FUNCTION OF THE ACTIVATED PROTEIN C (APC) AUTOLYSIS LOOP IN ACTIVATED FVIII INACTIVATION

SUMMARY

Activated protein C (APC) binds to its substrates activated factor V (FVa) and activated factor VIII (FVIIIa) with a basic exosite that consists of loops 37, 60, 70 and the autolysis loop. These loops have a high density of basic residues, resulting in a positive charge on the surface of APC. Many of these residues are important in the interaction of APC with FVa and FVIIIa. The current study focused on the function of the autolysis loop in the interaction with FVIIIa. This loop was previously shown to interact with FVa, and it inhibits APC inactivation by plasma serpins. Charged residues of the autolysis loop were individually mutated to alanine and the activity of these mutants was assessed in functional FVIIIa inactivation assays. The autolysis loop was functionally important for FVIIIa inactivation. Mutation of R306, K311, and R314 each resulted in significantly reduced FVIIIa inactivation. The inactivating cleavages of FVIIIa at R336 and R562 were affected equally by the mutations. Protein S and FV stimulated cleavage at R562 more than cleavage at R336, independent of mutations in the autolysis loop. Together, these results confirmed that the autolysis loop plays a significant role as part of the basic exosite on APC in the interaction with FVIIIa.

INTRODUCTION

Protein C (PC) is a pivotal serine protease in the down-regulation of the coagulation cascade. PC is activated by thrombin (activated factor (F) II, or FIIa) in the FIIa-thrombomodulin complex to form activated protein C (APC). Once activated, APC inactivates activated factor V (FVa) and activated factor VIII (FVIIIa) (Marlar et al, 1982) via limited proteolysis, resulting in a severe decrease of thrombin formation.

The structure of FVIII consists of an N-terminal heavy chain (A1 and A2 domains) and a C-terminal light chain (A3, C1 and C2 domains) that are connected by the B domain. FIIa activates FVIII via cleavages at R372, R740 and R1689 (Pittman & Kaufman, 1988; O’Brien et al, 1990), which results in removal of the B domain and separation of the A1 and A2 domains. The A1 domain remains tightly associated with the light chain, but the A2 domain is only associated with the rest of the FVIIIa molecule via weak electrostatic interactions. Consequently, the A2 domain readily dissociates from the rest of the molecule resulting in inactivation of FVIIIa (Lollar & Parker, 1991; Fay & Smudzin, 1992; Pipe et al, 1999).

Subsequently, FVIIIa is inactivated by APC via cleavages at R336 at the C-terminal end of the A1 domain and R562 in the A2 domain (Fay et al, 1991a), which enhances dissociation of the A2 domain. Cleavage at R336 is fast, but results in a partially active molecule. Cleavage at R562 is slower than R336, but completely inactivates FVIIIa. The APC cofactors protein S and FV enhance cleavage at both sites and both cofactors stimulate R562 cleavage more than R336 cleavage (Regan et al, 1994; O’Brien et al, 2000; Gale et al, 2008).

In comparison, FV is highly homologous to FVIII and has the same domain structure, with approximately 40% sequence homology (Church et al, 1984; Kane & Davie, 1986; Jenny et al, 1987). FVa is inactivated by APC via cleavages in the heavy chain at R306, R506, and R679 (Suzuki et al, 1983; Kalafatis et al, 1994). In this case, R306 is the slower cleavage that completely inactivates the molecule, whereas the faster R506 cleavage results in a partially active intermediate (Nicolaes et al, 1995). The cleavage at R679 has thus far not been attributed any physiological significance. Protein S stimulates cleavage of both R306 and R506, but has a larger effect on R306 cleavage (Rosing et al, 1995; Norstrom et al, 2003; Norstrom et al, 2006). The anticoagulant form of FV also stimulates FVa inactivation (Cramer et al, 2010), but thus far this has not been further characterized. Because FVa does not spontaneously inactivate, like FVIIIa, it is dependent on APC for proteolytic inactivation. A recent study suggested that FVIIIa has a significantly lower affinity for APC than does FVa (Nicolaes et al, 2010), which would support the importance of APC in FVa inactivation. Nevertheless, some evidence supports the physiological relevance of FVIIIa inactivation by APC proteolysis (Castoldi et al, 2004).

Analysis of the Gla-domainless crystal structure of APC elucidated a positively charged basic exosite on APC (Mather et al, 1996). This basic exosite contains loop 37 (loops are numbered in the chymotrypsin numbering system, loop 37 corresponds with PC residues 190–193), loop 60 (PC residues 214–222), loop 70 (PC residues 225–235; calcium binding loop) and loop 148 (PC residues 301–316; autolysis loop).

The autolysis loop of APC has been of special interest, because this loop is relatively large, highly flexible, and the sequence 306-REKEAKRNR-314 with five positively charged and two negatively charged residues, gives this loop an overall positive charge (Mather et al, 1996). A comparative study showed that this sequence is well conserved in mammals (Murakawa et al, 1994). However, four out of nine species studied (cat, cow, dog, and goat), have a truncated autolysis loop with residues 305–308 deleted, emphasizing the possible importance of residues 309–314.

