Transcriptome analysis of Aeromonas hydrophila infected hybrid sturgeon (Huso dauricus×Acipenser schrenckii)

The hybrid sturgeon (Huso dauricus × Acipenser schrenckii) is an economically important species in China. With the increasing aquaculture of hybrid sturgeon, the bacterial diseases are a great concern of the industry. In this study, de novo sequencing was used to compare the difference in transcriptome in spleen of the infected and mock infected sturgeon with Aeromonas hydrophila. Among 187,244 unigenes obtained, 87,887 unigenes were annotated and 1,147 unigenes were associated with immune responses genes. Comparative expression analysis indicated that 2,723 differently expressed genes between the infected and mock-infected group were identified, including 1,420 up-regulated and 1,303 down-regulated genes. 283 differently expressed anti-bacterial immune related genes were scrutinized, including 168 up-regulated and 115 down-regulated genes. Ten of the differently expressed genes were further validated by qRT-PCR. In this study, toll like receptors (TLRs) pathway, NF-kappa B pathway, class A scavenger receptor pathway, phagocytosis pathway, mannose receptor pathway and complement pathway were shown to be up-regulated in Aeromonas hydrophila infected hybrid sturgeon. Additionally, 65,040 potential SSRs and 2,133,505 candidate SNPs were identified from the hybrid sturgeon spleen transcriptome. This study could provide an insight of host immune genes associated with bacterial infection in hybrid sturgeon.


Differential expression analysis. To identify gene expression changes between bacteria infected and mock
infected samples, FPKM method was used to calculate the expression levels of genes. 2,723 genes were found to be expressed differently between bacteria infected and mock infected groups. 1,420 genes were up-regulated and 1,303 genes were down-regulated with an FDR ≤0.001 and ratios larger than 2. The up-regulated genes (red spots), down-regulated genes (green spots) and no differential expression genes (blue spots) distribution trends were shown in Fig. 4. Among these differently expressed genes, 1,233 genes were assigned to at least one of the Nr, Nt, Swiss-Prot, KEGG and COG databases, in which 700 genes were up-regulated and 533 genes were down-regulated.
Differential expression immune-relevant genes analysis. In this study, 283 annotated genes, including 168 up-regulated genes and 115 down-regulated genes, were found to be involving in various anti-bacterial . KEGG categories of unigenes. All unigenes were annotated using KEGG Automatic Annotation Server for pathway information. The categories GIP and EIP stand for genetic information processing and environmental information processing, respectively. The X axis is KEGG pathway category and Y-axis written in roman indicates the number of unigenes in each category respectively. immune-relevant pathways, they were further grouped into 6 sub-categories as follows: pattern recognition genes, complement system, inflammatory cytokines and receptors, T-cell and B-cell antigen activation, antigen presenting and regulators, adapters, effectors and signal transducers, other genes related to immune cell response (    ). The internal control was β-actin, and the log2 fold change obtained from q-PCR was measured with the 2 −ΔΔCT method and normalized by the median expression of β-actin. Even though the folds of changes were not exactly the same, however they showed the identical up-regulated or down-regulated patterns of these ten genes in both assays. This demonstrated the reliability of RNA-seq results and indicates the necessity for further identification of immune relevant genes in sturgeon.

Discussion
The transcriptome represents the complete repertoire of RNA transcripts in a cell investigated. Recently, many reports revealed that the transcriptome analysis is a great tool in deciphering the functional complexity of the genome and in obtaining a better understanding of cellular activities in organisms, including growth, development, disease, and immune defense 21,25 . In this study, the transcriptome profiles of the spleen from Aeromonas hydrophila infected and mock infected hybrid sturgeons (Huso dauricus × Acipenser schrenckii) were analyzed and compared. A total of 187,244 unigenes were obtained with a mean length of 1056 bp. The length distribution showed that 62.63% of the unigenes were clustered in a group with 200-1000 bp in length and 7.56% of the unigenes were longer than 3000 bp, which was similar to previous transcriptomes of the Amur sturgeon spleen and liver that were based on Illumina sequencing 19,26 . Hence, the transcriptomes of the hybrid sturgeon spleens were effectively established and could be utilized for further analysis. There were 2,723 differential expressed genes between the two groups, including 1,420 up-regulated and 1,303 down-regulated genes. The spleen is an important immune organ in fishes, the previous comparative transcriptome of Amur sturgeon spleen identified 125 DEGs between Yersinia ruckeri infection and mock-infection groups and these DEGs could be divided into 16 immune-related KEGG pathways 26 . Therefore, we focused on genes related to immune responses after Aeromonas hydrophila infected. 283 anti-bacterial immune related genes were found to be differently expressed between the two groups were scrutinized and discussed, including 168 up-regulated and 115 down-regulated genes.
