Abstract This paper reports a new flow-through fluo-roimmunosensor, the function of which is based on anti-bodies immobilized on an inmunoreactor of controlled-pore glass (CPG), for determination of digoxin, used inthe treatment of congestive heart failure and artery disease.The immunosensor has a detection limit of 1.20 µg L–1 andprovides high reproducibility (RSD=4.5% for a concen-tration of 0.0025 mg L–1, and RSD=6.7% for 0.01 mg L–1).The optimum working concentration range was found to be1.2×10–3–4.0×10–2 mg L–1. The lifetime of the immunosen-sor was about 50 immunoassays; if stored unused its life-time can be extended to three months. A sample speed ofabout 10–12 samples per hour can be attained. Possibleinterference from substances with structures similar to di-goxin (morphine, heroin, tebaine, codeine, pentazocine andnarcotine) was investigated. No cross-reactivity was seenat the highest digoxin: interferent ratio studied (1:100).The proposed fluoroimmunosensor was successfully usedto determine digoxin concentrations in human serum sam-ples.

Keywords Flow-through fluoroimmunosensor · Digoxin ·Permanent immobilization

Introduction

Digoxin (3-[o-2,6-dideoxy-β- D -ribo-hexapyranosyl(1→4)-O–2,6-dideoxy-β- D -ribo-hexapyranosyl(1→4)-O–2,6-di-deoxy-β- D -ribo-hexapyranosyl)oxy]-12,14-dihidroxy(3β,2β,12β)-card-20(22) is a cardiac or digitalic glycoside withspecific effects on the myocardium. This drug, extracted

from the leaves of Digitalis lanatus, is used in the treat-ment of congestive heart failure to increase circulation. Itis also used in patients with atrial fibrillation and flutter toslow the ventricular rate. The measurement of serumdigoxin concentration is necessary owing to the narrowtherapeutic range of the drug; there is a thin line betweentherapeutic and toxic levels (0.05–0.2 µg L–1) [1].

Many techniques have been used to determine bloodconcentrations of digoxin. These include gas chromatog-raphy [2], radioimmunoassays [1], enzyme immunoassaytechniques [3, 4], fluorescence polarization immunoassay[5], and electrochemical methods [6]. Some are these meth-ods are relatively tedious, often requiring sample pre-treat-ment, phase separation or even special technology to in-crease their sensitivity and selectivity [1]. Many methodsfor digoxin determination are based on gas chromatogra-phy or radiochemical and enzymatic immunoassay tech-niques. However, some possible problems of these meth-ods include the labile nature of certain enzymes and diffi-culty of making enzyme–antibody conjugate and also thepresence of inhibitors: caused by substance depletion orproduct inhibition. Nowadays, they are being replaced bymore practical immunoassays techniques, involving, forexample, fluorescent immunosensors that allow rapid, au-tomated and selective digoxin determinations.

Different and very sensitive radioimmunoassays existfor the measurement of digoxin in serum [7]. However, ra-dioisotopes are now being replaced by fluorescent labels,such as fluorescent labeled antibodies, which allow sim-ple, rapid techniques that can be automated.

When applied to immunoassay, flow injection (FI) of-fers some advantages, such as precise control of reactiontimes, the re-use of supports and reagents, and improvedprecision. Sensitivity is similar to that obtained with batchmethods [8].

Previous works we developed fluoroimmunosensor fordifferent pesticides (triazines) [8] and isoproturon [9].

This paper presents the development and characteriza-tion of a flow-through fluoroimmunosensor for digoxin.This sensor uses an anti-digoxin polyclonal antibody im-mobilized on CPG in a direct competitive assay. Sensor

P. Fernández · J. S. Durand · C. Pérez-Conde ·G. Paniagua

Permanently oriented antibody immobilization for digoxin determinationwith a flow-through fluoroimmunosensor

Anal Bioanal Chem (2003) 375 : 1020–1023DOI 10.1007/s00216-002-1746-4

Received: 30 August 2002 / Revised: 13 November 2002 / Accepted: 4 December 2002 / Published online: 28 February 2003

SPECIAL ISSUE PAPER

P. Fernández (✉) · J. S. Durand · G. PaniaguaDpto. Ciencias Analíticas, Facultad de Ciencias, Universidad Nacional de Educación a Distancia (UNED), C/ Senda del Rey nº 9, 28040 Madrid, Spaine-mail: pfhernando@ccia.uned.es

C. Pérez-CondeDpto. Química Analítica, Facultad de Químicas, Universidad Complutense de Madrid, 28040 Madrid, Spain

© Springer-Verlag 2003

optimization and analytical aspects are addressed. Finally,the optimized sensor was used to determine the digoxinconcentrations of human serum samples.

