Fibrinogen αC-regions are not directly involved in fibrin polymerization as evidenced by a "Double-Detroit" recombinant fibrinogen mutant and knobs-mimic peptides.

BACKGROUND
Fibrin polymerization, following fibrinopeptides A and B (FpA, FpB) cleavage, relies on newly exposed α- and β-chains N-termini (GPR, GHR; A-, B-knobs, respectively) engaging pre-existent a and b pockets in other fibrin(ogen) molecules' γ- and (B)β-chains C-terminal regions. A role for mostly disordered (A)α-chains C-terminal regions "bridging" between fibrin molecules/fibrils has been proposed.


OBJECTIVES
Fibrinogen Detroit is a clinically observed mutation (AαR19→S) with non-engaging GPS A-knobs. By analogy, a similar Bβ-chain mutation, BβR17→S, should produce non-engaging GHS B-knobs. A homozygous "Double-Detroit" mutant (AαR19→S, BβR17→S; DD-FG) was developed: with A-a and B-b engagements endogenously blocked, other interactions would become apparent.


METHODS
DD-FG, wild-type recombinant (WT-FG), and human plasma (hp-FG) fibrinogen self-association was studied by turbidimetry coupled with fibrinopeptides release HPLC/mass spectrometry analyses, and by light-scattering following size-exclusion chromatography (SE-HPLC).


RESULTS
In contrast to WT-FG and hp-FG, DD-FG produced no turbidity increase, irrespective of thrombin concentration. The SE-HPLC profile of concentrated DD-FG was unaffected by thrombin treatment, and light-scattering, at lower concentration, showed no intensity and hydrodynamic radius changes. Compared with hp-FG, both WT-FG and DD-FG showed no FpA cleavage difference, while ~50% FpB was not recovered. Correspondingly, SDS-PAGE/Western-blots revealed partial Bβ-chain N-terminal and Aα-chain C-terminal degradation. Nevertheless, ~70% DD-FG molecules bearing (A)αC-regions potentially able to associate were available. Higher-concentration, nearly-intact hp-FG with 500-fold molar excess GPRP-NH2 /GHRP-NH2 knobs-mimics experiments confirmed these no-associations findings.


CONCLUSIONS
(A)αC-regions interactions appear too weak to assist native fibrin polymerization, at least without knobs engagement. Their role in all stages should be carefully reconsidered.


