Fully synthetic human combinatorial antibody libraries (HuCAL) based on modular consensus frameworks and CDRs randomized with trinucleotides

By analyzing the human antibody repertoire in terms of structure, amino acid sequence diversity and germline usage, we found that seven V(H) and seven V(L) (four Vkappa and three Vlambda) germline families cover more than 95 % of the human antibody diversity used. A consensus sequence was derived for each family and optimized for expression in Escherichia coli. In order to make all six complementarity determining regions (CDRs) accessible for diversification, the synthetic genes were designed to be modular and mutually compatible by introducing unique restriction endonuclease sites flanking the CDRs. Molecular modeling verified that all canonical classes were present. We could show that all master genes are expressed as soluble proteins in the periplasm of E. coli. A first set of antibody phage display libraries totalling 2x10(9) members was created after cloning the genes in all 49 combinations into a phagemid vector, itself devoid of the restriction sites in question. Diversity was created by replacing the V(H) and V(L) CDR3 regions of the master genes by CDR3 library cassettes, generated from mixed trinucleotides and biased towards natural human antibody CDR3 sequences. The sequencing of 257 members of the unselected libraries indicated that the frequency of correct and thus potentially functional sequences was 61 %. Selection experiments against many antigens yielded a diverse set of binders with high affinities. Due to the modular design of all master genes, either single binders or even pools of binders can now be rapidly optimized without knowledge of the particular sequence, using pre-built CDR cassette libraries. The small number of 49 master genes will allow future improvements to be incorporated quickly, and the separation of the frameworks may help in analyzing why nature has evolved these distinct subfamilies of antibody germline genes.     

SDR grafting--a new approach to antibody humanization

A major impediment to the clinical utility of the murine monoclonal antibodies is their potential to elicit human anti-murine antibody (HAMA) response in patients. To circumvent this problem, murine antibodies have been genetically manipulated to progressively replace their murine content with the amino acid residues present in their human counterparts. To that end, murine antibodies have been humanized by grafting their complementarity determining regions (CDRs) onto the variable light (V(L)) and variable heavy (V(H)) frameworks of human immunoglobulin molecules, while retaining those murine framework residues deemed essential for the integrity of the antigen-combining site. However, the xenogeneic CDRs of the humanized antibodies may evoke anti-idiotypic (anti-Id) response in patients. To minimize the anti-Id response, a procedure to humanize xenogeneic antibodies has been described that is based on grafting, onto the human frameworks, only the specificity determining residues (SDRs), the CDR residues that are most crucial in the antibody-ligand interaction. The SDRs are identified through the help of the database of the three-dimensional structures of the antigen-antibody complexes of known structures or by mutational analysis of the antibody-combining site. An alternative approach to humanization, which involves retention of more CDR residues, is based on grafting of the 'abbreviated' CDRs, the stretches of CDR residues that include all the SDRs. A procedure to assess the reactivity of the humanized antibody to sera from patients who had been administered the murine antibody has also been described.     

Phage antibody fragments library combining a single human light chain variable region with immune mouse heavy chain variable regions

We describe the construction of a phage antibody fragments library which combines, in a single cloning step, a synthetic human light chain variable region (V(L)) with a diverse set of heavy chain variable regions, from a mouse immunized with the prostate specific antigen (PSA). Despite V(L) restriction, selection from this library rendered two different single chain Fv antibody fragments, specifically recognizing PSA. The human V(L), used as a general partner for mouse heavy chains, was constructed by linking the germline A27 gene and the J(K)1 minigene segment, both of which are prominently involved in human antibody responses. Our approach offers a fast and simple way to produce half-human molecules, while keeping the advantage of immunizing animals for high affinity antibodies.     

