Comparative proteomic profiling of newly acquired, virulent and attenuated Neoparamoeba perurans proteins associated with amoebic gill disease

The causative agent of amoebic gill disease, Neoparamoeba perurans is reported to lose virulence during prolonged in vitro maintenance. In this study, the impact of prolonged culture on N. perurans virulence and its proteome was investigated. Two isolates, attenuated and virulent, had their virulence assessed in an experimental trial using Atlantic salmon smolts and their bacterial community composition was evaluated by 16S rRNA Illumina MiSeq sequencing. Soluble proteins were isolated from three isolates: a newly acquired, virulent and attenuated N. perurans culture. Proteins were analysed using two-dimensional electrophoresis coupled with liquid chromatography tandem mass spectrometry (LC–MS/MS). The challenge trial using naïve smolts confirmed a loss in virulence in the attenuated N. perurans culture. A greater diversity of bacterial communities was found in the microbiome of the virulent isolate in contrast to a reduction in microbial community richness in the attenuated microbiome. A collated proteome database of N. perurans, Amoebozoa and four bacterial genera resulted in 24 proteins differentially expressed between the three cultures. The present LC–MS/MS results indicate protein synthesis, oxidative stress and immunomodulation are upregulated in a newly acquired N. perurans culture and future studies may exploit these protein identifications for therapeutic purposes in infected farmed fish.


Results
In vivo challenge. The control cohort remained negative for N. perurans for the duration of the trial, as confirmed by gill scoring, qPCR and histology. During gill scoring, both the control and attenuated cohort had no visible gross macroscopic lesions at 7 days post infection (dpi) and were assigned a gill score of 0, while two fish from the virulent cohort were gill score 1. One fish from the attenuated cohort and all fish from the virulent cohort sampled at 7 dpi were qPCR positive for N. perurans. Histological analysis revealed no pathology for fish sampled at 7 dpi. This observation is consistent with the early stage of the disease where gross gill pathology is not evident 32,33 , further strengthening the use of qPCR in identifying the presence of the parasite [32][33][34] . Gill scores > 1 were observed for each fish at 14 dpi for the virulent cohort, with qPCR analysis confirming the presence of N. perurans for each fish sampled. qPCR detected the presence of N. perurans in one fish sampled 14 dpi from the attenuated cohort, however no corresponding gills score or pathology was observed with this fish. Negative qPCR results were observed for each fish sampled in the attenuated cohort at 21 dpi. At 21 dpi, half the fish sampled from the virulent cohort were observed to have a gill score of 2 and the experiment was terminated. No AGD-like pathology was detected in the attenuated cohort (Fig. 1a). Histopathology for the virulent cohort revealed lamellar epithelium hyperplasia and fusion, with occasional formation of interlamellar lacunae (Fig. 1b).
16S Sequencing of attenuated and virulent N. perurans microbiome. Relative abundances of OTUS ( Fig. 2) were used to generate the resulting list of prominent taxa in both the attenuated and virulent N. perurans cultures. Gram negative bacteria were identified with Proteobacteria and Bacteroidetes being the dominant phyla in both cultures. A loss of microbial diversity and relative abundance is evident in the attenuated culture, presumably reflecting the lack of adaptation to prolonged culturing of these particular genera. This was most true for several identifications of Gammaproteobacteria in the virulent culture. Three independent identifications appeared only in the virulent culture suggesting the presence of this class warrants further investigation. An unknown Gammaproteobacteria had a higher relative abundance in the virulent culture and could potentially be implicated in AGD virulence. Differences in community structure were found in shifts of relative abundance of the species Thalassospira xiamenensis belonging to the class Proteobacteria from ~ 1% coverage in the virulent culture and ~ 70% coverage in the attenuated culture. The genus Vibrio was only identified in the virulent microbiome as was the phyla Chlamydiae, an interesting find due to both this genus and phylum's role causing marine finfish disease. Winogradskyella was also identified in both microbiomes, however in much lower relative abundance than other genera. The UniProt proteomes of three genera from this sequencing analysis were selected for creation of the 'bacterial' protein database used in the proteomic characterisation of the N. perurans cultures in this study. These genera were: Pseudoaltermonas, Fluviicola and Vibrio. Additionally, due to the large coverage of T. xiamenensis in the attenuated, this species was also included in the database. 