Identification, Characterization, and Functional Analysis of Tube and Pelle Homologs in the Mud Crab Scylla paramamosain

Tube and Pelle are essential components in Drosophila Toll signaling pathway. In this study, we characterized a pair of crustacean homologs of Tube and Pelle in Scylla paramamosain, namely, SpTube and SpPelle, and analyzed their immune functions. The full-length cDNA of SpTube had 2052 bp with a 1578 bp open reading frame (ORF) encoding a protein with 525 aa. A death domain (DD) and a kinase domain were predicted in the deduced protein. The full-length cDNA of SpPelle had 3825 bp with a 3420 bp ORF encoding a protein with 1140 aa. The protein contained a DD and a kinase domain. Two conserved repeat motifs previously called Tube repeat motifs present only in insect Tube or Tube-like sequences were found between these two domains. Alignments and structure predictions demonstrated that SpTubeDD and SpPelleDD significantly differed in sequence and 3D structure. Similar to TubeDD, SpTubeDD contained three common conserved residues (R, K, and R) on one surface that may mediate SpMyD88 binding and two common residues (A and A) on the other surface that may contribute to Pelle binding. By contrast, SpPelleDD lacked similar conservative residues. SpTube, insect Tube-like kinases, and human IRAK4 were found to be RD kinases with an RD dipeptide in the kinase domain. SpPelle, Pelle, insect Pelle-like kinases, and human IRAK1 were found to be non-RD kinases lacking an RD dipeptide. Both SpTube and SpPelle were highly expressed in hemocytes, gills, and hepatopancreas. Upon challenge, SpTube and SpPele were significantly increased in hemocytes by Gram-negative or Gram-positive bacteria, whereas only SpPelle was elevated by White Spot Syndrome Virus. The pull-down assay showed that SpTube can bind to both SpMyD88 and SpPelle. These results suggest that SpTube, SpPelle, and SpMyD88 may form a trimeric complex involved in the immunity of mud crabs against both Gram-negative and Gram-positive bacteria.


Introduction
Toll and/or Toll-like receptors (TLRs) are membrane protein molecules evolutionarily conserved from insects to mammalian [1]. As crucial pattern recognition receptors, TLRs can be activated directly by recognizing distinct pathogen-associated molecular patterns (PAMPs) or indirectly by a cascade reaction triggered by pathogens [2]. In Drosophila, the Toll signaling pathway serves a major function in the defense against Grampositive bacteria and fungi because it can be activated indirectly by interacting with the proteolytically cleaved ligand Spaetzle [3,4]. Moreover, the diaminopimelic acid-type peptidoglycan of Gramnegative bacteria can also activate the Toll signaling pathway and further generate antibacterial peptides [5]. Aside from Toll, only Toll7 out of the eight other Drosophila TLRs has been recently proven to be involved in immunity, participating in antiviral autophagy in Drosophila [6]. By contrast, most mammalian TLRs are involved in different immune signaling pathways through sensing distinct PAMPs. TLR5, TLR11, TLR4, and the heterodimers of TLR2-TLR1 or TLR2-TLR6 located at the cell surface can be activated by recognizing bacterial flagellin, uropathogenic bacteria, bacterial LPS, and bacterial lipoproteins, respectively. Meanwhile, TLR3, TLR7-TLR8, TLR9, and TLR13 reside at the surface of endosomes, where they can be trigged by sensing double-stranded viral RNA, single-stranded viral RNA, CpG-rich bacterial or viral DNA, and bacterial ribosomal RNA, respectively [7].
The adaptor Tube (Drosophila melanogaster Tube or DmTube) and kinase Pelle (D. melanogaster Pelle or DmPelle) are essential components of the Toll signaling pathway in Drosophila. Upon ligand binding, dMyD88 (D. melanogaster MyD88), an inflexible regulator anchoring at PIP2-rich regions of the plasma membrane, can recruit the activated Toll receptor and the cytosolic adaptor Tube to permit Toll signaling [8,9]. Subsequently, Pelle is recruited to the vicinity of the Tube, thereby forming a trimeric complex (dMyD88-Tube-Pelle) mediated by the Tube death domain (DD) [10]. This process activates Pelle by autophosphorylation and its disassociation from this complex [11]. Similarly, IRAK4 and IRAK1, the pair homologs of Tube and Pelle, are essential components of human TLR signaling pathways. IRAK4 can also be recruited by MyD88; hence, these two proteins form a complicated scaffold to recruit IRAK1. Finally, IRAK1 is activated with the involvement of IRAK4 and then released from this complex [12]. However, compared with Tube, IRAK4 with the C-terminal kinase domain is a protein kinase. Thus, it can more elaborately modulate the autophosphorylation activation of IKAK1 by dint of this kinase domain [13].
Studies on the innate immunity of crustaceans have attracted considerable attention because of the huge losses caused by various aquatic animal diseases in the past decades [14,15]. Currently, increasing evidence implies that TLR signaling pathways consisting of the homologs of TLR/MyD88/Tube/Pelle/TRAF6/NF-kB in Drosophila, which are involved in the defense against pathogen invasion, may exist in crustaceans. As the central components of TLR signaling pathways, TLRs have been identified in various crustaceans and have been proven to participate in immunity and defense against several pathogens [16][17][18][19][20]. Furthermore, the homologs of most other components in the Toll signaling pathway, such as MyD88, TRAF6, Cactus, and Dorsal (nuclear factor-kB), have also been characterized [21][22][23][24]. However, two essential components of Tube and Pelle homologs in crustaceans have not been identified or well-characterized. In addition, the mechanism by which these TLR signaling pathways function remains unclear.
In our previous work, we proposed a possible antibacterial model mediated by SpToll and SpMyD88 in mud crab [25]. In this model, SpMyD88 pre-docking at specific sites of the plasma membrane with its C-terminal extension (CTE) domain may recruit activated SpToll. Thus, SpMyD88 cooperating with SpToll is involved in immunity against Gram-negative bacteria. In this study, we further identified the downstream components of SpMyD88, namely, SpTube and SpPelle, a pair of homologs of Tube and Pelle in mud crabs, and analyzed their individual characteristics and likely immune functions.

