Structure determination of midkine a

STRUCTURE DETERMINATION OF MIDKINE-A MENG DAN NATIONAL UNIVERSITY OF SINGAPORE 2008 STRUCTURE DETERMINATION OF MIDKINE-A MENG DAN A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF BIOLOGICAL SCIENCES NATIONAL UNIVERSITY OF SINGAPORE 2008 Acknowledgement I would like to express my sincere appreciation to my supervisor, Associate Professor Yang Daiwen, for his kind guidance and trust during the two years in the course of my research project. Also thank Dr Mok Yu-Keung for his advice and critical suggestions. I would like to give special thanks to Dr Lin Zhi for his helpful advices on my project, and thank Lim Jack Wee and Dr Fang Jingsong for their help in the experiments. I also want to thank A/P Christoph Winkler and his student Yao Sheng for the functional study of Mdka. I had a pleasant learning experience here thanks to the friendship of my fellow graduates, lab mates and friends at NUS. Particularly, I am thankful to Professor Wong Sek Man, Dr Zhang Jingfeng, Yang Shuai, BC Karthik, Long Dong, Yong Yee Heng, Zheng Yu, Qin Haina, Zhang Xin, and Zhang Xiaolu. I am deeply grateful to my parents in China and my boyfriend in USA for their spiritual support and love to overcome the difficult times. Finally, I am thankful to NUS for offering me the research scholarship and the valuable opportunity here for postgraduate study. I Table of Contents Table of Contents ........................................................................................................ II Abbreviations .............................................................................................................. V List of Figures........................................................................................................... VII List of Tables............................................................................................................ VIII Summary.....................................................................................................................IX Chapter 1 Introduction................................................................................................1 1.1 Studies on Mdka and Mdkb of zebrafish: heparin binding growth factors .................................................................................................................................1 1.1.1 Biological background of zebrafish embryonic and neural development....................................................................................................1 1.1.2 Characterization of Mdka and Mdkb of zebrafish............................3 1.1.3 Biological activities of Mdka and Mdkb .............................................5 1.1.4 Zebrafish PTN: another member of Mikine family in zebrafish....10 1.2 Structural and functional studies on Midkine and PTN of human ..........10 1.2.1 Introduction.........................................................................................10 1.2.2 Protein structures................................................................................ 11 1.2.3 Receptors of MK and PTN.................................................................13 1.2.4 Biological activities..............................................................................14 1.2.5 Medical significance............................................................................15 1.3 Protein structure determination by NMR ..................................................16 1.3.1 Introduction to NMR spectroscopy ...................................................16 1.3.1.1 Development of NMR ...............................................................16 1.3.1.2 Basic theories of NMR ..............................................................17 1.3.1.2.1 Chemical shift .................................................................................17 1.3.1.2.2 J coupling ........................................................................................17 1.3.1.2.3 NOE .................................................................................................18 1.3.2 General strategies of protein structure determination by NMR ....18 1.3.2.1 Sample Preparation ..................................................................18 1.3.2.2 Spectrum collection and Resonance Assignments..................20 1.3.2.3 Collection of NMR Restraints..................................................20 1.3.2.4 Structure Calculations and Refinement..................................20 1.3.2.5 Evaluation of protein NMR structure .....................................21 1.3.3 Advantages of structure study by NMR............................................22 1.4 Objectives of this project..............................................................................22 II Chapter 2 Materials and Methods............................................................................24 2.1 NMR sample preparation.............................................................................24 2.1.1 Media....................................................................................................24 2.1.2 Preparation of DNA plasmid..............................................................24 2.1.3 Transformation of E .coli competent cells ........................................25 2.1.4 Protein expression and purification ..................................................25 2.1.4.1 Expression of unlabeled Mdka.................................................25 2.1.4.2 Expression of labeled Mdka.....................................................25 2.1.4.3 Purification of Mdka.................................................................26 2.1.5 General protein assays........................................................................26 2.1.5.1 Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) ..........................................................................................26 2.1.5.2 Protein quantitative assay ........................................................27 2.1.5.3 Protein circular dichroism (CD) ..............................................27 2.1.5.4 Protein dynamic light scattering (DLS) ..................................28 2.2 NMR experiments .........................................................................................28 2.2.1 1D NMR experiment...........................................................................28 1 15 1 13 2.2.2 2D H- N HSQC experiment.............................................................29 2.2.3 2D H- C HSQC experiment.............................................................29 2.2.4 H/D exchange experiment ..................................................................29 2.2.5 3D HNCA and 3D HNCOCA experiments .......................................29 2.2.6 3D MQ-CCH-TOCSY experiment ....................................................30 2.2.7 4D time shared NOESY experiment..................................................30 2.3 Data processing .............................................................................................30 2.4 Resonance assignment ..................................................................................31 2.4.1 Backbone assignment..........................................................................31 2.4.2 Sidechain assignment..........................................................................32 2.4.3 NOE assignment, structure calculation and refinement .................32 Chapter 3 Results and Discussion.............................................................................34 3.1 Mdka construct .............................................................................................34 3.2 Expression and purification of Mdka..........................................................34 3.2.1 Expression of GST-Mdka in E. coli ...................................................34 3.2.2 Purification of Mdka...........................................................................34 3.3 General properties of Mdka.........................................................................37 3.4 Mdka in vivo assay: zebrafish embryonic development activity ..............37 3.5 NMR assignment of Mdka ...........................................................................43 3.5.1 Backbone assignment of Mdka ..........................................................43 3.5.2 Sidechain assignment of Mdka ..........................................................43 3.5.3 Secondary structure characterization from backbone assignment 47 3.5.4 NOE assignment..................................................................................47 3.5.5 Structure calculation and refinement................................................48 3.5.6 Disulfide restraints determination.....................................................52 3.6 NMR structure of Mdka...............................................................................55 3.7 Structural comparison to other Midkine family members........................60 3.7.1 Structure comparison of Mdka and human MK .............................60 3.7.2 Structure comparison of Mdka with human PTN ...........................64 III Chapter 4 Conclusion and Future Work .................................................................69 Chapter 5 References.................................................................................................73 IV Abbreviations 1D One-dimensional 2D Two-dimensional 3D Three-dimensional 4D Four-dimensional a.a. Amino acid E. coli Escherichia coli EDTA Ethylenediamine tetraacetic acid HSQC Heteronuclear single quantum correlation spectrum IPTG Isopropyl β-D-thiogalactoside MQ Multiple-quantum NMR Nuclear magnetic resonance NOE Nuclear Overhauser effect NOESY Nuclear Overhauser enhanced spectroscopy Ppm Parts per million RMSD/rmsd Root mean square deviation SDS Sodium dodecyl sulphate Tris 2-amino-2-(hydroxymethyl-1,3-propanediol TOCSY Total Correlation Spectroscopy Ala, A Alanine V Arg, R Arginine Asn, N Asparagine Asp, D Aspartic acid Cys, C Cysteine Gly, G Glycine Glu, Q Glutamic acid Gln, E Glutamine His, H Histidine Ile, I Isoleucine Leu, L Leucine Lys, K Lysine Met, M Methionine Phe, F Phenylalanine Pro, P Proline Ser, S Serine Thr, T Threonine Trp, W Tryptophan Tyr, Y Tyrosine Val, V Valine VI List of Figures Fig1.1 Process of neurulation .................................................................................2 Fig1.2 Organization of the neural tube ...................................................................4 Fig1.3 Sequence of Mdka. ......................................................................................6 Fig1.4 Sequence of Mdkb. ......................................................................................7 Fig1.5 Sequence alignment of MK family..............................................................8 Fig1.6 Human Midkine family sequence alignment and structure. ......................12 Fig1.7 General strategy for structure determination by NMR..............................19 Fig3.1 Final construct of Mdka.............................................................................35 Fig3.2 Mdka expression and purification by GST beads......................................36 Fig3.3 Heparin affinity column purification of Mdka. .........................................38 Fig3.4 Q-TOF mass spectrum of Mdka. ...............................................................39 Fig3.5 CD spectrum of Mdka ...............................................................................40 Fig3.6 Apparent hydrodynamic MW of Mdka from DLS measurement..............41 Fig3.7 1D and 2D NMR experiments of Mdka. ...................................................42 Fig3.8 Backbone connectivity of Mdka................................................................44 Fig3.9 Backbone assignment of Mdka..................................................................45 Fig3.10 Assigned 1H-15N HSQC spectrum of Mdka backbone peaks. .................46 Fig3.11 Chemical shift index of Mdka. ................................................................49 Fig3.12 NOE assignment. .....................................................................................50 Fig3.13 Sequential- and medium-range NOE pattern of Mdka............................51 Fig3.14 Comparison of Mdka domain structures calculated with and without disulfide constraints. .....................................................................................53 Fig3.15 Solution structure of Mdka. .....................................................................57 Fig3.16 Ramachandran plot of Mdka. ..................................................................58 Fig3.17 Configuration of proline residues. ...........................................................59 Fig3.18 Comparison of Mdka with human MK....................................................61 Fig3.19 Sequence comparison of Mdka and human MK. ....................................62 Fig3.20 Sequence alignment of Mdka and PTN. ..................................................65 Fig3.21 Secondary structure of PTN. ...................................................................66 Fig3.22 Sequence comparison of Mdka and Mdkb. .............................................68 VII List of Tables Table 1 Distances between the Cβ atoms of nearby Cys residues in the 10 conformers. ..................................................................................................54 Table 2 Structural statistics of Mdkaa ...............................................................56 VIII Summary Midkine-a (Mdka) is a heparin-binding growth factor from zebrafish, which regulates the fish brain formation during embryogenesis. Its homologs in zebrafish, Midkine-b, and in human, named Midkine, have 58% and 68% amino acid identity with Mdka respectively, but show distinctively different expression and functional patterns from Mdka. In order to understand more about how this growth factor functions, and why similar molecules can function differently, we sought to investigate their structural and mechanistic characteristics. In this study, the solution structure of Mdka is solved based on the multi-dimensional NMR spectroscopy using a novel assignment method developed by our group, which shows that this method works as well for small proteins as larger ones. Mdka contains two domains, and each of them consists of three anti-parallel β-sheets, resembling that of human Midkine. The N domain backbone architectures of Mdka and human Midkine are essentially the same, but C domains have major differences, which could be the reason for the differential functions of the two proteins. Our results provide new insights into the functional mechanism of the Midkine family; and directions for future structural study of this protein include: 1) comparing the structures of Mdka and Mdkb to understand why they have different functions; and 2) mapping the heparin binding site of Mdka to identify the specific residues important for the function of the protein. IX Chapter 1 Introduction 1.1 Studies on Mdka and Mdkb of zebrafish: heparin binding growth factors 1.1.1 Biological background of zebrafish embryonic and neural development Zebrafish Danio rerio is a common and useful model organism for studies of vertebrate development and gene function (Mayden et al., 2007), since zebrafish embryos are large, robust, and transparent, and have the properties of developing rapidly and externally to the mother, characteristics which all facilitate experimental manipulation and observation (Dahm, 2006). Like other vertebrates’ embryo development, zebrafish development consists of stages of fertilization, cleavage, gastrulation, organogenesis and hatching. During gastrulation, three germ layers are formed in the early embryo: endoderm, mesoderm and ectoderm, which develop into different groups of the tissues. Among them, the ectoderm layer is supposed to develop into future neural systems by neurulation. Neurulation begins with the signals emitted from one discrete stripe of mesoderm to the ectoderm instructing the cells of the ectodermal tissue to change shape (Fig1.1). This process is coordinated as a whole, where the cells migrate to form a neural tube. The neural tube is the embryo's precursor to the central nervous system, which comprises the brain and spinal cord. During its formation, cells at the edges of the dorsal midline will sort themselves out by sticking to the non-neural ectoderm or sticking with the neural ectoderm. The former would develop into the epidermal epithelium while the latter would become part of the neural tube epithelium. 1 A Fig1.1 Process of neurulation. (Kristjan R. Jessen & Rhona Mirsky, 2005) 2 However, of unusual significance are some in-between cells that form a loose aggregate in the space when the two epitheliums separate. These are neural crest cells which quickly migrate during or shortly after neurulation. Following their migration the neural crest cells differentiate into a wide variety of cell types throughout the body (Bronner-Fraser, 1995). The junction between neural and non-neural ectoderm, also called neural plate border, also gives rise to another kind of cells called Rohon-Beard (RB) sensory neurons. Different from neural crest cells, RB sensory neurons do not migrate away from the neural tube upon its closure, but remain within the central nervous system instead, and eventually comprise bilateral stripes in the developing dorsal spinal cord (Roberts, 2000). In the neural tube, a signaling center called the roof plate is formed when the Bone morphogenetic protein (BMP) from the epidermis permeated in. Meanwhile, in the ventral floor of the neural tube, a floor plate is formed due to Sonic hedgehog (Shh) protein from the notochord. These two molecules form a double gradient by spreading through the neural tube, and the identity of neurons in the tube can be determined by the relative concentrations of the two factors (Fig1.2). Therefore, the two signaling centers, the floor plate and the roof plate, affect the cell differentiation according to the position in the neural tube, thus determine the pattern formation in the embryonic spinal cord. At the dorsal mesoderm, cells that adjacent to the notochord during vertebrate organogenesis form transient structures called somite, which define the segmental pattern of the embryo, and subsequently give rise to vertebrae and ribs, dermis of the back, and skeletal muscles of the back, body wall and limbs (Wikipedia). 1.1.2 Characterization of Mdka and Mdkb of zebrafish 3 Fig1.2 Organization of the neural tube. Mdka regulates floor plate formation. (Schäfer et al., 2005; Schäfer et al., 2007, Squire et al., 2003). 4 Mdka and Mdkb are the two heparin binding growth factors in zebrafish, while in mammals there is only one mdk gene encoding one secreted heparin binding growth factor. Mdkb was identified using an expression-cloning screen for neural inducing factors (Winkler and Moon, 2001). Using Mdkb as a query, Mdka was identified from searching the EST database, showing only ~70% nucleotide identity with Mdkb, 68% amino acid identity, and 72% similarity with Mdkb (Winkler et al., 2003). Both Mdka and Mdkb contain more basic residues than acid ones, being positive charged proteins with theoretic isoelectric point (PI) of 9.59 and 9.53 respectively. Both proteins are soluble proteins according to the hydropathy profiles (Fig1.3 & Fig1.4). The N terminals of both proteins are highly hydrophobic and the first 23 amino acids of each protein are supposed to form a signal peptide. The rest of the amino acids are mainly hydrophilic although there are two short hydrophobic segments at the amino acid positions of 43-47 and 110-116 in Mdka (Fig1.3). Mdka shows higher sequence identity (58%) with human Midkine compared to Mdkb (51%). Sequence comparison between the zebrafish Midkines and homologs from other species revealed the presence of 10 conserved cysteine residues, two clusters of basic residues (cluster I and cluster II) that are important for heparin binding, and a highly conserved hinge region that separates the amino-and carboxyterminal domains (Iwasaki et al., 1997; Winkler, 2003; Fig1.5). In all fish Midkines, a highly conserved Arg residue that is essential for binding of Midkine to its receptor PTPζ is also present (Maeda et al., 1999). Two glutamine residues that are important for the formation of covalently linked dimers by tissue-type 2 transglutaminases are conserved in zebrafish (Kojima et al., 1997). 1.1.3 Biological activities of Mdka and Mdkb 5 A B Fig1.3 Sequence of Mdka. (A) The nucleotide sequence of the Mdka cDNA and its deduced amino acid sequences. (B) Hydropathy profile of Mdka (Software from ExPSAy, ProtScale). 6 A B Fig1.4 Sequence of Mdkb. (A) The nucleotide sequence of the Mdkb cDNA and its deduced amino acid sequences. (B) Hydropathy profile of Mdkb (Software from ExPSAy, ProtScale). 7 Fig1.5 Sequence alignment of MK family. (Chang et al., 2004) heparin binding; cluster I a.a. for cluster II a.a. for heparin binding. 8 Winkler (2003) first reported the expression patterns and biological activities of Mdka and Mdkb, revealing their important roles in the early neural development of zebrafish. Both Mdka and Mdkb are expressed in very dynamic and regionally restricted patterns during embryogenesis. Spatially, Mdkb expression is restricted to the dorsal regions of the developing nervous system, but Mdka is strongly expressed in a broad area in the central neural tube. At the beginning, Mdka expression is found in the paraxial mesoderm, which indicates the role of Mdka in somite formation. At 16-h post fertilization (hpf), Mdka expression is found in epithelialized somites and also in a very dynamic fashion in the head region and in the spinal cord. Its expression is excluded from dorsal neural tube and the floor plate, which is the ventral most structure in the neural tube. In addition to spatial differences, the onset of transcription also differs for both genes. Mdkb expression starts during early gastrulation, about 4 h prior to Mdka that is first detectable at 10 hpf. Onset of Mdka expression is at tailbud stage in two lateral clusters of cells of the paraxial mesoderm (Fig 1.2B). Since Mdka and Mdkb show an overall 68% amino acid identity and the heparin binding as well as the hinge region are highly conserved, it opens to the question whether these two proteins share similar activities. To answer this question, an over-expression approach was used, which was to ectopically express mdka and mdkb genes after microinjection of in vitro-synthesized RNA into early zebrafish embryos. The resulting embryonic development was significantly different. Over-expression of Mdka affected the formation of somites and neural tube (Winkler et al., 2003) while injection of mdkb RNA resulted in generally posteriorized embryos lacking head structures rostral to the mid-hindbrain boundary (Winkler and Moon, 2001). Therefore, both secreted growth factors, although structurally related, show different functions, and it would be interesting to investigate the functions of each 9 protein and why such structurally related proteins would have functional divergence. Although the exact functional mechanisms are not known, Mdka is reported to be required for the formation of the medial floor plate (Schäfer et al., 2005). On the contrary, Mdkb is the downstream signaling factors of several pathways, most notably retinoic acid, although the processes are still unclear. Other possible pathways include FGF signaling pathway, Wnt signaling pathway, and BMP signaling pathway. In addition, Mdkb controls the cell determination at the neural plate border and affects the formation of neural crest cells and Rohon-Beard sensory neurons (Liedtke et al., 2008). As secreted growth factors, both proteins might bind to the transmembrane heparan sulfate proteoglycans. These receptors have the heparin-like carbohydrates; thus the growth factors for these kinds of receptors are classified as heparin-binding growth factors. 1.1.4 Zebrafish PTN: another member of Mikine family in zebrafish Zebrafish heparin-binding neurotrophic factor (HBNF or PTN) is another secreted heparin-binding protein which is highly basic and contains ten cysteine residues (Chang et al., 2004). The amino acid sequence comparison shows that it displays 60% identity to human PTN, while only 39%-40% identity to human midkine, Mdka, and Mdkb (Fig1.5). During zebrafish embryogenesis, the expression of zebrafish PTN was observed upon fertilization and onwards. In adult fish, it is highly expressed in brain and intestine. The in vivo biological function of zebrafish PTN is found to promote neurite outgrowth. 1.2 Structural and functional studies on Midkine and PTN of human 1.2.1 Introduction 10 Midkine protein (MK) and Pleiotrophin (PTN; also called HB-GAM) are the two members of the MK family in human. Both proteins were discovered two decades ago (Kadomatsu et al., 1988; Rauvala, 1989), and have been studied extensively in vitro since then. As extracellular signaling molecules, MK and PTN play pivotal roles in neural development, pathogenesis of neurodegenerative diseases, and cancer development (Kadomatsu et al., 2004). 1.2.2 Protein structures MK precursor protein with an intact signal peptide (22 amino acids) has a molecular mass of 15.6 kDa whereas the matured form of MK is calculated to be 13.2 kDa. Mature PTN protein, which shares 45% amino acid sequence identity with MK (Fig1.6A), has a molecular weight of 19 kDa. The individual domain structures of MK have been solved by NMR (Iwasaki et al., 1997; Fig1.6B), and the NMR analysis has demonstrated that the three dimensional structures of MK and PTN are very similar (Kilpeläinen et al., 2000). Both proteins are essentially composed of two domains, the N-terminally located domain and the C-terminally located domain. These two domains are held by five disulfide bridges from ten conserved cysteine residues. Each domain contains three anti-parallel β-sheets. It is noteworthy that the heparin binding site of MK is mainly located in the C-domain which is responsible for most of the Midkine activities (Muramatsu et al., 1994 ), while that of PTN is equally distributed in both N- and C-domains (Kilpelainen et al., 2000). In the C domain of MK, two heparin binding clusters have been identified (Fig1.6B). Cluster I contains residues K79 (βC2), R81 (βC2), and K102 (βC3), all of which are located on the two β-sheets. All these three residues are conserved in the MK family from invertebrates to vertebrates. Residues, K86, K87, and R89, from the long loop between βC2 and 11 A B N domain C domain Fig1.6 Human Midkine family sequence alignment and structure. (A) Sequence alignment of MK and PTN (Kadomatsu and Muramatsu, 2004). (B) NMR solution structure of human MK domains (Muramatsu, 2002). 