Impact of Bd endemism in amphibian communities in Flanders
In a first study, Bd prevalence was determined across our study area (Flanders, Belgium). We sampled 1483 amphibians belonging to 62 populations in 2015–2016 (Supplementary Fig. 1). To detect the presence of Bd, we collected swabs from the superficial skin surface of metamorphosed animals or the mouthparts of larval anurans. To study potential co-existence of Bd in the study region with small populations of a susceptible species (where negative effects are expected to be most obvious), in a second study, we sampled five breeding sites of midwife toads in Flanders for 4 consecutive years (Supplementary Table 2). In these breeding sites, larvae were counted once a year and their mouthparts were sampled for the presence of Bd. In a third field study, we selected 26 ponds across our study area containing at least a population of alpine newt (Ichthyosaura alpestris), being the European urodele most likely infected by Bd. Ponds were sampled with funnel traps three or four times (depending on the presence of water) with a 1-month interval (March–June 2019). An envisaged 30 newts per sampling per pond were swabbed for the presence of Bd (Supplementary Table 3), weighed to the nearest 0.1 g and the snout-vent length measured to the nearest mm. As an estimate of body condition, we used the scaled mass index (SMI)45, which adjusts the mass of all individuals to that which they would have obtained if they had the same body size. SMI was calculated using the equation of the linear regression of log‐body mass on log‐snout-vent length estimated by type 2 (standardized major axis; SMA) regression45. Eleven outliers were present (i.e. |standardized residual| >3). These observations were not used for deriving SMI relationships (as per Peig and Green45). The regression slope was 2.87, and average snout-vent length was 42.7 mm. We thus calculated the SMI as (body mass × (42.7/snout-vent length)2.87). The average number of individuals caught per fyke per pond was used as proxy for newt density. To test whether trends in newt density (i.e. average number of newts per fyke) differed between Bd positive and Bd negative ponds, a generalized linear mixed model (GLMM) was used specifying newt density as the dependent variable and the interaction between time (month) and Bd status (positive versus negative) as independent variables. Trends in newt density were better approximated by a quadratic relationship compared to a linear trend (delta AIC = 5.73). To test whether trends in SMI differed between Bd positive versus Bd negative newts, a GLMM was implemented using SMI as the dependent variable and time (month), Bd status (positive versus negative), newt sex (male versus female) and newt density (see above) as independent variables. The initial model contained all two-way interactions between the independent variables. For both GLMMs, pond was implemented as a random factor, and a Gaussian error structure was specified (model residuals were normally distributed, Shapiro–Wilk W > 0.90). A frequentist approach was adopted whereby initial models were reduced in a stepwise manner, by excluding the variable with the highest P value until only P < 0.05 predictors remained.
This field study was performed with approval of the Flemish government (derogation number ANB/BL/FF-V15-00015).
Animals were handled while wearing a fresh pair of non-powdered, disposable gloves. Equipment and field clothing were cleaned and disinfected between visits to sampling locations.
Detection of chytrid associated disease in the study region was done by postmortem examination of field cases of amphibian disease or mortality over a period of 4 years. Wildlife Health Ghent hosts an amphibian disease hotline where suspect cases of infectious disease are submitted (cases with obvious traumatic causes such as predation or traffic are not withheld). The dead amphibians are routinely examined for the presence of Bd, Bsal and Ranavirus using (q)PCR46,47,48.
Isolate collection
Isolates were obtained from wild amphibians collected July 2015–September 2016 in the Flanders region of Belgium. Isolation was carried out using the protocol described in Fisher et al.49. Using a sterile needle, small skin sections 1–2 mm were cleaned of surface-contaminating bacteria and fungi by dragging it through agar-medium. Subsequently, the tissue sample was transferred into a sterile TGhL medium with antibiotics (200 mg/L penicillin-G and 400 mg/L streptomycin sulphate) and incubated at 20 °C. Six isolates were genotyped and their topology depicted in the global phylogeny in O’Hanlon et al.1. These isolates belong to three clades. Full information on each isolate can be found in (Supplementary Table 4). All isolates were preserved in liquid nitrogen at passage 10. Before use in an experiment, isolates were passaged one time in TGhL broth at 20 °C.