Immune Challenge and Tissue Collection
Mud crabs (150 g to 180 g each) obtained from a farm in Chongming County (Shanghai, China) were cultured temporarily in flowing, aerated seawater in 400 L tanks in laboratory (This study did not involve in endangered or protected species, and no specific permissions were required for the animal experiments). The expression profiles after injecting 50 mL of Vibrio harveyi (2610 7 CFU), 50 mL of Staphylococcus aureus (2610 8 CFU) suspension, or 100 mL of White Spot Syndrome Virus (WSSV) inoculum (1610 5 copies) into the base of the right fifth leg of each crab were investigated. The control was challenged with 50 mL of phosphate-buffered saline (140 mM NaCl and 10 mM sodium phosphate; pH 7.4) or with the supernatant of normal tissues. After anesthetizing the crabs with ice for 15 min, hemolymph was collected from the base of the right chelate leg with an ice-cold anticoagulant buffer (0.14 M NaCl, 0.1 M glucose, 30 mM trisodium citrate, 26 mM citric acid, and 10 mM ethylenediaminetetraacetic acid; pH 4.6) [26] at 0, 2, 6, 12, 24, and 48 h after challenge with V. harveyi, S. aureus, or WSSV. The collected samples were centrifuged immediately at 8006g for 15 min at 4uC to isolate the hemocytes. The gill, hepatopancreas, heart, stomach, intestine, connective tissue, muscle, and eyestalk of the mud crabs were also collected for RNA isolation. At least three crabs from each sample were selected to eliminate individual differences.

RNA Extraction and cDNA Synthesis
The collected tissues from normal crabs, together with the hemocytes from normal or pathogen-challenged crabs at different times, were utilized to isolate the total RNA with TRIzol Reagent (Ambion, USA) following the manufacturer's instructions. Firststrand cDNA was synthesized with 5 mg of RNA as the template using a First-strand cDNA Synthesis kit according to the manufacturer's instructions.

