Patterns of host gene expression associated with harboring a foregut microbial community
© The Author(s). 2017
Received: 24 July 2017
Accepted: 31 August 2017
Published: 6 September 2017
Harboring foregut microbial communities is considered a key innovation that allows herbivorous mammals to colonize new ecological niches. However, the functions of these chambers have only been well studied at the molecular level in ruminants. Here, we investigate gene expression in the foregut chamber of herbivorous rodents and ask whether these gene expression patterns are consistent with results in ruminants. We compared gene expression in foregut tissues of two rodent species: Stephen’s woodrat (Neotoma stephensi), which harbors a dense foregut microbial community, and the lab rat (Rattus norvegicus), which lacks such a community.
We found that woodrats have higher abundances of transcripts associated with smooth muscle processes, specifically a higher expression of the smoothelin-like 1 gene, which may assist in contractile properties of this tissue to retain food material in the foregut chamber. The expression of genes associated with keratinization and cornification exhibited a complex pattern of differences between the two species, suggesting distinct molecular mechanisms. Lab rats exhibited higher abundances of transcripts associated with immune function, likely to inhibit microbial growth in the foregut of this species.
Some of our results were consistent with previous findings in ruminants (high expression of facilitative glucose transporters, lower expression of B4galnt2), suggestive of possible convergent evolution, while other results were unclear, and perhaps represent novel host-microbe interactions in rodents. Overall, our results suggest that harboring a foregut microbiota is associated with changes to the functions and host-microbe interactions of the foregut tissues.
Complex and intimate associations with fermentative gut microbes have been instrumental in the success of mammalian herbivores [1, 2]. In general, these associations take place in enlarged gut compartments that house both the symbiotic microbes and the food material being digested. The most notable of these, the rumen, can compose 10–25% of an animal’s body weight, and the microbes housed within can provide 50–80% of the daily energetic needs of the host . In addition to ruminants, a number of other animals exhibit foregut fermentation, such as some macropod marsupials, nonhuman primates, and rodents . Understanding the evolution and function of foregut fermentation chambers is of considerable interest to biologists, because of its key role in facilitating herbivory.
The phenotypic and functional variation observed between species is a consequence of differences in gene expression as well as gene content . Changes in gene expression profiles associated with hosting foregut microbial communities have been studied in a number of ways. First, developmental studies have compared gene expression in rumen epithelial tissue from suckling animals (before the establishment of fermentation) to adults [4–6]. Additionally, others have compared gene expression in tissues from various gut chambers, such as the fermentative rumen versus the acid-producing abomasum . However, the question of which changes in gene expression are associated with hosting a foregut microbial community has not been evaluated in a comparative sense. Furthermore, this question has only been investigated in ruminants, and thus studies into the foregut chambers of other mammals may reveal potential convergence in gene expression profiles.
Here, we compare gene expression profiles in the proximal stomach regions of two rodent species that differ in diet, stomach morphology, and density of associated microbes. Woodrats (Neotoma spp.) are small herbivorous rodents that inhabit North America, and exhibit a bilocular, or semi-segmented stomach morphology . We have demonstrated that the foregut harbors a dense microbial community (~1010 live cells/g contents) and has high concentrations of short chain fatty acids (~200 mM), which is a signature of microbial fermentation . Conversely, laboratory rats are omnivorous, have a single-chambered stomach with fewer microbial cells (103−104;  and much lower concentrations of short chain fatty acids (<5 mM; . The rat foregut region is thought to be largely for food storage, as surgical removal of this section does not impact food digestion . The foreguts of both species are nonglandular and lined by keratinized, stratified squamous epithelia . The current study focuses on a comparative analysis of foregut gene expression in order to gain insight into host-microbe interactions in the woodrat foregut.
Comparing the gene expression profiles from the foregut regions of woodrats and lab rats should reveal genes associated with hosting a foregut microbial community that participates in fermentation of food. We generated several hypotheses based on previous research in ruminant herbivores. First, given the evolution of bilocular morphology of the woodrat foregut, and the importance of muscular function in the mixing and passage of food material in ruminants [13, 14], we predicted altered expression of genes involved in smooth muscle function in the control of gastrointestinal motility. Second, the rumen of ruminants and foreguts of rats are both lined by keratinized, stratified squamous epithelium. In the rumen, tissues exhibit high levels of expression of genes in the mammalian epidermal development complex ; thus we predicted altered expression of genes associated with epidermal development and differentiation in woodrats compared to the lab rat foregut. Third, post-weaning development of the fermentative rumen coincides with increased expression of genes associated with fatty acid oxidation and ketogenesis [4–6]. Given the high concentration of short chain fatty acids in the woodrat foregut, we also predicted that this tissue would have higher expression of genes associated with fatty acid oxidation and ketogenesis. Fourth, post-weaning development of the rumen also produces significant changes in the expression of several solute carrier (Slc) genes, such as increases in the expression of transporters for glucose and urea [4, 6]. Therefore, we hypothesized that woodrats and lab rats would exhibit differential expression of Slc genes. Finally, we used untargeted techniques to investigate the data for differential abundances in other biological functions. Together, these approaches highlight convergent and divergent gene expression profiles associated with hosting a foregut microbial community.
