All samples for this study were provided by the Department of Experimental Animal Facilities and Biorisk Management of the Friedrich-Loeffler-Institut on the Isle of Riems within the H1N1pdm09 animal experiment under reference number 7221.3–1-035/17 (Schwaiger et al., 2019). For this study, 16 influenza virus A/Bayern/74/2009-infected and three mock-infected (serving as experimental control) German landrace pigs, which were 7 weeks of age at the beginning of the study, were analyzed. The sampling scheme is described in Table 1. Swine fecal samples were collected within 30 s of defecation for a period of 31 days to account for longitudinal shifts. Digesta from intestinal sections (IS) (ileum, proximal and distal colon from three healthy and three infected individuals) were sampled during necropsy on days 7 (n = 1), 25 (n = 2) and 30 (n = 3). All samples were immediately frozen on dry ice and subsequently stored at – 80°C.

For 16S rRNA gene sequencing and metaproteomics approach, fecal samples and intestinal sections (IS) were homogenized by the Covaris® CP02 CryoPrep™ instrument (Covaris Ltd., Brighton, United Kingdom) and processed as previously described in Gierse et al. (2020).

Nucleic acids were extracted from appr. 100 mg fecal powder using a bead-beating phenol chloroform extraction protocol (Berry et al., 2012) with subsequent precipitation of nucleic acids with 3 M Na-acetate and isopropanol as described in Gierse et al. (2020). Afterwards, the DNA pellets were washed with 70% v/v ethanol before being dried and re-suspended in DEPC-treated MilliQ water for downstream applications. The DNA content was quantified with Qubit™ dsDNA Broad Range Assay Kit (Invitrogen™). Next, the 16S rRNA amplification and Illumina MiSeq sequencing were performed using the V4 primer pair 515F (5`-GTG-YCA-GCM-GCC-GCG-GTA-A-3`) /806R (5`-GGA-CTA-CNV-GGG-TWT-CTA-AT-3`), as described in Gierse et al. (2020). The Illumina MiSeq resulted in an average output per sample from around 100,000 reads per sample and the sequences were submitted to European Nucleotide Archive (ENA) under the project name “Influence of an Influenza-A-Virus infection on the respiratory and gastrointestinal tract microbiome and archaeal community composition’ (project number PRJEB42450, accession number ERP126308).

Analysis was performed in R (R Core Team, 2020) using the dada2 pipeline package version 1.11.1 (R-version 3.6.1) for sequence annotation. Advanced bioinformatics processing (e.g., alpha-and beta diversity, Bray–Curtis dissimilarity and nonmetric multidimensional scaling) was done in R with packages “vegan,” “ggplot,” “phyloseq,”” plyr” and“reshape2” as described in Gierse et al. (2020). For statistical analysis, Kruskal-Wallis rank sum tests were performed (value of p ≤0.05) and the p values were corrected using the Benjamini and Hochberg false discovery rate (fdr) approach in R.

Metaproteomic analysis was performed as previously described in Gierse et al. (2020). Proteins were extracted from appr. 100 mg fecal powder using the TRIzol-based extraction protocol, and protein concentrations were measured via Pierce BCA Protein Assay (Smith et al., 1985; Kessler and Fanestil, 1986; Wiechelman et al., 1988; Brown et al., 1989). Subsequently, 30 μg protein were loaded on a 4–22% Criterion TGX precast gel (BioRAD, Hercules, CA, United States) and stained with Colloidal Coomassie Brilliant Blue G-250 (Neuhoff et al., 1988). Each lane was cut in 10 fractions, and each fraction was processed into smaller pieces, de-stained and purified according to the ZipTip manufacturer’s protocol (C18, Merck Millipore, Billerica, MA, United States). Peptide-containing solution was eluted in glass vials and vacuum centrifugation was performed until dry. Finally, the peptides were re-suspended in 10 μL of 0.1% (v/v) formic acid for mass spectrometric analysis as previously described (Gierse et al., 2020). Protein identification was performed using Mascot Daemon version 2.6.2 (Matrix Science Ltd., London, United Kingdom) against a custom archaea-specific database, which was built using the results of the 16S rRNA gene sequencing data from the corresponding fecal samples. The database contained all UniprotKB entries (access date: 09.06.2020) for every identified archaeal member of the gastrointestinal microbiome. Because Clostridium sp. CAG_221, Lactobacillus reuteri and Prevotella copri were identified as the most abundant species in the porcine GIT microbiome, these species were included and used as a reference for bacterial-originated proteins. Furthermore, all entries for Sus scrofa and influenza A virus were included in the database. Compared to a database including all members of the porcine gastrointestinal microbiome, the amount of archaeal originated protein groups was increased from ~0.4% to ~4% of all identified protein groups through use of the reduced database.