cDNA Cloning
A cDNA fragment of SpTube and a partial SpPelle cDNA sequence were identified by high-throughput transcriptome sequencing with a mixture of hemocytes and hepatopancreas. Four pairs of primers (SpTubeF1 and SpTubeR1; SpTubeF2 and SpTubeR2; SpPelleF1 and SpPelleR1; SpPelleF2 and SpPelleR2) were designed for polymerase chain reaction (PCR) and for verifying the harvested SpTube and SpPelle fragments. Two additional primers (SpTubeCF and SpPelleCF) paired with a 39 anchor R primer were also synthesized to further amplify the cDNA sequence 39 ends of SpTube and SpPelle, respectively. PCR was conducted under the following parameters: 94uC for 2 min; 35 cycles of 94uC for 35 s, 53uC for 45 s, and 72uC for 50 s; and 72uC for 10 min. The corresponding amplified fragments were subcloned into pMD-18T vectors prior to sequencing by a commercial company (Sangon, China). Complete cDNA sequences were obtained by overlapping their fragments of individual genes.

Bioinformatics Analysis
Alignments of amino acid sequences were performed using the ClustalX 2.0 program (http://www.ebi.ac.uk/tools/clustalw2) and GenDoc software. Online Basic Local Alignment Search Tool Program (BLASTP) (http://blast.ncbi.nlm.nih.gov/Blast.cgi) was used to analyze protein similarities. Translation of the amino acid sequence and prediction of the deduced protein were conducted on http://web.expasy.org/translate/. The putative domain was predicted by the Simple Modular Architecture Research Tool (http://smart.embl-heidelberg.de/). Theoretical isoelectric point (pI) and molecular weight (Mw) were calculated with online software (http://web.expasy.org/compute_pi/). MEGA4.0 was used to construct phylogenetic trees, and 1,000 bootstraps were selected for the neighbor-joining tree to assess reliability [27]. The three-dimensional (3D) structure of SpTube DD was predicted through the Phyre server [28] using the crystal structure of DmTube DD (PDB ID: 1d2z, chain B) as template. The 3D structure of SpPelle DD was modeled by homology using the SWISS-MODEL workspace (http://swissmodel.expasy.org/ workspace/) based on the templates of the solution structure of DmPelle DD (PDB ID: 1d2z, chain A). Using the same SWISS-MODEL online software, the 3D structures of both SpTube and SpPelle kinase domains were also predicted based on the crystal structure of the human IRAK4 kinase domain (PDB ID: 2nru, chain B). Comparative modeling of the 3D structures of DD or the kinase domain was generated by Swiss-PdbViewer V 4.0 software [29].

Real-time PCR
Using the synthesized cDNA as template, quantitative reverse transcription PCR (qRT-PCR) was conducted to examine the relative expression levels of SpTube and SpPelle in a real-time thermal cycler (ABI, USA) based on a previous protocol [25]. Two pairs of primers (SpTubeRF and SpTubeRR; SpPelleRF and SpPelleRR) were used to generate 191 and 147 bp amplicons, respectively (Table 1). Another pair of primers (18SRF and 18SRR) was used to amplify the corresponding fragment (121 bp) as reference. The total volume was 20 mL (10 mL of 26Premix Ex Taq, 2 mL of cDNA, and 4 mL of each primer). qRT-PCR was programmed as follows: 95uC for 5 min; 40 cycles of 95uC for 10 s and 60uC for 50 s; and a melt from 60uC to 95uC. All tests were conducted thrice using individual templates. The relative expression levels of SpTube and SpPelle in different tissues were calculated according to the 2 2gCT method. The algorithm of 2 2ggCT was applied in expression profile analysis.

Recombinant Expression and Purification
According to the full-length cDNA sequences of SpMyD88, SpTube, and SpToll, three primer pairs (SpMyD88DDEF and SpMyD88DDER; SpTubeDDEF and SpTubeDDER; SpPelleD-DEF and SpPelleDDER) were designed to amplify the sequences that encode their corresponding fragments, respectively ( Table 1). The amplified fragment of SpMyD88DD or SpPelleDD was digested by EcoR I and Xho I, and then inserted into a pET230a (+) vector. Meanwhile, the SpTubeDD fragment was ligated into a pGEX4T1 expression vector after being cut by EcoR I and Xho I. Three recombinant expression vectors, namely, pET-30a-MyD88DD, pET-30a-PelleDD, and pGEX4T1-TubeDD, were then transformed into competent Escherichia coli Rosetta (DE3) host cells. Isopropyl-b-D-thio-galactoside (IPTG) was added to induce protein expression. The recombinant SpTubeDD protein with a glutathione S-transferase (GST) tag and a GST protein was purified using glutathione Sepharose 4B chromatography (Gen-Script, USA) according to the manufacturer's instructions. Both SpMyD88DD and SpPelleDD proteins with an N-terminal His tag were harvested by His Bind resin chromatography (Novagen, USA) according to a previous method [30].

