Tenacibaculosis is a worldwide fish disease responsible for considerable farmed fish mortality events, but knowledge is lacking on the microbiome kinetics during infection and the concomitant host–pathogen interaction. The dual RNAseq method was chosen as provides unparalleled simultaneous data on the molecular features of the infection. It is particularly suitable for systems characterised by a massive pathogen burden with readily accessible material but for which cultures are not available [17, 63]. Here, we adapted this approach for tenacibaculosis in P. orbicularis fish skin samples with the goal for comprehensively assessing the genomic basis and kinetics of infection as well as associated resistance mechanisms.
Infection modulates a wide range of innate and adaptive host immune effectors.
Bath exposition of T. maritimum was highly efficient at inducing tenacibaculosis in juvenile orbicular batfish. The low survival rate and kinetics of infection support what is usually observed for other fish species [16, 20, 64]. Infected fish, sampled at the peak of infection (24 hpi), show large skin lesions characteristic of tenacibaculosis together with a high cortisol concentration in their scales [32]. Cortisol mediates changes in individual energy balance (e.g. mobilisation of energy stores, immunity, cognition, visual acuity or behaviour) [65, 66]. This initial cascade of physiological and behavioural changes enables the organism to cope with acute stressors by mobilising adequate bodily functions, while concurrently inhibiting non-essential functions (e.g. reproduction, digestion) [67]. Here, increasing cortisol level reflects a local stress response to an unfavourable environment and is most likely involved in triggering the fish rapid immune response [68].
As expected, fish immune response, especially the innate immune system is strongly solicited at 24 hpi. Infected fish show activation of acute inflammatory response, mainly through driver genes, including interleukin-8 (IL-8) [69], but also activation of pathogen recognition receptors (PRRs), chemokines and antimicrobial-related humoral effectors. For instance, infection triggers co-expression of the cascade Toll-like receptor 5 (TLR5) and Myeloid differentiation primary response protein (MyD88), as previously reported in bony fish during bacterial infection [70]. However, the diversity of fish immune actors combined with the relatively limited knowledge we have on specific effector functions significantly hampered the comprehensive understanding of the mechanisms involved in our non-model species. For instance, in parallel to the TLR5, several other TLRs show reduced expression in infected fish, including TLR2 type-1, TLR-8 and non-mammalian (‘fish-specific’) TLR21. Despite previous effort towards assessing diversity of TLR sequences, protein-specific function remains poorly known in teleosts [71]. Similar observations have be made for the complement system, specifically complement C3, a key component of the immune system involved in ‘complementing’ antibodies for bacterial cell killing [72], for which several isoforms are reported in the Platax transcriptome. The different isoforms here have divergent patterns of expression (both up- and down-regulated in the infected24h group), which support previously observed differences in target surface binding specificities [73].
Innate immune response is generally tightly linked to cellular homeostasis regulation and precedes adaptive immune response. The ability of the fish to maintain cellular homeostasis during infection is of primary importance when facing infection, and mechanisms include redox, biological quality control (autophagy) as well as ion level maintenance [74, 75]; all of which were found to be affected in Platax. For instance, infected fish largely activate effectors of iron ion homeostasis. Iron, albeit largely present in the environment, is poorly accessible to organisms and iron sequestration and maintenance is a major mechanism developed by the host to limit pathogen growth as well as to regulate macrophage cytokine production [76]. In parallel, infected24 individuals activated the (1->3)-beta-D-glucan binding process and thus contribute to the body of literature revealing receptor capacities in fish and pathway conservation (through PRRs, C-lectin and/or TLRs) across vertebrates and invertebrates [77, 78]. Indeed, supplementation of β-glucan stimulates immune response in fish and increases resistance of the host to viruses and other pathogens (probably by reducing bacterial adhesion through lectin binding [79]); it therefore represents a promising immunostimulant for aquaculture [78, 80]. Effects of β-glucan vary depending on species, exposure time, source of glucan, organs and markers monitored [77, 81] and further studies will be needed to evaluate its potential at the production scale. Nevertheless, β-glucan is also relevant in bridging inflammatory response and activation / differentiation of T-cells in the adaptive immune response [82].
The adaptive immune response was also modulated at 24 hpi and its fine-tuned orchestration offers the opportunity to separate the preferential immune paths that can fight against T. maritimum infection. We identified several hallmarks of differentiated T-cells, indicative of the specialisation of the adaptive immune response to T. maritimum infection. Among the main driver genes of the response to infection in Platax, we noted a reduced expression of foxp3 and gata-3 in infected24h. Both transcription factors are important regulators of the fate of Naïve CD4 + naïve T-cells, encouraging differentiation to T-regulatory (Treg) [83] and T-helper 2 (Th2) cells [56], respectively. Similarly, we showed reduced expression of T-bet transcription factors, a hallmark of Th1 cells [56]. Inversely, infected fish seem to activate Th17 cell differentiation, as suggested by simultaneous activation of the signal transducer and activator of transcription (STAT1-alpha/beta) and cytokine IL-17 [83]. Th17 cells are mainly dedicated to controlling bacterial and fungal entry [84]. In line with previous work [56, 85], our results suggest a complex orchestration of T-cell differentiation via antigen communication and associated cytokine regulatory network in Platax during T. maritimum infection. However, we cannot rule out the possibility that changes in transcript abundance might also be indicative of cell migration. Complementary approaches, including cellular imaging [86] would clarify the presence of T-reg cells in fish and improve our knowledge of their regulation.
