Fab preparations of sheep polyclonal anti-digoxin Abs have proven useful for reversal of the toxic effects of digoxin overdoses in patients. Unfortunately, the use of foreign species proteins in humans is limited because of the potential for immunological responses that include hypersensitivity reactions and acute anaphylaxis. Immunization of recently developed transgenic mice, whose endogenous μ heavy and κ light chain Ig genes are inactivated and which carry human Ig gene segments, with a digoxin-protein conjugate has enabled us to generate and isolate eight hybridoma cell lines secreting human sequence anti-digoxin mAbs. Six of the mAbs have been partially characterized and shown to have high specificity and low nanomolar affinities for digoxin. In addition, detailed competition binding studies performed with three of these mAbs have shown them to have distinct differences in their digoxin binding, and that all three structural moieties of the drug, the primary digitoxose sugar, steroid, and five-member unsaturated lactone ring, contribute to Ab recognition.

The cardiac glycosides digoxin and digitoxin, composed of a steroid core with a lactone ring (aglycone) and three β(1→4)-d glycoside-linked digitoxoses, are the active compounds in digitalis preparations derived from the foxglove plant and are commonly prescribed to treat chronic heart failure and some supraventricular arrhythmias (1, 2, 3). However, these compounds exhibit a very narrow therapeutic index, and digitalis toxicity is among the most prevalent of adverse drug reactions encountered by clinicians (2, 4, 5). Fortunately, the use of digitalis has been facilitated by the generation of Abs that can be used to monitor patients’ serum drug concentrations in order to maintain safe levels (6, 7). Also, the i.v. administration of affinity-purified Fab of sheep polyclonal anti-digoxin Abs can quickly reverse digoxin cardiotoxicity by binding to free drug in plasma and effecting the redistribution of the drug from a patient’s tissues back to the vascular circulation (8, 9). Further, because Fabs are relatively rapidly excreted in urine, high affinity Fabs that retain bound drug can provide a route of drug elimination as well as a means of its neutralization (10).

Because both the clinical usefulness of anti-digoxin polyclonal Fab treatment (11, 12, 13) and the feasibility of generating drug-specific mouse mAbs have been well established, this immunological treatment strategy is potentially applicable to additional drug overdose situations. Indeed, in addition to sheep polyclonal anti-digoxin Fabs, goat Fabs directed against colchicine (14) have been used to treat life-threatening overdoses of this toxic alkaloid. Additionally, therapeutic mouse mAbs directed against the tricyclic anti-depressant, desipramine (15), and the abused psychosis-inducing drug, phencyclidine (16), are under development. Unfortunately, in general, sheep and goat polyclonal Fabs can present problems with respect to the ability to reproducibly generate high affinity Abs, which then must be purified from collections of animal sera. Further, the use of these foreign proteins in patients leads to an immunological response that can cause hypersensitivity reactions and acute anaphylaxis. Mouse mAbs, while providing a single well-characterized product, have very short half-lives in humans and are especially prone to generate anti-idiotypic, inactivating Abs as well as anti-constant region responses. Therefore, drug-targeted immunological intervention has been restricted by the lack of reliable methods to produce drug-specific human Abs that would reduce the risk of sensitizing patients and allow for more than a one time only intervention during life-threatening crises. To address this general problem, in recent years, considerable efforts have been made to generate mouse-human chimeric, humanized, and primatized mAbs (17, 18, 19, 20) that are more suitable for human use.

As an alternative and potentially more useful approach, recently we (21, 22, 23, 24, 25, 26) and others (27, 28, 29, 30) have adopted the strategy of humanizing the mouse humoral immune system. We have been able to develop several unique strains of genetically altered mice with inactivated endogenous μ heavy and κ light chain loci and inserted human heavy (constant regions, Cμ and Cγ) and κ light chain transgenes (21, 22, 23, 24). This allows the generation of mouse B cells that are capable of responding to immunization and undergo heavy chain class switching and somatic mutation to generate human IgG1 κ Abs. Our initial results indicate that these animals are capable of generating human Abs against a variety of human and nonhuman proteins and that standard hybridoma technology can be used to obtain human mAbs with affinities and quantities that compare favorably with those of murine mAbs (24, 25, 26).

We now report the first use of one of these transgenic mouse strains, HC2/KCo5 (24), to generate hybridoma-secreted human mAbs directed against the low m.w., nonpeptidic hapten, digoxin. In this paper we describe the partial characterization of six of the eight human anti-digoxin mAbs we have obtained to date. Three of these mAbs have been studied more thoroughly, and determination of the ability of eight digoxin-related cardiotonic steroids and 10 steroid hormones to compete with digoxin for mAb binding has shown them to have a fine specificity of binding comparable to that of normal mouse mAbs. Further, radioligand binding studies have shown them to have low nanomolar binding affinities for digoxin that may prove sufficient for them to be clinically useful.

The mice [(C57BL/6J × CBA/J)F2] used in this study have undergone four distinct genetic modifications, resulting in double-transgenic/double-deletion mice that have been described previously (25). The disruption of the endogenous mouse μ heavy chain production (designated the CμD strain) results from the insertion of a neomycin resistance gene into the μ coding region (N. Lonberg, D. Fishwild, and L. D. Taylor, manuscript in preparation). The κ light chain disruption (JCκ D strain) results from recombinant deletion of Jκ and Cκ gene segments (24). The constructed human sequence heavy chain minilocus, transgene, designated HC2, that rescues B cell development in the mutant background animals, includes four VH, 16 D, six JH gene segments, and Cμ and Cγ1. The KCo5 light chain transgene contains the KCo4 transgene of four Vκ, with all five Jκ gene segments and Cκ as well as a 450-kb yeast artificial chromosome (YAC)3 that contains most of the remaining distal portion of the human Vκ region (25).

Digoxin was conjugated to BSA, chicken OVA, and keyhole limpet hemocyanin (KLH) using the method described by Butler and Chen (6). The approximate extent of digoxin covalent coupling per mole (or milligram) of protein was determined spectrophotometrically in 83% H2SO4, with the absorption of the hapten-protein conjugate at 388 and 465 nm compared with that of protein and digoxin (6). The extent of hapten coupling was 8 and 2 mol of digoxin/mol of BSA and OVA, respectively, and 0.1 mmol/mg KLH.