Pull-down Assay
Pull-down assay was performed according to a previous protocol with slight modifications [31]. Approximately 1 mL of purified SpMyD88DD or SpPelleDD (200 mg/mL) was incubated with His-Bind resin (1 mL) for 10 min at 4uC and then washed with 6 mL of binding buffer (0.5 M NaCl, 20 mM Tris-Cl pH 7.9, 5 mM imidazole). Afterward, SpTubeDD (1 mL) was added and then incubated for 30 min at 4uC. After being washed thoroughly with wash buffer (0.5 M NaCl, 20 mM Tris-Cl pH 7.9, 60 mM imidazole), the proteins were eluted with elution buffer (0.5 M NaCl, 20 mM Tris-Cl pH 7.9, 1 M imidazole) and then finally analyzed on 15% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). GST protein was used as the negative control in this assay.

Statistical Analysis
Significant differences were determined by one-way analysis of variance and Duncan's test for multiple range comparison using SPSS 13.0, with significant levels accepted at P,0.05. All experiments were repeated at least three times. Data are shown as means 6 standard deviation.

cDNA Cloning
A SpTube fragment and a partial SpPelle cDNA sequence harvested from transcriptome sequencing were verified further by PCR using the synthesized cDNA as template. Based on these two partial cDNA sequences, the 39 ends of the complete sequences were obtained through the Rapid Amplification of cDNA Ends technology. The full-length cDNA sequences of SpTube and SpPelle were determined by overlapping their individual fragments. The complete cDNA sequence of SpTube had 2052 bp, with a 159 bp 59 untranslated region (UTR), a 315 bp 39 UTR, and a 1578 bp open reading frame (ORF) encoding a protein with 525 aa (Genbank Accession No. KF155697) (Fig. 1). A DD (residues 14 to 129) and a serine/threonine kinase domain (residues 248 to 495) were predicted in the deduced protein; no signal peptide was found. The theoretical pI and Mw of SpTube were 4.91 and 58.2 kDa, respectively. The complete sequence of SpPelle had 3825 bp, with a 3420 bp ORF encoding a protein with 1140 aa, a 65 bp 59 UTR, and a 337 bp 39 UTR with a poly(A) tail (Genbank Accession No. KF155698). The nucleotide and deduced amino acid sequences are shown in Fig. 2. A DD (residues 11 to 101), two repeat motifs (residues 252 to 346, and residues 353 to 451), and a serine/threonine kinase domain (residues 601 to 888) were found
Based on the BLASTP results, a phylogenetic tree was constructed with Tube and Pelle homologs from insects, crustaceans, and humans to identify the evolutionary relationship among these homologs (Fig. 3). In this tree, Tube homologs without Kinase domain from D. melanogaster, Camponotus floridanus, Apis Mellifera, and Nasonia vitripennis formed a unique group, whereas other Tube and Pelle homologs were grouped into a large cluster. In this large group, all Pelle homologs (except for LvPelle) and IRAK1 formed a small meaningful cluster (the value at the nodes of the subgroups is over 70), suggesting that these Pelle homologs, including SpPelle, showed closer evolutionary relationship with human IRAK1. In addition, the SpTube and LvPelle in this tree formed another small cluster, and these two protein molecules exhibited closer evolutionary relationship with human IRAK4.

Alignments of DD and Kinase Domains
Alignments of DD sequences demonstrated that Tube and Pelle DD sequences showed clear and significant differences. Similar to Tube DD, the DDs of most insect Tube-like proteins contained three common conserved residues (R, K, and R) required for binding to MyD88s and several residues (E, A, GPXA) essential for binding to Pelles. SpTube and Tube shared the same three conserved residues (R, K, and R) that mediate binding to dMyD88 and two strongly conserved residues (A and A) that mediate binding to Pelle. By contrast, Pelle and Pelle-like protein sequences, except for LvPelle, lacked the said conserved residues (Fig. 4). In addition, alignments of kinase domain sequences  demonstrated that these proteins can be classified as RD or non-RD kinases based on a single position in sequence, which has been accepted as a criterion for distinguishing these two types of protein kinase [32]. Arthropod Tube-like kinases, human IRAK4, and LvPelle kinase were RD kinases with an RD dipeptide in the kinase domain; meanwhile, SpPelle, Pelle, arthropod Pelle-like kinases, and human IRAK1 were non-RD kinases lacking an RD dipeptide (Fig. 5).

