lable at ScienceDirect

Biomaterials 37 (2015) 1e12

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

Enzymes produced by autoactivation of blood factor XII in bufferA contribution from the Hematology at Biomaterial InterfacesResearch Group

Avantika Golas a, 1, Harit Pitakjakpipop a, 1, Matthew S. Rahn a, Christopher A. Siedlecki b, c,Erwin A. Vogler a, b, *

a Department of Materials Science and Engineering, The Pennsylvania State University, University Park, PA 16802, USAb Department of Bioengineering, The Pennsylvania State University, University Park, PA 16802, USAc Department of Surgery, Pennsylvania State University College of Medicine, Biomedical Engineering Institute, Hershey, PA 17033, USA

a r t i c l e i n f o

Article history:Received 15 August 2014Accepted 15 September 2014Available online 31 October 2014

Keywords:AutoactivationAutoamplificationAutohydrolysisFXIIPlasma coagulationConformers

* Corresponding author. Department of Materials SPennsylvania State University, University Park, PA 168

E-mail address: EAV3@PSU.EDU (E.A. Vogler).1 First authors.

http://dx.doi.org/10.1016/j.biomaterials.2014.09.0150142-9612/© 2014 Elsevier Ltd. All rights reserved.

a b s t r a c t

High-resolution electrophoresis of FXII-derived proteins produced by contact activation of FXII in buffersolutions (i.e. in absence of plasma proteins) with hydrophilic and silanized-glass activators spanning theobservable range of water wettability (hydrophilic to hydrophobic), shows no evidence of proteolyticcleavage of FXII into aFXIIa or bFXIIa. The autoactivation mixture contains only a single-chain proteinwitha molecular weight of ~80 kDa, confirming Oscar Ratnoff's previous finding of a single-chain activatedform of FXII that he called ‘HFea’. Functional assays have shown that these autoactivation products exhibitprocoagulant potential (protease activity inducing clotting of blood) or amidolytic potential (cleavesamino bonds in s-2302 chromogen but do not cause coagulation of plasma) or both amidolytic potentialand procoagulant potential. Some of these proteins also have the remarkable potential to ‘suppressautoactivation’ (i.e. suppress creation of enzymes with procoagulant potential). It is thus hypothesizedthat autoactivation of FXII in the absence of plasma proteins generates not just a single type of activatedconformer, as suggested by previous researchers, but rather an ensemble of conformer products withcollective activity that varies with activator surface energy used in contact activation of FXII. Further-more, reaction of aFXIIa with FXII in buffer solution does not produce additional aFXIIa by the putativeautoamplification reaction FXIIa þ FXII / 2FXIIa as has been proposed in past literature to account forthe discrepancy between chromogenic and plasma-coagulation assays for aFXIIa in buffer solution.Instead, net procoagulant activity measured directly by plasma-coagulation assays, decreases system-atically with increasing FXII solution concentration. Under the same reaction conditions, chromogenicassay reveals that net amidolytic activity increases with increasing FXII solution concentration. Thus,although autoamplification does not occur it appears that there is some form of “FXII self reaction” thatinfluences products of aFXIIa reaction with FXII. Electrophoretic measurements indicate that no pro-teolytic cleavage takes in this reaction leading us to conclude that change in activity is most likely due tochange(s) in FXII conformation (with related change in enzyme activity).

© 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Potentiation of the blood-plasma coagulation cascade by surfacecontact activation of the blood zymogen Factor XII (FXII, Hagemanfactor) is thought to be a cause of the relatively poor

cience and Engineering, The02, USA.

hemocompatibility of all currently known cardiovascular bio-materials. Two enzymes, aFXIIa and bFXIIa, have been identified asproducts of FXII activation with protease activity inducing clottingof blood plasma (procoagulant activity). These enzymes initiate theintrinsic pathway of coagulation that ultimately causes blood orblood plasma to undergo a phase transition from liquid to gel (i.e.coagulate or clot, see Ref. [1] and citations therein that generallysupport this section). And yet, in spite of extensive research effortover the last five decades, FXII surface activation (a.k.a. autoacti-vation) remains a mysterious reaction with unusual characteristicsrecently revealed by functional assays measuring enzyme activity

A. Golas et al. / Biomaterials 37 (2015) 1e122

of the mixture produced by FXII autoactivation with particulateactivators spanning a full range of surface chemistry/energy [2e5].

1. FXII autoactivation in buffer and protein solutions produces ahighly variable burst of procoagulant enzyme(s) with nomeasurable kinetics, sensitivity to mixing, or solution-temperature dependence. Autoactivation thus appears to be astochastic, mechanochemical-like reaction that does not exhibitcharacteristics of an ordinary (bio)chemical reaction.

2. Although autoactivation is definitively initiated by surfacecontact, no measurable dependence on activator surface areacan be discerned in buffer or protein solutions. That is to say,serial addition of activator surface area does not measurablyincrease procoagulant-enzyme yield.

3. Autoactivation appears to be a self-limiting reaction with anapparent procoagulant-enzyme yield converting less than 10%of starting FXII (at 30 mg/mL physiologic concentration) in thecontinuous presence of activator surface area. Thus, it appearsthat surface activation is somehow “shut off” after an initialcatalytic event.

4. The above observations are to be contrasted to the fact thatactivation of blood-plasma coagulation in vitro by contact withparticulate activator is demonstrably dependent on plasma-volume-to-activator-surface-area ratio.

5. FXII autoactivation in buffer produces an ensemble of as-yetunidentified proteins represented herein by FXIIact in thedescriptive chemical formula FXII ����!activator

surface FXIIact. FXIIactincludes proteins exhibiting either procoagulant potential(protease activity inducing clotting of blood) or amidolytic po-tential (cleaves amino bonds in s-2302 chromogen but do notcause coagulation of plasma) or both amidolytic potential andprocoagulant potential. Apparent yield of procoagulant en-zymes was less than purely amidolytic enzymes under all con-ditions studied. FXIIact also includes proteins that remarkablysuppress production of procoagulant enzymes but does notsuppress production of purely amidolytic enzymes.

6. The relative reactive proportions of proteins subsumed by FXIIactwere found to depend on activator surface chemistry/energy.Thus it appears that FXII undergoes some kind of chemicaltransformation by contact with any kind of activator surface,quite independent of hydrophilicity or surface chemistry; atleast in buffer solution.

The FXII-derived proteins mentioned in the above item (6)appear to be in a complex statistical ensemble of a large number ofstates, with some states of low local stability and others with highstability. Phenomena such as allosteric regulation are oftenascribed to arise from equilibrium between two major conforma-tions of dynamic proteins [6]. It is not proven herein that FXII existsin more than one conformer state, but change in conformationwithvarious perturbants has been observed [7].

Precise biochemical mechanisms for reactions involving FXII arenot well understood, but in-vitro activation of FXII has been vari-ously attributed to proteolytic cleavage [8e14], change in confor-mation [7,15e21], or a combination of these mechanisms [10,22].Speculation on the nature of these conformation changes ishampered by the fact that 3-D crystal structures of FXII and itscleavage products are not clear, with only a few studies conductedusing computer models [23].

In light of the above, it has become evident that a molecularinventory of proteins and conformers resulting from FXII autoac-tivation under different experimental conditions with activatorsspanning a full range of surface chemistry/energy is a necessarystep in the formulation of testable mechanisms of autoactivation.In turn, formulation of testable mechanisms is essential to the

long-sought ability to prospectively surface engineering cardio-vascular biomaterials with improved hemocompatibility. Thisinventory will compliment purely functional assays; hopefullyleading to an identification of those specific protein(s) comprisingFXIIact that is (are) responsible for the above-mentioned amido-lytic, procoagulant, and procoagulant-suppressive activity. Pre-sent work also sheds light on putative “self reactions” betweenFXII and aFXIIa in buffer solution in the absence of activator sur-faces (other than that of test tubes and cuvettes in which exper-imental solutions are contained). Herein, we report results from atwo-prong research strategy using high-resolution electropho-resis coupled with chromogenic and plasma-coagulation func-tional assays.