12 βC3 form cluster II (Iwasaki et al., 1997; Fig1.6B). Upon binding to heparin ligand, MK could form a dimer, and this dimerization is important for the activity of MK. Three glutamine residues in human MK have been identified to be responsible for the transglutaminase-mediated dimerization (Kojima et al., 1997). 1.2.3 Receptors of MK and PTN Heparin binding growth factors, like MK and PTN, could bind to a variety of transmembrane receptors on neurons and osteoblasts. Four kinds of receptors have been identified to have interactions with the MK family so far, including N-syndecan (Raulo et al., 1994; Mitsiadis et al., 1995; Kojima et al., 1996; Nakanishi et al., 1997), receptor-type protein tyrosine phophafase ζ (PTPζ) (Maeda et al., 1996; Maeda et al., 1998; Maeda et al., 1999), anaplastic leukemia kinase (ALK) (Stoica et al., 2001; Stoica et al., 2002), and low density lipoprotein receptor-related protein (LRP) (Herz et al., 2000). N-syndecan belongs to the syndecan family that comprises four transmembrane heparan sulfate proteoglycans. Binding of MK and PTN to syndecans is mediated by the heparin sulfate chain, which is the heparin-like domain. This domain consists of a variably sulfated repeating disaccharide unit. PTPζ is transmembrane protein with chondroitin sulfate chains, where MK and PTN bind, and an intracellular tyrosine phosphatase domain. LRP and ALK can bind MK and PTN via extracellular potion of the receptor, but how exactly they interact is not clear. These receptors might be differentially utilized for specific biological activities such as neurite outgrowth (Li et al., 1990; Rauvala et al., 1994; Kaneda et al., 1996a), nerve cell migration (Maeda et al., 1998; Maeda et al., 1999), neurodegenerative diseases (Wisniewski et al., 1996; Yasuhara et al., 1993), and cancer (Kadomatsu et al., 2004). Many of the down-regulation pathways are still unclear, but it is pointed 13 out that the intracellular signaling of N-syndecan family receptors involves cortactin-src pathway, PTPζ and ALK receptors’ signaling pathways involve PI3-kinase and Erk, and the LRP receptor is responsible for the anti-apoptotic activity of MK. 1.2.4 Biological activities MK and PTN are strongly expressed during embryogenesis, but their expression in adults are only restricted to kidney and tumors, and very low levels in brain (Kato et al., 2000; Mishima et al., 1997). Both MK and PTN have diverse biological functions, among which their roles in neural development and pathological aspects are of significant importance (Kadomatsu and Muramatsu, 2004). Both MK and PTN have been reported to be involved in the neurite outgrowth and nerve cell migration. MK also shows neuroprotective activity by preventing cell degeneration (Unoki et al., 1994). As to neurodegenerative diseases, both MK and PTN are reported to deposit at the senile plaques in Alzheimer’s patients. MK is also found at the neurofibrillary tangles in Alzheimer’s patients. Since MK and PTN can bind to LRP, and LRP has featured roles in Alzheimer’s disease, thus MK and PTN might be related to the pathogenesis of Alzheimer’s disease. Besides, MK knockout mice show that MK is involved in the pathogenesis of interstitial nephritis and vascular restenosis. The pathogenesis of both diseases might result from the recruitment of inflammatory cells by MK. But no receptors and signaling pathways have been discussed on this subject. Moreover, both MK and PTN are reported to be involved in cancer-related activities. For normal tissues of human adults, MK shows restricted expression, but for carcinoma specimens, MK expresses at a high level in a tissue type-independent manner, and its expression is more intensely and in a wider range of human 14 carcinomas than PTN. It is also reported that MK can promote cell growth, cell survival, and cell migration. And MK is shown to be highly expressed in all stages of neuroblastomas (malignant tumor that usually metastasizes quickly), and weakly in ganglioneuromas (benign tumors), while PTN expression can be high in both. Hitherto the exact signaling pathways involved in all these activities are still unknown, but it is possible that MK and PTN have different receptors or signal transduction pathways. 1.2.5 Medical significance Since MK and PTN are closely involved in several diseases, especially cancer, their medical significances were proposed by Kadomatsu and Muramatsu (2004). One possible application is that they could be used as tumor markers, since there are elevated blood levels of both proteins in cancer patients, and the protein level would be decreased when the tumors are removed. Besides, the blood level of MK is significantly correlated with prognostic factors of neuroblastomas, while PTN level is reported to be correlated with the prognosis in pancreatic carcinoma patients. Another possible application is that they might be used as gene therapy targets since a suicide gene under the control of the mdk promoter is identified as a highly potential strategy for curing carcinomas. The third application is that they could be the molecular targets of therapy for carcinomas. Since both in vitro and in vivo assays using mice have demonstrated that antisense MK oligodeoxyribonucleotides suppressed cell growth, anchorage-independent growth, tumor growth of mouse colorectal carcinoma cells, and pre-grown tumors in nude mice via atelocollagen-mediated gene transfer; disruption of MK or its signaling pathway could be a strong means of curing human carcinomas. Similar situations applies to PTN, except that it is PTN-targeted 15 ribozymes (enzymatically active RNAs, i.e. ribozymes, that were designed to specifically cleave the PTN mRNA thus leading to significantly reduced PTN levels) that can suppress the growth of tumors, and a part of the N-terminal of PTN shows a dominant-negative form in the growth of human breast cancer cells. Therefore, PTN and its signaling molecules could also be a molecular targeting candidate for cancer therapy. 1.3 Protein structure determination by NMR 1.3.1 Introduction to NMR spectroscopy 1.3.1.1 Development of NMR NMR development began in the area of physics in 1938 when Isidor Rabi (Nobel Prize 1944) firstly described and measured the resonance in molecular beams. Eight years later, Purcell et al., and Bloch independently refined the technique and reported the phenomenon of NMR, for which they shared the Noble Prize in Physics in 1952. In the 1940s and 1950s, many basic parameters and theories of NMR were discovered such as chemical shift, spin-spin coupling, spin relaxation, and Overhauser effect, which laid the foundation of NMR method development. In 1966, Fourier Transform (FT) techniques were introduced by Ernst, and later he developed the multi-dimensional NMR, thus began the era of FT NMR (Nobel Prize 1991). Its application in biological sciences was marked by Kurt Wüthrich when he developed the NMR spectroscopy for determining 3D structures of biological macromolecules in solution (Nobel Prize 2002). Today, with the accessibility of improved NMR hardware and better developed NMR methodology, NMR has gained wide application in the areas of chemistry, biology and medicine. In biological sciences, with the development of molecular biology and 16 biochemical methods for preparing and isolating isotope labeled biomolecules, NMR has become a powerful tool for biomolecule structure characterization, protein dynamics, and identification of bioactive compounds. 1.3.1.2 Basic theories of NMR 1.3.1.2.1 Chemical shift Neutrons and protons that composing an atomic nucleus have an intrinsic quantum mechanical property called spin. Spins within different chemical environments would have different resonance frequencies. This is because of the electrons that circulating about the direction of the applied external magnetic field. The circulation would cause a small magnetic field at the nucleus; thus the actual magnetic field at the nucleus is different from the applied one. As in an NMR spectrum, each nucleus gives rise to a resonance which is characterized by chemical shift reflecting its unique chemical environment. Chemical shift is an important parameter for identifying individual nucleus and assigning the resonances in the spectrum to their corresponding atoms in a molecule. Moreover, characteristic chemical shifts in an amino acid residue can be helpful for identifying individual residues in a protein sequence and determining protein secondary structure (Wishart et al., 1991). Chemical shift is measured in parts per million (ppm) in order that the value is independent of the static magnetic field strength. 1.3.1.2.2 J coupling Scalar couplings or J-couplings are mediated through chemical bonds between two spins. The energy levels of each spin are slightly altered depending on the spin state of a scalar coupled spin (α or β). This gives rise to a splitting of the resonance 17 lines. Scalar couplings are used in multidimensional (2D, 3D, 4D) correlation experiments to transfer magnetization from one spin to another in order to identify spin systems, e.g. spins which are connected by not more than three chemical bonds (Sattler, 2004). 1.3.1.2.3 NOE The nuclear Overhauser effect (NOE) is a result of cross-relaxation between dipolar coupled spins as a result of spin/spin interactions through space. The NOE allows to transfer magnetization from one spin to another through space, and scales 6 with the distance r between the two spins (NOE ~ 1/r ), e.g. two protons in a protein. When interproton distances are larger than 5 Å, the NOE may be too small to be observed. Therefore NOEs are related to the three-dimensional structure of a molecule (Sattler, 2004). 1.3.2 General strategies of protein structure determination by NMR 1.3.2.1 Sample Preparation The protein sample for NMR experiments is often prepared by overexpressing proteins in a host, like E. coli or yeast, and the media used to incubate the host is enriched with 13 C and 15 N isotopes. About 0.5 ml protein solution with a concentration of ~1 mM protein is usually required in structure elucidation. Different buffer conditions (pH, salt concentrations, and types of buffer) need to be tested to optimize the quality of the NMR spectra and to avoid protein aggregation. Sometimes, the native protein needs to be modified, such as cutting the protein into separate domains, or making constructs with various sizes and domain boundaries, in order to obtain samples which give good NMR spectra. Before labeling the protein with 13 C 18 Sample preparation Cloning, expression, purification, isotope labeling Resonance assignment Backbone, sidechain, NOE Distance and torsion angle constraints NOE, dihedral angles, hydrogen bond, disulfide bond Structure calculation and refinement DYANA, CNS, structure energy minimization Evaluation of structure quality Fig1.7 General strategy for structure determination by NMR. 19 and 15 N isotopes, unlabeled sample can be used to determine if a protein sample 1 prepared under a certain condition is folded or not using 1D H NMR experiment. 1.3.2.2 Spectrum collection and Resonance Assignments A set of heteronuclear multidimensional NMR experiments and NOESY experiments (From 2D to 4D) are recorded for the assignment of the chemical shifts of all spins and NOE peaks. The total time for NMR data acquisition varies from ten days to four weeks depending on the experiments adopted. The chemical shift assignments of backbone and sidechain atoms are typically obtained from the triple-resonance experiments, which can be done manually or automatically. The chemical shifts alone would allow the determination of protein secondary structure. H-H NOE peaks observed in the NOESY spectrum can be used to assign the inter-atomic distances. 1.3.2.3 Collection of NMR Restraints Besides the distance constraints derived from NOE, other structural 3 information can be obtained for structure calculation. These include J-coupling constants that provide dihedral angles information, residual dipolar couplings that show information about bond projection angles, and solvent exchange that measures α β α hydrogen bonds. Torsion angles can also be predicted based on C , C , H chemical shifts with the program TALOS (Cornilescu et al., 1999). 1.3.2.4 Structure Calculations and Refinement A restrained molecular dynamics/simulated annealing (MD/SA) or torsion angel dynamics (TAD) is chosen for structure calculation, which would use all the 20 available structural parameters mentioned above. Structure calculation with NOE assignment is an iterative process. The result of an NMR structure calculation is an ensemble of structures, all of which are consistent with the experimental NMR data. The calculated structures should converge to the same fold. Poor convergence may indicate problems of the experimental constraints and/or the MD/SA/TAD protocol (Sattler, 2004). 