Dysbiosis associated with acute helminth infections in herbivorous youngstock – observations and implications

A plethora of data points towards a role of the gastrointestinal (GI) microbiota of neonatal and young vertebrates in supporting the development and regulation of the host immune system. However, knowledge of the impact that infections by GI helminths exert on the developing microbiota of juvenile hosts is, thus far, limited. This study investigates, for the first time, the associations between acute infections by GI helminths and the faecal microbial and metabolic profiles of a cohort of equine youngstock, prior to and following treatment with parasiticides (ivermectin). We observed that high versus low parasite burdens (measured via parasite egg counts in faecal samples) were associated with specific compositional alterations of the developing microbiome; in particular, the faecal microbiota of animals with heavy worm infection burdens was characterised by lower microbial richness, and alterations to the relative abundances of bacterial taxa with immune-modulatory functions. Amino acids and glucose were increased in faecal samples from the same cohort, which indicated the likely occurrence of intestinal malabsorption. These data support the hypothesis that GI helminth infections in young livestock are associated with significant alterations to the GI microbiota, which may impact on both metabolism and development of acquired immunity. This knowledge will direct future studies aimed to identify the long-term impact of infection-induced alterations of the GI microbiota in young livestock.

The relationships between infection burdens, anthelmintic treatment and altered abundances of specific microbial taxa. Differences in the relative abundance of specific microbial taxa between groups were evaluated using Linear discriminant analysis Effect Size (LEfSe) and correlations between individual bacterial families, parasite infection burden and sample collection time points were represented by Pearson's www.nature.com/scientificreports www.nature.com/scientificreports/ correlation network analysis (Tables 1-3; Fig. 3). At D0, Bacteroidetes were reduced in C-high compared to C-low (LDA effect size 4.66), which was largely attributable to a reduced abundance of bacteria belonging to the families Prevotellaceae and Paraprevotellaceae, and genus Alistipes (Family Rikenellaceae) ( Table 1; Fig. 3a). However, in C-high, a relative expansion of bacteria belonging to the families Mogibacteriaceae and Leuconostocaceae (genus Weissella), and genus Paraeggerthella belonging to the class Clostridia (phylum Firmicutes) was observed when compared with C-low (Table 1; Fig. 3a). In the latter group, several taxa belonging to the class Clostridia, including Anaerotruncus, Pseudobutyrivibrio and Unclassified Acidaminobacteraceae, were increased compared with C-high (Table 1; Fig. 3a). When assessing the impact of anthelmintic treatment on the faecal microbial composition of yearlings enrolled in our study, LEfSe analysis revealed a significant decrease of bacteria belonging to the phylum Firmicutes in C-high at D14 when compared to D0, which was largely attributable to a reduction in Mogibacteriaceae, Dehalobacteriaceae, Lactobacillaceae, Streptococcaceae and Enterococcaceae (Table 2; Fig. 3b); in contrast, Prevotellaceae were increased in faecal samples collected from C-high at D14 compared with D0 ( Table 2; Fig. 3b). A transient expansion of bacteria belonging to the families Lachnospiraceae, Clostridiaceae and Succinovibrionaceae was also observed in C-high at D2 (Table 2; Fig. 3b). In C-low, Betaproteobacteria (order) and Verrucomicrobia (phylum) were reduced and increased, respectively, at D14 when compared to D0 and D2 (Table 3; Fig. 3c). Similar to samples collected from C-high, bacteria belonging to the families Lachnospiracaeae, Clostridiaceae and Succinivibrionaceae were increased at D2 compared with D0 and D14, respectively (Table 3; Fig. 3c). Of the several bacterial taxa whose abundance was significantly different in samples from C-high and C-low collected pre-and post-anthelmintic treatment, four (i.e. Mogibacteriaceae, Prevotellaceae, Paraprevotellaceae and Rikenellaceae; analysed individually using ANOVA at family level) were significantly affected at D14 following the administration of ivermectin in C-high, whilst remaining unchanged in C-low at the same time point (Fig. 4).
