Cyanogen bromide fragments of human serum albumin.

Abstract CNBr cleaves nonreduced human serum albumin into three large fragments, A, B, and C, which account for the total amino acid composition of albumin. Reduction and carboxamidomethylation of the free —SH groups produce four subfragments from A, identified according to their NH2-terminal amino acid residues as A-ProI (32 residues), A-AsxI (38 residues), A-ProII (109 residues), and A-Phe (92 residues). A small residue which contains no homoserine is unaccounted for in this sum of the subfragments in A. A-AsxI contains the COOH-terminal leucine of albumin. Reduction and carboxamidomethylation of fragment B produce two further subfragments, B-Ala (36 residues) and B-AspII (89 residues). These subfractions account for all amino acid residues in B. B-AspII contains the NH2-terminal Asp and free —SH group of albumin. Fragment C has only one peptide chain (164 residues) with an NH2-terminal Cys residue. Heterogeneity in disulfide linkages is evident in some preparations.

A small residue which contains no homoserine is unaccounted for in this sum of the subfragments in A. A-AsxI contains the COOH-terminal leucine of albumin.
Reduction and carboxamidomethylation of fragment B produce two further subfragments, B-Ala (36 residues) and B-Asp11 (89 residues). These subfractions account for all amino acid residues in B. B-Asp11 contains the NHz-terminal Asp and free -SH group of albumin. Fragment C has only one peptide chain (164 residues) with an NH2-terminal Cys residue. Heterogeneity in disulfide linkages is evident in some preparations.
Other than the determination of the sequence of the 24 amino acid residues adjacent to the NHs-terminal end by Bradshaw and Peters (1) and the determination of the sequence of 7 amino acid residues in the tryptic peptide containing the tryptophan residue by Swaney and Klotz (2), there appears to have been no organized structural studies undertaken on HSA. ' A major obstacle to structural studies of this protein has been the lack of an established fragmentation procedure. This latter has posed a number of problems since albumin contains 17 disulfide groups and has only one primary chain. This communication reports methods for cleavage and isolation of the seven CNBr fragments of HSA. The investigation was further supported by National Science Foundation Research Grants GB7224 and GB8101. A preliminary report was given at the 14th Annual Meeting, Biophysical Society, February 1970. 1 The abbreviations used are: HSA, human serum albumin; GuHCl-guanidine hydrochloride; SDS, sodium dodecyl sulfate; dansyl, 1-dimethylaminonaphthalene-5-sulfonylchloride. The CNBr fragments are named according to the letter designation of the major fragment from which they are derived, followed by their NHz-terminal amino acid residue.
King and Spencer (3) earlier reported success in initial fragmentation of bovine serum albumin by reaction of CNBr with nonreduced albumin and separation of the cleavage products into two major fragments by passage over Sephadex G-100.
1Ye have found it advantageous also to initia.llp react CNBr with nonreduced HSA. By use of Sephadex G-100 and carboxgmethylcellulose chromatography, three (rather than two) major fragments were resolved from the CNl3r-treated, nonreduced HSA preparation.
By reduction and blockage of the freed -SH groups in the major fragments and various resin column techniques, seven CNBr peptides of whole albumin have been isolated. Since the isolation procedures must, be closely followed for good resolutions, they are described in detail.

EXPERIMENTAL PROClWUHE
A number of sources of HSA were investigated for the cleavage studies. Albumin obtained from fresh plasma by modification of the methods of Kendall (4) and Janatova, Fuller, and Hunter (5) was by far the best preparation for the cleavage studies. All steps in these procedures were cnrricd out at room temperature. A crude preparation of albumin was first, obtained by removing from 125 ml of plasma the protein soluble in 45y0 saturated (NH&Sod, pH 6.5, but precipitated by 75% saturated (NH&-Sod, pH 4.5. ilfter centril'ugation the precipitate was removed, placed on a pad of filter paper for approsimately 30 min in order to remove excess moisture, scraped off, and solubilized by addition of a small amount of water and by adjusting the pH to 5.5 (total volume ~40 ml). The solution was passed through Sephadex G-100 (Pharmacia) in 0.05 M NaCl, column size 5 x 220 cm, flow rate 40 to 60 ml per hour.