Fifteen transgenic mice were immunized initially with either 100 μg of a digoxin-KLH conjugate suspended in CFA via i.p. injection or 50 μg of digoxin-KLH suspended in TiterMax (TM) via s.c. injection. The mice receiving immunogen in CFA were then immunized i.p. twice with 20 μg of digoxin-KLH in IFA followed by weekly or biweekly injections of 20 μg of digoxin-OVA in IFA. The mice receiving immunogen in TM were subsequently immunized s.c. approximately monthly with 50 μg of digoxin-OVA in TM. All mice received a final i.v. boost of 20 μg of digoxin-OVA in PBS 3 days before fusion. Splenic lymphocyte suspensions were fused to P3 × 63-Ag8.653 nonsecreting mouse myeloma cells (American Type Culture Collection, Manassas, VA; CRL 1480), and hybridomas selected as previously described (25).

An ELISA was used to screen for hybridoma-secreted human anti-digoxin Abs. Microtiter plate wells were coated overnight at 4°C with 100-μl aliquots of 10 μg/ml of digoxin-BSA in 10 mM PBS, pH 7.2. Thereafter, the ELISA was completed as previously described (24). The hybridomas from positive-testing plate wells were then subcloned by limiting dilution plating.

The stability of cloned hybridoma cell lines was established by several weeks of in vitro culturing and repeated testing of spent culture medium for the presence of secreted mAb. Samples containing ∼1–10 × 106 cells in log phase growth were injected i.p. into pristane-treated SCID mice for growth as ascites cells. The ascites fluids were then removed from animals, centrifuged, and sterile membrane filtered, and non-Ig proteins were removed by 0.8 M (NH4)2SO4 precipitation. The Ig fraction was equilibrated with 1 M (NH4)2SO4 and 0.1 M glycine, pH 8.0, and loaded onto a protein A-Sepharose column (Pharmacia, San Diego, CA). The mAbs were eluted with 0.1 M glycine and 0.1 M NaCl, pH 3.0, neutralized with 1 M Tris base, and dialyzed against PBS. The OD at 280 nm was determined, and an absorptivity coefficient of 1.4 was used to calculate the protein concentration (31). Lowry protein determinations (32) of mAb concentrations performed using BSA as standard matched well with the OD-calculated values.

The relative levels of serum human IgM and IgG anti-digoxin Abs were determined at weekly intervals during immunization using the ELISA described by Fishwild et al. (25). In addition, mouse serum samples obtained at the time of cell fusion were monitored for relative levels of both human and mouse IgG and IgM anti-digoxin Abs. Serum samples were exposed to plate-adsorbed digoxin-BSA, and biotinylated goat anti-mouse and anti-human heavy chain-specific secondary Abs were used to detect the hapten-bound Ab. A streptavidin-alkaline phosphatase conjugate was used to quantitate anti-digoxin Abs through its hydrolysis of the substrate p-nitrophenylphosphate. Total serum concentrations of mouse and human Abs were determined by adsorbing to the microtiter plates either goat anti-mouse Ig Fc specific or anti-human Igs Fc specific Abs to capture any mouse or human Ig chains present in the serum solutions. The captured Abs were probed using a collection of biotinylated goat anti-mouse and anti-human heavy and light chain-specific Abs.

Each cloned hybridoma cell line was tested for its production of a human anti-digoxin IgG κ Ab and the absence of any mouse chain Igs. Spent medium from hybridoma cultures was exposed to digoxin-BSA-coated ELISA plate wells, and bound Ab was probed with isotype- and species-specific, biotinylated Abs, including goat anti-human G, M, κ, and λ chains and goat anti-mouse G, M, A, κ, and λ chain-specific Abs. Next, the media were tested for any Ab binding to the carrier proteins coupled with digoxin, including BSA, chicken OVA, and KLH or to turkey OVA, chicken serum albumin, and casein. Finally, as described above for the serum samples, goat anti-mouse IgG Fc-specific and anti-human IgG Fc-specific Abs were adsorbed to the microtiter plates to capture for detection and quantitation any mouse or human Ig chains present in the hybridoma supernatant solutions.

Avidity determinations.

The avidities of the six mAbs isolated for digoxin were determined using an ELISA in which the digoxin-BSA conjugate (5 μg/ml) was adsorbed to microtiter plates, which were then blocked using a 5% BSA/PBS solution followed by varying concentrations of culture medium or purified mAb. Then, biotinylated goat anti-human IgG Ab was added, followed by a streptavidin-alkaline phosphatase conjugate and then substrate to detect bound human Ab.

Competitive binding ELISAs.

Determination of the relative binding specificities of the six mAbs, their binding to digoxin, additional derivatives, and a variety of steroids was accomplished through use of a competitive binding ELISA. Anti-human IgG Fc region-specific Ab was adsorbed to the microtiter plates, and then the human mAbs were captured to the plates. Subsequently, a fixed concentration of a digoxin-alkaline phosphatase conjugate was mixed with varying concentrations of digoxin, serving as the standard, or other competitors, and binding of the digoxin-AP conjugate was determined. Briefly, goat anti-human IgG Fc-specific IgG (5 μg/ml) was adsorbed onto polystyrene (Corning, Corning, NY) plates in 0.1 M NaHCO3, pH 9.6, followed by a blocking step with 0.5% casein in 10 mM PBS buffer, pH 7.6, and 0.02% sodium azide and then the addition of 100 μl of mAb (2 μg/ml). The captured mAbs were incubated for 1 h at 37°C with 100 μl of a mixture of a 1/50 dilution of a digoxin-alkaline phosphatase conjugate (O.E.M. Concepts, Toms River, NJ) and varying competitor concentrations in 10 mM PBS, pH 7.2, and 0.5% casein. Following washing of the plate, 50 μl/well of the substrate solution (1 mg/ml p-nitrophenylphosphate, 50 mM NaHCO3, pH 9.8, and 1 mM MgCl2), was added to the plates and incubated at room temperature for 10 min. Ab-bound digoxin-AP was quantitated colorimetrically (405 nm) after adding 50 μl/well of 0.1 N NaOH. Data were analyzed using a nonlinear regression curve-fitting program Inplot (GraphPad, San Diego, CA).

