Development and analytical validation of an enzyme linked immunosorbent assay for the measurement of canine gastric lipase immunoreactivity in serum

Abstract

The objective of this study was to develop and analytically validate an enzyme linked immunosorbent assay (ELISA) for measurement of canine gastric lipase immunoreactivity (cGLI). A sandwich ELISA was developed using canine gastric lipase (cGL) purified from canine stomachs and polyclonal antibodies directed against cGL, raised in rabbits and purified by affinity chromatography. The assay was validated by determination of sensitivity, working range, linearity, accuracy, precision, reproducibility, and the upper limit of the control range by determining the 97.5th percentile of serum cGLI concentration in 74 healthy canines. Sensitivity and working range in serum were 200 ng/L and 200 to 39 160 ng/L, respectively. Observed to expected ratios for dilutional parallelism for 3 serum samples and 3 dilutions ranged from 86.1% to 244.2% (mean ± standard deviation [s]; 125.4% ± 48.2%). Observed to expected ratios for spiking recoveries for 3 serum samples and 6 spiking concentrations ranged from 66.4% to 152.5% (mean ± s; 104.5% ± 22.9%). Intra-assay and interassay variabilities for 3 different serum samples were 25.5%, 9.4%, and 13.4% and 26.0%, 17.2%, and 14.4%, respectively. The upper limit of the control range for serum cGLI was 662 ng/L. We concluded that the ELISA for cGLI described here is highly sensitive and shows a wide working range. However, the validation characteristics for this assay are suboptimal and below values of approximately 2.000 ng/L the assay is more semiquantitative in nature. Despite its limitations, whether this assay is useful for the diagnosis of canine gastric disorders remains to be determined.

Introduction

Gastric disease is common in canines and the most common manisfestation is gastritis (1,2). Currently, a definitive diagnosis of gastritis relies on the use of gastroscopy and biopsy specimens (2). However, gastroscopy is fairly invasive and carries anesthetic risks. Thus, a non-invasive or minimally-invasive diagnostic test for canine gastritis is desirable. Gastric lipase and pepsinogen are both synthesized and secreted by gastric glands (3,4). In human beings the measurement of different isoforms of pepsinogens has been shown to be useful in diagnosing gastritis (5,6). Therefore, it would be reasonable to evaluate both gastric lipase and pepsinogen as potential markers for gastritis in the canine. Recently, canine pepsinogens have been purified and an assay for measurement of serum pepsinogen A in canine serum has been developed and validated (3). Unfortunately, serum pepsinogen A concentration did not appear to be useful as a marker for canine gastritis (7). The hope that gastric lipase immunoreactivity (GLI) might be of value in the diagnosis of gastric disorders in human patients was dampened when a French group was unable to detect GLI in serum from healthy human beings (personal communication; F. Carrière, LLE, CNRS, Marseille, France, 2000). However, it is possible that serum gastric lipase concentrations are higher in canines than they are in human beings, potentially resulting in a clinically useful marker in canines.

Serum lipase activity has been used for the diagnosis of pancreatitis in humans and canines for several decades. However, it has always been recognized that serum lipase activity is not specific for exocrine pancreatic function (8). This is because many different cells in different organs synthesize and secrete lipases (9). While these lipases may only share a small portion of their amino acid sequence and thus do not cross-react immunologically, they all share the property of lipolytic activity (9). This is the reason why lipases from different tissues may all be detected by a catalytic assay leading to the poor specificity of these assays for pancreatic function and disease. In contrast, immunoassays, such as radioimmunoassays and enzyme linked immunosorbent assays (ELISA) are specific for the surface structure of the analyte to be measured and thus immunoassays for a lipase of a specific tissular origin are expected to be specific for that particular lipase.

Therefore, the goal of this project was to develop and analytically validate an ELISA for the measurement of canine gastric lipase immunoreactivity (cGLI) as a prelude for evaluation of serum cGLI as a marker for gastritis in the canine. In order to accomplish this goal canine gastric lipase (cGL) was purified from canine stomachs, antiserum directed against cGL was raised in rabbits, and a sandwich-ELISA for cGLI was set up and validated.

Materials and methods

Purification of canine gastric lipase

Canine gastric lipase (cGL) was purified based on a protocol initially described by Carrière et al (4).

