Dr Charles T. Esmon, Howard Hughes Medical Institute, Oklahoma Medical Research Foundation, 825 NE 13th Street, Oklahoma City, OK 73104, USA. Tel.: +405 2717571; fax: +405 2713137; e-mail: Charles-Esmon@omrf.ouhsc.edu
Summary. Activated protein C (APC) serves as an ‘on demand’ anticoagulant. Defects in the APC anticoagulant pathway are underlying risk factors for the development of venous and arterial thrombosis. APC has recently been shown to significantly reduce mortality in patients with severe sepsis, presumably by virtue of its ability to down-regulate coagulation as well as inflammation. Our objective was to develop an assay that, for the first time, permits rapid detection of plasma APC. This assay will expedite studies of APC in a variety of vascular disease states including sepsis, severe atherosclerosis, diabetes, and vasculitis. By generating a highly APC-specific monoclonal antibody (HAPC 1555), we have developed an assay that, for the first time, allows rapid detection of plasma APC. The Kd measured for the interaction between APC and HAPC 1555 based on BIAcore studies and binding to immobilized HAPC on microtiter plates is 6.2 ± 0.9 and 8.8 ± 1.0 nmol L−1, respectively. The interaction between HAPC 1555 and APC is Ca2+-dependent, with a Ca2+ concentration of 313 ± 48 µmol L−1 required for half maximal binding. HAPC 1555 interferes with APC-mediated inactivation of factor (F)Va in the presence and absence of phospholipids, suggesting that HAPC 1555 binds to the FVa binding domain of APC. When HAPC 1555 was used in an APC enzyme capture assay, therapeutic APC levels could be measured in 1.5 h, and physiologic levels of APC could be detected between 3 and 19 h. APC levels were also shown to vary markedly in patients with severe sepsis. The rapidity of our APC assay makes APC detection in patients practical clinically. This assay will expedite studies of APC in a variety of vascular disease states including sepsis, severe atherosclerosis, diabetes, and vasculitis.
The protein C anticoagulant pathway is well established as a physiologically important mechanism for inhibiting blood coagulation . The pathway is initiated when thrombin binds to thrombomodulin (TM) on the endothelial cell surface. The thrombin–TM complex activates protein C to generate APC. APC, in combination with its cofactor protein S, acts as an anticoagulant by inactivating factor (F)Va and FVIIIa, key cofactors in coagulation. Protein C activation is augmented by the endothelial cell protein C receptor (EPCR), an endothelial cell-specific receptor that binds protein C and APC with comparable affinity. Physiologically, the APC plasma concentration appears to be proportional to both the thrombin concentration  and protein C concentration .
Defects in the protein C anticoagulant pathway are associated with an increased risk for venous thrombosis [4–6] and myocardial infarction [7–9]. The protein C pathway also plays a critical role in host defense against bacterial infection, particularly in preventing severe sepsis, a disease that impacts about 700 000 people in the USA annually and retains a mortality rate of 30–50% . In vivo, APC protects baboons from the lethal effects of Escherichia coli infusion . When protein C activation is blocked , APC function is impaired , or the interaction between EPCR and protein C/APC is blocked, a sublethal concentration of E. coli becomes lethal , resulting in an intense coagulopathy and elevated cytokine levels. The protective effects of APC probably result from its multifunctional properties. In addition to its anticoagulant activities, APC is profibrinolytic by virtue of its ability to inactivate plasminogen activator inhibitor-1 (PAI-1) [14,15], a process that is stimulated markedly by vitronectin . PAI-1 inactivation results in an increase in the plasma concentration of tissue plasminogen activator activity. In vitro studies have shown that APC is anti-inflammatory since it down-regulates production of proinflammatory cytokines in monocytes , blocks neutrophil interactions with selectins  and suppresses the expression of leukocyte adhesion molecules in endothelial cells .
