by McMenamy and Oncley (33) in 0.05 M Tris-HCl containing 0.10 M NaCl (pH 7.75). However, they carried out the binding studies with undefatted BPA, and the single sulfhy-dry1 group of BPA was not blocked, as is the case in the present work. To eliminate these variables, binding studies were carried out with undefatted BPA which had been freed of polymers by chromatography on Sephadex G-100. No difference was observed between the binding properties of undefatted BPA and defatted half-cystinyl-BPA.
Fairclough and Fruton (34) have also studied the binding of L-and n-tryptophan by BPA. Since their studies were carried out at a different pH from that used in the present work, it is not feasible to compare their values directly with the present ones. However, it is possible to compare the ratio of their binding constants, 21, with the present value which is 22. Teresi and Luck (32) have studied the octanoate binding by BPA in phosphate buffer at pH 7.6. They interpreted their binding data to indicate that BPA contained 4.2 primary sites with binding constants of 0.65 x lo* and 27 secondary sites with binding constants of 0.01 x lo*. These values differ significantly from the present ones. The discrepancy is not related to the fact that defatted half-cystinyl-BPA was used in the present work, since experiments with nondefatted BPA yielded the same result. The discrepancy may be a result of the different buffer anions used in the two studies.
The decrease in the primary binding constant of the tryptic fragment may be a result of one factor or a combination of several factors. (a) The fragment has an altered conformation as compared to its state in half-cystinyl-BPA so that it binds the ligand less efficiently. (b) The fragment has several conformational states, of which one is favored for binding. (c) The fragment has lost a portion of the primary binding site of half-cystinyl-BPA.
At present, there is no evidence to support the first two factors, but there is suggestive evidence for the third.
For simplification, the primary binding site of half-cystinyl-BPA may be considered to consist of two parts, one interacting with the hydrophobic portion of the ligand, and the other interacting with the anionic portion of the ligand. This consideration is indicated by the observations that BPA binds n-tryptophan methyl ester less tightly than it binds n-tryptophan (33), and that acetyl L-tryptophan amide is bound less tightly than acetyl-n-tryptophan (34). Since the fragment has the same steric and structural features for its binding site as half-cystinyl-BPA does, this will suggest that the two hydrophobic binding sites are the same. If that is the case, the decrease in the primary binding constant of the fragment may be due to the loss of the anionic binding site. The resistance of nondefatted halfcystinyl-BPA toward tryptic digestion may be interpreted to support the present hypothesis, that the stabilization is due to the ligand bridging the different segments of half-cystinyl-BPA molecule. It should be possible to test this hypothesis by comparing the binding properties of the fragment for L-tryptophan and its methyl ester.