Guided selection of human antibody light chains against TAG-72 using a phage display chain shuffling approach

To enhance therapeutic potential of murine monoclonal antibody, humanization by CDR grafting is usually used to reduce immunogenic mouse residues. Most humanized antibodies still have mouse residues critical for antigen binding, but the mouse residues may evoke immune responses in humans. Previously, we constructed a new humanized version (AKA) of mouse CC49 antibody specific for tumor-associated glycoprotein, TAG-72. In this study, to select a completely human antibody light chain against TAG-72, guided selection strategy using phage display was used. The heavy chain variable region (VH) of AKA was used to guide the selection of a human TAG-72-specific light chain variable region (VL) from a human VL repertoire constructed from human PBL. Most of the selected VLs were identified to be originated from the members of the human germline VK1 family, whereas the VL of AKA is more homologous to the VK4 family. Competition binding assay of the selected Fabs with mouse CC49 suggested that the epitopes of the Fabs overlap with that of CC49. In addition, they showed better antigen-binding affinity compared to parental AKA. The selected human VLs may be used to guide the selection of human VHs to get completely human anti-TAG72 antibody.     

A comparative analysis of the immunological evolution of antibody 28B4

In an effort to gain greater insight into the evolution of the redox active, catalytic antibody 28B4, the germline genes used by the mouse to generate this antibody were cloned and expressed, and the X-ray crystal structures of the unliganded and hapten-bound germline Fab of antibody 28B4 were determined. Comparison with the previously determined structures of the unliganded and hapten-bound affinity-matured Fab [Hsieh-Wilson, L. C., Schultz, P. G., and Stevens, R. C. (1996) Proc. Natl. Acad. Sci. U.S.A. 93, 5363] shows that the germline antibody binds the p-nitrophenyl ring of hapten 3 in an orientation significantly different from that seen in the affinity-matured antibody, whereas the phosphonate moiety is bound in a similar mode by both antibodies. The affinity-matured antibody 28B4 has more electrostatic and hydrophobic interactions with hapten 3 than the germline antibody and binds the hapten in a lock-and-key fashion. In contrast, significant conformational changes occur in the loops of CDR H3 and CDR L1 upon hapten binding to the germline antibody, consistent with the notion of structural plasticity in the germline antibody-combining site [Wedemayer, G. J., Patten, P. A., Wang, L. H., Schultz, P. G., and Stevens, R. C. (1997) Science 276, 1665]. The structural differences are reflected in the differential binding affinities of the germline Fab (K(d) = 25 microM) and 28B4 Fab (K(d) = 37 nM) to hapten 3. Nine replacement mutations were found to accumulate in the affinity-matured antibody 28B4 compared to its germline precursor. The effects of each mutation on the binding affinity of the antibody to hapten 3 were characterized in detail in the contexts of both the germline and the affinity-matured antibodies. One of the mutations, Asp95(H)Trp, leads to a change in the orientation of the bound hapten, and its presence is a prerequisite for other somatic mutations to enhance the binding affinity of the germline antibody for hapten 3. Thus, the germline antibody of 28B4 acquired functionally important mutations in a stepwise manner, which fits into a multicycle mutation, affinity selection, and clonal expansion model for germline antibody evolution. Two other antibodies, 20-1 and NZA6, with very different antigen specificities were found to be highly homologous to the germline antibody of 28B4, consistent with the notion that certain germline variable-region gene combinations can give rise to polyspecific hapten binding sites [Romesberg, F. E., Spiller, B., Schultz, P. G., and Stevens, R. C. (1998) Science 279, 1929]. The ultimate specificity of the polyspecific germline antibody appears to be defined by CDR H3 variability and subsequent somatic mutation. Insights into the evolution of antibody-combining sites provided by this and other structural studies are discussed.     

Dissection of the combining site in a humanized anti-Tac antibody.

