Abstract
To study the effects of feeding on taste receptor cell transcriptional regulation, we performed RNA-seq analysis of circumvallate taste buds isolated from mice before or after food consumption. Here we report and compare the taste bud transcriptomes obtained from food-deprived, satiated, and ad libitum fed control mice. Despite sample heterogeneity inherent to the whole taste bud transcriptome, bioinformatics analysis yielded 144 differentially expressed transcripts associated with immunity, cytoskeletal structure, and protein folding between these groups. We also profiled the transcriptome obtained from ad libitum fed control mice based on receptor related gene ontology terms, demonstrating a use of this dataset in the identification of novel TRC receptors. The data presented here suggest that transcriptional regulation of immune cytokine signaling occurs in TRCs shortly after meal consumption, though additional analysis is required to confirm this hypothesis.
Introduction
Taste buds are comprised of specialized epithelial cells known as taste receptor cells (TRCs) (Chaudhari and Roper 2010). These TRCs can be classified into 3 categories referred to as Type I, II, and III based on their morphological characteristics (Murray 1993; Pumplin et al. 1997; Yee et al. 2001). Type I TRCs are the proposed sodium sensitive cells, detected via expression of epithelial sodium channels (ENaC), which also serve a glial-like supporting role in the bud (Bigiani 2001; Dvoryanchikov et al. 2009; Vandenbeuch et al. 2008). Type II TRCs express different G-protein coupled receptors (GPCRs), which transduce sweet, bitter, and umami taste qualities (Adler et al. 2000; Mueller et al. 2005; Nelson et al. 2002; Nelson et al. 2001; Zhang et al. 2003; Zhao et al. 2003). The final cell type, Type III TRCs, detect sour tastants via cytoplasmic changes in pH (Huang et al. 2008; Ishimaru et al. 2006; Tu et al. 2018). While many genetic and electrophysiological studies have been conducted on the gustatory system, substantially less data on the TRC transcriptome is available. With the exception of a recent study which used RNA-seq to profile the transcriptome of Tas1r3 positive and Type III TRCs (Sukumaran et al. 2017), the few other TRC transcriptome studies used micro array technology as opposed to RNA-seq (Hevezi et al. 2009; Moyer et al. 2009). Acquisition of transcriptome data for all TRC subtypes is essential for the discovery of novel receptors, as well as obtaining a better understanding of TRC gene expression patterns during different circadian and digestive phases.
The functioning of the gustatory system is intimately linked with the functioning of the digestive system and as such, can influence and be influenced by the different phases of digestion. For example, during the cephalic phase of digestion, initial tastant detection by sweet responsive TRCs stimulates secretion of insulin from pancreatic islets (Tonosaki et al. 2007). Likewise, TRCs express receptors for a number of gastrointestinal peptides and hormones that are secreted into the circulation and saliva during the cephalic, gastric, and intestinal digestive phases (Awazawa et al. 2011; Elson et al. 2010; Hurtado et al. 2012; Kawai et al. 2000; La Sala et al. 2013; Martin et al. 2009; Martin et al. 2010; Shin et al. 2010). These include receptors for glucagon-like peptide 1 (GLP-1), peptide tyrosine-tyrosine (PYY), glucagon, and insulin among other modulatory peptides and hormones. Furthermore, satiety peptides PYY, GLP-1, and glucagon modulate the taste perception of specific taste qualities, presumably by interacting with cognate receptors expressed on TRCs (Elson et al. 2010; La Sala et al. 2013; Takai et al. 2015). While a modulatory effect of these peptides on taste perception has been demonstrated in animal models, these results are contextually dependent on the ingestive or sated state of the animals. In other words, levels of GI hormones known to influence the gustatory system are not static throughout the day. In the TRCs as well, there may be transcriptional or translational differences depending on the circadian, digestive, or ingestive phase of the animal. To gain a better understanding of the transcriptional differences in the peripheral gustatory system during periods of feeding or fasting, we performed RNA-seq on circumvallate taste buds isolated from wild-type (WT) C57BL6/J mice under different feeding regiments. Specifically, we performed RNA-seq on taste bud messenger RNA isolated from 3 groups of mice referred to a “satiated”, “hungry”, and “ad lib”, then identified differentially expressed genes between each group. We performed gene enrichment analysis of the differentially expressed genes and found enrichment in pathways associated with immunity, cytoskeletal structure, chemokines, chaperone proteins, and protease inhibitors. This suggests that these specific pathways may be regulated in TRCs in response to feeding, though further investigation into the role of these pathways is TRCs is needed.
