Differential Immune Responses Correlate With Chytridiomycosis Severity in Italian Crested Newts

ABSTRACT

In the midst of the current biodiversity crisis, amphibians are severely threatened by emerging diseases such as chytridiomycosis. Characterizing the mechanisms that underlie susceptibility to this disease is fundamental to improve amphibian conservation. Using a comprehensive multi‐omics approach, this study investigates the impact of an exposure to the chytrid fungus Batrachochytrium salamandrivorans (Bsal) on the gene expression of Italian crested newts ( Triturus carnifex ) and on the bacterial symbionts that constitute their skin microbiota. Exposure to Bsal affected multiple components of the newts' immunity, from limited structural changes in their microbiota and decreased expression of keratin‐encoding genes in the skin to the upregulation of genes involved in inflammation and adaptive immunity both at the site of infection (skin) and in their primary lymphoid organ (spleen). We found that chytridiomycosis severity was positively correlated with the downregulation of basal metabolism and with the upregulation of immune responses and of tissue restructuration. Together, our results suggest that in T. carnifex , Bsal susceptibility may be linked to a reallocation of energy resources from the maintenance of basal metabolism and tissue integrity towards elevated immunity and tissue restructuration.

Abbreviations

ASV

Amplicon Sequence Variant

Bd

Batrachochytrium dendrobatidis

Bsal

Batrachochytrium salamandrivorans

DAA

Differential Abundance Analysis

DEA

Differential Expression Analysis

ECM

Extracellular Matrix

EID

Emerging Infectious Disease

GO

Gene Ontology

GSEA

Gene Set Enrichment Analysis

LFC

Log‐Fold Change

MHC

Major Histocompatibility Complex

NTC

Non‐Template Control

1. Introduction

As globalization and climate change facilitate the global spread of pathogens to new niches and ecosystems, emerging infectious diseases (EID) pose an increasing threat to biodiversity (Schmeller et al. 2020). Fungal pathogens are notably threatening wildlife more than any other parasitic group, causing declines in many taxa worldwide (Fisher et al. 2012, 2020). One of the most infamous fungal EIDs known to this day is amphibian chytridiomycosis. This skin disease precipitated the extinction of over 90 amphibian species and contributed to the decline of hundreds of others, thereby causing the greatest vertebrate biodiversity loss attributable to a disease ever documented (Scheele et al. 2019). Chytridiomycosis is caused by two etiologic agents: the chytrid fungi Batrachochytrium dendrobatidis (Bd) and B. salamandrivorans (Bsal), which proliferate in the amphibian epidermis and disrupt its physiology (Voyles et al. 2009; Van Rooij et al. 2015). While Bd infects all orders of amphibians and is present on all continents, Bsal threatens salamanders and is currently geographically restricted to Asia and Northwestern Europe (Fisher and Garner 2020). Bsal puts urodeles at an unprecedented risk because of its high pathogenicity for numerous species (Martel et al. 2014; Gray et al. 2023) and its high invasion potential (Yap et al. 2017); thus, characterizing the mechanisms underlying susceptibility to Bsal is critical for salamander conservation.

Generally, multiple immune components form intertwined layers against pathogens in amphibians. To simplify briefly, constitutive defences (the physical and chemical barriers made by symbiotic bacteria in the skin microbiota, granular glands' secretions in the mucus and keratinized cells) along with the inflammasome and the complement pathway participate in the early innate immune response. Innate immune cells expressing major histocompatibility complex (MHC) molecules, such as dendritic cells and macrophages, can later present pathogen‐derived antigens, leading to the activation of the lymphocytes responsible for adaptive immunity (Grogan et al. 2020).

To date, most research on chytridiomycosis has focused on Bd, probably because of its global distribution, larger number of susceptible taxa and earlier discovery compared to Bsal. Depending on the host, Bd infection can trigger a response that involves the skin microbiota (Woodhams et al. 2015), the secretion of antimicrobial peptides (Jiménez et al. 2022), complement and pro‐inflammatory cascades (Rodriguez and Voyles 2020), and/or the activation of lymphocytes and macrophages (Grogan et al. 2020; Ruiz and Robert 2023). Variation in host response to infection among species and among individuals is complex and still poorly understood. For example, susceptibility to Bd can sometimes (Bates et al. 2022) but not consistently (Muletz‐Wolz et al. 2019) be predicted from microbiota profiles—although exposure to Bd consistently disrupts the amphibian skin microbiota (Jani and Briggs 2014; Walke et al. 2015; Muletz‐Wolz et al. 2019; Jani et al. 2021) and alters its beneficial roles for host health (Jiménez and Sommer 2017; Knutie et al. 2017). Transcriptomic approaches have also revealed contradictory patterns in innate and acquired immune responses between studies and model species: for example, survival to Bd is sometimes associated with an absence of change in gene expression following pathogen exposure (Eskew et al. 2018) or conversely, with increased (Ribas et al. 2009) or reduced (Ellison et al. 2015) inflammatory responses, or with the upregulation (Ellison et al. 2015) or downregulation of genes related to acquired immunity (Savage et al. 2020). Nevertheless, in susceptible species, Bd appears to consistently disrupt acquired immune responses, either by suppressing them (inhibition of lymphocytes proliferation; Fites et al. 2013, 2014; Rollins‐Smith et al. 2015) or by excessively enhancing them—to the point that the immune reaction itself might cause damage to the host (immunopathology hypothesis; Ellison et al. 2014; Eskew et al. 2018).

Given the disparity between Bd and Bsal pathogenesis (Farrer et al. 2017) and the fact that both chytrids diverged between 30 and 115 million years ago (Wacker et al. 2023), host immune responses to Bd may not be extendable to Bsal. The few studies which investigated the role of the microbiota in Bsal‐related chytridiomycosis dynamics showed that effects of Bsal on epidermal bacterial communities vary from subtle (Bletz et al. 2018; Fieschi‐Méric et al. 2025) to more pronounced but temporary disturbances (Bates et al. 2019). Transcriptomic responses to Bsal infection have only been investigated in two susceptible species and over a short timeframe: in Wenxian knobby newts ( Tylototriton wenxianensis ), a weak inflammatory response but no acquired immunity can be detected 10 days following exposure despite severe epithelial erosion and necrosis (Farrer et al. 2017), whereas in eastern newts ( Notophthalmus viridescens ) many innate and adaptive immune genes are upregulated 2 weeks following Bsal exposure—although that change in expression does not prevent mortality (McDonald et al. 2020). Given that adaptive immunity can take up to 6 weeks to develop in amphibians (Grogan et al. 2020), studies conducted on longer timeframes are necessary to fully characterize amphibian immune responses to Bsal. Furthermore, in contrast with these two studies conducted on susceptible species, the use of a model salamander with among‐individual variation in response to Bsal would facilitate the identification of chytridiomycosis‐resistance pathways.