First, radioligand binding assays using varying concentrations of [3H]digoxin were performed with the three purified anti-digoxin mAbs to obtain the dissociation constants (Kd). Next, competition binding studies were performed to obtain the IC50 or inhibition constants of digitoxin, digoxigenin, and progesterone relative to that of digoxin. These assays used a double Ab precipitation technique, previously described (33), to recover the [3H]ligand-mAb complexes on glass-fiber filters. For Kd determinations, 0.06–0.3 μg/0.5 ml assay of purified mAb was incubated at room temperature for 1 h with varying concentrations of [G-3H]digoxin (sp. act. = 15 Ci/mmol; DuPont-NEN, Boston, MA) in 0.5 ml of PBS, pH 7.4, containing 0.05% BSA. Then excess affinity-purified goat anti-human IgG Fc specific (ICN Biomedicals, Costa Mesa, CA) and rabbit anti-goat IgG Abs were added to each assay tube, and the binding reactions were allowed to go to completion (33). To determine nonspecific binding, a Na+,K+-ATPase-directed mAb (M7-PB-E9) (34) was substituted for the anti-digoxin mAbs. The Kd values for digoxin were obtained by analyzing the data using PRIZM (GraphPad, San Diego, CA), a nonlinear regression curve-fitting program. The competition curves were obtained by performing binding studies with a fixed 60-nM concentration of [3H]digoxin and varying concentrations of cold competitor. The IC50 values were obtained by fitting the radioligand binding data using Inplot (GraphPad). These IC50 values were converted to inhibitory dissociation constants (Ki values), using the Cheng and Prusoff (35) equation: Ki = IC50/(1 + [L]/Kd), where [L] is the concentration of [3H]digoxin, and Kd is the dissociation constant for digoxin.

In this work, HC2/KCo5 mice were immunized with the digoxin-carrier conjugates as described in Materials and Methods. All 15 animals generated a hapten-directed response, with eight being designated high responders by having serum titer values of >1/1250 when tested for human IgG Ab binding to digoxin-BSA by ELISA.

As illustrative of the animals’ responses, serum samples from two mice that were considered good producers of anti-digoxin Abs were characterized with respect to the timing and level of their responses. Mouse IgG as well as the human Ab responses were expected due to trans-switching from the recombined human VDJ sequence to the murine endogenous γ heavy chain and because the endogenous λ locus was not disrupted (24). As shown in Fig. 1, both initial human IgM and IgG anti-digoxin responses occurred within the first week after immunization, with the IgG levels peaking at about 2 wk and severalfold higher than the IgM response. In addition, the mouse IgG anti-digoxin response was monitored and was found to be about one-third to one-half that of the human IgG response. Further, serum samples obtained from these two transgenic animals at the time of cell fusions as well as that for two normal CB6F1/J mice (one immunized with digoxin-KLH and one control) were analyzed and compared. These results indicated that the sera from the transgenic mice 14739 and 14747 contained about 190 and 42 μg/ml, respectively, of total mouse IgG with essentially no mouse IgM. In comparison, the two normal mice had mouse IgG levels of 7.2 and 2.3 mg/ml and IgM levels of 0.89 and 0.39 mg/ml. The serum levels of human IgG for both transgenic mice were ∼11 μg/ml, while human IgM levels were 244 and 400 μg/ml, respectively. Thus, while the production of mouse IgG in the transgenic animals appeared to be reduced to only 2–3% that in normal mice, these levels still surpassed those of the human IgG. Endogenous IgM production was well below that of the human IgM. Interestingly, despite the fact that total endogenous mouse IgG levels surpassed those of the human IgG, the transgenic animals’ immunological response to immunization with the digoxin-carrier conjugate was predominately a human IgG response.

FIGURE 1.

Determination of digoxin-specific human Ab responses in transgenic mice. • and ▴, Human IgG anti-digoxin serum responses (dilutions, 1/1000) for mice 14739 and 14747, respectively, following immunization with digoxin-KLH conjugate. ○ and ▵, Human IgM responses.

FIGURE 1.

Determination of digoxin-specific human Ab responses in transgenic mice. • and ▴, Human IgG anti-digoxin serum responses (dilutions, 1/1000) for mice 14739 and 14747, respectively, following immunization with digoxin-KLH conjugate. ○ and ▵, Human IgM responses.

Close modal

The spleens from six of the 15 immunized animals were used for cell fusions. Hybridomas from two of the six animals generated detectable human IgG k anti-digoxin Abs in the culture medium. Altogether, nine anti-digoxin Ab-producing hybridoma cell populations were detected in parental wells, while all eight of the hybridomas chosen for further growth and subcloning were successfully isolated. Six of these hybridoma cell lines were then conditioned to grow as T-flask cultures, and late growth phase samples of the culture medium were tested for human mAb binding to the digoxin-BSA, -KLH, and -OVA conjugates. The human mAbs secreted by these cell lines bound equally well to all three hapten-protein conjugates. Further all six anti-digoxin mAbs were fully human IgG κ Igs, with no detection of mouse or human-mouse mixed chain Abs. In addition, no mAb binding was detected to bovine, chicken, or human serum albumin; KLH; chicken or turkey OVA; or casein. Ab binding dilution curves obtained using culture medium samples showed that the mAbs were secreted at concentrations that gave titer values, or half-maximal binding to carrier-linked digoxin at dilutions of ∼1/200 to 1/400. Ab concentrations were determined by capturing the mAbs from solution with plate-adsorbed Ab and comparing these levels to those obtained using purified human IgG as a standard. Human mAb concentrations in the medium were ∼5–25 μg/ml of human mAb. Dilution curves of mAb binding to digoxin-BSA showed all of them to have an estimated apparent binding affinity or avidity of about 0.5–1 nM (data not shown). These avidity values were in agreement with the values obtained for several high affinity mouse mAbs (obtained from Michael Margolies, Massachusetts General Hospital, Boston, MA) monitored using the same ELISA procedure (our unpublished observations).

Three mAbs, designated 5C2-4, 7F2-31, and 11E6-7 were obtained from mouse ascites samples and purified, and their binding avidities and specificities for digoxin were determined. As found previously for mAbs in cell culture supernatants, the ELISA procedure using digoxin-BSA as Ag gave essentially identical titer values of about 1 nM for all three mAbs (data not shown).

Next, a competitive ELISA protocol was used to determine the mAbs’ apparent affinities for digoxin. In this procedure, adsorbed anti-human IgG Fc region-specific Igs captured a fixed amount of mAb to the plates, and the IC50 values for digoxin were obtained by having digoxin compete in solution with digoxin-alkaline phosphatase for mAb binding. The IC50 values of mAbs 5C2-4, 7F2-31, and 11E6-7 for digoxin were ∼0.28, 0.15, and 0.15 μM, respectively (see representative Fig. 2 A). These results suggested that mAbs 7F2-31 and 11E6-7 were similar to each other but distinct from 5C2-4.

