Cell Biochem Biophys DOI 10.1007/s12013-017-0810-9

ORIGINAL PAPER

Integrated In Silico–In Vitro Identification and Characterization of the SH3-Mediated Interaction between Human PTTG and its Cognate Partners in Medulloblastoma Jiangang Liu1 Dapeng Wang1 Yanyan Li1 Hui Yao1 Nan Zhang1 Xuewen Zhang1 Fangping Zhong1 Yulun Huang1 ●

●

●

●

●

●

●

Received: 29 January 2016 / Accepted: 9 June 2017 © Springer Science+Business Media, LLC 2017

Abstract The human pituitary tumor-transforming gene is an oncogenic protein which serves as a central hub in the cellular signaling network of medulloblastoma. The protein contains two vicinal PxxP motifs at its C terminus that are potential binding sites of peptide-recognition SH3 domains. Here, a synthetic protocol that integrated in silico analysis and in vitro assay was described to identify the SH3-binding partners of pituitary tumor-transforming gene in the gene expression profile of medulloblastoma. In the procedure, a variety of structurally diverse, non-redundant SH3 domains with high gene expression in medulloblastoma were compiled, and their three-dimensional structures were either manually retrieved from the protein data bank database or computationally modeled through bioinformatics technique. The binding capability of these domains towards the two PxxP-containing peptides m1p: 161LGPPSPVK168 and m2p: 168 KMPSPPWE175 of pituitary tumor-transforming gene were ranked by structure-based scoring and fluorescencebased assay. Consequently, a number of SH3 domains, including MAP3K and PI3K, were found to have moderate or high affinity for m1p and/or m2p. Interestingly, the two overlapping peptides exhibits a distinct binding profile to these identified domain partners, suggesting that the binding selectivity of m1p and m2p is optimized across the

Electronic supplementary material The online version of this article (doi:10.1007/s12013-017-0810-9) contains supplementary material, which is available to authorized users. Jiangang Liu and Dapeng Wang contributed equally to this work. * Yulun Huang [email protected] 1

Department of Neurosurgery, the First Affiliated Hospital of Soochow University, Suzhou 215006, China

medulloblastoma expression spectrum by competing for domain candidates. In addition, two redesigned versions of m1p peptide ware obtained via a structure-based rational mutation approach, which exhibited an increased affinity for the domain as compared to native peptide. Keywords Pituitary tumor-transforming gene SH3 domain Peptide motif Medulloblastoma ●

●

●

Introduction Human pituitary tumor-transforming gene (PTTG) is an important paracrine growth factor involved in early lactotrope transformation and early onset of angiogenesis in pituitary hyperplasia [1]. Serves as an important oncogenic protein, PTTG was initially isolated from rat pituitary cells [2] and is abundantly expressed in endocrine-related tumors, such as pituitary adenomas and other neoplasms [3, 4]. Over the past decade, accumulated evidences suggested that the PTTG also plays an important role in the development, invasion and metastasis of medulloblastoma (MDB) [5], which has recently been recognized as a new therapeutic target for treatment of MDB and glioma [6]. Overexpression of PTTG causes in vitro cell transformation, results in vivo tumor formation, and stimulates basic fibroblast growth factor (FGF) expression and secretion [7, 8]. The C-terminal domain of PTTG is a proline-rich region that contains two PxxP motifs as potential binding sites of the SH3 domains of its partner proteins [9]. Mutation or deletion of the PxxP motifs can totally abolish transforming activity and transactivating ability of the protein, suggesting

Cell Biochem Biophys

that the motifs are required for the biological function of PTTG in transformation, FGF induction and angiogenesis [10, 11]. Serve as a multifunctional regulator, PTTG has been found to interact with various proteins in the cell signaling network of MDB. For example, some protein kinases, such as MAP3K and PI3K have been implicated in PTTG phosphorylation [12, 13]; they contain a noncatalytic SH3 domain that functions as regulatory module to determine kinase’s substrate specificity by selectively targeting PxxP motif present in the substrate protein. Here, we report a successful integration of in silico analysis and in vitro assay to construct a systematic binding profile of the two PTTG PxxP motifs against a variety of SH3 domains with high expression in MDB, aiming at understanding of the molecular mechanism and biological implication underlying the disease signaling network meditated by PTTG. We also performed peptide docking, molecular dynamics (MD) simulations, binding energetic analysis and fluorescence-based assay to elucidate the selectivity and specificity of PTTG PxxP peptides over the identified SH3 domain partners.

