Abstract

   Emerging evidence suggests a link between alterations in the gut
   microbiome and adverse health outcomes in the hosts exposed to
   environmental pollutants. Yet, the causal relationships and underlying
   mechanisms remain largely undefined. Here we show that exposure to
   biotoxins can affect gut pathobiome assembly in amphibians, which in
   turn triggers the toxicity of exogenous pollutants. We used Xenopus
   laevis as a model in this study. Tadpoles exposed to tropolone
   demonstrated notable developmental impairments and increased locomotor
   activity, with a reduction in total length by 4.37%–22.48% and an
   increase in swimming speed by 49.96%–84.83%. Fusobacterium and
   Cetobacterium are predominant taxa in the gut pathobiome of
   tropolone-exposed tadpoles. The tropolone-induced developmental and
   behavioral disorders in the host were mediated by assembly of the gut
   pathobiome, leading to transcriptome reprogramming. This study not only
   advances our understanding of the intricate interactions between
   environmental pollutants, the gut pathobiome, and host health but also
   emphasizes the potential of the gut pathobiome in mediating the
   toxicological effects of environmental contaminants.

   Keywords: Gut microbiome, Pathobiome, Biotoxin pollution, Developmental
   toxicity, Behavioral disorders

Graphical abstract

   [47]Image 1
   [48]Open in a new tab

Highlights

     * •
       The amphibians populations in paddies decline with increasing
       tropolone levels.
     * •
       Tropolone exposure leads to behavioral and developmental disorders
       in tadpoles.
     * •
       Core gut pathobiome taxa are identified in tropolone-exposed
       tadpoles.
     * •
       Gut pathobiome drivers toxicity from tropolone exposure.

1. Introduction

   Amphibians play an essential role in sustaining ecosystem fitness and
   stability [[49]1]. However, more than 500 amphibian species are
   currently facing population declines and extinctions at a global scale
   [[50]2,[51]3]. These ecological consequences have been proposed to be
   associated with various environmental pollutants, such as
   agrochemicals, heavy metals, and emerging biotoxins [[52][4], [53][5],
   [54][6], [55][7]]. As the largest agroecosystem and wetland ecosystem
   on Earth, the paddy ecosystem is a dominant habitat for amphibians
   [[56]8]. However, a wide array of phytopathogen-derived biotoxins, such
   as tropolone (TR) [[57]9], ustiloxins [[58]10,[59]11], and pyricuol
   [[60]12], have been frequently detected in soils and sediments,
   especially in aquatic compartments [[61]13]. Biotoxin pollution has
   become a growing global concern due to its adverse impact on public
   health, but its ecological effects on amphibian populations are still
   underestimated.

   TR is a predominant biotoxin produced by the agricultural pathogen
   Burkholderia plantarii (B. plantarii). It is characterized as a
   seven-membered aromatic ring compound with the molecular formula
   C[7]H[6]O[2] ([62]Fig. S1), displaying high solubility in both organic
   solvents and water [[63]9]. As one of the most commonly identified
   biotoxins in agroecosystems, TR has been consistently detected in most
   paddy fields, with concentrations ranging from 14 to 157 μg L^−1, as
   observed in a continental-scale pilot monitoring [[64]9]. This
   B. plantarii-derived biotoxin has been observed to exhibit varying
   degrees of acute toxicities toward microbes, higher plants, and mammals
   [[65][14], [66][15], [67][16]]. Furthermore, we recently observed a
   negative correlation between the population of frogs, the predominant
   amphibian species, and exposure to trace levels of TR (ranging from 0.3
   to 23.3 μg L^−1) in the paddy aquatic environment. ([68]Fig. 1a,
   [69]Fig. S2, Dataset 1). This suggests that exposure to this prevalent
   biotoxin during the developmental stages of frogs, particularly the
   sensitive early developmental stages, could adversely affect the
   overall population stability of frogs within paddy ecosystems. While
   tadpoles are commonly recognized as the most vulnerable early
   developmental stage to environmental contaminants throughout a frog's
   lifespan, the specific impacts of TR under environmentally relevant
   exposures remain largely unknown.

Fig. 1.

   [70]Fig. 1
   [71]Open in a new tab

   Alterations in the density of frogs in paddy fields and development and
   locomotor activity of the tadpoles upon exposure to biotoxin. a,
   Regression analysis of the density of the frog population and the
   biotoxin tropolone (TR) levels in different region-specific rice
   paddies (Zhejiang, Jiangsu, and Fujian). b–e, Effects of biotoxin
   tropolone (TR) exposure (TR at 0, 1, 10, and 100 μg L^−1) on the total
   length (b), hindlimb length (c), weight (d), and body length (e) of
   tadpoles (n = 6 replicates). f–g, Comparison of swimming speed (f) and
   average activity (g) of tadpoles in the control (CK: TR at 0 μg L^−1)
   and TR-exposed groups (TR at 1, 10, and 100 μg L^−1) (n = 18
   replicates). h, Visualization of the swimming speed and average
   activity of tadpoles in the control (TR at 0 μg L^−1) and TR-exposed
   groups (TR at 1, 10, and 100 μg L^−1). The whole tracking of tadpoles
   recorded for 10 min is shown. Different letters with error bars
   indicate a significant difference based on a one-way analysis of
   variance (ANOVA) with least significant difference (LSD) test
   (p < 0.05).