2DE and protein identification. A total number of 98 spots identified by the Same Spots software were found to be statistically different in intensity, (p ≤ 0.05) across the three sets of replicate gels from the newly acquired, virulent and attenuated protein extracts. Eleven spots with a maximum fold change of ≥ 2 were chosen for further analysis using LC-MS (Fig. 3), resulting in the identification of 24 proteins (Supplementary Table S2), the majority derived from the N. perurans database (Table 1), 4 N. perurans proteins of which shared homology to proteins of other amoebozoa species (Acanthamoeba spp, Planoprotostelium spp) and 1 protein shared homology with Thalassospira xiamenensis, the gram negative bacterium largely evident in the attenuated microbiome analysis. Six proteins were not N. perurans associated, 3 proteins of which were bacterial and the remaining 3 proteins were from the Amoebozoa group. Thirteen proteins were exclusively assigned to N. perurans only, therefore the proteins identified in this study are predominantly amoebic in nature and furthermore, are from N. perurans. GeneBank IDs for N. perurans are provided in Supplementary Table S1 www.nature.com/scientificreports/ to be upregulated in the newly acquired culture, except for spot 69 which was more upregulated in the attenuated culture in comparison to both the virulent and newly acquired cultures. Protein expression was highest in the newly acquired culture, followed by the virulent and attenuated cultures from 10 of the 11 spots. Four of the protein spots were upregulated (p < 0.05) in the newly acquired culture followed by low expression in the attenuated culture and virulent cultures, however the change between the attenuated and virulent cultures was not statistically significant (p > 0.05). The identity of differentially expressed proteins is listed in Table 1, along with the corresponding intensity fold change and molecular weights. The function of these proteins can be broadly attributed to cellular proliferation, metabolism and immunomodulation. Some of the described proteins are involved in oxidative defence, plausibly protecting the parasite from the hostile host response.

Discussion
Attenuation of parasitic virulence has been attributed to long term cultivation in vitro that negatively impacts the organisms capacity to establish infection in the host model [35][36][37] . A complete loss of virulence has been previously reported during experimental in vivo trials using a clonal culture of N. perurans that was maintained in vitro for 3 years 19,38 . In the present study, a 3-year culture of N. perurans had a markedly reduced ability to establish AGD in a challenge trial using naïve salmon smolts, confirming attenuation of N. perurans virulence over time.
Validation of virulence retention and loss in cultures was achieved using gill scoring, histopathology, and qPCR of gill samples from the in vivo AGD virulence challenge trial. Gill scoring of six fish at each time point throughout the duration of the trial revealed no gill pathology for the attenuated cohort. In comparison, fish sampled in the virulent cohort were assigned cumulative gill scores at each www.nature.com/scientificreports/ time point until reaching the pre-determined humane endpoint. At 21 dpi three out of the six fish sampled were at a gill score of 2, resulting in the termination of the trial. No AGD-like pathology was detected in the control and attenuated cohort for the duration of the trial. Assessment of virulent cohort histology ( Fig. 1) revealed the presence of AGD pathology with infected gills displaying lamellar epithelium hyperplasia and fusion 13,39 . qPCR assessment revealed that the majority of fish sampled in the virulent cohort were positive for N. perurans. In the attenuated cohort, only one in six fish sampled at each of 7 and 14 dpi was positive for N. perurans and all fish tested in this cohort were negative for the parasite at 21 dpi. These results confirm that using a combination of histology and qPCR is an invaluable approach in diagnosing AGD 32 , particularly in the early stage of the disease. Due to the abundance of literature suggesting N. perurans microbiome may influence AGD progression 26,27,29,30 , 16S rRNA sequencing was employed to clarify the role of the microbiome in the virulent and attenuated isolates. Figure 2, clearly shows that time in prolonged culture results in microbial community shifts. A much broader and diverse group of bacteria were present within the virulent isolate. No gram-positive bacteria were identified in either isolate, in contrast to the study by Benedicenti, et al. 30 where both gram positive and gram negative bacteria were identified in their polyclonal and two clonal cultures. Several gammaproteobacterial classes were    www.nature.com/scientificreports/ found in the virulent isolate but were absent in the attenuated isolate. The presence of the genera Vibrio and the Simkaniaceae family warrants further investigation as these two microbes are causative agents of disease in marine shellfish and finfish aquaculture 40,41 . Characterisation of the virulence molecules involved in AGD was achieved by comparing the proteomes of several N. perurans cultures. The soluble proteomic profile of a 3-year, 1-year and 70-day old N. perurans culture was analysed using 2D PAGE coupled with LC-MS/MS. The genome of N. perurans has recently been sequenced (Botwright, personal correspondence), enabling the comprehensive proteomic studies of N. perurans presented here. A total of 11 spots were excised from the newly acquired 2D PAGE gel (Fig. 3) based on the statistical criteria of ANOVA (p ≤ 0.05) and maximum fold change of ≥ 2 using the newly acquired culture as a reference. These 11 spots resulted in 24 protein identifications from the collated databases of N. perurans (Table 1), Amoebozoa and a discrete number of gram-negative bacteria chosen from the microbiome sequencing (Supplementary Table S2). The newly acquired culture showed the highest expression for the excised proteins, excluding spot 69 which was more upregulated in the attenuated culture. This protein may be considered a biomarker for reduced virulence in N. perurans as its expression shows that is markedly reduced in the virulent and newly acquired cultures.