Animals and tissue collection
Stephen’s woodrats (Neotoma stephensi) were collected near Wupatki National Monument, AZ, USA (35°30′ N, 111°27′ W) in October 2013. Woodrats were fed juniper foliage (Juniperus monosperma) during transport to the laboratory, for one evening in the laboratory, and were dissected the next day. Sprague Dawley lab rats were purchased from Harlan Laboratories (Madison, WI, USA), and individually housed in the open top cages in a room shared with woodrats (Neotoma spp., in separate open top cages). Lab rats were maintained in these conditions for ~6 months, being fed standard rat chow (Harlan Teklad formula 2018). While lab rats and woodrats were fed different diets in this experiment, the diets represent their natural foods. All animals used in our experiment were females (3 for each species). Lab rats used in this experiment were virgins; we do not know whether this was the case for wild-caught woodrats. All animals were euthanized with isoflurane. We collected a small section of the fornix ventricularis of the stomach (or the fundus of the stomach), which was immediately rinsed in ice-cold physiological saline buffer and placed in RNAlater solution. After 24 h, the RNAlater solution was removed and tissues were frozen at −80 °C.
Library generation and RNA-Seq
Total RNA from each animal was isolated from frozen tissue samples using Qiagen RNeasy kit according to manufacturer instructions (Qiagen, Valencia, CA). An amount of 1.0 μg of total RNA from each animal was used to construct strand-specific, paired-end sequencing libraries from polyA-containing mRNA molecules using the Illumina TruSeq Stranded mRNA sample Preparation Kit (Illumina, Inc., San Diego, CA). Individual libraries were multiplexed and sequenced on a single lane of an Illumina HiSeq 2000 platform generating ~84 million total paired-end reads of 101 bp in length. Read quality and duplication levels were evaluated with the FastQC tool (https://www.bioinformatics.babraham.ac.uk/projects/fastqc). The average Phred value of trimmed Fastq sequence was 38, and the level of de-duplicated reads was in the expected range (34.4% to 42.3%) for a moderate level of rRNA and repeat reads from a total RNA library. Paired-end reads were quality filtered and trimmed using Trimmomatic  and the subsequent overlapping paired-end reads were merged using FLASH , which aligned ~90% of the read pairs into single, directional reads; the average insert size of the libraries was 139 bp, (SD = 3 bp.). These composite reads were merged with the forward reads from the remaining unmerged read pairs, resulting in ~46 million merged N. stephensi reads and ~37 million R. norvegicus reads. All reads have been uploaded to the NCBI Gene Expression Omnibus under accession GSE84381.
A map and count pipeline using mouse CDS transcripts as the reference genome
Since de novo transcriptome assemblies are fragmented and incomplete, we chose first to align the composite reads from both N. stephensi and R.norvegicus to the protein-coding portion of the comprehensive gene annotation of Mus musculus from the GENCODE mouse C57BL6/J annotation project . The coding transcripts from the M7 release of the GENCODE mouse genome annotation (gencode.vM7.pc_transcripts.fa) were formatted as a nucleotide database and Illumina reads were aligned using a fast version of the Smith-Waterman algorithm implemented in cross_match  with a -minscore parameter of 50 and a -masklevel parameter of 101 to allow reads to align to multiple transcripts. Total read counts were generated from read alignments to the correct strand per CDS region using custom Perl scripts that collapsed transcript level alignments (ENSMUST identifiers) to gene level alignments (ENSMUSG identifiers). Reads were assigned to a ENSMUSG gene identifier by their maximum cross_match alignment score, and reads that aligned to multiple genes with equivalent scores were evaluated for representation of paralogous genes, pseudogenes, overlapping genes and duplicate gene annotations. We excluded 519 ENSMUSG identifiers from the transcript database (Additional file 1: Table S1) based on cataloging multiple read alignments, and then re-evaluated the multiple read alignments to arrive at a final set of uniquely mapped read counts.
Multi-mapping reads for several key paralogous families encoding highly abundant transcripts relevant to epidermal differentiation were considered in a secondary analysis. In mammals, genes involved in the synthesis of the insoluble cornified envelope (CE) epithelial layer are found in a large gene cluster known as the epidermal differentiation complex or EDC;  that includes the Late Cornified Envelope Group I genes (Lce1 gene symbols), the Late Cornified Envelope Group 3 genes (Lce3 symbols), and the Cornifins, also known as small proline-rich region proteins (Sprr1, Sprr2, Sprr3 and Sprr4 families). Keratins (all Krt# symbols) also occur in several large clusters of paralogous genes. Since cross-species mapping within these genes families is challenging due to gene duplications and homogenization within the EDC, we collapsed read counts to the family level for these genes. Differential gene expression analysis was performed on read counts using quasi-likelihood (QL F-test) methods in edgeR . The tests for differential expression (DE) between experimental groups used the glmQLFTest function, and DE genes were identified at a false discovery rate (FDR) of 1%, using the Benjamini-Hochberg method for multiple testing correction.