   Like most animals, gut microbiome homeostasis plays a vital role in
   maintaining the development and health of amphibians [[72][17],
   [73][18], [74][19]]. Antibiotics, drugs, and pesticides are examples of
   external stimuli that have been shown to negatively affect the
   homeostasis of the gut microbiome, resulting in a shift from a
   symbiotic microbiome to a dysbiotic microbiome [[75]20]. A dysbiotic
   microbiome is generally associated with a decreasing (or potentially
   decreasing) health status of hosts and with the onset of diseases and
   pathological intestinal microenvironments associated with obesity,
   malnutrition, and chronic inflammatory diseases [[76][21], [77][22],
   [78][23], [79][24]]. Recently, a dysbiotic gut microbiome resulting
   from the disruption of a symbiotic gut microbiome that promotes health
   and is environmentally stable has been conceptualized as the pathobiome
   [[80]25,[81]26]. Recent research has shown that the gut pathobiome
   influences immune modulation in sepsis patients following surgical
   injury [[82]27]. Preliminary data from several perspectives suggest a
   direct link between the gut pathobiome and a maladaptive immune
   response in incritical illness [[83]28]. Apart from the characterized
   associative relationship between the pathobiome and disrupted host
   health, little is known about the causal relationship between them,
   particularly upon exposure to extrinsic cues at the environmental
   levels.

   In this study, we utilized the amphibian frog model organism Xenopus
   laevis (X. laevis) to investigate how exposure to trace levels of TR
   influences host development and behavior, specifically focusing on gut
   microbiome alterations. Exposure to TR impaired development and
   increased locomotor activity in X. laevis. Further, through integrating
   high-throughput data, behavioral and transcriptomic analyses, and
   microbiome transplantation experiments, we uncovered the core taxa
   within the pathobiome and elucidated the mechanism underlying the
   pathobiome-mediated host dysfunction. Our study unravels the previously
   hidden role of the gut pathobiome in the ecotoxicity caused by
   environmental pollutants to the host.

2. Materials and methods

2.1. Chemicals and reagents

   Tropolone (TR), 2-hydroxy-2,4,6-cycloheptatrien-1-one, with a purity of
   98.0%, was obtained from TCI. Ethylacetate (EtOAc) and other organic
   solvents (chromatographic grade) were purchased from Merck. Agar powder
   was purchased from Wako. Tricaine mesylate (MS-222) and dimethyl
   sulfoxide (DMSO) were purchased from Sigma (St. Louis, USA). The DisQue
   kit, purchased from Waters, was used to extract and purify TR.

2.2. Investigation of the amphibian population and quantification of TR in
situ

   To investigate the relationship between TR concentrations and amphibian
   population, we conducted a pilot study in Zhejiang, Jiangsu, and Fujian
   Provinces in China. The population of frogs was calculated in 100
   plants per paddy field via the chessboard-type sampling method. The
   paddy water (2 L) was collected at each point by using a five-point
   sampling method and subjected to determination of TR after filtration
   through No. 101 filter paper (Advantec) in a Buchner's funnel
   [[84]9,[85]29,[86]30] ([87]Fig. S3). The extraction and purification of
   TR were performed according to previous methods [[88]8], and further TR
   was subjected to analysis as follows. The filtered paddy water
   (approximately 50 mL) was transferred into a 250 mL separatory funnel.
   After 1 min of homogenization, an aliquot of the upper layer (25 mL)
   was pipetted into a 50 mL DisQuE centrifuge tube preloaded with 6 g of
   MgSO[4] and 1.5 g of CH[3]COONa. The mixture was then centrifuged again
   at 4000 rpm for 1 min, vigorously shaken for 1 min, and subsequently
   centrifuged at 4000 rpm for 1 min. For further purification, the
   supernatant was concentrated by a rotatory evaporator at 45 °C. The
   residue was redissolved in 2 mL EtOAc and filtered through a 0.22 μM
   filter for quantitative analysis of TR by gas chromatography
   (GC)-triple quadrupole mass spectrometry (QqQ, MS/MS) (Agilent 7000C,
   USA). The qualification and quantification of TR were referred to the
   previously established methods in the laboratory [[89]7]. All
   procedures and operations in the field trial were conducted according
   to the code of Good Agricultural Practices that was issued by the
   Ministry of Agriculture of China and without the involvement of any
   endangered wildlife species or protected areas of land.

2.3. Exposure experiment

   X. laevis, as an amphibian model organism [[90]31], was selected for
   further exposure experiments. X. laevis tadpoles at stage 51 were
   selected for the TR exposure experiments at the real environmental
   levels (1 and 10 μg L^−1) and ten-fold higher level (100 μg L^−1) for
   21 days, following the Nieuwkoop and Faber system [[91]32]. Different
   concentrations of TR solution were prepared using DMSO. Both the
   control and exposure groups received identical concentration of DMSO
   (0.001%, v/v). Throughout the experiment, the tadpoles were maintained
   in 30 L glass tanks under stable conditions, with a 22 ± 1 °C
   temperature and a light–night cycle (12 h light, 12 h dark). The pH and
   conductivity of the exposure solution were maintained between 7.0 and
   8.0 and 315–330 μS cm^−1, respectively. In each glass tank, the
   dissolved oxygen concentration was maintained at or above 4 mg L^−1,
   and the ammonia concentration was less than 0.2 mg L^−1. The exposure
   media (20 L) was prepared using dechlorinated carbon-filtered water,
   and 50% of the tank volume was replaced with fresh exposure media daily
   to maintain TR concentrations. There were three replicates in each
   group, with 18 tadpoles per tank.

2.4. Phenotype measurements

   During the exposure experiment, phenotype measurements, wet weight,
   total length, body length, and hindlimb length were recorded after
   exposure for seven days and the end of the exposure experiment (day
   21), respectively. After the exposure experiment, the tadpoles were
   anesthetized with 175 mg L^−1 MS-222 and rapidly frozen in liquid
   nitrogen, then stored at −80 °C for the subsequent experiment. The care
   and maintenance of the tadpoles adhered to the established guidelines
   following the Amphibian Metamorphosis Assay [[92]33].