An alanine scanning study of the APC autolysis loop showed that basic residues R306, K311, R312, and R314 are the most important residues for the interaction with FVa (Gale et al, 2000). APC inactivates FVa via cleavages at R306, R506, and R679, of which R506 is fastest, but results in a partially active molecule. Cleavage at R306 is slower than R506, but results in full inactivation, and the slowest cleavage, at R679, has thus far not been attributed any physiological significance. Mutation of basic residues in the autolysis loop had a greater impact on cleavage at R506, whereas cleavage at R306 was affected far less (Gale et al, 2000).

In studies using a different approach, an APC mutant was created in which the autolysis loop was swapped with the autolysis loop of FXa, which is five residues shorter than the autolysis loop of APC, but still has four basic residues. This APC mutant showed an increased rate of FVa and FVIIIa inactivation, but also reduced the half-life of APC in plasma from 30 min to ~5 min, due to increased interaction with plasma serpins (Yang et al, 2005). However, mutation of the basic residues of the FXa autolysis loop in this APC mutant reduced the FVa inactivation rate relative to APC with the wild type FXa autolysis loop, and also reduced the interaction with antithrombin (Yang et al, 2007), which is the main serpin in plasma to inactivate plasma serine proteases (Gettins, 2002). Thus, regardless of the length of the autolysis loop, basic residues in this loop appear important for the interaction of APC with FVa and antithrombin. Also, substitution of the autolysis loop with the corresponding loop of bovine APC, which is shorter in the region 303–310 than its human counterpart, resulted in increased proteolytic activity towards FVa, FVIIIa, and small synthetic chromogenic substrates, and also modestly enhanced APC inactivation by serpins, further emphasizing the relevance of the length of this loop (Shen et al, 1999).

Studies of the other loops in the basic exosite of APC have shown that several charged residues in each loop play a significant role in FVa inactivation. In particular, residues K193, R222, D227, R229 and R230 are significant contributors to the interaction with FVa. Also in these studies, cleavage at R506 was severely impaired by mutation of the basic exosite, whereas cleavage at R306 was affected much less (Friedrich et al, 2001; Gale et al, 2002). FVIIIa inactivation is mediated by the same residues in APC, but contrary to the results found in FVa inactivation, the inactivating cleavages at R336 and R562 are affected equally (Manithody et al, 2003). The role of individual residues of the autolysis loop however, has not been studied in the interaction with FVIIIa.

Protein S enhances cleavage at R562 in FVIIIa more than cleavage at R336 (Regan et al, 1994; O’Brien et al, 2000; Gale et al, 2008), and cleavage at R306 in FVa more than cleavage at R506 (Rosing et al, 1995). Although the APC autolysis loop distinguishes between the two cleavages in FVa, protein S appears to function independent of the autolysis loop in FVa inactivation (Qureshi et al, 2008), and also independent of the other loops of the basic exosite in FVIIIa inactivation (Manithody et al, 2003).

As the autolysis loop plays an important role in the inactivation of FVa, and the same residues in the other loops of the basic exosite of APC are important for the inactivation of both FVa and FVIIIa, the current study was designed to determine the importance of the autolysis loop in FVIIIa inactivation, and to characterize the function of individual residues of this loop in the FVIIIa inactivation reaction. APC mutants in which individual residues of the autolysis loop were mutated to alanine were created and their activity in FVIIIa inactivation was assessed in a purified FXase assay. We found that individual mutation of R306, K311 and R314 resulted in significantly reduced FVIIIa inactivation, and cleavages of FVIIIa at R336 and R562 were affected equally by mutations in the autolysis loop. Both protein S and FV stimulated cleavage at R562 more than cleavage at R336, independent of mutations in the autolysis loop. These results confirmed that the autolysis loop plays a significant role as part of the basic exosite on APC in the interaction with FVIIIa.

MATERIALS AND METHODS

Reagents

FIXa, FX, protein S, and FIIa were from Enzyme Research Laboratories (South Bend, IN, USA). FV was from Haematologic Technologies (Essex Junction, VT, USA). Refacto^® B domain-deleted FVIII was from Wyeth Corporation (Collegeville, PA, USA). Synthetic unsaturated phosphatidylcholine, phosphatidylserine, and phosphatidylethanolamine were from Avanti Polar Lipids (Alabaster, AL, USA). Bovine serum albumin (BSA), hirudin and benzamidine were from Calbiochem/EMD Biosciences (La Jolla, CA, USA). FVIII-deficient plasma and normal pooled human plasma were from George King Biomedical Inc. (Overland Park, KS, USA). TriniCLOT aPTT HS activated partial thromboplastin time (APTT) reagent was from Trinity Biotech plc (Bray, Wicklow, Ireland). Pefachrome PCa chromogenic substrate was from Centerchem Inc. (Norwalk, CN, USA). Chromogenic FXa substrate S2765 (Z-D-Arg-Gly-Arg- pNA·2HCl) was from Chromogenix (Lexington, MA, USA). Vitamin K₁ and p-nitrophenyl p′-guanodino-benzoate hydrochloride (pNPGB) were from Sigma (St. Louis, MO, USA). Penicillin/Streptomycin/L-Glutamine (100× solution), Lipofectamine 2000, G-418, and Dulbecco’s modified Eagle medium (DMEM) Ham’s F12 50:50 with phenol red were from Invitrogen (Carlsbad, CA, USA). DMEM Ham’s F12 50:50 without phenol red was from Mediatech Inc. (Manassas, VA, USA). Fetal bovine serum was from Omega Scientific Inc. (Tarzana, CA, USA).