Three proteins, glyceraldehyde-3 phosphate dehydrogenase (spot 314), chaperone protein (spot 283), putative TonB dependent receptor (spot 69) were associated only with V. rumoiensis, Vibrio sp. C7 and P. haloplanktis respectively. The identity of the protein from spot 69 is aconitate hydratase AcnA, an iron sulphur regulatory protein, an essential enzyme in both the tricarboxylic acid cycle (TCA) and glyoxylate cycles and its presence is widely regarded as a biomarker for iron and redox stress 42 . Aconitate hydratase was identified from both N. perurans and the largely abundant T. xiamenensis (primarily of the attenuated microbiome) database therefore indicating this protein may be a bacterial protein derivative. The anaerobic bacterium, Escherichia coli maintains the use of isozyme AcnB in the TCA cycle, however upon growth in iron deficient and reactive oxygen species (ROS) stress, AcnB is deactivated and AcnA is expressed to relieve the metabolic block in the TCA cycle 43 . Aconitase AcnA elevated expression in the attenuated culture suggests long term cultured N. perurans withstands oxidative and/or iron deficiency stressors when isolated from the host. This is presumably from microbial community changes over time and plausibly as a result of non-optimal nutritional media. A putative TonB-dependent receptor from P. haloplanktis was also identified in this spot with a plausible role in signalling and siderophore activity. TonB -dependent receptors have been implicated in carbohydrate scavenging 44 , further suggesting N. perurans may face nutritional deficiencies during prolonged culture.
The proteins upregulated in the newly acquired culture can be attributed to cytoskeletal, oxidative and immunomodulatory roles. Numerous cytoskeleton associated proteins; actin cytoplasmic A3a isoform (spot 599), actin depolymerization factor-like-domain-containing protein (spot 487), tubulin beta chain (spot 314) and fragmin A (spot 283) were elevated in the newly acquired culture. Actin is a major cytoskeleton protein involved in motility, phagocytosis and tissue invasion in protozoan parasites [45][46][47] . Profilin (spot 582) and ADF-like-domain, an actin binding protein (ABP), were both found to be upregulated in the newly acquired culture. ABP's are recruited for the regulation of dynamic actin remodelling during phagocytosis in Entamoeba histolytica 48 , a key process relating to E. histolytica's pathogenesis. An allergic response to N. perurans has been postulated previously 49 and profilin is commonly regarded as an allergen protein in many pollen and foods capable of inducing the production of IgE and IgM 50 . It is therefore plausible to think that this protein may also induce a similar response in the gills of Atlantic salmon.
Song et al. 51 inoculated recombinant Acanthamoeba profilin intranasally in mice to determine the proteins potential for triggering allergic airway inflammation. Mice affected with the recombinant protein displayed classical signs of an allergic response; increased mucin production and hyperplasia of epithelial cells, accompanied by an elevated Th2 cytokine response. Apicomplexans require profilin for actin polymerisation and furthermore, profilin has been shown to be crucial in host tissue invasion, egress from the cell and stimulates host toll like receptors of the innate system 52 . Profilins role as an immunomodulator in AGD warrants extensive investigation based on its previous role as an immunogen in these protozoan and apicomplexan studies.