Lipase assay

Lipolytic activity was measured using tributyrin as a substrate. The assay was performed at pH 5.5, which is the optimal pH reported for cGL (4). A volume of 14.5 mL of reaction buffer (2 mM sodium taurodeoxycholate [Sigma Chemical, St. Louis, Missouri, USA], 150 mM NaCl [Sigma Chemical], 0.1 g/L bovine serum albumin [Sigma Chemical], pH 5.5) and 0.5 mL of tributyrin (Sigma Chemical), as a substrate, were placed into the heated reaction vessel of a pH stat titration system (Brinkmann Instruments, Westbury, New York, USA) at 37°C. The pH was readjusted to 5.5 and 20 μL of the sample solution was added to the reaction mixture. The pH was kept constant at 5.5 by the addition of 100 mM NaOH (Sigma Chemical) to titrate the free fatty acids released through lipid hydrolysis. After a waiting period of 60 s, the rate of use of NaOH per minute was determined for a time period of 2 min. Lipase activity was estimated from the amount of NaOH (in μmol/min) that was added in order to titrate an equal amount of free fatty acids released (1 μmol/min of fatty acids released equals 1 international lipase unit; U). Since butyric acid is not completely ionized at pH 5.5, the results obtained by pH titration were adjusted by multiplication with factor 1.12. This procedure was used to quantify lipolytic activity during the entire purification process.

Preparation of gastric soak

Stomachs were collected from several random source research canines euthanized for unrelated projects. The stomachs were removed whole and stored frozen at −20°C until purification. The health status of the canines used was unknown but the stomachs appeared grossly normal. Histopathology was not performed and gastric disease could not be excluded conclusively in these canines. A stomach was removed from the freezer and lightly thawed at room temperature until it could be cut open along the major curvature. Debris was removed manually and gastric tissue was cut into cubes. Gastric lipase was extracted from the canine stomachs by soaking approximately 60 g of canine stomach in 120 mL 20 mM glycine (Sigma Chemical), pH 2.5, for approximately 30 min at room temperature. During this time the pH was kept constant at 2.5 by titration with 1 M HCl (Sigma Chemical). The gastric soak was then centrifuged (Eppendorf 5810R centrifuge; Brinkmann Instruments) at 10 000 × g and 4°C for 20 min. The supernatant was filtered through a paper towel followed by filtration through a 0.45 μm pore-size filter (VWR Scientific, West Chester, Pennsylvania, USA).

Cation exchange chromatography

The filtered gastric soak was further purified by cation-exchange chromatography on a cation exchange resin (Source S; Amersham Pharmacia Biotech, Piscataway, New York, USA) packed into a column (Amersham Pharmacia Biotech) with a diameter of 10 mm and a column bed height of 100 mm on a fast liquid pressure chromatography system (Äkta purifier; Amersham Pharmacia Biotech). The column was equilibrated with 20 mM sodium acetate (Sigma Chemical), pH 4.0, and the entire sample volume was applied at a flow rate of 4 mL/min. Fractions of 4 mL each were collected. Then the column was washed with 20 mM sodium acetate, pH 4.0, until the reading of the UV detector at a wavelength of 280 nm had returned to the baseline. A linear salt gradient from 0 to 500 mM NaCl in 20 mM sodium acetate, pH 4.0 was applied over 1 h. All fractions that showed absorbance at 280 nm were evaluated for lipolytic activity, as discussed previously.

Anion-exchange chromatography

All fractions containing lipolytic activity were pooled and concentrated using a concentrating device (Centricon 10k; Amicon, Beverly, Massachusetts, USA). This was followed by buffer exchange to 10 mM Tris-HCl (Sigma Chemical), pH 8.0. The material was further purified by anionexchange chromatography on a prepacked column (Mono Q; Amersham Pharmacia Biotech). The column was equilibrated with 10 mM Tris-HCl, pH 8.0, at a flow rate of 1 mL/min and the partially purified product was applied. Fractions of 4 mL each were collected. The column was washed with 10 mM Tris-HCl, pH 8.0, until the reading of the UV detector at 280 nm had returned to the baseline reading and a linear salt gradient from 0 to 400 mM NaCl in 10 mM Tris-HCl, pH 8.0, was applied over 1 h. Lipolytic activity was measured in all fractions that showed absorbance at 280 nm.