Materials and Methods SH3 Collection More than 1000 proteins were observed to exhibit a significantly high expression profile in the transcriptomes of 20 different human MDBs as compared to normal cells [14], from which we identified 71 proteins containing 86 SH3 domains by manually examining and comparing them to the reported SH3-containing proteins in human genome [15]. Subsequently, cross sequence alignment between these identified domains were performed in a pairwise manner using ClustalW program [16], and those with sequence identity > 85% were removed to define a distinct set consisting of 69 structurally diverse, non-redundant SH3 domains, which come from 54 highly expressed proteins in MDB, including 28 kinases and six scaffold proteins (see Supplementary Materials Table S1); all of them are thought to participate directly or indirectly in the cellular signaling network and metabolic pathway of MDB. Structural Curation and Complex Modeling SH3 structure By examining the Protein Data Bank (PDB) database [17], 32 out of the 69 collected SH3 domains were found to have solved crystal or solution structures, while for others their structures were modeled using homology modeling approach [18]. The primary sequences of these 37 modeled

domains are tabulated in SM Table S1, with sequence length ranging between 50 and 70 amino acids long. A high-throughput, fully automated structure modeling method implemented in the SWISS-MODEL workspace [19] was carried out to fast predict atomic-level structures for the modeled domains, which performed sequence search against the PDB database [17] to select target templates that have high sequence identity with the queried domains. Considering that all SH3 domains are highly conserved that share similar primary sequence and structure architecture, the homologous modeling technique should be a good choice to construct three-dimensional SH3 domain structure models using experimentally determined structures of related domains as templates. The missing side chains of amino acid residues in crystal or solution structures were added with SCWRL program [20, 21] and then minimized with YASARA server [22]. SH3-peptide complex structure The C-terminal domain of PTTG contains two SH3-binding motifs 163PPSP166 and 170PSPP173. Considering that the residues flanking to the motifs can also contribute to SH3 binding affinity and specificity, each side of a motif was extended with two additional residues to define an 8-mer sequence as the potential binding peptides of SH3 domains. Consequently, two overlapping octapeptides m1p: 161 LGPPSPVK168 and m2p: 168KMPSPPWE175 separately covering the two motifs were generated [23]. Subsequently, the two peptides were docked to the recognition sites of the 68 SH3 domains in a standard SH3–peptide binding manner with PepCrawler server [24]. The obtained coarse-grained complex structures were then refined using the subangstrom FlexPepDock method [25], which refined the peptide-binding mode in SH3 pocket to high resolution, allowing full flexibility to the peptide backbone and to all side chains. MD Simulation and Binding Energetic Calculation Several selected SH3–peptide complex systems were subjected to MD simulations using AMBER ff03 force field [26] implemented in AMBER11 suite of programs [27]. The input files for the simulations were prepared in tleap module, in which the system was hydrated by TIP3P water box [28] and neutralized by Na+ counter ions. First, only water was relaxed by 500 steps of steepest descent and 1000 steps of conjugate gradient. Then, the entire system was minimized using 500 steps of steepest descent followed by 3000 steps of conjugated gradient. Heating of the system was conducted at 300 K using Langevin thermostat. The SHAKE method was employed to constrain covalent bonds involving hydrogen atoms [29], and the particle mesh