The genetically engineered "humanized" anti-Tac antibody (HAT) has been shown to bind the p55 chain of the human IL-2R with an affinity close to that of the murine anti-Tac. Although the HAT molecule contains all six mouse CDR, it was not known which, and to what extent, each of the CDR contributes to Ag binding. These questions were addressed by constructing a series of variant HAT antibodies, each substituting a single HAT CDR with a heterologous CDR. The association constants of the variant HAT antibodies to p55 were determined by competitive binding analysis. We find that CDR 1 and 3 of the H chain and CDR 3 of the L chain are essential for maintaining binding. The remaining three CDR appear to be involved to a lesser degree.     

A human and a mouse anti-idiotypic antibody specific for human T14(+) anti-DNA antibodies reconstructed by phage display

Little is known about human anti-idiotypic antibodies. Phage display methodology was used to reconstruct these antibodies from lupus patients, which recognize a subset (T14(+)) of anti-DNA antibodies. Antigen-specific B cells were isolated from the blood using a peptide based on a complementarity determining region (V(H)CDR3) of the prototypic T14(+) antibody. cDNA fragments of the V(H) and V(L) genes prepared from the cells were expressed as phage displayed single chain Fv (scFv) fragments using the pCANTAB-5E phagemid vector. From a reactive clone obtained, the Ig genes used were identified to be V(H)3, D5-D3, J(H)4b, V(kappa)I and J(kappa)2. The heavy chain was highly mutated, especially in CDR3, which bears mutations mostly of the replacement type; this region is also unusual in being extremely long due to a D-D fusion. In contrast, a mouse hybridoma antibody, made to the same T14(+) peptide and transformed as a scFv fragment, uses a short V(H)CDR3 comprising five amino acids, three of which are tyrosines. Tyrosines may be important for antigen binding because two of these also exist in the human V(H)CDR3. The light chains of both antibodies may also contribute to the specificity of the protein, because their V(L) segments, including the CDRs, are highly homologous to each other.     

Optimal design features of camelized human single-domain antibody libraries

We have constructed a human V(H) library based on a camelized V(H) sequence. The library was constructed with complete randomization of 19 of the 23 CDR3 residues and was panned against two monoclonal antibody targets to generate V(H) sequences for determination of the antigen contact residue positions. Furthermore, the feasibility and desirability of introducing a disulfide bridge between CDR1 and CDR3 was investigated. Sequences derived from the library showed a bias toward the use of C-terminal CDR3 residues as antigen contact residues. Mass spectrometric analyses indicated that CDR1-CDR3 disulfide formation was universal. However, surface plasmon resonance and NMR data showed that the CDR3 constraint imposed by the disulfide bridge was not always desirable. Very high yields of soluble protein products and lack of protein aggregation, as demonstrated by the quality of the (1)H-(15)N HSQC spectra, indicated that the V(H) sequence for library construction was a good choice. These results should be useful in the design of V(H) libraries with optimal features.     

Construction and characterization of single-chain variable fragment antibody library derived from germline rearranged immunoglobulin variable genes

Antibody repertoires for library construction are conventionally harvested from mRNAs of immune cells. To examine whether germline rearranged immunoglobulin (Ig) variable region genes could be used as source of antibody repertoire, an immunized phage-displayed scFv library was prepared using splenocytic genomic DNA as template. In addition, a novel frame-shifting PCR (fsPCR) step was introduced to rescue stop codon and to enhance diversity of the complementarity-determining region 3 (CDR3). The germline scFv library was initially characterized against the hapten antigen phenyloxazolone (phOx). Sequence analysis of the phOx-selective scFvs indicated that the CDRs consisted of novel as well as conserved motifs. In order to illustrate that the diversity of CDR3 was increased by the fsPCR step, a second scFv library was constructed using a single scFv clone L3G7C as a template. Despite showing similar binding characteristics towards phOx, the scFv clones that were obtained from the L3G7C-derived antibody library gave a lower non-specific binding than that of the parental L3G7C clone. To determine whether germline library represented the endogenous immune status, specific scFv clones for nucleocapsid (N) protein of SARS-associated coronavirus (SCoV) were obtained both from naïve and immunized germline scFv libraries. Both libraries yielded specific anti-N scFvs that exhibited similar binding characteristics towards recombinant N protein, except the immunized library gave a larger number of specific anti-N scFv, and clones with identical nucleotide sequences were found. In conclusion, highly diversified antibody library can be efficiently constructed using germline rearranged immunoglobulin variable genes as source of antibody repertoires and fsPCR to diversify the CDR3.     

