Characterization of Immunoglobulin G Bound to Latex Particles Using Surface Plasmon Resonance and Electrophoretic Mobility (original) (raw)

Characterization of Immunoglobulin G Bound to Latex Particles Using Surface Plasmon Resonance and Electrophoretic Mobility

J. L. Ortega-Vinuesa,* R. Hidalgo-Álvarez,* 11{ }^{11} F. J. de las Nieves, †\dagger C. L. Davey, ‡\ddagger D. J. Newman, ‡\ddagger and C. P. Price ‡\ddagger
*Biocolloid and Fluid Physics Group, Department of Applied Physics, Faculty of Sciences, University of Granada, Fuentenueva S/N, 18071 Granada, Spain; †Group of Physics of Complex Fluids, Department of Applied Physics, University of Almería, Cañada de San Urbano, 04120-E, Almería, Spain; and ‡\ddagger Department of Clinical Biochemistry of the St Bartholomew’s and the Royal London School of Medicine and Dentistry, Turner Street, London E1 2AD, United Kingdom

Received August 4, 1997; accepted April 14, 1998

Abstract

The main objective of this work was the investigation of passive adsorption and covalent coupling of a polyclonal IgG and a monoclonal preparation of IgG against HSA, to a carboxyl latex particle. The functional activity of the coupled protein was then assessed by quantitative immunoassays for the antigen. Sensitized particles, with different protein coverage, were fully characterized using a range of different technologies, including electrophoretic mobility (μe)\left(\mu_{\mathrm{e}}\right), photon correlation spectroscopy, and surface plasmon resonance (SPR). The antibody-labeled particles were studied with respect to electrokinetic behavior in pH and ionic strength titration, stability, antibody functionality, and their perfomance in immunoaggregation reactions. Important differences were observed between the two sets of particle preparations throughout the series of experiments. The differences could be attributed to the coupling of the IgG molecules to the particles by the two different adsorption protocols. When proteins were chemically bound to the polymer surface it was necessary to activate the carboxyl groups with a carbodiimide (CDI) moiety that in our case was positively charged. The differences in characteristics between the adsorbed and the coupled antibody particles are thought to be due to the fact that in the covalent coupling protocol some CDI molecules remained linked to the particles, which altered the average electrical state of the outer layer in comparison with those samples where antibodies were physically adsorbed. On the other hand, the isoelectric point of the monoclonal antibody was lower (5.4±0.1)(5.4 \pm 0.1) than the pJ of the polyclonal antisera (6.9±0.9)(6.9 \pm 0.9), which could explain why the IgG-latex complexes created with monoclonal molecules were colloidally more stable at neutral pH than those created with the polyclonal antisera. However, no immunoaggregation of antibody particles by the presence of antigen was found with the former. The use of SPR demonstrated that the equilibrium constants for the antibody-antigen recognition of the two antibody preparations were quite similar ( KA polyclonal tgG⁡=K_{\text {A polyclonal } \operatorname{tgG}}= 2.8106M−1;KA monoclonal tgG⁡=9.5107M−12.810^{6} \mathrm{M}^{-1} ; K_{\text {A monoclonal } \operatorname{tgG}}=9.510^{7} \mathrm{M}^{-1} ). These observations suggest that the lack of aggregation mediated by antigen demonstrated by the monoclonal antibody coupled to the latex particles may be due to this protein recognizing only one epitope in the HSA molecule. However, as the repulsive charge between antibodylatex particles counteracts the attractive forces between the antigen epitope and the antibody paratope, the greatest immunoag-

gregation was obtained when using latex particle-antibody complex with a low charge density (N)(N) in the external layer. © 1998 Academic Press

Key Words: polymer colloids; antibodies; particle-enhanced immunoassays; surface plasmon resonance; electrophoretic mobility; colloidal stability; latex.

INTRODUCTION

Light scattering immunoassays commonly employ colloidal particles to enhance sensitivity and extend the point of equivalence; they are based on the aggregation of polymeric colloidal microspheres (latex) that act as a carrier of antibody (or antigen) molecules. The agglutination process is initiated by the presence of the complementary antigen molecule (or antibody), respectively. The coupling of protein to this kind of colloidal particle, whether an antibody or an antigen, in order to develop diagnostic tests is an important area of research, which has been widely studied over the past 40 years (1-4). The popularity of this diagnostic technology is illustrated by the fact that in 1992 there were over 200 commercial reagents available employing this approach. The technology is simple to use, widely applicable, and nonhazardous and perhaps most importantly can be applied, by the virtue of the homogeneous format, to the measurement, with short reaction time, of many important analytes on routine automated clinical analyzers. The major performance characteristics of these assays, namely reproducibility, detection limit, analytical range, and reliability, depend on three important aspects that must be optimized: (i) the nature and characteristics of the antibody (or antigen)latex complexes; (ii) the nature of the reaction environment in which the immunoaggregation takes place; and (iii) the optical method used to detect the extent of immunoaggregate formation. The latter has been explored by several authors (5-8), who used light scattering theory to achieve major advances in the development and improvement of optical devices. Despite the improvements made in analytical performance through the use of improved optical instrumentation, less consideration has

been given to the theoretical aspects of optimization of reagents with empirical considerations often determining optimization strategies. However, there are now a range of analytical techniques, such as surface plasmon resonance (SPR) and photon correlation spectroscopy (PCS), which can provide an opportunity to characterize assay components more fully and compare both the functional and the structural aspects of pro-tein-particle interactions.

In this study the characterization was focused on the influence of binding of monoclonal and polyclonal preparations of immunoglobulin G to latex particles by either adsorption and covalent coupling using SPR and a functional assay while the surface characteristics of the particles were studied using electrokinetic and PCS technologies.

Previous work has identified important observations that must be taken into account when using reagents prepared by adsorption of antibodies onto colloidal particles. First, colloidal instability of latex particles covered by polyclonal IgG has been observed by different authors ( 9,10 ). This self-aggregation process is undesirable, as it is caused by the physicochemical conditions of the milieu and not by the presence of antigen-antibody binding. This is why different strategies have been proposed to increase the stability of the IgG-latex systems; most are based on the coadsorption of IgG and other macromolecules capable of acting as a stabilizer, such as surfactants (11), lipids (12), albumin (13, 14). However, it has been shown that working with monoclonal antibodies (15), or even with IgG fragments (i.e., F(ab′)2\mathrm{F}\left(\mathrm{ab}^{\prime}\right)_{2} or Fab) instead of the whole IgG molecule, can increase the stability of the system. This is true only when the isoelectric point of the monoclonal protein is far from the medium pH (15), since when IEP of the monoclonal IgG coincides with the medium pH the adsorbed protein layer would be uncharged and thus the system would become colloidally unstable. Secondly, polymer engineering has facilitated the synthesis of latex particles with a surface reactive group that enables covalent coupling of protein molecules to the particle. It has been shown that covalent linkage can increase the stability of a reagent minimizing the protein desorption that is seen when proteins are adsorbed onto the particles (16). Furthermore, covalent coupling is thought to improve orientation of antibody or ligand molecules and reduce the amount of adsorbed protein, a process thought to result in greater conformational change. In addition, the use of spacers between the particle and the immunoprotein can lead to more reactive complexes (17), as the IgG is made more available to react with its complementary antigen in the bulk of the reaction environment.

Taking into account these previous observations the work was designed to investigate the differences seen in the characteristics and functionality of both polyclonal and monoclonal antibody preparations for both adsorption and covalent coupling to latex particles. The various options were assessed using the following techniques:
(a) Electrophoretic mobility measurements versus both pH and ionic strength. From these data it is possible to determine the isoelectric point of the complexes, and estimate the average charge state of the adsorbed protein layer, respectively.
(b) PCS, which enables calculation of the average size of particles or aggregates. Therefore, this technique can be used to study the stability of the colloidal system in a saline solution.
© An immunoturbidimetric technique that enables detection of an antigen in an immunoaggregation reaction, i.e., assesses the functionality of the coupled antibody.
(d) SPR to obtain further detailed functional information about the antibody molecules, such as binding characteristics, affinities, and kinetic constants, with regard to the antigen union.

