Development and validation of a biosensor-based immunoassay

for progesterone in bovine milk

Els H. Gillis a, James P. Gosling b, Joseph M. Sreenan c, Marian Kane a,*

aNational Diagnostic Centre, National University of Ireland, Galway, IrelandbDepartment of Biochemistry, National University of Ireland, Galway, IrelandcAnimal Reproduction Department, Teagasc, Athenry, Co. Galway, Ireland

Received 30 January 2002; received in revised form 22 April 2002; accepted 2 May 2002

Abstract

We have developed a rapid automated immunoassay, using the BIACOREk surface plasmon resonance (SPR) biosensor, to

measure progesterone in bovine milk. The assay was designed as an inhibition assay with progesterone covalently immobilised

to the carboxymethyl dextran matrix of a CM5 sensor chip. A fixed amount of monoclonal anti-progesterone antibody 39C5H7

was mixed 9:1 with the sample and the amount of free antibody was then determined using biomolecular interaction analysis

(BIA) by injection of the mixture over the immobilised progesterone sensor surface. The assay was designed to cover the

concentration range 0.5 to 50 ng/ml. The limit of detection (LOD) was 3.56 ng/ml. Reproducibility of the assay was very good

with both intra-assay and inter-assay coefficients of variation < 5%. As results become available within minutes of injection and

the procedure involves fully automated instrumentation, we believe that this BIA assay for progesterone in milk could be used

in-line in the milking parlour and, thus, provide an important tool for reproductive management of dairy cattle to detect heat and

predict pregnancy.

D 2002 Elsevier Science B.V. All rights reserved.

Keywords: SPR; Biosensor assays; Progesterone; Heat detection

1. Introduction

Reproductive performance is a major factor affect-

ing the production and economic efficiency of dairy and

beef cow herds. According to Sreenan and Diskin

(1992), a dairy herd is reproductively efficient when

the calving interval is 365 days, when 90% of the cows

calve in a 10-week period andwhen 5%or less cows are

culled for reproductive failure. For herds using artificial

insemination (AI), failure to detect heat is one of the

major causes of a prolonged calving interval. Visual

observation is currently the primary method for heat

0022-1759/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.

PII: S0022 -1759 (02 )00166 -7

Abbreviations: AI, Artificial insemination; CV, Coefficient of

variation; BIA, Biomolecular interaction analysis; CMO, Carbox-

ymethyloxime; EDC, N-ethyl-NV(3-ethylaminopropyl) carbodiimide;

ELISA, Enzyme-linked immunoassay; DMF, N,N-dimethylforma-

mide; HBS-EP, HEPES-buffered saline; HEPES, N-[2-hydroxye-

thyl]piperazine-NV-[2-ethanesulfonic acid]; LOD, Limit of detection;

NHS, N-hydroxysuccinimide; SD, Standard deviation; SPR, Surface

plasmon resonance.* Corresponding author. Tel.: +353-91-586559; fax: +353-91-

586570.

E-mail address: marian.kane@nuigalway.ie (M. Kane).

www.elsevier.com/locate/jim

Journal of Immunological Methods 267 (2002) 131–138

detection, but this is time consuming and repetitive.

Farmers frequently achieve a heat detection rate of only

35–70% and up to 20% of cows presented for insemi-

nation are not in heat (Sreenan andDiskin, 1992; Diskin

and Sreenan, 2000).