In this context, we investigated the effects of Bsal infection on (1) the skin microbiota, (2) the gene expression in the skin and (3) the gene expression in the spleen of Italian crested newts ( Triturus carnifex ) 8 weeks after exposure to the fungus. Indeed, T. carnifex displays strong inter‐individual variation in susceptibility to Bsal (Fernández Meléndez et al. 2025), suggesting that this species is a promising candidate to characterize the factors leading to differential disease outcomes. Microbiota, gene expression and the regulation of biological pathways were compared between control and Bsal‐exposed individuals, as well as among Bsal‐exposed newts displaying variable susceptibility to the fungus, in order to thoroughly characterize host responses to Bsal infection.

2. Material and Methods

2.1. Experimental Setup

Twenty‐two captive‐bred Italian crested newts ( T. carnifex ; 8 females, 14 males) maintained in our collection under the same conditions for 1 year before the start of the experiment were used in this study: for more information about these animals, see Fernández Meléndez et al. (2025). In brief, newts were randomly divided into 2 treatment groups: exposed to Bsal (BSAL, n = 11) and sham‐treated negative controls (CTRL, n = 11). All animals were housed individually on moist tissue at 15°C. A hiding place was provided and crickets (Acheta domestica) and buffalo worms ( Alphitobius diaperinus ) were given ad libitum. Before the start of the experiment (Day 0) all newts were weighed and photographed both ventrally and dorsally on millimetre paper to later measure individual snout‐vent length (SVL) using ImageJ v.1.54d. BSAL newts were then exposed to Bsal (AMFP13/01) through a 24‐h bath in 1 mL of 8.8 × 10³ Bsal spores (Martel et al. 2014). Upon exposure, ventral (to maximize chytrid detection; Berger et al. 1998; Pessier et al. 1999; Hyatt et al. 2007) skin swabs were collected weekly from each individual to quantify their epidermal Bsal load (CLASSIQSwabs 160C, COPAN). The experiment ended 8 weeks after Bsal exposure (Day 56). At Day 56, each individual was held with a new pair of nitrile gloves and a whole‐body skin microbiota sample was collected as described in Fieschi‐Méric et al. (2023). The newts were weighed and photographed again for SVL measurements. Animals were then euthanized through an intracoelomic injection of 20% sodium pentobarbital (KELA; except for one individual that died on Day 47). A sample from the ventral skin, as well as the spleen of each newt, was collected and immediately stored in RNAlater (Sigma‐Aldrich, Saint‐Louis, USA) at −80°C until further processing. One BSAL individual died before the end of the experiment, and another BSAL individual did not become infected: samples for microbiota and transcriptomic analyses were not collected from these newts (Table S1).

The experiment was performed in accordance with European law and with the approval of the ethical committee of the Faculty of Veterinary Medicine, Ghent University (approval number: EC2023‐05).

2.2. Bsal Load Quantification and Susceptibility Metrics

DNA was extracted from the weekly‐collected swabs for fungal detection using PrepMan Ultra Sample Preparation reagent (Thermo Fisher Scientific, Waltham, USA). The number of Bsal zoospores was estimated using quantitative real‐time PCR (qPCR), following Blooi et al. (2015). Zoospore quantification is traditionally used as an indicator of Bsal pathogen burden, but it does not reflect the full severity of chytridiomycosis disease and its physiological consequences on the host (Smith 2007; Clare et al. 2016). Thus, we used two proxies of Bsal susceptibility: pathogen burden (total Bsal load), measured as the area under individual log₁₀‐transformed Bsal load curves generated from weekly qPCRs, and disease severity, an index built from the combination of pathogen burden (total Bsal load), infection kinetics (latency from exposure to peak Bsal load), and body condition (body mass index: mass (g)/snout‐vent length (mm)²) variation between the start and the end of the experiment (index previously published in Fieschi‐Méric et al. 2025), adapted to include a survival indicator (time between the end of the experiment and the day of natural death, divided by the duration of the experiment). The joint use of these two metrics provides a comprehensive view of the ramifications of the infection on host physiology. The R code to build this index is available at Figshare repository (https://figshare.com/s/7fcdb4b8daf55dcfa89a).

2.3. Skin Microbiota DNA Extraction and Quantification

DNA was extracted from the whole‐body swabs for microbiota analysis using the DNeasy PowerSoil Pro kit (QIAGEN, Hilden, Germany). All DNA extractions were conducted following the manufacturer's instructions and included three non‐template controls (NTCs) as well as a mock community of known composition (ZymoBIOMICS Microbial Community Standard, ZYMO research, Irvine, USA).

Microbiota load was quantified through real‐time polymerase chain reactions (qPCRs). In silico PCRs conducted with published primers on the SILVA 16S database (Klindworth et al. 2013) suggested that the forward (5′‐ACT CCT ACG GGA GGC AGC AG‐3′) and reverse (5′‐TTA CCG CGG CTG CTG G‐3′) primers designed by Clifford et al. (2012) detected most of the representative bacterial groups from an example dataset of salamander skin microbiota. These primers were used in a 0.5 μM concentration to perform the qPCR with 2 μL of template DNA and a SensiFAST SYBR No‐ROX Kit (Bioline, London, UK) on a CFX384 detection system (Bio‐Rad). Amplification consisted of a pre‐denaturation step at 95°C for 10 min, followed by 40 cycles of denaturation (95°C for 30 s) and annealing (60°C for 1 min), finished with an extension step (stepwise increase of temperature to 95°C, at 0.5°C/5 s). Samples and a 10‐fold dilution series of positive controls (DNA from Clostridium perfringens ) ranging from 0 (NTC) to 10⁷ GE (genomic equivalent) per 2 μL were included in triplicate in each qPCR plate. The software CFX Maestro v2.3 (Bio‐Rad) was used to retrieve quantification cycle (Cq) data and to calculate the starting quantity (SQ) of 16S rRNA copies in the 2 μL of DNA used for each qPCR reaction. When Cq values varied by more than 0.5 between triplicates, the outlier reaction was discarded from the dataset. Following Vences et al. (2022), average microbiota load per swab was estimated as: (average SQ per sample—average SQ of DNA extraction NTCs) * 25 (to account for the total amount of 50 μL DNA extraction yield per sample).

2.4. Skin Microbiota Sequencing

In addition to the qPCRs, extracted DNA from the whole‐body swabs was also used for 16S library preparation. Following the protocol outlined in Aguirre et al. (2019), the V3–V4 hypervariable region of the 16S ribosomal RNA gene (~464 bp) was first amplified using the S‐D‐Bact‐0341‐b‐S‐17 (5′‐CCT ACG GGN GGC WGC AG‐3′) and S‐D‐Bact‐0785‐a‐A‐21 (5′‐GAC TAC HVG GGT ATC TAA TCC‐3′) primers (Klindworth et al. 2013) flanked with Nextera adapters that included two phosphorothioate bonds on each end for increased stability against exonucleases. Amplicons were purified using 50% CleanNGS beads (CleanNA, Waddinxveen, the Netherlands) before a second PCR was conducted to attach dual barcodes and flow‐cell binding sequences to the amplicons (Nextera XT Index Kit, Illumina, San Diego, USA). The final PCR products were purified again, using 41% CleanNGS beads. These purified barcoded libraries were then quantified using a Quantus double‐stranded DNA assay (Promega, Madison, WI, USA) before being combined into an equimolar 10 nM pool and sequenced on an Illumina MiSeq system (Illumina) at a depth of 70,000 reads per sample (2 × 300 bp, paired‐end) by Macrogen (Amsterdam, Netherlands). All microbiota sequencing data generated for the current study are available in the NCBI Sequence Read Archive (BioProject ID: PRJNA1265256).

