Use of Microarray Technology to Improve DNA Vaccines in Fish Aquaculture – The Rhabdoviral Model

Fish rhabdoviral infections, specially those caused by novirhabdoviruses, can be tackled with commertial DNA vaccines such as the one against infectious haematopoietic necrosis virus (IHNV) (Salonius et al., 2007). Nevertheless, rhabdoviral diseases continue to pose a considerable threat to aquaculture because a number of practical problems regarding vaccination remain unsolved (for instance, mass delivery methods for small fish and requirements for safer vectors). Furthermore, some fish rhabdoviroses appear to be spreading to wild-type species.


Introduction
Fish rhabdoviral infections, specially those caused by novirhabdoviruses, can be tackled with commertial DNA vaccines such as the one against infectious haematopoietic necrosis virus (IHNV) (Salonius et al., 2007). Nevertheless, rhabdoviral diseases continue to pose a considerable threat to aquaculture because a number of practical problems regarding vaccination remain unsolved (for instance, mass delivery methods for small fish and requirements for safer vectors). Furthermore, some fish rhabdoviroses appear to be spreading to wild-type species.
Theoretically, efficient DNA vaccines could be used for any fish pathogen, such as other viruses (nodaviruses and orthomyxoviruses, for instance), bacteria and parasites. However, in practice, many fish DNA vaccines do not perform satisfactorily in most other pathogens than novirhabdoviruses. Therefore, fish novirhabdoviral vaccines are suitable models in which to study why similar DNA vaccines have not been successfully developed for other viruses (Gomez-Casado et al., 2011;Kurath, 2008). Those studies include the use of microarrays.
Most of our present knowledge about the factors that affect DNA vaccination efficacy in fish (vaccine dosage, delivery route, water quality, host species/size, time to challenge, severity of challenge, viral strain, etc) derives from work on fish rhabdoviral models (Kurath, 2008). Thus, the first fish DNA vaccines against IHNV were reported in 1996 (Anderson et al., 1996a;Anderson et al., 1996b) and against viral haemorrhagic septicemia virus (VHSV) in 1998 (Lorenzen et al., 1998). In 2005, Vical-Aqua Health Ltd. of Canada (Novartis APEXwww.intechopen.com Health and Environment in Aquaculture 252 IHN) received authorization to commertialize an IHNV DNA vaccine (Salonius et al., 2007). In support of their licensing, millions of salmon were vaccinated in British Columbia in 2004 and 2005. However, there are no reports on the efficacy of this vaccine against natural viral challenges (Kurath, 2008;Salonius et al., 2007).
When injected intramuscularly in each fish, plasmid-based G glycoprotein-coding rhabdoviral vaccines induce long-term (months) specific immunity, preceded by an early (4-8 days) non-specific protective response (Kim et al., 2000;Lorenzen, 2000;Lorenzen et al., 2002). Non-specific short-term protective immunity results from the induction of interferon-mx and related genes, while specific long-term protection may have this effect as a result of the induction of G glycoprotein gene-specific antibody or cellular responses (Kurath et al., 2007). However, most changes in gene expression that occur with resistance mechanisms in short-term and long-term immunity are not fully understood (Goetz & MacKenzie, 2008). Furthermore, more basic knowledge on mucosal immunity is required to move rhabdoviral DNA vaccines from the laboratory into the field, as existing vaccines still require either intramuscular injection in individual fish or stronger (adjuvanted) immune responses to facilitate mass delivery methods, such as those using oral- (delasHeras et al., 2010;Tian et al., 2008) or ultrasound-aided (Fernandez-Alonso et al., 2001) immunization. Studies using microarrays could greatly contribute to furthering this basic knowledge (Secombes, 2008).
Theoretically, for best performance an optimal vaccination should mimic viral infection steps such as entry and replication. For instance, since the entry of rhabdoviruses would be first detected by cellular membrane toll-like receptors (TLRs) through the G glycoprotein and their later cytoplasmic replication by endosomal TLRs through dsRNA intermediates, the question arises as to whether DNA vaccines should include not only the G glycoprotein gene but also dsRNA intermediates (ie.: RNA hairpins). Again, new data obtained from microarrays could shed some light on these possibilities.