Elongation factor 1 alpha (EF1α) (spot 283) is vital for cytoskeleton integrity 53 however it can serve numerous canonical functions in the cellular milieu hence being coined a moonlighting protein 54 . Elongation factors have been considered a virulence factor in both parasitic kinetoplastids and pathogenic bacteria as immunodominant antigens 55 . Macrophage deactivation occurs when Leishmania donovani EF1α binds to the hosts Src homology 2 domain containing protein tyrosine phosphatase-1 (SHP-1) and additionally inhibits inducible nitric oxide expression 56 . The protein's potential as a potent vaccine antigen in L. donovani has been evaluated by Sabur et al. 55 who generated liposomal forms of leishmanial elongation factor 1 vaccine and both the recombinant and truncated vaccine provided long term immunity in BALB/c mice against visceral leishmaniasis. Elongation factor thermo unstable (EF-Tu) (spot 283) was also upregulated in the newly acquired culture, and its identity was found in both N. perurans and Pseudoaltermonas spp databases. Ribosomal subunits as well as EF-Tu typically have roles in translation and this is conceivably the case here for N. perurans, however EF-Tu has also been reported as a cell surface protein that facilitates host cell attachment by binding of fibronectin 57 and plasminogen 58 . Bacteria only express EF-Tu therefore, it appears N. perurans has either acquired this genetic material through lateral gene transfer from its commensal microbiome , a described occurrence associated with amoebae and bacteria 59 , or this protein is of bacterial origin entirely; a very plausible consideration due to the homologous EF-Tu in Pseudoaltermonas spp.
Proteins that support a function in oxidative stress were found to be upregulated in the newly acquired culture, specifically superoxide dismutase (SOD) (spot 487). A correlation between oxidative stress in hyperplastic lesions and late stage AGD has been shown for naïve smolts 60 . It has been previously reported that E. histolytica can directly interfere with host derived neutrophil ROS production and can also rely on a secondary ROS scavenging mechanism using Fe-superoxide dismutase's 61  www.nature.com/scientificreports/ gill score of 2 from Marcos-Lopez et al. 60 ., study highlights the localised oxidative stress in the gill and suggests N. perurans is capable of tolerating the hostile environment by maintaining the expression of its own SOD. Further protection from the host immune response is provided by heat shock proteins (HSP) that facilitate protein synthesis and simultaneously offer protective measures during episodes of oxidative, nutritional and heat stress in parasites 62 . The presence of HSP8 (spot 326) from N. perurans, HSP85 (spot 326) from Perkinsela spp. and HSP70 (spot 334) from P. pallidum suggests the role of oxidative protection is crucial in the newly acquired culture before adapatation slows its production in the virulent and attenuated cultures. A chaperone protein (spot 283) from Vibrio sp. C7 was also found to be upregulated in the newly acquired isolate, potenially playing a similar role of oxidative protection. Malate dehydrogenase (spot 348) of N. perurans and glyceraldehyde-3 phosphate (spot 314), exclusively identified as Vibrio rumoiensis, are key enzymes involved in the TCA cycle that were upregulated in the newly acquired culture, highlighting the increased contribution of metabolism for N. perurans and its microbiome in the maintenance of virulence mechanisms.
Other metabolic proteins found to be upregulated in the newly acquired culture included citrate synthase (spot 283), ATP synthase (spot 511), ADP-ribosylation factor 4 and 1 (spot 511), and lipoxygenase (lox) homology domain-containing protein 1-like isoform X2 (spot 487). Citrate synthase is the first enzyme in the TCA cycle and ATP synthase is an essential nanomotor in the electron transport chain of the TCA cycle and thus, is responsible for the production of cellular ATP. Increased ATP production may be a virulence requisite for N. perurans as evidenced by the upregulation of these enzymes in the newly acquired culture with increased energy expenditure in the parasite due to replication and locomotion. Exploitation of ATP synthase is being considered as a drug target due to the novel subunits governing this protein in T. gondii 63 . This finding may offer a therapeutic avenue if N. perurans nanomotor shares homology with this ATP synthase structure. ADP-ribosylation factor 4 and 1 displayed a higher expression in the newly acquired culture with these GTP-binding proteins playing a role in protein tracking and cellular signalling 64 . Lipoxygenase constitutes a family of iron containing enzymes that serve as dioxygenases catalysts. In the nosocomial pathogen, Pseudomonas aeruginosa, lipoxygenase has been linked to increasing the bacterium's persistence in the host, destroying the lipid membrane of epithelial cells 65 and inhibiting the expression of chemokines 66 .
