Homology Modeling and Analysis of Human Proteins related to Disease Structural Investigations into Shwachman Bodian Diamond Syndrome SBDS using a Bioinformatics Approach 1Babu A. Manjasetty, 2Sunil Kumar, 3Andrew P. Turnbull, 4Niraj Kanti Tripathy 1Proteomics & Bioinformatics Research Group, Research & Industry Incubation Center, Dayananda Sagar Institutions, Bangalore 560 078 Karnataka, INDIA [email protected] 2Institute of Life Sciences, An Autonomous Institute of Department of Biotechnology, Govt. of India, Nalco Square, Bhubaneswar 751023, Orissa, INDIA 3Cancer Research Technology Ltd., Birkbeck College, University of London, London, WC1E 7HX, United Kingdom 4Department of Zoology, Berhampur University, Berhampur 760 007, Orissa, INDIA Abstract The functional correlation of missense mutations which cause disease remains a challenge to understanding the basis of genetic diseases. This is particularly true for proteins related to diseases for which there are no available three dimensional structures. One such disease is Shwachman Diamond syndrome SDS OMIM 260400, a multi system disease arising from loss of functional mutations. The Homo sapiens Shwachman Bodian Diamond Syndrome gene hSBDS is responsible for SDS. hSBDS is expressed in all tissues and encodes a protein of 250 amino acids SwissProt accession code Q9Y3A5. Sequence analysis of disease associated alleles has identified more than 20 different mutations in affected individuals. While a number of these mutations have been described as leading to the loss of protein function due to truncation, translation or surface epitope association, the structural basis for these mutations has yet to be determined due to the lack of a three- dimensional structure for SBDS. Moreover, SBDS is a highly conserved gene of unknown function, however, biochemical studies suggest that SBDS may be involved in RNA metabolism or ribosome assembly. Here, we describe a homology model for hSBDS which is based on the crystal structure of AF0491 from Archaeoglobus fulgidus PDB code 1P9Q that shares 24 percent sequence identity and 49 percent similarity with hSBDS. This model provides insights into how the different mutations are involved in translation, truncation, protein protein interactions, protein stability and function. Keywords: Homology modeling, Shwachman-Bodian-Diamond; Three-dimensional structure. 1. 1INTRODUCTION already been deposited in the protein data bank (PDB)[3] during the second phase (year 2006-2010). Structural proteomics (SP) is a global and rapidly This success is also evident in the increased growing research domain following the recent contribution from other world-wide SG/SP centers in successes in genome sequencing and technical terms of the number of structures being deposited into breakthroughs in macromolecular crystallography (MX) the PDB [4]. and nuclear magnetic resonance (NMR) [1]. MX is a Increased structural coverage of protein families very powerful technique allowing the structures of with high biomedical importance enables the proteins, nucleic acids and their complexes to be exploration of structure-function relationships and aides determined at atomic resolution. Worldwide SP computational modeling. Homology modeling is a initiatives have developed and implemented powerful technique that greatly enhances the value of technologies that facilitate automated, high-throughput experimentally determined structures by using the three-dimensional protein structure determination [2]. structural information from one protein to predict the The main focus of these initiatives is to determine the structures of related proteins. Protein families in which structures of a diverse set of proteins which span as at least one experimental structure has been determined many protein families as possible. With the advent of can act as a template for modeling other family advances in protein structure analysis, experimentally members (provided the level of sequence identity is at determined three-dimensional structures of least 25%). Current methodologies and technologies for representative members of large protein families are homology modeling can produce structural models with being generated at an ever increasing