In this context, the recent availability of fish microarrays , which allow the expression profiling of thousands of genes simultaneously, has provided new opportunities to further study fish immunological responses in several rhabdovirus/fish models.
Expressed sequence tag (EST)-based microarrays of the Japanese flounder, trout, salmon and zebrafish have been used in gene-discovery efforts. These studies included infections with IHNV (MacKenzie et al., 2008;Purcell et al., 2006a), VHSV (Byon et al., 2005;Byon et al., 2006;Encinas et al., 2010) and hirame rhabdovirus (HRV) (Yasuike et al., 2007) (Table  1). However, no studies have reported on the largest microarrays that have recently become available, such as the ~ 32 K cDNA of salmonids  and the ~ 37 K 60-mer oligos of trout (Salem et al., 2008), most probably due to the complexity of the interpretation of the data. 1 cDNA HRV-infected Leukocytes (Aoki et al., 1999) (Nam et al., 2000) European flounder Platichthys flesus 3 cDNA GENIPOL (Diab et al., 2008) Atlantic halibut In this review we focus on the data published on the use of microarrays for the identification of rhabdoviral-induced genes with properties that make them candidate adjuvants for the improvement of fish DNA vaccines.
Since the most effective vaccinations are obtained after infections with live or attenuated pathogens, several PAMPs from a unique pathogen (such as external glycoproteins and internally synthesized dsRNA/glycoproteins in rhabdoviruses) simultaneously stimulate several TLRs. In contrast, dead recombinant protein subunits and antigenic genes contain fewer PAMPs than live/attenuated pathogens. Nevertheless, single PAMPs have also been used to immunize against live pathogens, mostly with the help of adjuvants to replace the missing PAMPs. Therefore, the purpose of vaccine adjuvants is to increase the immune responses of otherwise weak individual PAMPs.
Most adjuvants in mammals are believed to target professional antigen-presenting cells, such as tissue DCs (De Gregorio et al., 2009;Lambrecht et al., 2009). The expression patterns of pro-inflammatory genes such as cytokines, chemokines, MHC and co-estimulatory molecules are altered in adjuvant-targeted DCs ( Figure 1). Subsequently, maturing DCs migrate to lymph nodes and activate naive CD4+ (helper) and CD8+ (cytotoxic) T cells to produce antigen-specific antibodies, cytotoxic cells, antimicrobial peptides and regulatory cytokines (Craig et al., 2009;Longhi et al., 2009;Manicassamy & Pulendran, 2009;Secombes, 2008). DCs also process PAMPs into peptides for presentation onto major histocompatibility (mhc) molecules to T cell receptors (tcr). Thus DCs are crucial for both adjuvant effects and innate/adaptive immune responses ( Figure 1).
Although most PAMP-derived vaccine adjuvants act through TLRs on mammalian DCs (Figure 1), other internal adjuvants, such as hmgb1 released from lysed cells, exert their action through cell damage molecules (Lambrecht et al., 2009). Artifitial TLR-independent adjuvants, such as those derived from particulate compounds administered together with mammalian vaccines (mineral salts, liposomes, microparticles, saponins, and emulsions) either increase antigen persistence or uptake by DCs. Traditionally, vaccine adjuvants have been empirically identified as enhancers of antibody responses to a co-administered antigen. However, new adjuvant candidates have also been found among molecules of the signalling cascades of DC activation. According to a recent review (Secombes, 2008), the molecules with potential capacity to act as fish vaccine adjuvants might be found among: i) cytokine/chemokine molecules; ii) co-stimulatory cluster differentiation (cd) antigen receptors; and iii) blocking molecules, which might inhibit negative regulators. Microarray analysis of rhabdoviral fish immunizations have identified some of these molecules, as it will be reviewed here.