The albumin fraction appeared in the elution volume of 1000 to 1400 ml. It followed a turbid region of high lipid content and was readily visible through the yellow color of protein-bound serum bilirubin. The albumin tubes were pooled; the pH of t,he solution was raised to 7.5 by addition of 1 N NaOH; 25 mg of iodoacetamide were added; the solution was allowed to stand for 1 hour, deionized through a column (2 X 40 cm) of mixed resins at a flow rate of -15 ml per hour (6), placed on a column (5 x 7 cm) of DEAE-cellulose (Whatman DE 52), and eluted at a flow rate of 300 to 400 ml per hour with a linear gradient of 0.02 M P04(Na+), pH 7.0 (3 liters), against 0.02 M POb(Na+), 0.1 M NaCl, pH 7.0 (3 liters).
The albumin fraction eluted between 500 and 5000 ml. The cut showing no impurities by rellulose acetate electrophoresis in pH 8.6 barbiturate buffer (characteristically in the elution position of 700 to 3500 ml) was isolated, adjusted to pH 5.5 by addition of 1 N HCl, freeze-dried, dissolved in the minimum amount of water to obtain solubility, desalted in 50-ml 12. H. McAlenamy, H. M. Dintxis, and F. Watson bat~chcs at a flow rate h 15 ml per min through a column (4 X 50 cm) of Sephsdes G-25 (course) in water, freeze-dried, and stored (yield -3 g).
Several lots of Mauri c:llrolnatographically isolated albumin, an albumin Fraction V preparation from Nutritional Biochemicals, and crystalline mcrcuptalbumin (Pentexj were also used in initial cleavage studies.
In a f&w studies thr --SlI group on albumin was not blocked by reaction with iodoacaetumide prior to CNBr treatment.
However, no differences in fragment resolutions were recognized in these untreated preparnt,ions.
Ch'Br was purified by sublimation. Formic acid, technical grade (97%: +), was purified by fractional crystallization according to the general technique described by \Vhitaker (7). In this technique, a 3-kg bottle of formic acid was slowly magnetically stirred at -20".
After 16 to 24 hours the bott,le was remo\-cd t,o a 4" environment and inverted, and the unfrozen solution was allowed to drain for several hours. The solid formic acid was thawed and the entire freezing, thawing, and draining steps were repeated two more times.
The melted solution was then placed at 6-10" without stirring.
The material which did not solidify in several days was further drained (avoiding supercooling).
A 50% yield of solid formic acid with a sharp melt,ing point was obtained. GuHCl (7 M) was prepared from twice crystallized guanidinium carbonate (250 g of the latter material were crystallized from 500 ml of water and 500 ml of methanol).
The GuHCl solution was prepared shortly before use by neutralization of the guanidinium carbonate in the approximate proportions of 9 ml of concentrated HCl to 10 g of guanidinium carbonate. The concentrated urea solutions (-8 M) were prepared by passage of a 10 M urea solution over a deionizing column (6). The urea solutions were stored frozen at -20" until ready for use.
The buffer solutions for gradients at, pll 2.0 and 2.7 were prepared from 1.0 M phosphate stock solutions.
Buffer solutions at pH 9.5 and 10.0 were prepared from 0.5 M borate and phosphate stock solutions, respectively.' The cation was always Naf and t,he pH listed was that of the concentrated stock solution.
CNBr Reaction-One gram of albumin isolated from fresh plasma was dissolved in 4 ml of HrO and 16 ml of formic acid were added. One gram of CNBr was added and the reaction was allowed to proceed for 20 to 24 hours at 4'. All further experiments were conducted at, room temperat,ure.
The solution was passed through :I, column (4 x 40 cm) of Sephadex G-25 (coarse) in 1% propionic acid, t,he protein was collected, and the solution was concentrated to --IO ml in a Diaflo apparatus (U&I 10 membrane ulbrafilter, Amicon Corporation) at a pressure of 50 p.s.i. The concentrated solution was fractionated on a Sephndex G-100 column as described in Fig. 1. Zone I from this column, diluted 1 : 1 with water and the pH raised to 3.2, was pljced on a carboxymethylcellulose column (Whatman CM 32) and eluted acacording to the conditions described in Fig.  2. Zone II was treated in a manner identical with Zone I except that the starting gradient contained 0.05 M NaCl in lieu of 0.075 M NaCl (Fig. 3).
