Three of the four probiotic strains used in this study (B11, DM14, and D16) were previously isolated and evaluated as a single combination treatment (22). The DM14 and D16 isolates were identified as different Pseudoalteromonas spp., and B11 as an Epibacterium sp (22). The fourth strain (ASW1) was later isolated from larvae cultured in autoclaved seawater using methods outlined previously (22). Genetic identification of ASW1 was completed using 16s rRNA gene sequencing and the NCBI’s BLAST suite (23). The 16S rRNA sequence was submitted to GenBank (Accession OQ595186). ASW1, in combination with B11, DM14, and D16, was evaluated for efficiency in improving the survival of C. gigas larvae challenged with Vibrio coralliilyticus using well-plate assays as described by Madison et al. (22). To obtain isolates for larval assays, cells from glycerol stocks stored at -80°C were streaked on plates with seawater-based Luria-Bertani agar (LBSw; 10 g of tryptone, 5 g of yeast extract, and 15 g of agar, per liter of 10 µm filtered seawater) and incubated at 25°C for 48 hours. Colonies were grown in 5 mL of LBSw broth (10 g of tryptone, 5 g of yeast extract per liter of filtered seawater) at 25°C and agitated at 40 RPM on a roller drum (New Brunswick TC-7; New Brunswick Scientific, Enfield, CT, USA) for 24 hours. Cultures were then washed twice with autoclaved seawater by centrifuging at 3900 x g for 5 min and resuspended in autoclaved seawater. The optical density was measured at 600 nm (OD₆₀₀) (Beckman DU 530, Beckman Coulter, Brea, CA, USA). Assuming that an OD₆₀₀ measurement of 1.0 equalled approximately 8.0 x 10⁸ CFU/mL, bacterial cultures were appropriately diluted to obtain target concentrations of 6.0 x 10³ CFU/mL for V. coralliilyticus and 7.5 x 10⁴ CFU/mL for each probiotic strain of the combination treatment used in larval challenge experiments.

Adult C. gigas were provided by the Molluscan Broodstock Program (MBP) at Oregon State University’s Hatfield Marine Science Centre. Oysters from genetically unrelated families were chosen for spawning. Oysters were held for two to four weeks in a conditioning system at 20°C to induce gametogenesis and continuously fed on a 50/50 (by cell concentration) mixed algal diet of Isochrysis galbana and Chaetoceros neogracile (total concentration of 10,000 cells/mL). When sexually mature, oysters were strip-spawned, and resulting gametes were fertilized following aseptic techniques (24). Larvae were hatched in sterile 250 mL Erlenmeyer flasks containing autoclaved seawater at 25°C, 32 ± 2 ppt salinity, and a pH of 8.2 ± 0.1. The seawater used to hatch larvae was pumped from the Yaquina Bay, Newport, passed through sand filters followed by 10-µm bag filters, and aerated with carbon-dioxide-stripped air overnight to adjust the pH before autoclaving. After the seawater was autoclaved, it was agitated and aerated on a shaker table (Benchmark, Sayreville, NJ, USA) at 80 RPM for 16 to 24 hours before adding fertilized eggs.

Our previous survival assays evaluated the effects of various probiotic combinations and concentrations in protecting developing C. gigas larvae against Vibrio coralliilyticus (22 and unpublished data). In this present study, the effects of the timing of probiotic additions on larval survival were evaluated by dosing larvae with the probiotics at different times post-fertilization of the eggs.

