These authors share first authorship
This article was submitted to Molecular Diagnostics and Therapeutics, a section of the journal Frontiers in Molecular Biosciences
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Ulcerative colitis (UC) is an inflammatory bowel disorder mainly affecting the large intestine. Most often, UC initiates from the rectum, affects the mucosal lining, and is limited to the innermost layers of the large intestine. Multiple pathogenic factors, including numerous susceptibility gene variants, environmental factors, changes in the gut microbiota, and a dysregulated immune response, are all associated with UC. Despite this recognition and the identification of apparently relevant factors, a complete understanding of UC pathogenesis still needs to be reached, and thus, treatment may not be optimal. An important reason for this unsatisfactory situation is the currently limited comprehension of the genuinely relevant components of UC immuno-pathogenesis. Thus, given the complex nature of UC, the study of animal models is crucial. A big challenge in using animal models for UC is whether the model’s genes and molecular pathways are analogous to humans. Although many coding and non-coding genes between humans and mice are highly similar (
Preclinical murine models have been used for
We, therefore, set out to characterize the transcriptomic landscape of the DSS-UC mouse model by performing transcriptomic profiling of PCGs, lncRNAs, and miRNAs from colon and whole blood using deep RNA sequencing. We identified differentially expressed lncRNAs and underscored their functional importance by annotating their expressed neighboring PCGs. Then we obtained the targets for the differentially regulated miRNAs and identified the biological processes that these targets are affecting. Further, we assessed the similarity between the experimental model and human UC. To this aim, several high-quality publicly available human UC colonic and blood transcriptomic studies, including our previous study, were selected, data obtained, processed, and cross-species UC transcriptomic landscape was compared. Our results highlight widespread differential regulation of coding and non-coding genes in the mouse and human UC models with marked similarities in both tissues, specifically in the colon. We also explored the underlying molecular pathways and networks through which these genes may contribute to disease development and progression. This study offers an excellent opportunity to investigate the role of novel and previously identified differentially regulated genes in UC. Moreover, cross-species high-resolution UC transcriptome comparison suggests new biomarkers and therapeutic targets and improves molecular phenotyping.
Ulcerative colitis was induced using dextran sulfate sodium salt (DSS) in C57BL/6 male mice. The control group was time-matched and received the same drinking water without DSS. The disease activity index (DAI) was recorded. RNA was extracted from the colon and whole blood, and total and small RNA were sequenced using the Illumina HiSeq 4000 system. Gene-level quantification corresponding to the total RNA-Seq data was obtained using Stringtie (
Functional enrichment analysis on the significantly differentially expressed (SDE) genes (FC > 2, padj ≤ 0.05) was performed using stringApp (
To suggest a functional role for SDE lncRNAs, we performed enrichment on their genomic neighbor genes situated within a span of 100 kb upstream and downstream of the lncRNA. For SDE miRNAs (padj ≤ 0.05), we retrieved a set of miRNA-target genes and performed enrichment on them (
StringApp was used to retrieve STRING networks (
To compare our mouse data with data from UC patients, we obtained and processed data from eight publicly available studies (
The expression of 19 selected SDE genes common in mouse and human colon and blood was validated by quantitative real-time PCR (qRT-PCR). Colon and blood cDNAs from 5 DSS-UC mice, five control mice, 10 UC patients, and six controls were used for validation. Detailed information on patient demographics and primer sequences used for qRT-PCR can be found in
Successful colitis induction in mice was confirmed by combined DAI score, measuring inflammatory markers, histological and micro positron emission tomography imaging analysis (
Numbers of SDE genes identified in colon and blood of DSS-UC mice versus control. (For PCGs and lncRNAs |log2FC| > 1, padj ≤ 0.05 and for miRNAs FC > 1.5, padj ≤ 0.05).
Colon | Blood | |
---|---|---|
PCGs | 2479 (▲1897, ▼582) | 518 (▲500, ▼18) |
lncRNAs | 282 (▲135, ▼147) | 44 (▲40, ▼4) |
miRNAs | 73 (▲39, ▼34) | 10 (▲3, ▼7) |
*Others | 300 (▲201, ▼99) | 54 (▲52, ▼2) |
*e.g. Pseudogenes, TEC, snoRNA, miscRNA, etc. Upregulation ▲, Downregulation ▼.
