Veterinary Immunology and lmmunopathology, 33 (1992) 145-154
145
Elsevier Science Publishers B.V., Amsterdam
An enzyme-linked immunosorbent assay (ELISA) for von Willebrand factor antigen (vWf-Ag) in canine plasma R.J. SlappendeP, R.A.J. Frielinka, J.A. MoP, A. Noordzij b a n d R. H a m e r ~ aDepartment of Clinical Sciences of Companion Animals and bDepartment of Immunology, Veterinary Faculty and CDepartment of Internal Medicine, Medical Faculty, Universityof Utrecht, Utrecht, Netherlands (Accepted 18 September 1991 )
ABSTRACT Slappendel, R.J., Frielink, R.A.J., Mol, J.A., Noordzij, A. and Hamer, R., 1992. An enzyme-linked immunosorhentassay (ELISA) for yon Willebrand factor antigen (vWf-Ag) in canine plasma. Vet. Immunol. Immunopathol., 33:145-154. A quantitative enzyme-linkedimmunosorbent assay (ELISA) has been developed to measure canine von Willebrand factor antigen (vWf-Ag) in plasma of the dog. A vWf-Ag antiserum was raised in rabbits and purified by preabsorption with the low molecular weight vWf-Ag-deficient fraction of canine cryoprecipitate, followed by affinity chromatography on protein-A Sepharose. The rabbit anticanine vWf-Ag IgG was used to bind the vWf-Ag of the test plasmas to the solid phase and to prepare the enzyme-antibody conjugate in ELISA. Normal rat serum was used as blocking agent. The standard curve was linear (r2> 0.98) and reproducible after loot-log transformation. The interassay coefficient of variation (CV) in test plasmas with various vWf-Ag concentrations was never greater than 7.7%. Assayed values in dilutions of pooled normal canine plasma added to canine vWf-Agdeficient plasma were linear between 0 and 100% ( r 2= 0.99 ) and indicated excellent analytical recovery of vWf-Ag. In 18 dogs with various internal diseases, including yon Willebrand's disease and haemophilia A, the coefficient of correlation between the results of the ELISA and those of electroimmunodiffusion (EID) was 0.93.
ABBREVIATIONS EID, electroimmunodiffusion; ELISA, enzyme-linked immunosorbent assay; FWP, fixed washed platelets; PPP, platelet poor plasma; vWD, yon Willebrand'sdisease; vWf-Ag, yon Willebrand factor antigen. Correspondence to: Dr. R.J. Slappendel, University Clinic for Companion Animals, Postbox 80.154, 3508 TD Utrecht, Netherlands.
© 1992 Elsevier Science Publishers B.V. All rights reserved 0165-2427/92/$05.00
146
R.J. SLAPPENDEL ET AL.
INTRODUCTION
Von Willebrand's disease (vWD) is a congenital bleeding disorder in which quantitative a n d / o r qualitative abnormalities of the factor VIII/yon Willebrand factor complex (FVIII-vWf) impair hemostasis. The von Willebrand factor moiety of the complex, called factor VIII-related antigen or more recently von Willebrand factor antigen (vWf-Ag) (Marder et al., 1986), can be precipitated with specific antisera. The occurrence of vWD in dogs has been well documented and a high incidence has been reported in some breeds (Johnson et al., 1988 ). Dogs with vWD provide useful models for the study of the FVIII-vWf complex (Bouma et al., 1976; Benson et al., 198 l; Brinkhouse et al., 1985). At present, the main tool to diagnose vWD in dogs is measurement of plasma vWf-Ag by rocket electroimmunodiffusion (EID) in agarose (Johnson et al., 1988). In our experience, the performance and interpretation of the rocket EID for vWf can be troublesome. Problems include low sensitivity, indistinct precipitation lines and failure to discriminate among extremely low and extremely high plasma vWf-Ag concentrations. The latter problem can be overcome by routinely performing tests in serial dilutions of test plasma in normal plasma, which is expensive, or by repeating a test in that way whenever readings are ambivalent, which is time-consuming. In addition, the rocket EID needs relatively large amounts of antiserum and is less appropriate for assaying large series. The technique applied in this paper is a sandwich method of an enzymelinked immunosorbent assay (ELISA) in which a specific IgG fraction of a rabbit anti-canine vWf serum is used to bind the vWf-Ag of the test plasma to the solid phase of microtitre wells. Soluble enzyme-linked antibody was used for labelling. The advantages of this technique are its relative technical simplicity, low cost, fair sensitivity, accuracy, and suitability for the screening of large series of test plasmas. MATERIALS AND METHODS
Plasma samples Venous blood samples were collected in plastic syringes and immediately anticoagulated with one part of sodium citrate (38 g 1-1 ) to nine parts of blood. The blood was centrifuged ( 15 min, 3000 × g, 18 °C ) and the plateletpoor plasma (PPP) was harvested, stored in aliquots of 0.5 ml in capped plastic vials at - 7 0 ° C, and thawed at 37 °C shortly before testing.