Materials and Methods
Animals
WT C57BL6/J mice used in this study were bread in house, housed at 22-24 oC with a 14/10 hr light/dark cycle, and given ad libitum access to food and water except where otherwise noted. In the 24 hrs prior to tissue collection for RNA-seq analysis, 2 groups of mice were placed on restricted feeding schedules; while a third control group, hear after referred to as “ad lib” (n=7), received ad libitum access to food throughout the experiment. The first group of food restricted mice, hear after referred to as “hungry” (n=7), were fasted for the 24 hrs prior to tissue collection. The second group of mice, hear after referred to as “satiated” (n=7), were fasted for 23 hrs, then given ad libitum access to food for 1 hr prior to tissue collection. All animals had ad libitum access to water throughout the experiments except where specifically noted. Only male mice, age 20-24 weeks, were used in order to eliminate influences of the estrus cycle on hormone levels and feeding behavior. As well, the average body weight of mice was kept relatively constant across each group. Procedures were done in accordance with principals set forth by the National Research Council’s guide for the Care and Use of Laboratory Animals. This study was approved by the University of Florida Institutional Animal Care and Use Committee (IACUC).
Circumvallate Taste Bud Isolation
Circumvallate taste buds were isolated from mouse tongues in accordance with protocols reported by Huang et al. 2005 (Huang 2005, Huang 2016). Briefly, mice were euthanized by CO2 asphyxiation followed by cervical dislocation after which, tongues were dissected. Tongues were washed in Tyrode’s buffer (140 mM NaCl, 5 mM KCl, 2 mM CaCl2, 1 mM MgCl2 10 mM HEPES, 10 mM glucose, 10 mM sodium pyruvate, and 5 mM NaHCO3 pH ~ 7.2) then, the lateral ends of the circumvallate papillae were each injected with 50 μl of protease cocktail (Collagenase A 1 mg/ml, Dispase II 2.5 mg/ml, Elastase 0.25 mg/ml, and DNAse I 0.5 mg/ml) and tongues were incubated in calcium/magnesium free Tyrode’s buffer. Tongues were pinned in Slygard gel, covered in Tyrode’s buffer, then using surgical scissors the circumvallate epithelium was removed, inverted, and pinned to a new Slygard gel. Circumvallate epithelia were incubated shortly with protease cocktail, washed with Tyrode’s buffer, then washed with calcium/magnesium free Tyrode’s buffer, before pinning in fresh Slygard gels and covering with Tyrode’s buffer containing 0.5 mg/ml DNAse I. Taste buds were collected by manual aspiration using 60-80 μm flame polished capillary pipettes, ejected onto Cell-Tak coated coverslips, then washed successively in droplets of Tyrode’s buffer. Taste buds from each animal were then collected into microfuge tubes for RNA extraction. All buds were collected 1 hr after onset of the dark cycle.
RNA extraction and cDNA library generation
RNA was extracted from taste buds using the Absolutely RNA Nanoprep Kit (Stratagene) and treated with RNAseOUT (ThermoFisher) to preserve RNA integrity. Quality control and total RNA amounts were determined using a Bioanalyzer, and only samples with RNA integrity numbers (RIN) greater than 7 were used for cDNA synthesis and subsequent sequencing. Poly-adenylated RNA was converted to cDNA using SuperScript III Reverse Transcriptase (ThermoFisher) and oligo(dT) primers. A cDNA library for Illumina sequencing was then constructed using the RNA-seq Library Preparation Kit (Kappa Biosystems). The cDNA libraries were transferred to the University of Florida Interdisciplinary Center for Biotechnology Research (ICBR) for next generation sequencing.