2. Methods and materials

2.1. Plasma and coagulation proteins

Citrated human platelet poor plasma (PPP) was prepared from unexpired pooledlots obtained from the M.S. Hershey Medical Center Blood Bank and prepared asdescribed previously [24e26]. This work was performed with a single batch of PPPaliquoted into 15 ml polypropylene tubes (Falcon, Becton Dickinson) and frozenat �20 �C until use. Experiments for this work were conducted within a month ofPPP preparation. Repeat trials yielded no significant variations in coagulation timeswithin this experimental time frame.

Human FXII and aFXIIa were used as received from Enzyme Research Labora-tories (South Bend, IN). Activity of both FXII and aFXIIa was specified by the vendorinmg/mL or traditional units of plasma equivalent units per mL (PEU/mL) [27]. Neat-buffer solutions of FXII and aFXIIa solutions were prepared in phosphate buffer sa-line (PBS; Sigma; 0.14 M NaCl, 3 mM KCL prepared from powder in 18 MU de-ionizedwater, pH 7.4).

2.2. Preparation and characterization of particulate activators

Test activator surfaces used in this work were 425e600 mm diameter glassparticles (Sigma Aldrich) in either cleaned or silanized form. The nominal specificarea used in this work was 5 � 10�3 m2/g (based on 512.5 mm mean diameter and168 mg/particle). The surface area measured by the Brunauer-Emmett-Teller (BET)method (Micromeritics ASAP 2000 using liquid nitrogen as the probe gas) was(8.5 ± 0.1) � 10�3 m2/g. Nominal surface area was used throughout this work as amatter of consistency with prior work and in recognition of the fact that surface areadid not influence conclusions based on comparisons among experiments describedherein using the same source of particulate activator with a fixed specific surfacearea.

Silanes (used as received from Sigma Aldrich) applied in this work were octade-cyltricholorosilane (OTS) and 3-aminopropyltriethoxysilane (APTES). OTS-treatedglass particles were coated in a 0.2% solution of 1,1-pentadecaflurooctylmethacrylatein tricholorotrifloroethane (“Nyebar”, Nye Lubricants, Fairhaven, MA) by immersionfollowed by air drying. Glass cover slips (Fisher 22 � 30 � 0.1 mm) were subjected toall surface treatments for particles as described above and further below, providing asubstrate suitable for reading phosphate buffer saline (PBS, Sigma) contact angles.

Glass particles and cover slips were first cleaned and activated by 30 min im-mersion in heated piranha solution (30% H2O2 in concentrated H2SO4 at approxi-mately 80 �C) followed by 3X sequential washes in each of 18 MU de-ionized waterand 100% ethanol. Piranha-solution oxidized glass was air dried and subsequentlyoxidized by air-plasma treatment (10 min at 100W plasma; Herrick, Whippany, NY)of a single layer of particles (or cover slip) held in a 15 mm Pyrex glass Petri dish,directly before use in silanization procedures or blood-plasma coagulation activa-tion measurements. Glass surfaces treated in this manner were found to be fullywater wettable and designated “clean glass”. Clean glass particles and cover slipsamples were silanized by 1.5 h reaction with 5% v/v OTS in chloroform. Silanizedsamples were 3X rinsed with chloroform before curing in a vacuum oven at 110 �Cfor 12 h. Cured OTS samples were optionally immersed in Nyebar solution for 10minand air dried to produce a surface slightly more hydrophobic than rendered by OTStreatment alone (see Table 1). APTES silanization was carried out by 20 min reactionof clean glass with 95:5 v/v ethanol-water solution with 5% APTES that had beenhydrolyzed overnight in the ethanol-water mixture before use. APTES treated glasswas 3X washed with ethanol and cured overnight in a vacuum oven at 110 �C.

PBS (see below) contact angles on glass cover slip witness samples weremeasured using an automated contact-angle goniometer (First Ten Angstroms Inc.,Portsmouth, VA) that employed the captive-drop method of measuring advancing/receding contact angles (see Refs. [28,29] for a comparison of goniometric tech-niques and discussion of experimental errors). Contact angles could not be readdirectly on glass particles but optical microscopy of the shape of the liquid meniscusof particles partly immersed in water on a microscope slide qualitatively confirmedthat the treated particles were not different from the cover slip witness samples.Surface chemistry of glass-particle surfaces has been previously assayed using a

Table 1Protein yields (as measured by automated electrophoresis) following autoactivation of FXII in the presence of HK and PK in pure-buffer solutions.

S. No. Trial I Trial II Control solutions

Protein molecularweighta (kDa)

Proteinyield (%)

Protein molecularweighta (kDa)

Proteinyield (%)

Protein molecularweighta (kDa)

Proteinyield (%)

Autoactivationtest solution

1 27.68 2.56 27.73 1.37 (FXII þ HK þ PK) 27.4 1.262 30.3 2.87 30.3 2.36 29.92 2.513 34.84 9.47 34.98 8.3 34.72 7.854 47.44 5.24 47.35 5.37 46.77 4.425 57.81 33.73 57.8 40.61 57.18 49.386 65.38 2.65 70.06 3.77 81.24 26.27 70.31 4.06 81.86 26.42 95.73 7.928 82.04 25.1 96.68 11.19 FXII 35.24 0.859 97.15 10.61 97.25 98.94

PK 92.28 19.7398.67 80.27

HK 171.45 100

a Molecular weight as read from the Experion unit. Degree of glycosylation affects molecular weight measurements in automated electrophoresis (see Section 3.1).

A. Golas et al. / Biomaterials 37 (2015) 1e12 3

Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFT) collected on aBruker IFS-66/S Tech as described in Ref. [30].

2.3. High-resolution electrophoresis

An automated electrophoresis system (Experion, Bio-Rad Laboratories, CA) wasused to perform electrophoresis of proteins. The automated electrophoresis systemperforms electrophoresis of samples within a microfluidic chip. Within each chip, aseries of micro-channels connects the sample wells to a separation channel andbuffer wells. A set of electrodes in the electrophoresis station applies a voltage acrossthe micro-channels, causing charged molecules in the samples to migrate into andthrough the separation channel. Samples are run sequentially, with a sufficient lagbetween each run to prevent cross-contamination. For separation, the micro-channels are filled with a proprietary gel-stain solution (GS) that acts as a sievingmatrix, and under denaturing conditions, in the presence of lithium dodecyl sulfate(LDS), the sample proteinsmigrate through the separation channel at a rate based ontheir molecular weight. Finally, proteins interact with a fluorescent dye duringseparation and are detected as they pass a laser and photodiode detector (laser-induced fluorescence).

For this work, 10 lane Experion Pro260 chips were used to separate and quantifyproteins. The vendor-defined protocol was used for sample preparation and runningthe electrophoresis. Chips were first primedwith 12 ml of a mixture of sieving matrix(gel) and fluorescent dye (stain). Test solutions (4 ml for each supernate) were mixedwith Pro260 sample buffer (2 ml for each sample) in 0.5 mL conical microtubes.Control solutions in neat-buffer, along with a vendor-provided protein ladder, werealso processed in a similar manner. The protein solutions were denatured by placingthe tubes in a water bath at 98 �C in for 5 min. The tubes were allowed to cool andlater centrifuged for 15 s to collect the condensate. A bolus of 84 ml of 18 MU de-ionized water (obtained from a Millipore Simplicity unit) was added to eachmicro-tube. The primed micro-chip was loaded with 12 ml gel and gel-stain solutionin designated wells. 6 ml of each sample, control and protein ladder solution wasadded to the wells, and loaded into the Experion station. Experion software plottedthe fluorescence intensity vs. time and produced an electropherogram and virtualgel image. In addition to test solution peaks, electropherograms also featured systempeaks (cluster of signals generated by small molecules that interact with LDSmicelles ¼ 8.3 kDa), as well as upper (260 kD) and lower marker (1.2 kD) peaks.