Size-exclusion chromatography

Once again, all fractions containing lipolytic activity were pooled; concentrated using a concentrating device; and the buffer was changed to 10 mM Tris-HCl, 150 mM NaCl, pH 8.0. Canine gastric lipase was further purified by sizeexclusion chromatography on a prepacked column (Hi Prep Sephacryl S-100HR; Amersham Pharmacia Biotech) at a flow rate of 1 mL/min using 10 mM Tris-HCl, 150 mM NaCl, pH 8.0, as a mobile phase. Again, fractions of 4 mL each were collected and all fractions containing lipolytic activity were pooled and the buffer was changed to phosphate buffered saline solution ([PBSS] 100 mM sodium phosphate, 150 mM NaCl, pH 7.2; BupHTM dry blend buffers; Pierce Chemical Company, Rockford, Illinois, USA), adjusted to an absorbance of 1.58 at 280 nm, and frozen at −80°C until further use.

Partial characterization

The molecular mass of purified cGL was estimated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) following manufacturer’s instructions (X cell II mini cell and powerease 500; Novex, San Diego, California, USA). The gel was stained with a Coomassie blue type stain (GelCode Blue; Pierce Chemical Company). The isoelectric point was estimated by isoelectric focusing following manufacturer’s instructions and the gel was stained with Coomassie blue following the manufacturer’s instructions. The N-terminal amino acid sequence for the first 25 amino acid residues was determined by an outside service laboratory using the Edman degradation procedure and an automated amino acid sequence analyzer.

Production of antiserum

Anti-cGL antiserum production

Two New Zealand white rabbits were used for the production of antiserum against canine gastric lipase. Because an animal care and use protocol for antiserum production was not in place at the time the antiserum was needed the antiserum production for cGL was out-sourced to another laboratory (Department of Biochemistry, Purdue University, West Lafayette, Indiana, USA). The exact vaccination protocol was not revealed by this laboratory. Both rabbits were initially vaccinated with cGL emulsified with complete Freund’s adjuvant (Sigma Chemical). Reinoculations with cGL emulsified with incomplete Freund’s adjuvant (Sigma Chemical) were performed at 21, 37, 44, 51, 62, 82, 103, 120, 140, and 152 d after the initial inoculation. Blood samples were collected at 51, 61, 71, 85, 91, 111, 120, 127, 134, 140, 148, and 159 d after the initial inoculation. Serum was separated from the blood samples and the antiserum was evaluated using a radioimmunoassay. Briefly, cGLwas iodinated with ¹²⁵I using the chloramine T method (10). A mini stir bar (8 mm × 1.5 mm; VWR Scientific) was placed in a polypropylene test tube (75 mm × 12 mm; Falcon polypropylene tubes; VWR Scientific) and 10 μL of 250 mM sodium phosphate (Sigma Chemical), pH 7.5, was added to the tube. The tube was suspended over a stir plate (VWR Scientific) and 10 μL ¹²⁵I (NaI, 0.1 mCi/μL at time of production; NEN Life Sciences Products, Boston, Massachusetts, USA) was added using a syringe (Hamilton syringe; VWR Scientific). Then 10 μg cGL in 10 μL PBSS, pH 7.2; 10 μL of 2 mg/mL chloramine T (Sigma Chemical) in 50 mM sodium phosphate, pH 7.5; 100 μL of 400 mg/mL sodium metabisulfite (Sigma Chemical) in 50 mM sodium phosphate, pH 7.5; and 860 μL of 2 mg/mL potassium iodide (Sigma Chemical) in 50 mM sodium phosphate, pH 7.5, were added to the tube in rapid succession. The protein fraction was separated from the free iodide by size-exclusion chromatography on a disposable size-exclusion chromatography column (PD-10; Amersham Pharmacia Biotech), following the manufacturer’s instructions. Radioimmunoassay buffer ([RIAB] 50 mM sodium phosphate, pH 7.5, with 5 g/L bovine serum albumin and 0.2 g/L sodium azide; Sigma Chemical) was used as the mobile phase and fractions of each 1 mL collected. The protein fraction with the highest radioactivity was collected and adjusted to approximately 40 000 counts/100 μL per minute. This working tracer was kept at 4°C until further use. The antiserum to be tested was diluted to 1 in 500 in the RIAB and a 1 in 2 serial dilution was prepared. A volume of 100 μL of each dilution step was placed into each of 2 polypropylene tubes. Also, 100 μL of RIAB and 100 μL of tracer were added to each of the tubes. Four additional polypropylene tubes did not receive any antiserum. The first 2 tubes (labeled TC for total counts) only received 100 μL of tracer each and the other 2 (labeled NB for nonspecific binding) received 100 μL of tracer and 200 μL of RIAB each. All tubes were thoroughly mixed using a vortex and incubated at room temperature for 2 h. Except for the first 2 tubes, 100 μL of carrier serum (1 mL of normal rabbit serum and 99 mL of RIAB) were added to each tube, followed by 1 mL of a commercially available precipitating solution (N6; Diagnostic Products Corporation, Los Angeles, California, USA). The tubes were mixed and centrifuged (GS-6R; Beckman Coulter, Anaheim, California, USA) at 3000 × g and 4°C for 20 min. The supernatant was carefully decanted and the tubes were counted for 1 min in a gamma counter (Riastar; Packard Instrument Company, Meriden, Connecticut, USA). Results for each antiserum dilution were expressed as percent binding = (mean count — mean NB)/(mean TC — mean NB) × 100.