RNA-Seq data analysis
Single end RNA-seq was performed on an Illumina HiSeq 2000 with average read length of 75 bp. A total of 7 biological replicates (n=7) were used for each experimental group with an average of 30 million reads obtained for each animal. Quality control of fastq files was conducted using FastQC (Andrews) after which additional trimming of adapter sequences was done using Trimmomatic (Bolger et al. 2014). Trimmed reads were then mapped to the GRCm38 p.4 mouse reference genome, downloaded from Ensembl (Zerbino et al. 2018), using Tophat2 running the Bowtie2 short read alignment algorithm (Trapnell et al. 2010). Samtools was used to convert sequence alignment map (SAM) output files to binary alignment map (BAM) format and transcript assembly was completed using Cufflinks (Li et al. 2009; Trapnell et al. 2010). A combined gene transfer format (GTF) assembly file was generated for differential expression analysis, using Cuffmerge, containing the reference GTF genome as well as the Cufflinks output for each sample (Trapnell et al. 2010). Cuffdiff was used to determine pairwise differential gene expression between the various experimental groups, and the cummeRbund R package was used to generate graphical representations of the RNA-seq data set (Trapnell et al. 2012; Trapnell et al. 2010).
DAVID Enrichment Analysis
We used DAVID Bioinformatics (DAVID) (Huang et al. 2009) to perform functional annotation enrichment analysis of the differentially expressed genes identified via Cuffdiff. DAVID identified annotation records for 128 of the 144 differentially expressed genes and used annotation information from the following databases for enrichment analysis: COG_ONTOLOGY, UP_KEYWORDS, UP_SEQ_FEATURE, GOTERM_BP_DIRECT, GOTERM_CC_DIRECT, GOTERM_MF_DIRECT, BIOCARTA, KEGG_PATHWAY, INTERPRO, PIR_SUPERFAMILY, and SMART. Using the functional annotation clustering program with high stringency classifications, similarity threshold of 0.85, multiple linkage threshold of 0.5, and an EASE enrichment threshold of 0.01, DAVID identified 5 enriched annotation clusters among the differentially expressed genes.
Panther Gene Ontology Profiling
To profile the transcript expression of receptor related genes in wild-type circumvallate taste buds, we used Panther gene ontology (GO) (Mi et al. 2017; Mi et al. 2013; The Gene Ontology Consortium 2017). All transcripts expressed at a frequency greater or equal to 2 transcripts per million (TPM) were used for gene ontology profiling. Panther identified a total of 6022 genes which were then profiled by molecular function GO (3789 total) and protein class GO (3540 total) terms. Using cummeRbund and the ggplot2 package in R, we generated expression plots and GO pie charts depicting transcripts associated with the GO molecular function subclass: G-protein coupled receptors. The same graphs were generated for transcripts associated with the Protein class subcategories: G-protein coupled receptors, cytokine receptors, ligand-gated receptors, and protein kinase receptors.
Results
Biological Replicate Variability and TRC Validation
To assess the intergroup variation of gene expression, we used cummeRbund to determine the squared coefficient of variance (CV2) as well as the gene dispersion of each group. The CV2 for each group was relatively high, indicating a large degree of variance between biological replicates within groups (Figure 1, panel A). As well, each group was overdispersed, further indicating a high degree of variation between biological replicates (Figure 1, panel B). Both models of variance showed a higher degree of variability at lower transcript expression levels as expected (Yoon and Nam 2017). To confirm enrichment of TRC RNA in our dataset, we queried the expression levels of multiple TRC markers, corresponding to all TRC subtypes, and confirmed that markers for Type I, II, and III TRCs are indeed expressed (Figure 2). There were no significant differences in the expression levels of the queried TRC markers between groups.
