Alternatively spliced down syndrome cell adhesion molecule (Dscam) controls innate immunity in crab

Alternatively-spliced hypervariable immunoglobulin domain-encoding molecules, called Down syndrome cell adhesion molecule (Dscam), have been widely detected as components of the arthropod immune system. Although its ability to specifically bind pathogens and enable phagocytosis of bacteria has been elucidated, the signal transduction mechanisms or effectors that activate post-Dscam–binding pathogens remain poorly characterized. Here, we reveal the alternative splicing exons of Dscam's cytoplasmic tail and its isoforms in the hemocytes of crab (Eriocheir sinensis), showing that the expression of Dscam was acutely induced after an immune challenge, which suggested its functioning for innate immunity. Significantly decreased expression levels of antimicrobial molecular peptides (AMPs) were detected in Dscam-silenced crab hemocytes in vitro, which coincided with their vulnerability to infection by Staphylococcus aureus and higher bacterial concentrations occurring in Dscam-silenced crabs in vivo. Further experimental investigation demonstrated that Dscam-regulated AMP expression via the Src homology (SH)3-binding domain in the first constant exon translated protein of the cytoplasmic tail bound with the SH3 domain of the Dock, an SH3/SH2 adaptor protein required for axon guidance. Dock promoted extracellular signal-regulated kinase (ERK) phosphorylation via indirect binding and then regulated dorsal phosphorylation and translocation from the cytoplasm to the nucleus, subsequently promoting AMP expression for the effective removal of bacteria. To the best of our knowledge, this comprehensive study is the first to highlight the critical role of the alternatively-spliced Dscam cytoplasmic tail in antimicrobial control activity. It also suggests possible cross-talk occurring between Dscam and other pattern recognition receptors.

The Down syndrome cell adhesion molecule (Dscam) 4 belongs to novel members of the immunoglobulin superfamily due to shared structural features with Igs (1,2). Dscam functions in the differentiation of the nervous system (1) and in the recognition and binding activity of the innate immune reaction in animals (2). Throughout the metazoa, Dscam's domain composition and physical arrangement has remained identical, namely 9(Ig)-4(fibronectin, FN)-1(Ig)-2(FN) (3). The most striking characteristic of Dscam is its alternative splicing ability, which can potentially generate 38,016 different isoforms through mutually exclusive splicing of four cassette exon clusters in Drosophila melanogaster (1,4). Furthermore, additional sources of diversity were also identified in the cytoplasmic tail (3,5,6), which may generate even more Dscam isoforms. Alternative splicing is an ancient and widespread mechanism used by eukaryotic organisms to enhance protein diversity and regulate gene expression (7). For example, up to 95% of the primary transcripts in humans are thought to undergo alternative splicing (8), which may contribute to subsequent functional differentiation among cells.
The extensive somatic diversification of their immune receptors is a hallmark of higher vertebrates (9,10). Nevertheless, recent studies have demonstrated the critical role of Dscam alternatively-spliced isoforms in providing immune protection for insects (4). In vivo silencing of Dscam reduced phagocytic ability and led to a proliferation of opportunistic bacteria, even in the hemolymph of unchallenged mosquitoes, which together suggested Dscam's key involvement in anti-bacterial immunity. Interestingly, after mosquito Sau5B cells were infected with bacteria, pathogen-induced Dscam showed a high-binding affinity against the originally induced bacteria, whereas specific Ig2 variant-Dscam isoform-silenced mosquitoes became more susceptible to infection when challenged with the same bacteria (4). Collectively, the body of research to date suggests a potentially critical role played by Dscam in constituting specific antibacterial immune responses. Furthermore, in vivo overexpression of parasite-specific Dscam isoforms was able to augment the mosquitoes' resistance to particular parasites, while in vivo silencing of a Plasmodium-specific isoform rendered the mosquitoes more susceptible to the parasite; those findings indicated specific Dscam isoforms may also contribute to immunity against parasites as well (4,11). Taken together, the evidence clearly supports the view that Dscam alternatively-spliced isoforms could specifically bind to different pathogens, leading to subsequent well-known immune reactions such as phagocytosis (2).
Yet, compared with the alternatively spliced Ig domains in Dscam, little is known about how its cytoplasmic tail may contribute to immune protection. Undoubtedly, far fewer alternatively spliced isoforms are derived in the cytoplasmic tail than Ig domains due to the generally limited exons in this region; but the combination of alternatively-spliced exons in Dscam cytoplasmic tails may provide an efficient mechanism by which different signal transduction pathways and functions are triggered following pathogen-specific binding events in immune responses.
In this respect, to date the best-studied and understood Dscam function is phagocytosis regulation in cellular immunity, now widely confirmed in different arthropods by multiple research groups (2,4,12). In contrast, whether Dscam can trigger an immune signaling pathway that contributes to humoral immunity, such as that achieved by antimicrobial peptides (AMPs) via the alternatively-spliced cytoplasmic tail, is a far from resolved issue. In this study, we show the genomic structure of Dscam's cytoplasmic tail in crab, test the immune protection role of Dscam in a bacterial challenge, identify the AMPs whose expression regulates Dscam both in vivo and in vitro, and uncover a novel antimicrobial-signaling pathway activated by the cytoplasmic tail of Dscam. As far as we know, this is the first reported evidence of a Dscam-triggered immune signaling pathway in nature.

