The present invention discloses ligand regulated protein- protein interaction systems and their use in diagnosis and ther apy, especially in tumour therapy.

Protein-protein interactions (PPIs) are the physical con tacts of high specificity established between two or more pro tein molecules as a result of biochemical events steered by electrostatic forces including the hydrophobic effect. Many are physical contacts with molecular associations between chains that occur in a cell or in a living organism in a specific bio- molecular context. PPIs have also been used in the prior art for establishing screening systems or defined switches for pharma ceutical purposes, especially in human medicine.

A specific example of PPIs is dimerization, especially chem ically induced dimerization (CID) by small molecules. There are several systems for CID of proteins by small molecules. In these systems PPIs, i.e. homo- or heterodimerization, can only occur in the presence of a small molecule, which thus acts as a chemi cal dimerizer. In most cases, the dimerizing proteins, or re spective domains thereof, are expressed as parts of fusion con structs containing the proteins of interest. Some of the systems are functional not only inside cells but also in the oxidative environment outside from cells. Until now CID systems mainly have been used for, e.g., regulating enzyme function, signal transduction, gene transcription, genome editing (by e.g. CRISPR/Cas), protein stability, and for generating logic gates. CID systems are also increasingly considered for clinical appli cations, as e.g., for regulating the function of T cells modi fied with a chimeric antigen receptor. The most frequently used system is based on FRB/FBKP-domains for heterodimerization or a mutant FKBP-domain for homodimerization. Several rapamycin de rivatives have been developed, but they have not yet solved the problem of simultaneously regulating two processes in the same cell. Such functionality has been introduced by the more recent development of CID systems, which are based on different pro teins and also different small molecules. These systems work well in vitro, however, their use in vivo is limited. A much higher impact would have the development of CID systems that are suited for in vivo and even clinical application and for simul- taneous control of multiple processes. The current systems are not suited for this purpose, since they are based on xenogeneic proteins and are thus expected to be highly immunogenic, and/or since the small molecules are not suited for in vivo applica tion.

The FKBP-based system for homodimerization is the most ad vanced system and has been in clinical use for several years for inducing apoptosis in administered T cells. It is based on a hu man protein, depends on an inert molecule and is fully orthogo- nalized, i.e., the dimerizer does not bind to endogenous pro teins and the mutated FKBP domain binds only the synthetic di merizer. This process of insulating a protein ligand pair, called "orthogonalization", prevents unwanted interaction in both directions and is essential for the use of CID systems in cellular systems. In the case of heterodimerization even the most advanced human protein based system (FKBP/FRB-system) is still not fully orthogonal. Engineering of the protein binding pocket and complementarily of the small molecule was performed only for the interaction of the rapalog with FRB but not with FKBP, which is still based on the wild-type FBKP domain. Due to the binding mode of rapalogues to FRB and FKBP (see PDB 1NSG for the structure) , redesigning the rapalogues for orthogonalization and for pharmacokinetic optimization poses a serious obstacle. Furthermore, the synthesis of rapalogues is difficult, potential contamination with rapamycin is a danger, and the molecules are not clinically approved.

Examples for CID systems have been disclosed in WO 2014/127261 A1 and WO 2017/032777 A1. Voss et al . (Curr. Op. Chem. Biol. 28 (2015): 194-201) disclose chemically induced di merization systems, mainly rapamycin-based CID systems, and their reversible and spatiotemporal control of protein function in cells by using such CIDs. As a result, however, it was re ported that these current CID systems, even these advanced ra pamycin-based CID systems, can still induce undesired heterodi merization of FKBP and FRB. DeRose et al . (Europ. J. Physiol. 465 (2013) : 409-417) review CID techniques to resolve problems in cell biology.

It is an object of the present invention to provide ligand regulated PPI systems (LRPPI) that can be regulated by small molecules for inducing dimerization of interaction partners and which are advantageous compared to existing systems. More spe cifically, novel LRPPI systems should be applicable in vivo, es pecially for the treatment of human patients, without the risk of adverse reactions or at least with reduced adverse reactions. It is a further object to provide means for tumour treatment, especially immunotherapy concepts for the treatment of tumours.

Therefore, the present invention provides a ligand regulated protein-protein interaction system based on a lipocalin-fold molecule comprising:

(a) a lipocalin-fold molecule

(b) a lipocalin-fold ligand with a low molecular weight of 1500 Da or below, and

(c) a lipocalin-fold binding interaction partner,

wherein the lipocalin-fold molecule can bind to the lipocalin- fold ligand; and

wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand binds to the lipocalin-fold binding interaction partner with an affinity which is at least 10-fold higher than the af finity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand,

and wherein the lipocalin-fold binding interaction partner is not a naturally occurring protein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule in the presence of any lipocalin-fold ligand.

The present invention also refers to a ligand regulated pro tein-protein interaction system based on a lipocalin-fold mole cule comprising:

(a) a lipocalin-fold molecule

(b) a lipocalin-fold ligand with a low molecular weight of 1500 Da or below, and

(c) a lipocalin-fold binding interaction partner,

wherein the lipocalin-fold molecule has at least a first confor mation when the lipocalin-fold ligand is not bound to the lipocalin-fold molecule and at least a second conformation when the lipocalin-fold ligand is bound to the lipocalin-fold mole cule; and

wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand in the second conformation binds to the lipocalin-fold binding interaction partner with an affinity which is at least 10-fold higher than the affinity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand in the first confor mation,

and wherein the lipocalin-fold binding interaction partner is not a naturally occurring protein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule in the presence of any lipocalin-fold ligand.

The present invention is based on the high flexibility of the lipocalin-fold molecule system for binding to its binding partners ("lipocalin-fold binding interaction partner") with and without involvements of other molecules, especially small mole cules which can be used as pharmaceutical agents. Although the system is based on naturally occurring counterparts, the systems of the present invention are artificial. This means that the lipocalin-fold molecule and/or the lipocalin-fold binding inter action partner may be derived from naturally occurring scaffolds by adapting the binding affinities and specificities of the three essential components of the present system ( lipocalin-fold molecule, lipocalin-fold ligand and the lipocalin-fold binding interaction partner) to the binding specificities needed, espe cially those which are required in a targeted and specific phar maceutical intervention on human patients.

The feature that the lipocalin-fold binding interaction partner is not a naturally occurring binding partner of a natu rally occurring lipocalin-fold molecule is important for reduc ing or avoiding significant cross-reactivity (which is disadvan tageous for the intended use of the present invention for phar maceutical application) . Such cross-reactivity could lead to side-effects which can eventually become severe (such side ef fects could also be hard to predict, if naturally occurring lipocalin-fold binding interaction partners were part of the "on-switch" system based on a lipocalin-fold molecule) .

In the course of the present invention, it was found out that LRPPI systems based on lipocalin-fold molecules have sur prisingly advantageous properties which make them excellently suitable for establishing LRPPI systems which have also the ca pability of working in vivo in human pharmaceutical therapy. This is not only based on the fact that lipocalin-fold molecule affinities for lipocalin-fold ligands are easily "tunable", spe cifically in the micro- and nanomolar range, but also due to the robust architecture of the central structure of the lipocalin- fold molecule (see further disclosure to the "lipocalin-fold", below) and the ability of some lipocalin-fold ligands to bind in the lipocalin-fold molecule. Due to these excellent properties, the LRPPI systems according to the present invention can be de signed very specific, robust and sensitive to allow use in human medicine .

The present LRPPI system is therefore based on three essen tial components: the lipocalin-fold molecule ("a" in Fig. 1), the lipocalin-fold ligand ("b" in Fig. 1) and the lipocalin-fold binding interaction partner ("c" in Fig. 1) .

The "lipocalin-fold molecule" according to the present in vention can be any naturally occurring lipocalin-fold molecule or derived version thereof that is part of the "lipocalin-fold" superfamily of proteins according to the Structural Classifica tion of Proteins (SCOP) database (version 1.75) (Murzin et al . J Mol Biol. 1995; 247 (4) : 536-540) . It is this structural motif of the "lipocalin-fold" proteins that allows the generation of the flexible LRPPI systems according to the present invention. In cluding the lipocalin-fold molecules in the LRPPI system accord ing to the present invention was thus a crucial step towards en abling flexible engineering of orthogonal LRPPI systems for clinical use. In the course of the present invention, this lipocalin-fold was identified as a uniquely shaped scaffold that can inherently bind a large variety of structurally different small molecules ("lipocalin-fold ligands" (or "ligands") accord ing to the present invention) and can be engineered even for binding of small molecules that initially cannot be captured. A lipocalin-fold protein contains a small characteristic 8- or 10- stranded up-and-down b-barrel motif, in which the antiparallel b-strands are arranged in a +1 topology and which has evolved to wrap around mostly hydrophobic small molecule ligands (Lakshmi et al . PLoS One. 2015;10(8): e0135507; Zhang et al . , PLoS One. 2012;7(5): e36772; Smathers et al . , Hum Genomics. 2011 ; 5 (3) : 170- 191; Grzyb et al . , J Plant Physiol. 2006 ; 163 ( 9) : 895- 915 ; Flower et al . , Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 9-24 ; Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) . In the SCOP database (version 1.75) the lipocalin-fold comprises only the lipocalin superfamily (Lakshmi et al . PLoS One. 2015;10(8): e0135507). Among the 9 families that are assigned to the lipocalin super family, retinol binding protein-like and fatty acid binding pro- tein-like proteins comprise the lipocalins (LCNs) and intracel lular lipid binding proteins (iLBPs) , respectively, and are the most relevant families. LCNs are 8-stranded b-barrel proteins (molecular mass roughly 20 kDa) (Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) and iLBPs are 10-stranded b-barrel proteins (molecular mass roughly 15 kDa) that comprise FABPs (fatty acid binding proteins) , CRBPs (cellular retinol binding proteins) and CRABPs (cellular retinoic acid binding proteins) (Smathers et al . , Hum Genomics. 2011 ; 5 (3) : 170-191 ) . It is assumed that lipocalin-fold proteins have evolved from a common ancestor by gene duplication and evolutionary divergence for adapting to capture a multitude of different small molecules. LCNs have al ready evolved in bacteria (>600 LCNs have been described among species; the human genome encodes at least 15 members), and iLBPs have likely evolved in animals after divergence from fungi and plants (the human genome encodes 10 FABPs and 6 retinoid binding proteins; Lakshmi et al . PLoS One. 2015;10(8): e0135507; Zhang et al . , PLoS One. 2012;7(5): e36772; Smathers et al . , Hum Genomics. 2011 ; 5 (3) : 170-191 ) . All these proteins have maintained a striking structural homology despite very low sequence homolo gy (for some FABPs around 20% and for many LCNs below 30%) , which illustrates the extraordinary high tolerance of the b- barrel structure of the lipocalin superfamily to mutations in virtually all regions of the structure for adaption to binding of different ligands. In fact, it has not been possible to de fine any sequence motif that is common to all members of the lipocalin superfamily, i.e. especially the lipocalins and iLBPs (Flower et al . , Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 9-24 ) .

This extraordinary flexible b-barrel structure of the lipocalin-fold is used according to the present invention as a broad platform (based on the "lipocalin-fold molecule") for gen erating a novel family of LRPPI systems that can be regulated by ligands which insert into the hydrophobic pocket of the lipocalin-fold . Such regulating lipocalin-fold ligands can be chosen from a large pool of possible molecules with different characteristics. The ligand-bound lipocalin-fold molecule can be used as a target molecule that is recognized by another engi neered protein, i.e. the lipocalin-fold binding interaction partner, in a ligand-dependent manner. That is, in this embodi ment, the lipocalin-fold binding interaction partner is engi- neered to specifically recognize the lipocalin-fold molecule with higher affinity if the ligand is inserted into the hydro- phobic pocket. This other engineered protein can be, but is not limited to, an antibody, an antibody fragment or any other pro tein that can be engineered for antigen binding. Alternatively, the ligand-bound lipocalin-fold molecule, i.e. the lipocalin- fold thereof, may be engineered for binding to another molecule, i.e. the lipocalin-fold binding interaction partner (e.g. a pro tein, carbohydrate, lipid, among others) in a ligand-dependent manner. That is, in this embodiment, in the ligand-bound state the engineered lipocalin-fold molecule is able to bind to the other interaction partner with strongly enhanced affinity. As a consequence, the ligand can be used for regulating the interac tion of the lipocalin-fold molecule with the lipocalin-fold binding interaction partner.

These regulating lipocalin-fold ligands can include clini cally applicable molecules with beneficial pharmacokinetics. Moreover, it is possible to select lipocalin-fold ligands that are orthogonal to each other, thereby even enabling separate regulation of multiple processes in parallel. All this founds firstly on the capacity of this b-barrel structure to inherently accommodate a multitude of different ligands in its calyx-like binding pocket and secondly on its unique tolerance to muta tions, which enables engineering of high affinity and specifici ty to a broad spectrum of non-natural ligands as previously dis closed (e.g. DE 19742706 Al, WO 99/016873 Al, EP 1 017 814 Bl, WO 2012/065978 Al, WO 2016/113203 Al, Skerra, Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 337-350 ; Korndorfer et al . , Proteins 2003;53(1):121-129; Korndorfer et al., J Mol Biol. 2003; 330 (2) : 385-396; Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576 ; Schlehuber et al . , Biophys Chem. 2002 ; 96 (2-3 ) : 213-228 ) . Just these two unique features of the b- barrel structure enable orthogonalization by engineering of the binding pocket for binding the ligands with high affinity and specificity, instead of redesigning the ligands, which signifi cantly reduces the clinical entry barrier and allows for choos ing molecules from a large pool of possible candidates. An addi tional degree of flexibility and specificity may be introduced by lipocalin-fold binding interaction partners that are engi neered for specifically recognizing the b-barrel containing lipocalain-fold molecules in the ligand-bound state. For this purpose, the ligand-loaded b-barrels (i.e. the ligand-loaded lipocalin-fold molecules) may be used as antigens for binder screening from appropriate libraries. Alternatively, the lipocalin-fold can also be engineered as ligand-dependent bind ers. The latter is based on the fact that lipocalin-fold mole cules can not only be engineered for small molecule binding but also for binding of large proteins. This has been meanwhile ex emplified in the form of so-called "anticalins" or "muteins of lipocalin" with a range of protein antigens and has been applied for generating soluble blocking agents and novel tumour binding moieties in chimeric antigen receptors (e.g. DE 19742706 Al, WO 99/016873 Al, EP 1 017 814 Bl, WO 2012/065978 Al, WO 2016/113203 Al, Richter et al . , FEBS Lett. 2014 ; 588 (2 ) : 213-218 ; Schonfeld et al . , Proc Natl Acad Sci U S A. 2009; 106 (20) : 8198-8203; Gebauer et al . , J Mol Biol 2013 ; 425 ( 4 ) : 780-802 ; Barinka et al . , Protein Eng Des Sel. 2016; 29 (3) : 105-115) . Accordingly, there are already numerous examples of lipocalin-fold molecules available in the prior art which may be applied in the LRPPI system according to the present invention, such as the naturally occurring lipocalins and engineered lipocalins ("anticalins", "muteins of lipocalin", etc.) . In contrast to the previous strategies pro vided on the basis of engineering LCN-based binder scaffolds, the present invention provides the engineering of lipocalin-fold based LRPPI systems for regulating PPIs by addition of small molecules .

Accordingly, any molecule comprising the lipocalin-fold as the central structural element may be used or adapted to be used in the LRPPI system according to the present invention. The LRPPI systems according to the present invention can easily be optimised and tuned with respect to affinity of the lipocalin- fold molecules to lipocalin-fold ligands and also with respect to differences in the affinity of lipocalin-fold molecules to lipocalin-fold binding interaction partners in absence and pres ence of lipocalin-fold ligands. In the present invention the "bound" or "unbound" state of the LRPPI system, i.e., a lipocalin-fold molecule "bound" or "unbound" to a lipocalin-fold ligand, is referred to a difference in affinity of the lipocalin-fold molecule to the lipocalin-fold binding interac tion partner of at least ten-fold. However, preferred embodi- merits apply an even more significant affinity difference between the (at least) two states of the lipocalin-fold molecules in ab sence and presence of lipocalin-fold ligands. This is why the difference in affinity is preferably at least 20-fold, especial ly at least 50-fold. The present invention allows differences in affinity to be designed and tuned even further, e.g. at least 100-fold, at least 200-fold, at least 500-fold or at least 1000- fold. This increase in affinity difference may be specifically advantageous in the human therapy environment, especially as a safety measure to exclude unwanted side effects or toxicity.

Although the LRPPI systems according to the present inven tion are based on the advantageous properties of the naturally occurring lipocalin-fold molecules, the LRPPI systems according to the present invention are artificial systems which are de signed to provide a suitable pharmaceutical system. This means that the LRPPI systems according to the present invention cannot have a naturally occurring counterpart (because this could not be used for the purpose intended by the present invention (since the natural systems have to fulfil their naturally intended pur pose) ) . In this context, it is important to note here that in the course of the present invention the surprising observation was made that the LRPPI systems according to the present inven tion are highly flexible regarding the type (i.e. the structure (or fold)) of the lipocalin-fold binding interaction partner. The LRPPI systems described in Example 1 of the present inven tion demonstrate that different types of proteins with structur ally very different binding sites (located on either rigid b- strands or loop regions, respectively) can be used as lipocalin- fold binding interaction partners. Moreover, the proteins used as lipocalin-fold binding interaction partners in Example 1 are not (and are not mutants of) any naturally occurring lipocalin- fold binding interaction partners. Instead, the lipocalin-fold binding interaction partners used in Example 1 are either mutat ed versions of the protein Sso7d (a DNA-binding protein from the archaeon Sulfolobus solfataricus) or mutated versions of the 10 ^th type III domain of human fibronectin (FN3) , demonstrating that structurally distinct proteins (or protein domains) derived from molecules with completely different original function can effi ciently act as lipocalin-fold binding interaction partners ac cording to the present invention. Accordingly, to create LRPPI systems which are more independent from naturally occurring PPI systems and to reduce the risk of disadvantageous cross reactivity with endogenous lipocalin-fold binding interaction partners, the lipocalin-fold binding interaction partners in the LRPPI systems according to the present invention are not (and are preferably also not derived from) naturally occurring lipocalin-fold binding interaction partners. More precisely, the lipocalin-fold binding interaction partner (or any domain of it that mediates binding to the lipocalin-fold molecule) is not and is preferably also not derived from a naturally occurring pro tein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule (in the presence of any lipocalin-fold ligand) , wherein "being derived from" is defined as containing at least one segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any segment of that naturally occurring protein and which has an affinity of <10 mM to any naturally oc curring lipocalin-fold molecule (in the presence of any lipocalin-fold ligand) .

Accordingly, a preferred embodiment of the LRPPI system ac cording to the present invention employs a lipocalin-fold bind ing interaction partner that does not contain a segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any segment of a naturally occurring protein and which has an affinity of <400 nM, preferably <2 mM, especially <10 mM to any naturally occurring lipocalin-fold molecule, especially in the presence of a lipocalin-fold ligand. Preferably, the lipocalin- fold binding interaction partner does not contain a domain of a naturally occurring protein that mediates binding to a naturally occurring lipocalin-fold molecule. This further safeguards lack of cross-reactivity. Preferably, the affinity of the lipocalin- fold binding interaction partner (used in the LRPPI systems ac cording to the present invention) to the lipocalin-fold molecule when bound to the lipocalin-fold ligand or in the second confor mation, respectively, is below 10 mM, preferably below 2 mM, es pecially below 400 nM.

In order to further lowering the risk of immunogenicity in a human patient, the ligand regulated protein-protein interaction system according to the present invention preferably comprises a lipocalin-fold binding interaction partner that is not a homolog of a different species than human of a naturally occurring human lipocalin-fold binding interaction partner. Even more preferred, the lipocalin-fold molecule is not even a homolog of a different species than human of a naturally occurring human lipocalin-fold molecule .

Preferred ligand regulated protein-protein interaction sys tems according to the present invention apply molecules as lipocalin-fold ligands which are useable in a pharmaceutical en vironment, preferably molecules which are suitable and appropri ate to be applied to humans. Accordingly, preferred embodiments of the present invention comprise a lipocalin-fold ligand which is a pharmaceutically active molecule, especially a pharmaceuti cally active molecule with a therapeutic activity in human pa tients. In this connection, molecules are preferred as lipocalin-fold ligands which can be effectively administered orally. This means that such molecules are suitable for oral ad ministration and are taken up by the individual to whom the mol ecule is administered through intestinal absorption. Also mole cules are preferred as lipocalin-fold ligands which can be ef fectively administered intravenously. This means that the mole cule can be administered intravenously without significant side effects to a patient. Specifically preferred lipocalin-fold lig ands are suitable for effective administration in both manners, intravenously and orally, to a human patient. Preferred lipocalin-fold ligands are therefore molecules which are (or have been) registered drugs for human use for which e.g. a valid marketing authorisation is present, e.g. in either the EU or the US, or both.

This artificial character of the LRPPI systems according to the present invention enables the proper regulation of the sys tem (e.g. by the small molecule ligand) also in vivo (as needed to solve the object of the present invention) . In a specifically preferred embodiment, both main system components of the ligand regulated protein-protein interaction system according to the present invention, namely, the lipocalin-fold molecule and the lipocalin-fold binding interaction partner and/or any domain of them that mediates binding to each other with an affinity (in the presence of a lipocalin-fold ligand) of <10 mM, preferably <2 mM, especially <400 nM, are not part of a naturally occurring LRPPI system, i.e. a biological pathway, wherein the physiologi cal function is performed by such a system. Accordingly, in such a preferred embodiment, both the lipocalin-fold molecule and the lipocalin-fold binding interaction partner are therefore not naturally occurring proteins, but mutated or artificially de signed non-natural proteins (i.e. proteins with no native coun terpart existing in nature) .

Moreover, in order to design an LRPPI system that is inde pendent from naturally occurring LRPPI or PPI systems, the LRPPI systems according to the present invention are artificial sys tems in which preferably the lipocalin-fold molecule and the lipocalin-fold binding interaction partner are not derived from naturally occurring molecules that bind to each other with an affinity of <10 mM. This means that (1) if the lipocalin-fold molecule used in a given LRPPI system according to the present invention is engineered, the naturally occurring lipocalin-fold molecule, that it is derived from, does not bind with an affini ty of <10 mM to the lipocalin-fold binding interaction partner used in that LRPPI system; or (2) if the lipocalin-fold binding interaction partner used in a given LRPPI system according to the present invention is engineered, the naturally occurring molecule, that this lipocalin-fold binding interaction partner is derived from, does not bind with an affinity of <10 mM to the lipocalin-fold molecule used in that LRPPI system; or (3) if both the lipocalin-fold molecule and the lipocalin-fold binding interaction partner used in a given LRPPI system according to the present invention are engineered, the naturally occurring molecules, which the lipocalin-fold molecule and the lipocalin- fold binding interaction partner are derived from, respectively, do not bind to each other with an affinity of <10 mM.

Moreover, to avoid any disadvantageous cross-reactivity, the lipocalin-fold binding interaction partner used in a given LRPPI system according to the present invention preferably does not bind with an affinity of <1 mM (neither in the presence nor in the absence of a lipocalin-fold ligand) to any naturally occur ring lipocalin-fold molecule except for (1) the naturally occur ring lipocalin-fold molecule used in that LRPPI system (if a naturally occurring lipocalin-fold molecule is used in that LRPPI system) or the naturally occurring lipocalin-fold molecule that the b-barrel sequence of the engineered lipocalin-fold mol- ecule used in that LRPPI system was derived from and except for (2) the naturally occurring lipocalin-fold molecules which are homologs (i.e. homologous lipocalin-fold molecules from other species) of the naturally occurring lipocalin-fold molecule used in that LRPPI system (if a naturally occurring lipocalin-fold molecule is used in that LRPPI system) or the naturally occur ring lipocalin-fold molecules which are homologs of the natural ly occurring lipocalin-fold molecule that the b-barrel sequence of the engineered lipocalin-fold molecule used in that LRPPI system was derived from.

A highly attractive field for clinical application of LRPPI systems based on lipocalin-fold molecules according to the pre sent invention is for regulating the function of T cells modi fied with chimeric antigen receptors (CARs) instead of using the FKBP-based systems for homo- and heterodimerization for regulat ing the CAR T cell function. Importantly, the function of cur rently applied CARs cannot be regulated in a clinically applica ble manner. Instead, inducing apoptosis in the CAR expressing effector cells is the only clinically applicable safety mecha nism in the CAR field until now (Jones et al . , Front Pharmacol. 2014;5:254). The most advanced strategies for regulating the function of the CAR molecule itself employ the FRB/FKBP-based system in different versions either for LRPPI (WO 2014/127261 Al, WO 2015/017214 Al, EP 3 087 101 Al, US 2017/0081411 Al) or for regulating protein stability (WO 2017/032777 Al) . As men tioned above, however, the FRB/FKBP system is associated with severe problems in any potential clinical application (Sun et al . , Cell Res. 2015; 25 (12) : 1281-1282) . Thus, alternatively, the lipocalin-fold molecule based LRPPI systems according to the present invention are highly attractive for integration into CARs in order to control T cell activation upon CAR mediated target antigen recognition by administering small molecules. In contrast to the rather problematic FRB/FBKP-system the lipocalin-fold based LRPPI systems according to the present in vention can pave the way for broad clinical application of switchable CARs.