MATERIAL AND METHODS

Latex Particles

To prepare carboxyl latex the steps developed by Guthrie (18) for surfactant-free emulsion polymerization of styrene were followed. Styrene monomer was obtained from Merck (Darmstadt, Germany) and was distilled at low pressure at 40∘C40^{\circ} \mathrm{C}. The purified monomer was stored at −5∘C-5^{\circ} \mathrm{C} until required. The initiator used in this work was the 4,4′4,4^{\prime}-azobis(4cyanopentanoic acid) (ACPA) from Aldrich (Sigma Chemical Company, St Louis, MO) and was used as received. Sodium hydroxide (Merck) was an analytical grade reagent and was also used without further purification. Double-distilled and deionized water was used throughout. It should be noted that the use of ACPA as initiator has the advantage that carboxyl terminal groups on the particle surface come from the initiator molecules, so the use of a second (or more) co-monomer(s) was avoided. An aliquot of water ( 720 ml ) was poured into a 1000−ml1000-\mathrm{ml} three-necked glass flask, followed by 25 g of styrene, 372 mg of ACPA, and 83 mg of sodium hydroxide. Polymerization was carried out for 7 h at 80∘C80^{\circ} \mathrm{C}. In order to maintain a continuous vigorous stirring ( 350 rpm ), a T-shaped stirrer ( 1×1 \times 5 cm ) was fitted 1 cm from the bottom of the vessel.

After the synthesized latex was cleaned with several washing and centrifugation cycles, the main characteristics of the sample were determined. Solid content of the stock was 5.65%(w/w)5.65 \%(\mathrm{w} / \mathrm{w}). The average size of the particles was obtained using two different methods: (i) transmission electron microscopy (TEM), and (ii) photon correlation spectroscopy, yielding very similar results: 264±8264 \pm 8 and 273±273 \pm 5 nm , respectively. The sample was highly monodispersed, its polydispersity index being 1.003±0.0011.003 \pm 0.001 by TEM. The surface charge density was calculated from acid-base titration data, and its value was −(16.2±0.3)μC/cm2-(16.2 \pm 0.3) \mu \mathrm{C} / \mathrm{cm}^{2}. The latex preparation used in all subsequent experiments will be referred to as JL3.

Proteins

The IgG samples were purchased from two different firms. An anti-human serum albumin (aHSA) goat-polyclonal antibody was purchased from Atlantic Antibodies, and in this work it will referred to as Pab. The second one was a protein A-purified HSA mouse-monoclonal IgG from Dade International (previously Du Pont USA, Glasgow, UK) (referred to as Mab). Both were used without further purification. The HSA used as antigen was from Behring (Marburg, Germany), and the bovine serum albumin (BSA) used in some experiments was from Sigma (Poole, Dorset). The IEP of both antibodies were determined by an isoelectric focusing (IEF) method performed with a Pharmacia Phast system (Uppsala, Sweden), and using a Phast Gel-1 IEF polyacrylamide medium covering the pH range 3-9. The IEP of the monoclonal antibody was lower (5.4±0.1)(5.4 \pm 0.1) than the IEP of the polyclonal antisera (6.9±0.9)(6.9 \pm 0.9).

Antibody Adsorption

The choice of protein loading was based on previous experience of the stability of particles with passively adsorbed antibodies. At neutral pH , the colloidal stability usually decreases very rapidly as IgG coverage increases. It has been found that 5.0 mg of IgG/m2\mathrm{IgG} / \mathrm{m}^{2} of polystyrene caused spontaneous aggregation of the system (15). Therefore, 1.0mg/m21.0 \mathrm{mg} / \mathrm{m}^{2} was chosen for the samples created by physical adsorption of antibodies. However, protein loading which may cause spontaneous agglutination of particles with covalently attached IgG had not been determined. Therefore, concentrations of 0.5,1.00.5,1.0, and 2.0mg/m22.0 \mathrm{mg} / \mathrm{m}^{2} were sought for the covalent coupling experiments. In both sets of experiments a blank was run, that is, particles without any protein, referred to as JL3.

Passive adsorption. A volume of 255μl255 \mu \mathrm{l} of the stock latex was added to 4.75 ml of protein solution in 15 mM borate buffer ( pH 8.0 , ionic strength equal to 0.002 M ); this should give a final concentration of 1.0mg/m21.0 \mathrm{mg} / \mathrm{m}^{2}. The samples were incubated and continuously shaken for 4 h at room temperature. Then, samples were spun, and the concentration of the protein that remained in supernatants was determined spectrophotometrically at 280 nm(ΣIgG=1.4mgml−1 cm−1)280 \mathrm{~nm}\left(\Sigma_{\mathrm{IgG}}=1.4 \mathrm{mg} \mathrm{ml}^{-1} \mathrm{~cm}^{-1}\right) (19, 20). Precipitates were redispersed in 5 ml of the same buffer used above. Final degrees of IgG coverage for both polyclonal and monoclonal antibody adsorption are shown in Table 1, namely “Pab ads” and “Mab ads.”

Covalent coupling. The covalent binding was also carried out at pH 8.0 . In order to achieve a chemical link between protein and polymer surface, the carboxyl groups of latex had to be activated; the carbodiimide (CDI) method was used for this purpose. The protocol has been described in detail elsewhere (21, 22). However, it should be noted that the CDI that was used, namely 1 ethyl-3-(3-dimethyl aminopropyl) carbodiimide, transforms a carboxyl group into a positively charged chemical group. It is also worth mentioning that, after centrifug

TABLE 1
Amounts of Adsorbed IgG in JL3 Particles Employing Two Methods for Quantification of Protein

Sample Jads 280 nm (mg/m2)\left(\mathrm{mg} / \mathrm{m}^{2}\right) Jads BCA (mg/m2)\left(\mathrm{mg} / \mathrm{m}^{2}\right) J ads (mg/m2)\left(\mathrm{mg} / \mathrm{m}^{2}\right) (Error <0.1mg/m2<0.1 \mathrm{mg} / \mathrm{m}^{2} )
Pab ads 1,03 - 1,0
Mab ads 1,26 - 1,3
Pab 0.5 0,48 0,55 0,5
Pab 1.0 1,06 1,12 1,1
Pab 2.0 1,81 1,78 1,8
Mab 0.5 0,51 0,40 0,5
Mab 1.0 0,78 0,60 0,7
Mab 2.0 1,53 1,46 1,5

Note. Pab and Mab refers to polyclonal and monoclonal antibodies, respectively. Samples with IgG passively adsorbed are denoted with the “ads” term; complexes with IgG chemically attached are denoted with a number (related to the desired adsorbed amount).
ugation, precipitates were first redispersed for 10 min in 2 ml of a glycine solution (0.1M)(0.1 \mathrm{M}) to block those activated carboxyl groups that had not reacted with amino groups on the protein molecules. To calculate the efficiency of the covalent coupling, 5 ml of a Tween 20 solution ( 1%1 \% ) was added to remove the physically adsorbed antibodies. This incubation lasted 14 h , and after this samples were again spun, precipitates redispersed in 5 ml of borate buffer ( pH 8.0 ), supernatants were filtered, and protein concentration was measured using two different methods: (i) direct UV spectrophotometry at 280 nm and (ii) the bicinchoninic acid reaction (BCA method, Pierce Reagent); details are given in Ref. (22). The BCA method was used to confirm the results obtained by the UV spectrophotometry measurements, as the presence of Tween 20 may cause some error, since it also adsorbs at 280 nm . Final protein loadings are also shown in Table 1. As can be seen both methods gave quite similar results, indicating that Tween 20 did not interfere in the UV calculations.