Cell Biochem Biophys

Ewald method was used to calculate the full electrostatic energy of a unit cell in a macroscopic lattice of repeating images [30]. Whole system equilibration was performed over a period of 1 ns. Production phase was carried out in the Number of particles (N), Pressure (P), and Temperature (T) ensemble at 300 K and 1 atm pressure. Nonbonded cutoff of 10 Å and step size of 2 fs was applied for the simulations [31]. The snapshots of domain–peptide complex were saved every 2 ps during the equilibration simulations, which were then subjected to molecular mechanics/ Poisson–Boltzmann surface area (MM/PBSA) analysis for binding energetic calculations [32]. The total binding free energy ΔGtotal of complex can be decomposed into the intermolecular interaction potential ΔGint between the domain and peptide, the desolvation effect ΔGdslv due to the binding, and the entropy penalty –TΔS upon the binding. The ΔGint was computed using force field-based approach using AMBER03 force field [26]. The ΔGdslv was described by numerical solution of nonlinear Poisson–Boltzmann equation for polar contribution plus surface area model for nonpolar contribution, and the – TΔS was estimated thought normal mode analysis with a distance-dependent dielectric constant was utilized to mimic solvent screening [33]. Fluorescence-based Affinity Assay The binding affinity between a SH3 domain and a peptide (m1p or m2p) was determined using fluorescence spectroscopy as described previously [34, 35]. The fluorescence emission spectra of Trp residues in the domain protein were used to monitor the changes in chemical microenvironment upon peptide binding. Fluorescence was measured using a phycoerythrin fluorescence spectrophotometer in room temperature. The experimental data were fitted to the equation F = F∞ – [x]/(KD + [x]) [36], where [x] is the

peptide concentration at each measurement point, F is the measured fluorescence intensity of domain protein at a peptide concentration, and F∞ is the observed maximal fluorescence intensity of the protein saturated with the peptide. Here, m1p and m2p peptides were synthesized using Fmoc solid phase chemistry and purified by RPHPLC. The GST-tagged, recombinant proteins of SH3 domains were obtained from Sigma.

Results and Discussion The Binding Profile of Peptide Ligands to MDB SH3 Domains The three-dimensional atomic structures of 69 structurally diverse, non-redundant SH3 domains that are highly expressed in human MDB were either retrieved from the PDB database [17] or modeled via bioinformatics approach. The binding modes of PTTG m1p and m2p peptides to these domain were predicted using peptide docking server PepCrawler [24], and the obtained coarse-grained complex structures were then refined using the sub-angstrom FlexPepDock method [25]. In this way, the complex structures of 69 MDB SH3 domains with m1p and m2p peptides were constructed, based on which the binding potency between the domains and peptides were scored using a QSAR-improved statistical potential PPRCP [37]. The scoring method combines unsupervised knowledge-based statistical potential derived from a arge number of protein–peptide complex structures and supervised quantitative structure–activity relationship (QSAR) modeling trained by protein–peptide interactions with known structure and affinity data. The PPRCP predicted binding score profile of m1p and m2p peptides to the 69 SH3 domains is shown in Fig. 1a, where the score is represented by logarithmic dissociation constant (pKd). As can be seen, most SH3 domains were

Fig. 1 a Histogram representation of the PPRCP scores of m1p and m2p peptides binding to 69 highly expressed SH3 domains in human MDB. b A complementary profile between the PPRCP scores of m1p and m2p peptides binding to 11 potential SH3 domain partners

Cell Biochem Biophys Fig. 2 a Superposition between the crystal and MD equilibrium structures of PI3K SH3 domain (rmsd = 0.29 Å). b Superposition between the apo and holo structures of PI3K SH3 domain (rmsd = 0.57 Å)

predicted to have only a modest or moderate binding capability towards both the two peptides (PPRCP scores < 7), suggesting that a majority of SH3-containing proteins with high expression in MDB are not the cognate partners of PTTG. This is expected if considering that PTTG should only recognize and interact with a few of key regulators in the cellular signaling pathway of MDB. In addition, a different binding profile between the m1p and m2p peptides to SH3 domains can be observed, although the difference is not very significant over all the 69 domains. However, there are 11 domains (Amphiphysin, EEN, Frk, Grap, Hck, MAP3K, Myosin, NESH, PI3K, RasGAP and TIM) indeed exhibit high or very high affinity scores to one or both of the two peptides with PPRCP scores > 9; these potent domains could be thought as potential candidates of the cognate binding partners of PTTG. Here, the PPRCP scores of m1p and m2p peptides to the 11 domains are shown in Fig. 1b. A complementary profile between the PPRCP scores of m1p and m2p peptides binding to these domains can be observed, indicating that the two overlapping peptides can bind to their cognate domains by competing to each other, that is, in most cases the domains can only selectively recognize and interact with one of the two peptides. However, there are two exceptions, that is, MAP3K and PI3K; both of them are Ser/Thr-specific protein kinases and were predicted to bind tightly to the two peptides (PPRCP scores = 11.9 and 10.2 for m1p and 10.6 and 9.4 for m2p, respectively). As known, the PTTG protein contains more than 30 phosphorylatable sites potential as the targets of protein kinases [38]. Thus, it is speculated that the MAP3K and PI3K kinases can recognize and catch PTTG substrate via their SH3 domain and then phosphorylate the substrate through their kinase domains [23].