Recently, recombinant APC has been shown to significantly reduce mortality in patients with severe sepsis, with treatment associated with a reduction in the relative risk of death of 19.4% . Part of the rationale for the use of APC to treat sepsis was that protein C levels drop markedly in severe sepsis [21,22] with the extent of the decrease correlating with an increased risk of death. The decrease in protein C levels can be rapid and often occurs hours before the symptoms of severe sepsis are manifested . Further compromising this system is the possibility of down-regulating the protein C activation complex. Both TM and EPCR have been shown to be down-regulated in cell culture by inflammatory cytokines like tumor necrosis factor α[23,24] and interleukin 1 β[24,25]. In a subset of patients with meningococcemia, major decreases in the ability to activate protein C were found . This appeared to be due to major decreases in TM and EPCR expression.
Given the potential for down-regulation of the protein C activation complex and the demonstrated roles for APC in host defense against sepsis, knowledge of the endogenous plasma levels of APC may be useful in the clinical management of patients with severe sepsis. These APC assays would allow evaluation of the extent to which the endogenous protein C activation complex has been compromised, which in turn might be useful in deciding whether to use protein C or APC therapeutically. Protein C therapy may be beneficial in patients with ongoing coagulopathy without generalized down-regulation of TM and EPCR on vascular endothelium, whereas APC therapy can bypass the endothelial defect because the enzyme does not require endothelial TM and EPCR for its function. APC assays may also be useful in monitoring the dose and duration of recombinant APC therapy in patients. The currently available enzyme capture assay for APC using the microtiter plates, while relatively accurate, is not useful for monitoring these patients since the assay may take 3 weeks to develop . The major reasons for the very long times required are the low levels of circulating APC (about 3 ng mL−1 plasma in normal individuals  and the very high concentration of protein C (3 µg mL−1) relative to the enzyme. With the relatively low capacity of the microtiter plates, the plasma must be diluted at least 30-fold to allow capture of most of the protein C and APC. This problem could be circumvented if the capture antibody exhibited a high degree of specificity toward APC.
In this study, we describe the generation and characterization of a monoclonal antibody with a high degree of specificity toward human APC. Such antibodies when adsorbed to surfaces such as microtiter plates have improved capacity for capturing APC from plasma and hence make the direct detection of APC much faster. This assay will expedite studies of APC plasma levels not only in sepsis, but also in a variety of vascular disease states including severe atherosclerosis, diabetes, and vasculitis, where evidence already exists for thrombomodulin down-regulation.
Human protein C (HAPC) and APC were prepared as described previously ). Human FVa was from Haematologic Technologies (Essex Junction, VT, USA). Spectrozyme PCa was from American Diagnostica. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine (PC) and 1-palmitoyl-2-oleolyl-sn-glycero-3-phosphatidylserine (PS) were from Avanti Polar Lipids, Inc. (Alabaster, AL, USA).
Generation of mouse antihuman APC monoclonal antibodies
Mouse monoclonal antibodies (mAbs) against human APC were developed by standard techniques . Male Balb/c mice were immunized with 25 µg of active-site blocked APC and 25 µg recombinant S195A human APC produced by standard approaches  mixed with an equal volume of Freund's complete adjuvant in a total volume of 100 µL. Mice were boosted for fusion with 20 µg of each antigen in PBS mixed to a total volume of 200 µL.
Screening for APC-specific mAbs by ELISA
Briefly, 96-well vinyl microtiter plates (Costar) were coated with 2 µg mL−1 of S195A APC or protein C in 20 mmol L−1 Tris-HCl, pH 7.5, 100 mmol L−1 NaCl, 5 mmol L−1 CaCl2. After overnight incubation at 4 °C, the plates were blocked for 1 h at room temperature with 300 µL of 1% bovine serum albumin (BSA) in 50 mmol L−1 Tris-HCl, pH 7.5, 100 mmol L−1 NaCl, 5 mmol L−1 CaCl2. The blocking solution was removed and 50 µL of each supernatant was applied to each well for 1 h at room temperature. After one wash in 50 mmol L−1 Tris, pH 7.5, 100 mmol L−1 NaCl, 5 mmol L−1 CaCl2, 0 .02% Tween 20, goat antimouse horseradish peroxidase-labeled antibody diluted into blocking solution was applied at 50 µL per well for 1 h. After three additional washes, ABTS [2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid)-diammonium salt] substrate (Pierce, Rockford, IL, USA) was added and observed for color development at 405 nm. Large-scale production of HAPC 1555 was performed in roller bottles using serum-free medium (Gibco BRL) and purification of HAPC1555 was accomplished using MEP Hyper-Cel resin (Gibco BRL).