The LRPPI system according to the present invention and shown in Figure 1 can be engineered using two strategies in principle: In strategy A, the lipocalin-fold binding interaction partner "c" (which may preferably be fused (+/- flexible linker) to a protein "II") is a binder, which binds to the barrel- containing structure of the lipocalin-fold molecule "a" with higher affinity when lipocalin-fold ligand "b" is present. The lipocalin-fold binding interaction partner "c" can e.g. be gen erated either by immunization of animals with ligand ("b") - loaded lipocalin-fold molecule, or by state of the art protein engineering methods such as phage display, yeast display, bacte rial display, mammalian cell display, ribosome display, mRNA display or covalent DNA display, among others (Sergeeva et al . , Advanced Drug Delivery Reviews 2006;58:1622-1654). In strategy B, the lipocalin-fold molecule "a" itself can be engineered as a binder that after loading with a lipocalin-fold ligand "b" can bind with higher affinity to a chosen lipocalin-fold binding in teraction partner "c" (which again can (according to a preferred embodiment) be fused to protein "II" or itself is a protein or non-protein antigen "II" (e.g. a tumor associated antigen)). The lipocalin-fold molecule "a" and the lipocalin-fold binding in teraction partner "c" can be fused to the N- or C-termini or al so internal sites of proteins "I" and "II" (see Fig. IB) . The lipocalin-fold molecule "a" may be engineered for increasing the affinity to a chosen small molecule lipocalin-fold ligand "b" and/or for decreasing the affinity towards endogenous natural lipocalin-fold ligands. Furthermore, the lipocalin-fold struc ture of lipocalin-fold molecule "a" may also be engineered for preventing or reducing interaction with natural interaction partners .

Preferably, the lipocalin-fold ligand induces a conforma tional change in the lipocalin-fold molecule "a", which facili tates the selection of a ligand-dependent binder ( lipocalin-fold binding interaction partner) "c" (or also the selection of a ligand-dependent lipocalin-fold molecule "a" when in turn used as a binder) . The lipocalin-fold ligand "b" can, but does not need to, interact directly with "c" or protein "II". However, in cases where binding of molecule "b" results only in rigidifica- tion of the structure of the lipocalin-fold molecule "a" with rather limited conformational selection, any direct contribution of molecule "b" for increasing the affinity of the interaction of "a" and "c" is beneficial.

Depending on the application, the LRPPI system can - accord ing to a specific embodiment - be further employed in a strate- gic variant for regulating dimerization of two identical or dif ferent lipocalin-fold molecules (the first one being the lipocalin-fold molecule "a" according to the present invention, the second one being a "lipocalin-fold binding interaction part ner" in the system according to the present invention (although being a "lipocalin-fold molecule", such second lipocalin-fold molecule would serve as the binding partner of the first lipocalin-fold molecule according to the present invention) ) . Accordingly, this specific embodiment uses lipocalin-fold lig ands "b" with two identical or different head groups, respec tively, to form a LRPPI with two lipocalin-fold molecules (dif ferent or the same (i.e. homodimerization or hetereodimeriza- tion) ) , in analogy to existing LRPPI systems (Rutkowska et al . , Angew Chem Int Ed Engl. 2012 ; 51 ( 33 ) : 8166-8176) . One of the two lipocalin-fold molecules serves then (formally) as a "lipocalin- fold binding interaction partner" within the definitions in the system according to the present invention.

As already outlined above, the lipocalin-fold molecule ac cording to the present invention may be any suitable molecule comprising the structural element of a lipocalin-fold . The lipocalin-fold of the lipocalin-fold molecule according to the present invention ("a") can be derived from the b-barrel of any known member of the LCN protein family or from the b-barrel of the iLBP protein family. This also includes LCN- and iLBP- variants, which deviate in the numbers of b-strands from the prototypic architecture as has been reported, e.g., for some LCNs (Papiz et al . , Nature. 1986 ; 324 ( 6095 ) : 383-385 ; Spinelli et al . , Eur J Biochem. 2002 ; 269 ( 10 ) : 2449-2456 ; Sevvana et al . , J Mol Biol. 2010 ; 404 ( 3 ) : 363-371 ) . As LCNs have evolved for extra cellular transport of small molecules and iLBPs have evolved for intracellular transport of many of the same molecules, the two b-barrel variants offer different advantages for application in oxidizing versus reducing environments outside and inside of cells. In principle, the lipocalin-fold of both families are suited for the flexible LRPPI systems according to the present invention as they have very similar structure and all can accom modate a large variety of lipocalin-fold ligands in their bind ing pocket with high affinity. However, lipocalin-fold molecules with known lipocalin-fold ligand induced conformational adaption and variants enabling a direct contribution of the lipocalin- fold ligand "b" in the interaction with the lipocalin-fold bind ing interaction partner "c" are preferred. The organisms of origin of the lipocalin-fold molecule can be selected according to the target organisms, in which the proteins are supposed to be expressed at maximum levels and with correct posttranslation- al modification. In this context LCNs have the advantage that they have evolved already in bacteria. However, the intended primary advantage of the b-barrel based LRPPI systems is their clinical applicability, thus, lipocalin-fold molecules of human protein origin are preferred.

Currently, human LCNs comprise 15 well characterized pro teins and a couple of yet elusive members (Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) . For the well characterized LCNs a large diversity in the size and shape of the binding pockets has been described as well as an accordingly highly diverse spectrum of lipocalin-fold ligands such as, e.g., simple fatty acids, glyco- and phospholipids, all-trans retinol, cholesterol, vanil lin, imatinib, staurosporine, or even large, complex molecules such as bacillibactin and heme. This adaptiveness has been found to particularly depend on the length and the sequence of the four loops at the entry site of the b-barrel (Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ; Skerra, Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 337-350 ; Korndorfer et al . , Proteins 2003;53(1):121-129; Korndorfer et al., J Mol Biol. 2003; 330 (2) : 385-396; Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576 ; Schlehuber et al . , Biophys Chem. 2002 ; 96 (2-3) : 213-228 ; Richter et al . , FEBS Lett. 2014 ; 588 (2 ) : 213-218 ) . Extensive conformational changes upon binding of different lipocalin-fold ligands so far have been re ported for human and bovine retinol binding protein 4 (RBP4), human tear lipocalin (TLC) and human apolipoprotein M (Ap- oM) (Berni et al . , FEBS Lett. 1992 ; 308 ( 1 ) : 43-45 ; Zanotti et al . , J Biol Chem. 1993 ; 268 ( 33 ): 24873-24879 ; Zanotti et al . , J Biol Chem. 1994 ; 269 ( 47 ) : 29613-29620 ; Pattanayek et al . , Protein Sci. 1999 ; 8 ( 10 ) : 2027-2032 ; Motani et al . , J Biol Chem. 2009 ; 284 ( 12 ) : 7673-7680 ; Gasymov et al . , Biochim Biophys Acta. 1998 ; 1386 ( 1 ) : 145-156 ; Breustedt et al . , Acta Crystallogr D Biol Crystallogr. 2009;65(Pt 10) : 1118-1125; Gasymov et al . , Biochem istry. 2012 ; 51 ( 14 ) : 2991-3002 ; Zhang et al . , Sci Rep. 2016; 6 : 30655; Christoffersen et al . , Proc Natl Acad Sci U S A. 2011 ; 108 (23 ) : 9613- 9618 ) . Among them, human TLC is characterized by elevated conformational flexibility and an extraordinary flexible binding behavior compared to other LCNs (Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ; Breustedt et al . , Acta Crystallogr D Biol Crystallogr. 2009;65(Pt 10) : 1118-1125;

Gasymov et al . , Biochemistry. 2012 ; 51 ( 14 ) : 2991-3002 ) . In the case of the particularly well characterized native human RBP4 there are two loop regions, which undergo conformational altera tion upon lipocalin-fold ligand binding and are involved in in teraction with natural interaction partners transthyretin (TTR) (via EF loop, residues 89-101) and the receptor STRA6 (via CD loop, residues 59-68; Redondo et al . , FASEB J. 2008 ; 22 ( 4 ) : 1043- 1054) . Meanwhile, several natural and synthetic retinoid and non-retinoid lipocalin-fold ligands for human RBP4 have been de scribed (i.e., Fenretinide, N-Ethylretinamide, all-trans retin oic acid, retinyl acetate, axerophthene, A1120 (PubChem CID 25138295)), which induce conformational changes resulting in dissociation from TTR (Berni et al . , FEBS Lett. 1992 ; 308 ( 1 ) : 43- 45; Zanotti et al . , J Biol Chem. 1993 ; 268 ( 33 ): 24873-24879 ; Za- notti et al . , J Biol Chem. 1994 ; 269 ( 47 ) : 29613-29620 ; Motani et al . , J Biol Chem. 2009 ; 284 ( 12 ) : 7673-7680 ; Coward et al . , Anal Biochem. 2009 ; 384 (2 ) : 312-320 ; Sharif et al . , Anal Biochem. 2009 ; 392 (2 ) : 162-168 ; ) . Reported crystal structures for human RBP4 illustrate that different lipocalin-fold ligands can induce different conformations in distinct loop regions of the LCN (e.g., protein data bank (PDB) 1RBP, 3FMZ and 2WR6 for retinol, A1120 and linoleic acid, respectively) . Similar effects have been reported for ApoM (PDB 2YG2 and 2WEW for sphingosine-1- phosphate versus myristic acid; Christoffersen et al . , Proc Natl Acad Sci U S A. 2011 ; 108 (23 ) : 9613- 9618 ) .

For selecting appropriate lipocalin-fold molecules according to the present invention, differences with regard to existence of glycosylation sites, free cysteines, disulfide bridges, oli gomerization behavior, ligand spectrum etc., and the necessity for removing more or less characterized interaction sites for preventing interaction with natural protein partners or lipid membranes can be considered (Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) . Among the human LCNs, RBP4, TLC and ApoM (UniProt IDs P02753, P31025, and 095445, respectively) are char acterized by already known lipocalin-fold ligand induced confor- mational adaption and are thus preferred members of the lipocalin family for generating LRPPI systems according to the present invention. One additional preferred lipocalin molecule is the human neutrophil gelatinase-associated lipocalin (NGAL) (UniProt ID P80188), for which a very detailed structural knowledge with respect to interacting amino acid residues and binding pocket engineering has accumulated (Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576 ; Schonfeld et al . , Proc Natl Acad Sci U S A. 2009; 106 (20) : 8198-8203; Barinka et al . , Protein Eng Des Sel. 2016; 29 (3) : 105-115; Gebauer et al . , J Mol Biol 2013 ; 425 ( 4 ) : 780-802 ; Bao et al . , RSC Adv. 2015; 5 (126) : 104363- 104374; Eggenstein et al . , J Struct Biol. 2014 ; 185 (2 ) : 203-214 ) . This is in particular due to the fact that its ligand binding pocket can easily be modified and was the basis of structurally resolved anticalins engineered for high affinity binding against different ligands.

Importantly, b-barrels of LCNs are not only an option for extracellular use but also for intracellular use of the LRPPI systems. This is based on the fact that, although all human LCNs have at least one disulfide bridge in their b-barrel structure, the b-barrel of, e.g., human TLC is functional and sufficiently stable also in the fully reduced state (Gasymov et al . , Biochim Biophys Acta. 2011; 1814 (5) : 671-683) . Other LCNs could be more affected under reducing conditions, however, these proteins can be stabilized by e.g. inserting stabilizing mutations. The lat ter has been exemplified for a zinc-binding mutant of human RBP4, which was stabilized by introducing the five mutations A43L, A55V, A57I, H104W and Q117I (Skerra; Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 337-35049 ; Schmidt, Untersuchungen zur Protein- faltung durch Protein-Design am Retinol-Bindungsprotein . Vol. ISBN 3-89675-314-2. Munchen: Herbert Utz Verlag; 1998).

The LCN-derived b-barrels preferably represent the full- length coding sequences; signal peptides can be replaced; both, the N- and the C-terminal ends of the full length or trimmed b- barrels (i.e. lipocalin molecules) are suited for fusion to pro tein partners (with or without linker sequences) , protein do mains, peptides or single amino acids since the termini of the b-barrels are not involved in ligand binding (Skerra; Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 337-35049) .

Possible sequence modifications can comprise: a) Preferably, the engineering of the binding pocket of the lipocalin-fold molecules for increasing the affinity to a chosen ligand "b" and/or lowering the affinity to other lipocalin-fold ligands by directed evolution including any sort of random muta genesis and subsequent selection or screening processes, such as phage display, yeast display, bacterial display, mammalian cell display, ribosome display, mRNA display or covalent DNA display, among others (Sergeeva et al . , Advanced Drug Delivery Reviews 2006;58:1622-1654). Alternatively, mutations can be based on in silico calculations and subsequently be introduced by site- directed mutagenesis (Whitehead et al . , Methods Enzymol. 2013;523:1-19; Strauch et al . , Proc Natl Acad Sci U S A. 2014 Jan 14 ; 111 (2 ) : 675-80 ) . Such engineering processes can require only limited mutagenesis at specific sites. If small molecules are chosen that initially do not bind the lipopocalin-fold, then more extensive mutagenesis in or nearby the center of the bind ing pocket of the b-barrel may be required for generating mu tants with binding capability (as described in DE 19742706 Al and exemplified for several small molecules in Korndorfer et al . , Proteins 2003; 53 (1) : 121-129; Korndorfer et al . , J Mol Biol. 2003; 330 (2) : 385-396; Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576 ; Schlehuber et al . , Biophys Chem. 2002 ; 96 (2-3 ) : 213-228 ) . In the preferred case, this process is compatible with screening for lipocalin-fold ligand induced con formational changes. Such a screening process is feasible by em ploying proteins that can bind to these lipocalin-fold molecules in the absence of any ligand but dissociate upon lipocalin-fold ligand binding for selecting lipocalin-fold molecule mutants, which have maintained conformational switch behaviour. Such a protein can be the natural protein TTR in the case of RBP4 or other natural binding partners that bind to the respective lipocalin-fold molecule. Alternatively, binders, which have been separately engineered for binding to a chosen lipocalin-fold molecule in the absence of a ligand, may also be used for such a screening process. In a typical screening process, the lipocalin-fold molecule libraries are alternately screened for mutants that are capable for binding to these proteins in the absence of any ligand. Then the capability for lipocalin-fold ligand binding and conformational switching of the lipocalin- fold molecule is concomitantly selected by screening of the lipocalin-fold molecule libraries for non-binding to these pro teins, or binding with lower affinity, in the presence of the lipocalin-fold ligand.

b) Additional mutagenesis (possibly, but not necessarily, after engineering of the binding pocket of the lipocalin-fold molecule for lipocalin-fold ligand binding) with a focus on the loop regions (as described in DE 19742706 Al), if the lipocalin- fold molecule is intended for use as a binder in the LRPPI sys tem (instead of using it as an antigen) ; preferably, the lipocalin-fold molecule is engineered by directed evolution in cluding any sort of random mutagenesis and subsequent selection or screening processes, such as phage display, yeast display, bacterial display, mammalian cell display, ribosome display, mRNA display or covalent DNA display, among others (Sergeeva et al . , Advanced Drug Delivery Reviews 2006; 58:1622-1654). Mainte nance of the capability for small molecule binding of the lipocalin-fold molecules may be warranted by alternating screen ing for antigen binding of the lipocalin-fold molecules in pres ence of the lipocalin-fold ligand and for non-binding (or bind ing with lower affinity) in the absence of the lipocalin-fold ligand .

c) Mutation/deletion/insertion of residues for preventing dimerization or interaction with other proteins or lipid mem branes (e.g., free cysteines, unprocessed signal peptide in Ap- oM, loop CD residues 59-68 in RBP4, etc.; see e.g. Skerra, Bio- chim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 337-350 ; Zhang et al . , Sci Rep. 2016; 6 : 30655; Redondo et al . , FASEB J. 2008 ; 22 ( 4 ) : 1043- 1054) .

d) Mutation/deletion/insertion of residues for preventing post translational protein modification.

e) Engineering the lipocalin-fold molecule for improved sta bility, e.g., under reducing conditions in the cytoplasm as has been demonstrated for antibody fragments (Worn et al . , J Biol Chem. 2000 ; 275 ( 4 ) : 2795-2803 ) . This can be achieved, for example, by rational design of stabilizing mutations or by directed evo lution experiments which select for improved stability (Traxlmayr et al, Biochim Biophys Acta. 2012 ; 1824 ( 4 ) : 542-549) . However, also any other method for stabilization of proteins is possible .

The human intracellular iLBP protein family comprises a group of 6 retinoid binding proteins termed CRBPs and CRABPs, and the group of 10 fatty acid binding proteins (FABPs) . Alt hough iLBPs are intracellular proteins, some of them, e.g., FABP4 may have a function also in the extracellular space (Ho- tamisligil et al . , Nat Rev Endocrinol. 2015; 11 (10) : 592-605) . All iLBPs have the same architecture with 10 anti-parallel b-strands connected by more or less elongated loops with the exception of an intervening helix-turn-helix motif between strands bA and bB (Lakshmi et al . PLoS One. 2015;10(8): e0135507; Zhang et al . ,

PLoS One. 2012;7(5): e36772; Smathers et al . , Hum Genomics.

2011 ; 5 (3) : 170-191 ) . Compared to LCNs their barrel structure is more compact and their ligands are hardly exposed to the solvent due to shielding by the helix-turn-helix motif at the entrance. However, like LCNs iLBPs have adapted to binding of a diverse spectrum of partially shared ligands and they also have in com mon the high tolerance to mutagenesis. The latter includes even the complete deletion of the helix-turn-helix motif at the bar rel entrance and substitution by a simple loop (Curto et al . , Protein Sci. 2009; 18 (4) : 735-746; Ogbay et al . , Protein Sci. 2004 ; 13 ( 5 ) : 1227-1237 ) . Among the iLBPs, FABP1 and FABP6 are characterized by higher backbone flexibility and a larger ligand entrance, resulting in their unique capacity to accommodate bulky molecules and two of each. Binding of the so far tested ligands generally revealed only small conformational changes due to ligand induced rigidification (Yu et al . , Sci Rep. 2016 ; 6 : 34171 ; Sharma et al . , J Biol Chem. 2011 ; 286 ( 36) : 31924- 31928; Cai et al . , Biophys J. 2012 ; 102 ( 11 ) : 2585-2594 ; Franzoni et al . , J Lipid Res. 2010 ; 51 ( 6) : 1332-1343 ; Vaezeslami et al . , J Mol Biol. 2006 ; 363 ( 3 ) : 687-701 ; Gillilan et al . , J Mol Biol. 2007 ; 372 ( 5 ) : 1246-1260 ; Menozzi et al . , J Struct Biol. 2017 ; 197 ( 3 ) : 330-339 ; Long et al . , Biophys J. 2010 ; 98 ( 12 ) : 3054- 3061) . Obviously, these changes are sufficient to control inter action with other proteins and to mediate, e.g., nuclear transport and interaction with nuclear receptors, as well as, e.g., shuttling retinoids within the cell by mediating interac tion with the receptor STRA6 in the cytoplasmic membrane in the case of CRBP-I (Gillilan et al . , J Mol Biol. 2007 ; 372 ( 5 ) : 1246- 1260; Armstrong et al . , J Biol Chem. 2014 ; 289 (21 ): 14941-14954 ; Amber-Vitos et al . , PLoS One. 2015 ; 10 ( 8 ) : eO 132138 ; Berry et al . , Mol Cell Biol. 2012 ; 32(15) :3164-3175; Sessler et al., Mol Cell. 2005; 18 (3) : 343-353; Hofer et al . , J Biol Chem. 2015; 290 (30) : 18438-18453; Furuhashi et al . , Nat Rev Drug Discov. 2008 ; 7 ( 6) : 489-503 ) . Some iLBPs thereby interact via the for mation/selection of a structural nuclear location signal (NLS) within the 2-helix at the entry of the calyx, which is elicited by some but not all of their small molecule ligands (Furuhashi et al . , Nat Rev Drug Discov. 2008 ; 7 ( 6) : 489-503 ) .

Like with LCNs, the human variants of iLBPs are preferred for use in a LRPPI system if applied in humans in vivo. Among them the retinoid binding proteins, in particular the very well characterized CRABP-II (UniProt ID P29373), are preferred due to their retinoid ligand specificity (Zhang et al . , PLoS One. 2012;7(5) : e36772; Franzoni et al . , J Lipid Res. 2010 ; 51 ( 6) : 1332-1343 ; Vaezeslami et al . , J Mol Biol. 2006 ; 363 ( 3 ) : 687-701 ; Menozzi et al . , J Struct Biol. 2017 ; 197 ( 3 ) : 330-339) . Furthermore, FABP2 (P12104), for which the helix-turn-helix substitution was exemplified, and FABP1 (P07148) due to their known low affinity to several clinically approved lipophilic drugs are also attractive iLBP members for engineering the binding pocket for recognition of clinically ap plicable ligands (Smathers et al . , Hum Genomics. 2011 ; 5 (3) : 170- 191; Curto et al . , Protein Sci. 2009 ; 18 ( 4 ) : 735-746 ; Ogbay et al . , Protein Sci. 2004 ; 13 ( 5 ) : 1227-1237 ; Velkov T, Chem Biol. 2007; 14 (4) : 453-465; Velkov T. PPAR Res. 2013 ; 2013 : 938401 ; Ber- inghelli T, PLoS One. 2015 ; 10 ( 7 ) : eO 132096 ; Chuang S, J Med Chem. 2008 ; 51 (13) :3755-3764) .

For integration into the LRPPI system according to the pre sent invention, fusion of the b-barrels of iLBPs to protein partners (with or without linker sequences) , protein domains, peptides or single amino acids is possible via the N- and the C- terminal ends (of full length or trimmed iLBP proteins) . The se quences are preferably modified for preventing unwanted interac tion with their natural protein partners in the cytoplasm by modifying or deleting the respective interacting sequence ele ments (e.g., the helix-turn-helix motif containing the hidden structural NLS in the 2-helix and other interaction sites; Gil- lilan et al . , J Mol Biol. 2007 ; 372 ( 5 ) : 1246-1260 ; Armstrong et al . , J Biol Chem. 2014 ; 289 (21 ) : 14941-14954 ; Amber-Vitos et al . , PLoS One. 2015 ; 10 ( 8 ) : eO 132138 ; Berry et al . , Mol Cell Biol. 2012; 32 (15) :3164-3175; Sessler et al . , Mol Cell. 2005; 18 (3) : 343- 353) . For increasing the affinity to selected lipocalin-fold ligands ("b") and/or for decreasing the affinity to their endog enous small molecule ligands, the iLBPs may be engineered simi larly to LCNs. To facilitate the engineering of lipocalin-fold ligand dependent recognition, it is possible to substitute the helix-turn-helix motif by a loop sequence (as exemplified by Curto et al . and Ogbay et al . (Curto et al . , Protein Sci. 2009 ; 18 ( 4 ) : 735-746 ; Ogbay et al . , Protein Sci. 2004 ; 13 (5) : 1227- 1237) .

The rationale of developing the LRPPI system according to the present invention (i.e. based on the "lipocalin-fold" of the lipocalin-fold molecule) is to maximize the freedom of choice in selecting the lipocalin-fold ligands ("b") . The crucial innova tion for enabling this freedom is based on proteins of the su perfamily of lipocalin proteins that contain the "lipocalin- fold" structure. Their key feature, i.e., the characteristic deep binding pocket with a calyx like shape and high structural flexibility is crucial for the flexibility and reliability of the present invention. As a consequence, these binding pockets can be engineered by relatively few mutations for binding to a broad range of small molecule ligands with diverse structural and biophysical characteristics. This has been exemplified with the proteins BBP and NGAL, which were engineered by substitution of 12 to 17 amino acid residues for high affinity binding to originally non-binding small molecules such as fluorescein, di- goxigenin, digitoxigenin and a diaminepentaacetic acid (DTPA) based chelator (Korndorfer et al . , Proteins 2003; 53 (1) : 121-129; Korndorfer et al . , J Mol Biol. 2003; 330 (2) : 385-396; Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576) . Thus, contrary to exist ing LRPPI systems, the novel system according to the present in vention is defined by employing lipocalin-fold containing pro teins (i.e., lipocalin-fold molecules) and neither by any defin itive list of possible small molecules (i.e., lipocalin-fold ligands) nor by any specific structural and chemical properties of such molecules.