De novo transcriptome assembly
As no reference genome sequence exists for Neotoma stephensi, we used a novel de novo algorithm, BinPacker , to assemble a foregut reference transcriptome. The sequence input was 45,992,608 composite reads from the 3 N. stephensi foregut libraries, with default parameters for single-end, stranded reads, resulting in 44,311 putative transcripts with a median length of 332 bp, and a total length of 58.6 Mb. Assembled N. stephensi transcripts were used to validate the read counts generated by the Mouse CDS map and count pipeline.
Testing a priori hypotheses
For smooth muscle genes, we used two transcriptome datasets from the mouse Smooth Muscle Cell Genome browser . RNA-Seq Illumina reads were downloaded from the NCBI GEO database for mouse jejunum tissue (GSM1388412) and for sorted primary smooth muscle cells isolated from mouse jejunum tissue (GSM1388406).
To compare the representation of genes of various biological processes, we used the g:Cocoa tool within g:Profiler . This tool compares gene lists and determines which biological functions are significantly enriched in those lists. The analysis can treat the gene lists as ordered, in which genes listed higher in the list are weighted more heavily . The g:Cocoa tool also performs a custom multiple testing correction procedure, called g:SCS . For our analysis we took the 1000 most differentially expressed genes from woodrats and lab rats, ordered from highest to lowest in terms of log2(fold change). We investigated all Biological Processes within Gene Ontology and used the default g:SCS multiple testing correction. For our a priori hypotheses, we investigated whether various Gene Ontology terms associated with (1) epidermal development and differentiation and (2) fatty acid oxidation and ketogenesis. Specific GO terms can be found in the Results section and Table 2.
Last, based on previous studies in ruminants, we were interested in differences in expression of solute carriers (Slc) genes. We determined the top five most differentially expressed Slc genes for each species, as determined by log(fold change), excluding mitochondrial transporters.
We also wanted to compare expression using an untargeted approach to uncover other biological processes that may differ between the foreguts of woodrats and lab rats. First, we identified the top 10 differentially expressed genes in each species. We also used g:Cocoa to investigate other biological processes that are differentially enriched in the two gene lists (using the top 1000 most differentially expressed genes from each species as described above). We used the “moderate filtering” setting to select the most significantly enriched GO term within each parent term (since GO terms are hierarchical). We then sorted these by P-values, and present the 10 most highly enriched GO terms for both woodrats and lab rats.
At the time of dissection, another section of the foregut stomach was prepared for histology by pinning the section to corkboard and then placing the pinned samples in glass vials containing 10% neutral-buffered formalin. Samples were shipped to the University of Iowa Comparative Pathology Laboratory for routine tissue processing, embedding, sectioning, and staining. Tissues were stained with 3,3′-diaminobenzidine (DAB) using a biotin-tagged Dolichos biflorus agglutinin (DBA) lectin. The DBA lectin binds to terminal, nonreducing N-acetylgalactosamine (GalNAc) residues produced by B4galnt2 . Additionally, we determined the relative muscle composition (muscularis mucosae, muscularis propria, and the sum of these two as an estimate of ‘total muscle’) of the foregut wall. For each tissue section, the area of each of these parameters was divided by the total area of the stomach wall to get a percentage of the total area.
Validation of analysis pipeline
It was important to judge the average nucleotide sequence divergence contained in woodrat:mouse versus the lab rat:mouse alignments. The mean nucleotide substitution per read from the cross_match alignments was 7.0% (s = 2.2%) for the N. stephensi alignments and 5.4% (s = 1.9%) for R. norvegicus, with similar distributions observed for mRNAs within the differentially abundant classes (Additional file 2: Figure S1), indicating that the cross-species mapping pipeline did not have global, systematic bias in mapping N. stephensi and R. norvegicus reads to mouse transcripts and annotations.
A priori hypotheses
Epidermal-specific genes showed a more complex pattern of specific changes (Fig. 1c). There was no significant enrichment for the following Gene Ontology terms: (GO:0009913: epidermal cell differentiation; GO:0008544: epidermis development) when using g:Cocoa to investigate enrichment in functional categories. However, the term “keratinocyte differentiation” (GO:0030216) was significantly enriched in the woodrat foregut compared to the lab rat foregut (P = 0.03). To focus on differentially expressed epidermal genes, we used the mouse jejunum smooth muscle RNA-Seq data as a filter to identify 510 transcripts not found in smooth muscle (Additional file 1: Table S3 and S4, Additional file 2: Figure S2). We observed 124 epidermal-specific transcripts that were differentially abundant. Specifically, we observed enrichment of cornefin genes (Sprr1 and Sprr3 family) abundance (Fig. 1c, Additional file 1: Table S5) in laboratory rat foregut, while woodrats exhibited higher expression of filaggrin and repetin (Fig. 1c, Additional file 1: Table S5). There was also highly abundant expression of fatty acid-binding protein (Fabp9) in the N. stephensi foregut, a gene that is restricted in expression to male germ-cells in lab mice . Woodrats exhibited high-level expression of the known epidermal fatty acid-binding protein (Fabp5). Phylogenetic analysis of the N. stephensi Fabp5 and Fabp9 genes confirmed the identity of the transcripts quantitated by RNA-Seq (Additional file 2: Figure S3).
