Characterization of a Novel Binding Protein for Fortilin/TCTP — Component of a Defense Mechanism against Viral Infection in Penaeus monodon

The Fortilin (also known as TCTP) in Penaeus monodon (PmFortilin) and Fortilin Binding Protein 1 (FBP1) have recently been shown to interact and to offer protection against the widespread White Spot Syndrome Virus infection. However, the mechanism is yet unknown. We investigated this interaction in detail by a number of in silico and in vitro analyses, including prediction of a binding site between PmFortilin/FBP1 and docking simulations. The basis of the modeling analyses was well-conserved PmFortilin orthologs, containing a Ca2+-binding domain at residues 76–110 representing a section of the helical domain, the translationally controlled tumor protein signature 1 and 2 (TCTP_1, TCTP_2) at residues 45–55 and 123–145, respectively. We found the pairs Cys59 and Cys76 formed a disulfide bond in the C-terminus of FBP1, which is a common structural feature in many exported proteins and the “x–G–K–K” pattern of the amidation site at the end of the C-terminus. This coincided with our previous work, where we found the “x–P–P–x” patterns of an antiviral peptide also to be located in the C-terminus of FBP1. The combined bioinformatics and in vitro results indicate that FBP1 is a transmembrane protein and FBP1 interact with N-terminal region of PmFortilin.


Introduction
The White Spot Syndrome Virus (WSSV) is a common pathogen that causes significant production losses in the shrimp industry around the world. The virus has a wide host range in a variety of crustaceans, both freshwater and seawater species, including penaied shrimp, crabs, and crayfish. A WSSV infection typically causes lethargic behavior, cessation of feeding, a pink to reddish-brown discoloration, and white inclusion of calcium salts embedded in the cuticle especially inside the surface of the carapace. The moribund shrimp swim near the surface at the edge of ponds. The populations of shrimp showing these signs have high mortality rates with cumulative mortalities reaching up to 100% within three to ten days. The transmission of disease is typically via cannibalism of the sick or dying prawns and directly through contaminated water. The virus can persist and retain infectivity in seawater for four to seven days. The histopathology of WSSV infected shrimp shows severe nuclear hypertrophy, chromatin margination, and eosinophilic to large basophilic intranuclear inclusions with variable multifocal necrosis in most tissues of ectodermal and mesodermal origin [1]. The WSSV also severely damages the stomach, gills, antennal gland, heart, and eyes. The infection can be classified into two types: Type I is an acute infection that causes high mortality within two weeks in species such as Penaeus monodon, Penaeus indicus, and Penaeus penicillatus. Type II is latent.
Based on the morphology and the genomic structures, WSSV has been classified to the genus Whispovirus and the family Nimaviridae, The virions are enveloped, ovoid to bacilliform 276626 nm long with a diameter of 12169 nm, containing double-stranded circular DNA with the size ranging from 290 kbp to 305 kbp, which covers approximately 185 open reading frames, ORFs [1][2][3]. The virion envelope contains two major proteins, VP28 and VP19 of estimated sizes 28 and 19 kDa, respectively. The nucleocapsid consists of three major proteins VP26, VP24 and VP15 of estimated sizes 26, 24, and 15 kDa, respectively [4,5]. A previous study indicates that the gene identified as ORF390 has anti-apoptotic properties, contains two putative caspase9 cleavage sites and a caspase3 cleavage site [6]. The WSSV can replicate and propagate in a number of cell types, such as haematopoietic cell cultures of freshwater crayfish, Pacifastacus leniusculus [7], hepatopancreas cells of Penaeus monodon [8] and ovarian cultures of Marsupenaeus japonicas [9]. The replication rate of the WSSV inside the nucleus of the haematopoietic cell cultures of freshwater crayfish is known to increase with increasing temperature [7].
Over the years, a lot of effort has been put into attempts to protect against WSSV infection in shrimp, such as the usage of random library peptides [10], RNA interference, siRNA, VP15 and VP28 genes and long dsRNA -shown to induce both a sequence-specific and sequence-independent antiviral immunity [11], injection of VP28-dsRNA alone [12], vaccination with viral structural proteins and a monoclonal antibody targeting VP28 [13].
Fortilin (also known as Translationally controlled tumor protein, TCTP) has attracted a lot of attention due to its implication in a variety of functions, i.e., control of cell cycle, cell growth and cell division [14,15], microtubule stabilization [16], and as a growth factor for B-cells [17]. Fortilin is present in both the nucleus and cytosol [18], is inducible by serum stimulation [19] and heavy metals [20]. Fortilin has a potent anti-apoptotic function [18,[21][22][23] through binding interactions with MCL1 (Myeloid cell leukemia 1) [24] and BCL-XL (B-cell leukemia XL) [25], belonging to the anti-apoptosis BCL-2 protein family. Fortilin can also bind to Ca 2+ [26][27][28][29] and thus prevent cytosolic Ca 2+ levels from increasing and activating Ca 2+ -dependent apoptosis pathways [28]. Fortilin is thought to be a modulator of GTPase activity, acting as a molecular switch for a vast number of cellular processes in all eukaryotes [30]. Fortilin can bind to Na, K 2 ATPase and regulates the function of Na, K 2 ATPase in cooperation with Nexin6 (SNX6) [31]. The Fortilin protein is also known to be involved in growth and development, it is encoded by abundant mRNA species, and was initially characterized in mouse Ehrlich ascites tumor cells and erythroleukemia cells [15,32]. Despite various functions, its wide range of distribution, and high level of conservation among many organisms, the distinct functions of Fortilin still remain unclear [16,20,[26][27][28]33].
Our group has previously identified Fortilin in Penaeus monodon, named PmFortilin (Accession No: AY186580.1) and shown it to have anti-apoptotic properties. This protein is well conserved, contains a Ca 2+ -binding domain, TCTP signatures similar to that of the Fortilin or TCTP genes reported in plants and animals [29]. We found it to be present at high levels during an onset of viral infection in P. monodon and based on a yeast two-hybrid screening assay and GST-pulldown, we further found Fortilin to interact with a previously unknown protein, named Fortilin Binding Protein 1 (FBP1) (Accession No: EU435133.1) [34]. This is a small peptide of 93 amino acids long with a molecular weight of 11 kDa, expressed solely in hemocytes and no homologs have yet been found. It was also of interest that we found transcripts of FBP1 to be up-regulated during WSSV infection with the highest level occurring at 48 h post-infection [34]. In this manuscript we report investigations into the interaction of PmFortilin with FBP1 and its function. From our results we suggest that FBP1 is a transmembrane protein and its likely function is to facilitate transport of PmFortilin across the cell membrane.