The lipocalin-fold ligand "b" according to the present in vention is a "small molecule", e.g. "small" compared to polypep tides and proteins, such as the lipocalin-fold molecule. Accord ingly, the lipocalin-fold ligand according to the present inven tion has a molecular weight of 1500 Da or less, preferably 1000 Da or less, especially 750 Da or less. It may be as small as, e.g. glycine (75 Da) or even below, provided that it allows a specific binding within the LRPPI system according to the pre sent invention (i.e. binding to the lipocalin-fold molecule) . Accordingly, preferred Mw ranges of the lipocalin-fold ligand of the present invention are 50 to 1500 Da, preferably 75 to 1500 Da, especially 150 to 750 Da. Preferably, the lipocalin-fold ligand can bind in the calyx of the lipocalin-fold molecule formed by the barrel and the loop regions of the lipocalin-fold structure. The lipocalin-fold ligand has a substantial and spe cific affinity to the lipocalin-fold molecule (in at least one conformation of the lipocalin-fold molecule) which may be 1 mM or lower, preferably 100 mM or lower, especially 10 mM or lower. Such affinity can be increased by directed evolution of the binding pocket of the lipocalin-fold molecules. Although up to 30 mutations or more may be applied to a given lipocalin-fold molecule (see, for example, Schonfeld et al . , Proc Natl Acad Sci U S A. 2009; 106 (20) : 8198-8203) , it is usually not necessary to exchange more than 25 or more than 20 to significantly increase the affinity to a given lipocalin-fold ligand (see, for example, Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576 ; Korndorfer et al . , Proteins 2003; 53 (1) : 121-129; Korndorfer et al . , J Mol Biol. 2003; 330 (2) : 385-396; Gebauer et al . , J Mol Biol 2013; 425 (4) : 780- 802) . Methods for generating lipocalin-fold molecules with an improved affinity towards the lipocalin-fold ligand are well available in the art (see above) . Usually, only a few mutations are necessary for a significant increase in affinity towards a given lipocalin-fold ligand, for example only one, two or less, three or less, four or less, five or less, six or less, seven or less, eight or less, nine or less, or ten or less. The capabil ity of a lipocalin-fold ligand "b" to induce a conformational alteration in the lipocalin-fold molecule "a" and/or to posi tively affect the affinity of a lipocalin-fold binding interac tion partner "c" by direct interaction can be predicted to some extent using pharmacophores from existing small molecule ligands with such function. Additionally, this functional capability can be screened for by employing binding proteins, which have a con formation specific binding affinity for the lipocalin-fold mole cule (detailed in Example 2, below, and in Coward et al . Anal Biochem. 2009; 384 (2) : 312-320, Sharif et al . , Anal Biochem. 2009; 392 (2) : 162-168 and Dobri et al . , Invest Ophthalmol Vis Sci. 2013; 54 (1) : 85-95) . In the case of RBP4, e.g., TTR can be used as a naturally occurring conformation specific binding protein. If there is no such protein available, as for example for TLC or ApoM, which have also been reported to undergo lipocalin-fold ligand dependent conformational switching (Gasymov et al . , Bio- chim Biophys Acta. 1998 ; 1386 ( 1 ) : 145-156 ; Breustedt et al . , Acta Crystallogr D Biol Crystallogr. 2009;65(Pt 10) : 1118-1125; Zhang et al . , Sci Rep. 2016; 6 : 30655; Christoffersen et al . , Proc Natl Acad Sci U S A. 2011 ; 108 (23 ) : 9613- 9618 ) , then it is possible to first generate a protein-based binder that interacts with the unloaded lipocalin-fold molecule, i.e. in the absence of lipocalin-fold ligands, by a process of alternating selection for binding to the lipocalin-fold molecule in the absence and non-binding (or binding with reduced affinity) in the presence of lipocalin-fold ligands known to shift conformational states of the lipocalin-fold molecule. This alternating selection strategy ensures that a binder is selected, which specifically recognizes the unloaded (i.e. not loaded with any lipocalin-fold ligand) state of the lipocalin-fold molecule. In Example 1 we exemplified this process of alternating screening in presence and absence of a lipocalin-fold ligand (A1120) in opposite di rection for selecting a lipocalin-fold binding interaction part ner, i.e., a protein that binds RBP4 (i.e. the lipocalin-fold molecule) with increased affinity in presence of the lipocalin- fold ligand A1120.

Typically, the process of selecting lipocalin-fold ligand molecules can start from:

1. any given molecule with or without initial binding affinity

2. any existing compound libraries for high throughput screening for binding affinity and/or function

3. any structural databases of compounds that can be used in virtual screening for binding and/or function

Ad 1: Initial binding affinity is not required in this ap proach, since the binding pocket of the lipocalin-fold molecule can be engineered for binding (see e.g. DE 19742706 Al) . The feasibility of engineering of lipocalin-fold molecules for high affinity binding to arbitrary, initially non-binding, molecules has been exemplified by engineering of BBP and NGAL for binding of fluorescein, digoxigenin and a DTPA-based chelator (Korn- dorfer et al . , Proteins 2003; 53 (1) : 121-129; Korndorfer et al . , J Mol Biol. 2003; 330 (2) : 385-396; Kim et al . , J Am Chem Soc. 2009 ; 131 ( 10 ) : 3565-3576) . Hence, clinically attractive lipocalin- fold ligand "b" candidates can be selected regardless of initial binding affinity, just considering pharmacological properties as, e.g., tolerability, bioavailability, plasma levels, pharma cokinetics, tissue penetrance etc. Such molecules could be, for example, Indicaxanthin and Vulgaxanthin, which are pharmacologi cally well investigated and tolerated and are contained in suf ficiently high amounts in a narrow segment of edible plants. There are, however, numerous examples of clinically approved drugs which have already been demonstrated to bind to certain lipocalin-fold molecules (e.g., human FABP1; Smathers et al . , Hum Genomics. 2011 ; 5 (3) : 170-191 ; Velkov et al . , Chem Biol. 2007; 14 (4) : 453-465; Velkov T., PPAR Res. 2013 ; 2013 : 938401 ; Ber- inghelli et al . , PLoS One. 2015 ; 10 ( 7 ) : eO 132096 ; Chuang at al . , J Med Chem. 2008 ; 51 ( 13 ) : 3755-3764 ) . Of course, such molecules are attractive starting points. The most attractive molecules are non-natural ligands, for which conformational effects have been proven and which have clinical application potential. For RBP4, for example, these ligands are A1120 and numerous derivatives as well as synthetic retinoids such as, e.g., fenretinide, N- Ethylretinamide, all-trans retinoic acid, retinyl acetate and axerophthene (Berni et al . , FEBS Lett. 1992 ; 308 ( 1 ) : 43-45 ; Zanot- ti et al . , J Biol Chem. 1993 ; 268 (33 ) : 24873-24879 ; Zanotti et al . , J Biol Chem. 1994 ; 269 ( 47 ) : 29613-29620 ; Motani et al . , J Bi ol Chem. 2009 ; 284 ( 12 ) : 7673-7680 ; Cioffi et al . , J Med Chem. 2014 ; 57 ( 18 ) : 7731-7757 ; Cioffi et al . , J Med Chem. 2015 ; 58 (15) :5863-5888) .

Ad 2: High throughput screening of molecule libraries for molecules with binding affinity can be performed by various as says (reviewed in Auld et al . , Receptor Binding Assays for HTS and Drug Discovery. In: Sittampalam et al . , eds . Assay Guidance Manual. Bethesda MD: Eli Lilly & Company and the National Center for Advancing Translational Sciences; 2004) . Screening of mole cule libraries was, e.g., employed for identifying A1120 as a high affinity ligand for RBP4 (Motani et al . , J Biol Chem. 2009 ; 284 ( 12 ) : 7673-7680 ) . A consequence of this strategy is that the identified lipocalin-fold ligands display already some bind ing affinity and thus can be further assayed for mediating lig- and-dependent binding or dissociation of a binder "c" without requirement of prior binding pocket engineering of the lipocalin-fold molecule. Of note, there are possibilities that enable high throughput screening for binding and function, like inducing conformational switching, in a single step. In the case of RBP4 this could be done, for example, by FRET based assays for detecting the conformation dependent interaction with TTR (Coward et al . , Anal Biochem. 2009 ; 384 (2 ) : 312-320 ; Sharif et al . , Anal Biochem. 2009 ; 392 (2 ) : 162-168 ; Dobri et al . , Invest Ophthalmol Vis Sci. 2013; 54 (1) : 85-95) ) . If desired, the affinity of the respective lipocalin-fold ligand to the lipocalin-fold molecule can be further increased by engineering (i.e. mutagene sis) of the binding pocket of the lipocalin-fold molecule.

Ad 3: Virtual screening of compound databases is attractive when there are known small molecule ligands, which mediate con formational alterations in a lipocalin-fold molecule. Then a modeled pharmacophore of the small molecule can be used to screen for molecules, which might similarly interact with the lipocalin-fold molecule and thereby show similar function (see Qing et al . , Journal of Receptor, Ligand and Channel Research 2014;7:81-92 for a review on virtual screening using pharmaco phore modeling) . In the optimal case, the crystal structure of the lipocalin-fold molecule charged with the modeled small mole cule ligand is also known, since this helps to improve the accu racy of the prediction algorithms. There are several examples of known structures for lipocalin-fold molecules bound to a small molecule ligand (Berni et al . , FEBS Lett. 1992 ; 308 ( 1 ) : 43-45 ; Za- notti et al . , J Biol Chem. 1993 ; 268 ( 33 ): 24873-24879 ; Zanotti et al . , J Biol Chem. 1994 ; 269 (47 ): 29613-29620 ; Pattanayek et al . , Protein Sci. 1999 ; 8 ( 10 ) : 2027-2032 ; Motani et al . , J Biol Chem. 2009 ; 284 ( 12 ) : 7673-7680 ; Gasymov et al . , Biochim Biophys Acta. 1998 ; 1386 ( 1 ) : 145-156 ; Breustedt et al . , Acta Crystallogr D Biol Crystallogr. 2009;65(Pt 10) : 1118-1125; Gasymov et al . , Biochem istry. 2012 ; 51 ( 14 ) : 2991-3002 ; Zhang et al . , Sci Rep. 2016; 6 : 30655; Christoffersen et al . , Proc Natl Acad Sci U S A. 2011 ; 108 (23) : 9613-9618 ; Coward et al . , Anal Biochem. 2009; 384 (2) : 312-320) . This strategy is exemplified in the exam ple section below with RBP4 charged with A1120, for which a pharmacophore model was generated in order to identify and pri oritize molecules for functional testing and for engineering of the binding pocket of RBP4 (described in Example 2, below) . Im portantly, the pharmacophore approach does not deliver a final complete list of potentially conformation inducing molecules due to limitations of pharmacophore methods (reviewed in Qing et al . , Journal of Receptor, Ligand and Channel Research 2014; 7:81— 92) and due to incomplete and not updated databases etc. Moreo ver, there are other molecules also known to alter the confor mation of RBP4, e.g., fenretinide, for which a pharmacophore based approach would lead to identification of yet another set of molecules in the chemical space. In fact, in preferred embod iments, in silico filtering of millions of molecules for func tion together with directed evolution of the highly flexible binding pocket of the lipocalin-fold structure in the lipocalin- fold molecules according to the present invention is a perfect combination and a crucial strength of the novel LRPPI system. Of note, virtual screening also enables identifying functional mol ecules, which do not have sufficient initial binding affinity, which is not a problem, since affinity can easily be generated by engineering of the lipocalin-fold molecule. In this case, lipocalin-fold molecule mutants are selected by use of a confor mation specific binding protein that binds to the lipocalin-fold molecule mutatants in the absence but not in the presence of a known conformation inducing ligand. Thus, the dissociation of this conformation specific binding protein enables the identifi cation of lipocalin-fold ligand candidates that bind to the lipocalin-fold molecule and that at the same time induce confor mational alterations (detailed in Example 2, below) .

Principally, the lipocalin-fold binding interaction partner "c" can be provided or engineered based on any available molecu lar binder scaffold including antibodies, antibody fragments [e.g. single-chain variable fragments (scFv) , antigen binding fragments (Fabs) , nanobodies, among others] and non-antibody based scaffolds such as affibodies, a further (or the same) lipocalin-fold molecule, preferably LCNs, especially anticalins; avimers, DARPins, fynomers, Kunitz domains, knottins, monobod ies, binders based on Sso7d, reduced charge Sso7d (rcSso7d) or Sac7d, among many others (Simeon et al . , Protein Cell. 2017; Gilbreth et al . , Curr Opin Struct Biol. 2012 ; 22 ( 4 ) : 413-420 ; Koide et al . , ACS Chem Biol. 2009 ; 4 (5 ) : 325-334 ; Traxlmayr et al . , J Biol Chem. 2016; 291 (43) : 22496-22508) . Meanwhile, many more non-antibody binding proteins have been reported (Pliick- thun, Alternative Scaffolds: Expanding the options of antibod ies. In: Little M, ed. New York: Cambridge University Press; 2009:244-271; Chapman et al . , Cell Chem Biol. 2016; 23 (5) : 543- 553; Binz et al . , Nat Biotechnol. 2005 ; 23 ( 10 ) : 1257-1268 ; Vazquez-Lombardi et al . , Drug Discov Today. 2015; 20 (10) : 1271- 1283) , and, in fact, synthetic library design and selection can be applied to any protein which then can potentially serve as lipocalin-fold binding interaction partner "c", too (Pliickthun, Alternative Scaffolds: Expanding the options of antibodies. In: Little M, ed. New York: Cambridge University Press; 2009:244- 271) . With the aim of clinical applicability of the LRPPI sys tems according to the present invention, lipocalin-fold binding interaction partners "c" are preferably derived from small human single protein domains (e.g., fibronectin type III domain (FN3) based Monobodies) with a minimized number of mutated amino acids for keeping immunogenicity as low as possible. It is of course also possible to attach such further molecules and scaffolds to the lipocalin-fold molecule (or to a lipocalin-fold binding in teraction partner) .

Accordingly, the lipocalin-fold molecule according to the present invention may be any protein that contains the structur al motif of a lipocalin-fold to which (or in which) the lipocalin-fold ligand binds and which enables binding of the lipocalin-fold molecule to the lipocalin-fold binding interac tion partner. Usually, the lipocalin-fold molecule according to the present invention is adapted to the specific needs for the lipocalin-fold molecule "a"/lipocalin-fold ligand "b'Vlipocalin- fold binding interaction partner "c" triangle by providing a "starting" lipocalin-fold molecule which may already have a cer tain affinity to a given lipocalin-fold ligand and/or for which clinically attractive ligands with functional activity could be identified by virtual screening. This starting lipocalin-fold molecule may then be engineered by one or more amino acid ex changes, insertions and/or deletions to optimize lipocalin-fold ligand binding, and in the case of strategy B of the lipocalin- fold based LRPPI system also for binding to a lipocalin-fold binding interaction partner "c". In strategy A the lipocalin- fold binding interaction partner "c" is generated by engineering of a protein (that is originally not able to bind to any natu- rally occurring lipocalin-fold molecule with an affinity of < 10 mM in the presence of any lipocalin-fold ligand) for binding to the (starting and/or engineered) lipocalin-fold molecule (bound to the lipocalin-fold ligand) . Such sequence optimization of a protein for generating a lipocalin-fold ligand dependent lipocalin-fold binding interaction partner is shown in the exam ple section of the present invention and well available for a person skilled in the art with the disclosure contained herein. In fact, it is established in the art of protein engineering that protein scaffolds (especially well-established scaffolds, such as antibodies and non-antibody-based scaffolds, especially lipocalins) can be adapted so that virtually any biological mol ecules (especially proteins) can be bound (see e.g. (for non antibody scaffolds) Vasquez-Lombardi et al . , Drug Discov. Today 20 (2015), 1271-1283; Pluckthun, Alternative Scaffolds: Expand ing the Options of Antibodies. In: Recombinant Antibodies for Immunotherapy, Melvyn Little, Cambridge University Press, New York (2009), pp . 244-271).

The (starting) lipocalin-fold molecule therefore preferably contains the structural lipocalin-fold of a naturally occurring protein of the lipocalin-fold superfamily (for clinical applica tions in human patients: preferably a naturally occurring human lipocalin-fold molecule) or a known variant thereof (e.g. an "anticalin" or "mutein of lipocalin" and the like; a vast number of such variants have already been disclosed in the prior art) . The lipocalin-fold molecule to be used in the LRPPI system ac cording to the present invention may then be adapted to the specificities of the lipocalin-fold ligand/lipocalin-fold bind ing interaction partner by the introduction of modifications, as disclosed above (e.g. a) to e) ) , by engineering (i.e. amino acid changes, insertions and/or deletions) of the binding pocket, amino acid positions in the b-barrel structure, and/or amino ac id positions in the regions adjoining the b-strands of the b- barrel structure, mutation/deletion/insertion of residues for preventing dimerization or interaction with other molecules or for preventing post-translational protein modification and/or engineering of the lipocalin-fold for improved stability.

As already stated above, the lipocalin-fold architecture is very robust and has high sequence flexibility, as demonstrated by the fact that it has not been possible to define any sequence motif that is common to all naturally occurring members of the lipocalin superfamily, including LCNs and iLBPs (Lakshmi et al . PLoS One. 2015;10(8): e0135507; Flower et al . , Biochim Biophys Acta. 2000 ; 1482 ( 1-2 ) : 9-24 ) . Apart from the generally high se quence variation in naturally occurring members of the lipocalin superfamily, the regions adjoining the structurally conserved b- strands not only show high variation in sequence, but also in structure (Flower et al . , Biochim Biophys Acta. 2000; 1482(1- 2):9-24; Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) . Given the extensive sequence variation among naturally occurring lipocalin superfamily members and/or structural variation in the regions adjoining the b-strands, it is not surprising that in the past lipocalin molecules have been successfully engineered by mutating more than 30 amino acid positions, corresponding to more than 15% of all amino acid positions in the respective pro tein, without any major structural changes in the structurally conserved b-strands of the characteristic b-barrel structure (Schonfeld et al . , Proc Natl Acad Sci U S A. 2009; 106 (20) : 8198- 8203) . Thus, it is well-established in the field that lipocalin- fold molecules can be extensively mutated without impairing the overall b-barrel fold (i.e. particularly the loop regions but also the b-strands of the b-barrel can be designed by exchang ing, inserting and/or deleting a considerable number of amino acids) . Therefore, in the present invention, which describes the usage of the lipocalin-fold structure in LRPPI systems, a lipocalin-fold molecule is defined as any naturally occurring molecule classified into the lipocalin superfamily in the SCOP database (version 1.75), or a mutant thereof. However, it is preferred to exchange only a limited number of amino acids. The LRPPI system according to the present invention therefore pref erably comprises (1) a lipocalin-fold molecule being identical to a naturally occurring member of the lipocalin superfamily or (2) an already existing variant thereof (e.g. an already exist ing anticalin, lipocalin mutein, iLBP with complete deletion of the helix-turn-helix motif at the barrel entrance etc.) or (3) a derivative of (1) or (2) with at least 70%, preferably at least 80%, especially at least 90% sequence identity to its counter part that it derives from.

According to a preferred embodiment, the lipocalin-fold mol ecule is a derivative of a naturally occurring or otherwise dis- closed (by its amino acid sequence) lipocalin-fold molecule with at least 70%, preferably at least 80%, especially at least 90% sequence identity in the b-barrel structure, whereby this b- barrel structure is defined as the regions which correspond structurally to amino acid positions 21-30, 41-47, 52-58, 71-78, 85-88, 102-109, 114-120 and 132-138 in human RBP4 (according to the amino acid residue numbering scheme in the PDB entry 1RBP) ; or to the amino acid positions 14-23, 37-43, 48-54, 62-69, 76- 79, 84-91, 96-102 and 111-117 in human tear lipocalin (TLC; as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) ; or to the amino acid positions 44-53, 69-75, 81-87, 96-103, 110- 113, 119-126, 131-137 and 142-148 in human apolipoprotein M (Ap- oM; as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) ; or to the amino acid positions 5-12, 41-45, 50-54, 61-65, 71-73, 81-87, 93-96, 108-112, 119-124 and 129-135 in human cellular retinoic acid binding protein II (CRABPII; ac cording to the amino acid residue numbering scheme in PDB entry 2FS6) ; or to the amino acid positions 5-12, 39-43, 48-52, 59-63, 69-71, 79-85, 91-94, 99-103, 109-114 and 119-125 in human fatty acid binding protein 1 (FABP1; according to the amino acid resi due numbering scheme in PDB entry 2F73) . Preferred embodi ments of the lipocalin-fold molecule according to the present invention also include lipocalin-fold molecules which comprise 1-30 amino acid exchanges and/or 1-50 amino acid deletions and/or 1-50 amino acid insertions of (1), (2) or (3) which (at least) contain the lipocalin-fold and are able to bind to the lipocalin-fold ligand and the lipocalin-fold binding interaction partner. Accordingly, the ligand regulated protein-protein in teraction system preferably comprises as lipocalin-fold molecule a molecule identical with a naturally occurring iLBP (intracel lular lipid binding protein) , a naturally occurring lipocalin or an anticalin, and derivatives of any of these molecules with 1- 30 amino acid exchanges and/or 1-50 amino acid deletions and/or 1-50 amino acid insertions.

Alternatively, the LRPPI system according to the present in vention may also comprise a (4) fragment of (1) or (2) or (3) with a length of at least 80, preferably at least 100, especial ly at least 120, amino acids in the case of LCNs [or with a length of of at least 80, preferably at least 85, especially at least 90, amino acids in the case of iLBPs (which are generally smaller proteins than LCNs) ] covering at least the structurally conserved b-barrel structure of the lipocalin-fold . This struc turally conserved b-barrel structure comprises or consists of amino acid positions which correspond structurally to the amino acid positions 21-30, 41-47, 52-58, 71-78, 85-88, 102-109, 114- 120 and 132-138 in human RBP4 (according to the amino acid resi due numbering scheme in the PDB entry 1RBP) ; or to the amino ac id positions 14-23, 37-43, 48-54, 62-69, 76-79, 84-91, 96-102 and 111-117 in human tear lipocalin (TLC; as defined by Schief- ner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) ; or to the amino acid positions 44-53, 69-75, 81-87, 96-103, 110-113, 119-126, 131-137 and 142-148 in human apolipoprotein M (ApoM; as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) ; or to the amino acid positions 5-12, 41-45, 50-54, 61-65, 71-73, 81- 87, 93-96, 108-112, 119-124 and 129-135 in human cellular retin oic acid binding protein II (CRABPII; according to the amino ac id residue numbering scheme in PDB entry 2FS6) ; or to the amino acid positions 5-12, 39-43, 48-52, 59-63, 69-71, 79-85, 91-94, 99-103, 109-114 and 119-125 in human fatty acid binding protein 1 (FABP1; according to the amino acid residue numbering scheme in PDB entry 2F73) .

Further preferred lipocalin-fold molecules in the LRPPI sys tem according to the present invention are derivatives of a nat urally occurring lipocalin or iLBP with up to 15, up to 30, or up to 50 amino acid deletions and/or up to 15, up to 30, or up to 50 amino acid insertions outside of the structurally con served b-barrel structure. Outside the structurally conserved b- barrel structure these molecules are very flexible to amino acid changes and therefore combinations of mutations, deletions and insertions are possible in these regions which correspond struc turally to amino acid residues 1-20, 31-40, 48-51, 59-70, 79-84, 89-101, 110-113, 121-131 and 139-183 in human RBP4 (according to the amino acid residue numbering scheme in the PDB entry 1RBP) , which define the regions adjoining the structurally conserved b- strands in human RBP4; or corresponding structurally to the re gions of amino acid residues 1-13, 24-36, 44-47, 55-61, 70-75, 80-83, 92-95, 103-110 and 118-158 in human TLC (according to the amino acid resi-due numbering scheme in Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 ) , which define the regions adjoin ing the structurally conserved b-strands in human TLC; or corre- sponding structurally to the regions of amino acid residues 1- 43, 54-68, 76-80, 88-95, 104-109, 114-118, 127-130, 138-141 and 149-188 in human ApoM (according to the amino acid residue num bering scheme in Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985), which define the regions adjoining the structurally con served b-strands in human ApoM; or corresponding structurally to the regions of amino acid residues 1-4, 13-40, 46-49, 55-60, 66- 70, 74-80, 88-92, 97-107, 113-118, 125-128 and 136-137 in human CRABPII (according to the amino acid residue numbering scheme in PDB entry 2FS6) , which define the regions adjoining the struc turally conserved b-strands in human CRABPII; or corresponding structurally to the regions of amino acid residues 1-4, 13-38, 44-47, 53-58, 64-68, 72-78, 86-90, 95-98, 104-108, 115-118 and 126-127 in human FABP1 (according to the amino acid residue num bering scheme in PDB entry 2F73) , which define the regions ad joining the structurally conserved b-strands in human FABP1.

