Technical note: Time-resolved fluoro-immunometric assay for
intact insulin in livestock species1
P. Løvendahl*2 and H. M. Purup†
Departments of *Animal Breeding and Genetics and †Animal Nutrition and Physiology,
Danish Institute of Agricultural Science, Research Center Foulum, DK 8830 Tjele, Denmark
the europium label is stabilized before measurement.
This gives a sensitivity of 3 pmol/L and a possible working range up 16,700 pmol/L. There is no cross-reactivity
with pro-insulin or IGF-I. Calibrators are prepared in
heat-inactivated serum from the relevant species. Porcine and bovine insulin have different calibration
curves; porcine insulin is more reactive and has a higher
background than bovine insulin. Validation results
show low CV values, parallel dilution of samples, and
a recovery ratio close to unity. Comparison with a commercial RIA shows good agreement, except at low concentrations, at which the RIA determinations are inaccurate. Plasma samples from other domestic species
(horse, sheep, goat, and mink) have also been assayed,
but it is emphasized that calibrators should be prepared
in heat-inactivated serum from the appropriate species,
and preferably insulin from that species should be used
for calibration.
ABSTRACT: Insulin levels in ruminants are often
very low and hence are difficult to measure with commercially available RIA kits designed for use with human serum or plasma samples. Those assays may also
have high cross-reactivity with nonintact insulin. An
assay originally invented for human insulin and based
on a pair of monoclonal antibodies binding to specific
parts of the insulin molecule was further developed
and validated for use with bovine or porcine plasma or
serum. The assay is of the sandwich type, with the
catching antibody coated to the solid phase of microtiter
plate wells and with the detecting antibody labeled with
europium, and measured as time-delayed fluorescence.
The assay protocol includes an incubation step in which
plasma samples of 50 ␮L are incubated with buffer and
detecting antibody for 3 h in coated wells, followed by
an enhancement step in which the fluorescence from
Key Words: Cattle, Horses, Insulin, Pigs
2002 American Society of Animal Science. All rights reserved.
Introduction
J. Anim. Sci. 2002. 80:191–195
concentrations so high sensitivity is required, but often
commercial assays are not available.
Although the RIA technique in general is reliable, its
use of radioactive chemicals is a drawback, because
radioactivity is a safety risk in the laboratory and to
the environment and the shelf life of RIA is short. Alternative assays without these drawbacks have become
available.
An assay based on two monoclonal antibodies, arranged as a sandwich using microtiter plates, was developed for human applications by Toivonen et al.
(1986) and Andersen et al. (1993). The assay showed
large cross-reactivity to insulin from other species but
was only validated for human samples. The purpose of
this study was to further develop and validate that
assay for use in various domestic animal species,
mainly cattle and swine, but also in horses, sheep, and
goats. Through use of europium as a label, this timeresolved fluoro-immunometric assay (TR-IFMA) for insulin offers high sensitivity, is fast and easy to run,
and uses no hazardous chemicals.
Measurements of insulin in plasma and serum from
animal experiments have by tradition employed RIA
based on polyclonal antibodies. A common problem is
that many RIA assays are unable to distinguish between the different forms of insulin, namely pro-insulin
(CIS RIA kit, St. Quentin, France; our observations),
C-peptide, and intact insulin, the last believed to be
the active form. Furthermore, the commercially available RIA products are essentially tailored to human
applications. Ruminants have low circulating insulin
1
We are grateful for the valuable support, ideas, protocols, and
peptides during development of the TR-IFMA from O. Blåbjerg, Odense Univ. Hospital, and R. Djurup, NovoNordisk A/S. We also want
to thank J. Adamsen, I. L. Sørensen, and L. R. Norup for their skilled
technical work and M. Vestergaard for comments on the manuscript.
2
Correspondence: P.O. Box 50 (phone: +45 8999 1338; fax: +45
8999 1300; E-mail: [email protected]).
Received November 13, 2000.
Accepted August 15, 2001.
191
192
Løvendahl and Purup
Materials and Methods
Antibodies. Paired monoclonal antibodies are used in
the assay. One was raised against human insulin (HUI018, NovoNordisk Farmaka A/S, Bagsværd, Denmark)
and the other raised against bovine insulin (OXI-005,
Novo Nordisk Farmaka A/S; Jørgensen et al., 1984).
