Grafting of thrombopoietin-mimetic peptides into cystine knot miniproteins yields high-affinity thrombopoietin antagonists and agonists Sebastian Krause 1, *, Hans-Ulrich Schmoldt 2, *, Alexander Wentzel 3,† , Matthias Ballmaier 4 , Karlheinz Friedrich 1 and Harald Kolmar 2,3 1 University of Jena Medical School, Institute of Biochemistry, Jena, Germany 2 Department of Molecular Genetics, Georg-August-University, Go ¨ ttingen, Germany 3 Selecore GmbH, Go ¨ ttingen, Germany 4 Department of Pediatric Hematology and Oncology, Medizinische Hochschule Hannover, Germany Protein grafting, the transfer of a peptide sequence onto the surface of another protein, is a potentially powerful technique to present peptides in defined and active three-dimensional conformations. Its applicabil- ity, however, has been hampered by low biological activity of the designed ligands and low tolerance of the recipient protein scaffold to surface substitutions [1,2]. To at least partially overcome these limitations, we established the concept of grafting peptides with desired biological activities onto the structural scaffold of small disulfide-rich inhibitors with cystine knot (CK) folds that display high intrinsic stability and rigidity and are small enough to be amenable both to chemical and to recombinant synthesis [3–5]. Keywords c-Mpl receptor; cystine knot proteins; peptide agonist; thrombocytopenia; thrombopoietin mimetics Correspondence H. Kolmar, Clemens-Scho ¨ pf Institute of Organic Chemistry and Biochemistry, Darmstadt University of Technology, Petersenstr. 22, D-64287 Darmstadt, Germany Fax: +49 6151 16 5399 Tel: +49 6151 16 3657 E-mail: Kolmar@Biochemie-TUD.de  Present address Institutt for Bioteknologi, NTNU, Trondheim, Norway *These authors contributed equally to this work (Received 11 August 2006, revised 2 Octo- ber 2006, accepted 2 November 2006) doi:10.1111/j.1742-4658.2006.05567.x Thrombopoietin is the primary regulator of platelet production. We exploi- ted two naturally occurring miniproteins of the inhibitor cystine knot fam- ily as stable and rigid scaffolds for the incorporation of peptide sequences that have been shown to act as high-affinity thrombopoietin antagonists. Several miniproteins that antagonistically block thrombopoietin-mediated receptor activation were identified using a microscale reporter assay. Cova- lent miniprotein dimerization yielded potent bivalent c-Mpl receptor agon- ists with EC 50 values in the low nanomolar or picomolar range. One selected miniprotein-derived thrombopoietin agonist was almost as active as natural thrombopoietin with regard to stimulation of megakaryocyte colony formation from human bone marrow mononuclear cells, and elici- ted doubling of platelet counts in mice. Our data suggest that dimeric cystine knot miniproteins have considerable potential for the future devel- opment of small and stable receptor agonists. This approach may provide a promising strategy for pharmaceutical interference with other receptors activated by ligand-induced dimerization. Abbreviations AGRP, Agouti-related protein; CK, cystine knot; c-Mpl, thrombopoietin receptor; EETI-II, Ecbalium elaterium trypsin inhibitor II; IL-4, interleukin-4; IL-4R, interleukin-4 receptor; rhuTPO, recombinant human thrombopoietin; SI, stimulation index; TPO, thrombopoietin. 86 FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS CK miniproteins are found in plants, animals and humans, and have a plethora of different functions [6,7]. A CK miniprotein of human origin is the C-ter- minal domain of human Agouti-related protein (AGRP) [8]. AGRP is an endogenous antagonist of the melanocortin-3 and melanocortin-4 G-protein-cou- pled receptors [9]. The C-terminal domain of AGRP [AGRP(87–132)] possesses five disulfide bonds and a well-defined three-dimensional structure that displays full activity as compared to the full-length protein [10]. Jackson et al. constructed a minimized 34 amino acid domain that autonomously folds into the CK motif [11]. Owing to its high intrinsic rigidity and stability, the CK motif is ideally suited for molecular engineering applications and development of lead compounds for drug design [6,12,13]. Protein engineering of CK mini- proteins takes advantage of the fact that the loops connecting the conserved cysteine residues tolerate sub- stitution of individual – and even the insertion of addi- tional ) amino acids. The CK is an embedded ring formed by two disulfides bonds that is penetrated by a third disulfide bond. Six conserved cysteines (numbered I–VI) form three disulfide bonds with connectivities I–IV, II–V and III–VI [14]. In the ICK fold, the I–IV and II–V disulfide bonds form the ring that is penetrated by the III–VI disulfide bond. The AGRP CK motif con- tains a fourth disulfide bond within a b-hairpin known to be essential for receptor binding (Fig. 1A). No natural CK miniprotein is known that exerts its biological function as a dimer. Many cytokine recep- tors are activated by ligand-induced dimerization. This prompted us to investigate whether homodimeric mini- proteins can be constructed by grafting onto the CK scaffold cytokine receptor-binding peptides that can act as cytokine receptor agonists. As a model system, we chose the thrombopoietin (TPO)–thrombopoietin receptor (c-Mpl) pair. TPO is a hematopoietic growth factor that serves as the primary regulator of thrombocytopoiesis [15,16]. It is mainly produced by the liver, and acts in several stages of platelet production, e.g. by promoting the proliferation of hematopoietic stem cells and by initiating specific maturation events in megakaryocytes. The mature cy- tokine comprises a 332 amino acid glycoprotein; the receptor-binding domain is located within the N-ter- minal 154 residues and folds into a four-helix bundle structure that is shared by a variety of hematopoietic cytokines [17]. The TPO receptor, c-Mpl, is expressed on early hematopoietic progenitors, megakaryocytes and platelets, and is activated by ligand-mediated homodimerization [18]. TPO and TPO mimetics are of considerable pharmaceutical interest for the treatment of low platelet counts [19]. However, administration of recombinant TPO has not become general clinical practice, due to the immunogenicity of the protein [20,21]. Two approaches have been used to overcome the antigenic problem of recombinant TPOs: either the use of dimeric antibody fragments that induce receptor dimerization and activation [22–24], or generation of dimeric agonistic TPO-mimetic peptides with amino acid sequences unrelated to TPO. Fig. 1. Schematic representation of the three-dimensional structures of CK miniprotein scaffolds used in this study. The disulfide bonds forming the CK architecture are indicated as sticks and numbered according to their occurrence in the sequence. b-Strands are shown as arrows. Sequences and disulfide connectivities are indicated below. Loops responsible for the function of the respective miniprotein are indi- cated by dashed lines in the sequence. (A) The minimized C-terminal fragment of AGRP(87–120) [11]. In AGRP*, a single lysine residue, highlighted in gray, is introduced for chemical dimerization plus two additional C-terminal amino acids for cloning purposes. (B) Squash tryp- sin inhibitor EETI-II from Ecballium elaterium [45]. EETI-II already harbors a single lysine residue suitable for dimerization, which is given as a stick representation. In EETI-II*, a serine residue has been added for cloning purposes. S. Krause et al. Cystine knot miniprotein TPO agonists FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS 87 In the present study, we transplanted c-Mpl-bind- ing peptides into two different CK miniproteins. We found that miniprotein dimerization through chem- ical linkage results in TPO-mimetic peptides with CK architecture. Results Generation of miniprotein derivatives containing TPO-mimetic peptides Using phage display technology, Cwirla et al. identified small peptides that competitively bind to the TPO receptor [22]. Among these binders, a 14-mer peptide, AF12505, devoid of sequence homology to TPO, was isolated that acted as a weak TPO receptor agonist and could be converted to a strong agonist by chem- ical homodimerization [22]. To investigate whether c-Mpl-binding peptides retain their receptor-binding properties when introduced into the structure of a CK miniprotein, peptides derived from AF12193, an 18-mer peptide that contains two cysteines forming an intramolecular disulfide bond, and from the linear 14-mer AF12505 were grafted onto two different miniprotein scaffolds. In the first set of constructs, the receptor-binding loop of a modified AGRP miniprotein domain (AGRP*; Fig. 1A), located between cysteines V and VIII (CYCRFFNAFCYC) [25], was replaced with the linear peptide (AF12505) or the disulfide bond-constrained peptide (AF12193), respectively (Table 1). In a second set of constructs, the inhibitor loop of the Ecballium elaterium trypsin inhibitor II (EETI-II, Fig. 1B), which is located between cysteines I and II and is known to tolerate a vast spectrum of amino acid replacements [3], was replaced with the TPO-binding peptide sequences. The EETI-II scaffold was employed because it is structur- ally similar to AGRP* but displays resistance against soluble and membrane-bound human proteases, a fea- ture that may be useful for oral application [5]. As no structural information on the TPO receptor- bound peptides is available, several variants with flank- ing extensions and truncations were generated to force the introduced peptide sequences into different loop conformations (AGTP-2, AGTP-4, AGTP-5, AGTP-6, ETTP-3; Table 1). Proteins were produced in Escheri- chia coli as fusions to an inactive version of the Bacil- lus amyloliquefaciens RNase (barnase¢). To evaluate the effect of peptide incorporation into the miniprotein structure, peptides AF12505 and AF12193 were also fused directly to the C-terminus of barnase¢ (LNTP-1, LNTP-2; Table 1). Monomeric miniprotein derivatives compete with recombinant human TPO for receptor binding The integration of the TPO-mimetic peptide sequences into rigid protein scaffolds is accompanied by struc- tural modifications and extensive restrictions in their conformational flexibility that may result in a loss of function. We first examined which of the generated miniprotein constructs permitted functional incorpor- ation of the peptide sequences. To this end, the mini- protein variants were tested for their inhibitory activity on TPO-mediated receptor activation in comparison with the respective peptides directly fused to barnase¢. To determine receptor binding and antagonistic activity, we employed a functional readout system based on an interleukin-4 receptor (IL-4R)-responsive reporter gene construct [26]. It rests on a functional c-Mpl ⁄ IL-4R receptor chimera comprising the ectod- omain of c-Mpl fused to the cytoplasmic part of the IL-4Ra chain. The chimeric receptor was transiently transfected into Ba ⁄ F3 cells along with the luciferase reporter gene construct pIeTATA-Luc (Fig. 2A). The transfected cells responded in a dose-dependent man- ner to recombinant human (rhu)TPO stimulation through induction of luciferase activity, with maximum response at 2–10 nm recombinant human thrombo- poietin (Fig. 2B). An rhuTPO concentration of 5 nm was chosen to assess the antagonistic properties of the miniprotein constructs under investigation. Transiently transfected Ba ⁄ F3 TPO reporter cells were supplemented with equal amounts of the miniproteins or control peptides and grown in the presence or absence of rhuTPO. Whereas neither the control peptides AF12193 and AF12505 nor the miniprotein constructs mediated activation of the hybrid receptor at 500 nm, some were able to act antag- onistically against TPO (Fig. 2C). We found that all Table 1. Structural scaffold and grafted peptide sequences of mini- protein derivatives used in this study. Construct name Miniprotein scaffold Grafted peptide Sequence family LNTP-1 – GGCADGPTLREWISFCGG AF12193 AGTP-1 AGRP* GGCADGPTLREWISFCGG AGTP-2 AGRP* YCADGPTLREWISFCY ETTP-1 EETI-II* GGCADGPTLREWISFCGG LNTP-2 – IEGPTLRQWLAARA AF12505 AGTP-3 AGRP* IEGPTLRQWLAARA AGTP-4 AGRP* GGTALAIEGPTLRQWLAARA AGTP-5 AGRP* GGTCLAIEGPTLRQWLCARA AGTP-6 AGRP* YCIEGPTLRQWLAACY ETTP-2 EETI-II* IEGPTLRQWLAARA ETTP-3 EETI-II* IEGPTLRQWLAA Cystine knot miniprotein TPO agonists S. Krause et al. 88 FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS constructs that had been derived from the low-affinity peptide AF12193 showed only marginal inhibitory effects at the concentration applied, whereas the high-affinity peptide AF12505, as well as some of its miniprotein derivatives, exerted strong inhibitory effects. Interestingly, when it was integrated into the EETI-II* miniprotein scaffold, the inhibitory activity of AF12505 was even enhanced compared to the linear AF12505 peptide barnase¢ fusion (compare ETTP-2 and ETTP-3 with LNTP-2). Inhibition of c-Mpl activation was also observed with two AGRP* miniprotein con- structs bearing the AF12505 peptide sequence (AGTP-3 and AGTP-4, with AGTP-4 applied only at 100 nm, due to limiting amounts of material), although their activity was lower compared to the EETI-II derivatives. Construction and functional analysis of TPO- mimetic miniprotein dimers The high-affinity peptide AF12505 described by Cwirla et al. displayed greatly increased potency in a cell pro- liferation assay upon dimerization by covalent linkage of two monomers [22]. Therefore, we investigated whe- ther dimerization of TPO receptor-binding minipro- teins would result in agonistic activity. From the characterized miniprotein monomers with antagonistic properties, we chose AGTP-4, AGTP-5, ETTP-3 and ETTP-2 for dimerization studies. AGTP-4 was chosen instead of AGTP-3, because it contains the complete AF12505 peptide sequence of AGTP-3 plus N-terminal flanking residues. The N-terminal Leu-Ala extension to the AF12505 peptide is present in the parent peptide AF12434, which was used by Cwirla et al. as a tem- plate for random mutagenesis and truncation to obtain AF12505 [22] and therefore might contribute to recep- tor binding. The barnase¢ expression tag was removed by chemical cleavage at the fusion junction [27], and covalent crosslinking by bis-succinimidyl suberate was achieved by employing the unique lysine residue within both the AGRP* and the EETI-II* scaffolds. The dimeric products designated AGTP-4d, AGTP-5d, ETTP-3d and ETTP-2d were separated from mono- AC B Fig. 2. Analysis of antagonistic activity of monomeric miniprotein compounds. (A) Schematic representation of the functional c-Mpl-ED ⁄ IL-4R hybrid receptor system introduced in this work. The signaling receptor complex is a homodimer of chimeric receptor chains comprising the c-Mpl ectodomain (striped boxes) and the transmembrane ⁄ cytoplasmic segment of IL-4Ra (gray boxes). (B) TPO-dependent reporter gene expression in transiently transfected Ba ⁄ F3 cells. Circles represent mean reporter gene activity at the indicated rhuTPO concentrations from triplicate measurements. (C) Inhibition of rhuTPO-induced reporter gene expression by monomeric miniprotein constructs. Upper panel: For each miniprotein, the normalized basal reporter gene activity and the normalized rhuTPO stimulation are shown as light gray bars and dark gray bars, respectively. The data were obtained from duplicate measurements; error bars indicate minimum and maximum values. Lower panel: Calculated relative inhibition of rhuTPO-induced reporter gene activity for each miniprotein applied. S. Krause et al. Cystine knot miniprotein TPO agonists FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS 89 mers by RP-HPLC and assayed in the Ba ⁄ F3 reporter cell assay for TPO-mimetic activity. Three of the four candidates (AGTP-4d, AGTP-5d, and ETTP-2d) strongly evoked c-Mpl-driven luciferase expression (Fig. 3A). ETTP-3d failed to induce significant c-Mpl activation, although its monomeric form showed strong interaction with c-Mpl in a competition assay (compare Fig. 2C). The maximum stimulation effect achieved with AGTP-4d and ETTP-2d was close to that of TPO, although the EC 50 values for both con- structs were approximately 10-fold higher than those for TPO (AGTP-4d, 1.4 nm; ETTP-2d, 1.6 nm; TPO, 0.16 nm). We next examined the ability of the dimeric mini- proteins to stimulate proliferation of the TPO-depend- ent megakaryoblastic cell line M-07e. AGTP-4d, AGTP-5d and ETTP-2d, which had displayed strong reporter gene activation in Ba⁄ F3 cells, also evoked sustained growth of M-07e cells (Fig. 3B). Even candi- date ETTP-3d, which was able to induce luciferase expression in Ba ⁄ F3 cells to only a marginal extent (compare Fig. 3A), elicited a significant degree of pro- liferation. Taken together, the results of the prolifer- ation tests are in line with those of the reporter gene assays on Ba ⁄ F3 cells; however, EC 50 values were gen- erally about 10-fold lower for the proliferation test (about 0.1 nm compared to 1 nm). Cell proliferation and in vivo activity of ETTP-2d ETTP-2d, which proved to be the variant with the low- est EC 50 in the M-07e proliferation assay (approxi- mately 0.14 nm), was also used to assess the megakaryocyte colony-forming activity on bone mar- row mononuclear cells. Human bone marrow mono- nuclear cells were cultured in semisolid media containing either no factor or different concentrations of rhuTPO or ETTP-2d. After 10 days in culture, the number of megakaryocyte colonies was determined. ETTP-2d at 10 nm elicited the formation of CD41-pos- itive colonies to a similar extent as obtained with rhu- TPO at 0.3 nm, with a shift in the distribution of colony size towards smaller colonies (Fig. 4B). The effect of ETTP-2d on thrombopoiesis in mice was also studied. To this end, ETTP-2d was administered sub- cutaneously in two daily injections for 5 consecutive days at 0.2 or 20 lg per animal, respectively. As can be seen from Fig. 4B, the low-dose regimen resulted in a two-fold increase in platelet count compared to the placebo control group, indicating a stimulatory effect on megakaryocytopoiesis. This doubling of platelet count was also observed when murine TPO or dimeric AF12505 peptide was used as the megakaryocytopoi- esis-stimulating agent [22]. Discussion In the present work, we designed c-Mpl-binding mini- proteins by grafting c-Mpl-binding peptide sequences onto small and rigid CK miniprotein scaffolds. Mini- protein dimerization by chemical linkage yielded deriv- atives with agonist properties. The dimeric miniprotein ETTP-2d displayed an activity close to that of the natural ligand TPO in vitro in a proliferation assay, Fig. 3. Analysis of TPO receptor agonistic activity of dimeric miniproteins. (A) Induction of reporter gene expression in Ba ⁄ F3 reporter cells. Ba ⁄ F3 TPO reporter cells were stimulated with the indicated concentrations of rhuTPO or dimeric miniproteins AGTP-4d, AGTP-5d, ETTP-2d, and ETTP-3d, and reporter gene activity was assayed. The data were obtained from duplicate measurements; error bars indicate minimum and maximum values. (B) Proliferation of the TPO-dependent cell line M-07e in response to stimulation by rhuTPO or miniproteins. M-07e cells were grown in the presence of the indicated concentrations of TPO or dimeric miniproteins AGTP-4d, AGTP-5d, ETTP-2d, and ETTP-3d for 72 h, and proliferation was determined photometrically. The chart shows relative stimulation of cell proliferation. The data represent mean values from duplicate measurements. Cystine knot miniprotein TPO agonists S. Krause et al. 90 FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS and significantly increased peripheral platelet numbers in mice. CK miniproteins constitute a family of small proteins with unique properties [6]. Their knotted disulfide arrangement renders them extremely stable against enzymatic digestion as well as thermal degradation [13,28,29], making them interesting scaffolds for pharma- ceutical applications. Moreover, recent studies indicate that miniproteins with CK architecture may act as phar- macophoric frameworks for oral peptide delivery [5]. We used two TPO-mimetic peptides that have been isolated using phage display technology and introduced them into the structural scaffolds of EETI-II, a trypsin inhibitor, and of human AGRP*. The basic scaffolds of both miniproteins are very similar. However, pep- tides were introduced into two different loop regions. Whereas in EETI-II the protease inhibitor loop spanning cysteines I and IV was replaced by the TPO- mimetic peptide sequence, in AGRP* the loop con- necting cysteines V and VIII was chosen. These natural loops differ from each other with respect to their loca- tion within the CK scaffold and their main chain con- formation. Whereas in EETI-II the distance between the Ca atoms of CI and CIV is 11.8 A ˚ and an exten- ded loop is formed by six intervening residues, in AGRP* the Ca atoms of cysteines V and VIII are 6.5 A ˚ apart and a b-hairpin is formed that is held together by an additional disulfide bond. Incorporation of both c-Mpl-binding peptide sequences AF12193 and AF12505 into EETI-II* yielded antagonists that were more potent than the AGRP* derivatives. This shows that structural con- straints imposed by the scaffold and ⁄ or neighboring residues contribute to receptor binding. Moreover, the antagonistic activity of the unconstrained peptide AF12505 could even be increased upon grafting onto the EETI-II* scaffold, indicating that the peptide thereby becomes rigidified in a biologically active con- formation. Structural data on the main chain confor- mation of the transplanted loop sequences within the context of the respective CK scaffold in free form and bound to the c-Mpl extracellular domain are required for a detailed analysis of entropic and enthalpic contri- butions to binding and affinity. By analogy to other receptors of the cytokine recep- tor family, the TPO receptor c-Mpl is believed to be activated upon homodimerization by a single TPO molecule. In line with this, Cwirla et al. demonstrated that covalent dimerization of the high-affinity peptide AF12505, which possesses mainly antagonistic activity, leads to a potent receptor agonist [22]. By chemical crosslinking, we converted receptor-binding derivatives of both AGRP* and EETI-II* into bivalent ligands. As the lysine residues that were used for chemical coupling are located at different positions within the respective scaffold, the integrated receptor-binding loops are spatially positioned relative to each other in different ways. Despite this fact, both dimeric conformations yielded potent TPO receptor agonists. In vitro, the most active miniprotein agonists, AGRP-4d and ETTP-2d, medi- ated TPO-specific responses with approximately 10-fold BA Fig. 4. Effect of ETTP-2d on human megakaryocyte formation and on platelet production in vivo. (A) Effect of ETTP-2d on megakaryocyte colony differentiation in comparison to rhuTPO. Human bone marrow mononuclear cells (10 5 ) (BM-MNCs) were incubated with rhuTPO or dimeric miniproteins as indicated. The CD41-positive colonies formed were counted by immunocytochemical staining and classified by size. (B) In vivo activity of ETTP-2d in NMRI mice. Mice were treated twice a day with 0.2 lgor20lg of ETTP-2d per individual on 5 consecutive days; on day 6, platelets were counted. S. Krause et al. Cystine knot miniprotein TPO agonists FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS 91 higher EC 50 values compared to the natural factor. However, there were obvious differences in the out- comes of the in vitro assays applied. In the Ba ⁄ F3 reporter cell readout, the natural ligand rhuTPO, as well as the dimeric TPO mimetic miniproteins, medi- ated receptor activation with bell-shaped dose depend- encies. This most likely results from a self-antagonistic effect at very high ligand concentrations, due to an unproductive one-to-one binding stoichiometry between ligand and receptor, as has been observed in a number of homodimeric cytokine receptors [30,31]. Consistent with this, the dose–response data could be well fitted to the mathematical model of receptor clus- tering by bivalent ligands [32]. Thus, this readout seems to reflect closely the formation of signaling receptor complexes. However, the sensitivity of the reporter gene response was significantly lower than that of the pro- liferative response of M-07e cells. Notably, M-07e pro- liferation remained maximal over a wide concentration range with TPO or the most active TPO mimetics (AGTP-4d, ETTP-2d). This observation suggests that only a minor fraction of receptors has to be activated to yield saturating proliferative signaling [33–35]. CK miniproteins are among the smallest proteins known to possess a defined three-dimensional struc- ture. Besides recombinant production, they are also amenable to chemical synthesis followed by in vitro oxidation [36–39]. They are generally remarkably resistant to serum proteases and membrane-bound pro- teases of the intestinal mucosa [5], which is also partic- ularly the case for ETTP, which displays a half-life for degradation by membrane-bound enzymes of rat intes- tinal mucosa of approximately 180 min (A. Bernkop- Schnu ¨ rch, unpublished results). Our data indicate that low doses (4.6 lg or 0.6 nmolÆkg )1 ) of dimeric ETTP- 2d miniprotein are sufficient to evoke a raise in platelet counts in mice. No efforts have yet been made to determine the minimum daily effective dose. It will be interesting to see whether covalent attachment of poly(ethylene glycol) to the N-terminus of the dimeric molecule results in delayed plasma clearance and an increased effect on megakaryopoiesis, as has been shown for pegylated recombinant human megakaryo- cyte growth and development factor, a nonglycosylated polypeptide encompassing the N-terminal sequence of TPO [40]. A number of growth factors (e.g. transforming growth factor-b, platelet-derived growth factor, nerve growth factor NGF) are members of the CK super- family of growth factors that display the same disulfide bond arrangement as miniproteins of the inhibitor CK family [41] but contain extended loop sequences and additional and longer secondary structure elements that render them less stable and less rigid than miniproteins. This underscores the outstanding potential of the CK structural motif as a convenient scaffold for agonistic modulation of diverse receptor systems. In this work, we demonstrate that potent agonists of a cytokine receptor can also be established through dimerization of functionalized CK miniproteins. No natural dimeric CK miniproteins are known, and this is the first report that miniprotein dimerization results in enhanced potency. It will be interesting to determine whether the concept of peptide rigidification by introduction into the CK miniprotein scaffold in combination with mini- protein dimerization will also be applicable for other disease-related receptors that underlie the mechanism of dimerization-induced signal transduction. Experimental procedures Materials rhuTPO of purity > 97% and an ED 50 in a cell prolifer- ation assay of 1–3 ngÆmL )1 was purchased from R&D Sys- tems (Minneapolis, MN). DNA constructs Construction of miniprotein fusion genes consisting of an enzymatically inactive variant of B. amyloliquefaciens RNase, barnase¢ and the coding sequence for the respective miniprotein was performed according to standard proce- dures [42]. The miniprotein-encoding genes, as well as the two linear variants LNTP-1 and LNTP-2, were assembled in a two-step PCR that was similar to a recently described procedure [43]. Expression vector pcDNA-cc/4Ra A cDNA fragment encoding the IL-4R a chain cytoplasmic domain fused to the ectodomain of the common c (cc) receptor chain was inserted into vector pcDNA3.1(+)neo (Invitrogen, Carlsbad, CA) via BamHI sites. The cc seg- ment was replaced via XhoI cleavage and ligation by an SfiI site containing XhoI linker followed by insertion of the PCR-amplified and SfiI-cleaved c-Mpl-ED encoding sequence (I Dusanter-Fourt, Institut Cochin, Paris, France). Protein production and purification Miniprotein production was performed as described recently, making use of an enzymatically inactivated variant of B. amyloliquefaciens RNase (barnase¢) as a purification handle [43]. Purified barnase¢ fusion proteins were cleaved overnight with 0.6 lLofa5m cyanogen bromide solution (Sigma-Aldrich, Schnelldorf, Germany) per mg of fusion Cystine knot miniprotein TPO agonists S. Krause et al. 92 FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS protein in 8 m urea ⁄ 0.2 m HCl or 6 m guanidinium hydrochloride ⁄ 0.2 m HCl and purified by RP-HPLC [43]. The purity was > 97%. Identity and disulfide bond forma- tion were confirmed by ESI MS. To obtain the dimeric vari- ants, two equivalents of the respective miniprotein were dissolved in dimethylformamide ⁄ dimethylsulfoxide (1 : 1). Triethylamine was added at a final concentration of 1% (v ⁄ v) together with one equivalent of sodium 3-(trimethylsi- lyl)propane-1-sulfonate (Pierce, Rockford, IL, USA). After overnight incubation at room temperature, the resulting dimeric crosslinked miniprotein was separated from the remaining monomers by RP-HPLC. Cell culture and transfection The human megakaryoblastic cell line M-07e was grown in RPMI 1640 ⁄ 10% fetal bovine serum ⁄ 50 lgÆL )1 gentamicin supplemented with 50 lgÆL )1 TPO (CellConcepts, Umkirch, Germany). Ba ⁄ F3 cells were cultured and transfected as previously described [26]. Each transfection batch was recovered in 3.5 mL of RPMI 1640 ⁄ 10% fetal bovine serum and seeded at 2 · 10 5 cells per 100 lL per well in a 96-well cell culture plate. Ba/F3 reporter gene assay Transfected Ba ⁄ F3 cells were left untreated for 1 h and then supplemented with various concentrations of rhuTPO or individual dimeric miniprotein variants, to a total volume of 200 lL. For rhuTPO competition assays, cell aliquots were preincubated with equal concentrations of individual monomeric barnase¢ miniprotein fusion proteins (500 nm, except for AGTP-4, which was applied at 100 nm) 1 h before stimulation with 5 nm rhuTPO. After incubation for 12 h at 37 °C and 5% CO 2 , cell lysates were pre- pared and luciferase activity was measured as previously described [27]. M-07e proliferation assay M-07e cells were washed twice in medium devoid of rhuTPO and seeded into a 96-well cell culture plate at 2 · 10 4 cells per 100 lL. After starvation for 2 h, cells were stimulated with various concentrations of rhuTPO or individual mini- protein variants for 72 h. Proliferative activity was deter- mined using the CellTiter kit (Promega, Mannheim, Germany). Wells were reacted with 25 lL of reagent for 4–6 h at 37 °C and 5% CO 2 , and subsequently read at 492 nm. Reporter gene activity was quantified by lumines- cence units. The stimulation index (SI) was derived from the reporter gene activity of stimulated TPO-responsive cells divided by the basal reporter gene activity of nonstimulated cells. Relative inhibition of TPO-mediated receptor activa- tion by a test compound X was calculated with the formula: %inhibition(X) ¼ SIðTPOÞÀSIðTPO:XÞ SIðTPOÞÀ1 Â 100% with SI(TPO:X) standing for TPO-associated SI in the pres- ence of the respective test compound X at a given concen- tration. Relative TPO-mimetic activities of dimeric miniprotein derivatives were derived from dose–response experiments. Data were fitted either to a sigmoidal model or to the model of receptor clustering by bivalent ligands described by Perelson and DeLisi [32]. Fitting was done in microcal origin (Northampton, MA, USA) using a nonlinear least- squares algorithm. Assays with hematopoietic progenitor cells and in vivo experiments The colony-forming unit–megakaryocyte assay was per- formed as previously described [44]. Briefly, 10 5 human bone marrow mononuclear cell were cultured for 120 days in a semisolid medium (37 °C, 5% CO 2 , 100% humidity) containing the growth factors specified in the results sec- tion. Collagen gels were dehydrated on slides and immuno- cytochemically stained with a primary antibody against CD41 and an alkaline phosphatase monoclonal anti-(alka- line phosphatase) detection system. For in vivo efficacy experiments, three groups of five NMRI mice were subcutaneously injected twice a day with either 0.2 lgor20lg of ETTP-2d per mouse or NaCl ⁄ P i buffer as a negative control. Blood samples were collected