2.4. Autoactivation reactions in buffer solution

Autoactivation of FXII in PBS by clean or silanized glass-particle activators wasthe primary phenomenon studied in this work. FXII, as received from the vendor,was diluted 4� in PBS to prepare a solution at a concentration of 0.28 mg/ml.Concentrations ~10� that of physiological concentration were used in electropho-retic studies to ensure optimum resolution of protein fragments formed. FXII acti-vation in PBS solutionwas carried out as previously described [3,5]. Briefly, 200 ml ofFXII in PBS at 0.28 mg/mL, at room temperature, was brought in contact with aspecific weight of test activator (detailed in Sections 2.4.1e2.4.4) for between 5 and70 min. Supernatant from these experiments were then analyzed by automatedelectrophoresis as detailed in Section 2.3. Each electrophoresis microchip alsocontained control samples of FXII in PBS (0.28 mg/mL) that were not incubated withparticulate activators.

2.4.1. Surface area titration of purified FXII in buffer solutionThe autoactivation protocol detailed above was carried out with incrementally

varying mass between 0 and 50 ± 0.1 mg (nominal surface area of 5.0 � 10�4 m2) ofclean glass activator weighed using a laboratory electronic balance. Autoactivationwas carried out by 30 min contact of 200 ml of FXII in PBS at 0.28 mg/mL with clean-

glass surfaces in 2 mL microtubes (VWR). Autoactivation supernate was analyzedusing automated electrophoresis as detailed in Section 2.3.

2.4.2. Autoactivation of FXII with procoagulants with varying surface energyTest solutions of FXII were prepared at concentrations of 0.28 mg/mL in 2 mL

microtubes (VWR). 200 ml of the solution was incubated with 100 mg of clean glass,APTES-silanized or Nyebar-coated-OTS-silanized glass for 10 min and 70 min. Su-pernatant from each microtube was evaluated in the Experion automated electro-phoresis unit as detailed in Section 2.3.

2.4.3. Concentration titration of FXIITest solutions with FXII in PBS were prepared at concentrations of 0.28 mg/mL.

Three additional concentrations of FXII (0.14, 0.07, 0.035 mg/mL) were prepared by50% serial-dilutions. FXII solutions at each concentration were incubated with100 mg of clean-glass activator for 30 min. Test supernatant and FXII control at eachconcentration were then analyzed by the Experion automated electrophoresis unitas described in Section 2.3.

2.4.4. Kinetics of FXII autoactivationFXII test-solutions in PBS (200 ml at 0.28 mg/mL) were incubated for durations

ranging from 10 min to 70 min with 100 mg of clean-glass activators and silanizedparticle activators. Test supernatant were then analyzed in the Experion unit asdescribed in Section 2.3.

2.5. Autoactivation of FXII with PK and HMWK

Test solution containing 0.15 mg/mL of FXII, 0.20 mg/mL of prekallikrein and0.43 mg/mL of high-molecular weight kininogen was prepared and 3X seriallydiluted by 50% to prepare four test solutions containing these proteins in molarratios similar to those found physiologically [1]. 200 ml of the test solutions wereincubated with 100 mg of clean glass for 60 min. Supernatant from each microtubewas evaluated in the Experion automated electrophoresis unit as detailed in Section2.3. Control solutions containing the above mixture of proteins incubated in theabsence of clean-glass activator, and solutions containing each of the reagentsindividually were also added to the microchip to compare results.

2.6. Autohydrolysis reactions in buffer solution

Autohydrolysis reactions were carried out by timed incubation of FXII and aFXIIain neat-buffer solutions. 200 ml test solutions containing varied [aFXIIa] (0.06, 0.13,0.20, 0.30, 0.50, 0.75 PEU/ml) and 0.03 mg/ml of FXII in PBS, were prepared by gentlepipette aspiration and allowed to stand undisturbed in 0.5 mL conical microtubes(Safe-lock microcentrifuge tubes, Eppendorf) at room temperature for 30 min.Autohydrolysis supernatant was subsequently evaluated by a coagulation time (CT)assay to measure procoagulant activity and a chromogenic assay to measure ami-dolytic activity using protocols detailed in Ref. [3]. The above protocol was repeatedwith FXII at 0.015 and 0.075 mg/ml, to evaluate the effect of varying FXII concen-tration on autohydrolysis. FXII plasma titrations were conducted to determinecontrol CT values. Difference in CT for autohydrolysis supernates and controlsupernates was calculated as DCT. Similarly a difference in absorbance rate (asmeasured by the UVeVis Spectrometer) was calculated as DAbsorbance Rate.

Electrophoretic identity of autohydrolysis products was established throughautomated electrophoresis. Serially diluted solutions starting with 0.20 mg/ml eachof FXII and FXII in neat-buffer solution were mixed by gentle pipette aspiration andincubated for 30 min at room temperature. Concentrations ~10� physiological [FXII]were used to ensure effective resolution of the FXII autohydrolysis suite.

A. Golas et al. / Biomaterials 37 (2015) 1e124

Autohydrolysis supernates were then analyzed by an automated-electrophoresissystem as detailed in Section 2.3.

3. Results

3.1. Electrophoresis of FXII and aFXIIa reagents

Electrophoresis of FXII and aFXIIa reagentswas carried out in theExperion system as outlined in Section 2.3. Experion softwareplotted the fluorescence intensity vs. time (electropherogram), asshown in Fig. 1, and generated a virtual gel image. Electrophoresisfor both reagents was conducted in reducing, as well as, non-reducing buffer (not shown). As is evident, vendor-sourced FXII(Panel B, Fig. 1) exhibited a single peak at ~82 kDa. Vendor-suppliedaFXIIa (Panel A, Fig. 1) exhibited three main peaks at 29, 52 and58 kDawith <1% of other proteins/contaminants. In addition to testsolution peaks, electropherograms also featured system peaks(cluster of signals generated by small molecules that interact withLDS micelles~8 kDa), as well as upper (260 kDa) and lower marker(1.2 kDa) peaks that were used by the Experion unit to normalizethe data.

An ~15e18% higher molecular weight for FXII and aFXIIa pro-teins was consistently observed with automated electrophoresiswhen compared with conventional SDS-PAGE. According to thevendor, this is purportedly due to glycosylation of FXII and aFXIIa,and has been observed with other glycosylated proteins [31].Electrophoresis with deglycosylated reagents was not conducted in

A 260 kDa

52.3 kDa 29.6 kDa

260 kDa

8.3 kDa

1.2 kDa

1.2 kDa

82.3 kDa

8.3 kDa

1.2 kDa

8.3 kDa

B

58.2 kDa

1.2 kDa

8.3 kDa

29

Fig. 1. Electropherogram and virtual-gel image, in reducing buffer, for aFXIIa (Panel A) and Ffeature system peaks (cluster of signals generated by small molecules that interact with LDSnormalize the data.

order to avoid artifacts. However, molecular weights read from theExperion system, were uniformly scaled down by 15% to avoidconfusion and maintain consistency with conventional SDS-PAGE.