2.5. Locomotion assays

   Locomotor activity was assayed with a Video-Track system (ViewPoint
   Life Science, Montreal, France) according to previously established
   methods [[93]34,[94]35]. The tadpoles were selected randomly and
   transferred into individual transparent culture tanks (150 × 150 mm) to
   measure the locomotor activity. After a 2-min acclimation period, the
   data of mobile frequency, mobile distance, and total movement tracking
   were collected per treatment for 10 min at ten frames per second, after
   which further data analysis was performed. Additionally, the locomotor
   behavior of tadpoles in the 0 μg L^−1 group was chosen as a control.

2.6. Total community DNA extraction

   The gut of tadpoles (three tissues per replicate) was performed to
   extract total community DNA with a FastDNA SPIN Kit for Soil (MP
   Biomedicals, USA) according to the manufacturer's instructions. The
   purity and concentration of DNA were measured by a NanoDrop ND-100
   spectrophotometer (ALLSHENG, Hangzhou, China), and the quality was
   assessed by agarose gel electrophoresis. Total community DNA extraction
   was stored at −20 °C for follow-up processing.

2.7. Amplicon libraries preparation and sequencing

   The V3–V4 region of the 16S rRNA gene fragments was amplified via
   polymerase chain reaction (PCR) following the Earth Microbiome
   guidelines [[95]36]. PCR amplification of 16S rRNA gene fragments for
   microbiome studies was performed using the primers listed in
   [96]Table S1. After quality control and adaptor ligation, equimolar DNA
   concentrations of each barcoded amplicon sample were subjected to
   next-generation sequencing on the Illumina HiSeq-PE250 platform.

2.8. Microbiome analyses

   The acquired data were processed using QIIME2 v.2020.11.0 [[97]37].
   Cutadapt was employed to assess the quality and demultiplex the quality
   of paired-end reads. The microbiome of the gut of the control and
   treatment tadpoles was subjected to detailed profiling, with six
   replicates and 85,074–92,170 reads per sample group. After filtering
   and splicing all the samples, the clean tags were clustered, and
   chimeras were further removed for clustering into the Operational
   Taxonomic Unit (OTU) by UPARSE [[98]38]. The representative sequences
   of the OTUs were selected and compared with the database to obtain
   species annotation information. Sequence counts ranged from 67,776 to
   76,601 reads per sample, and we obtained a total of 328 bacterial OTUs
   at 97% identity after sequence processing and filtering the clean tags.

2.9. Isolation of intestinal bacteria

   The gut of the tadpoles were ground into powder with liquid nitrogen,
   and sterilized ultrapure water was subsequently added. The sterilized
   ultrapure water was purified using a Milli-Q Water Purification System
   (Merck Millipore, Germany). After the mixture was settled, 1 mL of the
   supernatant was diluted to five concentration gradients. Subsequently,
   a fastidious anaerobe agar (FAA, Shanghai Rui Chu Biological Technology
   Co., Ltd., Shanghai, China) medium was then spread with 100 μL of the
   supernatant. The FAA medium was incubated in an AnaeroPack (Mitsubishi
   Gas Chemical America, New York, USA) containing 5% CO[2]
   AnaeroPack·CO[2] (Mitsubishi Gas Chemical America, New York, USA).
   Suitable culture conditions and media were selected to screen for
   specific bacteria based on microbiome analysis. According to the
   results of the morphological analysis, visible bacterial colonies were
   screened and streaked onto FAA media to isolate single colonies.

2.10. Identification of intestinal bacteria

   Genomic DNA was extracted with a Takara MiniBEST Bacterial Genomic DNA
   Extraction Kit Ver.3.0 (TaKaRa Biotechnology, Japan) according to the
   manufacturer's instructions. The purity and concentration of the DNA
   were measured using a NanoDrop ND-100 spectrophotometer. The quality
   and the integrity of the DNA were checked by gel electrophoresis. For
   bacterial identification, we performed PCR amplification of 16S rRNA
   gene fragments using universal primers ([99]Table S1). Subsequently, we
   conducted multiple alignments of the data and phylogenetic analysis of
   the sequences using MEGAX. A phylogenetic tree was constructed using
   the neighbor-joining method. Meanwhile, the alignment of the 16S RNA
   gene sequences of the isolated strains and the V3–V4 sequences of the
   OTU2 (Fusobacterium) and OTU12 (Cetobacterium) were consistent by
   Clustal X and GeneDoc software.

2.11. RNA-seq and transcriptome analysis

   As directed by the manufacturers, the total RNA of tadpoles was
   extracted using a standard TRIzol reagent (Invitrogen, America). Oligh
   (dT) beads are used to remove rRNA and enhance mRNA. The enriched mRNA
   was fragmented into short fragments (200–300 bp). The first-strand of
   cDNA was reverse transcribed with random hexamers, and the
   second-strand cDNA was synthesized. Subsequently, the cDNA libraries
   were prepared from the total RNA of tadpoles and further sequenced on
   an Illumina novaseq 6000 platform. The raw data were cut for adapter
   sequence, and low-quality reads were filtered out to generate clean
   data using FASTP ([100]https://github.com/OpenGene/fastp). With HISAT2
   ([101]http://ccb.jhu.edu/software/hisat2/index.shtml), the
   quality-filtered reads were mapped to the reference assembly accession
   (GCF_001663975.1), with a total of 34,034,694–38,678,970 mapped reads
   per sample. Fragments per kilobase per million fragments (FPKM) values
   were added, a pseudocount of 1, and the results were log[2] transformed
   to calculate the gene expression. The log[2] fold change was used to
   calculate the differences in fold changes between the control and
   treatment groups by the DESeq 1.32.0 package. Differentially expressed
   genes (DEGs) were identified based on the criteria of |log[2] fold
   change| > 1 and false discovery rate (FDR) < 0.05. For further
   analysis, the DEGs were subjected to the enrichment analysis of KEGG
   pathways and network analysis visualization using Cytoscape (3.8.0).