According to Senger (1994), an ideal system for

detecting oestrous should have the following charac-

teristics: (i) continuous surveillance of the cow; (ii)

accurate and automatic identification of the cow in

oestrous; (iii) operation for the productive life-time

of the cow; (iv) minimal labour requirements; and

(v) high accuracy and efficiency (95%) for identify-

ing the appropriate physiological events that corre-

late with oestrous or ovulation or both. Elevated

milk progesterone concentrations indicate luteal

dominance, while low amounts of progesterone are

associated with oestrous. Progesterone measurement

tests based on enzyme-linked immunoassays

(ELISA) are available as kits for on-farm use (Nebel,

1988). Most of these tests are, however, manual and

designed as qualitative tests for the confirmation of

oestrous and determination of pregnancy or non-

pregnancy, rather than providing a precise concen-

tration. Attempts at automation and in-line applica-

tion of quantitative enzyme immunoassays for

progesterone have been described (Koelsch et al.,

1994; Claycomb et al., 1998; Pemberton et al.,

2001), but, to date, such methods either suffer from

poor sensitivity or are still too cumbersome for use

in the milking parlour.

Biomolecular interaction analysis (BIA) from Bia-

core Ab uses the optical phenomenon of surface

plasmon resonance (SPR) to monitor biomolecular

interactions in real time without labelling. SPR meas-

ures changes in refractive index of the solution close

to the sensor surface, resulting from changes in the

mass concentration of molecules in the solution

(reviewed by Hashimoto, 2000; Markey, 2000). BIA-

CORE instrumentation was developed primarily for

use as a research tool, with a variety of applications

such as screening biological samples for binding

partners, the determination of kinetic and equilibrium

constants for complex formation, epitope mapping

and ligand fishing for known receptors. Recently, this

principle has also gained popularity as an analytical

tool for the determination of concentrations of analy-

tes in biological fluids. In 1993, Minunni and Mascini

described a BIACORE assay for the determination of

the pesticide, atrazine, in drinking water, while Ster-

nesjo et al. (1995) used a BIACORE to develop a

biosensor-based immunoassay for the detection of

sulfamethazine residues in milk. Since then, BIA

assays have been described for other veterinary drug

residues (Elliott et al., 1999; Gaudin and Maris,

2001), toxins (Mullett et al., 1998; Daly et al.,

2000) and vitamins (Bostrom Caselunghe and Linde-

berg, 2000; Indyk et al., 2000).

When compared to traditional techniques, BIA-

CORE technology offers several significant advan-

tages, such as high precision, speed, simplicity, and

automation (Mellgren et al., 1996). This paper

describes the development and validation, according

to principles set out by Wong et al. (1997), of a BIA

assay for real-time measurement of progesterone in

bovine milk.

2. Materials and methods

2.1. Instrumentation

A BIACORE 2000k biosensor instrument and

CM5 sensor chips (research grade) were used (Biacore

Ab, Uppsala, Sweden). The BIACORE 2000k was

controlled by BIACORE Control Software version

3.1.1 running underWindows 95. The software also in-

cludes an interface to the separate BIAevalution soft-

ware. The instrument running temperature was 25 jC.

2.2. Reagents

HEPES-buffered saline (HBS-EP) buffer pH7.4

(10 mM N-[2-hydroxyethyl]piperazine-NV-[2-ethane-sulfonic acid] (HEPES), 0.15 M NaCl, 3.4 mM

EDTA, 0.005% Surfactant P-20), and amine coupling

kit (containing N-hydroxysuccinimide (NHS), N-

ethyl-NV(3-ethylaminopropyl) carbodiimide (EDC),

and ethanolamine hydrochloride) were obtained from

Biacore Ab (Uppsala, Sweden). The Ridgeway milk

progesterone ELISA was purchased from Ridgeway

Science (Alvington, Gloucestershire, UK). Progester-

one-3-carboxymethyloxime (CMO) was purchased

from Steraloids (Wilton, NH, USA). Progesterone

(MW 314.5) and N,N-dimethylformamide (DMF)

were purchased from Sigma. All other reagents were

of analytical grade.

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138132

2.3. Monoclonal antibody

The monoclonal antibody used was that described

by O’Rorke et al. (1994). The antibody was raised

against 11a-hydroxyprogesterone hemisuccinate–

bovine serum albumin conjugate, which was prepared

by modification of the mixed anhydride method

(Erlanger et al., 1957). Ascites fluid from mice

infected with hybridoma cells secreting antibody

39C5H7 was diluted 1/5000 in HBS-EP before use.