Experimental

Apparatus and materials

Fluorescence intensity measurements were made on a Perkin–Elmer(Norwalk, CT, USA) LS5 spectrofluorimeter equipped with a 18 µLHellma (Jamaica, NY, USA) flow cell (optical path 1 cm) and anATT computer. All pH readings were made on a E516 TritrskoppH meter. A Gilson Minipulse 2 pump, Omnifit injection valve(six way) and PTFE tubes (0.5 mm id) were employed to build themanifold.

Chemicals and biochemicals

Digoxin sheep polyclonal antibody (3.6 g L–1) and digoxin labeledwith FITC (15 µmol L–1) were all purchased from Helena Bioscience(Sunderland, UK). Digoxin (95%) was purchased from Sigma–Aldrich (Madrid, Spain) which was immobilized on controlled-pore glass (of 37–74 µm particle size) supplied by Bio-Processing(Consett, Co Durham, UK). Phosphate buffer solution (PBS) pH 7.0was prepared by adding 8 g NaCl, 0.2 g KH2PO4, 0.2 g KCl, 2.9 gNaH2PO4, and 1 g MgCl2 (Merck, Darmstadt, Germany) to 1 Ldeionized water. Citric acid solution was 0.5 mol L–1 in 0.5 mol L–1

NaCl pH 3.0 (Merck) and (3-aminopropyl)triethoxysilane was fromMerck (Shuchardt, Germany). Deionized water from a NanopureSystem (Barnstead, UK) was used to prepare the solutions.

Procedures

Antibody immobilization by cross-linking to CPG

CPG was first alkylamined with (3-aminopropyl)triethoxysilaneaccording to the León-González and Townsend procedure [10] andthe periodate method of Wilson and Nakane [11]. Oxidized anti-body (500 µg) was allowed to react with 1.0 g of alkylaminedCPG. The product was incubated overnight at room temperature.The immobilization yield was determined by measuring the fluo-rescence emission of the antibody solution at 340 nm (λex=298 nm)before and after coupling.

System design

A diagram of the flow-through immunosensor manifold is shownin Fig. 1. The device consists of a peristaltic pump connected to asix-way valve with a 30 µL injection coil. The immunoreactor is aflow cell whose optical path is filled with CPG-antidigoxin placedin the spectrofluorimeter for “in situ” fluorescence detection. A fritwas placed in the optical path of the flow cell to prevent the carriersweeping the CPG-antibody away.

Sample preparation

Serum samples from digoxin-treated patients were supplied by thePuerta de Hierro Hospital, Madrid, diluted (1:1) with PBS and an-alyzed using the proposed assay protocol.

Assay protocol

The proposed method is based on the principle of a heterogeneouscompetitive fluorescence immunoassay, where the antibody is co-valently bound to the CPG. Solutions (0.2 µmol L–1, 30 µL) of

labeled antigen (digoxin-FITC) and unlabeled antigen (digoxin)are injected into the carrier solution at a 0.2 mL min–1 flow rate (10 mmol L–1 PBS, pH 7.2). Both antigens compete for the activesites of the antibodies, and antigen–antibody complexes are formed.

The excess antigen (labeled and unlabeled) was removed fromthe immunoreactor by the carrier solution. The fluorescence signalgenerated by labeled antigen–antibody complex formation wasmeasured “in situ” in the immunoreactor at λem=517 nm andλex=496 nm). Finally, a citric acid solution (0.5 mol L–1, pH 3.0)was pumped into the flow cell for immunoreactor regeneration. Acompetitive calibration curve was obtained by increasing the digoxinconcentration to 0.05 mg L–1.

Results and discussion

A suitable immunosensor should be able to perform analy-ses with high sensitivity, accuracy, speed, without interfer-ence, at low cost, and be of low maintenance. The choice ofoptimum assay conditions should take into account all theseconsiderations. In many cases, a compromise must beadopted between assay sensitivity, speed and reusability [12].