Real-time biomolecular interactions between APC and HAPC 1555 were studied using a BIAcore™ 1000 biosensor instrument (BIAcore Inc.). HAPC 1555 was covalently coupled to a carboxymethyl dextran (CM5) sensor chip through its primary amine groups according to the manufacturer's instructions. Binding of APC to immobilized HAPC 1555 was monitored by measuring changes in RU (1000 RU corresponds to ≈1 ng of bound protein mm−2). Unless otherwise stated, all experiments were performed at 25 °C at a flow rate of 10 µL min−1 in 20 mmol L−1 Hepes, pH 7.5, 150 mmol L−1 NaCl, 3 mmol L−1 CaCl2, and 0.005% surfactant P-20 (BIAcore grade).
For each set of experiments, APC was introduced onto the surface of a sensor chip that lacked immobilized HAPC 1555 (control sensor chip). The sensograms of the control sensor chip were subtracted from the sensograms of the HAPC 1555-containing flowcells to remove the effects of non-specific binding to the dextran surface. After each protein injection, the sensor chips were regenerated by the injection of 20 µL of 1 mol L−1 glycine, pH 2.5, followed by washing for 3 min with buffer before reinjecting APC for the next cycle. To determine the apparent Kd for the interaction between APC and HAPC 1555, the maximum RUs of the binding isotherms were plotted vs. APC concentration. The Kd values were calculated by non-linear regression analysis of the curves using the Michaelis–Menten equation in TableCurve (Jandel Scientific, San Rafael, CA, USA).
APC binding to HAPC 1555 on microtiter plates
The affinity of APC for immobilized HAPC 1555 was also measured on microtiter plates. Ninety-six well flat-bottom polystyrene microtiter plates (Dynex Technologies Inc., Chantilly, VA, USA) were coated with 100 µL of HAPC 1555 (5 µg mL−1) in coating buffer (20 mmol L−1 Hepes, pH 7.5, 150 mmol L−1 NaCl, 5 mmol L−1 CaCl2) for 2 h at 37 °C. The plates were then blocked for 1 h at 37 °C with 200 µL of blocking buffer (coating buffer containing 1% BSA). 100 µL of APC (from 50 to 2000 ng mL−1) in blocking buffer was added to the HAPC 1555-coated wells. After 30-min incubation at 23 °C, 100 µL of the unbound APC was removed from the wells and added to 100 µL of 1 mmol L−1 Spectrozyme PCa (in coating buffer). The amount of unbound APC was determined by comparing the chromogenic activity of the unbound APC to that of an APC chromogenic activity standard curve (the standard curve consists of APC concentrations from 50 to 2000 ng mL−1). Substrate hydrolysis of the unbound APC was monitored after 12-min incubation at 23 °C at 405 nm in end-point mode using a Spectromax plate reader (Molecular Devices). The incubation time of 12 min has previously been shown to be in the linear phase of chromogenic substrate hydrolysis for APC concentrations ranging from 50 to 2000 ng mL−1 (i.e. absorbance reading at 405 nm after 12 min incubation is proportional to the APC concentration). The amount of APC bound to each well was determined by adding 100 µL of 0.5 mmol L−1 Spectrozyme PCa (in coating buffer) to the wells. The amount of bound APC was determined by comparing the chromogenic activity of the bound APC to that of an APC chromogenic activity standard curve as described above. Graphs of bound APC vs. unbound APC were plotted, and the Kd value was determined by fitting non-linear binding isotherms with a hyperbolic equation using the TableCurve program.