faecal metabolite changes associated with high-and low infection burdens and anthelmintic treatment. A total of 28 faecal metabolites in all samples from C-high and C-low collected at D0 and D14 were identified and quantified by Proton Nuclear Magnetic Resonance ( 1 H-NMR). These were subjected to PCoA ( Supplementary Fig. S6) and CCA; the latter allowed us to identify differences between the faecal metabolomes of C-high and C-low at D0 (P = 0.059; F = 1.97) (Fig. 5a). Pearson's correlation heatmap analyses demonstrated increased abundances of isobutyrate, trehalose, leucine, phenylalanine, glutamate, glucose, lysine and propionate, and decreased nicotinate, valerate and butyrate, in C-high versus C-low (Fig. 6a); however, these differences were not statistically significant according to ANOVA following False Discovery Rate (FDR) correction for multiple testing (P > 0.05 for all metabolites) (Supplementary Table S2). CCA also detected a difference between Figure 2. Faecal microbial richness is reduced in animals with high parasite burdens, and increased postanthelmintic treatment. Microbial alpha diversity (based on Operational Taxonomic Unit [OTU] presence and abundance), measured by richness, evenness and Shannon index, in faecal samples from (a) C-high (≥100 eggs per gram (e.p.g.)) versus C-low (≤10 e.p.g.), (b) C-high at day 0 (D0) versus day 2 (D2) and day 14, and (c) C-low at D0 versus D2 and D14.
the faecal metabolomes of C-high at D0 and D14 (P = 0.001; F = 1.2), that however, for individual metabolites, was not statistically significant using ANOVA following FDR correction (P > 0.05 for all metabolites) ( Fig. 5b; Supplementary Table S3); nevertheless, a relative reduction in the short chain fatty acids (SCFAs) butyrate, isobutyrate and propionate was observed in samples from this group at D14 versus D0 via Pearson's correlation heatmaps (Fig. 6b). Conversely, in addition to a greater effect size according to CCA (P = 0.001; F = 5.88), clear differences were detected between the faecal metabolites identified in samples from C-low at D0 and D14 ( Fig. 5C; Supplementary Table S4). In particular, glucose, uracil, inosine, trehalose, leucine, butyrate and valine were more abundant in samples collected at D14 when compared to D0 (ANOVA; P < 0.001, P < 0.001, P = 0.001, P = 0.003, P = 0.022, P = 0.041 and P = 0.044, respectively) ( Fig. 6c; Supplementary Table S4). Partial Least Squares (PLS) analysis of metabolite and bacterial OTU data obtained from faecal samples at D0, with 'metabolites' as the dependent matrix and 'bacterial OTUs' as the independent matrix, did not identify any significant association between these two datasets (P = 0.96)

Discussion
In this study, acute helminth infections in a cohort of equine youngstock were associated with significant alterations in gut microbial profiles and diversity, which were reversed following parasiticide treatment. The faecal microbial profiles of the yearlings in our study were in line with those described previously for managed equines 54 www.nature.com/scientificreports www.nature.com/scientificreports/ nevertheless, a greater interindividual variation in taxonomic composition was observed in this group of animals when compared to a cohort of adult broodmares hosted on the same farm 40 . This discrepancy may be suggestive of a 'developing microbiome' in these young animals. Indeed, a recent study in human infants demonstrated the occurrence of three stages of microbial development in the GI tract during early life, which include an early phase characterised by rapid changes in the core GI microbiota over time, an intermediate transitional phase characterised by fewer changes over time, and a final and stable 'adult' microbial profile 4 . In support of our findings, a previous study conducted in equine youngstock demonstrated that, although gut microbiome re-modelling occurs mostly during the first 60 days of life, the gut microbial profiles of horses of 9 months of age is still significantly different from those of adult animals 7 . Of note, environmental and pathogenic stimuli which impact on the order of colonisation of the GI tract during the developmental phase may have significant repercussions on adult microbiome composition and homeostasis 1,28,74 . Hence, given the known roles of the gut microbiome in immune regulation and metabolism, it is likely that the alterations in faecal microbiome composition of equine youngstock observed in association with GI helminth infections may result in long-term implications for animal health and wellbeing.