2.4. Immobilisation of progesterone-3-CMO deriva-

tive

Simultaneous immobilisation of progesterone onto

all four flowcells of a CM5 sensor chip was performed

outside the biosensor system using conventional amine

coupling (Sternesjo et al., 1995). For covalent immo-

bilisation, carboxyl groups of the sensor chip surface

are activated by derivatisation with N-hydroxysuccini-

mide (NHS) mediated by N-ethyl-NV(3-ethylamino-

propyl) carbodiimide (EDC). The resulting active

ester groups will react spontaneously with amino

groups (Johnsson et al., 1991). EDC and NHS were

mixed (1:1) and 40 Al of the mixture were deposited on

the gold film for 10 min activation. An amine surface

was then prepared by placing 40 Al of 1M ethylene

diamine pH8.5 in contact with the activated surface for

20 min. Capping any remaining active carboxyl groups

was achieved by exposing the surface to 40 Al 1Methanolamine for 10 min. Progesterone-3-CMO (2 mg)

was dissolved in 900 Al DMF and mixed with 450 Al of10 mM sodium acetate solution pH4.6 containing 5 mg

EDC and 2 mg NHS. Progesterone was then immobi-

lised on the surface by placing 40 Al of this solution in

contact with the amine surface for approximately 15

min. Prior to use, the immobilised surface was condi-

tioned with repeated injections of 25 Al regenerationbuffer (10% acetonitrile in 50 mM NaOH) to remove

any non-covalently bound progesterone.

2.5. Milk samples

Milk samples were collected from dairy cows

throughout the oestrous cycle. Samples were collected

into 10 ml plastic containers, with one Lactab mark III

tablet (Thompson and Capper, Cheshire, UK) added as

a preservative or alternatively 15 Al of 0.068 M sodium

dichromate solution. Samples were then stored at 4 jC.Before analysis, samples were heated for 40 min in a

water bath at 21 jC and dispersed by gentle mixing on a

rock and roller for 10 min. No difference was noted

between the results obtained in the BIACORE assay

with preserved or unpreserved milk samples.

2.6. Assay

Progesterone standards were prepared by dilution in

preserved whole milk from an oestrous animal. The

concentrations used to construct standard curves were

0.5, 1, 2, 5, 10, 20, and 50 ng/ml (i.e. 3.18, 6.36, 15.9,

31.8, 63.6 and 159 nmol/l). The curves were fitted

using a four-parameter equation with BIAevaluation

software.

The assay was designed as an inhibition assay. A

fixed amount of the monoclonal anti-progesterone

antibody was mixed 9:1 with the standard/sample.

The amount of free antibodies remaining was then

determined by injection of 50 Al of the mixture over

the progesterone surface at a rate of 25 Al/min. After

each measurement the surface was regenerated by

injection of 25 Al of 10% acetonitrile in 50 mM

NaOH at the same flow rate. The analysis cycle time,

including regeneration was 5 min 8 s.

3. Results

3.1. Stability of sensor chip during regeneration

Regeneration performance was evaluated by re-

peated analysis of a negative whole milk sample using

the standard assay procedure. Baseline readings and

antibody binding responses are shown in Fig. 1. The

baseline response showed less than 1% change over

129 cycles and the binding capacity was also unaf-

fected, with less than 5% change being observed from

the beginning to the end of the analysis. The progester-

one sensor surface went through 3600 regeneration

cycles over a 9-month period before any loss in binding

capacity was observed.

3.2. Limit of detection

The assay was designed to cover the concentration

range 0.5 to 50 ng/ml. The standard curve in Fig. 2

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138 133

represents the mean of 10 curves, obtained separately,

eachwith triplicate concentrations of the standards. The

precision of the response for each standard was < 7%

coefficient of variation (CV). The limit of detection

(LOD), determined as the concentration corresponding

to the mean response for negative whole milk minus

three times the standard deviation (SD),was 3.56 ng/ml.