Ca2+ dependence of APC binding to HAPC 1555 antibody
One hundred microliter aliquots of APC (final concentration of 100 ng mL−1) was added to HAPC-coated wells (described above) and incubated at room temperature for 30 min the APC aliquots were made in 20 mL Hepes, pH 7.5, 150 mM NaCl, 1% BSA containing increasing concentrations of CaCl2 (from 0 to 5 mM). As a control, APC was added to microtiter wells coated with 100 µL of BSA (5 µg mL−1). After washing the wells twice, each time for 10 min, by gentle agitation at room temperature with wash buffer (coating buffer containing 0.05% Tween-20), the chromogenic activity of bound APC was determined by the addition of 100 µL Spectrozyme PCa (0.5 mmol L−1) in coating buffer. The plates were incubated for 2 h at 37 °C, and substrate hydrolysis was monitored at 405 nm in end-point mode. The incubation time of 2 h has previously been shown to be in the linear phase of chromogenic substrate hydrolysis (i.e. absorbance reading at 405 nm is proportional to APC concentration). Substrate hydrolysis of APC bound to BlA-coated wells (non-specific binding) was subtracted from that of APC bound to HAPC 1555-coated wells.
Effect of protein C on APC binding to immobilized HAPC 1555
One hundred microliters of APC (10 ng mL−1 in blocking buffer) containing increasing concentrations of protein C (0–50 µg mL−1) were added to HAPC 1555-coated wells (described above) and incubated at 23 °C for 30 min. The wells were washed (2 × 10 min) and the chromogenic activity of bound APC was determined by the addition of 100 µL Spectrozyme PCa (0.5 mmol L−1) in coating buffer. The plates were placed at 37 °C, and substrate hydrolysis was monitored at 405 nm in end-point mode over time.
Influence of HAPC 1555 on APC-mediated hydrolysis of chromogenic substrates
The chromogenic activity of 5 nmol L−1 APC in blocking buffer was determined with 0–1 mmol L−1 Spectrozyme PCa in the absence or presence of 400 nmol L−1 of HAPC 1555. The Km values were calculated by non-linear regression analysis using the Michaelis–Menten equation in TableCurve.
Influence of HAPC 1555 mAb on FVa Inactivation by APC
FVa (50 nmol L−1) was incubated at 37 °C with 0.2 nmol L−1 APC, 20 µg mL−1 PC:PS liposomes, and the absence or presence of HAPC 1555 (20 nmol L−1) in buffer containing 20 mmol L−1 Hepes, pH 7.5, 150 mmol L−1 NaCl, 5 mmol L−1 CaCl2, and 0.1% gelatin. As a control, the reaction was also performed in the presence of HPC4 (20 nmol L−1), a monoclonal antibody against human protein C that does not bind to APC . The reaction was stopped at 1, 2, 5, 10, or 30 min by the addition of 10 mmol L−1 benzamidine HCl (Sigma, St Louis, MO, USA). Remaining FVa activity was calculated by dividing the rate of thrombin formation in the presence of FVa treated with APC by the rate of thrombin formation with untreated FVa. The studies were also performed in the absence of PC:PS phospholipid vesicles using the APC and mAb concentrations of 10 nmol L−1 and 100 nmol L−1, respectively.
Quantification of APC levels in human plasma
Venous blood (4.5 mL) was drawn via 19-gauge needles from healthy volunteers or from patients with severe sepsis into syringes containing 0.5 mL of 0.105 mol L−1 trisodium citrate and 100 µL of 1 mol L−1 benzamidine HCl (i.e. 20 mmol L−1 benzamidine final). Benzamidine HCl, a reversible inhibitor of APC, is necessary at the time of blood collection to block irreversible inhibition of APC by plasma protease inhibitors. The benzamidine is removed during the APC assay wash steps, thus restoring the enzymatic activity of APC towards chromogenic substrates. Patients with severe sepsis were eligible for this study based on the criteria of . Plasma was stored as aliquots in − 80 °C. Informed consent was obtained under a study protocol approved by the Hamilton Health Sciences Research Ethics Board.