Indeed, the abundance of several bacterial taxa with key roles in metabolism and immune-regulation differed significantly between C-high and C-low. For instance, bacterial families belonging to the classes Clostridia and Bacteroidetes, were increased and decreased, respectively, in C-high when compared to C-low prior to anthelmintic treatment. Of note, several differences in faecal microbial profiles observed between groups before treatment were reversed following anthelmintic administration; for instance, bacteria belonging to the families Mogibacteriaceae (phylum Firmicutes; class Clostridia), Prevotellaceae (phylum Bacteroidetes; class Bacteroidetes) were similar in abundance in faecal samples from C-low and C-high by D14. A reduction in Bacteroidetes (genera Prevotella and Parabacteroides) was also observed in mice experimentally infected with the murine whipworm, T. muris 36 . Members of the family Prevotellaceae are commensals of mammalian mucosal surfaces, and are known potential pathobionts, particularly associated with oral infections 75 . Prevotella spp. have been demonstrated to promote pro-inflammatory Th17-mediated immune responses 75,76 ; thus, a decreased abundance of this bacterial family in the presence of GI helminth infections could be linked to the known immune-regulatory properties of these parasites 77 . Indeed, a reduction in Th17-inducing segmented filamentous bacteria has been previously reported in association with Nippostrongylus brasiliensis infection in a rodent model 34 , indicating that this may be a common immune-modulatory mechanism in GI helminth infections. Nevertheless, in herbivores, Prevotellaceae also play a key role in the break-down of indigestible fibres 78,79 ; moreover, in a study by Houlden et al. 36 , an observed increase in long chain fatty acids in the faeces of T. muris infected mice was attributed to inefficient breakdown of plant fibres as a result of reduced abundance of Prevotella and Parabacteroides. This dichotomy of function in Prevotella spp. has been well described 80 , and demonstrates that, whilst Prevotella are commensal bacteria with an important metabolic role, they can be pro-inflammatory in association with mucosal pathology. It is also worth reporting that, in contrast to our findings, a positive association between helminth infections and Prevotella spp. abundance was observed in the abomasum of 3 mo small ruminants colonised by the GI helminth H. contortus 46 . Whilst this apparent discrepancy may be linked to fundamental differences between the host:parasite systems under investigation and infection sites (stomach versus large intestine), this data calls for further investigations of the mechanisms governing the interactions between GI helminths and Prevotella spp. and their implications for the developing microbiome of young vertebrates.
Within the phylum Firmicutes, a number of families belonging to class Clostridia were significantly more abundant in the faecal microbiota of C-high when compared to C-low at D0, according to LEfSe (Mogibacteriaceae and Eubacteriaceae) and network (Dehalobacteriaceae and Clostridiacae) analyses. Clostridia have been frequently implicated in studies of helminth-microbiome interactions, with increased abundances being reported in the gut of mice experimentally infected with T. muris 36,81,82 , and in association with mixed   www.nature.com/scientificreports www.nature.com/scientificreports/ GI helminth infections in humans 82 . Clostridia are known to exert immune-regulatory functions, since they produce SCFAs with anti-inflammatory properties, including butyrate 83,84 . Hence, the increased abundances of Mogibacteriaceae and other Clostridiaceae observed in the faecal microbiota of youngstock with heavy parasite burdens could represent a mechanism by which helminths suppress host immune responses, thus reducing pathology and facilitating the establishment of chronic infections 83 . However, Mogibacteriaceae are not butyrate producers 85 ; in addition, metabolomic analysis of faecal samples collected in this study revealed lower levels of butyrate in faeces from C-high when compared to C-low prior to anthelmintic treatment. In accordance with this observation, a recent study investigating the faecal metabolome of human volunteers with chronic infections by Strongyloides stercoralis also detected lower levels of butyrate in samples from parasite-colonised individuals 42 . These data indicate that, whilst increases in SFCAs have been described previously in association with GI helminth infection 83 , this link may not be applicable to all host-parasite systems. Of note, Clostridia belonging to the family Mogibacteriaceae have also been associated with GI inflammation in periodontal disease 86 , and hence their increased abundance in animals with higher infection levels in this study may suggest a role for these bacteria in the onset of mucosal pathology. On the other hand, increased levels of Mogibacteriaceae have also been identified as a biomarker of health in studies comparing subjects with inflammatory bowel disease and irritable bowel disease, to healthy controls [87][88][89] . These apparent contradictions are analogous to those described above for Prevotella spp., and highlight the need for mechanistic studies to unravel the complex function of microbial species within the gut flora colonising different hosts, and in various disease states.
