Enantiomer Analysis by Competitive Binding Methods
C. Edgar Cook Research Triangle Institute, Research Triangle Park, North Carolina
It is only when presented with an asymmetric environment, for example, another chiral molecule, that differences in the structure of enantiomers permit their separation for analysis. Because antibodies and receptors are themselves chiral molecules, nature can provide the analyst with a variety of substances that would be expected to bind drug molecules in a selective manner based on their stereochemistry. The complexes formed by binding of individual enantiomers to a chiral immunoglobulin may be considered diastereoisomers and will have different physical properties, including affinity constants. Because a criterion for demonstrating a specific receptor for a given drug is its stereoselectivity, it is also obvious that receptors can be used to distinguish between enantiomers. Both receptors and antibodies can therefore be used in competitive binding assays to selectively analyze one enantiomer in the presence of the other.
This review will not attempt to be an exhaustive compilation of all references to competitive binding analysis of enantiomers. However, it is my intention to give sufficient examples of this area of research to give the reader some feel for the potential advantages and problems associated with enantiomers in competitive binding assays.
Various immunoassays and receptor assays have been referred to as being stereoselective, enantioselective, stereospecific, or enantiospecific. I generally prefer the suffix selective rather than specific, based on the fact that most binding proteins exhibit at least some affinity for isomers of a compound, even though that affinity may be minimal. Although from a practical point of view, binding assays can be specific for one isomer, the term selective is a more accurate one and I will use it throughout this chapter. The choice between the use of stereo- and enantio- may depend on the context. Enantiomers are a special case of stereoisomers. Stereoisomers differ in their relative stereochemistry at least at one point in the molecule, but only if two isomers are mirror images of each other are they enantiomers. Stereoselective may be used to describe the ability to distinguish between, for example, cis and trans isomers of olefins or cyclic compounds or between diastereoisomers. Thus, an antibody for quinidine (1) is properly termed stereoselective, since it was able to distinguish quinidine and its diastereomer quinine. Its enantioselectivity (ability to distinguish R- and S-quinidine) was not determined. Assays for R- and S-isomers, described below, are both stereoselective and enantioselective.
1(1. IMMUNOASSAYS FOR ENANTIOMERS A. Immunoassay Methodology
Immunoassays are a form of competitive binding assay in which a ligand such as a drug molecule interacts in a reversible manner with the binding site of a specific antibody. If the ligand bears a distinguishing label, then displacement by unlabeled drug may be measured and related to the concentration of drug present.
Since the introduction of radioimmunoassay (RIA) by Yalow and Berson in 1959 (2), numerous books and review articles have been written on immunoassays. The reader is referred to the literature for a selection of reviews on general immunoassay methodology (3), RIA (4-7), theory  and statistical analysis (9,10), synthesis of immunogens (5,6,11,12), enzyme immunoassays (13,14), fluorescence immunoassay techniques (15), miscellaneous labels (16) and separation techniques (17). A brief discussion of general techniques is presented below.
Basic requirements for a competitive binding assay are a selective, high-affinity binding site, a ligand with a detectable label, and a means of distinguishing bound and unbound labeled ligands. Given the availability of these factors, competitive binding assays have many advantages. These include sensitivity, speed of analysis, simplicity of procedures, and often high selectivity for a given analyte.
Antibody molecules can provide the required high-affinity binding site. Antibody binding sites may be obtained that can recognize any kind of organic compound ranging in molecular mass from perhaps 200 to 1000 Da and in any shape or conformation that such amount of matter can assume. For larger molecules, antibodies are capable of recognizing a portion of the molecule and binding to that. Antigen-antibody binding is noncovalent and involves hydrophobic, ionic, and van der Waals interactions, as well as hydrogen bonding and steric repulsive forces (18).
Small molecules such as drugs are not immunogenic per se, but conjugation of the small molecule (hapten) to a large molecule such as a protein (carrier) results in the formation of immunogenic material (19). When injected into an animal, this immunogen stimulates the formation of antibodies capable of binding the drug. Introduction of a linking group inevitably results in at least some change in the overall structure and electronic configuration of the small molecule, and its presence will influence the affinity and selectivity of the resulting antibodies. Although there seems to be no really definitive studies on the effects of fink structure on resulting antibody affinity and selectivity, it may be reasoned that the link should be relatively small in volume, rather uniform in structure, and should facilitate extension of the small molecule away from the surface of the protein.