The genomic bases of resistance in P. orbicularis.
Despite profound activation of the immune system response, most of the fish failed to contain the infection. At 96 hpi, less than 25% of the infected fish had survived the bacterial challenge. These surviving individuals did not display any skin lesions, suggesting that they resisted the T. maritimum penetration and/or limited initial bacterial adhesion. Considering the high bacterial concentration in the tanks during infection and the severity of the mortality event, it is very unlikely that resistant fish totally escaped contact with the pathogen. Indeed, T. maritimum were present in resistant96h but not in control96h individuals (see next section for details). Fish were thus able to maintain the integrity of their first barrier against pathogens, so we hypothesise that their differences in gene activity difference would reveal specific immune actors present and/or efficient enough (timing of regulation) to inhibit pathogen multiplication and entry in resistant fish [16]. We show that PRRs, specifically a C-type lectin receptor, was up-regulated in resistant96h, together with a T-cell receptor and Low affinity immunoglobulin gamma Fc receptor II-like, while fibronectin-coding genes were down-regulated. These gene products are known for binding, agglutinating and neutralising bacteria [87], as well as triggering humoral immune response [88] or providing extracellular structure for pathogen adhesion through fibronectin-binding proteins [89, 90]. Our experimental design did not permit us to tell whether there was a basal difference in expression in resistant96h (genomic basis of resistance per se) or if the difference at 96 hpi was the result of a delayed adaptive immune response (timing of gene expression). Nonetheless, in catfish, lectin expression makes it possible to differentiate families resistant and susceptible to Flavobacterium colummnare, another gram-negative bacteria of the Flavobacteriaceae family [91]. Further longitudinal studies monitoring gene expression of resistant fish throughout the entire infection (including before the challenge) should prove useful in identifying resistance-specific responses to infection. Ideally, these studies should also simultaneously look at different immune-specific organ and tissue compartments and integrate genome-scale genetic variants (not limited to coding regions) to infer putative genetic bases of resistance.
Microbiome dynamics and host–pathogen communication.
At 24 hpi, the microbiome community was dramatically affected by the over-dominance of T. maritimum. Abundance of T. maritimum was evident from metabarcoding data and contributed to significantly reducing species richness in infected24h fish. We went further and compared expression of T. maritimum in vivo (during infection) compared to in vitro, with the hypothesis that key drivers of pathogenicity would be called upon to enable the bacteria to thrive and break host defence barriers. There are at least two major challenges that T. maritimum must to overcome to successfully infect the host: Pathogens need 1) to compete (at the intra and interspecific levels) for resources to metabolise from the local environment, and 2) to resist the host immune responses and stressful conditions. During infection, T. maritimum enhances its glucan catabolic activity. Although this might only reflect differences due to changes in the local environmental conditions (host mucus and skin) and/or resource availability, it also reveals some major mechanisms explaining the success of T. maritimum at growing on fish skin. Among the genes involved in glucan catabolism, we report several key components linking alternative food and mineral supplies and putative virulence-related functions, such as several genes involved in specifically degrading and uptaking sialoglycan and ions, mainly iron [19]. Sialidase activity explains why Capnocytophaga canimorsus bursts when in contact with host cells as this allows the pathogen to mobilise sugar directly from host phagocytes [92]. Similarly, the tonB-coding gene, regularly reported as a gene relevant for pathogenicity, confers virulence on Edwardsiella XXXctalurid by making it possible to maintain growth in an iron-depleted medium [93]. In parallel, several stress resistance-related genes were activated during experimental infection, all of which are also involved in the antibiotic catalytic functions. These genes include katA and katG, coding for two catalase-peroxidases involved in resistance to reactive oxygen species (ROS) by detoxifying exogenous H2O2 produced by host macrophages as a defence mechanism [94]. Obviously, the identification of virulence-related genes cannot be limited to those differentiating in vitro versus in vivo infectious status and other actors might be involved in making T. maritimum pathogenic. For instance, siderophore-coding genes are constitutively expressed in vitro or during experimental infection. These genes are a determining factor of host–pathogen and pathogen–pathogen interactions in the so-called ‘race for iron’ [95–97], which suggests that T. maritimum is highly efficient at mobilising iron independently of the local environment.
Finally, we mostly explored expression level variations in the light of an exclusive interplay between the host and T. maritimum, which might be effective considering the over-dominance of T. maritimum in fish mucus. However, most of the infection systems report several pathogen co-occurrences and the presence and/or activity of other opportunistic pathogens that might also play an important role in host fate [15]. In infected fish, we found that so-called ‘opportunistic’ bacteria were relatively largely represented. Opportunistic bacteria, which include V. harveyi, are known for their pathogenicity to fish. V. harveyi is a ubiquitous bacterium and one of the most common pathogens inducing major disease outbreaks in fish farming [60]. The enrichment of vibrio ASVs in resistant96h mucosal communities and the absence of obvious associated physiological changes (cortisol, mortality, skin integrity) in this group, suggest that V. harveyi alone is not sufficient to induce mortality in Platax under our specific experimental conditions and associated bacterial burden.