Zones A, B, and C (obtained frorn t,he eluates described in Figs was not carried out until the reduction of the fragments could immediately follow. Reduction and Separation of Fragments-After the freezedrying, the A concentrate residue was dissolved in -5 ml of 7 M GuHCl and transferred to a reaction container, the gas phase was changed to Nz, 200 mg of dithiothreitol dissolved in 2 ml of 7 M GuHCl were added, the pH was adjusted to 10.0 by addition of 6 N NaOH, and the reaction was held under Nz at room temperature for 3 hours. The pH was adjusted to 8.5 by addition of 6 N HCl and 0.8 g of iodoacetamide dissolved in 3 ml of 7 M GuHCI solution was added. The reaction chamber was covered with aluminum foil and the mixture was held at pH 8.5 under NS for 1; hours. The pH was reduced to 2 to 2.5 by addition of 1 N HCl and the solution was immediately passed through a Sephadex G-25 column ( 2.0, at a Slow rate of N 10 ml per min. The peptide fraction was collected; the pH was raised to 3.1 by addition of 1 N NaOH; and the solution was diluted to 1:2 with water, placed on a cnrboxymethylcellulose column (Whatman CM 32)) and eluted according to the conditions given in Fig. 4. The A-ProI, A-AsxI, A-ProII, and A-Phe zones from this elution were adjusted to pH 4 to 4.5 by addition of 1 N NaOH and freeze-dried.
The dried residues were dissolved in 5 to 15 ml of 8 M urea; the solu-tions were passed through a column (2 X 90 cm) of Sephadex G-25 (coarse) in 1 y0 propionic acid at a flow rate N 10 ml per min, and the salt-free peptide fractions were collected and freezedried.
The B fragment concentrate was freeze-dried, dissolved in GuHCl solution, and reduced in the same manner as Fragment A, except that the pH for the reduction step was held at 9.3 and only 100 mg of dithiothreitol and 400 mg of iodoacetamide were used. The solution following reduction was passed through a column (2 x 90 cm) of Sephadex G-25 (coarse) in 0.01 M BOI (Na+), pH -10.0, at a flow rate -10 ml per min. The peptide fraction was collected, diluted 1: 1 with water, placed on a DEAEcellulose column (Whatman DE 52)) and eluted according to the conditions described in Fig. 5. The B-Ala fraction from the column was adjusted to pH 4.5 by addition of 1 N NaOH, freezedried, dissolved in 8 M urea, and desalted in the same manner as the fragments from the A reduction.
The B-Asp11 fraction was adjusted to pH 4 to 4.5 by addition of 1 N NaOH and freezedried, 15 ml of water were added to the residue, the pH was raised to -8 by addition of 1 N NaOH to obtain a clear solution, and the solution was desalted at a flow rate of -10 ml per min over a column (2 X 90 cm) of Sephadex G-25 (coarse) in water and the freeze-dried.
Alternately, the reduced B fragment was fractionated under conditions identical with those described except that a gradient at pH 10.0, consisting of 0.005 M PO* against 0.02 tin P04, 0.2 M NaCl was used.
Fragment C was reduced and passed through Sephadex G-25 at pH 2.0 in the same manner as A. The reduced C fragment was then placed on a carboxymethylcellulose column and eluted according to the conditions given in Fig. 6. The C-Cys fraction obtained from this elution was adjusted to pH 4 to 4.5 by addition of 1 N NaOH, freeze-dried, dissolved in 8 AI urea, and desalted over Sephades G-25 in the same manner as the fragments from A reduction.