No definitive study of the effects of protein carrier on response is available, and many carrier proteins have been used, including globulins, albumin, hemocyanin, thyroglobulin, and fibrinogen. The optimal number of haptens, or the epitope density, is also controversial, but a density of 8 to 25 haptens per bovine serum albumin molecule is probably optimal (12).
The classical studies of Landsteiner (19) and many reports since that time show that antibodies resulting from conjugates are generally most selective for those portions of the small molecule that are not involved in the link to protein. However, different animals injected with the same immunogen can still give rise to a wide range of antibody selectivity. Even with the same animal, a range of antibodies of varying selectivity and affinity will be produced on immunization. The antiserum composition will thus vary from one animal to the next and indeed from one bleeding of an animal to the next.
In principal, such problems may be overcome by the hybridoma technique of Kohler and Milstein (20), Immunization of a mouse or other animal will stimulate the production of a series of B cells producing antibodies directed at various portions of the immunogen. Such cells do not grow well in culture, but they may be fused with myeloma cells, which do grow readily. The resulting hybridoma cells have both the immortality of the myeloma cell line and the antibody-producing capability of the B cell. These cells can be isolated by growing them in a medium that kills the unfused myeloma cells and allows the unfused B cells to die. Cloning by limiting-dilution procedures and selection of the appropriate hybridoma by screening techniques then permit one to develop a cell line that can produce a desired uniform antibody. The celt line can be grown by tissue culture techniques or injection into mice, where it forms ascites. The resulting ascites fluid is rich in the desired antibody.
The hybridoma technique has had an enormous impact on immunology in general, and many antibody reagents are now based on monoclonal antibodies. However, much work is still done with polyclonal antibodies from whole animals. When one recognizes that a few milliters of good antiserum from a rabbit permit thousands or even hundreds of thousands of analyses, it is apparent that generation of polyclonal antibodies, which is much less labor-intensive than generation of monoclonal antibodies, will still be useful for development of immunoassays for several years to come. Nevertheless, the hybridoma technique has much to recommend it and will certainly increase in importance for development of selective immunoassays.
It is necessary to be able to distinguish the labeled ligand bound to antibody from that which is not bound. This may be done by either some physical separation of bound and unbound labeled ligand, or some change in property of the ligand upon binding to the antibody. Separation techniques have relied heavily on the use of adsorbents, second antibodies, etc. Centrifugation then separates the liquid and solid phases and the appropriate phase can be measured. Other techniques such as the use of magnetizable particles have come into being to facilitate automation of the separation procedure (17).
Although the most common label is still the radioactive hydrogen or iodine atom, the problems associated with radioactivity have led to a continual search for other techniques that may be in some ways preferable to it. Fluorescence, in principal, can be detected with the same sensitivity as radioactivity; however, unlike radioactivity, there is a very significant natural background fluorescence in biological material. A variety of procedures have been called on to get around this problem (15), and some of these have made their way into commercial instrumentation.
One commercial system is based on fluorescence polarization. Molecules absorbing polarized light emit polarized fluorescent light. The de gree of polarization is related to molecular size, which in turn controls molecular rotation. Thus, the degree of polarization from a fluorophor-labeled Iigand will change when it binds to a macromolecule such as an antibody. This procedure permits a homogeneous assay. Another system is based on the fact that the fluorescence of most serum components is very short-lived. Therefore, if a ligand is labeled with a fluorophor in which the fluorescence has a long lifetime, it is possible to pulse the sample with radiation, measure fluorescence emission after an appropriate interval, and thereby overcome the problem of background fluorescence (21).
The amplification ability of enzymes has led to many schemes for enzyme immunoassay (13,14). The enzyme may be labeled with a drug in such a manner that when the drug portion of the conjugate binds to an antibody, enzymatic activity is inhibited. By measuring the product of the enzyme reaction in the presence of added substrate, one may obtain a measure of the amount of free enzyme-drug conjugate present in solution, which will be related to the competition of this conjugate with the drug in solution.
Alternatively, one may attach the drug to an enzyme in such a way that enzymatic activity is not inhibited. If the conjugate is allowed to bind to the antibody on a solid phase, excess enzyme may be washed away, and then by measurement of enzymatic activity the concentration of enzyme conjugate (and by deduction, therefore, the concentration of competitive drug in the original solution) may be measured. The drug may also be labeled with the enzyme substrate, an enzyme cofactor, an enzyme inhibitor, an enzyme activator, or even with a prosthetic group for an enzyme. Alternatively, a second antibody may be labeled with an enzyme.