We found little evidence supporting the prediction that woodrats would exhibit higher expression of genes associated with fatty acid oxidation and ketogenesis. There were no significant differences between woodrats and lab rats in representation of several Gene Ontology terms related to ketogenesis and response to ketones (GO:1,901,654: response to ketone; GO:0042180: cellular ketone metabolic process; GO:0042181: ketone biosynthetic process; GO:1,901,655: cellular response to ketone). Similarly, we found no differences in several GO terms related to the metabolism of fatty acids (GO:0019395: fatty acid oxidation; GO:0006631: fatty acid metabolic process; GO:1,901,568: fatty acid derivative metabolic process; GO:0033559: unsaturated fatty acid metabolic process). However, woodrats were significantly enriched in genes associated with long-chain fatty acid metabolic process (GO:0001676; P = 0.045), and exhibited higher expression of fatty acid-binding protein 9 (Fabp9; Fig. 1c).
The five most differentially expressed solute carrier genes in woodrats and lab rats
Consistent with prediction from ruminants?
Monocarboxylic acid transporter
Facilitated glucose transporter
Facilitated glucose transporter
Proline IMINO transporter
Fatty acid transporter
The top ten differentially expressed genes from the foregut tissues of each species, as determined by log2(fold change)
Log2 (Fold Change)
Overexpressed in woodrats
Fatty Acid Binding Protein 9
Membrane Protein, Palmitoylated 1
Pregnancy Up-Regulated Nonubiquitous CaM Kinase
Calcium Binding And Coiled-Coil Domain 2
Ubiquitin-Conjugating Enzyme E2D 4
MANSC Domain Containing 4
Aldo-keto reductase family 1, member B7
Overexpressed in lab rats
Phospholipase A2, Group IIA
Retinol Binding Protein 7
Beta-1,4-N-Acetyl-Galactosaminyl Transferase 2
Ribonuclease, RNase A Family, 1
Amine Oxidase, Copper Containing, 1
GTP Binding Protein 6
Myosin Binding Protein C
The top ten most overrepresented Biological Functions in each species, determined using g:Cocoa
No. genes in term
No. genes overexpressed in woodrats
No. genes overexpressed in lab rats
Overrepresented in woodrats
Cellular component organization
Actin cytoskeleton organization
Intracellular signal transduction
Cellular component morphogenesis
Regulation of multicellular organismal process
Regulation of molecular function
Regulation of intracellular signal transduction
Overrepresented in lab rats
Single-organism metabolic process
Organic hydroxy compound metabolic process
Positive regulation of cell activation
Positive regulation of T cell activation
Nuclear DNA replication
L-serine biosynthetic process
Additionally, we found that genes associated with muscle activity were more highly expressed in the woodrat tissue (Smtnl1: Table 2, Table 3). We measured the proportion of the tissue wall in each species that was composed of various muscle tissues. Though our sample size was limited, we found that lab rats have a greater proportion of the muscularis mucosae (Fig. 4c), but woodrats have a greater proportion of muscularis propria and total muscle (Fig. 4d and e).
We compared gene expression in the foregut tissues of woodrats and lab rats, which differ in diet, stomach morphology, and density of associated microbes. The expression profiles of these two species were substantially different, highlighting the differential organization and function of the foregut tissue. We are unable to disentangle the many effects that could be underlying these differences in gene expression. They may be driven by diet, given that woodrats are herbivorous and lab rats are omnivorous. Resistant starches and dietary fiber can influence intestinal gene expression, though these differences may be modulated through the microbiota [29, 30]. Unfortunately, controlling for diet was not possible in this study, given that placing herbivorous rodents (including woodrats) on omnivorous diets can induce diabetes and other metabolic syndromes [31, 32]. Differential gut morphology also likely underlie the observed differences, given that woodrats have thicker muscle layer, and thus we observed higher abundance of genes associated with musculature (discussed below). Last, microbial communities are known to induce changes in host gene expression . However, germ-free woodrats have not been developed, so we cannot control for these effects. It should also be noted that microbial communities  and host physiology  can change as an effect of captivity. Regardless of the exact mechanisms, the differences in gene expression between these two species reveal differential biological functions. We discuss our findings in relation to previous work conducted in ruminant systems.
Genes related to muscle system processes were over-represented in woodrat foregut tissue. This is likely due to the fact that a larger proportion of the woodrat foregut wall is composed of muscle tissue when compared to lab rats. In ruminants, muscle tissue is important for the mixing and passage of food material [13, 14]. Woodrats do not have a physical separation between stomach regions, as is observed in ruminants . Thus, it could be that peristaltic movements in the woodrat foregut are important for the mixing and retention of food material in the foregut for microbial digestion. However, it should be noted that retention of food material in this chamber is relatively short (typically less than 1.5 h; ). Of particular interest is the increased level of smoothelin like-1, which regulates smooth muscle contraction and relaxation, and changes in expression levels are known to alter vascular response to exercise, as well as uterine smooth muscle activity during pregnancy . The up-regulation of Smtnl1 in woodrat foregut may therefore be a specific adaptation facilitating increased foregut motility.
