The experiments performed followed the National Animal Protection Guidelines and were approved by the German Animal Ethics Committee for the protection of animals (G0176/16). Female specific pathogen-free (SPF) and germfree C57BL/6 mice were kept in individually ventilated, filter-topped cages with autoclaved bedding, chow and water. Infections with 200 H. polygyrus larvae were performed aseptically in a laminar flow. H. polygyrus L3 were freshly isolated from fecal cultures of infected mice and treated for 1 week with an antibiotic cocktail (5 mg/ml streptomycin, 1 mg/ml ampicillin, 0.5 mg/ml gentamicin, 1 mg/ml neomycin, 0.5 mg/ml vancomycin; all from AppliChem, Darmstadt, Germany). L3 were shown to be free of aerobic microbes as determined by lack of bacterial growth in antibiotic-free LB medium. Infected and naïve control GF mice received antibiotics (as specified above) via the drinking water. To further reduce the risk of contamination, SPF and GF C57BL/6 mice were kept without bedding change until the dissection 2 weeks post-infection. The axenic status of GF mice was confirmed by qPCR of eubacterial 16s rRNA with colon content collected on the day of infection and dissection. Adult worms were removed from the small intestine, counted and eight females per mouse were kept at 37°C in RPMI-1640 medium containing 200 U/mL penicillin, 200 μg/mL streptomycin (all from PAN Biotech, Aidenbach, Germany) and 1% glucose for 24 h for the determination of individual egg counts. Female worm length was determined after culture.

Excretory/secretory products of H. polygyrus (HES) were collected from adult worms extensively washed before being cultured in phenol-red free RPMI-1640 medium containing 200 U/mL penicillin, 200 μg/mL streptomycin. After 24 h in culture, worms were washed extensively with antibiotic-free worm growth media (RPMI-1640 medium with 1% glucose) and maintained in this medium with daily media changes. Spent media from the first 48 h were discarded. Thereafter, supernatants were harvested every 48 h and sterile filtered through a 0.22 μm syringe-driven filter system, and stored at −20°C until further use.

The strains used to evaluate antibacterial activities of HES in the radial diffusion assay included Escherichia coli IMT19224, Salmonella enterica serovar Typhimurium ATCC14028, and Staphylococcus aureus IMT29828 obtained from the strain collection of the Institute of Microbiology and Epizoonotics, Freie Universität Berlin and Enterococcus faecium DSM20477 provided by Dr. Markus Heimesaat (Institute of Microbiology, Charité—Universitätsmedizin Berlin). E. coli IMT19224 was used to assess agglutinating activity of HES.

Antibacterial activities of HES were assessed using the radial diffusion assay (19). Overnight bacterial cultures were diluted 1:100 in Mueller-Hinton broth (Carl Roth, Karlsruhe, Germany) and incubated at 37°C with shaking at 250 rpm until reaching an optical density of 0.3–0.4 at 600 nm. Bacteria were washed and resuspended in cold sodium phosphate buffer (100 mM, pH 7.4) by centrifugation (880 × g, 10 min, 4°C). Bacteria were then resuspended in warm (50°C), sterile underlay agar [10 mM sodium phosphate buffer, 1% (v/v) Mueller-Hinton broth, 1.5 (w/v) agar] at 4 × 10⁵ colony forming units per mL. Fifteen milliliter of bacteria-infused underlay agar was poured into 120 mm square petri dishes and allowed to solidify. Evenly spaced wells (5 mm) were formed in the agar using the blunt ends of P10 pipet tips, and treatments and controls added (5 μL/well). Five microliter native HES corresponded to 5 μg protein. The antimicrobial peptide Pexiganan (kindly provided by Jens Rolff, Institute of Biology, Freie Universität Berlin, 0.0125 μg/well) was applied as positive control. PBS and RPMI-1640 medium were included as negative controls. Plates were incubated at 37°C for 3 h and then overlaid with double-strength Mueller-Hinton agar [4.2% (w/v) Mueller-Hinton broth, 1.5% agar]. Petri dishes were incubated for 18 h at 37°C and the growth inhibition zones around each well were measured. Antibacterial activity is represented as the diameter of the inhibition zone (mm) beyond the 5 mm well.

