MINIREVIEW Alternative binding proteins: Anticalins – harnessing the structural plasticity of the lipocalin ligand pocket to engineer novel binding activities Arne Skerra Lehrstuhl fu ¨ r Biologische Chemie, Technische Universita ¨ tMu ¨ nchen, Freising-Weihenstephan, Germany Lipocalins occur in many organisms, such as verte- brates, insects and plants, and even in bacteria, where their physiological role usually lies in the transport or storage of hydrophobic and ⁄ or chemically sensitive organic compounds, especially vitamins, lipids, steroids and other secondary metabolites [1]. Currently, the number of assigned lipocalin sequences has grown beyond 500 [2] and for more than 100 members of this family the 3D structure has been described [3]. In the human body up to 12 different lipocalins, which exert diverse physiological functions, have been identified [4]: a 1 -acid glycoprotein, a 1 -microglobulin, apolipopro- tein D (ApoD), apolipoprotein M, complement component 8c, the epididymal retinoic acid-binding protein, glycodelin, neutrophil gelatinase-associated lipocalin (NGAL, Lcn2), odorant-binding protein, prostaglandin D synthase, retinol-binding protein and tear lipocalin (Tlc, Lcn1). Keywords bacterial expression; b-barrel; CTLA-4; digitalis; fluorescein; ligand binding; lipocalin; molecular recognition; protein engineering; VEGF Correspondence A. Skerra, Lehrstuhl fu ¨ r Biologische Chemie, Technische Universita ¨ tMu ¨ nchen, An der Saatzucht 5, 85350 Freising-Weihenstephan, Germany Fax: +49 8161 714352 Tel: +49 8161 714351 E-mail: skerra@wzw.tum.de (Received 16 November 2007, revised 9 March 2008, accepted 22 March 2008) doi:10.1111/j.1742-4658.2008.06439.x Antibodies are the paradigm for binding proteins, with their hypervariable loop region supported by a structurally rigid framework, thus providing the vast repertoire of antigen-binding sites in the immune system. Lipoca- lins are another family of proteins that exhibit a binding site with high structural plasticity, which is composed of four peptide loops mounted on a stable b-barrel scaffold. Using site-directed random mutagenesis and selection via phage display against prescribed molecular targets, it is possi- ble to generate artificial lipocalins with novel ligand specificities, so-called anticalins. Anticalins have been successfully selected both against small hapten-like compounds and against large protein antigens and they usually possess high target affinity and specificity. Their structural analysis has yielded interesting insights into the phenomenon of molecular recognition. Compared with antibodies, they are much smaller, have a simpler molecu- lar architecture (comprising just one polypeptide chain) and they do not require post-translational modification. In addition, anticalins exhibit robust biophysical properties and can easily be produced in microbial expression systems. As their structure–function relationships are well understood, rational engineering of additional features such as site-directed pegylation or fusion with functional effector domains, dimerization mod- ules or even with another anticalin, can be readily achieved. Thus, antica- lins offer many applications, not only as reagents for biochemical research but also as a new class of potential drugs for medical therapy. Abbreviations ApoD, apolipoprotein D; BBP, bilin-binding protein; CDR, complementarity-determining region; CTLA-4, cytotoxic T-lymphocyte antigen-4; NGAL, neutrophil gelatinase-associated lipocalin; Tlc, tear lipocalin; VEGF, vascular endothelial growth factor. FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS 2677 The lipocalins share a structurally conserved b-barrel as their central folding motif, which is composed of eight antiparallel b-strands that wind around a central axis (Fig. 1). At its open end the cup-like structure supports four loops, which form the entrance to the ligand pocket. The opposite end of the b-barrel is closed by short loops, and densely packed amino acid side chains form the hydrophobic core in this region. As another typical feature, a C-terminal a-helix packs against the b-barrel from one side. Despite extremely low mutual sequence homology, the b-barrel is struc- turally highly conserved among the lipocalins. In con- trast, the loop region around the ligand-binding site exhibits large mutual differences, both in amino acid sequence and length, and in the conformation of the four polypeptide segments [5]. This structural property reflects the many ligand specificities observed for