In a preferred embodiment, the LRPPI system according to the present invention comprises as lipocalin-fold molecule a deriva tive of a naturally occurring member of the lipocalin superfami ly with at least one, two, three, four, five, six, seven, eight, nine, ten, eleven, twelve, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30 amino acid exchanges.

According to a further preferred embodiment, the lipocalin- fold molecule used in the LRPPI system according to the present invention is a lipocalin, i.e., a protein containing an eight- stranded up-and-down b-barrel arranged in a +1 topology, fol lowed by an -helix after the C-terminal end of the eighth b- strand, or a derivative of a lipocalin with 1-30 amino acid ex changes and/or 1-50 amino acid deletions and/or 1-50 amino acid insertions .

In designing the lipocalin-fold molecule according to the present invention, one parameter which may be significantly im proved in the derivatization/mutation process from a starting lipocalin-fold molecule to a molecule actually optimized for a given LRPPI system, is affinity to the lipocalin-fold ligand "b". Additionally or alternatively, in the case in which the lipocalin-fold molecule itself is engineered as a binder instead of the lipocalin-fold binding interaction partner, the parameter to improve is increasing the affinity difference between the two statuses of the lipocalin-fold molecule (bound or not bound to the lipocalin-fold ligand) for binding to the lipocalin-fold binding interaction partner "c". Selection of lipocalin-fold molecule/lipocalin-fold ligand pairs that are preferentially characterized by structural differences between the bound and unbound statuses (preferably resulting from conformational al terations) is the basis for maximum affinity differences of the two lipocalin-fold molecule statuses for a lipocalin-fold bind ing interaction partner "c". In the case, in which the lipocalin-fold molecule is used as an antigen (i.e. strategy A of the lipocalin-fold based LRPPI system) , the lipocalin-fold binding interaction partner "c" can be subsequently engineered for maximum selectivity between the two statuses of the lipocalin-fold molecule, i.e., maximum affinity difference. The engineering of the lipocalin-fold binding interaction partner and/or the lipocalin-fold molecule leads to LRPPI systems where in the affinity of the lipocalin-fold binding interaction part ner to the lipocalin-fold molecule in the lipocalin-fold ligand- bound state is at least 10-fold higher, preferably at least 20- fold higher, especially at least 50-fold higher than the affini ty of the lipocalin-fold binding interaction partner to the lipocalin-fold molecule in the unbound state (i.e. in the ab sence of the lipocalin-fold ligand) . In the course of the pre sent invention, the human lipocalin RBP4, which is known to un dergo conformational alteration upon binding of some ligands, has been investigated in more detail to prove the concept of the present invention in principle. Similarly to RBP4, the human lipocalins TLC and ApoM are also known to undergo significant ligand induced conformational alterations and it is thus pre ferred to use one of these three lipocalins as lipocalin-fold molecule or as starting lipocalin-fold molecule for the genera tion of a LRPPI system according to the present invention.

A preferred embodiment of the LRPPI system according to the present invention therefore applies a lipocalin-fold molecule that has a sequence identity with the native versions of human RBP4, TLC or ApoM of at least 70%, preferably at least 80%, es pecially at least 90%.

According to a preferred embodiment, the lipocalin-fold mol ecule according to the present invention has a sequence identity with human RBP4 of at least 70%, preferably of at least 80%, es pecially of at least 90% in the structurally conserved b-barrel structure including the regions of amino acid residues 21-30, 41-47, 52-58, 71-78, 85-88, 102-109, 114-120 and 132-138 accord ing to the amino acid residue numbering scheme in the PDB entry 1RBP, or a fragment thereof with at least 80, preferably at least 100, especially at least 120 amino acid residues and com prising the regions corresponding to amino acid residues 21-30, 41-47, 52-58, 71-78, 85-88, 102-109, 114-120 and 132-138 of hu man RBP4 (according to the amino acid residue numbering scheme in the PDB entry 1RBP) ; or (2) the lipocalin-fold molecule has a sequence identity with human tear lipocalin (TLC) of at least 70%, preferably of at least 80%, especially of at least 90% in the structurally conserved b-barrel structure including the re gions of amino acid residues 14-23, 37-43, 48-54, 62-69, 76-79, 84-91, 96-102 and 111-117 in human TLC as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 , or a fragment thereof with at least 80, preferably at least 100, especially at least 120 amino acid residues and comprising the regions corresponding to amino acid residues 14-23, 37-43, 48-54, 62-69, 76-79, 84-91, 96-102 and 111-117 of human TLC; or (3) the lipocalin-fold mole cule has a sequence identity with human apolipoprotein M (ApoM) of at least 70%, preferably of at least 80%, especially of at least 90% in the structurally conserved b-barrel structure in cluding the regions of amino acid residues 44-53, 69-75, 81-87, 96-103, 110-113, 119-126, 131-137 and 142-148 in human ApoM as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 , or a fragment thereof with at least 80, preferably at least 100, especially at least 120 amino acid residues and comprising the regions corresponding to amino acid residues 44-53, 69-75, 81- 87, 96-103, 110-113, 119-126, 131-137 and 142-148 of human ApoM.

In a preferred embodiment, the lipocalin-fold molecule of the LRPPI system according to the present invention has a se quence identity with human RBP4 of at least 95%, or a sequence identity with human tear lipocalin (TLC) of at least 95%, or a sequence identity with human apolipoprotein M (ApoM) of at least 95%.

The LRPPI system according to the present invention consists of the central lipocalin-fold molecule "a"/lipocalin-fold ligand "b"/lipocalin-fold binding interaction partner "c" triangle (see Fig. 1A) . However, this triangle can be further functionalized by linking further moieties to the lipocalin-fold molecule "a" and/or the lipocalin-fold binding interaction partner "c" (see e.g. Fig. IB) . The LRPPI triangle according to the present in vention can therefore be expanded into various architectures, e.g. as CARs (as disclosed below) . Again, it may be emphasised that the "triangles" according to the present invention differ from any naturally occurring (physiological) triangle, because the lipocalin-fold binding interaction partner (or any domain of it that mediates binding to the lipocalin-fold molecule) is not a naturally occurring lipocalin-fold binding interaction partner (i.e. a protein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule in the presence of any lipocalin-fold ligand) . Preferably, the lipocalin-fold binding interaction partner is also not derived from a naturally occur ring lipocalin-fold binding interaction partner. "Being derived from" is defined as containing at least one segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any seg ment of that naturally occurring protein and which has an affin ity of <10 mM to any naturally occurring lipocalin-fold molecule (in the presence of any lipocalin-fold ligand) . Preferably both the lipocalin-fold molecule and the lipocalin-fold binding in teraction partner are not naturally occurring molecules but ar tificially designed molecules (e.g. designed by recombinant technology and/or directed evolution) . In a preferred embodi ment, the lipocalin-fold binding interaction partner is or com prises a lipocalin-fold molecule and/or the lipocalin-fold bind ing interaction partner comprises an antigen, a cell surface re ceptor, an antibody, etc.

For the LRPPI system according to the present invention, it is also preferred that the affinity of the lipocalin-fold ligand to the lipocalin-fold molecule is in a range that allows binding of the lipocalin-fold ligand to the lipocalin-fold molecule at lipocalin-fold ligand concentrations that can be achieved in the physiological environment in the human body. Accordingly, it is preferred that the lipocalin-fold ligand has an affinity to the lipocalin-fold molecule of below 1 mM, preferably of below 100 mM, especially of below 10 mM. This affinity between the lipocalin-fold ligand and the lipocalin-fold molecule is defined as a K _d (dissociation constant) value and preferably determined by isothermal titration calorimetry (ITC) using an automated Mi- croCal PEAQ-ITC instrument (Malvern Instruments) .

The LRPPI system according to the present invention general ly relies on a substantial difference in the affinities of the lipocalin-fold molecule "a" to the lipocalin-fold binding inter action partner "c" depending on whether the lipocalin-fold lig and "b" is bound or not. It is also preferred that this affinity window (i.e. the affinities of the lipocalin-fold binding inter action partner to the lipocalin-fold molecule bound or not bound to the lipocalin-fold ligand, respectively) is present in a rea sonable affinity range which allows for regulation of this LRPPI system under physiological conditions. Therefore, it is pre ferred that the affinity of the lipocalin-fold binding interac tion partner to the lipocalin-fold molecule in the ligand-bound state is below 10 mM, preferably below 2 mM, especially below 400 nM. This affinity between the lipocalin-fold binding inter action partner and the lipocalin-fold molecule in the ligand- bound state is defined as a K _d value and preferably determined by surface plasmon resonance (SPR) using a BiacoreT200 instrument (GE healthcare) .

For the LRPPI system according to the present invention, the affinity difference of the lipocalin-fold molecule binding to the lipocalin-fold binding interaction partner between the states bound/unbound to the lipocalin-fold ligand should be as high as possible. This can be achieved by directed evolution of the interaction partner and screening for suitable affinity dif ferences (for increased affinity differences) . For being appli cable in an in vivo environment, the lipocalin-fold molecule has to have a sufficiently high affinity to the lipocalin-fold lig and. This enables functioning of the LRPPI system according to the present invention also in a patient in vivo. Here, it has also to be considered that drug concentrations in plasma (e.g. for the lipocalin-fold ligand) is usually a few mM. Accordingly, the affinity of the lipocalin-fold ligand to the lipocalin-fold molecule has to be sufficiently high that such plasma concentra tions can achieve appropriate binding of the ligand to the lipocalin-fold molecule. Moreover, the LRPPI system according to the present invention should be designed to minimize relevant binding of other substances to the lipocalin-fold molecules (which could elicit adverse reactions or hinder effectiveness of in vivo functioning of the system (e.g. by competing with bind- ing to the lipocalin-fold molecule) ) .

The lipocalin-fold based LRPPI system enables the highest flexibility with respect to choosing the lipocalin-fold ligand used for controlling PPI . Whereas certain embodiments of the present invention work with substances already known to have a certain affinity to (certain) lipocalin-fold molecule (s), also the lipocalin-fold molecule may be designed to bind a small mol ecule of interest, thereby enabling the small molecule to become a lipocalin-fold ligand by such engineering of a lipocalin-fold molecule. Accordingly, any interesting small molecule, interest ing for pharmaceutical purposes and thus preferred to be used as a lipocalin-fold ligand "b" according to the present invention (i.e. a substance enabling the binding of the lipocalin-fold molecule to the lipocalin-fold binding interaction partner) can be included in a LRPPI system according to the present inven tion. According to a preferred embodiment of the present inven tion, the lipocalin-fold ligand is fenretinide (PubChem CID: 5288209), N-Ethylretinamide (PubChem CID: 5288173), all-trans retinoic acid (PubChem CID: 444795), axerophthene (PubChem CID: 5287722), A1120 (PubChem CID 25138295), derivatives of A1120 (Cioffi et al . , J Med Chem. 2014 ; 57 ( 18 ) : 7731-7757 ; Cioffi et al . , J Med Chem. 2015; 58 (15) : 5863-5888) ) , 1 , 4-butanediol (Pub Chem CID: 8064), sphingosine-l-phosphate (Pubchem CID: 5283560), tetradecanoic acid (Pubchem CID: 11005) , indicaxanthin (Pubchem CID: 6096870 and 12310796), vulgaxanthin I (Pubchem CID: 5281217), Montelukast (Pubchem CID: 5281040), Cyclandelate (Pub chem CID: 2893), Oxolamine (Pubchem CID: 13738), Mazaticol (Pub- chemCID: 4019), Butoctamid (Pubchem CID: 65780), Tonabersat (Pubchem CID: 6918324), Novazin (Pubchem CID: 65734), Diphenidol (Pubchem CID: 3055), Neobornyval, Erlotinib (Pubchem CID: 92131336), Tanespimycin (Pubchem CID: 6505803), LMI070 (Pubchem CID: 85471316), Alloclamide (Pubchem CID: 71837), Diacetolol (Pubchem CID: 50894), Acotiamide (Pubchem CID: 5282338), Acoziborole (Pubchem CID: 44178354), Acumapimod (Pubchem CID: 11338127), Apalutamide (Pubchem CID: 24872560), ASP3026 (Pubchem CID: 25134326), AZD1480 (Pubchem CID: 16659841), BIIB021 (Pubchem CID: 16736529), Branaplam (Pubchem CID: 89971189), Brequinar (Pubchem CID: 57030), Chlorproguanil (Pubchem CID: 9571037), Clindamycin (Pubchem CID: 446598), Emricasan (Pubchem CID: 12000240), Enasidenib (Pubchem CID: 89683805), Enolicam (Pubchem CID: 54679203), Flurazepam (Pubchem CID: 3393), ILX-295501 (Pubchem CID: 127737), Indibulin (Pubchem CID: 2929), Metoclopramide (Pubchem CID: 12598248), Mevastatin (Pubchem CID: 64715), MGGBYMDAPCCKCT-UHFFFAOYSA-N (Pubchem CID: 25134326), MK0686 (Pubchem CID: 16102897), Navarixin (Pubchem CID: 71587743), Nefazodone hydrochloride (Pubchem CID: 54911), Pantoprazole (Pubchem CID: 4679), Pavinetant (Pubchem CID: 23649245), Proxazole (Pubchem CID: 8590) , Siccanin (Pubchem CID: 71902), Sulfaguanole (Pubchem CID: 9571041), Sunitinib (Pubchem CID: 5329102), Suvorexant (Pubchem CID: 24965990), Tiapride (Pubchem CID: 5467), Tonabersat (Pubchem CID: 6918324), VNBRGSXVFBYQNN-UHFFFAOYSA-N (Pubchem CID: 24794418), YUHNXU- AATAMVKD-PZJWPPBQSA-N (Pubchem CID: 44548240), Ulimorelin (Pub chem CID: 11526696), Xipamide (Pubchem CID: 26618), Tropesin (Pubchem CID: 47530), Triclabendazole (Pubchem CID: 50248), Triclabendazole sulfoxide (Pubchem CID: 127657), Triclabendazole sulfone (Pubchem CID: 10340439) and Trametinib (Pubchem CID: 11707110) etc. Preferred substances are also provided in Table 1, below.

An unmet need in the field of cellular immunotherapy is the reversible control of CAR function by clinically applicable small molecules. Thus, in a preferred application the LRPPI sys tem of the present invention is integrated into a CAR which is expressed in T cells or other effector cells such as, e.g., NK cells. The preferred version of the LRPPI system for controlling CAR function is where the lipocalin-fold ligand ("b") -loaded lipocalin-fold molecule "a" is used as antigen for a binder "c" (see schematic representation in Fig. 2 A and B) . In this ver sion, the CAR construct that mediates binding to the target cell is separated from the CAR construct that mediates signal trans duction, whereby the CAR construct that mediates binding to the target cell can be secreted by the effector cell or can be ad ministered/added exogeneously as soluble protein (LRPPI-CAR strategy A) or can be membrane-anchored (LRPPI-CAR strategy B) . If the target cell-binding CAR construct is membrane-anchored, it may also contain one or more intracellular signal transducing domains. Addition of lipocalin-fold ligand "b" induces interac tion of the two constructs (containing the lipocalin-fold mole cule "a" and the lipocalin-fold binding interaction partner "c") . The components "a" and "c" of the LRPPI system can be ex- tracellularly (preferred) or intracellularly integrated (with or without linker sequences) into the two CAR constructs. Shown in Fig. 2A and 2B is the fusion of the lipocalin-fold molecule "a" to the CAR construct that mediates target cell binding and fu sion of the lipocalin-fold binding interaction partner "c" to the signal transducing construct, which, however, could also be reverse. Binding to the target cell can be mediated by any pro tein capable of binding to a chosen antigen on the target cell surface. In an alternative version (LRPPI-CAR strategy C) the lipocalin-fold molecule itself can directly bind to the antigen, whereby the antigen then is part of the LRPPI system and acts as lipocalin-fold binding interaction partner "c". The lipocalin- fold molecule "a" in this case is engineered by in vitro di rected evolution using protein engineering technologies and sub sequent selection for lipocalin-fold ligand-dependent binding to the lipocalin-fold binding interaction partner "c", which in this case is an antigen of a target cell.

Examples of CAR architectures are well-known in the art (e.g. reviewed by Abate-Daga et al . , Mol Ther Oncolytics. 2016; 3:16014, and see e.g. WO 2014/127261 Al and WO 2015017214 Al) . In an embodiment, the signal transducing construct usually comprises one or two of the costimulatory signaling domains of the receptors CD27, CD28, CD134, CD137, ICOS, DAP12, activating NK cell receptors etc. (in any order from N- to C-terminus) , with or without a domain for transmitting signal 1 (e.g. CD3zeta) , or alternatively comprises the inhibitory cytoplasmic domains of the inhibitory receptors PD1, CTLA4, LAG3, TIM3, in hibitory NK cell receptors etc. Transmembrane domains and extra cellular membrane anchors can be derived from these receptors or from other proteins such as, e.g., CD3zeta, CD8alpha, CD28 etc.

In LRPPI-CAR strategy B (Figure 2B) , the signal transducing domain (s) do not need to be necessarily confined to one of the two constructs, but instead could be separated from each other by fusing them to the two different CAR constructs separately (not shown) . Further diversity arises due to the fact that both CAR constructs, the antigen binding and/or the signal transduc ing chain, can contain spacer domains comprised of fragments of the signal transducing proteins or of, e.g., IgG-Fc domains. Moreover, the antigen binding domains can be based on single chain Fv fragments, endogenous receptor- or ligand domains (e.g., NKG2D, IL13 etc.) or any other available binder scaffold (Simeon et al . , Protein Cell. 2017; DOI 10.1007/sl3238-017-0386- ) . If the lipocalin-fold molecule itself is the binding domain engineered for binding to a chosen target antigen (LRPPI-CAR strategy C, Figure 2C) , the antigen on the target cell itself is part of the LRPPI system, i.e., representing the lipocalin-fold binding interaction partner "c". Principally, the binding do mains of the CAR can be directed to any antigen, including non protein antigens.

Accordingly, a preferred embodiment of the LRPPI system ac cording to the present invention is a system, wherein the lipocalin-fold molecule is part of an ectodomain of a chimeric antigen receptor and wherein the lipocalin-fold binding interac tion partner is a cell surface antigen.

Due to the nature of the present invention, the lipocalin- fold molecule and the lipocalin-fold binding interaction partner are preferably provided as polypeptides, especially polypeptides obtained by recombinant technologies. The molecules according to the present invention can - at least theoretically - also be provided by chemical syntheses; however, molecular biology tech niques are, of course, the practically most relevant techniques for providing the lipocalin-fold molecule and the lipocalin-fold binding interaction partner. On the other hand, the lipocalin- fold ligand is preferably produced by chemical synthesis or by extraction from natural sources.

Accordingly, another aspect of the present invention relates to nucleic acid molecules comprising nucleotide sequences encod ing the lipocalin-fold molecule and/or the lipocalin-fold bind ing interaction partner according to the present invention. The nucleic acid according to the present invention will in some em bodiments be DNA or RNA, including, e.g., a recombinant expres sion vector. The nucleic acid molecules according to the present invention may also be provided in other form, e.g. in viral vec tors. The nucleic acid molecules may be active or conditionally active in cells and be present or presents in some embodiments as RNA, e.g., in vitro or in vivo synthesized RNA or RNA pack aged in a retrovirus, preferably RNA packaged in a lentivirus.

In some cases, the nucleic acid molecule of the present in vention comprises a nucleotide sequence encoding only the lipocalin-fold molecule (and not the lipocalin-fold binding in- teraction partner) . In some cases, the nucleic acid molecule of the present invention comprises a nucleotide sequence encoding only the lipocalin-fold binding interaction partner (and not the lipocalin-fold molecule) . In some cases, the nucleic acid mole cule of the present disclosure comprises a nucleotide sequence (or two separate nucleotide sequences) encoding both the lipocalin-fold molecule and the lipocalin-fold binding interac tion partner of the present invention.

In the case where the lipocalin-fold molecule and the lipocalin-fold binding interaction partner are encoded by dif ferent nucleic acid molecules, the present invention provides a kit of at least two nucleic acid molecules, wherein the first nucleic acid molecule comprises nucleotide sequences encoding the lipocalin-fold molecule according to the present invention and wherein the second nucleic acid molecule comprises sequences encoding the lipocalin-fold binding interaction partner accord ing to the present invention, wherein, again, the nucleic acids are preferably selected from DNA or RNA, more preferably in vitro transcribed RNA or RNA packaged in a retrovirus, especial ly RNA packaged in a lentivirus.

The present invention also provides a vector, e.g. a recom binant expression vector, comprising the nucleic acid molecules according to the present invention (i.e. encoding the lipocalin- fold molecule and/or the lipocalin-fold binding interaction partner) and/or the kit of nucleic acid molecules (encoding the lipocalin-fold molecule and the lipocalin-fold binding interac tion partner) .

Such a vector can include a selectable marker, an origin of replication, and other features that provide for replication and/or maintenance of the vector. Suitable vectors include, e.g., plasmids, viral vectors, and the like. Large numbers of suitable vectors and promoters are known to those of skill in the art; many are commercially available for generating the re combinant constructs according to the present invention. The following vectors are provided by way of example. Bacterial: pBs, phagescript, PsiX 174, pBluescript SK, pBs KS, pNH8a, pNH16a, pNH18a, pNH46a (Stratagene, La Jolla, Calif., USA); pTrc99A, pKK223-3, pKK233-3, pDR540, and pRIT5 (Pharmacia, Upp sala, Sweden). Eukaryotic: pWLneo, pSV2cat, pOG44, PXR1, pSG (Stratagene) pSVK3, pBPV, pMSG and pSVL (Pharmacia) . Vectors generally can have convenient restriction sites located near the promoter sequence to provide for the insertion of nucleic acid sequences encoding heterologous proteins. A selectable marker operative in the expression host may be present. Suitable vec tors include viral vectors (e.g. viral vectors based on vaccinia virus, poliovirus, adenovirus, adeno-associated virus, SV40, herpes simplex virus, human immunodeficiency virus, a retroviral vector (e.g., Murine Leukemia Virus, spleen necrosis virus, and vectors derived from retroviruses such as Rous Sarcoma Virus, Harvey Sarcoma Virus, avian leukosis virus, human immunodefi ciency virus, myeloproliferative sarcoma virus, and mammary tu mor virus) ; and the like) . Preferred vectors, due to the ability of efficiently integrating into the genome of the transduced cells, are retroviral vectors, especially gamma-retroviral vec tors and lentiviral vectors, i.e. vectors derived from at least a portion of a retrovirus genome. An example of a preferred ret roviral vector is a self-inactivating lentiviral vector (as pro vided in Milone et al . , Mol Ther. 2009; 17 (8) : 1453-1464) . Other examples of lentivirus vectors that may be used in the clinic include, e.g., the LENTIVECTOR® gene delivery technology from Oxford BioMedica, the LENTIMAX™ vector System from Lentigen and the like. Nonclinical types of lentiviral vectors are also available and would be known to one skilled in the art. Other types of preferred vectors that can efficiently integrate into the genome of transfected cells are transposon vectors, prefera bly PiggyBAC-based vectors and Sleeping beauty-based vectors. Further important non-viral strategies for integrating a gene of interest into the genome of a cell are based on site-specific nuclease technologies (e.g., based on Zinc-finger nucleases (ZFNs) or transcription activator-like effector nucleases (TALENs) ) or on CRISPR/Cas-technology (as described, e.g., by Gaj et al . , Trends Biotechnol. 2013; 31 (7) : 397-405; and Ren et al . , Protein Cell 2017 ; 8 ( 9) : 634-643) . These technologies allow for integration of defined nucleotide sequences from any DNA molecule (single stranded DNA or double stranded DNA; in the form of a vector, PCR amplicon etc.) and are attractive because the gene of interest can be integrated into the genome down stream of endogenous promoters (as described, e.g., by Eyquem et al., Nature. 2017 ; 543 ( 7643 ) : 113-117 ) .

The present invention also provides a kit of at least two vectors, wherein the first vector comprises a nucleic acid mole cule encoding the lipocalin-fold molecule according to the pre sent invention and wherein the second vector comprises a nucleic acid molecule encoding the lipocalin-fold binding interaction partner according to the present invention. The two vectors may be provided with the same or different regulation sequences in order to achieve expression in the same or different host sys tems (e.g. suitable cells where the vectors express the lipocalin-fold molecule and/or the lipocalin-fold binding inter action partner after transformation with the vector or propaga tion) .

