Crystallization of bi-functional ligand protein complexes Claudia Antoni a,b , Laura Vera a , Laurent Devel a , Maria Pia Catalani b,1 , Bertrand Czarny a , Evelyn Cassar-Lajeunesse a , Elisa Nuti b , Armando Rossello b , Vincent Dive a , Enrico Adriano Stura a,⇑ a CEA, iBiTec-S, Service d’Ingénierie Moléculaire des Protéines (SIMOPRO), Gif-sur-Yvette F-91191, France b Dipartimento di Farmacia, Università di Pisa, Via Bonanno 6, 56126 Pisa, Italy article info Article history: Received 19 February 2013 Received in revised form 26 March 2013 Accepted 27 March 2013 Available online 6 April 2013 Keywords: Crystallization Dimeric ligands Drug-design abstract Homodimerization is important in signal transduction and can play a crucial role in many other biological systems. To obtaining structural information for the design of molecules able to control the signalization pathways, the proteins involved will have to be crystallized in complex with ligands that induce dimer- ization. Bi-functional drugs have been generated by linking two ligands together chemically and the rel- ative crystallizability of complexes with mono-functional and bi-functional ligands has been evaluated. There are problems associated with crystallization with such ligands, but overall, the advantages appear to be greater than the drawbacks. The study involves two matrix metalloproteinases, MMP-12 and MMP- 9. Using flexible and rigid linkers we show that it is possible to control the crystal packing and that by changing the ligand-enzyme stoichiometric ratio, one can toggle between having one bi-functional ligand binding to two enzymes and having the same ligand bound to each enzyme. The nature of linker and its point of attachment on the ligand can be varied to aid crystallization, and such variations can also provide valuable structural information about the interactions made by the linker with the protein. We report here the crystallization and structure determination of seven ligand-dimerized complexes. These results suggest that the use of bi-functional drugs can be extended beyond the realm of protein dimerization to include all drug design projects. Ó 2013 Elsevier Inc. All rights reserved. 1. Introduction The protein is probably the most important variable in protein crystallization (Dale et al., 2003). Among the various strategies that can be envisaged to enhance the crystallizability of difficult pro- teins is complexation with ligands (Hassell et al., 2007). Varying the ligand is also effective since the ligand’s physical, chemical and biological properties affect the crystallization of the complex. In prior studies, rapid crystallization of the erythropoietin receptor was achieved with peptidic ligands that undergo spontaneous dimerization in the presence of the receptor (Livnah et al., 1996). Ligands that can induce homodimerization and those that can bind polyvalently are of particular interest in protein crystallization since they can modify the strength and the relative orientation of the interaction between the molecules that form the lattice. This affects, not only the static assembly, but also the dynamic of crys- tallization. The type of ligands that can be considered vary widely. The polymeric nature of oligosaccharides can mediate the associa- tion of carbohydrate binding proteins, and in some cased induce dimerization that can be visualized in the crystal structure (Flint et al., 2004). Unfortunately, some oligosaccharides, like heparin, can cause aggregation, which complicates but does not prevent crystallization (Stura et al., 2002a). For crystallization, ligand-in- duced protein association presents both advantages and draw- backs that need to be understood so as to select favorable systems and avoid unfavorable ones. Disulphide-mediated dimer- ization can be helpful in Fab crystallization where a cysteine con- taining peptide hapten can dimerize through the formation of a disulfide bond and modify the manner in which a Fab crystallizes (Stura et al., 2002b). On the other hand, it is generally accepted that for ease of crystallization, cysteine-linked F(ab’) 2 should be re- duced to give F(ab)’s and for crystallization. It is not uncommon for surface cysteines to be mutated to serine to prevent heteroge- neous aggregation (Hibi et al., 2002), while in other cases mutagen- esis of residues to cysteines can achieve the formation of intermolecular disulfide bonds to stabilize a crystal contact with the added advantage of an InlB mutant that is constitutively di- meric with exceptionally high signaling activity (Ferraris et al., 2010). Also additives, for example dioxane, can be extremely effec- tive in enhancing the rate of crystal growth when they strengthen a dimeric crystal interface (Ménétrey et al., 2007). 1047-8477/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jsb.2013.03.015 ⇑ Corresponding author. Fax: +33 169082603. E-mail addresses: antoni.claudia@cea.fr (C. Antoni), laura.vera@cea.fr (L. Vera), laurent.devel@cea.fr (L. Devel), mariapia.catalani@inwind.it (M.P. Catalani), bertrand.czarny@cea.fr (B. Czarny), evelyne.lajeunesse@cea.fr (E. Cassar- Lajeunesse), elisa.nuti@farm.unipi.it (E. Nuti), aros@farm.unipi.it (A. Rossello), vincent.dive@cea.fr (V. Dive), estura@cea.fr (E.A. Stura). 1 Present address: Aptuit s.r.l., Via A. Fleming, 4 CAP 37125 Verona, Italy. Journal of Structural Biology 182 (2013) 246–254 Contents lists available at SciVerse ScienceDirect Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi