Cellular signaling and gene expression profiles evoked by a bivalent macrocyclic peptide that serves as an artificial MET receptor agonist

Non-native ligands for growth factor receptors that are generated by chemical synthesis are applicable to therapeutics. However, non-native ligands often regulate cellular signaling and biological responses in a different manner than native ligands. Generation of surrogate ligands comparable to native ligands is a challenging need. Here we investigated changes in signal transduction and gene expression evoked by a bivalent macrocyclic peptide (aMD5-PEG11) capable of high-affinity binding to the MET/hepatocyte growth factor (HGF) receptor. Binding of aMD5-PEG11 to the MET extracellular region was abolished by deletion of the IPT3−IPT4 domain, indicating the involvement of IPT3−IPT4 in the binding of aMD5-PEG11 to the MET receptor. aMD5-PEG11 induced dimerization and activation of the MET receptor and promoted cell migration that was comparable to induction of these activities by HGF. Signal activation profiles indicated that aMD5-PEG11 induced phosphorylation of intracellular signaling molecules, with a similar intensity and time dependency as HGF. In 3-D culture, aMD5-PEG11 as well as HGF induced epithelial tubulogenesis and up-regulated the same sets of functionally classified genes involved in multicellular organism development. Thus, a non-native surrogate ligand that consisted of a bivalent macrocyclic peptide can serve as an artificial MET receptor agonist that functionally substitutes for the native ligand, HGF.


Results
Mapping of the MET receptor binding domain. The extracellular region of the MET receptor is composed of a semaphorin (SEMA) domain, a plexin-semaphorin-integrin (PSI) domain (similar in structure to the plexins, semaphorins and integrins), and four immunoglobulin-like plexin transcription (IPT) factor domains (similar in structure to the immunoglobulin-like fold shared by the plexins and transcription factors) 14,15 . The macrocyclic portion of aMD5-PEG11 is composed of 15 amino acids (Fig. 1A). aMD5 binds to the extracellular region of the MET receptor with a K D value of 2.3 nM 13 ; the binding region was not determined. To determine the binding domain of the MET receptor for aMD5, the association between aMD5 monomer and the full-length was analyzed with surface plasmon resonance (SPR) (Fig. 1B,C). A clear association between aMD5 and the fulllength MET receptor extracellular region was seen at a K D value of 11 nM, whereas the association was lost when the IPT3-IPT4 or IPT1-IPT4 domains were deleted from the full-length extracellular region (Fig. 1C). On the other hand, previous reports indicated that the SEMA domain is responsible for binding to HGF 16,17 . Therefore, the binding region in the MET receptor is different between HGF and aMD5-PEG11. Consistent with this, aMD5 neither competitively inhibited HGF-induced MET receptor phosphorylation in cultured cells (Fig. 1D) nor inhibited the protein-protein association between HGF and MET receptor-Fc (Fig. 1E).

MET receptor dimerization and activation triggered by bivalent macrocyclic peptides.
Phosphorylation of Y1234/1235 residues within the MET receptor tyrosine kinase domain is an initial hallmark of MET receptor activation 18 . EHMES-1 and normal human renal proximal tubular epithelial cells (RPTEC) were stimulated with aMD5-PEG11 or HGF, respectively, and MET receptor activation was analyzed using a cell-based enzyme-linked immunosorbent assay, which selectively detects phosphorylated Y1234/1235 in the MET receptor ( Fig. 2A). HGF triggered maximal MET receptor activation at concentrations of 1-10 nM, whereas MET receptor phosphorylation decreased from the maximal level with higher concentrations of HGF. aMD5-PEG11 increased MET receptor phosphorylation at higher concentrations compared with HGF. The maximal MET receptor activation by aMD5-PEG11 was seen at 120 nM, and the level was nearly comparable to that induced by HGF ( Fig. 2A, left). MET receptor phosphorylation decreased from the maximal level with higher concentrations of aMD5-PEG11. Similar to the result in EHMES-1 cells, aMD5-PEG11 induced maximal MET receptor activation  in a comparable manner as HGF and showed a bell-shaped profile in normal human renal proximal tubular epithelial cells (RPTEC) ( Fig. 2A, right). A critical first step for trans-phosphorylation of the MET receptor tyrosine kinase domain is considered to be dimerization of the MET receptor. We performed chemical cross-linking to analyze the MET receptor dimerization on the cell surface in cultured cells and detected MET receptor monomers and dimers with western blotting (Fig. 2B). Addition of aMD5-PEG11 induced dimerization of the MET receptor in a concentration-dependent manner. Maximal MET receptor dimerization was seen at 600 nM, whereas the level of dimerization clearly decreased at a higher concentration, showing a bell-shaped profile. Because MET receptor dimerization induced by aMD5-PEG11 showed similar concentration dependency and a bell-shaped profile as MET receptor tyrosine phosphorylation, the results indicate that MET receptor activation by aMD5-PEG11 is associated with the ability to induce dimerization of the MET receptor.
