The essential covalent structure of human fibrinogen evinced by analysis of derivatives formed during plasmic hydrolysis.

Abstract In an effort to determine the essential structural features of human fibrinogen, plasmic hydrolysis was carried out and degradation products were identified and analyzed. In agreement with previous work, the initial attack was shown to occur in the Aα chains with the release of portions of their COOH-terminal regions (designated /α chains in the terminology developed to describe the present findings), whereas simultaneous, but slower, degradation occurs in the NH2-terminal region of the Bβ/ chains) and results in the removal of remnants (designated Bβ/ chains) containing peptide B. These changes, plus internal cleavages leading eventually to the separation of Fragments D and E (terminology of Nussenzweig et al.), produce a series of high molecular weight derivatives which can be separated by polyacrylamide gel electrophoresis of unreduced samples in sodium dodecyl sulfate. (These were designated Bands II through VII in order of decreasing molecular size; Band VII was identified as Fragment X in the terminology of Marder et al.) Upon further hydrolysis derivative Band VII can lose the COOH-terminal portion of one of its Bβ chains (as remnant /β6, molecular weight 32,000) to form derivative Band VIII (Fragment Y in the terminology of Marder et al.) or, alternatively, can undergo cleavage to yield early forms of Fragments D and E (designated D1 and E1, respectively). Derivative Band VIII, in turn, is degraded to Fragment E1 plus a later form of Fragment D. Once cleaved from the parent molecule, these Fragments D continue to release /β chains. These degradative pathways lead to a series of Fragments D (D1 through D5) which differ in molecular weight. Additional cleavages occur in Fragment E1 to form a smaller fragment (E2) which, unlike E1, lacks peptide A. Immunochemical studies of derivatives obtained at successive phases of digestion demonstrated the presence of at least five antigenic determinants on the fibrinogen molecule. One of these is located on /α chains; another, on Fragment E; three others, on Fragment D. Of the latter, one is associated with /β6 chains; a second (designated F) is associated with Fragments F, which arise by further degradation of Fragments D; the third (designated D) is that lost when Fragment F is evolved. Electrophoretic behavior of plasmic derivatives before and after reduction revealed at least one intrachain disulfide bridge in each /α15 and /β6 chain. Studies of Fragment D subunit composition and recovery at various phases of digestion indicated that it is a dimeric structure containing substantial portions of the COOH-terminal region of both γ chains (as /γ1 remnant chains, each having a molecular weight of 42,000) and hence that only one such fragment (linked by at least five interchain disulfide bridges) can be generated by each fibrinogen molecule. These results, plus those reported by others, led to the conclusion that fibrinogen itself has a dimeric structure, the "backbone" of which consists of a pair of γ chains linked directly to each other by disulfide bridging in the NH2-terminal region and covalently linked (enter directly or indirectly) with the Aα and Bβ chains in at least two regions of the molecule.

can be generated by each fibrinogen molecule. These results, plus those reported by others, led to the conclusion that fibrinogen itself has a dimeric structure, the "backbone" of which consists of a pair of y chains linked directly to each other by disulfide bridging in the NH,-terminal region and covalently linked (enter directly or indirectly) with the Aa and BP chains in at least two regions of the molecule.

Endopeptidases
have been of great value in structural studies of a number of proteins (I).
In the case of plasma fibrinogen such an approach employing plasmin-catalyzed hydrolysis was used to establish the presence of at least two antigenically distinct regions, named D and E, in the protein "core" (2-4).
The antigcnicity of these plasmic fragments, which were chromatographically separable from one another, was preserved even after prolonged enzymic degradation. These discoveries provided both a stimulus and a rational basis for additional esperimentation.
More recent estimates indicate a molecular x-eight in the range of 49,000 to 60,000 (14,15), and, coupled with other structural studies (ll- 13), strongly indicate that the N&terminal regions of all six chains are involved in its structure. Plasmic core Fragment E has been conclusively shown to be dcrivcd from the same region of the molecule as the N-DSK. Illartier et al. (14) and others (16) have demonstrated an antigcnic identity between N-DSK and Fragment E, which itself is antigenically unique with respect to other regions of the molecule (3,17, 18 inter al&x). Furthermore Mills (19), employing sodium dodecyl sulfate electrophoretic analyses of Fragment E from relatively early digests, and Rlarder et al. (14), using NHnterminal analysis, have demonstrated the presence of thrombin susceptible bonds, probably due to the prescncc of peptide A.
Information regarding the structure of the remainder of the molecule is indirect and far less conclusive than that cited above for the N&terminal region. Two different hypotheses, both based upon analyses of plasmic digests, can be proposed.