FIGURE 2.

ELISA determination of the binding specificity of human monoclonal anti-digoxin Ab 7F2-31. Inhibition curves show the abilities of various compounds to inhibit mAb 7F2-31 binding to a digoxin-alkaline phosphatase conjugate. Each point is the average of three determinations made during a single experiment. SEs were <5%. A, Results with increasing concentrations of digoxin (▪), digitoxin (○), digoxigenin (▴), progesterone (□), and ouabain (♦). B, Results with increasing concentrations of digoxin (▪), acetyl-strophanthidin (⋄), progesterone (□), testosterone (•), oleandrin (♦), and β-estradiol (○).

FIGURE 2.

ELISA determination of the binding specificity of human monoclonal anti-digoxin Ab 7F2-31. Inhibition curves show the abilities of various compounds to inhibit mAb 7F2-31 binding to a digoxin-alkaline phosphatase conjugate. Each point is the average of three determinations made during a single experiment. SEs were <5%. A, Results with increasing concentrations of digoxin (▪), digitoxin (○), digoxigenin (▴), progesterone (□), and ouabain (♦). B, Results with increasing concentrations of digoxin (▪), acetyl-strophanthidin (⋄), progesterone (□), testosterone (•), oleandrin (♦), and β-estradiol (○).

Close modal

Once the mAbs’ affinity values for digoxin were obtained, their fine binding specificities for eight additional cardiotonic steroids (see Fig. 3) were investigated by having these compounds compete with the digoxin-alkaline phosphatase conjugate for binding to immobilized mAb. Table I (and representative data, Fig. 2, A and B) presents the compounds relative binding affinities. In brief, the three mAbs were similar in that each had about a 3-fold lower affinity for digitoxin (distinguished by the absence of the C12-OH group on steroid ring C) than digoxin. In addition, the digitoxose sugars clearly contributed to the binding of each mAb, since average 5- and 8-fold losses in the mAb binding affinities to the digoxin and digitoxin aglycones, respectively, were observed. The similarity between mAbs 7F2-31 and 11E6-7 was further extended in that they both were more sensitive to the reduction of the double bond in the lactone ring (dihydrodigitoxin) than 5C2-4, while 5C2-4 binding was more affected by acetylation of the aglycone, i.e., acetyl strophanthidin. Interestingly, all three mAbs showed a similar decreased affinity for the toad-derived bufalin with its six-member lactone, while 5C2-4 showed somewhat poorer binding to the oleander plant-derived glycoside, oleandrin (C-16 acetate). The fine specificities of these three mAbs for substituents on the steroid moiety of digoxin was further illustrated by the fact that the more hydrophilic, but still cardioactive, glycoside, ouabain (three additional hydroxyl groups and a single sugar), showed virtually no competition with digoxin.

FIGURE 3.

Chemical structures.

FIGURE 3.

Chemical structures.

Close modal
Table I.

Ratio of IC50 values of cardiac glycosides and derivatives relative to digoxina

Cardiac GlycosideAnti-Digoxin Human mAbs
5C2-47F2-3111E6-7
Digoxin 
Digoxigenin 4.6 5.5 4.5 
Digitoxin 3.1 2.8 2.4 
Digitoxigenin 21 24 20 
Dihydrodigitoxin 48 485 270 
Acetylstrophanthidin 35 15 18 
Bufalin 350 420 410 
Ouabain No binding No binding No binding 
Oleandrin No binding 1055 1260 
Cardiac GlycosideAnti-Digoxin Human mAbs
5C2-47F2-3111E6-7
Digoxin 
Digoxigenin 4.6 5.5 4.5 
Digitoxin 3.1 2.8 2.4 
Digitoxigenin 21 24 20 
Dihydrodigitoxin 48 485 270 
Acetylstrophanthidin 35 15 18 
Bufalin 350 420 410 
Ouabain No binding No binding No binding 
Oleandrin No binding 1055 1260 
a

The values reported here are the ratios of molar concentrations of the compound required to give 50% inhibition (IC50) of digoxin-alkaline phosphatase binding to each mAb, relative to the IC50 value for digoxin (competitive binding ELISAs are described in Materials and Methods). Ratios are the averages of at least two separate determinations.

In addition to determining the binding specificities of the mAbs to various cardiac glycosides we determined the extent to which these mAbs cross-react with steroid hormones (see Fig. 3). As shown in Table II (and representative data, Fig. 2, A and B), the mAbs showed little or no binding to either the steroid precursor cholesterol or most of the 10 steroid hormones tested. Progesterone and testosterone, however, were recognized by all three mAbs with about a 100-fold reduction in affinity relative to digoxin, while 5C2-4 was distinct from the other two mAbs, with a 2-fold higher affinity for corticosterone.

Table II.

Ratio of IC50 values of steroids relative to that of digoxina

SteroidsAnti-Digoxin Human mAbs
5C2-47F2-3111E6-7
Digoxin 
Cholesterol No binding No binding No binding 
Progestins/androgens/estrogens    
Progesterone 95 101 132 
Testosterone 217 116 132 
Androsterone 931 1875 1500 
Androstenediol 731 1750 1500 
β-Estradiol No binding No binding No binding 
Corticosteroids    
Corticosterone 150 333 367 
Dehydrocortisone (prednisone) 1667 No binding No binding 
Cortisone No binding No binding No binding 
Hydrocortisone (cortisol) 596 No binding 2670 
Aldosterone 676 1458 1500 
SteroidsAnti-Digoxin Human mAbs
5C2-47F2-3111E6-7
Digoxin 
Cholesterol No binding No binding No binding 
Progestins/androgens/estrogens    
Progesterone 95 101 132 
Testosterone 217 116 132 
Androsterone 931 1875 1500 
Androstenediol 731 1750 1500 
β-Estradiol No binding No binding No binding 
Corticosteroids    
Corticosterone 150 333 367 
Dehydrocortisone (prednisone) 1667 No binding No binding 
Cortisone No binding No binding No binding 
Hydrocortisone (cortisol) 596 No binding 2670 
Aldosterone 676 1458 1500 
a

See legend for Table I.