NHz-terminal Groups-They were determined by reaction with 1 -dimethylaminonaphthalene-5-sulfonylchloride (dansylation). To be as certain as possible that no groups were missed by improper dansylation conditions, dansylation of all fragments was carried out by two methods; one sample in the presence and the other in the absence of urea solution, as described in the procedures of Gros and Labrousse (8). The small fragments tended to be too soluble in urea solutions and did not precipitate on addition of trichloroacetic acid, resulting in their loss. On the other hand, the large fragments tended not to react well in the absence of urea. Reacting the same fragment under both conditions reduced chances of missing the end groups. The identification of the dansylated amino acids was performed by high voltage electrophoresis as described by Gray (9). Standards and samples were run at three concentrations; 1,2, and 4 nmoles. All questionable zones on the electrophoresis papers were cut out and their identity was made positive by re-electrophoresis in a second buffer. The possible presence of an end group unreactive with dansyl-chloride (e.g. the pyrrolidonecarboxyl group) could not be ruled out, although no fragments were isolated without identifiable end groups.
Amino Acid Analyses-Amino acid analyses were conducted with a Beckman 120B or 120C analyzer. For hydrolyzed samples the recovery of serine was taken as 0.95 and threonine as 0.98. Homoserine lactone values were added to the homoserine values. A value of 0.65 leucine eq was taken for the lactone peak (10). The lactone value was always small in No methio-weight of 15,000 to 20,000. All three fragments, A, 13, and C, nine, methionine sulfoxide, or methionine sulfone was found in move as single zones on cellulose acetate electrophoresis at pH amino acid analysis of unfractionated CNBr-treated albumin, 8.6. These fragments are soluble in water and in some salt or, for that matter, in the analysis of any CNBr fragments.
solutions throughout the pH range of 2 tr, 11. The aggregate In developing systems for separation of the fragments, the fraction and the HSA monomer fractions (Figs. 1 and 2) have initial screening for fragment purity was conducted by deter-essentially the same amino acid composition as whole albumin. mining the NHz-terminal groups present. In electrophoresis These fractions contained all of the NHz-terminal amino acid of the dansylated products as described, there was usually no residues found in the separate fragments, A, B, and C, and were difficulty in visualizing zones at I-nmole application.
Applica-clearly not subfractions. The aggregation of the fractions could tions of 2 and 4 nmoles were made primarily for determining in be due to any of several factors: the cyanogen bromide reaction initial studies the presence of contaminating fragments, or may have failed to cleave the peptide chain (15,16), disulfide providing evidence of incomplete reaction.
The seven CNBr interchange leading to cross-linkages among the fragments may fragments, separated as described, showed no NHz-terminal have occurred, or the lengthy peptide chains may simply have groups other than those reported, and the intensities of the zones become entangled in some manner. It is known that albumin were roughly in accord with the amount of peptide present.
itself, when placed in solution at low pH values, forms irreversible The amino acid analyses of the three large fragments, A, B, aggregates involving disulfide bonds as well as nondisulfide and C, are shown in Table I. These fragments account fairly bonds (5). We have not further investigated the aggregates. well for the total amino acids in albumin.
Fragment A contains Table II gives the analyses of the reduced subfragments found  in A: the A-ProI, A-AsxI, A-ProII, and A-Phe fragments. The sum of the amino acid residues in the four subfragments is somewhat less than that obtained from the analysis of A. Within the variability of the analyses, amino acids not totally accounted for are Arg, Asp, Glu, Gly, Ala, and Leu. This shortage suggests t.he loss of a small fragment after the reduction step. The lost fragment would contain no homoserine since the latter has been accounted for in the other fragments isolated.
The matter is being further explored by the use of other techniques.
Traces of glycine (up to 0.4 residue) were sometimes present in the A-Pro11 fragment.
A second reaction of this fragment with either CNBr or performic acid, followed by rechromatography on carboxymethylcellulose, did not change the level of glycine (or other amino acids) in the fragment recovered. We suspect that this trace level of glycine is related to the missing amino acid residues in A, described above, one of which is glycine. The A-Pro11 fragment in polyamide SDS electrophoresis sometimes showed a weak but easily dist,inguishable band trailing the major band. There were indications that the presence of this second band depended on the presence of glycine in the A-Pro11 analysis. Table III gives the amino acid analyses of fragments B-Ala and B-AspII, which were obtained on reduction of Fraction B. The sum of the amino acids in the two subfragments agrees well with the amino acid analysis of the whole B fragment. Also compared in this table are the amino acid analyses of C-Cys and the unreduced C fragment.