Amongst other bacterial taxa with well-known immune-modulatory functions, those belonging to the order Lactobacillales (phylum Firmicutes, class Bacilli), were also expanded in the faecal microbiota of C-high versus C-low at D0, and bacteria of the family Lactobacillaceae were significantly reduced following anthelmintic treatment in the former group. The existence of a mutualistic association between Lactobacillaceae and parasitic helminths has been hypothesised based on experimental evidence obtained from murine models of infection with the intestinal nematode Heligmosomoides polygyrus 38,90 . Indeed, in a key study by Reynolds et al. 90 , H. polygyrus infections were associated with significantly expanded populations of Lactobacillaceae in the gut of colonised mice; in addition, oral administration of L. taiwanensis prior to helminth exposure was followed by the onset of T regulatory cell mediated immune responses and significantly increased worm burdens 91 . Based on this knowledge, it is tempting to speculate that similar relationships may exist between cyathostomin parasites and equine hosts. Moreover, the increased abundance of bacterial taxa with immune-modulatory roles in animals with heavy burdens of helminth infections may alter the susceptibility of the equine hosts to colonisation by other, 'opportunistic' pathogens. Indeed, potential pathobionts such as Campylobacter jejuni and Pasteurellaceae were expanded in animals with high FEC, and reflected data collected in previous investigations of helminth-infected horses and pigs 92,93 . In humans and rodents, Lactobacillaceae are also generally considered to have positive health benefits for the host 94 ; however, the beneficial effects of Lactobacillaceae to the health and homeostasis of the equine GI microbiota is yet to be demonstrated 54,95 . A recent review of data collected from equine gut microbiome studies has highlighted the roles of other bacterial taxa, such as Lachnospiraceae (phylum Firmicutes, class Clostridia), in supporting intestinal health 54 . In particular, increased abundances of OTUs assigned to the family Lachnospiraceae have been reported in the faecal microbiome of healthy horses when compared with animals with colitis 51,68 ; furthermore, ponies with innate resistance to helminth infection, and consequently lower parasite burdens, have been shown to harbour larger populations of this bacterial family 93 . Accordingly, in our study, the genus Pseudobutyrivibrio (family Lachnospiraceae) was significantly more abundant in the faecal microbiome of C-low versus C-high prior to treatment. Together, this data points towards a detrimental effect of acute cyathostomin infections on the development and maturation of the equine gut microbiome, an effect which is highly likely to have repercussions on susceptibility to a range of infectious and non-infectious diseases 1,27,93,96 . In support of this hypothesis, faecal microbial richness (a measure of microbial alpha diversity and a proxy of gut 'health' 29,66,97-99 ) was significantly lower in samples from C-high when compared to C-low prior to treatment. Acute helminth infections have been frequently associated with decreased alpha diversity of microbial gut populations 36,42,43 and attributed primarily to alterations of the GI microenvironment in response to mounting immune reactions against the invading parasites. Nevertheless, these data contrast with our previously published work in which we reported a trend towards increased alpha diversity in the gut microbiome of chronically helminth-infected adult horses 40 . Differences between the systems investigated (i.e. acutely versus chronically infected horses) may provide a possible explanation, whereby infections of immunologically naïve animals and subsequent pro-inflammatory responses are accompanied by a significant decrease in gut microbial diversity; the latter may be restored (or increased; cf. 39,42,100 ) during chronic helminth infections, due to a synergic effect of host adaptive immune responses and immune-suppressive properties of helminth parasites 101 that contributes to the dampening of local inflammation 102 .
