Intestinal mucus was isolated from the small intestine (duodenum, jejunum, and ileum) of healthy chickens, turkeys, pigs, sheep, or cows euthanized at the National Animal Disease Center, Ames Iowa, in accordance with a protocol approved by the Institutional Animal Care and Use Committee. Animal sources of mucus included: 10 turkeys (2 years old), 20 chickens (14 weeks old), 5 sheep (8 months old), five pigs (ranged from 2 to 4 years old), and one cow (around 2 years old). The mucosal surfaces were gently rinsed with sterile PBS to remove ingesta, and subsequently scraped with sterile microscope slides for collection of mucus. Mucus from respective animal species was pooled and lyophilized. Each mucus prep was purified to enrich the mucus glycans, as previously described (Martens et al., 2008). Briefly, lyopholized crude mucosal scrapings were suspended at 25 g/L in 100 mM Tris pH 7.4. Proteinase K was added (100 mg/L) to each sample and incubated 16–20 h at 65°C. Samples were then centrifuged at 21,000 × g for 30 min at 4°C and the supernatant was saved. NaOH was added to 0.1 M and NaBH4 to 1M and the solution was incubated 16–20 h at 55°C. HCl (12M) was added until pH reached 7.0. Mucus preps were centrifuged at 21,000 × g for 30 min at 4°C and supernatants were saved and filtered stepwise to clarify (5 μm → 3 μm → 1 μm → 0.65 μm → 0.45 μm → 0.22 μm). Mucus preps were dialyzed at 1 kDa cutoff against ddH2O to remove salts. Tris was added to the glycans at 50 mM (pH 7.4) and passed over DEAE-Cellulose (CAS 9013-34-7, Santa Cruz Biotechnology) columns equilibrated in 50 mM Tris pH 7.4 and the flow-through was collected. The resulting fractions were dialyzed at 1 kDa cutoff against ddH2O and lyophilized for downstream use.

Glycan analysis of each mucus preparation was performed at the University of California, San Diego Glycotechnology Core facility. Analysis of the monosaccharide and sialic acid components of the glycans was done using high performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD) and Reverse phase HPLC Fluorescence, respectively. Profiling of isolated glycans was done by MALDI-TOF and analyzed using the GlycoWorkbench Software Tool (Damerell et al., 2012).

Bradford protein assays were performed on mucus from each host to quantitate the total protein in each sample (Bradford, 1976). A total of 10 μg of each mucus preparation was run on SDS gel and stained with Coomassie blue and then SYPRO Ruby. Mass spectroscopy was performed, as described previously (Reinhardt et al., 2012), on each host mucus preparation (10% wt/vol) to characterize the proteins present in the mucus glycan preparation.

Iron quantification was performed for each host mucus sample using the EnzyX Iron (II) colorimetric assay (Assay Biotechnology Sunnyvale, CA). Samples were run at 1, 2, and 10% wt/vol (w/v), following the manufacturer's recommendations.

C. jejuni NCTC 11168 was obtained from the American Type Culture Collection (ATCC 700819) and grown at 42°C under microaerophilic conditions (85% N₂, 5% O₂, and 10% CO₂) in MCLMAN (medium cysteine leucine methionine aspartic acid niacinamide) broth medium (Alazzam et al., 2011) for maintenance. Growth curves were generated (O.D.₆₀₀) by culturing C. jejuni NCTC11168 using the Bioscreen growth curve system (Growth Curves USA, Piscataway, NJ), with measurements taken each hour for 60 h. Cultures (10 replicates of each medium) were grown in 300 μl MCLMAN medium supplemented with 0.5% (w/v) purified mucus from chickens, turkeys, pigs, sheep, or cows, as well as MCLMAN without mucus.