Electrophoretic Mobility Measurements

A 7−μ17-\mu 1 aliquot of the above IgG-latex stocks was diluted in 8 ml of the desired buffered solution 20 min before the electrophoretic mobility was measured. All the buffers used for the study of the effect of the pH were of low ionic strength, equal to 0.002M;13.5mM0.002 \mathrm{M} ; 13.5 \mathrm{mM} acetic acid for pH4.0,3.15mM\mathrm{pH} 4.0,3.15 \mathrm{mM} acetic acid for pH5.0,1.79mMNaH2PO4\mathrm{pH} 5.0,1.79 \mathrm{mM} \mathrm{NaH}_{2} \mathrm{PO}_{4} for pH6.0,1.13mM\mathrm{pH} 6.0,1.13 \mathrm{mM} NaH2PO4\mathrm{NaH}_{2} \mathrm{PO}_{4} for pH 7.0 , and 15.0 mM boric acid for pH 8.0 . Buffers for the study of the effect of the ionic strength were prepared as follows. Different volumes of buffer pH 8.0 were mixed with different volumes of a 0.4 M KBr solution to give varying ionic strengths from 0.002 to 0.11 M . All reagents were of analytical grade. The measurements were carried out with a Zeta-Sizer IV (Malvern Instruments, Worcs., UK), by taking the average of three data readings at the stationary level in a cylindrical cell.

Stability

Colloidal stability of the systems was studied by two methods. First, the critical coagulation concentration (CCC), at which the particles rapidly coagulate in the presence of a salt (KBr)(\mathrm{KBr}) was determined. Experimental details have been described previously (21). It was important to know the stability of the reagents in the medium in which immunoassays were to be carried out, i.e., a saline borate BSA buffer. Therefore, the average size of particles (or aggregates) was determined by means of photon correlation spectroscopy. Data were acquired using a commercial device, the 4700c System (Malvern Instruments). This instrument incorporates an Argon laser (λo=488\left(\lambda_{o}=488\right. nm ), with 75 mW of power. The photomultiplier was set at a 60∘60^{\circ} position. Before sizes were measured, samples were highly diluted ( 7:40000,v/v7: 40000, \mathrm{v} / \mathrm{v} ) and incubated for 2 days in saline BSA borate which was the buffer used in the immunoassay experiments detailed below. Each value was obtained after the diluted sample was measured continuously for 1 h . As we have just mentioned, the measurement of particles sizes was performed after a long time interval ( 48 h ); the choice of this time is justified by the fact that samples were extremely diluted (a necessary condition to use the 4700 System instrument). Therefore, long incubation time is needed to form aggregates in case of any colloidal instability.

Immunoassays

Most of the assays were carried out in saline BSA borate buffer ( pH 8.0 borate ( 15 mM ), NaCl(150mM),NaN3\mathrm{NaCl}(150 \mathrm{mM}), \mathrm{NaN}_{3} as preservative ( 1.0mg/ml1.0 \mathrm{mg} / \mathrm{ml} ), and bovine serum albumin ( 1.0mg/1.0 \mathrm{mg} / ml ). The BSA blocks any bare surface of the latex particles in order to avoid bridging between IgG molecules and latex particles to form an aggregate; in addition, it improves the colloidal stability of the antibody latex particles as shown in previous studies (13, 14). The antibody latex stock particle reagent was diluted in saline BSA borate buffer to provide a working latex reagent. A series of solutions of HSA was prepared in saline BSA over a concentration of HSA 0.008−80.008-8 mg/L\mathrm{mg} / \mathrm{L}. It had been previously shown that neither of the antibody preparations employed in this work reacted with BSA.

Turbidimetric progress curves of absorbance over time were obtained using a Monarch centrifugal analyzer (Instrumentation Laboratory, Warrington, UK). Measurements of absorbance were made at 570 nm , for 10 min and 37∘C37^{\circ} \mathrm{C}, after mixing 50μl50 \mu \mathrm{l} of the antigen solution (at different concentrations) with 150μl150 \mu \mathrm{l} of the particle solution. The final particle concentration in the cuvette was 1.8×10101.8 \times 10^{10} particles /ml/ \mathrm{ml} throughout the series of experiments.

Antibody-Antigen Affinity

Measurements were made on a BIAcore surface plasmon resonance immunosensor (Biacore, Steverage, UK). All experiments were carried out using a running buffer of 10 mM

Hepes-buffered saline (HBS), at 5μl/min5 \mu \mathrm{l} / \mathrm{min} flow rate. The HBS included NaCl ( 150 mM ), ethylenediamine tetracetic acid (EDTA, 3.4 mM ), and P20 nonionic surfactant ( 0.05%0.05 \% ). The carboxymethyl dextran-modified gold sensor chips used in the BIAcore instrument were modified with HSA molecules attached on their surfaces, using an amine coupling kit from BIAcore.

The gold sensor chip was preactivated with 35μl35 \mu \mathrm{l} of a solution of 0.2 M CDI and 0.05 M N -hydroxylsuccinimide (NHS); 35μl35 \mu \mathrm{l} of 100 mM HSA in 10 mM sodium formate pH 3.6 was injected, followed by 35μl35 \mu \mathrm{l} of 1 M ethanolamine hydrochloride, pH 8.3 , to block the remaining NHS ester groups. Then, 10μl10 \mu \mathrm{l} of 50mMNaOH/20%50 \mathrm{mM} \mathrm{NaOH} / 20 \% acetonitrile was injected to remove all the noncovalently bound HSA. Regeneration of HSA-labeled sensor chip was accomplished by injection of 10 μl\mu \mathrm{l} of 50mMNaOH/20%50 \mathrm{mM} \mathrm{NaOH} / 20 \% acetonitrile, followed by reequilibration with HBS solution. After the gold sensor chip was equilibrated (previous sensitized by HSA) with HBS for 2 min , diluted solutions of IgG-JL3 particles (1:100) were poured through the cell for 4 min ; this does not mention regeneration. After that, pure HBS was injected again to minimize rebinding to facilitate measurement of kdiss k_{\text {diss }}. The antibody affinity was determined by passing five concentrations (in duplicate) of antibody over the HSA-sensitized chip. After each injection the surface was regenerated using the NaOH/\mathrm{NaOH} / acetonitrile procedure.

RESULTS AND DISCUSSION

Antibody Adsorption

The particle protein coverage data have already been shown in Table 1; it is worth noting the high agreement given by the two analyses used in the covalently coupled samples, indicating that Tween 20 did not interfere in the UV spectrophotometry measurements (provided that a blank is used). Also, it should be noted that final adsorbed amounts do not always coincide with those values sought. Thus the different particle preparations are denoted as Pab0.5,1.1\mathrm{Pab} 0.5,1.1, and 1.8 , and Mab 0.5 , 0.7 , and 1.5 as these were the protein loading calculated, whereas values of 0.5,1.00.5,1.0, and 2.0mg/m22.0 \mathrm{mg} / \mathrm{m}^{2} were actually sought.

As adsorption pH was equal to 8 , both protein molecules and particle surfaces were negatively charged, so adsorption was expected to be governed mainly by hydrophobic interactions (23).

Electrokinetic Behavior

The effect of the pH. The electrophoretic mobility (μe)\left(\mu_{\mathrm{e}}\right) was determined for each of the particle preparations in buffers from pH 4.0 to pH 8.0 in order to determine the isoelectric point (IEP) of every complex, as a tool to understand and explain the stability results. In addition, it was also possible to confirm from these measurements that the IEP of the monoclonal

img-0.jpeg

FIG. 1. Electrophoretic mobility versus pH for ( ()\mathbf{(}) bare latex, ( )\boldsymbol{)} ) Pab ads (1.0mg/m2)\left(1.0 \mathrm{mg} / \mathrm{m}^{2}\right), ( ()\mathbf{(}) ) Mab ads (1.3mg/m2)\left(1.3 \mathrm{mg} / \mathrm{m}^{2}\right), ( △\triangle ) bare latex treated with the CDI method, (○) Pab 1.8 complex, and ( □\square ) Mab 1.5 complex particles.
antibody was lower than the average IEP of the polyclonal IgG, as will be shown later. Figure 1 shows representative results of the pH titration for the passively adsorbed and covalently attached IgG preparations. At first glance, it is surprising that both sets of blanks behave quite differently. By extrapolation, the IEP of the bare latex would appear to be less than 3.0 (see Fig. 1), whereas the JL3 particles in the covalent series have a IEP around 4.7. Such an important difference can be attributed to the coupling method used to link proteins to the polymer surface. Thus the carboxyl groups on the particle surface were activated by CDI molecules, which exchanged every negatively charged group into a positively charged one. Although all of these activated groups should have reacted with either protein or glycine molecules, it seems likely that some of them that remain as positive organic groups. We have previously demonstrated (21) that a carboxyl group activated by the CDI method is not as reactive as might be expected. In the same work we also showed that activation by CDI affects the overall charge of the latex particle, due to the presence of attached CDI molecules on the surface. It may be almost impossible to completely remove CDI molecules, even when incubating the activated samples in an ethanolamine solution for over 15 h (21). Therefore, the covalent coupling procedure leaves a certain number of positively charged groups, together with the carboxyl groups placed in the particles, thereby explaining why both protein-free latex particles differ in their electrokinetic behavior.