Immunosensor optimization and characterization

Cross-linking efficiency was evaluated by analyzing theamount of unbound antibody present in all fractions col-lected during the immobilization procedure by spectroflu-orimetry at 340 nm (λex=280 nm). In all cases, cross-linkingefficiency was greater than 70% (see Table 1 for a sum-mary of cross-linking efficiency).

The capacity of the immobilized antibodies to bind hap-ten depends of the amount of active sites available, that is,on good steric accessibility to active binding sites. Thiswas expressed as the maximum amount of hapten bound

1021

Fig. 1 Flow-injection system

Table 1 Efficiency of antibody (Ab) loading on the solid support

Immuno- µg Ac added/ µg Ac loaded/ % Efficiency sensor g CPG g CPG

1 180 128 71.1 2 360 292 81.1 3 900 794 88.2 4 1800 1607 89.3 5 3600 3101 86.1

to antibody immobilized on 1 g of CPG. For this purpose,30 µL of digoxin-FITC at different concentrations were in-jected and the fluorescence intensity of each monitored at517 nm (λex=496 nm).

A systematic study was made of the parameters affect-ing the antibody–hapten reaction on the CPG. The effectof binding solution flow rate (0.1–0.4 mL min–1) on theanalytical signal was also studied and 0.2 mL min–1 waschosen as the optimum carrier flow rate. The influence ofpH on the binding and regeneration solutions was investi-gated by varying the pH of both solutions within theranges 6.8 to 8.5 and 2.0 to 3.5, respectively. Optimum re-sults were obtained for pH 7.2 and pH 3.0.

Also to be optimized was the amount of antibody pre-sent in the reactor. This depends on the amount of CPG,on the antibody density described above, and the amountof antigen tracer used in each assay. Both parameters mustbe minimal if good assay sensitivity is to be attained, butthey should also be large enough to procedure acceptable

signals. Different competitive calibrations were tested us-ing antigen tracer within the range 0.2 to 0.05 µmol L–1

(for results see Fig. 2). The best compromise was found at0.2 µmol L–1 concentration of digoxin-tracer, the samevalue required to obtain an acceptable analytical signal.

Table 2 summarizes the optimum conditions for theimmunosensor.

Analytical performance

Competitive calibration curves were produced under opti-mum conditions using standards at concentrations rangingfrom 0 to 0.05 mg L–1. The normalized signals were plottedas (B/Bo) vs. digoxin concentration, where B is the peak ofthe fluorescent complex at different standard concentrationsof digoxin and Bo is the blank sample. The experimentalpoints were fitted to a quadratic polynomial equation.

The reproducibility of the method was tested by mea-suring two digoxin standard concentrations six times forseveral days. The relative standard deviation (RSD) ofnormalized signals for 0.0025 mg L–1 and 0.01 µg L–1

digoxin standards were 4.5% and 6.7% respectively. Re-producibility between calibrations was determined on twodifferent days. The graphs for these calibrations are al-most identical (Fig. 2). Statistical comparison showed nosignificant differences at the 95% confidence level.

The detection limit for this assay, expressed as the leastdetectable dose (LDD), was determined as the concentra-tion which provided twice the standard deviation from themean blank signal measurement. The LDD under optimumconditions was 1.20 µg L–1 digoxin. The assay dynamicrange (DR), defined as the analyte concentrations that in-hibited the maximum signal by 20% and 80%, was be-tween 1.20 × 10–3 and 4 ×10–2 mg L–1.

For interference studies, competition curves were pro-duced with digoxin-related (similar structures) compounds,and their corresponding I50 values determined. Cross-re-activity was then calculated as the ratio I50 digoxin/I50 re-lated compound (both I50 values expressed in µg L–1). The

1022

Fig. 2 Calibration curves for digoxin immunosensor carried outon different concentrations of antigen tracer

Fig. 3 Calibration curves for the digoxin immunosensor performedon two different days

Table 2 Summary of optimum conditions for the digoxin fluo-roimmunosensor

Parameter Optimum value

Excitation wavelength (Ag*) 496 nm Emission wavelength (Ag*) 517 nm Excitation and emission slits 0.5, 2.5 nm, respectively Binding solution 10 mmol PBS in