3.3. Reproducibility

Repeated assays in one run (n= 15) of three quality

control samples resulted in the following concentra-

tion-dependent intra-assay CVs: at 3.5 ng/ml, 3.67%;

at 18.41 ng/ml, 1.87% and at 56.03 ng/ml, 3.95%.

Inter-assay variation (n = 10) was determined over 10

consecutive assays using another set of control sam-

ples and was 4.79%, 4.57%, and 3.85% at 3.25, 23.15

and 48.81 ng/ml, respectively. The chip-to-chip vari-

ability was examined by comparison of the standard

Fig. 2. Standard curve of progesterone in bovine milk, produced on

the BIACORE 2000k. The standard curve is derived from results

for 10 sets of standards analysed in triplicate. The error bars indicate

the standard deviations.

Fig. 3. Relationship between the effective volume of milk assayed

and the concentration of progesterone measured. Milk samples were

diluted 1/2, 1/5, 1/10 and 1/20 in negative whole milk before assay.

Fig. 1. Baseline readings (.) and antibody binding response (o)

following 129 successive antibody binding– regeneration cycles

over the progesterone coated surface with a negative whole milk

sample.

Fig. 4. Regression analysis of the correlation between the Ridgeway

ELISA and the BIACORE assay for progesterone in milk samples

taken at various times during the oestrous cycle (n= 70). The inset

graph shows the analysis of a subset of samples (n= 24) with

progesterone concentrations lower than 15 ng/ml. The dotted lines

represent the lines of perfect agreement and the solid lines the

correlation plots.

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138134

curves obtained with three sensor chips coated at

different times. The standard curves obtained were

identical when overlaid (data not shown).

3.4. Linearity/parallelism

Milk samples from five animals were diluted 1/2,

1/5, 1/10 and 1/20 in negative whole milk before

assay. The amount of progesterone measured was

directly related to the effective milk volume assayed

over the range examined (Fig. 3), indicating minimal

interference of the milk matrix. The lowest effective

volume resulted in an over-estimation of progesterone

for some samples but these were below the LOD of

the assay. Regression analysis of the correlation

between actual and measured values of diluted sam-

ples confirmed linearity of the assay with R2 values

> 0.98.

3.5. Recovery

Samples containing various levels of endogenous

progesterone were spiked with 2, 5, 10 and 20 ng/ml

progesterone. The total concentration of progesterone

present in each sample was then determined to calcu-

late the recovery in each case. The percentage recov-

ery ranged from 88.9% to 112.4%, the mean was

101.7% (F 7.8 SD, n = 12).

3.6. Correlation with Ridgeway ELISA

Seventy samples collected at various times through-

out the oestrous cycle were analysed by both the

Fig. 5. Comparison of standard curves obtained by the BIA assay

(.) and the Ridgeway ELISA (o) for progesterone in bovine milk.

Duplicate readings for each assay are indicated on the graph.

Fig. 6. Progesterone levels estimated by the BIA (.) and the Ridgeway assay (o) in daily milk samples. These milk samples were collected for

26 days after artificial insemination. Heat was detected by visual observation.

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138 135

BIACORE assay and the Ridgeway milk progesterone

ELISA. Fig. 4 shows a fairly good correlation between

both assays (R2 = 0.75). Greatest deviation from the

perfect line of agreement occurred at values >15 ng/ml.

Although it has a lower detection limit, the dynamic

range of the Ridgeway assay is limited, with little

discrimination power above 15 ng/ml (Fig. 5). A much

better correlation between the two assays existed

(R2 = 0.93; n = 24) when samples with progesterone

concentrations higher than 15 ng/ml were omitted.