Recombinant protein C production, purification, and characterization

Mutations in the PC autolysis loop were made using site directed mutagenesis as described (Gale et al, 2000) and the cDNA was subcloned into the pcDNA3.1+ expression vector (Invitrogen). The construct was then stably transfected into HEK293 cells using Lipofectamine 2000, G-418 was used to select stable clones. Conditioned medium was collected and stored at −80°C. PC was purified essentially as described (Gale et al, 1997). Briefly, after thawing the conditioned medium, 5 mM benzamidine and 5 mM ethylenediaminetatraacetic acid (EDTA) were added, and run over a 12 ml Fast Flow Q (FFQ)-sepharose (Amersham Biosciences, GE Healthcare, Piscataway, NJ, USA) column at 4°C, which was equilibrated with buffer A (20 mM Tris, 100 mM NaCl, pH 7.4). PC was eluted in a step elution with buffer C (20 mM Tris, 50 mM CaCl₂, pH 7.4). The presence of PC was assessed in a PC activation assay with Protac (Aniara, Mason, OH, USA), and the chromogenic substrate Pefachrome PCa. Fractions containing PC were pooled and loaded on a second FFQ-sepharose column (5 ml), washed with buffer A, and PC was eluted with a NaCl gradient (0.1 – 1 M NaCl). Fractions containing PC were pooled and concentrated with an Amicon^® Ultra concentrator with a 30000 Da molecular weight cut-off from Millipore Corporation (Billerica, MA, USA). After PC purification, concentration was determined by absorption at 280 nm, using the extinction coefficient E = 14.5 per cm (for a 1% solution).

Protein C activation by FIIa

PC was activated by FIIa at 37°C, at a 300-fold molar excess of PC over FIIa. APC activity was monitored over time; the activation reaction was stopped by adding excess hirudin to inactivate the FIIa when the chromogenic activity of activation mixture towards Pefachrome PCa no longer increased (average 130 min over all preparations, range was from 90 to 165 min). APC was then further purified by loading the sample on a 1 ml FFQ-sepharose column and eluted in 1 step with a high salt buffer (20 mM Tris, 640 mM NaCl, pH 7.4). The APC preparations were then concentrated and stored at −80°C. Purity was assessed on silver stained gels, using the SilverXpress kit from Invitrogen. Active site concentrations of APC variants, FIXa and FXa were determined with active site titrations adapted from Chase & Shaw (1967).

Phospholipid vesicles

Phospholipid vesicles composed of 40% phosphatidylcholine, 20% phosphatidylserine and 40% phosphatidylethanolamine (M/M/M) (PCPSPE) were prepared as previously described (Mesters et al, 1991).

Recombinant disulfide crosslinked B domain truncated FVIII production, purification, and characterization

M662C/D1828C FVIII, M662C/D1828C/R336Q FVIII, and M662C/D1828C/R562 FVIII (Gale et al, 2006; Gale et al, 2008) were produced in HEK293 cells and purified as described (Gale & Pellequer, 2003). FVIII characterization was performed as previously described (Gale et al, 2006; Gale et al, 2008) using the Asserachrom Factor VIIIC:Ag ELISA kit from Diagnostica Stago (Parsippany, NJ, USA) to determine FVIII antigen level. FVIII activity in u/ml was based on APTT assays using a FVIII standard curve with Refacto^® FVIII (13,350 u/mg).

The functional concentration of FVIIIa-binding sites for FIXa was determined with FVIIIa and FIXa titrations for each recombinant FVIII as described (Gale et al, 2006), with the following exceptions. FVIII was activated for 1 min with 1.6 u/ml FIIa, followed by addition 6.4 u/ml hirudin to inactivate the FIIa. For the FVIIIa titrations, 2.56 nM was activated for 1 min with 1.6 u/ml FIIa, followed by 4.8 u/ml hirudin. FX was added to a final concentration of 0.2 μM. FXa generation data were fitted to a hyperbolic curve to derive the K_½ for the formation of the FVIIIa-FIXa complex, and active FVIII concentration was calculated as described (Gale et al, 2006).

FVIIIa inactivation experiments in the FXase assay

For FVIIIa inactivation reactions 1 nM M662C/D1828C FVIII (or variants) was activated for 1 min with 0.8 u/ml FIIa in the presence of 33 μM PCPSPE in FXase buffer (50 mM HEPES, 150 mM NaCl, 5 mM CaCl₂, 0.1 mM MnCl₂, 5 mg/ml BSA, 0.5 mg/mL NaN₃, pH 7.4), followed by addition of 4.8 u/ml hirudin to inactivate the FIIa. Protein S and FV were added together as a solution of 700 nM protein S and 84 nM FV, to a final concentration of 100 nM and 12 nM respectively, followed by APC as a solution of 11-fold of final concentration to start the time course of APC inactivation. 3 μl aliquots were taken at time points and were added to 27 μl of 0.39 nM FIXa/25 μM PCPSPE mix in a 96-well plate, immediately followed by addition of 5 μl 875 nM FX. The FX activation reaction was stopped after 30 s by addition of 50 μl AAB/EDTA (20 mM Tris-base, 100 mM NaCl, 50 mM EDTA, 5 mg/ml BSA, 0.2 mg/ml NaN₃, pH 8.3). Final concentrations for the FX activation reaction were 65 pM FVIIIa, 0.3 nM FIXa, 125 nM FX, 25 μM PCPSPE. The amount of FXa formed was assessed by adding 35 μl of 1 mM chromogenic FXa substrate S2765 followed immediately by a three minute kinetic OD read at 405 nm.