2.5. Microbiota Bioinformatics

Demultiplexed bacterial sequences were processed using DADA2 v.1.8 (Callahan et al. 2016). Forward and reverse reads were truncated at decreasing quality (respectively 280 and 250 bp), and chimeric Amplicon Sequence Variants (ASVs) were removed by reconstruction against more abundant parent ASVs. Taxonomy was assigned to representative sequences using the SILVA database (Pruesse et al. 2007), clustering at 97% identity at the genus level, and at 100% at the species level (Edgar 2018). Preprocessing of the bacterial sequences was carried out using the R package phyloseq (McMurdie and Holmes 2013). Only bacterial sequences were kept, and 36 contaminant ASVs identified from NTCs were removed using the R package decontam (Davis et al. 2018). The resulting ‘raw dataset’ comprised 1679 ASVs and was used to compute Chao1 (estimated ASV richness) and Shannon (estimated ASV evenness) alpha diversity (within‐sample diversity) indexes, and to carry out Differential Abundance Analyses (DAA) using DESeq2 (Love et al. 2014). Spurious ASVs present in < 2 samples were filtered out (Bokulich et al. 2013), and because the sequencing depth varied by less than 2.4 times between samples, cumulative sum scaling normalization was applied to all samples to correct for overdispersion (Paulson et al. 2013). The resulting ‘normalized dataset’ comprised 654 ASVs and was used to measure beta diversity using Jaccard (compositional similarity), Bray–Curtis (abundance‐based compositional similarity), and weighted Unifrac distances (phylogenetic distance weighted by species abundance information) to investigate differences in community structure among samples (Lozupone et al. 2011). The sequences of relevant ASVs identified through DAA were blasted manually on the BLAST web tool (NCBI database) to verify taxonomy and collect additional information.

2.6. RNA Extraction and Sequencing

A subset of 7 BSAL (4 males, 3 females) and 7 CTRL (5 males, 2 females) individuals was selected for transcriptomics. Total RNA was extracted from 10 mg of each skin and spleen sample using the RNeasy Mini Kit (QIAGEN) with DNase I digestion (Bio‐Rad, Hercules, USA), as per manufacturer's instructions and including one NTC. Following quality check of the RNA extracts on a TapeStation (Agilent, Santa Clara, USA), mRNA libraries were constructed using the mRNA Library Prep Kit (Watchmaker Genomics, Boulder, USA) with stubby adapters in a 4 μM concentration (IDT, Newark, USA) and Unique Dual Indexes (xGen UDI barcoded adapters, IDT, Newark, USA). The quality of a subset of the barcoded libraries was checked on a TapeStation; they were then quantified using a Quantus double‐stranded DNA assay (Promega) and were pooled in an equimolar 45 nM pool, to be sequenced at a depth of 14,500,000 reads per sample (2 × 150 bp, paired‐end) on a NovaseqX 25B platform (Illumina) by Macrogen (Seoul, South Korea). All RNA sequencing data generated for the current study is available in the NCBI Sequence Read Archive (BioProject ID: PRJNA1266396).

2.7. RNA Bioinformatics

Raw reads were visualized using FastQC v.0.11.9 (Babraham Bioinformatics, Cambridge, UK) and cleaned using TrimGalore! v.0.6.10 to remove adapter content, delete reads shorter than 50 bp, and trim bases below a Q5 quality score, which is considered optimal for transcriptome assembly (MacManes 2014). A consensus reference transcriptome was assembled de novo from the cleaned reads using Trinity v.2.15.2 with all samples as input for a comprehensive species‐specific reference (Haas et al. 2013). The quality of this transcriptome assembly was refined using part of the TransPi pipeline (specifically, the evigene step; Rivera‐Vicéns et al. 2022) and was evaluated with BUSCO (Seppey et al. 2019; Table S2). Sequences were annotated with Entap v.2.0.0, both using the vertebrate translation code, in addition to a separate translation of remaining transcripts to identify mitochondrial genes, which use a distinct translation code. Any fungal or bacterial sequences were also removed at this step. EggNOG was used to cluster genes into orthologous groups and assign them with functional annotations using Gene Ontology (GO) Biological Process (BP) terms (Ashburner et al. 2000). BPs are most relevant for interpreting functional biological changes, so other GO‐terms (Molecular Function (MF) and Cellular Component (CC)) were excluded to reduce redundancy. Finally, SALMON v.1.10.3 was used for quantification, using only annotated genes as a reference. We made our T. carnifex transcriptome available on the NCBI Transcriptome Shotgun Assembly database (GLNC00000000).

The final RNA‐seq dataset consisted of 394,388,865 raw reads across 14 skin (199,147,414 raw reads) and 13 spleen (195,241,451 raw reads) tissue samples. The 14th spleen sample could not be exploited because it was contaminated with liver tissue during sampling (Table S1).

2.8. Statistical Analysis

Statistical analyses were performed using R version 4.3.2 (R Core Team 2023); all data and code are publicly available at Figshare repository (https://figshare.com/s/7fcdb4b8daf55dcfa89a). Statistical tests were deemed significant if associated with p‐values (adjusted following Benjamini–Hochberg false‐discovery rate correction) below a 0.05 threshold. Filtering thresholds for differential analyses were defined in line with the recommendations from the DESeq2 vignette. Specifically, we excluded bacterial taxa/genes with a count below 10 in less than x samples, where x was equal to the smallest group size for tests conducted on categorical variables, and x was equal to half of the total sample size for discrete variables. To ensure our conclusions were not an artefact of the chosen filtering thresholds, we conducted sensitivity analyses across a range of thresholds (x = 5 to x = 10; Tables [Link], [Link], [Link]) and focused our interpretation on features that were consistently significant across multiple thresholds.

In order to investigate the influence of Bsal exposure and of sex on average skin microbiota alpha diversity, a series of ANOVA tests were conducted on linear models that included microbiota load (log₁₀‐transformed), Chao1 or Shannon indexes as response variables, and treatment group (CTRL vs. BSAL) and sex as explanatory variables, in the form Microbiota~Treatment + Sex. Differences in inter‐individual variance (homoscedasticity) in microbiota alpha diversity between treatment groups or sexes were explored using F‐tests. Similarly, the influence of Bsal exposure and sex on microbiota structure was tested with PERMANOVAs on models that included a beta diversity index (Jaccard, Bray–Curtis, or weighted Unifrac distances) as the response variable, and treatment group and sex as the explanatory variables. Differences in within‐group variation in community structure between treatment groups or sexes were investigated using betadisper tests. Phylotypes that significantly differed in abundance between the microbiota of CTRL vs. BSAL individuals, and between male vs. female newts were identified through DAA using a log‐fold change (LFC) threshold of 1 (i.e., a minimum 2‐fold difference in abundance between groups) and poscounts normalization to account for the scarcity of microbiota data. Bacterial taxa with < 10 reads in < 9 (smallest group size: here, n BSAL newts) samples were excluded from DAA on the effect of Bsal, and taxa with < 10 reads in < 7 (n females) samples were excluded from DAA on the effect of sex.