3.2. Autoactivation as a function of procoagulant surface energy

Products of FXII autoactivation with clean-glass and silanized-glass activators were analyzed in reducing buffer using the Expe-rion automated electrophoresis system as detailed in Section 2.3.Fig. 2A is the virtual-gel image generated by the instrument. Upperand lower bands at 260 kDa and 1.2 kDa are present in the vendor-provided buffer and the band at 8.3 kDa is system generated. Theonly band visible from test-solutions was found to be at ~80 kDa foractivation supernatants from both clean-glass and silanized-glassactivators. Proteins in this band accounted for >99% of theautoactivation supernatant. Fig. 2 B plots the concentration of allprotein fragments as measured by the Experion system. Concen-tration of the 80 kDa fragment, in case of clean-glass and Nyebarautoactivation, appeared marginally lower than that for FXII con-trol. However, the percentage of total for the 80 kDa band (outputprovided by the Experion system) in all test-solutions was >99%,and differences appeared to be due to error in loading proteins. Theother band observed in some test samples was ~29 kDa(<0.001 mg/ml). However, this was found in similar low concen-trations within FXII-control samples as well, and deemed to be anunknown contaminant.

260 kDa

82.3 kDa

260 kDa

.6 kDa

52.3 kDa 58.2 kDa

XII (Panel B). In addition to test solution peaks, electropherograms and gel-images alsomicelles ~8.3 kDa), as well as upper (260 kDa) and lower marker (1.2 kDa) peaks that

Nyebar APTES CleanGlass

80 kDa

Appa

rent

pro

tein

sol

utio

nco

ncen

trat

ion

(mg/

ml x

10-

1 )

Protein molecular weight (kDa)

A

B

8.3 kDa

260 kDa

1.2 kDa

0 20 40 60 80 1000.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Fig. 2. Autoactivation as a function of procoagulant surface energy. Panel A shows virtual gel images of autoactivation supernatant in reducing buffer, obtained by contacting neat-buffer solutions of FXII (0.28 mg/ml) with clean-glass and silanized-glass activators, for 70 min. The only band visible is at ~80 kDa which represents >99% of autoactivation productin all three experiments. Panel B plots the concentrations of these proteins (medium grey: APTES; dark grey: clean glass; solid black: Nyebar) in comparison to FXII control (lightgrey). Data and error bars represent mean and standard deviation of N ¼ 2 measurements. A marginally lower concentration of the 80-kDa fragment was observed in case of clean-glass and Nyebar activator particles, however there was no concomitant increase in other protein fragments, and 80 kDa protein yield in all cases was ~99%. Evidence of proteolysiswas thus not seen with clean and silanized glass activators.

A. Golas et al. / Biomaterials 37 (2015) 1e12 5

3.3. Effect of varying experimental parameters (clean glass activatorsurfacearea, FXII concentrationand incubation time)onautoactivation

Fig. 3 illustrates the percentage of 80 kDa protein in totalautoactivation-supernatant, as a function of clean-glass activatorsurface area. Apparent yields following autoactivation were closeto 100% and not statistically distinct from those in pure-buffersolutions of FXII. No evidence of FXII proteolysis was thusobserved by varying clean-glass activator surface area.

Continuous incubation with clean-glass and silanized particleactivators exhibited a similar trend (not shown). The 80 kDa banddid not exhibit a statistically significant change in concentrationeven after 70 min of continuous incubation of FXII solution withclean glass and silanized particle activators. The ~80 kDa bandrepresented >99% of the autoactivation suite, and the concentra-tion of the other minor fragments/contaminants in test-solutions(<1%; not shown) was not significantly different than concentra-tion in control solutions of FXII.

0.0 0.5 1.0 1.5 2.0 2.5 3.0

20

40

60

80

100

120

80 kDa `

App

aren

t 80

kDa

prot

ein

yiel

d (%

)

Activator Surface Area (m2 x 10-4)

A

B

1.2 kDa

8.3 kDa

260 kDa

FXII 0.25 0.75 0.5 1.0 1.25 1.75 1.5 2.0 2.5

Fig. 3. Surface-area dependence of FXII autoactivation in buffer solution. Panel B plots the percentage of the 80 kDa fragment in autoactivation supernatant, produced with varyingsurface-area of clean-glass activator. Dashed line and grey area shows mean and standard deviation of the 80 kDa fragment in 0.28 mg/mL FXII control. There was no dependence onactivator surface area, and no significant difference in the percentage of the 80 kDa fragment appeared on autoactivation when compared to FXII control. Bands suggestive of theformation of aFXIIa or bFXIIa did not appear at clean-glass activator surface areas used in this study. Panel A is the corresponding virtual gel image generated by the Experionautomated electrophoresis unit.

A. Golas et al. / Biomaterials 37 (2015) 1e126

Apparent concentrations of the 80 kDa protein in totalautoactivation-supernatant, as a function of FXII solution concentra-tion (not shown) were within in the 98e100% range and were sta-tistically indistinguishable from those inpure-buffer solutions of FXII.

3.4. FXII contact activation in the presence of PK and HMWK

To simulate reciprocal-activation in-vitro, we used an experi-mental environment that included FXII, high-molecular weightkininogen (HMWK) and prekallikrein (PK) in PBS incubated with

clean-glass activators. This system was simplistic owing to theabsence of other substrates and inhibitors for each reagent, butoffered incremental increase in biochemical complexity comparedto FXII autoactivation studies in buffer without exogenous proteinsand had the advantage of easy analysis by electrophoresis.

Products of FXII autoactivation with clean-glass in the presenceof HMWK and PK in PBS were analyzed using the Experion auto-mated electrophoresis system as detailed in Section 2.3. Fig. 4 (Test)presents the virtual-gel image generated by the instrument forautoactivation with clean-glass whereas Fig. 4 (Control) has the

260 kDa

8.3 kDa

1.2 kDa

82* kDa 97* kDa

58* kDa

34* kDa

HMWK PK FXII260 kDa`

1.2 kDa 8.3 kDa

171* kDa

98* kDa

Test solution Control solution

260 kDa

8.3 kDa

1.2 kDa

82* kDa 97* kDa

58* kDa

34* kDa

Fig. 4. FXII contact-activation in the presence of prekallikrein and high-molecular weight kininogen. Asterisk (*) indicates that molecular weight measurements for this experimentare as read from the Experion system and not scaled down as for other figures. Test and Control solutions show protein bands <80 kDa which indicate proteolysis in both solutions.Inset shows virtual gel images for Prekallikrein, High-molecular weight Kininogen and FXII in PBS as measured separately by the Experion system. Gel-images also feature systempeaks (cluster of signals generated by small molecules that interact with LDS micelles ~8.3 kDa), as well as upper (260 kD) and lower marker (1.2 kD) peaks that normalize the data.

A. Golas et al. / Biomaterials 37 (2015) 1e12 7

virtual-gel image for a mixture of the reagents incubated withoutclean-glass. The image in the upper-left box is the virtual-gel forindividual reagents. Upper and lower bands at 260 kDa and 1.2 kDaare present in the vendor-provided buffer and the band at 8.3 kDa issystem generated. A 15e25% higher molecular weight for reagents,test and control proteins, was reported by the instrument pur-portedly due to glycosylation of proteins. We did not scale downthis data because it was difficult to determine the sources ofcleavage fragments and the degree of glycosylation appropriate toeach band. Table 1 summarizes data from contact-activation andcontrol-solution yields (as read from the Experion unit). Evidencesuggestive of proteolysis was found in both the test and controldata, even though the correct identity of each protein band couldnot be ascertained due the impact of glycosylation.

3.5. Plasma coagulation assay for procoagulant activity inautohydrolysis supernate

Increasing concentrations of aFXIIa, ranging from0.06 PEU/ml to0.75 PEU/ml were incubated with 0.03 mg/ml of FXII for timeranging from2min to 30min. Autohydrolysis supernates were thenassessed in a CT assay (Section 2.6). Using this assay, it was foundthat the autohydrolysis reaction was effectively instantaneouswithin the minimum elapsed-time resolution of the experiment

(about 2 min) and exhibited no statistically-significant trend over30 min of testing (not shown).