Radioactivity, however, is still a very sensitive means of measuring the presence or absence of a given material. Assay methodology has now come full circle, to the development of an ultrasensitive enzyme RIA. In this technique, an antigen is bound to a solid phase. Antibody will bind to the antigen, which could be a drug-protein conjugate, and the presence of bound antibody is detected by means of a second antibody coupled to alkaline phosphatase. So far this is the standard enzyme-linked immunosorbent assay (ELISA). However, if the substrate is tritium-labeled adenosine monophosphate, it is converted by the enzyme to tritium-labeled adenosine, which may be readily separated and measured. The detection limit for this assay for cholera toxin is approximately 600 molecules of the toxin (22).
5. Enantioselective Antisera: Historical
As is true for many of the procedures and concepts on which immunoassays are based, Landsteiner's classical studies first pointed the way toward the use of antibodies for analysis of enantiomers. He produced antisera to D-tartaric acid and showed that the antibodies could discriminate between D- and Lrtartaric acid (19). Although this gives a clear indication of the possibilities of enantioselective immunoassay, for many years after the introduction of the immunoassay technique there was little, if any, systematic effort to explore this particular aspect of antibody properties.
B. Development and Use of Enantioselective Antisera from Racemic Immunogens
When an animal is immunized with a drug-protein conjugate, a mixture of heterogeneous antibodies may be formed. The immune system may recognize haptenic determinants from the drug itself, or those involving portions of the drug molecule and the protein to which it is conjugated. However, in a typical immunoassay, further selectivity may be achieved by the choice of the labeled ligand. This component of the assay permits the analyst to observe only those antibodies for which it has high affinity.
Conjugation of a racemic drug to a protein increases the possibility for heterogeneous antisera because at least two haptens (R and S) are introduced. The immune system could respond by producing antibodies to only one enantiomer, or varying amounts of antibodies to each enantiomer. Each antibody may also have affinity for the opposite enantiomer. The use of racemic radioligands further complicates the situation. These problems have been discussed by Maeda and Tsuji (23), Cook et al. (24), and Rominger and Albert (25).
2. Cyclazocine: A Compound with Multiple Asymmetric
Rabbits were immunized with a conjugate of racemic cyclazocine (Fig. 1 [la]) and competitive binding studies carried out with tritium-labeled d, /-cyclazocine (23). Various types of antibody mixtures were obtained from different animals. One animal immunized with the compound in the form [lc] shown in Fig. 1 exhibited almost no affinity for the d isomer of cyclazocine. In such an instance one would expect that the displacement curve for the d,l mixture would be displaced to higher concentrations than the displacement curve for the I isomer, and this indeed was observed. If the antibodies produced were completely nonenantioselective, then displacement curves for d, I, and the d,l mixture should be identical. This case was not observed.
Antiserum from an animal immunized with the conjugate [le] shown
*Denotes asymmetric center a. R1 = -OH. R2 = -CH2^7 , R3 = H
e. R1 ~ -OH. R2 = -CH2^7 , R3 = -CH2 NH-(BSA)Vn figure 1 Cyclazocine and immunogens based on cyclazocine.
in Fig. 1 gave displacement curves in which both the d and / isomers displaced radioligand at higher concentrations than was observed for the racemate. Such an antiserum is consistent with a model in which there are equal amounts of d- and /-selective antibodies, each of which cross-reacts to some extent with the opposite enantiomer. In a third animal (immunized with the form of cyclazocine [la] shown in Fig. 1), the displacement curve for the racemate showed the normal S-shaped characteristics of a standard RIA curve; however, the curves for the individual d and I isomers, although displaced to higher concentrations, became saturated at lower concentrations than occurred with the d,l mixture. This suggests the presence of a mixture of highly enantioselective antibodies. Thus, each antibody binds only its own specific enantiomeric radioligand, and saturation occurs at binding percentages that correspond to the ratio of d- and /-selective antibodies in the antiserum (23).
Although the three different types of antisera described above each was derived by challenge with a different immunogen, there are insufficient data to determine whether this is done to the immunogen structure or the individual animal.
When dogs were administered cyclazocine and their serum was analyzed by RIA, markedly different values were obtained depending on which antiserum was used. Such a result is, of course, to be expected and serves as a warning to those who develop immunoassays from racemic immunogens. For example, in animals administered d,i-cyclazocme, a peak plasma concentration of somewhat over 600 pg/mL was observed 0.5 hr after administration when the /-selective antiserum was used. Analysis with an antiserum that contained an apparent mixture of crossreacting d and I antibodies indicated a peak concentration of about 400 pg/mL at 1 hr after drug administration (23).
