48 BioProcess International FEBRUARY 2003

TT he production ofrecombinant proteins astherapeutics requirescareful assessment ofprotein quality, including

measurements of identity, quantity,purity, and activity. Because eachprotein has its unique set of activities,assays must be customized for eachproduct — and activity assays areoften the most difficult to develop.

Serp-1 represents the first of anew class of therapeuticantiinflammatory proteins derivedfrom viruses. Viruses have evolvedmany mechanisms to prevent,disable, or subvert a host’s innateimmune and inflammatory defensesystems. These mechanisms includeproducing viral proteins that affectchemokine, cytokine, andcomplement signaling cascades andthe signals that induce apoptosis andmajor histocompatibility complex(MHC) class-1 surface expression (1,2). Serp-1 was originally isolated

from myxoma virus and shown to be a member of the serpinsuperfamily of serine proteinaseinhibitors (3). Biochemical studieshave demonstrated Serp-1 to be anefficient inhibitor of human uPA(urokinase type plasminogenactivator), tPA (tissue typeplasminogen activator), and plasmin(4). Studies in animal model systemshave demonstrated that Serp-1 is ahighly effective inhibitor of restenosisand of the vasculopathy that leads toorgan failure during chronic rejectionof organ transplants (5–7).

Recombinant Serp-1 has beenexpressed in engineered CHO(Chinese hamster ovary) cells andpurified using standard columnchromatography methods (4). Toensure the quality of the purifiedproduct, a number of Serp-1 activityindicating assays have beendeveloped. We describe thedevelopment of three assays, two ofwhich measure the interaction orinhibition of the serine proteinaseuPA. The third assay is a captureELISA (enzyme-linkedimmunosorbent assay) that uses aconformation-specific monoclonalantibody. This antibody is specificfor the active form of Serp-1, whichenables the ELISA to function as asurrogate Serp-1 activity-indicatingassay. All three activity-indicatingassays produced similar profiles in aseries of Serp-1 stability studies.

METHODS AND MATERIALS

Band-shift assay. The stock uPA(Sigma-Aldrich, U1131) was diluted to 2.7 unit/mL (equal to

10 µg/mL) in reaction buffer (100 mM NaCl, 2 mM CaCl2,0.005% Triton X-100, 100 mMTris-HCl, pH 7.5). Serp-1 wasdiluted in the same reaction bufferto three concentrations: 40 µg/mL,20 µg/mL, and 10 µg/mL. A 5-µLsample of uPA was mixed with 5 µLof the three different Serp-1concentrations and incubated atroom temperature for two hours. A10-µL aliquot of 2� SDS-PAGE(sodium dodecyl sulfate-polyacrylamide gel electrophoresis)loading buffer (0.1 M Tris, 4% SDS,and 20% glycerol, pH 6.8) wasadded to stop the reaction. Themixture was separated byelectrophoresis on a 12% SDS-polyacrylamide gel. The separatedproteins were electrophoreticallytransferred from the gel to aHybond-C membrane (Amersham

Three Activity-Based Assays for Serp-1A Viral-Derived Serine Proteinase InhibitorXing Li, Elaine King, Colin Macaulay, Alexandra Lucas, Grant McFadden

PRODUCT FOCUS: ANTI-INFLAMMATORY VIRAL PROTEIN

PROCESS FOCUS: ASSAY

DEVELOPMENT

INTENDED AUDIENCE: LABORATORY

RESEARCHERS, QA/QC

KEY WORDS: ACTIVITY ASSAY

DEVELOPMENT; CONFORMATION-SPECIFIC MONOCLONAL ANTIBODY;VIRAL SERPIN, ELISA

LEVEL OF COVERAGE: INTERMEDIATE

AUDIENCE IN ASSAY DEVELOPMENT

PHOTODISC (WWW.PHOTODISC.COM)

FEBRUARY 2003 BioProcess International 49

Biosciences, RPN303E). Themembrane was blocked by mixing inblotto using 5% nonfat milk in 1�phosphate-buffered saline (PBS) and0.1% Tween for one hour at roomtemperature.

The Serp-1 protein was detectedby first incubating the membranewith an anti-Serp-1 monoclonalantibody (VT-mAbA) diluted inblotto, washing the membrane inPBS, and then incubating with ahorseradish peroxidase (HRP)-conjugated goat antimouse IgG(Pierce Biotechnology, P21124)diluted 1:20,000 in blotto. After aone-hour incubation, the blot wasrinsed in PBS, the antibodiesdetected using a chemiluminescencekit (Amersham, 1059243), and theimages recorded on Kodak scientificfilm (NEN Life Science Products,NNK-8646770). Densitometry andimage analysis were performed withQuantity One software (Bio-RadLaboratories).