Pattern recognition genes. The pattern recognition receptors (PRRs) could recognize specific surface components of microorganisms 27 . The Toll like receptor (TLR) is one of the most important classes of PRRs that play crucial roles in initiating inflammatory responses and shaping adaptive immunity 28 . In mammals, the human genome contains 10 functional TLRs whereas the mouse genome contains 12 TLRs, and many more TLR genes are identified in several fish species (TRL18, 19,20,22,25,26) than in mammalian species due to the presence of duplicated TLRs and fish-specific TLRs 18,29-32 . In mammalian, TRL 1, 2, 4, 5, 6 and 10 are located on the cell membrane, while TRL 3, 7, 8 and 9 functions within the cytoplasm 33 . Activation of TLRs stimulates NF-kappa B pathway through MyD88-dependent pathway and finally induces pro-inflammatory cytokine (IL, tumor necrosis factor TNF) 33 . In this report, TRL 1-9, 13 and 22 were detected by transcriptome analysis, they belong to six TLR families (TLR1, TLR3, TLR4, TLR5, TLR7, and TLR11) found in vertebrate taxa. TRL 1-9 and 13 are normally found in mammals, while TLR22 is "fish-specific" TLR. Among them, TRL 2, 5 and 8 were up-regulated significantly in the spleen of Aeromonas hydrophila infected hybrid sturgeons, suggesting that the TLR pathway is activated in the infected fish. TRL5 was significantly up-regulated in Yersinia ruckeri -infected Amur sturgeon through transcriptome analysis 19 . Fish have both a membrane and a soluble form of TLR5 that senses bacterial flagellin. TLR5 is involved in recognizing bacterial flagellin and after binding, it triggers myeloid differentiation primary response gene 88 (MyD88)-dependent signaling pathway to induce pro-inflammatory cytokines 34 . It is noteworthy that Aeromonas hydrophila possesses flagellin, we speculated that TLR5 was up-regulated by triggering of flagellin in this study. Interestingly, in the present study, the majority of the DEGs associated with "NF-kappa B signaling pathway" were strongly induced in infected sturgeon. There are 13 genes up-regulated (IL-1β, IL8, TNF receptor 1, interleukin-1 receptor-associated kinase 1 (IRAK 1), TNF receptor associated factor (TRAF) 1/2, TRAF 2/5, TRAF 2/6, TRAF 2, TRAF 3, epidermal growth factor receptor kinase (Btk), prostaglandin H synthase (COX2), vascular cell adhesion protein (VCAM) 1, C-C motif chemokine 19), 4 genes down-regulated (TCR, BCR, IRAK4, IL8) and eighteen IL (1β, 6, 8, 11, 12β) genes up-regulated in the infected fish (Table 7). These results suggested that TLRs activate NF-kappa B pathway induced inflammatory response to defense against Aeromonas hydrophila infection in hybrid sturgeon. The TLR mechanisms are conserved from fish to mammals throughout vertebrate evolution. A putative draft of TLR signaling pathways in hybrid sturgeon based on the knowledge of TLR signaling in teleost fish was constructed (Fig. 8). Mannose receptor (MR), belonging to C-type lectin family, is also a class of PRRs. MR is a macrophage surface receptor that recognizes surface polysaccharides of a wide range of Gram-negative and Gram-positive bacteria, yeast, parasites and mycobacteria 35 . MR expression is modulated by immunoglobulin receptors, cytokines, pathogens and their products, which suggests that the expression of the MR may correlates with macrophage activation 36 . The MR is a 180 kDa transmembrane protein that has five domains: the cysteine-rich region, the fibronectin type II repeat domain, the eight tandem lectin-like carbohydrate recognition domains (CRDs), the transmembrane domain and the cytoplasmic carboxy-terminal domain 36 . The recognition of mannose and fucose is restricted to the CRDs, and the degree of homology between mouse and human CRDs ranges is 76-92% 36,37 . The cooperativity of the MR with TLRs has recently been described for variety of human pathogens 35 . MR also plays an important role in adaptive immunity, including antigen processing 36 . The role of MR has been described extensively in mammals, but rarely in fish. Here, two types of mannose receptor transcripts (C-type mannose receptor 1 and 2) were identified in fish 18,38 . The C-type mannose receptors 1 and 2 of large yellow croaker also have been characterized, which are up-regulated following bacterial infection 38 . DGE analysis showed that MR1 and 2 of hybrid sturgeon are up-regulated (Table 7). These results suggest mannose receptor pathway is activated in response to Aeromonas hydrophila infection in hybrid sturgeon. It is be interesting to follow the relationship between MR and TLRs expression following bacterial infection.