We identified 3,061 and 623 SDE genes from the total RNA-Seq and 73 and 10 SDE miRNAs from the small RNA-Seq data comparing UC mice colon and blood with control, respectively (
DSS-UC mouse transcriptional signature. Heatmaps of SDE PCGs, lncRNAs, and miRNAs, based on z-scores of normalized log counts for
From the SDE genes in the colon, we detected 116 genes with a human ortholog situated within 100 kb distance of the known Informatory bowel disease (IBD)-risk loci (
To investigate the roles of the SDE PCGs in the UC colon and blood, we performed a functional enrichment analysis (
For the SDE lncRNAs, we performed functional enrichment of their genomic neighbors expressed in our samples (
In the colon, for 59 out of 73 SDE miRNAs, we identified 2,330 target candidates expressed in our samples, and for blood, for eight out of 10 SDE miRNAs, we identified 1,641 targets expressed in our samples (
To identify genes commonly differentially regulated in the colon and blood of UC mice with importance in the disease phenotype, 284 genes were retrieved (
Commonly dysregulated SDE genes in UC mouse colon and blood.
Collectively, the common SDE genes define distinct inflammatory, immune system, and connective tissue-related gene expression signatures. We observed the enrichment of genes in processes related to the immune system, specifically the regulation of inflammatory response and leukocyte activation and cytokine-cytokine receptor interactions (
To obtain reliable sets of SDE genes in human UC and compare them with the mouse in both colon and blood, we identified and combined SDE genes from several public human datasets. For the colon, 2 RNA-Seq (
Public human UC datasets used in our study.
Accession | Description | Platform | Method | Number of samples | |
---|---|---|---|---|---|
Colon Total RNA-Seq | |||||
GSE109142 | Pediatric cohort, treatment-naive | RNA-Seq | Illumina TruSeq mRNA-Seq, 75 paired-End | 206 UC, 20 CO | |
GSE117993 | Pediatric cohort, treatment-naive | RNA-Seq | Illumina TruSeq mRNA-Seq, 75 Single-End | 43 UC, 55 CO | |
GSE59071 | Adult cohort | Microarray | Affymetrix GeneChip Human Gene 1.0 ST array | 74 UC, 11 CO (colon) | |
GSE67106 | Adult cohort | Microarray | Agilent Custom 8 × 60K format lncRNA expression microarray | 15 UC, 9 CO (colon, rectum) | |
Colon Small RNA-Seq | |||||
GSE89667 | Adult cohort | Small RNA-Seq | Illumina HiSeq 2500 | 10 UC, 18 CO (diverticular disease) | |
GSE48957 | Adult cohort | Microarray | Affymetrix Multispecies miRNA-2 Array | 10 UC, 10 CO | |
Blood Total RNA-Seq | |||||
GSE112057 | Pediatric cohort | RNA-Seq | Illumina HiSeq 2000, 100 paired-End | 15 UC, 12 CO | |
PRJEB28822-C | Pediatric cohort | RNA-Seq | Ion AmpliSeq Transcriptome | 34 UC, 35 CO | |
PRJEB28822-A | Adult cohort | RNA-Seq | Ion AmpliSeq Transcriptome | 37 UC, 32 CO |
The combined sets of human SDE PCGs and lncRNAs contained ∼11,000 genes in the colon and ∼2,000 genes in the blood (padj ≤ 0.05 in at least two datasets and consistently differentially regulated across datasets). We observed approximately equal proportions of up- and downregulated PCGs in the colon and a predominant (∼75%) upregulation in blood. However, in both colon and blood, lncRNAs were mainly (∼60%) downregulated. Furthermore, ∼2,500 genes in the colon and ∼130 genes in the blood were inconsistently differentially regulated between the datasets and thus not included in the combined SDE gene sets. By combining the two public small RNA-Seq datasets in the colon (
Numbers of SDE genes in human colon and blood combined datasets. (For all padj ≤ 0.05 in min 2 datasets).
Colon | Blood | |
---|---|---|
PCGs | ▲4658 ▼4909 ✗1531 | ▲1382 ▼511 ✗177 |
lncRNAs | ▲454 ▼673 ✗812 | ▲12 ▼16 ✗6 |
miRNAs | ▲14 ▼19 | NA |
*Others | ▲450 ▼181 ✗204 | ▲16 ▼16 ✗6 |
*e.g. Pseudogenes, TEC, snoRNA, miscRNA, etc. Upregulation ▲, Downregulation ▼, Inconsistent ✗.