Affinity chromatography of antiserum

An affinity chromatography column (Hi Trap; Amersham Pharmacia Biotech) for the purification of monospecific polyclonal antibodies against cGL was prepared following the manufacturer’s instructions. Antisera were prepared for purification by lipoprotein precipitation. Ten milliliters of anti-serum were added to 10 mL 1M CaCl₂ (Sigma Chemical) and 400 μL 10% dextran sulfate (Sigma Chemical), mixed for 20 min using a rocking plate (VWR Scientific) at 4°C, and centrifuged at 10 000 × g and 4°C for 30 min. The supernatant was filtered through a paper towel and the buffer changed using a disposable size-exclusion column (PD-10). The column was prepared by washing with 20 mL of 75 mM Tris-HCl, 150 mM NaCl, pH 8.0. Then 2.5 mL of the serum supernatant was applied and the eluent discarded. The column was washed with 3.5 mL of 75 mM Tris-HCl, 150 mM NaCl, pH 8.0, and the eluent collected. For further purification of the antiserum, the ligated column (HiTrap) was washed with 75 mM Tris-HCl, 150 mM NaCl, pH 8.0, and the collected eluent from the column (PD-10) was applied. The column was, once again, washed with 75 mM Tris-HCl, 150 mM NaCl, pH 8.0, until the absorbance at a wavelength of 280 nm had returned to the baseline value. Then the column was washed with 100 mM glycine, 500 mM NaCl, pH 3.0, until the absorbance had once again reached the baseline value. The eluent was collected into tubes containing 200 μL of 1 M Tris-HCl, pH 8.0, for each mL of eluent. Fractions containing the initial peak after buffer exchange were collected and concentrated using a concentrating device. The buffer was changed to PBSS, pH 7.2, the antibody adjusted to a concentration of approximately 1 mg/mL, and frozen at −20°C until further use.

Biotinylation

For use as a reporter antibody a portion of the purified monospecific polyclonal antibody directed against cGL was biotinylated. A 20- to 50-fold molar excess of biotin (EZ-Link Sulfo-NHS-LC-Biotin; Pierce Chemical Company) freshly diluted in deionized water to a concentration of 10 mg/mL was added to the antibody in PBSS, pH 7.2, and incubated for a variable time at room temperature in a 10k dialysis cassette (Slide-A-Lyzer; Pierce Chemical Company). Then the cassette was dialyzed against 4 hourly changes of 800 mL of PBSS, pH 7.2, at 4°C. The biotinylation coefficient was evaluated by a 2-(4′-hydroxyazobenzene) benzoic acid (HABA) avidin assay kit (HABA avidin assay kit; Pierce Chemical Company) according to the manufacturer’s instructions. Depending on the biotinylation coefficient the biotinylation procedure was repeated adjusting the molar excess of biotin and the incubation time until a biotinylation coefficient of approximately 3 to 4 was reached. The biotinylated antibody was adjusted to a solution of approximately 1 mg/mL and frozen at −20°C until further use.