Cell migration induced by the HGF-MET receptor pathway plays a critical role in embryonic development and wound healing [19][20][21][22][23] . To determine if MET receptor activation by aMD5-PEG11 produced biological responses, we evaluated the ability of aMD5-PEG11 to induce cell migration. HGF potently facilitated migration of HuCCT1 cells in a dose-dependent manner (Fig. 2C). aMD5-PEG11 also facilitated cell migration with a maximal activity that was comparable to HGF. The comparable activity of aMD5-PEG11 to HGF seemed to be consistent with MET receptor activation.
Bell-shaped profiles for MET activation were also observed in previous studies 13,24,25 , suggesting a basic profile for receptor activation by bivalent ligands. [R 2 L], the concentration of dimerized receptor complexed with a bivalent ligand, changes according to the equation (Fig. 2D, right). Because activation of a receptor is proportional to Cellular signaling evoked by bivalent macrocycles. MET receptor activation triggers downstream cellular signaling cascades involving phosphorylation of protein kinases. Previous studies demonstrated that HGF facilitates phosphorylation of a variety of kinases, including ERK1/2 and its downstream molecule CREB 26,27 , AKT and its downstream molecules such as PRAS40 28 , and STAT3 29 . PRAS40 Thr246 phosphorylation by AKT activates the mTORC1 pathway, thereby connecting the AKT and mTOR signaling pathways 30,31 . STAT3 and CREB are signal-transducing transcription factors involved in cell proliferation and survival 27,32,33 . Prior to detailed characterization and comparison of signal activation between aMD5-PEG11 and HGF, we confirmed the specificity of aMD5-PEG11 for the MET receptor by utilizing a selective inhibitor of the MET receptor. aMD5-PEG11 as well as HGF induced MET tyrosine phosphorylation and downstream AKT phosphorylation (Fig. 3A). The selective MET tyrosine kinase inhibitor PHA665752 almost completely blocked the activation of the MET receptor and AKT to basal levels. This result indicates that aMD5-PEG11 activates AKT through selective activation of the MET receptor.
To investigate the differences in characteristics of intracellular signal activation between aMD5-PEG11 and HGF, we assessed the time-dependent phosphorylation/activation involving ERK, AKT, PRAS40, STAT3, and CREB with western blotting (Fig. 3B,C). aMD5-PEG11 induced strong phosphorylation of ERK and AKT after 10 min, followed by a decrease over time. HGF induced similar time-dependent phosphorylation of ERK and AKT with a strength and profile comparable to aMD5-PEG11. Furthermore, the phosphorylation kinetics for the downstream molecules PRAS40, STAT3, and CREB induced by aMD5-PEG11 was also very similar to that obtained with HGF.
To systematically assess the activation of cellular signaling through protein phosphorylation, we analyzed the changes in the phosphorylated signaling proteins in EHMES-1 cells treated with aMD5-PEG11 or HGF, using a phospho-kinase array (Fig. 4A). Both aMD5-PEG11 and HGF enhanced phosphorylation of a variety of signaling proteins, and clear enhancement was seen with STAT3, PRAS40, and WNK1 (Fig. 4A,B). Overall, we did not find signaling pathways that were selectively enhanced by either aMD5-PEG11 or HGF (Fig. 4B). Collectively, aMD5-PEG11 elicited activation/phosphorylation of a variety of intracellular signaling molecules at a similar strength and with similar kinetics as HGF.
Changes in the gene expression profile. To further evaluate the functional compatibility between aMD5-PEG11 and HGF, the gene expression profile was analyzed with the Whole Human Genome Oligo Microarray. Because induction of epithelial tubulogenesis is unique to the HGF-MET receptor pathway 34 , we assayed this biological response to analyze the gene expression profile. When normal human renal proximal epithelial tubular cells were cultured in collagen gel for 8 days, aMD5-PEG11 as well as HGF induced tubulogenesis, indicating a MET receptor-mediated dynamic biological response (Fig. 5A).