The ATP binding cassette (ABC) membrane transporter proteins (spot 198) in protozoan parasites were best described as exporters of parasite waste, until T. gondii was shown to import host derived cholesterol 67 , which therefore suggests that transport may be bidirectional. They also represent an aspect of multi drug resistance in parasites 68 , therefore their characterisation is imperative in how future therapeutics may be effectively designed.
A characteristic of protozoan parasitic pathogenesis is the variation in antigenic representation to fulfil their parasitic lifecycle and gene management strategies remain at the forefront of this process. Chromatin is central to this role and histones are the building blocks of the nucleosomes that create chromatin structure. The elevation of histone H2B (spot 511) seen here in the newly acquired culture may highlight the underestimated role histones undertake in regulating transcription dependent on the parasites environment 69 . The most peculiar protein upregulated in the newly acquired culture was peptidase C53 family protein (spot 198), a protease homologue from pestiviruses, that would be typically located in the extracellular matrix as an excreted protease. The presence of a protease family in N. perurans however is not surprising as proteases assist in promoting pathogenesis through tissue destruction, invasion and facilitate immune evasion 70 . The identification of this protein in the cytoplasmic fraction suggests this protein remained in its pro-peptide form when it was isolated, a form seen when the protein is translocating towards the plasma membrane for extracellular secretion and release.
In conclusion, these findings suggest the attenuated N. perurans culture has a markedly reduced ability to establish disease in the host model as revealed by qPCR, histopathology and gill score. Microbial characterised by 16S rRNA gene sequencing successfully identified differences in diversity and relative abundances between the attenuated and virulent isolate. 2DE and LC-MS/MS successfully revealed the identity of 11 spots, of which 24 proteins were characterised and discussed here. Proteins in ten of the 11 spots were found to be significantly upregulated in the newly acquired culture and proteins in one spot, spot 69 containing bacterial protein derivatives; aconitate hydratase and TonB-dependent receptor proteins, was found to be significantly upregulated in the attenuated cohort. Many of the differential proteins followed a linear trend of expression, starting with the highest expression in the newly acquired, followed by intermediate expression in the virulent culture and the lowest expression in the attenuated culture. The current study focused on the soluble, cytoplasmic proteins of the N. perurans parasite and their association with pathogen virulence. This is supported by the upregulated proteins identified from the newly acquired culture which include the biological processes of cell cytoskeletal re-organisation, protein synthesis, oxidative stress and immunomodulation, all functions which ultimately suggest their involvement in acute AGD virulence. To further the identification of all proteins associated with AGD pathogenesis, future studies will investigate the proteomic profile of membrane-bound and extracellular proteins from the N. perurans cultures presented in this report.

Materials and methods
All methods described were carried out in accordance with ARRIVE guidelines 71  www.nature.com/scientificreports/ was subsequently plated and maintained xenically at 16 °C on marine yeast agar plates (MYA; 0.01% malt, 0.01% yeast, 2% Bacto Agar, sterile sea water at 30 ppt salinity) overlaid with 7 mL of 0.2 µm filtered sea water 72 . Inoculated plates were washed weekly with 7 mL of sterile seawater to control bacterial growth. The amoebae were sub-cultured weekly by transferring free-floating cells to fresh MYA plates. Confirmation of N. perurans identity was performed using real time PCR as previously described by Downes et al. 32 . Parasites from cultures established at 3-years, 1-year and 70 days are referred to as attenuated, virulent and newly acquired, respectively. The attenuated culture of amoeba was established in October 2015, while the virulent culture was propagated for 1 year prior to inoculation of naïve smolts. These two cultures were used in the in vivo challenge trial. The newly acquired culture was propagated for 70 days prior to harvesting for proteomic analysis.
In vivo challenge trial. Naïve Atlantic salmon smolts (n = 120) weighing approximately 120 g were transported from a commercial hatchery to the animal housing unit in Galway-Mayo Institute of Technology. Fish were divided into three cohorts, namely the control cohort (n = 40), the attenuated cohort (n = 40) and virulent cohort (n = 40). These cohorts were distributed into six circular 1000 L (n = 20) circular tanks and maintained for a 2-week acclimation period prior to inoculation. Tanks were monitored for dissolved oxygen, salinity and water quality. Tanks were maintained at 16 °C and salinity of 30 ppt. All fish used in this study were fed daily to satiation using a commercial food pellet.
