Replacing the complementarity-determining regions in a human antibody with those from a mouse

Immunotoxins: new therapeutic reagents for autoimmunity, cancer, and AIDS

Immunotoxins consist of cell-reactive ligands coupled to toxins or their toxic subunits. The ligands are usually antibodies, hormones, or growth factors and the toxins are of bacterial or plant origin. In vitro studies using A chain-containing immunotoxins specifically to kill tumor cells were successful and led to further experiments in vivo. Such studies, carried out over the past 5 years in both animals and humans, have demonstrated that the efficacy of immunotoxins in vivo is often poor, due to problems involving instability of the conjugate, inferior potency, inaccessibility of tumor cells, nonspecific binding to cells other than the target cells, and survival of antigen-negative mutants. In addition, immune responses against both the ligand and the A chain are usually elicited, precluding repeated therapy. During the past several years, there have been attempts to solve these problems and develop more effective immunotoxins.

Identification of human antibody fragment clones specific for tetanus toxoid in a bacteriophage lambda immunoexpression library

We have applied a molecular biology approach to the identification of human monoclonal antibodies. Human peripheral blood lymphocyte mRNA was converted to cDNA and a select subset was amplified by the polymerase chain reaction. These products, containing coding sequences for numerous immunoglobulin heavy- and kappa light-chain variable and constant region domains, were inserted into modified bacteriophage lambda expression vectors and introduced into Escherichia coli by infection to yield a combinatorial immunoexpression library. Clones with binding activity to tetanus toxoid were identified by filter hybridization with radiolabeled antigen and appeared at a frequency of 0.2% in the library. These human antigen binding fragments, consisting of a heavy-chain fragment covalently linked to a light chain, displayed high affinity of binding to tetanus toxoid with equilibrium constants in the nanomolar range but did not cross-react with other proteins tested. We estimate that this human immunoexpression library contains 20,000 clones with high affinity and specificity to our chosen antigen.

Framework residue 71 is a major determinant of the position and conformation of the second hypervariable region in the VH domains of immunoglobulins

Analysis of the immunoglobulins of known structure reveals systematic differences in the position and main-chain conformation of the second hypervariable region of the VH domain (H2). We show that the major determinant of the position of H2 is the size of the residue at site 71, a site that is in the conserved framework of the VH domain. It is likely that for about two thirds of the known VH sequences the size of the residue at this site is also a major determinant of the conformation of H2. This effect can override the predisposition of the sequence, as in the case of the H2 loop of J539, which is an exception to the rules relating sequence and conformation of short hairpin loops. Understanding the relationship between the residue at position 71 and the position and conformation of H2 has applications to the prediction and engineering of antigen-binding sites of immunoglobulins.

Heavy and light chain contributions to antigen binding in an anti-digoxin chain recombinant antibody produced by transfection of cloned anti-digoxin antibody genes

We used immunoglobulin gene transfection to study the effect that substituting an homologous light (L) chain for a parental L chain has on antigen fine specificity and affinity. High-affinity monoclonal anti-digoxin antibodies 26-10 and 40-100 were selected for study because their L chains are 92% homologous (although the H chains differ), and their binding with digoxin and digoxin analogs show very different properties. In order to generate a recombinant transfectoma, the genes encoding the 26-10 H and L chains were cloned. After the sequenced clones had been shown to contain the V gene and the transcriptional control elements, the H and L chain V region genes were subcloned into different expression vectors. Both constructs were transfected into myeloma J558L, a lambda 1 chain producer, to verify that the genetic constructs expressed correctly. The recombined 26-10 antibody was identical to parental 26-10 antibody in fine specificity and affinity. The 26-10 L chain construct was then transfected into a cell line, CR-101, that expresses the 40-100 H chain and a lambda 1 chain. The transfectoma 1E6, secreting 40-100 H chain and 26-10 L chain, was selected. Appropriate gene expression in 1E6 was proven by polymerase chain reaction cloning and sequencing. The fine specificity properties of the 1E6 recombinant derive from both the 40-100 and 26-10 antibodies; however, the affinity of 1E6 is 130 times less than that of the parental antibodies. We conclude that, in 1E6, the H and L chains are codominant in their influence on antigen specificity and that homologous pairing of H and L chains is required for optimal affinity.