The first, consistent with the struct'ural symmetry indicat'ed by electric birefringcnce mcasurcments on fibrinogen (20, al), suggests that the COOH-terminal regions of fibrinogen chains, like those in its NH*-terminal region, are covalcntly linked to form a single dimeric structure. Support for this may be drawn from the data of Fletc+er and his co-workers (I 7, 22)) who sl~owctl that, a large variety of intermediate core spcxcies WB formed during the degradation. Among them were scvcral species immunochemically identifiable as Fragment D ; these rangctl in size from 150,000 or more to the final plasmin resistant core fragment of 88,000. Since Fragment D is derived from the COOH-terminal region of the molecule, tho relatively high molecular weight of its early forms supports the "dimeric" hypothesis because only a single structure of this size can be obt,ained from 1 molecule of fibrinogen.
Though the data of these investigators has been disputed (18), and few subsequent studies have been directly interpreted as supporting such a model (23) As a portrayal of the COOI3-terminal region of the fibrinogen molecule the "dimcric" and "monomeric" models are not corn-patihlc.
The studies to be reported J\-ere designed to test these models critically by studying the sequence of degradation catalyzed by plasmin. TERMINOLOGY The degradation of fibrinogen by plasmin has been divided into Stages 1, 2, and 3 according to the terminology introduced by Rlnrder and co-workers (18). Although there is con~itlerablc overlapping in the transition from one "stage" to another, each phase can be distinguished from the others by virtue of the products present in the digest. Stage 1 is characterized by the presence of coagulable species; Stage 3, by that of core fragments containing either D or E antigenie determinants, but not both.
In Stage 2 thcrc exist core species with botll D and E antigenic determinants, but no (or relatively few) coagulable species.
The subfractions (comprisin g core derivatives) isolated from Stage 1 digests of Fraction I-4 have been designated in a manner related to that employed for plasma fibrinogen Subfractions I-8 and I-9 (37), which consist primarily of circulating fibrinogen catn.bolitcs (38). They arc identifiable by their relatively high solubilit,irs and delayed aggregation rates (39).
To differentiate an i?z vitro plasmic derivative subfraction from the fraction of corresponding solubility isolated from plasma, the suffix "D" is added to the name of the former, and the clottnbility of the digest mixture from which it was isolated is shown as a superscript.
For example, a derivative subfraction essentially equivalent in solubility l;o plasma Fraction I-9, but obtained from a digest whose clottability had fallen to 88y0 at the time of sanpling, is designated pertains to densitometric scans. The relative scan area of a given region of the gel, normalized for the amount of total sample applied, is placed to the right of the gel (gel and immunoelectrophoretic analysis of 89% clottable sample not shown). l . * e l , Gels 7 to 9 were analyzed in was dissolved to a concentration of 3 to 8 mg per ml in a ureaphosphate-sodium dodecyl sulfate buffer (8 M, 10 mM, 1 %, pH 7) and reduced with DTT (14 mM).
An alternate labeling method, which did not require DTT reduction immediately prior to electrophoresis, consisted of dissolving the protein in 8.4 M guanidine. HCl buffered with N-ethylmorpholine at pH 8.5, reducing with DTT and subsequently adding dansyl chloride.
(The amount of dansyl chloride added was equal to the calculated amount for the protein plus an amount equivalent to the quantity of DTT in the mixture.) Sodium dodecyl sulfate electrophoresis was carried out (typically 24 gels per run) at sample loads of 150 to 400 fig per gel. After completion of the experiment, the fluorescent bands were identified, sliced out, pooled appropriately, homogenized, eluted with 10 M urea, dialyzed against 2% acetic acid, and freezedried.
The NHp-terminal acids of the resulting material were then determined as described above.

RESULTS
Acrylamide gel electrophoresis of unreduced digest samples in the presence of sodium dodecyl sulfate was a precise and reproducible way of establishing which stage of degradation had been reached.
In Fig. 1 is shown such an analysis (in 9% gels) of serial samplings from plasmic digests.
Gels of this concentration were selected to demonstrate digest components whose annarent molecular size was in the range of 100.000 to __ 10,000. These included derivative chains released from the core (e.g. /(YIP to /air), as well as the smaller core fragments (e.g. Dg, E).
Gels of 5% concentration (cf. Fig. 8) were more suitable for resolution of larger core fragments. Fig. 1 also includes digest clottability, parallel immunoelectrophoretic analyses done before and after heat precipitation, and results of certain gel scanning experiments.
This figure thus serves not only as a primary data source but also as a reference to which other analyses of digest samples will be compared.  at earlier phases than those examined here also have enzymically identifiable Acu/ core remnants. Fragment EQ showed no evidence of peptide A and, in addition, was smaller than El even after the latter had been subjected to reptilase or thrombin treatment. This suggested not only that a fragment containing the A peptide had been released, but also that hydrolytic attack had occurred at other sites (cf. El, Fig.  13). The failure to demonstrate the presence of peptide B enzymically or by radioimmunoassay and the known plasmin resistance of the thrombin-susceptible site on the BP chain (7, 10, 12) indicated that a derivative chain containing peptide B is released from the core at a relatively early phase (BP/13, Table  I; CJ". Fig. 13). Confirmation for this conclusion was obtained from studies of other derivative subfractions like I-9D40 (see below).