Development and validation of cGLI ELISA

ELISA development

Ninety-six-well flat bottom ELISA plates (Combiplate8; Labsystems Oy, Atlsinki, Finland) were coated with 200 ng/well of affinity purified monospecific anti-cGL-antibodies (cGL-AB) in 100 μL/well carbonate-bicarbonate buffer (BupHTM dry blend buffers; Pierce Chemical Company), pH 9.4. The plate was incubated for 1 h at 37°C under constant shaking and washed 4 times with 200 μL/well PBSS, pH 7.2. Nonspecific binding sites were blocked with 200 μL/well of a milk-free blocking solution (Superblock; Pierce Chemical Company). Again, the plate was incubated for 1 h at 37°C under constant shaking and washed 4 times, as described above. Phosphate-buffered saline solution, pH 7.2; with 1% bovine serum albumin (BSA); and 0.05% polyoxyethylenesorbitan monolaurate (buffer A) (Tween 20; Sigma Chemical) was used for blanks. Standards were prepared by a 1 in 2 serial dilution of a solution of 5000 ng/L cGL in buffer A. Standards of 5000.0, 2500.0, 1250.0, 625.0, 312.5, 156.25, 78.13, 39.06, 19.53, and 9.77 ng/L were produced and frozen in aliquots of 300 μL at −20°C. Standards were thawed immediately prior to loading. A volume of 100 μL/well of blank or standard solution was used. All samples (control samples and unknown samples) were prepared as a 1 in 10 dilution with buffer A and the wells were loaded with 100 μL/well of diluted sample. The plate was incubated for 2 h at 37°C without shaking and washed 4 times, as previously described. Then the plates were incubated with the secondary antibody solution, containing biotinylated antibodies against cGL, at 200 ng/well in 100 μL/well of buffer A. After incubation for 1 h at 37°C under constant shaking the plate was once again washed 4 times and incubated with 100 μL/well horseradish peroxidase-conjugated streptavidin solution (200 ng/mL; ImmunoPure Streptavidin HRP conjugated; Pierce Chemical Company) in buffer A. The plate was incubated for another hour at 37°C under constant shaking and washed 4 times. Plates were developed for 10 min with 100 μL/well of a 3,3′ 5,5′-tetramethylbenzidine dihydrochloride (TMB) substrate solution (ImmunoPure TMB substrate kit; Pierce Chemical Company) prepared immediately prior to use, following the manufacturer ’s instructions. The reaction was stopped by adding 100 μL/well 4 M acetic acid (EM Science, Gibbstown, New Jersey, USA), and 0.5 M sulfuric acid (EM Science). Plates were read with an ELISA plate reader (UVMAX kinetic microplate reader; Molecular Devices, Sunnyvale, California, USA) at a wavelength of 450 nm. Standard curves were calculated using a 4-parameter curve fit: y = (A – D)/(1 + [x/C]^B) + D, where D is the y-value corresponding to the asymptote at high values on the x-axis, A the y-value corresponding to the asymptote at low values on the x-axis, C the x-value corresponding to the midpoint between A and D, and B describes how rapidly the curve makes its transition from the asymptotes in the center of the curve; all 4 parameters are calculated using an algorithm based on the Levenberg-Marquardt Method (SOFTMAX PRO software package; Molecular Devices).

Assay validation

The assay was validated by determination of assay sensitivity, working range, dilutional parallelism, spiking recovery, intra-assay variability, and interassay variability. Assay sensitivity was determined by loading 10 sets of blanks as unknown samples and determining the result of the mean, plus 3 standard deviations of the 10 mean results, on the standard curve. The lower limit of the working range was defined by the following higher integer above the sensitivity that was determined. The upper limit of the working range was determined by the absorbance value, which equals the mean maximum absorbance minus 3 times the standard deviation of the maximum absorbance. The maximum absorbance of the assay was determined by measuring 10 duplicate wells containing approximately 50 000 ng/L cGL. Three serum samples were used for validation of the assay, 2 of which were random samples that were left over from other studies. The 3rd sample had to be mixed from 2 other samples in order to achieve a cGLI concentration between the other 2 samples. To determine dilutional parallelism, each sample was evaluated in dilutions of 1 in 10 as the full strength solution (1 in 1) and dilutions of this solution at 1 in 2, 1 in 4, and 1 in 8. Spiking recovery was determined by adding 0.0 ng/L, 195.3 ng/L, 390.6 ng/L, 781.3 ng/L, 1562.5 ng/L, 3250.0 ng/L, and 6250.0 ng/L to each of the 3 serum samples that had been diluted 1 in 10. Intra-assay variability was determined by evaluating the 3 1 in 10 diluted serum samples 10 times within the same assay run. Interassay variability was determined by evaluating the 3 1 in 10 diluted serum samples in 10 consecutive assay runs.