pace. For highly accurate backbones and an intermediate level of example, the protein structure initiative (PSI) produced accuracy for side chains and loops [5,6,7]. Given that over 1,300 structures during its pilot phase (year 2000- the number of experimental structures available is far 2005) and, to date, approximately 3,000 structures have 84 Structural Investigations into SBDS … lower than the huge number of sequenced proteins, high detailed view of the disease-related mutations within quality models can be produced for a proportion of the structural domains of the protein. these proteins of unknown structure and reliable models can be generated for up to 60% of the remaining 2. MATERIALS AND METHODS proteins encoded by a particular genome. Homology Template Search and Sequence Alignment: The modeling plays an important role in filling the huge gap protein sequence of Shwachman-Bodian-Diamond between experimentally determined structures and syndrome (Accession no. AAN77490) of Homo sapiens sequenced proteins, especially for the structures of was retrieved from the NCBI database. A Basic Local proteins related to human diseases which prove difficult Alignment Search Tool (BLAST) [13] search was to determine experimentally. Thus, the vast majority of performed against the Protein Data Bank (PDB) [14] biologists are increasingly reliant on homology models using default parameters to find suitable templates for of proteins of interest. Additionally, the protein homology modeling. Based on the maximum identity structures of disease-related proteins, experimental or with high score and low e-value, 1P9Q at 2.0 Å predicted, are of great research interest to both resolution was selected as the most appropriate experimental and computational biologists. Their high template for homology modeling. The sequence identity value in medicine and human health stems from the fact and similarity are 24% and 49% between the template that they provide informative clues as to how they (1P9Q) and the target sequence respectively. The function in the cell and a molecular snapshot of the Clustal-W (http://www.ebi.ac.uk/clustalw) [15] disease process which aides rational drug design [8]. program was used to produce the sequence alignment This is particularly true for diseases caused by between hSBDS and 1P9Q. alterations of proteins for which no three-dimensional 3D Structure Generation: The academic version of structure is available. With the help of homology MODELER (http//:www.salilab.org/modeler) [16] was modeling, disease-related mutations can also be studied used for three-dimensional structure generation based at the atomic level. To date, thousands of proteins have on the information obtained from the sequence been implicated in various human diseases. Frequently, alignment. Out of 20 models generated by mutations in these proteins have been identified in MODELLER, the one with the best G-score in patients suffering from a particular disease. Mutational PROCHECK [17] and the best VERIFY3D [18] profile data often allows these diseases to be studied at the was subjected to energy minimization. Using the molecular level by generating the protein structure parameters as a distance-dependent dielectric constant either through experimental procedures or by homology  = 1.0 and non-binding cutoff of 14 Å, CHARMM modeling. In this report, a homology model for the Homo [19] force field and CHARMM-all-atom charges, sapiens Shwachman-Bodian-Diamond Syndrome initially a 600 step steepest descent algorithm was used (hSBDS) gene is presented. SBDS is responsible for to remove close Van der Waals contacts, followed by Shwachman-Diamond syndrome (SDS; OMIM an 800 iteration conjugate gradient minimization until 260400), a multi-system disease arising from loss-of- the energy showed stability in sequential repetition. All function mutations [9]. SDS is a rare autosomal hydrogen atoms were included during the calculation. recessive marrow failure syndrome associated with The energy minimization described above was started exocrine pancreatic insufficiency and leukemia with the core main chain and then extended to all the predisposition. Bone marrow failure typically manifests core side chains. All calculations