Structural analysis of PmFortilin and FBP1
Experimentally determined structures are not available for either PmFortilin or FBP1. For this reason, we have predicted the structures by homology modeling. The best scoring 3-D structure of PmFortilin contains three sheets, six beta hairpins, two beta bulges, three helices and one helix-helix interaction. The model displays one helical domain, stretching from residue 77 to 126 and one right handed hook (RHH) of the disulphide type ( Figure 1). The model of FBP1 contains two helix domains, 16 beta turns and 12 gamma turns. The two helix domains are located at residues 8 to 21 and residues 23 to 25, forming disulphide bonds between the Cys59 and Cys76 (Figure 2).
To assess the accuracy of the structure predictions, we performed a ramachandran analysis, for PmFortilin, 108 (72.50%) residues were in the most favored regions, additional allowed regions contained 33 (22.10%) residues, generously allowed regions contained seven (4.70%) residues and disallowed regions contained only a single (0.70%) residue ( Figure S1). For the predicted FBP1 structure, 31 (49.20%) residues were plotted in the most favored regions, additional allowed regions contained 21 (33.30%) residues, generously allowed regions contained six (9.50%) residues and five (7.90%) residues were located in the disallowed regions ( Figure S2).
PmFortilin is highly conserved; the chain A has the solution structure of human translationally controlled tumor protein, the chain A with the translationally controlled tumor associated protein p23fyp from Schizosaccharomyces pombe, the chains A, B, C, and D with the crystal structure of human translationally controlled tumor associated protein (hTCTP) mutant E12V, the chain A with the crystal structure of translationally controlled tumor associated protein from Plasmodium knowlesi and with the Fortilin structure of Drosophila melanogaster (PDB ID: 2HR9, 1H6Q, 3EBM, 1TXJ, and 1YZ1 respectively). A further comparison of multiple sequences reveals that PmFortilin contains conserved regions that correspond to the helical domain in the predicted 3-D structure. The amino acid residues 76-110, part of the helical domain of PmFortilin are contained within a Ca 2+ -binding domain, forming a helical domain between the H2-helix (residues 81-99) and the H3-helix (residues 106-122) producing an EF-hand structure ( Figure 1). Importantly, a small section at the N-terminus (residues 1-10) contains protein transduction domains (PTDs), 1-MKVFKDMLTG-10, allowing for the delivery of active molecules into the cells through the lipid bilayer [35][36][37]. PmFortilin contains two signature patterns. The  (Table S1).
For FBP1, we were unable to find homologous sequences in the existing public databases. However, we used Motif Scan (http:// myhits.isb-sib.ch/cgi-bin/motif_scan) to detect signatures. The search revealed some interesting signatures: Amino acid residues 4-6 at the N-terminal represented by ''[ST]-x-[RK]'' pattern at protein kinase C phosphorylation site and by ''x-G-[RK]-[RK]'', which is a known amidation pattern, were found to be located at the residues 90-93 at the C-terminal end of FBP1 (Table S1). Two segments of compositionally biased regions or Low Complexity Regions (LCRs), were located at the amino acid residues 27-48 and 59-87 using SMART server (http://smart.embl-heidelberg. de) (Table S1 and S2), previously implied in the context of proteinprotein interactions [38][39][40]. The residues 1-24 are likely to be a signal peptide (scored 99.00%) with a cleavage site at Ala24 and Thr25 (scored 87.30%), determined by the Signal P 3.0 server (http://www.cbs.dtu.dk/services/SignalP). It was of some interest that FBP1 contains a number of ''x-P-P-x'' signature sequences found in antiviral peptides. The longest such pattern is located at residues 59-76 [34]. In addition, we performed combinatorial structure analyses using GANGSTA and GANGSTA+ servers (http://agknapp.chemie.fu-berlin.de/gplus) (Table S3).