Serum cGLI concentration was measured in 74 healthy canines (45 research canines and 29 canine pets). Median serum cGLI concentrations were compared between canine pets and research canines by a two-sided Mann Whitney test. A P-value < 0.05 was considered statistically significant. The upper limit of the control range for serum cGLI concentration was determined by calculating the upper 97.5th percentile in these 74 healthy canines.

Results

Canine gastric lipase was successfully purified from canine stomachs with a yield of approximately 60% (Table I). The SDS-PAGE showed a single band of the pure material (Figure 1). The molecular mass of cGL was estimated from the SDS-PAGE to be approximately 49 kDa (Figure 1). The isoelectric point of cGL was estimated to be between 6.8. and 6.9 (results not shown). N-terminal amino acid sequencing of the first 25 amino acid residues showed the sequence Leu-Phe-Gly-Lys-Leu-His-Pro-Thr-Asn-Pro-Glu-Val-Thr-Met-?-Ile-Ser-Gln-Met-Ile-Thr-Tyr-Trp-Gly-Tyr (? = amino acid residue could not be determined).

Table I.

Sequential purification of canine gastric lipase (cGL) from canine stomach

Purification stage	Volume (mL)	A280	Protein contenta (mg)	Specific lipolytic activity (U/mg)	Total lipolytic activity (U)	Yield (%)
Gastric soak	120.0	N/Ab	N/Ab	N/Ac	12 720c	100
After cation-exchange chromatography	22.5	1.546	22.0	453	9966	78.3
After anion-exchange chromatography	15.5	1.326	13.0	625	8125	63.9
After size-exclusion chromatography	12.0	1.424	10.8	715	7722	60.7

^aProtein content was estimated by assuming a specific absorbance of A₂₈₀ = 1.58 for canine gastric lipase

^bAbsorbance of gastric soak was not determined because this material was slightly turbid

^cLipolytic activity was determined on a per volume basis

Figure 1.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDSPAGE) of canine gastric lipase (cGL) at different purification stages. The gel was stained with a Coomassie blue type stain. Lanes: 1: marker (molecular mass: 250, 148, 60, 30, 22, and 17 kDa); 2: empty; 3: gastric soak; 4: partially purified cGL after cation-exchange chromatography; 5: partially purified cGL after anion-exchange chromatography; and 6: purified canine gastric lipase; molecular mass of standard proteins and cGL are shown in lane 2.

Both of the rabbits inoculated with cGL showed antibody titers directed against cGL. Monospecific antibodies against cGL were successfully purified by affinity chromatography. The specific yield for antibody purification was 0.33 mg antibodies against cGL per 1 mL rabbit antiserum. A total of 6.3 mg of antibodies against cGL was purified, of which 2.9 mg was biotinylated. Use of a 20-fold molar excess of biotin and incubation for 80 min at room temperature led to a biotinylation coefficient of 0.64. Biotinylation was continued with a 50-fold molar excess of biotin and 90 min of incubation at room temperature leading to a final biotinylation coefficient of 3.45.

Figure 2 shows a typical standard curve for an ELISA for cGLI. The sensitivity of the assay was 1.97 ng/L resulting in a lower limit of the working range of 2 ng/L. However, the lowest standard of 9.77 ng/L was not routinely detectable in the assay, but the second lowest standard of 19.53 ng/L was. Therefore, the practical sensitivity of the assay was set at 20 ng/L. The maximum was determined to be 3916.2 ng/L leading to an upper limit of the working range of 3916 ng/L. Thus the working range for this assay is 20 to 3916 ng/L or 200 to 39 160 ng/L cGLI in serum when taking into account the dilution of serum samples of 1 in 10. Observed to expected ratios for dilutional parallelism ranged from 86.1% to 244.2% (mean ± s; 125.4% ± 48.2%; Table II). Observed to expected ratios for spiking recovery of 3 serum samples ranged from 66.4 to 152.5% (mean ± s; 104.5% ± 22.9%; Table III). Coefficients of variation for intra-assay variability of 3 serum samples were 25.5%, 9.4%, and 13.4% (Table IV). Coefficients of variation for interassay variability of 3 serum samples were 26.0%, 17.2%, and 14.4% (Table IV).

Figure 2.

A typical standard curve for an enzyme linked immunosorbent assay (ELISA) for canine gastric lipase immunoreactivity (cGLI). The standard curve was calculated using a 4-parameter curve fit (y = (A – D)/(1 + (x/C)B) + D; A = 0.013, B = 1.173, C = 1655.174, D = 3.683).

