Development and validation of a specific radioimmunoassay for equine osteocalcin

2.1. Purification of equine osteocalcin 

Equine OC was extracted and purified from long bones of a foal by acid demineralization, reverse phase chromatography, gel filtration over Superdex-75 column and ion-exchange chromatography over Mono Q column, as previously described [5], [20]. Complete amino acid sequencing was performed by automated Edman degradation using an automated Beckman sequencer. Chemical cleavages were performed with trypsin, chemotrypsin and CnBr and the peptide fractions were subsequently subjected to RP-HPLC analysis. The dialyzed Mono Q purified equine OC used for immunization, iodination, and standards showed a single band on SDS–polyacrylamide electrophoresis.

2.2. Preparation of the antisera 

Two rabbits were immunized against equine OC according to the method of Vaitukaitis et al. [21]. Each animal received in 2-week intervals intradermal injections of 250μg purified equine OC, emulsified in complete Freund’s adjuvant or incomplete Freund’s adjuvant. After 6 weeks, serum samples were drawn in 2-week intervals, the blood was allowed to clot, was centrifuged for 20min at 1000×g and was subsequently stored at −21°C. Each serum obtained was tested for the antibody titer to equine OC by binding to radioactively labeled equine OC.

2.3. Equine serum osteocalcin-free samples 

Equine serum-free samples were obtained by jugular venipuncture of an adult gelding, which was treated with corticosteroids for reasons unrelated with this study. The serum-free blood samples were allowed to clot at room temperature and were subsequently centrifuged. The serum was stored at −21°C in various aliquots till use.

2.4. Radiolabeling of equine osteocalcin 

Equine -OC was prepared according to the Iodogen method using Na- and an iodinating reagent (Iodogen® Perbio Science, Erembodegem-Aalst, Belgium). Ten micrograms of equine OC were dissolved in 40μL of phosphate buffer 0.2M, pH 7.5 and were incubated with 20μg of iodogen and 1mCi of Na- (specific activity 17.4mCi/μg iodide; Amersham Pharmacia Biotech UK Ltd., Little Chalfont, UK). The reaction period was 5min at room temperature. Radiolabeled OC was separated from free Na- by gel filtration on a Sephadex G-25 column (cm, Amersham Pharmacia Biotechnology, Uppsala, Sweden), using 0.025M Tris, 0.01M MgCl2 at pH 7.5, containing 0.1% bovine serum albumin (BSA, ICN Biomedicals Inc., OH, USA) and 0.01% neomycin sulfate (Sigma–Aldrich, Steinheim, Germany) as eluting buffer (Tris–BSA). The eluted fractions were collected in 1mL aliquots. The radioactivity of each fraction was measured and different fractions of the tracer were tested with the antiserum. The best tracer-fractions were chosen, depending on non-specific binding (NSB) value and bound/total (B/T) ratio, and were stored at −21°C in 100μL aliquots.

2.5. Radioimmunoassay procedure 

The assay was developed in Tris–BSA buffer. All measurements were performed in duplicate, in polystyrene tubes, under equilibrium conditions at room temperature. Equine standards were prepared from homogeneously pure equine OC. NSB measurements were performed with 0.3mL Tris–BSA, 0.1mL serum-free equine OC sample (1:10 dilution) and 0.1mL tracer. Total count (TC) measurements were performed with 0.1mL tracer containing approximately 25,000cpm of -OC. Standard tubes (0–50ng/mL) contained 0.1mL Tris–BSA buffer, 0.1mL serum-free equine OC (1:10 dilution), 0.1mL standard OC (1:10 dilution), 0.1mL of antiserum and 0.1mL tracer. Unknown sample measurements were performed with 0.2mL Tris–BSA buffer, 0.1mL unknown sample (1:10 dilution), 0.1mL of antiserum and 0.1mL tracer. All tubes were vortexed and incubated overnight at room temperature.