Differential Gene Expression between Ad Lib, Hungry, and Satiated Mice
Due to the high variance between biological replicates, we anticipated a relatively low number of differentially expressed genes between ad lib, hungry, and satiated mice. However, using a false discovery rate (FDR) of 5%, we identified a total of 144 differentially expressed genes between the 3 groups. Pairwise volcano plots indicate differentially expressed genes between ad lib, hungry, and satiated mice respectively (Figure 3). The significance expression matrix (Figure 4) displays the number of differentially expressed genes identified pairwise between each group. To look for patterns of gene expression across groups, genes were clustered by FPKM and expression was shown as a heatmap (Figure 5, panel A). Based on the heatmaps alone, there are only small clusters of genes with similar expression patterns. A heatmap showing fold change in FPKM of differentially expressed genes for hungry and satiated mice, relative to ad lib mice was constructed to observe changes in gene expression from the control state (Figure 5, panel B). There are small clusters of genes in satiated and hungry mice with opposing expression patterns, relative to ad lib mice, indicating that their expression may be associated with a pre or post-ingestive state (Figure 5). However, this type of comparison assumes that the ad lib mice are in an intermediary state relative to the satiated and hungry mice, which may not be the case.
Functional Clustering of Differentially Expressed Genes using DAVID
To cluster differentially expressed genes based on their biological function, we used DAVID bioinformatics to perform functional annotation clustering using annotation information from the default DAVID databases (Huang et al. 2009). Using high stringency enrichment criteria with an EASE threshold of 0.01, DAVID identified 5 functional annotation clusters. Heatmaps depicting genes and associated annotation terms for each functional annotation cluster are shown in Figure 6. Functional annotation clusters 1-5 largely contain cytokeratin, chemokine, serpin protease inhibitor, immune related, and chaperone related genes, respectively (Figure 6). Expression levels of clustered genes were plotted across experimental groups to observe the pattern of gene expression changes (Figure 7). For each functional annotation cluster, most genes display the same expression pattern between ad lib, satiated, and hungry mice (Figure 7). Specifically, most enriched genes are upregulated in satiated mice relative to hungry or ad lib (Figure 7).
Profiling of Taste Bud Transcriptome Using Keyword Gene Ontology Terms
Because of the limited transcriptome data available for murine taste buds, we thought it useful to determine an expression profile of transcripts associated with key GO annotations; specifically, those associated with receptor function. Using Panther gene ontology, we profiled the whole circumvallate taste bud transcriptome, from ad lib mice, for genes with expression levels greater than or equal to 2 transcripts per million (TPM). We specifically looked for the presence of genes with “receptor” associated GO terms. Pie charts, generated using Panther and the ggplot2 package in R, display either molecular function (3789 transcripts total, Figure 8 panel A) or protein class (3540 transcripts total, Figure 8 panel B) GO categories for the ad lib transcriptome at different annotation levels. Bar graphs depict expression of transcripts associated with receptor related GO terms, namely: GO:0004872, PC00084, PC00021, PC00141, and PC00194 (Figure 9). These data serve to validate previously reported findings, as well as identify uncharacterized receptors in the gustatory system which could serve as potential research targets.
Discussion
In this study, we used RNA-seq analysis to generate circumvallate taste bud transcriptomes from hungry and satiated mice. These transcriptomes contain RNA isolated from heterogeneous cell populations comprised primarily of Type I, II, and III TRCs. Because of the inherent heterogeneity present in the taste bud, as well as contamination with non-taste tissue present from taste bud isolation, we observed high sample variability as evident by dispersion and CV2 statistical analysis. Despite this variance, using a large number of biological replicates (n=7), we successfully identified 144 genes that are differentially expressed between satiated, hungry, and ad lib mice. Using gene enrichment analysis we identified 5 gene clusters associated with specific cellular pathways, which are differentially regulated between these groups. Additionally, we profiled the ad lib transcriptome based on receptor related GO terms in order to identify receptors novel to the gustatory system.