ELISA FOR vWf-Ag IN CANINE PLASMA
147
Animals Fifty-one apparently healthy, privately owned adult male and female dogs of various breeds and mixed breeding were used to establish the normal range and to constitute the normal plasma pool, which was stored in aliquots of 50 #l at - 70°C and used as the reference standard. Dobermann pinschers, German shepherds, Golden retrievers, terriers, and miniature Schnauzers, being breeds in which a high prevalence of vWD has been reported in the USA (Johnson et al., 1988 ), were not included. Plasma was also obtained from eight dogs with hemophilia A and three dogs with vWD. Hemophilia was diagnosed by a previously described method (Slappendel, 1975 ) and vWD was diagnosed by defective ristocetin-induced agglutination of human fixed washed platelets (FWP) in a semi-quantitative macroscopic tilt tube test (Rosborough et al., 1980) and by a lack of detectable amounts ofvWf-Ag in a rocket EID (Laurell, 1966 ) in agarose (Seakem, Bausch and Lomb, Rochester, NY, USA) with a commercially available goat anti-human vWf-Ag serum (Dakopats, Copenhagen, Denmark). The vWf-Ag-deficientplasma used for recovery studies was obtained from a dog which had vWD and severe hemorrhagic diathesis. This plasma did not precipitate in rocket EID with a rabbit anti-canine vWf-Ag serum ('Utrecht antiserum', generously provided by Dr. B.N. Bouma, Academic Hospital, Utrecht, Netherlands) (Benson et al., 1975), nor did it support ristocetininduced agglutination of FWP. The correlation between results of the ELISA and rocket EID assays was studied in the plasmas from 17 dogs with various internal diseases, including four dogs with hemophilia A and three dogs with vWD. The same frozen plasma standard from a pool of normal dogs was used in both methods.
Purification of v Wf-Ag VWf-Ag was purified from a cryoprecipitate from pooled plasma of eight fasted dogs by a modification of the method of van Mourik and Mochtar (1970). The rapidly frozen plasma was thawed at 4°C and the cryoprecipitate was isolated by centrifugation for 30 min at 10 000 × g (4 ° C ). The pellet was redissolved in 0.017 M barbital buffer (pH 7.0) containing 0.125 M NaC1 (pH 7.0), 0.01 M benzamidine (Sigma, St. Louis, MO, USA), and 0.01 M epsilon-amino-caproic acid (EACA; Sigma, St. Louis, MO, USA) and this mixture was centrifuged for 5 min at 1000×g to remove undissolved materials. The vWf-Ag was isolated from the clear supernatant by gel filtration on a (5 cm×50 cm) Sepharose C1-4B column (Pharmacia Fine Chemicals, Uppsala, Sweden) in barbital buffer containing 0.001 M benzamidine and 0.001 M EACA. The vWf-Ag was precipitated from the void volume fractions by dialysis against 1.9 M ammonium sulphate (pH 7.0) for 16 h at 4°C and
148
R.J. SLAPPENDEL ET AL.
stored at - 7 0 ° C until use. The protein content of the purified vWf-Ag solution was 0.03 Mg m l - l
Preparation of antiserum and IgG fraction Purified canine vWf-Ag emulsified in Freund's complete adjuvant was injected intramuscularly into rabbits. Then, a mixture of the antigen and Freund's incomplete adjuvant was injected at 4-week intervals until, after 6 months, antibody titres were sufficiently high. The antiserum was purified by preabsorption with the vWf-Ag deficient low molecular fraction ( < 600 kDa) of canine cryoprecipitate, which was covalently bound to Sepharose CL-4B using the CNBr method according to the manufacturer's instructions (Pharmacia Fine Chemicals, Uppsala, Sweden). Its IgG fraction was isolated by affinity chromatography on a protein-A Sepharose-CL-4B column (Pharmacia Fine Chemicals), eluted with 0.1 M glycine HC1 in 0.5 M NaC1 buffer, pH 2.5, neutralized with 0.5 M Tris buffer, and stored in capped plastic cups at - 70 ° C until used.
Immunoconjugate Aliquots of the purified IgG fraction were conjugated with horseradish peroxidase ( H R P ) (Sigma, St. Louis, MO, USA) (Nakana and Kawaoi, 1974) and stored at - 7 0 ° C until used.
ELISA procedure One hundred microliltres of the anti-canine vWf-Ag IgG in PBS ( 1% v/v ) were added to the wells of polystyrene microtitre plates (ELISA microtitre plates with high affinity (Art. 655061 ), Greiner, Frickenhausen, Germany). The plates were covered, incubated overnight at 37 ° C, thoroughly washed in running tap water to remove unadsorbed materials, and shaken dry. Remaining binding sites were blocked by the addition of 100/tl of rat serum ( 1% v / v ) in PBS to each well. After incubation for at least 10 min at room temperature, 100 ~tl of each sample to be tested was added in various dilutions (vide infra) to the rat serum. The plates were covered, incubated for 1 h at 37 oC, washed in running tap water to remove unbound material, and shaken dry. Then, 100 ~tl of the antibody-enzyme conjugate, diluted 1/300 in PBS, was added to each well. Plates were incubated for 1 h at 37°C, washed to remove unbound conjugate, and dried again. To measure peroxidase activity, 100/~1 of enzyme substrate was added to each well. The ELISA substrate was prepared just before use by adding 34/tl hydrogen peroxide, 35% ( v / v ) , to freshly prepared solution (pH 5.95) of 5-amino-2-hydroxybenzoic acid (Merck, Darmstadt, Germany ) (0.6 g 1-1 ) in distilled water.
149
ELISA FOR vWf-Ag IN CANINE PLASMA
Following 45 min incubation at room temperature, the reaction was stopped by the addition of 50 al NaOH (0.1 N), and quantitated within 30 min by measurement of the optical density at 405 n m (OD 405 ) with a spectrophotometer (Titertek multiscan, Flow Laboratories, Ayrshire, U K ) . Test and reference plasmas were assayed in duplicate at dilutions of 1:4, 1:8, and 1:16 in PBS. Non-specific absorption was determined by incubation of the antibody-enzyme conjugate in wells that were not preincubated with canine plasma. A standard curve was obtained by logit-log transformation of the absorbance at 405 n m vs. the dilution of the reference pool. U n k n o w n concentrations were extrapolated from a least-squares regression line of the logitlog curve and expressed as a percentage of the undiluted reference pool. RESULTS A standard curve for the determination of vWf-Ag was constructed by serial dilution of a pooled reference plasma. Fig. 1 shows a typical standard curve. After logit transformation (Fig. 2 ), the curve of the absorbance vs. the log of the dilution of the reference plasma was consistently linear from 1.6 to 100%. For 11 standard curves, each one constructed on a different day, the regression coefficients varied from 0.997 to 0.980 (mean 0.990). The coefficient of variation of 12 replicates of a normal plasma (mean vWfA g = 97.6%) in the dilutions routinely used for the construction of the reference curve ( 1 : 4, 1 : 8 and 1 : 16) was 3.9, 5.7 and 3.9%. The detection limit of the assay, calculated as the vWf-Ag concentration corresponding on the ref-
....---,
tD O
O
0
. . . . . . . 1
i 10
, ,
. . . . .
i 1QO
vWf-Ag (%)
Fig. 1. Dose-responsecurve for the ELISAofvon Willebrandfactor antigen (vWf-Ag) in pooled normal canine plasma. Depicted is the percentageof pooled plasma in PBS vs. the absorbance at 405 nm. The absorbance in undiluted plasma was arbitrarily defined as 100%vWf-Ag.
150
R.J. SLAPPENDEL ET AL. 1.50
0.75
0.00
-0.75
- 1.50
. . . . . . .
i
,
10 vWf-Ag
.
.
.
.
.
.