Treatment of 150 cases of life-threatening digitalis intoxication with digoxin-specific Fab antibody fragments. Final report of a multicenter study

One hundred fifty patients with potentially life-threatening digitalis toxicity were treated with digoxin-specific antibody fragments (Fab) purified from immunoglobulin G produced in sheep. The dose of Fab fragments was equal to the amount of digoxin or digitoxin in the patient's body as estimated from medical histories or determinations of serum digoxin or digitoxin concentrations. The youngest patient received Fab fragments within several hours of birth, and the oldest patient was 94 years old. Seventy-five patients (50%) were receiving long-term digitalis therapy, 15 (10%) had taken a large overdose of digitalis accidentally, and 59 (39%) had ingested an overdose of digitalis with suicidal intent. The clinical response to Fab was unspecified in two cases, leaving 148 patients who could be evaluated. One hundred nineteen patients (80%) had resolution of all signs and symptoms of digitalis toxicity, 14 (10%) improved, and 15 (10%) showed no response. After termination of the Fab infusion, the median time to initial response was 19 minutes, and 75% of the patients had some evidence of a response by 60 minutes. There were only 14 patients with adverse events considered to possibly or probably have been caused by Fab; the most common events were rapid development of hypokalemia and exacerbation of congestive heart failure. No allergic reactions were identified in response to Fab treatment. Of patients who experienced cardiac arrest as a manifestation of digitalis toxicity, 54% survived hospitalization.(ABSTRACT TRUNCATED AT 250 WORDS)

Use of deuterium labelling in NMR studies of antibody combining site structure

Antibody molecules secreted by B-lymphocytes play a central role in the immune defense systems of higher organisms. The major function of the antibody molecule is to bind specifically to foreign molecules (antigens) and to effect their inactivation and/or removal. Antibody molecules exist in millions of different forms, each with a unique amino acid sequence and combining site structure. Collectively called immunoglobulins (abbreviated as Ig), they form one of the major classes of proteins present in the blood, constituting 20% of the total plasma protein by weight.

Engineering human prolactin to bind to the human growth hormone receptor

A strategy of iterative site-directed mutagenesis and binding analysis was used to incorporate the receptor-binding determinants from human growth hormone (hGH) into the nonbinding homolog, human prolactin (hPRL). The complementary DNA for hPRL was cloned, expressed in Escherichia coli, and mutated to introduce sequentially those substitutions from hGH that were predicted by alanine-scanning mutagenesis and other studies to be most critical for binding to the hGH receptor from human liver. After seven rounds of site-specific mutagenesis, a variant of hPRL was obtained containing eight mutations with an association constant for the hGH receptor that was increased more than 10,000-fold. This hPRL variant binds one-sixth as strongly as wild-type hGH, but shares only 26 percent overall sequence identity with hGH. These studies show the feasibility of recruiting receptor-binding properties from distantly related and functionally divergent hormones and show that a detailed functional database can be used to guide the design of a protein-protein interface in the absence of direct structural information.

Engineering human prolactin to bind to the human growth hormone receptor

A strategy of iterative site-directed mutagenesis and binding analysis was used to incorporate the receptor-binding determinants from human growth hormone (hGH) into the nonbinding homolog, human prolactin (hPRL). The complementary DNA for hPRL was cloned, expressed in Escherichia coli, and mutated to introduce sequentially those substitutions from hGH that were predicted by alanine-scanning mutagenesis and other studies to be most critical for binding to the hGH receptor from human liver. After seven rounds of site-specific mutagenesis, a variant of hPRL was obtained containing eight mutations with an association constant for the hGH receptor that was increased more than 10,000-fold. This hPRL variant binds one-sixth as strongly as wild-type hGH, but shares only 26 percent overall sequence identity with hGH. These studies show the feasibility of recruiting receptor-binding properties from distantly related and functionally divergent hormones and show that a detailed functional database can be used to guide the design of a protein-protein interface in the absence of direct structural information.

Bispecific antibody: a tool for diagnosis and treatment of disease

Antibodies with two distinct binding specificities have great potential for a wide range of clinical applications as targeting agents for in vitro and in vivo immunodiagnosis and therapy, and for improving immunoassays. They have shown great promise for targeting cytotoxic effector cells, delivering radionuclides, toxins or cytotoxic drugs to specific targets, particularly tumour cells. We discuss potential applications of bispecific antibodies, the theoretical basis and problems associated with their production and purification, cell fusion and chemical conjugation techniques, and propose a new manufacturing strategy by genetic engineering. This approach has enormous potential applications for producing tailor-made bispecific antibodies, and will enable widespread clinical uses of these antibodies both for diagnostic purposes and therapy.