Following egg fertilization, embryos were transferred from a concentrated egg suspension in a single sterile 1 L beaker to sterile 250 mL flasks containing autoclaved seawater at a final concentration of 50 embryos per mL. This concentration allowed for the rearing of sufficient numbers of healthy normal larvae for the assays while considering handling constraints. The flasks were agitated and aerated on a shaker table at 50 RPM throughout the culturing the of larvae. Treatment groups were designated by the timing of probiotic addition. Probiotics were provided to larvae at 2 (2 hr PB Only), 6 (6 hr PB Only), 12 (12 hr PB Only), 18 (18 hr PB Only), or 24 hpf (24 hr PB Only). The four probiotics (ASW1, B11, DM14, D16) were added to larval cultures as a combination treatment at a concentration of 7.5 x 10⁴ CFU/mL per probiotic strain (22). Each of these five probiotic treatment groups was cultured in triplicate flasks. Twenty-four hours post-fertilization (hpf), D-larvae from each flask were poured onto a sterile 40 µm sieve, rinsed with autoclaved seawater, and resuspended in probiotic-free autoclaved seawater, except for treatment groups receiving probiotics at 24 hpf. This 24 hpf treatment represented a “standard” method used in our earlier studies (22), whereby the probiotics were added to the larval cultures at 24 hpf, but the larvae were not resuspended in probiotic-free autoclaved seawater before the addition of V. coralliilyticus. Therefore, by adding the probiotic combination at time points before 24 hpf, we investigated if the probiotic additions could influence the larvae at different points early in larval development. The transfer of larvae into probiotic-free autoclaved seawater also limits potential interactions between the probiotics and pathogen within the culture environment and focuses on interactions within the gut of the host where infection begins.

After transferring larvae from the 2, 6, 12, and 18 hpf treatments to probiotic-free autoclaved sea water, the larvae were resuspended at a concentration of 35 larvae/mL. Then one mL of larvae suspension was transferred to each well of a 24-well plate ( Figure 1A ). Larval concentrations were reduced in the well plates to better compensate for the addition of bacteria and subsequent influence on abiotic factors (22). Once transferred into well plates, the 24 hpf probiotic treatment groups received the probiotic addition.

At 48 hpf, probiotic-treated larvae were challenged with 6.0 x 10³ CFU/mL of V. coralliilyticus strain RE22 and incubated for 48 hours (these treatment groups were designated “2 hr PB + Vcor”, “6 hr PB + Vcor”, “12 hr PB + Vcor”, “18 hr PB + Vcor”, and “24 h PB + Vcor”). A positive control consisted of larvae infected with V. coralliilyticus alone without exposure to probiotics (Vcor Only). In addition, a negative control of larvae that received neither the probiotic treatment nor the pathogen was included (Larvae Only). Each treatment group consisted of six replicate wells ( Table 1 ). At 96 hpf, larvae were preserved with 10 µL of 10% phosphate-buffered formalin at a pH of 8. Well-plates were observed under a 40X objective inverted light microscope. Larvae were counted as either live or dead, with larvae classified as dead, having less than 90% of their tissue remaining in their shells (22).

One liter flasks containing 500 mL autoclaved seawater were stocked with 1 x 10⁴ embryos. This concentration was chosen to ensure that adequate amounts of live larvae were collected for RNA analysis (4). The flasks were agitated and aerated on a shaker table at 50 RPM throughout the culturing of the larvae. The experiment was conducted with four replicates per treatment group. The probiotic combination of ASW1, B11, DM14, and D16 was added to the embryos at 18 hpf (18 hr PB Only) at a total concentration of 3.0 x 10⁵ CFU/mL. Unlike in the survival assay ( Figure 1A ), the larvae cultures were not rinsed and added to probiotic-free autoclaved seawater to minimize experimental effects on transcriptional changes and potential contamination and to increase the timing accuracy of bacterial exposures across all treatment groups and replicates. Vibrio coralliilyticus was then added to probiotic-treated larvae at 48 hpf (18 hr PB + Vcor). The larvae of the positive control received no probiotics, but V. coralliilyticus was added at 48 hpf (Vcor Only) ( Figure 1B ). A “no exogenous bacteria” negative control of larvae receiving no probiotic additions or V. coralliilyticus (Larvae Only) was maintained throughout this experiment.

All treatment and control groups were sampled at 72 hpf to ensure sufficient numbers of live larvae for RNA analysis after exposure to V. coralliilyticus. Five thousand larvae per treatment group were rinsed onto a sterile 40-µm sieve with autoclaved seawater and then rinsed with 2 mL of RNAlater (Thermo Scientific, Waltham, MA, USA). The larvae were then transferred to a microcentrifuge tube containing 1.5 mL RNAlater. In addition, samples of up to 100 larvae were collected from each of the four replicates into 10 mL shell vials (Thermo Scientific, Waltham, MA, USA) and preserved with 10% phosphate-buffered formalin (pH of 8) to evaluate survival, as described above.