| INTRODUC TI ON
Fibrinogen is a central player in blood coagulation, with important roles in pathological situations such as thrombosis, 1,2 atherosclerosis, 3,4 and cancer metastasis. 5 It is a high-molecular-weight (~340 000 Da), elongated (~45 nm) glycoprotein circulating as an inactive precursor in the blood at ~3 to 5 mg/mL. 6 Fibrinogen is composed of two pairs of three polypeptide chains, Aα, Bβ, and γ (Aα 2 Bβ 2 γ 2 ; in human form 610, 461, and 411 amino acids, respectively 7 ). All chains' N-terminal ends are bundled together by S-S bridges in a central "E-region," from which two triple coiled-coil connectors depart in opposite directions, each held in register by two disulfide rings between the Aα-, Bβ-, and γ-chains. 8 At the end of the connectors, the Bβ-and γ-chain C-terminal parts form two outer D-regions, within which each chain folds independently. 9 Instead, the >400 C-terminal residues of the Aα chains first reverse direction forming a fourth strand up to about halfway on the coiled-coils connectors, 10 and then protrude as mainly disordered appendages ("AαC-regions"), 11 within which a small partially ordered subdomain (Aα425-503 in the human sequence) has been identified. [12][13][14] Thrombin converts fibrinogen into a reactive species by cleaving two pairs of short peptides, called fibrinopeptides A and B (FpA and FpB, 16 and 14 residues, respectively), from the N-termini of the Aα and Bβ chains in the central E-region, generating the α 2 β 2 γ 2 fibrin monomer. [15][16][17] The resulting N-termini in the α-and β-chains, with initial sequence GPR and GHR, are called the A and B "knobs," respectively. 18 They engage very tightly, mainly by electrostatic interactions, into pre-existing and readily available a and b "holes" in the D-region's C-terminal parts of the γ-and (B)β-chains, respectively, in other fibrin(ogen) molecules. 18 Rapid polymerization ensues, first forming elongated (proto)fibrils, 19,20 which by subsequent branching and lateral aggregation give rise to a three-dimensional network, the clot scaffold that stabilizes the initial platelet plug during blood coagulation (see 7,17 ). FpA release is the key initial event, with A-a interactions governing (proto)fibril formation in a final half-staggered, double-stranded arrangement. 20,21 FpB is released by thrombin later in the process, and the B-b engagement enhances the lateral thickening of the fibers. 22,23 There is also evidence of promiscuity between the A and B knobs toward the a and b holes, probably derived from the common evolutionary origin of the fibrinogen chains. 24 Several important aspects of fibrin polymerization have been elucidated over the years, but some key questions still remain. In particular, it has been proposed that the (A) αC-regions interact with each other, and with the central E-region in the fibrinogen molecule, and that they are released following fibrinopeptide cleavage, more likely after FpB removal. 11,23 The released αC-regions have been postulated to assist fiber assembly by intermolecular binding between parallel protofibrils. 25,26 However, proving this αC-regions release mechanism at the level of individual fibrin molecules is difficult, as they rapidly polymerize, and only large amounts of knobs-mimic peptides inhibitors such as GPRP-NH 2 and GHRP-NH 2 (at ≥500-fold molar ratio) can block this process. 27 Because the B-b engagement induces changes in the relative orientation of the β-and γ-chains C-terminal subdomains, 28 binding of knobs-mimics can have difficult to evaluate consequences at a structural level. However, they could still be employed to reveal other potential interactions between fibrin(ogen) molecules.
Among the many clinically observed fibrinogen mutations affecting fibrin formation (http://site.geht.org/base-fibri nogene 29 ), fibrinogen Detroit (AαR19 → S) 30 is of particular interest. In this mutant, FpA can be cleaved by thrombin, but the resulting mutated A-knob, GPS, is unable to bind either the a or b holes, leading to severely impaired fibrin formation, only partially rescued by the GHR normal B-knobs binding to their cognate b holes. 31 On this basis, we hypothesized that a similar mutation in the B-knob, BβR17 → S, would stop it binding to either holes b or a. A mutant carrying both AαR19 → S and BβR17 → S substitutions should therefore reveal any other potential interaction between fibrin monomers following cleavage of both fibrinopeptides.
Conclusions: (A)αC-regions interactions appear too weak to assist native fibrin polymerization, at least without knobs engagement. Their role in all stages should be carefully reconsidered.

K E Y W O R D S
fibrin, fibrinogen, mutation, polymerization, thrombin

Essentials
• αC-regions are thought to actively complement knobhole interactions during fibrin assembly.
• Defective knobs in a recombinant "Double Detroit" mutant (DD-FG) should impede fibril formation.
• Thrombin-treated DD-FG alone or fibrinogen with knobs-mimics showed a total lack of associations.
• A more passive role of the α-chains C-terminal regions in fibrin assembly is proposed.
Here we report the development of this recombinant human fibrinogen mutant, that we have termed Double-Detroit fibrinogen (DD-FG), and its characterization before and after thrombin treatment. As it unfortunately sometimes happens with recombinant fibrinogen production in mammalian cells, we have encountered degradation issues with the DD-FG mutant and wild-type fibrinogen (WT-FG), despite the addition of protease inhibitors during the purification procedures. This resulted in cleavage of portions of the AαCregions and of the first ~50N-terminal amino acids of the Bβ-chain, in a manner reminiscent of the formation of the so-called fragment X by plasmin action. 32 Nevertheless, the amount of intact or just slightly degraded species was sufficient to allow clear-cut results to be obtained. Namely, we found that, despite thrombin cleavage of the fibrinopeptides, DD-fibrin monomers showed no signs of polymerization whatsoever, neither by turbidity analysis, nor by time-resolved static and dynamic light scattering. Similar results were obtained with a fibrinogen fraction with mostly intact AαC-regions and in the presence of a large excess of both GPRP-NH 2 and GHRP-NH 2 . Overall, these data failed to reveal any contribution of the αC-regions, while confirming the fundamental role of knob-hole interactions in powering fibrin polymerization. DD-FG will also provide an essential new tool for the study of the molecular properties of fibrin monomers after their generation from fibrinogen by thrombin, without the interference of polymerization or the formation of a clot.