were performed using with neutropenia, but anemia, thrombocytopenia, or the ACCELRYS DS Modeling 2.0 (Accelrys Inc. San aplastic anemia may also develop. Other clinical Diego, CA 92121, USA) software suite. During these features include skeletal, immunologic, hepatic and steps, the quality of the initial model improved cardiac disorders [9-12]. significantly. STRIDE [20], which uses hydrogen bond The SBDS gene is ubiquitously expressed and energy and main chain dihedral angles to recognize encodes a protein of 250 residues (SwissProt accession helix, coils and strands, was subsequently used to code Q9Y3A5). The function of this protein is unknown predict the secondary structure composition of the and it has no primary sequence similarity to any other modeled SBDS. The weighted root mean square protein or structural domain that could indicate a deviation (RMSD) of the modeled protein was possible function. However, biochemical studies calculated using the combinatorial extension (CE) suggest that SBDS protein may be involved in RNA algorithm [21]. The modeled structure was then metabolism or ribosome assembly. Many of the superimposed on the crystal template without altering disease-associated mutations and truncations have also the coordinate systems of atomic position in the been identified. The model presented here may serve as template. The RMSD for the superimposition is 2.1Å. a valuable reference to improve the understanding of The residue profiles of the three-dimensional models the molecular basis of this disease and to provide a were further checked using VERIFY3D [18]. InterJRI Science and Technology, Vol. 1, Issue 2, July 2009 8 5 PROCHECK [17] analysis was performed in order to comprises -helices arranged in helix-turn-helix motifs assess the stereo-chemical properties of the three- suitable for RNA or DNA binding [24]. The C-terminal dimensional models and Ramachandran plots were domain resembles a ferrodoxin-like fold which calculated. No residues reside in the disallowed regions commonly exists in a large number of proteins [25]. of the Ramachandran plot. The structural superposition of the C atom positions of α the hSBDS homology model and SfSBDS (Figure 4) gives an RMSD of 2.1Å with a significant Z-score of 3. RESULTS AND DISCUSSION 6.8. Structural differences stretching throughout the Homology Model model represent conformational flexibility between the The X-ray structure of an SBDS homologue in homologues. Archaeoglobus fulgidus, AF0491 [22], was used as the structural template (PDB 1P9Q) for the homology Implications for SDS modeling protocol. The sequence alignment between The wide occurrence of SBDS homologues human SBDS (hSBDS) and Archaeoglobus fulgidus provides opportunities to glean insights into the AF0491 (AfSBDS) is shown in Figure 1. All conserved molecular basis of this disease [26]. The structure of residues are coloured grey. The alignment is AfSBDS was recently determined and indicated that the characterized by some insertions and deletions in loop tertiary structure of this protein is arranged in three regions. Since the first 9 N-terminal residues and the domains, comprising a novel N-terminal FYSH domain last 9 C-terminal residues (residues 241–250) in SBDS followed by two domains that are highly conserved in do not have corresponding residues in 1P9Q, the other structural domains [22,23,26]. On the other hand, modeling was carried out from residues 10-240, biophysical and biochemical functional data from followed by a rigorous refinement of the model by model organisms clearly provides clues to functional means of Energy Minimization (EM - see the Materials conservation in family members and a phylogeny study and Methods section for the protocol used). The three- reveals that the characteristics of a functionally dimensional conformation of the predicted model of coherent group of SBDS proteins is highly suggestive SBDS is presented in Figure 2. The stereo-chemical of a role in translation and/or RNA metabolism [27]. quality of the modeled SBDS was evaluated using The structural and functional analyses of SBDS PROCHECK. The main chain phi/psi angles for 94.9% orthologues, coupled to its localization within the of residues fall in the most favored regions, 4.7% of nucleus, suggest an involvement in RNA processing residues lie in the additionally allowed regions and [22,28]. 