Average coagulation times (CT) for autohydrolysis supernates,were compared to coagulation times (CT) for an aFXIIa titration ofplasma. For each concentration of aFXIIa, the difference in coagu-lation time between the experiment and control was calculated asDCT. Fig. 5 (inset) shows that DCT scales in an exponential-likemanner with aFXIIa concentration (best fit line through the data).Whereas the difference DCT is ~6 min at a concentration of 0.06PEU/ml of aFXIIa, an asymptote plateaus to ~2 min is obtained withincreasing concentrations of aFXIIa. Furthermore, as is evident inFig. 5, the plot of DCT/[aFXIIa] reveals an exponential decrease,reaching a plateau at ~3 min per-unit [aFXIIa]. Data was fit toDCT ¼ c þ ae�b[aFxIIa] by non-linear regression, with c ¼ 1.81 ± 2.65,a ¼ 183.4 ± 19.4, b ¼ 11.9 ± 1.5; R2 ¼ 0.99.

As illustrated in Fig. 6, DCT increases as the concentration of FXIIincreases. While no consistent trend was observed for DCT, anexponential decrease in DCT per-unit [aFXIIa] was observed for allthree [FXII] tested. Datawas fit to DCT¼ cþ ae�b[aFxIIa] by non-linearregression, c ¼ 10.0 ± 2.3, a ¼ 164.7 ± 10.7, b ¼ 9.2 ± 0.9, R2 ¼ 0.99for autohydrolysis with [FXII] ¼ 0.07 mg/ml; c ¼ 3.9 ± 2.5,a ¼ 205.8 ± 18.8, b ¼ 12.0 ± 1.3, R2 ¼ 0.99 for autohydrolysis with[FXII] ¼ 0.03 mg/ml; c ¼ 3.6 ± 1.6, a ¼ 205.9 ± 15.1, b ¼ 13.6 ± 1.1,R2 ¼ 0.99 for autohydrolysis with [FXII] ¼ 0.015 mg/ml.

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

0.0

20.0

40.0

60.0

80.0

100.0

120.0

CT /[

FXIIa

] (m

ins.

ml/P

EU)

Concentration of FXIIa (PEU/ml) C

T (m

ins)

Concentration of α FXIIa (PEU/ml) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

Fig. 5. Increase in coagulation time (CT) following autohydrolysis in pure-buffer. For each concentration of aFXIIa, the difference in coagulation time for autohydrolysis supernateand aFXIIawas calculated as DCT. Fig. 5(inset) illustrates that the difference in CT scales in an exponential-like manner with aFXIIa concentration. While the difference is ~6 min at aconcentration of 0.06 PEU/ml of aFXIIa, it plateaus to ~2 min at higher concentrations of aFXIIa. A plot of DCT/[aFXIIa] reveals an exponential trend, plateauing at ~3 min per unitaFXIIa. Data points and error bars represent mean and standard deviation of n ¼ 3 trials. Solid lines through data are best-fit lines obtained by non-linear regression.

A. Golas et al. / Biomaterials 37 (2015) 1e128

3.6. Chromogenic assay for amidolytic activity

Average absorbance rates for autohydrolysis supernates, werecompared to absorbance rates for aFXIIa. For each [aFXIIa], thedifference in absorbance rates between the two was calculated as

0.0 0.1 0.2 0.3 0-20.0

0.0

20.0

40.0

60.0

80.0

100.0

120.0

140.0

CT

/[FX

IIa] (

min

s.m

l/PEU

)

Concentration of

Fig. 6. Decrease in net procoagulant activity with increasing FXII solution concentration. Dcoagulation time observed upon autohydrolysis, in comparison to aFXIIa alone, as measurfunction of aFXIIa (inset), however the decrease in procoagulant activity with increasing FXI0.03 and 0.07 mg/ml respectively. Data points and error bars represent mean and standard deregression.

DAbsorbance Rate. Fig. 7 (inset) illustrates that the absorbance rates ofautohydrolysis supernatant with FXII at 0.07mg/ml and 0.03mg/mlwere higher than control values with aFXIIa alone. DAbsorbance Rate

also decreases in a linear manner with aFXIIa concentration. A plotof DAbsorbance Rate/[aFXIIa] exhibited an exponential decrease in

.4 0.5 0.6 0.7 0.8

[FXII] mg/ml 0.07 0.03 0.015

FXIIa (PEU/ml)

CT (m

ins)

Concentration of FXIIa (PEU/ml) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

-2.0

0.0

2.0

4.0

6.0

8.0

10.0

CT/[aFXIIa] increases with increasing FXII concentration, where DCT is the increase ined by a coagulation time (CT) assay. No consistent trends were observed for DCT as aI is evident. Solid triangles, open circles, and solid squares correspond to [FXII] at 0.015,viation of n ¼ 3 trials. Solid lines through data are best-fit lines obtained by non-linear

C t ti f F12 (PEU/ l)

0.0 0.1 0.2 0.3 0.4 0.5 0.6-10.0

0.0

10.0

20.0

30.0

40.0

50.0

0.0 0.1 0.2 0.3 0.4 0.5 0.6-2

-1

0

1

2

3

Abs

orba

nce

Rat

e /[

FXIIa

](a

bsor

banc

e un

it.m

l/PEU

.min

)

Concentration of FXIIa (PEU/ml) A

bsor

banc

e R

ate

(abs

orba

nce

unit/

min

)

Concentration of FXIIa (PEU/ml)

[FXII] mg/ml

0.07 0.03

0.015

[FXII] mg/ml

0.07

0.03

0.015

Fig. 7. Average Absorbance rates for autohydrolysis supernates, compared to absorbance rates for aFXIIa. DAbsorbance rate (inset) represents the difference in absorbance rates.DAbsorbance rate decreases linearly with aFXIIa concentration while DAbsorbance Rate/[aFXIIa] exhibits an exponential decrease. Autohydrolysis with 0.015 mg/ml of FXII yielded similar-to-lower absorbance rates when compared to [aFXIIa]. DAbsorbance Rate values were therefore negative. Solid squares represent FXII ¼ 0.07 mg/ml, empty circles representFXII ¼ 0.03 mg/ml, and solid triangles represent FXII ¼ 0.015 mg/ml. Solid lines are best fit to data.

A. Golas et al. / Biomaterials 37 (2015) 1e12 9

these two cases. However autohydrolysis with 0.015mg/ml of [FXII]yielded similar-to-low absorbance rates as compared to aFXIIaalone, quite possibly representing threshold values for autohy-drolysis products.

3.7. Electrophoresis of autohydrolysis products

Fig. 8 illustrates the electrophoretic identity of autohydrolysisproducts measured in reducing buffer (scaled down by 15%). Themolecular suite of autohydrolysis proteins matches closely with thesuite observed for the reagents. A comparison in concentration foreach of these fragments reveals that with the exception of the58 kDa fragment, concentrations of other fragments were not sta-tistically different. Solid black bars establish electrophoretic iden-tity of the autohydrolysis proteins, whereas grey bars represent theelectrophoretic identity of reagents. The 58 kDa fragment wasapproximately 50% higher in concentration after hydrolysis. How-ever, no concomitant decrease in concentration for any otherfragment, which could account for the appearance of the 58 kDafragment after autohydrolysis, was observed.

4. Discussion

At least three distinct reactions for FXII have been described inhematology literature.

1. Autoactivation, represented as FXII ����!activatorsurface FXIIa takes

place when FXII comes in ‘contact’ with an activating surface[1,10,15,32,33]. While aFXIIa and bFXIIa are relatively wellknown products of autoactivation, the possibility of the forma-tion of minor fragments has been suggested [9,34]. OscarRatnoff also described a single-chain-activated Hageman factor,formed by the incubation of FXII with sephadex-ellagic acid

[20,35,36]. The conformer (termed HFea exhibited amidolyticpotential, but showed no evidence of proteolytic scission whenvisualized by gel electrophoresis. Thus autoactivation may bemore accurately represented asFXII ����!activator

surface FXIIact, whereFXIIact subsumes all FXII autoactivation products including pro-tein fragments and conformers (see Section 4.1).