from the vena cava caudalis into K 2 -EDTA containing ves- sels. The number of platelets was measured with a Sysmex (Sysmex, Kobe, Japan) blood cell counter. Animal studies were done at the Fraunhofer Institute for Toxicology and Experimental Medicine, Hannover, FRG on a service-for- fee basis, and were carried out in accordance with the European Communities Council Directive of 24 November 1986 (86 ⁄ 609 ⁄ EEC). Acknowledgements We thank I. Dusanter-Fourt for providing the cDNA of human c-Mpl, and B. Wiederanders for helpful dis- cussions. This work was supported by the Deutsche Forschungsgemeinschaft (DFG) through SFB 416. References 1 Vita C, Drakopoulou E, Vizzavona J, Rochette S, Mar- tin L, Menez A, Roumestand C, Yang YS, Ylisastigui L, Benjouad A et al. (1999) Rational engineering of a miniprotein that reproduces the core of the CD4 site interacting with HIV-1 envelope glycoprotein. Proc Natl Acad Sci USA 96, 13091–13096. S. Krause et al. Cystine knot miniprotein TPO agonists FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS 93 2 Domingues H, Cregut D, Sebald W, Oschkinat H & Serrano L (1999) Rational design of a GCN4-derived mimetic of interleukin-4. Nat Struct Biol 6, 652–656. 3 Christmann A, Walter K, Wentzel A, Kra ¨ tzner R & Kolmar H (1999) The cystine knot of a squash-type protease inhibitor as a structural scaffold for Escherichia coli cell surface display of conformationally constrained peptides. Protein Eng 12, 797–806. 4 Reiss S, Sieber M, Oberle V, Wentzel A, Spangenberg P, Claus R, Kolmar H & Lo ¨ sche W (2006) Inhibition of platelet aggregation by grafting RGD and KGD sequences on the structural scaffold of small disulfide- rich proteins. Platelets 17, 153–157. 5 Werle M, Schmitz T, Huang HL, Wentzel A, Kolmar H & Bernkop-Schnu ¨ rch A (2006) The potential of cystine- knot microproteins as novel pharmacophoric scaffolds in oral peptide drug delivery. J Drug Targeting 14, 137– 146. 6 Chiche L, Heitz A, Gelly JC, Gracy J, Chau PT, Ha PT, Hernandez JF & Le-Nguyen D (2004) Squash inhi- bitors: from structural motifs to macrocyclic knottins. Curr Protein Pept Sci 5, 341–349. 7 Craik DJ (2001) Plant cyclotides: circular, knotted pep- tide toxins. Toxicon 39, 1809–1813. 8 Bolin KA, Anderson DJ, Trulson JA, Thompson DA, Wilken J, Kent SB, Gantz I & Millhauser GL (1999) NMR structure of a minimized human agouti related protein prepared by total chemical synthesis. FEBS Lett 451, 125–131. 9 Wilczynski AM, Joseph CG & Haskell-Luevano C (2005) Current trends in the structure–activity relation- ship studies of the endogenous agouti-related protein (AGRP) melanocortin receptor antagonist. Med Res Rev 25, 545–556. 10 McNulty JC, Thompson DA, Bolin KA, Wilken J, Barsh GS & Millhauser GL (2001) High-resolution NMR structure of the chemically-synthesized melano- cortin receptor binding domain AGRP(87–132) of the agouti-related protein. Biochemistry 40 , 15520–15527. 11 Jackson PJ, McNulty JC, Yang YK, Thompson DA, Chai B, Gantz I, Barsh GS & Millhauser GL (2002) Design, pharmacology, and NMR structure of a mini- mized cystine knot with agouti-related protein activity. Biochemistry 41, 7565–7572. 12 Craik DJ, Daly NL & Waine C (2001) The cystine knot motif in toxins and implications for drug design. Toxi- con 39, 43–60. 13 Clark RJ, Daly NL & Craik DJ (2005) Structural plasti- city of the cyclic cystine knot framework: implications for biological activity and drug design. Biochem J 394, 85–93. 14 Isaacs NW (1995) Cystine knots. Curr Opin Struct Biol 5, 391–395. 15 Kaushansky K, Lok S, Holly RD, Broudy VC, Lin N, Bailey MC, Forstrom JW, Buddle MM, Oort PJ, Hagen FS et al. (1994) Promotion of megakaryocyte progenitor expansion and differentiation by the c-Mpl ligand thrombopoietin. Nature 369, 568–571. 16 de Sauvage FJ, Hass PE, Spencer SD, Malloy BE, Gur- ney AL, Spencer SA, Darbonne WC, Henzel WJ, Wong SC, Kuang WJ et al. (1994) Stimulation of megakaryo- cytopoiesis and thrombopoiesis by the c-Mpl ligand. Nature 369, 533–538. 17 Feese MD, Tamada T, Kato Y, Maeda Y, Hirose M, Matsukura Y, Shigematsu H, Muto T, Matsumoto A, Watarai H et al. (2004) Structure of the receptor-bind- ing domain of human thrombopoietin determined by complexation with a neutralizing antibody fragment. Proc Natl Acad Sci USA 101 , 1816–1821. 18 Methia N, Louache F, Vainchenker W & Wendling F (1993) Oligodeoxynucleotides antisense to the proto- oncogene c-mpl specifically inhibit in vitro megakaryo- cytopoiesis. Blood 82, 1395–1401. 19 Kuter DJ & Begley CG (2002) Recombinant human thrombopoietin: basic biology and evaluation of clinical studies. Blood 100, 3457–3469. 20 Basser RL, O’Flaherty E, Green M, Edmonds M, Nichol J, Menchaca DM, Cohen B & Begley CG (2002) Development of pancytopenia with neutralizing antibo- dies to thrombopoietin after multicycle chemotherapy supported by megakaryocyte growth and development factor. Blood 99, 2599–2602. 21 Li J, Yang C, Xia Y, Bertino A, Glaspy J, Roberts M & Kuter DJ (2001) Thrombocytopenia caused by the development of antibodies to thrombopoietin. Blood 98, 3241–3248. 22 Cwirla SE, Balasubramanian P, Duffin DJ, Wagstrom CR, Gates CM, Singer SC, Davis AM, Tansik RL, Mattheakis LC, Boytos CM et al. (1997) Peptide agon- ist of the thrombopoietin receptor as potent as the natural cytokine. Science 276, 1696–1699. 