Data conversion for FVIIIa inactivation data

FVIIIa activity was expressed as a percentage of FXa production in the FXase assay at time point zero. To calculate the reaction rate constant, the percentage data for M662C/D1828C/R336Q FVIIIa inactivation were fit to an exponential decay curve (equation 1):

$$\text{A}={\text{A}}_0\times e^{(-\text{k}\times \text{t})}$$

in which A is the FVIIIa activity in percent, A₀ is the FVIIIa activity in percent at time zero, k is the rate constant per min, and t is time in min. M662C/D1828C/R562Q FVIIIa was not fully inactivated due to cleavage by APC at only R336, and reached a plateau of approximately 20% residual activity. Therefore the data for M662C/D1828C/R562Q FVIIIa were fitted to equation 2 that incorporates this plateau:

$$\text{A}={\text{A}}_0\times e^{(-\text{k}\times \text{t})}+0.2\times {\text{A}}_0\times \left(1-e^{(-\text{k}\times \text{t})}\right)$$

The FVIIIa inactivation reactions did not follow pseudo-first order kinetics under these conditions, and therefore the rate constant was calculated for only the initial part of the curve fit that appeared linear on a semi log plot. The rate constants were then converted to an initial reaction rate with the unit nM FVIII × /min × /nM APC by multiplying the rate constant by the concentration of FVIIIa in the inactivation reaction (0.754 nM) and dividing by the concentration of APC. Therefore the data with the same conditions, independent of APC concentration, could be pooled and averaged to give the initial reaction rate. Inactivation of M662C/D1828C FVIIIa should follow a double exponential decay curve, due to cleavage at both inactivation sites. However, due to the lack of pseudo-first order kinetics, two cleavages give too many variables to confidently calculate inactivation rates. Therefore the inactivation data of M662C/D1828C FVIIIa were presented as qualitative curves only, and rates were not calculated.

Western blotting

M662C/D1828C FVIIIa inactivation reactions for Western blot analysis were done as follows. 5 nM M662C/D1828C FVIII was activated with 2 u/ml FIIa for 10 min at room temperature, followed by addition of 8 u/ml hirudin to inactivate the FIIa. The sample at time zero was removed and 20 nM wt-APC and 100 nM protein S were added to the FVIIIa. Then samples were taken at time points as indicated and prepared by adding lithium dodecyl sulfate sample buffer (Invitrogen), 1 mg/mL dithiothreitol (Sigma) and 1.5 mg/mL iodoacetamide (Sigma), and heating at 80°C for 15 min. The samples were run on 10% Bis-Tris gels in MOPS (3-(N-morpholino)propanesulfonic acid) buffer (Invitrogen) and transferred to Immobilon-FL^™ polyinylidene difluoride membrane (Millipore). Membranes were probed with 58.12 anti-A1 domain monoclonal IgG (Wood et al, 1984) and developed according to LI-COR instructions with IRDye^® 680 goat anti-mouse (LI-COR Biosciences, Lincoln, NE, USA) as secondary antibody. Subsequently, the membrane was stripped (Yeung & Stanley, 2009) and reprobed with R8B12 anti-A2 domain monoclonal IgG (Fay et al, 1991b) and IRDye^® 680 goat anti-mouse (both primary antibodies were gifts from Bayer Corporation, Berkeley, CA, USA). Quantification was done with the LI-COR Odyssey Infrared Imaging System software.

RESULTS

FVIIIa rapidly loses activity due to dissociation of the A2 domain (Lollar & Parker, 1991; Fay & Smudzin, 1992; Pipe et al, 1999), which makes the study of FVIIIa inactivation by APC in a purified system technically challenging. In the published literature this problem has usually been circumvented by using very high concentrations of FVIIIa to create a larger timeframe in which FVIIIa activity is measurable. As an alternative approach, our laboratory has developed a stable FVIII molecule in which the A2 domain was linked to the A3 domain with a disulfide bond. This was achieved by mutation of residues M662 and D1828 to cysteine (M662C/D1828C FVIII), resulting in a disulfide bridge between these two residues, prevention of A2 domain dissociation, and stabilization of the FVIIIa molecule (Gale et al, 2006; Radtke et al, 2007). The specific activity of the M662C/D1828C FVIII preparations was assessed in APTT clotting assays using Refacto FVIII as standard curve. M662C/D1828C FVIII showed a specific activity of 10,900 u/mg, M662C/D1828C/R336Q FVIII showed a specific activity of 8,900 u/mg, and M662C/D1828C/R562Q FVIII showed a specific activity of 14,500 u/mg. Mutations of the respective APC cleavage sites in M662C/D1828C/R336Q FVIII and M662C/D1828C/R562Q FVIII were confirmed with analysis of proteolysis products on silver stained SDS-PAGE gels (Gale et al, 2008).