this protein family and resem- bles the hypervariable region that forms the antigen- binding site of antibodies [6]. In the immunoglobulins, six hypervariable loops, also called complementarity- determining regions (CDRs), are supported by the structurally rigid b-sandwich framework of the paired variable domains of the light and heavy chains. These CDRs come together at the tips of the Y-shaped mole- cule to form a contiguous interface for antigen bind- ing. On the basis of this structural resemblance, lipocalins should offer the same potential for molecu- lar recognition as do antibodies. In contrast, natural li- pocalins cannot benefit from the mechanisms of somatic gene recombination and hypermutation, which lead to the vast number of different antibodies gener- ated by the immune system. However, the methods of combinatorial biochemistry can be employed in order to engineer artificial lipocalins with novel specificities for prescribed targets, which were hence dubbed ‘anticalins’ [7,8]. Properties and potential of anticalins Engineered lipocalins offer several advantages over immunoglobulins. Their size, of < 20 kDa, is much smaller than that of antibodies, whose extended molec- ular dimensions hamper efficient tissue penetration. RBP ApoD Tlc NGAL Superposition of Lipocalins Fig. 1. Molecular architecture of human lipocalins and structural variability of their binding sites. Ribbon representation of the crystal struc- tures of four human lipocalins: retinol-binding protein (RBP; PDB entry 1RBP), apolipoprotein D (ApoD; PDB entry 2HZQ), tear lipocalin (Tlc; PDB entry 1XKI) and neutrophil gelatinase-associated lipocalin (NGAL; PDB entry 1L6M). Lipocalins share a conserved b -barrel of eight antiparallel b-strands (cyan). The four exposed loops at its open end (red), which form the natural ligand-binding site, exhibit high structural variability, which is illustrated by the superposition shown to the right. Anticalins A. Skerra 2678 FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS Furthermore, lipocalins have a rather simple composi- tion, which is based on a single polypeptide chain. In contrast, antibodies comprise two different polypep- tides (i.e. the light and heavy chains), which leads to unstable domain association when dealing with small Fv fragments and which also requires complicated cloning steps for recombinant expression. With four structurally variable loops, the binding site of lipoca- lins is less complex and easier to manipulate [5] than the CDR of antibodies, which is composed of alto- gether six non-sequential loop segments from both immunoglobulin chains [6]. Naturally, lipocalins lack the constant Fc region, which mediates immunological effector functions but often causes undesired interactions of antibodies while being crucial only for a few biopharmaceutical applica- tions. Finally, many lipocalins lack glycosylation and can thus be produced as authentic proteins in micro- bial expression systems, whereas the manufacture of glycosylated full-size antibodies requires expensive eukaryotic cell culture, whose optimization and fer- mentation is time-consuming and prone to limited capacities [9]. While some of these benefits have also been claimed for engineered single-chain variable frag- ments of antibodies or isolated VHH domains of cam- eloid immunoglobulins, for example their practical applicability compared with intact antibodies, espe- cially for medical purposes, is still unclear [10]. Similarly to the immunoglobulins, human lipocalins occur as soluble proteins in the plasma and other tissue fluids, with concentrations up to approximately 1.0 mgÆmL )1 . Most of the lipocalins are freely distrib- uted in the body, where they exert a ligand buffer or transport function. This predestines this family of proteins not only as carrier vehicles or scavengers for pharmaceutically active compounds but also, especially when engineered for novel binding functions, as thera- peutic drugs on their own [11]. Both natural and engineered lipocalins are often surprisingly stable, with melting temperatures above 70 °C [12], and they are easily produced in Escherichia coli in a functional state [4]. The recombinant lipoca- lins can be recovered as soluble monomeric proteins, even when lacking natural glycosylation (eg: ApoD and NGAL). Lipocalins are typical secretory proteins, both in vertebrates and in lower organisms such as insects, and thus they often carry one or two disul- phide bonds. Consequently, bacterial production via a secretory route is the