Total RNA was extracted from each sample and DNase treated using the Zymo Direct-Zol RNA Microprep kit (Zymo Research, Irvine, CA, USA) with the addition of Trizol reagent (Thermo Scientific, Waltham, MA, USA). Total RNA extracted from each sample was quantified using the Qubit Broad Range RNA kit (Thermo Scientific, Waltham, MA, USA). The quality of total RNA was assessed using Agilent 2100 BioAnalyzer (Agilent, Santa Clara, CA, USA) with the Agilent RNA 6000 Nano kit (Agilent, Santa Clara, CA, USA). Typically, one of the four replicates within each treatment group had poor quality or quantity of total RNA; therefore, the three replicates from each treatment group with the highest quality and quantity of RNA were identified and prepared for sequencing.

The first step in sequence library preparation was the isolation of mRNA from the total RNA using the NEBNext Poly(A) Magnetic Isolation Module (New England BioLabs, Ipswich, MA, USA). Subsequently, complementary DNA (cDNA) libraries were constructed using the NEBNext Ultra II RNA Library Prep kit for Illumina (New England BioLabs, Ipswich, MA, USA), per the manufacturer’s instructions. Complementary DNA library quantification, normalization, and sequencing were then completed by Oregon State University’s Center for Quantitative Life Sciences (Corvallis, OR). Libraries were sequenced as 100-bp single-end reads with the NextSeq 2000. The P2 flow cell was utilized, allowing for 400 million reads to be sequenced per run. Four sequencing runs were completed, resulting in a sequencing depth of 32 million to 44 million reads per sample ( Supplementary Table 1 ).

Raw sequencing reads were analysed for quality using FastQC 0.11.9 (25). Any identified poor-quality bases (quality scores lower than 20) and adapter sequences were removed using fastp (26). Processed reads were aligned to a C. gigas reference genome (version 1.0, Genbank GCA_902806645.1) via Hisat2 2.1.0 (27). Transcriptome assembly was performed using Stringtie (28) using default parameters. The overall alignment rate to the C. gigas reference genome using HISAT2 ranged from 45-80% for transcriptomes ( Supplementary Table 1 ). Differential gene expression analysis was performed by comparing transcript counts between the treatment groups (“18 hr PB Only”, “18 hr PB + Vcor”, and “Vcor Only”) and the control (Larvae Only) using DESeq2, which was used also to normalize libraries to control for differences in sequence depth and remove outliers (29). Transcripts displaying a log fold change of ≥ 2 or ≤ −2 and a Benjamini-Hochberg adjusted p-value <.05 were considered significantly differentially expressed. A principal component analysis (PCA) was performed to observe the level of variation between treatment groups due to their level of differentially expressed genes. For significance testing of the PCA, a PERMANOVA was completed using the adonis2 function (30). Differentially expressed genes were annotated using KEGG’s GhostKOALA (KEGG Orthology and Links Annotation) program (31). Defence-related genes were annotated using Uniprot (32).

The “Larvae Only” negative control was used to normalize larval mortalities that were unrelated to experimental treatments. Relative percent survival (RPS) was calculated as RPS = [1-(percent mortality of treatment group/percent mortality of untreated control group)] x 100. Relative percent survival values were arcsine square root transformed before analysis. Statistical analyses were conducted using R statistical software (Version 4.0.3, R Project for Statistical Computing). Normality was assessed using the Shapiro-Wilk test and Q-Q plots. The homogeneity of variance was assessed using Levene’s test.

Nonparametric methods were used if violations of normality and variance assumptions were observed. For multiple comparisons of treatment groups, the Kruskal-Wallis one-way ANOVA was conducted. When significant differences (P <.05) were found, Dunn’s test with the Benjamini-Hochberg correction was used for pairwise comparisons. Comparisons between the two treatment groups were conducted using the Mann-Whitney U test.

In evaluating ASW1 as a potential addition to the probiotic combination treatment, ASW1 did not result in any mortalities when added to larvae at 24 hpf ( Supplementary Table 2 ). Larvae that received ASW1 at 24 hpf and V. coralliilyticus at 48 hpf had an average relative percent survival of 76.9 ± 38.9%, which was 56.1% higher than that of larvae exposed to V. coralliilyticus alone ( Supplementary Table 2 ). A BLASTN suite search identified ASW1 as an Alteromonas sp. based on 16S rRNA sequencing. Specifically, Alteromonas oceani was identified as the best match with a 99% query coverage and 98.71% percent identity.