Microarrays in the study of the flatfish/HRV/VHSV models
Traditional sequencing, annotation and estimation of frequencies of each rhabdovirallyinduced transcript in flatfish, is one of the strategies designed to identify genes transcribed after rhabdoviral infections (pathogen-induced gene approach) (Aoki et al., 2011). Thus, the first attempts to identify HRV-induced genes were made in the Japanese flounder Paralichthys olivaceus by sequencing 300-596 expressed sequence tag (EST) clones from leukocytes 2-5 days after infection. The frequencies of each EST were estimated within a short 1 to 10 range (Aoki et al., 1999;Nam et al., 2000).  Gregorio et al., 2009;Manicassamy & Pulendran, 2009;Secombes, 2008). External incoming or internally synthesized rhabdoviral molecules (dsRNA, glycoprotein G, other viral proteins, etc) activate dendritic cells (DCs). These recognize rhabdoviral molecules and are activated either through toll-like receptors (TLRs) or cytokines produced by other cells (monocytes, granulocytes, macrophages, mast cells, natural killer cells, stromal cells, muscle cells, etc). Each combination of rhabdoviral molecules induces simultaneous stimulation of DCs to induce the expression of secreted cytokines and costimulatory membrane cds. The induced membrane cds, together with other signals (blue), induce differentiation of cd4+ cells to T helper cells (Th1, Th2, Th17 and/or Threg). Each differentiated Th cell produces a series of cytokines (red), which are required to make antibodies, cytotoxic lymphocytes, antimicrobial peptides and molecules involved in the regulation of other Ths. Theoretically, any of the up-or down-regulatory molecules that increase defensive responses could be candidate molecular adjuvants for vaccines.
The new mass-sequencing technologies, such as those offered by the Genome Sequencer FLX (454 Life Sciences, Branford, CT, USA), Illumina Solexa (Illumina Inc., San Diego, CA, USA) or ABI SOLiD (Applied Biosystems, Foster City, CA, USA), could improve the pathogen-induced transcript frequency-estimation strategy. Thus the mass-sequencing technologies produce millions of sequences per run, facilitating significant statistical data for the quantitation of each sequence frequency. However correct annotation of such a mass of new sequences continues to be a problem (Goetz & MacKenzie, 2008). For instance, of 58 million cDNA sequences of ~ 100 bp from largemouth bass, only 31391 unique sequences could be annotated (Garcia-Reyero et al., 2008). Although, the recent production of longer sequence sizes (200-400 bp), will facilitate their annotation, comparison of transcripts from many samples by this ultra-high-throughput sequencing technology is still not economically feasible. Massive sequencing could be used as a first approach, while a more focused microarray developed with selected genes could then be used for quantification of larger numbers of samples (Goetz & MacKenzie, 2008).
Japanese flounder EST-derived cDNA microarrays were applied to in vitro kidney cell cultures 3-6 h after HRV infection (Kurobe et al., 2005). The number of expressed transcripts changed in 20.8 % of the genes after HRV infection. The 91 immune-related genes of the microarray were preliminarily categorized into 8 clusters on the basis of their known pattern of gene expression. After 3 h of HRV infection, several genes included in the chemotaxis, apoptosis, cell growth and antigen-presenting clusters were increased while the expression of some genes, including mx, decreased. Among the genes of unknown function that changed after HRV infection, 13 showed a similar response profile to that of the genes of known function mentioned above. This observation may be indicative of their association.
Improved versions of Japanese flounder EST-derived cDNA microarrays (779 spots containing 228 immune-related genes) were used for in vivo differential gene expression after intramuscular injection of DNA vaccines containing the G gene of VHSV (Byon et al., 2006) and/or HRV (Yasuike et al., 2007). The differential expression of their transcripts was studied in kidney tissue 1, 3, 7 and 21 days after vaccination. The greatest number of differentially expressed genes ( Figure 2) was observed 3 days after injection (91.4 % were increased, of which 31 % were known genes). Genes with increased expression/transcription include those related to the non-specific immune responses, such as tnf, il1r, ccr, and mx, transcription factors, and even a few genes associated with the late specific antibody response, such as cd20. Many interferon-inducible genes including mx and interferon regulatory genes were the most strongly induced genes 3 and 7 days after injection. The expression of a number of unknown genes was also increased (Aoki et al., 2011). Among these, the LB3(8) gene increased a maximum of 56-fold 3 days after infection and then remained increased during one week (Byon et al., 2005).