method of choice [4,13], albeit several recombinant lipocalins were also successfully isolated from the soluble cytoplasmic extract of E. coli [14,15]. As the disulphide bridges are not buried in the hydrophobic interior of lipocalins – but rather serve for cross-linking the N- and C-terminus to the b-barrel [5] – they are not as crucial for folding as is the case for immunoglobulins. Indeed, several natural lipocalins devoid of disulphide bonds exist (eg: the human epi- didymal and the bacterial lipocalins), and in other li- pocalins (e.g. Tlc) the single disulphide bond can be eliminated without much loss of protein stability. Interestingly, especially among the human lipocalins, many members carry an additional free Cys residue. Its reactive thiol side chain sometimes serves for cross-linking to other plasma proteins, although the physiological function is often not known. Thus, for application as research reagents, or in medical diagnos- tics as well as therapy, it is usually advisable to substi- tute the unpaired Cys residue with an inert amino acid, such as Ser [4]. On the other hand, a free Cys res- idue can be used for the site-specific covalent attach- ment of functional groups via maleimide chemistry, including fluorescent labels or poly(ethylene glycol), which can serve for plasma half-life extension [16]. Lipocalins are also well suited for the construction of functional fusion proteins. The fusion of anticalins with alkaline phosphatase, for example, leads to useful reporter reagents [17]. Anticalins may even be fused with each other, yielding either bivalent or bispecific- binding proteins, so-called ‘duocalins’ [18]. Anticalins recognizing small molecules Initially, the structurally and biochemically well char- acterized bilin-binding protein (BBP) of Pieris brassi- cae [19] was employed to engineer an artificial binding site for ligands such as fluorescein and digoxigenin, as well as other small molecules and peptides. This lipoc- alin comprises 174 residues and exhibits a rather wide and shallow ligand pocket, where biliverdin IX c is complexed as natural ligand. Sixteen residues distrib- uted across all four loop segments, whose side chains form the centre of the binding site, were identified by molecular modelling and subjected to concerted random mutagenesis, followed by phagemid display selection for variants with novel binding activities [7]. In the case of fluorescein, which was chosen as a well-known immunological hapten, several variants with high specificity and dissociation constants as low as 35.2 nm were identified. Following X-ray structural analysis of the complex between the engineered lipoca- lin and this ligand [20], improved variants with K D values for fluorescein of approximately 1 nm were rationally engineered just by optimizing two side chains in the binding pocket [21]. Thus, it was demon- strated that engineered lipocalins with novel specifici- ties (i.e. anticalins) can provide hapten-binding A. Skerra Anticalins FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS 2679 proteins with affinities in a range that was so far con- sidered typical for antibodies. Notably, the BBP vari- ants appeared to recognize fluorescein – or other small molecule targets – as true haptens, without measurable context-dependence concerning the carrier protein that served for ligand immobilization during the phage-dis- play panning process. With their ability to provide deep and highly complementary ligand pockets, antica- lins distinguish themselves from most other protein scaffolds that are currently under investigation [22]. From the same BBP mutant library an anticalin with specificity for the cardiac steroid digoxigenin was selected [17]. Its initially moderate affinity was subse- quently raised by selective random mutagenesis of the first hypervariable loop, followed by phagemid display and colony screening under more stringent conditions, thus resulting in a 10-fold improved K D value of 30.2 nm. Attempts to raise the affinity for digoxigenin even further were made in a combinatorial approach using a ‘loop-walking’ randomization strategy [12] and also by rational protein design based on the crystal structure of this engineered lipocalin [23]. These approaches allowed the identification of several point mutations, leading to K D values as low as 800 pm for digoxin (i.e. the natural glycosylated derivative of digoxigenin) [11]. The crystal structures of these first anticalins in com- plex with their ligands – and in one instance also as apo-protein – which were solved at resolutions of 2.0 A ˚ or better [20,23], provided interesting insight into the mechanism and specificity of molecular recognition by engineered lipocalins (Fig. 2). Most importantly, the extensive replacement of side chains, affecting 10% of all residues in the BBP, did not impair the b-barrel fold. The randomized loops, on the other hand, adopted dramatically altered conformations compared with the wild-type lipocalin. Both fluorescein and digoxigenin are bound at the bottom of the cleft that harbours biliverdin IX c in the BBP [24]. Thus, while the overall topology of the lipocalin, comprising the b-barrel with the a-helix attached to it, remained con- served for both anticalins, the set of four loops at the entrance to the ligand pocket exhibited pronounced conformational differences in comparison with each other and with the BBP. These structural changes seem to be triggered by the amino acid substitutions that were introduced during the combinatorial engineering of the anticalins rather than by complex formation with the ligand, thus illustrating the inherent structural plasticity of the lipocalin loop region. Indeed, the mechanism of complex formation, at least with low-molecular-weight ligands, appears to be similar to the interaction between antibodies and hap- tens, except that the ligand can be buried more deeply in the engineered lipocalin pocket. Shape complemen- tarity is mainly generated by means of aromatic side chains, and specific interactions arise from suitably placed hydrogen-bond donors or acceptors, sometimes mediated by buried water molecules. Notably, in the case of the digoxigenin-binding anticalin, DigA16, the bound steroid ligand is sandwiched between one Trp and two Tyr side chains, very similar to a monoclonal antibody directed against digoxin, which provides a nice example of ‘convergent’ in vitro evolution [23]. In addition, a His side chain at the bottom of the ligand pocket displays an induced fit upon complex formation with digoxigenin, an effect so far regarded as common in antibodies. Further to the pronounced backbone plasticity in the loop region, comparison of the pri- mary sequences of many engineered lipocalins revealed that all randomized amino acid positions essentially tolerate the entire set of natural side chains. Apart from these fundamental insights into the struc- ture–function relationships of lipocalins and their simi- larity to immunoglobulins, the resulting anticalin, which was designated Digical, may be applicable as a therapeutic agent for the treatment of digitalis intoxica- tions. Although digitalis is widely applied in conjunc- tion with heart insufficiency and arrhythmias [25], it has a very narrow therapeutic window, and precise adjustment of digoxin plasma levels is mandatory to prevent poisoning with fatal outcome. Indeed, when Digical was employed for studies in a guinea-pig animal model of digitalis intoxication, the anticalin appeared N Loop #4 Loop #1 C Loop #2 Loop #3 Fig. 2. 3D structure of an anticalin in complex with its cognate ligand. Ribbon representation of the crystal structure of the digoxi- genin-binding anticalin DigA16 (PDB entry 1LKE). The bound ligand is shown in a space-filling representation in yellow, whereas the 16 amino acid side chains in the four hypervariable loops – as well as the adjoining regions of the b-barrel – which were randomized in the naive combinatorial library derived from the BBP used for the anticalin selection, are depicted in orange. The N-terminus (N) and the C-terminus (C) of the polypeptide chain are labelled. Anticalins A. Skerra 2680 FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS to be effective in reversing the digoxin-induced toxicity after administering just a moderate stoichiometric excess [11], thus demonstrating the acute protective effect of this anticalin on the cardiovascular system and its suitability as an antidote against digoxin. Furthermore, the anticalin FluA, which possesses high affinity for fluorescein, has the interesting prop- erty of almost completely quenching the fluorescence emission of this widely applied reagent [7]. The reason for the disappearance of the stationary ligand fluores- cence seems to be an ultrafast electron transfer between the excited fluorescein dianion and a Trp side chain in the binding site of the engineered lipocalin, which closely packs against the xanthenolone moiety [26]. This phenomenon opens interesting applications in biophysics. Such an ‘anti-fluorescent’ protein could also be