Continuous-Flow System for Automation of Latex Immunoassay by Particle. Counting. Alfred M. Bernard and Robert. R Lauwerys'. Latex immunoassay.
CLIN. CHEM. 29/6, 1007-1011 (1983)
Continuous-Flow System for Automation of Latex Immunoassay by Particle Counting Alfred
M. Bernard and Robert
R Lauwerys’
Latex immunoassay is a nonisotopic method based on agglutination, by protein, of calibrated latex particles coated with a specific antibody. The assay has been automated in a
Materials and Methods
simple
Buffers. The glycine-buffered saline (GBS) used throughout the assay is prepared by diluting 10-fold a stock solution containing, per liter, 1 mol of glycine, 1.7 mol of NaCl, and 76 mmol of NaN3 and adjusting the pH to 9 with NaOH. The same stock buffer, but adjusted to pH 10.1, is also used for the stabilization of antibody-coated particles. All the buffers are ifitered through a 0.22-gm pore-size membrane (Millipore Corp., Bedford, MA 01730) and stored at 4 #{176}C. Standards. f32-Microglobulin and retinol-binding protein were purified as described previously (8). Human serum albumin was obtained from Fluka, Buchs, Switzerland, and ferritin (spleen) from Calbiochem Behring Corp., San Diego, CA 92112. Antibodies. Rabbit immunoglobulins against human ferritin, -microglobulin, retinol-binding protein, and albumin were obtained from Dako Immunoglobulin, Copenhagen, Denmark. The immunoglobulin concentration, determined by measuring the absorbance at 280 nm, ranged from 15 to 18 g/L. Latex. Polystyrene latex particles (0.8 m in diameter, 100 gIL, Estapor K109) were kindly supplied by Dr. J. C. Daniel (Rh#{244}ne-Poulenc, Aubervilliers, France).
continuous-flow
system
by incubating
the reaction
mixture in a heated mixing coil for 25 mm and measuring the agglutination with a cell counter. No external shaking of the latex suspension and no additional reagent is required for the agglutination. The method can accurately and precisely quantify a wide variety of proteins in plasma and urine, including human ferritin, f32-microglobulin, retinol-binding protein, and albumin. Depending on the antigen-antibody system, the detection limit ranges from 10b0 to about 10- 12 mol/L. Within- and between-assay CVs are 98 3.8 96-99 Sigma no. A 2lS3 Calbiochem no. 126601 e 5.0 >95 >95 5.5 Fluka no. 05480 #{149} Taken from manufacturers specifications. 5Drift observed with an anti-p2-microglobulincoated latex. Fatty acid free, no immunoglobulindetected by Ouchterlony technique. dSigma Chemical Co, St Louis, MO 63178.
a3
B
Eiectrophoretlc purity, %
Source
#{149}Prepared under nondenaturing conditions; native fatty acid profile reflecting endogenous lipids.
shown in Table 1, the quality of the BSA used as stabilizing agent. The best stability, as estimated from the drift over time of the zero-standard peaks, was observed with the most nearly pure BSA (Calbiochem no. 126609). With this preparation, drift in the zero standard peaks ranged from less than 1 to 2% per hour, depending on the antigen-antibody system. However, in the assay of urinary albumin we used BSA from Sigma or Fluka (Table 1), which showed the lowest cross reactivity with anti-human albumin antibody. The characteristics of the BSA used in the dilution buffer (GBSBSA buffer) have no influence on the stability of the particles. The stability of antibody-coated particles depends also on the characteristics of the container; they are much more stable in plastic vials than in glass vials. Standard curves. Figure 4 shows typical calibration curves for the automated determination by LIA of f3microglobulin, retinol-binding protein, ferritin, and albumin. These curves are composites of seven curves run successively during 2.5 h. Depending on the antigen-anti-
80
>< LU -J
60
g3
0 LU
-J I-
z I-
40
-J 0
0
z
-#{149}--
20#{176}C
---
37#{176}C
-.--
50#{176}C
>
( U.) -J
Within-run Mean (and
___________
10
100 -
RADIOIMMUNOASSAY
1.000
50.000
W.000
(pg/LI
Fig. 5. Correlation between automated LIA and radloimmunoassay for the determination of ferritin in 86 sera The value above 50 000 ug/L was not used for calculating the correlation coefficient. The standarddeviationof the slope is 0.04
1010
Urine, 1.tg/L
CLINICAL CHEMISTRY, Vol. 29, No. 6, 1983
18.9 74 217
Serum,
mg/L
SD)
(1.16) (2.38) (13.9)
448 (24.3) 1.38 (0.051)
Between-run CV, % 6.1 3.2 6.4 5.4 3.7
Urmnesdiluted 20-fold, serum 5000-fold; n
=
CV,% 21.6 (1.4) 6.5 77.4 (6.27) 8.1 212 (17.5) 8.3 434 (41.6) 9.6 1.35 (0.102) 7.5 10 at each concentration. Moan (and SD)