It is a well-known feature that when a protein covers colloidal particles, the IEP of such complexes gradually tends to the IEP of the pure biomolecule when increasing the protein loading on the particles (24-27). In the “physically adsorbed” samples the IEP of those microspheres having monoclonal antibodies is slightly lower than the particles covered by the polyclonal antisera, even though the protein loading in the former is higher than for the latter. In the “chemically bound”
case the difference in IEP is much clearer, the Mab sample having a value very near that of the blank. All these data agree qualitatively with the isoelectric points of the monoclonal and polyclonal IgG.

The effect of the ionic strength. The titrations were undertaken in pH 8.0 buffers of varying ionic strengths from 0.002 to 0.11 . Results obtained with the Pab samples are depicted in Fig. 2, together with results obtained with the Mab 1.5 complex. According to the Ohshima and Kondo theory, fitting the experimental μe\mu_{\mathrm{e}} data of the protein-latex particles to theoretical predictions, one can obtain valuable information about some of the characteristics of the adsorbed protein layer, such as thickness, charge density, and frictional coefficient. The above authors give different expressions for the electrophoretic mobility of colloidal particles covered by surface charged layers (28-30). All of them are dependent on the adsorbed polyelectrolyte layer thickness (d)(d), its charge density (N)(N), and a term (λ)(\lambda) directly related to the frictional coefficient of such a layer. In order to obtain these values from our systems, we decided to use the following theoretical equation (28);

μe=−ϵrϵdkTve⁡η[ln⁡(∣σ∣2 vend +[(σ2 vend )2+1]1/2)]+ϵrϵdkTve⁡η×([(2 vend σ)2+1]1/2−2 vend ∣σ∣)×(κλ)[1+(σ/2 vend )2]1/4tanh⁡λd−1(κ/λ)2[1+(σ/2 vend )2]1/2−1−∣σ∣ηdλ2(1−1cosh⁡λd)\begin{aligned} \mu_{\mathrm{e}}= & -\frac{\epsilon_{\mathrm{r}} \epsilon_{\mathrm{d}} k T}{\operatorname{ve} \eta}\left[\ln \left(\frac{|\sigma|}{2 \text { vend }}+\left[\left(\frac{\sigma}{2 \text { vend }}\right)^{2}+1\right]^{1 / 2}\right)\right]+\frac{\epsilon_{\mathrm{r}} \epsilon_{\mathrm{d}} k T}{\operatorname{ve} \eta} \\ & \times\left(\left[\left(\frac{2 \text { vend }}{\sigma}\right)^{2}+1\right]^{1 / 2}-\frac{2 \text { vend }}{|\sigma|}\right) \\ & \times \frac{(\kappa \lambda)\left[1+(\sigma / 2 \text { vend })^{2}\right]^{1 / 4} \tanh \lambda d-1}{(\kappa / \lambda)^{2}\left[1+(\sigma / 2 \text { vend })^{2}\right]^{1 / 2}-1} \\ & -\frac{|\sigma|}{\eta d \lambda^{2}}\left(1-\frac{1}{\cosh \lambda d}\right) \end{aligned}

which is useful when the external polyelectrolyte layer is
img-1.jpeg

FIG. 2. Electrophoretic mobility versus ionic strength for (X) bare latex, ( ⊙\odot ) Pab ads (1.0mg/m2)\left(1.0 \mathrm{mg} / \mathrm{m}^{2}\right), ( ()\mathbf{(}) Pab 0.5 , ( )\boldsymbol{)} ) Pab 1.1, ( ()\mathbf{(}) Pab 1.8, and ( □\square ) Mab 1.5 reagents. Solid lines represents the theoretical μe\mu_{\mathrm{e}} trends given by Eq. [1] using the following values: d=100A˚,λ−1=3.3 nmd=100 \AA, \lambda^{-1}=3.3 \mathrm{~nm}, and N=25,15,12,10N=25,15,12,10, and 14 mM , respectively.

negatively charged, and the colloidal particles are moving in a liquid containing a symmetrical electrolyte of valence " vv " and bulk concentration " nn ". In the above equation, " ϵr\epsilon_{\mathrm{r}} " is the dielectric constant of electrolyte solution, " ϵ0\epsilon_{0} " is the permittivity of the vacuum, " kk " is the Boltzman constant, " TT " the absolute temperature, " η\eta " is the viscosity of the medium, " σ\sigma " is the amount of fixed charges contained in the surface charged layer per unit area ( σ=−eNd\sigma=-e N d ), " κ\kappa " the DebyeHückel parameter of the electric double layer, and " λ\lambda " is a hydrodynamic parameter equal to

λ=(γη)1/2\lambda=\left(\frac{\gamma}{\eta}\right)^{1 / 2}

where " η\eta " and " γ\gamma " are the viscosity of the fluid and the frictional coefficient of the adsorbed polyelectrolyte layer, respectively. λ\lambda is also related to the degree of flow penetration into the adsorbed protein layer. λ−1\lambda^{-1}, which has dimensions of length, can be used as a measure of flow penetration (30). The choice of Expression [1] is not arbitrary; it is based on a very detailed study (31) performed with carboxyl latex covered with F(ab′)2\mathrm{F}\left(\mathrm{ab}^{\prime}\right)_{2}. From our previous knowledge and experience (22,31)(22,31), we took a constant value of 3.3 nm for the λ−1\lambda^{-1} parameter and decided to use the charge density of the outer polyelectrolyte (protein) layer as the fitting parameter for our experimental results. The thickness of the IgG layer was also chosen to remain constant, " dd " being equal to 100A˚(22,30)100 \AA(22,30). The theoretical curves obtained by Eq. [1] are also shown in Fig. 2. As can be seen, the NN values that best fit the experimental μe\mu_{\mathrm{e}} data are 25 mM (for Pab ads complex), 15 mM (Pab 0.5), 12 mM (Pab 1.1), 10 mM (Pab 1.8), and 14 mM (Mab 1.5).

The great difference between the “physically adsorbed” sample and the “chemically bound” complexes is caused by the presence of attached CDI molecules, which drastically reduces the charge density of the external layer for the latter particle preparations. It is worth noting that the NN value decreases for increasing protein loading, a result that has also been found previously (22). These data are consistent with the idea of polyclonal IgG molecules having a low charge at pH 8 . In addition, comparing the NN values for the Mab 1.5 and the Pab complexes (mainly with the Pab 1.8, as it has similar IgG coverage to that of Mab 1.5), one can conclude that the monoclonal antibody-conjugated particles are more charged than the polyclonal species at pH 8.0 .