Functional analysis of Fortilin/TCTP binding protein (FBP1)
The predicted 3-D model of FBP1 contains two alpha helix domains. Such segments are common protein structure elements that cross biological membranes. To further assess the topology and possible localization of FBP1, we utilized HMMTOP (http:// www.enzim.hu/hmmtop) and TMHMM2 servers (http://www. cbs.dtu.dk/services/TMHMM). According to these predictions, the helix residues 7-26 are transmembrane segments (Table S1), the residues 1-6 that correspond to the N-terminal phosphorylation site found by Motif Scan are located in the cytosol and the residues 27-93, containing LCRs and ''x-P-P-x'' signature sequences, are extracellular ( Figure 2B).
To find further support to the hypothesis that the FBP1 is a transmembrane protein, we transfected Sf9 cells with a plasmid containing an FBP1-GFP construct. An empty phMGFP plasmid and mock-transfected cells were used as negative controls.
Confocal laser scanning microscopy was used to observe the fluorescence patterns in the cells 48 h after transfection.
The signals in the Sf9 cells are clearly located at the plasma membrane, whereas the signals in the cells transfected only with GFP are dispersed. The mock-transfected cells show no GFP signal ( Figure 3). In addition plasma membranes stained with 1,19-Dioctadecyl-3,3,39,39-Tetramethylindocarbocyanine Perchlorate (DiI) confirmed the localization of FBP1 ( Figure 3, lower panel).

The Co-localization of PmFortilin and FBP1 by immunocytochemistry
Immunofluorescence studies were performed to investigate the localization of PmFortilin in the presence of FBP1. A confocal laser scanning microscopy was used to observe the fluorescence patterns in the cells 48 h after transfection. We observed that FBP1-GFP was located at the plasma membrane ( Figure 4A) while the staining with the anti-PmFortilin antibody appeared in the whole cell ( Figure 4B). According to previous reports, TCTP/Fortilin was mainly localized in the cytosol and also in the nucleus [14,18]. In this experiment, the co-localization of PmFortilin and FBP1 was detected as a yellow color only at the plasma membrane ( Figure 4C).

PmFortilin/FBP1 docking simulations and interaction analysis
To further analyze the PmFortilin and FBP1 interaction in detail, we predicted the PmFortilin/FBP1 binding site residues and performed docking simulations utilizing the ClusPro 2.0 server (http://cluspro.bu.edu), including the PIPER docking software based on the Fast Fourier Transform (FTT) method [41], using FBP1 as a receptor protein and PmFortilin as a ligand. The models were predicted using four separate modes: Balance, Electrostatic, Hydrophobic and VdW+Elec mode. The short interacting elements were extracted from the best scoring interaction complex and collected to produce a refined model ( Figure 5). The best, i.e. the lowest docking score of the balance mode is 2993. 40 (Table S7).
According to the docking simulations, the flexible region of PmFortilin binds to the C-terminus of FBP1. These interactions are represented in two major conformations A and B ( Figure 5A). The balance mode simulation resulted in 25 separate models. The conformation A is favored over conformation B with frequencies 18/25 (72.00%) and 7/25 (28.00%) respectively ( Figure 5B). Four of the top five ranking conformations are described in Table S4. The electrostatic mode yields the frequencies 17/23 (73.91%) and 6/23 (26.09%) for the conformations A and B respectively ( Figure 5C), the hydrophobic mode yields the frequency 6/23 (26.09%) for conformation A and 17/23 (73.91%) for conformation B ( Figure 5D) and the VdW+Elec mode yields the frequencies 22/23 (95.65%) and 1/23 (4.35%) for conformations A and B, respectively ( Figure 5E). The frequencies seem to be in good agreement, except for the hydrophobic mode, where the conformation B is the most favored.
These interactions coincide with three domains: (1)