Later versions of the Japanese flounder EST-derived cDNA microarrays of up to 1187 unique flounder ESTs (691 identified genes) were then used to compare the injection of recombinant G protein (non-protective) with the G gene (protective) (Byon et al., 2006). A number of IFN-related genes (including the unknown LB3(8)) and mx increased 7 days after injection, thereby confirming the observations made in previous studies using reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR) (Acosta et al., 2005;Robertsen, 2008). Further studies included differential gene expression in kidneys from Japanese flounder injected with the HRV G gene (protective) in comparison with the N gene  (Encinas et al., 2010). ■, VHSV infection-byimmersion of zebrafish and expression in internal organs (head kidney, liver and spleeen) (Encinas et al., 2010). □, Injection of trout with the IHNV G gene and expression in muscle tissue with p>0.01 (Purcell et al., 2006b). ∆, injection of Japanese flounder with the HRV G gene and expression in head kidney (Yasuike et al., 2007). ○, Injection of Japanese flounder with the VHSV G gene and expression in head kidney (Byon et al., 2006).
(non protective). Results confirmed that the IFN-inducible genes, LB3(8) and mx, were also increased 7 days after vaccination but only when the G gene was used (Yasuike et al., 2007). Furthermore, it was shown that the LB3(8) gene has an homologous domain to that of a mammal IFN-inducible protein. Thus, this gene is an example of how new genes involved in rhabdoviral immunization can be discovered by the microarray approach.
However, in all the series of experiments on flounder commented above, only transcripts from pooled organs from 3-5 fish were compared. Biological replicates were not reported and therefore statistical biological variation could not be estimated. Furthermore, the number of genes in the microarrays were relatively small and their collected data has not been deposited in any known banks to allow for independent or comparative analysis. Nevertheless, two main conclusions can be drawn from these experiments. Firstly, the largest number of differentially expressed kidney genes after fish rhabdoviral (VHSV or HRV) immunization-by-injection occurs 2-3 days after vaccination ( Figure 2) and, secondly, IFN-induced gene responses are stimulated after 3-7 days (Byon et al., 2005;Byon et al., 2006;Kurobe et al., 2005).
In these earliest experimentations, microarrays based on cDNAs (100-500-mer) rather than oligos (60-70-mer) were used. Because one of the greatest concerns with cDNA arrays is cross-hybridization between similar genes or between repeated elements of different genes as a result of the pseudotetraploidy of many fish, the use of oligo microarrays would increase specificity . However, in contrast to cDNA microarrays, oligo microarrays have a poorer performance when used for other related species. The current tendency appears to favour the use of the former. Thus, by using oligo microarrays, the printing layouts, total number of sequences and number of sequence replicates can be modified to meet any formats. Furthermore, oligo microarrays do not required maintenance of collections of bacterial clones coding for cDNAs. In addition, oligonucleotides can be selected and used in a range of various formats suitable for each experimental design. Improved sensitivity, increased dynamic range, lower variance and fewer outliers have also been demonstrated when using oligo rather than cDNA microarrays. Correlation between cDNA and oligo microarray results has been demonstrated, although some discrepancies have also been reported (Salem et al., 2008). High density oligo microarrays have been developed in other fish such as salmonids , rainbow trout (Dios et al., 2008;Salem et al., 2008) and zebrafish (Cameron et al., 2005) (Table 1).

Microarrays in the study of the salmonid/IHNV/VHSV models
Large-scale genomic projects for salmon have been initiated by groups in Canada, the USA, the UK, Norway and France. As a result there are many physical and genetic maps, large collections of ESTs and a growing number of genomic sequences and derived microarrays. Thus three projects have developed salmonid microarrays. The first salmonid 16K cDNA microaarray appeared in 2004. This array was developed by the Genomic Research on Atlantic Salmon Project (GRASP) (von Schalburg et al., 2005a) and led to the most recent 32K cDNA  and the first 5K oligo DNA of 70-mer  microarrays. The high sequence similarity (~ 86 %) between salmonids (9 genera and 68 species) indicates that cDNA microarrays may be suitable for studies involving any member of this fish family. Transcriptome Analysis of Important Traits of Salmon (TRAITS) and the Norwegian Salmon Genome Project (SGP) also developed a 16K cDNA microarray (http://www.abdn.sfirc/salmon) based on two independent collections of their bacterial clones kept in ARK, Genomics Facility at Roslin Institute, UK and at SGP Genetics Laboratory at the University of Oslo, respectively. The TRAITS-SGP cDNA array was obtained from ESTs from 15 tissues (pathogen-induced libraries, trait-specific substractive EST, starvation-induced libraries, diet-response libraries, smoltification-response libraries and well-known genes). This array was conceived as a preliminary tool to develop an oligo microarray for routine health monitoring of Atlantic salmon. The first results found some artefactual expression patterns caused by cross-hybridization of similar transcripts and underlined the greater relevance of biological over technical replicates (Taggart et al., 2008).