Agglutinating activity of HES was assessed as described previously (20) using E. coli IMT19224. Bacteria were collected at mid-logarithmic phase by centrifugation at 880 × g for 5 min, then washed and resuspended in Tris-buffered saline (50 mM Tris-HCl, 150 mM NaCl, pH 7.5) at approximately 10⁹ cells/mL. Thirty microliter of bacteria were mixed with 30 μL of treatments in the presence and absence of 10 mM CaCl₂ and incubated for 1 h at room temperature on a glass slide. Concanavalin A from Canavalia ensiformis (Con A) and Lectin from Triticum vulgaris (Wheat germ agglutinin; WGA, both from Sigma-Aldrich) were included as positive controls. Samples were then visualized and photographed using the 40X objective on a Leica DM750 microscope equipped with an ICC50HD digital camera (Leica Microsystems, Wetzlar, Germany).

Small intestines and the bulk of removed parasites were kept in ice-cold physiological NaCl solution. Thirty worms (15 males/15 females) were quickly isolated from three individual SPF and GF mice, washed repeatedly in cold physiological NaCl solution, inspected for physical integrity, and absence of host tissue and then snap frozen in liquid nitrogen before storage at −80°C. Samples were homogenized using shredder columns filled with 200 mg sterile sea sand and the FastPrep®-24 instrument (MP Biomedicals, Eschwege, Germany) at 5 m/s for 35 s. Supernatants of homogenized worms were further processed for RNA isolation (InnuPREP RNA isolation, Analytik Jena AG, Germany), DNase treatment (Analytik Jena AG, Germany), and RNA quality control (Agilent 2100 Bioanalyzer, RNA 6000 Nano Kit, Agilent Technologies, Waldbronn, Germany). All RNA samples displayed RIN values of 10.

For transcriptome sequencing on an Illumina platform a TruSeq RNA library generation was utilized. The library was generated by using the TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA, USA) following the manufacturer's instructions. The library was quantified by using the KAPA Library Quantification Kit for Illumina (Kapa Biosystems, Wilmington, MA, USA). The library size was determined by using the High Sensitivity DNA Analysis Kit for the 2100 Bioanalyzer Instrument (Agilent Technologies, Waldbronn, Germany). Libraries were adjusted to a concentration of 12 pM and sequenced on a HiSeq 1500 instrument (Illumina, San Diego, CA, USA) in rapid mode. For cluster generation, the TruSeq Rapid PE Cluster Kit v2 was used. Cluster generation was performed on board. For sequencing the HiSeq Rapid SBS kit v2 was used to sequence 100 + 100 bases.

We sequenced three isolates from SPF and GF mice with a mean library size of 40.15 million paired-end reads and a standard deviation of 10.74. Raw reads were subjected to quality control and trimming via the QCumber pipeline (version 1.0.14, https://gitlab.com/RKIBioinformaticsPipelines/QCumber) utilizing FastQC (v0.11.5, https://www.bioinformatics.babraham.ac.uk/projects/fastqc/), Trimmomatic (0.36) (21) and Kraken (0.10.5-beta) (22). On average, 91.77% of reads remained after trimming.