Although the data are not shown, the (nonpurified) mAbs from the three additional cloned hybridomas, designated 3E4, 10B1, and 5D8 were similarly tested for their binding avidities to the plate-adsorbed digoxin-BSA conjugate, and their binding specificities were determined using the competitive ELISA procedure. These Abs each gave similar, ∼1 nM titer values for binding to the digoxin-BSA conjugate and IC50 ELISA values of 0.17, 0.12, and 0.18 μM, respectively, for digoxin. In the competition ELISA they had similar digoxin relative IC50 ratios for digitoxin and digoxigenin of ∼3.3 and 1.7, respectively, with no binding to ouabain. For the steroids, determination of their relative IC50 values showed testosterone to have an ∼150-fold lower affinity than digoxin, with cortisone being weakly inhibitory. In addition, cholesterol, β-estradiol, and androsterone showed essentially no competition with digoxin. Thus, pending a more complete repetition of these ELISAs with purified Ab and determination of their Kd values and amino acid sequences, these three mAbs, while not identical, were more similar to 7F2-31 and 11E6-7 than to 5C2-4.

To obtain true Kd or affinity values of the purified mAbs for digoxin, we used a radioligand binding assay, which employed a double Ab aggregation step to generate a [3H]ligand-three Ab complex that was recovered on glass-fiber filters. As shown, in Table III (and illustrated in Fig. 4), the affinities of mAbs 7F2-31, 11E6-7, and 5C2-4 for digoxin were 2.5, 4.5, and 22 nM, respectively. Determination of their Ki values for digoxin, digoxigenin, digitoxin, and progesterone showed the three mAbs to have nanomolar affinities for digoxin, digoxigenin, and digitoxin and micromolar values for progesterone. mAb 5C2-4 was clearly distinguished from the other two mAbs by having the poorest affinity for all four compounds tested, while 7F2-31 showed nearly identical affinities for digoxin and digitoxin. Further, there were sufficient differences between the Ki values obtained for 7F2-31 and 11E5-7 to suggest that, while similar, they are distinct mAbs. Interestingly, all three mAbs have considerably less cross-reactivity with the steroid progesterone than was apparent from the competitive ELISA. These results suggest that in any in vivo usage these mAbs would not alter endogenous steroid hormone levels.

Table III.

The determination of affinity and inhibition constants for human anti-digoxin mAbsa

Anti-Digoxin Human mAbsKd (nM), DigoxinKi (nM)
DigoxinDigitoxinDigoxigeninProgesterone
5C2-4 22 11 50 140 2400 
7F2-31 2.5 2.6 3.3 7.2 1200 
11E6-7 4.5 2.7 14 27 670 
Anti-Digoxin Human mAbsKd (nM), DigoxinKi (nM)
DigoxinDigitoxinDigoxigeninProgesterone
5C2-4 22 11 50 140 2400 
7F2-31 2.5 2.6 3.3 7.2 1200 
11E6-7 4.5 2.7 14 27 670 
a

The values given represent the averages obtained from two competitive radioligand binding assays (see Materials and Methods).

FIGURE 4.

Determination of inhibition constants (IC50) for cardiac glycosides and hormone binding to human mAb 7F2-31. The curves show the ability of increasing concentrations of digoxin (▪), digoxigenin (▴), digitoxin (•), and progesterone (□) to compete with [3H]digoxin (60 nM) for binding to mAb 7F2-31. Each point is the average of three determinations from a single experiment.

FIGURE 4.

Determination of inhibition constants (IC50) for cardiac glycosides and hormone binding to human mAb 7F2-31. The curves show the ability of increasing concentrations of digoxin (▪), digoxigenin (▴), digitoxin (•), and progesterone (□) to compete with [3H]digoxin (60 nM) for binding to mAb 7F2-31. Each point is the average of three determinations from a single experiment.

Close modal

A sizable number of patients have been successfully rescued from digoxin overdoses by the infusion of one of two commercial preparations of polyclonal anti-digoxin Fabs (Digibind, Glaxo Wellcome, Research Triangle Park, NC; and Digidot, Boehringer Mannheim, Indianapolis, IN) (10, 11, 36, 37). In addition, because of the diversity of Abs present in these preparations they have also been used to treat poisoning by toad venoms (38), a Chinese herbal medicine/aphrodisiac, and the oleander plant (3). However, despite their clinical successes these preparations appear underutilized, perhaps due to their cost and concerns about the use of foreign species Abs in patients (39). In addition, many mouse anti-digoxin mAbs have been generated and used for the quantitation of serum digoxin levels in patients. A few mouse mAbs such as 35-20 and 26-10 isolated by Margolies et al. (40, 41) have been also been shown to be effective for the reversal of digoxin intoxication in animal models (42, 43). However, no anti-digoxin mAb has been used in the clinic.

In this work we report the production of eight mouse hybridomas, which secrete human sequence anti-digoxin mAbs, and initial avidity and fine binding specificity determinations for six of these mAbs. ELISA competition binding studies using digoxin and related cardioactive steroids as well as steroid hormones demonstrated the importance of all three structural components of digoxin, the digitoxose sugars, steroid moiety, and lactone ring for mAb recognition. In addition, radioligand binding studies showed three of the mAbs, those designated 5C2-4, 7F2-31, and 11E6-7, to have affinities for digoxin from about 2–20 nM with relatively good recognition of digoxigenin and digitoxin but greatly reduced affinities for progesterone, the steroid hormone that showed the highest degree of mAb binding.

Analysis of the ELISA competition binding data obtained for the three more fully characterized mAbs enables us to make some initial conclusions about their binding specificities. These results show that the sugar moiety (at least the primary sugar) contributes to binding of the mAbs, because the aglycones, digoxigenin and digitoxigenin, had ∼5- and 8-fold lower binding affinities than digoxin and digitoxin, respectively. As for the contributions of substituents on the steroid moiety of digoxin, the removal of the C-12 hydroxyl group (digitoxin) results in an ∼3-fold reduction in the binding affinity for all three mAbs. In contrast, with acetylstrophanthidin, the presence of an acetyl and hydroxyl group at the aglycone’s C-3 and C-5 groups (ring A) and the conversion of the C-19 methyl to a keto group had surprisingly little additional effect on mAb binding. The results obtained with ouabain and oleandrin, however, show that the presence of additional hydroxyl groups (C-1, C-3, and C-11, ouabain) or an acetate group at the steroid C-16 (oleandrin) essentially abolishes recognition by these Abs. The data obtained with the steroids then showed that the addition of a hydroxyl group (corticosterone vs hydrocortisone) at C-17 or the oxidation of the C-18 methyl to a keto group (corticosterone vs aldosterone) results in a 4- to 10-fold drop in mAb affinities from that for corticosterone. In addition, removal of the C-19 methyl group and conversion of the steroid ring A from a keto-hexyl (testosterone) to an aromatic phenol (β-estradiol) essentially abolished the binding for all three mAbs. Therefore, these human mAbs seem to have extensive complementary shape interactions with much of what, as conventionally presented in Fig. 3, would be the upper side or face of the steroid moiety of digoxin.