2.12. Real-time quantitative polymerase chain reaction (RT-qPCR) analyses

   To extract RNA from the intestinal tissues of tadpoles, a homogenate of
   three tissues per replicate was prepared on ice with TRIzol reagent.
   The extracted of RNA was subjected to reverse transcription using
   Primer-Script RT reagent kit (TaKaRa Biotechnology, Japan).
   Subsequently, RT-qPCR was conducted on the ABI 7500 real-time PCR
   system (Bioer Technology, Shanghai, China) following the manufacturer's
   instructions. The specific primers for RT-qPCR were designed using
   Primer Premier 5.0 software and the National Center for Biotechnology
   Information (NCBI) online tool, as shown in [102]Table S1. To ensure
   that the input RNA in each sample was the same for RT-qPCR, the 18S
   ribosomal RNA (18S rRNA) gene was selected as intestinal reference
   genes. The 2^−△△Ct method was chosen to calculate the relative gene
   expression.

2.13. Quantification PCR of the microbial taxa

   To quantify the relative abundance of Fusobacterium sp. and
   Cetobacterium sp. in the gut of tadpoles, FastDNA SPIN Kit for Soil (MP
   Biomedicals, USA) was used to extract total community DNA and specific
   primers ([103]Table S1) was used to quantify the amplification of the
   16S rRNA gene by qPCR. The qPCR was performed using ChamQ Universal
   SYBR qPCR Master Mix (Vazyme, China).

2.14. Colonization experiments of bacteria in the gut of tadpoles

   X. laevis tadpoles at stage 51 were selected for the colonization
   experiments. After acclimatization to the environment, the tadpoles
   were randomly divided into three groups, each equally divided into
   three replicates with 18 tadpoles per tank. The diet preparation was
   performed as described previously with appropriate modifications
   [[104]39]. The experimental groups were administered Fusobaterium and
   Cetobacterium feeds, respectively. Only sterilized feeds were fed to
   the control group. Throughout the experiment, the exposure media (20 L)
   was prepared using sterilized dechlorinated carbon-filtered water, and
   50% of the tank volume was replaced with fresh exposure media daily.
   Sterilized water was detected that there were no bacteria in it.

2.15. Statistical analyses

   The data were analyzed via One-way Analysis of Variance (ANOVA) and
   student's t-test analysis (p < 0.05) to evaluate statistical
   significance. Version 26.0 of the Statistical Package for the Social
   Sciences (SPSS) was used to evaluate and calculate the significance of
   the differences between the control and treatment groups. Cytoscape
   (3.8.0) was used for network visualization. Alpha diversity and
   principal coordinate analysis (PCoA) analyses were performed on Tutools
   plantform ([105]https://www.cloudtutu.com).

2.16. Data availability

   All the raw sequence data have been deposited in the Sequence Read
   Archive of NCBI. The gut microbiome data of tadpoles was deposited
   under the accession PRJNA910976. Transcriptome datasets were deposited
   under accession PRJNA910979. All other raw data for all figures and
   tables are available from the corresponding authors upon reasonable
   request.

3. Results

3.1. Impaired development of the host upon exposure to the biotoxin

   To investigate the effect of TR on frog development, we measured a
   series of fundamental development indexes, including the total length,
   hindlimb length, body length, and wet weight of tadpoles exposed to TR
   at the real environmental levels (1 and 10 μg L^−1) and at a ten-fold
   higher level (100 μg L^−1). After seven days of TR exposure, there was
   a downward trend (the p values for total length, body length, hindlimb
   length, and weight were 0.334, 0.777, 0.051, and 0.284, respectively)
   in the 1 μg L^−1 group compared with those in the control group (in the
   absence of TR) ([106]Fig. S4). After further continuous observation
   until the end of the exposure experiment, compared with the control
   group, we observed statistical differences in the development indexes
   between the control and the TR-exposed groups at both 1 and
   10 μg L^−1 TR. In the control group, the total length of tadpoles
   measured 7.10 ± 0.33 cm, while it dose-dependently decreased to
   6.09 ± 0.50 and 5.50 ± 0.79 cm under 1 and 10 μg L^−1 of TR,
   respectively ([107]Fig. 1b). Interestingly, an increase in total length
   was observed in the tadpoles upon exposure to 100 μg L^−1 TR in
   comparison with those upon exposure to 1 or 10 μg L^−1 TR, but the
   difference was not statistically significant compared with that in the
   control group ([108]Fig. 1b).

   TR exposure induced a similar pattern in the total length, hindlimb
   length, and wet weight of the tadpoles ([109]Fig. 1b–d). Furthermore,
   the body length of the 1 and 10 μg L^−1 groups was reduced compared
   with the control group, while the body length of the 100 μg L^−1 group
   was not significantly different from the control group ([110]Fig. 1e).
   Compared with that in the control group, there was a developmental
   delay observed in the treatment groups ([111]Table S2). These results
   indicated that exposure to TR at the environmental levels could
   imparied the development of tadpoles.

3.2. Host locomotor activity increases upon exposure to TR

   In the TR exposure experiment (equivalent to the environmental exposure
   level), the behavior of the tadpoles changed as well, and their
   swimming speed showed a visually increasing trend in contrast to the
   control group. Because locomotor activity is also a sensitive indicator
   of developmental toxicity, we conducted a precise behavioral assay. As
   a result, we found that exposure to 1, 10, and 100 μg L^−1 TR
   significantly increased the average swimming speed of the tadpoles.
   Specifically, the swimming speed of the tadpoles exposed to 1, 10, and
   100 μg L^−1 TR significantly increased by 49.96%, 84.83%, and 57.58%,
   respectively ([112]Fig. 1f).