Gene structure and expression pattern of Dscam
The transmembrane region and cytoplasmic tail were both unique in the membrane-bound Dscam compared with soluble Dscam (2). Previous studies focused on alternative splicing exons in the Ig domain that are shared by membrane-bound and soluble Dscam (2,4,13,14); however, information about the alternatively-spliced cytoplasmic tail still remains limited. To resolve this outstanding issue, the gene structure of Dscam was determined through whole-genome sequencing of Eriocheir sinensis, 5 followed by the exons' predicting and cloning of many cDNAs. Results revealed nine exons in the cytoplasmic tail of Dscam, named exons 32-40 (Fig. S1), followed by the 31 exons in the outer membrane region. This cytoplasmic tail can generate four isoforms (Fig. S2), including exons 32-40, 35 Ϫ/ Ϫ 36 Ϫ/Ϫ , 33 Ϫ/Ϫ , and 33 Ϫ/Ϫ 35 Ϫ/Ϫ 36 Ϫ/Ϫ by the alternatively spliced exons 33 and 35/36, while the signature domain of the SH3-and SH2-binding domains were located in the constant exons 32 and 37. The signature motif of endocytosis, polyproline, and ITIM-like were located in alternatively spliced exon 33, constant exon 38, and constant exon 40, respectively (Fig. 1A). The phylogenetic analysis demonstrated that the Dscam of crustaceans had diverged from that of insects, albeit the relationship of the two Daphnia species' Dscam was ambiguous. Moreover, the EsDscam clustered with another mud crab, Scylla paramamosain, which formed a sister group with other crustaceans (Fig. 1B).
To investigate the expression profiles of Dscam isoforms, primers were designed across constant exon 37 and exon 38 as it is shared between all membrane-bound Dscam (Fig. 1C). The LPS (Fig. 1D) and Staphylococcus. aureus (Fig. 1, E and F) challenge induced high expression levels of Dscam, both in vitro and in vivo, and its expression peaked at 24 h post-S. aureus infection in vitro (Fig. 1E) and 12 h post-S. aureus stimulation in vivo (Fig. 1F). 24 h since the primary cultured hemocytes of E. sinensis was S. aureus-stimulated, the expressions of 10 AMPs (ALF1, ALF2, ALF4, LYZ1, LYZ2, LYZ3, LYZ4, Crus1, Crus3, and DWD) were also significantly induced (Fig. 1G). This result pointed to a possible linkage between Dscam and the expression of AMPs.
To elucidate the role of Dscam in vivo, crabs were injected with dsGFP or dsDscam for 12 h and then received an S. aureus infection and were injected with dsRNA again at day 2 to extend the effect of RNAi. Crab survival was monitored from days 1 to 4, and bacterial CFU were analyzed at day 3 (Fig. 2D). Dscam mRNA and protein expression were significantly knocked down by the dsRNA injection for at least 3 days (Fig. 2E), accompanied by significantly reduced survival (Fig. 2F), as well as significantly increased bacterial concentration in the hemolymph compared with the dsGFP-treated group of crabs (Fig.  2G). Together, this suggested Dscam might control antimicrobial activity of crab via regulation of AMPs.