Human embryonic INT-407 intestinal epithelial cells were used to test adherence and invasion by C. jejuni (Negretti and Konkel, 2017), with the following modifications. In some experiments, INT-407 cells were infected with C. jejuni and turkey or pig mucus was added only during the time of infection, and in others, mucus was present while culturing C. jejuni, as well as during the infection. The day before the assay, 1.5 × 10⁵ INT-407 cells were cultured at 37°C in 5% CO₂ environment in wells of a 24 well plate using DMEM with 10% FBS containing penicillin and streptomycin (1X) and 10 mM HEPES buffer. For each treatment, INT-407 cells were plated in quadruplicate. The next day, cells were washed 3X in HBSS buffer containing calcium and magnesium prior to adding the inoculum. C. jejuni was grown on Mueller-Hinton plates at 42°C in a microaerophilic gas environment. At least 5 colonies were inoculated into the broth phase of biphasic liquid and agar (2% w/v) MCLMAN media. In some cases, the liquid phase of the biphasic media contained 0.5% (w/v) turkey or pig mucus. Inoculum was statically cultured for 48 h at 42°C in a microaerophilic gas environment. Motility of the inoculum was assessed by dark field microscopy. The inocula were prepared by adjusting the OD₅₄₀ to 0.03 (3 × 10⁷ cfu/mL) in HBSS containing 1% BSA (fraction V) with or without the addition of 0.5% (w/v) turkey or pig mucus. Adherence and invasion was also tested without the addition of mucus to the inocula. A MOI of approximately 100:1 was used by adding 500 μL of the inocula to each well. Infection was synchronized by centrifuging the plate at 800 × g for 5 min at room temperature. Cells were incubated for 3 h at 37°C with 5% CO₂ cells were then washed 3X in 1mL of HBSS and lysed with the addition of 200 μL of Triton X-100 for 5 min at 37°C. Cell associated C. jejuni (adherent and invaded) were enumerated by serial dilution and culture on Mueller-Hinton plates. After 3 h incubation, invasion was measured by removing the media from cells plated and inoculated in the same manner as described above, washing cells 3X with HBSS, and adding 1 mL of DMEM containing 1% FBS and 250 μg/mL gentamicin back to each well. Cells were incubated for 3 h and then washed 3X in HBSS. Cells were lysed as described above, bacteria were enumerated, and reported as the number of invaded bacteria.

C. jejuni was grown, while shaking under microaerophillic conditions, in 5.0 ml of MCLMAN liquid medium supplemented with 0.5% (w/v) of each host mucus and MCLMAN without mucus (four C. jejuni culture replicates of each). After a period of 24 h, the cultures were removed from the incubator and RNAprotect (QIAGEN, Germantown, MD) was added to each culture followed by RNA extraction using the RNeasy extraction kit (QIAGEN), following the manufacturer's instructions. DNA was removed using Turbo DNase (Ambion, Austin, TX). Total RNA quality was assessed with the 2100 Bioanalyzer, using RNA Nano chips (Agilent Tech., Santa Clara, CA). Ribosomal RNA was removed using the RiboZero system according to the manufacturer's instructions (Illumina, San Diego, CA), to enrich for total mRNA. Libraries were constructed using directional TruSeq RNA library kit and were sequenced on a HiSeq 2500 sequencer using the 100-cycle, high output mode (Illumina Inc., San Diego, CA).

Messenger RNA used for RNA-Seq analysis from C. jejuni grown on chicken mucus was sequenced using PacBio's long read technology to evaluate if antisense RNA resulted from pervasive transcription extending from adjacent upregulated mRNA. Total RNA was isolated as described above. Transcripts were polyadenylated followed by removal of rRNA using the RiboZero kit (Illumina, San Diego, CA, United States). After rRNA depletion, cDNA was synthesized and amplified using SMARTer cDNA system (Clontech Laboratories, Mountain View, CA, United States), and low and high molecular weight fractions were selected using the BluePippin instrument (Sage Science, Beverly, MA, United States). Low and high molecular weight libraries were sequenced using PACBIO RS II sequencer, implementing the Iso-Seq pipeline (Pacific Biosciences, Menlo Park, CA, United States).

Sequence data was imported into CLC Genomic workbench V 8.0, and mapped to the C. jejuni NCTC 11168 reference sequence NC_002163 (Gundogdu et al., 2007). Reads were iteratively mapped to the genome in the following order: Mapped to the gene regions in the sense orientation, then the unmapped reads were mapped in the gene regions in the antisense orientation, and finally any remaining reads were mapped to intergenic regions. Expression values were calculated using only reads that mapped uniquely to a single location on the reference genome. Empirical analysis of differential gene expression was performed using the EdgeR statistical test, implemented in CLC Genomic workbench (Robinson et al., 2010). The data were normalized by scaling to the mean for graphical representations and presentation. A custom Pearl script was written to identify and bin all examples of differentially expressed genes [fold change >4.0, False Discovery Rate corrected (FDR) p < 0.01] in the sense orientation that were adjacent in the genome to differentially expressed genes in the antisense orientation. The criteria for these antisense-sense associations required the differentially expressed gene to be either the same gene as the antisense gene, or adjacent to it in the genome. If an association was identified, the next gene in the genome was evaluated. This continued until our fold change and FDR p-value threshold was not met.