23 Orita T, Tsunoda H, Yabuta N, Nakano K, Yoshino T, Hirata Y, Ohtomo T, Nezu J, Sakumoto H, Ono K et al. (2005) A novel therapeutic approach for thrombo- cytopenia by minibody agonist of the thrombopoietin receptor. Blood 105, 562–566. 24 Deng B, Banu N, Malloy B, Hass P, Wang JF, Cava- cini L, Eaton D & Avraham H (1998) An agonist murine monoclonal antibody to the human c-Mpl receptor stimulates megakaryocytopoiesis. Blood 92, 1981–1988. 25 Joseph CG, Bauzo RM, Xiang Z, Shaw AM, Millard WJ & Haskell-Luevano C (2003) Elongation studies of the human agouti-related protein (AGRP) core dec- apeptide (Yc[CRFFNAFC]Y) results in antagonism at the mouse melanocortin-3 receptor. Peptides 24, 263– 270. 26 Krause S & Friedrich K (2005) A microscale assay for the identification of TGF-b antagonists based on func- tional coupling of the heterodimeric TGF-b receptor to Cystine knot miniprotein TPO agonists S. Krause et al. 94 FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS STAT6-driven promoter activation. Signal Transduction 4, 177–183. 27 Krause S, Wu ¨ rdemann D, Wentzel A, Christmann A, Fehr H, Kolmar H & Friedrich K (2004) Bacteria dis- playing interleukin-4 mutants stimulate mammalian cells and reflect the biological activities of variant soluble cytokines. Chembiochem 5, 804–810. 28 Colgrave ML & Craik DJ (2004) Thermal, chemical, and enzymatic stability of the cyclotide kalata B1: the importance of the cyclic cystine knot. Biochemistry 43, 5965–5975. 29 Clark RJ, Fischer H, Dempster L, Daly NL, Rosengren KJ, Nevin ST, Meunier FA, Adams DJ & Craik DJ (2005) Engineering stable peptide toxins by means of backbone cyclization: stabilization of the alpha-cono- toxin MII. Proc Natl Acad Sci USA 102, 13767–13772. 30 Schneider H, Chaovapong W, Matthews DJ, Karkaria C, Cass RT, Zhan H, Boyle M, Lorenzini T, Elliott SG & Giebel LB (1997) Homodimerization of erythropoie- tin receptor by a bivalent monoclonal antibody triggers cell proliferation and differentiation of erythroid precur- sors. Blood 89, 473–482. 31 Fuh G & Wells JA (1995) Prolactin receptor antagonists that inhibit the growth of breast cancer cell lines. J Biol Chem 270, 13133–13137. 32 Perelson A & DeLisi C (1980) Receptor clustering on a cell surface. I. Theory of receptor cross-linking by ligands bearing two chemically identical functional groups. Mathemat Biosciences 48, 71–110. 33 Stephenson RP (1956) A modification of receptor the- ory. Br J Pharmacol Chemother 11, 379–393. 34 Stefanich E, Senn T, Widmer R, Fratino C, Keller GA & Fielder PJ (1997) Metabolism of thrombopoietin (TPO) in vivo: determination of the binding dynamics for TPO in mice. Blood 89, 4063–4070. 35 Li J, Xia Y & Kuter DJ (1999) Interaction of thrombo- poietin with the platelet c-mpl receptor in plasma: bind- ing, internalization, stability and pharmacokinetics. Br J Haematol 106, 345–356. 36 Avrutina O, Schmoldt HU, Kolmar H & Diederichsen U (2004) Fmoc-assisted synthesis of a 29-residue cystine-knot trypsin inhibitor containing a guaninyl amino acid at the P1-position. Eur J Org Chem 2004, 4931–4935. 37 Daly NL, Love S, Alewood PF & Craik DJ (1999) Che- mical synthesis and folding pathways of large cyclic polypeptides: studies of the cystine knot polypeptide kalata B1. Biochemistry 38, 10606–10614. 38 Tam JP, Lu YA, Yang JL & Chiu KW (1999) An unu- sual structural motif of antimicrobial peptides contain- ing end-to-end macrocycle and cystine-knot disulfides. Proc Natl Acad Sci USA 96, 8913–8918. 39 McNulty JC, Jackson PJ, Thompson DA, Chai B, Gantz I, Barsh GS, Dawson PE & Millhauser GL (2005) Structures of the agouti signaling protein. J Mol Biol 346, 1059–1070. 40 Harker LA, Marzec UM, Hunt P, Kelly AB, Tomer A, Cheung E, Hanson SR & Stead RB (1996) Dose– response effects of pegylated human megakaryocyte growth and development factor on platelet production and function in nonhuman primates. Blood 88, 511–521. 41 Sun PD & Davies DR (1995) The cystine-knot growth- factor superfamily. Annu Rev Biophys Biomol Struct 24, 269–291. 42 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 43 Schmoldt HU, Wentzel A, Becker S & Kolmar H (2005) A fusion protein system for the recombinant pro- duction of short disulfide bond rich cystine knot pep- tides using barnase as a purification handle. Protein Expr Purif 39, 82–89. 44 Ballmaier M, Germeshausen M, Schulze H, Cherkaoui K, Lang S, Gaudig A, Krukemeier S, Eilers M, Strauss G & Welte K (2001) c-mpl mutations are the cause of congenital amegakaryocytic thrombocytopenia. Blood 97, 139–146. 45 Heitz A, Chiche L, Le-Nguyen D & Castro B (1989) 1H, 2D NMR and distance geometry study of the fold- ing of Ecballium elaterium trypsin inhibitor, a member of the squash inhibitors family. Biochemistry 28, 2392– 2398. S. Krause et al. Cystine knot miniprotein TPO agonists FEBS Journal 274 (2007) 86–95 ª 2006 The Authors Journal compilation ª 2006 FEBS 95 . Grafting of thrombopoietin- mimetic peptides into cystine knot miniproteins yields high-affinity thrombopoietin antagonists and agonists Sebastian Krause 1, *, Hans-Ulrich Schmoldt 2, *, Alexander. concept of grafting peptides with desired biological activities onto the structural scaffold of small disulfide-rich inhibitors with cystine knot (CK) folds that display high intrinsic stability and rigidity. 2006) doi:10.1111/j.1742-4658.2006.05567.x Thrombopoietin is the primary regulator of platelet production. We exploi- ted two naturally occurring miniproteins of the inhibitor cystine knot fam- ily as stable and rigid scaffolds