| DD-FG and WT-FG expression, purification, and quality control
Recombinant human AαR19S/BβR17S fibrinogen (DD-FG) and WT-FG were prepared as previously described. 33 Detailed protocols, including for the enzyme-linked immunosorbent assay tests, can be found in the Appendix S1. Final concentrations were determined spectrophotometrically at λ = 280 nm (ε 280 = 1.51 mL mg −1 cm −1 ) 34 and the purity of each recombinant fibrinogen batch was assessed by SDS-PAGE under reducing conditions using 10% polyacrylamide (PAA) gels. 35 Further characterization was conducted by Western-blot analysis after SDS-PAGE, using the mouse monoclonal antibody Y18 specific for the N-terminal end of the Aα-chains, 36  done essentially as previously reported 35 (for details, see the Appendix S1).

| Turbidity coupled to fibrinopeptides release assays
These experiments were conceived to simultaneously monitor turbidity in a spectrophotometer and fibrinopeptide release by HPLC on the same sample, the latter at long time-points. In addition to WT-FG and DD-FG, fibrinogen purified from human plasma (hp-FG; type FIB3, Enzyme Research Laboratories) was used. An hp-FG fraction with mostly intact AαC-regions (HMW-FG) was prepared as previously described, 35,37,38  Samples were checked by SDS-PAGE before and after these turbidity experiments. Detailed protocols are provided in the Appendix S1.

| Size-exclusion chromatography
Size-exclusion chromatography (SE-HPLC) was performed both analytically, to check for the presence of high-and low-molecular weight components in the fibrinogen preparations, and in a semipreparative way to isolate sufficient monomeric fractions for the static and dynamic light scattering experiments (see Appendix S1).

| Characterization of recombinant and plasmaderived fibrinogens
Reduced samples of the proteins were run on SDS-PAGE, and were found to be consistent with apparently pure preparations, with bands corresponding to the standard Aα, Bβ, and γA chains ( Figure   S1A). However, a more detailed Western blot analysis using a monoclonal antibody specific for the N-terminal end of the Aα-chain revealed that in both WT-and DD-FG, up to ~70% of the Aα chains presented varying levels of degradation in the AαC region ( Figure   S2 and Table S2). Furthermore, from the analyses of a Western-blot stained with a polyclonal antibody against the C-terminal region of the Bβ-chain, two groups of bands could be discerned ( Figure S3 and Table S3). The constituents of the first group had approximate molecular weights close to that of intact Bβ chains with up to two sialic acids in its single carbohydrate chain (theoretical 53 900-54 450), whereas the second group could result in the hp-FG sample by the loss of the N-terminal 1-42 residues (mol. wt. ~4600), a classic plasmin-degradation event. 39 The corresponding DD-FG and WT-FG bands in this second group appeared to have slightly higher molecular weights (Table S3), suggesting that a different process might have generated a similar N-terminal degradation of the Bβ chains. In these particular batches analyzed, about 25% (DD-FG) and 30% (WT-FG) of the Bβ chains appeared missing the N-terminal residues (Table   S3), which include the FpB and the B-knob residues. Given the size shift, it is unlikely that the proteolysis took place at the C-terminal end. Attempts to prevent this degradation with additional protease inhibitors have so far been unsuccessful, suggesting that a different cell line or expression system might be necessary in the long run to obtain more pristine products. However, the main self-interaction domain within the (A)αC-regions has been identified within residues Aα425-503. 14 Because the Aα1-503 stretch has a predicted molecular weight of 54 589, this value was used as a cutoff to conservatively calculate from the SDS-PAGE/Western blot data (Table S2) the percentage of molecules bearing AαC regions potentially able to interact. About 70% of DD-FG molecules (and ~50% of both hp-FG and WT-FG) were found to contain the AαC self-interaction region, allowing us to perform meaningful polymerization studies. Indeed, in initial assays at both low and high thrombin concentrations, WT-FG behaved as a typical fibrinogen sample, whereas DD-FG did not show any increase in turbidity ( Figure S1B). In addition, preliminary fibrinopeptide release experiments indicated cleavage of both FpA and FpB from the recombinant FGs (data not shown).