The preparations of M662C/D1828C FVIII used for this study were inactivated slower by APC than those used in our previous study, yielding somewhat different inactivation rates (Gale et al, 2008). This was probably due to changes in experimental conditions, and possibly also due to some batch-to-batch variability. Therefore comparison between different batches of FVIII, and direct comparison of the current results to those obtained in previous studies, could not be performed. However, the studies presented here were all done with a single preparation of each FVIII variant, and therefore these experiments were internally consistent.

All protein C variants were produced in HEK293 cells, and purified yielding 0.6 to 10 mg PC per litre conditioned media. Following activation by FIIa and purification of APC, the purity of each single amino acid variant was assessed on silver stained gels and was estimated at 90% – 95%. R306A/K311A/R312A/R314A-APC expressed very poorly at approximately 40 μg/l, and was estimated on silver stained gels as approximately 75% pure. Active concentration of APC was determined with active site titrations adapted from Chase & Shaw (1967).

Titrations with the chromogenic substrate Pefachrome PCa were used to assess if the mutations in the autolysis loop affected the catalytic site of APC. K_m and k_cat values were derived and did not differ significantly from wt-APC. Only K311A-APC showed a slight decrease in k_cat and for R306A/K311A/R312A/R314A-APC both K_m and k_cat were slightly increased (Table I). These values were consistent with previously published data (Gale et al, 2000; Gale et al, 2002). Unfortunately we were only able to do the titration once with R306A/K311A/R312A/R314A-APC due to the limited amount of protein available.

Table I.

Kinetic properties of APC variants towards the chromogenic substrate Pefachrome PCa.

APC variant	KM (μM ± sd)	kcat (/s ± sd)	kcat/KM (/s/M)
wt	288 ± 1	63.3 ± 14.1	2.2 × 105
306A	271 ± 13	55.7 ± 1.4	2.1 × 105
307A	288 ± 26	54.4 ± 5.7	1.9 × 105
308A	277 ± 13	62.2 ± 1.0	2.2 × 105
311A	300 ± 20	42.3 ± 1.1	1.4 × 105
312A	307 ± 40	61.6 ± 0.1	2.0 × 105
314A	236 ± 25	71.2 ± 3.5	3.0 × 105
R306A/K311A/R312A/R314A	370	71.3	1.9 × 105

Measurements are the average of two separate experiments done in duplicate. Data from R306A/K311A/R312A/R314A-APC was from one experiment done in duplicate. sd, standard deviation.

M662C/D1828C FVIIIa was inactivated by wt-APC and the APC autolysis loop mutants following activation by thrombin (Fig 1). Residual FVIIIa cofactor activity for FIXa was measured in a FXase assay in a time course as described in materials and methods. In M662C/D1828C FVIIIa both R336 and R562 cleavages contribute to the inactivation of FVIIIa. In the absence of APC, M662C/D1828C FVIIIa and the M662C/D1828C FVIIIa cleavage site mutants described below lost approximately 20% activity over the 30 min time course due to incomplete formation of the engineered disulfide crosslink bond, as has been described earlier (Gale et al, 2006). This was subtracted from the inactivation data shown in the relevant figures.

Figure 1. M662C/D1828C FVIIIa inactivation by APC autolysis loop mutants.

FVIIIa residual activity is expressed as percentage of the initial activity at t = 0. A and B, FVIIIa inactivation by 20 nM APC (final concentration). C and D, FVIIIa inactivation by 2.5 nM APC (final concentration) in the absence (open symbols) and presence (filled symbols) of 100 nM protein S. E and F, FVIIIa inactivation by 1 nM APC (final concentration) and 100 nM protein S in the absence (open symbols) and presence (filled symbols) of 12 nM FV. A, C and E, ◇, ◆ = wt-APC; □, ■ = 306A-APC; ○, ● = 307A-APC; △, ▲ = 308A-APC. B, D and F, ◇, ◆ = wt-APC; □, ■ = 311A-APC; ○, ● = 312A-APC; △, ▲ = 314A-APC; ▽, ▼ = R306A/K311A/R312A/R314A-APC. Data represent the average of 3 to 5 individual experiments. Data for R306A/K311A/R312A/R314A-APC was from 2 individual experiments. Error bars were excluded for aesthetic reasons.

After 30 min of incubation with 20 nM wt-APC, ~21 % FVIIIa activity was left (Fig 1A). R306A-APC, K311A-APC, and R314A-APC were all moderately impaired in FVIIIa inactivation compared to wt-APC (Figs 1A & B). K308A-APC was slightly less impaired, but still showed significantly less FVIIIa inactivation than wt-APC. The quadruple mutant, R306A/K311A/R312A/R314A-APC, had the lowest FVIIIa inactivation activity (Fig 1B). E307A-APC and R312A-APC showed no significant difference from wt-APC under any of the experimental conditions (Fig 1) and were therefore not further investigated.

In order to measure FVIIIa inactivation in the presence of protein S, the APC concentration was reduced to 2.5 nM. At this concentration of APC, and in the absence of protein S, the difference between wt-APC and the mutants was minimal as FVIIIa inactivation was minimal. In the presence of 100 nM protein S the difference between wt-APC and mutants became apparent again (Figs 1C & D filled symbols). R306A-APC, K311A-APC and R314A-APC all showed a similar decrease, whereas the K308A-APC mutant showed the same FVIIIa inactivation as wt-APC. The R306A/K311A/R312A/R314A-APC mutant showed severely reduced FVIIIa inactivation. Initially, R306A/K311A/R312A/R314A-APC inactivated FVIIIa similar to K311A-APC and R314A-APC but, in the presence of protein S, FVIIIa activity remained stable after the 4-minute time point, at 75–90%.