In ruminants, the epithelial lining is composed of keratinized and stratified squamous epithelium, which has histological similarities to mammalian skin . Rumen tissue exhibits high expression of a number of genes associated with epidermal development, especially a homolog of small proline-rich proteins (Sprr) and a trichohyalin-like gene . Interestingly, the Sprr genes are highly expressed in sheep rumen, but not skin tissue . In the current study, we found that the woodrats and lab rats had differential abundance of a number of genes associated with epidermal differentiation and development. From the EDC gene cluster, the Sprr1 and Sprr3 genes and the involucrin gene showed higher expression in the lab rat foregut. Since the Sprr proteins play known roles in modulating the barrier function and biomechanical properties of the epithelial surface, these abundance differences may also contribute to functional differences in barrier response to gut microflora. The woodrat foregut exhibited higher expression of repetin (Rptn), filaggrin (Flg), and cystatin A1 (Csta1) genes associated with the cornified envelope [38–40]. Together, these results suggest the molecular basis of this cornification may differ between lab rats and woodrats.
In ruminants, the rumen tissue itself utilizes short chain fatty acids (SCFAs) as energy substrates , and thus exhibits high expression of genes associated with metabolizing SCFAs [4–6]. We found no support for the hypothesis that woodrat tissues would have higher expression of genes related to ketogenesis and metabolism of fatty acids, even though this chamber contains high concentrations of SCFAs (~200 mM; ) relative to lab rats. Though, concentrations in the gut lumen are the result of both fatty acid production by microbes and absorption by the host. Rumen tissue exhibits higher rates of fatty acid absorption compared to the equivalent stomach regions in pigs and horses , and in horses, the gastric muscosa exhibits higher rates of SCFA absorption than the foregut region . Thus, one potential explanation could be that there is low capacity for SCFA absorption in the woodrat foregut, and thus SCFAs may be absorbed and utilized by subsequent gut regions. Alternatively, SCFAs could be absorbed in the woodrat foregut, and be passed into the blood stream and metabolized in the liver, as occurs in other non-ruminant species . This process might be facilitated by the high expression of fatty acid binding proteins, which are largely considered to be lipid chaperones . In addition to high-level expression of Fabp5 in both rodent species (a known epidermal fatty acid binding protein), woodrats specifically expressed high levels of Fabp9, whose tissue specific expression in other mammals is restricted to testis, where it is the most abundant protein of the sperm perinuclear theca . Physiological studies investigating the absorption and utilization rates of SCFAs by woodrat foregut tissues would disentangle these possibilities.
We observed a number of differences in the types of transporters expressed in the foregut tissues of woodrats and lab rats. Woodrats exhibited higher expression of a monocarboxylic acid transporter (Slc16a10). Proteins from the Slc16a family are thought to be important for the absorption of SCFAs in rumen tissue , and thus similar absorption may be occurring in the woodrat foregut. Additionally, woodrats exhibited higher expression of two facilitative glucose transporters (Slc2a10, Slc2a12), which transport glucose across the basolateral membrane into the blood stream . The expression of facilitative glucose transporters increases over development of ruminants, concordant with the development of fermentation . High expression of these transporters in ruminants (and woodrats) is puzzling, given that apical absorption of glucose is presumed to be low in ruminants, since simple carbohydrates would largely undergo fermentation in the rumen . Therefore, the role of facilitative transporters in the foregut epithelia of ruminants and woodrats remains unclear.
The foregut region of lab rats was highly enriched in genes associated with immune responses. The rats used in our study were healthy and not immune-challenged, and thus we assume this difference represents a higher basal expression of immune-related genes in this tissue. Tolerance, or ‘unresponsiveness’ to microbes is presumed to be important for allowing gut microbes to reside within the gastrointestinal tract . Therefore, woodrats may have a decreased immune response to microbes in the foregut, allowing the resident microbial community to flourish. Consistent with this notion is the higher expression of ribonuclease (Rnase1) in the lab rat foregut, which is considered to have antimicrobial activity in the gut . The Pla2g2a gene encoding the Group IIA secretory phospholipase A2 (sPLA2 IIA) is a well-known biomarker for inflammatory disease and was highly expressed in the lab rat while undetectable in the woodrat tissues. sPLA2 enzymes convert phospholipids used in the formation of extracellular lamellar membranes and participate in the acidification of the stratum corneum, important for antimicrobial defense , and mammalian sPLA2 IIA also specifically displays potent enzymatic antibacterial activity . Together, these immune-related and antimicrobial genes may be important in suppressing microbial growth in the foregut chamber of lab rats. Gene knock-out studies in lab mice could be conducted to better investigate the connections between immune-related genes and microbial abundance in the foregut chamber.
Another interesting gene expressed at higher levels in the lab rat foregut tissue compared to woodrats was beta-1,4-N-acetyl-galactosaminyl transferase 2, or B4galnt2. This gene plays a role in the formation of host glycans, and is known to have long-term balancing selection of expression variation in rodents [51, 52]. Host glycans are thought to be important for the regulation of microbial communities , and indeed, knockout mice (B4galnt2 −/− ) harbor distinct gut microbial communities compared to wild-type mice . Differential B4galnt2 expression in lab rats was supported by DBA lectin staining, which also revealed differential localization, such that lab rats maintain GalNAc residues on the epithelial surface, while woodrats do not. Interestingly, there is variation in the localization of B4galnt2 expression and GalNAc residue localization across wild populations of mice, which is thought to be due to balancing selection between susceptibility to gut pathogens and a bleeding disorder [51, 52]. Additionally, in ruminants expression of B4galnt2 significantly decreases with the development of a fermenting rumen . The function and consequences of reduced B4galnt2 expression in the woodrat foregut is an interesting area of future research.