Dscam regulated the signaling pathway that led to expression of AMPs
The crucial role of the Toll pathway in regulating AMPs' expression following Gram-positive bacterial infection of invertebrate species has been widely confirmed (15). Hence, we speculated Dscam likewise regulated AMPs' expression in crab; a hypothesis tested here by knocking it down in vitro. Not only was dorsal phosphorylation obviously reduced when compared with the siGFP-treated group ( Fig. 3A and Fig. S5A), the dorsal translocation from cytoplasm to nucleus was also suppressed in the Dscam-silenced hemocytes (Fig. 3B), which is consistent with Dscam regulating the Toll pathway. ERK reportedly promotes dorsal translocation (16), so we also considered whether Dscam could regulate the MAPK/ERK pathway. To test this, the phosphorylation and total protein levels of ERK and JNK were determined, which demonstrated a clearly inhibited ERK phosphorylation rate, as well as a total ERK protein level in Dscam-silenced hemocytes at both the 30-and 60-min post-S. aureus stimulation ( Fig. 3C and Fig. S5B) that was supported by the suppressed p-ERK fluorescent signal in Dscam-silenced hemocytes (Fig. 3D). Unfortunately, antibody-binding sites of the commercial p-38 antibody shared low similarities with p-38 in crab, and the specific E. sinensis p-38 antibody could not be synthesized, hence their omission in this study.

ERK regulates expression of AMPs via the Toll pathway
To test whether ERK could regulate AMPs expression, ERK was knocked down in crab hemocytes (Fig. 4A). Doing so significantly decreased the expression of AMPs (ALF1, ALF2, ALF4, LYZ1, LYZ3, LYZ4, Crus1, and DWD) detected before as regulated by Dscam upon S. aureus stimulation, when compared with the siGFP-treated group (Fig. 4B). We further speculated that ERK may regulate AMPs' expression via dorsal. To test this hypothesis, ERK phosphorylation was inhibited by the MEK1/2 inhibitor PD98059, and the 10 M concentration clearly affected ERK phosphorylation inhibition ( Fig. 4C and Fig. S5C); moreover, this concentration also produced a conspicuous effect after the S. aureus stimulation ( Fig. 4D and Fig. S5D). By using the ERK phosphorylation inhibition system, reduced dorsal phosphorylation was

Dscam controls innate immunity in crab
found compared with the inhibitor-treated group ( Fig. 4E and Fig. S5E). To verify this result in mRNA levels, ERK was knocked down in hemocytes followed by S. aureus stim-ulation, and reduced dorsal phosphorylation was confirmed compared with the siGFP-treated group ( Fig. 4F and Fig.  S5F).  The green fluorescence signal indicated the distribution of phosphorylated dorsal, whereas blue indicated hemocytes' nucleus stained with DAPI. C, knockdown of Dscam reduced both total ERK and phosphorylated ERK expression in crab hemocytes after 30 or 60 min of S. aureus stimulation, but not JNK, in the Western blotting. D, Dscam RNAi reduced phosphorylated ERK in both the cytoplasm and nucleus following S. aureus stimulation, as detected by immunocytochemical staining. The green fluorescence signal indicated the distribution of phosphorylated ERK, whereas blue indicated the nuclei of hemocytes stained with DAPI. All data shown are from three independent experimental repeats. Where applicable, siGFP or siGFP plus stimulation served as the control.

Dscam controls innate immunity in crab Dock regulates antimicrobial activities
That Dock protein binds with Dscam in neuron cells is known (1), so we investigated the effect of Dock in Dscamregulated ERK phosphorylation because the immune functions of Dock have not been reported. To test whether Dock regulated innate immunity in crab, the cDNA full-length of the Dock gene in E. sinensis was cloned (Fig. S3), and its bioinformatics analysis suggested three SH3 domains and one SH2 domain in the Dock protein, with the Dock protein highly similar within the arthropod species ( Fig. 5A and Fig. S4). The expression level of Dock was significantly induced post-S. aureus stimulation in hemocytes, especially at 12 h (Fig. 5B), which indicated its immune functioning in crab. Next, siRNA was transfected into hemocytes for the RNAi Dock expression (Fig. 5C). Compared with the siGFP-treated group, this resulted in a clearly decreased ERK phosphorylation signal in hemocytes (Fig. 5D), a suppressed dorsal translocation from cytoplasm to nucleus (Fig. 5E), and significantly reduced AMPs' expression levels regulated by Dscam in Dock-silenced hemocytes upon S. aureus stimulation (Fig. 5F). Collectively, these results suggested Dock could regulate AMPs expression via ERK/dorsal signaling in the crab.