Additions of the probiotic combination treatment at 2, 6, 12, 18, and 24 hpf were evaluated for the ability to reduce larval mortalities due to V. coralliilyticus. The negative control of larvae (Larvae Only) that received no bacterial additions and was used to calculate the relative percent survival yielded a survival rate of 99.7%. The positive control receiving only V. coralliilyticus (Vcor Only) resulted in an average relative percent survival of 8.7 ± 26.7% ( Figure 2 ). When probiotics were added before pathogen exposure at 2, 6, 12, 18, or 24 hpf (PB + Vcor), the average relative percent survival significantly increased to 87.5 ± 18.9%, 89.3 ± 11.4%, 91.1 ± 12.9%, 85.9 ± 23.3%, and 83.6 ± 18.1%, respectively (P <.05) ( Figure 2 ; Supplementary Table 3 ). There was no significant difference in survival among the different timings of probiotic additions ( Supplementary Table 3 ); therefore, the 18-hour probiotic addition time was chosen as the focus for the RNA sequencing results to limit the possibility for changes in bacterial composition prior to the larvae being transferred into culture 24-well plates at 24 hpf.

Larvae infected with V. coralliilyticus without the probiotic supplementation (Vcor Only; positive control for disease) had 267 differentially expressed genes (DEGs) compared to larvae that received no bacterial additions (Larvae Only; negative control) ( Figure 3A ). One hundred and twenty-nine DEGs were annotated. When the probiotics were added to larvae at 18 hpf without any pathogen (18 hr PB Only), 535 DEGs were identified, including 227 annotated DEGs ( Figure 3A ). The same number, though not an identical set, of DEGs were identified when V. coralliilyticus was added in addition to the probiotics (18 hr PB + Vcor) compared to the negative control (Larvae Only) ( Figure 3A ) ( Supplementary Table 4 ). The “Vcor Only” treatment group had 125 total unique DEGs, 72 being annotated. Comparatively, the “18 hr PB Only” and “18 hr PB + Vcor” each had two unique annotated DEGs and two unique uncharacterized DEGs. Between all three treatment groups, a total of 141 DEGs were shared, 57 of them being annotated genes.

When only defence-related DEGs were assessed, 32 DEGs were identified in the “Vcor Only” control compared to the “Larvae Only” control ( Figure 3B ). In comparison, there were 44 defence-related DEGs in both the “18 hr PB Only” and “18 hr PB + Vcor” treatment groups, 30 of which were identical (shared between both of the probiotic supplementation treatments). Conversely, the “Vcor Only” treatment group had 19 defence-related DEGs that were uniquely expressed in that treatment ( Figure 3B ). Twelve defence-related DEGs were shared between all three treatment groups ( Supplementary Table 4 ).

Defence-related genes that were identified as significantly differentially expressed were organized into four categories: pattern recognition receptors, immune signalling, immune effectors, and other immune or inflammatory genes. These groups describe various components of the larva’s innate immune system that recognize foreign material and respond accordingly. For instance, recognition of bacteria is accomplished by pattern recognition receptors, which include both extra- and intracellular protein receptors. Subsequently, the immune signalling category includes genes coding for proteins that contribute to relaying a signal throughout various immune pathways involving toll-like signalling, nuclear factor- kappa B, mitogen-activated protein kinase, complement cascade, and others. Genes identified as immune effectors respond to upstream signalling and contribute to the defence response. Finally, the genes grouped into the “other” category may contribute to the defence response through other routes, such as apoptosis or phagocytosis.

The addition of ASW1 to a previously identified probiotic combination treatment of three bacterial isolates (22) further improved the relative percent survival of V. coralliilyticus-challenged C. gigas larvae compared to an infection control with no probiotics. While the relative percent survival between the three-strain and the four-strain combination was not statistically different ( Supplementary Table 3 ), the latter increased the survival of infected larvae after an exposure of only six hours.

The probiotic treatment may directly or indirectly affect immune priming, with the latter possibly resulting from the probiotics influencing the early development of the oyster gut microbiome. It has previously been observed that C. gigas larvae begin ingesting small particles around 22 hpf (unpublished data), which allows the larvae two hours to ingest the probiotic bacteria before the larvae are rinsed with autoclaved seawater.