Hygiene and biosafety protocols
The animal experiments were performed under strict BSL2 conditions. During the fieldwork, each individual was handled with a new pair of nitrile gloves. At the end of each field visit, boots, dipnets, funnel traps and other equipments were disinfected with a 1% Virkon® solution for at least 5 min50.
Phenotypic characterisation of local Bd isolates
All cultures (BdBE1, 2, 3, 4, 5, 6, 7, 8, 9, 10 and BdJEL423) were grown on TGhL agar at 20 °C, for 5 days. Plates were washed with sterile distilled water, and the water containing the zoospores was passed over a sterile mesh filter with pore size 10 mm (Pluristrainer, PluriSelect). Using a haemocytometer and inverted microscope (Nikon Eclipse TS100, Nikon) the spores were counted. Bd spores (1 × 106 spores per well) were seeded in TGhL medium (1.6% tryptone, 0.4% gelatin hydrolysate and 0.2% lactose in H2O) supplemented with 50% H2O into 24-well tissue culture plates. Wells were photographed using an Olympus CKX41 with an attached camera (Olympus SC50, Olympus) with a (2560 × 1920) pixel field, ~72, 96, 120, 144 and 168 h after inoculation. The experiment was performed in quadruplicate. Images were analysed using ImageJ 1.52d software, with the following measurements taken: (1) NZOOSPORE = number of spores in the central 1000 × 1000 pixels of the well per image, (2) NSPOR = number of sporangia in the central 1000 × 1000 pixels, (3) ASPOR = area of the largest 10 sporangia, (4) AZOOSPORE = area of 10 random spores. To measure fecundity, at day 7, the following formula was used ((average NZOOSPORE/average NSPOR)/average ASPOR) as described in Fisher et al.26.
The number of zoospores and the number of sporangia were modelled using generalized linear models (GLM) with Poisson error distributions, while the area of sporangia and the calculated fecundity were modelled using linear models, all with isolate as the response variable. For the area of sporangia, a random effect at the replicate level was included to account for pseudo-replication. The significance of pairwise differences among isolates was assessed using Tukey’s HSD test. In addition to differences among isolates for individual characteristics, their overall similarity was assessed using two-dimensional non-metric multidimensional scaling (NMDS). NMDS maps the relationship between the dissimilarity matrix (Bray–Curtis index) to locate each sample along two coordinates in the ordination space51. All analyses were performed in R52.
Infection trials
The animal experiments were performed with the approval of the ethical committee of the Faculty of Veterinary Medicine (Ghent University EC2016/20, EC2015/86). Only captive-bred animals were used in standardized experiments, using identical environmental conditions that allow comparison of isolate virulence and host susceptibility. All animals were housed individually in terraria at 15 °C on moist tissue with access to a hiding place. All animals (males/females) were clinically healthy and derived from breeding colonies that are free of Bd, Bsal and Ranavirus as assessed by sampling the skin using cotton-tipped swabs and subsequent performing qPCR or PCR46,47,53. Individuals were randomly assigned to treatments. All animals were clinically inspected daily. Skin sampling was done weekly and the swabs were analysed for the presence of Bd using qPCR described by Hyatt et al.53 with the respective isolate used as standard. Sample analysis was blinded. Animals were euthanized by an overdose of pentobarbital. Humane endpoints were set at the loss of self-righting ability and/or change in posture.