Secondly, we investigated whether microbiota profiles of Bsal‐exposed newts were related to chytridiomycosis susceptibility, with tests in the form Microbiota~Susceptibility. To this end, Pearson correlation tests (or Kendall tests, when assumptions of normality were violated) were used to investigate the relation between Bsal susceptibility metrics (pathogen burden and disease severity) and microbiota load or alpha diversity. A series of PERMANOVA tests were used to examine the relation between each microbiota beta diversity index and each Bsal susceptibility metric. To identify ASVs potentially associated with Bsal susceptibility, DAAs were conducted for each susceptibility metric with the same parameters as described above, with adaptations for continuous response variables: an LFC of 0 and filtering out taxa with < 10 reads in < 4 (n newts/2) samples.

To investigate the influence of Bsal exposure and sex on gene expression, differential expression analyses (DEA) were conducted using DEseq2 with an LFC of 1, on linear models with treatment group (CTRL vs. BSAL) and sex as explanatory variables, in the form Gene expression ~Treatment + Sex. A separate model was built for each tissue type (skin vs. spleen). Genes with < 10 reads in < 5 (n female newts) samples were excluded from the DEA. A principal components analysis (PCA) was used to visualize global patterns of gene expression and samples clustering, using normalized gene counts (variance stabilizing transformation). Gene Set Enrichment Analyses (GSEA; Subramanian et al. 2005) based on GO‐terms were performed with clusterProfiler (Yu et al. 2012) to identify enriched biological processes in exposed newts compared to negative controls. To this end, all of the expressed genes were ranked based on the magnitude of change in their expression (signed LFC) between treatment groups (CTRL vs. BSAL): this ranked list served as input to the GSEA function, using default limits for gene sets size and 10,000 permutations.

Similarly, DEA with an LFC of 0 were conducted to determine whether gene expression in each tissue type was correlated with Bsal susceptibility (pathogen burden or disease severity), with models in the form Gene expression ~Susceptibility. Genes with < 10 reads in < 5 (n newts/2) samples were excluded from these DEAs. GSEA was carried out as described above to identify significant biological processes related to differences in susceptibility to Bsal.

3. Results

3.1. Microbiota and Transcriptomes Overview

Skin microbiota load varied between [309,118–7,917,951] 16S rDNA copies per swab. mRNA sequencing yielded between 5,669,467 and 48,239,204 mapped reads per individual tissue sample. The concatenated transcriptome was composed of 15,728 transcript isoforms clustered into 12,007 presumptive genes, of which 11,048 (92%) were successfully annotated with GO‐terms. The number of presumptive genes expressed varied between tissues: on average, 11,835 different genes were expressed in skin samples, while 11,520 were expressed in spleen samples. A total of 20,193 different GO‐terms were associated with the genes expressed in our dataset.

3.2. Susceptibility to Bsal Varies Among Individuals in Italian Crested Newts

Among BSAL newts, one individual did not become infected (0 GE at all times of sampling) and was therefore excluded from the following calculations. Over the 8 weeks following exposure, individual Bsal loads varied between [0–31,550] GE, with an average load of 967 GE and a median load of 4 GE per individual per week. Peak load (median: 1706 GE per individual) was reached at 6 weeks following exposure to Bsal. Pathogen burden (log₁₀‐transformed total Bsal load over 8 weeks following exposure) varied between [10.99–144.27] and disease severity varied between [−5.33 to 3.52]. One newt died 7 weeks after exposure to Bsal: this individual had the highest Bsal peak load, the greatest pathogen burden, and the highest disease severity index of all newts. All other individuals survived until the end of the experiment (Figure S1).

3.3. Exposure to Bsal Elicits Changes in Microbiota Structure, but Not in Alpha Diversity nor Phylogenetic Diversity

Average microbiota load was not influenced by Bsal exposure (F _(1,17)  = 3.07, p‐value = 0.097) nor by sex (F _(1,17)  = 1.70, p‐value = 0.210). Among‐individual variance in microbiota load was not different between treatment groups (F _(8,10)  = 0.99, p‐value = 0.996) nor between sexes (F _(6,12)  = 0.58, p‐value = 0.524) either. Average microbiota alpha diversity was not influenced by Bsal exposure (Chao1: F _(1,17)  = 0.19, p‐value = 0.666; Shannon: F _(1,17)  = 0.43, p‐value = 0.520) nor by sex (Chao1: F _(1,17)  = 0.01, p‐value = 0.915; Shannon: F _(1,17)  = 0.03, p‐value = 0.866). However, the among‐individual variance in microbiota alpha diversity was significantly higher in Bsal‐exposed newts compared to controls (Chao1, F _(8,10)  = 5.82, p‐value = 0.012; Shannon, F _(8,10)  = 9.02, p‐value = 0.002). Alpha diversity variance did not vary significantly between males and females (Chao1, F _(6,12)  = 1.93, p‐value = 0.313; Shannon, F _(6,12)  = 1.27, p‐value = 0.683; Figure S2A–C).

Microbiota structure was significantly affected by Bsal exposure in terms of compositional diversity (Jaccard: F _(1,19)  = 2.31, R ² = 0.11, p‐value < 0.001; Bray–Curtis, F _(1,19)  = 3.07, R ² = 0.15, p‐value < 0.001), but not in terms of phylogenetic diversity (PERMANOVA; weighted Unifrac, F _(1,19)  = 1.59, R ² = 0.08, p‐value = 0.144). Sex did not influence microbiota beta diversity (Jaccard: F _(1,19)  = 0.98, R ² = 0.05, p‐value = 0.453; Bray–Curtis, F _(1,19)  = 0.95, R ² = 0.05, p‐value = 0.505; weighted Unifrac, F _(1,19)  = 0.91, R ² = 0.05, p‐value = 0.463). Regardless of the beta diversity metric used, the compositional variance of the microbiota was homogenous between Bsal‐exposed and control individuals (Jaccard: F _(1,18)  = 1.39, p‐value = 0.253; Bray–Curtis: F _(1,18)  = 1.53, p‐value = 0.225; weighted Unifrac: F _(1,18)  = 1.05, p‐value = 0.329). The variance in microbiota structure was also similar between males and females (Jaccard: F _(1,18)  = 0.41, p‐value = 0.550; Bray–Curtis: F _(1,18)  = 0.10, p‐value = 0.773), except when measured with weighted Unifrac distances: in that latter case, the variance in beta diversity was higher in females compared to males (weighted Unifrac: F _(1,18)  = 6.41, p‐value = 0.023; Figure S2D–F).

Regardless of their treatment and sex, the skin microbiota of the newts was dominated by Proteobacteria (52%), Bacteroidota (32%) and Actinobacteriota (6%; Figure 1A; Tables [Link], [Link], [Link]). DAA suggested that a few ASVs were more abundant in Bsal‐exposed individuals compared to control newts: notably, Comamonas korrensis and one Flavobacterium sp. (Table S9). No taxa significantly differed in abundance between males and females.