Subunit Structure of I-9D40-The elucidation of the degradative sequence, especially as it involved the subunit structure of Fragment D, was facilitated by the knowledge that the Aa! chain is the first to be degraded (31- 34,38,59) and that the hydrolysis results in the release of peptide fragments from the COOHterminal region of this chain (38,39). That is, it was possible to prepare plasmic derivative subfractions whose Acr chain population consisted almost entirely of Aa/ remnants smaller than the y chain, but which still retained some or most of its BP and y chain populations in an intact form. The subfraction selected for study was one which possessed these properties, namely, I-9D from a 40% clottable digest (I-9D40). CM-cellulose chromatography (Fig. 3) of the S-sulfo derivative of I-9D40 yielded an elution profile which was more complex than that obtained for the S-sulfa derivative of the precursor material, Fraction I-4. The "y" chain position (Zone II) was enriched relative to that of the "BP" (Zone IV) and "A&' (Zones  (38). b The mean molecular weight of the /P4, /?I position in I-9D40 was 42,200, whereas that of the /yr position in more advanced Stage 3 digest mixtures was 41,800. Comparison of these means suggested that the differences were not statistically significant (P < 0.3). The molecular weight of the position at all degradative phases has thus been taken as 42,000.
c Migration in sodium dodecyl sulfate gels was the same whether sample was reduced or not.
d Estimated by the difference between the intact chain and the larger derivative chain (i.e. /OS or /yi, respectively). 8 In our previous study (38) the molecular weight of the y chain was probably somewhat overestimated because it was computed as the mean of values obtained by electrophoresis of whole fractions (e.g. I-9) as well as isolated chromatographic peaks (cf. Fig. 3). It seems likely in retrospect that the putative y chain contaminant (Band 5 in those experiments) identified in Aor chain preparations was probably of Bp origin (i.e. /pa) and therefore of somewhat higher molecular weight than y chains. The present value for y chains has thus been recomputed with these factors taken into account.
V and VI) chain positions.
Furthermore, material eluted between the y and BP chain positions (Zone III) did not correspond in position to any peak in the precursor material. Though almost all identifiable ,4cr remnant chains occupied the Aa/ and Aa/ positions (Fig. 3, Gels IO to 18), a very small amount of an Ae! remnant occupying the B@ position in sodium dodecyl sulfate gels (i.e. AQ!/~) was detectable by enzymic analysis of chromatographic Zones V and VI (e.g. Gels 10 to 12). The relative amount of this derivative chain was too low to be detectable in the starting material (Gels 16 to 18). All other derivative chains which were not. reactive with reptilase but were larger than a y chain (namely //?I, /& l/33) were therefore remnants of Each tube contained 4.5 f 0.1 ml. after cross-linking unmodified I-9D40. The /p3 derivative chain In the case of S-sulfo I-4, peaks in the elution pattern are desigwas not found in the sample obtained from the clot, but was evinated according to the chains present, consistent with previous dent in analyses of the whole cross-linked sample or of the clot data and nomenclature (38,60 Fig.  8, Gel 11) suggests that /& also arises by attack in the NH2terminal region (cf. Fig. 13). The /& chain is measurably larger (Table I)   rivative is demonstrated not only by its migration in sodium dodecyl sulfate gels but also by its chromatographic difference from the y chains ( Fig. 3; cf. Gels I to S and 10 to 15) and its failure to undergo cross-linking (not shown). The position designated /&, /?I, BP/5 (Fig. 3) is occupied by at least three types of derivatives from BP or y chains; B/3/6 is readily identifiable by enzymic criteria (Gels 1 to S). Proof of of the y chain origin of l-r1 is developed later; however, the presence of this derivative is shown by the fact that the y chain (except for the cleavage resulting in the removal of its NHzterminal region as part of Fragment E, cf. Fig. 13) is plasminresistant and /?I continues to occupy this position during digestion of other portions of the molecule (see Fig. 8, Gels 11 through 15). By contrast /&, like other B/3 core derivatives, undergoes progressive degradation (see following section and Figs. 4 and  8). (I-4) digest whose clottability had fallen to 88% (a) or 31Tn (6).
The samples were analyzed separately (Gels 1 to 4) or pooled (Gels 5 and 6) before and after reduction with DTT.
The arrow with the question mark indicates an unidentified contaminant (possibly albumin) present in this preparation.

Release of Large Peptide Containing COOH-terminal Region of B/3 Chain (/&-Sodium
dodecyl sulfate gel electrophoretic analysis of unreduced samples from a digest of I-4 ( Fig. 1) showed that large peptides identifiable as Aa! chain remnants were released from the core at an early phase.
