• Log in with Facebook Log in with Twitter Log In with Google      Sign In    
  • Create Account
  LongeCity
              Advocacy & Research for Unlimited Lifespans

Photo

Ageing-associated changes in the expression of lncRNAs in human tissues reflect transcriptional modulation in ageing ...

aging transcriptomics lncrna ncrna

  • Please log in to reply
No replies to this topic

#1 Engadin

  • Guest
  • 198 posts
  • 580
  • Location:Madrid
  • NO

Posted 13 November 2019 - 04:50 PM


.

 

 

C O M P L E T E   T I T L E :  Ageing-associated changes in the expression of lncRNAs in human tissues reflect a transcriptional modulation in ageing pathways.

 

 

 

F U L L   T E X T   S O U R C E :   ScienceDirect

 

 

 

 

Highlights

 
  •Ageing-associated changes in the expression of lncRNAs are highly tissue-specific
 
  •Ageing-associated lncRNAs are associated with similar functions across tissues
 
  •Ageing-associated changes in lncRNAs mirror those observed for protein coding genes
 
 
Abstract
 
Ageing-associated changes in the protein coding transcriptome have been extensively characterised, but less attention has been paid to the non-coding portion of the human genome, especially to long non-coding RNAs (lncRNAs). Only a minority of known lncRNAs have been functionally characterised; however, a handful of these lncRNAs have already been linked to ageing-associated processes. To gain more information on the effects of ageing on lncRNAs, we identified from GTEx data lncRNAs that show ageing-associated expression patterns (age-lncRNAs) in 29 human tissues in 20-79-year-old individuals.
 
The age-lncRNAs identified were highly tissue-specific, but the protein coding genes co-expressed with the age-lncRNAs and the functional categories associated with the age-lncRNAs showed significant overlap across tissues. Functions associated with the age-lncRNAs, including immune system processes and transcription, were similar to what has previously been reported for protein coding genes with ageing-associated expression pattern. As the tissue-specific age-lncRNAs were associated with shared functions across tissues, they may reflect the tissue-specific fine-tuning of the common ageing-associated processes. The present study can be utilised as a resource when selecting and prioritising lncRNAs for further functional analyses.
 
 
1. Introduction
 
Despite the small proportion of protein coding sequences (3%) in the human genome, the great majority of the genome is pervasively transcribed, producing a versatile pool of non-coding RNA molecules (ncRNAs) (ENCODE, 2012; Derrien et al., 2012; Djebali et al., 2012; Iyer et al., 2015). ncRNAs over 200 nt in length are classified as long non-coding RNAs (lncRNAs), and their number is at least comparable to that of protein coding genes and may be as high as 200 000 (Iyer et al., 2015; Xu et al., 2017; Hon et al., 2017). The processing of lncRNA molecules shares majority of features with mRNAs (Samudyata and Bonetti, 2018; Jarroux et al., 2017), but lncRNAs are typically expressed at a lower level as compared to protein coding genes. lncRNAs are very tissue- or lineage-specific and typically show highly specific spatio-temporal expression patterns (Djebali et al., 2012; Cabili et al., 2011; Ward et al., 2015; Li et al., 2015; Hon et al., 2017).
 
As everything over 200 nt in length and without protein coding potential is classified as a lncRNA, lncRNAs are a very heterogenous group. Only a minority of the identified lncRNAs have been functionally characterised, but the ones that have been characterised have been shown for example to regulate gene expression, post-transcriptional maturation, translation and epigenetics. Mechanisms by which lncRNAs bring about their function include interaction with other RNA species or DNA, scaffolding of subcellular domains or complexes and regulation of protein activity or abundance. In addition to the lncRNA transcript itself being functional, there is evidence showing that the act of transcription from the lncRNA locus can affect nuclear structure, epigenetic landscape or the expression of nearby genes (Ulitsky & Bartel, 2013; Kaikkonen & Adelman, 2018; Schmitz et al., 2016; Cech & Steitz, 2014; Yang et al., 2014; Melé and Rinn, 2016). lncRNAs have been shown to play a role for example in cellular pluripotency, cell differentiation, lineage specification, maintenance of cell identity, developmental patterning, dosage compensation and imprinting as well as cell migration (Flynn & Chang, 2014; Ransohoff et al., 2018).
 
Previously lncRNAs have been shown to be associated with processes important for various ageing-associated diseases, including cancer (He et al., 2019; Kondo et al., 2017), cardiovascular diseases (Bink et al., 2019, Zhou et al., 2016), type II diabetes (He et al., 2017) and neurodegenerative diseases such as Alzheimers disease (Pereira Fernandes et al., 2018; Idda et al., 2018). In model organisms, changes in the expression of lncRNAs with ageing have been reported (Wood et al., 2013). In humans, ageing-associated changes in the protein coding transcriptome have been extensively studied (Gomez-Verjan et al., 2018; Frenk & Houseley, 2018), but the ageing-associated changes in the non-coding transcriptome remain to be extensively studied. In the present study, we characterised the ageing-associated changes in lncRNA expression in 29 healthy human tissues between the ages 20 and 79 years, using the data available in GTEx database (gtexportal.org).
 