1.3.2.5 Evaluation of protein NMR structure Factors for evaluation of protein NMR structures include the root-mean-square deviation (RMSD), Distance constraints violation, dihedral angle violation, electrostatic potential term, van der Waals force, and the Ramachandran plot. The RMSD indicates the coordinate precision of the structure ensembles calculated. The smaller the value of RMSD between the calculated conformers and the mean structures (usually less than 1 Å), the better the quality of the structures calculated. Distance constrains violation should be within 0.5 Å, while dihedral angle violation should be within 5°. As for the electrostatic potential term, negative results indicate good structure quality. Besides, the number of bad contacts, e.g. the number of restraints with too short atom-atom distances in a molecule should be small, otherwise the experimental constraints might be applied too strong and cause van der Waals clashes. Furthermore, the Ramachandran plot, which specifies the fraction of backbone φ,ψ angles which correspond to favored, additionally allowed, generally allowed and disallowed conformations based on statistical analysis of high-resolution crystal structures, is a good quality measure. Most φ,ψ angles should be within the allowed regions, while poor quality structures usually have a large number of φ,ψ angles in forbidden regions in the Ramachandran plot. 21 1.3.3 Advantages of structure study by NMR Nuclear Magnetic Resonance, X-ray Crystallography, and Cryo-electron Microscopy are the only three methods to solve protein structures so far. Only Nuclear Magnetic Resonance and X-ray Crystallography can be used to solve high resolution structures of proteins, nuclear acids and their complexes. The uniqueness of NMR methods is that the samples are in solution. The pH, salt concentration and temperature can all be adjusted so that the biomoleucles’ physiological condition can be mimicked. Besides, NMR can be used to study the dynamics of biomoleucles, to investigate structural, thermodynamic and kinetic aspects of interactions between proteins and other solution components. 1.4 Objectives of this project Mdka and Mdkb are new heparin binding growth factors found in zebrafish seven years ago, and many functions of the proteins are still unclear. These two proteins share high sequence identity, but perform different expression patterns and functions. On the one hand, Mdka, expressed in the paraxial mesoderm, is involved in the medial floor plate formation. On the other hand, the role of Mdkb is proposed to be specifying the formation of neural crest cells and Rohon-Beard sensory neuron by acting downstream of several signaling pathways at the neural plate border near the roof plate. Therefore it is interesting to understand why two growth factors with high sequence similarity would have different functions in the embryogenesis of zebrafish. In this study, the structure of Mdka is characterized. The major objective of this study is to elucidate the structural information of this protein, and compare the structure of Mdka with those of other homologs to understand the functional prosperities of Mdka. 22 The structural study of Mdka would lay foundation for the functional study of this protein family, such as the binding properties of the protein with heparin ligand and receptors. By comparing the structures of Mdka and Mdkb, we might gain new insights into the functions of the two proteins. 23 Chapter 2 Materials and Methods 2.1 NMR sample preparation 2.1.1 Media LB Broth: 1 L culture contains 10 g tryptone, 5 g of yeast extract, and 10 g of NaCl and 100 μg/ml ampicilin LB Agar: 1 L culture contains 10 g tryptone, 5 g yeast extract, 10 g NaCl, 15 g agar and 100 μg/ml ampicilin M9 minimal medium: 1 L culture contains 1 g NH4Cl, 2 g Glucose, 0.0147 g CaCl2(H2O), 0.493 g MgSO4(7H2O), 7.52 g Na2HPO4, 3 g KH2PO4, 0.5 g NaCl and 100 μg/ml ampicilin 2.1.2 Preparation of DNA plasmid The gene encoding Mdka protein was obtained from our collaborator, Prof Winkler (Dept. of Biological Sciences, NUS), in the template plasmid pCS2 Zmk-2. Oligonucleotide primers were used to amplify the mdka gene. The sense primer is 5’-CGCGGATCCAAAAACAAGAAAGAGAAGAA-3’ and antisense primer is 3’-GGGGATATCACTCAGCATAAT-5’. DNA gel electrophoresis (Agarose from Bio-Rad) was subsequently performed to separate the template plasmid from the desired PCR products. The recovered PCR products were digested with BamHI and XholI restriction enzymes (New England Biolabs). After purification using the QIAquick Gel Extraction Kit 250 (QIAGEN), the mdka DNA fractions were ligated to the pGEX-4T-1 vector, which has a GST tag. The obtained Plasmids were sequenced (reaction Kit from Perkin Elmer) using the automated DNA sequencer for verification. 24 2.1.3 Transformation of E .coli competent cells The mdka plasmids were transformed into E. coli BL21 (DE3) competent cells and were spread onto LB agar plates containing the ampicilin antibiotic selective for the plasmid. The plates were incubated for 12~16 hours at 37 °C to enlarge the bacterial colony. 2.1.4 Protein expression and purification 2.1.4.1 Expression of unlabeled Mdka The construct containing the mdka gene was expressed in the E. coli Strain BL21 (DE3) using LB medium. Procedure begins with inoculating one single colony containing mdka plasmid in 5 ml LB culture with 100 μg/ml ampicilin at 37 °C shaking overnight at 200 rmp. Then the overnight culture was transferred to 1 L fresh LB broth to inoculate at the same condition until the OD600 was around 0.4. To have an expression control, 1 ml of the culture was taken out, centrifuged, and the pellet was re-suspended with 50 μl SDS-PAGE sample buffer. The rest of the culture was induced with 0.05 mM IPTG at 16 °C for 12 hours. 1 ml of the induced culture was taken out, centrifuged, and the pellet was re-suspended with 50 μl SDS-PAGE sample buffer to test the protein expression level. The rest of the culture was centrifuged and pellets were collected for protein purification. 2.1.4.2 Expression of labeled Mdka The expression protocol was quite similar as that of unlabeled Mdka except 15 that the culture used was M9 medium. The culture used for N labeled Mdka was M9 15 15 13 medium with N labeled NH4Cl, and that for N and C labeled Mdka was medium with 15 N labeled NH4Cl and 13 C labeled glucose. After the cell density reached 25 OD600 around 0.6, 0.3 mM IPTG was added to induce protein expression at 20 °C for 20 hours. 2.1.4.3 Purification of Mdka The cell pellets collected by centrifuge were re-suspended with lysis buffer (20 mM Tris buffer, pH 8.0, 150 mM NaCl). After sonication thoroughly, the cell lysate was centrifuged to remove the unsonicated pellets. The supernatant was mixed with GST beads which have been wash with water and equilibrated with lysis buffer. The mixture was shaken at 4 °C for one hour followed by separating the resin from the flow-through. The resin was washed with lysis buffer four times to wash out the unspecific bindings. Then 10 U thrombin was added to the resin with 5 ml thrombin cleavage buffer (20 mM Tris buffer, pH 8.0, 150 mM NaCl, 2.5 mM CaCl2) to cleave GST tag from Mdka protein. Finally, the untagged protein was eluted with lysis buffer. After GST beads purification, there were still a few impurities. Therefore, heparin affinity column (HiTrap Heparin HP Column, GE Healthcare) was used with linear elution to further purify Mdka. The starting buffer was 20 mM Tris buffer, pH 8.0, 150 mM NaCl, 2 mM EDTA, while the elution buffer was 20 mM Tris buffer, pH 8.0, 2 M NaCl, 2 mM EDTA. The peak fractions collected from the FPLC machine were combined and concentrated for further protein assays. 2.1.5 General protein assays 2.1.5.1 Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) 12% and 15% SDS-PAGE were used to analyze the protein samples in 1 × sample buffer with electrophoresis apparatus (Bio-Rad). The 5 × sample buffer used 26 was 0.2 M Tris-HCl, pH 6.8, 25% (v/v) glycerol, 25% (v/v) SDS, 12.5% (v/v) 2-mercaptoethanol, 0.005% (w/v) bromophenol blue. The running buffer was 25 mM Tris, 192 mM glycine and 0.1% (v/v) SDS. After mixing the protein sample with the sample buffer, the mixture was heated for 5 min before loading into the well of the SDS-PAGE. When the gel finished running, coomassie-blue solution [0.05% (w/v) Coomassie brilliant blue R, 25% (v/v) isopropanol, 10% (v/v) acetic acid] was used to stain the gel for 30 min followed by distaining with 10% acetic acid. 2.1.5.2 Protein quantitative assay The protein concentration could be determined by UV absorbance at 280 nm, which was convenient and fast to use. The concentrated protein was diluted so that the UV absorbance was between 0.1 and 1.0 to ensure accuracy of the method. The concentration of the protein was calculated by the equation: A = εIC, where A stands -1 -1 for absorbance of UV, ε is the molar absorption coefficient (M cm ), I denotes the path length (cm), and C is the protein concentration (M). The value of ε can be estimated using the ProtParam in the ExPASy Proteomics Server. 2.1.5.3 Protein circular dichroism (CD) Far UV Circular Dichroism Spectroscopy was used to evaluate the secondary structure of Mdka. Experiments were carried out on a Jasco-810 spectropolarimeter (Jasco International Co. Ltd.) with a 0.1 nm quartz cuvette with a minimal sample volume of 250 μl. The protein sample was prepared by dialyzing against 10 mM phosphate buffer, and the protein concentration was within the range of 10~100 μM. The result was the average of 10 scans so that the background noise could be reduced. 27 2.1.5.4 Protein dynamic light scattering (DLS) To see if Mdka was still in the monomer form when the protein concentration was as high as 1 mM, Dynamic Light Scattering (Protein Solutions Inc) was used to evaluate the molecular weight of the protein particle. The protein sample was centrifuged at 15,000×g for 30 min at 4 ºC before adding the sample into a cuvette. The experiment was carried out at 30 ºC with 1 mM Mdka in the same buffer as the final NMR sample (20 mM phosphate buffer, pH 6.5, 100 mM NaCl, 2 mM EDTA). The maximal acquisition time was 10 sec and the data were analyzed using Dynamics 5.0 software (Moradian-Oldak et al., 1998), where the standard curve of globular protein was selected for MW model. 2.2 NMR experiments Bruker Avance 500 MHz and 800 MHz spectrometer equipped with pulse field gradient units and actively shielded cryoprobes were used to perform NMR experiments. Software used to process and analyze NMR data involved NMRPipe software (Delaglio et al., 1995) and NMRView (Johnson, 2004) on Linux workstations. A homemade NMRview plug-in was installed to assist the 4D NOE 1 15 1 13 assignment (Written by Lin Zhi). 2D H- N HSQC, 2D H- C HSQC, 3D HNCA, 3D HNCOCA, 3D CCH-TOCSY, and 4D time shared NOESY experiments were used to do backbone, side-chain, and NOE assignments (Xu et al., 2006). 2.2.1 1D NMR experiment 1D NMR experiments were used to test whether the protein was folded properly in in-vitro condition so that the following NMR experiments could be performed. Unlabeled Mdka sample was used to acquire data at 20 ºC in H2O 28 (10%D2O) eluted using 20 mM Tris buffer, 100 mM NaCl, 2 mM EDTA at pH 8.0, while experiment on labeled sample was recorded at 30 ºC. 1 15 2.2.2 2D H- N HSQC experiment 1 15 A 2D H- N HSQC experiment was used to show the dispersion of Mdka backbone amide signals, from which we could tell whether the protein was in 15 aggregation form. It was also used as a reference for backbone assignment. Both N 13 15 labeled and C, N labeled samples were used to record the experiments on the 800 Hz and 500 Hz NMR at 30 ºC. 1 13 2.2.3 2D H- C HSQC experiment 1 13 A 2D H- C HSQC experiment with aliphatic C-H correlation information 13 was performed, which was used as a reference for the sidechain assignment. A C, 15 N labeled sample was used to record the experiments on the 800 Hz NMR at 30 ºC. 2.2.4 H/D exchange experiment Hydrogen bond restraints were obtained by observing the presence of any hydrogen-bonded amide proton peaks in the HSQC spectrum. These amide protons 1 15 were resistant to exchange with deuterons in D2O. A H- N HSQC spectrum was recorded within 20 min after changing the sample buffer to D2O buffer. 2.2.5 3D HNCA and 3D HNCOCA experiments An HNCA experiment, which is often used in tandem with an HNCOCA experiment, can provide linkage information for the adjacent residues in a protein 29 sequence. In HNCA experiment, the magnetization of the amide proton of an amino acid residue is transferred to the amide nitrogen, and then to the alpha carbons of both the starting residue and the preceding residue in the protein's amino acid sequence. In contrast, the complementary HNCOCA experiment transfers magnetization only to 13 15 the alpha carbon of the preceding residue (Bax et al., 1991). A C, N labeled sample was used to record the experiments on the 800 Hz NMR at 30 ºC. 2.2.6 3D MQ-CCH-TOCSY experiment Like the traditional 3D HCCH-TOCSY experiment, the 3D MQ-CCH-TOCSY can correlate the sidechain aliphatic protons with the carbon spins in the same residue. 13 15 But the 3D MQ-CCH-TOCSY is more sensitive (Yang et al., 2004). A C, N labeled sample was used to record the experiment on the 800 Hz NMR at 30 ºC. 2.2.7 4D time shared NOESY experiment A time-shared 4D NOESY experiment (Xu et al., 2007) that records HN-HN NOE, HN-HC NOE, HC-HN NOE, and HC-HC NOE was used, which greatly reduced the time needed for data collection. The time-shared 4D NOESY spectrum can be split into three sub-spectra (4D NOESY and 4D 13 13 15 15 N, N-edited NOESY, 4D 13 13 15 C, N-edited C, C-edited NOESY) during data processing. A C, 15 N labeled sample was used to record the 4D NOESY on the 500 Hz NMR at 30 ºC. 2.3 Data processing NMR data were processed with NMRPipe, and NMRDraw was used to phase the spectra. All processed spectra were converted to the NMRView format for further analysis by NMRView. 