Given the impact of helminth infections on the composition and diversity of the youngstock gut microbiome, we sought to investigate the associations between such changes and the composition of the faecal metabolome. A multivariate analyses of faecal metabolite levels in samples from C-high and C-low prior to and following ivermectin administration revealed only moderate differences between the faecal metabolic profiles of these two groups. A greater abundance of the SCFA isobutyrate, the amino acids leucine, lysine and phenylananine, and the products of carbohydrate breakdown glucose and trehalose were observed in samples from C-high versus C-low prior to ivermectin administration. Of note, increases in selected amino acid abundance have also been described anthelmintic treatment. Differences in relative abundance of selected microbial taxa in faecal samples from C-low (FEC < 10 egg per gram (e.p.g.)) prior to treatment (D0), and 2 (D2) and 14 (D14) days post-treatment. Light grey: effect size of 3.5-4.5; dark grey: effect size >4.5. www.nature.com/scientificreports www.nature.com/scientificreports/ in the faecal metabolome of both mice 36 and humans infected by GI helminths 42 . This consistent observation may indicate a reduction in the absorption of the products of microbial metabolism, for example, as a consequence of ongoing intestinal inflammation caused by helminth colonisation. This hypothesis is supported by the lack of significant correlations between the abundance of specific gut bacteria and metabolites detected by PLS analysis, which might indicate that host-related factors, such as malabsorption, might have been responsible for the observed changes. In the context of livestock management, a reduction in amino acid absorption is likely to have considerable implications for animal performance and production. Thus, the relationships between metabolite production and absorption should be further investigated via metabolomics analyses of faeces and other biofluids (e.g. urine and blood) of equines and other livestock species infected by parasitic helminths 103 . Administration of anthelmintics to C-high did not result in significant alterations of concentrations of faecal metabolites detected   www.nature.com/scientificreports www.nature.com/scientificreports/ before treatment, suggesting that any malabsorptive component of the observed increase in amino acids and glucose, did not resolve over the time-scale of the study. In contrast, concentrations of glucose, uracil, inosine and trehalose in faecal samples from C-low were significantly increased post-treatment. As these animals did not harbour significant parasite burdens prior to anthelmintic treatment, this finding supports a possible role of other environmental factors, such as grass quality and/or anthelmintic administration, on faecal metabolite and gut bacteria abundance over time.

Scientific RepoRtS
concluding remarks. Data from both bacterial 16S rRNA sequences and 1 H-NMR analyses of faecal samples from equine youngstock with high versus low parasite burdens revealed a number of helminth-associated perturbations to GI microbial composition and metabolism, with likely repercussions on medium-to long-term host susceptibility to a range of infections and diseases which should be further investigated. In particular, we advocate for in-depth studies of the impact that early colonisation of young production animals by parasites (e.g. calves, lambs and piglets) exerts on host gut microbial composition and metabolism, given the likely severe economic impact of alterations of the gut homeostasis of these species. In addition, data from our study calls for further explorations of the role/s that parasite-associated changes in gut microbiota may play in the immunopathology of helminth infections in children from endemic areas, and their susceptibility to other parasitic, bacterial and viral infectious agents in areas of poor sanitation.

Methods ethics statement.
This study was approved and carried out in strict accordance and compliance with the guidelines of the Institutional Ethical Review Committee, Department of Veterinary Medicine, University of Cambridge, UK (Ref. No. CR190). Written informed consent was obtained from the stud farm from which study samples were collected.

Study population and diagnostic procedures.
A cohort of TB yearlings was recruited from a stud farm in eastern England, UK. The stud hosts ~50 yearlings each year, which are kept at pasture in groups of 2-8 across 486 hectares. Routine control of parasite infections in the stud relies on administration of targeted anthelmintic treatments (with ivermectin and fenbendazole), based on evaluation of individual parasite burdens (inferred by FEC, that consist in counting the number of parasite e.p.g. of faeces) at 2 monthly intervals. In addition, praziquantel is administered to each yearling three times a year for tapeworm control, whilst a single moxidectin treatment is administered in late November for control of cyathostomins. Samples used in this study were collected in April 2017; all yearlings had received ivermectin (0.2 mg/kg) and fenbendazole (10 mg/kg) in February 2017, and praziquantel (1.5 mg/kg) in March 2017. A total of 53 TB yearlings, between 12-16 months of age at the time of www.nature.com/scientificreports www.nature.com/scientificreports/ sampling, were screened for infection by cyathostomins. Briefly, duplicate faecal samples were collected from all horses on D0; aliquots of each sample were subjected to (i) FEC analysis using a centrifugal floatation technique sensitive to one e.p.g. 104 , and (ii) screening for infections with the common equine cestode Anoplocephala perfoliata using a double sugar flotation technique 105 , and (iii) larval culture and microscopic examination to screen for infections with large strongyle nematodes, e.g. Strongylus vulgaris. Horses were recruited in the study if they satisfied the following criteria: (i) FEC of ≥100 e.p.g. (=C-high) or ≤10 e.p.g. (C-low) in duplicate faecal samples at D0; (ii) negative for co-infections with other GI helminths (i.e. Strongylus spp., Parascaris equorum, Strongyloides westeri, A. perfoliata) and no history of or clinical signs of infection by Oxyuris equi; (iii) no antibiotic treatment for at least 2 months prior to sampling. Of the 53 horses screened, 23 matched these criteria, of which 9 were enrolled into the C-high group, and 14 into C-low (Supplementary Table S1). A power calculation was performed based on a bacterial 16S rRNA amplicon sequence dataset from a previous study conducted using samples from equines on the same farm 40 . A Wilcoxon-Mann-Whitney test for comparing OTU-specific abundances between two samples, as described by Mattiello et al. 106 (https://fedematt.shinyapps.io/shinyMB), demonstrated that the current study with 23 samples had sufficient power (0.81) to detect an effect size of 2, in at least 20 OTUs between C-high and C-low.