0.5% fall in the generously allowed regions. No More than 20 mutations have been identified in residues are present in unfavourable regions of the hSBDS which are highlighted in reverse type in Figure Ramachandran plot (Figure 3). The overall 1. The functional consequences of SBDS mutations PROCHECK G-factor for the homology modeled have not yet been characterized, however, mutations in structure was – 0.05. Thus, statistical analysis suggests the affected patients result from gene conversion that the backbone conformation of the predicted model between SBDS and an SBDSP pseudogene [29]. Most of SBDS is almost as good as that of the template. mutations are present in the N-terminal FYSH of VERIFY3D provided the plot which indicates that the hSBDS suggesting that this domain plays an important final protein model is correct. The secondary structure physiological role. The most important point mutation, elements of SBDS were predicted using the program C31W, is located within the hydrophobic core of the STRIDE, which uses hydrogen bond energy and main FYSH domain (Figure 5). Strikingly, C31 has the chain dihedral angles to recognize helix, coils and potential to form a disulfide bond with another mutated strands which are well defined in the modeled SBDS residue, C84, to enhance structural stability. The structure. structural consequences of the dual C31 and C84 mutations will be to destabilize the protein which will Overall Structure adversely affect protein function. Therefore, these The core features of the three dimensional SBDS mutations are responsible for protein stability and will homolog (AfSBDS) structure are retained in the hSBDS eventually cause SDS. model (Figure 2)[22, 23]. The subunit structure of A detailed understanding of the functions of hSBDS is hSBDS is organized into three domains: an N-terminal, not known, however, based on the structural features central and C-terminal domain. The N-terminal domain identified in the modeled structure, one can speculate was first identified in the AfSBDS structure and potential roles. In this regard, residues located on the displayed no structural similarity to other protein folds surface of SBDS proteins are likely to be important for in the PDB. Therefore, this domain represented a novel interacting with partner proteins and any mutation of fold and was termed the Fungal, Yhr087wp, these residues will affect the surface electrostatic Shwachman domain (FYSH). The central domain potential. Some of the missense mutations such as 86 Structural Investigations into SBDS … E44G, K118N, E28A and R100A, which are implicated [8] Ye Y, Li Z & Godzik A (2006) Modeling and in SDS disease, are localized on the surface of hSBDS analyzing three-dimensional structures of human and have been mapped onto the modeled structure disease proteins. Pacific Symposium on Bio- computing, 11:439-450. (Figure 6). Based on proteomics data, an SBDS- [9] Ball HL, Zhang B, Riches J, Gandhi R, Li J, interactome has been developed and SBDS binding Rommens JM & Myers JS (2009) SBDS is a multi- partners with diverse molecular functions, notably functional protein implicated in cellular stress components of the large ribosomal subunit and proteins responses. Human Molecular Genetics, July 14 [Epub involved in DNA metabolism, have been reported [30]. ahead of print]. The in-depth comparison of experimental data with the [10] Shwachman H, Diamond LK, Oski FA, Khaw KT. structural data will provide further insights into SDS. (1964) The syndrome of pancreatic insufficiency and bone marrow dysfunction. J. Pediatr., 65: 645-663. [11] Bodian M, Sheldon W, Lightwood R. (1964) 4. CONCLUSION Congenital hypoplasia of the exocrine pancreas. Acta To date, thousands of genes (proteins) have been Paediatr., 53: 282-293. identified to be associated with various human diseases. [12] Burroughs L, Woolfrey A & Shimamura A. (2009) Shwachman-Diamond Syndrome: A Review of the Frequently, mutations in these proteins have been Clinical Presentation, Molecular Pathogenesis, identified in patients suffering from a particular