A highly conserved interspecies V(H) in the human genome

Idiotype conservation between human and mouse antibodies has been observed in association with various infectious and autoimmune diseases. We have isolated a human anti-idiotypic antibody to a mouse monoclonal anti-IgE antibody (BSW17) suggesting a conserved interspecies idiotype associated with an anti-IgE response. To find the homologue of BSW17 in the human genome we applied the guided selection strategy. Combining V(H) of BSW17 with a human V(L) repertoire resulted in three light chains. The three V(L) chains were then combined with a human V(H) repertoire resulting in three clones specific for human IgE. Surprisingly, one clone, Hu41, had the same epitope specificity and functional in vitro activity as BSW17 and V(H) complementarity-determining regions identical with that of BSW17. Real-time PCR analysis confirmed the presence of the Hu41 V(H) sequence in the human genome. These data document the first example of the isolation of a human antibody where high sequence similarity to the original murine V(H) sequence is associated with common antigen and epitope specificity.     

In vitro characterisation of a monovalent and bivalent form of a fully human anti Ep-CAM phage antibody

 Antibodies to tumour-associated antigens are increasingly being used as targeting vehicles for the visualisation and for therapy of human solid tumours. The epithelial cell adhesion molecule (Ep-CAM) is an antigen that is overexpressed on a variety of human solid tumours and constitutes an attractive target for immunotargeting. We set out to obtain fully human antibodies to this antigen by selecting from a large antibody repertoire displayed on bacteriophages. Two single-chain variable antibody fragments (scFv) were identified that specifically bound recombinant antigen in vitro. One of the selected antibodies (VEL-1) cross-reacted with extracellular matrix components in immunohistochemistry of colon carcinoma, whereas the other scFv (VEL-2) specifically recognised colon cancer cells. The latter antibody was further characterised with respect to epitope specificity and kinetics of antigen-binding. It showed no competition with the well-characterised anti Ep-CAM MOC-31 monoclonal antibody and had an off-rate of 5 × 10⁻² s⁻¹. To obtain an antibody format more suitable for in vivo tumour targeting and to increase the apparent affinity through avidity, the genes of scFv VEL-2 were re-formatted by fusion to a human (γ1) hinge region and CH3 domain. This “minibody” was expressed in Escherichia coli, specifically bound the Ep-CAM antigen and showed a 20-fold reduced off-rate in surface plasmon resonance analysis. These results show that phage antibody selection, combined with antibody engineering, may result in fully human antibody molecules with promising characteristics for in vivo use in tumour targeting. Received: 13 July 2000 / Accepted: 12 October 2000     

Function blocking antibodies to neuropilin-1 generated from a designed human synthetic antibody phage library

Non-immune (na?ve) antibody phage libraries have become an important source of human antibodies. The synthetic phage antibody library described here utilizes a single human framework with a template containing human consensus complementarity-determining regions (CDRs). Diversity of the libraries was introduced at select CDR positions using tailored degenerate and trinucleotide codons that mimic natural human antibodies. Neuropilin-1 (NRP1), a cell-surface receptor for both vascular endothelial growth factor (VEGF) and class 3 semaphorins, is expressed on endothelial cells and neurons. NRP1 is required for vascular development and is expressed widely in the developing vasculature. To investigate the possibility of function blocking antibodies to NRP1 as potential therapeutics, and study the consequence of targeting NRP1 in murine tumor models, panels of antibodies that cross-react with human and murine NRP1 were generated from a designed antibody phage library. Antibody (YW64.3) binds to the CUB domains (a1a2) of NRP1 and completely blocks Sema3A induced neuron collapse; antibody (YW107.4.87) binds to the coagulation factor V/VIII domains (b1b2) of NRP1 and blocks VEGF binding and VEGF induced cell migration. YW107.4.87 inhibits tumor growth in animal xenograft models. These antibodies have provided valuable tools to study the roles of NRP1 in vascular and tumor biology.     