To improve our understanding of the relationship between asRNA, iron acquisition, and oxidative stress genes, data from a study evaluating PerR (global regulator for oxidative stress response) or Fur (global regulator for iron acquisition) single or double knockouts on C. jejuni 11168 gene expression under iron limited or replete conditions was analyzed (Butcher et al., 2015). Briefly, the cDNA libraries of the transcripts had been generated in a strand specific manner by the authors, but the antisense data was not evaluated for proximity or linkages to the sense transcripts. After downloading the reads from NCBI (accession number SRP044881), the directional reads were mapped and analyzed as described above. Antisense RNA and gene expression within our dataset were identified as Fur or PerR repressed based on comparisons with the results from the published gene knockout study.

Analysis was performed using CLC Genomic workbench V 8.0 and PRIMER7 software v7.0.13 (Gorley and Clarke, 2008). PAST statistical analysis software (Hammer et al., 2001) was used for Permutational multivariate analysis of variance (PERMANOVA) tests. Differences with a False Discovery Rate (FDR) corrected P-value < 0.01 with at least a 4-fold change in gene expression was considered to be significant. Data are deposited in NCBI's Short Read Archive (SRA) associated with BioProject PRJNA411826.

C. jejuni grown in MCLMAN broth supplemented with intestinal mucus from various species (0.5% w/v) showed a significant enhancement in growth and terminal O.D. (P < 0.001) compared to defined media alone, suggesting that mucus components were used as a growth substrate (Figure 1). The terminal O.D.s varied slightly among mucus sources, but no significant differences were detected between avian or mammalian mucus. Cultures supplemented with mucus reached terminal stationary phase after 28 h of growth, while it took 38 h on MCLMAN without mucus.

Addition of pig mucus [0.5% (w/v)] during the incubation stage of INT-407 cells with C. jejuni significantly enhanced (P < 0.001) the number of cell-associated and cell-invaded C. jejuni, compared to C. jejuni exposure with turkey mucus or without mucus (Figures 2A,B). In subsequent experiments, C. jejuni were cultured for 48 h in a defined media biphasic culture containing 0.5% (w/v) pig or turkey mucus, or no mucus in the liquid phase. Mucus was also present during the incubation stage of the assay. Culturing C. jejuni in either pig or turkey mucus significantly enhanced cell association and invasion, as compared to a control lacking mucus (Figures 2C,D). The number of invaded C. jejuni cultured in the presence of pig mucus was significantly different (P = 0.0013) compared to C. jejuni cultured in turkey mucus (Figure 2D).

The crude mucus processing liberated mucus glycans from the protein backbone and enriched for O-linked neutral glycans (Martens et al., 2008). The total amount of protein present in each of the mucus preparations was low as determined by Bradford assays after purification [percentage/dry weight: chicken (0.12%), cow (0.15%), pig (0.08%), sheep (0.13%), and turkey (0.09%)], and compositional and structural analysis identified glycan sialic acid and differences between the avian and mammalian mucus (Table S1 and Figure S1). In each of the mucus samples, Fucose, N-Acetyl-galactosamine, N-Acetyl-glucosamine, Galactose, Glucose, and Mannose were present as glycan components. The sialic acids N-Glycolylneuraminic acid and N-Acetylneuraminic acid were both present in the glycans from cow, pig, and sheep mucus but N-Acetylneuraminic acid was the only sialic acid present in the chicken and turkey mucus (Table S1). Iron was not detectable in any of the mucus extractions by colorimetric assay (data not shown). The protein composition of each preparation was determined by MS/MS analysis (Table S2 and Figure S2), however no mucus proteins were identified, suggesting that glycans make up a large portion of the dry weight of the material isolated from each mucus (Figure S1).

In total, 417,448,571 sequence reads were obtained after quality filtration. 311,343,359 reads (74.6%) were mapped to protein coding regions in the sense orientation, 14,659,287 reads (3.5%) mapped to intergenic regions and 843,531 reads (0.2%) mapped to protein coding regions in the antisense orientation (Table 1).