Chromogenic uPA inhibition assay.The activity of each lot of uPA wasdetermined using the substrateChromozym U (Roche AppliedScience, 836583). The amount ofuPA that gave an absorbancereading (OD405 nm) of 0.90 �0.10 after one hour at 37 ºC wasselected for use in this assay. A 25 µL aliquot of uPA at thisconcentration was mixed with a 25 µL aliquot of Serp-1 andincubated at room temperature fortwo hours in a 96-well microplate.Typically, six serial dilutions of Serp-1 (0.2 µg/mL, 0.4 µg/mL,0.8 µg/mL, 1.2 µg/mL, 1.6 µg/mL, and 2.0 µg/mL) werefirst prepared in reaction buffer(100 mM NaCl, 2 mM CaCl2,0.005% Triton X-100, 100 mMTris-HCl, pH 7.5). After the two-hour incubation, a 100-µL aliquotof 0.5 mM Chromozym U wasadded to the uPA mixture. The platewas placed in a 37 ºC microplatereader (Thermo LabSystems), andthe absorbance at 405 nm wasdetermined for one hour.

Capture ELISA. A 50-µL aliquot ofanti-Serp-1 monoclonal antibodydiluted to 1.6 µg/mL in coatingbuffer (Sigma, C-3041) was

adsorbed to the surface of a 96-wellmicroplate (Corning, 3590) by anovernight incubation at 4 ºC. Eachwell was then coated with blockingbuffer (1 � PBS containing 2% BSAand 0.05% Tween-20) to preventnonspecific binding. The Serp-1 testsolution was diluted to a properrange and added to the antibody-coated plate. After a one-hourincubation at room temperature,unbound proteins were washed awayand each well incubated with asecond biotin-labeled anti-Serp-1antibody for one hour at roomtemperature. The wells were washedand each well incubated with HRP-conjugated streptavidin (Pierce,P21124) for one hour at roomtemperature. The unboundstreptavidin was washed away, andthe bound HRP-streptavidin wasdetected with the chromogenicsubstrate TMB (3, 3´, 5, 5´tetramethy benzidine chromgen,Pierce P34021). Color developmentwas stopped with 2N H2SO4, andthe absorbance (450nm) values wererecorded using a microplate reader(Bio-Tek Instruments).

The Serp-1 concentration in asample was calculated by comparingto a standard curve created usingpurified Serp-1. All samples andstandards were run in duplicate. Theassay linear range was between 1 and 20 ng/mL.

RESULTS AND DISCUSSION

Band-shift assay. Serpins share aconserved core structure and a

common mechanism of serineproteinase inhibition (8). A targetproteinase binds to the serpin’sreactive center loop (RCL) and triesto cleave the peptide bond betweenthe P1 and P1´ residues. The P1residue is therefore critical fordetermining proteinase specificity.As soon as the P1-P1´ peptide bondis cleaved, the serpin snaps into amore stable conformation. Thischange in conformation traps theserpin/proteinase complex at anintermediate step in the cleavageprocess in which a covalent linkageis formed between the P1 residue ofthe serpin and the active site serineof the proteinase. Thisserpin/proteinase complex is stableto the detergent SDS and can beseparated from both the free serpinand proteinase by SDS-PAGE.

When Serp-1 is mixed with uPA,the Serp-1/proteinase complex isformed. This complex has a slowermobility on a polyacrylamide gelcompared to that of the free Serp-1,resulting in an apparent band-shift(Figure 1). In this particularexample, a monoclonal antibodyagainst Serp-1 (VT-mAbA) is usedto detect both the native and shiftedforms on an immunoblot. Thismakes the band-shift assay verysensitive in that very little Serp-1(50 ng) is required for detection.When the molar ratio of Serp-1 anduPA is 1:1 (Figure 1, lanes 3, 6, and9), nearly 100% of the Serp-1 isshifted. This indicates that nearly allof the Serp-1 is active, because any