The Croonian lecture, 1989. Antibodies: a paradigm for the biology of molecular recognition

The hallmark of the antibody response to antigenic challenge is its remarkable specificity. In his Croonian Lecture in 1905, Ehrlich recognized it as a biological puzzle, but considered it inconceivable that animals could produce substances capable of specific recognition of toxins that the species had never encountered before. It took the largest part of the following 70 years to begin to understand the chemical base of the biological puzzle. Even more recently, the genetic base of the underlying events has been clarified. Unique genetic rearrangements of the DNA initiate the biological diversity of somatic cells; this provides an initial source of antigen recognition. The remarkable specificity is the result of an antigen-driven Darwinian selection of proliferating clones, operating on further diversity that is generated by a high rate of point mutations in specific genes. Although the complexity of the biological events underlying the process remain largely unknown, the knowledge gained so far provides insights into alternative approaches to the production of new antibodies.

A rapid method of cloning functional variable-region antibody genes in Escherichia coli as single-chain immunotoxins

We have devised a strategy based on polymerase chain reaction (PCR) for the rapid cloning of functional antibody genes as single-chain immunotoxins. RNA from a hybridoma producing an antibody (OVB3) that reacts with ovarian cancer cells was used as a template to make the first strand of a cDNA. Then a second strand was synthesized and amplified by using two sets of DNA primers that (i) hybridized to the ends of the light- and heavy-chain variable regions, (ii) encoded a linker peptide, and (iii) contained appropriate restriction enzyme sites for cloning. After 30 cycles of PCR, the DNA fragments containing sequences encoding the light- and heavy-chain variable regions were cloned into an Escherichia coli expression vector containing a portion of the Pseudomonas exotoxin gene. Clones encoding recombinant single-chain immunotoxins were expressed in E. coli and the protein product was assessed for its ability to bind to or kill cells bearing the OVB3 antigen. By using this approach it should be possible to rapidly clone the functional variable region sequences of many different antibodies from hybridoma RNA.

Antibody mediated targeting of radioisotopes, drugs and toxins in diagnosis and treatment

The recent resurgence of interest in site specific delivery of radioisotopes, chemotherapeutic drugs and toxins for the diagnosis and treatment of cancer, and for the selective manipulation of the immune system, can be directly related to the need for improved diagnosis and the fact that for many cancers, for example lung, colon and gastric, the conventional treatments of surgery, radiotherapy and chemotherapy have reached a plateau in terms of the number of patients cured. To date, because of their specificity, the major emphasis has been on the use of antibodies as carriers and extensive in vitro, in vivo preclinical and clinical evaluation is underway. The aim of this article is to review recent progress, highlight avenues being explored to overcome limitations and to indicate new approaches that are evolving in antibody mediated targeting.

Conformations of immunoglobulin hypervariable regions

On the basis of comparative studies of known antibody structures and sequences it has been argued that there is a small repertoire of main-chain conformations for at least five of the six hypervariable regions of antibodies, and that the particular conformation adopted is determined by a few key conserved residues. These hypotheses are now supported by reasonably successful predictions of the structures of most hypervariable regions of various antibodies, as revealed by comparison with their subsequently determined structures.

High-level expression of chimeric antibodies using adapted cDNA variable region cassettes

A rapid and generally applicable method for the modification of immunoglobulin cDNAs was developed so that the variable (V) regions could be expressed as cassettes, together with a variety of constant regions. Murine cDNAs were isolated, sequenced and the V regions joined to short oligonucleotides providing both splice donor sites and unique restriction sites for insertion into an expression vector. Using this strategy we have expressed the V regions of several murine antibodies, together with the human gamma 1 constant region. Although most of these chimeric antibodies were readily expressed, one murine light-chain cDNA sequence could not be expressed in transfected hybridoma cells. Reconstruction experiments indicate that the sequence created by the fusion of the murine leader and variable region blocked expression at the level of RNA accumulation. The methods described, as well as the potential problems of expression, are applicable to both traditional cDNA fragments and those obtained by in vitro amplification techniques.

The immunogenicity of chimeric antibodies

Mice were immunized with model xenogeneic (both the VH frameworks and the CH domains of human origin), chimeric (just VH frameworks human), or self antibodies, and the antiantibody responses were dissected. Only the self antibody did not elicit a response. A strong response was elicited by the most xenogeneic antibody with approximately 90% against the C and approximately 10% against the V. The anti-V response was not attenuated in the chimeric antibody, demonstrating that foreign VH frameworks can be sufficient to lead to a strong antiantibody response. The magnitude of this xenogeneic anti-VH response was similar to that of the allotypic response elicited by immunizing mice of the Igha allotype with an Ighb antibody. Thus, although chimerization can diminish antiantibody responses, attention should be paid both to V region immunogenicity and to polymorphism.