30 2.4 Resonance assignment 2.4.1 Backbone assignment Peak picking of all NMR spectra was performed using the automatic function 1 15 in NMRview. All cross peaks from H- N HSQC, HNCA, HNCOCA, and 4D 13 15 15 1 C, N-edited NOESY sub-spectrum were grouped into clusters based on their H, N chemical shifts with tolerances of 0.02 ppm for HN from intralist and interlist, 0.2 ppm for nitrogen chemical shifts and 0.3 ppm for carbon chemical shifts. Chemical 1 15 shifts of the aliased peaks in the spectra were corrected according to H- N HSQC. Each cluster usually contains one HNCA peak of residual i, its preceding residual i-1, and a few HC-HN NOE peaks. Among these NOE peaks, those belonging to residual i and i-1 could be identified and differentiated from long range NOE peaks by comparing the TOCSY slices defined by the (H, C) chemical shifts of each NOE peak. α Those NOE peaks that have matched C chemical shifts with the HNCA and α HNCOCA of residual i belong to the intra-residual and sequential NOEs. Since C chemical shifts alone can be ambiguous for assignment, MQ-CCH-TOCSY that provides sidechain chemical shift is used together to confirm assignment. Spin-system classification by residue type was done according to the chemical shifts of the spins in a spin-system. Four types of amino acid residues (Ala, Thr, Ser, and Gly) have α β characteristic C and/or C chemical shifts, which were unambiguously recognized. Val and Leu were easily classified based on their characteristic sidechain chemical shifts, and Arg, Lys and Pro were also distinguished according to sidechain chemical shifts. The spin-systems were linked to build fragments when each two have identical α C chemical shifts, the largest number of matched NOE peaks and no inconsistent (C, 31 H) chemical shifts. Finally, fragments containing the characteristic residues can be mapped onto the amino acid sequence of the Mdka by matching the positions of those characteristic residues. A general protocol for backbone sequential assignment is available online at http://www.natureprotocols.com/2006/11/09/noesybased_strategy_ for_assign.php. 2.4.2 Sidechain assignment Most of the sidechain assignments were already done when doing backbone assignment (Xu et al., 2005). The H and C chemical shifts were extracted from the 4D 13 15 C, N-edited NOESY subspectrum. Some sidechain chemical shifts were assigned based on the NOE peaks of two adjacent spin-systems. The 3D MQ-CCH-TOCSY spectrum was used in combination with the 4D 13 13 15 C, N-edited NOESY and 4D 13 C, C-edited NOESY sub-spectra to perform or confirm the assignments. 2.4.3 NOE assignment, structure calculation and refinement 13 15 15 15 The NOE peaks in 4D C, N-edited NOESY, 4D N, N-edited NOESY and 4D 13 13 C, C-edited NOESY sub-spectra were assigned when a relatively complete chemical shift list was available. NOE distance constraints generated by the unambiguously assigned NOE peaks from 4D 13 13 15 C, N-edited NOESY and 4D 13 C, C-edited NOESY sub-spectra were used for the initial structure calculation. The distance upper bounds of the constraints were calibrated to three categories (2.8, 3.5, 5.5 angstroms) according to the peak volumes. Dihedral angle constraints were generated by TALOS (Cornilescu, 1999), which empirically predicts the φ and ψ α α β backbone torsion angles using a combination of three kinds (H , C , C ) of chemical 32 shifts for a given protein sequence. Only constraints that were classified as good prediction results were used for the structure calculation. Disulfide bond restraints were added only in the later stage, but not at the beginning. The initial structure calculation was done with DYANA (Herrmann et al., 2002), which started structure calculation from 50 conformers with random torsion angle values by simulated annealing with 10,000 steps per conformer. The software MOLMOL (Koradi, 1996) was used to display the structures. Energy minimization of the 5 conformers with the lowest target function values was selected. NOE peaks that gave distance violations >0.4 Å were checked and the range of angular constraints was set to two times of that predicted range by TALOS if specific dihedral angle violations were constantly larger than 5º. Initial structures were refined to minimize the violations. When a relatively good quality initial structure was acquired, ambiguous NOE peaks were then assigned according to the initial structure. Finally, 100 conformers with random torsion angle values by simulated annealing with 15,000 steps per conformer were calculated and the best 10 structures were selected for analysis. 33 Chapter 3 Results and Discussion 3.1 Mdka construct Analysis of Mdka shows that the first twenty-three amino acids are highly hydrophobic, and alignment of Mdka and human MK indicates that these amino acids are the signaling peptide of this growth factor (Fig1.5). Therefore in the construct of Mdka for protein structure elucidation, the signal peptide was removed to get the mature protein directly. With signal peptide deleted, the final plasmid of Mdka in the vector of pGEX4T-1 was successfully constructed, which was used in the expression of Mdka with a GST fusion tag at the N terminal of the protein (Fig3.1). 3.2 Expression and purification of Mdka 3.2.1 Expression of GST-Mdka in E. coli Both labeled and unlabeled Mdka with a GST tag attached to the N terminal of the protein could be easily over-expressed by E. coli BL21 (DE3), although its expression was slightly lower in M9 medium than in LB medium. 2 L M9 culture was used to obtain a sample of 500 μl 1 mM double labeled Mdka. Since GST tag has a molecular weight of around 27 kDa and the molecular weight of Mdka is 13.5 kDa, GST tagged Mdka shows a molecular weight of around 40 kDa (Fig3.2A & B). 3.2.2 Purification of Mdka After sonication, most of the protein was in supernatant (Fig3.2C). After 4 times of washing, most of the unspecific bindings could be washed out except for a few impurities (Fig3.2D). The GST tag could be easily cleaved with 1 U thrombin cleaving about 1 mg protein within 1 hour. After GST beads purification, relatively pure protein was obtained, but it was not pure enough for NMR experiments until 34 A B Fig3.1 Final construct of Mdka. (A) Final amino acid sequence of the expressed Mdka in the E. coli expression system. (B) Amino acid components of Mdka. 35 Fig3.2 Mdka expression and purification by GST beads. (A) Expression in LB media. Lane 1: marker; Lane 2: pre-induction when the cell density reached OD600 0.6, which was expression control; Lane 3: post-induction with 0.05 mM IPTG at 16 °C for 12 hours, the GST-tagged Mdka shows a molecular weight of around 40 kDa. (B) Expression in M9 media. Lane 1: post-induction with 0.3 mM IPTG at 20 °C for 20 hours; Lane 2: protein marker. (C) Solubility of Mdka. Lane 1: marker; Lane 2: pellets after sonication; Lane 3: supernatant after sonication. (D) GST purification. Lane 1: thrombin cleavage of Mdka for 3 hours (1 mg Mdka/ 1 U thrombin); Lane 2: thrombin cleavage of Mdka for 1 hour; Lane 3: marker; Lane 4: Mdka on GST beads after washing and before thrombin cleavage; Lane 5: the first time washing of Mdka with lysis buffer; Lane 6: the fourth time washing with lysis buffer; Lane 7: GST tag on GST beads after elution of Mdka from beads; Lane 8: the first time elution of Mdka from GST beads; Lane 9: the second time elution of Mdka from GST beads; Lane 10: the third time elution of Mdka from GST beads. 36 further purification of Mdka with a heparin affinity column (Fig3.3). 3.3 General properties of Mdka Mdka is a small protein with a molecular weight of 13.5 kDa, which was verified by Q-TOF mass spectrum (Fig3.4). The theoretical molecular weight for Mdka is 13521.4 Da (Cys in reduced form; software ProtParam), thus if Cys residues are oxidized, the molecular weight should be 13511.4 Da. The Q-TOF mass spectrum shows that the molecular weight is 13510.6 Da, which indicates that the Cys residues of Mdka could be in the oxidized form. Besides, the CD spectrum profile (Fig3.5), especially the positive hump at the position of 230 nm, suggests that Mdka has a β-turn structure. The solubility of the protein could be as high as 1 mM as measured with UVA280. When the protein is concentrated to 1 mM, it is still in the monomer form as evidenced by the DLS result (Fig3.6). Furthermore, 1D NMR and 2D HSQC experiments demonstrate that the recombinant Mdka purified from E. coli in this way is well structured (Fig3.7). Mdka is also a stable protein that can stand 30 °C for one week (Fig3.7). Since during the heparin affinity column purification, high concentration of NaCl is needed to elute the protein out of the column, this suggests that the recombinant protein has strong binding affinity to heparin (Fig3.3A). 3.4 Mdka in vivo assay: zebrafish embryonic development activity The activity of recombinant Mdka was tested in vivo using zebrafish embryo by our collaborator (Winkler, Dept of Biological Sciences, NUS). The recombinant Mdka was injected into zebrafish embryos, and the development of these embryos at different stages are compared with the wild type embryos as well as those injected with in-vitro-synthesized mdka RNA (Winkler et al., 2003). As a result, embryos 37 A volume (ml) Fig3.3 Heparin affinity column purification of Mdka. (A) Heparin column profile of Mdka. Mdka peak is indicated by Peak 2. (B) Peak fracture of Mdka from heparin affinity column. Lane 1: marker; Lane 2: from Peak 1; Lane3-7: from Peak 2. (C) Final concentrated Mdka for NMR experiment. 38 Fig3.4 Q-TOF mass spectrum of Mdka. The protein was desalted and prepared in water. The molecular weight of Mdka is 13510.6 Da. 39 Fig3.5 CD spectrum of Mdka. Sample was prepared with 10 mM phosphate buffer at pH 6.5. Spectrum was recorded at room temperature. 40 A Fig3.6 Apparent hydrodynamic MW of Mdka from DLS measurement. (A) Size distribution of Mdka protein particle in the buffer of 20 mM phosphate, 100 mM NaCl, pH 6.5, at 30 °C. The average size of Mdka protein is 2.02 nm. (B) Average molecular weight (MW) of Mdka. 20 MW values were analyzed, and the average MW is 13.7 kDa with a standard deviation of 0.66 kDa. 41 Fig3.7 1D and 2D NMR experiments of Mdka. (A) 1D NMR spectrum was acquired on a 500 MHz NMR machine at 20 ºC in 20 mM Tris buffer, 100 mM NaCl, 2 mM EDTA at pH 8.0. (B) Overlay of two sets of 2D 1H-15N HSQC spectra (magenta: acquired one week before, black: acquired one week later). The sample was stored at 30 ºC for one week. Spectra were recorded at 800 MHz NMR at 30 ºC in 20 mM phosphate buffer, pH 6.5, 100 mM NaCl, 1 mM EDTA. 42 injected with recombinant Mdka show the same development phenotypes as RNA-injected embryos, which indicates that the recombinant protein is properly folded and has the proposed biological functions (Winkler, data not published). 3.5 NMR assignment of Mdka 3.5.1 Backbone assignment of Mdka Based on the 3D HNCA, 3D HNCOCA, 4D 13 15 C, N-edited NOESY and 3D CCH-TOCSY spectra, all the spin systems in the two domain regions of Mdka were connected together, and the loop regions were assigned as long as the peaks appeared 1 15 in the H- N HSQC. Totally, there are about ten peaks missing (Fig3.9B) in the 1 15 H- N HSQC among the 125 amino acids of Mdka (with GS extension at the N terminal of the protein resulting from thrombin cleavage). Figure 3.8 is a part of the 1 15 backbone assignment result for residues from E39 to T41. The H- N HSQC spectrum of assigned Mdka backbone peaks was shown in figure 3.10. 3.5.2 Sidechain assignment of Mdka When backbone assignment was finished, the spin systems were identified, and therefore most of the sidechain chemical shifts were available. Some sidechain assignments need to compare two adjacent 4D 13 15 C, N-edited NOESY planes to decide the resonances. Some need to compare the 4D 13 13 C, C-edited NOESY subspectrum. The chemical shifts of the CH groups that have no NOE on the 4D 13 15 C, N-edited NOESY could be assigned by examining the CCH-TOCSY spectrum and 4D 13 13 C, C-edited NOESY. The assigned proton chemical shifts should be in consistent with the chemical shifts statistics of the twenty amino acids. The 43 Fig3.8 Backbone connectivity of Mdka. (A) Cα connectivity for a stretch of residues from E39 to T41. (B) 4D 13C,15N-edited NOESY slices of E39, G40 and T41. Each 1 H-13C slice is labeled with proton identities with sequential connectivity information. 15 N and 1H frequencies in ppm with assigned residues are indicated at the top of each slice. (Labels are in the NMRView format.) 44 B Fig3.9 Backbone assignment of Mdka. (A) Spin system identification. T41 was indentified because of its characteristic Cβ and Cγ chemical shifts. G40 was indentified according to its small Cα chemical shift and two HαCα peaks. G40 and T41 form a fragment, and when comparing with the sequence of Mdka, there are only two GT fragments (EGT and TGT). Since E39 clearly is not Thr, thus, it could be identified to be Glu. (B) Assigned backbone peaks (magenta) and missing backbone peaks (black). (Labels are in the NMRView format.) 45 Fig3.10 Assigned 1H-15N HSQC spectrum of Mdka backbone peaks. Spectrum was acquired at 800 MHz and 30 ºC on a sample of 0.8 mM protein and pH 6.5. The three Trp sidechain peaks are assigned. The labeled sequence numbers correspond to those in the final construct, and are consistent with NMRView format. 46 assignment of sidechain proton and carbon spins could further be verified by 1H-13C HSQC spectrum. Sidechain assignment for the loop regions was not as complete as the domain region. Totally, 78.3% aliphatic sidechain resonances were assigned. Aromatic sidechain assignment was also finished. 