Yeast two-hybrid assay -N-terminal region of PmFortilin is responsible for interaction with FBP1
According to the docking simulations, the PmFortilin binds FBP1 at residues 37-63 located on the flexible loop of the PmFortilin Nterminus and contacts FBP1 at residues 77-88. To confirm the docking results, we sub-cloned PmFortilin in three separate fragments: PmFT1 (residues 1-70), PmFT2 (residues 71-120) and PmFT3 (residues 121-168). The interaction of FBP1 with each of the three fragments was tested by the yeast two-hybrid assay. The yeast growth in YPDA was monitored to ensure that there were no growth defects in all recombinant clones ( Figure 8A). After co-transformation, the clones that harbored the binding proteins were selected from a selective medium, at this step there were no clones found from the yeast harboring BD-PmFT2+ AD-FBP1 (Table 1). This indicated that these two proteins did not interact with each other. The bgalactosidase filter assay was then used to test the yeast harboring BD-PmFT1+AD-FBP1 and that harboring BD-PmFT3+AD-FBP1. The result confirmed an interaction only for BD-PmFT1+AD-FBP1, thereby supporting the docking results ( Figure 8B). Further investigations are required to understand the binding of PmFortilin and FBP1.

Discussion
In previous report FBP1 was identified and cloned from the library by yeast two hybrid screening using PmFortilin as the bait.  The interaction of these two proteins was confirmed by GST pull down [34]. Both PmFortilin and FBP1 transcripts have been show to be up regulated during a viral infection, thus implying a role in a defense mechanism against viruses. Here we try to accelerate a way to design proper biochemical experiments for the functional analysis of a novel gene product, FBP1 by using in silico methods. We have identified FBP1 as a membrane binding protein and analyzed its interaction with PmFortilin. The prediction was supported by the co-localization of Fortilin and FBP1 at the same position on plasma membrane. The contact sites of PmFortilin are is to protect the C-terminus from being modified by proteases. Further investigation by using site directed mutagenesis at the predicted locations are required to support this conclusion.
In addition to the above analysis results, given that the signal peptides and transmembrane helixes contain hydrophobic amino acids, the existence of an N-terminal phosphorylation site and the cleavage site at Ala24 and Thr25, the results indicate that the function of FBP1 is to facilitate the transport of PmFortilin through the cell membrane. Additional docking simulations further indicate that PmFortilin can bind two FBP1s simultaneously and thus may be active as a trimer. However, further investigations are needed to decipher the detailed mechanism of transport, phosphorylation and cleavage.

Molecular modeling
The homology modeling for the 2-D and 3-D structures were performed with SWISS-MODEL [42] and I-TASSER [43]. The best template was automatically selected based on multiplethreading simulations and used for the structural model. Each of the predicted models was validated based on the best scoring model given by the PROCHECK software [44]. These were evaluated by ramachandran plot analysis utilizing ProFunc [45] (http://www.ebi.ac.uk/thornton-srv/databases/profunc) and PROCHECK software. The 3-D template structures were brought from the PDB database (http://swissmodel.expasy.org).

Functional and molecular homology analysis
The GANGSTA and GANGSTA+ [46] were used for comparisons of the 3-D structures of PmFortilin and FBP1. Motif Scan [47] and SMART [48] were used for scanning signature domains of the PmFortilin and FBP1 with the default parameters, including outlier homologs and homologs of known structures, Pfam domains, signal peptides, internal repeats and intrinsic protein disorder databases.
The ProFunc server was used to identify the potential functions within the 3-D structures. The HMMTOP [49] with the 'reliable' mode and Baum-Welch iteration and TMHMM servers [50] were utilized to predict helical transmembrane segments and the topology with default parameters. Signal peptides and cleavage site patterns were predicted using the SignalP server [51], trained on eukaryotes. Multiple sequence alignments were constructed by using Clustal [52].