Structural Analysis, Energetic Calculation and Dynamics Simulation The crystal structures of MAP3K and PI3K SH3 domains can be retrieved from the PDB database [17] under the accession codes 2RF0 and 3I5R, respectively, which were then subjected to 120-ns MD simulations to reach at equilibrium state. Here, superposition of the crystal and MD equilibrium structures of PI3K SH3 domain is shown in Fig. 2a. As can be seen, only a modest difference between the two structures can be observed, with atomic root-meansquare deviation (rmsd) of 0.29 Å, indicating that the crystal structure is already in relaxed state and dynamics simulations would not cause large conformational change in the domain. Next, the structures of MAP3K and PI3K SH3 domains in complex with m1p and m2p peptides were modeled by peptide docking and structural refinement, as described in method section. The four resulting complex systems were separately subjected to 1.5-μs MD simulations, from which the dynamics trajectory was extracted and monitored. The trajectory characterizing distance fluctuation between SH3 and peptide revealed that these complexes are dynamically stable and no considerable conformational change can be observed during the simulations. Superposition of the apo and holo structures of PI3K SH3 domain is shown in Fig. 2b, from which it is evident that the peptide binding can only address a moderate effect on the domain conformation, with rmsd = 0.57 Å. This is expected since a previous structure-based analysis of the crystallographic data of peptide–protein interactions unraveled that most peptides do not induce substantial conformational changes on their partners upon binding, thus minimizing the entropic cost of binding [39]. However, the

Cell Biochem Biophys

side chains of certain residues such as Tyr14, Arg18, Trp55, and Thr72 in the peptide-binding pocket of SH3 domain have a considerable motion upon the peptide binding, although their backbone and location alter limitedly. After MD simulations, it is found that the complex systems of m1p peptide with both PI3K and MAP3K SH3 domains can be well kept in a standard SH3–peptide binding mode, where the peptide ligand is folded into a polyproline II (PPII) helix [40] and tightly bound in peptide-binding pocket of the domains. As listed in Table 1, the binding free energy ΔGtotal of m1p peptide to the two domains was calculated as −4.5 and −6.8 kcal/mol, respectively, by using MD-based MM/PBSA and NMA analysis, indicating a spontaneous binding in the domain–peptide recognition. The energy can be decomposed into intermolecular interaction potential (ΔEint = −63.9 and −72.3 kcal/mol), desolvation effect (ΔGdslv = 22.7 and 31.5 kcal/mol) and entropy penalty (−TΔS = 36.7 and 34.0 kcal/mol). It is revealed that interaction potential contributes positively to the domain–peptide binding, Table 1 The calculated binding energetics and experimental binding affinity of peptide ligands to MAP3K and PI3K SH3 domains Domain–peptide complex

Binding energetics (kcal/mol) ΔEint

Affinity Kd (µM)

ΔGdslv –TΔS ΔGtotal

PI3K–m1p

−63.9 22.7

36.7

−4.5

174 ± 18

MAP3K–m1p

−72.3 31.5

34.0

−6.8

87 ± 9

PI3K–m2p

−38.2 21.3

15.7

−1.2

n.d.a

MAP3K–m2p

−46.9 23.3

20.1

−3.5

149 ± 12

PI3K–m1p (L161R)

−67.4 25.0

35.2

−7.2

31 ± 4

PI3K–m1p (V167I)