Dscam binding with Dock
The SH3-binding domain in Dscam may act as a potential target for the SH3 domain in Dock to affiliate with it (1). To elucidate how binding might occur between Dscam and Dock, the full-length of Dock and four truncated Dscam cytoplasmic tail cDNA plasmids (consisting of all constant exons for 891 bp, exon 33-36/38 -40 for 807 bp, exon 34/37-40 for 720 bp, and full-length for 1182 bp) were constructed and transfected into HEK293T cells (Fig. 6A). The co-IP results demonstrated that each plasmid was well-expressed in these cells; the full-length Dscam cytoplasmic tail (FLAG-1182) could strongly bind with the Dock protein; all constant exons containing the isoform (FLAG-891) had a strong binding activity with the Dock protein; isoform without a constant exon 32 and exon 37 (FLAG-807) was unable to bind with the Dock protein; isoform con-

Dock regulates ERK phosphorylation via indirect binding
We then wondered whether Dock could regulate the phosphorylation of ERK, as this would enable it to later participate in the process of Dscam-regulated antimicrobial activities. To test this, Dock was knocked down in hemocytes, for which the Western blotting results showed an obviously reduced ERK phosphorylation at both 30 and 60 min post-S. aureus stimulation ( Fig. 6C and Fig. S5G). Moreover, both Dock and ERK plasmids were transfected into S2 cells for overexpression, which resulted in a greater ERK phosphorylation rate in Dockoverexpressed cells (Fig. 6D and Fig. S5H). In sum, these results demonstrated Dock regulated ERK phosphorylation. To explore whether Dock did so via direct binding, the plasmids of these two genes were transfected into HEK293T cells, but the co-IP results revealed no evidence for direct binding (Fig. 6E). Moreover, this result was confirmed by co-immunoprecipitation assay by using a specific antibody against EsERK and EsDock in E. sinensis hemocytes (Fig. 6F).

Discussion
The Dscam gene is an extraordinary example of biomolecular diversity; by combining alternatively spliced exons, thousands of isoforms can be produced from this single gene (1,17,18). Such diversity has so far only been found in insects and crustaceans (19), and its essential involvement in neural networks' wiring is now well-characterized for D. melanogaster (20). Pioneering studies of multiple arthropod species provide convincing evidence that Ig domains in the outer membrane region of Dscam specifically bind to pathogens, subsequently activating phagocytosis in hemocytes (2,4). These exciting findings showed that via processes of somatic diversification, arthropod species harbor an unexpectedly great potential diversity of immune molecules. Here, we hypothesized Dscam