   Moreover, compared with the control group, tadpoles exposed to 1, 10,
   and 100 μg L^−1 concentrations of TR showed 3.11-, 4.71-, and 3.74-fold
   increases in average activity, respectively ([113]Fig. 1g). The
   locomotor activity of the tadpoles increased for the entire tracking
   duration of 10 min ([114]Fig. 1h) or every 10 s ([115]Fig. S5).
   Interestingly, TR at 1–100 μg L^−1 did not exert a dose-dependent
   impact on either speed or activity ([116]Fig. 1f and g), and the
   maximum impact of TR was observed at the level equivalent to the
   environmental exposure level ([117]Fig. 1f and g). These results
   indicate that the behavior of tadpoles exposed to TR at the
   environmental concentration levels was drastically altered,
   characterized by increased locomotor activity.

3.3. Host developmental disorders are associated with a shift in the gut
symbiome to the pathobiome

   To investigate whether the gut microbiome changed in the
   development-impaired hosts, we further analyzed the community structure
   and composition of the gut microbiome. Analysis of alpha diversity,
   including the Chao 1 richness index and Shannon index, showed that the
   abundance and diversity of the gut microbiome in tadpoles exposed to TR
   (10 μg L^−1) were significantly lower than those in the control group
   ([118]Fig. 2a). Using beta diversity-based PCoA, we observed a
   significant difference in community composition between the control and
   exposure groups ([119]Fig. 2b, Dataset 2). In addition, through a
   rank-abundance curve, it was found that the control group had higher
   species abundance and more uniform species composition than the
   TR-exposed groups ([120]Fig. S6a). Remarkably, the exposure group had a
   higher proportion of dominant bacteria, and the diversity of species
   was lower than that of the control group ([121]Fig. S6b).

Fig. 2.

   [122]Fig. 2
   [123]Open in a new tab

   Characterization of the gut microbial community in the biotoxin-exposed
   tadpoles. a–b, Analysis of the gut microbial community structure based
   on alpha (a) and beta (b) diversities. Significant differences were
   observed between the control (TR at 0 μg L^−1) and TR-exposed groups
   (10 μg L^−1). n = 6 replicates. c–d, Comparison of the gut microbial
   community composition between the control (TR at 0 μg L^−1) and
   TR-exposed group (10 μg L^−1) at the phylum (c) and genus (d) levels.
   n = 6 replicates. c and f indicate unassigned taxa at the taxonomic
   levels of class and family, respectively. e, UpSet plot of the gut
   microbial community in the TR-exposed (10 μg L^−1) tadpoles at the
   genus level. The length of the red bars (bottom left) indicates the
   total size sets of the genera. The blue symbols, connected with blue
   lines, represent the intersections between the sets and the number. The
   yellow columns indicate the frequency of these intersections.

   We further analyzed the composition of the microbial community in the
   control and exposure groups. At the phylum level, the gut microbiome of
   tadpoles in the control and exposure groups was dominated by
   Bacteroidetes (mean, 25.70% and 21.05% in the control and exposure
   groups, respectively) and Firmicutes (mean, 23.72% and 26.49% in the
   control and exposure groups, respectively) ([124]Fig. 2c, [125]Fig. S7a
   and b, [126]Table S3). Consequently, the further detailed resolution
   analysis revealed a distinct gut microbial composition in the exposure
   group ([127]Fig. S8a and b, [128]Fig. S9, Dataset 3), with a marked
   change at the genus level ([129]Fig. 2d and e). This indicates that
   exposure to TR results in a shift from a healthy intestinal microbiome
   to a pathobiome in tadpoles.

   To infer the potential core taxa of the pathobiome, a co-occurrence
   network of microbial communities in the tadpoles’ gut was visualized
   ([130]Fig. 3a). We defined the taxa with relative abundances higher
   than 0.01 as the core taxa of the gut microbiome. The network linked 12
   core taxa, including Fusobacterium, Cetobacterium, Parabacteroides,
   Eubacterium, Bacteroides, Desulfovibrio, Akkermansia, Odoribacter,
   Romboutsia, Acetobacterium, Lachnospiraceae, and Tannerellaceae
   ([131]Fig. 3a), among which Acetobacterium, Cetobacterium,
   Fusobacterium, and Lachnospiraceae were inferred as the core taxa of
   the pathobiome due to their significant enrichment in the TR-exposed
   group ([132]Fig. 3b–[133]Table S4.

Fig. 3.

   [134]Fig. 3
   [135]Open in a new tab

   Identification of core taxa of the gut pathobiome in biotoxin-exposed
   tadpoles. a, Co-occurrence network of the microbial community of the
   gut pathobiome in the tadpoles upon exposure to 10 μg L^−1 TR. Nodes
   indicate the top 20 most abundant taxa in each treatment, with node
   size indicating relative abundance. The pie graphs at nodes show the
   proportion of each treatment, and labels demonstrate taxa. c, o, and f
   indicate unassigned taxa at the taxonomic levels of class, order, and
   family, respectively. b, Analysis of the relative abundance of the core
   taxa of the gut pathobiome in the tadpoles. Student's t-test
   (two-tailed). ∗p < 0.05, ∗∗p < 0.01. The values are presented as the
   means ± s.d. (shown as error bars, n = 6).

3.4. Core pathobiome taxa mediate the TR-induced developmental disorders in
the host

   To untangle the causality between the TR-induced developmental
   impairment and increased locomotor activities in tadpoles and the shift
   of the symbiome toward a pathobiome, we conducted a preliminary
   comparison involving the tadpoles fed isolated bacteria in the absence
   of TR. Using culture-dependent isolation, we obtained 154 bacterial
   isolates from the gut of developmentally disordered tadpoles.
   Interestingly, only 16 bacterial isolates had effects similar to those
   observed under TR-induced excitability ([136]Fig. 4a, [137]Fig. S10).
   Phylogenetic analysis based on the 16S rRNA sequence showed high
   homology with Fusobacterium sp. (ten isolates) and Cetobacterium sp.
   (six isolates) ([138]Figs. S11 and 12), respectively. Furthermore, the
   16S rRNA gene sequences of these isolates were consistent with the core
   pathobiome taxa identified in the microbiome profiling ([139]Figs. S13
   and 14, Dataset 4).