Together, these results inform us about the underlying changes in gene expression that are associated with hosting a foregut microbial community. We observed a number of differences that are consistent with previous findings in ruminants (high expression of facilitative glucose transporters, lower expression of B4galnt2). Additionally, a number of differences were observed between ruminants and rodents. For example, the molecular basis of cornification requires further study, and it is still unclear whether the capacity of woodrat foregut tissue to absorb and oxidize SCFAs is similar to that of the rumen. It would be interesting to conduct studies similar to this in other rodents with more developed foregut chambers. For example, a number of rodent genera actually have macrovilli in their stomachs that support microbial attachment of symbiotic microbes; , as well as in foregut fermenting primates and macropod marsupials. Together, these studies would reveal the shared changes in gene expression associated with hosting a foregut microbial community and shed light on the evolution of these microbial symbioses.
This research was supported by the National Science Foundation (Doctoral Dissertation Improvement Grant, DEB 1210094, to M.D.D. and K.D.K.; DEB 1342615 to M.D.D.; and DBI 1400456 to K.D.K) and the National Institutes of Health (T32DK007673 Training Grant to K.D.K.).
Availability of data and materials
The datasets generated during and/or analysed during the current study are available in the NCBI Gene Expression Omnibus repository under accession GSE84381.
KDK, KFO, and DD generated RNA libraries for sequencing. DKM performed histological analysis. KDK, KFO, and RBW conducted data analysis and interpretation. CD, RBW, and MDD oversaw the study and aided with interpretation and conclusions. KDK and RBW wrote the paper, and all authors provided feedback and have approved the final manuscript.
All procedures described in this paper were approved by the University of Utah’s Animal Care and Use Committee under protocol #12–12,010.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Mackie R. Mutualistic fermentative digestion in the gastrointestinal tract: diversity and evolution. Integr Comp Biol. 2002;42:319–26.View ArticlePubMedGoogle Scholar
- Stevens CE, Hume ID. Comparative physiology of the vertebrate digestive system. Cambridge: Cambridge University Press; 1995.Google Scholar
- Brawand D, Soumillon M, Necsulea A, Julien P, Csárdi G, Harrigan P, Weier M, Liechti A, Aximu-Petri A, Kircher M, et al. The evolution of gene expression levels in mammalian organs. Nature. 2011;478:343–8.View ArticlePubMedGoogle Scholar
- Connor EE, Baldwin VIRL, Li C-J, Li RW, Chung H. Gene expression in bovine rumen epithelium during weaning identifies molecular regulators of rumen development and growth. Funct Integr Genomics. 2013;13:133–42.View ArticlePubMedGoogle Scholar
- Lane MA, Baldwin VIRL, Jesse BW. Developmental changes in ketogenic enzyme gene expression during sheep rumen development. J Anim Sci. 2002;80:1538–44.View ArticlePubMedGoogle Scholar
- Naeem A, Drackley JK, Stamey J, Loor JJ. Role of metabolic and cellular proliferation genes in ruminal development in response to enhanced plane of nutrition in neonatal Holstein calves. J Dairy Sci. 2012;95:1807–20.View ArticlePubMedGoogle Scholar
- Jiang Y, Xie M, Chen W, Talbot R, Maddox JF, Faraut T, Wu C, Muzny DM, Li Y, Zhang W, et al. The sheep genome illuminates biology of the rumen and lipid metabolism. Science. 2014;344:1168–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Carleton MD. A survey of gross stomach morphology in new world Cricetinae (Rodentia, Muroidea), with comments on functional interpretations. In. Ann Arbor, Michigan: Museum of Zoology, University of Michigan; 1973.Google Scholar
- Kohl KD, Miller AW, Marvin JE, Mackie R, Dearing MD. Herbivorous rodents (Neotoma spp.) harbour abundant and active foregut microbiota. Environ Microbiol. 2014;16:2869–78.View ArticlePubMedGoogle Scholar
- Elliott SN, Buret A, McKnight W, Miller MJS, Wallace JL. Bacteria rapidly colonize and modulate healing of gastric ulcers in rats. Am J Physiol Gastrointes Liver Physiol. 1998;275:G425–32.Google Scholar
- Elsden SR, Hitchcock MWS, Marshall RA, Phillipson AT. Volatile acid in the digesta of ruminants and other animals. J Exp Biol. 1946;22:191–202.PubMedGoogle Scholar