Dscam controls innate immunity in crab
could provide the mechanistic underpinnings of such specific immune responses.
As a pathogen recognition receptor, the cytoplasmic tail of membrane-bound Dscam provides the structural basis for signaling transduction upon receiving dangerous signals by Ig domains situated in the outer membrane region. Surprisingly, however, this process has not been investigated in previous studies. Dscam isoforms with different cytoplasmic tails were identified in some species, such as D. melanogaster (5), Daphnia magna (3), and Penaeus monodon (6), with varied patterns of alternative splicing. For example, the four cytoplasmic tail variants found in D. magna arise from the inclusion or exclu-sion of exons 27 and 30, and the latter's absence would shift the downstream reading frame (3). Similarly, four cytoplasmic tail variants were also found in D. melanogaster due to the presence or absence of exons 19 and 23, with the downstream reading frame shifted if exon 23 was excluded from mature mRNA (5). By contrast, in crab Dscam, there are two individually spliced exon regions leading to four cytoplasmic tail variants, no matter which exon is excluded, so the downstream reading frame remains unchanged. Similarly, the combination of six diverse elements (putative exons) led to different cytoplasmic tail isoforms in P. monodon, whose downstream reading frame is unaffected by a sequence deletion (6). The diverse functional assay showed the interaction between differently-truncated Dscam cytoplasmic tail proteins and Dock in HEK293T cells. Anti-V5 and anti-FLAG antibodies were used to detect Dock and truncated Dscam isoforms in co-infected HEK293T cells, respectively. The anti-V5 and anti-FLAG antibodies were incubated with the cell lysates and then isolated using Protein G-FLAG-m-beads. C, Western blotting detected the reduced phosphorylation of ERK in Dock-silenced crab hemocytes stimulated by S. aureus at 30 and 60 min, respectively (siGFP ϭ control). D, ERK phosphorylation increased in cultured S2 cells overexpressing both Dock and ERK proteins. E and F, co-immunoprecipitation (IP) assay did not shown any interaction between Dock and ERK protein both in the HEK293T cells (E) and crab hemocytes (F). B and E, anti-V5 and anti-Myc antibodies were used to detect Dock and ERK in co-infected HEK293T cells, respectively. The anti-V5 and anti-FLAG antibodies were incubated with the cell lysates and then isolated using Protein G-FLAG-m-beads. The data shown are from three independent experimental repeats.

Dscam controls innate immunity in crab
motifs in such exons might be associated with different signal transduction pathways, perhaps corresponding to the neuron development and immune response functions of Dscam. In a study of Drosophila neuronal morphogenesis, when compared with the full-length endodomain, those isoforms without exons 19 and 23 played a more important role in post-embryonic stages (5). For immune reactions, however, the functions of spliced exons and endodomains in the cytoplasmic tail of EsDscam and even among well-studied arthropods are unknown.
Dock is an adaptor protein containing three SH3 domains and a single SH2 domain, and it is closely related to mammalian Nck (21,22). Dock mutants often show defects in axon guidance in the adult fly's visual system and its embryonic nervous system (1). In previous studies of D. melanogaster, Dscam was isolated by virtue of its affinity to Dock, in that Dscam can directly and strongly bind to its SH3 domain but only early to its SH2 domain (1). Genetic studies have revealed that Dscam, Dock, as well as Pak, a serine/threonine kinase, may act together to direct path-finding of Bolwig's nerve, which contains a subclass of sensory axons, to an intermediate target in the embryo (1). Interestingly, we found Dock expression was substantially induced following bacterial stimulation, with silencing of the Dock gene in hemocytes resulting in substantially reduced AMPs' expression, ERK phosphorylation, and dorsal translocation, all of which convincingly demonstrate for the first time the critical role played by Dock in innate immunity. More importantly, Dock acted as a bridge to transduce the signaling from the Dscam cytoplasmic tail to ERK in the hemocytes, which contributed immensely to antimicrobial activities. This novel phenomenon indicates Dscam may regulate innate immunity via a Dock-activated signaling pathway that differs with nervous development.
Admittedly, it is difficult to distinguish soluble and membrane-bound Dscam because they share the same outer membrane regions. Yet, nonalternatively spliced exons in the cytoplasmic tail of Dscam provided a good target for us to study the unique functions of membrane-bound Dscam isoforms. In our study, significantly enhanced Dscam expression in crab hemocytes suggests it performs key immune-related functions, as both in vitro and in vivo experiments demonstrated critical antimicrobial activities of Dscam in the E. sinensis crab. We also found that the SH3-binding domain in constant exon 32 translated protein binding with the SH3 domain in Dock, to activate ERK phosphorylation, promoting the phosphorylation and translocation of dorsal, which subsequently enhanced the expression of AMPs. Exon 32 in the cytoplasmic tail was fixed in membrane-bound Dscam; hence, its function in regulating AMPs' expression is likely a common feature of all membranebound Dscam isoforms.
For the cytoplasmic tail of Dscam in E. sinensis crab, alternatively-spliced exons can generate four isoforms, with the SH2-/ SH3-binding domain, endocytosis motif, polyproline motif, and ITIM-like motif in the alternatively and nonalternatively spliced exons. We interpret this as evidence for some conserved functions shared among Dscam cytoplasmic tail isoforms. In particular, the endocytosis motif is located in an alternatively spliced exon, which indicated isoforms with versus without this exon may differ considerably in their signaling capacity. The endocytosis motif is involved in determining the exact site of viral release at the surface of infected mononuclear cells and promotes endocytosis (HAMAP database), which may be linked with the common functions of Dscam-regulated phagocytosis. Building on this premise, we hypothesize that besides the common function of regulating AMPs' expression, isoforms with alternatively spliced exons may have other immune-related functions linked to pathogen-specific recognition capacity of Ig domains in the outer-membrane region of Dscam. If so, this should be able to generate "specific immunity" in invertebrates.
The results from this comprehensive study together demonstrate that Dscam's cytoplasmic tail can generate four isoforms via alternative splicing, with all isoforms able to bind to the SH3 domain in the Dock protein because of the SH3-binding domain in the first constant exon. Dock then promotes ERK phosphorylation via indirect binding, and it regulates dorsal phosphorylation and translocation from the hemocytes' cytoplasm to nucleus. This enables dorsal to regulate the expression of AMPs resulting in the effective clearance of pathogenic bacteria in crab (Fig. 7). Collective data indicate the comprehensive function of Dscam apart from regulating phagocytosis and suggest possible cross-talk occurring between Dscam and other pattern recognition receptors.