A humanized antibody that binds to the interleukin 2 receptor

The anti-Tac monoclonal antibody is known to bind to the p55 chain of the human interleukin 2 receptor and to inhibit proliferation of T cells by blocking interleukin 2 binding. However, use of anti-Tac as an immunosuppressant drug would be impaired by the human immune response against this murine antibody. We have therefore constructed a "humanized" antibody by combining the complementarity-determining regions (CDRs) of the anti-Tac antibody with human framework and constant regions. The human framework regions were chosen to maximize homology with the anti-Tac antibody sequence. In addition, a computer model of murine anti-Tac was used to identify several amino acids which, while outside the CDRs, are likely to interact with the CDRs or antigen. These mouse amino acids were also retained in the humanized antibody. The humanized anti-Tac antibody has an affinity for p55 of 3 x 10(9) M-1, about 1/3 that of murine anti-Tac.

Homologous recombination in hybridoma cells: heavy chain chimeric antibody produced by gene targeting

We demonstrate that murine myeloma cells can efficiently mediate homologous recombination. The murine myeloma cell line J558L was shown to appropriately recombine two transfected DNA molecules in approximately 30% of cells that received and integrated intact copies of both molecules. This activity was then exploited to direct major reconstructions of an endogenous locus within a hybridoma cell line. Production of antigen-specific chimeric heavy chain was achieved by targeting the human IgG1 heavy chain constant region (C gamma 1) exons to the genomic heavy chain locus of a hybridoma cell line secreting antibody specific for a human tumor-associated antigen. The frequency of productive genomic recombinations was approximately 1 in 200 transfectants, with accumulation of the chimeric protein reaching greater than 20 micrograms/ml in culture supernatants.

Strategy for the production of human monoclonal antibodies using in vitro activated B cells

An up-to-date strategy to optimally produce human monoclonal antibodies by primary in vitro immunization or by secondary in vivo stimulation of immunized donors is discussed in detail. The effect of a lysosomotropic amino acid dipeptide on the B cell suppression by lysosome-rich cytolytic cells and the subsequent Epstein-Barr virus transformation of immune B lymphocytes is explained. The described strategy allows a routine production of human hybridomas, derived from peripheral blood lymphocytes and exhibiting a productivity in the range of 20-50 micrograms Ig/24 h per 10(6) cells. Furthermore, the possibilities to modulate antibody isotype and affinity by molecular biological methods is reviewed.

Binding activities of a repertoire of single immunoglobulin variable domains secreted from Escherichia coli

In antibodies, a heavy and a light chain variable domain, VH and VL, respectively, pack together and the hypervariable loops on each domain contribute to binding antigen. We find, however, that isolated VH domains with good antigen-binding affinities can also be prepared. Using the polymerase chain reaction, diverse libraries of VH genes were cloned from the spleen genomic DNA of mice immunized with either lysozyme or keyhole-limpet haemocyanin. From these libraries, VH domains were expressed and secreted from Escherichia coli. Binding activities were detected against both antigens, and two VH domains were characterized with affinities for lysozyme in the 20 nM range. Isolated variable domains may offer an alternative to monoclonal antibodies and serve as the key to building high-affinity human antibodies. We suggest the name 'single domain antibodies (dAbs)' for these antigen binding demands.

C1q binding to chimeric monoclonal IgG3 antibodies consisting of mouse variable regions and human constant regions with shortened hinge containing 15 to 47 amino acids

We have constructed four different deletion mutants of a chimeric mouse-human IgG3 anti-(4-hydroxy-3-nitrophenyl)acetyl/(5-iodo-4 hydroxy-3 nitrophenyl) acetyl (NP/NIP) antibody lacking one or more of the four exons coding for the hinge region. The mutant variants all retained intact hinge region epitopes since they all reacted with IgG3 hinge-specific antibodies. Surprisingly, all the deletion mutants bound C1q equally well or even better than the wild type. Thus the high C1q binding activity of IgG3 compared to IgG1 is apparently not due to the total length of the IgG3 hinge, which is 62 amino acids, nor is it due to the length of the upper hinge which is the stretch from the end of CH1 to the first inter-heavy chain disulfide bond.