Figure 1: Results of a band-shift assay. The activity of three different lots of Serp-1 (A, B, andC) was compared. Serp-1 at various amounts (200 ng: lanes 1, 4, 7; 100 ng: lanes 2, 5, 8; 50 ng: lanes 3, 6, 9) was mixed with uPA (50 ng) and analyzed in immunoblot. Serp-1 bandswere probed using an anti-Serp-1 monoclonal antibody (VT-mAbA) and an HRP-conjugatedgoat anti-mouse secondary antibody. The Serp-1/uPA complex is seen in the shifted position,and the native Serp-1 is seen in the unshifted position.

inactive Serp-1 would appear at theunshifted position. When the molarratio of Serp-1 to uPA is 2:1 (lanes2, 5, and 8) or 4:1 (lanes 1, 4, and7), then the portion of Serp-1 thatcannot react with uPA remains atthe unshifted position. The use of aSDS-polyacrylamide gel allows forthe separation of the native Serp-1from the Serp-1/proteinase complexand from any proteolyticallydigested Serp-1. Using such a band-shift assay, numerous proteinaseshave been screened for their abilityto form SDS-stable complexes (4,9). In addition to uPA, both tPAand plasmin can also efficiently formSDS-stable complexes with Serp-1.

One advantage of the band-shiftassay is that active Serp-1 isphysically separated from inactiveSerp-1. Any treatment that affectsSerp-1 activity or the proportion ofshifted to unshifted Serp-1 is easilymonitored. One disadvantage is thatthe assay is not preciselyquantitative. Although immunoblotscan be scanned and the ratio ofshifted to unshifted Serp-1calculated, this has a limited linearrange. In addition, the antibodyused to detect the Serp-1 may nothave the same reactivity toward thenative Serp-1 and the Serp-1/proteinase complex.

Chromogenic uPA inhibition assay.The amount of proteinase in

solution is commonly quantitatedusing a proteinase-specificchromogenic substrate. As thesubstrate is cleaved, theconcentration of the chromophorereleased into solution is quantitatedspectrophotometrically. ChromozymU has been used as a chromogenicsubstrate for uPA. A quantitativeassay of Serp-1 activity wasdeveloped by mixing serial dilutionsof Serp-1 with a singleconcentration of uPA. After a two-hour incubation, excesschromogenic substrate was addedand the mixture incubated at 37 °C.The amount of residual uPA activitywas determined by measuring theamount of chromophore releasedinto solution over time (Figure 2a).The amount of uPA activity isrepresented by the increase ofchromophore (OD405) with time,or the slope of the curves shown inFigure 2a. As the concentration ofSerp-1 increases, the amount ofresidual uPA activity (slope)decreases. The slopes of all curveswere linear, indicating that theamount of substrate was notlimiting during the time course ofthis reaction.

The amount of Serp-1 activitycan be expressed in a number ofways. The simplest way is todetermine the concentration ofSerp-1 that inhibits half of the uPA

activity (IC50). Figure 2b shows aplot of the slopes of the curves inFigure 2a versus the concentrationof Serp-1 in the reaction.Uninhibited uPA activity isindicated by the slope value at aSerp-1 concentration of zero. TheIC50 can be determined byextrapolating the Serp-1concentration needed to give aslope that is 50% or half the value ofthe uninhibited uPA. This typicallygives an IC50 of ~1 µg/mL for pureSerp-1 with an interassay coefficientof variation of less than 10% (datanot shown).

For this assay to workconsistently, it is important tocontrol the amount of uPA that isused in the reaction. The amount ofuPA is always determined by thecontrol reaction without Serp-1.Alternatively, one can use a standardpreparation of uPA with a definednumber of international units (IU),which can be obtained from theNIBSC (National Institute forBiological Standards and Controls).Assuming that one unit of Serp-1inhibits one unit of protease, thenan IC50 of 1 µg/mL corresponds toa Serp-1 specific activity of ~60,000Units/mg.

The chromogenic assay has anumber of advantages over theband-shift assay: It can be adaptedto a high throughput plate-basedformat; it provides a quantitativedetermination of Serp-1 activity;and it is highly reproducible.