In the vector or kit of vectors of the present invention, the nucleic acid molecules encoding the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner can be op- erably linked to a transcriptional control element, yielding an expression vector. Such a transcriptional control element can be e.g. a promoter, an enhancer, etc., wherein suitable promoter and enhancer elements are known in the art. For expression in a bacterial cell, suitable promoters include lacl, lacZ, T3, T7, gpt, lambda P and trc. For expression in a eukaryotic cell, suitable promoters include light and/or heavy chain immunoglobu lin gene promoter and enhancer elements, cytomegalovirus immedi ate early promoter, herpes simplex virus thymidine kinase pro moter, early and late SV40 promoters, promoter present in long terminal repeats from a retrovirus (e.g. the 5 ' -LTR of a gamma retrovirus or a promoter sequence comprising subelements R and U3 of the 5 ' -LTR of the Moloney murine leukaemia virus (MMLV) ) , promoter present in the murine stem cell virus (MSCV) , mouse metallothionein-I promoter, EFl-alpha with or without intron, promoter of phosphoglycerate kinase (PGK) , and various art-known tissue specific promoters. Suitable reversible promoters, in cluding reversible inducible promoters are known in the art. Such reversible promoters may be isolated and derived from many organisms, e.g. eukaryotes and prokaryotes. Modification of re versible promoters derived from a first organism for use in a second organism, e.g. a first prokaryote and a second a eukary ote, a first eukaryote and a second a prokaryote, etc., is well known in the art. Such reversible promoters, and systems based on such reversible promoters but also comprising additional con trol proteins, include alcohol regulated promoters (e.g. alcohol dehydrogenase I (alcA) gene promoter, promoters responsive to alcohol transactivator proteins (AlcR) , etc.) _/ tetracycline reg ulated promoters, (e.g. promoter systems including TetActiva- tors, TetON, TetOFF, etc.), steroid regulated promoters (e.g. rat glucocorticoid receptor promoter systems, human estrogen re ceptor promoter systems, retinoid promoter systems, thyroid pro moter systems, ecdysone promoter systems, mifepristone promoter systems, etc.), metal regulated promoters (e.g. metallothionein promoter systems, etc.), pathogenesis-related regulated promot ers (e.g. salicylic acid regulated promoters, ethylene regulated promoters, benzothiadiazole regulated promoters, etc.), tempera ture regulated promoters (e.g., heat shock inducible promoters (e.g. HSP-70, HSP-90, soybean heat shock promoter, etc.), light regulated promoters, synthetic inducible promoters, and the like .

In some instances, the locus or construct or transgene con taining the suitable promoter can be irreversibly switched through the induction of an inducible system. Suitable systems for induction of an irreversible switch are well known in the art, e.g., induction of an irreversible switch may make use of a Cre-lox-mediated recombination. Any suitable combination of re- combinase, endonuclease, ligase, recombination sites, etc. known to the art may be used in generating an irreversibly switchable promoter. Methods, mechanisms, and requirements for performing site-specific recombination, described elsewhere herein, find use in generating irreversibly switched promoters and are well known in the art. In some cases, the promoter is a CD8 cell- specific promoter, a CD4 cell-specific promoter, a neutrophil- specific promoter, or an NK-specific promoter. For example, a CD4 gene promoter can be used. As another example, a CD8 gene promoter can be used. NK cell-specific expression can be achieved by use of a Neri (p46) promoter. In some embodiments, e.g. for expression in a yeast cell, a suitable promoter is a constitutive promoter such as an ADH1 promoter, a PGK 1 promot er, an ENO promoter, a PYK 1 promoter and the like; or a regu- latable promoter such as a GAL1 promoter, a GAL10 promoter, an ADH2 promoter, a PH05 promoter, a CUP1 promoter, a GAL7 promot er, a MET25 promoter, a MET3 promoter, a CYC1 promoter, a HIS3 promoter, an ADH1 promoter, a PGK promoter, a GAPDH promoter, an ADC1 promoter, a TRP1 promoter, a URA3 promoter, a LEU2 promot- er, an ENO promoter, a TP1 promoter, and AOX 1 (e.g. for use in Pichia) . Selection of the appropriate vector and promoter is well within the level of ordinary skill in the art. Suitable promoters for use in prokaryotic host cells include a bacterio phage T7 RNA polymerase promoter; a trp promoter; a lac operon promoter; a hybrid promoter, e.g. a lac/tac hybrid promoter, a tac/trc hybrid promoter, a trp/lac promoter, a T7/lac promoter; a trc promoter; a tac promoter, and the like; an araBAD promot er; in vivo regulated promoters, such as an ssaG promoter or a related promoter, a pagC promoter, a nirB promoter, and the like; a sigma70 promoter, e.g. a consensus sigma70 promoter (see, e.g., GenBank Accession Nos. AX798980, AX798961, and AX798183); a stationary phase promoter, e.g. a dps promoter, an spv promoter, and the like; a promoter derived from the patho genicity island SPI-2; an actA promoter; an rpsM promoter; a tet promoter; an SP6 promoter; and the like. Suitable strong promot ers for use in prokaryotes such as Escherichia coli include Trc, Tac, T5, T7, and PLambda. Examples of operators for use in bac terial host cells include a lactose promoter operator (Laci re pressor protein changes conformation when contacted with lac tose, thereby preventing the Laci repressor protein from binding to the operator) , a tryptophan promoter operator (when complexed with tryptophan, TrpR repressor protein has a conformation that binds the operator; in the absence of tryptophan, the TrpR re pressor protein has a conformation that does not bind to the op erator) , and a tac promoter operator.

According to a preferred embodiment of the present inven tion, the vector or the kit of at least two vectors, or at least one, preferably a least two, of the vectors (in the kit) com prise a T lymphocyte-specific promoter or an NK cell-specific promoter operably linked to the nucleotide sequences encoding the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner.

According to a further aspect, the present invention also relates to a genetically modified cell which has been modified to produce the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner according to the present invention. Preferably, a cell is provided which has been modified to pro duce both, the lipocalin-fold molecule and the lipocalin-fold binding interaction partner according to the present invention. As an alternative, the present invention also provides a kit of at least two cells, wherein the first cell is genetically modi fied to produce the lipocalin-fold molecule according to the present invention and wherein the second cell is genetically modified to produce the lipocalin-fold binding interaction part ner according to the present invention.

The cells according to the present invention are designed to be capable of expressing the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner. With the ligand reg ulated protein-protein interaction system according to the pre sent invention, protein interaction can be regulated and steered within such a cell. As a research tool, all cells (which are in principle capable of being transformed with the nucleic acid molecules according to the present invention) as well as organ isms comprising such cells may be designed with the ligand regu lated protein-protein interaction system according to the pre sent invention to address versatile biological and biochemical questions. The cells of the present invention may also be used to produce the vectors of the present invention (e.g. as virus or plasmid supernatant) from where they may then be further pu rified and provide these vectors in amplified and purified form.

According to a preferred embodiment, the cell (or the cells of the kit) are mammalian cells which are genetically modified to produce the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner according to the present invention. Preferred mammalian cells are stem cells, progenitor cells, or cells derived from a stem cell or a progenitor cell. Further preferred cells to be genetically modified according to the pre sent invention are primary cells and immortalized cell lines. For pharmaceutical uses, human primary cells and human trans formed cell lines are specifically preferred. However, also non human cells and cell lines may be suitable cell types, especial ly for addressing scientific questions with the system according to the present invention, e.g. non-human primate cell lines, ro dent (e.g., mouse, rat) cell lines, and the like.

Further preferred cells according to the present invention may be HeLa cells (e.g., American Type Culture Collection (ATCC) No. CCL-2), CHO cells (e.g., ATCC Nos. CRL9618, CCL61 , CRL9096), 293 cells (e.g., ATCC No. CRL-1573) , Vero cells, NIH 3T3 cells (e.g., ATCC No. CRL-1658), Huh-7 cells, BHK cells (e.g., ATCC No. CCL10) , PC12 cells (ATCC No. CRL1721), COS cells, COS-7 cells (ATCC No. CRL1651), RATI cells, mouse L cells (ATCC No. CCL1.3), human embryonic kidney (HER) cells (ATCC No. CRL1573), HLHepG2 cells, Hut-78, Jurkat, HL-60, NK cell lines (e.g., NKL, NK92, and YTS) , and the like. In some preferred instances, the cell according to the present invention is not an immortalized cell line, but is instead a cell (e.g. a primary cell) obtained from an individual. For example, in some cases, the cell is an immune cell obtained from an individual. As an example, the cell is a T lymphocyte obtained from an individual. As another exam ple, the cell is a cytotoxic cell obtained from an individual. As another example, the cell is a stem cell or progenitor cell obtained from an individual.

According to a specifically preferred embodiment, the mamma lian cell according to the present invention, which is trans formed with a vector or a kit of at least two vectors encoding the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner according to the present invention, is a T cell or an NK cell.

According to a further aspect, the present invention relates to a pharmaceutical preparation which comprises a nucleic acid molecule or a kit of nucleic acid molecules according to the present invention, a kit according to the present invention, a vector or a kit of vectors according to the present invention, or a cell or a kit of cells according to the present invention.

According to another aspect, the present invention also re lates to a non-human animal comprising a cell (or cells) accord ing to the present invention, which is capable of expressing the lipocalin-fold molecule and the lipocalin-fold binding interac tion partner and which is therefore able to provide the system according to the present invention by e.g. the external addition of the lipocalin-fold ligand. With such a transgenic animal mod el, the ligand regulated protein-protein interaction system can be used in an in vivo model for addressing various biological questions both, scientifically and as test system for various industrial uses. Preferred animal models are those that are typ ically used in scientific research, such as primates, pigs, sheep, rodents, especially mice, rabbits and rats; chicken, frog (Xenopus laevis) , insects, such as drosphila; zebrafish, etc..

Specifically preferred animal models are those with an immune system which is similar to the human immune system, especially if the system according to the present invention is established as CAR (see above) . Establishment of the ligand regulated pro tein-protein interaction system is also possible in plant cells. Accordingly, another aspect of the present invention relates to a plant comprising a cell (or cells) according to the present invention, which is also capable of expressing the lipocalin- fold molecule and the lipocalin-fold binding interaction partner and which is therefore also able to provide the system according to the present invention. Preferred plants wherein the system according to the present invention can be established are Ara- bidopsis, but also crop plants, such as corn, wheat, potato, to mato, soy, etc.. Other preferred organisms are yeast and bacte ria .

The present invention is further described by the following examples and the figures, yet without being limited thereto.

Fig. 1 shows the schematics of the present invention as an LRPPI system based on a lipocalin-fold (A) and a preferred em bodiment thereof (B) . (A) lipocalin-fold molecule ("a") can ac commodate in its calyx a small molecule lipocalin-fold ligand ("b") and can bind to a lipocalin-fold binding interaction part ner "c" with higher affinity in the presence of "b"; (B) the lipocalin-fold molecule ("a") may be fused (+/- flexible linker) to terminal or internal sites of a protein "I" and also accommo dates in its calyx a small molecule lipocalin-fold ligand ("b") and can bind to a lipocalin-fold binding interaction partner "c", which may be preferably a protein (that may be fused (+/- flexible linker) to terminal or internal sites of a further pro tein) , with higher affinity in the presence of "b".

Fig. 2 shows a schematic example of lipocalin-fold molecule- based LRPPI systems according to the present invention integrat ed into CARs. "a" can be part of a soluble protein (secreted or exogenously added) or of a transmembrane construct and "c" can be part of a signal transducing construct (shown in A and B) . Of course, "a" and "c" can be integrated in the constructs vice versa (not shown) . There are many more possible arrangements (not shown) . For example, the signal transducing domains do not need to be necessarily confined to one of the two constructs, and "a" and "c" can alternatively be part of the cytoplasmic do mains of the constructs. (C) shows an example in which "c" is an antigen .

Fig. 3 shows A1120 dependent binding of increasing concen trations of human RBP4 to differrent rcSso7d- and FN3-based lipocalin-fold binding interaction partners displayed on the surface of yeast. The assay was performed in the presence (5 mM) or absence of the lipocalin-fold ligand A1120. RBP4-binding was detected with an anti-penta-His antibody and subsequently ana lyzed by flow cytometry.

Fig. 4 shows binding of rcSso7d- and FN3-based lipocalin- fold binding interaction partners (displayed on the surface of yeast) to RBP4 loaded with different small molecules. All small molecules were present at a concentration of 5 mM. RBP4-binding was detected with an anti-penta-His antibody and subsequently analyzed by flow cytometry.

Figs. 5A and 5B show sequences of rcSso7d- and FN3-based lipocalin-fold binding interaction partners. The amino acids differing from the original rcSso7d or FN3 sequence are high lighted in blue.

Figs. 6A and 6B show schematics of the signal transducing constructs, i.e. the part of the CAR, in which either the lipocalin-fold molecule or the lipocalin-fold binding interac tion partner is fused to the signaling domains.

Figs. 7A, 7B, 7C and 7D show sequences of the signal trans ducing CAR constructs containing rcSso7d- and FN3-based RBP4 binding interaction partners.

Figs. 8A and 8B show expression of the signal transducing CAR constructs in primary T cells: (A) Expression of constructs containing different rcSso7d-based RBP4 binding interaction partners; (B) Expression of constructs containing different FN3- based RBP4 binding interaction partners or alternatively RBP4 fused to the signaling domains.

Fig. 9 shows schematics of different antigen binding CAR contructs, i.e., the part of the CAR, in which an antigen bind ing domain is fused to lipocalin-fold molecule or the lipocalin- fold binding interactin partner. In the shown example either a scFv directed against CD19 or an rcSso7d directed against EGFR was used as antigen binding domain.

Fig. 10 shows sequences of the antigen binding CAR con structs .

Fig. 11 shows the binding of antigen binding constructs (fu- sion proteins A-D) to antigen expressing target cells (A) and signaling CAR construct expressing primary human T cells (B) :

(A) CD19 positive Nalm-6 cells with and without stable expres sion of a truncated EGFR (tEGFR) were co-incubated with superna tants of Jurkat cells expressing fusion proteins A-D, as indi cated; (B) Fusion proteins were either co-expressed in primary human T cells "co-electroporated") or added to primary T cells as supernatants obtained from Jurkat cells that expressed the fusion proteins ("supernatant added") . The primary T cells ex pressed a signaling transmembrane CAR construct that was either RS3 short CAR or RS3 long CAR, as indicated. Staining of fusion protein binding was performed in the presence of 5 mM A1120.

Fig. 12 shows lipocalin-fold ligand (in this case 5 mM A1120) dependent function of different CAR variants in primary T cells, whereby the soluble CAR constructs (fusion proteins A and B) were expressed in Jurkat cells and the supernatants therefrom were added to the primary T cells as indicated. Depicted in (A) and (B) is the capacity for triggering IFN-g production and cy totoxicity, respectively.

Fig. 13 shows lipocalin-fold ligand (in this case 5 mM A1120) dependent function of the soluble fusion protein B when co-expressed in primary T cells together with CAR construct "RS3 long" ("RS3 long + co-elpo B") . T cells without CAR ("no con struct") and T cells expressing only the signal transducing con struct but no antigen binding construct ("RS3 long only") served as negative controls. T cells expressing an anti-CD19 CAR served as positive control. Depicted in (A) and (B) is the capacity for triggering IFN-g production and cytotoxicity, respectively. Sta tistical significance was calculated using the paired two-tailed and the ratio paired two-tailed Student's t test for specific lysis and IFN-g levels, respectively (* = p < 0.05; ns = p > 0.05).

Fig. 14 shows analysis of RBP4 binding interaction partners RS3, RS5 and RF2 (in this case the lipocalin-fold binding inter action partners) by differential scanning calorimetry (DSC) for evaluation of the melting temperature (T _m) .

Fig. 15 shows analysis of the affinity of the RBP4 binding interaction partner RS3 to RBP4 by surface plasmon resonance (SPR) . (A) Binding of RS3 to RBP4 in the presence of 5 mM A1120.

(B) and (C) Binding of RS3 to RBP4 in the absence of A1120. Ap- plied concentrations of RS3 are indicated in the plots. Steady state analysis of K _d values (dissociation constants) is shown in the right panel of A and C.

Fig. 16 shows the elution profiles of RBP4 binding interac tion partners RSI, RS3, RS5, RF1, RF2 and RF3 analyzed by size exclusion chromatography (SEC) .

Fig. 17 shows analysis of the affinities of the RBP4 binding interaction partners RS3 (A) , RS5 (B) and RF2 (C) to RBP4 in the presence (50 mM) and absence of A1120 as determined by isother mal titration calorimetry (ITC).

Fig. 18 shows the overview of all performed affinity meas urements for the RBP4 binding interaction partners RS3, RS5 and RF2 for binding to RBP4 in the presence and absence of A1120 with three different methods (*n.a., not analyzable) .

Fig. 19A shows binding of TTR to RBP4 displayed on the sur face of yeast. Fig. 19B shows binding of TTR to yeast-displayed RBP4 in the presence (50 mM) or absence (PBSA) of different po tential lipocalin-fold ligands.

Fig. 20 shows binding of TTR to yeast-displayed RBP4 in the presence (50 mM) or absence (PBSA) of different potential lipocalin-fold ligands.

Examples :

Example 1: RBP4-based LRPPI system

For bovine and human RBP4 there are a series of ligands known to induce conformational changes in the loop regions, which result in dissociation of the natural protein partner TTR (see prior art cited above) . In the present example, it is demonstrated by using human RBP4 and its synthetic non-retinoid ligand A1120 (PubChem CID 25138295) how such a ligand-induced conformational switch of a lipocalin-fold molecule can be used as an element in LRPPI. For this purpose, His-tagged full length RBP4 (UniProt ID P02753) was employed as antigen in an alternat ing screening process in presence (5 mM) and absence of A1120, respectively, in which ligand-dependent lipocalin-fold binding interaction partners were selected from libraries of two scaf folds with very different protein structure, i.e., FN3- and Sso7d-based scaffolds (Traxlmayr et al . , J Biol Chem. 2016 ; 291 ( 43 ) : 22496-22508 ; Chen et al . , Methods Enzymol. 2013;523:303-326). As described below, this process yielded lipocalin-fold binding interaction partners which bind RBP4 in an A1120-dependent manner. Figure 3 shows the affinity of these lipocalin-fold binding interaction partners in the presence (5 mM) and absence of the RBP4-ligand A1120. The diagrams in Fig. 3 show the flow cytometric analysis of the binding intensity at different concentrations of human RBP4 in the presence (5 mM) or absence of A1120 to selected rcSso7d- or FN3-based lipocalin- fold binding interaction partners, which were displayed on the surface of yeast. RBP4-binding was detected by using an anti- penta-His antibody (Qiagen) , followed by flow cytometric detec tion. rcSso7d-based lipocalin-fold binding interaction partners are termed RSI through RS5, respectively, whereas FN3-based lipocalin-fold binding interaction partners are termed RF1, RF2 and RF3. The calculated K _d values are displayed below the dia grams (mean of three measurements) . These data in Fig. 3 clearly show that it is possible to engineer LRPPI systems based on lipocalin-fold molecules, i.e. that the affinity of a lipocalin- fold molecule "a" (in this case RBP4) to a lipocalin-fold bind ing interaction partner "c" (in this case different mutants based on the protein scaffolds rcSso7d or FN3) is increased in the presence of a lipocalin-fold ligand "b" (in this case A1120) . Moreover, the data in Fig. 3 also show that LRPPI sys tems based on lipocalin-fold molecules can be engineered by us ing structurally diverse lipocalin-fold binding interaction partners (in this example based on either rcSso7d or FN3, re spectively) containing binding sites that are structurally very different (in FN3-based binders the binding sites are composed of loop regions, whereas in rcSso7d-based binders the binding sites are composed of rigid b-strands; (Traxlmayr et al . , J Biol Chem. 2016 ; 291 ( 43 ) : 22496-22508 ; Chen et al . , Methods Enzymol. 2013;523:303-326)). This means that LRPPI systems based on lipocalin-fold molecules are not crucially dependent on a par ticular structure (i.e. fold) of the lipocalin-fold binding in teraction partner and therefore such LRPPI systems are highly flexible regarding the choice of the lipocalin-fold binding in teraction partner.

The possibility to engineer lipocalin-fold binding interac tion partners that can specifically recognize conformational changes in the lipocalin-fold molecule induced by one lipocalin- fold ligand but not or much less the conformations induced by other lipocalin-fold ligands is a key advantage in engineering clinically applicable LRPPI systems that are orthogonal to each other and can work in parallel. Figure 4 shows that the lipocalin-fold binding interaction partners selected for the A1120-induced conformation of RBP4 indeed have only very low or absent affinity to RBP4 loaded with other ligands known to in duce conformational switching of RBP4. This was determined by flow cytometric analysis of the binding intensity (mean of three measurements) of different concentrations of human RBP4 in the presence (5 mM) or absence of the indicated ligands to selected rcSso7d- or FN3-based lipocalin-fold binding interaction part ners, which were displayed on the surface of yeast. The sequenc es of the respective lipocalin-fold binding interaction partners are shown in Figure 5A and 5B.

Expression and purification of human RBP4

Human His-tagged RBP4 (residues 19-201) in pPICZ-alpha-A vector was kindly provided by John Findlay (Marie Curie lab for membrane proteins, Department of Biology, Maynooth University, Co Kildare, Ireland (Wysocka-Kapcinska et al . , Protein Expr Pu- rif. 2010 ; 71 ( 1 ): 28-32 ) ) and the sequence was verified using Sanger-sequencing . The protein was expressed in P. pastoris strain KM71H. Cells were cultivated in YPG (2% peptone, 1% yeast extract, 1% glycerol) supplemented with Zeocin (100 yg/mL) at 30°C and 180 rpm overnight. Cells were diluted on the following day and further cultivated until Oϋboo of 2 was reached. Protein expression was induced by centrifugation of the cells and resus pension in YP-medium supplemented with 1% methanol and further incubation at 20°C and 180 rpm for 3 days. Every day fresh meth anol was added to a final concentration of 1% to enhance protein expression and subsequent secretion. After 3 days, the superna tant was harvested by 2-step centrifugation (1500 g, 15 min, 4°C and 12200 g, 25 min, 4°C) to first remove the cells and further small particles, respectively. Subsequently, diafiltration was performed to exchange the medium with 50 mM phosphate buffer, pH 7.5. The following purification protocol was adapted from Wysocka-Kapcinska et al . (Protein Expr Purif. 2010 ; 71 ( 1 ): 28-32 ) .

Briefly, the diafiltrated supernatant was supplemented with 5 mM imidazole and applied to a HisTrap FF column (GE healthcare) connected to an AKTA FPLC purifier system to enable binding of the His-tagged RBP4 to the column. After washing with 50 mM phosphate buffer (pH 7.5 containing 500 mM NaCl and 5 mM imidazole) , elution was performed using a linear imidazole gra dient (from 5% to 100% with 50 mM phosphate buffer, pH 7.5, con taining 500 mM NaCl and 500 mM imidazole) . Absorbance at 280 nm was detected to observe elution of the protein and the corre sponding fractions were analysed by SDS-PAGE. Those fractions showing both absorbance and the presence of a protein band at the expected size (22 kDa corresponding to the His-tagged RBP4), were pooled and concentrated using Amicon Ultra-15 10K Centrifu gal filters (Merck Millipore) . In addition, buffer was exchanged to phosphate buffered saline (PBS, pH 7.4) to prepare the pro tein solution for size exclusion chromatography (SEC) with a Su- perdex 200 column (10 mm x 300 mm, GE Healthcare) .

Before SEC, a fraction of RBP4 was labelled with biotin us ing the EZ-Link Sulfo-NHS-LC-LC-Biotin kit (Thermofisher Scien- tifc) . 10 mM biotin solution was added in a molar excess of 5:1 to the protein solution and incubated at room temperature for 1 hour while stirring. Subsequent preparative SEC was performed to remove remaining unbound biotin molecules and possible aggre gates of RBP4, which could only be observed to a limited amount.