Using one-to-one and one-to-many orthology assignments, we compared the combined sets of SDE genes in humans with the sets of SDE genes in mouse colon and blood (
Numbers of overlapping genes between UC mouse and UC human colon and blood (For all padj ≤ 0.05). ⇈: Upregulation in both mouse and human, ⇊: Downregulation in both mouse and human, ⇅: Opposite directions between mouse and human
Colon | Blood | |||||
---|---|---|---|---|---|---|
Human SDE genes with mouse orthologs | 8446 (+ 116 miRNAs) | 1723 | ||||
Mouse SDE genes with human orthologs | 3066 (+ 70 miRNAs) | 512 | ||||
Common SDE gens in Human and Mouse | 1937 (⇈1082, ⇊360, ⇅495) | 159 (⇈154, ⇊3, ⇅2) | ||||
⇈ | ⇊ | ⇅ | ⇈ | ⇊ | ⇅ | |
PCGs | 1061 | 345 | 489 | 154 | 3 | 2 |
lncRNAs | 5 | 2 | 1 | 0 | 0 | 0 |
miRNAs | 9 | 12 | 4 | NA | NA | NA |
*Others | 7 | 1 | 1 | 0 | 0 | 0 |
*(e.g. Pseudogenes, TEC, snoRNA, miscRNA, etc).
Though ∼25% of common SDE genes in the colon are contra-regulated, 356 genes with upregulation in mice and downregulation in humans showed to be mainly involved in lipid metabolism, ion transport, regulation of localization, and trans-synaptic signaling. This could reflect species-specific differences in biological processes’ gene regulation. No enrichment was found for the 133 genes, downregulated in mice and upregulated in humans.
Overall, 51 genes were identified as commonly SDE between the colon and blood of humans and mice compared with the healthy controls (
Comparison between SDE genes of human and mouse UC.
From the 51 genes common between humans and mice in both colon and blood, 19 were selected for qPCR validation in a separate human UC cohort and the UC mice. Selected genes were either novel in a UC phenotype setting or not previously reported as therapeutic targets or biomarkers. IPA and Pharos were used to assess the therapeutic and biomarker applications of the 51 genes (
Over the last few years, several transcriptional profiling studies have provided evidence regarding the differential regulation of PCGs, lncRNAs, and miRNAs in UC (
Here, we performed colon and blood transcriptional profiling of UC mice for coding and non-coding RNAs and investigated the affected biological pathways. Our analyses showed that most differentially regulated genes are upregulated and involved in immunological and inflammatory responses and connective tissue, organ abnormality, and injuries. This points to the pathological nature of UC and the importance of inflammatory processes in destroying epithelium and aggravation of the disease. Moreover, alterations and mainly upregulation of genes that suppress inflammation were detected in the colon and blood. Genes, which demonstrated significant downregulation, were mainly genes with basic metabolic functions involved in cellular growth and proliferation, lipid, vitamin, amino acid, and mineral metabolism. Most of the genes common between the colon and blood of UC mice are regulated in the same direction (upregulated) and are related to immune or inflammatory processes. Among these genes,
Here we investigated the transcriptional landscape of both lncRNAs and miRNAs in the colon and blood of the DSS-UC mice model and identified widespread differential regulation of non-coding RNAs. Numerous SDE lncRNAs and miRNAs were identified in both tissues, many of which have not been reported previously. Several SDE lncRNAs in the colon have been widely studied previously, including
Among the SDE miRNAs, several IBD-related miRNAs, including miR-30, miR-223, miR-21, and miR-142, were upregulated, and miR-192 was downregulated in the colon, while miR-223, miR-16, and miR-126 were upregulated in blood. miR-223-3p is the only miR upregulated in both colon and blood of the UC mice. It is upregulated in neutrophils and monocytes and acts as a controller of NLRP3 inflammasome activity, regulating the intestine inflammatory process by affecting IL-1β production (
By comparing our UC mouse to human UC data, we identified similar widespread directional differential regulation of genes. The overlap of a substantial number of differentially regulated genes in mouse and human UC suggests common fundamental mechanisms responsible for disease onset and progression. To date, few colon studies have compared the transcriptional changes in UC mouse models and human UC (
Although we recognize that separate analysis of common genes in the colon and blood may provide additional information, we focus on the genes that were common between both tissues in the mouse and human, thus, may represent the universal markers of UC. Ascribed to the pathophysiology of the UC and the fact that for diagnosis purposes, it is difficult to access the colon in comparison to blood, identification of biomarker candidates showing a similar differential expression in the colon and blood helps us diagnose UC faster and without the need for invasive assessments. Moreover, conserved same-direction differentially expressed genes in two distinct species might serve as a more reliable disease-specific biomarker compared with genes that are not expressed in one species or expressed differently in different species under the same pathogenic condition.