Study of arthropods and its primary cultured hemocytes
Animal experiments were performed according to a protocol approved by the East China Normal University Animal Care and Use Committee (protocol license number: AR2012/12017), following the animal care guidelines of the Ministry of Science and Technology of the People's Republic of China. Healthy E. sinensis crabs (100 Ϯ 10 g per adult, and 15 Ϯ 2 g each for juvenile crabs in a life-stage, nonantibiotic or antifungal fed) were obtained from the Songjiang aquatic farm in Shanghai, China. After quickly transferring them to the Biological Experiment Station at East China Normal University, all crabs were maintained in filtered and aerated freshwater with an abundance of oxygen, and they were fed daily with a commerciallyformulated diet free of any antibiotics.

Immune stimulation and sample collection
S. aureus bacteria were gifted to us by the National Pathogen Collection Center for Aquatic Animals (Stock No. BYK0113, Shanghai Ocean University, Shanghai, China). Bacteria were cultured, collected, and resuspended in sterile PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 2 mM KH 2 PO 4 , pH 7.4). The bacterial counts were determined by plating each diluted suspension onto agar plates.

Dscam controls innate immunity in crab
For in vitro bacterial stimulation, cultured S. aureus (1 ϫ 10 7 microbes per dish, 50 l) were added separately to the hemocyte-cultured dish, with sterile PBS (50 l) serving as the control. For in vitro pathogen associated molecular pattern stimulation, lipopolysaccharide (Sigma) was added separately to a 60-mm dish to a final concentration of 1 g/ml, with sterile PBS (50 l) as the control. Total RNA was collected from hemocytes at specific time points post-stimulation. At least three crabs were used for each sample. First-strand cDNA was synthesized with a reverse transcriptase kit (Takara, Osaka, Japan) according to the manufacturer's instructions.
For in vivo bacterial infection, S. aureus (1 ϫ 10 8 CFU per crab, 200 l) were injected into hemolymph from the nonsclerotized membrane of the posterior walking leg, with sterile PBS (200 l) used as the control. Total RNA was collected from hemocytes at specific time points post-infection. At least three crabs were used for each sample.