Screening for RBP4 binding interaction partners using yeast display

Two different scaffolds were used for screening of RBP4 binding interaction partners ("binders") , one of them being re duced charge Sso7d (rcSso7d) , a small (7 kDa) thermostable DNA- binding protein from Sulfolobus solfataricus in which excess positive charges have been minimized previously (Traxlmayr et al . , J Biol Chem. 2016; 291 (43) : 22496-22508) . The second binder scaffold is the tenth type III domain of human fibronectin (FN3) , with a molecular weight of 10 kDa (Chen et al . , Methods Enzymol. 2013;523:303-326). Yeast libraries based on rcSso7d (rcSso7d-ll and rcSso7d-18), containing 1.4 x 10 ⁹ transformants each and the library G4 based on the FN3 domain (Hackel et al . , JMB 2010;401:84-96), containing 2.5 x 10 ⁸ transformants, were used. S. cerevisiae cells (strain EBY100) were grown in SD-CAA medium (20 g/L glucose, 6.7 g/L yeast nitrogen base, 5 g/L bacto casamino acids, 11.85 g/L sodium citrate dihydrate and 7.4 g/L citric acid monohydrate) at 30°C overnight while shaking. On the following day, cells were sub-cultivated to an Oϋboo of 1 in SD- CAA and growth was monitored by measuring the Oϋboo· After reach ing a maximum Oϋboo of 4, yeast cells were centrifuged and resus pended in SG-CAA medium (2 g/L glucose, 20 g/L galactose, 6.7 g/L yeast nitrogen base, 5 g/L casamino acids, 10,2 g/L disodium hydrogen phosphate and 4,82 g/L sodium phosphate monobasic) for induction of surface expression of rcSso7d- or FN3-mutants, re spectively. Cells were incubated at 20°C overnight with shaking and harvested by centrifugation. Two cycles of bead selection were performed with magnetic streptavidin-coated Dynabeads (Life technologies) loaded with biotinylated RBP4. To avoid the selec tion for non-specific binders or binders which are specific for streptavidin, negative selection (with bare beads) was performed between cycles of positive selection. In all bead selections, the RBP4 ligand A1120 was present at a concentration of 5 mM. After the second bead-selection cycle, error prone PCR (epPCR) was conducted in order to increase mutations in rcSso7d and FN3 genes which might contribute to binding. The plasmid DNA was subjected to 19 cycles of PCR using nucleotide analogs (2 mM of 8-O _XO -2 ' -deoxyguanosine-5 ' -triphosphate, 8-oxo-dGTP, and 2mM 2 deoxy-p-nucleoside-5 ' -triphosphate, dPTP) and the primers epSso_fwd (5 ' -GGCTCTGGTGGAGGCGGTAGCGGAGGCGGAGGGTCGGCTAGC-3 ' ) and epSso_rev (5 ' -CTATTACAAGTCCTCTTCAGAAATAAGCTTTTGTTCGGATCC-3 ' ) for the rcSso7d-based lipocalin-fold binding interaction partners; and the primers FN3_fwd (5'-

CGACGATTGAAGGTAGATACCCATACGACGTTCCAGACTACGCTCTGCAG-3 ' ) and

FN3_rev ( 5 ' -ATCTCGAGCTATTACAAGTCCTCTTCAGAAATAAGCTTTTGTTCGGAT CC-3') for the FN3-based lipocalin-fold binding interation part ners. The resulting product was used as a template for amplifi cation by a second PCR. Afterwards, EBY100 cells were trans formed with linearized pCTCON2 vector (Chao et al . , Nat Protoc. 2006 ; 1 (2 ) : 755-768 ; Angelini et al . , Methods Mol Biol. 2015;1319:3-36) and insert (PCR product) using the square wave protocol (single pulse, 500 V, 15 ms) and Bio-Rad Gene Pulser Xcell (Bio-Rad) . After a third round of bead-selection (compris ing 3 negative and 1 positive selection) , libraries were further enriched by fluorescence activated cell sorting (FACS) . Staining of yeast cells was performed in PBS supplemented with 0.1% bo vine serum albumin (BSA) in the presence of 5 mM A1120 (Sigma- Aldrich) using 300 nM biotinylated antigen (RBP4) and 5 yg/mL mouse anti-c-myc antibody 9E10 (Thermo Fisher Scientific) and incubation at 4°C for 1 hour while shaking. After a washing step, secondary staining was performed with 20 yg/mL streptavi- din-Alexa Fluor 647 and 20 yg/mL anti-mIgG-AF488 (both from Thermo Fisher Scientific) for 20 minutes at 4°C while shaking. After a final washing step, cells were sorted using a FACS Aria Fusion (BD) . In some sorting rounds, cells were stained with non-biotinylated RBP4 as primary reagent followed by secondary staining with 1 yg/mL anti-HA-Alexa Fluor 647 (clone 16B12, Bio- Legend) and 5 yg/mL Penta-His-Alexa Fluor 488 (Qiagen) , to avoid enrichment of binders recognizing biotinylated epitopes. Between the second and third FACS selection, another epPCR was per formed. One round of negative selection was performed in which no A1120 was present and mutants with strongly reduced binding signal were sorted. In total, 6 or 7 rounds of FACS were per formed for selection of RBP4-binding interaction partners.

Soluble expression of selected RBP4 -binding interaction partners

After the last selection round, 96 clones were sequenced. Based on the sequence, 16 RBP4-binders (9 rcSso7d- and 7 FN3- based lipocalin-fold binding interaction partners) were chosen and used for transformation of EBY100 cells. The affinity of single clones was determined by flow cytometry and titration of RBP4. Based on the affinity and the expression level, 8 binders (5 rcSso7d- and 3 FN3-based lipocalin-fold binding interaction partners) were further chosen for soluble expression. For this purpose, binders were subcloned into the pE-SUMO-vector (Life- Sensors) and expressed as fusion proteins with His6-tagged small ubiquitin-like modifier (SUMO) protein. After transformation of Rosetta (DE3) E. coli cells (Merck Millipore) with the sequence- verified plasmids, cultures were incubated in LB medium supple mented with kanamycin (50 yg/mL) and chloramphenicol (34 yg/mL) at 37°C overnight while shaking. Cells were diluted in terrific broth (12 g/L tryptone, 24 g/L yeast extract, 4% glycerol, 2,31 g/L KH ₂ PO ₄ and 16,43 g/L K ₂ HP0 ₄ *3H ₂ 0) supplemented with kanamycin (50 yg/mL) and chloramphenicol (34 yg/mL) and further incubated at 37°C until an Oϋboo of 2 was reached. Expression was induced by addition of 1 mM isopropyl b-D-l-thiogalactopyranoside (IPTG) and cells were further cultured overnight at 20°C. On the fol- lowing day, cells were harvested by centrifugation (5000 g, 20 min, 4°C) and resuspended in sonication buffer (50 mM sodium phosphate, 300 mM NaCl, 3% glycerol, 1% Triton X-100, pH 8) . Af ter sonication (2 x 90 s, duty cycle 50%, amplitude set to 5) and centrifugation (20000 g, 30 min, 4°C), purification of the supernatants containing the HIS6-SUMO-fusion proteins was per formed using the TALON metal affinity resin (Clontech) . Before application to the column, imidazole was added to the superna tants to a final concentration of 10 mM to avoid unspecific binding. Afterwards, supernatants were applied to the column twice, followed by several washing steps with equilibration buffer (50 mM sodium phosphate, 300 mM NaCl, pH 8) containing increasing amounts of imidazole (5-15 mM) . Elution was performed with equilibration buffer supplemented with 250 mM imidazole. The buffer was exchanged to PBS using Amicon Ultra-15 10K Cen trifugal filters (Merck Millipore) and the concentration was de termined by measuring the absorbance at 280 nm with a Nanodrop instrument. A portion of the fusion-proteins was directly frozen at -80°C while the rest was digested overnight at room tempera ture with SUMO-protease 1 (1 mg enzyme was used for 100 mg fu sion-protein) , resulting in cleavage of the hexahistidine tag and SUMO protein from the respective binder. The digested pro tein was again purified with TALON metal affinity resin (Clon tech) . His-tagged SUMO protein and His-tagged SUMO protease 1 bound to the resin while cleaved rcSso7d- or FN3-based lipocalin-fold binding interaction partners were found in the flow-through, which was further analyzed by SDS-PAGE. The ab sorbance at 280 nm of purified rcSso7d- and FN3-based lipocalin- fold binding interaction partners was determined by Nanodrop analysis and the concentration was calculated using the corre sponding extinction coefficients. For biophysical measurements, the buffer was exchanged with PBS using Amicon Ultra-15 3K cen trifugal filters (Merck Millipore) and the proteins were frozen at -80 °C .

Biophysical characterization of RBP4 -binding interaction partners

The stability of RBP4-binding interaction partners is evalu ated by determining the melting temperature (T _m) by differential scanning calorimetry (DSC) using MicroCal VP-DSC capillary cell microcalorimeter (MicroCal) . Briefly, 30 mM protein solution is exposed to an increasing temperature ranging from 20-110°C with a heating rate of l°C/min. Buffer baseline is subtracted and the resulting data are normalized for protein concentration, fol lowed by fitting with a non-two-state thermal unfolding model.

Surface plasmon resonance (SPR) is used to determine the af finity of the interaction between RBP4 and rcSso7d- or FN3-based lipocalin-fold binding interaction partners and is performed with BiacoreT200 (GE healthcare) . The antigen (RBP4) is coated onto a chip and incubated with various concentrations of binder solution in PBS, followed by a dissociation phase in PBS only. All measurements are performed both in the presence and absence of A1120 or other ligands. Finally, k _onr k _off and K _d values are obtained by global fitting.

For analytical SEC analysis, a total amount of 25 yg rcSso7d or FN3 mutants in running buffer (PBS with 200 mM NaCl) is fil tered through 0.1 ym Ultrafree-MC filter (Merck Millipore) , ap plied to a Superdex 200 10/300 GL column (GE healthcare) con nected to an HPLC prominence LC20 system (Shimadzu) and eluted with a flow rate of 0.75 mL/min at 25°C. In addition, multi angle light scattering (MALS) is used for determining the molec ular mass using WYATT Heleos Dawn8+ plus QELS, a refractive in dex detector RID-10A (Shimadzu) and a diode array detector SPD- M20A (Shimadzu) .

For the thermodynamic characterization of the interaction between RBP4 and rcSso7d- or FN3-based lipocalin-fold binding interaction partners, isothermal titration calorimetry (ITC) is conducted using an automated MicroCal PEAQ-ITC instrument (Mal vern Instruments) . The samples are centrifuged (17,000 g, 10 min, 20°C) and filtered (0.1 ym Ultrafree-MC filter, Merck Mil lipore) . RBP4 protein solution is applied to the sample cell and the respective binders are titrated with varying intervals. Dif ferent concentrations both for RBP4 and binders are used and in some experiments, binders are applied to the sample cell and RBP4 is titrated. The experiments are performed both in the presence and absence of the ligand A1120 or other ligands. Data analysis is performed with the MicroCal PEAQ-ITC analysis soft ware .

Example 2 : Identification of clinically applicable lipocalin- fold ligands with capacity for inducing conformational switching in a lipocalin-fold

By using human RBP4 and its native interaction partner TTR we exemplified a strategy for identifying novel lipocalin-fold ligands with capacity for inducing conformational switching of a lipocalin-fold. With the aim of identifying clinically applica ble lipocalin-fold ligands we performed a virtual screening of several databases (World Drug Index, KEGG Medicus, KEGG Ligands, DrugBank, Human Metabolome Database, ChEBI, ChEMBL, MDDR) with different stringency by using a pharmacophore model deduced from the RBP4/A1120 complex (PDB 3FMZ) . The result of this virtual screening is shown in Table 1, which contains a series of ap proved and experimental drugs and other molecules potentially attractive for clinical and non-clinical applications.

Table 1 :

Nr database HMDB pentenyl) anthraquinone Trihydroxy-3-prenylchalcone

1 Nafcillin 17 Sorafenib beta-D- 36 Dulxanthone H

2 Gerberinol Glucuronide 37 Judeol

3 Montelukast 18 Heterophyllin 38 Artonin E

4 Flurazepam 19 2 ' -0- 39 Kanzonol T

5 Quinidine barbitu Methylphaseollinisoflavan 40 Fragransol A rate 20 Tiapride 41 Dulxanthone F

6 Glyceollin I 21 Gluten exorphin C 42 Mulberrofuran T

7 Glyceollin II 22 Mulberrofuran M 43 Garcimangosone A

8 (-) -Shinpterocarpin 23 Dulxanthone G 44 Artonin B

9 Kanzonol W 24 Sclareol 45 Asteltoxin

10 6alpha- 25 Colupdox a

Hydroxyphaseollin 26 Kanzonol F database KEGG

11 (la, 5b, 6a) -7- 27 Mangostinone 46 Oxyfedrine

Protoilludene-1 ,5,6,14- 28 Gancaonin X 47 Profluthrin tetrol 14- (2, 4-dihydroxy-6- 29 Rubraflavone D 48 Momfluorothrin methylbenzoic acid) 30 Cyclokievitone hyd 49 Xyloylsulfamine

12 4 ' -O-Methylkanzonol rate 50 Cetotia ine hydro W 31 Glyceollidin II chloride hydrate

13 Cyclokievitone 32 Cyclandelate 51 Dicethia ine hydro

14 Licofuranone 33 Dulxanthone E chloride hydrate

15 Armillatin 34 Morusin 52 Diceta in

16 2- (4-Methyl-3- 35 (E) -2 ' , 4, 4 ' - 53 Oxola ine 54 Oksalamin 87 DIARBARONE 119 MENOXYMYCIN-B

55 Crisnatol mesylate 88 DICLOFOP 120 PENICILLIN-S

56 lotlubenzamide I 89 EPIPHENETHICILLIN 121 PSORALIDIN

57 Ceritinib 90 FTALIL-MEDEYOL 122 RUBIGINONE-Cl

58 Zykadia 91 SALETAMIDE HYDRO 123 SECOPSEUDOPTEROSIN-E

59 I iprothrin CHLORIDE 124 ASADISULFIDE

60 Triclabendazole 92 TIAPRIDE HYDROCHLO 125 BARANGCADOIC-ACID-A

61 Fasinex RIDE 126 BEPHEDON

62 Brivanib alaninate 93 TRUNCULIN-A 127 MELLEDONAL-B

63 Transfluthrin 94 DEOXYPENTALENYLGLU- 128 NERAMINOL

64 Enolica sodium CURON 129 PHOMOPSOLIDE-A

65 Enolicam sodium 95 LAFLUNIMUS 130 ROBUSTIC-ACID monohydrate 96 MERAZOLAM 131 ZOPFIELLAMIDE-A

66 Dibucaine 97 DIBUSADOL 132 CYSTODYTIN-B

67 Cinchocaine 98 PHENETICILLIN POTAS 133 DICLOXACILLIN

68 Nupercaine SIUM 134 FLUOROPROPRANOLOL

69 Trametinib 99 PRANOSAL 135 ILIOCICOLIN-B

100 DALEFORMIS 136 INDICANINE-B database WDI 101 DIPHENICILLIN 137 UACAREUBIN

70 FLUCLOXACILLIN 102 FENOTEROL HYDROCHLO 138 KOTTAMIDE-C

71 S- RIDE 139 MEXOLAMINE

FARNESYLTHIOSALICYLIC ACID 103 GIGANTIC-ACID 140 MYCAPEROXIDE-H

72 2-FLUOROTROPAPRIDE 104 HALOLITORALIN-B 141 OTOGIRIN

73 BUTAMPICILLIN 105 ISOPROPICILLIN 142 OXOLAMINE HYDROCHLO

74 DICLOXACILLIN SUL 106 PROGUANIL HYDROCHLO RIDE

FATE RIDE 143 OXOPROPALINE-A

75 FUROXACILLIN 107 PYRANOKUNTHONE-B 144 PURVALANOL-A

76 IBUCILLIN 108 TUBEROSIN 145 RUBIGINONE-C2

77 PRAZOCILLIN 109 ZOPFIELLAMIDE-B 146 TERACRYLSHIKONIN

78 SALETAMIDE 110 CLOXACILLIN SODIUM 147 CLODINAFOP-

79 TETRACHLORSALICYLAN- 111 LETIMIDE HYDROCHLO PROPARGYL-ESTER

ILIDE RIDE 148 MAZATICOL

80 CARBENICILLIN 112 BAIGENE-B 149 SETHOXYDIM

81 DICLOMETIDE 113 BIDWILLON-B 150 SULFAGUANOLE

82 METHYL-SULFOMETURON 114 CARBENICILLIN DISO 151 BALAPERIDONE

83 TRICHLOROSALICYLANI- DIUM 152 FLUCLOXACILLIN SODI LIDE 115 HYDROXYPROCAINE UM

84 CLOMETOCILLIN 116 OCHRATOXIN-A 153 GEODIAMOLIDE-TA

85 CYSTODYTIN-F 117 THEROX 154 LUCANTHONE-SULFOXIDE

86 DETANOSAL 118 MACARANGAFLAVANONE-B 155 MELLEOLIDE-D 156 NADOXOLOL HYDROCHLO 190 GARCIGERRIN-A 18,20

RIDE 191 0- 224 EUGLOBAL-G2

157 BECLOBRIC-ACID- DEMETHYLCHLOROTHRICIN 225 IDARUBICIN HYDRO GLUCURONIDE 192 SANGGENON-C CHLORIDE

158 CLOXACILLIN 193 TETRAPTEROL-G 226 MUTISICOUMARANONE-A

159 HYPERGUINONE-B 194 CHLORSULFURON 227 3-0-

160 OLIGOSPOROL-A 195 DEXAMETHASONE- METHYLCALOPOCARPIN

161 PROPOXYCAINE HYDRO DIETHYLAMINOACETATE 228 ALLOCLAMIDE HYDRO CHLORIDE 196 DIETHYXIME CHLORIDE

162 RONIFIBRATE 197 PYRIDOVERICIN 229 ANOPTERINE

163 SUDAN-BLUE-GN 198 SUBENDAZOLE 230 DIOXATION

164 TRICHODERMAMIDE-B 199 THIOCAINE 231 EUGLOBAL-I IC

165 BOTRYLLAMIDE-A 200 TRAPEZIFOLIXANTHONE 232 ILICICOLIN-C

166 CARFECILLIN 201 TRICLAZAN 233 MYCINOLIDE-I I

167 CLETHODIM 202 3 ' , 4 ' - 234 SUILLIN

168 DUTADRUPINE DICHLOROBENZAMIL 235 TOCOTRIENOL-GAMMA

169 EPICOCHLIOQUINONE-B- 203 CHAETOVIRIDIN-C 236 INDOMYCINONE-BETA 14 204 CYCLOGREGATIN 237 ISOTHIORBAMINE

170 FLURAZEPAM 205 FLURAZEPAM MONOHY 238 MEBEVERINE

171 HYDROXYFLUCLOXACIL- DROCHLORIDE 239 MUTISICOUMARANONE-D LIN 206 FUSIDILACTONE-B 240 NYMPHAEOL-C

172 RHINACANTHIN-C 207 GRISEOCHELIN-METHYL- 241 PRUSOCAIN

173 TEFLUBENZURON ESTER 242 ZIMET-20-84

174 XENYSALATE 208 ISOBUTYL-SHIKONIN 243 2, 4-D-BUTOXYPROPYL

175 ANTIMYCIN-A8A 209 MICINICATE 244 ABYSSINONE-IV

176 ARTOINDONESIANIN-U 210 AJUDAZOL-B 245 BAIGENE-A

177 BRONCHOCAINE 211 CYSTODYTIN-E 246 CHRYSOCHLAMIC-ACID

178 ENOLICAM 212 DEMETHYLPRAECANSONE- 247 EPOLONE-B

179 IPAZILIDE A 248 ETHOXYCAINE

180 MORDANT-BROWN-1 213 DESTRUXIN-A 249 FLUORENAMIL

181 PROPRANOLOL PHENO- 214 DIFENIDOL EMBONATE 250 GUAIACOL-MEFENAMATE BARBITAL 215 DISCOKIOLIDE-B 251 MAXIMA-ISOFLAVONE-C

182 PUGHI ININ-A 216 IRUMANOLIDE-2 252 SARCODICTYIN-B

183 AMPHIBINE-H 217 LACTOQUINOMYCIN 253 TRICLABENDAZOLE

184 ARMILLARIC-ACID 218 NEOBORNYVAL 254 ACHYOFURAN

185 CARINDACILLIN 219 SAROTHRALEN-D 255 ASTERRIQUINONE

186 CHLOROBIOCATE 220 ABYSSINONE-V 256 DIETHYLGLYCOLATE-

187 CHLORPROGUANIL 221 AXITIROME TOLYHYDRAZIDE

188 CYLINDROL-B 222 CHLORFLUAZURON 257 MALLOTOPHILIPPENS-D

189 ECLIPTALBINE 223 CHONDRILLIDIENE- 258 NAFTOXATE 259 PACHYDICTYOL-A- 291 PENTAMOXANE HYDRO 325 TRICHOPOLYN-I I EPOXIDE CHLORIDE 326 3-HYDROXYTOLUFAZEPAM

260 RATJADONE 292 SCANDENIN 327 DIMETPRAMIDE

261 SALVERINE HYDROCHLO 293 ACTINOPYRONE-C 328 FENOTEROL HYDROBRO RIDE 294 AMITON (als Bsp. filr MIDE

262 4-MENAHYDROQUINONE Gift ! ) 329 PHENKAPTON

263 BOTRYLLAMIDE-C 295 CRATOXYARBORENONE-C 330 CRATOXYARBORENONE-B

264 ELSAMICIN 296 CYMOPOL 331 ETHOMOXANE HYDRO

265 FENFLUTHRIN 297 DOXYCYCLINE-HYCLATE CHLORIDE

266 MUTISIPHENONE-A 298 FLAVIDULOL-C 332 ISOCENTRATHERIN

267 SANGGENON-A 299 FRAN-12 333 KALIMANTACIN-A

268 SCHWEINFURTHIN-A 300 MYRIAPORONE-3 334 HISPAGLABRIDIN-A

269 VANYLIDILOL 301 ORINOCINOLIDE 335 LANKACYCLINOL

270 4-O-METHYLMELLEOLIDE 302 TONABERSAT 336 MACLURAXANTHONE

271 ACETYLVISMIONE-F 303 VISMIONE-B 337 PIVAMPICILLIN HYDRO

272 ALFENTANIL- 304 AMIKHELLINE CHLORIDE

HYDROCHLORIDE 305 BUTAMOXANE HYDRO 338 PLURAFLAVIN-A

273 ASPERTETRONIN-A CHLORIDE 339 SIGMOIDIN-I

274 BETA- 306 CHLOROBIOCIN 340 FENALAMIDE

HYDROXYISOVALERYLSHIKONIN 307 CYCLOCOMMUNIN 341 HYDROXYACLACINOMY-

275 CARBISOCAINE 308 FENAZAFLOR CIN-M, 2

276 CHLORPROGUANIL HY 309 VISMIONE-D 342 SOPHORADIN

DROCHLORIDE 310 3-TRICHLORMETAPHOS 343 ASCOCHLORIN

277 CRYPTOPHYCIN-52 311 AMINOPROPYLONE PHE 344 BETA-HYDROXYSANSHOOL

278 DIDEMETHYL- NYLBUTAZONE 345 BISTRAMIDE-K TOCOTRIENOL 312 DIOCLENOL 346 CHLORTETRACYCLINE

279 FLUOSOL-DA 313 GRIFOLIN BORATE

280 PYRIDOXYPHEN 314 HYPERJOVINOL-A 347 DEHYDROASCOCHLORIN-

281 TIAZESIM HYDROCHLO 315 TAMOLARIZINE 8+ , 9+

RIDE 316 TOMENTOL 348 DIMETHIALIUM CHLO

282 BUTOCTAMIDE 317 TRANS-DELTA- RIDE

283 DEOXYSHIKONIN TOCOTRIENOLOIC-ACID 349 MACARANGIN

284 GAMBIERIC-ACID-A 318 DIACETOLOL HYDRO 350 MARCELLOMYCIN

285 LICOFURANONE CHLORIDE 351 BENZOYLGOMISIN-H

286 PREDNYLIDENE- 319 DUTOMYCIN 352 CLINDAMYCIN HYDRO

DIETHYLAMINOACETATE 320 LIGUROBUSTOSIDE-O CHLORIDE

287 PSEUDOPTEROSIN-G 321 RAISPAILOL-B 353 HYDROXYASCOCHLORIN-

288 TRIKENDIOL 322 ERECTQUIONE-A 8+

289 ZOAPATANOL 323 SAROTHRALEN-A 354 NEOCARZ ILIN-A

290 ERGOKININ-C 324 SITAMAQUINE 355 CHAETOVIRIDIN-B 356 CHLORTETRACYCLINE 411 Brequinar

BISULFATE database MDDR 412 Chlorproguanil

357 NAPYRADIOMYCIN-Cl 385 RUBIGINONE C2 413 E ricasan

358 SAROASPIDIN-B 386 FLOBUFEN 414 Enasidenib

359 SARRACINE 387 TETRONOTHIODIN 415 Enolica

360 ERECTONE-B 388 LAFLUNIMUS 416 Flurazepam

361 HOMOHARZIANIC-ACID 389 MYCAPEROXIDE A 417 ILX295501

362 ASTERRIQUINONE-B1 390 FLURAZEPAM HYDRO 418 Indibulin

363 DETAMID CHLORIDE 419 Metoclopramide

364 RUBRAXANTHONE 391 RUBIGINONE Cl 420 Mevastatin

365 DOLASTATIN-19 392 CRISNATOL MESYLATE 421 MK0686

366 EDETOL 393 DOLASTATIN D 422 Navarixin

367 PHASEOLLIDIN 394 EPOLACTAENE 423 Nefazodone

368 GEDOCARNIL 395 LEXIPAFANT 424 Pantoprazole

369 MANUMYCIN-B 425 Pavinetant

370 SANTALOL-BETA- database CHEMBL 426 SCYX-7158

SALICYLATE 396 CARBENICILLIN 427 Siccanin

371 COWANIN 397 DICLOFOP 428 Sulfoguanole

372 INDIBULIN 398 EPIPHENITICILLIN 429 Sunitinib

373 COWANOL 399 FLOBUFEN 430 Suvorexant

374 KARATAVICIN 400 GIGANTIC ACID 431 Tiapride

375 PICLOXYDINE 401 PHENETICILLIN POTASSIUM 432 Tonabersat

376 PROXAZOLE 402 DIPHENICILLIN 433 Ulimorelin

377 STROBILURIN-E 434 Xipairmide

378 ERECTQUIONE-B database Pubchem 435 MGGBYMDAPCCKCT-

379 SURICAINIDE 403 Acotia ide UHFFFAOYSA-N

380 DIETHYLAMINOMETHYL- 404 Acoziborole 436 VNBRGSXVFBYQNN- RUTIN 405 Acu api od UHFFFAOYSA-N

381 TROPESIN 406 Apaluta ide 437 YUHNXUAATAMVKD-

382 PICLOXYDINE HYDRO 407 ASP3026 PZUWPPBQSA-N

CHLORIDE 408 AZD1480 438 LMI070

383 HEX-1 409 BIIB021 439 Erlotinib

384 DECLOVANILLOBIOCIN 410 Branpla 440 Tanespimycin

The capacity of some of these ligands for inducing conforma- tional switching of RBP4 is detected by exploiting the confor- matron dependent binding of TTR, which is inhibited by ligands with capacity for inducing conformational switching of RBP4. Flow cytometric detection of ligand induced conformational switching of RBP4

RBP4-encoding DNA is isolated out of P.pastoris using the zymoprep yeast plasmid miniprep kit II (Zymo Research) followed by amplification by PCR using the primers picZalpha_fwd (5'- ATTGCCAGCATTGCTGCTAAAGAAG-3 ' ) and picZalpha_rev (5'- GCAAATGGCATTCTGACATCC-3 ' ) . The DNA is purified using the illus- tra GFX PCR DNA and Gel Band Purification Kit (GE healthcare) and sequenced. For cloning into the vector pCTCon2, the RBP4- encoding DNA is amplified with PCR using primers encoding Nhel/BamHI restriction sites homologous to the vector. After di gestion with Nhel and BamHI (both New England Biolabs) , the RBP4 encoding sequence is ligated into the linearized vector and electroporated in E.coli for verification of the sequence. The different domains of the resulting fusion protein have the order Aga2p - HA-tag - (GlySer Linker - RBP4 - c-myc Tag. S.cerevisiae strain EBY100 is then transformed with the Quick and easy transformation kit (Takara) .