The class A scavenger receptor (cA-SR) is another important class of PRRs, and can recognize low-density lipoproteins and bacteria 39,40 . These receptors are extracellular glycoproteins, which mediate phagocytosis of negatively charged ligands 41 . The cA-SR family consists of five members: the scavenger receptor class A (SR-A), macrophage-associated receptor with collagenous structure (MARCO), scavenger receptor class A member 3 (SCARA3), scavenger receptor class A member 4 (SCARA4) and scavenger receptor class A member 5 (SCARA5). SR-A, MARCO and SCARA5 possess a scavenger receptor cysteine-rich (SRCR) domain, while SCARA3 and SCARA4 lack the domain. MARCO and SCARA3/5 were discovered in Chinese sturgeon 18 . Using DGE analysis, it found that eight MARCOs and three cA-SRs contained SRCR domain were up-regulated, which indicated the activation of the class A scavenger receptor pathway (Table 7). Interestingly, the DGE analysis showed that "phagosome" pathway, which is the downstream of scavenger receptor pathway, is the most induced pathway associated with Aeromonas hydrophila defense with 27 DGEs up-regulated and 30 DGEs down-regulated. The protein identification and concrete expression profile analysis of these 57 genes is shown in     of phagosome are involved in bacteria degradation, pathogens killing and the regulated processing of their proteins for antigen presentation 24 . These results suggest that the hybrid sturgeon cA-SRs could recognize bacteria, which then activated phagocytosis to defense against Aeromonas hydrophila infection including bacteria degradation and antigen presentation for further immune response. The SRCR domain may have a potential role in bacteria recognition, which is yet to be determined. These results could provide new insight into hybrid sturgeon anti-bacterial immunity. To clarify the functions of this pathway, other components need to be identified, and the interaction among these components needs to be explored as soon as possible.

Genen name Forward primer (5′-3′) Revers primer (5′-3′)
Complement system. The complement C is an important part of innate immune system, which facilitates the ability of antibodies and phagocytic cells to clear pathogens. The mammalian complement system has three different activation pathways which include classical, alternative and mannose-binding lectin. Activation of the classical pathway is triggered by C1 that binding to antibodies antigen complex on the surface of bacterial. The C1 complex consists of C1q, C1r and C1s, which are also found in Chinese sturgeon 18,42 . The lectin pathway is initiated by binding mannose-binding lectins (MBL), associated with MBL-associated serine proteases (MASP) to an array of carbohydrate groups on the surface of bacterial 42 . The alternative pathway is triggered spontaneously, and primarily depends on recognition of host-associated molecular patterns (HAMPs) 43 . In this study, C1q (components of classical pathway), complement factor B and D (component of the alternative pathway) were all up-regulated (Table 7), which suggest activation of classical pathway and alternative complement pathway following Aeromonas hydrophila infection. We also identified the complement receptor (CR), both C1q and CR were up-regulated in hybrid sturgeon following Aeromonas hydrophila infection (Table 7). This result was different from Yersinia. ruckeri infection of Amur sturgeon 19 .