Overall, 51 common PCGs in mouse-human colon and blood with strong inflammatory and immunological profiles showed to be consistently differentially regulated. Several of these genes, including
No long non-coding RNAs made it to the 51 final gene list. The major reason is that only a few lncRNAs are assigned orthology relationships across species in general (e.g., only 131 of all mouse lncRNAs are assigned a human ortholog based on the HCOP database). However, we have identified seven lncRNAs that were SDE in the colon in both humans and mice. Apart from H19, little is known about the other six lncRNAs in IBD, which clearly shows a need for detailed analysis. We also identified 21 miRNAs SDE in the colon in humans and mice. Among them, miR-146a-5p, miR -155-5p, miR-192-5p, miR-194-5p, miR-196b-5p, miR-200c-3p, miR-223-3p, and miR-223-5p were frequently shown to be differentially regulated in IBD previously, which makes them a promising candidate for further analysis.
Given the complexity of the whole colon and blood cell types, one could isolate different cell populations and perform single-cell sequencing to distinguish pathogenic mechanisms with higher resolution. One of the limitations of our study is that whole colon and blood gene expression profiling was performed. Albeit, the primary rationale was to preserve the natural state of the disease as much as possible. In addition, the disadvantages of isolation of different cell types are the technical fractionation procedures and the time duration from tissue collection to sample processing, potentially affecting gene expression statuses specifically for non-coding RNAs in cells.
In conclusion, high-throughput transcriptome analysis provides a unique tool for discovering new therapeutic and diagnostic targets by identifying relationships between gene expression and disease phenotype. Here, we showed that a one-to-one comparison of the transcriptome of the DSS-UC mouse model to human UC could provide novel disease pathophysiology information by identifying genes and pathways with essential contributions to UC. Our data suggest that prospective therapeutic interventions and diagnostic applications should target multiple major gene regulators involved in UC pathogenesis and propagations and combine several genes for valid biomarker applications. Moreover, targeting genes with conserved functional roles in the disease pathogenesis may offer a reliable UC treatment approach compared to targets with different functional roles in different tissues and organisms.
The data underlying this article are available in the article and its online
The studies involving human participants were reviewed and approved by the Regional Ethical Committee (H-4-2012-030). The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Danish Animal Experiments Inspectorate, Ministry of Food, Agriculture, and Fisheries, Denmark, at Odense University Hospital under the 2015-15-0201-00730 license number.
RY designed the study, established the mouse model, performed the experiments, interpreted the results, and wrote the manuscript (Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Resources, Writing—Original Draft, Visualization, Project administration). OP processed and analyzed the RNA-Seq and public datasets and contributed to writing the manuscript (Software, Formal analysis, Data Curation, Writing—Original Draft, Visualization). ND contributed to analyzing the data and to writing the manuscript (Software, Formal analysis, Writing—Original Draft, Visualization). CA provided support in analyzing the data and revised the manuscript (Formal analysis). BP and UH provided tools and help for establishing the mouse model (Methodology, Resources). CS performed the qPCRs (Validation). AM prepared the human samples and revised the manuscript (Resources). TL provided essential insights and revised the manuscript (Conceptualization). CB-B and MV prepared the human samples and revised the manuscript (Resources). JG and LJ provided essential tools and insights in designing the experiment supervised the study and revised the manuscript (Conceptualization, Supervision, Funding acquisition). FP contributed to designing the experiment, provided essential tools and insights, supervised the study, and revised the manuscript (Conceptualization, Supervision, Project administration, Funding acquisition). All authors read, reviewed, and approved the final manuscript.
This work was supported by the Independent Danish Research Foundation, Technology, and Production, grants 4005-00443 and 8020-00300B, the Novo Nordisk Foundation, grant NNF14CC0001, Lundbeck Foundation, grant R303-2018-3148, and the Sehested Hansen foundation.
We want to thank Charlotte Aaberg Poulsen from the University of Southern Denmark for providing the animal study license for the UC mouse model establishment. We also would like to thank Professor Poul Flemming Høilund-Carlsen and Christina Baun (Research radiographer) from Clinical Physiology and Nuclear Medicine at Odense University Hospital for performing microPET scanning of the UC mouse models. Moreover, thanks to Dr. Giulia I. Corsi from the Center for non-coding RNA in Technology and Health for help with the visualization of enrichment analysis.
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.
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
The Supplementary Material for this article can be found online at:
Primers for qPCR and Patients info.
SDEGs in colon and blood.
List of IBD loci genes.
Functional enrichment for SDE PCGs in colon and blood.
lncRNA-neighbors and functional enrichment.
miRNA-targets and functional enrichment.
Common SDEGs in mouse blood and colon.
Confident sets SDEGs in Human UC.
Colon common SDEGs in mouse and human.
Blood common SDEGs in mouse and human.
Final 51 SDEGs, centrality ranking and applications.
Supplementary methods and figures.