A repertoire of monoclonal antibodies with human heavy chains from transgenic mice

The introduction of human immunoglobulin gene segments in their unrearranged configuration into the germ line of mice might allow the production of a repertoire of human antibodies. Such transgenic mice could be used for the production of human monoclonal antibodies against human antigens. To test the feasibility of this approach, mice were created that carry a human heavy-chain minilocus comprising unrearranged immunoglobulin variable, diversity, and joining elements linked to a human mu-chain gene. The gene segments of this minilocus are rearranged in a large proportion of cells in thymus and spleen but not in nonlymphoid tissue. Some 4% of the B lymphocytes synthesize human mu chains resulting in a serum titer of about 50 micrograms of transgenic IgM antibody per ml. Hybridomas were established from the transgenic mice that stably secreted several micrograms of antibodies containing human mu heavy chains per milliliter.

Antibody engineering

The antibody molecule is a therapeutic agent, designed by nature to bind to a wide range of antigen molecules and to trigger effector functions, such as complement lysis and cell-mediated killing. The genes encoding antibodies can be manipulated in vitro, allowing the binding sites for antigen and effector molecules to be dissected, and new properties to be engineered. The future for the application of engineered antibodies in medicine is reviewed in the context of the past century.

Characterization of a chimeric aequorin molecule expressed in myeloma cells

We have constructed a chimeric aequorin consisting of a fragment of the anti-NP immunoglobulin gene fused to the aequorin gene. Expression in a myeloma cell line has produced a Fab'-like molecule that has the ability to bind NIP specifically and generate bioluminescent activity. It takes approximately 8 h at 4 degrees C in the presence of 2-mercaptoethanol and coelenterazine to regenerate luminescent activity. While the flash kinetics of this recombinant molecule are similar to native aequorin, its quantum efficiency is ten times lower. Preliminary studies have been conducted to ascertain its usefulness for immunoassays. We have shown for this chimeric aequorin 7 x 10(-19) moles can be detected in solution, also it can be used in a solid-phase assay and is stably stored at -70 degrees C for at least 2 months.

Cloning immunoglobulin variable domains for expression by the polymerase chain reaction

We have designed a set of oligonucleotide primers to amplify the cDNA of mouse immunoglobulin heavy and light chain variable domains by the polymerase chain reaction. The primers incorporate restriction sites that allow the cDNA of the variable domains to be force-cloned for sequencing and expression. Here we have applied the technique to clone and sequence the variable domains of five hybridoma antibodies and to express a mouse-human chimeric antibody that binds to the human mammary carcinoma line MCF-7. The technique should also lead to the cloning of antigen-binding specificities directly from immunoglobulin genes.

Immunotoxins up to the present day

Immunotoxins consist of monoclonal or polyclonal antibodies conjugated to bacterial or plant toxins. The toxins used are typically of the A-B type in which a toxic A chain is coupled to a B chain responsible for cell binding and facilitation of A chain entry into the cytosol. Two broad strategies have been followed: coupling intact toxins, or A chains alone, to antibodies. This review examines current progress in in vitro and in vivo research, including recent clinical studies, concentrating principally on ricin or ricin A chain conjugates. The future role of conjugates using membrane-acting toxins, immunolysins, is also discussed.

Cloning and characterization of cDNAs coding for the heavy and light chains of a monoclonal antibody specific for Pseudomonas aeruginosa outer membrane protein I

A set of seven monoclonal antibodies (MAb) directed against outer membrane proteins of Pseudomonas aeruginosa has been examined by Western blot analysis, indirect immunofluorescence tests and subclass typing. The hybridoma cell line secreting MAb 6A4, which reacts with outer membrane protein I, belongs to the IgG2a subclass and crossreacts with the 17 P. aeruginosa serotypes as listed in the International Antigenic Typing System, was selected as source for the preparation of poly(A)+RNA which in turn was used as template for cDNA synthesis and cloning. Full length cDNA clones of the gamma heavy chain as well as the kappa light chain were obtained and characterized by nucleotide sequence analysis. The complete cDNA sequences coding for the heavy and light chains will be the prerequisite for the construction and heterologous expression of a chimeric human-mouse monoclonal antibody which might be used in therapy of P. aeruginosa infections.

Single-chain antigen-binding proteins

Single-chain antigen-binding proteins are novel recombinant polypeptides, composed of an antibody variable light-chain amino acid sequence (VL) tethered to a variable heavy-chain sequence (VH) by a designed peptide that links the carboxyl terminus of the VL sequence to the amino terminus of the VH sequence. These proteins have the same specificities and affinities for their antigens as the monoclonal antibodies whose VL and VH sequences were used to construct the recombinant genes that were expressed in Escherichia coli. Three of these proteins, one derived from the sequence for a monoclonal antibody to growth hormone and two derived from the sequences of two different monoclonal antibodies to fluorescein, were designed, constructed, synthesized, purified, and assayed. These proteins are expected to have significant advantages over monoclonal antibodies in a number of applications.