FIGURE 1.

Mean relative abundance of the main phyla composing T. carnifex skin microbiota, averaged across all samples (A). Dominant phyla are identified in the legend. Correlations between Bsal disease severity and microbiota load (B) and evenness (C).

3.4. Chytridiomycosis Severity May Correlate With Microbiota Load, Evenness and Composition

Microbiota load and alpha diversity were not correlated with pathogen burden (Microbiota load: t = 1.86, df = 7, p‐value = 0.105; Chao1: t = −0.46, df = 7, p‐value = 0.659; Shannon: t = −0.85, df = 7, p‐value = 0.424; Figure S3A–C). Correlation tests suggested that disease severity was not correlated with microbiota richness (Chao1: T = 10, p‐value = 0.119; Figure S3D) but was positively correlated with microbiota load (z = 1.99, p‐value = 0.046) and negatively correlated with microbiota evenness (Shannon: T = 7, p‐value = 0.023; Figure 1B,C).

Microbiota structure was neither correlated with pathogen burden (Jaccard: F _(1,7)  = 1.24, p‐value = 0.098; Bray–Curtis: F _(1,7)  = 1.33, p‐value = 0.116; weighted Unifrac: F _(1,7)  = 0.92, p‐value = 0.466) nor with disease severity (Jaccard: F _(1,7)  = 1.25, p‐value = 0.063; Bray–Curtis: F _(1,7)  = 1.39, p‐value = 0.063; weighted Unifrac: F _(1,7)  = 0.57, p‐value = 0.708).

DAA suggested that the abundance of one Sphingobacterium sp. was positively correlated with pathogen burden (Table S10) and that 11 ASVs were correlated with disease severity: 2 taxa were negatively correlated, while 9 others (of which the same Sphingobacterium sp. as with pathogen burden) were positively correlated with disease severity (Table S11).

3.5. Bsal Exposure Is Associated With Enhanced Expression of Immunity Genes and Downregulation of Protein Synthesis and Skin Maintenance

Variation in gene expression between skin samples was primarily captured by the first principal component of the PCA, which corresponded with Bsal exposure (Figure S4A). DEA revealed that 10 genes (< 0.1%) were significantly more expressed in the skin of Bsal‐exposed newts compared to unexposed controls: these upregulated genes were mostly involved in immune processes, through mucous secretions (MUC5B), interferon responses (IRF1), MHC activation (CIITA, TAP1) and T‐cell proliferation (ARG1, IL2RA; Figure 2A; Figure S4C; Table S12). Only one gene significantly differed in expression between sexes, being more expressed in males (PKHD1L1; Figure S4B; Table S13).

FIGURE 2.

Heat‐map of differentially expressed genes (rows) in the skin of each newt (columns) with hierarchical clustering of samples (A). Tiles are coloured based on Z‐scores, from blue (lower than average expression) to red (higher), and samples are coloured by treatment (green: CTRL; purple: BSAL). Dotplot displaying the five most significantly enriched and five most significantly suppressed biological functions in the skin of Bsal‐exposed newts compared to controls (B). Dots are coloured based on p‐values and ranked by gene ratio (percentage of genes in the set that are enriched/suppressed in Bsal‐exposed newts). Gene count refers to the number of genes included in each GO‐term.

GSEA suggested that biological processes associated with innate (GO:0019221, GO:0001819) and adaptive immunity upregulation (GO:0002694, GO:0050778 and GO:0002253) were significantly enriched in the skin of Bsal‐exposed newts compared to controls, whereas skin development (GO:0043588, GO:0070268), keratinization (GO:0031424) and protein synthesis (GO:0002181, GO:0006413) were suppressed (Figure 2B; Figure S4D). These suppressed processes linked to skin development and keratinization notably included genes involved in epidermal differentiation (ALOXE3), keratinocyte proliferation (POU2F3), and the production of keratins and other intermediate filament proteins involved in cornification (KRT3, KAZN and RPTN; Figure S5). Overall, more GO‐terms were enriched (86%) than suppressed (14%) in the skin of Bsal‐exposed individuals compared to that of controls (Table S14).

Exposure to Bsal was largely associated with variation in gene expression between spleen samples, primarily along the first principal component of the PCA (Figure S6A). DEA conducted on spleen samples revealed that 15 genes (< 0.2%) were significantly more expressed in the spleen of Bsal‐exposed newts compared to controls: most of these genes were involved in inflammatory signalling (ADAM8) and T‐cells activation and apoptosis (CD300LF, IL2RA, IL2RB, FASLG, GZMA, LAT2, TARP, TRDC and TNFRSF25; Figure 3A; Figure S6C; Table S15). Two genes (PKHD1L1 and LACTB2) were significantly more expressed in males than in females (Figure S6B; Table S16).

FIGURE 3.

Heat‐map of differentially expressed genes (rows) in the spleen of each newt (columns) with hierarchical clustering of samples (A). Tiles are coloured based on Z‐scores, from blue (lower than average expression) to red (higher), and samples are coloured by treatment (green: CTRL; purple: BSAL). Dotplot displaying the five most significantly enriched and five most significantly suppressed biological functions in the spleen of Bsal‐exposed newts compared to controls (B). Dots are coloured based on adjusted p‐values and ranked by gene ratio (percentage of genes in the set that are enriched/suppressed in Bsal‐exposed newts). Gene count refers to the number of genes associated with each GO‐term.

In the spleen of Bsal‐exposed newts, the most significantly enriched biological processes were associated with cell‐signalling (GO:0038023, GO:0009986) and with the activation of the immune system (GO:0050778, GO:0002253, GO:0019221 and GO:0002694), whereas suppressed processes were predominantly related to protein synthesis (GO:0002181, GO:0006412 and GO:0005840; Figure 3B; Figure S6D). As for skin GSEA, more biological processes were enhanced (78%) than suppressed (22%) in BSAL compared to CTRL individuals (Table S17).

3.6. In the Skin, Bsal Susceptibility Correlates With the Upregulation of Immune Responses and Tissue Remodelling, and the Downregulation of Basal Metabolism

In the skin, pathogen burden was positively correlated with the expression of two genes involved in gene silencing (AGO1) and alternative splicing (RBFOX3), and negatively correlated with the expression of three genes related to homeostasis (OTOP2), vesicle trafficking (RAB9B), and immunity downregulation (KLRG1; Figure 4A; Table S18). Pathogen burden was also positively associated with GO‐terms related to the extracellular secretion of a collagen matrix (GO:0062023, GO:0005201) and to cellular division (GO:0007059, GO:0006260), and was negatively associated with GO‐terms related to protein synthesis (GO:0022626, GO:0006614; Figure S7A; Table S19).

FIGURE 4.

Skin expression of the top 5 genes most significantly correlated to Bsal susceptibility, measured in terms of pathogen burden (A) and disease severity (B) in T. carnifex . Splenic expression of the top 5 genes most significantly correlated to Bsal susceptibility, measured in terms of pathogen burden (C) and disease severity (D). Each dot represents a newt.