The major remnants at this stage (/al5 to /CYST) ranged in molecular weight from 40,500 to 48,700 and accounted, by their size, for the appearance of Aa core remnants like Aa/il and Aac/i? (molecular weight 25,000 and 22,600, respectively, see Table I).
These first appeared after the /a remnants and continued to increase in amount at a time when ACY chains or ACY core remnants of sufficient size to generate such fragments had been consumed.
The possible presence of disulfide bridges in these derivative chains was investigated after they had first been separated from larger thermolabile core fragments by heat precipitation (Fig. 4). The slower migration of derivative chains larger than 32,000 (i.e. /PC or larger) after DTT reduction strongly suggested that these chains contained intrachain bridges, but there was no evidence of interchain bridges The demonstration that the /& and /& chains contained no disulfide bridges (Fig. 4) indicated that the bridge(s) in the /& chain was present in a region which was lost during the formation of /& or /&. Comparison of the molecular weights of these remnants (Table I) permitted the conclusion that the disulfide bridge(s) lies within 100 amino acid residues of the end of the /fir, chain.
Isolation of the //?s chain on a preparative scale was accomplished in the following way. During plasmic hydrolysis of 700 mg of a mixture of I-9D7" and I-9D40, clottability was allowed to fall from 85% (starting mixture) to 62%. Plasmin was then inhibited with Kunitz' pancreatic trypsin inhibitor and the core fragments were removed by heat precipitation.
The supernatant solution was freeze-dried, redissolved, dialyzed against 0.005 M Tris-phosphate buffer, and subjected to gradient elution chromatography on DEAE-cellulose (Fig. 5). Zone IV, containing the //36 chain was then chromatographed (not shown) on Sephadex G-100 to separate it from lower molecular weight fragments (Fig.  5, Gel 7).
Demonstration of the Up chain origin of //36 was as follows. Sequential sodium dodecyl sulfate electrophoretic analyses (analogous to those shown in Fig. 1) were carried out on plasmic digests (not shown) of fractions containing core fragments. Digestion of fractions which were devoid of Aa chains or ACY/ remnants sufficiently large to be the precursor of such a peptide (cf. Fig. 3) nevertheless resulted in the progressive evolution of a chain migrating in the /PC position. This indicated that this derivative was of either Up or y origin. Furthermore, the derivative chain must have been released from the COOH-terminal end of the molecule because the nature of the interchain disulfide bridges in the NHz-terminal region (12) precludes the release of a derivative chain of this size from the NH2-terminal end.
Immunoelectrophoretic experiments (Fig. 1) have demonstrated the appearance during Stages 1 and 2 of thermostable antigenic material.5 The nature of these antigenic species was explored by comparing digest supernatant fractions and purified /fly chains (Fig. 6). All Stage 1 and 2 samples tested formed immunoprecipitates with antifibrinogen serum. The precipitin line 5 In immunoelectrophoretic analyses of sequential digest samples, the precipitin arcs formed by Acu and subsequently by Bp chain remnants released from the core become less intense and finally undetectable (Fig. 1) Fig. 1; the digest clottability at the time of sampling is indicated.
formed by heated digest samples consisting primarily ofi/oc remnants (i.e. 89% to 80 '% digest, Fig. 6) was qualitatively different than that formed by the //3~ chain, though the pattern of later samples suggested that these two antigens were immunochemically related. When these samples were reacted against antifibrinogen serum which had been absorbed with S-sulfo BP chains, a precipitin line still formed against the /a! remnants present in the 89% to 80% digest samples (not shown), but none formed against later digest samples or against the /& chain (cf. Fig. 9). These results showed a qualitative difference between the antigenic sites on /ar and //I chains and indicated that the /OL antigenic site was lost during plasmic hydrolysis. In contrast to the capacity of S-sulfo BP chains to absorb antibody against /&, chains, antifibrinogen serum absorbed with S-sulfo y chains still formed a precipitin line against this derivative (not shown), demonstrating antigenic nonidentity of /& and y chains.
Samples of the /& chain yielded tryptic peptide maps (not shown) very similar to those given by preparations of /& plus /& though some spots observed in //& maps were relatively faint in the latter.
Whereas peptide maps of /&, preparations exhibited similarities to maps of both the 130 and Acr chains, they were distinguishable from those of y chains ( Fig. 7 and Reference 38). The nonidentity of y chains and //& shown by immunodiffusion and peptide mapping experiments was further substantiated by studies of the subunit composition of Fragment D, which are described in the next section.