Comparisons of the Predicted 3D Structures
Pairwise sequence alignments showed that the DDs of SpTube and SpPelle shared 28% and 35% identities with Tube and Pelle DDs, respectively. The SpTube kinase domain shared 43% identity with the human IRAK4 kinase domain, whereas the SpPelle kinase domain shared only 34% identity with its IRAK4 counterpart. Thus, the SWISS-MODEL workspace was utilized to construct the 3D models of the kinase domains of SpTube and SpPelle, as well as the DD of SpPelle, because of the high identities with their corresponding individual templates (more than 30%). Given that the identity between the two DDs of SpTube and DmTube was below 30%, Phyre server was recommended for predicting the 3D model. The SpTube DD (from 14 aa to 144 aa) and SpPelle DD (from 13 aa to 109 aa) aligned well with the crystal structure of Tube and Pelle DDs, respectively (Figs. 6A and 6B). Furthermore, similar to the spatial location of the three conserved core residues interacting with dMyD88 at one binding surface and that of several core residues interacting with Pelle at the other in Tube DD, SpTube DD also had several common conserved core residues at two separate binding surfaces with similar spatial locations. In addition, both the SpTube kinase domain (from 251 aa to 488 aa) and SpPelle kinase domain (from 540 aa to 825 aa) can merge well with the crystal structure of the human IRAK4 kinase domain (Figs. 6C and 6D). These findings suggest that these DDs and kinase domains may have similar structures and further exhibit similar functions with their corresponding counterparts in fruitfly or human.

Structure Elements in Tube and Pelle Homologs
Based on the diagrams of Tube and Pelle homologs (Fig. 7), most of the Tube and Pelle homologs, including human IRAK4 and IRAK1, were identified as kinase proteins containing a DD and a kinase domain. D. melanogaster and N. vitripennis Tube sequences were non-kinase proteins lacking a kinase domain. In addition, the ''Tube repeat motif'' present in most insect Tube homologs but not in insect Pelle homologs was absent in crustacean Tube homologs (SpTube and DpTube) and human IRAK4. However, this motif was present in the SpPelle sequence. These two motifs that share 43% identity with each other were the first identified ''Tube repeat motifs'' in Pelle homologs.

Tissue Distributions and Expression Profiles After Challenge
qRT-PCR was performed to analyze the tissue distributions of SpTube and SpPelle. As illustrated in Fig. 8A, SpTube was mainly  distributed in the gills, hemocytes, hepatopancreas, and eye stalks. SpPelle was highly expressed in hemocytes, gills, hepatopancreas, and connective tissues (Fig. 8D). In addition, the transcripts of these two genes could also be detected in all other examined tissues.
Given that both SpTube and SpPelle were highly expressed in hemocytes, we further examined the temporal expression postinjection with pathogens. After the challenge with V. harveyi, the SpTube transcripts in the hemocytes were up-regulated from 12 h to 48 h post-injection and elevated by approximately threefold at 24 h (Fig. 8B). Similarly, the SpPelle transcripts in the hemocytes were up-regulated at 12 h. They reached their highest levels (approximately threefold increase) at 24 h post-injection and then recovered to their normal levels at 48 h post-injection (Fig. 8E). Moreover, slight decreases were also found for both SpTube and SpPelle transcripts at 2 h post-injection. The expression of SpTube in the S. aureus-challenged experiments was remarkably elevated from 6 h to 12 h and then recovered gradually (Fig. 8B). The expression of SpPelle significantly increased from 2 h to 12 h after the challenge (Fig. 8E). These results imply that both SpTube and SpPelle may participate in the immunity against bacteria. Although SpPelle expression decreased from 24 h to 48 h after challenge with WSSV, it was elevated from 6 h to 12 h post-injection. By contrast, SpTube was down-regulated by WSSV injection from 6 h to 48 h, suggesting that SpPelle migh be also involved in antiviral immune responses (Figs. 8C and 8F).