Inflammatory cytokines and receptors. Cytokines, which are cell-signaling proteins involved in many physiological processes including the regulation of immune and inflammatory responses, could be divided into interferons (IFNs), interleukins (ILs), tumor necrosis factors (TNFs), colony stimulating factors and chemokines 44 . Interleukins (IL-1β, 6, 8, 11, 12β) and interleukin receptors (IL-1R2) were up-regulated after Aeromonas hydrophila infection, which indicated that the sturgeon employed these interleukins to defense against the infection. Chemokines are divided into four major subfamilies in mammals: CC, CXC, CX3C, and C, while only two groups (CC and CXC) have been identified in most teleost species 45 . CC and CXC chemokines (CCL3, CCL14, CCL19, CCL20, CCL21, CCL28, SCYA118, CXCL3, CXCL9-12, CXC14, and CXCB2) were identified in Chinese sturgeon 18 . CCL19, CCL21 and IL8 (also known as CXCL8) were up-regulated and CCL13 and CXC chemokine receptor 4 (CXCR4) were down-regulated in Yersinia. ruckeri-infected Amur sturgeon 19,26 . In this study, CCL4, CCL19 and IL8 were up-regulated while CXCL10, CXCR1 and CCR5 were down-regulated, suggesting the complex interaction of chemokine signaling pathway response to Aeromonas hydrophila infection of hybrid sturgeon. It is interesting to note that, the haematopoietic cytokines angiopoietin-1 46 was down-regulated and matrix metalloproteinase-10, 13, 19, 28 (MMP-10, 13, 19, 28) which is critical for the vessel formation 47,48 were up regulated. The similar results were observed in Flavobacterium columnare infected Mandarin fish 17 . These results imply the potential relationship between haematopoietic and vessel formation in Aeromonas hydrophila infection.

T-cell and B-cell antigen activation. T lymphocytes and B lymphocytes are the main cellular components
of the adaptive immune response. Fish adaptive immunity is relative primitive due to limited immunoglobulins and secondary lymphoid tissue necessary for adaptive immunity 49 . Recent, T and B cells receptors (TCR, BCR, CD3, CD4, and CD8), antigen restriction molecules (MHC I, MHC II, and DC-SIGN/CD209), co-stimulatory factors (CD80/86, CD83, CD154, and CD40) and immunoglobulins (IgM, IgD, and IgZ/T) have been identified in teleost fish, which provided evidence for the existence of adaptive immunity in fish 21 . In Chinese sturgeon,   TCRα/β/γ/δ, CD3ε/ζ, CD4, and CD8, CD40, CD83, and CD80/86-CD28/CTLA4, IgM, IgD and IgL were discovered 18 . The TCR, BCR, CD4, CD8 and immunoglobulins were rearranged under the RAG1 and RAG2 regulation during T cell and B cell development 50 . DGE analysis demonstrated that RAG1, TCRδ, TCRγ, CD3, CD4, Ig light chain, Ig heavy chain were significantly down-regulated in bacterial infected sturgeon, while non-functional variable lymphocyte receptor A (VLRA) was up-regulated. These results indicate the complex interaction of adaptive immune after Aeromonas hydrophila infection of hybrid sturgeon.

T-cell and B-cell Antigen Activation
Antigen processing and regulators. Antigen processing is an immunological process that prepares antigens for presentation to T cells of the immune system 19 . The key components of antigen processing and regulation were present, such as class I and class II MHC molecules, integrin α/β, TNF receptor (TNR). The CD8 + T cells recognize protein-derived peptides in association with MHC class I (MHC-I) molecules, whereas CD4 + T cells recognize peptides bound to MHC class II (MHC-II) molecules 51 . MHC class I was down-regulated after Yersinia. ruckeri infection of Amur sturgeon 19 , while MHC class Ia and MHC class IIβ were up-regulated after Aeromonas hydrophila infection (Table 7). These results imply that the regulation of antigen processing may be different between Aeromonas hydrophila infected hybrid sturgeon and Yersinia. ruckeri infected of Amur sturgeon.

Adapters, Effectors and Signal Transducers.