Milk samples were collected daily for 26 days after

insemination of a cow in heat and were analysed using

the BIACORE and the Ridgeway assay. Similar

patterns were obtained with both assays (Fig. 6).

Progesterone levels remained high after 21 days,

indicating pregnancy had occurred, which was later

confirmed by ultrasonography.

4. Discussion

The commercially available surface plasmon reso-

nance biosensor BIACOREk holds great potential

for in-line monitoring as it integrates sampling with

analysis. The BIACORE 2000k is a fully automated

system, including an autosampler for precision liquid

handling, and integrated microfluidics and detection,

which allows precise, rapid and simple-to-operate

analysis of samples.

This study has shown that the BIACOREk can be

used for progesterone analysis in raw bovine milk. A

simple and rapid assay was developed, which

required minimal hands-on time. The LOD was 3.56

ng/ml and intra-assay and inter-assay coefficients of

variation were lower than 5%. The sensitivity of the

BIACORE assay is not as low as for other progester-

one assays reported (Groves et al., 1990; Pemberton

et al., 2001), but this proved not to be a problem as

the assay showed a much greater precision and

accuracy and a good correlation (R2 = 0.93, n = 24)

was observed with the established Ridgeway ELISA

for progesterone concentrations lower than 15 ng/ml.

Preliminary results suggested that it would be possi-

ble to improve the assay sensitivity by changing the

milk to antibody ratio while keeping all other con-

ditions constant. This would, however, narrow the

range of the calibration curve, which was not consid-

ered desirable, as progesterone concentrations would

no longer be directly measurable over the entire oes-

trous cycle.

Regression analysis of the correlation between

actual and measured values of diluted samples con-

firmed linearity of the assay with R2 values >0.98,

while the mean percentage recovery of progesterone

added to milk was 101.7% (F 7.8 SD, n = 12), indi-

cating minimal interference of the milk matrix.

Another important validation parameter examined

was the regeneration performance. Regeneration

should remove all non-covalently bound material

from the sensor without affecting the binding activity

of the immobilised ligand. Poor regeneration can

result in a drifting baseline, which will affect assay

performance. The baseline response showed less than

1% change over 129 cycles and a less than 5% change

in binding capacity was observed from the beginning

to the end of the analysis, which was within the limits

recommended by Wong et al. (1997). The fact that the

BIACORE assay gave a similar profile as the Ridge-

way assay over the oestrous cycle supports its suit-

ability for application in heat and pregnancy detection.

Also, as the analysis cycle time, including regener-

ation, was only 5 min 8 s, concentration values could

be available to the farmer within minutes of injection.

This, to our knowledge, is the first SPR-based

biosensor assay for progesterone in bovine milk. Other

biosensor assays for milk progesterone with different

detection systems have been reported recently. An

amperometric biosensor was described by (Pemberton

et al., 1998, 1999, 2001), in which single-use, dispos-

able screen-printed carbon electrodes were coated with

antibody. The assay involved a 30-min competitive

binding step, during which sample/standard progester-

one competed with enzyme-labelled progesterone for

binding to antibody, followed by continuous flow

electrochemical detection of bound enzyme. Although

the sensitivity of this assay appears to be better than the

BIACORE assay described here, the authors refer to the

need for a significant reduction in assay time and

improvement in precision. Others report the develop-

ment of a biosensor for progesterone, based on a rapid

competitive enzyme immunoassay using covalently

immobilized antibody, horseradish peroxidase as the

enzyme label, and tetramethylbenzidine as the sub-

strate. The dynamic range of the assay was 0.1 to 5 ng/

ml and the assay time was 8 min (Claycomb and

Delwiche, 1998; Claycomb et al., 1998). However,

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138136

the re-usability of the test well was very poor (approx-

imately 15–20 cycles) due to noise from residual en-

zyme activity.