Plasmin-resistant "Backbone" Structure oj' Fragment D-Immunoelectrophoretic analyses of digest mixtures showed the separation of the D and E antigenic determinants (Fig. 1) and were used to identify core fragments possessing the D antigen and lacking E. Further identification was made on the basis of characteristic chromntographic behavior on DEAE-cellulose (Fig. 11)   The y chain had been isolated by CMcellulosechromatography (cf. Fig. 3). The/r1 derivativeprepared from Fragment D had been separated from smaller chains by chromatography on Sephadex G-100 in the presence of 2 M guanidine 1 HCl.
resolved into a doublet (e.g. Fig. 10, Gel l), but no additional designation was assigned. The larger core species were abundant during earlier phases, whereas the smaller species were predominant in the more advanced phases of Stage 3 ( Figs. 1 and 8).
Sodium dodecyl sulfate gel electrophoresis of reduced samples of Fragment D (Fig. 8) showed a plasmin-resistant derivative chain, molecular weight 42,000, which was identified as a y chain derivative (i.e. /-ri) on the following grounds. First, the known hydrolytic sites of cleavage of Aac and BP chains (see above) preclude the possibility that a plasmin-resistant chain of 42,000 could arise from either Acr or BP chains. Secondly, confirmation of the y chain origin of the derivative was obtained from comparison of tryptic peptide maps of the X-sulfo derivatives of y chains and that of /TX (Fig. 7).
Analysis of reduced Fragment D species (Fig. 8, Gels 11 to 15) showed the absence of derivatives larger than intact y chains. The subunit structure of the most advanced Fragment D species, namely Db (Fig. 8, Gel 15 Fig. 8, Gel 7). The major NHe-terminal residues were aspartic acid, valine, methionine (in agreement with the findings of Marder et al. (14)), and threonine, plus small amounts of serine, alanine, and glycine.
Since sodium dodecyl sulfate gel electrophoresis of reduced, dansyl-labeled samples yielded essentially the same band pattern as did unlabeled, reduced samples, elution and analysis of the fluorescent bands could be used to identify the NHz-terminal acid(s) in any given band. The gels (5% acrylamide) in the upper portion of the figure are of unmodified Fraction I-4 or sequential digest samples thereof.
The samples for the lower gels (9% acrylamide) were obtained by elution of the given band in preparative electrophoretic experiments (Gels 9 to 11), by heat precipitation of digest samples (Gels 12 and 13), by a combination of DEAE-cellulose chromatography and gel chromatography on Sephadex G-100 (Gels 14 and 15), or by these chromatographic procedures after Fig. 1, Gel 6), and the advanced Stage 3 digest having the NHsterminal residues given above. Alanine and small amounts of aspartic acid were identified in the /rl, /PJ position in the Stage 1 and Stage 2 samples.
However, aspartic acid was the sole NHt-terminal residue in this position in the Stage 3 sample. This suggested that the sites of cleavage in the formation of /fly and a precursor of /rr were Lys-Ala or Arg-Ala bonds or both.
The finding of NHz-terminal aspartic acid in this position at advanced phases of Stage 3 indicated that an additional bond (Lys-Asp or Arg-Asp), close to the initial cleavage site, had been broken to form /rl (see Footnote b, Table I).
NHz-terminal residues in other fluorescent regions of the gels were methionine, valine, and threonine, but no (or only a trace of) aspartic acid was detected.
The observation that aspartic acid was the NHz-terminal residue of /yl, but of no other derivative chain in this type of preparation, indicated that quantitative determination of NH%-terminal aspartic acid could be used as a meas-thermal denaturation to remove large core fragments (Gel 16). Gels not obtained from the same electrophoretic experiment were aligned with respect to known bands rather than with respect to the gel origins.
Gels of Bands VII and VIII are aligned with that of a reduced sample of Fraction I-9D50 (containing intact y chains) run in the same experiment (Gels 8 to 10). The recovery of Fragment D determined for the phase of degradation studied is shown above the corresponding gel.
ure of the number of /yr chains present in Fragment D. Two such analyses indicated the presence of 1 mole of aspartic acid for every 36,000 and 34,000 g of protein, respectively.
(In the second analysis bovine albumin, which is known to have 1 mole of NHz-terminal aspartic acid per 67,000 g, was used as a control. The experimentally determined value was 1.0 to 1.5 moles /67,000 g.) The results were thus consistent with the previous conclusion that two /rr derivative chains exist in each Fragment D species.
Formation oj Antigen&z Fragments F from Fragment D-At least one other discrete process was evident from sodium dodecyl sulfate gels of unreduced digest samples. This process was characterized by the formation of 4 to 5 bands, collectively termed Fragments F, which migrated at a greater rate than did Fragment E (Figs. 1, 8, and 11) and which contained interchain disulfide bridges.
