The host phenotype and microbiome varies with infection status, parasite origin and parasite microbiome composition

Megan Hahn Stony Brook University https://orcid.org/0000-0001-9266-8232 Agnes Piecyk Max Planck Institute for Evolutionary Biology: Max-Planck-Institut fur Evolutionsbiologie Fatima Jorge University of Otago Robert Cerrato Stony Brook University Martin Kalbe Max Planck Institute for Evolutionary Biology: Max-Planck-Institut fur Evolutionsbiologie Nolwenn M. Dheilly (  nolwenn.dheilly@stonybrook.edu ) Stony Brook University https://orcid.org/0000-0002-3675-5013


7
determining the host immune response [46,47]. In particular, anti-inflammatory interleukins, 112 foxp3 and tgf-β appear to be at the heart of the interplay between bacteria and the immune system 113 [48,49]. For example, the microbiome modulates the innate immune response of mice exposed to 114 influenza [50]. In this study, a microbiota-induced expression of IL-1β and IL-18 was found to be 115 associated with better outcome, and a distal inoculation of LPS to the colon was sufficient to 116 restore the immune response to influenza virus in the lung. Thus, to understand the impact a 117 parasite has on its host holobiont, it is essential to consider both immune function and microbiome 118 composition, and to investigate how they interact with each other. 119 Parasites can also host microbes and for some parasites, the role of bacterial symbionts in 120 virulence and pathogenesis has been thoroughly investigated. Many nematodes depend on 121 Wolbachia for normal development and fertility, and the bacteria also contributes to inflammation 122 and adverse reaction to anti-filarial drugs [51]. Similarly, the bacteria Neorickettsia has high 123 prevalence among digenean trematodes and is often transmitted to parasitized hosts causing    Tail), and GPS (Blue Tail) separated by a 164 mesh divider from 5 Wolf fish , 5 GPS fish, and 4 Walby fish that had been exposed to parasites 165 (right). (D) Upon dissection, the success of infection was assessed, and exposed individuals were 166 classified as exposed non-infected (ENI) or exposed successfully infected (ESI). All successfully 167 infected fish were processed, and corresponding ENI and control non-infected fish from the same 168 tanks and fish origin were processed as controls.  on plerocercoid surface confirmed that the parasite harbors an endomicrobiome ( Figure S6).

188
Cross infection experiment 189 Following experimental infections with hosts and parasites of different origin (Figure 1), 190 we quantified and sequenced the 16S genes of a total of 42 control non-infected sticklebacks (CNI), 191 35 exposed but non-infected sticklebacks (ENI), and 71 exposed and successfully infected 192 11 sticklebacks (ESI), and corresponding S. solidus (Ss). Our results confirmed that S. solidus and G. 193 aculeatus harbor a distinct microbiome, and that exposure and infection alters the host microbiome 194 ( Figure S7-8).

195
The microbiome of threespine sticklebacks varied with exposure, infection, host origin, 196 and parasite origin ( Figure S9, Figure 2). Comparisons of the microbiome composition of CNI fish 197 revealed constitutive differences between Alaskan and European sticklebacks ( Figure 2A).

198
Exposure to S. solidus was associated with small changes in beta diversity ( Figure 2B), but resulted 199 in an increase in differences in microbiome diversity metrics among fish of all three origins ( Figure   200 2C). Parasite origin played a less profound role and limited differences in diversity were found 201 between fish exposed to Walby and SKO parasites ( Figure 2C). Successful infection with S. 202 solidus was associated with an increase in bacterial load that varied with parasite origin ( Figure   203 2B and 2D). The microbiome of infected fish was dominated by more taxa than non-infected fish 204 ( Figure 2B). Finally, parasite origin, but not host origin, was associated with differences in 205 microbiome composition among infected sticklebacks ( Figure 2D Unifrac Axis 3 p=0.041). Differences were found between alpha diversity of fish exposed to Walby  their corresponding fish host revealed an absence of relationship ( Figure 3A, Figure S8). In total, 245 93% of the ASVs, and 9.4% of the families present in ESI sticklebacks were never found in S.  We used DESeq2 to identify differentially abundant bacteria phylotypes ( Figure S11). Cyanobacteria.