APC concentration was further reduced to 1 nM to accommodate for measurements of FVIIIa inactivation in the presence of both protein S and FV (Figs 1E & F). In the presence of 100 nM protein S, the mutations of the APC autolysis loop had the same effect as they did at 2.5 nM APC, although FVIIIa inactivation was obviously slower at 1 nM APC (compare Figs 1C & D filled symbols with 1E & F open symbols). In the presence of FV, R306A-APC mediated FVIIIa inactivation was reduced significantly compared with wt-APC, but only for the first two min of the reaction, after which it no longer differed from wt-APC. The K308A-APC mutant was not different from wt-APC. K311A-APC and R314A-APC showed equally reduced activity in the FVIIIa inactivation reaction. Again R306A/K311A/R312A/R314A-APC showed severely reduced FVIIIa inactivation. FV enhanced FVIIIa inactivation by all APC variants considerably. Overall, residues R306, K311, and R314 were the most important residues in the APC autolysis loop for inactivation of FVIIIa (Fig 1).

Cleavage of M662C/D1828C FVIIIa was confirmed by Western blot analysis. The A1 domain was observed at 48 kDa, and the cleavage product following R336 cleavage was seen at 43 kDa (Fig 2A). The intact A2 domain was 41 kDa and its cleavage product following R562 cleavage appeared at 20 kDa (Fig 2C). A strong band between the 51 and 64 kDa molecular weight marks, and several high molecular weight bands (>97 kDa), were from the BSA present in the samples at ~2000-fold higher concentration than FVIII (indicated with *). The faint band at ~80 kDa in both blots was consistent with a minor fraction of intact heavy chain. The A1 and A2 band intensities were quantified (Figs 2B, D). Intensity values of the observed bands were normalized against the intensity of the heavy chain band at time zero. Cleavages by wt-APC in both domains were significantly faster than cleavages by R306A/K311A/R312A/R314A-APC, which had virtually no cleavage (Figs 2B, D). This confirmed that differences between the APC autolysis mutants in the functional assays of M662C/D1828C FVIIIa inactivation were due to differences in cleavage rates.

Figure 2. Western blot of M662C/D1828C FVIIIa inactivation time course.

A, Membrane developed with the 58.12 antibody against the FVIII A1 domain. B, Quantification of band intensities of the A1 domain, averages of two experiments. C, Membrane from A was stripped, and reprobed with the R8B12 antibody against the FVIII A2 domain. D, Quantification of band intensities of the A2 domain, averages of two experiments. B, D, Several error bars are too small to be visible. ◇ = wt-APC. ▽ = R306A/K311A/R312A/R314A-APC. Wt, wild-type-APC; 306/314, R306A/K311A/R312A/R314A-APC; *, bands from BSA; A1, A1-domain (48 kDa); A1n, n-terminal cleavage product of A1 domain (43 kDa); A2, A2 domain (41 kDa); A2c, C-terminal cleavage product of A2 domain (20 kDa).

The effect of the APC autolysis loop mutants on the individual cleavages in FVIIIa were investigated with APC cleavage site mutants of M662C/D1828C FVIII, under the same experimental conditions as used in Fig 1. Mutation of the R562 cleavage site to glutamine in disulfide crosslinked FVIII (M662C/D1828C/R562Q FVIII) enabled us to isolate the R336 cleavage as sole event for the inactivation of FVIIIa, and mutation of the R336 cleavage site to glutamine in disulfide crosslinked FVIII (M662C/D1828C/R336Q FVIIIa) allowed us to isolate R562 cleavage. For comparison of the APC mutants with wt-APC, the FVIIIa inactivation data were fit to an exponential decay curve and an initial reaction rate constant was derived in nM FVIIIa × /min × /nM APC (as described in materials and methods), and were displayed in Fig 3 as relative rates normalized to wt-APC.

Figure 3. Relative initial reaction rates of R336 and R562 cleavage.

A and B, FVIIIa inactivation by APC alone. C and D, FVIIIa inactivation by APC in the presence of 100 nM protein S. E and F: FVIIIa inactivation by APC in the presence of 100 nM protein S and 12 nM FV. A, C and E, R336 cleavage in M662C/D1828C/R562Q FVIIIa. B, D and F, R562 cleavage in M662C/D1828C/R336Q FVIIIa. Data represent averages of 3 individual experiments. Error bars are standard error of the mean. Labels on X-axis represent the APC mutants, 306/314 is R306A/K311A/R312A/R314A-APC.

In the absence of either cofactor, R306A-APC, K311A-APC and R314A-APC showed a reduction of reaction rate of R336 of ~50% compared to wt-APC, whereas K308A-APC was only slightly reduced (Fig 3A). In the absence of either cofactor, all the single mutants showed a cleavage rate at R562 that did not significantly differ from wt-APC, but R306A-APC, K311A-APC and R314A-APC had somewhat reduced activity (Fig 3B). When FVIIIa was inactivated by APC alone, the autolysis loop seemed to play an important role for the R336 cleavage, but less so for the R562 cleavage. However, under these conditions R562 cleavage was very slow, even for wt-APC, so conclusions are limited.