3. WR 171,669: A Compound with a Single Chiral Center
Cyclazocine has three asymmetric centers and thus the d and I isomers differ in stereochemistry at three separate points on the molecule. However, a similar situation has been shown to exist in a compound (Fig, 2 [2a]) with a single asymmetric carbon atom; l-(l,3-dichloro-6-trifluoromethyl-9-phenanthryl)-3-N,N-dibutyI-aminopropan-l-ol (WR 171,669), an antimalarial compound (24), Conjugation of the hemisuccinate ester of the racemic drug with bovine thyroglobulin gave an immunogen, shown to contain equal amounts of the d- and /-WR 171,669 (Fig. 2 [2b]), which was injected into rabbits. When tested with tritium-labeled racemic WR 171,669 as the radioligand, antisera from two rabbits challenged with the immunogen exhibited equivalent 50% displacement values for the race-mate and I isomer, with markedly higher values for the d isomer (Fig. 3A). A third Tabbit showed a similar trend but differences were smaller.
Comparison of displacement curves from one antiserum was made by
Figure 2 WR 171,669 and related immunogens.
use of the program ALLFIT (26). This program permits simultaneous analysis of families of sigmoidal curves that may be fitted by the general logistic equation y = D + (A - D)/[l + (x/C)B], Initially, individual parameters are fit to each curve without constraint. The curves can then be forced to share one or more parameters in common. If the constraints chosen do not significantly degrade the quality of fit (F test), the curves are assumed to have those parameters in common. This test showed that (as expected from inspection) the d and d,l displacement curves could not be constrained to share all the parameters (p < 0.001). Even the / and d,l
Figure 3 Competition with binding of racemic tritium-labeled WR 171,669 expressed as counts per minute of radioligand bound as function of log of competitor weight (A) or reciprocal of counts per minute of radioligand bound as a function of competitor weight (C). X = binding in absence of competitor; + = nonspecific binding;B= rf-, A = /,• = rf,/-WR 171,669. (From Ref. 23, used with permission.)
displacement curves could not share all parameters (p < 0.001), Although they have a superficial resemblance and almost identical values for parameter C (the 50% displacement intercept), B (the slope parameter) differed significantly (0,76 vs. 0.99).
Cook et al. (24) also applied the method of Pratt et al. (27) for evaluating cross-reaction. In this procedure the reciprocal of bound radioligand is plotted vs. competitor concentration. This provides a very sensitive test for divergence of standard curves, which may not be observed in the normal standard displacement plot. The Pratt plot showed rapid divergence of all three lines (d, d,l, and /), although divergence between d,l and / lines was less than between d,l and d lines (Fig. 3C).
For individual rabbits the relative enantioselectivity of the antisera appeared to remain fairly constant over time. However, an insufficient number of animals were studied for definitive conclusions.
4. Mathematical Description of Multiple Antibodies and Ligands
A mathematical description of the situation involving two enantio-selective antibodies and a racemic radioligand in the RIA procedure is complex. Rominger and Albert (25) have examined this situation from a theoretical basis and have reported equations that can be used to describe the situation, although they cannot be solved explicitly. By assuming two different antibody concentrations, different affinities of the antibodies for the corresponding enantiomers of the tracer and of the ligand [one antibody with a very low cross-reactivity and one with a relatively high cross-reactivity (10%)], Rominger and Albert arrived at a plot for the d, I, and d,l mixture displacement curves that resembles closely the experimental plot (Fig. 3A) presented by Cook et al, (24).
The above work indicates that a wide variety of responses may be obtained using antisera from immunization of animals with a racemic immunogen and competitive binding studies with racemic radioligand. This point seems not to have been considered by all investigators. In a sampling of 20 papers in which immunization was carried out with the d,l-immunogen, 7 made no mention of cross-reactions of the individual isomers in the binding studies; 3 simply made the statement that there was no discrimination of isomers; 3 reported that, based on 50% inhibition of binding, the d, I, and d,l forms had equivalent displacement of racemic radioligand; 2 indicated that the d and / forms showed 50% cross-reaction when compared with the racemate; and 3 papers studied these interactions in more detail.