Cell signaling is a complex system of communication that regulates basic cellular activities and coordinates cell actions. Hundreds of proteins are involved in the many signaling pathways, some of them are important adaptors, effectors and signal transducers 17 . The key components of adapters, effectors and signal transducers of anti-bacterial immune pathways were present, such as TNF receptor-associated factor (TRAF), calmodulin and nuclear factor κB (NF κB). TRAF and NF κB are the adaptor protein, effectors and transcription factors in TLR pathway 21 . The calmodulin is versatile messenger that transduces calcium signals by binding calcium ions, and mediates many crucial processes in the immune responses during the infection, such as inflammation and apoptosis 52 . During the hybrid sturgeon was infected with Aeromonas hydrophila, these three genes were up-regulated, implying the triggering of TLRs and calcium signals. Other Genes Related to Immune Response. The heat shock proteins (Hsps) play a role in both innate and adaptive immunity, including (1) elicit T-cell response specific against antigenic peptide they chaperone; (2) modulate innate response that are independent of chaperoned peptides 53 . In the present study, down-regulation of Hsp70 was similar to Yersinia. ruckeri infected of Amur sturgeon19, indicating that Hsp might be regulated by bacterial pathogens. It is noticeable that the expression of hepcidin, ferritin and serotransferrin were up-regulated following Aeromonas hydrophila infection, these three genes were associated with iron transport. Iron is an essential element for the growth of fish and bacterial species. Bacteria snatch the iron from their hosts, while the hosts inhibit the bacteria growth by limiting their usage of iron 54 . Bacteria have evolved many strategies to compete iron with hosts, including releasing iron-binding molecules and scavenging iron from hemoglobin and transferring 54 .
The similar hepcidin up-regulation was also observed in Flavobacterium columnare infected Mandarin fish, which suggest that the fish produced high level of hepcidin to decrease iron level and limited the bacterial infection 17 .

Conclusions
Based on the transcriptome sequencing results in the present study, 213 DEGs from spleen of bacterial infected hybrid sturgeon were found to be associated with pattern recognition proteins, complement system, inflammatory cytokines and receptors, antigen presenting and regulators, adapters, effectors and signal transducers and other genes related to immune cell response. Selected genes are also verified by RT-PCR. TLRs pathway, NF-kappa B pathway, class A scavenger receptor pathway, phagocytosis pathway, mannose receptor pathway and complement pathway were up-regulated in Aeromonas hydrophila infected hybrid sturgeon. Moreover, SSRs and SNPs were identified in hybrid sturgeon spleen transcriptome, which would be helpful for genetic linkage and QTL analysis. This study shed significant light on the anti-bacterial immune system of hybrid sturgeon, and which could lead to better understanding of interactions between the pathogen and host.

Materials and Methods
Ethics statement. All  cfu/ml. Ten hybrid sturgeons in infected group were injected intraperitoneally (i.p.) with 0.1 ml of bacterial suspension (1.7 × 10 8 cfu/ml). Ten fish in mock infected group were injected (i.p.) with an equal volume of sterile DPBS. All the fish were returned into water tank after injection. At 7 h post-injection, the fish were anaesthetized by 0.05% MS-222 (Sigma, USA) when the infected sturgeons showed the symptom of abnormal swimming, ascites and cloacal hemorrhaging. The spleen from ten individual sturgeon from each group were collected and kept in liquid nitrogen for RNA extraction. The whole experiment was repeated twice, and two sets of bacterial infected spleens and the other two sets of mock infected spleens were collected. And they sequenced in separate lanes.
Total RNA isolation and cDNA library construction. Total RNA was isolated from spleen tissue using Trizol Reagent (Invitrogen, USA), and genomic DNA was removed by RNase-free DNaes I (Qiagen, German). These four RNA samples (two sets of bacterial infected and the other two sets of mock infected spleens) were sent to Beijing Genomics Institute-Shenzhen (BGI, Shenzhen, China) for the Illumina deep sequencing. RNA degradation and contamination were detected by 1.5% (w/v) agarose gels. The quality and quantity of RNA were measured using Agilent 2100 Bioanalyzer (Agilent Technologies, USA). To avoid amplification of the RNA samples, the amount of total RNA for each sample used was kept at 4.0 μg with the RNA integrity number (RIN) >8.0. Poly-A-containing mRNA was further sorted by oligo-dT-attached magnetic beads and fragmented into small pieces using divalent cations under elevated temperature. Cleaved RNA fragments were converted into first-strand cDNA using reverse transcriptase and random primers, followed by second-strand cDNA synthesis using DNA polymerase I and RNase H. After the end repair process and adapter ligation step, the products were purified and amplified to generate the final cDNA libraries using the TruSeq RNA sample preparation kit.