Anthelmintic treatment and sampling. Individual, freshly voided, faecal samples were collected from the centre of the faecal mass from all C-high and C-low animals at D0. Immediately following sample collection, an anthelmintic treatment (Eqvalan: ivermectin 0.2 mg/kg) was administered to each animal. Sampling was repeated as above at D2 and D14. A 100 g aliquot of each faecal sample was snap frozen, transported to the laboratory and stored at −80 °C within 2 h of collection, prior to genomic DNA extraction, high-throughput sequencing of the bacterial 16S rRNA gene and metabolite extraction; the remainder was kept fresh and subjected to FEC analysis as described above. 40,42 bacterial 16S rRNA high-throughput sequencing protocols and bioinformatics analyses of sequence data were adapted for this study. Briefly, genomic DNA was extracted from each faecal sample, as well as from five negative 'blank' (=no DNA) controls, using the PowerSoil ® DNA Isolation Kit (Qiagen, Carlsbad, CA, USA), according to the manufacturers' instructions. High-throughput sequencing of the V3-V4 hypervariable region of the bacterial 16S rRNA gene was performed by Eurofins Genomics on an Illumina MiSeq platform according to the standard protocols with minor adjustments. Briefly, the V3-V4 region was PCR-amplified using universal primers, that contained the adapter overhang nucleotide sequences for forward (TACGGGAGGCAGCAG) and reverse primers (CCAGGGTATCTAATCC). Amplicons were purified using AMPure XP beads (Beckman Coulter) and set up for the index PCR with Nextera XT index primers (Illumina). The indexed samples were purified using AMPure XP beads (Beckman Coulter) and quantified using the Fragment Analyzer Standard Sensitivity NGS Fragment Analysis Kit (Advanced Analytical) and equal quantities from each sample were pooled. The resulting pooled library was quantified using the Agilent DNA 7500 Kit (Agilent), and sequenced using the v3 chemistry (2 × 300 bp paired-end reads, Illumina).

DNA extractions and bacterial 16S rRNA gene Illumina sequencing. Previously published
Bioinformatics and statistical analyses of 16S rRNA sequencing data. Raw paired-end Illumina reads were trimmed for 16S rRNA gene primer sequences using Cutadapt (https://cutadapt.readthedocs.org/en/ stable/) and sequence data were processed using the Quantitative Insights Into Microbial Ecology 2 (QIIME2-2018.4; https://qiime2.org) software suite 107 . Successfully joined sequences were quality filtered (Read cut-off: 17; 286 and 17; 255 for forward and reverse, respectively), dereplicated, chimeras identified, and paired-end reads merged in QIIME2 using DADA2 108 . A phylogenetic tree was generated for diversity analysis, followed by calculation of alpha and beta diversity metrics using the 'core-metrics-phylogenetic command' in QIIME2. Sequences were assigned to taxonomy using the feature classifier: Greengenes 13_8 99% OTUs full-length sequences. A feature table with the assigned taxonomy was exported from QIIME2 alongside a weighted UniFrac distance matrix for downstream biostatistical analysis. Statistical analyses were executed using the Calypso software 109 (cgenome.net/calypso/); total sum scaling (TSS) normalisation was applied, followed by square route transformation to account for the non-normal distribution of taxonomic counts data. Microbial community profiles were ordinated using PCoA (Bray-Curtis distances) and supervised CCA including infection status and sample collection time point as explanatory variables. Differences in bacterial alpha diversity (richness, evenness and Shannon Index) between study groups (C-high and C-low at D0; as well as D0, D2 and D14 for each C-high and C-low) were evaluated based on rarefied data (read depth of 25,868) using ANOVA. Differences in beta diversity (weighted UniFrac distances) were measured using PERMDISP 110 . Differences in the abundances of individual microbial taxa between groups were assessed using the LEfSe workflow 111 , accounting for the paired nature of samples pre-and post-anthelmintic treatment. Furthermore, networks of correlation were constructed using the Calypso software 109 to identify clusters of co-occurring bacteria based on their association with helminth infection status. Taxa and explanatory variables were represented as nodes, taxa abundance as node size, and edges represented positive