Stability

Most of the assays based on latex technology are carried out in media where ionic strength ( II ) and pH are similar to those of the physiological fluids, that is pH around 7.2 and I≈0.16I \approx 0.16. However, in most instances colloidal particles only covered by IgG are unstable under these physicochemical conditions (10, 11) and to employ these particles in a light scattering immunoassay would be inappropriate. To avoid erroneous diagnosis,

TABLE 2
Data Indicating the Stability of the Antibody Latex Preparations Employing Two Techniques: (i) Assessment of “Critical Coagulation Concentration” (CCC) Using Ionic Strength Titration, and (ii) Size of Complexes after a 2-Day Incubation in Saline BSA Borate Buffer by Photon Correlation Spectroscopy

Sample Average size in saline BSA (nm) CCC (mM of KBr)
JL-3 269±6269 \pm 6 95050
Pab ads (1.0) 280±7280 \pm 7 40050
Mab ads (1.3) 263±3263 \pm 3 >1150>1150
Pab 0.5 274±12274 \pm 12 -
Pab 1.1 370±40370 \pm 40 -
Pab 1.8 2300±10002300 \pm 1000 -
Mab 0.5 271±7271 \pm 7 -
Mab 0.7 279±9279 \pm 9 -
Mab 1.5 309±17309 \pm 17 -

the properties of the medium must not provoke the self-coagulation of the IgG-latex complexes (i.e., nonspecific aggregation). Thus, the stability of a sensitized latex used for immunoassay purposes is one of the most important aspects in any optimization strategy.

In our case, immunoreactions were carried out in saline BSA borate buffer. For this reason we were interested in studying the stability in saline solutions where pH was equal to 8.0 . We attempted calculation of the CCC for all of our samples using KBr as the agglutinating agent. It should be noted that aggregation kinetics are usually very fast in this kind of experiment, and it has been found preferable to detect the coagulation process by means of a PC spectrophotometer capable of analyzing changes in light scattering over short reaction intervals (typically less than 30 s ). The CCC values for the “physically adsorbed” samples are shown in Table 2 (last column). However, the CCC values for the particles with covalently attached protein could not be calculated using this method because the changes in absorbance observed for these particles were very small over the range of KBr concentration used (up to 1.15 M ). It could be concluded that microspheres with IgG covalently bound are more stable than those with the protein passively adsorbed, which is unexpected as the IEP of the former are higher than for the latter due to the effect of the CDI molecules. Therefore, it is concluded that there must be Tween 20 molecules adsorbed to bare polystyrene patches, as these particles were incubated for 14 h with a Tween 20 solution during the sensitization protocol. It is well known that emulsifiers are good stabilizers of colloidal systems. Thus, these adsorbed nonionic molecules may be responsible for the stability of all the “covalent” preparations. Nevertheless, systems using covalently linked antibody latex particles may coagulate in KBr solutions but as this surfactant stabilizes by means of steric hindrance, the process could be very slow and the aggregation could not be detected in the monitoring time frame, i.e., 30 s .

However, if there was no Tween 20 on these particles, it would be expected that the CCC values would be lower for these samples than for the “physically adsorbed” ones, due to their differences in IEP.

It is worthwhile to highlight from the CCC data the high stability found in the Mab ads sample. While at pH 8.0 the stability of JL3 particles decreases as they are being covered by polyclonal IgG, it increases when loading with the monoclonal antibody. This can only be explained by a difference in the state of charge of the adsorbed protein layer. Therefore, these stability results confirm once more the idea that the monoclonal antibody used in this work possesses a IEP lower than the IEP of the polyclonal molecules.

We also studied the stability of samples in the same medium employed for immunoassays, namely saline BSA borate buffer. As noted under Materials and Methods, in this kind of experiment very diluted samples were incubated in such a solution for a long time ( 48 h ), and then the average size of the complexes was determined. Values are also shown in Table 2, first column. When comparing the polyclonal sample to those of the monoclonal data, one can check that the same pattern described above is obtained; that is, in all cases Mab reagents are more stable than corresponding Pab preparations even at very high loading levels. This result can be easily explained if the state of charge of the external protein layer is taken into account. However, it is worth noting the stability of physically adsorbed antibody latex particles, which is in some cases higher than that found in the “covalent” reagents (even with the adsorbed Tween 20 molecules). This is due to the presence of BSA molecules in the reaction mixture, and protein thus being an excellent stabilizing agent at pH8.0(13,14)\mathrm{pH} 8.0(13,14) by virtue of its binding to any bare polystyrene patches on the latex particles with the passively adsorbed form of protein loading. They are not, however, able to attach on the polymer surface of the “covalent” reagents because of the presence of previously adsorbed surfactant molecules. It should also be noted that BSA is better than Tween 20 as a colloidal stabilizer, as particles covered by the protein are stabilized by means of both osmotic and electrostatic interactions (32,33)(32,33), while those covered by the nonionic emulsifier are only stabilized by steric hindrance effects, although there will be some charge effect due to the protein present.

It is also worth reiterating that the higher the protein coverage (and thus the lower the Tween 20 loading), the lower the stability becomes. This result is much more evident for the Pab samples, due to the difference in IEP of monoclonal and polyclonal antibodies. Furthermore, the average sizes of Pab 0.5(274 nm)0.5(274 \mathrm{~nm}), Mab 1.5(309 nm)1.5(309 \mathrm{~nm}), Pab 1.1(370 nm)1.1(370 \mathrm{~nm}), and Pab 1.8(2.3μ m)1.8(2.3 \mu \mathrm{~m}) found in the stability experiments correlate quite well in relation to the charge density values (N)(N) of the protein layer, obtained using the Ohshima and Kondo theory: 15, 14, 12 , and 10 mM , respectively. This observation further emphasizes that colloidal stability largely depends on the electrical state of the external layer in protein-latex complexes.
img-2.jpeg

FIG. 3. (a) Absorbance changes over a 10 -min reaction versus antigen concentration. ( ◯\bigcirc ) Pab ads ( 1.0mg/m21.0 \mathrm{mg} / \mathrm{m}^{2} ), ( ←\leftarrow ) Pab 0.5 , ( ∼\boldsymbol{\sim} ) Pab 1.1, and ( □\square ) Pab 1.8. (b) Resonance units (RU) variations for (⋯ )(\cdots) JL3, (⋯ )(\cdots) Pab 0.5, (—) Pab 1.1, and (-) Pab 1.8 particles binding immobilized HSA.

Immunoassays and SPR Experiments

As shown earlier in Table 2, the Pab 1.8 was the only particle preparation which showed instability in saline BSA borate buffer, as measured by its size, after incubation for 2 days in such a medium. This might suggest a tendency to nonspecific aggregation in an immunoassay. However, it could also suggest that it is the most likely IgG-JL3 preparation to give a specific aggregation reaction, as the particles would be able to approach near enough to each other to facilitate agglutination in the presence of antigen (HSA in our case). All the immunoassays performed in saline BSA are depicted in Fig. 3a. Results present the typical “bell-shape” precipitin curve (34). However, none of the Mab particle reagents gave any immunoaggregation response throughout the HSA concentration range studied. As can be seen, the highest levels of immunoaggregation were obtained when polyclonal antibody was covalently attached to the particles, the level of aggregation increasing with protein loading. However, the response with the polyclonal antibody physically adsorbed on latex

particles was also significantly lower than the covalently coupled antibody. This result might seem a little odd, taking into account that IgG coverage is almost equal to that of Pab 1.1 complex and knowing that reactivity of any latex reagent largely depends on the antibody coverage (35,36)(35,36), provided that it is stable in absence of antigen. However, as we have just pointed out, the lower electrical repulsive interaction between particles, the easier it is for the particles to approach. Thus, as can be seen, the level of aggregation response seen in Fig. 3a also shows the same change with the electrical state of the adsorbed protein layer. The NN values obtained by using the Ohshima and Kondo theory ( NPab-adx >NPab 0.5>NPab 1.1>N_{\text {Pab-adx }}>N_{\text {Pab } 0.5}>N_{\text {Pab } 1.1}> Neat 15N_{\text {eat } 15} ) show an inverse relationship to the apparent immunoreactivity of these particle reagents. It is worth mentioning that all reagents were stable in saline BSA in the absence of HSA during the period of the time immunoassays reaction studied (10 min). This does not contradict the size value given in Table 2 for the Pab 1.8 sample, as this one was obtained after a 2-day incubation in such a buffer while immunoreactions were short time period experiments.