Because HAPC 1555 is a Ca2+-dependent antibody, prior to the APC assays the trisodium citrate benzamidine-containing plasma samples were anticoagulated and recalcified by supplementing with heparin, Hepes pH 7.5, and CaCl2 to final concentrations of 2 U mL−1, 20 mmol L−1, and 10 mmol L−1, respectively. Alternatively, hirudin (1.5 µmol L−1) can be used as an anticoagulant instead of heparin (the stability of APC is the same whether heparin or hirudin are used). Plasma samples (100 µL) were then transferred to HAPC 1555-coated microtiter plates and incubated at room temperature for 30 min. The wells were washed (2 × 10 min) and the chromogenic activity of bound APC was determined by the addition of 100 µL Spectrozyme PCa (0.5 mmol L−1) in coating buffer. The plates were placed at 37 °C, and substrate hydrolysis was monitored at 405 nm in end-point mode over time.
A standard curve for the quantitative measurement of APC in plasma was generated as follows. Increasing amounts of APC (from 0 to 100 ng mL−1) were spiked into 20 mmol L−1 Hepes, pH 7.5, 2 U mL−1 heparin, 20 mmol L−1 benzamidine, 5 mmol L−1 CaCl2, 1% BSA. 100 µL samples were incubated with immobilized HAPC 1555 at room temperature for 30 min.
Determination of binding capacity of HAPC 1555-coated microtiter wells for APC
One hundred microliter aliquots of APC (from 0 to 10 µg mL−1) in blocking buffer was added to HAPC 1555-coated wells (described above) and incubated at room temperature for 30 min. The wells were washed (2 × 10 min) and the chromogenic activity of bound APC was determined by the addition of 100 µL Spectrozyme PCa (0.5 mmol L−1) in coating buffer. The amount of APC bound to the wells was determined by the comparing the chromogenic activity of the bound APC to that of an APC chromogenic activity standard curve.
Quantification of protein C and F1 + 2 levels in human plasma
ELISAs were employed to measure levels of protein C antigen (Affinity Biologicals Inc.) and F1 + 2 (Dade Behring, Inc.) in plasma collected into trisodium citrate.
Determination of the affinity of HAPC 1555 for APC
HAPC 1555, a monoclonal antibody relatively specific for APC was generated as described in Methods. HAPC 1555 is an IgG 1 Kappa antibody. The affinity of HAPC 1555 for APC was determined in two ways: (i) by BIAcore studies and (ii) by direct binding studies to immobilized HAPC 1555 bound to microtiter plates. In the BIAcore studies, various concentrations of APC (from 8.1 to 68.5 nmol L−1) were passed over a sensor chip containing immobilized HAPC 1555. The overlaid dose–response binding curves are shown in Fig. 1. To determine the apparent Kd for the interaction between APC and HAPC 1555, the maximum RUs of the binding isotherms shown in Fig. 1 were plotted vs. APC concentration. HAPC1555 bound to APC with a Kd of 6.2 ± 0.9 nmol L−1. The Kd value represents the mean and standard error of two separate experiments. In the absence of CaCl2, there is no binding of either APC or protein C to immobilized HAPC 1555, suggesting that there is obligatory Ca2+ binding to antigen and/or antibody (data not shown).
As an alternative method, increasing amounts of APC were added to microtiter wells coated with HAPC 1555 and the amount of bound and unbound APC was determined as described under Methods. Graphs of bound APC vs. unbound APC were plotted (data not shown) and the Kd value of APC binding to immobilized HAPC 1555 was determined by non-linear regression analysis. The Kd value is 8.8 ± 1.0 nL, based on three separate experiments.
Ca2+ dependence of APC binding to HAPC 1555 antibody
The Ca2+ dependence of APC binding to HAPC 1555 antibody was determined by incubating APC with immobilized HAPC 1555 in the presence of increasing concentrations of CaCl2. The Ca2+ concentration required for half maximal binding to HAPC 1555 is 313 ± 48 µmol L−1 (Fig. 2) based on three separate experiments.