The BIACORE assay developed in this study is

based on covalently immobilised progesterone rather

than immobilised antibody, which accounts for the

much greater stability of the sensor surface compared

to the above methods. During this study, the sensor

withstood 3600 regeneration cycles before any

decrease in binding was observed, which would

enable twice daily measurements to be made on a

herd of 50 cows for a period of well over 1 month.

Since there are four flowcells, these results suggest

that the progesterone sensor surface has the potential

for at least 14000 regeneration cycles, which would,

however, have to be verified. The fact that the SPR

technology is incorporated into fully automated

instrumentation, which is commercially available, also

means that the assay described here could be much

more easily set up in the milking parlour for in-line

testing. Although the instrument model (BIACORE

2000k) used in this study is very expensive, it is

intended as a research tool. However, a less costly

BIACORE Qk system is also available. This is an

instrument dedicated for routine concentration analy-

sis, with greater automation, higher throughput rates

and immediate read-out of results and would be ideal

for on-farm applications.

The BIACORE assay has the potential to con-

tribute significantly to improved reproductive man-

agement of dairy herds by daily monitoring of milk

progesterone levels and, hence, accurate identifica-

tion of animals in heat, early pregnancy diagnosis

and detection of ovarian failure. Also, the introduc-

tion of BIACORE technology into the milking

parlour would facilitate continuous monitoring of

other analytes in milk, of relevance to both produc-

tive and reproductive performance and to milk

quality.

Acknowledgements

EHG is a recipient of a Walsh fellowship from

Teagasc. The BIACORE 2000k was purchased by

a research grant from the Irish Higher Education

Authority through its PRTLI-cycle 1 programme.

The authors would like to thank Dr. Chris Elliot

and Imelda Traynor, Chemical Surveillance Depart-

ment, Veterinary Sciences Division, Belfast, North-

ern Ireland for assistance with coating of the sensor

chip and for useful discussions.

References

Bostrom Caselunghe, M., Lindeberg, J., 2000. Biosensor-based de-

termination of folic acid in fortified food. Food Chem. 70, 523–

532.

Claycomb, R.W., Delwiche, M.J., 1998. Biosensor for on-line meas-

urement of bovine progesterone during milking. Biosens. Bio-

electron. 13, 1173–1180.

Claycomb, R.W., Delwiche, M.J., Munro, C.J., BonDurant, R.H.,

1998. Rapid enzyme immunoassay for measurement of bovine

progesterone. Biosens. Bioelectron. 13, 1165–1171.

Daly, S.J., Keating, G.J., Dillon, P.P., Manning, B.M., O’Kennedy,

R., Lee, H.A., Morgan, M.R., 2000. Development of surface

plasmon resonance-based immunoassay for aflatoxin B(1). J.

Agric. Food Chem. 48, 5097–5104.

Diskin, M.G., Sreenan, J.M., 2000. Expression and detection of

oestrus in cattle. Reprod. Nutr. Dev. 40, 481–491.

Elliott, C., Baxter, G., Crooks, S., McCaughey, W., 1999. The de-

velopment of a rapid immunobiosensor screening method for

the detection of residues of sulphadiazine. Food Agric. Immu-

nol. 11, 19–28.

Erlanger, B.F., Borek, F., Beiser, S.M., Lieberman, S., 1957. Ste-

roid–protein conjugates: I. Preparation and characterization of

conjugates of bovine serum albumin with testosterone and with

cortisone. J. Biol. Chem. 228, 127–713.

Gaudin, V., Maris, P., 2001. Development of a biosensor-based

immunoassay for screening of chloramphenicol residues in

milk. Food Agric. Immunol. 13, 77–86.

Groves, D.J., Sauer, M.J., Rayment, P., Foulkes, J.A., Morris, B.A.,

1990. The preparation of an ovine monoclonal antibody to pro-

gesterone. J. Endocrinol. 1990, 217–222.

Hashimoto, S., 2000. Principles of BIACORE. In: Nagata, K., Han-

da, H. (Eds.), Real-Time Analysis of Biomolecular Interac-

tions—Applications of BIACORE. Springer-Verlag, Tokyo, p.