Conformation-specific captureELISA. A capture ELISA often isused to quantitate the level of aspecific antigen in solution. A pairof antigen-specific antibodies isrequired for this type of assay (10).The first antibody is immobilized onthe surface of a microtiter plate,which is then exposed to antigen-containing solution during thecapture phase of the assay. The firstantibody binds the antigen, andunbound material is washed fromthe plate. A second antibody is thenadded to detect the presence of theimmobilized antigen. The secondantibody is usually labeled with anenzyme or marker that can then bequantitated by absorbance,

Figure 2: Results of a chromogenic uPA inhibition assay. A given concentration of uPA wasmixed with Serp-1 at 0.0 µg/mL, 0.2 µg/mL, 0.4 µg/mL, 0.8 µg/mL, 1.2 µg/mL, 1.6 µg/mL,and 2.0 µg/mL (a, top to bottom). After two hours, 0.5 mM Chromozym U was added toeach sample, and the absorbance (405 nm) monitored at 37 ºC for one hour. The reactionrates of seven samples are presented (a). A replot of the slopes of each sample in (a) versusthe concentration of Serp-1 is presented in (b). The IC50 is extrapolated from the slope valuethat is one half the maximum slope value in the uninhibited control.

50 BioProcess International FEBRUARY 2003

52 BioProcess International FEBRUARY 2003

fluorescence, or chemiluminescence.The level of antigen in a solution isdetermined by comparing theELISA signal to a standard curvecreated using pure antigen ofknown concentration.

Twelve anti-Serp-1 monoclonalantibodies were available forconformation-specific ELISAdevelopment. Two antibodies (VT-mAbC and VT-mAbD) were testedas capture antibodies; the two otherantibodies (VT-mAbA and VT-mAbB) were labeled with biotin andtested as detection antibodies. Thesefour antibody pairs (A/C, A/D,B/C, and B/D) were tested fortheir ability to detect differentconformational forms of Serp-1.Three different conformationalforms of Serp-1 were prepared:native Serp-1 as typically purified;Serp-1 in an inhibitory complexwith uPA; and heat-denatured Serp-1 (75 ºC, 5 minutes). Thethree conformational forms of Serp-1 were adjusted inconcentration from 50 ng/mL to0.75 ng/mL and tested in thecapture ELISA. Figure 3 presentsthe results.

All four antibody pairs were ableto detect native Serp-1 (Figure 3a).The A/C pair was the most sensitive,the B/D pair the least sensitive, andthe B/C pair had the widest linearrange (0.75–24 ng/mL). None ofthe antibody pairs were able todetect heat-denatured Serp-1 (Figure3c). It is known that VT-mAbA candetect Serp-1 on an immunoblot(Figures 1 and 4a), suggesting it maybe able to detect heat-denaturedSerp-1, but that neither of thecapture antibodies (VT-mAbC andVT-mAbD) appeared to bind to heatdenatured Serp-1. Two of theantibody pairs (A/C and A/D) wereable to detect Serp-1 while it was in acomplex with uPA (Figure 3b). Theother two pairs (B/C and B/D)could not detect Serp-1 while it wasin a complex with uPA, suggestingthat the epitope for the detectionantibody VT-mAbB was masked oraltered by the presence of the bounduPA. Table 1 summarizes theconformational specificity of eachantibody. The B/C pair of antibodies

apparently can be used in a captureELISA that is specific for native Serp-1 and would not detect Serp-1 thatwas denatured or in a complex with atarget proteinase. The detection limitfor this assay is ~2 ng/mL with acoefficient of variation less than 20%(data not shown). Thus, the capture

ELISA with this pair of MAbs will benot only an assay to measure Serp-1concentrations, but a surrogate assayto indicate Serp-1’s activity.

SERP-1 STABILITY STUDIES

Serp-1 stability has been examinedby subjecting the purified protein to

Figure 3: Results of a capture ELISA detection of three conformational forms of Serp-1. Thefour combinations of capture and detection antibodies (A/C, A/D, B/C, and B/D) were usedto detect different conformational forms of Serp-1: (a) Serp-1 in its native conformation. (b)Serp-1 that was incubated with uPA for 2 hours. (c) Serp-1 that was heated to 75 ºC for 5minutes. All three Serp-1 solutions were adjusted to the same concentrations (X axis) anddetected in ELISA assays. The OD (450 nm) values were measured (Y axis).

54 BioProcess International FEBRUARY 2003

a variety of stress conditionsincluding heating, cycles of freezingand thawing, and exposure to lowpH conditions. Absorbance at 280nm, the band shift assay, thechromogenic uPA inhibition assay,and the capture ELISA using theB/C pair of antibodies were all usedto monitor Serp-1 during thevarious treatments.