The functionality of the antibody particle preparations was also assessed using real-time biospecific interaction analysis: surface plasmon resonance, which can provide information about the properties of biomolecules, such as binding patterns, affinities, and kinetic rate constants (37-40). However, to our knowledge nobody has used SPR to assess these characteristics for proteins adsorbed on latex particles. SPR is a very modern technique which has been developed during the past decade. This technique transforms the interaction between biomolecules into the position of a dark spot on a photodiode array, which occurs for a SPR-resonance angle called ‘resonance signal.’ The ‘resonance signal’ varies as analyte (antibody molecules or sensitized polysterene particles) captures a ligand (HSA) previously immobilized in the matrix of the gold sensor chip. This is because the SPR response is determined by changes in the optical mass of the sensing layer. The device we used, a BIAcore from Pharmacia, expressed the ‘resonance signal’ in ‘resonance units’ (’ ‘RU’), that is, a measurement of the position of the SPR minimum (‘dark spot’) in the photodiode array, and not the SPR-resonance angle itself. A more detailed and deeper information about surface plasmon resonance is shown by I. Lundström in a recent review (37). Figure 3b shows how the antiHSA-IgG molecules attached to the latex particles bind to immobilized HSA molecules located on the sensor chip surface. The deposition of IgG-latex complexes on the gold sensor chip by means of the antibodyantigen recognition gives a change in the response of the BIAcore apparatus. The resonance units increase as the mass of deposited IgG-latex complexes increases. It is possible to distinguish three different zones in the experiments shown in Fig. 3b. After the gold sensor chip (previously sensitized with HSA) was equilibrated with HBS buffer for 2 min (first part of the curve), diluted solutions of IgG-JL3 particles (1:100) were put through the cell for 4 min (second part). Then, pure HBS

TABLE 3
Increments of “Resonance Units” ( ΔRU\boldsymbol{\Delta R U} ) and Absorbance Change Values ( Δ\Delta Abs) for SPR and Immunoaggregation Experiments

Sample ΔRU (in HBS + P20) \begin{gathered} \Delta \mathrm{RU} \\ \text { (in HBS }+ \text { P20) } \end{gathered} Δ Abs (570 nm) (in HBS + P20) \begin{gathered} \Delta \text { Abs }(570 \mathrm{~nm}) \\ \text { (in HBS }+ \text { P20) } \end{gathered} ΔRU (in saline-BSA) \begin{gathered} \Delta \mathrm{RU} \\ \text { (in saline-BSA) } \end{gathered} Δ Abs (570 nm) (in saline-BSA) \begin{gathered} \Delta \text { Abs }(570 \mathrm{~nm}) \\ \text { (in saline-BSA) } \end{gathered}
Pab ads (1.0) 768 30 329 25
Mab ads (1.3) 402 0 10 0
Pab 0.5 261 19 230 27
Pab 1.1 529 112 249 103
Pab 1.8 628 165 377 170
Mab 0.5 45 0 49 0
Mab 0.7 438 0 100 0
Mab 1.5 799 0 250 0
Pab 1461 - 1023 -
Pab + HSA 93 - 0 -
Mab 1591 - 1479 -
Mab + HSA 552 - 362 -

Note. HSA concentration used to agglutinate the systems were 0.25μ g/ml0.25 \mu \mathrm{~g} / \mathrm{ml}.
buffer was injected again (third part). It can be seen that higher increments in RU are obtained for increasing IgG coverage. Therefore, immunoaggregation rates (Fig. 3a) and binding of antibody particles to antigen on the sensor chip (Fig. 3b) demonstrate a similar relationship with respect to the different particle preparations studied, namely increasing signal with increasing coverage of particles by antibody. This is a behavior similar to that found in the immunoassay reaction shown in Fig. 3a. It should be noted that both techniques give information about the biological activity of the antibody molecules attached on latex particles. However, responses obtained by this two methods are different, although those supplied by one complements the results obtained using the other. However, we obtained important and significant differences between the SPR and immunoaggregation experiments. In Table 3 we show the increments of resonance units given by the BIAcore, after incubating our samples in HBS and saline BSA media for 4 min , versus the absorbance increments in immunoagglutination assays (which lasted 10 min ). In addition, we also performed a set of experiments in which free IgG molecules (namely “Pab” and "Mab’’ in Table 3) were introduced to the BIAcore in the absence of latex particles, binding to the HSA on the sensor chip surface. These antibodies were also mixed with a solution of HSA (’ ‘Pab + HSA’’ and ’ ‘Mab + HSA’’ in Table 3), previously to inject them into the device, to observe whether the reaction was inhibitable, which gives an indication of whether the reaction was specific. As can be seen at the bottom of Table 3 (last four rows), all the reactions were inhibitable, although some activity remained in the monoclonal sample.

With respect to the antibody particle preparations two important features (considering the results obtained in either HBS or saline BSA buffer) should be noted. (i) According to SPR data all the monoclonal-latex samples are reactive against

HSA, and the activity (i.e., binding) increases with antibody loading. However, there is no aggregation of these antibodyparticle preparations when they are incubated with free HSA. As can be observed, both the Mab and the Pab particle preparations gave very similar RU increments on the BIAcore which implies that there were similar quantities of viable paratopes available for binding HSA under immunoassay conditions. However, as pointed out earlier, particles with adsorbed monoclonal IgG exhibited no immunoagglutination. The reason for this observation may lie in the fact that immunoaggregation is a process that depends on the ability of two antibody molecules adsorbed on different particles to bind to different epitopes on the same antigen molecule, in this case HSA. As Mab is a monoclonal sample, it needs two (or more) distinct but identical epitopes on the HSA molecule. The combined results given by SPR and immunoassay experiments suggest that the monoclonal IgG we have used in this work only recognizes a single epitope in the antigen molecule. It would also be possible that there were two or more “reactive” or “recognizable” epitopes present on the HSA, but they may be spatially close such that the binding of one monoclonal antibody attached on a particle prevents the subsequent binding of another monoclonal antibody adsorbed on another latex microsphere, either by steric hindrance or by masking the second epitope, (ii) the SPR data with the physically adsorbed polyclonal IgG particle preparation (Pab ads) shown in Table 3 indicate a large change in resonance units, i.e., binding of antibody particle, which contrasts strikingly with the data using this particular preparation in the immunoassay format (see Fig. 3a and Table 3, third and fifth columns). It is also worth comparing this complex with the covalently coupled polyclonal preparation Pab 1.1, as both of them possess quite similar IgG coverage. The ΔRU\Delta \mathrm{RU} values are higher for the particles with physically adsorbed IgG , in either HBS or saline BSA medium. Therefore, the important differences between both samples obtained in the immunoaggregation experiments are not caused by the presence or absence of covalent links between the proteins and the polymer surface, but the use of different components (i.e., CDI and/or Tween 20) during the sensitization process. Then, the affinity of the attached antibodies were not being determined by whether the molecules are physically adsorbed or chemically bound. Therefore, it is concluded that there is no significant difference with regard to the molecular orientation or deformation during the spreading process, as it has been speculated to justify reactivity differences found in “physical adsorption” versus "chemical binding’ studies based on immunoassay experiments. Perhaps the use of “spacers” in covalent coupling procedures may help to increase the availability of attached antibodies to bind antigens (17), but probably not in this case. We can conclude that the differences obtained between adsorbed (Pab ads) and covalently coupled (Pab 1.1) preparations (in immunoaggregation assays) are caused by their differences in the charge density of the external polyelectrolyte layer (see Fig. 2). Fur-
thermore, the origin of this difference was determined by the absence or presence of CDI molecules on particles with physically adsorbed or covalently bound IgG, respectively. Thus, the different protocols used to attach the antibody to the JL3 latex are, in the end, responsible for all the differences observed between the two sets of particle reagents throughout the series of experiments for evaluation of IEP and NN (electrophoretic mobility), size (PCS), CCC, and immunoaggregation (turbidimetry). However, as can be seen in the present work, SPR experiments demonstrate that covalent coupling of protein to polymer surface does not cause any structural or conformational difference with regard to passive adsorption. What happens is that as NN decreases the repulsion between sensitized particles also diminishes, and thus the aggregation process is favored.