These F fragments from an advanced Stage 8 digest were concentrated in a chromatographic peak (Peale F, Fig. 11 with Fragment E was evident from the immunoelectrophoretic pattern (Fig. 1) and from immunodiffusion experiments in which chromatographically isolated Fragments F were compared with other plasmic derivatives (Fig. 9). F formed a line of partial identity with Dh, showing that these fragments possessed at least one common antigenic determinant (Determinant F). This pattern was not altered by absorption of the antiserum with B/3 chains. Thus, neither Determinant F nor the determinant which was unique to Fragment D (Determinant D) was related to that associated with /& (Fig. 9). Unequivocal evidence that Fragment D was the precursor of Fragment F was provided by an experiment in which isolated Fragment D (Fig. 11) was degraded by the addition of plasmin (Fig. 10). The fact that degradation was associated with a progressive decrease in staining intensity of D fragments (namely D4 and DJ suggested that the formation of the F fragments occurred at the expense of Fragment D and that the transition from Dq to Ds was the consequence of another process (i.e. release of B/3 derivative chains). The main subunit chain of reduced Fragment F preparations migrated at the same rate as the /ar/la derivative seen in reduced Fragment D (Fig. 8, Gels 11 to 16), suggesting that they both represent the same chain.
The slower, faintly staining bands (e.g. Fig. 8 arrow, Gel 16) may have arisen from /yl, a possibility which will require confirmation.
Difference between Derivatives VII and VIII-Samples of Derivatives VII and VIII were prepared from Stage 2 digests (e.g. Fig. 8, Gel .2) by elution from sodium dodecyl sulfate gels. Analyses of reduced samples showed, in both casts, considerable depletion of the intact y chain position as well as an absence of to the sample (4 mg of protein per ml, final concentration, buffered at pH 8.6) which was then incubated at 37". After 7 hours a sample was removed, an additional increment of plasmin was added, and the mixture was incubated at 37" for another I6 hours.
intact BP and Aa! chains (Fig. 8, Gels 9 and 10). The y and j& chains did not resolve; nevertheless, the presence of some intact y chains was established by the finding that y dimer, of the size expected from the cross-linking of two intact y chains, was formed in cross-linking experiments (not shown) with unmodified samples.  This implied that the essential difference between VII and VIII was due to the loss of a BP remnant like that previously shown ( Figs. 1 and 4) to be cleaved from the core during this phase of digestion (namely l&J.

Recovery of Fragment D from Plasmic
Digests-The following experiments were designed to examine the basis for the monomeric D hypothesis by determination of the recovery of Fragment D at successive phases of degradation.
During early and intermediate phases of the third stage of degradation of Fraction I-4 (e.g. Gels S to 6, Fig. 8), thermal denaturation resulted in complete precipitation of Fragment D species free of other digest components.
Determination of the recovery of Fragment D from these mixtures could therefore be made spectrophotometrically by comparing the absorbance of the washed, solubilized (in alkaline urea) precipitate fraction obtained from a thermally denatured digest sample with that from an equivalent amount of starting material (i.e. Fraction I-4). Corrections were applied for the absorbance coefficient of I-4 (A:2m, 282 nm = 16.0) and that of a preparation of Fragment D isolated from this phase of degradation (A&,, 282 run = 17.9). Under these conditions, the recovery ranged from SO%, at an early phase of Stage 3 when Derivatives VII and VIII had been virtually all consumed and Dr was the main species of Fragment D (Fig. 8, Gel S), to 48.1% at an intermediate phase of Stage 3 when considerable amounts of smaller species (i.e. Da to DE) had formed (Fig. 8, Gel 6).
At more advanced degradative phases thermal denaturation procedures were inadequate.
Estimation of recovery was therefore made by gradient elution chromatography (Fig. 11) Fig. 11) and by their positions in sodium dodecyl sulfate gels. The absorbance of the chromatographic peaks containing these fragments was used to estimate their apparent recovery (Fig. 11, bracketed numbers).
Subsequently, for each peak, the proportion of absorbance which was not due to core Fragments D and E was determined by gel chromatography on Sephadex G-100 and appropriate correction of the apparent recovery was made. It seems likely that the F fragments which had formed at this phase were also included in the estimated recovery of D (Fig. 11, Gels S and 4). For the peak containing both Fragments D and The immunodiffusion patterns of selected column fractions (indicated by tube number) which had been reacted against antifibrinogen serum are shown as insets adjacent to the DEAE-cellulose chromatogram. Sodium dodecyl sulfate gel (9y0 acrylamide) electrophoresis of the unmodified starting material (Gel 1) and various core fragments (Gels 2 to 6) is shown adjacent to the Sephadex G-100 chromatograms.
The distribution of eluate absorbance is shown between bracketed arrows. E (i.e. Peak G), an additional minor correction was made by visual estimation of the relative distribution of D, E, and F (85'% D plus F, 15% E) in sodium dodecyl sulfate gels (Fig. 11 The diagram is not intended to imply the molar content of any given fragment but rather is intended to indicate the region(s) of the molecule whose cleavage results in the formation of another derivative.