Disease severity was significantly correlated with changes in the expression of 83 genes in the skin: it was positively associated with the expression of 26 genes, such as immunity genes (WFD3, ZNF346), and negatively correlated with that of 57 genes notably involved in metabolism (PGAM2, PGK1; Figure 4B; Table S20). Disease severity was positively associated with biological processes related to cellular division (GO:0051052, GO:0006260) and immunity (GO:0019221, GO:0045089), and negatively related to processes involved in muscle cell production (GO:0043292, GO:0030017; Figure S7B; Table S21).

3.7. In the Spleen, Bsal Susceptibility Correlates With the Upregulation of Immune Responses and Tissue Remodelling, and the Downregulation of Protein Synthesis

Pathogen burden was positively correlated with the splenic expression of three genes, including the same alternative splicing gene observed in skin samples (RBFOX3). Nine genes were negatively correlated with pathogen burden in the spleen, notably genes related to immune regulation (TRIB1, IGJ and KLRG1), protein synthesis, and metabolism (EEF1A2, LACTB2; Figure 4C; Table S22). GSEA on spleen samples showed that biological processes related to the extracellular secretion of a collagen matrix and cell‐adhesion (GO:0062023, GO:0043062) were positively associated with pathogen burden, whereas processes related to protein synthesis were negatively associated with pathogen burden (GO:0022626, GO:0006412; Figure S7C; Table S23).

Disease severity was associated with changes in expression of 205 genes in the spleen: the expression of 117 genes related to metabolism (ABHD8), antibacterial defences (BPI) and the inflammasome (IRL2B, MMP8) was positively correlated with disease severity. In contrast, 88 genes, notably related to haemoglobin (HBE1, HBA), were negatively correlated with that metric (Figure 4D; Table S24). Biological processes related to cellular interactions and cell‐adhesion (GO:0009986, GO:0043062) were positively associated with disease severity, while processes related to protein synthesis were negatively associated with disease severity (GO:0006412, GO:0044391; Figure S7D; Table S25).

4. Discussion

This study sheds new light on the response of salamander hosts and of their resident microbial symbionts to chytrid pathogens. To the best of our knowledge, this is the first attempt to simultaneously investigate how skin microbiota and host gene expression are affected by Bsal exposure, in both a binary (controls vs. exposed newts) and continuous (among newts of variable Bsal susceptibility) manner. Our findings demonstrate that Bsal exposure affects amphibian hosts—with the upregulation of genes involved in immune activation, metabolism reduction and tissue remodelling in the host skin and spleen—but has limited effects on their resident skin symbionts. Together, these results lay the groundwork for future research on the determinants of variation in susceptibility to the deadly chytridiomycosis, which could open new perspectives for amphibian conservation.

4.1. Exposure to Bsal Affects Host Expression of Genes Related to Immunity, Metabolism Maintenance and Skin Keratinization, but Has Limited Effects on Host Microbiota

Exposure to Bsal affected all components of T. carnifex immunity, from changes in constitutive defences (enhanced mucous secretions, modifications of skin microbiota structure) to the upregulation of genes involved in innate immune responses (interferons, inflammatory cytokines cascades) and in adaptive immunity (MHC stimulation, leucocytes proliferation), both in the skin (site of infection) and spleen (primary lymphoid organ). This corroborates previous studies conducted on shorter terms, which reported subtle (Farrer et al. 2017) to strong (McDonald et al. 2020) upregulation of innate immunity genes following Bsal exposure in salamanders. Our study shows that adaptive immune responses also get activated on a longer term after exposure.

Despite this positive immune response to Bsal, it should be noted that the upregulation of immunity genes is not always associated with favourable disease outcomes: several intra‐ and interspecific studies on anurans suggest that sustained immune responses can be ineffective against chytrids, as they are often enhanced at the cost of cellular maintenance and of the upkeep of metabolic pathways (Eskew et al. 2018; Savage et al. 2020). Indeed, survival may depend on the maintenance of skin integrity and basal metabolism rather than on increased immune responses (Poorten and Rosenblum 2016; Ellison et al. 2020). Here, GSEA suggested that Bsal exposure suppresses metabolic maintenance, notably translation processes, both in the skin and spleen of T. carnifex . Our study does not allow us to determine whether this results from a strategy of the host to redirect its energy resources towards immunity activation, or from a direct weakening effect of Bsal on the gene expression of its host. In addition, we found that Bsal exposure suppressed the expression of genes encoding type I and II keratins, as well as proteins involved in the cornification and terminal differentiation of keratinocytes (repetin, kazrin; Alibardi 2022) and in skin development. This is coherent with the symptoms of skin erosion and massive tissue ulceration typically observed in Bsal‐infected animals and could also result from a host response to limit the proliferation of the fungus, since Bsal thalli often invade skin keratinocytes (Martel et al. 2013). As with any transcriptomic studies on non‐model organisms, it is important to interpret our results cautiously. Indeed, given the lack of T. carnifex ‐specific genomic resources, the identification and functional categorization of expressed genes in our study relies on annotations from other vertebrates: thus, genes or isoforms unique to our model species could be misannotated or not annotated at all, and functional assignments may sometimes be inaccurate. Moreover, while GSEA allows the interpretation of the extensive gene lists generated by DEA, this approach also comes with limitations: results are uncontrollably biased towards large and well‐annotated pathways, the presence of single multi‐functional genes can inflate the enrichment of multiple GO‐terms, and non‐coding RNA genes currently lack systematic annotations and are therefore excluded from pathway enrichment analyses (Reimand et al. 2019).

In addition to its effects on gene expression, Bsal exposure was associated with changes in skin microbiota structure, but only at deep taxonomic levels (as suggested by Unifrac distances). Prior studies showed that Bsal exposure alters microbiota composition and taxa abundance (Bates et al. 2019) without affecting its diversity or phylogenetic structure (Bletz et al. 2018), in contrast with Bd (Jani and Briggs 2014; Muletz‐Wolz et al. 2019). Whether this subtle Bsal‐associated microbiota dysbiosis results from the direct interaction between Bsal and bacterial communities or from the erosion of the keratinized epidermis which the microbiota interacts with remains to be determined. Given the limited influence of the skin microbiota on Bsal‐driven disease severity (Fieschi‐Méric et al. 2025), these changes may not be consequential for salamander hosts. Interestingly, one Flavobacterium sp. was more abundant in the microbiota of Bsal‐exposed individuals, but the limited taxonomic resolution of genomic reference databases and inconclusive BLAST alignment of this taxon did not allow its identification at the species level. The genus Flavobacterium comprises several pathogens for freshwater fish and amphibians (Green et al. 1999; Nematollahi et al. 2003; Xie et al. 2009) and is strongly associated with skin wounds during limb regeneration in axolotls (Altın et al. 2024), but also comprises protective symbionts with antifungal properties (Lauer et al. 2008; Shaw et al. 2014), so further research will be necessary to characterize the various roles of phylotypes in this genus in relation to amphibian skin diseases. Overall, we acknowledge that the ecological relevance of our findings regarding the microbiota may be limited, as our experimental newts were maintained ex‐situ for their whole lives and captivity is known to strongly alter the amphibian skin microbiota (Bates et al. 2019; Fieschi‐Méric et al. 2023).