It is also evident that the unsaturated five-member lactone ring is an important binding determinant as reduction of the C20-C22 double bond (dihydrodigitoxin) reduces digitoxin’s affinity for 5C2-4, 7F2-31, and 11E6-7, 16-, 170-, and 100-fold, respectively. Also, substitution with a six- rather than a five-member unsaturated lactone as occurs in bufalin (a digitoxigenin, congener) results in an ∼18-fold decrease in affinity for all three mAbs. A further indication that these mAbs may have defined binding pockets for the lactone within their Ag-combining sites analogous to those determined, for mouse mAbs 26-10 (44) and 40-50 (45), is the fact that cholesterol has essentially no ability to compete with digoxin, but replacement of its eight-carbon alkyl chain on ring D (C-17) with either a ketomethyl (progesterone) or a hydroxyl group (testosterone) enables the steroid moiety to inhibit digoxin binding, albeit with a 100-fold lower affinity than digoxin.

In the absence of confirmatory sequence information we do not know the extent to which the six mAbs reported here differ or are closely related, because they originate from one animal. However, 5C2-4 appears distinct from 7F2-31 and 11E6-7 as its IC50 and Kd values for digoxin were about 2- and 10-fold higher than those for the other two mAbs. Moreover, 7F2-31 and 11E6-7 appear distinct based on their Ki values for digoxigenin, digitoxin, and progesterone. Further testing of all six mAbs will allow us to determine whether this mouse yielded only three or up to six distinct mAbs, a possible testament to the repertoire diversity of the transgenic animal.

We can, however, make some general comparisons between these mAbs and the mouse anti-digoxin mAbs characterized by others. First, the low nanomolar affinities of these human mAbs for digoxin compare very well with mAbs obtained from the BALB/c mouse strain (46, 47). These values are, however, at least 10-fold poorer than the majority of mAbs obtained by Margolies and colleagues (40, 41) from the A/J strain of mice, which they report to be unique with respect to its high affinity anti-digoxin responses. Next, we can compare the competition binding data for the human mAbs with the results obtained by Margolies et al. (40) with 14 mouse mAbs. First, with respect to the ability of the human mAbs to distinguish between digoxin and digitoxin we found that the 3-fold loss in the affinity of the human mAbs for digitoxin more than matches the average 1.4-fold difference in affinity observed for nine of the mouse mAbs. Then, with respect to the recognition of the sugar moiety by the human mAbs we found that the 5- and 8-fold reductions in their binding to the digoxin and digitoxin aglycones, respectively, are comparable with those of a group of six mouse mAbs, which averaged ∼2.6- and 9-fold decreases in their affinities for the aglycones. Further, the decreased affinities of the three human mAbs for acetylstrophanthidin with essentially no binding to ouabain, oleandrin, and most steroids are also consistent with the range of specificities observed for the A/J mouse-derived Abs. Interestingly, a chimpanzee-derived anti-digoxin mAb (48) has also been characterized and found to have a 4-fold affinity differential between digoxin and digitoxin but no recognition of the digitoxose sugars.

Finally, we would like to consider the likelihood of mAb 7F2-31 having the potential to be clinically useful for digoxin detoxification. Although most animal detoxification studies have used mouse mAbs (42, 43) with about a 10-fold higher affinity than mAb 7F2-31, one study has directly compared the in vivo effects of three mouse mAbs with differing digoxin affinities. Cano et al. (46) tested the abilities of these mAbs to raise total plasma digoxin and reduce free digoxin levels by reversing the drug’s normal tissue distribution in rats. Interestingly, they report that at a stoichiometric mAb/drug ratio, mAb 6C9 with a digoxin affinity of 3.2 nM was able to reduce free plasma levels by 90% compared with the 99% achieved by mAb 1C10, which has an affinity of 0.17 nM. Furthermore, by raising the concentration of mAb 6C9 to a 5:1 ratio vs drug, it was found to be equally as effective as 1C10. In contrast, the third mAb, with a Kd value of 40 nM, was essentially ineffective.

Comparison of the affinity of mAb 7F2-31 with clinically used Fab preparations provides encouraging information. Sheep Fabs with affinities ranging from 1–0.01 nM (49, 50) have all been effective when administered at stoichiometric concentrations relative to patient’s plasma drug levels. In addition, Fab preparations that may have had Kd values above 1 nM have proven effective (37). Therefore, mAb 7F2-31 may prove adequate for initial treatment of the massive drug overdose levels of digoxin (or digitoxin), for which anti-digoxin Fab intervention is currently recommended (50). Clearly, both mAbs 7F2-31 and 11E6-7 have affinities that are at least 10- to 30-fold greater than the Kd values of the Na+ pump for digoxin under in vivo ligand conditions (51, 52), and they should compete effectively with the Na+ pump for digoxin.

We thank Mr. Lamar Gerber for his technical assistance and construction of Fig. 3.

1

This work was supported by National Institutes of Health Grant 1R41 HL57039 (to W.J.B. and D.F.) and National Institutes of Health Training Grant 5T32HL07382 (to R.K.).

3

Abbreviations used in this paper: YAC, yeast artificial chromosome; KLH, keyhole limpet hemocyanin; TM, Titer-Max.