I
100 (%)
Fig. 2. Standard reference curve obtained from the dose-response curve (Fig. 1) after logit transformation of the OD 405 nm readings. erence curves to an absorbance at two standard deviations above the mean blank, measured in triplicate on 11 days, was 1.6%. Intra-assay variation was 10.7, 8.1, 4.6, 2.4 and 12.9% for 15 replications on a single plate in five test plasmas representing high (192%), normal (105%), borderline normal (52%), low (19%), and very low (6%) vWf-Ag concentrations. Interassay variation was 5.8, 7.7, 1.6, 4.2 and 7.5% in five test plasmas assayed in duplicate on 3 different days representing high (242%), high normal (130%), normal 97% ), borderline normal (52%), and very low (6%) vWf-Ag. Analytical recovery studied with different proportions of pooled normal plasma in vWf-Ag-deficient plasma (vWf-Ag less than 1.6%) was excellent for concentrations between 1.'6 and 100% when assayed at 1:4 to 1 : 16 dilutions in PBS (Fig. 2). At lower dilutions, recovery was incomplete and not linear. The correlation of results obtained by ELISA and Laurell EID in the plasmas of 17 dogs with various vWf-Ag concentrations is shown in Fig. 3. The vWf-Ag concentrations obtained in 51 clinically normal dogs ranged from 46 to 249% (mean 94.9%, SD 36.9%, median 92%). The 2.5 and 97.5 percentile confidence limits ( P < 0.05 ) calculated by a distribution-free method for use as clinical reference values (Riimke and Bezemer, 1972) were 48 and 172%, in close agreement with normal values reported by others (Rosborough et al., 1980; Johnson et al., 1988 ). The vWf-Ag concentrations estimated with ELISA in the plasmas of eight German shepherd dogs with acute or subacute bleeding problems associated with hemophilia A ranged from 84 to 535% (mean 202% ). High plasma vWf-
ELISAFORvWf-AgIN CANINEPLASMA
151
120
100
J
4
8O
6O
I
>
40 20
0
i
I
I
!
I
20
40
60
80
100
% Pool in v W f - A e - d e f i c i e n t
plasma
Fig. 3. Analytical recovery of von Willebrand factor antigen (vWf-Ag) in serial dilutions of pooled normal plasma in vWf-Ag-deficient plasma from a dog with severe yon Willebrand's disease.
Ag concentrations associated with hemophilia have been reported before in man (Denson, 1973)and dogs (Benson et al., 1975). DISCUSSION
ELISA techniques for vWf-Ag assay have been used for more than a decade in human medicine (Bartlett et al., 1976; Mazurier et al., 1977; Ness and Perkins, 1979; Yorde et al., 1979; Ingerslev, 1987) but have not been described yet for the canine species. Antisera against human vWf-Ag can be used for the detection of canine vWf-Ag, at least in the Laurell EID, but species-specific antiserum is mandatory for accurate quantitation, especially at low vWf-Ag concentrations (Benson et al., 1983 ). A species-specific problem associated with the isolation of canine vWf-Ag and the subsequent production of the antiserum is physiologic post-prandial hyperlipaemia of the donor dogs, as this may result in a low yield of cryoprecipitate and generation of heterospecific antibodies (Benson and Dodds, 1982). This was avoided by the use of plasmas from fasted dogs and by ultracentrifugation of the plasma. Another species-related problem consists of the aspecific heterologous immunoreactivity of canine sera for albumin and other proteins in the sera of many species. The reactivity with the proteins in the rat serum used as blocking agent in the ELISA was low, as indicated by the low OD 405 of the blanks, corresponding to 1.6% vWf-Ag (mean + 2 SD ) on the reference curve.
152
R.J. SLAPPENDEL ET AL.
A standard canine vWf-Ag preparation or international reference preparation for the assessment of accuracy is lacking, but our reference plasma consistently showed good linearity and reproducibility at seven serial dilutions from 1.6 to 100% (Fig. 1 ). The high correlation coefficient of the regression line (r2> 0.98 ) allowed reduction to three dilutions for the routine construction of the standard curve, thus saving time and materials. The intra-assay and interassay variation in test plasmas with various concentrations of vWf-Ag were good and compared well with those of canine vWf-Ag assays by the EID technique (Benson et al., 1983). The detection limit of the assay was considerably higher for ELISA (1.6%) than reported for optimal conditions of the EID (9%) (Benson et al., 1983). Results of ELISA and Laurell EID correlated very well over a wide range of plasma vWf-Ag concentrations (Fig. 4) but according to the regression line, values were higher by EID than by ELISA. This is probably due to the inaccuracy of the non-species-specific antiserum used in the EID, which gives inaccurate results at lower vWf-Ag values (Benson et al., 1983 ). It is probably also due to the EID technique itself, which gives artifactual variations in the vWf-Ag levels of plasma from dogs with vWD and hemophilia A, depending on the source of the agarose used (Benson et al., 1975). The high values obtained in hemophilic dogs also confirm that our ELISA measures the vWfpart rather than the F VIII-part of the F VIII-vWfcomplex. In conclusion, the ELISA assay for canine vWf-Ag, which is easy to per300
J
•
O
200 r~ CO
•
0
% o~ .