Effect of protein C on APC binding to immobilized HAPC 1555
Increasing concentrations of protein C were analyzed for their ability to compete with 10 ng mL−1 APC for binding to the immobilized HPAC 1555, as described in Methods. As shown in Fig. 3, the IC50 value for inhibition of APC binding to HAPC 1555 by protein C is 13.4 ± 0.5 µg mL−1. Protein C at physiologic concentrations (i.e. 3–4 µg mL−1) blocked approximately 20% of the binding of APC to immobilized HAPC 1555. Thus, in this setting which is a purified system mimic of the plasma enzyme capture assay, physiologic concentrations of protein C have little effect on blocking the capture of APC. In most septic patients, the inhibition will be less due to the decreased protein C levels.
Effect of HAPC 1555 on FVa inactivation by APC
The effect of HAPC 1555 on FVa inactivation by APC was determined in the presence (Fig. 4a) or absence (Fig. 4b) of phospholipid. HAPC 1555 inhibited FVa inactivation under both conditions indicating that the antibody probably blocks the FVa-APC binding site.
Development of an APC enzyme capture assay for the measurement of plasma APC levels
The principle of an APC enzyme capture assay is that APC is captured from solution using an immobilized antibody and the amount of APC captured is quantified by measuring the amidolytic activity of the immobilized APC toward a chromogenic substrate. HAPC 1555 does not alter the kinetic parameters of APC toward Spectrozyme PCa, making HAPC 1555 a useful monoclonal antibody for the APC assay (Fig. 5). The results show that the kinetic parameters of APC toward the synthetic substrate are not affected by the presence of HAPC 1555. An APC standard curve for the enzyme capture assay was made as described in Methods. The specificity of the enzyme capture assay for APC was assessed in two ways. First, immobilization of HPC4 (an antiprotein C monoclonal antibody that does not bind to APC) or BSA onto the microtiter wells reduced the chromogenic activity of APC to that of the background, suggesting that HAPC 1555 is necessary to capture the APC present in solution (data not shown). Second, preincubation of 50 ng mL−1 APC with 500-fold molar excess of HAPC 1555 also reduced the amidolytic activity to that of background, which likely reflects solution phase competition for APC binding to the plate-bound HAPC 1555 (data not shown).
We then determined the speed at which our APC enzyme capture assay can detect therapeutic and physiologic concentrations of APC. As shown in Fig. 6(a), the assay is sensitive to APC levels of 10–50 ng mL−1 with a color development time of ≤ 1 h. For reference, the therapeutic dose of recombinant APC used in the PROWESS sepsis study  was 24 µg kg body weight h−1, which corresponds to circulating concentration of 70–80 ng mL−1. Thus, this APC assay permits the rapid detection of therapeutic doses of APC in approximately 1.5 h (i.e. 1-h incubation time, 2 × 10-min washes, 15-min color development time).
Based on previous studies, the physiologic concentration of APC in healthy adult volunteers ranges from 1.4 to 3.2 ng mL−1[27,32]. In this study, the physiologic concentration of APC in 10 healthy adults ranges from 0.32 to 1.78 ng mL−1 (described in more detail below). As shown in Fig. 6(b), the assay is sensitive to APC levels of ≤ 2 ng mL−1 with a color development time of ≤ 3 h. For APC concentrations of ≤ 0.5 ng mL−1, color development times of ≤ 19 h are required. Thus, to detect physiologic concentrations of APC, the APC assay requires color development times of between 3 and 19 h.
Quantification of plasma APC levels in healthy volunteers and in patients with severe sepsis
To demonstrate the utility of the enzyme capture assay in quantifying circulating levels of APC, we measured physiologic plasma APC levels in 10 healthy volunteers and in six patients with severe sepsis. The mean plasma APC levels in 10 healthy volunteers is 1.30 ± 0.54 ng mL−1 (range from 0.32 to 1.78 ng mL−1), consistent with values obtained by other methods [27,32]. The APC levels in six patients with severe sepsis are shown in Fig. 7, presented as percentages relative to APC levels in normal pooled plasma. The absolute APC levels in these patients range from 0.5 to 3.17 ng mL−1. The coefficient of variation for intra-assay variability is 7.4%, from samples run five times within one run. The coefficient of variation for interassay variability is 14.9%, from samples run on five separate days.