For measuring the stability ofSerp-1 to heat, samples wereincubated at different temperaturesfor five minutes and then returnedto 4 ºC. The effect of the heattreatments on the ability of Serp-1to form an SDS-stable complexwith uPA was measured using theband-shift assay (Figure 4a). Nodifference in Serp-1 activity wasseen between 4 ºC and 45 ºC, butinactivation started to occur at

55 ºC. At 65 ºC and higher, noSerp-1 activity was detected usingthe band-shift assay. The level ofSerp-1 activity can be quantifiedusing a densitometer by measuringthe ratio of shifted to unshiftedSerp-1. When this ratio iscompared to that of the controlsample that was stored at 4 ºC, theremaining Serp-1 activity can beexpressed as a percent of thecontrol sample, and this is plottedin Figure 4b.

In addition to the band-shiftassay, a simple measurement ofabsorbance at 280 nm wasperformed on each heated sample.The absorbance increased withtemperature. This could be due toan increase in light scattering dueto denaturation and aggregation ofthe heated protein. The amount of

Serp-1 activity in each sample wasalso measured using thechromogenic uPA inhibition assayand expressed in terms of a percentof the control (4 ºC) sample(Figure 4b). Similarly, the level ofSerp-1 protein in each sample wasdetermined using the captureELISA with the B/C antibodypair. This ELISA used unheatedSerp-1 for the standard curve, andthe level of Serp-1 detected in eachheated sample was expressed as apercent of the untreated (4 ºC)control sample (Figure 4b). Thereis a clear correlation between theloss of Serp-1 activity as measuredby both the band-shift assay andthe chromogenic uPA inhibitionassay, and the loss of Serp-1detected by the capture ELISA.This suggests that the captureELISA could be used as a sensitivesurrogate activity-indicating assayfor Serp-1.

To determine the stability ofSerp-1 to repeat freezing andthawing, test samples were treatedto five cycles of freeze/thaw. Thisconsisted of freezing to less than–60 ºC and thawing to 22 ºC.After each freeze-thaw cycle, theamount of Serp-1 activity wasmeasured using the chromogenicuPA inhibition assay, and the levelof active Serp-1 protein wasdetermined using the conformation-specific capture ELISA with theB/C antibody pair (Table 2). Nosignificant change was seen in eitherSerp-1 activity or protein level,

Figure 4: Thermal stability of Serp-1. Serp-1 was incubated at various temperatures for fiveminutes and analyzed using either: the band shift assay; the chromogenic uPA inhibitionassay; the conformation-specific capture ELISA with the (B/C) MAb pair; or absorbance at280nm. (a) An image of the X-ray film showing the shifted and unshifted positions of Serp-1in the band shift assay. (b) A summary of the Serp-1 activity data from the capture ELISAassay, the band shift uPA banding assay, and the chromogenic uPA inhibition assay. Serp-1activity was expressed as percentages (left Y axis) of control samples, which were kept at 4 °C. A second axis indicates the values (OD280nm) for each sample.

Table 1: Interaction of four anti-Serp-1monoclonal antibodies (VT-mAbA, VT-mAbB,VT-mAbC, and VT-mAbD) with Serp-1 of threedifferent conformational forms in captureELISAs; (+) indicates that the MAb couldrecognize Serp-1 molecules.

Detecting MAb Coating MAb

A B C DNative Serp-1 + + + +

Serp-1/uPAcomplex + – + +

HeatedSerp-1 *+ – – –

* See Figure 4a for the detection of Serp-1 byimmunoblot using VT-mAbA.

FEBRUARY 2003 BioProcess International 55

indicating that the purified proteinis stable to at least five cycles offreeze thaw treatment.

Exposure to low pH, eitherduring elution from an affinitymatrix or as a means of virusinactivation, is a relatively commonoccurrence during the production ofa protein therapeutic. To examinethe stability of Serp-1 in an acidicenvironment, purified Serp-1 wasexposed to a number of acidicconditions for 60 minutes and thenneutralized (to ~pH 7.0). Theamount of Serp-1 activity remainingin each sample was measured usingthe band-shift assay, and the level ofactive Serp-1 protein wasdetermined using the conformation-specific ELISA (Figure 5). Asignificant drop in Serp-1 activity

was seen when the pH was less than3.7, and by pH 3.3, virtually allSerp-1 activity was lost. A similaractivity profile was seen using theconformation-specific ELISA.