Fig. 4.

   [140]Fig. 4
   [141]Open in a new tab

   Core gut pathobiome taxa responsible for the altered development and
   locomotor activity of the biotoxin-exposed tadpoles. a, Whole tracking
   of tadpoles recorded for 10 min is shown for the control group and
   those transplanted with Fusobacterium or Cetobacterium groups. b–c,
   Quantitative analysis of the colonization of Fusobacterium (b) and
   Cetobacterium (c) in the gut of the tadpoles after microbial
   transplantation. n = 3 replicates. d–g, Effects of the transplantation
   of Fusobacterium and Cetobacterium on the total length (d), body length
   (e), hindlimb length (f), and weight (g) of the tadpoles. n = 12
   replicates. h–i, Effects of the transplantation of Fusobacterium and
   Cetobacterium on the swimming speed (h) and average activity (i) of the
   tadpoles. n = 19 for each. Different letters with error bars indicate a
   significant difference based on ANOVA with LSD test (p < 0.05).

   To clarify whether TR directly affects the tadpoles or indirectly
   affects them through the gut pathobiome, the tadpoles were fed with the
   inferred core taxa in the gut pathobiome. After quantification
   polymerase chain reaction (qPCR) confirmed the successful colonization
   of Fusobacterium and Cetobacterium in the intestinal tissue
   ([142]Fig. 4b and c). Further precise quantitative analysis revealed
   that the total length of the tadpoles fed Fusobacterium and
   Cetobacterium were 5.50 ± 0.18 and 5.08 ± 0.14 cm, respectively, which
   were lower than that of in contrast with the control group
   (5.71 ± 0.31 cm) ([143]Fig. 4d). Similarly, there was a marked decrease
   in the body length, hindlimb length, and wet weight of the tadpoles in
   the treatment groups ([144]Fig. 4e–g). In addition, we found that the
   average swimming speed and average activity of tadpoles increased in
   the groups fed Fusobacterium and Cetobacterium. Specifically, the
   average speed of the tadpoles increased obviously by 31.78% and 31.09%,
   respectively ([145]Fig. 4h), and excitability was significantly induced
   ([146]Fig. 4i). These results together indicate that the developmental
   disorders induced by TR were mediated by the core pathobiome taxa
   inhabiting the tadpoles’ gut.

3.5. The gut pathobiome core taxa-reprogrammed transcriptome underlies host
developmental and behavioral disorders

   We further explored the global gene expression in the
   development-disordered tadpoles by transcriptomic analyses. Each pair
   of groups under the same treatment had an r-value of 1.00 in the
   heatmap with correlation coefficient analysis ([147]Fig. S15a). The
   data repeatability for subsequent analysis was indicated by an r value
   of 0.9923 for log[2]-transformed normalized gene expression between the
   control and treatment groups by a linear regression analysis
   ([148]Fig. S15b). Each sample's gene body coverage profile demonstrated
   no bias in the control or treatment groups ([149]Fig. S16a-f). The
   principal component analysis (PCA) of the RNA sequencing revealed that
   the TR-exposed samples significantly differed from those not exposed to
   TR ([150]Fig. S17a). The UpSet plot and volcano plot illustrated the
   significant differences in gene expression between the control and
   treatment groups ([151]Fig. S17b and c). Furthermore, compared with the
   control group, there were alterations in the expression of 1403 genes
   in the TR treatment group based on RNAseq analysis; among them, 961
   genes were significantly upregulated (log[2]-transformed fold change
   >1, p < 0.05) (Datasets 5), while 442 genes were significantly
   downregulated (log[2]-transformed fold change < −1, p < 0.05) (Dataset
   6). We used all the DEGs in the top 10 KEGG pathways to generate a
   column clustering heatmap, showing obvious gene expression differences
   between the control and treatment groups ([152]Fig. S18, Dataset 7).
   The up- and down-regulated genes were further subjected to KEGG pathway
   enrichment analysis. We found 75 pathways enriched by significantly
   downregulated genes, five of which were significantly enriched
   ([153]Fig. 5a, Dataset 8). In contrast, 115 pathways were enriched by
   upregulated genes, among which 15 pathways were significantly enriched
   ([154]Fig. 5a, Dataset 9). A scatter plot of the enriched KEGG pathways
   showed 16 significantly enriched pathways among all DEGs
   ([155]Fig. S19, Dataset 10).

Fig. 5.

   [156]Fig. 5
   [157]Open in a new tab

   Transcriptome reprogramming in the gut pathobiome-harboring tadpoles.
   a, KEGG pathways with enrichment of genes whose expression was
   significantly upregulated and downregulated genes in the TR-treated
   group (10 μg L^−1), respectively. The number on the x-axis represents
   the p values of the enrichment of the KEGG pathway by
   log[10]-transformed. The quantity of DEGs in each enrichment pathway is
   exhibited by the numbers in the plot, and the proportion of all genes
   is also indicated in the pathway (in parentheses). b, Network analysis
   of the DEGs associated with the top ten KEGG enrichment pathways. The
   circle size denotes the relative fold change (log[2]-transformed) of
   gene expression. c–d, The colonization of Fusobacterium (c) and
   Cetobacterium (d) in the gut of the tadpoles upon exposure to
   10 μg L^−1 TR. n = 3 for replicates. e–f, The expressions of gene cldn2
   (e) and fabp3 (f) in the TR-exposed tadpoles (10 μg L^−1). n = 5 for
   replicates. g–h, The gene cldn2 (g) and fabp3 (h) expressions in the
   tadpoles fed with Fusobacterium and Cetobacterium, respectively. n = 5
   for replicates. The values are presented as the means ± s.d. Student's
   t-test (two-tailed), p values were shown on the top of the paired
   columns.