Free and antibody bound -OC were separated by the addition of 1.0mL of precipitation buffer, containing Tris–HCl 0.025M, pH 7.5, 1% sheep antiserum to rabbit IgG (CER, Marloie, Belgium), cellulose 0.05% (Merck KGaA, Darmstadt, Germany), polyethylenglycol 6000 4% (PEG, Vel, Leuven, Belgium) and BSA 0.4%. After an additional hour of incubation at room temperature, 2mL of Tris–BSA were added to each tube. All tubes were subsequently centrifuged at 2890×g for 20min at +4°C. The supernatant was discarded, total and antibody bound labeled -OC were determined with a gamma counter having a counting efficiency of 75% (LKB Wallac 1261 Multigamma automatic counter, Breda, The Netherlands).

2.6. Specificity 

The affinity of the antiserum for equine OC and not for other serum and bone constituents has been determined by a gel filtration elution profile of equine serum and of bone extract. According to Price et al. [22] bone proteins were extracted using 0.5M EDTA, pH 8.0 at +4°C for 8 days and subsequently were dialyzed against 5mM NH4HCO3, and lyophilized. Equine serum (1mL) and bone extract (10mg) were separately subjected to a Superdex-75 column (cm) equilibrated in 100mM Tris–HCl, pH 8.0. One milliliter fractions were collected. The quantity of total proteins was analyzed by UV spectroscopy at 280nm and the quantity of equine OC was analyzed with the equine-specific RIA.

2.7. Sensitivity, precision, reproducibility and accuracy 

The assay’s lowest detection limit of the equine OC RIA was determined by calculating the mean value of 20 B0 measurements minus 2 SD. The assay’s precision and reproducibility were determined by calculating the intra- and inter-assay coefficient of variation (CV). The intra-assay precision was obtained by calculating the CV of eight measurements of a high-, medium- and low-concentration OC pool analyzed within the same assay. The inter-assay precision was obtained from duplicate samples of the high-, medium-, and low-concentration OC pool measured in five consecutive assays. Potential non-specific effects of equine serum on the equine OC RIA were tested by recovery and linearity tests. For the recovery test serum samples containing equine OC were spiked with known concentrations of equine OC (1, 3, 5, 7, 9, 11, 13, and 15ng/mL). In the linearity test, two serum samples were diluted with OC-free serum.

2.8. Experiments 

2.9. Statistical analysis 

The data were analyzed with a commercially available software program (Statistical Analysis Systems, Cary, NC, USA). Mean values (x) and SD were calculated. The inter- and intra-CV were calculated as the SD divided by the mean value. Mean recoveries at each concentration were calculated as a percentage of the expected value. Pearson’s coefficient of correlation analysis and linear regression analysis were performed for OC RIA assay comparisons.

3.1. Radiolabeled equine osteocalcin and antisera 

The iodination of equine OC by the Iodogen method produced a -radiolabeled equine OC tracer with a specific radioactivity of 500Ci/mmol. All immunized rabbits produced antibodies to equine OC within 6 weeks. The -radiolabeled OC bound at an antiserum final dilution of 1:25,000. The tracer allowed a highly sensitive OC RIA with low NSB (<4%) and high-specific binding.

3.2. Standard curve and sensitivity 

The typical standard curve ranged from 0.1 to 50ng/mL and is illustrated in Fig. 1. The assay’s lowest detection limit of the equine OC RIA was 0.2ng/mL.

View Large Image Download to PowerPoint

Fig. 1. RIA for equine OC. The fraction of -radiolabeled equine OC bound to antibodies (B/B0) at increasing levels of equine OC. Serial dilutions of two equine serum samples diluted at 1:10, 1:20, 1:40, 1:80, and 1:160 were parallel to the standard curve.

3.3. Specificity 

The analysis of all fractions of the chromatography profile by spectroscopy and equine OC RIA showed that serum proteins and bone proteins, which did not coelute with OC, did not bind significantly to OC antiserum (Fig. 2, Fig. 3). Larger and smaller sized proteins did, therefore, not cross-react non-specifically with the OC assay. In addition, about 4% of the total activity detected in the bone extract were found to be associated with fractions eluting at the void volume.