2. FXII is cleaved by kallikrein, in plasma, which presumably can begenerated by FXIIact -mediated hydrolysis of prekallikrein[8,21,37]. This mutual activation of prekallikrein and FXII isreferred to as reciprocal activation. Reciprocal activation ac-counts for 75% of the FXIIact generated in plasma [38].

3. Autohydrolysis of FXII is a putative “self-reaction”, where FXII is asubstrate for the FXII enzyme. Mathematical models of theintrinsic pathway of blood-plasma coagulation suggest thatautohydrolysis is an insignificant reaction in plasma, but sig-nificant in pure buffer solutions of FXII and aFXIIa [26,38].Autoamplification appears to be a particular type of autohy-drolysis reaction that presumably generates aFXIIa by reactionof aFXIIa with FXII, as described by the chemical formulaFXII þ FXIIa ����!activator

surface 2FXIIa. Autoamplification has beendiscussed in past literature (see Ref. [39] and citations therein)and, in particular, has been proposed to account for differencesbetween chromogenic and plasma-coagulation assays for aFXIIain buffer solution [39]. Subsequent work [3,5] casts doubt onthis latter proposal, suggesting instead that a less specificautohydrolysis reaction of the typeFXII þ FXIIa ����!activator

surface amidolytic proteins accounts for thediscrepancy between chromogenic and plasma coagulation.

Prekallikrein is a zymogen composed of two molecular-weight-variants of a single polypeptide chain: 88 kDa and 85 kDa.Following reciprocal activation, prekallikrein is converted to kalli-krein, which consists of a heavy chain (52 kDa) linked by disulfide

20 30 40 50 60 70 80 900

5

10

15

20

25

30

35

Appa

rent

sol

utio

n pr

otei

n co

n cen

tra t

ion

(mg/

ml x

10-

2 )

Protein molecular weight (kDa)

260 kDa

80.6 kDa58.2 kDa

29.4 kDa

8.3 kDa1.2 kDa

52.1 kDa

Fig. 8. Comparison of protein fragments formed with autohydrolysis to those in FXII and aFXIIa reagents. Solid black bars represent autohydrolysis protein electrophoresis suite,while grey bars represent reagent electrophoresis suite. Data points and error bars are mean and standard deviation for n ¼ 5 measurements. No statistically significant differenceobserved for fragments with the exception of 58 kDa fragment.

A. Golas et al. / Biomaterials 37 (2015) 1e1210

bonds to a light chain of either 36 kDa or 33 kDa. A variant called b-kallikrein has a cleaved heavy chain broken down into molecularweights of 28 kDa and 22 kDa. Kallikrein has a wide variety ofsubstrates in plasma in addition to factor XII which include high-molecular weight kininogen (HMWK), plasminogen, C3 and pro-renin. C1 inhibitor and a2macroglobulin account for over 90% of theinhibitory activity of kallikrein in normal plasma [40]. FXII activatesto form aFXIIa by cleaving the bond connecting Arg353eVal354 andgenerating a two-chain molecule composed of a heavy chain52 kDa and a light chain 28 kDa, held together by a disulfide bond.Proteolytic cleavage further proceeds on aFXIIa to yield a majoractive product at a molecular weight of 40 kDa, as well as bFXIIa,which appears as two closely related molecular species of molec-ular weight 28 kDa and 30 kDa. Aminor active product of molecularweight at 70 kDa is also observed. Other minor fragments of FXIIhave also been hypothesized as well as observed [3]. The FXII suiteis inhibited by C1eINH, a2-antiplasmin, a2-macroglobulin andantithrombin III.

Changes in conformation accompanying activation of purifiedFXII were first studied by Revak and Cochrane [8,21]. They foundthat activation of purified Hageman factor with negatively chargedsurfaces (so-called ‘solid phase activation’) resulted in a putativeconformational change in the molecule, while activation withkallikrein and plasmin caused proteolytic cleavage [41e43].

McMillin et al. [7,17] and Samuel et al. [16,19] used circular-dichroism spectroscopy to observe changes in conformationfollowing activation of FXII. Samuel et al. further proposed thatactivation with dextran sulfate was biphasic, with a first rapidphase associated with small but significant changes in proteinsecondary structures, followed by a second slow phase of proteo-lytic cleavage also accompanied with changes in conformation.Oscar Ratnoff too observed that FXII exposed to sephadex-ellagicacid in the absence of proteolytic agents exhibited amidolytic ac-tivity, but had not undergone scission [20], leading him to concludethat chain-scission may not be necessary for clot-promotingproperties of FXII in plasma. He called this single-chain activated

form ‘HFea’ [35,36,44]. Later Heimark et al. [45] proposed a tentativemechanism for the surface activation of blood coagulation withkaolin which suggested that the initial event in the activation ofblood coagulation involves a substrate-induced catalysis by single-chain FXII ‘bound’ to kaolin, suggesting that kaolin-activated single-chain FXII has sufficient enzymic activity to propagate the intrinsiccascade. Finally, Citarella et al. [46] summarized two views of theidea that change in conformation of FXI in-vitro (see Ref. [46] andcitations therein) led to activation. The ‘autoactivation’ hypothesis,postulates that binding of FXII to a negatively charged surface in-duces conformational changes in FXII, and is the key event thatinitiates autoactivation. An alternate ‘susceptibility’ hypothesisstates that FXII bound to a surface does not autoactivate but ratherundergoes a conformational change that enhances its susceptibilityfor cleavage by kallikrein.

4.1. FXII contact-activation in the absence of plasma proteins

Experimental evidence discussed in Sections 3.2 and 3.3 usinghigh-resolution electrophoresis, indicates that FXII does not un-dergo proteolytic cleavage following autoactivation with clean-glass and silanized-glass activators in the absence of plasma pro-teins. However, given that procoagulant and amidolytic activityhave been observed and quantified previously under similar in-vitro experimental conditions [3,47], we are led to hypothesize, butnot yet finally conclude that FXII undergoes changes in conforma-tion following contact-activation with an activator surface (irre-spective of its surface energy or chemistry), in the absence ofplasma proteins, as hypothesized by earlier studies. Section 3.3indicates that conformer formation exhibits no measurable ki-netics and does not depend upon activator surface-area or FXIIconcentration. It is not certain whether all available FXII is con-verted into active conformers. The self-limiting nature of autoac-tivation, along with low yields of the autoactivation suite (<10%) asmeasured by chromogenic and plasma coagulation assay seem to

A. Golas et al. / Biomaterials 37 (2015) 1e12 11

indicate either partial conversion of FXII into conformers, or weakprocoagulant and amidolytic potential of conformers.

Moreover, as described in Section 3.1, conformer activity varieswith surface chemistry/energy of activator surfaces. We cannotspeculate based on this work, on the nature of changes that takeplace in the 3-D conformation of the FXII molecule, but evidencepresented herein seems to suggest that FXII can indeed exist inmore than one conformer state, possibly depending on vicinalwater properties [48] (here “vicinal water refers to the interfacialaqueous phase separating the physical surface from the bulk so-lution that is hypothesized to have different properties than thesolution.