Table II.

Dilutional parallelism of enzyme linked immunosorbent assay (ELISA) for canine gastric lipase immunoreactivity (cGLI) shown for 3 serum samples at dilutions of 1 in 1, 1 in 2, 1 in 4, and 1 in 8. Observed (O) to expected (E) ratios are given in %

	Sample 1
Dilution	Observed	Expected	% O/E
1 in 1	347
1 in 2	149	173	86.1
1 in 4	119	87	136.8
1 in 8	105	43	244.2

	Sample 2
Dilution	Observed	Expected	% O/E
1 in 1	1892
1 in 2	984	946	104.0
1 in 4	670	473	141.6
1 in 8	259	236	109.7

	Sample 3
Dilution	Observed	Expected	% O/E
1 in 1	5445
1 in 2	3030	2722	111.3
1 in 4	1228	1361	90.2
1 in 8	713	681	104.7

Table III.

Spiking recovery of enzyme linked immunosorbent assay (ELISA) for canine gastric lipase immunoreactivity (cGLI) shown for 3 serum samples. Observed (O) to expected (E) ratios are given in %

	Sample 1
Concentration added (ng/L)	Observed	Expected	% O/E
0.0	575
195.3	753	770	97.8
390.6	943	965	97.7
781.3	1407	1356	103.8
1562.5	2049	2137	95.9
3125.0	2860	3700	77.3
6250.0	4530	6825	66.4

	Sample 2
Concentration added (ng/L)	Observed	Expected	% O/E
0.0	1326
195.3	1579	1521	103.8
390.6	2615	1716	152.4
781.3	2147	2107	101.9
1562.5	2856	2888	98.9
3125.0	3928	4451	88.2
6250.0	6373	7576	84.1

	Sample 3
Concentration added (ng/L)	Observed	Expected	% O/E
0.0	4336
195.3	6911	4532	152.5
390.6	5882	4727	124.4
781.3	5863	5118	114.6
1562.5	7127	5899	120.8
3125.0	8580	7461	115.0
6250.0	8970	10 586	84.7

Table IV.

Intra-assay and interassay variability shown for 3 serum samples

Variability	Sample 1	Sample 2	Sample 3
Intra-assay variability
Number of repeats	10	10	10
Mean (ng/L)	385	2017	5061
Standard deviation (ng/L)	98	189	677
Coefficient of variation (%)	25.5	9.4	13.4
Interassay variability
Number of repeats	10	10	10
Mean (ng/L)	657	2034	5530
Standard deviation (ng/L)	171	349	797
Coefficient of variation (%)	26.0	17.2	14.4

Serum cGLI concentrations were measured in 29 healthy canine pets and 45 healthy research canines (Figure 3). Serum cGLI concentrations were not significantly different between healthy canine pets and healthy research canines (P = 0.1326). The upper limit of the control range for serum cGLI concentration in canine serum, as determined by the 97.5th percentile in these 74 healthy canines, was 662 ng/L.

Figure 3.

Serum canine gastric lipase immunoreactivity (cGLI) concentrations in 74 healthy canines. This figure shows serum cGLI concentrations in 74 healthy canines in the 1st column. In the 2nd column serum cGLI concentrations for 29 of these 74 canines are shown. These canines were kept as pets, while serum cGLI concentrations for 45 of the total 74 canines that were for research purposes are shown in the 3rd column. Many of the canines, in either group, have undetectable (below the practical sensitivity of cGLI in serum of 200 ng/L) serum cGLI concentrations. Therefore, datapoints from these canines can not be separated as single data-points but form a single line at the bottom of the graph.