By using all the tentative consensus sequences available at the Rainbow Trout Gene Index (RTGI) data base, a 37K oligo microarray was constructed (Salem et al., 2008), which is available at Agilent (design number 16271, deposited on the GEO with the GPL6018 number). The new rainbow trout (Oncorhynchus mykiss) high-density, oligonucleotide microarray was developed using 37394 specific 60-mer oligonucleotide probes assembled from 244984 ESTs from 12 tissues (http://compbio.dfci.harvard.edu/tgi/cgibin/tgi/gimain.pl?gudb¼r_trout). The specificity of each probe was checked for possible non-specific mRNA cross-hybridization by comparing all individual probes with all rainbow trout transcriptome sequences. Approximately 91 % of the sequences used for this microarray matched a previously annotated sequence in the GenBank.
Few attempts have been made to use these microarrays to study the rhabdoviral immunization of salmonids. In homozygous trout, the 16 K cDNA GRASP microarray was used to profile 7-day muscle transcripts after intramuscular injection of the IHNV G gene (Purcell et al., 2006a;von Schalburg et al., 2005b). After immunization, irf3, mx, vig1, and vig8 transcripts were increased (Purcell et al., 2006a). Genes associated with antigenpresenting cells, lymphocytes, leukocytes, inflammation, antigen presentation, and interferon pathways were also augmented. The increased levels of transcripts associated with type I IFN pathways in systemic organs (gill, spleen and kidney) were corroborated by RT-qPCR. These observations confirmed that, when intramuscularly injected, the hostexpressed viral G gene induces a systemic non-specific type 1 IFN innate immune response.
Using a 1.8K cDNA salmonid microarray, comparison of infection-by-injection with IHNV and attenuated IHNV in rainbow trout after 1 and 3 days showed an IHNV-dependent change in differential transcription in kidney towards adaptive immunity genes . Thus, the rapid spread of the IHNV infection inhibited tnfa, mhc1, and several other gene markers while favouring mhc2 and ig responses. The molecular mechanism for the development of late (months) specific cytotoxic T or B cell-mediated humoral responses has not been addressed by means of microarrays (Kurath, 2008;Kurath et al., 2006).
More recently, trout families with low (32% survival following challenge) and high susceptibility to VHSV (18% survival following challenge) were infected with VHSV by bath exposure and transcriptional data from internal organs were analyzed with the 16K GRASP microarray from day 3 post-challenge (Jorgensen et al., 2011). In total, 939 genes were differentially expressed between infected and non-infected fish. The genes increased in infected fish belonged to the following categories: stress and defence response, NFkappaB signal transduction, response to non-self, antigen processing and presentation, and proteasome complexes. Most were also increased among the 642 differentially expressed genes in the low-susceptibility trout family but not among the 556 differentially expressed genes in the high-susceptibility family. These results suggest that the innate immune system of internal organs plays a crucial role in eliciting an effective immune response to VHSV infection in rainbow trout (Jorgensen et al., 2011).

Microarrays in the study of the VHSV/zebrafish model
The zebrafish Danio rerio is one of the most suitable models in which to carry out microarray studies because, compared to other fish, its genome sequence is one of the most advanced. Furthermore, ~ 40 K annotated quantitative polymerase chain reaction (qPCR) arrays and annotated oligo microarrays are available. In addition, large-scale experimentation with zebrafish is easier than with other fish models and zebrafish are susceptible to several viruses, most of these belonging to the fish rhabdoviral family (Sullivan & Kim, 2008). Of these, VHSV (Novoa et al., 2006) was chosen in a recent study using microarrays (Encinas et al., 2010) over IHNV , snake-head rhabdovirus (SHRV) (Phelan et al., 2005) and spring viremia of carp (SVC) (Sanders et al., 2003), because only in the VHSV/zebrafish model have infection-by-immersion (the natural route of infection) and successful vaccination been described (Novoa et al., 2006).