1
Repke, K. R. H., R. Megges, J. Weiland, R. Schön.
1995
. In search of ideal inotropic steroids.
Angew Chem. Int. Ed. Engl.
34
:
282
2
Kernan, W. N., J. Castellsague, G. D. Perlman, A. Ostfeld.
1994
. Incidence of hospitalization for digitalis toxicity among elderly Americans.
Am. J. Med.
96
:
426
3
Marcus, F. I..
1992
. Use and toxicity of digitalis.
Heart Dis. Stroke
1
:
27
4
Smith, T. W., E. M. Antman, P. L. Friedman, E. M. Blatt, J. G. Marsh.
1984
. Digitalis glycosides: mechanisms and manifestations of toxicity. III.
Prog. Cardiovasc. Dis.
26
:
413
5
Mahdyoon, H., G. Battilana, H. Rosman, S. Goldstein, M. Gheorghiade.
1990
. The evolving pattern of digoxin intoxication: observations at a large urban hospital from 1980 to 1988.
Am. Heart J.
120
:
1189
6
Butler, V. P., Jr, J. P. Chen.
1967
. Digoxin-specific antibodies.
Proc. Natl. Acad. Sci. USA
57
:
71
7
Smith, T. W., V. P. Butler, Jr, E. Haber.
1969
. Determination of therapeutic and toxic serum digoxin concentrations by radioimmunoassay.
N. Engl. J. Med.
281
:
1212
8
Schmidt, D. H., V. P. Butler, Jr.
1968
. Immunological protection against digitalis toxicity.
Circulation
38(Suppl. 6)
:
VI-174
9
Smith, T. W., E. Haber, L. Yeatman, V. P. Butler, Jr.
1976
. Reversal of advanced digoxin intoxication with Fab fragments of digoxin-specific antibodies.
N. Engl. J. Med.
294
:
797
10
Butler, V. P., Jr, D. H. Schmidt, T. W. Smith, E. Haber, B. D. Raynor, P. DeMartini.
1977
. Effects of sheep digoxin-specific antibodies and their Fab fragments on digoxin pharmacokinetics in dogs.
J. Clin. Invest.
59
:
345
11
Antman, E. M., T. L. Wenger, V. P. Butler, Jr, E. Haber, T. W. Smith.
1990
. Treatment of 150 cases of life-threatening digitalis intoxication with digoxin-specific Fab antibody fragments: final report of a multicenter study.
Circulation
81
:
1744
12
Hickey, A. R., T. L. Wenger, V. P. Carpenter, H. Tilson, M. Hlatky, C. Furberg, C. Kirkpatrick, H. Strauss, T.W. Smith.
1991
. Digoxin immune Fab therapy in the management of digitalis intoxication: safety and efficacy results of an observational surveillance study.
J. Am. College Cardiol.
17
:
590
13
Ujhelyi, M. R., S. Robert.
1995
. Pharmacokinetic aspects of digoxin-specific Fab therapy in the management of digitalis toxicity.
Clin. Pharmacokinet.
28
:
483
14
Baud, F., A. Sabouraud, E. Vicaut, P. Taboulet, J. Lang, C. Bismuth, J. Rouzioux, J.-M. Scherrmann.
1995
. Brief report: treatment of severe colchicine overdose with colchicine-specific Fab fragments.
N. Engl. J. Med.
332
:
642
15
Pentel, P. R., D. E. Keyler.
1995
. Drug-specific antibodies as antidotes for tricyclic antidepressant overdose.
Toxicol. Lett.
82/83
:
801
16
Owens, S. M..
1997
. Antibodies as pharmacokinetic and metabolic modifiers of neurotoxicity.
NIDA Res. Monogr.
173
:
259
17
Tempest, P., P. Bremmer, M. Lambert, G. Taylor, J. Furze, F. Carr, W. J. Harris.
1991
. Reshaping a human monoclonal to inhibit human RSV infection in vivo.
BioTechnology
9
:
266
18
Jones, P. T., P. H. Dear, J. Foote, M. S. Neuberger, G. Winter.
1986
. Replacing the complementarity-determining regions in a human antibody with those from a mouse.
Nature
321
:
522
19
Rieckmann, L., M. Clark, H. Waldmann, G. Winter.
1988
. Reshaping human antibodies for therapy.
Nature
332
:
323
20
Solinger, A. M., A. Trunch, J. A. Lipani, R. A. Newman.
1997
. The therapeutic potential of a primatized anti-CD4 monoclonal antibody. W. J. Harris, Jr, and J. R. Adair, Jr, eds.
Antibody Therapeutics
341
CRC Press, New York.
21
Taylor, L. D., C. E. Carmack, S. R. Schramm, R. Mashayekh, K. M. Higgens, C.-C. Kuo, C. Woodhouse, R. Kay, N. Lonberg.
1992
. A transgenic mouse that expresses a diversity of human sequence heavy and light chain immunoglobulins.
Nucleic Acids Res.
20
:
6287
22
Tuaillon, U., L. D. Taylor, N. Lonberg, P. W. Tucker, J. D. Capra.
1993
. Human immunoglobulin heavy-chain minilocus recombination in transgenic mice: gene-segment use in μ and γ transcripts.
Proc. Natl. Acad. Sci. USA
90
:
3720
23
Taylor, L. D., C. E. Carmack, D. Huszar, K. M. Higgins, R. Mashayekh, G. Sequar, S. Schramm, C.-C. Kud, S. O’Donnell, R. Kay, et al
1994
. Human immunoglobulin transgenes undergo rearrangement, somatic mutation and class switching in mice that lack endogenous IgM.
Int. Immunol.
6
:
579
24
Lonberg, N., L. D. Taylor, F. A. Harding, M. Trounstine, K. M. Higgins, S. R. Schramm, S. Bernhard, D. Jones, R. Kay, K. Higgins, S. Schramm, N. Lonberg.
1994
. Antigen-specific human antibodies from mice comprising four distinct genetic modifications.
Nature
368
:
856
25
Fishwild, D. M., S. O’Donnell, T. Bengoechea, D. J. Hudson, F. Harding, et al
1996
. High-avidity human IgG κ monoclonal antibodies from a novel strain of minilocus transgenic mice.
Nat. Biotechnol.
14
:
845
26
Fishwild, D. M., S. O’Donnell, D. Hudson, T. Bengoechea, T. Hida, M. Katayama, T. Yamauchi, S. Qin, C. Mackey and N. Londerg. 1997. High avidity human IgG mAbs from novel transgenic mice. In Antibody Engineering, Vol. 2: New Technology, Application and Commercialization. W. Hori, ed. IBC, Southborough, MA, p. 187.