View Large Image Download to PowerPoint

Fig. 2. Gel filtration of equine OC from serum. Undiluted equine serum was chromatographed on a Hiload Superdex-75 column (cm) equilibrated in 100mM Tris–HCl, pH 8.0. The elution profile of the serum sample was monitored by UV spectroscopy (dotted line). The elution profile of serum OC was determined by the equine OC RIA (○).

View Large Image Download to PowerPoint

Fig. 3. Gel filtration of equine OC from bone. Ten milligrams of bone extract were chromatographed on a Hiload Superdex-75 column (cm) equilibrated in 100mM Tris–HCl, pH 8.0. The elution profile of the proteins was monitored by UV spectroscopy (dotted line). The elution profile of OC was determined by the equine OC RIA (○).

3.4. Reproducibility and accuracy of the radioimmunoassay 

The inter- and intra-assay CV were summarized in Table 1. The accuracy of the equine OC RIA is shown in Table 2. Serial dilutions of equine serum samples showed curves which paralleled the standard curve (Fig. 1).

Table 1. Intra- and inter-assay precision of the equine-specific OC RIA

		Within assay			Between assay
	n	8	8	8	5	5	5
	Mean (ng/mL)	89.6	42.7	18.1	93.1	43.2	20.2
	SD	5.5	2.0	0.9	7.6	3.2	1.0
	CV (%)	6.2	4.7	5.1	8.2	7.3	5.0

n, number of measurements; SD, standard deviation; CV, coefficient of variation.

Table 2. Recovery of added OC to equine serum samples

	Expected value (ng/mL)	Observed value (ng/mL)	Recovery (%)
	1	1.079	107.90
	3	3.094	103.13
	5	4.981	99.62
	7	7.459	106.56
	9	8.808	97.87
	11	11.812	107.38
	13	13.571	104.39
	15	14.082	93.88

Equine serum was spiked with known quantities of equine OC. Mean recoveries at each concentration were calculated as a percentage of the expected value.

3.5. Experiments 

3.5.1. Equine osteocalcin values in normal horses 

Serum OC values ranged from 3.68 to 127.31ng/mL. The effect of age on serum OC values (n=31) is shown in Fig. 4. Serum creatinine values were in the normal range (90–140μmol/L).

View Large Image Download to PowerPoint

Fig. 4. Serum OC values (ng/mL) of clinically healthy horses (n=31) as a function of age (years).

3.5.2. Radioimmunoassay comparison 

Serum OC levels were evaluated in 39 horses using the equine-specific OC RIA as well as two bovine-specific OC RIA kits. A tight correlation was obtained between the equine-specific OC RIA and bovine OC RIA kit #1 (r=0.976; P<0.0001), respectively, bovine OC RIA kit #2 (r=0.952; P<0.0001). There was also a tight correlation between bovine OC RIA kit #1 and bovine OC RIA kit #2 (r=0.947; P<0.001). Linear regression analysis of all three tests is shown in Fig. 5a–c. However, mean (±SD) OC values obtained with the equine-specific OC RIA (x=21.017±18.6ng/mL) were higher than those obtained with the bovine OC RIA kit #1 (x=17.17±12.64ng/mL) and bovine OC RIA kit #2 (x=5.69±3.55ng/mL).

View Large Image Download to PowerPoint

Fig. 5. (a) Linear regression analysis of equine serum sample results (n=39) analyzed with the equine-specific OC RIA and the bovine-specific OC RIA kit #1 (OC DSL-6900, DSL Inc., Webster, TX, USA). (b) Linear regression analysis of equine serum sample results (n=39) analyzed with the equine-specific OC RIA and the bovine-specific OC RIA kit #2 (OC125l RIA Kit, Incstar Corp., Stillwater, MN, USA). (c) Linear regression analysis of equine serum sample results (n=39) analyzed with the bovine-specific OC RIA #1 and the bovine-specific OC RIA #2.