Damage and epithelial cell death immediately after VHSV infection in the surface portals of entry of these viruses, such as the fins (Harmache et al., 2006), should alert surrounding cells to promote epithelial cell division to replace dead cells, recruit inflammatory cells to the infection site, and send signals to internal immune organs. However, viral-induced signals to inhibit the most relevant host responses have also been detected. Detection of natural early responses may contribute to identifying vaccine adjuvants. Thus, the expression of the 636 immune-related transcripts that were increased after VHSV infection, as estimated by hybridization to oligo microarrays (confirmed by RT-qPCR arrays), was higher in fins than in organs. In contrast, the number of decreased transcripts was higher in organs than in fins ( Figure 3). Therefore, an upregulated response of immune-related genes was greatest in fin tissues, while a downregulated response was most detected in the internal organ responses. The latter might be targets of viral inhibitory signals early after infection (Encinas et al., 2010). These results showed that 2 days after infection-by-immersion, VHSV had not yet caused an strong response from zebrafish internal organs, which contrasts with reports in other fish at later times after infection-by-injection (such as ifn1, mx, il1b, tnfa, etc) (Acosta et al., 2006;Samuel, 2001;Tafalla et al., 2007;Tafalla et al., 2005) or infection-by-immersion (Jorgensen et al., 2011;Zhang et al., 2009).
The zebrafish are refractory to rhabdoviral infection-by-immersion at high temperatures or without acclimatation to low temperatures with IHNV (LaPatra et al., 2000), VHSV (Novoa et al., 2006) or SVC (Sanders et al., 2003). Therefore, a temperature-dependent response mechanism(s) that inhibits rhabdoviral infection and spread may occur. While these preliminary findings shed some light on the earliest effects of VHSV infection at the molecular level, some of the new immune-related genes identified might be suitable candidate adjuvants for fish vaccines (Rajcani et al., 2005;Secombes, 2008).

Comparative microarray study of fish/rhabdoviral models
To best detect innate immune responses, early times after rhabdoviral infection should be studied. Thus, according to the data obtained from flatfish, salmonid and zebrafish studies, the maximal number of >2-fold differentially expressed genes in microarrays was detected 2-3 days after rhabdoviral infection (Aoki et al., 1999;Byon et al., 2005;Byon et al., 2006;Kurobe et al., 2005;MacKenzie et al., 2008;Nam et al., 2000;Purcell et al., 2006a;von Schalburg et al., 2005b;Yasuike et al., 2007) (Figure 2). Table 2 shows a list of some of the differentially transcribed immune-related genes detected using microarrays after rhabdoviral immunization, independently of immunization mechanism, fish species, rhabdoviruses and organs. Among the gene list, ifn and irf -related genes were expected to be present; however, their presence was scarce. As with many other Use of Microarray Technology to Improve DNA Vaccines in Fish Aquaculture -The Rhabdoviral Model 261 Fig. 3. Comparison of the number of differentially expressed genes 2 days after VHSV infection in zebrafish fins and organs (modified from Encinas et al., 2010). Groups of 10 zebrafish (Danio rerio) were infected with VHSV. To perform the hybridization to oligo microarrays, 636 selected immune-related sequences from the 44K microarray (Appligene) were used to analyze mRNA levels. Normalization was made with the ribosomal phosphoprotein p0 (rplp0) gene. Folds were calculated by the following formula: mRNA levels in VHSV-infected zebrafish / mRNA levels in non-infected zebrafish. The total number of genes increasing (+) or decreasing (-) expression > 2-fold and p<0.05 was represented as the percentage of the total number of immune-related genes assayed (n = 4). Percentages were calculated by the formula, 100 x the number of differentially expressed genes / total number of immune-related genes assayed (Encinas et al., 2010).
viruses and host species, an increase in ifn1 expression is one of the first responses to the injection of any DNA vaccine (Acosta et al., 2006;Samuel, 2001) and to rhabdoviral infections (Samuel, 2001;Theofilopoulos et al., 2005).