These antibodies bind to epitopes available only in intact insulin (Andersen et al., 1993).
Coating of Microtiter Plates. A coating solution containing 0.5 ␮g of HUI-018 per 200 ␮L of buffer (50 mmol
of K2HPO4, 150 mmol of NaCl, pH = 9.2, sterile) was
prepared in polypropylene bottles and allowed to equilibrate binding to surfaces of containers and other utensils for 0.5 h before coating of plates began. Microtiter
plate wells (1244-550, Wallac OY, Turku, Finland) were
coated with 200 ␮L of coating solution. Plates were
incubated overnight at room temperature (20°C),
washed three times with wash solution (1380-1088,
Wallac OY), and blocked with assay buffer spiked with
0.5% BSA (assay buffer: 100 mmol/L of Tris-HCl, 150
mmol/L of NaCl, 7.7 mmol/L of NaN3, 20 ␮mol/L of
diethylene triamine penta acetic acid [DTPA], 0.5%
BSA [A 1662, Sigma Chemical, St. Louis, MO], and
0.1% Tween-20) for 1 h while shaking and then aspirated, and the wells were filled with 50 ␮L of blocking
buffer. Plates were covered with adhesive film and
stored cold until use (4°C, up to 4 wk) in boxes in the
presence of a damp cloth containing 3 mL of 1% Germall
(Wallac OY).
Labeling of antibody OXI-005 with europium (Eu3+)
followed the instructions for the labeling kit provided
by the manufacturer (1244-302, Wallac OY). Purification of Eu3+-labeled antibody was on PD-10 columns
(Pharmacia, Uppsala, Sweden). After characterization,
labeled antibody was stored cold (4°C) in aliquots and
lasted at least 12 mo.
Heat-Inactivated Serum. Serum for preparation of calibrators was inactivated by heating it to 56 ± 1°C for 3
h, followed by removal of precipitates by filtration on
a glass filter and centrifugation (4,000 × g, 4°C, 30 min).
Calibrators. Monocomponent bovine or porcine insulin was used to prepare calibrators in heat-inactivated
serum. The biological activities for bovine and porcine
insulin were 26.9 IU/mg and 28.5 IU/mg, respectively.
Calibrators were stored frozen (−20°C) for up to 6 mo.
Assay Procedure. Coated microtiter plates were
washed once with wash solution. To each well was
added 50 ␮L of calibrator or sample and 200 ␮L of assay
buffer containing 25 ng of Eu3+-OXI-005 label. Plates
were incubated at room temperature in one of two ways:
1) 3 h with continuous shaking or 2) 10 min of shaking,
overnight resting, and 10 min of shaking. The porcine
assay was not improved by incubation for more than 2
h of continuous shaking, so 2 h was used. Plates were
washed six times. Bound Eu3+ was dissociated by 250
␮L of enhancement solution (1244-104, Wallac OY) during 10 min of shaking and 10 min of rest. Time-resolved
fluorescence was counted on a time-resolved fluorome-
Figure 1. Calibration curves for the time-resolved fluoro-immunometric assay (TR-IFMA), with bovine and
porcine monocomponent insulin in heat-inactivated bovine or porcine plasma. Curves for bovine b-pro-insulin
and human IGF-I show a low degree of cross-reactivity.
ter (DELFIA 1234 Research Fluorometer, Wallac OY).
Standard curves were fitted using a smoothing spline
algorithm on log-log values (Multicalc, Wallac OY).
Validation. Comparable assays for human insulin
were validated by Toivonen et al. (1986) and as an
ELISA by Andersen et al. (1993). The modified assay
was validated for cross-reactivity to bovine pro-insulin
(NovoNordisk Farmaka A/S) and IGF-I (Bachem Feinchemikalien, Bubendorf, Switzerland) and linearity of
dilution for recovery of insulin in spiked samples and
for variation within and between assays. Furthermore,
results were compared with those obtained with a commercial RIA (CIS) on plasma from calves subjected to
an i.v. glucose tolerance test.