Discussion

Canine gastric lipase was purified from canine stomach. The purification protocol was highly efficient. Characteristics of cGL purified here were similar to those previously reported (4). The SDS-PAGE, under denaturing and reducing conditions, showed a major band of an approximate molecular mass of 49 kDa. This is identical to the molecular mass of 49 kDa reported by Carrière and coworkers (4). The isoelectric point estimated for cGL from isoelectric focusing was 6.8 to 6.9. This is slightly higher than the 6.3 to 6.5 reported by Carrière et al (4). This difference is probably due to the use of different systems for isoelectric focusing. The finding of a range of isoelectric points is not surprising for a glycoprotein, such as cGL. As previously discussed, similar ranges have been identified for the isoelectric point of other preduodenal lipases, including human and rabbit gastric lipase (11–15). The N-terminal amino acid sequence found was identical to one previously reported (4). As mentioned, the amino acid residue in position 15 could not be identified (4). Reduction and alkylation using β-mercaptoethanol and 4-vinylpyridine did not reveal the identity of this amino acid, suggesting that this residue is unlikely to be cysteine but rather some other amino acid residue. The sequencing of cDNA encoding human and cGL predicts an Asn residue in this position (13,16). This Asn residue could serve as a glycosylation site, which would prevent proper identification during N-terminal amino acid sequencing. Comparing the N-terminal amino acid sequence of the first 25 amino acid residues, cGL shares a sequence homology of 92% with human gastric lipase, 79% with rat lingual lipase, 71% with rabbit gastric lipase, but only 4% with canine pancreatic lipase (11,13–15,17).

The sensitivity of the ELISA for cGLI was determined to be 2 ng/L with a working range from 2 to 3916 ng/L. However, the lowest standard of 9.77 ng/L was not consistently detectable by the assay, while the next higher standard of 19.53 was consistently measurable. Therefore, the practical sensitivity of this assay was set at 20 ng/L and the working range was defined as 20 to 3916 ng/L. Taking into account the dilution of serum samples of 1 in 10, this translates into a practical working range of 200 to 39 160 μg/L for serum samples. Even this adjusted working range shows a wide range and a high sensitivity that is not routinely reached by using ELISAs.

The samples chosen for validation are unusual in that all 3 samples are in the lower area of the working range. These samples were chosen because the preliminary studies had suggested that this range would encompass the area of interest for a clinical assay. A range of observed to expected ratios of 86.1% to 244.2% for dilutional parallelism would be unacceptable for any ELISA intended for clinical use. However, if measurements less than the practical lower range of the working range (assay concentrations of < 20 ng/L or serum concentrations of < 200 ng/L) were excluded, the new range of observed to expected values was 90.2% to 141.6%, which is still suboptimal. This would suggest that the assay described here shows a suboptimal linearity, especially in the lower area of the working range. It is suspected that this lack of linearity is, in part, due to the sensitivity of the assay that makes the assay more vulnerable to external disturbances. However, it should be pointed out that this suboptimal linearity may well be acceptable for clinical use of this assay. The final determination of this issue has to await clinical studies that would determine the difference between serum cGLI concentrations in healthy canines and in canines with gastric disorders. If the difference in cGLI concentration between these 2 groups of canines were very large, then the degree of linearity determined here may be acceptable. The range of observed to expected values for spiking recovery experiments ranged from 66.4% to 152.5% (mean ± s; 104.5% ± 22.9%). This range is wider than observed with other ELISAs. However, scientifically based minimal recommendations for this performance parameter are not available. Overall, the addition of cGL to all 3 samples led to an increase in the cGLI concentration measured. It is also interesting to note that recoveries of added cGL decrease when larger concentrations of cGL are added. Similar effects have been found for the spiking recovery experiments with other immunoassays (18,19). These effects are usually explained with serum factors that bind the analyte of interest and render it unavailable for the detection system. It is intriguing to speculate that cGL, when added to serum, could be bound by a serum factor; form multimers; or form aggregates with serum lipids, rendering them unavailable for measurement by the assay. However, these speculations were not further evaluated at this point.

Coefficients of variation for intra-assay variabilities for the 3 samples were 25.5%, 9.4%, and 13.4% and would suggest that the assay shows suboptimal precision in the lower area of the working range. Coefficients of variation for interassay variabilities for the 3 samples were 26.0%, 17.2%, and 14.4% suggesting that the assay shows suboptimal reproducibility, especially in the lower range of the assay.

In summary, the ELISA for the cGLI described here shows a wide working range and is highly sensitive. Unfortunately, validation characteristics for the assay are suboptimal. However, for serum concentrations above approximately 2.000 ng/L the assay is linear, precise, and reproducible. For values below that, the assay is more semiquantitative in nature. It remains to be determined whether future studies using monoclonal antibodies will be able to overcome these performance issues. However, such further studies will only be justifiable if preliminary clinical studies using this assay support the clinical usefulness of serum cGLI determination for the diagnosis of gastric disorders in the canine. Currently, the clinical usefulness of this assay is unknown and the assay should not be employed in the diagnosis of gastric disorders in the canine.