For validation of Mascot data, Scaffold 4.8.7 and X!Tandem (version X! Tandem Alanine, 2017.2.1.4) were used. Metaproteome annotation pipeline Prophane¹ was employed for taxonomic and functional protein analysis (Schiebenhoefer et al., 2020) with the same parameters as previously described (Gierse et al., 2020). The mass spectrometry proteomics data were deposited to the ProteomeXchange Consortium via the PRIDE (Perez-Riverol et al., 2019) partner repository with the data set identifiers PXD044365.

The analyzed samples were provided from the same infection trial (reference number 7221.3–1-035/17; Schwaiger et al., 2019) in which swine were infected with influenza virus A H1N1pdm09. Fecal and rectal swab samples from the swine were collected on days 0 (day of first infection), 4, 21 (day of second infection) and finally day 30 (control group) or 31 (infection group).

The preparation of the samples for sequencing was performed as published (Wylezich et al., 2018) with a few modifications. Prior to RNA extraction, the fecal samples were diluted threefold (2 g of feces and 4 mL water) and homogenized using the IKA Ultra Turrax Tube Drive System (IKA, Staufen, Germany) in an ST-20 Tube with a stirring device of 5 steel balls (5 mm) to ensure representative subsamples. Subsequently, the coarse fiber components were separated from the homogenates with single-use nylon sieves (mesh size 1 mm; Carl Roth, Karlsruhe, Germany). The rectal swab samples (Rayon and Polyester Dryswabs™, Medical Wire and Equipment, England) were incubated via shaking in 1 mL distilled water at 4°C and 1,400 rpm for 4 min (ThermoMixer®, Eppendorf, Wesseling-Berzdorf, Germany). The supernatant was then used for RNA extraction (see below).

To extract the nucleic acids, 200 μL of the feces homogenates or 250 μL of the swab supernatants were disintegrated using cryoPREP (Covaris Inc., Woburn, United States). The pulverized material from feces homogenates or swab supernatants was suspended in 800 μL or 1 mL AL lysis buffer (Qiagen, Hilden, Germany), respectively, and preheated to 56°C. Subsequently, RNA was extracted from the lysate using TRIzol Reagent (Thermo Fisher Scientific Inc., Waltham, United States) in combination with RNeasy Mini Kit (Qiagen) including on-column DNase digestion (Qiagen) according to the manufacturer’s instructions. The RNA was reverse-transcribed into cDNA with the cDNA-Synthesis System (Roche, Mannheim, Germany) and hexanucleotide primers (Roche). After fragmentation with a M220 Focused-ultrasonicator (Covaris), Ion-Torrent compatible DNA libraries were prepared using IonXpress barcode adapters (Life Technologies, Darmstadt, Germany) and a GeneRead DNA Library L Core Kit (Qiagen) according to the manufacturer’s instructions. Thereafter, size exclusion, quality control with a Bioanalyzer 2,100 and a High Sensitivity-Chip (Agilent Technologies, Santa Clara, CA, United States), and library quantification using the KAPA Library Quantification Kit (Roche) were performed. Deep sequencing of pooled sequence libraries was completed with Ion 530 Chips in 400 Bp mode on an Ion Torrent S5 XL instrument (Thermo Fisher Scientific Inc.) according to the manufacturer’s instructions.