−66.4 25.5

34.6

−6.3

76 ± 5

a

not detectable

Fig. 3 The MD equilibrium structures of PI3K SH3 domain in complex with m1p a and m2p b peptides. The peptide-binding pocket of domain consists of residues Tyr12, Asp13, Tyr14, Glu17, Arg18, Asp21, Glu51, Ile53, Gly54, Trp55, Asp68, Pro70, Gly71, Thr72, and Tyr73 for m1p peptide and residues Tyr14, Lys15, Lys16, Glu17, Arg18, Glu20, Asp21, Gly35, Ser36, Ala39, Leu40, Trp55, Leu56, Asn57, Arg66, Gly67, Asp68, and Phe69 for m2p peptide

which, however, could be largely offset by the unfavorable desolvation effect and entropy penalty. Consequently, a moderate affinity ΔGtotal between the two domains and m1p peptide was predicted with theoretical approach. This can be confirmed by fluorescence-based assays; the peptide was measured to bind with PI3K and MAP3K SH3 domains with Kd values of 174 and 87 µM, respectively, suggesting a moderate affinity in the domain–peptide recognition. According to the simulations, a considerable difference between the MD equilibrium conformations of m1p and m2p peptides in complex with PI3K SH3 domain can be observed; the m1p always holds in its initial binding mode with a PPII helix conformation (Fig. 3a), while the m2p cannot bind stably to the domain in its initial state; the peptide ligand moves by a large distance during simulations and finally stabilizes in a side region of the domain (Fig. 3b). MM/PBSA and NMA analysis revealed a weak interaction potential (ΔGtotal = −1.2 kcal/mol) between the domain and peptide, which is composed of intermolecular interaction potential ΔEint = −38.2 kcal/mol, desolvation effect ΔGdslv = 21.3 kcal/mol and entropy penalty –TΔS = 15.7 kcal/mol. No binding between the PI3K domain and m2p peptide was observed by fluorescence-based assays (Kd = n.d.), while a moderate affinity was revealed for the peptide binding to MAP3K domain (Kd = 149 µM) (Table 1).

Mutational Modification and Redesign of m1p Peptide It is demonstrated that the binding selectivity of m1p and m2p peptides is optimized across the MDB expression spectrum by competing for their common domain candidates, and the Ser/Thr-specific protein kinases PI3K and

Cell Biochem Biophys Fig. 4 a The modeled complex structure of PI3K SH3 domain with m1p peptide. b Hydrophobic potential map on the domain surface, where the N terminus, central part and C terminus of the peptide ligand correspond to hydrophilic, hydrophobic and amphiphilic regions on the domain surface, respectively

MAP3K exhibit differential affinity for the two peptides. The PI3K is a critical signal transduction system linking oncogenes and multiple receptor classes to many essential cellular functions, which is perhaps the most commonly activated signaling pathway in human cancer [41]. Here, we found that the m1p peptide can bind PI3K SH3 domain with a PPII helix comformation and exhibits a moderate affinity between them (Kd = 174 µM). Thus, a mutational modification and redesign of the peptide was performed, in order to improve its domain-binding capability. According to the computationally modeled complex structure of the domain with the peptide (Fig. 4a), the positively charged Lys168 residue at peptide C terminus can form several salt bridges with domain residues Asp13 and Glu51. A number of hydrogen bonds are also observed across the backbones of domain and the core region 163PPSP166 of peptide, conferring large stability and strong specificity to the complex system. By examining hydrophobic potential map on domain surface, it is evident that the peptide-binding pocket of SH3 domain can be divided into hydrophilic, hydrophobic and amphiphilic regions corresponding to the N terminus, central portion and C terminus of peptide ligand, respectively (Fig. 4b). The central portion covers the core binding motif PxxP of peptide ligand, which defines hydrophobic interactions and a network of hydrogen bonds across the domain–peptide complex interface. Considering that the PxxP motif is critical for peptide binding, the central portion of peptide was fixed during the redesign. The N terminus of peptide is very close to a negatively charged region (Asp13 and Glu51) of the domain, and we therefore considered mutating the nonpolar residue Leu161 to a positively charged residue Arg at the N terminus to match the electrostatic complementarity between domain and peptide, resulting in a mutated version of m1p peptide, namely m1p (L161R): 161RGPPSPVK168. On the other side, the Cterminal residue of Lys168 can also form several electrostatic salt bridges with the domain, but the vicinal nonpolar

residue Val167 seems to contact the hydrophobic patch of domain surface. Here, we considered mutating the residue to a nonpolar, bulky and flexible amino acid Ile to generate the m1p (V167I): 161LGPPSPIK168. The complex structures of PI3K SH3 domain with the two redesigned versions of m1p peptides were computationally modeled by using peptide docking, structural refinement, and MD simulations. As might be expected, the redesigned peptides can be stably bound with the domain in a PPII conformation during the simulations, which were theoretically predicted to have an improved binding energy as compared to the wild-type m1p peptide, with ΔGtotal increase from −4.5 to −7.2 and −6.3 kcal/mol for the m1p (L161R) and m1p (V167I) peptides, respectively. To verify the redesigning, binding affinities of the two mutant peptides were measured fluorescently as Kd = 31 and 76 μM, respectively (Table 1). In particular, the affinity of m1p (L161R) increased considerably relative to wild-type m1p (Kd = 31 vs. 174 μM), confirming the feasibility and practicality of rational redesign of SH3-binding peptides.