In the presence of protein S the R306A, K311A, and R314A mutations in APC each resulted in ~40–50% reduction of reaction rate for either cleavage, whereas the cleavage rates with the K308A-APC mutant did not significantly differ from wt-APC (Figs 3C and 3D).

When both FV and protein S were added to the reaction, R306A-APC, K311A-APC, and R314A-APC showed ~50% reduced reaction rate for R336 cleavage, and the K308A-APC mutant showed a reduction of ~30% (Fig 3E). Cleavage at R562 in the presence of FV and protein S showed reduced reaction rates for R306A-APC, K311A-APC and R314A-APC by ~60%, ~55%, and ~40% respectively. The K308A-APC mutant however, did not show a significant difference from wt-APC (Fig 3F).

Under all conditions the R306A/K311A/R312A/R314A-APC mutant showed a reduction of ~90% of reaction rate compared to wt-APC. However, due to the very low reaction rates, which resulted in a relatively large amount of scatter, these data did not yield good curve fits with the exponential decay equations, and therefore these numbers were approximate. Still, it was obvious that cleavage of both R336 and R562 by R306A/K311A/R312A/R314A-APC was severely reduced.

Table II shows the fold increase of the initial reaction rates induced by the APC cofactors protein S and FV for the individual cleavages of FVIIIa. The fold increase was calculated by dividing the rate (nM FVIIIa × /min × /nM APC) in presence of the cofactor by the rate in absence of the cofactor. Overall these data showed no significant difference between the APC mutants and wt-APC for either the protein S or FV cofactor effects for either cleavage. However, it was apparent that both protein S and FV had a larger stimulatory effect on cleavage at R562 than at R336. Unfortunately due to the low reaction rate and relatively high amount of scatter in the data obtained with R306A/K311A/R312A/R314A-APC, we were unable to confidently determine a fold increase in rate due to the presence of the APC cofactors for this mutant.

Table II.

Fold increase of initial reaction rate due to presence of protein S or FV.

APC variant	M662C/D1828C/R562Q FVIIIa R336 cleavage		M662C/D1828C/R336Q FVIIIa R562 cleavage
	Prot. S	Prot. S + FV	Prot. S	Prot. S + FV
wt	7.7 ± 1.1	3.4 ± 0.6	61 ± 21	7.6 ± 1.4
306A	7.1 ± 1.0	3.4 ± 0.7	37 ± 12	6.6 ± 1.5
308A	9.3 ± 1.6	2.4 ± 0.4	43 ± 13	7.0 ± 1.3
311A	11.3 ± 3.1	2.6 ± 0.6	36 ± 14	7.7 ± 1.4
314A	7.1 ± 1.0	4.1 ± 0.7	48 ± 24	9.6 ± 1.7

Values are the ratios of the rate in the presence of the cofactor divided by the rate in the absence of the cofactor, ± SEM.

The results of Figs 1 and 3 have been summarized in Table III, which is intended as a reference as to which residue of the APC autolysis loop is important for which cleavage in FVIIIa inactivation.

Table III.

Reference guide for significant effect of charged residue in the APC autolysis loop.

APC residue	wt-FVIIIa			R336 cleavage			R562 cleavage
	APC	+ PS	+ PS/V	APC	+ PS	+ PS/V	APC	+ PS	+ PS/V
R306	+	+	+	+	+	+	x	+	+
E307	x	x	x	nd	nd	nd	nd	nd	nd
K308	+	x	x	+	x	+	x	x	+
K311	+	+	+	+	+	+	x	+	+
R312	x	x	x	nd	nd	nd	nd	nd	nd
R314	+	+	+	+	+	+	x	+	+

+, mutation of this residue to Ala resulted in decreased inactivation of FVIIIa under this condition. x, mutation of this residue to Ala had no effect on FVIIIa inactivation under this condition. nd, not determined.

A mutant of APC with residue E309 mutated to alanine was also made and assayed in the same experiments as the other APC mutants described in this report. However, analysis of E309A-APC on silver stained gels and Western blots revealed high levels of heavy chain degradation, and therefore this APC mutant was excluded from the current study. Wt-APC can be cleaved at two positions during purification, after H10 in the light chain, and after K308 in the heavy chain (Baker et al, 1998). We have found degradation of the heavy chain in E309A-APC and low molecular weight fragments that reacted with antibodies directed against the heavy chain, but no fragments of the light chain or degradation of the light chain, which suggested that K308 cleavage was enhanced in this mutant. Further investigations will be required however to confirm these speculations.