Expression of RBP4 on yeast is induced by culturing the cells in SG-CAA medium (2 g/L glucose, 20 g/L galactose, 6.7 g/L yeast nitrogen base, 5 g/L casamino acids, 10,2 g/L disodium hy drogen phosphate and 4,82 g/L sodium phosphate monobasic) over night at 20°C and 180 rpm. For detection of RBP4 expression on the surface of yeast cells, cells are stained with either mouse anti-c-myc antibody 9E10 (Thermo Fisher Scientific) followed by secondary staining with anti-mIgG-AF488 (Thermo Fisher Scien tific) or with anti-HA-Alexa Fluor 647 (BioLegend) .

Flow cytometric detection of ligand induced conformational switching of RBP4 is based on employing the conformation depend ent binding behaviour of TTR. Hereto, recombinant TTR (labelled with Alexa Fluor 488 NHS ester kit, Thermo Fisher Scientific) is added to RBP4-expressing yeast in absence or presence of differ ent concentrations of various small molecule ligands in the staining buffer (PBS supplemented with 0.1% BSA) .

Example 3: Integration of an RBP4-based LRPPI system into a CAR

The third example demonstrates the applicability of a LCN- fold based LRPPI for regulation of CAR function by using the RBP4-based LRPPI system described in example 1. The schematics of the CAR constructs shown in Figures 6A, 6B and 9 illustrate some tested possibilities of integrating such a lipocalin-fold based LRPPI into a CAR. Figures 7A, 7B, 7C, 7D and 10 show the sequences of the tested constructs and Figures 8A, 8B and 8C il lustrates the expression of signaling CAR constructs in primary human T cells. Primary human T cells were electroporated with 5 yg mRNA for each construct and CAR expression was detected 20 h after electroporation via Strep II Tag or in the case of "RS3 CAR long without c-myc Tag" and "RF2 long CAR" via the Fc domain by using a biotinylated anti-human-IgGl-antibody as primary an tibody and a PE-conjugated streptavidin as secondary staining reagent. Figures 11A and 11B show the binding of CAR constructs containing the antigen binding domain (fusion proteins A-D, de tected via the integrated His Tag) to target cells and to prima ry human T cells expressing different transmembrane CAR signal ing constructs in the presence of 5 mM A1120 (20 h after elec troporation of 5 yg mRNA) . The fusion proteins were either co expressed in the primary T cells or added to the primary T cells as supernatants obtained form Jurkat cells that expressed the fusion proteins. The ligand-dependent function of the resulting CARs, as determined by induction of cytotoxicity and cytokine producinon of primary human T cells in absence and presence of 5 mM A1120, is shown in Figures 12 and 13.

Design of CARs

Selected rcSso7d- and FN3-based RBP4-binding interaction partners were incorporated into a second-generation CAR signal ing backbone, comprising a CD8 stalk, a 4-1BB costimulatory do main and a CD3zeta signaling domain ("8a-BBz") , or an IgG-Fc spacer fused to a CD28 costimulatory domain and a CD3zeta sig naling domain ("Fc-28z") , respectively. RBP4-binding interaction partners based on rcSso7d (RS1-RS5) and Fibronectin (RF1-RF3) were fused to the CAR backbones via a Strep-Tag and one or two repeats of G4S (4x glycine and lx serine) amino acid residues as linker. The composition of the constructs is given in the sche matics of Figure 6A and 6B and the sequences are shown in Figure 7A, 7B, 7C and 7D. The secreted antigen binding CAR constructs contained either a FMC63-based scFv directed against human CD19 or rcSso7d directed against human EGFR fused to either IgGl-Fc- RBP4, or RBP4 or alternatively a lipocalin-fold binding interac tion partner rcSso7d RS3. The schematics and sequences are shown in Fig. 9 and 10, respectively. CAR constructs were constructed by Gibson Assembly (NEB), using 0.02-0.5 pmol of 2-3 PCR- amplified DNA fragments and 0.2-1 pmol of 4-6 PCR-amplified DNA fragments for assembly, respectively. The resulting constructs were amplified by PCR and subsequently used for in vitro tran scription .

In vitro transcription and electroporation of mRNA

In vitro transcription was performed with 50-200 ng of puri fied PCR product using the mMessage mMachine T7 Ultra Kit (Ambi- on) according to the manufacturer's instructions. The resulting mRNAs were column purified with an adapted protocol using the RNeasy Kit (Qiagen) . According to this protocol, RLT buffer from the kit and 1% beta-mercaptoethanol were added followed by addi tion of absolute ethanol. The mixture was loaded onto an RNeasy column and purification was performed following the manufactur er's protocol. Purified mRNAs were frozen at -80°C until elec troporation. For CAR expression, Jurkat cells and primary T cells were electroporated with 5 or 10 yg mRNA using the Gene Pulser (Biorad) (square wave protocol, 500 V, 5 ms, 4 mm cu vettes) .

Detection of the expression of signaling CAR constructs

Expression of rcSso7d- and FN3-based CARs was detected via the Strep-II Tag using an anti-Strep-II Tag antibody (clone 5A9F9, GenScript) as primary antibody and a PE-conjugated sec ondary antibody (eBioscience) . CARs containing an IgG spacer were detected by using a biotinylated anti-human-IgGl-antibody as primary antibody and a PE-conjugated streptavidin as second ary staining reagent. Finally, cells were analyzed by flow cy tometry .

Detection of binding of the antigen binding CAR constructs fu sion proteins A-D

Binding of fusion proteins A-D, which were either co expressed with the signaling CAR constructs or produced by Jurkat cells, was detected by flow cytometry. Accordingly, 25xl0 ³ cells were either first incubated with fusion protein-containing supernatant for 45 min at 4°C in the presence of 5 mM A1120 ( Sigma-Aldrich) and thereafter washed in the presence of 5 mM A1120 before the second step staining with anti-penta-His anti body (Qiagen) , or were directly stained with the anti penta-His antibody for 25 min at 4°C. Before flow cytometric analysis, two washing steps were performed again in the presence of A1120.

Cytotoxicity assay

Analysis of the cytotoxic potential of CAR T cells was per formed by a luciferase-based cytotoxicity assay. Therefore, lu- ciferase-expressing tumor target cells were co-cultured with CAR T cells at different effector : target cell ratios (1:1, 2:1, 5:1, 10:1, 25:1, 100:1) in round-bottom 96 well plates for 4 h or 24 h at 37°C. Remaining living cells were then quantified by addi tion of luciferin (150 yg/mL final concentration; Perkin Elmer) and luciferase activity was measured by using the ENSPIRE Multi- mode plate reader. The percentage of specific lysis was deter mined with the following formula:

% killing = 100 - ( (RLU from well with effector and target cell co-culture) / (RLU from well with target cells only) x 100)) .

Cytokine release by CAR T cells and CAR-expressing Jurkat cells

Secretion of cytokines into supernatants was assessed by co culturing the target cells with effector cells at effec- tor:target (E:T) ratios of 1:1 or 2:1 in flat-bottom 96 well plates for 4 or 24 h at 37°C. The supernatants were centrifuged (1600 rpm, 7 min) to remove remaining cells and were frozen at - 20°C. For analysis of secreted cytokines, IFN-y ELISA was per formed using the Human IFN gamma ELISA Ready-SET-Go ! ^® (eBiosci- ence) according to the manufacturer's instructions. Secreted IL- 8 was quantified by IL-8 Ready-SET-Go ! ^® ELISA (Thermo Fisher Sci entific) . Measurements were conducted using the ENSPIRE Multi- mode plate reader.

Example 4 : Biophysical characterization of RBP4 binding interaction partners

The RBP4 binding interaction partners of example 1 (in this case the lipocalin-fold binding interaction partners) were fur ther characterized with different biophysical methods. DSC anal ysis clearly demonstrated that the RBP4 binding interaction partners RS3, RS5 and RF2 are stable proteins. Moreover, SEC analysis showed that the RBP4 binding interaction partners RS3, RS5 and RF2 are well folded, monomeric proteins.

Furthermore, the affinities between RBP4 binding interaction partners (RS3, RS5 and RF2) and RBP4 were analyzed by isothermal titration calorimetry (ITC) and surface plasmon resonance (SPR) . These measurements clearly demonstrated that the affinities of those lipocalin-fold binding interaction partners (in this case RS3, RS5 and RF2) to the lipocalin-fold molecule (in this case RBP4) strongly depend on the presence of the lipocalin-fold lig and (in this case A1120) . Figs. 15, 17 and 18 show that addition of the lipocalin-fold ligand A1120 increases the affinity be tween those lipocalin-fold binding interaction partners and the lipocalin-fold molecule RPB4 by up to several hundred-fold.

Characterization of RBP4 binding interaction partners by DSC

The thermal stability of RBP4 binding interaction partners was evaluated by determining the melting temperature (T _m) by dif ferential scanning calorimetry (DSC) using the PEAQ Differential Scanning Calorimeter Automated (Malvern Panalytical) . Briefly, 80 mM protein solution was exposed to an increasing temperature ranging from 20-110°C with a heating rate of l°C/min. Buffer baseline was subtracted and the resulting data were normalized for protein concentration, followed by fitting with a non-two- state thermal unfolding model. Figure 14 shows the determined T _m values for the RBP4 binding interaction partners RS3, RS5 and RF2 (mean of 3 independent experiments +/- s.d.) .

Characterization of the interaction between RBP4 binding interaction partners and RBP4 by SPR

Surface plasmon resonance (SPR) was used to determine the affinity of the interaction between RBP4 and rcSso7d- or FN3- based lipocalin-fold binding interaction partners and was per formed with BiacoreT200 (GE Healthcare) . The antigen (RBP4) was coated onto a CM5 chip (GE Healthcare) and incubated with vari ous concentrations of lipocalin-fold binding interaction partner solution in HBS-EP (GE Healthcare) , followed by a dissociation phase in HBS-EP. All measurements were performed both in the presence (5 mM) and absence of A1120. Finally, K _d values (disso ciation constants) were obtained by steady-state analysis. Fig- ure 15A shows a single-cycle kinetics (SCK) SPR experiment with RBP4 immobilized on a sensor chip and titrated with the lipocalin-fold binding interaction partner RS3 in the presence of 5 mM A1120. Figure 15B shows an SCK experiment in the absence of A1120 with the same RS3 concentrations used in (A) . Since al most no signal could be observed with these concentrations, they were elevated. Figure 15C shows an SCK experiment with higher RS3 concentrations in the absence of A1120. i¾ values were calcu lated by steady-state analysis (right diagram in A and C) . One representative experiment out of four independent experiments is shown .

Characterization of RBP4 binding interaction partners by SEC

For analytical SEC analysis, a total amount of 25 yg of RBP4 binding interaction partners in running buffer (PBS with 200 mM NaCl) was filtered through a 0.1 ym Ultrafree-MC filter (Merck Millipore) , applied to a Superdex 200 10/300 GL column (GE healthcare) connected to an HPLC Prominence LC20 system (Shimad- zu) and eluted with a flow rate of 0.75 mL/min at 25°C. Figure 16 shows the elution profiles of 6 different RBP4 binding inter action partners (RSI, RS3, RS5, RF1, RF2 and RF3) .

Characterization of the interaction between RBP4 binding interaction partners and RBP4 by ITC

For the thermodynamic characterization of the interaction between RBP4 and rcSso7d- or FN3-based lipocalin-fold binding interaction partners (in this case RBP4 binding interaction partners), isothermal titration calorimetry (ITC) was conducted using the PEAQ Isothermal Titration Calorimeter Automated (Mal vern Panalytical) . The samples were centrifuged (17000 g, 10 min, 20°C) and dialysed against the same buffer (PBS) . RBP4 pro tein solution was applied to the sample cell and the respective RBP4 binding interaction partners were titrated with varying in tervals. Different concentrations both for RBP4 and RBP4 binding interaction partners were used. The experiments were performed both in the presence and absence of the lipocalin-fold ligand A1120. In the experiments where A1120 was present, the ligand was added both to the RBP4 protein and RBP4 binding interaction partner solution to achieve the same buffer compositions. Data analysis was performed with the MicroCal PEAQ-ITC analysis soft- ware and the resulting data were fitted to a one-set-of-sites binding model. Figure 17A shows the analysis of the interaction of one RBP4 binding interaction partner (RS3) with RBP4 in the presence (50 mM) and absence of A1120 as determined by ITC. The integrated data were fitted to a one-set-of-sites binding model to calculate the i¾ values shown in Figure 18. The signature plots show the thermodynamic parameters of the interaction be tween RBP4 and RS3. One representative experiment out of four independent experiments is shown. Figure 17B shows binding of the RBP4 binding interaction partners RS5 and RF2 to RBP4 in the presence (50 mM) and absence of A1120. The integrated data were fitted to one-set-of-sites binding model to calculate the i¾ val ues in Figure 18. Signature plots show the thermodynamic parame ters of the interaction. One representative experiment out of four independent experiments is shown.

Figure 18 summarizes the affinity measurements for RBP4 and the RBP4 binding interaction partners RS3, RS5 and RF2 by yeast surface display (n=3) , SPR (n=4) and ITC (n=4) . Mean values +/- s.d. are shown.

Example 5: Screening of clinically applicable lipocalin-fold ligands with capacity to induce conformational switching in a lipocalin-fold molecule

The capacity of some of the ligands described in example 2 to induce conformational switching of RBP4 was detected by ex ploiting the conformation-dependent binding of TTR to RBP4. That is, if a lipocalin-fold ligand induces a conformational switch in RBP4, this can result in reduction (or elevation) of TTR af finity to RBP4, which can be detected by flow cytometric analy sis of TTR binding.

Subcloning of RBP4 into the vector pCTCON2 and yeast transformation

RBP4-encoding DNA was isolated from P. pastoris using the zymoprep yeast plasmid miniprep kit II (Zymo Research) followed by PCR amplification using the primers picZalpha_fwd (5"- ATT- GCCAGCATTGCTGCTAAAGAAG-3 ' ) and picZalpha_rev (5'- GCAAATGGCATTCTGACATCC-3 ' ) . The DNA was purified using the Illus- tra GFX PCR DNA and Gel Band Purification Kit (GE healthcare) and sequenced. For cloning into the vector pCTCON2, the RBP4- encoding DNA was amplified by PCR using primers encoding Nhel/BamHI restriction sites. After digestion with Nhel and Bam- HI (both New England Biolabs) , the RBP4 encoding sequence was ligated into the linearized vector and E.coli cells were elec troporated with the resulting construct for verification of the sequence. The different parts of the resulting fusion protein have the order Aga2p - HA-tag - (GlySer Linker - RBP4 - c-myc Tag. S. cerevisiae strain EBY100 was then transformed with the sequence-verified construct using the Quick and easy transfor mation kit (Takara) .

Expression of RBP4 on the surface of yeast was induced by culturing the cells in SG-CAA medium (2 g/L glucose, 20 g/L ga lactose, 6.7 g/L yeast nitrogen base, 5 g/L casamino acids, 10,2 g/L disodium hydrogen phosphate and 4,82 g/L sodium phosphate monobasic) overnight at 20°C and 180 rpm. For detection of RBP4 expression on the surface of yeast cells, cells were stained with either mouse anti-c-myc antibody 9E10 (Thermo Fisher Scien tific) followed by secondary staining with anti-mouse IgG-AF488 (Thermo Fisher Scientific) or with anti-HA-Alexa Fluor 647 (Bio- Legend) .

Expression and purification of TTR

Human StrepII-tagged TTR was cloned into the pET52b+ vector and E. coli pLys cells were transformed with the construct. For the expression, LB medium (consisting of lOg/L peptone, 5 g/L yeast extract and 5 g/L sodium chloride) containing ampicillin (100 pg/mL) was inoculated with E. coli cells containing the se quence-verified plasmid and cultivated overnight at 37°C while shaking (180 rpm) . Cells were diluted on the following morning and further cultivated until an Oϋboo of 1 was reached. Induction of protein expression was performed by addition of IPTG (Isopro- pyl^-D-thiogalactopyranoside, 1 mM) . In addition, the tempera ture was decreased to 20°C and cells were cultivated overnight at 180 rpm. Cells were harvested by centrifugation (5000 g, 15 min) , the pellet was resuspended in lysis buffer (PBS supple mented with 150 mM NaCl, 10 mM imidazole, 5 mM b- mercaptoethanol , pH 7.4) and the protease inhibitor PMSF (Phe- nylmethylsulfonylfluorid, 1 mM) was added to prevent protein degradation. Cells were lysed by sonication (frequency 50 kHz, 2 times 90 s, amplitude: 5) using the Vibra Cell sonicator (Son- ics&Materials Inc.) followed by centrifugation (10 000 g, 20 min) . The supernatant was filtered through a 0.45 pm filter (Merck Millipore) . A StrepTrap HP column (GE Healthcare) con nected to an AKTA FPLC purifier system was equilibrated with binding buffer (PBS supplemented with 150 mM NaCl, pH 7.4) and the sample was applied with a flow rate of 1 mL/min. After wash ing with binding buffer, elution was performed with elution buffer (binding buffer with 2.5 mM desthiobiotin) and the frac tions were analysed by NanoDrop and SDS-PAGE. Those fractions showing both absorbance at 280 nm and the presence of a protein band at the expected size were pooled and concentrated using Amicon Ultra-15 10K Centrifugal filters (Merck Millipore) .

Flow cytometric detection of lipocalin-fold ligand-induced conformational switching of RBP4

Flow cytometric detection of lipocalin-fold ligand induced conformational switching of RBP4 is based on employing the de pendence of the RBP4-TTR interaction on the RBP4-conformation . For this purpose, lxlO ⁶ yeast cells displaying RBP4 were incubat ed with PBSA (PBS supplemented with 0.1% BSA) containing 100 mM of the potential lipocalin-fold ligand and incubated for 20 min at 4°C while shaking. Afterwards, different concentrations of purified TTR (ranging between 0 and 1000 nM) were added in pres ence (50 mM) or absence of the potential lipocalin-fold ligand and incubated for 1 h at 4°C while shaking. Two washing steps were performed followed by incubation (25 min at 4°C) with a bi otinylated anti-Strep II antibody (Genscript) , which recognizes the StrepII-tagged TTR. After two washing steps, tertiary stain ing was performed using Streptavidin-PE (Biolegend) and anti-HA- Alexa Fluor 488 (Biolegend) and incubation for 25 min at 4°C. After two final washing steps, cells were kept as a pellet and resuspended in PBSA just before flow cytometric analysis. Figure 19A shows dotplots from one representative experiment with stained yeast cells displaying RBP4 and binding to TTR. Figure 19B shows single measurements of binding of TTR (Median fluores cence intensity) to RBP4 displayed on the surface of yeast in the presence of different potential lipocalin-fold ligands (50 mM) or in the absence of any lipocalin-fold ligand (PBSA) .

Figure 20 shows binding of TTR (Median fluorescence inten- sitiy) to RBP4 displayed on the surface of yeast in the presence of various potential lipocalin-fold ligands (50 mM) or in the absence of any lipocalin-fold ligand (PBSA) at two different concentrations of TTR (100 and 300 nM) . Mean values of tripli cate measurements +/- s.d. are shown.

Accordingly, the present invention discloses the following preferred embodiments:

1. A ligand regulated protein-protein interaction system based on a lipocalin-fold molecule comprising:

(a) a lipocalin-fold molecule

(b) a lipocalin-fold ligand with a low molecular weight of 1500 Da or below, and

(c) a lipocalin-fold binding interaction partner,

wherein the lipocalin-fold molecule can bind to the lipocalin- fold ligand; and

wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand binds to the lipocalin-fold binding interaction partner with an affinity which is at least 10-fold higher than the af finity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand,

and wherein the lipocalin-fold binding interaction partner is not a naturally occurring protein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule in the presence of any lipocalin-fold ligand.

2. A ligand regulated protein-protein interaction system based on a lipocalin-fold molecule comprising:

(a) a lipocalin-fold molecule

(b) a lipocalin-fold ligand with a low molecular weight of 1500 Da or below, and

(c) a lipocalin-fold binding interaction partner,

wherein the lipocalin-fold molecule has at least a first confor mation when the lipocalin-fold ligand is not bound to the lipocalin-fold molecule and at least a second conformation when the lipocalin-fold ligand is bound to the lipocalin-fold mole cule; and

wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand in the second conformation binds to the lipocalin-fold binding interaction partner with an affinity which is at least 10-fold higher than the affinity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand in the first confor mation,

and wherein the lipocalin-fold binding interaction partner is not a naturally occurring protein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule in the presence of any lipocalin-fold ligand.

3. A ligand regulated protein-protein interaction system accord ing to embodiment 1 or 2, wherein the lipocalin-fold binding in teraction partner does not contain any segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any segment of a naturally occurring protein which has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule, especially in the presence of a lipocalin-fold ligand.

4. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 3, wherein the affinity of the lipocalin-fold binding interaction partner to the lipocalin-fold molecule when bound to the lipocalin-fold ligand or in the second conformation, respectively, is below 10 mM, preferably below 2 mM, especially below 400 nM.

5. A ligand regulated protein-protein interaction system accord ing to any one of embodiments 1 to 4, wherein the lipocalin-fold binding interaction partner does not contain any segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any segment of a naturally occurring protein and wherein this segment of at least 50 consecutive amino acids has an affinity of <400 nM to any naturally occurring lipocalin-fold molecule, especially in the presence of a lipocalin-fold ligand.

6. A ligand regulated protein-protein interaction system accord ing to any one of embodiments 1 to 4, wherein the lipocalin-fold binding interaction partner does not contain any segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any segment of a naturally occurring protein and wherein this segment of at least 50 consecutive amino acids has an affinity of <2 mM to any naturally occurring lipocalin-fold molecule, es pecially in the presence of a lipocalin-fold ligand.

7. A ligand regulated protein-protein interaction system accord ing to any one of embodiments 1 to 4, wherein the lipocalin-fold binding interaction partner does not contain any segment of at least 50 consecutive amino acids with an amino acid sequence that is at least 98% identical with the amino acid sequence of any segment of a naturally occurring protein and wherein this segment of at least 50 consecutive amino acids has an affinity of <10 mM to any naturally occurring lipocalin-fold molecule, especially in the presence of a lipocalin-fold ligand.

8. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 7, wherein the lipocalin- fold binding interaction partner is not a homolog of a different species than human of a naturally occurring human lipocalin-fold binding interaction partner.

9. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 8, wherein the lipocalin- fold molecule is an artificial lipocalin-fold molecule which has no natural counterpart.

10. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 9, wherein the lipocalin- fold binding interaction partner is engineered to specifically recognize the lipocalin-fold molecule with higher affinity if the lipocalin-fold molecule is bound to the lipocalin-fold lig and compared to the affinity to the lipocalin-fold molecule not bound to the lipocalin-fold ligand, and wherein the lipocalin- fold molecule is optionally engineered for binding with higher affinity to the lipocalin-fold ligand.

11. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 10, wherein the lipocalin-fold molecule is not a homolog of a different species than human of a naturally occurring human lipocalin-fold mole cule . 12. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 11, wherein the lipocalin-fold ligand is a pharmaceutically active molecule, es pecially a pharmaceutically active molecule with a therapeutic activity in human patients.

13. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 12, wherein the lipocalin-fold ligand is a molecule which can be effectively ad ministered orally.

14. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 12, wherein the lipocalin-fold ligand is a molecule which can be effectively ad ministered intravenously.

15. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 14, wherein the lipocalin-fold ligand is a molecule which can be effectively ad ministered intravenously and/or orally to a human patient.

16. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 15, wherein the lipocalin-fold binding interaction partner does not contain a domain of a naturally occurring protein that mediates binding to a naturally occurring lipocalin-fold molecule with an affinity of <10 mM, especially in the presence of a lipocalin-fold lig and .

17. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 16, wherein the lipocalin-fold molecule is a molecule identical with a naturally occurring iLBP (intracellular lipid binding protein) , a natural ly occurring lipocalin or an anticalin, or is a derivative of any of these molecules with 1-30 amino acid exchanges and/or 1- 50 amino acid deletions and/or 1-50 amino acid insertions.

18. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 17, wherein the lipocalin-fold molecule is a derivative of a naturally occurring or otherwise disclosed lipocalin-fold molecule with at least 70%, preferably at least 80%, especially at least 90% sequence identity in the b-barrel structure.

19. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 18, wherein the lipocalin-fold molecule is a derivative of a naturally occurring lipocalin or iLBP with at least one, two, three, four, five, six, seven, eight, nine, ten, 25 or 30 amino acid exchanges.

20. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 19, wherein the lipocalin-fold molecule is a derivative of a naturally occurring lipocalin or iLBP with up to 15, up to 30, or up to 50 amino ac id deletions and/or up to 15, up to 30, or up to 50 amino acid insertions outside of the structurally conserved b-barrel struc ture, preferably corresponding structurally to the regions of amino acid residues selected from

amino acid residues 1-20, 31-40, 48-51, 59-70, 79-84, 89- 101, 110-113, 121-131 and 139-183 in human RBP4, which define the regions adjoining the structurally conserved b-strands in human RBP4 according to the amino acid residue numbering scheme in the PDB entry 1RBP;

amino acid residues 1-13, 24-36, 44-47, 55-61, 70-75, 80-83, 92-95, 103-110 and 118-158 in human TLC according to the amino acid residue numbering scheme in Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 , which define the regions adjoining the structurally conserved b-strands in human TLC;

amino acid residues 1-43, 54-68, 76-80, 88-95, 104-109, 114- 118, 127-130, 138-141 and 149-188 in human ApoM according to the amino acid residue numbering scheme in Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 , which define the regions adjoining the structurally conserved b-strands in human ApoM;

amino acid residues 1-4, 13-40, 46-49, 55-60, 66-70, 74-80, 88-92, 97-107, 113-118, 125-128 and 136-137 in human CRABPII ac cording to the amino acid residue numbering scheme in PDB entry 2FS6, which define the regions adjoining the structurally con served b-strands in human CRABPII;

amino acid residues 1-4, 13-38, 44-47, 53-58, 64-68, 72-78, 86-90, 95-98, 104-108, 115-118 and 126-127 in human FABP1 ac- cording to the amino acid residue numbering scheme in PDB entry 2F73, which define the regions adjoining the structurally con served b-strands in human FABP1.

21. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 20, wherein the lipocalin-fold molecule is a derivative of a naturally occurring lipocalin or iLBP with at least 70%, preferably at least 80%, especially at least 90% sequence identity in the b-barrel struc ture, whereby this b-barrel structure is defined as the regions preferably corresponding structurally to the regions of amino acid residues selected from

- amino acid residues 21-30, 41-47, 52-58, 71-78, 85-88, 102-

109, 114-120 and 132-138 in human RBP4 according to the amino acid residue numbering scheme in the PDB entry 1RBP, which de fine the structurally conserved b-strands in human RBP4;

- amino acid residues 14-23, 37-43, 48-54, 62-69, 76-79, 84-91,

96-102 and 111-117 in human tear lipocalin (TLC) as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 , which define the structurally conserved b-strands in human TLC;

- amino acid residues 44-53, 69-75, 81-87, 96-103, 110-113, 119-

126, 131-137 and 142-148 in human apolipoprotein M (ApoM) as de fined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985, which define the structurally conserved b-strands in human ApoM;

- amino acid residues 5-12, 41-45, 50-54, 61-65, 71-73, 81-87,

93-96, 108-112, 119-124 and 129-135 in human cellular retinoic acid binding protein II (CRABPII) according to the amino acid residue numbering scheme in PDB entry 2FS6, which define the structurally conserved b-strands in human CRABPII;

- amino acid residues 5-12, 39-43, 48-52, 59-63, 69-71, 79-85,

91-94, 99-103, 109-114 and 119-125 in human fatty acid binding protein 1 (FABP1) according to the amino acid residue numbering scheme in PDB entry 2F73, which define the structurally con served b-strands in human FABP1;

22. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 21, wherein the lipocalin-fold molecule is a fragment of a naturally occurring lipocalin or a derivative thereof with a length of at least 80, preferably at least 100, especially at least 120, amino acids covering at least the structurally conserved b-barrel structure of the lipocalin-fold, or wherein the lipocalin-fold molecule is a fragment of a naturally occurring iLBP or a derivative thereof with a length of at least 80, preferably at least 85, especially at least 90, amino acids covering at least the structurally con served b-barrel structure of the lipocalin-fold,

wherein the structurally conserved b-barrel structure comprises or consists of amino acid positions preferably corresponding structurally to the regions of amino acid residues selected from

- amino acid residues 21-30, 41-47, 52-58, 71-78, 85-88, 102- 109, 114-120 and 132-138 in human RBP4 according to the amino acid residue numbering scheme in the PDB entry 1RBP, which de fine the structurally conserved b-strands in human RBP4;

- amino acid residues 14-23, 37-43, 48-54, 62-69, 76-79, 84-91, 96-102 and 111-117 in human tear lipocalin (TLC) as defined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985 , which define the structurally conserved b-strands in human TLC;

- amino acid residues 44-53, 69-75, 81-87, 96-103, 110-113, 119- 126, 131-137 and 142-148 in human apolipoprotein M (ApoM) as de fined by Schiefner et al . , Acc Chem Res. 2015 ; 48 ( 4 ) : 976- 985, which define the structurally conserved b-strands in human ApoM;

- amino acid residues 5-12, 41-45, 50-54, 61-65, 71-73, 81-87, 93-96, 108-112, 119-124 and 129-135 in human cellular retinoic acid binding protein II (CRABPII) according to the amino acid residue numbering scheme in PDB entry 2FS6, which define the structurally conserved b-strands in human CRABPII;

- amino acid residues 5-12, 39-43, 48-52, 59-63, 69-71, 79-85, 91-94, 99-103, 109-114 and 119-125 in human fatty acid binding protein 1 (FABP1) according to the amino acid residue numbering scheme in PDB entry 2F73, which define the structurally con served b-strands in human FABP1;

23. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 22, wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand binds to the lipocalin-fold binding interaction partner with an affin ity which is at least 20-fold higher, preferably at least 50- fold higher, than the affinity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand. 24. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 23, wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand binds to the lipocalin-fold binding interaction partner with an affin ity which is at least 100-fold higher, preferably at least 200- fold higher, especially at least 500-fold higher, than the af finity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand.

25. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 24, wherein the lipocalin-fold molecule bound to the lipocalin-fold ligand binds to the lipocalin-fold binding interaction partner with an affin ity which is at least 1000-fold higher than the affinity of the lipocalin-fold molecule not bound to the lipocalin-fold ligand.

26. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 25, wherein the lipocalin-fold ligand has a molecular weight of 1500 to 75 Da, preferably of 750 Da to 150 Da.

27. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 26, wherein the lipocalin-fold binding interaction partner is or wherein the lipocalin-fold molecule and/or the lipocalin-fold binding inter action partner comprises an antigen, a cell surface receptor, an antibody, an antibody fragment, or a non-antibody based scaf fold, preferably an affibody, a lipocalin-fold molecule, prefer ably an iLPB or a LCN, especially an anticalin; an avimer, a DARPin, a fynomer, a Kunitz domain, a knottin, a monobody, a Sso7d-based binder, reduced charge Sso7d (rcSso7d) -based binder or Sac7d-based binder.

28. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 27, wherein the lipocalin-fold binding interaction partner and/or the lipocalin- fold molecule comprises an antigen, a cell surface receptor, an antibody, an antibody fragment, or a non-antibody based scaf fold, preferably an affibody, a lipocalin-fold molecule, prefer ably an iLPB or a LCN, especially an anticalin; an avimer, a DARPin, a fynomer, a Kunitz domain, a knottin, a monobody, a Sso7d-based binder, reduced charge Sso7d (rcSso7d) -based binder or Sac7d-based binder.

29. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 28, wherein the lipocalin-fold ligand has an affinity to the lipocalin-fold mol ecule of below 1 mM, preferably of below 100 mM, especially of below 10 mM.

30. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 29, wherein the lipocalin-fold ligand is a ligand selected from Table 1, espe cially fenretinide (15- [ (4-hydroxyphenyl) amino] retinal) , N- Ethylretinamide (PubChem CID: 5288173), all-trans retinoic acid (PubChem CID: 444795), axerophthene (PubChem CID: 5287722), A1120 (PubChem CID 25138295) and derivatives thereof, 1,4- butanediol (Pubchem CID: 8064), sphingosine-l-phosphate (Pubchem CID: 5283560), tetradecanoic acid (Pubchem CID: 11005), in- dicaxanthin (Pubchem CID: 6096870 and 12310796), vulgaxanthin I (Pubchem CID: 5281217), Montelukast (Pubchem CID: 5281040), Cyclandelate (Pubchem CID: 2893), Oxolamine (Pubchem CID: 13738), Mazaticol (PubchemCID: 4019), Butoctamid (Pubchem CID: 65780), Tonabersat (Pubchem CID: 6918324), Novazin (Pubchem CID: 65734), Diphenidol (Pubchem CID: 3055), Neobornyval, Erlotinib (Pubchem CID: 92131336), Tanespimycin (Pubchem CID: 6505803), LMI070 (Pubchem CID: 85471316), Alloclamide (Pubchem CID: 71837), Diacetolol (Pubchem CID: 50894), Acotiamide (Pubchem CID: 5282338), Acoziborole (Pubchem CID: 44178354), Acumapimod (Pubchem CID: 11338127), Apalutamide (Pubchem CID: 24872560), ASP3026 (Pubchem CID: 25134326), AZD1480 (Pubchem CID: 16659841), BIIB021 (Pubchem CID: 16736529), Branaplam (Pubchem CID: 89971189), Brequinar (Pubchem CID: 57030), Chlorproguanil (Pubchem CID: 9571037), Clindamycin (Pubchem CID: 446598), Emricasan (Pubchem CID: 12000240), Enasidenib (Pubchem CID: 89683805), Enolicam (Pubchem CID: 54679203), Flurazepam (Pubchem CID: 3393), ILX-295501 (Pubchem CID: 127737), Indibulin (Pubchem CID: 2929), Metoclopramide (Pubchem CID: 12598248), Mevastatin (Pubchem CID: 64715), MGGBYMDAPCCKCT-UHFFFAOYSA-N (Pubchem CID: 25134326), MK0686 (Pubchem CID: 16102897), Navarixin (Pubchem CID: 71587743), Nefazodone hydrochloride (Pubchem CID: 54911), Pantoprazole (Pubchem CID: 4679), Pavinetant (Pubchem CID: 23649245), Proxazole (Pubchem CID: 8590) , Siccanin (Pubchem CID: 71902), Sulfaguanole (Pubchem CID: 9571041), Sunitinib (Pubchem CID: 5329102), Suvorexant (Pubchem CID: 24965990), Tiapride (Pubchem CID: 5467), Tonabersat (Pubchem CID: 6918324), VNBRGSXVFBYQNN-UHFFFAOYSA-N (Pubchem CID: 24794418), YUHNXU- AATAMVKD-PZJWPPBQSA-N (Pubchem CID: 44548240), Ulimorelin (Pub chem CID: 11526696), Xipamide (Pubchem CID: 26618), Tropesin (Pubchem CID: 47530), Triclabendazole (Pubchem CID: 50248), Triclabendazole sulfoxide (Pubchem CID: 127657), Triclabendazole sulfone (Pubchem CID: 10340439) and Trametinib (Pubchem CID: 11707110) .

31. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 30, wherein the lipocalin-fold molecule is part of an ectodomain of a chimeric antigen receptor and wherein the lipocalin-fold binding interac tion partner is a cell surface antigen.

32. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 30, wherein the lipocalin-fold molecule and the lipocalin-fold binding interac tion partner are parts of intra- and/or extracellular domains of a chimeric antigen receptor.

33. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 32, wherein the lipocalin-fold molecule is a lipocalin or a derivative thereof with 1-30 amino acid exchanges and/or 1-50 amino acid deletions and/or 1-50 amino acid insertions.

34. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 33, wherein the lipocalin-fold molecule has a sequence identity with human RBP4 of at least 95%.

35. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 33, wherein the lipocalin-fold molecule has a sequence identity with human tear lipocalin (TLC) of at least 95%.

36. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 33, wherein the lipocalin-fold molecule has a sequence identity with human apolipoprotein M (ApoM) of at least 95%.

37. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 36, wherein both, the lipocalin-fold molecule and the lipocalin-fold binding interac tion partner are not part of a naturally occurring ligand regu lated protein-protein interaction system.

38. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 37, wherein the lipocalin-fold ligand is selected from the group fenretinide (15- [ (4-hydroxyphenyl) amino] retinal) , N-Ethylretinamide (PubChem CID: 5288173), all-trans retinoic acid (PubChem CID: 444795), axerophthene (PubChem CID: 5287722), A1120 (PubChem CID 25138295) and derivatives thereof, 1 , 4-butanediol (Pubchem CID: 8064), sphingosine-l-phosphate (Pubchem CID: 5283560), tetra- decanoic acid (Pubchem CID: 11005), indicaxanthin (Pubchem CID: 6096870 and 12310796), vulgaxanthin I (Pubchem CID: 5281217), Montelukast (Pubchem CID: 5281040), Cyclandelate (Pubchem CID: 2893), Oxolamine (Pubchem CID: 13738), Mazaticol (PubchemCID: 4019), Butoctamid (Pubchem CID: 65780), Tonabersat (Pubchem CID: 6918324), Novazin (Pubchem CID: 65734), Diphenidol (Pubchem CID: 3055), Neobornyval, Erlotinib (Pubchem CID: 92131336), Tanespi- mycin (Pubchem CID: 6505803), LMI070 (Pubchem CID: 85471316), Alloclamide (Pubchem CID: 71837), Diacetolol (Pubchem CID: 50894), Acotiamide (Pubchem CID: 5282338), Acoziborole (Pubchem CID: 44178354), Acumapimod (Pubchem CID: 11338127), Apalutamide (Pubchem CID: 24872560), ASP3026 (Pubchem CID: 25134326), AZD1480 (Pubchem CID: 16659841), BIIB021 (Pubchem CID: 16736529), Branaplam (Pubchem CID: 89971189), Brequinar (Pubchem CID: 57030), Chlorproguanil (Pubchem CID: 9571037), Clindamycin (Pubchem CID: 446598), Emricasan (Pubchem CID: 12000240), Enasidenib (Pubchem CID: 89683805), Enolicam (Pubchem CID: 54679203), Flurazepam (Pubchem CID: 3393), ILX-295501 (Pubchem CID: 127737), Indibulin (Pubchem CID: 2929), Metoclopramide (Pubchem CID: 12598248), Mevastatin (Pubchem CID: 64715), MGG- BYMDAPCCKCT-UHFFFAOYSA-N (Pubchem CID: 25134326), MK0686 (Pub chem CID: 16102897), Navarixin (Pubchem CID: 71587743), Nefazo- done hydrochloride (Pubchem CID: 54911), Pantoprazole (Pubchem CID: 4679), Pavinetant (Pubchem CID: 23649245), Proxazole (Pub chem CID: 8590) , Siccanin (Pubchem CID: 71902), Sulfaguanole (Pubchem CID: 9571041), Sunitinib (Pubchem CID: 5329102), Su- vorexant (Pubchem CID: 24965990), Tiapride (Pubchem CID: 5467), Tonabersat (Pubchem CID: 6918324), VNBRGSXVFBYQNN-UHFFFAOYSA-N (Pubchem CID: 24794418), YUHNXUAATAMVKD-PZJWPPBQSA-N (Pubchem CID: 44548240), Ulimorelin (Pubchem CID: 11526696), Xipamide (Pubchem CID: 26618), Tropesin (Pubchem CID: 47530), Triclabendazole (Pubchem CID: 50248), Triclabendazole sulfoxide (Pubchem CID: 127657), Triclabendazole sulfone (Pubchem CID: 10340439) and Trametinib (Pubchem CID: 11707110).

39. A ligand regulated protein-protein interaction system ac cording to any one of embodiments 1 to 38, wherein the lipocalin-fold molecule and the lipocalin-fold binding interac tion partner are polypeptides.

40. A nucleic acid molecule comprising nucleotide sequences en coding the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner according to any one of embodiments 1 to 39, wherein the nucleic acid is preferably selected from DNA or RNA, more preferably in vitro transcribed RNA or RNA packaged in a retrovirus, especially RNA packaged in a lentivi- rus .

41. A kit of at least two nucleic acid molecules, wherein the first nucleic acid molecule comprises nucleotide sequences en coding the lipocalin-fold molecule according to embodiment 39 and wherein the second nucleic acid molecule comprises sequences encoding the lipocalin-fold binding interaction partner accord ing to embodiment 39, wherein the nucleic acids are preferably selected from DNA or RNA, more preferably in vitro transcribed RNA or RNA packaged in a retrovirus, especially RNA packaged in a lentivirus .

42. A nucleic acid molecule or a kit of nucleic acid molecules according to embodiments 40 or 41, wherein the nucleic acid mol ecules are present in a vector and preferably packaged as DNA or RNA into an infectious virus particle.

43. A nucleic acid molecule or a kit of nucleic acid molecules according to any one of embodiments 40 to 42, wherein the nucle ic acid sequences are linked to a sequence mediating strong and stable transgene expression in lymphocytes, wherein such a se quence preferably comprises the 5'-LTR of a gamma retrovirus or subelements R and U3 of a 5 ' -LTR of the Moloney murine leukaemia virus (MMLV) or the promoter of the murine stem cell virus (MSCV) or the promoter of phosphoglycerate kinase (PGK) or even more preferably the human elongation factor 1 (EF-1) alpha pro moter .

44. A recombinant expression vector comprising the nucleic acid molecule according to embodiment 40 or the kit of nucleic acid molecules according to embodiment 41.

45. A kit of at least two recombinant expression vectors, where in the first recombinant expression vector comprises a nucleic acid molecule encoding the lipocalin-fold molecule according to embodiment 40 and wherein the second recombinant expression vec tor comprises a nucleic acid molecule encoding the lipocalin- fold binding interaction partner according to embodiment 40.

46. A recombinant expression vector or a kit of at least two recombinant expression vectors, wherein the vector or at least one, preferably at least two, of the vectors comprise a T lym phocyte-specific promoter or an NK cell-specific promoter opera- bly linked to the nucleotide sequences encoding the lipocalin- fold molecule and/or the lipocalin-fold binding interaction partner according to embodiment 39.

47. A vector comprising a nucleic acid molecule or a kit of nu cleic acid molecules according to any one of embodiments 40 to 43.

48. A kit of at least two vectors comprising nucleic acid mole cules according to any one of embodiments 41 to 43, wherein the first vector comprises a nucleic acid molecule encoding the lipocalin-fold molecule and wherein the second vector comprises a nucleic acid molecule encoding the lipocalin-fold binding in teraction partner.

49. A vector or a kit of vectors according to embodiments 47 or 48, wherein a vector is a recombinant adeno-associated virus (rAAV) vector or a transposon vector, preferably a Sleeping Beauty transposon vector or PiggyBac transposon vector, or wherein a vector is a retroviral vector, preferably a gamma- retroviral vector or a lentiviral vector.

50. A vector or a kit of at least two vectors according to any one of embodiments 47 to 49, wherein the vector or at least one, preferably at least two, of the vectors are expression vectors, preferably expression vectors in which the nucleotide sequences encoding the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner according to any one of embodiments 1 to 39 are operably linked to a sequence mediating strong and stable transgene expression in lymphocytes, wherein such a se quence preferably comprises the 5'-LTR of a gamma retrovirus or subelements R and U3 of a 5 ' -LTR of the Moloney murine leukaemia virus (MMLV) or the promoter of the murine stem cell virus (MSCV) or the promoter of phosphoglycerate kinase (PGK) or even more preferably the human elongation factor 1 (EF-1) alpha pro moter .

51. A cell genetically modified to produce the lipocalin-fold molecule and/or the lipocalin-fold binding interaction partner according to embodiment 39.

52. A cell genetically modified to produce the lipocalin-fold molecule and the lipocalin-fold binding interaction partner ac cording to embodiment 39.

53. A cell modified in vitro or ex vivo with a nucleic acid mol ecule or a kit of nucleic acid molecules according to any one of embodiments 40 to 43 or with a vector or a kit of vectors ac cording to any one of embodiments 47 to 50 to produce the lipocalin-fold molecule and/or the lipocalin-fold binding inter action partner according to any one of embodiments 1 to 39, or a kit comprising two or more of said modified cells.

54. A kit of at least two cells according to embodiment 53, wherein the first cell is genetically modified to produce the lipocalin-fold molecule according to embodiment 39 and wherein the second cell is genetically modified to produce the lipocalin-fold binding interaction partner according to embodi ment 39.

55. A cell according to embodiment 53 or a kit of cells accord ing to embodiments 53 or 54, wherein the cell is a prokaryotic or eukaryotic cell, preferably a mammalian cell, more preferably a hematopoietic stem cell, a progenitor cell, or a cell derived from a hematopoietic stem cell or a progenitor cell, especially a T cell or an NK cell.

56. A cell or a kit of cells according to any one of embodiments 53 to 55, wherein the cell is transfected or transformed with a vector or a kit of at least two vectors according to any one of embodiments 47 to 50.

57. A cell or kit of cells according to any one of embodiments 53 to 56, wherein the cell has stably integrated the nucleotide sequences encoding a lipocalin-fold molecule and/or a lipocalin- fold binding interaction partner into its genome.

58. A cell or kit of cells according to any one of embodiments 53 to 57, wherein the cell has stably integrated the nucleotide sequences encoding a lipocalin-fold molecule and/or a lipocalin- fold binding interaction partner into its genome by the use of site directed nuclease technology, preferably by the use of zinc finger nucleases or TALENs, or even more preferably CRISPR/Cas technology .

59. A cell transformed with a recombinant expression vector or a kit of at least two recombinant expression vectors according to any one of embodiments 44 to 46, preferably a mammalian cell, especially a T cell or an NK cell.

60. A pharmaceutical preparation comprising a nucleic acid mole cule or a kit of nucleic acid molecules according to any one of embodiments 40 to 43, and/or a vector or a kit of vectors ac cording to any one of embodiments 47 to 50, and/or a cell or a kit of cells according to any one of embodiments 53 to 58.

61. A pharmaceutical preparation according to embodiment 60, wherein the vectors are contained in infectious virus particles.

62. A method of making a cell according to any one of embodi ments 53 to 58, the method comprising introducing into the cell, preferably stably integrating into the genome of the cell, in vitro or ex vivo a nucleic acid molecule or a kit of nucleic ac id molecules according to any one of embodiments 40 to 43, or a vector or a kit of vectors according to any one of embodiments 47 to 50.

63. A non-human animal comprising a cell according to any one of embodiments 53 to 58.

64. A plant comprising a cell according to any one of embodi ments 53 to 58.