Conclusions PTTG plays an important role in the pathological process of human MDB. Understanding of the molecular mechanism and biological implication underlying the recognition and association of PTTG with its cognate partners is fundamentally important for design of new anti-MDB agents with high affinity and selectivity to target the PTTG-mediated cancer signaling network. Here, we have systematically investigated the intermolecular interaction profile of two PTTG PxxP-containing peptides m1p and m2p against 69 structurally diverse, non-redundant SH3 domains that are highly expressed in MDB. It is found that the binding selectivity of m1p and m2p is optimized across the MDB expression spectrum by competing for domain candidates. In order to improve the binding potency of peptide ligands

Cell Biochem Biophys

to SH3 domains we also redesigned the m1p peptide using a structure-based rational strategy, resulting in two peptide mutants with considerably increased affinity. Acknowledgements This study was supported by the Health and Family Planning Commission of Jiangsu Province Youth Research Subject (No. Q201606), the Six Talent Peaks Project in Jiangsu Province (No. 2014-wsw-021), and the Suzhou Applied Basic Research (No. Sys201535). Compliance with Ethical Standards Conflict of Interest ing interests.

The authors declare that they have no compet-

References 1. Xiao, J. Q., Liu, X. H., Hou, B., Yao, Y., Deng, K., Feng, M., Xing, B., Lian, W., Wang, R. Z., & Feng, F. (2014). Correlations of pituitary tumor transforming gene expression with human pituitary adenomas: A meta-analysis. PLoS One, 9(3), e90396. 2. Pei, L., & Melmed, S. (1997). Isolation and characterization of a pituitary tumor-transforming gene (PTTG). Molecular Endocrinology, 11(4), 433–441. 3. Zhang, X., Horwitz, G. A., Heaney, A. P., Nakashima, M., Prezant, T. R., Bronstein, M. D., & Melmed, S. (1999). Pituitary tumor transforming gene (PTTG) expression in pituitary adenomas. Journal of Clinical Endocrinology and Metabolism, 84(2), 761–767. 4. Salehi, F., Kovacs, K., Scheithauer, B. W., Lloyd, R. V., & Cusimano, M. (2008). Pituitary tumor-transforming gene in endocrine and other neoplasms: A review and update. Endocrine Related Cancer, 15(3), 721–743. 5. Liu, J., Wang, Y., He, H., Jin, W., & Zheng, R. (2015). Overexpression of the pituitary tumor transforming gene upregulates metastasis in malignant neoplasms of the human salivary glands. Experimental and Therapeutic Medicine, 10(2), 763–768. 6. Cui, L., Xu, S., Song, Z., Zhao, G., Liu, X., & Song, Y. (2015). Pituitary tumor transforming gene: A novel therapeutic target for glioma treatment. Acta Biochimica et Biophysica Sinica, 47(6), 414–421. 7. Yu, R., Heaney, A. P., Lu, W., Chen, J., & Melmed, S. (2000). Pituitary tumor transforming gene causes aneuploidy and p53dependent and p53-independent apoptosis. Journal of Biological Chemistry, 275(47), 36502–36505. 8. McCabe, C. J., Khaira, J. S., Boelaert, K., Heaney, A. P., Tannahill, L. A., Hussain, S., Mitchell, R., Olliff, J., Sheppard, M. C., Franklyn, J. A., & Gittoes, N. J. (2003). Expression of pituitary tumour transforming gene (PTTG) and fibroblast growth factor-2 (FGF-2) in human pituitary adenomas: Relationships to clinical tumour behaviour. Clinical Endocrinology, 58(2), 141–150. 9. Boelaert, K., Yu, R., Tannahill, L. A., Stratford, A. L., Khanim, F. L., Eggo, M. C., Moore, J. S., Young, L. S., Gittoes, N. J., Franklyn, J. A., Melmed, S., & McCabe, C. J. (2004). PTTG’s Cterminal PXXP motifs modulate critical cellular processes in vitro. Journal of Molecular Endocrinology, 33(3), 663–677. 10. Domínguez, A., Ramos-Morales, F., Romero, F., Rios, R. M., Dreyfus, F., Tortolero, M., & Pintor-Toro, J. A. (1998). hpttg, a human homologue of rat pttg, is overexpressed in hematopoietic neoplasms. Evidence for a transcriptional activation function of hPTTG. Oncogene, 17(17), 2187–2193.