145
Elsevier Science Publishers B.V., Amsterdam
An enzyme-linked immunosorbent assay (ELISA) for von Willebrand factor antigen (vWf-Ag) in canine plasma R.J. SlappendeP, R.A.J. Frielinka, J.A. MoP, A. Noordzij b a n d R. H a m e r ~ aDepartment of Clinical Sciences of Companion Animals and bDepartment of Immunology, Veterinary Faculty and CDepartment of Internal Medicine, Medical Faculty, Universityof Utrecht, Utrecht, Netherlands (Accepted 18 September 1991 )
ABSTRACT Slappendel, R.J., Frielink, R.A.J., Mol, J.A., Noordzij, A. and Hamer, R., 1992. An enzyme-linked immunosorhentassay (ELISA) for yon Willebrand factor antigen (vWf-Ag) in canine plasma. Vet. Immunol. Immunopathol., 33:145-154. A quantitative enzyme-linkedimmunosorbent assay (ELISA) has been developed to measure canine von Willebrand factor antigen (vWf-Ag) in plasma of the dog. A vWf-Ag antiserum was raised in rabbits and purified by preabsorption with the low molecular weight vWf-Ag-deficient fraction of canine cryoprecipitate, followed by affinity chromatography on protein-A Sepharose. The rabbit anticanine vWf-Ag IgG was used to bind the vWf-Ag of the test plasmas to the solid phase and to prepare the enzyme-antibody conjugate in ELISA. Normal rat serum was used as blocking agent. The standard curve was linear (r2> 0.98) and reproducible after loot-log transformation. The interassay coefficient of variation (CV) in test plasmas with various vWf-Ag concentrations was never greater than 7.7%. Assayed values in dilutions of pooled normal canine plasma added to canine vWf-Agdeficient plasma were linear between 0 and 100% ( r 2= 0.99 ) and indicated excellent analytical recovery of vWf-Ag. In 18 dogs with various internal diseases, including yon Willebrand's disease and haemophilia A, the coefficient of correlation between the results of the ELISA and those of electroimmunodiffusion (EID) was 0.93.
ABBREVIATIONS EID, electroimmunodiffusion; ELISA, enzyme-linked immunosorbent assay; FWP, fixed washed platelets; PPP, platelet poor plasma; vWD, yon Willebrand'sdisease; vWf-Ag, yon Willebrand factor antigen. Correspondence to: Dr. R.J. Slappendel, University Clinic for Companion Animals, Postbox 80.154, 3508 TD Utrecht, Netherlands.
© 1992 Elsevier Science Publishers B.V. All rights reserved 0165-2427/92/$05.00
146
R.J. SLAPPENDEL ET AL.
INTRODUCTION
Von Willebrand's disease (vWD) is a congenital bleeding disorder in which quantitative a n d / o r qualitative abnormalities of the factor VIII/yon Willebrand factor complex (FVIII-vWf) impair hemostasis. The von Willebrand factor moiety of the complex, called factor VIII-related antigen or more recently von Willebrand factor antigen (vWf-Ag) (Marder et al., 1986), can be precipitated with specific antisera. The occurrence of vWD in dogs has been well documented and a high incidence has been reported in some breeds (Johnson et al., 1988 ). Dogs with vWD provide useful models for the study of the FVIII-vWf complex (Bouma et al., 1976; Benson et al., 198 l; Brinkhouse et al., 1985). At present, the main tool to diagnose vWD in dogs is measurement of plasma vWf-Ag by rocket electroimmunodiffusion (EID) in agarose (Johnson et al., 1988). In our experience, the performance and interpretation of the rocket EID for vWf can be troublesome. Problems include low sensitivity, indistinct precipitation lines and failure to discriminate among extremely low and extremely high plasma vWf-Ag concentrations. The latter problem can be overcome by routinely performing tests in serial dilutions of test plasma in normal plasma, which is expensive, or by repeating a test in that way whenever readings are ambivalent, which is time-consuming. In addition, the rocket EID needs relatively large amounts of antiserum and is less appropriate for assaying large series. The technique applied in this paper is a sandwich method of an enzymelinked immunosorbent assay (ELISA) in which a specific IgG fraction of a rabbit anti-canine vWf serum is used to bind the vWf-Ag of the test plasma to the solid phase of microtitre wells. Soluble enzyme-linked antibody was used for labelling. The advantages of this technique are its relative technical simplicity, low cost, fair sensitivity, accuracy, and suitability for the screening of large series of test plasmas. MATERIALS AND METHODS
Plasma samples Venous blood samples were collected in plastic syringes and immediately anticoagulated with one part of sodium citrate (38 g 1-1 ) to nine parts of blood. The blood was centrifuged ( 15 min, 3000 × g, 18 °C ) and the plateletpoor plasma (PPP) was harvested, stored in aliquots of 0.5 ml in capped plastic vials at - 7 0 ° C, and thawed at 37 °C shortly before testing.