4.2. Disease Susceptibility Correlates With Enhanced Expression of Immune and Tissue Remodelling Genes and With Reduced Skin Microbiota Diversity

Bsal disease susceptibility was correlated with the upregulation of genes involved in immune responses and extracellular matrix (ECM) production, and with the downregulation of translational activity. The ECM plays key roles in immunity, by acting as a barrier and facilitating the infiltration of immune cells (Theocharis et al. 2019; Sutherland et al. 2023; Diehl et al. 2025), and is also essential to the maintenance of tissue integrity, wound healing and regeneration processes in amphibians (Geng et al. 2015; Geyer et al. 2022). The positive association between disease severity and the enhanced expression of genes involved in cellular interactions, cell division and ECM production suggests that tissue remodelling may be a response to skin erosion in highly susceptible individuals. Our results suggest that individuals with moderate translational responses to infection developed less severe disease compared to newts that strongly redirected their energy resources towards the expression of genes related to immunity and tissue restructuration. In fact, an increased production of ECM may alter tissue integrity, and elevated immune responses could cause self‐inflicted damage to the host beyond the damage caused by the pathogen itself, and reduce the resources necessary to maintain homeostasis in the face of infection—a phenomenon known as ‘immunopathology’ (Grogan et al. 2020). However, given our limited sample size, our findings should be taken cautiously before additional work confirms these results.

Among its various functions, the ECM also mediates the interactions between microorganisms and their hosts via its structural components, including collagens, glycosaminoglycans and glycoproteins (Mouw et al. 2014). For example, ranaviruses attach to their amphibian hosts by binding to the glycosaminoglycans present in their ECM (Liu et al. 2020). Bacteria are also known to interact with animal ECMs (Chagnot et al. 2012): thus, the negative correlation between microbiota diversity and disease severity observed in our experimental newts could also partially result from the changes in their ECM structure. The positive relation between Bsal disease susceptibility and microbiota load is coherent with reports of bacterial overgrowth in Bsal‐infected salamanders (Martel et al. 2013; Vences et al. 2022), but more comprehensive reference databases are necessary to determine if these taxa are pathogenic or opportunistic. In particular, we found that disease severity may be related to the abundance of a Sphingobacterium sp.: although we could not identify that phylotype at the species level, a member of this genus is reported to be negatively correlated with Bd resistance in anurans (Bates et al. 2022). We emphasize again that further work with larger sample sizes and multiple host species should confirm our preliminary insights in the relation between Bsal susceptibility and host immunity.

The diverse roles of animal ECMs are still being uncovered, and more research is needed to further characterize their functions in amphibians. Our results suggest that modifications of the ECM may have important implications in chytridiomycosis dynamics in salamanders. This finding corroborates a recent study that demonstrated the interaction between Bsal and salamanders' epidermal glycoproteins (Wang et al. 2021). Determining whether ECM dysregulation may result from an impaired host response against tissue erosion, or from the direct manipulation of host gene expression by Bsal to enhance substrate availability constitutes a promising avenue for future research.

5. Conclusion

This study demonstrates that Bsal exposure affects multiple components of T. carnifex immunity, from limited structural changes of the microbiota to enhanced expression of immune‐related genes at the site of infection (skin) and in their primary lymphoid organ (spleen). Interestingly, disease severity may be associated with the upregulation of genes involved in immunity and in tissue restructuration, suggesting a reallocation of resources from metabolic maintenance towards immune responses and tissue remodelling. The functional consequences of these transcriptional shifts remain to be tested experimentally. Through its multi‐omics approach, this study provides a comprehensive foundation for future investigations into chytridiomycosis susceptibility—an essential research area for amphibian conservation.

Author Contributions

Léa Fieschi‐Méric: investigation, methodology, formal analysis, writing – original draft. Kevin P. Mulder: investigation, formal analysis, writing – review and editing. Eduardo Fernández Meléndez: methodology. Sarah Van Praet: methodology. Sofie De Bruyckere: methodology. Michael Fahrbach: methodology. Frank Pasmans: conceptualization, supervision, resources, writing – review and editing. An Martel: conceptualization, supervision, methodology, resources, writing – review and editing.

Funding

This work was supported by ERC (ERC Advanced Project: 101096163—GLOSSI) and EFSA funding (Project AMPHIDEB: OC/EFSA/SCER/2021/12). K.P.M. was supported by an FWO postdoctoral fellowship under grant number 1224223N.

Disclosure

Benefit‐sharing statement: This research addresses a priority concern for biodiversity conservation: the emerging infectious disease Bsal that causes amphibian declines. Benefits from this research accrue from the sharing of our data and results on public databases as described above.

Ethics Statement

This work was performed following European law and with the approval of the ethical committee of the Faculty of Veterinary Medicine, Ghent University (approval number: EC2023‐05).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Individual values of Bsal load over time (A) and of Bsal‐susceptibility calculated at the end of the experiment, as pathogen burden (B) and disease severity (C). Each individual is identified by a unique colour throughout the three plots.

Figure S2: Influence of Bsal exposure and sex on the alpha‐diversity (A–C) and beta‐diversity (D–F) of T. carnifex skin microbiota. Dots colour indicate the treatment (green: negative controls; purple: Bsal‐exposed) and dots shape indicates the sex of the newt (rounds: females; triangles: males).

Figure S3: Relation between Bsal susceptibility (pathogen burden and disease severity) and microbiota load and alpha‐diversity (Chao1 and Shannon) following Bsal exposure. Relations between disease severity and pathogen burden/microbiota evenness are presented in Figure 1.

Figure S4: Principal component analysis showing differences between newts based on their epidermal gene expression profiles (A). Differentially expressed genes in the skin of females compared to males (B) and in Bsal‐exposed compared to control newts (C). Each dot represents a newt. Boxes display the lower and upper quartiles, and the line splitting boxes shows the median normalized gene count. Network plot displaying the biological functions (yellow pies) most significantly affected by Bsal‐exposure and their associated genes (D). Blue hues indicate the fold‐change in expression of these genes in the skin of BSAL newts compared to CTRL.

Figure S5: Epidermal expression of the genes included in the GO‐term ‘keratinization’ (A) and ‘skin development’ (B) in Bsal‐exposed compared to control newts. Each dot represents a newt (box: lower and upper quartiles; line: median normalized gene count).

Figure S6: Principal component analysis showing differences between newts based on their splenic gene expression profiles (A). Differentially expressed genes in the spleen of females compared to males (B) and Bsal‐exposed compared to control newts (C). Each dot represents a newt. Boxes display the lower and upper quartiles, and the line splitting boxes shows the median normalized gene count. Network plot displaying the biological functions (yellow pies) most significantly affected by Bsal‐exposure and their associated genes (D). Blue and red hues indicate the fold‐change in expression of these genes in the spleen of BSAL compared to CTRL.