To analyze changes in gene expression, the cells were cultured in collagen gel, treated with aMD5-PEG11 or HGF for 8 h, and total RNA was applied to the Microarray. Genes that showed a fold change larger than 1.5 (increase or decrease) were selected, and then further selected when the p value was less than 0.05 between two independent samples of HGF-and aMD5-PEG11-treated groups. Heat maps were prepared, and Gene Ontology analysis was performed based on these processed data. aMD5-PEG11 and HGF increased the same sets of functionally classified genes, including genes involved in multicellular organism development, anatomical structure development, multicellular organism processes, developmental processes, and system development (Fig. 5B)  morphogenesis and organization of functional renal development, the result indicates that aMD5-PEG11 can orchestrate the gene expression network involved in a complex multicellular process, in a comparable manner to HGF. In addition to genes involved in development and morphogenesis, both aMD5-PEG11 and HGF increased expression of genes involved in negative regulation of protein kinases. Because MET receptor activation by aMD5-PEG11 and HGF triggered phosphorylation/activation of a variety of protein kinases (Fig. 4), the induction of these genes may represent a feedback mechanism following activation of multiple pathways. The heat map diagram indicates that aMD5-PEG11 increased or inhibited expression of a variety of genes ( Fig. 5C) (raw data for the heatmap are provided by Dataset 1). Changes in several genes were distinguishable between aMD5-PEG11 and HGF. Overall, significantly up-regulated or down-regulated genes by aMD5-PEG11 overlapped those affected by HGF. We conclude that aMD5-PEG11 can elicit the authentic MET receptor signaling cascade and function as a mimetic of the physiological ligand, HGF.

Discussion
Different types of non-native ligands capable of inducing dimerization/oligomerization of growth factor/ cytokine receptors can evoke receptor-mediated signal activation. The apparent permissiveness and flexibility in dimer architecture needed for signal activation allowed us to test the possibility that different receptor-ligand dimer architectures can influence the signal strength and induce distinct signaling, leading to distinct biological responses. For example, covalently linked fragments of the VH/VL variable domain of antibodies against the extracellular domain of the EPO receptor generate agonistic bivalent molecules called "diabodies". EPO receptors dimerized by diabodies with distinct epitopes elicit biased and distinct activation of signaling pathways compared to the native ligand, EPO 5 . The different dimer architectures and flexibility of growth factor/cytokine receptors may allow distinct signaling, as well as uneven signal strength.
A previous study showed that bivalent monoclonal antibodies with different epitopes in the extracellular domain of the MET receptor induce different biological responses: one monoclonal antibody enhances only cell motility, but the other monoclonal antibody triggers cell motility, proliferation, and epithelial tubulogenesis 35  receptor and enhances cell migration and proliferation with less potency compared to HGF 36 . NK2, another variant of HGF, promotes cell motility and survival but does not promote cell proliferation 37 . Artificial agonistic ligands for the MET receptor were generated by oligonucleotide-based screening 38 . The maximal activity of the oligonucleotide-based ligand exhibits maximal biological activities for enhanced MET receptor activation, cell proliferation, and migration, with a comparable ability as HGF. These results indicate that distinguishable activation of signaling molecules and subsequent biological responses can be elicited by distinct ligands that may induce different dimer/oligomer conformations of the MET receptor.
In the present study, we investigated the details of signal transduction and gene expression profiles evoked by the PEG-linked macrocyclic peptide, aMD5-PEG11. We obtained evidence that this PEG-linked bivalent macrocyclic peptide can induce activation of MET receptor-mediated signal transduction pathways and gene expression profile in a manner that was mostly indistinguishable from that of the native HGF protein. aMD5 binds to the structure determined by IPT3−IPT4 domains but does not bind to the SEMA domain to which HGF binds. The structural basis for MET receptor activation by the artificial surrogate ligand, aMD5-PEG11, and the native ligand, HGF, may be explained by several possibilities. MET receptor dimer conformation induced by aMD5-PEG11 may be either (i) similar to that induced by HGF even if the binding regions are different, or (ii) different from that induced by HGF, if the MET receptor has structural plasticity for dimer conformation that allows optimal/suboptimal tyrosine phosphorylation events in the MET receptor intracellular region, at least to some extent.