- Gärtner K, Pfaff J. The forestomach in rats and mice, a food store without bacterial protein digestion. Zbl Vet Med A. 1979;26:530–41.View ArticleGoogle Scholar
- Wyburn RS. The mixing and propulsion of the stomach contents of ruminants. In: Ruckebusch Y, Thivend P, editors. Digestive physiology and metabolism in ruminants; 1979. p. 35–51.Google Scholar
- Okine EK, Mathison GW, Hardin RT. Effects of changes in frequency of reticular contractions on fluid and particulate passage rates in cattle. J Anim Sci. 1989;67:3388–96.View ArticlePubMedGoogle Scholar
- Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30:2114–20.View ArticlePubMedPubMed CentralGoogle Scholar
- Magoč T, Salzberg SL. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics. 2011;27:2957–63.View ArticlePubMedPubMed CentralGoogle Scholar
- Mudge JM, Harrow J. Creating reference gene annotation for the mouse C57BL6/J genome assembly. Mamm Genome. 2015;26:366–78.View ArticlePubMedPubMed CentralGoogle Scholar
- Gordon D, Abajian C, Green P. Consed: a graphical tool for sequence finishing. Genome Res. 1998;8:195–202.View ArticlePubMedGoogle Scholar
- Robinson MD, McCarthy DJ. Smyth GK: edgeR: a bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–40.View ArticlePubMedGoogle Scholar
- Liu J, Li G, Chang Z, Yu T, Liu B, McMullen R, Chen P, Huang X. BinPacker: packing-based de novo transcriptome assembly from RNA-seq data. PLoS Comput Biol. 2016;12:e1004772.View ArticlePubMedPubMed CentralGoogle Scholar
- Lee MY, Park C, Berent RM, Park PJ, Fuchs R, Syn H, Chin A, Townsend J, Benson CC, Redelman D, et al. Smooth muscle cell genome browser: enabling the identification of novel serum response factor target genes. PLoS One. 2015;10:e0133751.View ArticlePubMedPubMed CentralGoogle Scholar
- Reimand J, Arak T. Vilo J: g:profiler—a web server for functional interpretation of gene lists (2011 update). Nucl Acids Res. 2011;39:W307–15.View ArticlePubMedPubMed CentralGoogle Scholar
- Reimand J, Kull M, Peterson H, Hansen J. Vilo J: g:profiler–a web-based toolset for functional profiling of gene lists from large-scale experiments. Nucl Acids Res. 2007;35:193–200.View ArticleGoogle Scholar
- Mohlke KL, Purkayastha AA, Westrick RJ, Smith PL, Petryniak B, Lowe JB, Ginsburg D. Mvwf, a dominant modifier of murine von Willebrand factor, results from altered lineage-specific expression of a glycosyltransferase. Cell. 1999;96:111–20.View ArticlePubMedGoogle Scholar
- McCarthy DJ, Chen P, Smyth GK. Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. Nucl Acids Res. 2012;40:4288–97.View ArticlePubMedPubMed CentralGoogle Scholar
- Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15–21.View ArticlePubMedGoogle Scholar
- Borman M, Freed TA, Haystead TA, MacDonald JA. The role of the calponin homology domain of smoothelin-like 1 (SMTNL1) in myosin phosphatase inhibition and smooth muscle contraction. Mol Cell Biochem. 2009;327:93–100.View ArticlePubMedPubMed CentralGoogle Scholar
- Selvaraj V, Asano A, Page JL, Nelson JL, Kothapalli KSD, Foster JA, Brenna T, Weiss RW, Travis AJ. Mice lacking FABP9/PERF15 develop sperm head abnormalities but are fertile. Dev Biol. 2010;348:177–89.View ArticlePubMedPubMed CentralGoogle Scholar
- Conlon MA, Kerr CA, McSweeney CS, Dunne RA, Shaw JM, Kang S, Bird AR, Morell MK, Lockett TJ, Molloy PL, et al. Resistant starches protect against colonic DNA damage and alter microbiota and gene expression in rats fed a western diet. J Nutr. 2012;142:832–40.View ArticlePubMedPubMed CentralGoogle Scholar
- Haenen D, Zhang J, Souza da Silva C, Bosch G, van der Meer IM, van Arkel J, van den Borne JJGC, Gutiérrez OP, Smidt H, Kemp B, et al. A diet high in resistant starch modulates microbiota composition, SCFA concentrations, and gene expression in pig intestine. J Nutr. 2013;143:274–83.View ArticlePubMedGoogle Scholar
- Schmidt-Nielsen K, Haines HB, Hackel DB. Diabetes mellitus in the sand rat induced by standard laboratory diets. Science. 1964;143:689–90.View ArticlePubMedGoogle Scholar
- Post DM, Hawkins BL, Eldridge JA. The effects of nutrition-induced abnormal food metabolism in the Southern Plains woodrat (Neotoma micropus): comparisons of variations of the western diet. J Anim Physiol Anim Nutr. 2015;99:29–36.View ArticleGoogle Scholar
- Hooper LV, Wong MH, Thelin A, Hansson L, Falk PG, Gordon JI. Molecular analysis of commensal host-microbial relationships in the host. Science. 2001;291:881–4.View ArticlePubMedGoogle Scholar