Quality-control plasma was obtained from three
calves and three pigs. Plasma samples containing low
concentrations of insulin were obtained after overnight
feed deprivation, medium values from animals with ad
libitum intakes, and high values following an i.v. bolus
injection of glucose. Quality-control plasma was stored
frozen at (−20°C) in Cryo-vials (Nunc, Roskilde,
Denmark).
Glucose Tolerance Test Data. Twenty 9-mo-old calves
were subjected to an i.v. glucose tolerance test following
an overnight feed deprivation. A bolus dose of glucose
(0.2 g/kg BW) was administered through a jugular cannula. Blood was sampled serially into heparinized tubes
at −30, −15, −5, 5, 10, 15, 20, 30, 45, and 60 min relative
to glucose administration. Plasma was obtained by centrifugation (2,000 × g, 4°C, 20 min), and stored frozen
(−20°C) until it was assayed by RIA and TR-IFMA.
Results and Discussion
Standard curves for bovine and porcine insulin are
shown in Figure 1. The lower detection limit was 3
193
Time-resolved insulin assay
pmol/L (bovine and porcine). Standard curves for bovine
and porcine insulin were not parallel, indicating a matrix effect of the plasma and also different immunoreactivities of bovine and porcine insulins. The working
range of the assay covers up to 16,700 pmol/L. At higher
concentrations we observed decreasing fluorescence
counts, the so-called hook effect (not shown). For routine assays of bovine insulin, we have used seven calibrators for the standard curve (3.34, 6.68, 13.36, 26.72,
106.9, 427.5, and 1,710 pmol/L) and a similar curve for
porcine insulin.
The sensitivity of the current assay (3 pmol/L ∼ 0.4
␮IU/mL) is as good or better than comparable commercial assays (e.g., DPC Immulite, 2.0 ␮IU/mL or DSL
ELISA, 0.3 ␮IU/mL, producers’ data sheets). The assay
is easy and fast to run because it has few pipetting
steps and a short incubation time, yielding results in
less than 4 h. Small aberrations in temperature and
incubation time do not impair performance (our unpublished observations). Furthermore, the shelf life is long
and we have so far not experienced “down” periods.
Cross-reactivity with bovine pro-insulin, calculated
as “measured insulin” divided by concentration of proinsulin, was undetectable up to 100 pmol/L and below
1% at 1,000 and 10,000 pmol/L (Figure 1). Cross-reactivity with IGF-I was nonsignificant (Figure 1). Crossreactivity with insulin from other animal species has
been studied using plasma samples. Besides samples
from cattle and pigs, the assay also works with plasma
from horses, sheep, goats and mink (our unpublished
data).
Recovery of quantitative amounts of bovine insulin
added to pools of plasma was close to unity (Table 1).
Dilution of plasma with heat-inactivated serum gave
measurements parallel to the calibration curve, if the
measured concentrations were within the range of the
assay (Table 1).
Variation in pools of quality-control plasma (QC) was
assessed during routine use. Each assay contained a
single 96-well plate. On bovine plates, QC were placed
following the calibrators and again in a position at the
end of the plate. Porcine plates had QC placed following
the calibrators only. At each position on the plate, QC
were assayed in duplicate. By analysis of variance, the
total variance in QC was split into components of between assays, within assay between positions, and
within position (Table 2).
Comparison with RIA. The assay was compared with
a commercial RIA kit (CIS) that is based on human
insulin calibrators and expresses results in terms of IU
of human insulin at a specified biological activity of
22.9 IU/mg. When five bovine calibrators were run in
the RIA (Figure 2), their concentration was modeled by
a linear regression, using picomoles/liter on both axes:
Y = 30.038 + 0.987 X,
R2 = 0.996
However, the heat-inactivated serum (0-calibrator) was
measured as 39 pmol/L in the RIA, showing that bovine
serum interferes with the RIA. Furthermore, pro-insulin showed significant binding in the RIA with a crossreactivity close to 50% (data not shown).