Susceptibility of native amphibians to BdGPL isolates under standardized laboratory conditions
Five captive-bred individuals of seven species (fire salamander (Salamandra salamandra), common spadefoot (Pelobates fuscus), natterjack toad (Epidalea calamita), common toad (Bufo bufo), alpine newt (Icthyosaura alpestris), great crested newt (Triturus cristatus) and European tree frog (Hyla arborea)) were exposed to the local BdGPL isolate BdBE1 or the hypervirulent isolate BdJEL423. In addition, two midwife toads (A. obstetricans) were included as BdJEL423-infected positive control group (the limited number of midwife toads in this experiment were merely included as positive controls, extensive experiments in this species were conducted separately). For inoculum, isolates were grown on TGhL agar at 20 °C, for 5 days. Plates were washed with sterile distilled water, and the water containing the zoospores was passed over a sterile mesh filter with pore size 10 mm (Pluristrainer, PluriSelect), then counted using a haemocytometer and inverted microscope (Nikon Eclipse TS100, Nikon). Spores were resuspended at a concentration of 1 × 106 spore ml−1. Individuals were inoculated with 1 ml of 1 × 106 fresh spores, in separate Petri dishes and placed in climate-controlled (~15 °C & ~80% relative humidity) room for 24 h. Individuals were then transferred to individual plastic containers containing damp tissue and a hide. Small crickets were given ad libitum, providing a constant food supply. Clinical examination of the animals was carried out daily, and tissue replacement and swabbing weekly. Species that did not show infection for over 3 weeks (3 weeks negative in qPCR and no clinical signs) in any individual were considered uninfected and removed from experiment at week 8. Absence of Bd infection was confirmed by histopathology and immunohistochemistry. Sample analysis was blinded. We compared the proportion of individuals of each species becoming infected from either isolate using a generalized linear model with binomial error distribution.
Differential BdGPL isolate pathogenicity for midwife toads
Four local phenotypically and genotypically1 different isolates originating from different species (Fig. 1, Supplementary Table 5) (BdBE1, BdBE3, BdBE4 and BdBE5) and the hypervirulent isolate BdJEL423 were selected to inoculate midwife toads (A. obstetricans). This species is used as model species for susceptibility to virulent Bd. Negative control animals were sham inoculated with water. Thirty-six newly metamorphosed A. obstetricans were randomly assigned to groups of six for each treatment. Individuals were exposed to 1 ml of 1 × 106 of fresh spores, or distilled water (prepared as previously) for 24 h. Individuals were then transferred to individual plastic containers containing damp tissue and a hiding place in a climate-controlled (~15 °C & ~80% relative humidity) room. Small crickets were given ad libitum, providing a constant food supply. Clinical examination of the animals was carried out daily, and tissue replacement and swabbing weekly. From day 15 onwards, animals were weighed weekly. Individuals that died during the trial or were euthanized due to reaching a humane end point (see above) had skin samples taken for histology and DNA extraction for qPCR. Since after the first inoculation none of the toadlets became positive for BdBE5, in week 5, an additional five BdBE5 and three BdJEL423 individuals were inoculated and added to the experiment. Sample analysis was blinded.
The probability of an individual dying as a result of infection with different isolates was modelled using a binomial GLM. Penalized regression (package brglm2 in R54) was used to account for perfect separation in some groups, i.e. where all or no animals died within a given treatment group. Since death occurred for all BdJEL423-infected animals and for only one of the animals infected with the local isolates, parametric survival regression model could not be fitted to estimate the mortality rate over time. The relationship between probability of death and the infection load sustained was modelled using a binomial penalized GLM (as above). Infection load was always modelled as the log10 of the genomic equivalent. Mean, maximum and total load over the course of infection were all assessed as predictors.
Protective capacity of low-virulence BdGPL against hypervirulent BdGPL challenge in midwife toads
After completion of the multi-isolate A. obstetricans infection trial, all remaining individuals were exposed to 1 ml of 1 × 106 of fresh spores of BdJEL423 (prepared as previously) for 24 h. Individuals were then transferred to individual plastic containers containing damp tissue and a hiding place in a climate-controlled (~15oC & ~80% relative humidity) room. Small crickets were given ad libitum, providing a constant food supply. Clinical examination of the animals was carried out daily, and tissue replacement and swabbing weekly. Sample analysis was blinded.
As for the initial infection trials, penalized binomial GLMs were used to estimate whether the probability of death following infection with BdJEL423 depended on (1) the isolate individuals were exposed to in the initial experiment and (2) the mean/maximum/total infection load sustained during the initial experiment. Parametric survival regression was also used to assess whether the survival rates of BdJEL423-re-infected individuals depended on (1) the isolate they had been exposed to in the initial trial and (2) on the mean load of infection in the initial trial. In both cases, the response variable (death) was modelled using a Weibull distribution while accounting for censoring. We repeated all analysed grouping individuals initially infected with local isolates BdBE1/BdBE3 and BdBE4/BdBE5, since these two pairs of isolates exhibited similar characteristics in previous experiments.