- Kohl KD, Skopec MM, Dearing MD. Captivity results in disparate loss of gut microbial diversity in closely related hosts. Conserv Physiol. 2014;2:cou009.View ArticlePubMedPubMed CentralGoogle Scholar
- Kuhlman JR, Martin LB. Captivity affects immune redistribution to skin in a wild bird. Funct Ecol. 2010;24:830–7.View ArticleGoogle Scholar
- Lontay B, Bodoor K, Sipos A, Weitzel DH, Loiselle D, Safi R, Zheng D, Devente J, Hickner RC, McDonnell DP, et al. Pregnancy and smoothelin-like protein 1 (SMTNL1) deletion promote the switching of skeletal muscle to a glycolytic phenotype in human and mice. J Biol Chem. 2015;290:17985–98.View ArticlePubMedPubMed CentralGoogle Scholar
- Dobson MJ. A histological study of the organization of the rumen epithelium of sheep. Q J Exp Physiol Cogn Med Sci. 1956;41:247–53.PubMedGoogle Scholar
- Huber M, Siegenthaler G, Mirancea N, Marenholz I, Nizetic D, Breitkreutz D, Mischke D, Hohl D. Isolation and characterization of human repetin, a member of the fused gene family of the epidermal differentiation complex. J Invest Dermatol. 2005;124:998–1007.View ArticlePubMedGoogle Scholar
- Simon M, Haftek M, Sebbag M, Montézin M, Girbal-Neuhauser E, Schmitt D, Serre G. Evidence that filaggrin is a component of cornified cell envelopes in human plantar epidermis. Biochem J. 1996;317:173–7.View ArticlePubMedPubMed CentralGoogle Scholar
- Takahashi M, Tezuka T, Katunuma N. Phosphorylated cystatin α is a natural substrate of epidermal transglutaminase for formation of skin cornified envelope. FEBS Lett. 1992;308:79–82.View ArticlePubMedGoogle Scholar
- Pennington RJ. The metabolism of short-chain fatty acids in the sheep. 1. Fatty acid utilization and ketone body production by rumen epithelium and other tissues. Biochem J. 1952;51:251–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Argenzio RA, Southworth M, Stevens CE. Sites of organic acid production and absorption in the equine gastrointestinal tract. Am J Phys. 1974;226:1043–50.Google Scholar
- Hocquette J-F, Bauchart D. Intestinal absorption, blood transport and hepatic and muscle metabolism of fatty acids in preruminant and ruminant animals. Reprod Nutr Dev. 1999;39:27–48.View ArticlePubMedGoogle Scholar
- Furuhashi M, Hotamisligil GS. Fatty acid-binding proteins: role in metabolic diseases and potential as drug targets. Nat Rev Drug Discov. 2008;7:489–503.View ArticlePubMedPubMed CentralGoogle Scholar
- Graham C, Gatherar I, Haslam I, Glanville M, Simmons NL. Expression and localization of monocarboxylate transporters and sodium/proton exchangers in bovine rumen epithelium. Am J Phys. 2007;292:R997–R1007.Google Scholar
- Boyer S, Sharp PA, Debnam ES, Balwin SA, Srai SKS. Streptozotocin diabetes and the expression of GLUT1 at the brush border and basolateral membranes of intestinal enterocytes. FEBS Lett. 1996;396:218–22.View ArticlePubMedGoogle Scholar
- Kelly D, Conway S, Aminov R. Commensal gut bacteria: mechanisms of immune modulation. Trends Immunol. 2005;26:326–33.View ArticlePubMedGoogle Scholar
- Mukherjee S, Vaishnava S, Hooper LV. Multi-layered regulation of intestinal antimicrobial defense. Cell Mol Life Sci. 2008;65:3019–27.View ArticlePubMedGoogle Scholar
- Ilic D, Bollinger JM, Gelb M, Mauro TM. sPLA2 And the epidermal barrier. Biochim Biophys Acta - Mol Cell Biol Lipids. 1841;2014:416–21.Google Scholar
- Weiss JP. Molecular determinants of bacterial sensitivity and resistance to mammalian group IIA phospholipase A2. Biochim Biophys Acta Biomembr. 1848;2015:3072–7.Google Scholar
- Staubach F, Künzel S, Baines AC, Yee A, McGee BM, Bäckhed F, Baines JF, Johnsen JM. Expression of the blood-group-related glycosyltransferase B4galnt2 influences the intestinal microbiota in mice. ISME J. 2012;6:1345–55.View ArticlePubMedPubMed CentralGoogle Scholar
- Linnenbrink M, Johnsen JM, Montero I, Brzezinski CR, Harr B, Baines JF. Long-term balancing selection at the blood group-related gene B4galnt2 in the genus Mus (Rodentia; Muridae). Mol Biol Evol. 2011;28:2999–3003.View ArticlePubMedPubMed CentralGoogle Scholar
- Hooper LV, Gordon JI. Glycans as legislators of host-microbial interactions: spanning the spectrum from symbiosis to pathogenicity. Glycobiol. 2001;11:1–10.View ArticleGoogle Scholar
- Vorontsov NN. Macromutations and evolution: fixation of Goldschmidt's macromutations as species and genus characters. Papillomatosis and appearance of macrovilli in the rodent stomach. Russ J Genet. 2003;39:422–6.View ArticleGoogle Scholar