In order to consolidate this hypothesis and explore the effect of antibody coupling to particles, we tried to calculate the binding constants for the antibody-antigen union in four cases: free polyclonal and monoclonal antibodies and covalently coupled polyclonal (Pab 1.8) and monoclonal (Mab 1.5) preparations. In all cases antigen molecules (HSA) were immobilized on the surface of the gold sensor chip in order to obtain SPR results. It was considered that differences between the free in the bulk and the adsorbed proteins might exist as the latter should have a different conformation (partially denatured) due to spreading processes on the particle surface. Therefore, the SPR experiments were conducted to determine the association rate ( kass k_{\text {ass }} ), dissociation rate ( Kdiss K_{\text {diss }} ) and equilibrium ( KAK_{\mathrm{A}} ) constants of the following reaction:

IgG⁡+ HSA:IgG-HSA \operatorname{IgG}+\text { HSA:IgG-HSA }

where

d[IgG−HSA]/dt=kass [IgG][HSA]d[IgG−HSA]/dt=kdiss [IgG][HSA]\begin{aligned} & d[\mathrm{IgG}-\mathrm{HSA}] / d t=k_{\text {ass }}[\mathrm{IgG}][\mathrm{HSA}] \\ & d[\mathrm{IgG}-\mathrm{HSA}] / d t=k_{\text {diss }}[\mathrm{IgG}][\mathrm{HSA}] \end{aligned}

and

KA=kass /kdiss =[IgG−HSA]/[IgG][HSA]K_{\mathrm{A}}=k_{\text {ass }} / k_{\text {diss }}=[\mathrm{IgG}-\mathrm{HSA}] /[\mathrm{IgG}][\mathrm{HSA}]

Solutions of free antibody and antibody particle preparation solutions were passed through the BIAcore device at different analyte concentrations. RU increments obtained for the polyclonal samples are shown in Fig. 4, namely Fig. 4a for free IgG and Fig. 4b for adsorbed antibodies. As can be seen, both sets of experiments exhibit different behavior. The respective results for the monoclonal antibody reagents exhibited a very similar pattern (figures not shown). kass k_{\text {ass }} and kdiss k_{\text {diss }} can be obtained after some mathematical treatments from the first derivative of RU with regard to time,

img-3.jpeg

FIG. 4. (a) Resonance units (RU) variations for different free in the bulk polyclonal IgG concentrations: (⟶)50nM,(⋯ )100nM,(⋯ )200nM(\longrightarrow) 50 \mathrm{nM},(\cdots) 100 \mathrm{nM},(\cdots) 200 \mathrm{nM}, (−−−)400nM(---) 400 \mathrm{nM}, and ( - ) 800 nM . (b) Resonance units (RU) variations for different Pab1.8\mathrm{Pab} 1.8 complex dilution in HBS ( v/v\mathrm{v} / \mathrm{v} ): (⟶)1:200,(⋯ )1:150,(⋯ )(\longrightarrow) 1: 200,(\cdots) 1: 150,(\cdots) 1:100, (—) 1:75, and (——) 1:50.

(dRU/dt)=kassC(RUmax⁡−RU)−kdissRU(d \mathrm{RU} / d t)=k_{\mathrm{ass}} C\left(\mathrm{RU}_{\max }-\mathrm{RU}\right)-k_{\mathrm{diss}} \mathrm{RU}

where " CC " is the concentration of analyte (free IgG or IgGlatex complexes) and RUmax \mathrm{RU}_{\text {max }} is the maximum RU value obtained for a given " CC ". kass k_{\text {ass }} was obtained from the increasing part of the RU vs time curve. kdiss k_{\text {diss }} was obtained from the part of the curve when no analyte was passed through the system (third part of the curves).

The measurement of rate constants on the BIAcore has been described previously (41). Experimentally it was necessary to immobilize the antigen (HSA) on the sensor surface and this is not ideal for determination of rate constants. This is because of the bivalency of the antibody which precludes the univalent binding interaction that is essential to kinetic analysis. The valence is further increased with the antibody-particle conjugates, and thus the values determined could only be used in a comparative, qualitative sense. The values obtained for the two unconjugated antibodies Pab and Mab do give a meaningful comparison of relative kass k_{\text {ass }} but obtaining the kass k_{\text {ass }} values for the
antibody-particle conjugates was experimentally impossible. The mathematical treatment of data when IgG-latex particles bound to the HSA deposited on the sensor chip always led to erroneous results (i.e., negative kass k_{\text {ass }} values). Some evidence of functionality was found (see Fig. 4b), but no KAK_{\mathrm{A}} value could be obtained although experiments were performed in triplicate during different days. The data for the free antibody preparations are shown in Table 4. As can be seen, the affinity for the polyclonal IgG - HSA binding (see kass k_{\text {ass }} data) was somewhat higher than the affinity of the monoclonal sample. These values depend only on the own nature of the IgG pools. As both KAK_{\mathrm{A}} values are of the same order of magnitude, differences with regard to the immunoaggregation process between Pab-latex and Mab-latex complexes must be justified as noted above. Therefore, KAK_{\mathrm{A}} values corroborate that the lack of aggregation mediated by the monoclonal IgG coupled to the latex particles may be due to this antibody recognizing only one epitope in the antigen (HSA) molecules. Mab-latex complexes bind to HSA in the SPR experiments but HSA cannot bridge two single particles in the latex-enhanced immunoassays.

Finally, in order to confirm that electrical repulsion interactions between sensitized particles may decrease the extent of agglutination in immunoassay reactions, we decided to perform a set of immunoassays in which the electrical state of the external protein layer was modified by changing the medium pH and ionic strength, and repulsion between colloidal particles diminished by adding increasing amounts of KBr to the reaction mixture. Experiments were carried out with the Pab 1.8 samples in pH 7.0 (phosphate buffer), pH 8.0 , and pH 9.0 (borate buffers), with KBr concentrations of 50, 100, and 150 mM . Particle concentrations and reaction times were exactly the same as in the previous experiments. The results are shown in Fig. 5, indicating that as the pH increases the antibody molecules become more positively charged leading to a higher charge density (N)(N) in the sensitized particles, and thus to a progressive diminution in the immunoresponse. On the other hand, as ionic strength increases, the screening effect becomes more important and it leads to increasing agglutination. It is worth noting that there is almost no turbidimetric response at the low ionic strength experiments. Therefore, best results were obtained at pH 7.0 (where adsorbed IgG molecules have a low charge) and 150 mM of KBr . One must be very careful to choose a reaction medium which facilitates the coagulation by means of antigen molecules, but avoids the self-aggregation of

TABLE 4
Kinetic and Equilibrium Constants for the Reaction IgG⁡+HSA:IgG⁡−HSA\operatorname{IgG}+\mathrm{HSA}: \operatorname{IgG}-\mathrm{HSA}

Sample kass (M−1 s−1)k_{\text {ass }}\left(\mathrm{M}^{-1} \mathrm{~s}^{-1}\right) kdiss (s−1)k_{\text {diss }}\left(\mathrm{s}^{-1}\right) KA(M−1)K_{\mathrm{A}}\left(\mathrm{M}^{-1}\right)
Pab 36500 1.310−41.310^{-4} 2.81062.810^{6}
Mab 12400 1.310−41.310^{-4} 9.51079.510^{7}

img-4.jpeg

FIG. 5. Absorbance changes over a 10-min reaction versus HSA concentration obtained with the Pab⁡1.8\operatorname{Pab} 1.8 complex. KBr concentration was equal to ( ()\mathbf{(}) 50mM,(∙)100mM50 \mathrm{mM},(\bullet) 100 \mathrm{mM}, and (□)150mM(\boldsymbol{\square}) 150 \mathrm{mM}.
particles caused by the physicochemical properties of such a solution, i.e., pH and ionic strength.

SUMMARY

The use of different and new technologies, such as SPR or electrophoretic mobility, provides a means of characterizing antibody-latex complexes used in light scattering immunoassays. SPR has become a very useful tool for explaining and confirming different results obtained by other techniques. At least, SPR has allowed to us to understand why the monoclonal antibody used in this work did not appear to retain its functionality in immunoaggregation experiments and (in combination with μe\mu_{\mathrm{e}} measurements) why the reagents with IgG chemically bound are better than those with passively adsorbed antibodies.