This is an area that should receive considerably more attention. The data of Maeda and Tsuji (23) show the very significant errors in plasma concentrations that can result from use of inappropriately enantioselective sera in immunoassay of plasma from animals given racemic drug. The data of Cook et al. (24) demonstrate that even in cases in which the cross-reaction calculated at 50% inhibition is identical between an isomer and the racemate, the slopes of the two curves may differ significantly. Thus, it would behoove those who generate "nonselective" antisera from racemic immunogen to show that the resulting antisera will indeed give correct values for total concentration of the drug under the conditions of the experiment. Simple 50% cross-reaction data may not be sufficient to establish coincidence of curves, and an approach such as the use of ALLF1T (26) or Pratt-type plots (27) should be carried out to establish lack of enantioselectivity of an assay.
In connection with the above recommendation, it should be recognized that the purpose of an assay is to give valid data. An enantio-merically mixed assay can still give valid results if the ratio of enantiomers in the biological system measured does not vary appreciably from unity. Rominger and Albert (25) suggest determining whether the isomer ratio differs from unity by analysis of serial dilutions of the sample vs. the racemate standard curve. This is a standard approach to establishing identity of analyte and standard and may be helpful in cases in which it is not known whether the enantiomer ratio varies due to differences in enantiomer pharmacokinetics.
For compounds with a single asymmetric center, it might in principle be possible to generate antisera that are not enantioselective by using an immunogen that is a nonchiral analog of the drug in question (24). Thus, when the tetrahedral secondary alcohol carbon of WR 171,669 was converted to a trigonal carbon analog (Fig. 2 ) and this substance used for immunization, the resulting antisera showed relatively little enantioselectivity in a competitive binding assay with the d,l radioligand. Because antibodies are chiral molecules, they could nevertheless exhibit enantioselectivity even when generated to a nonchiral conjugate. In the case of WR 171,669, the enantioselectivity was small. This indicates that the binding site was sufficiently flexible to accommodate the change from the essentially planar trigonal template of the immunogen to the tetrahedral structure of the analyte, and that the binding site was insensitive to changes in the drug molecule close to the site of the hapten-protein link in the immunogen. This flexibility was observed even though the apparent affinity constant remained high (approximately 1010 L/mol). This approach may be useful in other instances where a single asymmetric center is present (24), Obviously, when two or more asymmetric centers are present, the likelihood of being able to devise a nonchiral analog for immunogen formation becomes much smaller (25).
C. Development and Use of Antisera for a Single Enantiomer
In a number of cases, investigators have prepared immunogens from one of two enantiomers and used these for analysis of that particular enantiomer. In some cases, the radioligand was enantiomerically pure; in others, racemic radioligand was used. Usually, this was done because of some specific information regarding the differential pharmacological effect of the enantiomer. In other cases, the drug itself is already administered as a single enantiomer and thus is often readily available for conjugation to protein. Most RIAs for steroids are developed by means of immunogens incorporating the naturally occurring or hormonally effective steroid enantiomer. The voluminous literature on this type of assay will not be further discussed. Likewise, a number of natural products that are generally obtained in the form of a single enantiomer have been subjected to RIA development. Unless there is special information regarding the cross-reaction of enantiomers, this also will not be covered,
A 1974 report described the development of an antiserum for (¡-amphetamine, Inhibition studies were carried out by means of displacement of tritium-labeled d-amphetamine, and the cross-reaction for /-amphetamine was 4.5% (28). Later an antiserum to rf-methamphetamine was reported to have no significant cross-reaction with the I isomer (29).
RIA for L-a-acetyl methadol (LAAM) has been reported (30). The O-acetyl group of this compound (Fig. 4 [4a]) was replaced by an O-suc-cinyl moiety, which was coupled to bovine thyroglobulin by mediation with a soluble carbodiimide reagent to give an immunogen (Fig. 4 [4b]). Tritium-labeled LAAM was used as a radioligand. A sensitivity of 50 pg was recorded with no serum interference. LAMM has two asymmetric centers, both of the S configuration. The diastereoisomer formed by change of the 3-S configuration (which bears the acetate residue and was involved in linkage to the protein) to the 3-R configuration exhibited a cross-reaction 25% that of LAMM. When the 6 carbon atom in the chain (Fig. 4) was changed from S to R, the cross-reaction dropped to 0.033% and the 3-R, 6-R compound (enantiomer of LAAM) had a cross-reaction of 0.17%. Thus, in this case, we see the general trend toward greater differen-
Was this article helpful?
All Natural Immune Boosters Proven To Fight Infection, Disease And More. Discover A Natural, Safe Effective Way To Boost Your Immune System Using Ingredients From Your Kitchen Cupboard. The only common sense, no holds barred guide to hit the market today no gimmicks, no pills, just old fashioned common sense remedies to cure colds, influenza, viral infections and more.