associations, while nodes were coloured according to infection status. Taxa abundances were associated with infection status using Pearson's correlation and nodes were then coloured based on the strength of this association. Networks were generated by first computing associations between taxa using Pearsons's rho and the resulting pairwise correlations were converted into dissimilarities and then used to ordinate nodes in a two-dimensional plot by PCoA. Therefore, correlating nodes were located in close proximity and anti-correlating nodes were placed at distant locations in the network. www.nature.com/scientificreports www.nature.com/scientificreports/ Metabolite extraction. Metabolites were extracted from 200 mg aliquots of each faecal sample using a methanol-chloroform-water (2:2:1) procedure, as described previously 42 . In particular, 600 μl of methanolchloroform mix (2:1 v:v) were added, samples were homogenised using stainless steel beads and sonicated for 15 min at room temperature. 200 μl each of chloroform and water were added, samples were centrifuged and the separated aqueous and lipid phases were collected. The procedure was repeated twice, and the aqueous fraction from each extraction were pooled. The aqueous fraction was dried in a vacuum concentrator (Concentrator Plus, Eppendorf).

H-NMR analysis of aqueous extracts.
Protocols of 1 H-NMR metabolite analysis have been described previously 42 . Briefly, the dried aqueous fractions were re-dissolved in 600 μl D 2 O, containing 0.2 mM sodium-3-(tri-methylsilyl)-2,2,3,3-tetradeuteriopropionate (TSP) (Cambridge Isotope Laboratories, MA, USA) as an internal standard and phosphate buffer (40 mM NaH 2 PO 4 /160 mM Na 2 HPO 4 ). The samples were analysed using an AVANCE II + NMR spectrometer operating at 500.13 MHz for the 1 H frequency and 125.721 MHz for the 13 C frequency (Bruker, Germany) using a 5 mm TXI probe. The instrument is equipped with TopSpin 3.2. Spectra were collected using a solvent suppression pulse sequence based on a one-dimensional nuclear Overhauser effect spectroscopy (NOESY) pulse sequence to saturate the residual 1 H water signal (relaxation delay = 2 s, t1 increment = 3 us, mixing time = 150 ms, solvent pre-saturation applied during the relaxation time and the mixing time). One hundred and twenty-eight transients were collected into 16 K data points over a spectral width of 12 ppm at 27 °C. In addition, representative samples of each data set were also examined by two-dimensional Correlation Spectroscopy (COSY), using a standard pulse sequence (cosygpprqf) and 0.5 s water presaturation during relaxation delay, 8 kHz spectral width, 2048 data points, 32 scans per increment, 512 increments. Peaks were assigned using the COSY spectra in conjunction with reference to previous literature and databases and the Chenomx spectral database contained in Chenomx NMR Suite 7.7 (Chenomx, Alberta, Canada). 1D-NMR spectra were processed using TopSpin. Free induction decays were Fourier transformed following multiplication by a line broadening of 1 Hz, and referenced to TSP at 0.0 ppm. Spectra were phased and baseline corrected manually. The integrals of the different metabolites were obtained using Chenomx. Metabolites were normalised to faecal dry matter, total area and differential abundance of metabolites between samples from C-high and C-low, at D0 and D14. Faecal metabolite abundances from each sample were ordinated by PCoA according to infection status (C-high and C-low) and sample collection time point (D0 and D14). Predicted associations among metabolites identified in the faecal metabolome of each sample group were also identified by Pearson's correlation heatmaps in Calypso 109 (cgenome.net/calypso/). In particular, heatmaps were constructed to identify associations between metabolite abundance and infection status (i.e. C-high and C-low) and time point pre-and post-treatment (i.e. D0 and D14). Differences in individual metabolite abundance between groups were evaluated for statistical significance using ANOVA with FDR correction for multiple comparisons. In order to identify linear correlations between metabolites and bacterial OTUs identified in faecal samples, PLS analysis was performed on data from all samples at D0 (pre-treatment) in Simca-P v15, with metabolites as the dependent and OTU as the independent matrix, respectively.