The main conclusions can be summarized as follows.
(a) Antibodies can be covalently bound to carboxyl latex particles using the carbodiimide method. However, some of the carboxyl-CDI activated groups do not react with either protein or glycine molecules and remain as positively charged organic groups. This fact is responsible for the main differences found between the antibody latex preparations achieved by covalent coupling or adsorption of immnuglobulin. In addition, the presence of Tween 20 also causes some differences in the physicochemical properties of both sets of reagents.
(b) Electrophoretic mobility studies are very useful for determining the IEP of protein-latex preparations. Moreover, according to the Ohshima and Kondo theory, it is possible to estimate the charge density of the adsorbed polyelectrolyte layer. As we have shown, this parameter is one of the most
important to take into account when interpreting and understanding the immunoaggregation results.
© The use of monoclonal antibodies with low IEP provides more stable latex reagents (in a “colloidal” sense), which may be helpful when seeking to achieve antibody loading. However, the use of an immunoactive monoclonal IgG may lead to non-aggregating complexes in case Mab only recognizes a rare epitope and thus the crosslinking does not take place, as has been shown in this paper.
(d) SPR has been extremely useful in demonstrating that the covalent coupling of IgG molecules does not cause a conformational change in the attached protein molecules in comparison with the physically adsorbed antibodies, demonstrating similar binding patterns (see Table 3). Therefore, the important differences found between them are caused by the presence or absence of CDI molecules on the polymer surface, which leads to different NN values. SPR has also been useful for explaining the antibody functionality differences between the polyclonal and monoclonal IgG used in this work, although both of them had similar equilibrium constants.

Thus, it is the electrical state of the external layer of the antibody-latex particle that plays a major role in the absent functionality of the bound antibody, since the extent of the immunoaggregation largely depends on the charge density of this layer (N)(N). However, care must be taken to ensure that adequate colloidal stability is achieved to avoid nonspecific aggregation. In order to achieve this, an optimization strategy must be designed including protein coverage of the particle, pH , and ionic strength of the reaction medium, which constitute the critical variables.

ACKNOWLEDGMENTS

Financial support from the British Council (Spanish-British Integrated Action 247 B) is gratefully acknowledged. Also thanks are due to the ALFA Programme (Contract Number ALR/B7-3001/94.04-3.0242.8). C. L. Davey and D. J. Newman were supported by a grant from Dade International.

REFERENCES

  1. Singer, J. M., and Plotz, C. M., Am. J. Med. 21, 888 (1956).

  2. Bangs, L. B., and Kenny, M. T., Ind. Res. 18, 46 (1976).

  3. Hechemy, K. E., and Michaelson, E. E., Lab. Man. 22, 27 (1984).

  4. Price, C. P., and Newman, D. J., "Principles and Practice of Immunoassay.‘’ Stockton Press, New York, 1991.

  5. Heller, W., and Pangonis, W. J., J. Chem. Phys. 26, 498 (1954).

  6. Von Schulthess, G. K., Giglio, M, Cannell, D. S., and Benedek, G. B., Mol. Immunol. 17, 81 (1980).

  7. Skoug, J. W., and Pardue, H. L., Clin. Chem. 34, 300 (1988).

  8. Ortega, J. L., Molina, J. A., Peula, J. M., and Hidalgo-Álvarez, R., J. Immunol. Methods, in press.

  9. Galisteo, F., Puig, J., Martín, A., Serra, J., and Hidalgo-Álvarez, R., Colloids Surf. B 2, 435 (1992).

  10. Nakamura, M., Ohshima, H., and Kondo, T., J. Colloid Interface Sci. 154, 393 (1992).

  11. Lichtenbelt, J. W. Th., Heuvelsland, W. J. M., Oldenzeel, M. E., and Zsom, R. L. J., Colloids Surf. B 1, 75 (1993).

  12. Ortega, J. L., Gálvez, M. J., and Hidalgo-Alvarez, R., J. Mater. Sci. Mater.Med. 6, 754 (1995).

  13. Ortega, J. L., and Hidalgo-Alvarez, R., Biotechnol. Bioeng. 47, 633 (1995).

  14. Peula, J. M., Hidalgo-Alvarez, R., and de las Nieves, F. J., J. Biomater. Sci. Polym. Sci. 7, 231 (1995).

  15. Galisteo, F., “Adsorción de Proteínas Sobre Modelos Coloidales,” Ph.D. Dissertation, University of Granada, 1992.

  16. Ortega, J. L., Bastos, D., and Hidalgo-Alvarez, R., J. Colloid Interface Sci. 184, 331 (1996).

  17. Kitano, H., Iwai, S., Okubo, T., and Ise, T., J. Am. Chem. Soc. 109, 7608 (1987).

  18. Guthrie, W. H., “New Free Radical Initiators and Their Use in the Preparation of Polystyrene Polymer Colloids,” Ph.D. Dissertation, University of Lehigh, PA, 1985.

  19. Galisteo-González, F., Puig, J., Martín-Rodríguez, A., Serra-Domènech, J., and Hidalgo-Alvarez, R., Colloids Surf. B 2, 435 (1994).

  20. Peula, J. M., Puig, J., Serra, J., de las Nieves, F. J., and Hidalgo-Alvarez, R., Colloids Surf. B 2, 435 (1994).

  21. Bastos, D., Ortega, J. L., de las Nieves, F. J., and Hidalgo-Alvarez, R., J. Colloid Interface Sci. 176, 232 (1995).

  22. Ortega, J. L., Bastos, D., and Hidalgo-Alvarez, R., J. Colloid Interface Sci. 176, 240 (1995).

  23. Norde, W., Adv. Colloid Interface Sci. 25, 267 (1986).

  24. Bagchi, P., and Birnbaum, S. M., J. Colloid Interface Sci. 83, 460 (1981).

  25. Elgersma, A. V. “Competitive Adsorption of Albumin and Monoclonal γ\gamma-Globulin Molecules on Polystyrene Surfaces.” Ph.D. Dissertation, Agricultural University, Wageningen, The Netherlands, 1990.

  26. Suzawa, T., and Shirahama, H., Adv. Colloid Interface Sci. 35, 139 (1991).

  27. Ortega, J. L., Gálvez, M. J., and Hidalgo-Alvarez, R., Langmuir 12, 3211 (1996).

  28. Ohshima, H., and Kondo, T., J. Colloid Interface Sci. 116305 (1987).

  29. Ohshima, H., and Kondo, T., J. Colloid Interface Sci. 130, 281 (1989).

  30. Ohshima, H, Nakamura, M. and Kondo, T., Colloid Polym. Sci. 270, 873 (1992).

  31. Ortega, J. L., and. Hidalgo-Alvarez, R., J. Non-Equilib. Thermodyn. 21, 339 (1996).

  32. Ortega, J. L., “Estabilidad Coloidal de Interfases Estructuradas: Aplicación a Inmunosensores.” Ph.D. Dissertation, University of Granada, Spain, 1995.

  33. Ortega, J. L., Martín, A., and Hidalgo-Alvarez, R., J. Colloid Interface Sci. 184,259 (1996).

  34. Heidelberger, M., and Kendall, F. W., J. Exp. Med. 62, 467 (1935).

  35. Peula, J. M., Hidalgo-Alvarez, R., and de las Nieves, F. J., J. Biomater. Sci. Polym. Edn 7, 241 (1995).

  36. Ortega, J. L., Molina, J. A., and Hidalgo-Alvarez, R., J. Immunol. Methods 190, 29 (1996).

  37. Lundström, I., Biosensors Bioelectron. 9, 725 (1994).

  38. Flanagan, M. T., and Pantell, R. H., Electron. Lect. 20, 968 (1984).

  39. Kooyman, R. P. H., Kolman, H., Van Gent, J., and Greve, J., Anal. Chem. Acta 213, 35 (1988).

  40. Liedberg, B., Lundström, I., and Stenberg, E., Sensors Actuators B 11, 63 (1993).

  41. Karlsson, R, Michaelsson, A., and Mattson, L., J. Immunol. Methods 145, 229 (1991).