Resulting RNA sequences were submitted to European Nucleotide Archive (ENA) under the project name “Influence of an Influenza-A-Virus infection on the respiratory and gastrointestinal tract microbiome and archaeal community composition’ (project number PRJEB42450, accession number ERP126308).

RNA sequences were quality-checked with FastQC and subsampled into rRNA and non-rRNA fractions using SortMeRNA (Kopylova et al., 2012). The resulting non-rRNA from all samples were pooled and aligned against the NC_nr database (accessed 12/03/2020) using DIAMOND (Buchfink et al., 2015). Taxonomic and functional analysis was performed in MEGAN6 Ultimate using the lowest common ancestor (LCA) algorithm (min score 50; top percent 4; min support 1; Huson et al., 2016). Taxonomic binning of the mRNA reads was done to the phylum Archaea, family Methanobacteriaceae and the genera Methanobrevibacter and Methanosphaera, respectively. Within these taxonomic bins, mRNA sequences assigned to the KEGG category “methane metabolism” were considered as methanogenesis transcripts of the respective taxa (Kanehisa and Goto, 2000).

The microbial richness of the total prokaryotic community between the feces of naïve and infected animals demonstrates no significant difference, while the microbial diversity within the same experiment showed significant alterations in the high abundant families Prevotellaceae, Clostridiaceae and Lachnospiraceae (Gierse et al., 2021). In contrast, the comparison between feces and intestinal sections (summarized within naïve and infected individuals) indicated significant differences (p < 0.001). The number of identified taxa (represented by the amplicon sequence variant ASV) was highest in the colonal sections (avg. colon proximal 947 ASVs and colon distal 1,069 ASVs) and lowest in ileum (avg. 577 ASVs) with significant differences to intermediate fecal data (avg. 868 ASVs) (Figure 1A).

The percentage of archaeal ASVs was highest in fecal samples (avg. 0.72%), with negligible lower amounts in the three different intestinal samples (colon distal avg. 0.66%, colon proximal avg. 0.65% and ileum avg. 0.62%; Figure 1B). Although the colonel sections had higher rates of prokaryote colonization, the relative number of archaea was rather similar between the intestinal samples.

A deeper look into the gastrointestinal compartments in Figure 1C revealed Firmicutes and Bacteroidetes as predominant phyla. This is consistent with the previously published study on the healthy cohort of an animal trial (Gierse et al., 2020) as well as other studies (Holman et al., 2017; Le Sciellour et al., 2019). Additionally, Spirochaetes and Euryarchaeota were found in the intestinal tract. The small intestine, represented by the ileum, was colonized by members of the genera Clostridium, Terrisporobacter and Romboutsia, whereas the compartments of the large intestine and feces are primarily dominated by Lactobacillus, Ruminococcus and Prevotella spec. These intestinal section-dependent shifts were also observed within the archaeal community composition. While the relative abundances of the three dominant archaeal genera in the small intestine (Methanobrevibacter, Methanosphaera and Methanomethylophilaceae) were below 0.5%, between 4 and 7% of these members were found in the colonal sections and feces. The overall increasing relative abundance in methanogenic archaea in the biological course of the intestinal tract was characterized by a significant increase of Methanobrevibacter (p < 0.05) and Methanosphaera (p < 0.05), especially between the ileum and feces. In contrast, Methanomethylophilaceae showed stable but very low relative abundances ranging from 0.075% up to 0.1% along the natural course of the intestinal tract.

Within the archaeal community, the relative abundance of Methanobrevibacter increases significantly (p = 0.028) from 44.2% in the ileum to 59% in feces (Figure 1D). A possible consequence of this shift is that methane formation via hydrogenotrophic CO₂ reduction increases relatively along the natural course of the intestines. In a recent study, Mi et al. (2019) sequenced mcrA gene amplicons and identified similar abundances of colonic digesta of Chinese pigs, with a Methanobrevibacter abundance 57%. In our study, methylotrophic Methanosphaera were present in 40.2% of the ileum samples, 46.8% of colonic sections (44% proximal and 49.7% distal) and 39.3% of fecal samples. Although the differences between the intestinal sections were insignificant at the 0.05 threshold, one could assume higher rates for methylotrophic methanogenesis in the large intestines compared to fecal samples. Mi et al. (2019) described a lower relative abundance for Methanosphaera of 14% and a higher relative abundance of Methanomassiliicoccales of 15%.