DISCUSSION

The present study has shown that the autolysis loop of APC contributes significantly to FVIIIa inactivation. Mutation of residues R306, K311 or R314 in this loop each resulted in significant decrease of FVIIIa inactivation by APC. A previous study has shown that those same residues are also important for inactivation of FVa (Gale et al, 2000). Residues E307 and R312 however, which are important for FVa inactivation, did not affect FVIIIa inactivation. Residue K308 of the autolysis loop was a minor contributor to FVIIIa inactivation

The APC autolysis loop did not differentiate between the two inactivating cleavages in FVIIIa, at R336 and R562. This correlates well with other studies of the basic exosite of APC in FVIIIa inactivation. Manithody et al. (2003) investigated loop 37, loop 60, and loop 70 of the basic exosite by alanine scanning studies, and showed that both cleavages in FVIIIa were affected equally by mutations of basic residues in these loops. This is distinct from the impact of the basic exosite of APC on inactivation of FVa, where mutations in all loops resulted in a larger reduction of cleavage at R506 than at R306 (Gale et al, 2000; Friedrich et al, 2001; Gale et al, 2002). This collection of studies used the same APC mutants, and used similar experimental settings in purified assays. Thus, in spite of the homology of FVa and FVIIIa, these two substrates of APC are distinct.

Compared to other coagulation serine proteases, the autolysis loop of APC is five residues longer, except for the FIIa autolysis loop which is one residue longer than that of APC (Yang et al, 2007). Several studies in which the APC autolysis loop was replaced with the shorter one of FXa, have shown that the length of this loop plays a pivotal role in the interaction with substrate and plasma serpins (Yang et al, 2004; Yang et al, 2005). The shorter autolysis loop resulted in a faster inactivation of FVa and FVIIIa and a higher amidolytic activity towards chromogenic substrate (Yang et al, 2005), indicating that the APC autolysis loop is not optimally configured for the inactivation of these substrates. But this mutant APC also showed a severely shortened half-life in a plasma milieu due to improved interaction with plasma serpins. This suggests an advantage of the wt-APC autolysis loop over that of other coagulation serine proteases by allowing for a longer half-life of APC (Yang et al, 2004; Yang et al, 2005; Yang et al, 2007).

The charge of the autolysis loop also plays a role in the different interactions. Mutation of individual charged residues in the APC autolysis loop affected the interaction with FVa (Gale et al, 2000). But also mutation of the charged residues in the APC variant with the FXa autolysis loop resulted in impaired FVa inactivation (Yang et al, 2007). Furthermore in FXa itself, individual mutations to Ala of the positively charged residues resulted in reduced inactivation by antithrombin and tissue factor pathway inhibitor, and impaired activation of prothrombin in the prothrombinase complex (Manithody et al, 2002). Similar results were found for the autolysis loop of FIXa (Yang et al, 2003). Together these studies suggested that the specific sequence of the autolysis loop in these plasma serine proteases is important for the interactions with their various substrates and inhibitors. Our current results are in agreement with these studies, as single mutations of the charged residues of the APC autolysis loop inhibited FVIIIa inactivation.

Protein S stimulated the inactivation of FVIIIa significantly, independent of mutations in the autolysis loop. It has been previously suggested that protein S does not modulate the APC autolysis loop (Qureshi et al, 2008) or the other loops of the basic exosite (Manithody et al, 2003), and our current results support this idea. The fold increase of initial reaction rate for the R336 cleavage (in M662C/D1828C/R562 FVIIIa) was similar to previously published results (Gale et al, 2008). Cleavage at R562 however (in M662C/D1828C/R336 FVIIIa), appeared to be stimulated by protein S much more than we previously reported, comparing a 7-fold increase of previous measurements (Gale et al, 2008) to a 60-fold increase in our current results. The difference between these results and those previously published cannot be fully explained. However, in our previous study (Gale et al, 2008), FVIIIa was inactivated with plasma-derived APC, rather than the recombinant APC that was used in the current study. Although the recombinant APC was required in order to allow direct comparison between wt-APC and the autolysis loop mutants of APC, there may be differences between plasma-derived and recombinant APC that influence APC activity toward macromolecular substrates. Another source of variation may be the new (synthetic) phospholipids, that we were forced to switch to because the natural phospholipids used in our previous study (Gale et al, 2008) were no longer commercially available. Nevertheless, as we previously described, the cofactors protein S and factor V enhance R562 cleavage more than R336 cleavage. And, our primary conclusion, that protein S and factor V cofactor effects are independent of the autolysis loop, is unchanged.

This study, combined with the current literature, suggests that the basic exosite of APC is not modulated by either protein S or FV. This is consistent with reports suggesting that APC interaction with protein S is mediated primarily through interaction with residues 35–39 of the Gla domain of APC (Smirnov et al, 1998; Preston et al, 2006; Harmon et al, 2008). The specific binding interactions between anticoagulant factor V and APC or protein S have not yet been described.

In addition to the presence of protein S, the APC cofactor effect of FV requires cleavage of FV by APC at R506. Mutations in the autolysis loop however, did not change the FV cofactor effect significantly under our experimental conditions. Although R506 cleavage in activated FV (FVa) was profoundly influenced by the autolysis loop (Gale et al, 2000), our results may suggest that cleavage of intact FV at R506 was not significantly affected, because no significant difference in FV cofactor effect was observed between the APC autolysis loop mutants.

In summary, our results show that the autolysis loop of APC is involved in the interaction with FVIIIa. In particular, residues R306, K311 and R314 strongly contribute to the inactivation of FVIIIa, and K308 to a lesser extent. The autolysis loop did not differentiate between the R336 and R562 cleavages in FVIIIa, as both cleavages were affected equally by mutation in this loop. Furthermore, protein S and FV enhanced cleavage at R562 more than cleavage at R336, and did so similarly in all APC mutants. Thus, the autolysis loop is probably not modulated by either protein S or FV.