without agitation is far better than a shorter incubation with shaking (ms. in preparation). The validity of any immunoassay based on latex particle agglutination depends on the quality of the principal reagent, i.e., the antibody-coated particles. Several conditions regarding its agglutinability, specificity, stability, and reproducibility must be fulfilled. The ability of antibody-coated particles to be agglutinated by a given antigen-i.e., their agglutinability-depends on both the affinity of the antibody for that antigen and the amount of antibody adsorbed on the particles. The optimal antibody loading must be defined for each antibody-antigen system and may vary with the affinity and the titer of the antibody used. Also, there is an inverse relationship between the agglutinability of antibody-coated particles and their stability. For instance, if one increases the stability of the particles during the assay by increasing the pH or decreasing the ionic strength of the GBS-BSA buffer, the sensitivity of the assay is substantially reduced. The weakest point of LIA is certainly its specificity. Because no competitive antigen is used, the specificity of the assay depends entirely on the specificity of the antibody. Each time a new source of antibody is tested, its specificity must be carefully checked. The problem of specificity becomes particularly critical when one is analyzing trace proteins in serum. In the assay for ferritin, interferences from complement and rheumatoid factor were successfully eliminated by heat and reduction of the sample. The use of F(ab’)2-coated latex particles, proposed by Limet et al. (7), probably is also an efficient solution to these interferences. Interferences at low dilutions of serum may also appear when antibody-coated particles are very unstable during the assay (e.g., as a result of an excessive antibody loading). This interference stabilizes the particles, which apparently decreases their agglutination. This can be prevented by selecting appropriate stability conditions for the assay. A distinction must be made between the stability of antibody-coated particles during storage and their stability during assay. Although immunoglobulins are not covalently bound to the particles, antibody-coated latex is very stable during storage. For instance, we observed no marked change in the agglutinability of an anti-f3-microgiobulincoated latex preparation stored for seven months at 25 #{176}C; at 4 #{176}C, the shelf life of this reagent certainly exceeds one year. This high stability is achieved by storing antibody-coated latex at conditions of pH and ionic strength that preserve the affinity of the immunoglobulins for the latex surface (the pH of the stock preparation of antibody-coated latex particles is between 7 and 7.5). The tendency of immunoglobulinincubation
coated particles to agglutinate spontaneously during the assay is overcome by introducing negative charges on the latex surface, as described in Results. The use of excess BSA for this has the advantage of saturating the remaining protein adsorption capacity of the latex particles. The LIA method has been used in our laboratory for several years. We are convinced that antibody-coated latex is a reagent with sufficiently reproducible characteristics to have a great potential for use in the clinical laboratory. Because of its simplicity and low consumption of antibody (more than 10000 duplicate analyses with only 1 mL of Dako antibody), the automated nonisotopic assay described here is a practical and reliable method for the determination of trace proteins in biological fluids, if each application is carefully evaluated according to the above-mentioned criteria. We gratefully acknowledge the technical assistance of X. Dumont. We thank Dr. J. C. Daniel, Rh#{244}ne-Poulenc, France, for providing us with batches of polystyrene latex particles (ESTAPOR K 109). A. M. B. is a Charg#{233} de Recherches du Fonda National Beige de Ia Recherche Scientifique.
References 1. Bernard A, Vyskocil A, Lauwerys H. Determination of 132microglobulin in human urine and serum by latex immunoassay. Clin Chem 27, 832-837 (1981). 2. Bernard A, Moreau D, Lauwerys R. Latex immunoassay of retinol-binding protein. Clin Chem 28, 1167-1171 (1982). 3. Bernard A, Lauwerys R. Latex immunoassay of urinary albumin. J Clin Chem Clin Biochem 21, 25-30 (1983). 4. Canibiaso CL, Leak AE, De Steenwinkel F, et al. Particle counting immunoassay (PAcIA). I. A general method for the determination of antibodies, antigens and haptens. J Immunol Methods 18, 33-44 (1977). 5. Bernard A, Lauwerys R. Comparison of turbidimetry with particle counting for the determination of human 132-microglobulin by latex immunoassay (uA). Clin Chim Acta 119, 335-339 (1982). 6. Collet-Cassart D, Magnusson CGM, Ratcliffe JG, et al. A fully automated non radioisotopic assay for alpha-fetoprotein using the PACIA technique. Clin Chem 27, 64-67 (1981). 7. Limet J, Collet-Cassart D, Magnusson CGM, et al. Particle counting immunoassay of ferritin. J Clin Chem Clin Biochem 20, 141-144 (1982). 8. Bernard A, Lauwerys R, Starace V, Masson PL. Isolation of a new low molecular weight 13-globulin from urine of a worker with chrome cadmium poisoning. Biochem Biophys Res Commun 93, 535-543 (1980).
CLINICAL CHEMISTRY, Vol. 29, No. 6, 1983
1011