Gene expression profiles from C. jejuni grown on defined media or defined media supplemented with intestinal mucus from different hosts significantly differed [Table S3, P < 0.001, Permutational multivariate analysis of variance test (PERMANOVA)]. A principal component analysis plot (PCA) of all transcriptomes showed differences in total transcriptomes between C. jejuni grown in the presence of mucus compared to without mucus (Figure 3A). The increased growth rate of all the mucus-containing cultures likely contributed to many of the expression differences observed. The mucus-containing cultures were in the late-log growth phase at the 24-h sampling, while the defined media were still in early-log phase. Because of these growth-rate differences, discussion of genes differentially expressed between mucus-containing and defined media will be limited to genes known to be impacted by the presence of mucus. A set of genes, part of the fucose utilization genetic island present in several C. jejuni strains including NCTC11168, were up regulated (9–19 fold) for all the mucus cultures. This gene cluster (Cj0481-Cj0489) includes fucose utilization proteins and a transporter.

Mucus-containing C. jejuni cultures were each in the late-log growth phase at the 24-h sampling for transcriptomic analysis (Figure 1). PCA plots of the C. jejuni transcriptomes didn't show clear separation among mucus from different hosts (Figure 3A). Despite overlap in expression profiles, there were 73 genes differentially expressed between C. jejuni grown in avian (chicken and turkey) and mammalian (pig, cow, and sheep) mucus (Table 2, Figure 3B). Most of the genes differentially expressed had increased expression in the avian mucus (71 genes). The 2 genes that were upregulated in the mammalian mucus were rrc and fdxa (both associated with oxidative stress response) and both were ~14-fold higher when compared to the avian mucus. Most of the C. jejuni genes upregulated in the avian mucus are involved in iron acquisition, oxidative response, or were membrane proteins. The genes that were most increased in the avian mucus, including Cj1383c, chuC, Cj0177, exbD1, kata, are involved in iron acquisition and oxidative stress response (all were over 100-fold higher in the avian mucus compared to mammalian mucus). In all, there were 35 genes associated with iron acquisition, 10 oxidative stress genes, 13 putative membrane proteins, and 13 genes of various functions (such as the energy taxis response gene cetB), upregulated in C. jejuni due to avian mucus, compared to mammalian mucus (Table 2).

All comparisons among C. jejuni grown on different host mucus identified genes differentially expressed between each animal mucus, with the exception of cow mucus compared to pig mucus, which had no significant differences (Table S4). Other than the defined media alone, chicken mucus was the strongest driver of C. jejuni gene expression differences, yielding 157, 152, 115, and 113 genes differentially expressed from cow, pig, sheep, and turkey mucus, respectively. The turkey mucus also diverged from the mammalian mucus resulting in 46, 49, and 47 differentially expressed genes from cow, pig, and sheep mucus, respectively. While there were no detected differences between cow and pig mucus, they had 23 and 21 genes differentially expressed with sheep mucus, respectively. Many of the genes identified with these comparisons reflect the avian or mammalian origin of the mucus. For example, iron acquisition and oxidative stress genes were again increased in C. jejuni grown in the chicken or turkey mucus, compared to pig, cow or sheep mucus. Some of these iron-associated genes were increased ~1,000 or ~2,000 fold (see chicken comparisons with cow or pig mucus, Table S4) in the avian mucus. The same 2 genes (rrc and fdxa) that were upregulated in the mammalian mucus when compared to avian, were also increased in the individual mammalian host comparisons with each avian host. Surprisingly, many of the same iron acquisition genes that were increased in the avian mucus, were also increased in the sheep mucus (i.e., chuABD, tonB2, exbD2, exbB2, cfbpAB, p19, ceuB, etc), when sheep was compared to the other mammalian mucus (pig and cow), highlighting differences among animal species.

Antisense sequences were separated from the C. jejuni transcriptome for parallel analysis. These transcripts were identified in the directional RNA-seq dataset by mapping the reads to genes in the antisense orientation. In all, 641 asRNAs, that mapped to the noncoding strand of genes, were identified in the dataset (had at least 10 reads). These asRNAs expression patterns segregated by source of mucus (host) in the C. jejuni growth media (Figure 4A). The C. jejuni grown on avian mucus (chicken or turkey) had asRNA expression profiles (PERMANOVA test) significantly different from the mammalian mucus' (cow, pig, or sheep); and both were different from that of the defined media, (p < 0.01), but there were no differences within groups (avian or mammalian). Many asRNAs were differentially present among C. jejuni grown in each host mucus as well as between defined media compared to all mucus types (Figure 4B). In many cases the asRNAs displayed atypical patterns, mapping to only portions of genes and intergenic regions.