Figure S7: Dotplots displaying the five biological functions in the epidermis most positively/negatively associated with pathogen burden (A) and disease severity (B). Dotplots displaying the five biological functions in the spleen most positively/negatively associated with pathogen burden (C) and disease severity (D). Dots are coloured based on adjusted p‐values and ranked by gene ratio (percentage of genes in the set that are affected). Gene count refers to the number of genes associated with each GO‐term.

Table S1: Information about the samples used in this study. Read counts are raw (not normalized). The spleen sample of Newt 19 was excluded due to contamination with liver tissue during sampling.

Table S2: BUSCO assessment scores of completeness for our T. carnifex transcriptome assembly.

Table S3: Sensitivity analysis showing the impact of different DAA filtering thresholds on the bacterial taxa detected as significantly differing in abundance across tests. The column ‘ASV’ indicates the differentially abundant taxa after bacteria with a count below 10 in less than x samples were filtered out, where x is indicated in the column ‘Filtering threshold’.

Table S4: Sensitivity analysis showing the impact of different DAA filtering thresholds on the skin genes detected as significantly differing in abundance across tests. The column ‘Genes’ indicates the differentially abundant genes after those with a count below 10 in less than x samples were filtered out, where x is indicated in the column ‘Filtering threshold’.

Table S5: Sensitivity analysis showing the impact of different DAA filtering thresholds on the spleen genes detected as significantly differing in abundance across tests. The column ‘Genes’ indicates the differentially abundant genes after those with a count below 10 in less than x samples were filtered out, where x is indicated in the column ‘Filtering threshold’.

Table S6: Average relative abundance of each bacterial phylum in the skin microbiota of T. carnifex , regardless of their experimental group or sex.

Table S7: Average relative abundance of each bacterial phylum in the skin microbiota of T. carnifex , per experimental group (CTRL or BSAL).

Table S8: Average relative abundance of each bacterial phylum in the skin microbiota of T. carnifex , by sex.

Table S9: Differentially abundant bacterial phylotypes in the microbiota of Bsal‐exposed newts compared to negative controls. The sign of the log₂‐fold change in abundance of these bacteria (log₂foldchange) indicates whether these phylotypes are more (positive sign) or less (negative sign) abundant in BSAL compared to CTRL individuals.

Table S10: Differentially abundant bacterial phylotype in the microbiota of Bsal‐exposed newts with various pathogen burdens. The sign of the log₂‐fold change (log₂foldchange) indicates that the abundance of this phylotype was positively correlated with pathogen burden (Bsal load).

Table S11: Differentially abundant bacterial phylotypes in the microbiota of Bsal‐exposed newts with various disease severity. The sign of the log₂‐fold change (log₂foldchange) indicates whether the abundance of these phylotypes was positively (positive sign) or negatively (negative sign) correlated with disease severity.

Table S12: Differentially expressed genes in the skin of Bsal‐exposed newts compared to negative controls. The sign of the log₂‐fold change in expression (log₂foldchange) indicates that these genes are over‐expressed (positive sign) in BSAL compared to CTRL individuals.

Table S13: Differentially expressed gene in the skin of female compared to male newts. The sign of the log₂‐fold change in expression (log₂foldchange) indicates that this gene is under‐expressed (negative sign) in females compared to males.

Table S14: Differentially enriched biological processes in the skin of Bsal‐exposed newts compared to negative controls. The sign of the normalized enrichment score (NES) indicates whether these processes (GO‐terms) are enhanced (positive sign) or suppressed (negative sign) in BSAL compared to CTRL individuals. Set size refers to the number of studied genes associated with each function.

Table S15: Differentially expressed genes in the spleen of Bsal‐exposed newts compared to negative controls. The sign of the log₂‐fold change in expression (log₂foldchange) indicates that these genes are over‐expressed (positive sign) in BSAL compared to CTRL individuals.

Table S16: Differentially expressed genes in the spleen of female compared to male newts. The sign of the log₂‐fold change in expression (log₂foldchange) indicates that these genes are under‐expressed (negative sign) in females compared to males.

Table S17: Differentially enriched biological processes in the spleen of Bsal‐exposed newts compared to negative controls. The sign of the normalized enrichment score (NES) indicates whether these processes (GO‐terms) are enhanced (positive sign) or suppressed (negative sign) in BSAL compared to CTRL individuals. Set size refers to the number of studied genes associated with each function.

Table S18: Differentially expressed genes in the skin of newts with different pathogen burdens. The sign of the log₂‐fold change in expression (log₂foldchange) indicates whether the expression of these genes was positively (positive sign) or negatively (negative sign) correlated with pathogen burden (Bsal load).

Table S19: Differentially enriched biological processes in the skin of Bsal‐exposed newts with different pathogen burdens. The sign of the normalized enrichment score (NES) indicates whether these processes (GO‐terms) are positively or negatively associated with pathogen burden (Bsal load). Set size refers to the number of studied genes associated with each function.

Table S20: Differentially expressed genes in the skin of newts with different disease severities. The sign of the log₂‐fold change in expression (log₂foldchange) indicates whether the expression of these genes was positively (positive sign) or negatively (negative sign) correlated with disease severity.

Table S21: Differentially enriched biological processes in the skin of Bsal‐exposed newts with different disease severities. The sign of the normalized enrichment score (NES) indicates whether these processes (GO‐terms) are positively or negatively associated with disease severity. Set size refers to the number of studied genes associated with each function.

Table S22: Differentially expressed genes in the spleen of newts with different pathogen burdens. The sign of the log₂‐fold change in expression (log₂foldchange) indicates whether the expression of these genes was positively (positive sign) or negatively (negative sign) correlated with pathogen burden (Bsal load).

Table S23: Differentially enriched biological processes in the spleen of Bsal‐exposed newts with different pathogen burdens. The sign of the normalized enrichment score (NES) indicates whether these processes (GO‐terms) are positively or negatively associated with pathogen burden (Bsal load). Set size refers to the number of studied genes associated with each function.

Table S24: Differentially expressed genes in the spleen of newts with different disease severities. The sign of the log₂‐fold change in expression (log₂foldchange) indicates whether the expression of these genes was positively (positive sign) or negatively (negative sign) correlated with disease severity.

Table S25: Differentially enriched biological processes in the spleen of Bsal‐exposed newts with different disease severities. The sign of the normalized enrichment score (NES) indicates whether these processes (GO‐terms) are positively or negatively associated with disease severity. Set size refers to the number of studied genes associated with each function.

Fieschi‐Méric, L. , Mulder K. P., Fernández Meléndez E., et al. 2026. “Differential Immune Responses Correlate With Chytridiomycosis Severity in Italian Crested Newts.” Molecular Ecology 35, no. 13: e70438. 10.1111/mec.70438.

Data Availability Statement

Our de novo assembled T. carnifex transcriptome has been deposited in the NCBI Transcriptome Shotgun Assembly (TSA; accession number: GLNC00000000) and raw sequence reads in the Sequence Read Archive (SRA; BioProject ID: PRJNA1265256 & PRJNA1266396). Metadata and R code are publicly available at Figshare repository (https://figshare.com/s/7fcdb4b8daf55dcfa89a).