Genomic sequencing, assembly, annotation, and phylogenetic analysis
Total DNA from the muscle of healthy E. sinensis was sent to the Novogene Co. (Beijing, China) for whole-genome sequencing on the Illumina HiSeq 2000 and PacBio platforms. The cytoplasmic tail of the EsDscam gene sequence was captured from genomic data by alignment and searching within the EsDscam cDNA sequence. From genomic data that encoded a protein with SH3 and SH2 domains, a fragment containing a full ORF was identified and designated as EsDock. The full-length EsDscam and EsDock cDNAs were then each amplified by a pair of gene-specific primers and re-sequenced to confirm the correct sequences. The similarity analysis was conducted using Clustalx from EMBL, with signal peptide(s) predicted by SignalP version 4.1 Server (http://www.cbs.dtu.dk/services/ SignalP/) 6 (24) and the domain architecture predicted by SMART (http://smart.embl-heidelberg.de/). 6 The MUSCLE method in MEGA version 6.0 software was used for the alignment of multiple amino acid sequences between the EsDscam cytoplasmic tail (ANQ45731.1) and other orthologs of representative pancrustacea in GenBank: Aedes aegypti (EAT37388.1); Apis mellifera (NP_001014991.1); Tribolium castaneum (KYB25260.1); D. magna (ACC65887.1); D. melanogaster (AF260530.1); Daphnia pulex (EFX86769.1); S. paramamosain (ANN87808.1); Pacifastacus leniusculus (AEC50084.1); and Cherax quadricarinatus (AGK90306.1). A nonrooted neighborjoining phylogenetic tree based on 1000 bootstrap replications was constructed in MEGA version 6.0.

RNAi assay
The cDNA fragments of EsDscam, EsERK, and EsDock were first amplified by PCR with primers linked to the T7 promoter (Table S1, and then were used as the template to produce 6 Please note that the JBC is not responsible for the long-term archiving and maintenance of this site or any other third party hosted site.

Dscam controls innate immunity in crab
siRNA with an in vitro T7 transcription kit (Fermentas, Burlington, Canada). The corresponding control, GFP dsRNA, was purchased from GenePharma Co. (Shanghai, China). For in vitro RNAi, the siRNA was dissolved in RNase-free water and transfected into E. sinensis primary-cultured hemocytes using Lipofectamine 3000 (Thermo Fisher Scientific) at a final concentration of 10 nM. For in vivo RNAi, the dsRNA (1 g/g) was injected into the hemolymph from the nonsclerotized membrane of the posterior walking leg of each crab. To strengthen the RNAi effect, a second injection was provided 24 h after the first injection. The efficiency of RNAi was determined via realtime RT-PCR and Western blotting, with dsGFP RNA used as the control. After setting up the RNAi assay, S. aureus was used to stimulate the gene-silenced hemocyte in vitro or was injected into the gene-silenced crab in vivo.

Gene expression profile analysis
The relative expression levels of EsDscam, EsDock, EsERK, and AMP expression in differently-treated hemocytes were determined by real-time RT-PCR using the CFX96 TM Real-Time System (Bio-Rad) and SYBR Premix Ex Taq (Tli RNaseH Plus; TaKaRa, Japan) with the primers shown in Table S1. The reaction conditions were 94°C for 3 min, followed by 40 cycles of 94°C for 10 s and 60°C for 1 min, and then melting from 65 to 95°C. The 2 Ϫ⌬⌬CT method was applied to the obtained data, which were normalized to the control group's samples. Three independent experiments were carried out, for which their results as the mean Ϯ S.D. is presented.

Assessment of survival rates and bacterial clearance assay
Sixty crabs were randomly assigned to two groups (n ϭ 30 per group). Each crab of the first group was injected with EsDscam dsRNA, and 15 g of GFP dsRNA was injected into crabs of the second group. After EsDscam was knocked down by the dsRNA injection, all 60 crabs were injected with S. aureus (1 ϫ 10 9 CFU per crab, 200 l). The number of dead crabs in each group was recorded daily, and their respective survivorship over 4 days was calculated. Twenty four h after the bacteria injection, the hemolymph from each live crab was collected, diluted, and then cultured on solid LB plates overnight, after which its bacterial colonies were counted.

