Binders Based on Dimerised Immunoglobulin V H Domains Jorge Sepu ´ lveda 1 , Hulin Jin 1 , Daniele Sblattero 2 , Andrew Bradbury 3 and Oscar R. Burrone 1 * 1 International Centre for Genetic Engineering and Biotechnology, Molecular Immunology, Area Science Park, Padriciano 99, 34012 Trieste, Italy 2 Department of Biology University of Trieste, Trieste Italy 3 Bioscience Division MS-M888, TA-43 HRL-1 Building 1, Los Alamos National Laboratory, Los Alamos, NM 87545, USA Antibody binding to antigen is mediated by the surface formed by the association of the two variable (V) regions of the L (V L ) and H (V H ) chains. The capacity of V L to dimerise and the high structural similarity of V L and V H domains suggested the possibility that V H could also associate. We show here that spontaneous formation of V H dimers (VHD) is in many cases permissive, producing stable molecules with antigen binding speci- ficity. VHD were displayed on filamentous phages for the selection of antigen-specific binders. VHD were expressed and secreted efficiently from both bacteria and mammalian cells in different formats, including single-chain (V H (1)-linker-V H (2)), double chain ((V H ) 2 ) and IgG analogues having the V L replaced by V H . The affinity (K d,app ) achieved with a V H dimer expressed in the IgG format, specific for a glutenin subunit was around 30 nM measured by two different methods, which was about 20 times higher than that corresponding to the V L /V H counterpart. q 2003 Elsevier Ltd. All rights reserved. Keywords: antibody engineering; binders; immunoglobulin domains *Corresponding author Introduction The antigen-binding site of antibody molecules is formed by the surface derived from the parallel association of the two variable domains (V L and V H ) of the light (L) and heavy (H) chain, respect- ively. Both domains resemble each other in second- ary and tertiary structure, each having two layers of b-sheets supporting three loops, termed comple- mentarity determining regions (CDR1, CDR2, CDR3), also designated L1, L2 and L3 for the L chain and H1, H2 and H3 for the H chain, that con- stitute the basic elements for antigen recognition. In most cases, the contribution of V H to antigen binding is more relevant than that of V L . 1 The bur- ied surface of the antigen-binding site is occupied mainly by V H -derived residues, especially from loop H3. 1,2 H3 differs from the other five loops, 3–9 in having a greater variation in sequence, length and conformational heterogeneity. 3,10 Moreover, transgenic mice with only one V H and two V L gene segments, but full H3 diversity, are able to mount highly specific antibody responses to most antigens. 11 Many hydrophobic residues responsible for V L / V H inter-domain interaction have been identified to be highly conserved in humans and mice, 12,13 and are found mainly in FR2 and FR4. Interest- ingly, V region genes of IgG1 of camelids that do not associate with L chains have these residues substituted by hydrophilic residues, thus contri- buting to V H solubility. 14 The conservation of the V L /V H interface, involving an area of around 1000 – 1700 A ˚ 2 , 2,15 allows a very permissive associ- ation between the many different L and H chains, 16 although some unknown aspects can determine biased V L /V H combinations. 17 It has been suggested that the V L /V H interface is an adaptable structure, modulated both by the CDRs and by interaction with antigen. 18 Upon antigen binding, the two V regions can vary their relative orien- tation by as much as 168. 19 An unusual mode of domain–domain association is the V L –V L interaction, found in most cases of Bence –Jones disease and many myelomas lacking the H chain. This closely mimics the interface 0022-2836/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. E-mail address of the corresponding author: burrone@icgeb.org Abbreviations used: V L , and V H , variable domains of the light (L) and heavy (H) chain, respectively; CDR1, CDR2 and CDR3, complementarity-determining regions also designated L1, L2 and L3 for the L chain and H1, H2 and H3 for the H chain; dc, double chain; sc, single- chain. doi:10.1016/j.jmb.2003.08.033 J. Mol. Biol. (2003) 333, 355–365