ELISA FOR vWf-Ag IN CANINE PLASMA
147
Animals Fifty-one apparently healthy, privately owned adult male and female dogs of various breeds and mixed breeding were used to establish the normal range and to constitute the normal plasma pool, which was stored in aliquots of 50 #l at - 70°C and used as the reference standard. Dobermann pinschers, German shepherds, Golden retrievers, terriers, and miniature Schnauzers, being breeds in which a high prevalence of vWD has been reported in the USA (Johnson et al., 1988 ), were not included. Plasma was also obtained from eight dogs with hemophilia A and three dogs with vWD. Hemophilia was diagnosed by a previously described method (Slappendel, 1975 ) and vWD was diagnosed by defective ristocetin-induced agglutination of human fixed washed platelets (FWP) in a semi-quantitative macroscopic tilt tube test (Rosborough et al., 1980) and by a lack of detectable amounts ofvWf-Ag in a rocket EID (Laurell, 1966 ) in agarose (Seakem, Bausch and Lomb, Rochester, NY, USA) with a commercially available goat anti-human vWf-Ag serum (Dakopats, Copenhagen, Denmark). The vWf-Ag-deficientplasma used for recovery studies was obtained from a dog which had vWD and severe hemorrhagic diathesis. This plasma did not precipitate in rocket EID with a rabbit anti-canine vWf-Ag serum ('Utrecht antiserum', generously provided by Dr. B.N. Bouma, Academic Hospital, Utrecht, Netherlands) (Benson et al., 1975), nor did it support ristocetininduced agglutination of FWP. The correlation between results of the ELISA and rocket EID assays was studied in the plasmas from 17 dogs with various internal diseases, including four dogs with hemophilia A and three dogs with vWD. The same frozen plasma standard from a pool of normal dogs was used in both methods.
Purification of v Wf-Ag VWf-Ag was purified from a cryoprecipitate from pooled plasma of eight fasted dogs by a modification of the method of van Mourik and Mochtar (1970). The rapidly frozen plasma was thawed at 4°C and the cryoprecipitate was isolated by centrifugation for 30 min at 10 000 × g (4 ° C ). The pellet was redissolved in 0.017 M barbital buffer (pH 7.0) containing 0.125 M NaC1 (pH 7.0), 0.01 M benzamidine (Sigma, St. Louis, MO, USA), and 0.01 M epsilon-amino-caproic acid (EACA; Sigma, St. Louis, MO, USA) and this mixture was centrifuged for 5 min at 1000×g to remove undissolved materials. The vWf-Ag was isolated from the clear supernatant by gel filtration on a (5 cm×50 cm) Sepharose C1-4B column (Pharmacia Fine Chemicals, Uppsala, Sweden) in barbital buffer containing 0.001 M benzamidine and 0.001 M EACA. The vWf-Ag was precipitated from the void volume fractions by dialysis against 1.9 M ammonium sulphate (pH 7.0) for 16 h at 4°C and
148
R.J. SLAPPENDEL ET AL.
stored at - 7 0 ° C until use. The protein content of the purified vWf-Ag solution was 0.03 Mg m l - l
Preparation of antiserum and IgG fraction Purified canine vWf-Ag emulsified in Freund's complete adjuvant was injected intramuscularly into rabbits. Then, a mixture of the antigen and Freund's incomplete adjuvant was injected at 4-week intervals until, after 6 months, antibody titres were sufficiently high. The antiserum was purified by preabsorption with the vWf-Ag deficient low molecular fraction ( < 600 kDa) of canine cryoprecipitate, which was covalently bound to Sepharose CL-4B using the CNBr method according to the manufacturer's instructions (Pharmacia Fine Chemicals, Uppsala, Sweden). Its IgG fraction was isolated by affinity chromatography on a protein-A Sepharose-CL-4B column (Pharmacia Fine Chemicals), eluted with 0.1 M glycine HC1 in 0.5 M NaC1 buffer, pH 2.5, neutralized with 0.5 M Tris buffer, and stored in capped plastic cups at - 70 ° C until used.