To confirm the protective capacity of BdBE3, 40 newly metamorphosed A. obstetricans were randomly assigned to two groups and exposed to 1 ml of 1 × 106 of fresh spores of BdBE3, or distilled water (prepared as previously) for 24 h. Animals were then housed individually in plastic containers containing tissue, a hiding place and water dish in a climate-controlled (~20 °C & ~80% relative humidity) room and examined as previously described. Four weeks later, all animals were exposed to 1 ml of 1 × 106 of fresh spores of BdJEL423 (prepared as previously) for 24 h. Individuals were again transferred to individual plastic containers containing tissue, a hiding place and water dish. Small crickets were given ad libitum, providing a constant food supply. Clinical examination of the animals was carried out daily, and tissue replacement and swabbing weekly. Sample analysis was blinded. The readout for this experiment was the comparison of the proportion of Bd infected animals after challenge with the highly virulent BdJEL423 in each of the two treatment groups. Significance of protection was calculated using two-tailed Fisher’s exact test.
Bd growth in amphibian mucosomes
Mucosomes were collected from healthy midwife toads (Alytes obstetricans) and fire salamander (Salamandra salamandra) using the bathing method described by Woodhams et al.55. For each species, mucosomes were collected from ten animals (animals were not used in the previous experiments). Briefly, each animal was bathed in a Petri dish with HPLC water for 1 h, the volume of water was calculated according to animal surface56,57. The collected mucosome solutions were filtered through a 0.2 μm unit filter (Whatman, GE Healthcare Life Sciences) and immediately used in the next steps. Two BdGPL isolates, the hypervirulent BdJEL423 and the local BdBE1 were used in this experiment. For each isolate, spores were collected from culture flasks with sporulating sporangia. The medium was collected and filtered using a sterile filter with pore size 10 μm (Pluristrainer, PluriSelect). To achieve the target concentration of 1 × 106 spores per mL, the spore suspension was diluted with TGhL broth. Finally, 100 µl of this spore suspension was used to obtain 105 spores per well of a 96-well flat-bottom plate (Greiner Bio-One).
In each well, 100 μL of the mucosome solution was added to 105 spores. As positive control, sterile water was added instead of mucosome and the negative control consisted of heat-killed spores (10 min at 100 °C). All the samples were performed in triplicate. After incubation at 20 °C for 5 days, the bottom of each well was scraped with a 100 μL tip and the liquid was transferred to a 1.5 ml Eppendorf tube for DNA extraction and qPCR58. Growth of BdBE1 and BdJEL423 (corrected GE values) was shown by subtracting the mean GE loads of the negative control from the original GE values. Multiple comparisons on the original values were assessed by a Kruskal–Wallis analysis, followed by pairwise Mann–Whitney U-test with a Bonferroni-corrected P value of 0.017 (P value = 0.05/3 pairwise comparisons) (SPSS version 26).