| Turbidity coupled to Fps release studies
After prolonged treatment with thrombin (see below), both WT-FG and DD-FG showed a complete, small but noticeable reduction in the sizes of both the Aα-and Bβ-chains, attributable to normal cleavage of FpA and FpB ( Figure 1A). Although both hp-FG and WT-FG displayed a typical turbidity profile, 40  For WT-FG ( Figure 2B) the FpAP and FpAY peaks were practically absent, as previously noted with recombinant fibrinogens expressed in CHO cells. 43 As a result, the FpA peak was higher than for hp-FG. Only desArgB appeared to be present among the FpB variants, but the FpB peak was noticeably lower than its counterpart from hp-FG. Similar to hp-FG, the release of all fibrinopeptides for WT-FG appeared complete after 1-hour incubation with thrombin.
The DD-FG fibrinopeptides analyses showed some differences ( Figure 2C). FpAP appeared to be present, confirmed by MS-MS analyses (data not shown), and FpB was apparently released more slowly, with the 2-hour thrombin incubation peak still being lower than that observed after 3 hours.
Approximate absolute amounts of fibrinopeptides were then determined for the 3 hours timepoints, using calculated ε 211 molar extinction coefficients (see Appendix S1), as shown in For hp-FG, the FpA/FpB ratio was close to 1 (Table 1), as expected.
However, for both WT-FG and DD-FG, the FpA/FpB ratio was higher (~1.6 and ~2.3, respectively; Table 1), indicating that about one-third and one-half of FpB, respectively, were not recovered. By careful MS-MS analysis of all peaks present in our chromatograms, no alternative FpB form that could account for the missing amounts was found.
Moreover, by SDS-PAGE analyses ( Figure 1A) all FpB is removed from the Bβ-chains, excluding the presence of noncleavable FpB in the recombinants. The Western blot analyses could provide only a partial explanation. As shown in Table S3, ~31% and ~26% of the intact Bβchains were found missing in these WT-FG and DD-FG preparations, respectively. For WT-FG, this was reasonably close to the missing FpB amount in the HPLC analyses, whereas for DD-FG another ~30% was unaccounted for. Unfortunately, the amount of DD-FG sample taken from this experiment was insufficient for Western blot following the regular SDS-PAGE analysis, and this particular material-and time-consuming experiment was not repeated. It is conceivable, however, given the variability observed in the degradation of DD-FG batches, that this batch had an even higher amount of N-terminally cleaved Bβ-chains.
Although this issue is being further investigated, the combined results still support the notion that both Fps were cleaved from DD-FG, and that the absence of clot formation was due to the defective, non-binding A and B knobs.
Experiments with GPRP-NH 2 and GHRP-NH 2 knobs mimics were also conducted using hp-FG. As shown in Figure S5  600 × molar excess of GPRP-NH 2 alone (magenta trace) greatly delayed but not completely abolished lateral aggregation, likely because of a "rescue" effect by the B:b engagement. When GHRP-NH 2 was also added in a 290 × molar excess ( Figure S5, blue trace) no lateral aggregation took place. It was found, however, that while an excess of GPRP-NH 2 had no influence on FpA cleavage by thrombin, GHRP-NH 2 did delay FpB cleavage (data not shown). This effect could be overcome by a 10-fold increase in thrombin concentration (data not shown).  Table 2  ~333 000 g/mol. 43 Both data confirmed optimal performance of the SLS/DLS set-up.