useful as a reagent for the specific quenching of background signals that arise from fluorescein groups surrounding a cell, for example when deter- mining the topology of a site-specifically labelled membrane protein. Anticalins directed at proteins Considering medical applications, extracellular proteins or cell-surface receptors are the predominant class of biomolecules that currently provide relevant targets for biopharmaceuticals such as antibodies. Consequently, in recent years anticalin libraries were specifically developed for the recognition of such protein ‘anti- gens’. In addition, to reduce immunogenic side effects upon prolonged treatment, these libraries were con- structed on the basis of natural human lipocalins, in particular ApoD [27,28], NGAL [29] and Tlc [30] (Fig. 1). To this end, 16–24 amino acid residues located at exposed positions, close to the tips of the four hyper- variable loops, were subjected to random mutagenesis in order to allow tight contact formation with a mac- romolecular target, which cannot penetrate as deeply into the ligand-binding site as a small molecule. Using these libraries, anticalins with high specificity and affinities in the subnanomolar range were successfully selected against a variety of disease-related protein antigens, including immunological receptors such as cytotoxic T-lymphocyte antigen-4 (CTLA-4) [11] and soluble growth factors such as vascular endothelial growth factor (VEGF) [31]. Recently, the crystal structure of the complex between a cognate anticalin and the extracellular domain of CTLA-4 was solved, demonstrating that a macromolecular ‘protein antigen’ can be effectively bound at the cup-shaped binding site of an engineered lipocalin, even though its natural counterparts almost exclusively recognize low-molecular-weight substances. All four randomized loops of NGAL – which had served as a lipocalin scaffold in this case – contribute to the formation of the molecular complex, thus vali- dating the design of the anticalin library. CTLA-4 (CD152) is an activation-induced, trans- membrane T-cell coreceptor with an inhibitory effect on T-cell-mediated immune responses [32]. CTLA-4 antagonizes the CD28-dependent costimulation of T cells, whereby CTLA-4 and CD28 share the same counter-receptors on antigen-presenting cells (i.e. B7.1 and B7.2). Notably, the bound anticalin shields the CTLA-4 epitope that is involved in the interaction both with B7.1 and B7.2. Indeed, an antagonistic activity of the anticalin towards CTLA-4 was con- firmed in several in vitro cell culture tests, where T-cell proliferation was stimulated in a manner comparable to that of commercially available antibodies directed against the same target. Thus, the CTLA-4-specific anticalin is a promising drug candidate for the immu- notherapy of cancer, similarly to corresponding anti- bodies that are already in clinical trials [33]. Apart from its much smaller size and probably better tissue penetration, the lack of immunological effector func- tions – which reside in the antibody Fc region – for the anticalin should limit off-target toxicity because only the antagonistic activity is needed. In fact, this is the case for many relevant targets involved in the regu- lation of the immune response and inflammation as well as neoangiogenesis. Another promising drug candidate is an anticalin with strong antagonistic activity towards VEGF. VEGF is a well-characterized mediator of tumor angiogenesis and other neovascular diseases [34], for example age-related macular degeneration (AMD). The selected anticalin exhibits a favorable binding and activity profile in direct comparison with currently approved VEGF antagonists [31]. A half-life extended version of the anticalin has demonstrated excellent effi- cacy in three animal models assessing VEGF-induced enhanced vascular permeability, angiogenesis and anti- xenograft tumor activity. As immunological effector functions again appear to be irrelevant for biomedical activity, an anticalin with proven VEGF-antagonistic function should offer an interesting alternative to full- size antibodies, especially in the light of its presumably better distribution. Conclusions and prospects Engineered lipocalins offer binding sites with surpris- ingly high structural plasticity and an extended A. Skerra Anticalins FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS 2681 molecular interface for target recognition which is comparable in size to that of antibodies. Anticalins with high specificity and affinity, down to