Recombinant Expression and Purification
SpTubeDD was expressed as a soluble protein after IPTG induction. It was conveniently purified by glutathione Sepharose 4B chromatography. The purified protein comprised a DD (theoretical Mw, 13.2 kDa) and an approximately 26 kDa GST tag expressed by the plasmid pGEX4T1, which was roughly consistent with the size (approximately 39.2 kDa) of the unique band in the purified protein lane (Fig. 9A). Both SpMyD88DD and SpPelleDD were purified by His-Bind resin chromatography because each of these two recombinant proteins contained an additional N-terminal His Tag. Therefore, the theoretical Mw values of the purified SpMyD88DD and SpPelleDD were larger than those of their separate DDs, i.e., approximately 15.3 and 16.5 kDa, respectively, which were approximately consistent with their individual SDS-PAGE results (Figs. 9B and 9C). Binding Activity of SpTube DD The potential binding activities of SpTubeDD with SpMyD88DD and SpPelleDD were examined by a pull-down assay. GST-tagged SpTubeDD but not GST protein could bind to his-tagged SpMyD88DD (Fig. 10). In addition, His-tagged SpPelleDD but not GST protein could interact with GST-tagged SpTube DD (Fig. 11). These results indicate that SpTube DD could interact with both SpMyD88DD and SpPelleDD.