Preprocessed reads were mapped to a reference genome (as specified below) and corresponding sequence features using the TopHat split-read mapper (v2.1.1) (23) and reference as well as novel features were extracted and merged with the aid of Cufflinks and Cuffmerg (24) (v2.2.1) to obtain one integrated and unified transcriptome for H. polygyrus samples. The H. polygyrus draft genome nHp_v2.0 was applied as reference genome (database version WBPS10, annotation version 2016-09-WormBase), as available at WormBase ParaSite (25). For each sample, raw expression values were created by counting uniquely mapped reads on gene level using featureCounts (v1.5.0-p3) (26). To identify differentially expressed genes (DEGs) between SPF and GF mice isolates, respectively, DESeq2 (1.12.4) (27) was applied with a classic pairwise design model and a p-value threshold of 0.05. In addition, normalized and transformed expression values were extracted from DESeq2 (regularized log transformation) and corrected for batch effects via Limma (3.28.21, removeBatchEffect) (28) to allow for sample comparison with clustered heatmaps and principal component analysis (PCA).

Reference as well as novel transcripts were functionally (re-)annotated using an iterative annotation strategy. First, transcripts were either first-frame translated (reference) or examined for ORFs (novels, Cuffcompare class code “u”) using EMBOSS transeq (6.6.0.0) (29) and TransDecoder (v2.1), respectively. Next, resulting protein sequences were passed through a series of database searches until successfully annotated with Gene Ontology (GO) terms (30), either via blastp (2.6.0+) (31), and Blast2GO (4.0.7) (32) or by a final InterProScan (33). Databases used for annotation included (in this order) the UniProt (34) Heligmosomoides polygyrus bakeri proteome (UP000050761, downloaded at 07.04.2017), UniProt Swiss-Prot Nematoda proteins, UniProt TrEMBL Nematoda proteins as well as the complete Swiss-Prot database and the complete TrEMBL database (all downloaded at 16.02.2017).

Lymph node single cell suspensions and small intestinal tissue digestion for the isolation of siLP cells were performed as described previously (35). Cultures were kept for 6 h with brefeldin A added after 1 h before surface and intracellular staining. Surface and intracellular markers were stained according to the manufacturer's instructions with the following antibodies obtained from ThermoFisher/eBioscience, if not stated otherwise: CD4-PerCP/-BV510/-A700 (RM4-5), Foxp3-FITC/-PerCP-Cy5.5 (FJK-16s), GATA-3-A660/-PE/-PE-eF610 (TWAJ), T-bet-PE/-PE-Cy7 (eBio4B10), RORγt-BV421 (Q31-378, BD biosciences), IL-10-APC (JES5-16E3), IL-4-PE/-PE-Cy7 (11B11), and IL-17A-PerCP-Cy5.5 (eBio17B7). Live/dead discrimination was performed using fixable viability dye eF780 (ThermoFisher/eBioscience). Unspecific binding was prevented by addition of 20 μg/ml FcgRII/III blocking antibody (2.4G2).

Formalin-fixed, paraffin-embedded sections (1–2 μm) of duodenum were de-waxed and stained with hematoxylin and eosin for overview, with periodic acid Schiff for goblet cell quantification and by Direct red 80 (Sigma) for the detection of eosinophils. Enteritis was scored using hematoxylin and eosin-stained section as described before (16). PAS⁺ goblet cells were counted along five villi per section. Images were acquired using the AxioImager Z1 microscope (Carl Zeiss MicroImaging, Inc., Göttingen, Germany). All evaluations were performed blinded.

Data were assessed for normality using GraphPad Prism software (La Jolla, CA, USA). For comparison between two groups, an unpaired T-test was used. Testing of multiple groups was performed using a one-way analysis of variances followed by Tukey's multiple comparison or the Kruskall-Wallis test combined with Dunn's multiple comparison test.

Infection with H. polygyrus alters the composition of the intestinal microbiota alongside the intestine, including an increase in gram⁻ Enterobacteriaceae (2, 17, 36). Similar changes occurred in IL-4Rα^−/− mice, hence independently of Th2-mediated changes in gut physiology (2). As both free-living and parasitic nematodes defend themselves against potentially harmful microbes by the production of antimicrobial factors (7), we asked if H. polygyrus releases active antimicrobials, possibly interfering with its microbial environment. We used the radial diffusion assay to test the antibacterial activity of H. polygyrus excretory/secretory products (HES) in comparison to the antimicrobial peptide Pexiganan. Five micrograms of native HES collected from H. polygyrus cultures inhibited the growth of gram⁻ and gram⁺ bacteria, including E. coli, S. enterica var. Typhimurium, E. faecium, and S. aureus (Table 1).