4.2. FXII contact-activation in the presence of prekallikrein andhigh molecular-weight kininogen

Products of FXII contact-activation with clean-glass in thepresence of high-molecular weight kininogen (HMWK) and pre-kallikrein (PK) in PBS differed significantly from those in case of FXIIcontact-activation in the absence of proteins. In this experiment,evidence of proteolysis was apparent in both experiment andcontrol vials as these showed multiple protein bands with molec-ular weight <80 kDa. Proteolytic fragments with average molecularweights of ~58 kDa (as read from Experion unit and not correctedfor error due to glycosylation) constituted the highest percentage inthe suite (see Table 1). It is important to note that the distribution offragments was similar in both experiments (with clean-glass acti-vator) and control (without clean-glass activator surface) and couldbe due to the presence of minor concentrations of either aFXIIa orKallikrein, which trigger the reciprocal activation pathway. Alter-natively, since FXII autoactivation, and subsequent conformer for-mation, does not exhibit surface area dependence and is efficienteven with hydrophobic surfaces, the small surface area of poly-styrene test-tubes might have been sufficient to produce smallquantities of FXII conformers which in turn trigger the reciprocal-activation route. This finding seems to support earlier results byKaushik et al. (citation) which indicated that prekallikrein hydro-lysis is not localized to an activation complex on the procoagulantsurface [24]. Either scenario is indicative of the importance ofreciprocal-activation and its role as the major aFXIIa generator asshown previously [38].

The protocol used for this experiment, while simplistic doesindicate significant potential in a systematic analysis of contact-activation of FXII in the presence of plasma proteins.

4.3. Procoagulant activity following autohydrolysis in buffersolution

As is evident from Figs. 5 and 6, autohydrolysis does not lead toproduction of proteins with procoagulant activity, where activity ismeasured in terms of plasma coagulation time (CT). Whereas thisfunctional assay does not identify the number or kind of proteolyticproteins produced by autohydrolysis, and proteolytic activity doesnot in itself measure concentrations of putative protein fragments/conformers, the measurable decrease in coagulation time is evi-dence for to the formation of proteins that directly affect aFXIIaprocoagulant activity, or procoagulant-stimulus transmission downthe cascade.

Whereas the specific biochemical mechanism for autohydrolysisremains open to speculation, possible mechanisms can involve theFXIImolecule itself (i.e. denaturation or changes in conformation ofaFXIIa) that render it less effective; action of aFXIIa on target-proteins (i.e. blockage/inhibition of proteolytic cleavage of FXI andprekallikrein; or other components of the cascade downstream thatinfluence the coagulation time CT). It is reasonable to speculate that

the cause of the aforementioned effects is (are) either proteinfragments, conformers of FXII and/or aFXIIa, or complexes of thesetwo proteins. Inhibition is maximal at lower concentrations ofaFXIIa and plateaus to a minimum of ~2 min as [aFXIIa] increases.This is possibly because at higher [aFXIIa], procoagulant stimulus>> threshold value required for sufficient fibrin formation. Henceeven if a small portion of aFXIIa is inhibited, the final effect wouldbe much less pronounced. Increasing [FXII] leads to an increase inDCT and DCT/[aFXIIa] as shown in Fig. 6, pointing to a direct increasein autohydrolysis products with an increasing [FXII].

4.4. Amidolytic activity following autohydrolysis in buffer solutions

Fig. 7 illustrates that DAbsorbance rate decreases in a linear mannerwith increasing aFXIIa. The decrease is linear rather than asymp-totic (as in plasma) most likely because the PBS/chromogen systemdoes not suffer from stimulus-processing limitations as in humanplasma [5].

These results along with that of Section 4.2, point to the for-mation of FXII-derivative proteins that have amidolytic activity, butnot procoagulant activity. These proteins may possibly be similar toFXIIact proteins which exhibit amidolytic activity but not procoa-gulant activity, and additionally have activation-suppressing po-tential [3].

4.5. Electrophoretic identity of autohydrolysis products

Figs. 1 and 8 illustrate the electrophoretic identity of reagents(FXII and aFXIIa) and autohydrolysis proteins respectively. A carefulcomparison of the electrophoretic-fragment profile in the twofigures reveals little evidence of proteolytic cleavage followingautohydrolysis. Moreover, a systematic comparison of fragmentconcentrations revealed that most protein fragment concentrationsfor autohydrolysis were not statistically different from a simplesum of the reagents. We did observe an isolated increase in the58 kDa fragment following autohydrolysis, but did not see aconcomitant change in any other band. The differences betweenthe two could thus be due to experimental differences in amount ofprotein loaded and assayed.

We are thus led to hypothesize, but not conclude, that autohy-drolysis leads to the formation of conformers of FXII that haveamidolytic activity but no procoagulant activity. We cannot proveor disprove the possibility of the formation of additional con-formers with neither amidolytic nor procoagulant activities thatinhibit aFXIIa procoagulant stimulus and transmission down thecascade.

5. Conclusions

High-resolution electrophoresis of FXII autoactivation proteinsin neat-buffer solution does not show bands indicative of proteo-lytic fragment formation. Change in experimental parameters suchas activator-surface area, activator surface-energy, FXII solutionconcentration and incubation time did not lead to formation ofeither aFXIIa or bFXIIa as evidenced by automated electrophoresis.Correlating this with previous experimental findings which indi-cate that FXII autoactivation in the absence of plasma proteinscreates proteins that exhibit procoagulant and/or amidolytic po-tential, the relative proportions of which depend on activator sur-face chemistry/energy, we are led to hypothesize thatautoactivation creates an ensemble of single-chain active FXIIconformers. These conformers may possibly be an intermediatestep in the contact-activation reactions as they undergo proteolyticcleavage in the presence of prekallikrein and high-molecularweight kininogen, as suggested by earlier investigators.

A. Golas et al. / Biomaterials 37 (2015) 1e1212

Furthermore, we have found that net procoagulant activity (pro-tease activity inducing clotting of plasma) measured by a plasma-coagulation assay, decreases systematically with increasing FXIIsolution concentration. This implies that aFXIIa activity is inhibitedby an unknown reaction between FXII and aFXIIa. Under the samereaction conditions, chromogenic assay reveals that net amidolyticactivity (cleavage of amino acid bonds in s-2302 chromogen) in-creases with increasing FXII solution concentration. This shows thatthe discrepancy between chromogenic and plasma-coagulationassays for aFXIIa in buffer solution is not due to autoamplificationbut rather an autohydrolysis reaction of the type suggested by thechemical formula FXIIa þ FXII / amidolytic enzyme(s). High-resolution electrophoresis of autohydrolysis products did notshow evidence of proteolytic cleavage of either FXII or aFXIIa. Weare thus led to hypothesize that the autohydrolysis reaction gen-erates a change in conformation in either the FXII or aFXIIa mole-cule, which alters further aFXIIa procoagulant activity.

Acknowledgments

This work was supported by National Institute of Health grantPHS 5R01HL069965. Authors appreciate support from the De-partments of Bioengineering and Materials Science and Engineer-ing, The Pennsylvania State University.

References

[1] Vogler EA, Siedlecki CA. Contact activation of blood-plasma coagulation. Bio-materials 2009;30:1857e69.

[2] Golas A, Parhi P, Dimachkie ZO, Siedlecki CA, Vogler EA. Surface-energydependent contact activation of blood factor XII. Biomaterials 2010;31:1068e79.

[3] Golas A, Joshua Yeh C-H, Siedlecki CA, Vogler EA. Amidolytic, procoagulant,and activation-suppressing proteins produced by contact activation of bloodfactor XII in buffer solution. Biomaterials 2011;32:9747e57.

[4] Yeh C-HJ, Dimachkie ZO, Golas A, Cheng A, Parhi P, Vogler EA. Contact acti-vation of blood plasma and factor XII by ion-exchange resins. Biomaterials2011;33:9e19.

[5] Golas A, Yeh C-HJ, Pitakjakpipop H, Siedleck CA, Vogler EA. A comparison ofblood factor XII autoactivation in buffer, protein cocktail, serum, and plasmasolutions. Biomaterials 2013;34:607e20.

[6] Fastrez J. Engineering allosteric regulation into biological catalysts. Chem-biochem 2009;10:2824e35.

[7] McMillin CR, Saito H, Ratnoff OD, Walton AG. Secondary structure of humanHageman-factor (Factor-XII) and its alteration by activating agents. J ClinInvest 1974;54:1312e22.