Fish were transferred to a 300 L saltwater bath at 30 ppt and temperature maintained at 16 °C for inoculation with the parasite for 4 h. The control group were placed in a 300 L saltwater bath, minus the addition of the parasite, serving as the negative control. The attenuated cohort was challenged with 2000 cells/L, achieved by cell count using a Sedgewick rafter counting chamber (API Supplies), of a 3-year old N. perurans culture that was maintained since October 2015. The virulent cohort was challenged with 2000 cells/L of a 1-year old N perurans culture that was maintained since October 2019. After the bathing period, fish were placed back into their respective 1000 L tanks for the duration of the trial.

Validation of virulence.
Six fish from each cohort (n = 18) were removed at 0, 7, 14 and 21-days post infection (dpi) and euthanized in an overdose of anaesthetic containing 400 mg/L of tricaine methane sulfonate (MS-222) for lethal sample collection. Gross macroscopic gill scores were assessed according to Taylor et al. 73 . Gill samples from the first gill arch were excised for histology and quantitative polymerase chain reaction (qPCR) detection of N. perurans. Samples for histological examination were placed into 10% neutral buffered formalin (Sigma-Aldrich) and were processed by the Agri-Food & Biosciences institute (AFBI Stormont). Image analysis of histology samples was undertaken using Olympus cellSens Standard ver 1.12 (http:// www. olymp us-lifes cience. com/ en/ softw are/ cells ens/) and Image J (National Institutes of Health, https:// imagej. nih. gov/ ij/ downl oad. html) software. Samples for qPCR detection of N. perurans were excised, flash frozen and analysed as described by Downes et al. 32 .
RNA sequencing of N. perurans microbiome. DNA Extraction from N. perurans cultures. DNA was extracted using the AllPrep DNA/RNA Mini Kit (QIAGEN, Hilden, Germany) from the attenuated and virulent N. perurans cultures in triplicate. Three plates of the attenuated and virulent cultures were individually prepped for DNA extraction by mechanically scraping cells from MYA plates into a 50 mL falcon tube, yielding a suspension of adherant and floating cells in seawater. The suspensions were briefly vortexed to homogenise. One mL of each N. perurans cell suspension was added to 600 μL of RLT lysis buffer followed by vortexing for 10 min. Fresh lysozyme dilution (20 μL of 10 mg/mL) was added to each tube, prior to 30 min incubation at 37 °C on a shaker platform. Subsequently, 10 μL of Proteinase K (> 600 mAU/ml; QIAGEN, Hilden, Germany) was added into each tube followed by 30 min incubation at room temperature. The rest of the DNA extraction was performed following manufacturer's instructions (AllPrep DNA/RNA Mini Kit [QIAGEN, Hilden, Germany]). All DNA concentrations after the extraction processing were measured using Qubit dsDNA BR Assay Kit (ThermoFisher Scientific, Paisley, United Kingdom), following its instructions.
16S rRNA gene sequencing in Illumina. The sequencing targeted the V4 and ITS1-spanning regions from 16S rRNA genes in the DNA samples using the primers 515F (5′-GTG CCA GCMGCC GCG GTAA-3′) and 806R (5′-GGA CTA CHVGGG TWT CTAAT-3′) 74 . All sequencing was carried out in FISABIO's facilities (FISABIO / Avda. de Catalunya, 21/46020 Valencia, Spain). PCR reaction tubes contained (25 µL final volume): 12.5 ng (genomic DNA), 0.2 µM (final concentration) 515F primer, 0.2 µM 806R, and 12.5 µL 2 × KAPA HiFi Hot Start Ready Mix (0.5 U per 25 µL reaction; Roche). All PCR reactions followed the same program: 1) 95 °C for 3 min, 2) 25 cycles containing: 95 °C for 30 s 55 °C for 30 s and 72 °C for 30 s; 3) 72 °C for 30 s. After a PCR clean-up on each sample, Illumina sequencing adapters were added using Nextera XT Index Kit (Cambridge, United Kingdom) following instructions from the company. Prior to DNA quantification (Agilent Technologies 2100 Bioanalyzer), a second PCR cleaning-up was done. A DNA normalization (to 4 nM) protocol was done, followed by libraries pooling with unique indices. Pools were denatured (using 0.2 N of NaOH), diluted and heat denatured before sequencing. Resulting amplicons were sequenced (Illumina MiSeq platform 75 using 5% PhiX as an internal control. Sequencing analyses. All sequenced demultiplexed FASTQ data from samples were analysed using QIIME 2.2019.10 (Quantitative Insight Into Microbial Ecology) following the "Moving Pictures" tutorial workflow from the official QIIME 2 website (https:// docs. qiime2. org/ 2019. 