The good agreement between these latter two amino acid analyses is evidence that reduction has not further cleaved the C fragment.
With albumins from fresh plasma, Fragments A, B, and C were obtained in -75% yield.
The yields of the smaller fragments ranged for 20% to 500/,.
After coverage of the free --SII group in fresh plasma albumin with a '4C-carboxamidomethyl group, the amount of radioactivity in the separate fragments showed that the -SH group is totally in the B-Asp11 fragment.
The B-Asp11 fragment is also from the NHz-terminal end of whole albumin. The solution containing this fragment turns blue on addition of Cu++, a reaction which, as has been shown by Shearer ef al. (17), occurs only in the presence of the NIIz-terminal end of albumin. The other fragment with NH*-terminal AXsxI contains no COOH-terminal homoserine and is clearly identified as a fragment at the COOHterminal end of albumin.
The B fraction elution profile from the carboxymethylcellulose column sometimes showed distortions suggesting the presence of more than one component.
We suspect that this was due to the attachment of different radicals to the -SH group in this fragment: carboxamidomethyl (which we add), cysteine, glutathionine, or possible other adducts (18) might well be present in mixtures at this site. Partial format'ion of a disulfide with cysteine was confirmed by the elution of free dansyl-cysteic acid from performic acid-oxidized, but unhydrolyzed, dansylated B fragment.
Free dansyl-cysteic acid was also found when whole albumin was similarly treated.
The profile for the reduced U preparation nearly always showed two (sometimes three) fragments wit,h KHz-terminal alanine (Fig. 5). These fragments gave the same amino acid analyses and the same end groups, i.e. Ala. It suggests that the differ-0 0 0 ences in mobility are probably due to differences in the amide content of the fragments. The most troublesome separations were those required for the purification of Fragments A-Pro11 and A-Phe. When evaluated by SDS polyamide electrophoresis these fragments were obtained pure when albumin isolated from fresh plasma was used for the cleavage studies, and then only by following the precise steps indicated.
It was found to be very important that the pH of A fragment solution be adjusted to 4.0 to 4.5 before the solution was freeze-dried.
It was also important to prevent the exposure of the dried A residue to air for any length of time. For this reason the A fragment was reduced immediately after it was taken off the lyophiliaer.
The commercial albumin preparations after CNBr reaction gave elution profiles which frequently were different from those of fresh plasma albumin.
Thus, after CNBr cleavage of the Nutritional Biochemicals preparation, a prominent peak was consistently found between Peaks I and II (Fig. 1). Upon purification of the fragment in this peak, it was found by end group and amino acid analyses to have the composition of Fragment B. The position of elution of this peak suggested that it was dimerized B fragment, presumably with the attachment via the -SH groups.
With most commercial albumins a small peak was found between the A-AsxI and A-Pro11 zones in the elution profile of the reduced A fragment.
This zone was identified as the B-Ala fragment by amino acid analysis and end group analy-sis. In albumin from fresh plasma this zone was found only in fraction B. With the CNBr-treated crystalline mercaptalbumin preparation, traces of A-Pro1 and 8-AsxI fragments were found in the elution of the reduced C fragment from the carboxymethylcellulose column.
Again, this mixing of small fragments among the major fragments was never found in the albumin preparations isolated from fresh plasma.
The implication here is that analogous disulfide interchange has occurred in albumin preparations from various commercial sources. Disulfide interchange has been implicated as the primary cause of at least a part of the microheterogeneous behavior of albumin (19). Little is gained by preliminary purification of the commercial albumins.
Almost all of the aggregates in such preparations appear as an aggregate in the elution of the CNBr fragments from Sephadex G-100.
In preparations in which the aggregat.e content was high, however, the B fragment was reduced in yield more so than the C fragment,. This is consistent with the B fragment, which contains a free -SH group, being more involved in aggregate formation than the C fragment.