27
Wagner, S. D., A. V. Popov, S. L. Davies, J. Xian, M. S. Neuberger, M. Bruggemann.
1994
. The diversity of antigen-specific monoclonal antibodies from transgenic mice bearing human immunoglobulin gene miniloci.
Eur. J. Immunol.
24
:
2672
28
Popov, A. W., C. Butzler, J.-P. Frippiat, M.-P. Lefranc, M. Bruggemann.
1997
. Assembly and extension of yeast artificial chromosomes.
Gene
177
:
195
29
Green, L. L., M. C. Hardy, C. E. Marynard-Currie, H. Tsuda, D. M. Louie, M. J. Mendy, H. Abderrahim, M. Noguchi, D. Smith, Y. Zeng, et al
1994
. Antigen-specific human monoclonal antibodies from mice engineered with human Ig heavy and light chain YACs.
Nat. Genet.
7
:
14
30
Mendez, M., L. Green, J. Corvalin., X.-C. Jia, C. E. Maynard-Currie, X.-D. Yang, M. Gallo, D. Louie, D. Lee, K. Erickson, et al
1997
. Functional transplant of megabase human immunoglobulin loci recapitulates human antibody response in mice.
Nat. Genet.
15
:
146
31
Harlow, E., and D. Lane. 1988. Antibodies: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, p. 290.
32
Lowry, O. H., N. J. Rosenbrough, A. L. Farr, R. J. Randall.
1951
. Protein measurement with the Folin phenol reagent.
J. Biol. Chem.
193
:
265
33
Kasturi, R., J. Yuan, L. McLean, M. Margolies, W. J. Ball.
1998
. Identification of a model cardiac glycoside receptor.
Biochemistry
37
:
6658
34
Malik, B., G. Jamieson, W. J. Ball.
1993
. Identification of the amino acids comprising a surface-exposed epitope within the nucleotide-binding domain of the Na+,K+-ATPase using a random peptide library.
Protein Sci.
2
:
2103
35
Cheng, Y.-C., W. H. Prusoff.
1973
. Demonstration of the heterogeneity of nucleoside diphosphokinase in rat tissues.
Biochem. Pharmacol.
22
:
3099
36
Wenger, T., V. P. Butler, E. Haber, T. W. Smith.
1985
. Treatment of 63 severely digitalis-toxic patients with digoxin-specific antibody fragments.
J. Am. College Cardiol.
5
:
118A
37
Schaumann, W., B. Kaufmann, P. Neubert, A. Smolarz.
1986
. Kinetics of the Fab fragments of digoxin antibodies and of bound digoxin in patients with severe digoxin intoxication.
Eur. J. Clin. Pharmacol.
30
:
527
38
Brubacher, J. R., P. R. Ravikumar, T. Bania, M. B. Heller, R. S. Hoffman.
1996
. Treatment of toad venom poisoning with digoxin-specific Fab fragments.
Chest
110
:
1282
39
Borron, S., C. Bismuth, J. Muszynski.
1997
. Advances in the management of digoxin toxicity in the older patient.
Drugs Aging
10
:
18
40
Mudgett-Hunter, M., M. N. Margolies, H. Ju, E. Haber.
1982
. High affinity monoclonal antibodies to the cardiac glycoside digoxin.
J. Immunol.
129
:
1165
41
Mudgett-Hunter, M., W. Anderson, E. Haber, M. N. Margolies.
1985
. Binding and structural diversity among high-affinity monoclonal anti-digoxin antibodies.
Mol. Immunol.
22
:
477
42
Margolies, M. N., M. Mudgett-Hunter, T. W. Smith, J. Novotny, E. Haber.
1981
. Monoclonal antibodies to the cardiac glycoside digoxin. G. Hammerling, Jr, and U. Hammerling, Jr, and G. F. Kearney, Jr, eds.
Monoclonal antibodies and T-cell hybridomas
367
Elseiver, Amsterdam.
43
Lechat, P., M. Mudgett-Hunter, M. N. Margolies, E. Haber, T. W. Smith.
1984
. Reversal of lethal digoxin toxicity in guinea pigs using monoclonal antibodies and Fab fragments.
J. Pharmacol. Exp. Ther.
229
:
210
44
Jeffrey, P. D., R. K. Strong, L. C. Suker, C. Y. Chang, R. L. Campbell, G. A. Petsko, L. Sieker, G. Petsko, E. Haber, M. Maryolies, et al
1993
. 26–10 Fab-digoxin complex: affinity and specificity due to surface complementarity.
Proc. Natl. Acad. Sci. USA
90
:
10310
45
Jeffrey, P. D., J. F. Schildbach, C. Y. Chang, P. H. Kussie, M. N. Margolies, S. Sheriff.
1995
. Structure and specificity of the anti-digoxin antibody 40–50.
J. Mol. Biol.
248
:
344
46
Cano, N. J., A. E. Sabouraud, K. Benmoussa, F. Roquet, I. Navarro-Teulon, J.-C. Mani, J.-M. G. Scherrmann.
1995
. Monoclonal digoxin-specific antibodies induce dose- and affinity-dependent plasma digoxin redistribution in rats.
Pharm. Res.
12
:
709
47
Wahyono, D., M. Piechaczyk, C. Mourton, J.-M. Bastide, B. Pau.
1990
. Novel anti-digoxin monoclonal antibodies.
Hybridoma
9
:
619
48
Ehrlich, P., Z. Moustafa, J. Justice, K. Hartfeldt, I. Gadi, L. Sciorra, F. Uhl, C. Isaacson, L. Ostberg.
1988
. Human and primate monoclonal antibodies for in vivo therapy.
Clin. Chem.
34
:
1681
49
Smith, T. W., V. P. Butler, E. Haber.
1970
. Characterization of antibodies of high affinity and specificity for the digitalis glycoside digoxin.
Biochemistry
9
:
331
50
1999. Physicians’ Desk Reference. 53rd Ed. R. Arky, Med. consult. Medical Economics Data Production Co., Montvale, NJ, pp. 1110–1112.
51
Sun, Y., W. J. Ball, Jr.
1992
. Determination of Na+-K+-ATPase α- and β-isoforms and kinetic properties in mammalian liver.
Am. J. Physiol.
262
:
C1491
52
Wallick, E. T., B. J. Pitts, L. K. Lane, A. Schwartz.
1980
. A kinetic comparison of cardiac glycoside interactions with Na+,K+-ATPases from skeletal and cardiac muscle and from kidney.
Arch. Biochem. Biophys.
202
:
442