picomolar dissociation constants, can be readily generated against haptens, peptides and proteins. Thus, regarding their range of addressable targets they surpass other protein scaffolds that are presently pursued [22]. Available structural and functional data suggest that anticalins are able to recognize a diverse set of epitopes on differ- ent target proteins and therefore have considerable potential as specific antagonistic reagents in general. Consequently, anticalins constitute promising reagents for therapeutic applications. As anticalins can be derived from human lipocalin scaffolds, the risk of immunogenicity is minimized and further reformatting – such as by CDR grafting for the ‘humanization’ of antibodies – is not required. The absence of immunological effector functions prevents many potential side effects known for antibodies. Furthermore, the monovalent binding activity of anticalins decreases the risk of intermolecular cross- linking of cellular receptor targets that could lead to unwanted signal triggering. Natural lipocalins, as well as engineered lipocalins, are quickly cleared by renal filtration, as a result of their small size of approximately 20 kDa, if they circu- late as monomeric proteins. When conjugated with radioactive isotopes for in vivo diagnostics, for exam- ple, such properties should lead to images of high contrast soon after administration. Nevertheless, for medical indications that require prolonged treatment, the simple architecture and robustness of the lipocalin scaffold facilitates the preparation of fusion proteins or of site-directed conjugates to decelerate clearance. In principle, several established techniques are avail- able to extend the plasma half life of anticalins, for example by the production of fusion proteins with serum albumin, with an albumin-binding domain or peptide or via pegylation. Anticalins display both their N-terminus and C-ter- minus in an accessible manner and remote from the binding site, which differs from the situation with sin- gle-chain variable fragments of antibodies, where the N-terminus often forms part of the paratope. Thus, anticalins are well suited for fusion with other func- tional domains without compromising their engineered binding activities. Fusion proteins of anticalins that address a specific receptor on solid tumors with enzymes which generate a cytotoxic compound from an inactive precursor (prodrug) might be of special interest as an alternative to antibody-directed enzyme prodrug therapy (ADEPT). Furthermore, a dimeric binding mode utilizing either a duocalin that has twice the same target specificity or a fusion protein between an anticalin and a dimerization domain may be employed to enhance binding avidity. Hence, owing to their adaptable binding site and their simple and robust molecular architecture, specifi- cally engineered anticalins promise a future as versatile reagents for research, biotechnology and medicine. References 1A ˚ kerstro ¨ m B, Borregaard N, Flower DA & Salier J-S (2006) Lipocalins. Landes Bioscience, Georgetown, TX. 2 Finn RD, Mistry J, Schuster-Bockler B, Griffiths-Jones S, Hollich V, Lassmann T, Moxon S, Marshall M, Khanna A, Durbin R et al. (2006) Pfam: clans, web tools and services. Nucleic Acids Res 34, D247–D251. 3 Berman HM, Westbrook J, Feng Z, Gilliland G, Bhat TN, Weissig H, Shindyalov IN & Bourne PE (2000) The Protein Data Bank. Nucleic Acids Res 28, 235–242. 4 Breustedt DA, Scho ¨ nfeld DL & Skerra A (2006) Com- parative ligand-binding analysis of ten human lipoca- lins. Biochim Biophys Acta 1764, 161–173. 5 Skerra A (2000) Lipocalins as a scaffold. Biochim Bio- phys Acta 1482, 337–350. 6 Skerra A (2003) Imitating the humoral immune response. Curr Opin Chem Biol 7, 683–693. 7 Beste G, Schmidt FS, Stibora T & Skerra A (1999) Small antibody-like proteins with prescribed ligand specificities derived from the lipocalin fold. Proc Natl Acad Sci U S A 96, 1898–1903. 8 Skerra A (2001) ‘Anticalins’: a new class of engineered ligand-binding proteins with antibody-like properties. J Biotechnol 74, 257–275. 9 Werner RG (2004) Economic aspects of commercial manufacture of biopharmaceuticals. J Biotechnol 113, 171–182. 10 Holliger P & Hudson PJ (2005) Engineered antibody fragments and the rise of single domains. Nat Biotech- nol 23, 1126–1136. 11 Schlehuber S & Skerra A (2005) Lipocalins in drug dis- covery: from natural ligand-binding proteins to ‘antica- lins’. Drug Discov Today 10, 23–33. 