22.

Indyk, H.E., Evans, E.A., Bostrom Caselunghe, M.C., Persson, B.S.,

Finglas, P.M.,Woollard, D.C., Filonzi, E.L., 2000. Determination

of biotin and folate in infant formula and milk by optical biosen-

sor-based immunoassay. J. AOAC 83, 1141–1148.

Johnsson, B., Lofas, S., Lindquist, G., 1991. Immobilisation of

proteins to carboxy methyl dextran-modified gold surface for

biospecific analysis in surface plasmon resonance sensors. Anal.

Biochem. 198, 268–277.

Koelsch, R.K., Aneshansley, D.J., Butler, W.R., 1994. Milk proges-

terone sensor for application with dairy cattle. J. Agric. Eng.

Res. 58, 115–120.

Markey, F., 2000. Principles of surface plasmon resonance. In: Na-

gata, K., Handa, H. (Eds.), Real-Time Analysis of Biomolecular

Interactions—Applications of BIACORE. Springer-Verlag,

Tokyo, p. 14.

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138 137

Mellgren, C., Sternesjo, A., Hammer, P., Bjork, L., Heeschen, W.,

1996. Comparison of biosensor, microbiological immunochem-

ical and physical methods for detection of sulfamethazine resi-

dues in raw milk. J. Food Prot. 58, 1223–1226.

Minunni, M., Mascini, M., 1993. Detection of pesticide in drinking

water using real-time biospecific interaction analysis (BIA).

Anal. Lett. 26, 1441–1460.

Mullett, W., Lai, E.P., Yeung, J.M., 1998. Immunoassay of fumo-

nisins by a surface plasmon resonance biosensor. Anal. Bio-

chem. 258, 161–167.

Nebel, R.L., 1988. On-farm milk progesterone tests. J. Dairy Sci.

71, 1682–1690.

O’Rorke, A., Kane, M.M., Gosling, J.P., Tallon, D.F., Fottrell, P.F.,

1994. Development and validation of a monoclonal antibody

enzyme immunoassay for measuring progesterone in saliva.

Clin. Chem. 40, 454–458.

Pemberton, R.M., Hart, J.P., Foulkes, J.A., 1998. Development of a

sensitive, selective electrochemical immunoassay for progester-

one in cow’s milk based on a disposable screen-printed ampero-

metric biosensor. Electrochim. Acta 43, 3567–3574.

Pemberton, R.M., Hart, J.P., Stoddard, P., Foulkes, J.A., 1999. A

comparison of 1-naphthyl phosphate and 4-aminophenyl phos-

phate as enzyme substrates for use with a screen-printed am-

perometric immunosensor for progesterone in cow’s milk.

Biosens. Bioelectron. 14, 495–503.

Pemberton, R.M., Hart, J.P., Mottram, T.T., 2001. An electrochem-

ical immunosensor for milk progesterone using a continuous

flow system. Biosens. Bioelectron. 16, 715–723.

Senger, P.L., 1994. The estrus detection problem: new concepts,

technologies, and possibilities. J. Dairy Sci. 77, 2745–2753.

Sreenan, J., Diskin, M., 1992. Breeding the Dairy Herd Teagasc,

Dublin.

Sternesjo, A., Mellgren, C., Bjorck, L., 1995. Determination of

sulfamethazine residues in milk by a surface plasmon reso-

nance-based biosensor assay. Anal. Biochem. 226, 175–181.

Wong, R.L., Mytych, D., Jacobs, S., Bordens, R., 1997. Validation

parameters for a novel biosensor assay which simultaneously

measures serum concentrations of a humanized monoclonal

antibody and detects induced antibodies. J. Immunol. Methods

209, 1–15.

E.H. Gillis et al. / Journal of Immunological Methods 267 (2002) 131–138138