11. Zhang., X., Horwitz, G. A., Prezant, T. R., Valentini, A., Nakashima, M., Bronstein, M. D., & Melmed, S. (1999). Structure, expression, and function of human pituitary tumor-transforming gene (PTTG). Molecular Endocrinology, 13(1), 156–166. 12. Pei, L. (2000). Activation of mitogen-activated protein kinase cascade regulates pituitary tumor-transforming gene transactivation function. Journal of Biological Chemistry, 275(40), 31191–31198. 13. Vlotides, G., Cruz-Soto, M., Rubinek, T., Eigler, T., Auernhammer, C. J., & Melmed, S. (2006). Mechanisms for growth factorinduced pituitary tumor transforming gene-1 expression in pituitary folliculostellate TtT/GF cells. Molecular Endocrinology, 20 (12), 3321–3335. 14. Boon, K., Edwards, J. B., Siu, I. M., Olschner, D., Eberhart, C. G., Marra, M. A., Strausberg, R. L., & Riggins, G. J. (2003). Comparison of medulloblastoma and normal neural transcriptomes identifies a restricted set of activated genes. Oncogene, 22(48), 7687–7694. 15. Kärkkäinen, S., Hiipakka, M., Wang, J. H., Kleino, I., VähäJaakkola, M., Renkema, G. H., Liss, M., Wagner, R., & Saksela, K. (2006). Identification of preferred protein interactions by phage-display of the human Src homology-3 proteome. EMBO Reports, 7(2), 186–191. 16. Larkin, M. A., Blackshields, G., Brown, N. P., Chenna, R., McGettigan, P. A., McWilliam, H., Valentin, F., Wallace, I. M., Wilm, A., Lopez, R., Thompson, J. D., Gibson, T. J., & Higgins, D. G. (2007). ClustalW and ClustalX version 2. Bioinformatics, 23(21), 2947–2948. 17. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N., & Bourne, P. E. (2000). The protein data bank. Nucleic Acids Research, 28(1), 235–242. 18. Canutescu, A. A., & Dunbrack, Jr., R. L. (2005). MollDE: A homology modeling framework you can click with. Bioinformatics, 21(12), 2914–2916. 19. Bordoli, L., Kiefer, F., Arnold, K., Benkert, P., Battey, J., & Schwede, T. (2009). Protein structure homology modeling using SWISS-MODEL workspace. Nature Protocols, 4(1), 1–13. 20. Canutescu, A. A., Shelenkov, A. A., & Dunbrack, Jr., R. L. (2003). A graph-theory algorithm for rapid protein side-chain prediction. Protein Science, 12(9), 2001–2014. 21. Krivov, G. G., Shapovalov, M. V., & Dunbrack, Jr., R. L. (2009). Improved prediction of protein side-chain conformations with SCWRL4. Proteins, 77(4), 778–795. 22. Krieger, E., Koraimann, G., & Vriend, G. (2002). Increasing the precision of comparative models with YASARA NOVA — A self-parameterizing force field. Proteins, 47(3), 393–402. 23. Li, Y., Li, M., Min, W., Han, G., Wang, L., Chen, C., Li, Z., Zhang, Y., Li, J., & Yue, Z. (2017). Characterization, identification and disruption of the intermolecular interaction between protein kinases and human pituitary tumor-transforming gene 1. General Physiology and Biophysics, 36(1), 23–30. 24. Donsky, E., & Wolfson, H. J. (2011). PepCrawler: A fast RRTbased algorithm for high-resolution refinement and binding affinity estimation of peptide inhibitors. Bioinformatics, 27(20), 2836–2842. 25. London, N., Raveh, B., Cohen, E., Fathi, G., & Schueler-Furman, O. (2011). Rosetta FlexPepDock web server--high resolution modeling of peptide-protein interactions. Nucleic Acids Research, 39(Web Server Issue), W249–W253. 26. Duan, Y., Wu, C., Chowdhury, S., Lee, M. C., Xiong, G., Zhang, W., Yang, R., Cieplak, P., Luo, R., Lee, T., Caldwell, J., Wang, J., & Kollman, P. (2003). A point-charge force field for molecular mechanics simulations of proteins based on condensed-phase quantum mechanical calculations. Journal of Computational Chemistry, 24(16), 1999–2012.