   The DEGs enriched in the top ten KEGG pathways were subjected to
   network analysis ([158]Fig. 5b, Dataset 11). The major clusters were
   the peroxisome proliferator-activated receptor (PPAR) signaling pathway
   and tight junction (TJs), while the remaining clusters were related to
   metabolism. The high induction of a set of PPAR and TJ genes was
   associated with a dramatic increase in the population density of
   Fusobacterium and Cetobacterium ([159]Fig. 5c and d). These genes,
   included claudin 2 L homeolog (cldn2) and fatty acid binding protein
   3 L homeolog (fabp3), were induced which are associated with the tight
   junction pathway and PPAR signaling pathway, respectively ([160]Fig. 5e
   and f, Dataset 12). Through RT-qPCR, we further found that PPAR (fabp3)
   and TJ (cldn2) gene expressions were both induced by the
   transplantation of either Fusobacterium or Cetobacterium, verifying the
   causative impact of Fusobacterium and Cetobacterium on the induction of
   PPAR (fabp3) and TJ (cldn2) gene expression ([161]Fig. 5g and h). These
   results together profiled the gut pathobiome-triggered transcriptome
   reprogramming that underlies developmental and behavioral disorders in
   the host.

4. Discussion and conclusion

   As climate change continues, there is a growing concern that the
   prevalence of phytopathogens and the associated biotoxin pollution in
   aquatic environments may persist or worsen [[162]2,[163]5]. For a long
   time, research on biotoxin pollution and other environmental pollutants
   (intentionally or unintentionally introduced into the environment) has
   focused on the direct harm to exposed plant and animal communities.
   Recently, a pioneering study shed light on the disruption of the normal
   metabolic rhythm of the gut microbiome in host mammals with a high-fat
   diet, indicating a potential link between the gut pathobiome and
   diseases caused by extrinsic factors [[164]40]. A study suggested that
   the gut pathobiome could exacerbate systemic inflammation and
   behavioral disorders in the neurological disease CADASIL [[165]41].
   However, the causality between the gut pathobiome and host diseases,
   especially upon exposure to emerging biotoxin pollution, remained
   largely unexplored and elusive. Based on the model amphibian organism
   X. laevis, we uncovered a previously unknown causal role of the gut
   microbiome in frog population decline attributed to biotoxin pollution.
   In this study, the biotoxin TR was found to drive the assembly of the
   gut pathobiome, which further led to locomotor excitability and
   developmental disorders in the tadpoles.

   In animals, locomotor activity is critical in growth and development,
   which was previously reported to be modulated by the gut microbiome
   [[166]42,[167]43]. Our current study found that increased locomotor
   activity ([168]Fig. 1f and g) and induced developmental disorders
   ([169]Fig. 1b–e) were present in the tadpoles upon exposure to the
   biotoxin TR. Furthermore, it was highlighted that the gut microbiome
   was markedly altered in development-disordered tadpoles upon TR
   exposure, in which Acetobacterium, Cetobacterium, Fusobacterium, and
   Lachnospiraceae were identified as significantly enriched taxa in the
   pathobiome ([170]Fig. 2, [171]Fig. 3). To investigate whether TR
   affects tadpoles directly or indirectly through the gut pathobiome, we
   performed a colonization experiment by feeding tadpoles bacterial
   isolates corresponding to the inferred gut pathobiome taxa. In the
   absence of TR, Fusobacterium and Cetobacterium similarly caused
   developmental disorders and increased locomotor activity ([172]Fig. 4),
   suggesting that they are the causal core taxa of the gut pathobiome. A
   variety of clinical presentations identified the genus Fusobacterium as
   a causative agent in serious diseases, such as colitis and periodontal
   disease [[173][44], [174][45], [175][46]]; while the genus
   Cetobacterium was also reported as a possible pathogen in fish species,
   such as Nile tilapia, zebrafish [[176]47], and colorectal cancer in
   humans [[177]48]. Our study has shed light on the causality of these
   core taxa in the gut pathobiome, which triggered the excitability of
   locomotor activity and developmental disorders.

   By transcriptome profiling-guided analysis, we explored the mechanism
   underlying the disordered activities and development triggered by the
   pathobiome, and observed a significant enrichment of the PPAR signaling
   pathway along with the TR-induced pathobiome ([178]Fig. 5a,b,
   [179]Fig. S19). PPARs can influence the gut mucosal homeostasis and are
   pivotal in modulating host pathogenesis by influencing the gut and
   whole-body immune response [[180]49]. Interestingly, PPARs also engage
   in binary crosstalk with the gut microbiome [[181]50]. For instance,
   Firmicutes and Fusobacteria were potent activators of PPARs in the gut
   [[182]50], while the dysregulation of PPARs was involved in the
   dysbiosis of the gut microbiome in Trichinella spiralis-infected mice
   [[183]51]. Further analysis revealed fabp3, a representative PPAR gene
   expressed in the brain, was markedly enriched in both the TR- and
   pathobiome-exposed groups ([184]Fig. 5f–h), along with the accumulation
   of dopamine (DA) ([185]Fig. S20) and the downregulation of D2 dopamine
   receptor (drd2) B isoform X3 (Dataset 13). Activation of drd2 generally
   decreases dopamine release and further reduces locomotor activity, and
   dopamine-related locomotor activity is controlled by the binding of
   fabp3 to drd2 in the brain [[186]52,[187]53]. Hence, these results
   indicated that the core taxa of the gut pathobiome induced the
   promotion of locomotor activity, possibly through the tuning of the
   fabp3-associated dopamine and PPAR signaling pathways through the
   gut–brain axis ([188]Fig. 6).