Ultimately, more work is needed to better understand the effects of probiotic additions on the possible development of the oyster gut microbiome and to determine the role of these probiotic bacteria in stimulating immune processes. Regardless, this work highlights a potentially beneficial and practical tool for promoting the health of larvae in oyster hatcheries and provides evidence for potential explanatory mechanisms that can be experimentally tested.

The current study’s results reveal that there is a generalized response to all exogenous additions of bacteria, regardless of whether they involved probiotic or pathogenic microbes. In addition to this generalized response to Gram-negative bacteria, we observed specific gene expression responses to each different bacterial treatment (i.e., “Vcor only”, “PB only”, and “PB + Vcor”).

It is generally understood that the first step of the bivalve innate immune response to foreign materials is for pattern recognition receptors to recognize and bind bacterial cell wall components, including lipopolysaccharides and peptidoglycans (12). Accordingly, the expression of multiple pattern recognition receptors involved in toll-like signalling was influenced by bacterial exposure in the following ways: 1) larvae exposed to any exogenous bacteria decreased expression of TLR6; 2) larvae that received probiotics at 18 hpf, regardless of whether they were exposed to V. coralliilyticus or not, demonstrated an additional decreased expression of TLR2-2 and TLR4. The coordinated expression of these receptors is supported by previous reports that TLR2 cooperates with TLR6 (35) and heterodimerizes with TLR4 in response to Gram-negative bacterial exposure (36); however, our results contradict previous studies, which reported upregulation of said toll-like receptors when C. gigas were exposed to live bacterial components, including peptidoglycan and lipopolysaccharides (37–39). Furthermore, larvae exposed to pathogenic V. coralliilyticus without any probiotic addition (Vcor Only) decreased expression of TLR3. In contrast to TLR2-2, TLR4, and TLR6, TLR3 proteins are found intracellularly, localized within the endosome organelle where they recognize viral double-stranded DNA, signalling for it to be internalized and then transported to the lysosome (40). The reasons for decreased expressions of the TLR genes are unclear; however, possible explanations include the increased expression of genes that code for proteins responsible for the negative regulation of TLRs. For instance, the leucine-rich repeat-containing G-protein coupled receptor 4 (LGR4) protein is hypothesized to be a negative regulator of toll-like signalling and was found to have elevated expression in all three probiotic treatment groups (“18 hr PB Only”, “18 hr PB + Vcor”, and “Vcor Only”) ( Figure 5 ). In addition, larvae infected with V. coralliilyticus that did not receive any probiotic addition (Vcor Only) experienced an upregulation of the toll-like signalling negative regulator, tumour necrosis factor alpha-induced protein 3-like (TNFAIP3), compared to the larvae that did not receive any bacterial additions (Larvae Only). The differential expression of this protein seems unique to the larval response to V. coralliilyticus, as it was not differentially expressed in response to the probiotics. In contrast, both probiotic and pathogen exposure increased the expression of CAD, another negative regulator of toll-like receptors. The CAD protein works as a negative regulator of toll-like receptors by stimulating the expression of TNFAIP3 via CAD-dependent production of reactive oxygen species (ROS), leading to suppressed expression of toll-like receptors ( Figure 5 ). Previously, itaconic acid, a by-product of cis-aconitate decarboxylase, has been found to inhibit ROS produced by phagocytes, therefore, regulating the innate immune response (41, 42). However, whether itaconic acid was increased in response to bacterial additions in this study is unknown. Van Nguyen & Alfaroo (43) found that adult mussels infected with a V. coralliilyticus/neptunius-like isolate experienced significantly higher ROS levels than non-infected individuals 6, 18, and 60 hours post-infection but that itaconic acid levels did not increase until 60 hours post-infection. Therefore, it is possible that increased expression of itaconic acid could occur in infected larvae, but not before 60 hours after exposure to V. coralliilyticus (108 hours post-fertilization).