272
Host genotype and parasite genotype both contributed to differences in relative abundance 273 of bacterial families ( Figure S11). In CNI, host genotype was associated with variation in origin, but not host origin was associated with significant differences in the strength of the 298 correlation between immune gene expression and microbiome composition ( Figure 4B). These 299 correlations appear to be driven by a subset of bacterial families, among which some were 300 positively correlated with gene expression whereas others were negatively correlated ( Figure 4C).

301
The most significant correlations involved Treg-inducing genes stat4, stat6 and il16, Treg 302 associated gene foxp3, complement factor cfb, anti-microbial innate regulatory genes cd97 and 303 marco, and the regulator of inflammation tnfr1 ( Figure 4C, Figure S12). More specifically, the   345 Our results provide the first set of evidence of an endomicrobiome in the cestode S. solidus. 346 We collected S. solidus plerocercoids from the body cavity of G. aculeatus, so that the parasite 347 was no longer in contact with the host gut microbiota, limiting the potential for contamination [76]. 348 We did not culture any bacteria after spreading freshly sampled plerocercoids on agar suggesting 349 the absence of a surface microbiome. impact the maintenance of these microbial communities.

382
Both exposure and infection influence the host microbiome composition 383 We observed an impact of both exposure and infection by S. solidus on the fish gut       For dissection, an incision was made along the lateral line of the fish body, around the bony pelvis.

531
The cut extended from the pectoral fins to just anterior of the anus to avoid cutting the intestine.

532
The sex of the fish was assessed by visual inspection of gonads at the time of dissection and then 533 confirmed using PCR with sex specific primers as described in [100]. The presence of S. solidus that any bacteria found in S. solidus were indeed part of an endomicrobiome and not contamination 539 from the fish body cavity or surface of the parasite. All samples were stored at -80°C until use. 541 In Spring of 2016, threespine sticklebacks were caught, as described above, from Wolf lake albidus copepods from a laboratory stock were singly infected with S. solidus procercoids as 561 previously described [102]. Fish were starved for 24 hours before being fed either one singly 562 infected copepod or one non-exposed copepod (sham control). After 2 days, fish were transferred 563 into 16L aquaria. Fish exposed to a given parasite family were held together in the same tank. Each 564 tank held five exposed fish from Wolf, five exposed fish from GPS and four exposed fish from 565 Walby, in addition to one control fish per fish population (17 fish/tank). The common garden also calculated if the fish was infected [107]. 585 Head kidney RNA was extracted with a NucleoSpin® 96 kit (Macherey-Nagel) following 586 the manufacturer's protocol. RNA concentration and purity were determined 587 spectrophotometrically (NanoDrop1000; Thermo Scientific). We used the Omniscript RT kit 588 (Qiagen) according to the manual but used 0.2 µl of a 4-unit RNase inhibitor (Qiagen) per reaction. individuals) that were either (i) sham-exposed (42 individuals), (ii) exposed but non-infected by S. MiSeq following the manufacturer's guidelines. Sequence data were initially processed to join 629 forward and reverse reads and remove barcodes. bacterial load) as the dependent variable for a given model [120]. We began by testing the impact 681 on fish population in control non-exposed fish (CNI) (Figure 2A). We conducted model selection

691
To investigate the impact of infection status on these indices, we used infection status, 692 which included the levels CNI (control non-infected), ENI (exposed non-infected), and ESI 693 (exposed and successfully infected), sex, and their interaction as potential fixed factors, and 694 random factors were fish population, parasite population, tank, and round. Model selection was 695 performed as described above and the best-fit model was chosen to calculate p-values ( Figure 2).

36
Following this we tested the impact of fish and parasite population in exposed non-infected fish 697 (ENI), in successfully infected fish (ESI), and parasites (Ss). After conducting model selection as 698 described above, we tested for the role of fish population and parasite population separately ( Figure   699 2). When testing fish population effects, we used fish population, sex, and their interaction as fixed 700 factors, and random factors were parasites population, tank, and round.