Transcripts encoding several forms of il17 were detected as differentially expressed only in one of the studies using microarrays (Encinas et al., 2010). Il17 is produced by T helper 17 (Th17) cells ( Figure 1) and acts together with il22 on epithelial cells (Trifari et al., 2009) and other types of skin cells to trigger il1b and tnf. These responses induce neutrophil/macrophage recruitment in epithelial surfaces (Qiu et al., 2009), stimulate keratinocytes (Nograles et al., 2008) and increase the production of antimicrobial peptides such as hepcidin (hamp1) and defensin ß like-2 (defbl2). Both hamp1 and defbl2 were found

www.intechopen.com
Use of Microarray Technology to Improve DNA Vaccines in Fish Aquaculture -The Rhabdoviral Model 263 to be differentially expressed in some studies (Table 2) (Liang et al., 2006;Yu & Gaffen, 2008).
Co-stimulatory cell membrane cluster differentiation antigens (cd) molecules responsible for the antigen-presenting cell interactions with T cells were not differentially expressed, except those belonging to mhc1 and mhc2 molecules (Table 2). Nevertheless, the use of cds as vaccine adjuvants has been described for the cd154 gene in zebrafish (Gong et al., 2009); the reasoning being that co-expression of cds with antigen in the same cell might accelerate specific immune responses. Thus, specific antibody responses obtained using cd154 and the pMCV1.4 plasmid coding for the G gene of VHSV (Ruiz et al., 2008) were increased 3-4-fold with respect to the plasmid alone (Gong et al., 2009).
TLRs, immunoglobulin chains, TNF -related molecules, and high mobility proteins were also found amongst the differentially expressed genes in several studies using microarrays and thus might deserve some consideration as adjuvant candidates (Table 2).

Future research
Microarray and mass sequencing technologies have opened up new avenues to analyze gene expression profiles. The data obtained by these technologies might facilitate the discovery of new immune-related genes (immunogenomics), clarify the molecular mechanisms of immunity and identify new candidates for vaccine adjuvants. Nevertheless, some problems remain in the application of these technologies to the amelioration of fish rhabdoviral vaccines. For instance there is a need to improve comparison of the data obtained from different models, to complete present gene annotations, to confirm transcriptional data with protein data and to develop mathematical models to facilitate interpretation of the abundant data.
In addition, the number of immune-related genes on zebrafish, trout, salmon, human and mice microarrays (Table 3), shows that more fish immune-related genes might have to be included in future microarray designs. The number of immune-related genes are still much lower in other cultured fish species (turbot, sea bream, sea bass, etc).  The use of microarray technology stems from the availability of genome, mRNA and EST (http://www.ncbi.nlm.nih.gov/dbEST) sequences to build representative annotated (geneidentified sequences) microarrays. For most commertial fish species, there is a lack of information on the annotated genome or known mRNA sequences and thus most microarrays used for these species mostly apply EST sequences. However, correct and complete annotation continues to be a bottleneck.
At present, it is quite difficult to compare data from distinct microarrays, even between salmonid microarrays such as GRASP, TRAITS-SGP and RTGI. We consider that reanalysis of the data deposited in data banks, for instance by the NCBI Gene Expression Omnibus (GEO) (http://www.ncbi.nlm.nih.gov/geo), to identify similar or identical genes should be undertaken.
For the most studied fish species, such as zebrafish, trout and salmon, identified mRNA sequences present in current GenBanks can alternatively be used to design focused microarrays enriched in some gene classes. For instance, the zebrafish and trout microarrays that are currently available omit a number of immune-related genes, the number of which vary depending on each microarray. A possible alternative to this problem in fish species for which abundant annotated mRNA sequences are held in GenBanks is to search for keyword-selected sequences to build up the corresponding microarrays.