Cell Biochem Biophys 27. Case, D. A., Cheatham, T. E., Darden, T., Gohlke, H., Luo, R., Merz, K. M., Onufriev, A., Simmerling, C., Wang, B., & Woods, R. J. (2005). The Amber biomolecular simulation programs. Journal of Computational Chemistry, 26(16), 1668–1688. 28. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., & Klein, M. L. (1983). Comparison of simple potential functions for simulating liquid water. Journal of Chemical Physics, 79(2), 926–935. 29. Ryckaert, J. P., Ciccotti, G., & Berendsen, H. J. C. (1977). Numerical integration of the Cartesian equations of motion of a system with constraints: Molecular dynamics of n-alkanes. Journal of Chemical Physics, 23(3), 327–341. 30. Darden, T., York, D., & Pedersen, L. (1993). Particle mesh Ewald: An N.log(N) method for Ewald sums in large systems. Chemical Physics, 98(12), 10089–10092. 31. Gangwal, R. P., Dhoke, G. V., Damre, M. V., Khandelwal, K., & Sangamwar, A. T. (2013). Structure-based virtual screening and molecular dynamic simulation studies to identify novel cytochrome bc1 inhibitors as antimalarial agents. Journal of Computational Medicine, 2013(1), 637901. 32. Homeyer, N., & Gohlke, H. (2012). Free energy calculations by the molecular mechanics Poisson-Boltzmann surface area method. Molecular Informatics, 31(2), 114–122. 33. Yang, C., Zhang, S., He, P., Wang, C., Huang, J., & Zhou, P. (2015). Self-binding peptides: Folding or binding? Journal of Chemical Information and Modeling, 55(2), 329–342. 34. Pisabarro, M. T., & Serrano, L. (1996). Rational design of specific high-affinity peptide ligands for the Abl-SH3 domain. Biochemistry, 35(33), 10634–10640.

35. He, P., Tan, D. L., Liu, H. X., Lv, F. L., & Wu, W. (2015). The auto-inhibitory state of Rho guanine nucleotide exchange factor ARHGEF5/TIM can be relieved by targeting its SH3 domain with rationally designed peptide aptamers. Biochimie, 111 (4), 10–18. 36. Schweimer, K., Hoffmann, S., Bauer, F., Friedrich, U., Kardinal, C., Feller, S. M., Biesinger, B., & Sticht, H. (2002). Structural investigation of the binding of a herpesviral protein to the SH3 domain of tyrosine kinase Lck. Biochemistry, 41(16), 5120–5130. 37. Han, K., Wu, G., & Lv, F. (2013). Development of QSARimproved statistical potential for the structure-based analysis of protein-peptide binding affinities. Molecular Informatics, 32(9), 783–792. 38. Tong, Y., & Eigler, T. (2009). Transcriptional targets for pituitary tumor-transforming gene-1. Journal of Molecular Endocrinology, 43(5), 179–185. 39. London, N., Movshovitz-Attias, D., & Schueler-Furman, O. (2010). The structural basis of peptide-protein binding strategies. Structure, 18(2), 188–199. 40. Adzhubei, A. A., Sternberg, M. J., & Makarov, A. (2013). A. Polyproline-II helix in proteins: Structure and function. Journal of Molecular Biology, 425(12), 2100–2132. 41. Liu, P., Cheng, H., Roberts, T. M., & Zhao, J. J. (2009). Targeting the phosphoinositide 3-kinase pathway in cancer. Nature Reviews Drug Discovery, 8(8), 627–644.