Conversion Formulas. Ten plasma samples from each
of 20 calves (n = 200) subjected to a glucose tolerance
test were measured in both assays in order to develop
a conversion formula, taking the different units and
peptide immunoreactivities into account (Figure 3):
Table 1. Recovery of quantitative amounts of bovine insulin added to bovine plasma
samples and linearity of dilution with heat-inactivated seruma
Recovery experiment
Sample ID
Neat
b-Lowb
17.03
b-High
453.4
Spiked to
Measured
32.03
182.33
33.90
182.20
696.7
3,702.7
646.5
3,742.0
Recovery ratio
1.06
1.00
0.93
1.01
Dilution experiment
Sample ID
Dilution
b-Low
Neat
1+1
21.21
11.19
b-High
Neat
1+9
646.5
74.82
748.2
1.16
b-Medium
Neat
1+1
1+3
1+7
221.6
117.4
60.0
26.3
234.8
240.2
210.6
1.06
1.09
0.95
Measured
Calculated
22.38
Linearity ratio
1.06
a
The two experiments were run in separate single plate assays, with samples run as triplicates. Results
are concentrations in pmol/L, measured or calculated.
b-Low, etc. = low concentration of bovine insulin, etc.
b
[A]
194
Løvendahl and Purup
Table 2. Coefficients of variation, between assays (total), between positions on plate
within assay plate, and within and between duplicates within position on plate
(residual), estimated from variance components; porcine assays had
quality-control plasma (QC) in only one position on the plate
QC-id
Bovine
b-Lowa
b-High
Porcine
p-Low
p-Medium
p-High
Assays,
no.
Wells,
no.
Mean
concentration,
pmol/L
Coefficient of variation, %
Total
Within
Residual
81
81
323
327
18.82
601.4
13.9
7.9
9.9
5.5
6.8
3.5
9
9
9
64
61
62
23.63
250.8
510.9
13.8
9.9
10.9
8.7
3.4
5.1
b-Low, etc. = low concentration of bovine insulin, etc. p-Low, etc. = low concentration of porcine insulin, etc.
a
Y (pmol/L) = −33.91 + 6.686 X (␮IU/mL),
R2 = 0.901.
[B]
Formula B closely resembled formula A derived from
the calibrators. In both formulas the intercept expresses the matrix effect, and the regression coefficient
expresses the conversion of micro-IU/milliliter to picomoles/liter. The conversion factor is close to the expected values obtained from the molar weight of bovine
insulin (Mr = 5,733 g/mol) and the biological activity of
25 to 27 IU/mg in bovine insulin preparations. In general, RIA and TR-IFMA values were more equal at high
than at low concentrations because the RIA gave inaccurate results at low concentrations.
Figure 2. Bovine monocomponent insulin calibrators
assayed by RIA kit. The conversion formula excludes the
0-calibrator.
Limitations. The europium tracer is sensitive to
strong chelators, so EDTA and citrate are not suitable
anticoagulants for blood samples; consequently, heparin is recommended. However, it is possible to split
the incubation into two steps if only EDTA-containing
samples are available, but so far we have validated the
one-step procedure only.
Implications
This time-resolved immuno fluoro-metric assay for
insulin in domestic animals has high sensitivity and
Figure 3. Bovine insulin measured by RIA and timeresolved fluoro-immunometric assay (TR-IFMA) in 200
plasma samples from 20 calves subjected to an i.v. glucose
tolerance test. The regression line gives the conversion
formula.
195
Time-resolved insulin assay
low cross-reactivity with pro-insulin and other insulinrelated peptides. A commercial version of the assay has
now become available as a kit for use with human sera
(B080-101, Wallac OY), and may be used for animal
samples with appropriate modifications to calibrators.
Results of this study emphasize the need for calibrators
and matrix (i.e., serum) to come from the species under study.
Literature Cited
Andersen, L., N. Dinesen, P. N. Jørgensen, F. Poulsen, and M. E.
Røder. 1993. Enzyme immunoassay for intact human insulin in
serum or plasma. Clin. Chem. 39:578–582.
Jørgensen, P. N., C. Y. Wu, P. C. Pedersen, S. Patkar, and J. Zeuthen.
1984. Specificity and cross-reactivity of monoclonal antibodies
against bovine insulin. Dev. Biol. Stand. 57:313–319.
Toivonen, E., I. Hemmilä, J. Marniemi, P. N. Jørgensen, J. Zeuthen,
and T. Lövgren. 1986. Two-site time-resolved immunofluorometric assay of human insulin. Clin. Chem. 32:637–640.