Immunoconjugate Aliquots of the purified IgG fraction were conjugated with horseradish peroxidase ( H R P ) (Sigma, St. Louis, MO, USA) (Nakana and Kawaoi, 1974) and stored at - 7 0 ° C until used.
ELISA procedure One hundred microliltres of the anti-canine vWf-Ag IgG in PBS ( 1% v/v ) were added to the wells of polystyrene microtitre plates (ELISA microtitre plates with high affinity (Art. 655061 ), Greiner, Frickenhausen, Germany). The plates were covered, incubated overnight at 37 ° C, thoroughly washed in running tap water to remove unadsorbed materials, and shaken dry. Remaining binding sites were blocked by the addition of 100/tl of rat serum ( 1% v / v ) in PBS to each well. After incubation for at least 10 min at room temperature, 100 ~tl of each sample to be tested was added in various dilutions (vide infra) to the rat serum. The plates were covered, incubated for 1 h at 37 oC, washed in running tap water to remove unbound material, and shaken dry. Then, 100 ~tl of the antibody-enzyme conjugate, diluted 1/300 in PBS, was added to each well. Plates were incubated for 1 h at 37°C, washed to remove unbound conjugate, and dried again. To measure peroxidase activity, 100/~1 of enzyme substrate was added to each well. The ELISA substrate was prepared just before use by adding 34/tl hydrogen peroxide, 35% ( v / v ) , to freshly prepared solution (pH 5.95) of 5-amino-2-hydroxybenzoic acid (Merck, Darmstadt, Germany ) (0.6 g 1-1 ) in distilled water.
149
ELISA FOR vWf-Ag IN CANINE PLASMA
Following 45 min incubation at room temperature, the reaction was stopped by the addition of 50 al NaOH (0.1 N), and quantitated within 30 min by measurement of the optical density at 405 n m (OD 405 ) with a spectrophotometer (Titertek multiscan, Flow Laboratories, Ayrshire, U K ) . Test and reference plasmas were assayed in duplicate at dilutions of 1:4, 1:8, and 1:16 in PBS. Non-specific absorption was determined by incubation of the antibody-enzyme conjugate in wells that were not preincubated with canine plasma. A standard curve was obtained by logit-log transformation of the absorbance at 405 n m vs. the dilution of the reference pool. U n k n o w n concentrations were extrapolated from a least-squares regression line of the logitlog curve and expressed as a percentage of the undiluted reference pool. RESULTS A standard curve for the determination of vWf-Ag was constructed by serial dilution of a pooled reference plasma. Fig. 1 shows a typical standard curve. After logit transformation (Fig. 2 ), the curve of the absorbance vs. the log of the dilution of the reference plasma was consistently linear from 1.6 to 100%. For 11 standard curves, each one constructed on a different day, the regression coefficients varied from 0.997 to 0.980 (mean 0.990). The coefficient of variation of 12 replicates of a normal plasma (mean vWfA g = 97.6%) in the dilutions routinely used for the construction of the reference curve ( 1 : 4, 1 : 8 and 1 : 16) was 3.9, 5.7 and 3.9%. The detection limit of the assay, calculated as the vWf-Ag concentration corresponding on the ref-
....---,
tD O
O
0
. . . . . . . 1
i 10
, ,
. . . . .
i 1QO
vWf-Ag (%)
Fig. 1. Dose-responsecurve for the ELISAofvon Willebrandfactor antigen (vWf-Ag) in pooled normal canine plasma. Depicted is the percentageof pooled plasma in PBS vs. the absorbance at 405 nm. The absorbance in undiluted plasma was arbitrarily defined as 100%vWf-Ag.
150
R.J. SLAPPENDEL ET AL. 1.50
0.75
0.00
-0.75
- 1.50
. . . . . . .
i
,
10 vWf-Ag
.
.
.
.
.
.