| SE-HPLC and SLS/DLS studies
An additional experiment was performed with the SE-HPLCpurified hp-FG sample, after dilution to 0.22 mg/mL, similar to that of DD-FG (see below ). As shown in Figure 4A    The difference between these two values is statistically significant at the 99% confidence level (P < 0.001, one-tailed Student t test).
c Values extrapolated to c = 0 using the second virial coefficient of HMW-FG determined by Raynal et al. 43 Values within round brackets were not extrapolated to c = 0 (see Appendix S1 for details).
complete removal of all fibrinopeptides was ascertained both by SDS-PAGE/Western blots ( Figure S6) and RP-HPLC ( Figure S8). Experiments with the SE-HPLC purified DD-FG sample are reported in Figure 4C, where a 0. 25   the Aα252-610 C-terminal region, has a similar absence of non-specific interactions even without fibrinopeptides removal. 43 Because at pH 7.4 a reduction of net charge from −22 to −12 happens upon fibrinopeptides removal, and α251-FG has a net charge of −18, perhaps a common charge-related mechanism is responsible for this effect.

| D ISCUSS I ON
A considerable body of evidence exists regarding the proposed involvement of the AαC-regions in "helping" fibrin assembly at certain stages (early reviews 49,50 ). First, electron microscopy studies suggested a "release" mechanism following fibrinopeptide cleavage based on differences in the (A)αC-regions location between not-cleaved and enzyme-treated fibrinogen. 11,25 Additional studies showed concentra- CI) in HMW-FG solutions with peptide knobs mimics. These calculations suggest that interactions between αC-regions in native fibrin(ogen), or any other interactions between fibrin molecules, are too weak to be able to lead to any assembly following fibrinopeptide cleavage in the absence of knob-hole engagement.
A possible explanation that will reconcile our findings with the existence of αC-αC interactions, is to reverse the logic behind the currently accepted mechanistic view. That is, it could be conceivable that it is the lateral thickening of the fibrils that brings the αC-regions in sufficiently close proximity to each other and allows their reciprocal binding. This will allow immediate reinforcement of the fibers in terms of mechanical strength and resistance to proteolysis, both of which are later further enhanced by factor XIIIa-mediated crosslinking. As for what regulates the dramatic fiber thickening that follows the fibrin assembly lag phase, other mechanisms could be prevalent, from the change in the D-regions/coiled-coils relative orientation following B-b engagement, 28 to the collapse of hyperbranched fibrils, 43 or their combination. Clearly, more work is necessary to better understand this mechanism.
In conclusion, overall, our data strongly support that formation of the fibrin clot is critically dependent only on the residues residing in the A-and B-knobs that are exposed after thrombin cleavage of fibrinogen. Although we cannot exclude that undetected issues could affect the recombinant fibrinogens behavior, it is the combined results obtained with DD-FG, having similar degradation as "normal" plasma fibrinogen, and with HMW-FG plus knobs-mimics, with nearly intact Aα-and Bβ-chains, that preclude a relevant role for other interactions in fibrin formation. Importantly, the DD-FG described in this study provides a novel crucial tool compound with which, once degradation issues are resolved, we will be able to study monomeric fibrin structural and functional properties, such as the proposed αC-regions release and other conformational changes following thrombin treatment, in the absence of polymer formation, and in the absence of peptide mimics to bind the polymerization pockets, which by themselves may affect the fibrin monomer conformation.

ACK N OWLED G M ENTS
We are indebted to the late Professor RF Doolittle (University of California San Diego, La Jolla, CA) for insights about potential Bknob mutations affecting its binding to the b-hole. We thank V.
Fontana (IRCCS Ospedale Policlinico San Martino, Genova, Italy) for assistance in the statistical analyses. Supported by a British Heart Foundation Programme Grant (RG/13/3/30104) and by a grant from the Italian Ministry of Health (5 × 1000 Funds 2013 to AP).

CO N FLI C T O F I NTE R E S T S
All authors declare no conflicts of interest.