Structure of antibody hypervariable loops reproduced by a conformational search algorithm

The antigen-combining site of antibody molecules consists of six separate loops supported by a conserved beta-sheet framework; antibody specificity arises from length and sequence variation of these 'hypervariable' loops and can be manipulated by transferring sets of loops between different frameworks. Irregular loops are the most difficult parts of protein structure to understand and to model correctly. Here, we describe two computer experiments where all the hypervariable loops were deleted from X-ray structures of mouse immunoglobulins and reconstructed using the conformational search program CONGEN. A protocol was developed for reconstruction of the hypervariable loops in McPC 603 antibody. Calculated loop conformations were generated and a model of the combining site was built from selected low-energy conformations. We then modelled hypervariable loops in another antibody molecule, HyHEL-5. Both models agreed well with the known crystal structures. Our results hold out promise for the success of future modelling studies of complete antigen-combining sites from amino acid sequences.

Expression of an antibody Fv fragment in myeloma cells

The antigen binding site on antibodies is fashioned by loops at the tips of the beta-sheet framework of both heavy and light chain variable domains. A heterodimer of both variable domains (Fv fragment), incorporating loops from an anti-lysozyme antibody, was expressed and secreted from myeloma cells in good yield (8 mg/l in supernatant from roller bottles), and shown to bind lysozyme. The two subunits were found to be in dynamic equilibrium but are overwhelmingly associated at neutral pH. The small size of Fv fragments (25 x 10(3) Mr) make them attractive for structural studies, in vivo imaging, and therapy.

New strategies in the development of thrombolytic agents

Recombinant DNA technology has allowed large-scale production of the physiological, fibrin-specific, plasminogen activators tissue-type plasminogen activator (t-PA) and single-chain urokinase-type plasminogen activator (scu-PA). The results of clinical trials with these agents, mainly for the treatment of acute myocardial infarction, have revealed a limited fibrin specificity at the large therapeutic doses required for efficient thrombolysis. Mutants and variants of t-PA and scu-PA have given important information on structure-function relationships in these proteins and have resulted in rt-PA variants with significantly prolonged half-lives in vivo. Construction of chimaeric plasminogen activators containing various portions of t-PA and scu-PA has produced functionally active enzymes, however with a lower fibrin-affinity than wild-type t-PA. The promise of antibody targeting and the use of synergistic combinations of thrombolytic agents remains to be further investigated. We anticipate that eventually these research lines will yield artificial plasminogen activators with improved efficacy, risk/benefit and cost/benefit ratios.

Protein engineering of antibody binding sites: recovery of specific activity in an anti-digoxin single-chain Fv analogue produced in Escherichia coli

A biosynthetic antibody binding site, which incorporated the variable domains of anti-digoxin monoclonal antibody 26-10 in a single polypeptide chain (Mr = 26,354), was produced in Escherichia coli by protein engineering. This variable region fragment (Fv) analogue comprised the 26-10 heavy- and light-chain variable regions (VH and VL) connected by a 15-amino acid linker to form a single-chain Fv (sFv). The sFv was designed as a prolyl-VH-(linker)-VL sequence of 248 amino acids. A 744-base-pair DNA sequence corresponding to this sFv protein was derived by using an E. coli codon preference, and the sFv gene was assembled starting from synthetic oligonucleotides. The sFv polypeptide was expressed as a fusion protein in E. coli, using a leader derived from the trp LE sequence. The sFv protein was obtained by acid cleavage of the unique Asp-Pro peptide bond engineered at the junction of leader and sFv in the fusion protein [(leader)-Asp-Pro-VH-(linker)-VL]. After isolation and renaturation, folded sFv displayed specificity for digoxin and related cardiac glycosides similar to that of natural 26-10 Fab fragments. Binding between affinity-purified sFv and digoxin exhibited an association constant [Ka = (3.2 +/- 0.9) x 10(7) M-1] that was about a factor of 6 smaller than that found for 26-10 Fab fragments [Ka = (1.9 +/- 0.2) x 10(8) M-1] under the same buffer conditions, consisting of 0.01 M sodium acetate, pH 5.5/0.25 M urea.