We also measured levels of protein C and F1 + 2 (a marker of thrombin generation) in these samples. The mean plasma protein C and F1 + 2 levels in 10 healthy volunteers is 0.89 ± 0.15 U mL−1 and 0.84 ± 0.19 nmol L−1, respectively. Relative to normal pooled plasma, plasma from all six patients had elevated F1 + 2 levels (indicative of thrombin generation) and reduced protein C levels.
Determination of the maximum binding capacity of HAPC 1555-coated microtiter wells for APC
As mentioned above, the therapeutic dose of recombinant APC used in the PROWESS sepsis study is 70–80 ng mL−1. Since our APC assay protocol uses 100 µL of plasma samples, 100 µL plasma samples from septic patients receiving recombinant APC therapy could contain up to 7–8 ng APC. The maximum binding capacity of a microtiter well coated with 100 µL of 5 µg mL−1 HAPC 1555 for APC (described in Methods) is 22 ± 2 ng APC (data not shown). Thus, plasma samples from patients receiving recombinant APC therapy do not need to be diluted prior to adding to HAPC-coated microtiter wells.
As endogenous APC generation occurs on the endothelial cell surface, vascular endothelium dysfunction can potentially reduce the capacity to generate APC through the down-regulation of EPCR and TM levels, as has been shown in the case of meningococcemia . In a phase three clinical trials, recombinant APC infusion was shown to significantly reduce mortality in patients with severe sepsis , presumably by virtue of its ability to inhibit coagulation as well as inflammation. Many of the patients in the treatment group who ultimately died, did so after the treatment was terminated. Furthermore, to a variable extent, fibrin degradation products rose after APC infusion was halted. We hypothesize that some of these patients have badly impaired protein C activation potential and would benefit from prolonged infusion of APC. To test this hypothesis in this and other groups of patients, there is a need for a convenient laboratory assay that can evaluate a patient's capacity to endogenously generate APC.
Because EPCR and TM are expressed on endothelium, it is not possible to determine directly how well they are functioning without removal of blood vessels. An alternative approach is to measure the levels of APC in circulation. However, there has not been an APC assay that was both convenient and relatively rapid [27,32]. The APC assay described by Orthner et al.  utilizes a monoclonal antibody that recognizes the Gla-domain of APC, a region that is highly conserved among members of the vitamin-K dependent plasma proteins. The assay utilizes immunosorbent beads which results in a more rapid and sensitive assay as compared to antibody-coated microtiter wells, but the assay is not easily automated. A different assay strategy measures the complexes formed between APC and its two major plasma inhibitors, protein C inhibitor and α1-antitrypsin, in the presence of heparin. The APC:inhibitor complexes formed are subsequently measured by specific ELISAs. The main limitations of this method are that (i) variations in the levels of these inhibitors could potentially affect the recovery of APC; and (ii) the assay requires several hours to perform.
HAPC 1555 exhibits a high degree of APC specificity and a relatively high affinity. The Kd for the interaction between APC and HAPC 1555 determined by BIAcore studies and by direct binding analysis to HAPC 1555 immobilized on microtiter plates is 6.2 ± 0.9 and 8.80 ± 1.0 nmol L−1, respectively. The interaction between HAPC 1555 and APC is Ca2+ dependent, with a Ca2+ concentration of 313 ± 48 µmol L−1 required for half maximal binding of APC to HAPC 1555 (Fig. 2). Because HAPC 1555 interferes with APC-mediated inactivation of FVa both in the presence and absence of phospholipids (Fig. 4), the antibody likely contacts the FVa binding site of APC. Recently, a three-dimensional molecular model of FVa bound to APC was constructed which predicts specific interactions between acidic residues in the FVa A2 domain and an extended basic exosite of APC that includes the autolysis loop in the enzyme's protease domain . The calcium binding site in APC is on nearly the opposite side of APC at the end of an extended groove that has been suggested to be a FVa binding site . Molecular models of protein C  and some biochemical experiments reviewed therein are compatible with the concept that the protein C activation peptide region of the zymogen is located very close to the calcium binding site and may interact with this site. Based on this modeling, the calcium dependence for an APC specific antibody could be explained easily if the epitope were near the calcium binding site where the activation region of the zymogen would interfere with antibody access. Since binding to APC is calcium dependent, this interpretation also implies that the activation peptide is near a FVa recognition site on APC that undergoes a calcium dependent structural change.