Two C. jejuni genes (prfA and flip) had many antisense reads (relative to other asRNA) that mapped to the noncoding stand in the avian mucus cultures, and were also significantly increased in the avian mucus, compared to the mammalian mucus. Upon closer examination, it was observed that prfA and flip, while not differentially expressed in the sense orientation, were adjacent to differentially expressed genes in the sense orientation, within the same samples (Cj1613c next to prfA and Cj0819 next to fliP). Global analysis of all of the sense and antisense data showed that many asRNAs that were upregulated in the avian transcriptomes, were adjacent coding mRNAs that were upregulated in those same samples (Table 3). This phenomenon was not only observed in the avian samples, but was often associated with upregulated genes.

In addition to the positive associations among antisense and sense RNAs described above, negative associations among asRNA and their adjacent sense mRNA was observed with mucus comparisons with the defined media. The eight-gene fucose utilization operon (Cj0481-Cj0489), upregulated in all mucus cultures, had antisense reads mapping to the adjacent transcriptional regulator for the operon, Cj0480c, in the defined media cultures. Antisense RNA that mapped to Cj0480c was significantly increased while the rest of the operon was downregulated in the defined media. The close association between C. jejuni asRNAs and differentially expressed genes may suggest that these non-coding transcrips are associated to regulatory activities.

Many of the asRNAs upregulated in the avian mucus in this study were adjacent to genes associated with iron acquisition or oxidative stress response. To evaluate the role of the transcriptional regulators for iron acquisition (Fur) and oxidative stress response (PerR), public data were retrieved and re-analyzed from a report by Butcher et al. (2015) to evaluate the association between the asRNA and genes part of the Fur or PerR regulatory networks. In their study, Butcher et al. characterized C. jejuni NCTC11168 wild type, ΔFur, ΔPerR, and ΔFurΔPerR mutants under both iron-replete and iron-limited conditions to identify the Fur and PerR regulons. In doing so, they identified the Fur or PerR regulated genes under iron limited and iron replete conditions. The researchers used strand-specific RNA-seq, allowing the sense or antisense nature of the RNAs to be preserved in the data. We identified many of the same asRNAs, observed in our dataset, in our re-analysis of their data, and confirmed the associations of the asRNA in our study are part of the Fur or PerR regulons for iron acquisition or oxidative stress protection, respectively. The differentially expressed genes and asRNAs shared with our study are noted with an asterisk on Table 3.

We suspected that the detected asRNAs might be caused by delayed termination of mRNA, resulting in transcriptional read-through because the differentially expressed antisense reads correspond to genes arranged in the opposite direction. Read-through of RNA polymerase would create transcripts with an antisense portion if transcription continues into adjacent genes arranged in the opposite orientation. To test this, one of the replicates of C. jejuni grown in chicken mucus was subjected to full length cDNA sequencing on a PacBio instrument to evaluate the nature of the observed antisense RNAs. While the sequencing depth was low, we obtained sequences spanning regions where asRNA were adjacent to sense mRNAs. The most abundant asRNAs in the avian RNA-seq data mapped to the opposite stand of prfA, adjacent to the upregulated gene Cj1613c in the genome. Long reads spanning gene boundaries were observed in this region, suggesting that some antisense RNAs were being generated by failure of transcriptional termination to occur at gene boundaries, continuing into the non-coding stand of the open reading frame of the adjacent gene (Figure S3).

This transcriptional read-through did not always occur and did not appear to be proportional to transcriptional levels. For example, porA was the most highly expressed gene across all transcriptomes (1,151,741 reads average per sample, compared to 65,688 average reads mapping to Cj1613c in the avian samples). Immediately downstream of porA is dnaJ, in the opposite orientation as porA (similar to orientation of Cj1613c to prfA). Despite this, dnaJ only has on average only 5 antisense reads mapping to it in our dataset, compared to 6,264 antisense reads mapping to prfA in the avian mucus C. jejuni transcriptomes.

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.