3.5.3 Secondary structure characterization from backbone assignment α β Since the C and C chemical shifts of an amino acid could be used as an indicator of the local secondary structure information, the secondary structure of α β Mdka was analyzed by comparing the C and C chemical shifts of each amino acid of α β Mdka with those of the random coil. The deviation ΔC /ΔC equals the chemical shift α β of the observed C /C for an Mdka residue minus that for the same residue in a α random coil structure. Positive secondary shift for C indicates that the secondary structure is helix, while negative result means that it is β-strand conformation. This β α β trend is opposite in sign for C secondary shift. Thus ΔC -ΔC could give an enhanced effect of the secondary structure information. Chemical shift index (CSI) of Mdka α β (Fig3.11) was obtained by plotting ΔC -ΔC against the amino acid sequence. From the CSI, it can be inferred that the protein is mainly composed of β-sheets in the two domain regions, and there are three long loops at the two ends of the two domains as well as in-between. 3.5.4 NOE assignment When the backbone and sidechain assignments were finished, most of the NOE peaks in the 4D 13 13 15 C, N-edited NOESY, 4D 15 15 N, N-edited NOESY and 4D 13 C, C-edited NOESY sub-spectra could be unambiguously assigned (Fig3.12). 47 Ambiguous NOE peaks were assigned based on the initial structure calculated with the unambiguously assigned NOE peaks. HN-HN NOE peaks were assigned on the 4D 15 15 N, N-edited NOESY subspectrum where the C chemical shift could be converted into the N chemical shift with the equation: N (ppm) = (C-Cref)Csf/Nsf (ppm) + Nref (ppm) (Fig3.14C). Totally, there are 2085 NOE peaks: 926 intra-residual NOEs, 618 sequencial NOEs, 152 medium-range NOEs, and 389 long-range NOEs. As shown in the analysis of sequential- and medium-range NOEs (Fig3.13), strong α α NOE peaks between H C i-1 and HNi were observed, which are consistent with the β-structure of Mdka. Successive strong long-range NOEs of HN-HC and HC-HC were observed, which are from the opposite strands of the antiparallel β-sheet. 3.5.5 Structure calculation and refinement Initial structures calculated using the unambiguously assigned NOE peaks (distance constraints) and torsion angle constraints initially showed relatively large average target function. The main reason is because of the overlap of NOE peaks that results in larger peak volumes than they should be. After all the NOE peaks were assigned, 100 structures were calculated, and the best 10 structures show an average target function of 1.49. The N domain and C domain RMSD values between the 10 conformers and the mean structures for the backbone heavy atoms are 0.76 and 1.65 respectively. Each domain contains three anti-parallel β-sheets (Fig3.14). The β distances between the C atoms of the two nearby Cys residues of the ten structures are shown in Table 1. As can be seen, except the pair of C75-C107 and C65-C97, all β the Cys pairs show distances of the C atoms between 3 Å to 8 Å. The C65-C97 β β shows very close distances because both of the Cys H C protons have NOEs with 48 Fig3.11 Chemical shift index of Mdka. ΔCα/ΔCβ equals Cα/Cβ of Mdka amino acid minus that of the same amino acid in a random coil structure. 49 A C B 1 15 T87 ( HN:8.27ppm, N: 115.65ppm) Fig3.12 NOE assignment. (A) Representative slices of HN-HC assignment from the 4D 13C,15N-edited NOESY subspectrum. (B) HC-HC NOE assignment from the 4D 13 13 C, C-edited NOESY subspectrum. (C) HN-HN assignment from the 4D 15 15 N, N-edited NOESY subspectrum. The C chemical shifts and N chemical shifts can be transferred using the equation of Ncs = (Ccs-Cref)Csf/Nsf+Nref (Labels are in the NMRView format.) 50 Fig3.13 Sequential- and medium-range NOE pattern of Mdka. The NOE intensities are represented by the thickness of the solid bars. 51 L88 HδCδ proton. C75-C107 pair shows relatively larger distances than the other pairs, ranging from 7.3 Å to 12.3 Å. 3.5.6 Disulfide restraints determination β β Although there is no direct NOE between the H C of any Cys residues observed, the ten Cys residues in Mdka should form five pairs of disulfide bonds, which was determined empirically and experimentally. The sequence alignment of the Midkine family shows that the Cys residues are highly conserved among the family (Fig1.5), and Mdka has sequence identity as high as 58% with human MK. Thus it is expected that the overall structure of the two proteins would be more or less the same, which indicates that the disulfide linkage of the two proteins would be identical. Therefore the five pairs of the disulfide bridge should be: C18-C42, C26-C51, C33-C55, C65-C97, and C75-C107 according to the sequence of Mdka construct in this study. In addition, Q-TOF mass spectrum shows that the molecular weight of the recombinant Mdka is 13510.6 Da. Theoretically, the molecular weight of Mdka is 13511.4 Da if five pairs of disulfide bond are formed, and 13521.4 Da if not (software ProtParam). This suggests that there are five pairs of disulfide bridges in the β recombinant Mdka protein. Besides, NMR analysis shows that the C chemical shifts of all Cys in the protein sequence are larger than 30 ppm, which also indicates disulfide bonds formation (Sharma and Rajarathnam, 2000). Moreover, the initial β structure of Mdka shows that the distances between the C atoms of the Cys pairs are all around 3.5 Å to 8 Å (C75-C107 within 7.3 Å to 12.3 Å ), which is near enough to form disulfide bonds. Therefore, structures with disulfide constraints were calculated. The target function of the structures calculated with these five pairs of disulfide constraints is similar with that calculated without. Therefore, although there is no 52 A B Fig3.14 Comparison of Mdka domain structures calculated with and without disulfide constraints. (A) Initial Mdka domain structures without disulfide constraints. Cys residues are colored in blue. (B) Mdka domain structures calculated with disulfide constraints. 53 Table 1 Distances between the Cβ atoms of nearby Cys residues in the 10 conformers. 54 β β β β direct H C -H C NOE observed among the ten Cys residues, it can be concluded that the disulfide bonds should be formed in this way. 3.6 NMR structure of Mdka The final structure of Mdka was calculated with distance constraints, dihedral angle constraints, hydrogen bond constraints, and disulfide bond constraints. The final structural statistics are listed in Table 2. G1-D17 is the N terminal loop; C18-C55 forms the N domain; N56-D64 is the liker between the two domains; C65-C107 is the C domain; and T108 to the end is the C terminal tail. Since there is no confirmed interaction observed among the three loops and the two domains, these parts are supposed to be able to move freely from each other (Fig3.15). The RMSD between the 10 conformers and the mean structures are 0.56 Å (L21-C55) and 0.54 Å (Y67-C107, excluding the long loop) for backbone heavy atoms, and 1.12 Å (L21-C55) and 0.82 Å (Y67-C107, excluding the long loop) for all non-hydrogen atoms in the well-defined regions, respectively. N domain (C18-C55) consists of three antiparallel β-sheets, βN1 (L22-A28), βN2 (G36-C42), βN3 (T46-V53), which are connected by two loops. C domain (C65-C107) also consists of three antiparallel β sheets: βC1 (W67-D76), βC2 (T81-Q87), βC3 (T100-K105). There are also two loops connecting these three β strands, but the loop between βC1 and βC2 is longer than the others. All the three proline residues are in the trans configuration as evidenced by γ γ α α the observation of NOE connectivity of proline H C and H C of the preceding residues (Fig3.17). Hydrophobic, charged, and other polar residues are colored on the space filling model of Mdka (Fig3.15). Given that Mdka is highly basic charged, and it has strong binding ability to the negative charged heparin molecule, the basic 55 Table 2 Structural statistics of Mdkaa a 10 energy-minimized 10 structures were investigated. bDihedral angle constraints were generated by TALOS. cHydrogen bonds were identified on the basis of NH-D2O exchange experiment. Two constraints per hydrogen bond (dHN-O ≤ 2.2 Å and dN-O ≤ 3.2 Å) were included in the final structure calculations by visual inspection of preliminary structures derived from the NOE data. dPROCHECK-NMR was used to assess the quality of the structures. 56 Mdka N domain Mdka C domain Fig3.15 Solution structure of Mdka. (A/E) Superposition of the backbone heavy atoms of the 10 best calculated Mdka structures. The Mdka conformers are fitted for minimal RMSD of the backbone atoms of all residues in both domains. (B/F) Ribbon drawing of the best conformer of Mdka. (C/G) Space filling model of Mdka domains with annotations of basic (blue), acidic (red), hydrophobic (pink), and polar (purple) residues on the surface. (D/H) Back view of C/G. (I) Ribbon model of the best conformer of the whole protein. 57 Fig3.16 Ramachandran plot of Mdka. 58 Fig3.17 Configuration of proline residues. 4D are listed on the left. 13 C,13C-edited NOESY subspectra 59 residues in the two domains of Mdka colored in blue seem to form clusters and could be the possible binding sites for heparin. 3.7 Structural comparison to other Midkine family members 3.7.1 Structure comparison of Mdka and human MK The three dimensional domain structures of Mdka were compared with those of human MK. The N domain backbone architectures of the two proteins are relatively similar, but C domains have larger differences (Fig3.18). The loop between βC2 and βC3 in human MK is bent towards the three β sheets because NOEs are detected among them. In Mdka no such NOE is observed, and the position of the βC1 seems to prevent the long loop from bending over. The linear sequence alignment of the two proteins shows an overall amino acid identity of 58% (Fig1.5), with 57% and 65% identity for the N domain and C domain respectively. The conserved residues as well as non-conserved ones are colored on the transparent surface diagram of Mdka, where a cluster of non-conserved residues is found on one side in the C domain of Mdka (Fig3.19 C). These residues might be responsible for the different functions between Mdka and human MK. In contrast, a cluster of identical residues is found on the opposite side in the C domain, where basic residues cluster together (Fig3.22 D), which might form the heparin binding site of Mdka. In addition, there are some similarities between the structures of Mdka and human MK. Both the N domain and C domain of MK display a β-bulge-like structure because of Pro22 in the N domain and Asn68-Trp69-Gly70 fragment in the C domain. While the corresponding position of MK Pro22 is changed to Ser25 in Mdka, yet the bulge is still there. The twist in the βN1 (L22-A28) of both proteins is also similar. Resembling of the C domain of human MK, Mdka has a β-bulge-like structure due to 60 Fig3.18 Comparison of Mdka with human MK. (A/D) N/C domain backbone superimposition of Mdka (cyan) and MK (magenta). (B/E) Ribbon diagram of Mdka N domain and C domain. (C/F) Ribbon diagram of human MK N domain and C domain. [Human MK structure is from PDB file 1mkc and 1mkn. Ribbon structures are from the reference (Muramatsu, 2002)]. 61 B C 1800 1800 Fig3.19 Sequence comparison of Mdka and human MK. (A) Sequence alignment of N domain and C domain of Mdka and human MK. {Note: red: small [small + hydrophobic (incl.aromatic -Y)], blue: acidic, magenta: basic, green: hydroxyl + amine + basic – Q; "*" means that the residues in that column are identical in all sequences in the alignment; ":" means that conserved substitutions have been observed, according to the color table available at http://www.ebi.ac.uk/Tools /clustalw2/scorestable_frame.html; "." means that semi-conserved substitutions are observed.} (B) N domain (C) C domain of Mdka represented by the transparent surface diagram with the ribbon digram beneath it. Residues are colored according to the above sequence alignment result. (Note: blue: identical residues; yellow: conserved substitutions have been observed in human MK; orange: semi-conserved substitutions are observed in human MK; red: non-conserved residues; green: glutamine residue that might mediate dimerization in Mdka.) 62 Gly70-Asn71-Trp72 fragment. Sequence homology that is observed in human MK between the βN2 (Arg35-Glu36-Gly37-Thr38) and βC2 (Arg81-Gln82-Gly83-Thr84) as well as βN3 (Ile46-Arg47, Pro51-Cys52) and βC3 (Ile98-Arg99, Pro103-Cys104) is not found in Mdka. Instead, a homology is found between the N terminal loop (Gly15-Ala16-Asp17 -Cys18) and the end of the linker (Gly62-Ala63-Asp64-Cys65). But we have no idea about whether the existence of these repetitive fragments in this growth factor family is a coincidence or results from evolution. The C domain heparin binding site of MK was mapped using NMR perturbation by adding heparin 12mer solution. Iwasaki et al. discovered that K79 (βC2), R81 (βC2), and K102 (βC3) from the C domain β-sheets as well as K86, K87, and R89 from the long loop between βC2 and βC3 are the key residues involving in heparin binding. The corresponding six residues in Mdka are: K82 (βC2), R84 (βC2), K105 (βC3), Q89, K90, and L92. Therefore, the three charged residues on the β-sheets are very likely to be involved in heparin binding, but whether residues from the long loop participate in this activity needs further validation. Moreover, the stoichiometry of MK-heparin oligosaccharide complex was investigated by laser light scattering experiments, and it was reported that MK formed a dimmer upon binding to heparin of >20 monosaccharide units. Kojima et al. also reported the property of MK dimmerization, and three glutamine residues, although not conserved among species through evolution, had been identified to be responsible for the transglutaminase-mediated dimer formation of MK. Among the three residues, two are conserved in Mdka (Q45 and Q98), and one is changed to R47. The two glutamine residues are not far away from the non-conserved residues (Fig3.19 B & C). So far we have no idea whether Mdka would dimerize when it binds to heparin. 