Illumina sequencing and transcriptome annotation. RNA deep sequencing was conducted using Illumina Hiseq2000 (paired reads, 90 bp). Raw reads were first cleaned by removing adaptor sequences and low quality sequences (Q > 20) using the Filter_fq program. The clean reads were assembled into contigs firstly, and then assembled into unigenes using the Trinity software as described for de novo transcriptome assembles without reference genome 55 . For homology annotation, the unigenes were compared with the NCBI non-redundant protein (Nr) database using the BLASTx algorithm, with a cut-off E value of ≤10 −5 . Meanwhile, the unigenes were compared with non-redundant nucleotide (Nt) database using the BLASTn algorithm, with a cut-off E value of ≤10 −5 . Gene Ontology (GO) terms were extracted from the best hits obtained from the BLASTx against the Nr database using Blast2GO 56 . These results were then sorted by GO categories using in-house Perl scripts. BLASTx was also used to align unique sequences to the Swiss-Prot database, Clusters of Orthologous Groups (COG) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (with the E value of 10 −5 ) to predict possible functional classifications and molecular pathways 57 .
Differently expressed genes identification. To identify the differential expression between infected and mock infected groups, fragments per kilobase of transcript per million fragments sequenced (FPKM) were used to normalized the gene expression levels 58 . The differential expression was analyzed by RSEM 59 and edgeR softwares 60 . For each gene, the p value was computed, and then Benjamini-Hochberg false discovery rate (FDR) was applied to correct the results for p value. The transcripts that were increased or decreased at an estimated absolute log 2 -fold change of ≥1 and FDR adjusted p value ≤ 0.001 were considered to be differently expressed. Then, GO and KEGG pathway enrichment of differential unigenes were analyzed. The repeated samples were analyzed and merged into one result by NOIseq software.
Detection of SSRs and SNPs. MicroSAtellite (MISA) (http://pgrc.ipk-gatersleben.de/misa/misa.html) was used to analyze the microsatellite (SSR) distribution 61 . The minimum number of repeats for SSR detection was six for di-nucleotide and five repeats for tri-, quad-, penta-, and hexa-nucleotides. Potential single nucleotide polymorphisms (SNPs) were detected using SAMtools 62 and VarScan 63 with the following criteria: (1) total coverage and the number of reads to cover a candidate SNP (>8 reads); (2) the base quality where base calls with low Phred quality (<25) were removed from the coverage; (3) frequency of mutated bases higher than 30% among all reads covering the position. Quantitative real-time PCR analysis. To validate the RNA-seq differential expression studies, ten genes differently expressed between the infected and mock infected groups, including Toll like receptor 5 (TLR5), complement C1q (C1), interleukin-1 beta (IL-1β), MHC class Ia chain (MHC Ia), MHC class II beta chain (MHC IIβ), cathelicidin-OH antimicrobial peptide (CAMP), ferritin, cell death activator-3 (CIDE-3), cathepsinS and heat shock protein 70a (Hsp70a) were selected for RT-qPCR assay, using the same RNA sample used for the RNA-seq sequencing. Beta-actin (β-actin) was included as an internal reference gene to normalize the variations of input total cDNA template among samples. Briefly, total RNA was converted into first-strand cDNA using the RevertAid First Strand cDNA Synthesis Kit (Fermentas, Thermo Fisher Scientific, USA) with oligo-dT primers. Real-time PCR amplification reactions were carried out in a final volume of 20 μ l, which contained 10 μl 2 × SYBR Green PCR Master Mix (Toyobo, Japan), 1 μl diluted cDNA template, and 0.4 nM each of the forward and reverse primers. PCR amplification was performed under the following: 95 °C for 30 s, 95 °C for 15 s, 60 °C for 20 s and 72 °C for 35 s; steps 2-4 were repeated for 40 cycles 64 . All primers used for real-time PCR are listed in Table 6. Each sample was analyzed in triplicates. The relative expression level of target genes was measured with the 2 −ΔΔCT method and normalized by the median expression of β-actin. Results are displayed as means ± S.D. and were compared using an unpaired sample t-test, a p ≤ 0.05 was considered to be significant.