disease. Diagnosis, and Treatment. Hematology/Oncology Homology modeling is a valuable tool to help to Clinics of North America, 23(2):233-248. understand how these mutations cause disease. Here, a [13] Altshul SF, Gish W, Miller W, Myers EW, Lipman DJ homology model of hSBDS has been generated using a (1990) Basic Local Alignment Search Tool. J. Mol. bioinformatics approach. Our analysis provides insights Biol., 215:403-410. into the distribution of the mutations implicated in SDS [14] Bermen HM, Westbrook J, Feng Z, Gilliland G, Bhat and highlights how they can influence protein stability TN, Weissig H, Shindyalov IN & Bourne PE (2000) and alter the surface electrostatic potential. We plan to The Protein Data Bank. Nucleic Acids Research, 28:235-242. extend the current study to the prediction of the [15] Thompson JD, Higgins DG, Gibson TJ (1994) structures of complexes of disease proteins. The CLUSTAL W: Improving the sensitivity of resulting complex models would provide an important progressive multiple sequence alignment through resource for studying the mechanism of disease. sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res., 22:4673-4680. REFERENCES [16] Sali A & Blundell TL (1993) Comparative protein [1] Norvel JC & Berg JM (2007) Update on the Protein modeling by satisfaction of spatial restraints. J. Mol. Structure Initiative, Structure, 15(12): 1519.1522. Biol., 234:779-815. [2] Manjasetty BA, Turnbull AP, Panjikar S, Büssow K & [17] Laskoswki RA, MacArthur MW, Moss DS, Thornton Chance MR (2008) Automated technologies and novel JM (1993) PROCHECK: a program to check the techniques to accelerate protein crystallography for sterochemical quality of protein structures. J. Appl. structural genomics, Proteomics, 8(4):612-625. Cryst., 26:283-291. [3] Berman H, Henrick K, Nakamura H & Markley JL [18] Eisenberg D, Luthy R & Bowie JU (1997) (2007) The worldwide Protein Data Bank (wwPDB): VERIFY3D: Assessment of protein models with ensuring a single, uniform archive of PDB data. three-dimensional profiles. Methods Enzymol., Nucleic Acids Res., 35:D301–D303. 277:396-404. [4] Nair R, Liu R, Soong TT, Acton B, Everett JK, [19] Brooks BR, Bruccoleri RE, Olafson BD, States DJ, Kouranov A, Fiser A, Godzik A, Jaroszewski L, Swaminathan S and Karplus M (1993) CHARMm: A Orengo C, Montelione GT & Rost B (2009) Structural program for macromolecular energy minimization and genomics is the largest contributor of novel structural dynamics calculations. J. Comp. Chem., 4:187-217. leverage. J.Struct.Funct.Genomics, 10(2):181-191. [20] Frishman D & Argos P (1995) Knowledge-based [5] Bhattacharya A, Wunderlich Z, Monleon D, Tejero R protein secondary structure assignment. Proteins, & Montelione T (2007) Assessing model accuracy 23:566-579. using the homology modeling automatically software. [21] Shindyalov IN & Bourne PE (1998) Protein structure Proteins: Structure, Function, and Bioinformatics, alignment by incremental combinatorial extension 70(1):105-118. (CE) of the optimal path. Protein Engineering, [6] Wang Q, Canutescu AA & Dunbrack RL Jr. (2008) 11:739-747. SCWRL and MoIIDE: Computer programs for side [22] Savchenko A, Krogan N, Cort JR, Evdokimova E, chain conformation prediction and homology Lew JM, Yee AA, Sánchez-Pulido L, Andrade MA, modeling. Nature Protocols, 3(12):1832-1847. Bochkarev A, Watson JD, Kennedy MA, Greenblatt J, [7] Eswar N, Eramian D, Webb B, Shen M & Sali A Hughes T, Arrowsmith CH, Rommens JM, & (2008) Protein structure modeling with MODELLER. Edwards AM (2005) The The Shwachman-Bodian- Methods in molecular biology: Structural proteomics, Diamond Syndrome Protein Family Is Involved in 426:145-159. InterJRI Science and Technology, Vol. 1, Issue 2, July 2009 8 7 RNA Metabolism. The Journal of Biological Dr. Babu A Manjasetty is Professor & Chief Scientist at Chemistry, 280:19213-19220. Research & Industry Incubation Center (RIIC), Dayananda [23] Shammas C, Menne TF, Hilcenko C, Michell SR, Sagar Institutions (DSI), Bangalore. He leads the Proteomics Goyenechea B, Boocock GRB, Durie PR, Rommens & Bioinformatics Research Group at RIIC. Prior to joining JM & Warren AJ (2005) Structural & mutational DSI, he worked at Case Proteomics Center, Brookhaven analysis