C-type lectin domain-containing proteins are known to agglutinate bacteria and are important in nematode immune defense against microbial infection (37). As H. polygyrus produces a C-type lectin protein (38) we tested the agglutinating activity of nematode products by treating E. coli with increasing amounts of native HES in the presence and absence of CaCl₂. We observed dose- and calcium-dependent agglutinating activity (Figure 1), suggestive of C-type lectin-mediated bacterial agglutination. These data indicate that H. polygyrus employs defense mechanisms via released products during its interactions with microbes which may contribute to shaping its microbial environment in the murine gut.

Having demonstrated the ability of nematode products to influence bacterial growth, we sought to investigate if intestinal nematodes sense their microbial environment and hence asked if the complete absence of microbes in the host gut resulted in altered parasite gene expression. To that end, we infected germfree (GF) and conventional (specific pathogen-free; SPF) mice and performed RNA-sequencing with parasites isolated 2 weeks post-infection. Samples clearly clustered according to SPF vs. GF parasite origin (Figures 2A,B). We found that a surprisingly small set of 52 genes was differentially expressed in adult worms isolated from GF compared to SPF mice (Supplementary Table 1). The majority of genes were upregulated, comprising a venom-like allergen (VAL-1), chitinase-1, lysozyme-3, and orthologs of putative Caenorhabditis elegans glutathione S-transferase and C. elegans/C. briggsae UDP-glucuronosyl- transferases, amongst others (Supplementary Table 1). Only four of ten genes downregulated in parasites isolated from GF mice were annotated, including a putative C. elegans UDP-glucuronosyl-transferase. Hence, parasitic nematodes reared in a germ-free environment display a distinct gene expression pattern.

Previous studies reported on impeded survival and fecundity of intestinal nematodes in the absence of gut microbes (39–41); therefore, we assessed if parasite burden and fitness were altered depending on the host microbial status. While adult worm burdens were similar in SPF and GF mice at 2 weeks post-infection (Figure 3A), female worms developing in GF mice were significantly smaller and produced fewer eggs (Figures 3B,C). Importantly, H. polygyrus resides in the proximal small intestine harboring few microbes and the parasite mainly relies on host tissue as food source (42). Thus, we investigated next if the reduced parasite fitness in GF mice coincided with immune changes.

The microbiota supports the induction and maintenance of regulatory T cells (Treg) (43–46) and infections with H. polygyrus lead to the activation and expansion of regulatory T cells suppressing local immunopathology, but also host protective Th2 immunity (1, 15–17). Therefore, we surveyed if Treg expansion, phenotype and cytokine production in H. polygyrus infected mice differed depending on the microbial status.

The overall frequencies of Foxp3⁺ Treg were similar in mLN of naive SPF and GF mice and did not change significantly upon infection (Figure 4A). While Treg frequencies in the small intestinal lamina propria (siLP) were stably maintained in infected SPF mice, Treg frequencies dropped significantly in the small intestines of infected GF mice compared to the respective naive controls (Figure 4B).

Intestinal Foxp3⁺ Treg form a functional heterogeneous population comprising subsets marked by the elevated expression of GATA-3 or RORγt, respectively (47). While GATA-3 expression is necessary for Treg stability under inflammatory conditions (48, 49), RORγt⁺ Treg exhibit a highly activated phenotype and limit the Th2-driven control of helminth infection and immune pathology in intestinal inflammation (43, 50). Hence, we investigated if the reduced fitness of worms isolated from GF mice was associated with phenotypic alterations in the Treg population. Fewer Treg in mLN and siLP of naïve and infected GF mice expressed RORγt compared to the respective SPF controls (Figures 4C,D). Steady state GATA-3 expression by Treg and the expansion of GATA-3⁺ Treg upon infection was similar in mLN of SPF and GF mice (Figure 4C). Upon infection, the increase in GATA-3⁺Treg reached significance in the small intestine of SPF mice (Figure 4D). Thus, naïve and infected GF mice harbored significantly less RORγt⁺ Treg compared to SPF mice, while GATA-3⁺Treg expanded similarly.