Molecular docking simulations and analysis of the interaction model
The molecular interaction and docking simulations were performed using the ClusPro 2.0 server [53]. The interaction   Construction of recombinant FBP1-GFP and confocal microscopy FBP1 was cloned into the phMGFP vector (Promega, catalogue no. E6421) to generate an FBP1-GFP fusion protein. PCR was performed with forward primer FBP1-F: 59-GCTAGCAT-GAAGTTCTCATGT; containing NheI site and reverse primer FBP1-R: 59-CCCGGGCTTCTTGCCCTTACT; containing the XmaI site without the stop codon at its C-terminal. The fragments were blunt-end inserted into phMGFP amplified by transformation into the high efficiency Escherichia coli Top10F'competent (Invitrogen) by the heat shock method. The plasmids were extracted and purified with QIAprep Spin Miniprep Kit (Qiagen). The Sf9 cells were seeded at a density of 5610 5 cells onto the cover slips in Sf9-S2 medium (PPA). The phMGFP and phMGFP-FBP1 plasmids were transfected into Sf9 cells using Transfast TM Transfection reagent (Invitrogen) according to the product instructions and cultured in a 28uC incubator. After 48 h of transfection, the cells were washed by PBS twice. The plasma membrane was stained with 2 mM DiI (Invitrogen) for 15 min at 4uC and the Sf9 cells were observed directly with the confocal laser scanning microscope.

Localization of PmFortilin in the presence of FBP1 by immunocytochemist
Sf9 cells, at a density of 5610 6 cells, were grown on glass coverslips in Sf9-S2 medium (PPA), then cotransfected with phMGFP-FBP1 and pCDNA-PmFortilin using Transfast TM Transfection reagent (Invitrogen) according to the product instructions. Cells were cultured in a 28uC incubator. 48 h posttransfection, the cells were washed twice with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde for 10 min followed by washing with 0.1% Triton x-100 in PBS for 3 min at 4uC. Fixed cells were washed again with PBS before blocking with 5% skim milk at 4uC for 1 h. After incubating, cells were washed and incubated with specific rabbit anti-PmFortilin antibody (dilution of 1:500 in PBS) at 4uC for 16-18 h. The cells were washed 3 times with PBST and incubated with Alexa flour 647conjugated goat anti-rabbit IgG antibody (Invitrogen, 1:1,000 in PBS) at room temperature for 2 h. After washing 3 times with PBST, the cover glasses were mounted with 10% glycerol in PBS. The fluorescence signal was detected using a confocal laser scanning microscope (Olympus FV300).

Yeast two-hybrid assay
The yeast two-hybrid screens were performed with the Clontech Matchmaker GAL4 Two-Hybrid System 3 (Clontech Laboratories, Inc., Mountain View, CA, USA). PmFortilin was divided into 3 fragments: PmFT1 (residues 1-70), PmFT2 (residues 71-120) and PmFT3 (residues 121-168). Plasmids pGBKT7 (named BD vectors) were used for construction of BD-PmFortilin. The plasmids were constructed by inserting into EcoRI and SalI sites on a BD-bait vector. Each fragment was amplified by PCR using the following primers; All plasmids constructed were verified by sequencing. Competent yeast cells, Saccharomyces cerevisiae strain AH109 were co-transformed either with BD-PmFT1 and AD-FBP1, BD-PmFT2 and AD-FBP1, BD-PmFT3 and AD-FBP1. Cells were then spread on SD selective medium lacking tryptophan, leucine, histidine and adenine (SD-TLHA). The plates were incubated at 30uC until positive colonies were visible after 2-4 days. The transformations were confirmed by PCR. The positive clones were in addition assayed by a filter lift bgalactosidase assay for LacZ activity as described in the manufacturer's protocols. Figure S1 The ramachandran analysis of the Fortilin model from Penaeus monodon. The number of residues in most favored regions [A, B, L] is 108 (72.50%), the number of residues in additional allowed regions [a, b, l, p] is 33 (22.10%), the number of residues in generously allowed regions [,a, ,b, ,l, ,p] is 7 (4.70%) and one (0.70%) residue in disallowed regions. The number of non-glycine and non-proline residues is 149 (88.69%), the number of end-residues (excl. Gly and Pro) is 2 (1.19%), the number of glycine residues (shown as triangles) is 11 (6.55%) and the number of proline residues is 6 (3.57%). (TIF) Figure S2 The ramachandran analysis of the FBP1 model. The number of residues in most favoured regions [A, B, L] is 31 (49.20%), the number of residues in additional allowed regions [a, b, l, p] is 21 (33.30%), the number of residues in generously allowed regions [,a, ,b, ,l, ,p] is 6 (9.50%), the number of residues in disallowed regions is 5 (7.90%). The number of nonglycine and non-proline residues is 63 (67.74%), the number of end-residues (excl. Gly and Pro) is 2 (2.15%), the number of glycine residues (shown as triangles) is 3 (3.23%) and the number of proline residues is 25 (26.88%).