The derivative species shown are those identified in our sodium dodecyl sulfate gel electrophoretic and related experiments (e.g. D1 to Ds, /LY, etc.). For clarity the analogous derivatives X and Y identified by Marder et al. (14) in acidic gels are indicated parenthetically. 7925 and covalently linked (either directly or indirectly) to the Aar and BP chains in at least two regions. The experimental basis and logic employed to derive such conclusions follow.
The present data and that from other studies (31-34, 38, 59) have established that plasmin initially attacks the Aa! chains, removing portions of the COOH-terminal region of these chains (e.g. /ais to jai,) from the core (38,39). Hydrolysis of the BP chains occurs simultaneously, though at an appreciably lower rate (31-34, 38, 59). Shainoff and co-workers (64-66) as well as other investigators (33,61) and our group (38,67) have reported evidence, confirmed by this study, that the early attack on the BP chains is in the NHz-terminal region and results in the release of remnants containing peptide B (e.g. B/3/13, Table I). This is consistent with the amino acid sequence of the NH?terminal portion of the BP chain (7,10,12), which includes several plasmin-susceptible sites. Cleavage at two of these (producing core remnants //31 and /&) has been demonstrated in the present study. The virtual absence of intact BP chains or core remnants containing peptide B in derivative bands like VII and VIII and the absence of detectable amounts of peptide B in all forms of Fragment E (Fig. 2, Footnote 3, and Reference 66) indicate that these cleavages are essentially complete before substantial amounts of Fragment E are formed. Thus, release of /a! and B/3/ derivative chains plus internal hydrolytic attack leading to the eventual separation of Fragments D and E (e.g. those producing core remnants /& and /-yJ constitute the essential steps in the formation of Bands II through VII.2 The present studies have demonstrated other B/3 chain attack which occurs at a still lower rate and results in the release of large derivatives. The amino acid sequence and disulfide bridges of the B/3 chain (12) preclude the possibility of plasmin releasing a peptide with a molecular weight of more than 7,000 from the NHz-terminal region. Thus, release of a single stranded BP derivative chain with molecular weight of 20,000 or more (namely l/36, /&, /Be) demonstrates that these derivatives had Disulfide bridges whose precise location has been reported (12) are indicated by solid lines (-) ; those whose presence has been established by the present studies but whose precise location is still uncertain are indicated by hatched lines (lZ ). Certain of the known cleavage sites are indicated by vertical arrows ( J ) which are numbered to correspond to sites previously established (38)   Ea isolated from more advanced digests lacked this peptide ( Fig. 2 and Footnote 3) confirms the observations of Mills (19) and is consistent with known plasmin-susceptible sites (12, 13). The following results indicated that the backbone of fibrinogen includes both y chains linked in a second region to form a dimeric structure which appears as Fragment D during plasmic hydrolysis.
Tryptic peptide maps of X-sulfo /ri chains prepared from Fragment D and those of intact y chains were very similar (Fig. 7) and were unlike those of B/3 or Acr chains.
Furthermore, the degradative sequence of B/3 and Aac chains elucidated in this study precluded the possibility that the plasmin-resistant chain identified as /?I (Fig. 8) could arise from any chain other than y. Since aspartic acid was shown to be the NH&erminal residue of /yl, but of no other derivative chain present in advanced Stage 3 digests, the finding of 1 mole of NHs-terminal aspartic acid for each 34,000 to 36,000 g of protein was evidence that two such chains were present in each Fragment D. That is, data available in the literature indicate that the smallest Fragment D species (i.e. D5) is not likely to have a molecular weight of less than 80,000 (63) ; it may prove to be considerably larger. Sodium dodecyl sulfate gels of reduced Fragment DS exhibited only two bands (Fig. 8, Gel 15), one attributable to /yl (molecular weight 42,000) and the other, to /(11/18 (molecular weight 6,700). A molecular weight of 80,000 or higher thus does not appear possible unless two such y chain derivatives are present in the fragment.
The same conclusion arises even when molecular weights obtained by sodium dodecyl sulfate gel electrophoresis of unreduced Fragment D (16,19) are used, despite the fact that these conditions almost certainly lead to an underestimation of the molecular weights of disulfide bridged proteins (68-70). In addition to the disulfide bridges known to exist in the N-DSK (12), intrachain bridges have now been demonstrated in other regions of both Acr and BP chains ( Fig. 4; cf. Fig. 13), thus accounting for at least four of the 28 to 29 disulfide bridges in fibrinogen (71). Moreover, at least five interchain bridges must exist in that portion of the molecule which forms Fragment D (cf. Fig. 8, Gels 1s to 15 and Fig. 13) scission of at least one peptide bond in each of the six constituent chains, the subunit structure of derivative Bands VII and VIII (Fig. 8) suggests that of these critical bonds, those in the y chains (y + yIZ + /?I) are cleaved faster than those in the other chains.