The APC assay described herein permits the detection of therapeutic doses of APC (approximately 70–80 ng mL−1 plasma ) in approximately 1.5 h (i.e. 1-h plasma incubation time, 2 × 10-min washes, 15-min color development time). To detect physiologic concentrations of APC, this assay requires color development times of between 3 and 19 h. We demonstrated the utility of our APC enzyme capture assay by quantifying physiologic plasma APC levels in six patients with severe sepsis (Fig. 7). Despite increased thrombin generation, as measured by F1 + 2 levels, some patients have APC levels that are similar or even lower than that in normal pooled plasma. Since the amount of APC formed in healthy non-human primates is proportional to thrombin levels , the dissociation between APC and F1 + 2 levels observed in four out of six patients may reflect vascular endothelial dysfunction that impairs protein C activation through the down-regulation of TM and EPCR. Furthermore, in humans, APC levels under basal conditions are roughly proportional to the protein C levels . Thus, the disparity in APC to F1 + 2 levels does not appear to be due to the decreased levels of protein C alone. We hypothesize that impaired ability to generate APC in a subpopulation of septic patients probably reflects impaired protein C activation through the down-regulation of TM and EPCR levels on injured endothelium. Patients in the first group (i.e. patients 1 and 2 in Fig. 7) may benefit from protein C infusion, whereas patients in the second group (i.e. patients 3–6, in Fig. 7) have impaired protein C activation and might require APC infusion.
The APC assay described in this study has several clinical applications. First, since therapeutic levels of APC can be detected with a total assay time of ∼1.5 h, the assay may be useful in monitoring the dose and duration of recombinant APC therapy at the bedside. As mentioned previously, many of the septic patients in the APC treatment group who ultimately died , did so after the therapy was terminated, suggesting that some of the patients would benefit from prolonged infusion of recombinant APC. Second, our APC assay may be used to measure physiologic APC levels in patients with severe sepsis at base-line, which, together with levels of F1 + 2, might be useful in deciding the choice (i.e. protein C vs. APC) of therapy.
Other clinical applications of our APC assay include studies of vascular disease states such as severe atherosclerosis, diabetes, and vasculitis. Vascular endothelial TM and EPCR are down-regulated in coronary arteries with atherosclerosis, presumably shifting the hemostatic balance in favor of increasing thrombosis . In support of this hypothesis, abnormal TM gene expression is associated with an increased risk of myocardial infarction [8,38]. Vascular endothelial dysfunction is also associated with insulin resistance in diabetic patients  and with disease severity in patients with small vessel vasculitis (Wegener's granulomatosis) . Thus, circulating levels of APC, together with markers of thrombin generation (e.g. F1 + 2), may be a valuable serologic parameter in the diagnosis of vascular disease states. Improved substrates or enzyme attachment to the captured APC are all approaches that could result in increased sensitivity and rapidity of the APC assay in the future.
We thank Dr Naomi L. Esmon for critical reading of the manuscript and for many helpful discussions. We also thank Shelley Schmidt, RN for assistance in the identification of patients with severe sepsis. We wish to thank Dr Alireza R. Rezaie for use of the S195A APC protein, constructed by him. We are grateful to Nici Barnard in the preparation of this manuscript.
This research was supported by a Specialized Centers of Research grant awarded by the National Institutes of Health (grant P50 HL54502). Charles T. Esmon is an Investigator for the Howard Hughes Medical Institute. Patricia CY Liaw is a recipient of a Career Award from the Department of Medicine, McMaster University.