[8] Revak SD, Cochrane CG, Bouma BN, Griffin JH. Surface and fluid phase activ-ities of 2 forms of activated Hageman-factor produced during contact acti-vation of plasma. J Exp Med 1978;147:719e29.

[9] Dunn JT, Silverberg M, Kaplan AP. The cleavage and formation of activatedhuman Hageman-factor by autodigestion and by kallikrein. J Biol Chem1982;257:1779e84.

[10] Silverberg M, Dunn JT, Garen L, Kaplan AP. Auto-activation of humanHageman-factor - demonstration utilizing a synthetic substrate. J Biol Chem1980;255:7281e6.

[11] Dunn JT, Silverberg M, Kaplan AP. Structural-changes accompanyingHageman-factor activation by proteolysis. Clin Res 1982;30. A315-A.

[12] Miller G, Silverberg M, Kaplan AP. Autoactivatability of human Hageman-factor (Factor-XII). Biochem Biophys Res Commun 1980;92:803e10.

[13] Silverberg M, Kaplan AP. Enzymatic-activities of activated and zymogen formsof human Hageman-factor (Factor-XII). Blood 1982;60:64e70.

[14] Vanderkamp K, Vanoeveren W. Factor-XII fragment and kallikrein generationin plasma during incubation with biomaterials. J Biomed Mater Res 1994;28:349e52.

[15] Maas C, Oschatz C, Renne T. The plasma contact system 2.0. Semin ThrombHemost 2011;37:375e81.

[16] Samuel M, Pixley RA, Villanueva MA, Colman RW, Villanueva GB. Humanfactor-XII (Hageman-factor) autoactivation by dextran sulfate - circular-dichroism, fluorescence, and ultraviolet difference spectroscopic studies.J Biol Chem 1992;267:19691e7.

[17] McMillin CR, Walton AG. Circular-dichroism technique for study of adsorbedprotein structure. J Colloid Interf Sci 1974;48:345e9.

[18] Griep MA, Fujikawa K, Nelsestuen GL. Possible basis for the apparent surfaceselectivity of the contact activation of human-blood coagulation factor-XII.Biochemistry-US 1986;25:6688e94.

[19] Samuel M, Samuel E, Villanueva GB. The low pH stability of humancoagulation-factor-XII (Hageman-factor) is due to reversible conformationaltransitions. Thromb Res 1994;75:259e68.

[20] Ratnoff OD, Saito H. Amidolytic properties of single-chain activated Hageman-factor. P Natl Acad Sci U S A 1979;76:1461e3.

[21] Revak SD, Cochrane CG, Johnston AR, Hugli TE. Structural-changes accompa-nying enzymatic activation of human Hageman-factor. J Clin Invest 1974;54:619e27.

[22] Pixley RA, Colman RW. Factor-XII e Hageman-factor. Proteolytic enzymes incoagulation, fibrinolysis, and complement activation, part A, vol. 222; 1993.p. 51e65.

[23] Henriques E, Floriano W, Reuter N, Melo A, Brown D, Gomes J. The search for anew model structure of beta-factor XIIa. J Comput Aided Mol Des 2001;15:309e22.

[24] Chatterjee K, Thornton JL, Bauer JW, Vogler EA, Siedlecki CA. Moderation ofprekallkrein-factor XII interactions in surface activation of coagulation byprotein-adsorption competition. Biomaterials 2009;30:4915e20.

[25] Chatterjee K, Vogler EA, Siedlecki CA. Procoagulant activity of surface-immobilized Hageman factor. Biomaterials 2006;27:5643e50.

[26] Guo Z, Bussard KM, Chatterjee K, Miller R, Vogler EA, Siedlecki CA. Mathe-matical modeling of material-induced blood plasma coagulation. Biomaterials2006;27:796e806.

[27] Friberger P. Synthetic peptide substrate assays in coagulation and fibrinolysisand their application on automates. Semin Thromb Hemost 1983;9:281e300.

[28] Krishnan A, Liu YH, Cha P, Woodward R, Allara D, Vogler EA, et al. An eval-uation of methods for contact angle measurement. Colloid Surf B 2005;43:95e8.

[29] Lander LM, Siewierski LM, Brittain WJ, Vogler EA. A systematic comparison ofcontact-angle methods. Langmuir 1993;9:2237e9.

[30] Parhi P, Golas A, Barnthip N, Noh H, Vogler EA. Volumetric interpretation ofprotein adsorption: capacity scaling with adsorbate molecular weight andadsorbent surface energy. Biomaterials 2009;30:6814e24.

[31] Nguyen MSW. Application of the experion automated electrophoresis systemto glycoprotein visualization and analysis: Bio_Rad Laboratories.

[32] Espana F, Ratnoff OD. Activation of Hageman-factor (Factor-XII) by sulfatidesand other agents in the absence of plasma proteases. J Lab Clin Med 1983;102:31e45.

[33] Revak S, Cochrane C, Griffin J. Binding and cleavage characteristics of humanHageman-factor during contact activation e comparison of normal plasmawith plasmas deficient in factor-XI, prekallikrein, or high molecular-weightkininogen. J Clin Invest 1977;59:1167e75.

[34] Dunn J, Kaplan A. Formation and structure of human Hageman-factor frag-ments. J Clin Invest 1982;70:627e31.

[35] Ratnoff OD. Studies on the inhibition of ellagic acid activated Hageman-factor(Factor-XII) and Hageman-factor fragments. Blood 1981;57:55e8.

[36] Ratnoff OD. The relative amidolytic activity of Hageman-factor (Factor-XII)and its fragments - the effect of high-molecular-weight kininogen and kaolin.J Lab Clin Med 1980;96:267e77.

[37] Wuepper KD, Cochrane CG. Effect of plasma kallikrein on coagulation in-vitro.P Soc Exp Biol Med 1972;141. 271-&.

[38] Chatterjee K, Guo Z, Vogle EA, Siedlecki CA. Contributions of contact activationpathways of coagulation factor XII in plasma. J Biomed Mater Res A 2009;90A:27e34.

[39] Zhuo R, Vogler EA. Practical application of a chromogenic FXIIa assay. Bio-materials 2006;27:4840e5.

[40] Colman RW, Wachtfogel YT, Kucich U, Weinbaum G, Hahn S, Pixley RA, et al.Effect of cleavage of the heavy-chain of human-plasma kallikrein on itsfunctional-properties. Blood 1985;65:311e8.

[41] Revak SD, Cochrane CG, Griffin JH. Multiple forms of active Hageman-factor(HF) (coagulation Factor-XII) produced during contact activation. Fed Proc1977;36:329.

[42] Revak SD, Cochrane CG. Hageman-factor e its structure and modes of acti-vation. Thromb Haemost 1976;35:570e5.

[43] Revak SD, Cochrane CG, Griffin JH. Plasma-proteins necessary for cleavage ofhuman Hageman-factor during surface activation. Fed Proc 1976;35:692.

[44] Ratnoff OD, Saito H. The evolution of clot-promoting and amidolytic activitiesin mixtures of Hageman-factor (factor-XII) and ellagic acid e the enhancingeffect of other proteins. J Lab Clin Med 1982;100:248e60.

[45] Heimark RL, Kurachi K, Fujikawa K, Davie EW. Surface activation of blood-coagulation, fibrinolysis and kinin formation. Nature 1980;286:456e60.

[46] Citarella F, Wuillemin WA, Lubbers YTP, Hack CE. Initiation of contact systemactivation in plasma is dependent on factor XII autoactivation and not onenhanced susceptibility of factor XII for kallikrein cleavage. Br J Haematol1997;99:197e205.

[47] Zhuo R, Siedlecki CA, Vogler EA. Autoactivation of blood factor XII at hydro-philic and hydrophobic surfaces. Biomaterials 2006;27:4325e32.

[48] Vogler EA. Protein adsorption in three dimensions. Biomaterials 2012;33:1201e37.