10/ tutor ials/ moving-pictu res/) 76 . Raw sequences were imported using the "Casava 1.8 paired-end demultiplexed fastq" procedure. They were assembled and quality filtered at a phred score limit of Q25. Resulting files were denoised using the Dada2 command. Sequences were aligned and clustered based on a 97% identity, later each cluster was taxonomically classified using the Gel image analysis. Gels images were acquired using the Gel Doc EZ Gel Documentation System (Bio-Rad, CA) and images analysed using the SameSpots 5.1.0.0 software (TotalLab, Non-linear Dynamics, UK). Triplicate gels used to analyse the proteins extracted from the newly acquired, virulent and attenuated cultures were automatically aligned, and normalization of spots was performed to detect differentially expressed proteins. The software normalises volumes of each spot and detects differential proteins based on a difference in spot intensity using the 'between subject' experimental design, which performs an ANOVA calculation. Differential spots were chosen for further analysis based on maximum fold change of ≥ 2 and ANOVA statistical significance (p < 0.05) using the software's statistical analysis function.
Spot preparation and mass spectrometry. Enzymatic in-gel digestion was performed as described by Shevchenko et al. 78 . Gels were rinsed with deionised water and spots of interest were excised and cut into pieces using a sterile scalpel. Gel pieces were microcentrifuged to collect excess water, destained with 100 µL of 100 mM ammonium bicarbonate/acetonitrile (1:1, v/v) and incubated for 30 min with vortexing every 10 min. Following the addition of ammonium bicarbonate, 500 µL of neat acetonitrile (ACN) was added and incubated at room temperature until gel pieces were destained and decreased in size. The gel pieces were dried and rehydrated with trypsin buffer overnight at 37 °C (13 ng/µl trypsin in 10 mM ammonium bicarbonate containing 10% [v/v] ACN on a thermomixer at 350 rpm. To terminate enzymatic digestion, 1% acetic acid (AA) was added to the samples. Peptides were desalted using C18 stage tips with Equilibration Buffer (0.1% trifluoroacetic acid [TFA] in MS grade water) and Elution Buffer (50% ACN 0.1% TFA in MS grade water). After drying by vacuum centrifugation, peptides were acidified by AA, desalted with C18 STAGE tips 79 , and resuspended in 2.5% ACN, 0.5% AA. Peptide fractions were analysed on a quadrupole Orbitrap (Thermo Scientific Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer) mass spectrometer equipped with a reversed-phase NanoLC UltiMate 3000 HPLC system (Dionex LC Packings, now Thermo Scientific). Peptide samples were loaded onto C18 reverse phase columns (10 cm length, 75 µm inner diameter) and eluted with a linear gradient from 1 to 27% buffer B containing 0.5% AA 97.5% ACN in 58 min at a flow rate of 250 nL/min. The injection volume was 5 μL.
Raw data from the Orbitrap Q-Exactive (Thermo Scientific Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer) was processed using MaxQuant version 1.6.6.0 for identification of proteins 80 , incorporating the Andromeda search engine and MaxQuants contaminants fasta file 31 . To identify peptides and proteins, MS/MS spectra were matched to a combined custom database comprised of N. perurans (20,887 proteins [v2, 07/08/2019, CSIRO]); UniProt reference proteome databases of Amoebozoa (109,415 proteins) and bacteria selected from the microbiome analysis (148,582 proteins) as well as a database of Paramoeba proteins from a taxonomy search on UniProtKB (5001 proteins). All databases except for N. perurans were downloaded on May 8th 2020 from UniProt 81 and further information on individual species protein counts are in Supplementary  Tables S3, S4 and S5. All searches were performed with tryptic specificity allowing two missed cleavages. The database searches were performed with carbamidomethyl (C) as fixed modification and acetylation (protein N terminus) and oxidation (M) as variable modifications. Mass spectra were searched using the default setting of MaxQuant with a false discovery rate of 1% at the peptide and protein level.