I
100 (%)
Fig. 2. Standard reference curve obtained from the dose-response curve (Fig. 1) after logit transformation of the OD 405 nm readings. erence curves to an absorbance at two standard deviations above the mean blank, measured in triplicate on 11 days, was 1.6%. Intra-assay variation was 10.7, 8.1, 4.6, 2.4 and 12.9% for 15 replications on a single plate in five test plasmas representing high (192%), normal (105%), borderline normal (52%), low (19%), and very low (6%) vWf-Ag concentrations. Interassay variation was 5.8, 7.7, 1.6, 4.2 and 7.5% in five test plasmas assayed in duplicate on 3 different days representing high (242%), high normal (130%), normal 97% ), borderline normal (52%), and very low (6%) vWf-Ag. Analytical recovery studied with different proportions of pooled normal plasma in vWf-Ag-deficient plasma (vWf-Ag less than 1.6%) was excellent for concentrations between 1.'6 and 100% when assayed at 1:4 to 1 : 16 dilutions in PBS (Fig. 2). At lower dilutions, recovery was incomplete and not linear. The correlation of results obtained by ELISA and Laurell EID in the plasmas of 17 dogs with various vWf-Ag concentrations is shown in Fig. 3. The vWf-Ag concentrations obtained in 51 clinically normal dogs ranged from 46 to 249% (mean 94.9%, SD 36.9%, median 92%). The 2.5 and 97.5 percentile confidence limits ( P < 0.05 ) calculated by a distribution-free method for use as clinical reference values (Riimke and Bezemer, 1972) were 48 and 172%, in close agreement with normal values reported by others (Rosborough et al., 1980; Johnson et al., 1988 ). The vWf-Ag concentrations estimated with ELISA in the plasmas of eight German shepherd dogs with acute or subacute bleeding problems associated with hemophilia A ranged from 84 to 535% (mean 202% ). High plasma vWf-
ELISAFORvWf-AgIN CANINEPLASMA
151
120
100
J
4
8O
6O
I
>
40 20
0
i
I
I
!
I
20
40
60
80
100
% Pool in v W f - A e - d e f i c i e n t
plasma
Fig. 3. Analytical recovery of von Willebrand factor antigen (vWf-Ag) in serial dilutions of pooled normal plasma in vWf-Ag-deficient plasma from a dog with severe yon Willebrand's disease.
Ag concentrations associated with hemophilia have been reported before in man (Denson, 1973)and dogs (Benson et al., 1975). DISCUSSION
ELISA techniques for vWf-Ag assay have been used for more than a decade in human medicine (Bartlett et al., 1976; Mazurier et al., 1977; Ness and Perkins, 1979; Yorde et al., 1979; Ingerslev, 1987) but have not been described yet for the canine species. Antisera against human vWf-Ag can be used for the detection of canine vWf-Ag, at least in the Laurell EID, but species-specific antiserum is mandatory for accurate quantitation, especially at low vWf-Ag concentrations (Benson et al., 1983 ). A species-specific problem associated with the isolation of canine vWf-Ag and the subsequent production of the antiserum is physiologic post-prandial hyperlipaemia of the donor dogs, as this may result in a low yield of cryoprecipitate and generation of heterospecific antibodies (Benson and Dodds, 1982). This was avoided by the use of plasmas from fasted dogs and by ultracentrifugation of the plasma. Another species-related problem consists of the aspecific heterologous immunoreactivity of canine sera for albumin and other proteins in the sera of many species. The reactivity with the proteins in the rat serum used as blocking agent in the ELISA was low, as indicated by the low OD 405 of the blanks, corresponding to 1.6% vWf-Ag (mean + 2 SD ) on the reference curve.
152
R.J. SLAPPENDEL ET AL.
A standard canine vWf-Ag preparation or international reference preparation for the assessment of accuracy is lacking, but our reference plasma consistently showed good linearity and reproducibility at seven serial dilutions from 1.6 to 100% (Fig. 1 ). The high correlation coefficient of the regression line (r2> 0.98 ) allowed reduction to three dilutions for the routine construction of the standard curve, thus saving time and materials. The intra-assay and interassay variation in test plasmas with various concentrations of vWf-Ag were good and compared well with those of canine vWf-Ag assays by the EID technique (Benson et al., 1983). The detection limit of the assay was considerably higher for ELISA (1.6%) than reported for optimal conditions of the EID (9%) (Benson et al., 1983). Results of ELISA and Laurell EID correlated very well over a wide range of plasma vWf-Ag concentrations (Fig. 4) but according to the regression line, values were higher by EID than by ELISA. This is probably due to the inaccuracy of the non-species-specific antiserum used in the EID, which gives inaccurate results at lower vWf-Ag values (Benson et al., 1983 ). It is probably also due to the EID technique itself, which gives artifactual variations in the vWf-Ag levels of plasma from dogs with vWD and hemophilia A, depending on the source of the agarose used (Benson et al., 1975). The high values obtained in hemophilic dogs also confirm that our ELISA measures the vWfpart rather than the F VIII-part of the F VIII-vWfcomplex. In conclusion, the ELISA assay for canine vWf-Ag, which is easy to per300
J
•
O
200 r~ CO
•
0
% o~ .