12 Schlehuber S & Skerra A (2002) Tuning ligand affinity, specificity, and folding stability of an engineered lipoca- lin variant – a so-called ‘anticalin’ – using a molecular random approach. Biophys Chem 96, 213–228. 13 Mu ¨ ller HN & Skerra A (1993) Functional expression of the uncomplexed serum retinol-binding protein in Escherichia coli. Ligand binding and reversible unfold- ing characteristics. J Mol Biol 230, 725–732. 14 Coles M, Diercks T, Muehlenweg B, Bartsch S, Zolzer V, Tschesche H & Kessler H (1999) The solution struc- ture and dynamics of human neutrophil gelatinase- associated lipocalin. J Mol Biol 289, 139–157. Anticalins A. Skerra 2682 FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS 15 Holzfeind P, Merschak P, Rogatsch H, Culig Z, Feicht- inger H, Klocker H & Redl B (1996) Expression of the gene for tear lipocalin ⁄ von Ebner’s gland protein in human prostate. FEBS Lett 395, 95–98. 16 Harris JM & Chess RB (2003) Effect of pegylation on pharmaceuticals. Nat Rev Drug Discov 2, 214–221. 17 Schlehuber S, Beste G & Skerra A (2000) A novel type of receptor protein, based on the lipocalin scaffold, with specificity for digoxigenin. J Mol Biol 297, 1105–1120. 18 Schlehuber S & Skerra A (2001) Duocalins: engineered ligand-binding proteins with dual specificity derived from the lipocalin fold. Biol Chem 382, 1335–1342. 19 Schmidt FS & Skerra A (1994) The bilin-binding pro- tein of Pieris brassicae. cDNA sequence and regulation of expression reveal distinct features of this insect pigment protein. Eur J Biochem 219, 855–863. 20 Korndo ¨ rfer IP, Beste G & Skerra A (2003) Crystallo- graphic analysis of an ‘‘anticalin’’ with tailored specific- ity for fluorescein reveals high structural plasticity of the lipocalin loop region. Proteins Struct Funct Genet 53, 121–129. 21 Vopel S, Mu ¨ hlbach H & Skerra A (2005) Rational engi- neering of a fluorescein-binding anticalin for improved ligand affinity. Biol Chem 386, 1097–1104. 22 Skerra A (2007) Alternative non-antibody scaffolds for molecular recognition. Curr Opin Biotechnol 18, 295– 304. 23 Korndo ¨ rfer IP, Schlehuber S & Skerra A (2003) Struc- tural mechanism of specific ligand recognition by a lipocalin tailored for the complexation of digoxigenin. J Mol Biol 330, 385–396. 24 Huber R, Schneider M, Mayr I, Mu ¨ ller R, Deutzmann R, Suter F, Zuber H, Falk H & Kayser H (1987) Molecular structure of the bilin binding protein (BBP) from Pieris brassicae after refinement at 2.0 A ˚ resolu- tion. J Mol Biol 198, 499–513. 25 Hauptman PJ & Kelly RA (1999) Digitalis. Circulation 99, 1265–1270. 26 Go ¨ tz M, Hess S, Beste G, Skerra A & Michel-Beyerle ME (2002) Ultrafast electron transfer in the complex between fluorescein and a cognate engineered lipocalin protein, a so-called anticalin. Biochemistry 41, 4156– 4164. 27 Eichinger A, Nasreen A, Kim HJ & Skerra A (2007) Structural insight into the dual ligand specificity and mode of high density lipoprotein association of apolipo- protein D. J Biol Chem 282, 31068–31075. 28 Vogt M & Skerra A (2004) Construction of an artificial receptor protein (‘‘anticalin’’) based on the human apolipoprotein D. ChemBioChem 5, 191–199. 29 Goetz DH, Holmes MA, Borregaard N, Bluhm ME, Raymond KN & Strong RK (2002) The neutrophil lipocalin NGAL is a bacteriostatic agent that interferes with siderophore-mediated iron acquisition. Mol Cell 10, 1033–1043. 30 Breustedt DA, Korndo ¨ rfer IP, Redl B & Skerra A (2005) The 1.8-A ˚ crystal structure of human tear lipoca- lin reveals an extended branched cavity with capacity for multiple ligands. J Biol Chem 280, 484–493. 31 Hohlbaum AM & Skerra A (2007) Anticalins: the lipoc- alin family as a novel protein scaffold for the develop- ment of next-generation immunotherapies. Expert Rev Clin Immunol 3, 491–501. 32 Leach DR, Krummel MF & Allison JP (1996) Enhance- ment of antitumor immunity by CTLA-4 blockade. Science 271, 1734–1736. 33 Tarhini AA & Kirkwood JM (2007) Tremelimumab, a fully human monoclonal IgG2 antibody against CTLA4 for the potential treatment of cancer. Curr Opin Mol Ther 9, 505–514. 34 Ferrara N, Gerber HP & LeCouter J (2003) The biol- ogy of VEGF and its receptors. Nat Med 9, 669–676. A. Skerra Anticalins FEBS Journal 275 (2008) 2677–2683 ª 2008 The Author Journal compilation ª 2008 FEBS 2683 . MINIREVIEW Alternative binding proteins: Anticalins – harnessing the structural plasticity of the lipocalin ligand pocket to engineer novel binding activities Arne. by the structurally rigid b-sandwich framework of the paired variable domains of the light and heavy chains. These CDRs come together at the tips of the