In vitro antibody evolution targeting germline hot spots to increase activity of an anti-CD22 immunotoxin

Recombinant immunotoxin BL22, containing the Fv portion of an anti-CD22 antibody, produced complete remissions in most patients with drug-resistant hairy cell leukemia but had less activity in leukemias with low CD22 expression. Complementarity-determining region (CDR) mutagenesis is used to increase antibody affinity but can be difficult to perform successfully. We previously showed that antibodies with increased affinity and immunotoxins with increased activity could be obtained by directing mutations at specific DNA residues called hot spots. Because hot spots can arise either by somatic mutation or be present in the germline, we examined which type of hot spot is preferred for increasing antibody affinity. Initially, a second generation antibody phage-display library targeting a germline hot spot (Ser(30)-Asn(31)) within CDR1 of the antibody light chain was mutated. Substitution of serine 30 or asparagine 31 with arginine produced mutant immunotoxins with an affinity (0.8 nM) increased 7-fold over BL22 (5.8 nM) and 3-fold over the first generation mutant HA22 (2.3 nM). More importantly, a 10-fold increase in activity over BL22 and a 2-3-fold increase over HA22 were observed in various B lymphoma cell lines including WSU-CLL that contains only 5500 CD22 sites per cell. For comparison, two phage-display libraries targeting non-germline hot spots in heavy chain CDR1 and CDR3 were generated but did not produce Fv with increased affinity. Our results demonstrate that germline hot spots but not non-germline hot spots are effective for in vitro antibody affinity maturation.     

Molecular cloning and chimerisation of an inhibitory anti-cathepsin B antibody and its expression in Chinese hamster ovary cells

The lysosomal cysteine protease cathepsin B is one of several proteases that have been linked to tumour progression. Its increased expression and secretion in tumour cells may facilitate the degradation of extracellular matrix proteins, leading to tumour cell invasion and metastasis. Specific inhibitory monoclonal antibodies are a possible alternative to synthetic inhibitors as a therapeutic tool for cancer treatment. An inhibitory monoclonal antibody, which binds to an epitope near the active site of cathepsin B and inhibits its proteolytic activity, was prepared and its effect on invasion of ras-transformed MCF-10A neoT cells was tested in vitro. Here we present the nucleotide sequences of the heavy and light chains of the inhibitory antibody and compare them to the murine immunoglobulin germline sequences for possible somatic hypermutations. Since no harmful mutations were found, a mouse/human chimeric antibody was constructed by fusing murine V(H) and V(L) variable regions of the inhibitory antibody with human gamma 1 and K constant regions, respectively. Chinese hamster ovary K1 cells were co-transfected with expression vectors pcD-NA3L and pcDNA3H and the reactivity of the isolated chimeric antibody was tested by ELISA and Western blotting. We could demonstrate an inhibitory effect of the chimeric antibody.     