CONCLUSIONS

Each protein therapeutic requires acustomized set of assays to monitorits biological activity. For Serp-1, wehave developed three assays, two ofwhich are based on the ability ofSerp-1 to inhibit the serineproteinase uPA. The third assay is acapture ELISA that uses aconformation-specific monoclonalantibody. In a series of stabilitystudies, any loss in Serp-1 activitythat was measured using the uPA-based assays was also reflected by aloss in Serp-1 protein using theconformation-specific ELISA. Theseresults indicate that this ELISA isspecific for active Serp-1 and that itcan be used as a surrogate activity-indicating assay. The high sensitivityand high throughput abilities of theELISA will facilitate the developmentof Serp-1 as a protein therapeutic.

ACKNOWLEDGMENTS

We are grateful to Heather Schneider andLing Chen for their excellent technicalassistance. The monoclonal antibodies usedin this study were provided by Biogen.

REFERENCES1 Alcami, A; Koszinowski, UH. Viral

Mechanism of Immune Evasion. Immunol.Today 2000, 9, 447–455.

2 McFadden, G; Murphy, PM. Host-Related Immunomodulators Encoded ByPoxviruses and Herpesviruses. Curr. Opin.Microbiol. 2000, 3 (4), 371–378

3 Upton, C; et al. Myxoma Virus andMalignant Rabbit Fibroma Virus Encode aSerpin-Like Protein Important for VirusVirulence. Virology 1990, 179, 618–631.

4 Nash, P; et al. Inhibitory Specificity ofthe Anti-Inflammatory Myxoma VirusSerpin, SERP-1. J. Biol. Chem. 1998, 273(33), 20982–20991.

5 Lucas, A; et al. TransplantVasculopathy: Viral Anti-Inflammatory SerpinRegulation of Atherogenesis. J. Heart LungTransplant. 2000, 19 (11), 1029–38

6 Miller, LW; et al. Inhibition ofTransplant Vasculopathy in a Rat AorticAllograft Model After Infusion of Anti-Inflammatory Viral Serpin. Circulation2000, 101, 1598–1605.

7 Hausen, B; et al. Viral SerineProteinase Inhibitor (Serp-1) EffectivelyDecreases the Incidence of GraftVasculopathy in Heterotopic HeartAllografts. Transplantation 2001, 72,364–368.

8 Silverman, GA; et al. The Serpins Arean Expanding Superfamily of StructurallySimilar but Functionally Diverse Proteins. J.Biol. Chem. 2001, 276 (36), 33293–33296.

9 Lomas, DA; et al. Inhibition ofPlasmin, Urokinase, Tissue PlasminogenActivator, and C1s By a Myxoma VirusSerine Proteinase Inhibitor. J. Biol. Chem.1993, 268, 516–521.

10 Baker, K; et al. Rapid Monitoring ofRecombinant Protein Products: A Comparisonof Current Technologies. Trends inBiotechnology 2002, 20 (4), 149–156. ��

Corresponding author Dr. Xing Li SeniorScientist, Elaine King is Laboratory andManufacturing Manager, Dr. ColinMacaulay is Research and DevelopmentManager, Dr. Alexandra Lucas is ChiefMedical Officer, and Dr. GrantMcFadden is Chief Scientific Officer, atViron Therapeutics, Inc., UWOResearch Park, 100 Collip Circle, Suite103, London, ON, Canada N6G 4X8,xli@vironinc.com. Additionally, Drs.Lucas and McFadden are associatedwith the Vascular Biology ResearchGroup of the John P. Robarts’ ResearchInstitute, London, ON, Canada; and theDepartment of Microbiology andImmunology, University of WesternOntario (UWO), London, ON, Canada.

Figure 5: The pH stability of Serp-1. Six separate aliquotes of Serp-1 (pH 7.2 in PBS) wereadjusted by the addition of acetic acid to a final pH of 4.00; 3.75; 3.70; 3.60; 3.40; or 3.30;and held for one hour before being neutralized to ~pH 7.2 with Tris-HCl. Each sample wastested using either the band-shift assay or the capture ELISA. Serp-1 activity in samples wasexpressed as percentages of the activity of the untreated control sample.

Table 2: Freeze–thaw stability of Serp-1 overfive cycles, measuring the Serp-1 activity usingthe capture ELISA with the MAb pair of B/Cand chromogenic uPA inhibition assay(measured as specific activity)

Freeze– Adsorbance ELISA Activitythaw cycle (OD280nm) (mg/mL) (unit/mg)

1 0.895 0.968 594062 0.913 —* 582523 0.924 0.900 576924 0.942 0.971 —*5 0.978 0.911 63830

*Not tested