Ultimately, even though multiple toll-like receptors had a lower expression level in the bacteria-treated larvae, downstream signalling molecules whose activation is typically induced via toll-like receptors, including interleukin-17 (IL-17), had significantly increased expression. IL-17 transcripts are expressed in the gills of Mytilus galloprovincialis in response to infection by Vibrio splendidus, suggesting a contribution to mucosal immunity (44). Additionally, IL-17 is known to be directly responsive to bacterial LPS and contributes to the activation of both myeloid differentiation primary response 88 (MyD88) and the NF-kB pathway ( Figure 5 ). MyD88, found with elevated expression in larvae exposed to the probiotics, is directly stimulated by most toll-like receptors and plays a central role in the innate immune response, activating IL-1R associated kinases (IRAK) and subsequently the NF-kB and mitogen-activated protein kinase (MAPK) pathways (45). Therefore, MyD88 might be stimulated by long-term bacterial exposure as opposed to toll-like receptors whose expression may decline with continuous bacterial exposure (46).

In the absence of probiotics, infection by V. coralliilyticus caused an increase in expression of the 26S proteasome non-ATPase regulatory subunit 10 protein (PSMD10) and serine/threonine-protein kinase RIO3 (RIOK3), both of which have been found to inhibit NF-kB activation ( Figure 5 ). For example, PSMD10 has been hypothesized to retain NF-kB in the cytoplasm of cells, subsequently inhibiting NF-kB activity, whereas RIOK3 has been shown to inhibit TNF-alpha and caspase-10 induced activation of the NF-kB pathway (47, 48). However, when larvae were exposed to both the probiotics and V. coralliilyticus, neither the RIOK3 nor PSMD10 genes were found to be differentially expressed. Therefore, this result may be due to the probiotics directly inhibiting V. coralliilyticus growth, virulence, or both, consequently preventing the initial upregulation of these two genes. Alternatively, some of the numerous uncharacterized DEGs influenced by the probiotics may function as regulators of RIOK3 and PSMD10. Further progress in the characterization of the C. gigas genome will be required to evaluate this possibility.

In addition to immune signalling pathways, treatment with both probiotic and pathogenic bacteria influenced the expression of various inflammation and effector molecules, including mucins. Elevated Mucin C1 (Muc.C1) expression was seen in all larvae exposed to bacteria; however, larvae treated with probiotics had particularly elevated levels of Mucin 2 (Muc2) and Mucin 5AC (Muc5AC) expression ( Figure 5 ). These results correspond with previously identified localizations and proposed functions of each mucin. For example, Muc.C1 has been recognized as a transmembrane protein in many cell types, including human immune cells (49). Furthermore, Muc.C1 has been observed to behave as a binding site for bacterial pathogens, such as Pseudomonas aeruginosa. Following the binding of bacteria, Muc.C1 contributes to an anti-inflammatory response characterized by the inhibition of toll-like signalling (50). The resulting decreased expression of toll-like receptors identified in this study agrees with Muc.C1’s role within the anti-inflammatory process. In contrast, Muc2 and Muc5AC are mainly localized to the digestive tract and contribute to maintaining a mucosal barrier that will protect tissue surfaces and aid in removing unwanted material, including bacteria (49, 51, 52).

Aside from immune effectors, several proteins involved in apoptosis experienced differential expression due to bacterial exposure ( Figure 5 ). For example, the cell-death-inducing p53 (CDIP53) protein had elevated expression in all bacteria-exposed larvae; this protein induces cellular apoptosis through the intrinsic pathway (53). In contrast, probiotic exposure seems to prevent apoptosis as there was a reduced expression of GTPase immune-associated proteins 4 and 7 (GIMAP4, GIMAP7) (apoptosis accelerators) and cell-death abnormality protein 1-like (a protein responsible for enabling phagocytes to engulf apoptotic cells) (54–57). Reduced apoptosis regulators suggest a reduced need for cell death as a defence mechanism. These results align with previous studies that evaluated the impacts of probiotic bacteria on bivalve larvae, including the Eastern oyster, Crassostrea virginica (4). Additionally, larvae infected with V. coralliilyticus displayed increased expression of various baculoviral IAP repeat-containing proteins (BIRCs), which are negative regulators of apoptosis. BIRCs present anti-apoptotic features by mediating multiple caspases, including caspase-3, 7, and 9 (58). The lack of self-induced apoptosis by V. coralliilyticus-infected cells might allow the pathogen to replicate and overwhelm this defence mechanism.