BdGPL gene expression
We compared gene expression of a selection of virulence genes as identified by Farrer et al.5. Using the RNeasy mini kit (Qiagen), total RNA was isolated from fresh Bd spores (2.5 × 107 spores per condition) and Bd spores (2.5 × 107 spores per condition) that were incubated for 2 h at 20 °C with chytrid negative skin of A. obstetricans collected with a skin biopsy punch (6 mm). Bd spores were collected by putting sterile distilled water on a culture flask containing mature sporangia. Once the zoospores were released, the water containing the zoospores was collected. In order to reduce the percentage of mature cells, the water containing the zoospores was passed over a sterile mesh filter with pore size 10 µm (Pluristrainer, PluriSelect). The flow-through was used as the zoospore fraction (>90% purity). All conditions were analysed with Bd spores originating from BdJEL423, BdBE1-BdBE10 and they were tested in fourfold (biological replicates n = 4). Total RNA (1 μg) was reverse transcribed to cDNA with the iScript cDNA synthesis kit (Bio-Rad). The housekeeping genes α-centractin, APRT, TUB and Ctsyn1 were included as reference genes59. The list of genes and sequences of the primers used for quantitative PCR analysis can be found in Supplementary Table 10. Real-time quantitative PCR reactions were run in triplicate (technical replicates n = 3) and the reactions were performed in 10 μl volumes using the iQ SYBR Green Supermix (Bio-Rad). The experimental protocol for PCR (40 cycles) was performed on a CFX384 RT-PCR cycler (Bio-Rad) and data were analysed using the Bio-Rad CFX manager 3.1. The results are shown as fold changes of mRNA expression relative to the mRNA expression levels in fresh spores. Fold changes were calculated using the cycle threshold (ΔΔCT) method, and they were analysed in SPSS version 25 (SPSS Inc., Chicago, IL, USA). Multiple comparisons were assessed by a Kruskal–Wallis analysis, followed by pairwise Mann–Whitney U-tests adjusted for multiple testing with a Benjamini–Hochberg correction60, setting an adjusted P value of 0.05 as significant. Correlation between the relative expression of each isolate to BdJEL423 for CRN-like genes of fresh spores exposed to midwife toad skin and colonisation capacity from the individuals from the multi-isolate A. obstetricans infection trial was assessed by Spearman’s rank correlation in R54.
In vitro infection of A6 cells
Here, we compared the invasive capacity between hypervirulent and low-virulence local BdGPL isolates using a cell culture model. The Xenopus laevis kidney epithelial cell line A6 (ATCC-CCL 102) was grown in 75 cm2 cell culture flasks and maintained in complete growth medium (74% NCTC 109 medium, 15% distilled water, 10% fetal bovine serum (FBS) and 1% of a 10,000 U ml−1 penicillin-streptomycin solution (P/S)) and the cells were incubated at 26 °C and 5% CO2 until they reached confluence. Using trypsin, the cells were detached, washed with 70% Hanks’ Balanced Salt Solution without Ca2+, Mg2+ (HBSS−) by centrifugation for 5 min at 1500 rpm and resuspended in the appropriate cell culture medium for invasion assays, which were performed as described in Verbrugghe et al.61. To assess the germ tube formation, A6 cells were stained with 3 µM CellTrackerTM Green CMFDA, seeded (105 cells per well) in 24-well tissue culture plates containing collagen-coated glass coverslips and they were allowed to attach for 2 h at 20 °C and 5% CO2. After washing three times with 70% HBSS+, they were inoculated with Bd zoospores in invasion medium, at a MOI of 1:10. Two hours p.i., the cells were washed three times with 70% HBSS+ and the invasion medium was replaced by staining medium. Four hours p.i., the infected cells were washed three times with HBSS+ and they were incubated with Calcofluor White stain (1 µg ml−1 in 70% HBSS+) for 10 min. After washing three times with 70% HBSS+, the cells were fixed, mounted and analysed using fluorescence microscopy. To assess the invasive growth, A6 cells were seeded and inoculated with Bd zoospores as described above. Two days p.i., the infected cells were stained with 3 µM CellTrackerTM Green CMFDA, washed three times with 70% HBSS+ and they were incubated with Calcofluor White stain (10 µg ml−1 in 70% HBSS+) for 10 min. After washing three times with HBSS+, the infected cells were fixed, permeabilized for 2 min with 0.1% triton and incubated for 60 min with a polyclonal antibody against Bd (1/1000), which was obtained by immunizing rabbits with Bd-antigen62. After washing three times with 70% HBSS+, the samples were incubated with a goat anti-rabbit Alexa Fluor 568 (1/500). After an incubation of 1 h, the samples were washed three times with 70% HBSS+, mounted and analysed using fluorescence microscopy and Leica Application Suite (LAS) software X. The Alexa Fluor 568 targeting Bd and Calcofluor White stainings are used in concert to assess the ability of Bd to penetrate the host cell. Calcofluor White is not internalized by A6 cells, whereas the Alexa Fluor 568 targeting Bd staining was applied after permeabilisation of the host cells. As such, intracellular Bd will only be targeted by the Alexa Fluor 568 and extracellular Bd bodies will be bound with both the Alexa Fluor 568 and Calcofluor White stain. Overlay pictures were made with ImageJ 1.52d software. To assess the in vitro infection dynamics of different BdGPL isolates, three independent in vitro experiments were conducted with every condition being tested in triplicate, with similar results.