of the SBDS Protein Family. Journal of National Laboratory, New York, USA & as an Alexander von Biological Chemistry, 280(19):19221-19229. Humboldt (AvH) Fellow, Germany at Max-Delbrück Center [24] Luscombe NM, Austin SE, Berman HM & Thornton for Molecular Medicine, Berlin, Germany. He obtained his JM (2000) An overview of the structures of protein- Ph.D in Physics from the University of Mysore under DNA complexes. Genome Biol. 1, 1–3. University Grant Commission (UGC) fellowship. [25] Murzin, A. G., Brenner, S. E., Hubbard, T., and Chothia, C. (1995) SCOP: a structural classification of Sunil Kumar is Information Officer at Institute of Life proteins database for the investigation of sequences Sciences, An autonomous Institute of Department of and structures. J. Mol. Biol., 247, 536–540. Biotechnology, Govt. of India, Bhubaneswar. He is looking [26] Leong CN, Waterman DG, Koonin EV, Walters AD, after the Bioinformatics Centre. He holds masters degree in Chong JPJ, Isupov MN, Lebedev AA, Bunka DHJ, Biotechnology and Advanced Diploma in Bioinformatics. He Stockley PG, Ortiz-Lombardia M & Antson AA is pursuing Ph.D in Bioinformatics from Berhampur (2009) Conformational Flexibility and molecular University, Berhampur, India. His research interest is interactions of an archaeal homologue of the Structural Bioinformatics and Sequence analysis. Swachman-Bodian-Diamond syndrome protein. BMC Structural Biology, 9:32. Dr Andrew Turnbull is Principal Scientist of structural [27] Boocock GR, Marit MR & Rommens JM (2006) biology at Cancer Research Technology Ltd., London, United Phylogeny, sequence conservation and functional Kingdom. He set up and leads the new crystallography unit complementation of the SBDS protein family. responsible for determining the structures of pharmaceutically Genomics, 87:758-771. important protein-ligand complexes to aid CRT’s small [28] Austin KM, Leary RJ, Shimamura A. (2005) The molecule drug discovery programs. Dr Turnbull obtained his Shwachman-Diamond SBDS protein localizes to the PhD in Biochemistry from the University of Sheffield with nucleolus. Blood, 106:1253-1258. work on amino acid dehydrogenases. [29] Boocock GR, Morrison JA & Popovic M, et al. (2003) Mutations in SBDS are associated with Shwachman - Prof. Niraj Kanti Tripathy is Professor in Zoology at Diamond syndrome. Nature Genetics, 33:97-101. Berhampur University, Berhampur, Orissa (India) He is [30] Ball HL, Zhang B, Riches J, Gandhi R, Rommens JM having 35 years of research and Teaching experience. He & Myers JS (2009) SBDS is a multi-functional protein worked in Europe for 5 years. implicated in cellular stress responses. Hum. Mol. Genet., 91(4):475-83. BIOGRAPHY hSBDS MSIFTPTNQI RLTNVAVVRM KRAGKRFEIA CYKNKVVGWR SGVEKDLDEV 50 S fSBDS --------MV SLDKAVIARL RKGGEEFEVL VDPYLARDLK EGKEVNFEDL 42 hSBDS LQTHSVFVNV SKGQVAKKED LISAFGTDDQ TEICKQILTK GEVQVSDKER 100 S fSBDS LAAEEVFKDA KKGERASVDE LRKIFGTDDV FEIARKIILE GEVQITAEQR 92 hSBDS HTQLEQMFRD IATIVADKCV NPETKRPYTV ILIERAMKDI HYSVKTNKST 150 S fSBDS REMLEAKRKQ IINFISRNTI DPRTNAPHPP SRIERALEEA KVHIDIFKSV 142 hSBDS KQQALEVIKQ LKEKMKIERA HMRLRFILPV NEGKKLKEKL KPLIKVIESE 200 S fSBDS EAQVKDIVKA LKPILPLKFE EMEIAIKIPP EHTGRAISAL YNFGGVTR-E 191 hSBDS DYGQQLEIVC LIDP--GCFR EIDELIKKET KGKGSLEVLN LKDVEEGDEK FE 250 S fSBDS EWQRDGSWIC VMRIPSGMYG DLMDLLGKVA KGEALTKVLR RIG------- -- 234 88 Structural Investigations into SBDS … Figure 1. Sequence Alignment between Human SBDS (hSBDS) and AfSBDS. Identical residues are highlighted in grey (total of 24% identity). The positions of the human SBDS-associated mutations are highlighted in reverse type. Figure 2. Schematic representation of the modeled SBDS structure. Figure prepared using the program PyMOL. InterJRI Science and Technology, Vol. 1, Issue 2, July 2009 8 9 Figure 3. Ramachandran plot for the hSBDS model. Figure 4. C trace of the hSBDS model (dark blue) superposed on AfSBDS (light blue). Figure prepared using the program PyMOL. 90 Structural Investigations into SBDS … Figure 5. The key point mutation, C31W, is located within the hydrophobic core of the FYSH domain and resides within close proximity to C84 Figure 6. Some of the surface localized mutations involved in interactions highlighted in cyan