Discussion
In the past few years, although increasing evidence provided by different research groups suggests that TLR signaling pathways may exist in crustaceans, no such pathway has been established in any crustacean species to date [33]. One of the original causes is the fact that some essential components of TLR signaling pathways, such as Tube and Pelle homologs, have not been identified or well characterized in crustaceans. In this study, we identified the first pair of crustacean homologs of Tube and Pelle, namely, SpTube and SpPelle, and analyzed their individual characteristics and likely immune functions. SpTube was found to interact with both SpMyD88 and SpPelle, which are involved in the immunity against both Gram-negative and Gram-positive bacteria.
In Drosophila, Tube and Pelle are two closely related but distinct components of Toll signaling pathway. A former study revealed that Tube and Pelle may arise by duplication and divergence of an ancestral gene [32]. One remarkable difference between these two protein molecules is that Tube DD is bivalent, which mediates specific and essential interactions with dMyD88 at one binding surface and with Pelle at the other [32]. In our study, we found that it was SpTube DD, instead of SpPelle DD, that shared several common conserved core residues at two separate binding surfaces with Tube DD. Furthermore, these conserved residues showed similar spatial locations with those in Tube. These results indicate that SpTube, similar to Drosophila Tube, may also be a bivalent protein that possesses potential binding activity with both its upstream component SpMyd88 and downstream component SpPelle.
In addition, although Drosophila Tube lacks the kinase domain, most of the Tube and Pelle homologs from insects and humans are a pair of kinase proteins with a kinase domain in their individual C-terminals: one is an RD kinase and the other is a non-RD kinase. A previous study demonstrated that insect Tube-like kinases and human IRAK4 are RD kinases, whereas Pelle, insect Pelle-like kinases, and human IRAK1 are non-RD kinases [32]. In humans, the presence of a pair of protein kinases is involved in activating the downstream components through both phosphoregulation and phosphorelay activities [12]. Given that SpTube is an RD kinase and SpPelle is a non-RD kinase, SpTube and SpPelle may be their individual counterparts of human IRAK4 and IRAK1 in mud crab. Thus, the mechanism by which these two proteins function in the likely-existing TLR signaling pathway may be similar to that of their individual counterparts in humans.
Until recently, Drosophila Tube and Pelle have been considered as an orthologous pair to human IRAK-4 and IRAK-1 [32]. However, before this report, no mammalian counterpart for Drosophila Tube has been identified because of the lack of a catalytic domain for Tube, which once had been a long-standing confusion regarding Toll signaling in flies and mammals. Furthermore, IRAK4 was once considered as the counterpart of Pelle [34]. This assumption may mislead the following study about how to identify these two genes in other species. For example, the recently identified crustacean LvPelle [35], which shares higher identity with human IRAK4 (29.3%) and Drosophila Pelle (27.0%) than with Drosophila Tube (14.7%) and contains a DD and a kinase domain, has been considered as a ''Pelle'' homolog. However, this protein should be a Tube rather than a Pelle homolog because its DD possesses the typical characteristics of a Tube homolog, i.e., a bivalent DD and a RD kinase domain. Furthermore, LvPelle has shown close evolutionary relationship with Tube homologs, such as human IRAK4 and SpTube.
In addition to the DD and kinase domain, many insect Tubelike proteins contain several copies of the conserved sequence motif ''Tube repeat motif,'' which is also a distinctive feature for Tube molecules because this motif exists only in insect Tubes but not in Pelle or Pelle-like proteins in previous reports [32]. Intriguingly, two copies of the conserved repeat motifs were found in SpPelle but not in SpTube. This finding indicates that the conserved repeat motifs were no longer a unique feature for Tube and Tube-like molecules, which were once considered as a marker to distinguish Tube from Pelle. Although we did not test the exact function of these conserved repeat motifs, a previous report showed that these repeat motifs of Drosophila Tube participate in the binding activity to Dorsal and that Pelle can bind to Dorsal [36]. Based on this result, we speculated that SpPelle binds to SpDorsal and that SpPelle repeat motifs contribute to this binding activity. This speculation further implies that SpTube and SpPelle have important functions in the signaling transduction of the possibly-existing TLR signaling pathway in mud crab.
LvPelle (now identified as Tube ortholog) in hepatopancreas is upregulated by Vibrio alginolyticus [35]. In the present study, SpTube and SpPelle were elevated by V. harveyi, another species of Gramnegative bacteria. These two molecules can also be induced by the Gram-positive bacteria S. aureus. These results indicate that both Tube and Pelle orthologs of crustaceans may participate in the immunity against both Gram-negative and Gram-positive bacteria. In addition, SpTube DD and SpPelle DD have similar 3D structures with their respective counterparts of Drosophila Tube and Pelle. Furthermore, SpTube was also proven capable of binding to SpMyD88 and SpPelle via its bivalent DD. These findings suggest that SpMyD88, SpTube, and SpPelle may form a trimeric complex involved in the immunity against both Gram-negative and Grampositive bacteria.
In the Drosophila Toll signaling pathway, dMyD88 recruits its upstream-activated Toll receptor and downstream cytosolic adaptor Tube [8], thereby providing a platform to permit Pelle incorporation and forming a trimeric complex (dMyD88-Tube-Pelle). The forming of this complex further activates Pelle, which consequently activates Dorsal [37]. Similarly, in our previous work, we proposed a plausible antibacterial model mediated by SpToll and SpMyD88. In this model, SpToll cooperating with SpMyD88 is involved in the immunity against Gram-negative bacteria [25]. In this study, we further complement this likely antibacterial model. Upon challenge with Gram-negative bacteria, SpMyD88, which pre-anchored the plasma membrane via its CTE domain, would recruit its upstream-activated SpToll and down-   stream cytosolic adaptor SpTube. As a scaffold protein, SpTube would further recruit SpPelle, thereby forming a trimeric complex. In this case, SpPelle may be finally activated via the kinase domain of SpTube in a similar manner to IRAK1 activated by IRAK4 [13].
In conclusion, we identified the first pair of Tube and Pelle homologs in crustaceans and further analyzed their individual characteristics. SpTube is a RD kinase with a bivalent DD and a C-terminal kinase domain, whereas SpPelle is a non-RD kinase with an N-terminal DD, two repeat motifs, and a C-terminal kinase domain. SpTube was quite different from Drosophila Tube in sequence structure, which contains a DD and repeat motifs but lacks a C-terminal kinase domain. The conserved ''Tube repeat motif'' was once found in only insect Tube or Tube-like molecules; however, the similar ''Tube repeat motif'' appeared in SpPelle.
These findings well characterize SpTube and SpPelle, as well as the Tube and Pelle homologs in crustaceans and insects. In addition, SpTube can bind to both SpMyD88 and SpPelle, which are involved in the immunity against both Gram-negative and Grampositive bacteria. Therefore, these findings further complement our previously proposed antibacterial model that may exist in mud crab.