Specificity of resistance and geographic patterns of virulence in a vertebrate host-parasite system

Background Host genotype - parasite genotype co-evolutionary dynamics are influenced by local biotic and abiotic environmental conditions. This results in spatially heterogeneous selection among host populations. How such heterogeneous selection influences host resistance, parasite infectivity and virulence remains largely unknown. We hypothesized that different co-evolutionary trajectories of a vertebrate host-parasite association result in specific virulence patterns when assessed on a large geographic scale. We used two reference host populations of three-spined sticklebacks and nine strains of their specific cestode parasite Schistocephalus solidus from across the Northern Hemisphere for controlled infection experiments. Host and parasite effects on infection phenotypes including host immune gene expression were determined. Results S. solidus strains grew generally larger in hosts coming from a population with high parasite diversity and low S. solidus prevalence (DE hosts). Hosts from a population with low parasite diversity and high S. solidus prevalence (NO hosts) were better able to control the parasite’s growth, regardless of the origin of the parasite. Host condition and immunological parameters converged upon infection and parasite growth showed the same geographic pattern in both host types. Conclusion Our results suggest that NO sticklebacks evolved resistance against a variety of S. solidus strains, whereas DE sticklebacks are less resistant against S. solidus. Our data provide evidence that differences in parasite prevalence can cause immunological heterogeneity and that parasite size, a proxy for virulence and resistance, is, on a geographic scale, determined by main effects of the host and the parasite and less by an interaction of both genotypes. Electronic supplementary material The online version of this article (10.1186/s12862-019-1406-3) contains supplementary material, which is available to authorized users.


SI.1 Supplementary information on infection rates
If not stated otherwise, infection rates were calculated by using the number of infected individuals as proportional data in generalized mixed effects models (GLMMs) with binomial error structure and logit link function using the glmer() function of the lme4 R package (Bates et al., 2014). Significantly different groups were identified with glht() post hoc tests from the multcomp package (Hothorn et al., 2008). Infection rates differed considerably between parasite sibships and fish families; some parasite sibships failed to infect any fish. According to our experimental design, however, we did not test for fish family or parasite sibship effects.
Parasite sibship was included as a random factor in analyses of infection rates in copepods; the random term 'round' (i.e. parasite sibship x fish family combination) was included in all analyses of the interaction between S. solidus and its fish hosts.
As expected for the unspecific first intermediate host, S. solidus from every origin managed to infect M. albidus copepods. We tested for potential differences in infection rates in copepods between the two years of the experiment by using data from parasite sibships that were used in both years (Table S2). Indeed, infection rates of parasites from NU and SKO were towards sibship-rather than origin-effects. Interestingly, Pacific (ECH) parasites had the highest infection rates in copepods and the lowest infection rates in sticklebacks. However, overall, and consistent with previous publications (Hammerschmidt and Kurtz, 2005), infection rates in copepods did not influence infection rates in fish.
Infection rates in fish did not differ significantly between the two years of the experiment (DE data; Χ 2 5 = 9.42, p = 0.094). S. solidus origin influenced the infection rates in NO hosts (Χ 2 8 = 21.619, p = 0.006). This was driven by significant differences between infections with NU versus ECH parasites (z = -3.446, p = 0.016). NU S. solidus had the overall highest infection rate (average: 40 %) and ECH S. solidus had the lowest infection rate (average: 9 %). The variance terms for the random effect differed between the experiments, which indicates different parasite sibship x fish family effects; namely, lower variance in DE in contrast 1. Fish from the naturally highly parasitized Norwegian (NO) population ate considerably less infected copepods than DE fish, so we tested for a possible link between the number of ingested copepods and infection success. There was no consistent pattern; the number of infected copepods correlated with an increase or decrease of the infection rates, dependent on the origin of the parasite and the fish population (not shown). Accordingly and in line with the literature (Wedekind and Milinski, 1996), our data does not indicate avoidance behaviour.   rates were analyzed as proportional data (accounting for the copepods that were not ingested) with binomial error structure. We tested for differences between the years by using data of hosts that were exposed to the same sibships in the two years of the experiment. The respective generalized linear model (GLM) included 'round' (fish family x parasite sibship combination) and the interaction with S. solidus origin as an explanatory. Host and parasite effects were analyzed with GLMMs including 'round' as random effect. includes the effect of the random term and was calculated according to (Nakagawa and Schielzeth, 2013;Johnson, 2014;Lefcheck, 2016).  Figure S1. Phenotypic differences between NO (orange) and DE (violet) sticklebacks (contrast 1). The fish were either sham-exposed or infected with single S. solidus parasites from the Baltic (NST:

Splenosomatic index in German and Norwegian fish
We determined the overall condition (condition factor, CF, the ratio between the observed weight W (in g) and the expected weight at a given length L (in cm): CF = 100 * W/L b . The expected weight depends on the exponent b, which is characteristic for each fish population and was calculated by regression analysis of logarithm-transformed data of the length and the weight of all fish from each experiment, (Frischknecht, 1993)) and estimates of metabolic reserves (hepatosomatic index, HSI = 100 * W L / W, with W L representing the weight of the liver, (Chellappa et al., 1995)) and immunological activity (splenosomatic index, SSI = 100 * W S / W, with W s representing the weight of the spleen, (Seppänen et al., 2009); head kidney index, HKI, the weight of the head kidney in relation to body weight). Numbers of granulocytes and lymphocytes in 0.5 mL head kidney leukocyte (HKL) cell suspensions were used to calculate the granulocyte to lymphocyte (G/L) ratio as a proxy for the activity of the innate versus the adaptive immune system. Relative light units (RLUs) in a lucigenin-enhanced chemiluminescence assay quantify the production of reactive oxygen species (ROS) and hence phagocytic capacity of HKL.
Cell suspensions of HKL were prepared by forcing tissue samples through a 40 µm nylon mesh (BD Falcon, USA). The cells were transferred to a 96 deep well plate and rinsed twice in R-90 (90% (v/v) RPMI 1640 in distilled water) at 600 g for 10 min at 4 °C. Total cell numbers were determined by a modified protocol (Scharsack et al., 2004) of the Standard cell dilution assay (Pechhold et al., 1994). Therefore, each sample was supplemented with 2 mg/L propidium iodide (Sigma Aldrich) and 3 x 10 4 green fluorescent reference particles (4 µm, Polyscience, USA). FSC/SSC characteristics were measured in linear mode for one minute or for up to 10,000 events using a Becton Dickinson FACS Calibur and BD CellQuest™ pro software (Version 6.0). Propidium iodide positive (i.e. dead) cells and cellular debris (low FSC characteristics) were excluded from further analyses. Granulocytes and leukocytes were identified according to their FSC/SSC profiles. The numbers of viable granulocytes and lymphocytes in 0.5 mL were used to calculate the granulocyte to lymphocyte ratio (G/L ratio) . A lucigenin-enhanced chemiluminescence (CL) assay (Scott and Klesius, 1981;Kurtz et al., 2004) was used for functional analysis of innate immune activity. The CL assay measures the phagocytic capacity of HKL by quantifying the respiratory burst reaction in relative luminescence units (RLUs). Briefly, 10 5 live cells per sample were supplemented with (Sigma Z 4250) was added at a final concentration of 0.75 µg/µL to stimulate the production of reactive oxygen species (ROS). Chemiluminescence was measured every 3 min for 3.5 hours (Berthold Technologies luminometer) and the area under the kinetic curve (calculated with Win Glow 2000 professional software) was used for analyses. The RLU was standardized by division by the mean RLU of the negative controls (wells containing buffer without head kidney cells) for each day and by division by the number of vital granulocytes of the respective sample. Unfortunately, we could not obtain enough cells from every fish (data was missing from 13 samples) and thus analyzed production of reactive oxygen species of a total of 1430 different samples. Controls (medium without cells) were missing for one round in 2015. Values for those controls were inferred from data from empty wells in relation to controls.
Testing these condition and immunity related indices in each experiment, DE sticklebacks (contrast 2; Figure S2) showed significantly elevated immune parameters if they were infected with Pacific S. solidus: the head kidneys were larger (LMM; p < 0.001), the G/L ratio was significantly higher in comparison to all but SKO-infected fish (LMM; p < 0.001) and the head kidney's potential to produce reactive oxygen species was higher in comparison to controls Head kidneys were larger in ISC-infected fish than in control fish (LMM; p < 0.001) and the G/L ratio was significantly higher in SKO-infected fish than in control fish (LMM; p < 0.001) ( Figure S3). Relative to the control, NO sticklebacks had significantly lower Hepatosomatic indices when they were infected with sympatric (SKO-) S. solidus parasites (LMM; p < 0.001).
The Splenosomatic index was higher in ISC-parasite infected fish in comparison to controls and NST-parasite infected fish (LMMs; each p < 0.001). Head kidney related immune parameters did not differ between infected and uninfected NO sticklebacks ( Figure S4). Figure S2. Phenotypic differences between sham-exposed and S. solidus infected DE sticklebacks (contrast 2).  Figure S3. Phenotypic differences between sham-exposed and S. solidus infected DE sticklebacks (DE in contrast 1).

SI.4.1 Stickleback immune gene expression differences between populations
Differentially expressed genes are marked in bold letters if significant after FDR correction.  The R 2 includes the effect of the random term and was calculated according to (Nakagawa and Schielzeth, 2013;Johnson, 2014;Lefcheck, 2016). Post hoc tests are based on Tukey's all pair comparisons. includes the effect of the random term and was calculated according to (Nakagawa and Schielzeth, 2013;     ECH-infected DE sticklebacks had significantly higher expression of three genes of innate immunity, one gene of adaptive immunity (foxp3) and complement c9; RNA levels of tcr-β and mhcII were significantly lower than in controls (Table S17). NU-infected DE sticklebacks had significantly higher expression of five innate immune genes and two complement components; again, tcr-β was significantly lower expressed than in controls (Table S18). SKO-infected DE sticklebacks had significantly lower expression of the genes igm and tcr-β (Table S19).
Each dot represents one individual; data from infected fish is colored. Ellipses represent 95% confidence intervals. P-values are shown if significant after FDR-correction. showing multivariate data from 9 genes of the adaptive immune system of infected and sham-exposed (CTRL) DE sticklebacks (2014). Each dot represents one individual; data from infected fish is colored.
Ellipses represent 95% confidence intervals. P-values are shown if significant after FDR-correction. dimensions showing multivariate data from three genes of the complement system of infected and shamexposed DE sticklebacks (2014). Each dot represents one individual; data from infected fish is colored.
Ellipses represent 95% confidence intervals. P-values are shown if significant after FDR-correction. immune genes of infected and sham-exposed (CTRL) NO sticklebacks.
Each dot represents one individual; data from infected fish is colored.
Ellipses represent 95% confidence intervals. P-values are shown if significant after FDR-correction. Euclidian distances and two dimensions showing multivariate data from 12 innate immune genes of infected and sham-exposed (CTRL) NO sticklebacks. Each dot represents one individual; data from infected fish is colored. Ellipses represent 95% confidence intervals.