63 3.7.2 Structure comparison of Mdka with human PTN The structure of E.coli-expressed human PTN was analyzed by NMR (Kilpeläinen et al., 2000; Fig3.19 & Fig3.20), and its heparin binding activity was also investigated. Although no three dimensional structure of PTN is found in PDB, its secondary structure could be compared with that of Mdka. The main difference lies in the N domain, which has amino acid identity of 47%; the loop connecting βN2 and βN3 is longer in PTN than Mdka. The C domains of the two proteins are highly conserved with sequence identity as high as 60%. Similar features of the two proteins include the antiparallel β-sheet structures of the two domains, a highly conserved linker region, and the Lys-rich N-terminal and C-terminal tails. The heparin binding activity of PTN was analyzed using NMR and heparin affinity chromatography. The results show that both the N- and C-β-sheet domains bind to heparin in solution. NMR results also suggest that the Lys-rich tails, though highly charged, do not contribute to the binding of PTN to heparin and remain unstructured in the PTN-heparin complex. Upon binding to heparin, PTN undergoes structure changes according to the CD spectroscopy, especially prominent at wavelengths of 200 and 230 nm. However, how the structure changed is still unknown. Since human MK could form a dimmer when binding to heparin, the CD spectrum profile changes of PTN might be due to the dimmerization as well. The disulfide bonds in PTN are essential to maintain the native structure since reduction of disulfide bonds dramatically reduced heparin binding. Therefore, it can be inferred that disulfide bonds in Mdka play pivotal role to the formation of the β-sheet structure. Heparin binding activity of Mdka might also be due to the β-sheet domain structure rather than the charged N-terminal and C-terminal tails. 64 Fig3.20 Sequence alignment of Mdka and PTN. β sheets of PTN were shown as white arrows (Kadomatsu and Muramatsu, 2004). {Note: red: small [small + hydrophobic (incl.aromatic -Y)], blue: acidic, magenta: basic, green: hydroxyl + amine + basic – Q; "*" means that the residues in that column are identical in all sequences in the alignment; ":" means that conserved substitutions have been observed, according to the colour table available at http://www.ebi.ac.uk/Tools /clustalw2/scorestable_frame.html; "." means that semi-conserved substitutions are observed.} (Software ClustalW2, EBI, ExPASy). 65 Fig3.21 Secondary structure of PTN. Curved arrows show the location of disulfide bonds. Straight arrows indicate long-range NOE interactions (Kilpeläinen et al., 2000). 66 3.7.3 Comparison of Mdka with Mdkb The linear sequence alignments of the two domains of Mdka and Mdkb suggest that the sequence identity of the two proteins is very high, with 71% and 69% identity for the N domain and C domain respectively. Thus the three dimensional structures of the two proteins might be of high similarity. Residues that are conserved between the two proteins as well as those have substitutions are colored according to the sequence alignment result (Fig3.22). One conserved glutamine residue, which is involved in the dimerization of MK, is in the vicinity of the non-conserved residue in the C domain of Mdka (Fig3.22 D). From the sequence comparison, nine basic residues are identified in the C domain of Mdkb, among which six are conserved in Mdka and the other three are substituted by neutral amino acids in Mdka. These nine basic residues form two basic clusters, and might be the heparin binding site of Mdkb. In contrary to C domain, one basic residue in the N domain of Mdkb has changed to a neural one. Therefore, the functional difference of Mdka and Mdkb might lie in the fact that the charges of the C domains of the two proteins are different. Since Midkine family could bind to the transmembrane heparan sulfate proteoglycans to induce intracellular signaling pathways, the charge difference of the C domains of Mdka and Mdkb might possibly lead them to bind to different kinds of transmembrane receptors, thus inducing different signaling pathways. 67 B C 1800 1800 D 1800 Fig3.22 Sequence comparison of Mdka and Mdkb. (A) Sequence alignment of N domain and C domain of Mdka and Mdkb. (Note: residues are colored in the same way as in Fig3.20) (B) N domain (C) C domain of Mdka represented by the transparent surface diagram with the ribbon diagram beneath it. Residues are colored according to the above sequence alignment result. (Note: residues are colored in the same way as in Fig3.20.) (D) Basic residues in the sequence of Mdkb are colored accordingly in the C domain of Mdka. Hot pink: residues conserved between Mdka and Mdkb. Purple: basic residues in Mdkb that are substituted by neural ones in Mdka. Green: glutamine residue that might mediate dimerization in Mdka and Mdkb. 68 Chapter 4 Conclusion and Future Work Recombinant Mdka is expressed as a well structured soluble protein using E. coli BL21 (DE3) strain, and is suitable for NMR study. It has the same disulfide linkage pattern as human MK and PTN. As a heparin binding growth factor, it shows strong binding to heparin in vitro. Zebrafish embryos after injection of the E. coli-expressed Mdka show the same development phenotypes as those injected with in-vitro-synthesized RNA, which means that the recombinant protein is properly folded and has the desired biological functions. The solubility of Mdka is quite high; when it is concentrated to 1 mM, the protein still exists in monomer form. Besides, it is also a stable protein that can stand 30 °C for one week. All these properties of Mdka make it suitable for structure elucidation by NMR. Moreover, sequence alignment of Mdka with human MK reveals many long loops present in Mdka, which suggests that it would be extremely difficult to crystallize the protein. Thus, NMR becomes the only way to solve the structure of this protein. In this study, the solution structure of recombinant Mdka is determined based on the multidimentional NMR using a novel backbone assignment method developed by our group. The 4D NMR method is originally designed for structure elucidation of larger proteins (~40-50 kDa), this work shows that this method can also be used to solve high resolution structures for small proteins despite the fact that the 4D NMR method may lose some peak intensity. The structure of Mdka shows that there are two long tails and two domains. Connecting the two domains is a linker with amino acids highly conserved among the MK family. We are not sure whether the linker has specific functions, and it seems that the linker is also flexible. Each domain is composed of three antiparallel β sheets, resembling that of human MK. Comparing 69 individual domains of Mdka with human MK indicates that the N domain backbone architectures of the two proteins are more or less the same; while C domains show larger differences between the two proteins, especially for the position of the long loop between βC2 and βC3. In human MK, this long flexible loop is bent towards the three β sheets. In Mdka, the position of the βC1 seems to prevent the long loop from bending over. Therefore, the C domain of human MK has a more globular shape, while the C domain of Mdka seems more flat. By comparing Mdka with other members of the MK family, we can infer that C domain of Mdka might contribute to most of the heparin binding activity. The highly positive charged N terminal and C terminal tails might not be involved in heparin binding. Instead, charged residues on the C domain could be the possible heparin binding sites. Besides, Mdka could possibly form a dimer upon binding to heparin. With the structure of Mdka available, future work can be carried on in the following directions. First of all, to understand why the two Midkine proteins in zebrafish, Mdka and Mdkb, have different expression and functional patterns, the structures of the two proteins can be compared (Lim Jack Wee in our group is currently solving the structure of Mdkb). Since the sequence alignment of the two proteins shows that the domain regions have high sequence identity (70%), it is very likely the two proteins have similar structural architecture. If the structure is similar, then it would be interesting to pursue the reason why similar protein structures can lead to different functions. Of interesting notice is that there is charge difference between the C domains of the two proteins; Mdkb has three more basic residues than Mdka does in the possible heparin binding site. This charge difference might possibly be the reason why Mdka and Mdkb perform differently during zebrafish 70 embryogenesis. Besides, the heparin binding sites of Mdka need to be identified to understand where the key residues are that directly involve in the functions of this protein. Methods such as titration of heparin to Mdka, NMR perturbation, and site-directed mutagenesis of Mdka in binding with heparin can be used to pursue this target. Since Mdka has neural development function as well as heparin binding activity, it is possible that these two functions are performed via two different binding sites. Since the C domain of human MK is well studied and it is believed to be essential for heparin binding activity, the same situation might also be applied to the C domain of zebrafish Mdka. Therefore, charged residues on the Mdka C domain might be the candidates to do mutagenesis analysis to map the binding sites to heparin. Yet the N domain function of this family is still unclear. Whether the N domain is responsible for the neural development activity needs further validation. In addition, if possible, the binding complex of Mdka to its partner can be analyzed to get a view of the complex structure. So far the most commonly used ligand to evaluate the activity of MK family is heparin. Native heparin is a polymer with a molecular weight ranging from 3 kDa to 40 kDa, although the average molecular weight of most commercial heparin preparations is in the range of 12 kDa to 15 kDa. Iwasaki et al. found that heparin of >20 monosaccharide units would induce MK dimerization, and Kaneda (1996b) et al. reported that the minimize size for heparin to inhibit MK from binding to neurons is approximately 22 monosaccharide units (7 kDa). Complex structure would provide answers to questions such as how the dimer is fomed, how the two domains are orientated when binding to the ligand, and what the position of the linker and the terminal tails are. Moreover, to fully understand how the two zebrafish Midkines function, why 71 two proteins are needed in zebrafish while in the higher vertebrates only one homolog is present, and what the factors are to differentiate these two proteins, the signaling pathways of the two growth factors involved should be studied before these questions can be answered. 72 Chapter 5 References Bax, A., Ikura, M. (1991) An efficient 3D NMR technique for correlating the proton and 15 N backbone amide resonances with the alpha-carbon of the preceding 15 13 residue in uniformly N/ C enriched proteins. J. Biomol. NMR 1: 99-104. Bronner-Fraser, M. (1995) Origins and developmental potential of the neural crest. Experimental Cell Research 218: 405-417. Cornilescu, G., Delaglio, F., and Bax, A. (1999) Protein backbone angle constraints from searching a database for chemical shift and sequence homology. J. Biomol. NMR 13: 289-302. Chang, M.H., Huang, C.J., Hwang, S.P., Lu, I.C., Lin, C.M., Kuo, T.F., and Chou, C.M. (2004) Zebrafish heparin-binding neurotrophic factor enhances neurite outgrowth during its development. Biochem. Biophys. Res. Commun. 321: 502-509. Dahm, R. (2006) The zebrafish exposed. American Scientist 94: 446-453. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995) NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6: 277-293. Herrmann, T., Guntert, P., and Wuthrich, K. (2002) Protein NMR structure determination with automated NOE assignment using the new software CANDID and the torsion angle dynamics algorithm DYANA. J. Mol. Biol. 319:209-227. Herz, J. and Beffert, U. (2000) Apolipoprotein E receptors: linking brain development and Alzheimer’s disease. Nat. Rev. Neurosci. 1: 51–58. Iwasaki, W., Nagata, K., Hatanaka, H., Inui, T., Kimura, T., Muramatsu, T., Yoshida, 73 K., Tasumi, M. and Inagaki, F. (1997) Solution structure of midkine, a new heparin-binding growth factor. The EMBO Journal 16: 6936-6946. Jessen, R. K. and Mirsky, R. (2005) The origin and development of glial cells in peripheral nerves. Nature Reviews Neuroscience 6: 671-682 . Johnson, B. A. (2004) Using NMRView to visualize and analyze the NMR spectra of macromolecules. Methods Mol. Biol. 278: 313-352. Kadomatsu, K., Tomomura, M. and Muramatsu, T. (1988) cDNA cloning and sequencing of a new gene intensely expressed in early differentiation stages of embryonal carcinoma cells and in mid-gestation period of mouse embryogenesis. Biochem. Biophys. Res. Commun. 151: 1312-1318. Kadomatsu, K., Muramatsu, T. (2004) Midkine and pleiotrophin in neural development and cancer. Cancer Letters 204: 127-143. Kaneda, N., Talukder, A.H., Nishiyama, H., Koizumi, S. and Muramatsu, T. (1996a) Midkine, a heparin-binding growth/differentiation factor, exhibits nerve cell adhesion and guidance activity for neurite outgrowth in vitro. J. Biochem. 