Tissue-specific disruption of the MET receptor revealed indispensable or important roles for the HGF-MET receptor pathway in the regeneration and protection of tissues, including the liver, kidney, skin, nervous system tissue, pancreas, etc 12 . Administration of HGF is a therapeutic option in preclinical disease models, including models of fulminant hepatitis, liver cirrhosis, amyotrophic lateral sclerosis, and spinal cord injury 11 . Cytokine and growth factor drugs exhibit distinct therapeutic actions because of their intrinsic biological activities, but they have general disadvantages in medical use. For instance, recombinant HGF rapidly disappears from circulating blood with a half-life of 3-5 min following intravenous administration 39 . Manufacturing of protein drugs is costly, whereas artificial growth factors/cytokines such as macrocycles can be manufactured by chemical synthesis. Because of their smaller size, they may show superior tissue permeability compared to protein drugs. Artificial MET receptor agonists such as macrocyclic peptides have the potential to be developed as novel biological drugs manufactured by chemical synthesis. Finally, RaPID selection can be applied to various membrane proteins [40][41][42][43] , including other receptor tyrosine kinases, so that we will be able to obtain macrocycles that specifically bind such designated targets. As many of these transmembrane receptors are dimerized or heterodimerized following an interaction with their cognate ligand to activate signaling pathways 44,45 , our approach for generating dimeric macrocycles as non-protein ligands for cell surface receptors may be useful for developing surrogate ligands with a broad range of potential applications.

Truncated MET receptor extracellular proteins.
For the expression of Fc-fused MET receptor extracellular domains, regions corresponding to residues 1-931 (SEMA-IPT4), 1-741 (SEMA-IPT2), and 1-563 (SEMA-PSI) were PCR amplified from the wild-type full-length human MET receptor cDNA and cloned into a pcDNA3.1-based vector containing the human IgG1 Fc fragment. The expression constructs were used to transiently transfect Expi293F cells (Thermo Fisher) using the ExpiFectamine 293 Transfection Kit (Thermo Fisher), and secreted proteins were purified from the culture supernatant using Protein A-Sepharose (GE Healthcare). All proteins were buffer exchanged to 20 mM Tris, 150 mM NaCl, pH 8.0, and concentrated to 0.2-0.5 mg/ml using an Amicon Ultra centrifugation device (Merck-Millipore, MWCO 30 kDa).

SPR analysis.
For biotin labeling of macrocycles, aMD5-Lys(Mmt)-NH-resin was synthesized by the Fmoc solid-phase peptide synthesis method, and the Mmt group was then deprotected by a solution of 98% dichloromethane, 1% trifluoroacetic acid, and 1% triisopropylsilane. The resulting aMD5-Lys-NH-resin was equilibrated with 20% N,N-diisopropylethylamine in N-methylpyrrolidone and treated with 0.2 M NHS-biotin in N-methylpyrrolidone. The modified peptide was deprotected/cleaved and macrocyclized, as described above, followed by purification with reverse-phase high-performance liquid chromatography and in vacuo lyophilization. C-terminal biotinylated macrocycles were captured on a streptavidin-immobilized SPR sensor chip at a surface density of 200-500 RU using a Biotin Capture Kit (GE Healthcare). The binding constants of the macrocycles to the MET receptor extracellular domains were determined by SPR analysis using Biacore T200 (GE Healthcare). The running buffer was HBS EP+ buffer (10 mM Hepes pH 7.4, 150 mM NaCl, 3 mM ethylenediaminetetraacetic acid, and 0.05% (v/v) SurfactantP20) containing 0.1% dimethylsulfoxide. MET receptor binding was tested by injecting varying concentrations of the MET receptor extracellular domain-Fc fusion proteins at 30 μl/min and quantifying with the single-cycle kinetics method. All data were fitted to the standard 1:1 binding model.

Cell-based phospho-MET receptor measurement. EHMES-1 cells were seeded at 8 × 10 3 cells per
well in a 96-well black micro-clear plate (Greiner Bio-One) and cultured for 24 h. The cells were treated with HGF or MET receptor-binding dimeric macrocycle in RPMI-1640 medium supplemented with 10% FBS for 10 min, washed once with ice-cold phosphate-buffered saline (PBS), and fixed with 4% paraformaldehyde in PBS for 30 min at room temperature. After washing three times with PBS, the cells were blocked with 5% goat serum, 0.02% Triton X-100 in PBS for 30 min at room temperature and incubated with anti-phospho-MET (Tyr1234/1235) (D26) XP rabbit mAb (1:1000 in PBS with 1% goat serum) at 4 °C overnight. The cells were washed three times with PBS and incubated in horseradish peroxidase (HRP)-conjugated goat anti-rabbit antibody (1:1000 in PBS with 1% goat serum) for 1 h. After washing four times with PBS, tyrosine phosphorylated MET receptor was detected with ImmunoStar LD reagent (Wako) and measured using ARVO MX (Perkin Elmer). The relative MET receptor phosphorylation level was calculated as (sample chemiluminescence unit − mock control chemiluminescence unit)/(highest chemiluminescence unit − mock control chemiluminescence unit). Normal human renal proximal tubule epithelial cells were seeded at 2 × 10 4 cells per well in a 96-well black micro-clear plate and cultured for 24 h. The cells were treated with aMD5-PEG11 or HGF for 10 min. The subsequent procedure was the same as described above. Mathematical modeling for receptor activation by a bivalent ligand. A model for complex formation between the receptor (R) and ligand (L) is expressed by two reactions:

Binding assay between fluorescein-HGF and MET-receptor
where K 1 and K 2 are the equilibrium constants. Equilibrium equations are , Western blot analysis. EHMES-1 cells were seeded at 9 × 10 5 cells per well in 6-well plates and cultured overnight. To treat the cells with the MET receptor tyrosine kinase inhibitor, EHMES-1 cells were seeded at 3 × 10 5 cells per well in 24-well plates and cultured with or without 100 nM PHA665752 for 12 h. After serum starvation for 4 h, cells were stimulated with HGF or dimeric macrocycle peptide. After washing with ice-cold PBS, cells were lysed in 200 μl 1 × SDS-PAGE Laemmli sample buffer and treated with ultrasonification (Vibra-cell). Cell lysates were analyzed by SDS-PAGE with a 7.5% polyacrylamide gel and electroblotted onto a PVDF membrane. The membrane was treated with primary antibodies (1:1000), followed by HRP-conjugated secondary antibodies (Dako) (1:2000). Chemiluminescence was visualized and quantitated using ImmunoStar LD (Wako). Original blots are provided by supplementary info.
Phospho-kinase array. The phospho-kinase array was performed using the Proteome Profiler Human Phospho-Kinase Array Kit (R&D Systems). EHMES-1 cells were cultured at 1.5 × 10 6 cells per plate in 60-mm dishes for 12 h. After serum starvation for 4 h, the cells were stimulated with dimeric macrocycle peptide or HGF for 10 min. Cells lysates were prepared, and phosphorylated protein kinases were detected, according to the manufacturer's instructions. Chemiluminescence was detected using ImageQuant LAS 350 (GE Healthcare).
Tubulogenesis assay. Normal human renal proximal tubule epithelial cells (2 × 10 4 cells) were seeded in 400 μl Cellmatrix type IA collagen solution (Nitta Gelatin) in a 48-well plate and allowed to stand for 60 min at 37 °C for gelling, according to the manufacturer's instructions. After gelation, 500 μl renal epithelial basal medium supplemented with 0.5% FBS and 2.4 mM l-alanyl-l-glutamine, with or without HGF or dimeric macrocycle peptide, was added to each well. Cells were cultured for 8 days, and the culture medium (with or without HGF or aMD5-PEG11) was refreshed every 2 days. mRNA profiling array. For the mRNA array, normal human renal proximal tubule epithelial cells were seeded at 2 × 10 5 cells per well in 1.2 ml Cellmatrix type IA collagen solution (Nitta Gelatin) in a 12-well plate and allowed to stand for 60 min at 37 °C for gelling, according to the manufacturer's instructions. Culture medium (800 μl) supplemented with 0.5% FBS and 2.4 mM l-alanyl-l-glutamine was added, and the cells were cultured for 12 h. Then, the cells were cultured in the absence or presence of dimeric macrocycle peptide or HGF for 8 h. Cells were harvested, and total RNA was prepared using Seposal-RNA I Super G (Nacalai Tesque), according to the manufacturer's instructions. Microarray analyses were performed using the Whole Human Genome (4 × 44 k, G2565BA) Oligo Microarray, according to the Agilent 60-mer Oligo Microarray Processing Protocol (Agilent Technologies). Total RNA samples (200 ng) were used to prepare Cy3-labeled cRNA using a Low RNA Input Fluorescent Linear Amplification Kit (Agilent Technologies). Fluorescence-labeled cRNAs were purified using an RNeasy RNA Purification Kit (Qiagen Inc., Hilden, Germany). RNA samples obtained by duplicate experiments performed independently were used to confirm the reproducibility of the microarray analyses. The images were analyzed using Feature Extraction Software (Ver. 10.7.3.1) and GeneSpring GX 11.5 software (Agilent Technologies). Normalization was performed as follows: (i) intensity-dependent Lowess normalization; (ii) data transformation, with measurements set to ≤0.01; (iii) per-chip 75 th percentile normalization of each array; and, (iv) per-gene: normalized to the median of each gene. Genes with more than a 1.5-fold difference in expression between aMD5-PEG11-treated and HGF-treated groups were selected and used for the Gene Ontology enrichment analysis, which was performed using DAVID Bioinformatics Resources 6.7. The raw and processed data were deposited in the Gene Expression Omnibus database (access ID: GSE111542).