Next, we asked if Treg activation differed depending on the microbial status and hence assessed their cytokine production. IL-10 production by Treg in mLN increased similarly and strongly in SPF and GF mice upon infection (Figure 4E). IL-10 production by siLP Treg of SPF mice did not change in response to infection (Figure 4F). Small intestinal Treg of GF mice were rather poor IL-10 producers at steady state and upon infection (Figure 4F). As intestinal RORγt⁺ Treg have been reported as superior in IL-10 production compared to other gut Treg (43), we next surveyed IL-10⁺ Treg of SPF and GF mice for co-expression of RORγt and GATA-3. Expectedly, the reduced frequencies of RORγt⁺ cells in the Foxp3⁺Treg pool (Figures 4C,D) was reflected by their underrepresentation in the IL-10 producing Treg population of naïve and infected GF mice (Figures 4G,H). GATA-3⁺Treg expanding in mLN of infected SPF and GF mice (Figure 4C) clearly dominated the IL-10-expressing Treg pool in both groups upon infection (Figure 4G). Reflecting their high frequencies in the total siLP Treg population (Figure 4D), GATA-3⁺Treg dominated in the small intestinal IL-10⁺ population irrespective of microbial and nematode-infection status (Figure 4H). Finally, we investigated if the reduction in RORγt⁺ Treg in the intestine and the poor IL-10 expression by gut Treg in GF mice was associated with differences in local immunopathology. Duodenal enteritis scores were, however, similar in SPF and GF nematode-infected mice (Figure 4I).

Taken together our data show that Treg activation in gut-associated lymphoid tissue seen as increased IL-10 production occurs independently of the presence of gut microbes. Naïve and nematode-infected GF mice display a reduction in RORγt⁺ Treg in the gut and gut-draining lymph nodes, while GATA-3⁺ Treg expanded similarly in SPF and GF mice and formed the major IL-10 producing Treg subset upon infection irrespective of the microbial status.

To see if the reduction in RORγt⁺ Treg and the lower IL-10 expression by small intestinal Treg in GF mice coincided with deregulated Th2 responses we quantified Th2 cells based on GATA-3 expression and IL-4 expression. Significantly more GATA-3⁺ Th2 cells were present in the small intestines of infected GF mice compared to SPF mice, while IL-4 production was significantly increased in mLN (Figures 5A,B). Calculating the ratios of Th2 cells to Treg based on their frequencies in CD4⁺ T cells, we found significantly elevated Th2 effector to Treg ratios in the small intestine of infected GF compared to SPF mice (Figure 5C). We have previously shown that intestinal nematode infections lead to the differentiation of GATA-3⁺ Th2 and GATA-3⁺T-bet⁺ Th2/1 hybrid cells (51, 52). Th2/1 cells developed in infected SPF as well as GF mic, hence microbial signals were dispensable for their induction (Figures 5D,E). The increase in intestinal GATA-3⁺Th2 cells coincided with trends of increased goblet cell and eosinophil counts in the duodenum (Figures 5F,G). As the microbiota supports Th17 differentiation (53, 54) we assessed RORγt and IL-17A expression by Foxp3⁻CD4⁺ T effector cells. Expectedly, GF mice harbored very few Th17 cells in mLN and small intestine (Figure S1). In conclusion, nematode-induced local Th2 responses were significantly increased in the absence of gut microbes and decreased parasite fitness in GF mice was associated with elevated Th2 to Treg ratios at the site of infection.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.