Gene enrichment analysis of differentially expressed genes indicated that pathways associated with chemokines, immunity, cytokeratins, serpin protease inhibitors, and chaperone proteins may be differentially regulated between hungry or sated animals. Differential regulation of chemokine and immune related transcripts between hungry and satiated mice is particularly intriguing, as immune cytokine signaling seems to be involved in taste cell communication (Feng et al. 2014b; Hevezi et al. 2009). Specifically, subpopulations of TRCs express interferon-γ, tumor necrosis factor, interlukin-10, and their cognate receptors (Feng et al. 2014a; Feng et al. 2012; Shi et al. 2012; Wang et al. 2007). They also express multiple types of toll-like receptors (Wang et al. 2009), which are responsible for detecting pathogenic microbial proteins. Furthermore, excessive inflammation in the taste bud is associated with taste dysfunction and increased TRC apoptosis, highlighting the importance of these pathways in taste bud homeostasis (Wang et al. 2009). Differential expression of cytokines and interferon inducible genes between hungry and satiated mice suggests that up or downregulation of these pathways, which are important for cell differentiation, proliferation, and survival (Feng et al. 2014b), may be induced shortly after meal consumption. This idea is further supported by the presence of cytokeratin proteins in the enrichment analysis, as some cytokeratins, such as K14 and K5, are expressed in subsets of taste progenitor cells (Okubo et al. 2009). Increased cell proliferation and differentiation may also explain the differential regulation of Hsp40 chaperone proteins, as these aid in protein folding and cellular transport (Cyr and Ramos 2015). However, further investigation into the role of these genes in TRC function is required to validate and support this hypothesis. Furthermore, we cannot exclude the possibility that some of the differentially expressed transcripts identified, may in fact be expressed in the contaminating non-taste tissue rather than in TRCs. It is likely that this would only be true for a small number of differentially expressed transcripts as the dataset is enriched for TRC RNA.
Because of the limited transcriptome data available for TRCs and the significance of receptors in TRC signaling, we used GO annotations to profile receptor expression in ad lib circumvallate taste buds. Various transcripts with G-protein coupled receptor functional annotations and/or receptor protein class annotations were present in the taste bud transcriptome. In addition to the traditional taste receptor genes, less studied receptors such as proposed fatty acid receptor Ffar4 (Cartoni et al. 2010), as well as the orphan G-protein coupled receptor GPR85 (Yuri et al. 2011), are also present. While mere detection of these receptors does not indicate functional significance, this serves to validate the presence of debated receptors such as Ffar4 and highlights receptors like the orphan GPR85 which may play a novel role in the taste bud.
In summary we present here, circumvallate taste bud transcriptomes for fed, fasted, and ad libitum fed WT C57BL6 mice. Using gene enrichment analysis we showed that differentially expressed genes are enriched in pathways associated with immunity, chemokines, serine protease inhibitors, cytokeratin structural proteins, and chaperone proteins. We also profiled the ad lib transcriptome based on receptor related GO terms, in order to demonstrate the use of this transcriptome dataset in the identification of novel receptors. Overall, the data presented here suggest that specific cellular pathways may be regulated in TRCs with respect to food intake. However, further study is needed to understand the implications of these finding in a broader biological context.
Conflict of Interests
At the time of study, author C. D. Dotson was employed at the University of Florida. He has since been hired at the Coca Cola company, and has provided only editorial guidance since his hire at Coca Cola. This change of employment in no way influenced funding of the study or the results obtained. All other authors have no conflict of interests to declare.
Funding
This work was supported by the National Institutes of Health [F31 DC015751 and R01 DC012819]
Acknowledgements
The authors would like to acknowledge the University of Florida’s Interdisciplinary Center for Biotechnology Research core facility for their assistance during the RNA-seq analysis.