Western blotting
Protein samples obtained from hemocytes were quantified by the Coomassie Plus Protein Assay Reagent (Thermo Fisher Scientific). Whole-cell lysates were obtained by lysing cells with an RIPA buffer containing protease and phosphatase inhibitor mixtures (Invitrogen). Protein concentration was measured with the Pierce BCA protein assay kit (Thermo Fisher Scientific). Next, the protein was separated by 10% SDS-PAGE and transferred onto a nitrocellulose membrane. This membrane was blocked for 1-2 h with 3% nonfat milk in Tris-buffered saline (TBS: 10 mM Tris-HCl, pH 8.0, 150 mM NaCl) and then incubated with 1:200 diluted antiserum against the proteins of interest, EsDscam, Esp-dorsal, EsT-dorsal, p-ERK, T-ERK, p-JNK, T-JNK, V5, FLAG, Myc, and actin, for 3 h in the same TBS solution after confirming these antibodies generated nonspecific signals. After washing three times with TBS, alkaline phosphatase-conjugated goat anti-rabbit/mouse IgG (1:10,000 diluted in TBS) was added; and following its 3-h incubation with the membrane, any unbound IgG was washed away. The membrane was then dipped into the reaction system and visualized in the dark for 5 min. Antibodies recognizing the phosphorylated forms and total protein of ERK and JNK were purchased from Cell Signaling Technology, and both total protein and antibody-binding sites shared high similarities with these proteins in E. sinensis. Antibodies recognizing the phosphorylated forms and total protein of dorsal were produced (WuXi Apptec Co., Wuxi, China) against the specific dorsal protein in E. sinensis. Antibodies recognizing the total protein of V5, FLAG, Myc, and actin were purchased from Abcam (Cambridge, UK). All images were collected by the Odyssey CLx (LI-COR, Lincoln, NE).

Immunocytochemical staining
To examine ERK phosphorylation and dorsal translocation in crab hemocytes, immunocytochemical staining was applied. Briefly, pretreated hemocytes were blocked with 3% bovine serum albumin (BSA) for 30 min at 37°C and then incubated overnight at 4°C with an EsDorsal antibody or a p-ERK antibody (1:100 in blocking buffer). After washing with PBS, these hemocytes were incubated with 3% BSA for 10 min, after which the second antibody, goat anti-mouse Alexa Fluor 488 (1:1000 dilution in 3% BSA), was added. The reaction was kept in the dark for 1 h at 37°C and then washed with PBS. Finally, the hemocytes were stained with Ј,6-diamidino-2-phenylindole dihydrochloride (DAPI, AnaSpec Inc., San Jose) for 10 min at room temperature, washed again, and observed under a Revolve Hybrid Microscope (Echo, San Diego).

Co-immunoprecipitation
For co-IP experiments, human embryonic kidney cells (HEK 293T) were seeded in 60-mm dishes (1 ϫ 10 7 cells/dish) overnight and transfected with a total of 10 mg of empty plasmid or various expression plasmids. At 48 h post-transfection, the medium was carefully removed, and the cell monolayer washed twice with ice-cold PBS. Then the cells were lysed with 500 ml of RIPA Lysis Buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycholate, sodium orthovanadate, sodium fluoride, EDTA, and leupeptin) (Beyotime, Beijing, China) containing a protease mixture (Yeasen, Shanghai, China). Lysates were centrifuged at 14,000 ϫ g for 15 min, and the ensuing supernatant was transferred to a fresh tube and precipitated for 2 h at 4°C with 30 ml of either an anti-FLAG, anti-V5, or anti-Myc affinity gel (Biotool, Houston, TX). The affinity gel used was washed with cold TBS four times and eluted by boiling for 10 min with TBS and a 6ϫ SDS loading buffer (2% SDS, 60 mM Tris-HCl, pH 6.8, 10% glycerol, 0.001% bromphenol blue, 0.33% ␤-mercaptoethanol). The cell lysates were also eluted with the same 6ϫ SDS loading buffer and likewise boiled. Proteins isolated from the beads and cell lysates were separated by SDS-PAGE and analyzed by Western blotting using the indicated antibodies. To test the binding between ERK and Dock in crab, 500 l of hemocytes protein from E. sinensis was incubated with EsERK or EsDock antibody overnight at 4°C. Protein A/G magnetic beads were added and incu-

Dscam controls innate immunity in crab
bated to capture the antigen-antibody complexes for 3 h at 4°C. The beads were collected by centrifugation and washed three times with PBS. The mixture was then resuspended with SDS-PAGE loading buffer and boiled at 100°C for 5 min, subsequently centrifuged again, and analyzed by Western blotting. Purified rabbit IgG was used as a control. Odyssey CLx (LI-COR) instrument was used to collect all the images.

Statements
All authors agree with the submission of this study for publication. This work has not been published or submitted for publication elsewhere, either completely or in part, or in another form or language. No material has been reproduced from another source. Experiments using invertebrate animal were approved by local authorities.