Structural insights into parallel strategies for germline antibody recognition of lipopolysaccharide from Chlamydia

The structure of the antigen-binding fragment from the monoclonal antibody S64-4 in complex with a pentasaccharide bisphosphate fragment from chlamydial lipopolysaccharide has been determined by x-ray diffraction to 2.6 ? resolution. Like the well-characterized antibody S25-2, S64-4 displays a pocket formed by the residues of germline sequence corresponding to the heavy and light chain V gene segments that binds the terminal Kdo residue of the antigen; however, although S64-4 shares the same heavy chain V gene segment as S25-2, it has a different light chain V gene segment. The new light chain V gene segment codes for a combining site that displays greater affinity, different specificity, and allows a novel antigen conformation that brings a greater number of antigen residues into the combining site than possible in S25-2. Further, while antibodies in the S25-2 family use complementarity determining region (CDR) H3 to discriminate among antigens, S64-4 achieves its specificity via the new light chain V gene segment and resulting change in antigen conformation. These structures reveal an intriguing parallel strategy where two different combinations of germline-coded V gene segments can act as starting points for the generation of germline antibodies against chlamydial antigens and show how anti-carbohydrate antibodies can exploit the conformational flexibility of this class of antigens to achieve high affinity and specificity independently of CDR H3.     

Immunologic memory to phosphocholine keyhole limpet hemocyanin. Recurrent mutations in the lambda 1 light chain increase affinity for antigen.

Anti-phosphocholine (PC)-keyhole limpet hemacyanin hybridomas representative of a memory response that express the lambda 1 L chain isotype have a high reactivity to PC-protein. A common feature of these hybridomas possessing high affinity for PC-protein is the occurrence of somatic mutations resulting in replacement changes in three CDR2 positions of the lambda 1 L chain. The influence of each of these three positions on the Ag binding properties of these antibodies was examined by site-specific mutagenesis and expression of recombinant antibody molecules by transfected cells. Affinity measurements and fine specificity profile determinations demonstrated the importance of the three lambda 1 CDR2 positions in Ag binding. Compared to antibodies expressing germline lambda 1, including one with an additional junctional serine that is not encoded by V or J, those antibodies possessing critical changes in CDR2 would have a strong selective advantage based on affinity differences for Ag. Sequence analysis of a group of clonally related hybridomas expressing mutated lambda 1 genes allowed construction of a hypothetical genealogic tree that suggests selection based on changes in CDR2 of lambda 1 in the absence of H chain mutations. The results are consistent with stepwise acquisition of mutations and selection based on affinity constraints.     

Two induced anti-Z-DNA monoclonal antibodies use VH gene segments related to those of anti-DNA autoantibodies

The cDNA for H and L chain V regions of two anti-Z-DNA mAb, Z22 and Z44, were cloned and sequenced. These are the first experimentally induced anti-nucleic acid antibody sequences available for comparison with autoantibody sequences. Z22 and Z44 are IgG2b and IgG2a antibodies from C57BL/6 mice. They recognize different facets of the Z-DNA structure. They both use VH10 family genes and share 95% sequence base sequence identity in the VH and leader sequences; however, they differ in the 5'-untranslated region of the VH mRNA, indicating they arise from different germline genes. Both use JH4 segments. They differ from each other very extensively in the CDR3 of both H and L chains. The most closely related H chains in the current GenBank/EMBL data base are two mouse IgG anti-DNA autoantibodies, one from an MRL-lpr/lpr mouse (MRL-DNA4) and one from an NZB/NZW mouse (BV04-01). Z22 and Z44 share 95% sequence identity with these antibodies in the VH segment. In addition, Z22 is identical to MRL-DNA4 at 91% of the positions in the 5'-untranslated region of the H chain mRNA. The two antibodies share 95% base sequence identity in the V kappa segment. The most closely related L chains, with 97 to 98% sequence identity, are the V kappa 10b germline gene for Z22 and the V kappa 10a germ line gene, which is associated with A/J anti-arsonate antibodies and BALB/c anti-ABO blood group substance antibodies, for Z44. Z22 and Z44 share several structural features (similarities in VH, JH, and V kappa) but differ very markedly in the L chain CDR1 and both H and L chain CDR3 sequences; these regions may determine the differences in their specific interactions with Z-DNA.     