Fig. 6.

   [189]Fig. 6
   [190]Open in a new tab

   Proposed schematic illustration of developmental and behavioral
   disorders mediated by biotoxin exposure-driven assembly of the gut
   pathobiome. The left panel of the figure shows that the biotoxin TR
   exposure drives the shift of the normal gut microbiome to the
   pathobiome, which causes developmental and behavioral disorders in the
   tadpoles. The detailed mechanism underlying the gut pathobiome-mediated
   disorder is shown in the right panel of the figure. In the reprogrammed
   transcriptome, a series of genes in the gut–brain axis involving fabp3,
   cldn2, cldn15, and drd2, differentially express and trigger the
   excitability of locomotor activity and morphogenesis impairment in
   tadpoles. The upregulated genes are indicated with red ellipses, and
   the downregulated genes are denoted with green. The red arrow
   demonstrates an increased level of dopamine.

   Moreover, along with the TR-induced pathobiome, we also noticed that
   the TJ pathway was enriched in the transcriptome network
   ([191]Fig. 5b). In fact, some TJ protein-coding genes, such as claudin
   3, are reported to be transcriptionally regulated by PPARs [[192]54].
   Tight junctions are crucial in maintaining epithelial barrier function
   [[193]55] and are composed of cytoplasmic plaque proteins and integral
   transmembrane proteins. These proteins include zonular occludens (ZOs),
   which link the TJs to the cytoskeleton and occludin, junctional
   adhesion molecules (JAMs), and claudins, all of which define the
   properties of the paracellular pore [[194]56]. Some claudins, such as
   claudin-2, are known to form pores that allow specific ions to pass
   preferentially, while others, such as claudin-1, -3, -4, -5, and -7,
   prevent the formation of specific ions and enhance the intestinal
   barrier [[195][57], [196][58], [197][59]]. An increase in paracellular
   permeability is a causal factor of chronic diseases, such as
   inflammatory bowel disease (IBD) [[198]60]. A previous study suggested
   that claudin-2 expression was elevated in IBD and IBD-associated
   dysplasia [[199]61]. Extracellular vesicle-derived claudin-2 serves as
   a biomarker to predict colorectal cancer liver metastases in humans
   [[200]62]. The gut microbiome can upregulate the expression of cldn2
   (claudin-2) in the gut of mice by secreting Staphylococcal enterotoxin
   B (SEB) and cholera toxin (CT) [[201]63]. Interestingly, we observed
   that cldn2 (claudin 2 L homeolog) was upregulated in the TR- and
   pathobiome-exposed groups ([202]Fig. 5e–g, Dataset 13). This is the
   possible reason that the genera Fusobacterium and Cetobacterium induce
   inflammation in the gut by regulating cldn2.

   In addition, claudin-15 (cldn15) is an essential regulator of
   paracellular Na^+ flux and transcellular nutrient absorption
   [[203]64,[204]65], and is upregulated in celiac disease (CD) [[205]64].
   Previous studies have demonstrated that the dysfunction of tight
   junctions (TJs) caused by disordered expression of cldn2 and cldn15 is
   involved in the pathogenesis of diarrheal diseases
   [[206]58,[207]66,[208]67]. Diarrheal diseases are characterized by
   increased gut permeability, regulated by TJ proteins [[209]68].
   Diarrheal diseases or malabsorptive diseases are leading causes of
   malnutrition and can even lead to death in severe cases. Our study
   found that the expressions of cldn2 and cldn15 were significantly
   upregulated in the TR-exposed tadpoles. These results indicate that
   dysfunction of the TJ pathway, induced by the core taxa of the gut
   pathobiome in tadpoles, can result in malnutrition and further
   developmental impairment ([210]Fig. 6).

   In the past decade, biotoxin pollution has emerged as a growing global
   concern, impacting public health and potentially affecting frog
   populations. In our current study, we have discovered the role of the
   gut pathobiome in mediating locomotor activity and developmental
   impariments in frogs exposed to environmental biotoxin pollution.
   Additionally, our research has revealed that two core pathobiome taxa,
   Fusobacterium and Cetobacterium, can directly disrupt frog health even
   without biotoxin exposure, potentially by interfering with critical
   pathways in the gut–brain axis. These present findings not only
   emphasize the gut pathobiome as the endogenous trigger for the toxicity
   of exogenous pollutants, but also provide a fresh perspective on the
   ongoing decline of frog populations in the context of global change.

Ethics approval and consent to participate

   All animal procedures were approved by the Experimental Animal Ethics
   Committee of Zhejiang University.

CRediT authorship contribution statement

   Qianqian Pan: Conceptualization, Data Curation, Formal Analysis,
   Methodology, Software, Visualization, Writing - Original Draft.
   Tianxing Lv: Data Curation, Investigation. Haorong Xu: Data Curation,
   Investigation. Hongda Fang: Data Curation. Meng Li: Methodology.
   Jiaping Zhu: Investigation, Methodology. Yue Wang: Investigation.
   Xiaoyan Fan: Methodology. Ping Xu: Investigation. Xiuguo Wang:
   Investigation. Qiangwei Wang: Conceptualization, Methodology. Haruna
   Matsumoto: Conceptualization, Software, Supervision, Writing - Review &
   Editing. Mengcen Wang: Conceptualization, Formal Analysis, Funding
   Acquisition, Project Administration, Writing - Review & Editing.

Declaration of competing interest

   The authors declare that they have no known competing financial
   interests or personal relationships that could have appeared to
   influence the work reported in this paper.

Acknowledgements