0.1 mol L–1 NaCl Binding solution pH 7.2 Regeneration solution 0.5 mol L–1 citric acid in

0.5 mol L–1 NaCl Regeneration solution pH 3.0 Binding flow rate 0.2 mL min–1

Antibody concentration 50 µL (3.6 g L–1) Labeled antigen concentration 0.2 µmol L–1

Fluorescence measurement time 300 s Cross-linking efficiency 96–99%

Ac: antibodyAg: antigenAg*: labeled antigen

compounds studied were narcotine, heroine, tebaine, mor-phine, codeine and pentazocine. Figure 3 shows calibra-tion curves for each compound at concentrations 100 timeshigher than the analyte. In all cases, the cross-reactivitywas less than 1%, which implies there is virtually no com-bination with the antibody over the range studied (Fig. 4).No interference in the assay is therefore expected for thisparticular antigen.

The reusability of the immunosensor prepared on a solidsupport is a major problem in sensor development. Underoptimum conditions, immunosurface activity remainedconstant for 50 assays. When not in use, the immunosen-sor was kept at 4 °C in PBS, and was operative for at least2 months. The total time required for each immunoassaywas 300 s.

Serum samples analysis

The proposed immunosensor was used to determine digoxinconcentrations in serum samples from four patients. Nosample treatment was necessary. The results, as well asthose obtained by the radiochemical method used at thepatient’s hospital, are shown in Table 3. No significant dif-ferences were seen between the values obtained by thesetwo methods (95% CI).

Conclusions

This paper proposes a fast and reliable flow-through fluo-roimmunosensor whose function is based on a heteroge-

neous competitive assay, to determine digoxin in serumsamples. The proposed sensor is thought to be highly se-lective; cross-reactivity of compounds with structures sim-ilar to digoxin was negligible. The described method hasadvantages over other existing chromatographic techniquesand allows sensitive and rapid sample assessment ofdigoxin without sample pre-treatment. Compared to otherimmunoassays for determining digoxin, it provides thebenefits of being more rapid and requiring minimal sam-ple handling.

The developed immunosensor was successfully used todetermine digoxin in human serum samples.

Acknowledgements This project was support by the Plan de Pro-moción-UNED. The authors wish to thank Adrian Burton for re-vising the manuscript and the Puerta de Hierro Hospital for sup-plying serum samples.

References

1. Chen I-W, Heminger LA (1987) Digoxin and Digitoxin. In:Pesce AJ, Kaplan LA (eds) Methods in clinical chemistry. CVMosby, St Louis, pp 897–902

2. Watson E, Kalman SM (1971) J Chromatogr 56:209–2183. Eriksen PB, Andersen O (1979) Clin Chem 23:1694. Kaneki N, Xu Y, Kumari A, Halsall H, Heineman WR,

Kissinger P (1994) Anal Chim Acta 287:253–2585. Leflar CC, Freytag JW, Powell LM (1984) Clin Chem 30:

1809–18116. Suzana T, Ikariyama Y, Aizawa M (1995) Bull Chem Soc Jpn

68:165–1717. Gortner L, Hellenbrecht D (1985) Clin Chem 31:155–1568. Turiel E, Fernández P, Pérez Conde C, Gutiérrez A, Cámara C

(1998) Talanta 47:1255–12619. Pulido-Tofiño P, Barrero-Moreno JM, Pérez-Conde MC (2000)

Anal Chim Acta 417:85–9410. León-González ME, Townshend A (1990) Anal Chim Acta

236:267–27211. Wilson NB, Nakane PK (1978) In: Knapp W, Holubar K, Wick

G (eds) Immunofluorescence and related stained techniques,Elsevier–North Holland, p 125

12. Gonzalez-Martinez MA, Morais S, Puchades R, Maqueira A,Abad A, Montoya A (1997) Anal Chim Acta 347:199–205

1023

Table 3 Results of human serum analysis

Sample Fluoroimmunosensor Reference method (X̄±SD) (mg L–1) (X) (mg L–1)

1 (4.05±0.10)×10–3 4.00×10–3

2 (2.34±0.27)×10–3 2.47×10–3

3 (1.43±0.51)×10–3 1.70×10–3

4 (4.43±0.09)×10–3 4.40×10–3

Fig. 4 Cross-reactivity ofdigoxin with the above inter-ferent substances