Structural insights into humanization of anti-tissue factor antibody 10H10

Murine antibody 10H10 raised against human tissue factor is unique in that it blocks the signaling pathway, and thus inhibits angiogenesis and tumor growth without interfering with coagulation. As a potential therapeutic, the antibody was humanized in a two-step procedure. Antigen-binding loops were grafted onto selected human frameworks and the resulting chimeric antibody was subjected to affinity maturation by using phage display libraries. The results of humanization were analyzed from the structural perspective through comparison of the structure of a humanized variant with the parental mouse antibody. This analysis revealed several hot spots in the framework region that appear to affect antigen binding, and therefore should be considered in human germline selection. In addition, some positions in the Vernier zone, e.g., residue 71 in the heavy chain, that are traditionally thought to be crucial appear to tolerate amino acid substitutions without any effect on binding. Several humanized variants were produced using both short and long forms of complementarity-determining region (CDR) H2 following the difference in the Kabat and Martin definitions. Comparison of such pairs indicated consistently higher thermostability of the variants with short CDR H2. Analysis of the binding data in relation to the structures singled out the ImMunoGeneTics information system® germline IGHV1-2*01 as dubious owing to two potentially destabilizing mutations as compared to the other alleles of the same germline and to other human germlines.     

Design, construction, and characterization of a large synthetic human antibody phage display library

Advances in proteomic research allow the identification of several hundred protein components in complex biological specimens. Structural information is typically lost during proteomic investigations. For this reason, the rapid isolation of monoclonal antibodies specific to proteins of interest would allow the study of structurally intact biological specimens, thus providing complementary proteomic information. Here, we describe the design, construction, characterization, and use of a large synthetic human antibody phage display library (ETH-2-Gold) containing three billion individual antibody clones. A large repertoire of antibodies with similar biochemical properties was produced by appending short variable complementarity-determining region 3 (CDR3) onto three antibody germline segments (DP47, DPK22, and DPL16), which are frequently found in human antibodies. The ETH-2-Gold library exhibits efficient display of antibody fragments on filamentous phage, as assessed by immunoblot. Furthermore, the library is highly functional, since >90% of clones express soluble antibodies in bacteria and since good quality monoclonal antibodies have been isolated against 16 different antigens. The usefulness of the library as a tool for generating monoclonal antibodies for biomedical applications was tested using the C-domain of tenascin-C (a marker of angiogenesis) as antigen and showing that specific antibodies to this target were able to stain vascular structures in tumor sections.     

Biophysical properties of human antibody variable domains

There are great demands on the stability, expression yield and resistance to aggregation of antibody fragments. To untangle intrinsic domain effects from domain interactions, we present first a systematic evaluation of the isolated human immunoglobulin variable heavy (V(H)) and light (V(L)) germline family consensus domains and then a systematic series of V(H)-V(L) combinations in the scFv format. The constructs were evaluated in terms of their expression behavior, oligomeric state in solution and denaturant-induced unfolding equilibria under non-reducing conditions. The seven V(H) and seven V(L) domains represent the consensus sequences of the major human germline subclasses, derived from the Human Combinatorial Antibody Library (HuCAL). The isolated V(H) and V(L) domains with the highest thermodynamic stability and yield of soluble protein were V(H)3 and V(kappa)3, respectively. Similar measurements on all domain combinations in scFv fragments allowed the scFv fragments to be classified according to thermodynamic stability and in vivo folding yield. The scFv fragments containing the variable domain combinations H3kappa3, H1bkappa3, H5kappa3 and H3kappa1 show superior properties concerning yield and stability. Domain interactions diminish the intrinsic differences of the domains. ScFv fragments containing V(lambda) domains show high levels of stability, even though V(lambda) domains are surprisingly unstable by themselves. This is due to a strong interaction with the V(H) domain and depends on the amino acid sequence of the CDR-L3. On the basis of these analyses and model structures, we suggest possibilities for further improvement of the biophysical properties of individual frameworks and give recommendations for library design.