One example of using some of the most obvious keywords corresponding to immunerelated genes for the trout O. mykiss and the zebrafish D. rerio is shown in Table 4. Thus, this table shows that the number of some immune-related genes extracted from GenBank data are 2-24-fold higher than their corresponding numbers in the microarrays of trout and zebrafish commertially available. Furthermore, trout and zebrafish 2-4K microarrays designed on unique sequences selected from GenBank immune-related genes (using ~50 keywords in GenBanks) are enriched an average of 2-3-fold in immune-related sequences with respect to the more general 37-44K commertial alternatives (data not shown). The design of smaller, focused (ie: immune-related) microarrays based on existing GenBank sequences could contribute to making the experiments less expensive and their results easier to interpret.
Although the advent of mass sequencing technologies might soon change this scenario, correct annotation will still require a considerable research effort for most fish species.
Care must be taken not to over interpret differential transcript gene expression. Thus, some rhabdoviral-induced changes might involve protein cleavages (complement c3) and/or protein post-translation modifications (hmgb1) in which transcriptional control may not be essential. Although in most reports some of the microarray results were confirmed using RT-qPCR, true confirmation would require estimation of its corresponding protein levels by parallel proteomic studies. Thus, although the differential expression of some fin proteins  Table 4. Comparison of some immune-related genes found in commertially available microarrays with those obtained from GenBank sequences. GenBank at http://www.ncbi.nlm.nih.goc/nuccore (transferrin, hemopexin, annexin, ATP binding, alpha actin, and kinesin) show a parallel variation with their transcript levels, in most of them, the changes in the differential expression of proteins do not correlate with their corresponding transcript changes (Encinas et al., 2010). This observation suggests that regulation of their expression is not at the transcriptional level, at least in that study. Although correlation of gene and protein expression has been found in some plants (Gallardo et al., 2007;Joosen et al., 2007), most studies found no correlation, including a recent report on individual E. coli cells (Taniguchi et al., 2010). Correlation values comparing gene/protein expression levels in several systems are consistently very low (Hack, 2004), suggesting that mRNA levels are poor indicators of the expression of their corresponding protein. Therefore, the study of mRNA levels is justified only when protein levels cannot be detected by the proteomic approach because of their low concentrations or short lives.
Finally, mathematical modelling of microarray data may shed light on gene changes and be useful for testing new hypotheses. From the first symposium held on 2003 (Petrovsky et al., 2003), some progress has been reported on the use of mathematical modelling for early response genes (Lawrence et al., 2007), whole immune responses (Ahmed & Hashish, 2006;Kalita et al., 2006), immunity to infectious diseases, including microarray data (Morel et al., 2006) and future perspectives (Li et al., 2009;Ta'asan & Gandlin, 2009). Mathematical modelling is expected to develop further since there are few other alternatives available to interpret the massive amount of information generated by microarrays.

Conclusions
Novirhabdoviroses are among the few fish viral diseases for which efficacious DNA vaccines are available; however, they continue to affect aquacultured fish worldwide.
Despite DNA vacines being commertial in Canada, the actual method of delivery by fish-tofish intramuscular injection and safety concerns are the major bottle necks to wide www.intechopen.com Use of Microarray Technology to Improve DNA Vaccines in Fish Aquaculture -The Rhabdoviral Model 267 acceptance of DNA vaccination. In addition, a complete understanding of the molecular events induced after rhabdoviral fish infection and immunization may contribute to improving DNA vaccines not only for rhabdoviroses but also for other fish infections for which there are no current remedies.
Knowledge about infection, vaccination and adjuvant mechanisms in mammal models, together with high throughput genomic techniques, such as hybridization to microarrays (cDNA or oligo, wide or focused) and new massive sequencing technologies (largely unexplored in fish), offer the opportunity to gather a considerable amount of new transcriptional data in fish models.
Indeed, microarrays have already been used to quantify fish gene expression as well as to discover new genes involved in defense in several fish rhabdovirus models, such as flatfish, salmonid (salmon and trout) and zebrafish.
Genes that show increased transcription after infection (hypothetically signalling internal organs to react against the viral invasion) and also genes whose transcription is inhibited (possibly due to viral shut-off of critical host defences) might help researchers in their quest to identify new adjuvant candidates for fish vaccines.