We further investigated this longitudinal dynamic in highly abundant genera in the dissimilarity analysis at the end of section 3.2.

We have analyzed the temporal dynamics of the archaeal microbiome of pigs over the course of 1 month, which was a substantial part of the animal’s life span both in typical animal experiments (Kim et al., 2011; Le Sciellour et al., 2019; Mi et al., 2019; Wang et al., 2019), and livestock swine (Gutierrez et al., 2015). There were clear treatment-dependent differences between the archaeal community composition of healthy and influenza A virus-infected individuals (Figure 2).

The majority of ASVs in the porcine intestinal tract belong to the species Methanobrevibacter millerae, followed by Methanosphaera cuniculi and Methanobrevibacter boviskoreanii, together representing 80–90% of the archaeal community. In contrast to pigs, predominant phylotypes of the human intestines were Methanobrevibacter smithii and Methanosphaera stadtmanae (Miller et al., 1982; Dridi et al., 2009). As in our porcine model organism, Methanosphaera cuniculi was recently reported in human intestines (Chibani et al., 2022).

Initial structural and functional indications of a respiratory tract infection with influenza A virus on the gastrointestinal microbiome via metaproteome analysis were reported in a related study of Gierse et al. (2021). Microbial richness, bacterial diversity and central enzymes were significantly affected by the infection and indicate intestinal disturbances, most likely caused by the IAV infection. In our dataset from the same animal trial (Schwaiger et al., 2019; Gierse et al., 2021) the number of Methanobrevibacter boviskoreanii (ASV4) reads was significantly reduced by first Influenza A virus infection (d0, p < 0.01), whereas the naïve animals showed rather stable read counts. On the other hand, the unique ASV7, Methanosphaera stadtmanae, only occurred in the infected animals on day 2 post-infection and was absent in the naïve animals, which is possibly indicative of a treatment dependent effect (p < 0.001). Additional significant differences (p < 0.01) between the naïve and the H1N1 infected animals were observed for the predominant Methanosphaera cuniculi (ASV2), Methanobrevibacter millerae (ASV5) and within the family Methanomethylophilaceae (ASV6).

A phylogenetic analysis of methanogens in swine feces by Mao et al. (2011) indicated that the majority of present phylotypes belonged to Methanobrevibacter spp., followed by Methanosphaera stadtmanae and, with a lower percentage in sequence similarity, Aciduliprofundum boonei. The study of Guindo et al. (2021) identified Methanobrevibacter smithii as the most prevalent and Methanobrevibacter millerae as a minor dominant in the pig’s digestive tract via real-time PCR and PCR sequencing. Contrary to this study, we observed a minor prevalence for Methanobrevibacter smithii (ASV11) but identified Methanobrevibacter millerae (ASV1, ASV3, ASV5, and ASV12) as the most dominant phylotype in our dataset.

Similarly, Federici et al. (2015) described a strong significant shift from Methanobrevibacter smithii to Methanobrevibacter boviskoreanii due to weaning, which could be an explanation for the high abundance of Methanobrevibacter boviskoreanii and slightly decreasing occurrence of Methanobrevibacter smithii, particularly within the healthy pigs of our study.

The absence of Methanosphaera stadtmanae (ASV7), an H2-and methanol-dependent methanogen (Saengkerdsub and Ricke, 2014), in the healthy cohort (see Figure 2) could lead to the assumption of a decreased level of the methylotrophic methanogenesis pathway compared to the infected cohort. However, the predominant ASV2 Methanosphaera cuniculi was significantly higher in the healthy cohort, which could have a balancing effect on methane formation between these two groups.