Furthermore, the separation of D and E occurs more rapidly than the release of /PG. Therefore, chains of BP origin continue to be released from the COOH-terminal region after all core species which could have generated Fragment E have been consumed (Figs. 1 and 8 (Fig. 12).
In addition to size heterogeneity of Fragment D, electrophoretic heterogeneity has also been observed under conditions in which charge was an influential factor (23,28,62,73,74 On the basis of gel electrophoresis and immunochemical analyses, Fisher el al. (17,22) concluded that there were early forms of Fragment D with molecular weights of 150,000 or more.
Our results suggest that early forms of Fragment D have a molecular weight in the range of that reported by JIills (19) and Fisher el al. (17) or higher.
The present studies have demonstrated (Figs. 8 and 11) that the recovery of core Fragments D varies widely (i.e. from 60% to 28.7% or potentially even lower) and is a function not of the molar content of D but rather of the degree of enzymic degradation which has taken place.
Therefore, any valid calculation of the number of core fragments must take into account the molecular weight of the species of Fragment D existing at that phase of degradation.
There has been no direct evidence to support the structure proposed for Fragment Y, i.e. lacking one Fragment I>. Our analyses of Band VIII (Y) indicate that this derivative is formed by the release of a single //I chain from a core fragment of higher molecular weight (Band VII Apparent discrepancies between these studies and our present work are primarily a function of constraints imposed by the assumed monomeric model. The main differences can be resolved by the following considerations. In certain studies, the molecular weight of Fragment D was calculated by summing molecular weights of the subunit chains detected in sodium dodecyl sulfate gels, a calculation which involved the assumption of monomolecular contribution by each subunit chain (16,19,31,35,77). Molecular weight estimates based on the migration of uureduced fragments have also been considered compatible with the monomeric model (16,32,77) or, as in the case of Mills (19), were discounted when the estimated upper value for D was found to be above 100,000 (namely 130,000).
The latter investigator discounted the values because he assumed that these disulfide bridged species behaved as reduced protein (i.e. a relatively noncompact structure). The unlikelihood of such behavior has been emphasized in several recent studies (6870) as well as ours (e.g. Fig. 4), by the observation that disulfide bridged molecules migrate as if they had a more. rather than less, compact structure than do those which have beeu reduced.
The coexistence, during inter-mer1iat.e phases of plasmic hydrolysis.
of different chains with the same migration properties (e.g. /fis and y; /& and /?I) was apparently not recognized by a number of investigators. This ~vould seem to account for the incorrect identification or assigumcnt or both of relative plasmic resistance of the BP and y chain remnants occurring iu Fragment D (31, 32,34,35 Because of the persistence of the COOI-I-terminal region of the BP chain (and thus its antigenic determinant) in all but the most advanced forms of Fragment I>, it appears likely that these investigators had discovered the determinant associated with /a chains (~1. Fig. 6).
In addition to the antigenic determinant in the COOHterminal region of the BP chain (uamely /Pa), which is present in early and intermediate forms of Fragment I), two other detrrminants (designated 1) and F, respectively) have been itleiitified in this fragment.
The F fragment is derived from Fragment I> during advanced degradative phases (Figs. 1, 8, and 10) and was apparently responsible for immunoelectrophoretic precipitin arcs observed by other investigators (79,80). The designation I) has been assigned to that antigenic region iu Fragment 1) whicli is altered or otherwise lost (luring the evolution of Fragmeiits F (Fig. 9).
The COOKterminal portions of Xar am1 B/3 chains, though not identical, appear to be antigcnically related and both are distinguishable from y chaiils. Similar observations have bccu reported by Gollwitzer et ~2. (81), who showed that izol and I%@ chains from bovine fibrinogen were immunochemically relatetl to each other but not to y chains.
They suggested a large degree of homology between Acr and l%p chains, a result consistent with our own immunochemical and tryptic peptitle mapping esperiments (Reference 38 and present work) on human Aa! and B/3 chains.
The recognition of three separate antigenic determinants in Fragment D (namely /&, D, F) raises the question of the relationship between these determinants and those described recently by Plow and Edgington (36,82,83). Though these in vestigators have assumed the monomeric D hypothesis, their data are also consistent with the dirneric structure proposed above.
That is, Plow and Edgington mere able to identify two distinct antigenic regions in Fragment D (36). One of these, termed "neoantigen" (36,82,83), n-as detected with an anti-I> serum which had been absorbed with fibriuogen; the other, detected with unabsorbed anti-D serum was termed "total I)fragment-associated antigen" (36). They found tn-o of the latter antigenic sites in Fragment X and only one iu Fragment Y. However, they found only a single neoantigenic site per mole cule of Fragments X, Y, or I), although two shoultl have beeu found in X to conform to the monomeric hypothesis.
To :ucount for this apparent paradox they proposed that only one of the neoaritigciiic sites iii X \vas expressed, whereas tlic other was "latent." Our results would suggest that the total I)-fmgment-