Whole-genome sequence analysis
WGS read data were downloaded from NCBI Bioproject PRJNA413876 (SRA sample accession numbers: BdBE1; SRA: SRS2757215, BdBE3; SRA: SRS2757203, BdBE4; SRA: SRS2757202, BdBE5; SRA: SRS2757217, BdJEL423; SRA: SRS2757141). Reads were aligned to the reference BdJEL423 assembly (BioProject PRJNA13653) using BWA mem version 0.7.1763 and the aligned bam files were processed using Picard tools version 2.21.1 (http://picard.sourceforge.net/) AddOrReplace, MarkDuplicates, SortSam, CreateSequenceDictionary and ReorderSam. Variants were called using GATK v4.1.4.064 HaplotypeCaller, the output GVCFs were combined using CombineGVCFs, genotyped using GenotypeGVCFs, separated into SNP and Indel variants for filtering using SelectVariants and filtered using VariantFiltration with filters QD < 2.0 || FS > 60.0 || MQ < 40.0 || MQRankSum < −12.5 || ReadPosRankSum < −8.0 for SNPs and QD < 2.0 || FS > 200.0 || ReadPosRankSum < −20.0 for Indels. Variants were annotated using SNPeff v4.365 and the BdJEL423 reference protein annotation (BioProject PRJNA13653), with those variants annotated as HIGH impact on function used for further analysis. Variants that were only present in all local isolates or only present in the BdJEL423 isolate were identified as the genes of interest for the virulence/local phenotype. Similarly, variants that were only present in BdBe1, BdBe3 and BdJEL423 while absent in BdBe4 and BdBe5, or variants present in BdBe4 and BdBe5 while absent in BdBe1, BdBe3 and BdJEL423 were identified as variants of interest for the invasive/epibiotic phenotype. The BdJEL423 reference protein annotation was further annotated using SignalP 4.0 server66, the TMHMM Server 2.067 and the HMMER 3.2.1 server68 using the gathering threshold. Candidate Crinkler proteins were identified using Blastp (BLAST+ 2.9.0)69, all candidates listed displayed 100% identity and e-value of 0. PFAM domain enrichment in variant-effected groups was identified with right-tailed Fisher’s exact test using q-value False Discovery Rate. Copy number variation comparisons were made using CNVkit70 version 0.9.5 batch, segment, scatter, heatmap (with –d flag to de-emphasize low-amplitude segments) and genemetrics commands, utilising the JEL423 reference assembly and annotation, specifying the BdJEL423 read aligned bam as the normal control compared to the local isolate samples.
Protective capacity of low-virulence BdGPL against highly virulent Bsal in three urodele species
Clinically healthy, Bd, Bsal and ranavirus free fire salamanders (27), ribbed newts (Pleurodeles waltl) (24) and marbled newts (Triturus marmoratus) (20) were divided ad random in three treatment groups (every animal was housed individually). Two groups of each species were exposed to local BdGPL (1 ml of BdSP11 106 spores for 24 h), 3 weeks later one of these groups to Bsal (1 ml of AMFP18/02 103 spores for 24 h). At this 21-day time point, a Bsal control group of each species was exposed to Bsal (1 ml of AMFP18/02 103 spores for 24 h). All animals were clinically inspected daily. Skin sampling was done weekly and the swabs were analysed for the presence of Bsal using qPCR as described by Blooi et al.46,47 with the respective isolate used as standard. Bsal infection was confirmed using histopathology. Sample analysis was blinded. Statistical analyses were carried out using generalized linear models and survival regression, following methods described for Bd infection trials above (with treatment group in place of isolate).
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Source: Ecology - nature.com