119: 1150-1156. Kaneda, N., Talukder, A. H., Ishihara, M., Hara, S., Yoshida, K. and Muramatsu, T. (1996b) Structural characteristics of heparin-like domain required for interaction of midkine with embryonic neurons. Biochem. Biophys. Res. Commnun. 220: 108–112. Kato, M., Shinozawa, T., Kato, S., and Terada, T. (2000) Divergent expression of midkine in the human fetal liver and kidney: Immunohistochemical analysis of developmental changes in hilar primitive bile ducts and hepatocytes. Liver 20: 475-481. Kilpeläinen, I., Kaksonen, M., Kinnuneni, T., Avikainen, H., Fath, M., Linhardt, R. J., 74 Rauloi, E. and Rauvalai, H. (2000) Heparin-binding growth-associated molecule contains two heparin-binding β-sheet domains that are homologous to the thrombospondin type I repeat. J. Biol. Chem. 275: 13564-13570. Kojima, T., Katsumi, A., Yamazaki, T., Muramatsu, T., Nagasaka, T., Ohsumi, K., and Saito, H. (1996) Human ryudocan from endothelium-like cells binds basic fibroblast growth factor, midkine, and tissue factor pathway inhibitor. J. Biol. Chem. 271: 5914–5920. Kojima, S., Inui, T., Muramatsu, H., Suzuki, Y., Kadomatsu, K., Yoshizawa, M., Hirose, S., Kimura, T., Sakakibara, S. and Muramatsu, T. (1997) Dimerization of midkine by tissue transglutaminase and its functional implication. J. Biol. Chem. 272: 9410-9416. Koradi, R., Billeter, M., and Wuthrich, K. (1996) MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14:51-32. Li, Y. S., Milner, P. G., Chauhan, A. K., Watson, M. A., Hoffman, R. M., Kodner, C. M., Milbrandt, J. and Deuel, T. F. (1990) Cloning and expression of a developmentally regulated protein that induces mitogenic and neurite outgrowth activity. Science 250: 1690-1694. Liedtke, D. and Winkler, C. (2008) Midkine-b regulates cell specification at the neural plate border in zebrafish. Developmental Dynamics 237: 62-74. Maeda, N., Nishiwaki, T., Shintani, T., Hamanaka, H., and Noda, M. (1996) 6B4 Proteoglycan/phosphacan, an extracellular variant of receptor-like protein-tyrosine phosphatase ζ/RPTPβ, binds pleiotrophin/heparin-binding growth-associated molecule (HB-GAM). J. Biol. Chem. 271: 21446–21452. Maeda, N. and Noda, M. (1998) Involvement of receptor-like protein tyrosine phosphatase ζ/RPTPβ and its ligand pleiotrophin/heparin-binding 75 growth-associated molecule (HB-GAM) in neuronal migration. J. Cell Biol. 142: 203-216. Maeda, N., Ichihara-Tanaka, K., Kimura, T., Kadomatsu, K., Muramatsu, T. and Noda, M. (1999) A receptor-like protein-tyrosine phosphatase PTPζ/RPTPβ binds a heparin-binding growth factor midkine. Involvement of arginine 78 of midkine in the high affinity binding to PTPζ. J. Biol. Chem. 274: 12474-12479. Mayden, R. L., Tang, K. L., Conway, K. W., Freyhof, J., Chamberlain, S., Haskins, M., Schneider, L., Sudkamp, M., Wood R. M., Agnew, M., Bufalino, A., Sulaiman, Z., Miya, M., Saitoh, K., He, S. (2007) Phylogenetic relationships of Danio within the order Cypriniformes: a framework for comparative and evolutionary studies of a model species. J. Exp. Zool. (Mol. Dev. Evol.) 6: 17554749. Mishima, K., Asai, A., Kadomatsu, K., Ino, Y., Nomura, K., Narita, Y., Muramatsu, T., and Kirino, T. (1997) Increased expression of midkine during the progression of human astrocytomas. Neurosci. Lett. 233: 29-32. Mitsiadis, T., Salmivirta, M., Muramatsu, Muramatsu, T. H., Rauvala, H., Lehtonen, E., Jalkanen, H., and Thesleff, I. (1995) Expression of the heparin-binding cytokines, midkine (MK) and HBGAM (pleiotrophin) is associated with epithelial–mesenchymal interactions during fetal development and organogenesis. Development 121: 37–51. Moradian-Oldak, J., Leung, W., and Fincham, A. G. (1998) Temperature and pH-dependent supramolecular self-assembly of amelogenin molecules: a dynamic light-scattering analysis. J. Struct. Biol. 122: 320-327. Muramatsu, H., Inui, T, Kimura, T, Sakakibara, S., Song, X., Murata, H. and Muramatsu, T. (1994) Localization of heparinbinding, neurite outgrowth and antigenic regions in midkine molecule. Biochem, Biophys. Res. Commun. 203: 76 1131. Muramatsu, T. (2002) Midkine and pleiotrophin: two related proteins involved in development, survival, inflammation and tumorigenesis. J. Biochem. 132: 359-371. Nakanishi, T., Kadomatsu, K., Okamoto, T., Ichihara-Tanaka, K., Kojima, T., Saito, H., Tomoda, Y., and Muramatsu, T. (1997) Expression of syndecan-1 and -3 during embryogenesis of the central nervous system in relation to binding with midkine. J. Biochem. 121: 197–205. Rabi, I. I., Zacharias, J. R., Millman, S. and Kusch, P. (1938) A new method of measuring nuclear magnetic moment. Physical Review 53: 318-318. Raulo, E., Chernousov, M.A., Carey, D.J., Nolo, R., and Rauvala, H. (1994) Isolation of a neuronal cell surface receptor of heparin binding growth-associated molecule (HB-GAM). Identification as N-syndecan (syndecan-3). J. Biol. Chem. 269: 12999–13004. Rauvala, H. (1989) An 18-kd heparin-binding protein of developing brain that is distinct from fibroblast growth factors. EMBO J. 8: 2933-2941. Rauvala, H., Vanhala, A., Castren, E., Nolo, R., Raulo, E., Merenmies, J. and Panula, P. (1994) Expression of HB-GAM (heparin-binding growth-associated molecules) in the pathways of developing axonal processes in vivo and neurite outgrowth in vitro induced by HB-GAM. Brain Res. Dev. Brain Res. 79: 157-176. Roberts, A. (2000) Early functional organization of spinal neurons in developing lower vertebrates. Brain Res. Bull. 53: 585-593. Sattler, M. (2004) Introduction to biomolecular NMR spectroscopy. (http://www. embl-heidelberg.de/nmr/sattler/teaching/teaching_pdf/predoc_intro.pdf) Schäfer, M., Rembold, M., Wittbrodt, J., Schartl, M. and Winkler C. (2005) Medial 77 floor plate formation in zebrafish consists of two phases and requires trunk-derived Midkine-a. Genes & Dev. 19: 897-902. Schäfer, M., Kinzel, D., and Winkler, C. (2007) Discontinuous organization and specification of the lateral floor plate in zebrafish. Dev Biol. 301: 117-29 Sharma, D. and Rajarathnam, K. (2000) 13 C NMR chemical shifts can predict disulfide bond formation. Journal of Biomolecular NMR 18: 165-171. Squire R. L., McConnell, K. S., Zigmond, J. M. Fundamental neuroscience. Edition: 2, Published by Academic Press, 2003 Stoica, G. E., Kuo, A., Aigner, A., Sunitha, I., Souttou, B., Malerczyk, C., Caughey, D. J., Wen, D., Karavanov, A., Riegel, A. T., and Wellstein, A. (2001) Identification of anaplastic lymphoma kinase as a receptor for the growth factor pleiotrophin. J. Biol. Chem. 276: 16772–16779. Stoica, G. E., Kuo, A., Powers, C., Bowden, E. T., Sale, E. B., Riegel, A. T., and Wellstein, A. (2002) Midkine binds to anaplastic lymphoma kinase (ALK) and acts as a growth factor for different cell types. J. Biol. Chem. 277: 35990–35998. Unoki, K., Ohba, N., Arimura, H., Muramatsu, H. and Muramatsu, T. (1994) Rescue of photoreceptors from the damaging effects of constant light by midkine, a retinoic acid responsive gene product. Invest. Opthalmol. Vis. Sci. 35: 4063-4068. Wikipedia: http://en.wikipedia.org/wiki/Main_Page Winkler, C. and Moon, R.T. (2001) Zebrafish mdk2, a novel secreted midkine, participates in posterior neurogenesis. Dev. Biol. 229: 102-118. Winkler, C., Schäfer, M., Duschl, J., Schartl, M. and Volff, J. (2003) Functional divergence of two zebrafish midkine growth factors following fish-specific gene duplication. Genome Res. 13: 1067-1081. Wishart, D. S., Sykes, B. D., and Richards, F. M. (1991) Relationship between nuclear 78 magnetic resonance chemical shift and protein secondary structure. J. Mol. Biol. 222: 1-33. Wisniewski, T., Lalowski, M., Baumann, M., Rauvala, H., Raulo, E., Nolo, R., and Frangione, B. (1996) HB-GAM is a cytokine present in Alzheimer’s and Down’s syndrome lesions. Neuroreport 7: 667–671. Xu, Y., Lin, Z., Chien, H., Yang, D. (2005) A general strategy for the assignment of 13 15 aliphatic side-chain resonances of uniformly C, N-labeled large proteins. J. Am. Chem. Soc. 127: 11920-11921. Xu, Y., Zheng, Y., Fan, J., Yang, D. (2006) A novel strategy for structure determination of large proteins without deuteration. Nature Methods 3: 931-937. Xu, Y., Long, D., and Yang D. (2007) Rapid data collection for protein structure determination by NMR spectroscopy. J. Am. Chem. Soc. 129: 7722-7723. Yang, D., Zheng, Y., Liu, D., Wyss, D. F. (2004) Sequence-specific assignments of methyl groups in high-molecular weight proteins. Journal of the American Chemical Society 126: 3710-3711. Yasuhara, O., Muramatsu, H., Kim, S. U., Muramatsu, T., Maruta, H. and McGeer, P. L. (1993) Midkine, a novel neurotrophic factor, is present in senile plaques of Alzheimer disease. Biochem. Biophys. Res. Commun. 192: 246-251. 79 [...]... Na2HPO4, 3 g KH2PO4, 0.5 g NaCl and 100 μg/ml ampicilin 2.1.2 Preparation of DNA plasmid The gene encoding Mdka protein was obtained from our collaborator, Prof Winkler (Dept of Biological Sciences, NUS), in the template plasmid pCS2 Zmk-2 Oligonucleotide primers were used to amplify the mdka gene The sense primer is 5’-CGCGGATCCAAAAACAAGAAAGAGAAGAA-3’ and antisense primer is 3’-GGGGATATCACTCAGCATAAT-5’... (Li et al., 1990; Rauvala et al., 1994; Kaneda et al., 199 6a) , nerve cell migration (Maeda et al., 1998; Maeda et al., 1999), neurodegenerative diseases (Wisniewski et al., 1996; Yasuhara et al., 1993), and cancer (Kadomatsu et al., 2004) Many of the down-regulation pathways are still unclear, but it is pointed 13 out that the intracellular signaling of N-syndecan family receptors involves cortactin-src... al., 1999) 1.3.2.4 Structure Calculations and Refinement A restrained molecular dynamics/simulated annealing (MD/SA) or torsion angel dynamics (TAD) is chosen for structure calculation, which would use all the 20 available structural parameters mentioned above Structure calculation with NOE assignment is an iterative process The result of an NMR structure calculation is an ensemble of structures, all... development (Kadomatsu et al., 2004) 1.2.2 Protein structures MK precursor protein with an intact signal peptide (22 amino acids) has a molecular mass of 15.6 kDa whereas the matured form of MK is calculated to be 13.2 kDa Mature PTN protein, which shares 45% amino acid sequence identity with MK (Fig1. 6A) , has a molecular weight of 19 kDa The individual domain structures of MK have been solved by NMR (Iwasaki... specifies the fraction of backbone φ,ψ angles which correspond to favored, additionally allowed, generally allowed and disallowed conformations based on statistical analysis of high-resolution crystal structures, is a good quality measure Most φ,ψ angles should be within the allowed regions, while poor quality structures usually have a large number of φ,ψ angles in forbidden regions in the Ramachandran plot... cDNA and its deduced amino acid sequences (B) Hydropathy profile of Mdka (Software from ExPSAy, ProtScale) 6 A B Fig1.4 Sequence of Mdkb (A) The nucleotide sequence of the Mdkb cDNA and its deduced amino acid sequences (B) Hydropathy profile of Mdkb (Software from ExPSAy, ProtScale) 7 Fig1.5 Sequence alignment of MK family (Chang et al., 2004) heparin binding; cluster I a. a for cluster II a. a for heparin... violation, dihedral angle violation, electrostatic potential term, van der Waals force, and the Ramachandran plot The RMSD indicates the coordinate precision of the structure ensembles calculated The smaller the value of RMSD between the calculated conformers and the mean structures (usually less than 1 Å), the better the quality of the structures calculated Distance constrains violation should be within... signaling pathway, Wnt signaling pathway, and BMP signaling pathway In addition, Mdkb controls the cell determination at the neural plate border and affects the formation of neural crest cells and Rohon-Beard sensory neurons (Liedtke et al., 2008) As secreted growth factors, both proteins might bind to the transmembrane heparan sulfate proteoglycans These receptors have the heparin-like carbohydrates;... dorsal mesoderm, cells that adjacent to the notochord during vertebrate organogenesis form transient structures called somite, which define the segmental pattern of the embryo, and subsequently give rise to vertebrae and ribs, dermis of the back, and skeletal muscles of the back, body wall and limbs (Wikipedia) 1.1.2 Characterization of Mdka and Mdkb of zebrafish 3 Fig1.2 Organization of the neural... 1999), anaplastic leukemia kinase (ALK) (Stoica et al., 2001; Stoica et al., 2002), and low density lipoprotein receptor-related protein (LRP) (Herz et al., 2000) N-syndecan belongs to the syndecan family that comprises four transmembrane heparan sulfate proteoglycans Binding of MK and PTN to syndecans is mediated by the heparin sulfate chain, which is the heparin-like domain This domain consists of a variably ... used to amplify the mdka gene The sense primer is 5’-CGCGGATCCAAAAACAAGAAAGAGAAGAA-3’ and antisense primer is 3’-GGGGATATCACTCAGCATAAT-5’ DNA gel electrophoresis (Agarose from Bio-Rad) was subsequently... vertebrae and ribs, dermis of the back, and skeletal muscles of the back, body wall and limbs (Wikipedia) 1.1.2 Characterization of Mdka and Mdkb of zebrafish Fig1.2 Organization of the neural tube... quality structures usually have a large number of φ,ψ angles in forbidden regions in the Ramachandran plot 21 1.3.3 Advantages of structure study by NMR Nuclear Magnetic Resonance, X-ray Crystallography,