 
2. Materials & Methods
 
2.1. Identifying ageing-associated lncRNAs
 
The data used for the analyses described in this manuscript were obtained from GTEx project, dbGaP accession number phs000424.v7. The GTEx data set consists of tissue-specific gene expression data from non-diseased tissues (gtexportal.org). Only samples from subject who died in a ventilator (Hardy scale 0) were included. We decided to restrict our analysis to subjects with the same death circumstance, as we found that death circumstance has an effect on gene expression profiles that could obscure true ageing-associated effects (described in more detail in Supplementary file 1). The sample number for each tissue studied varied from 6 to 733 (median 154) (Table 1).
 
 
Table 1.jpg
 
 
Table 1Details of sample counts, total number of expressed genes and lncRNAs as well as numbers of identified ageing associated genes and age-lncRNAs. Tissues with >10 age-lncRNAs were included in further analyses (shown here in bold).
 
 
For each tissue, we identified differentially expressed genes with age using the following linear regression model:
 
Yij = αAgei + βSexi +ɛij
 
where Yij is the expression level of gene j in sample i, Agei denotes the age of sample i, Sexi denotes the sex of sample i and ɛij denotes the error term. The dataset provided the age of each subject as an age range (20-29, 30-39, 40-49, 50-59, 60-69 and 70-79); we approximated the age of each sample to be 25, 35, 45, 55, 65 and 75, respectively. To remove genes with low expression values, we excluded genes with expression less than 1 count per million (cpm) in more than 30 percent of samples. Raw read counts were normalized using TMM normalization and were voom transformed to remove heteroscedasticity from the count data. The linear model for each gene was generated by using the limma package in R (version 3.36.5). Genes were considered significantly associated with age with empirical Bayes moderated t-statistics and their associated adjust P-value (Benjamini-Hochberg method) < 0.05 and absolute fold change across 50 years (from 25 to 75 years old) > 1.5.
 
Biotypes of the ageing-associated genes were identified using the R package biomaRt (Durinck et al. 2009), based on Ensembl release 92 (April 2018). Protein coding genes as well as immunoglobulin genes were removed from the results, and the remaining non-coding genes were used in the following analyses. The total number of ageing-associated genes ranged from 0 to 2042 across tissues (0 to 10.8% of all expressed genes) and the number of age-lncRNAs ranged from 0 to 346 (0 to 33.3% of all ageing associated genes), for each tissue the numbers of ageing associated genes and ageing-associated lncRNAs (age-lncRNAs) are shown in Table 1.
 
2.2. Tissue specificity index (Tau)
 
For all genes in the GTEx data, a τ tissue specificity index was calculated. The τ index is an indicator of how specifically or broadly expressed a gene is, with a τ of 1 indicating expression specific to only one tissue, and a τ of 0 indicating equal expression across all tissues (Yanai et al., 2005). The τ index for a given gene can be calculated using the following equation:
 
Formula.jpg
 
where N is the number of tissues being studied and xi is the expression profile component for a given tissue, normalised by the maximal component value for that gene (i.e. the expression of that gene in the tissue it is most highly expressed in).
 
 
2.3. Genes co-expressed with age-lncRNAs
 
Genes co-expressed with age-lncRNAs were identified with GeneFriends RNA-seq (v3.1) (van Dam et al., 2015). This co-expression analysis describes which genes tend to be activated along with the age-lncRNAs of interest, which can be thus assumed to be under similar transcriptional regulation and to participate in similar functions. It does not, however, suggest regulatory relationship between the lncRNAs themselves and genes co-expressed with them. For each tissue, and separately for up- and down-regulated, age-lncRNAs were used as input (for numbers of age-lncRNAs, see Table 1) (data downloaded on 12.3.2019). The resulting list of co-expressed genes was trimmed to include only genes expressed in the tissue in question in GTEx data (median expression TPM > 1). The top 5% of the co-expressed genes were used in further analyses.
 
The overlap of lists of co-expressed genes was analysed using the R package OrderedLists (Yang et al., 2018), which yields a similarity score of the two lists, giving emphasis to genes in the top ranks. For the comparisons, an empirical p-value is calculated based on random shuffling of the original list. Tissues were compared to each other pairwise, with two.sided set to “FALSE” in order to only compare the top members of the list. 10000 permutations were performed to estimate empirical p-values, with p-values < 0.05 considered significant.
 
2.4. Functional enrichment and semantic similarity of GO terms
 
Enrichment of GO terms was analysed with R package topGO (Alexa & Rahnenfuhrer, 2018), which takes into account the hierarchical structure of GO terms. For each tissue, the enrichment was analysed for the genes co-expressed with the age-lncRNAs using the “weight01” method and using BH-corrected p-value of 0.05 as a cut-off.
 
The semantic similarity of lists of enriched GO terms between tissues was analysed with G-SESAME (James et al., 2007). This method takes into account the ancestors of each GO term. For the analysis, default settings were used, semantic contribution factors for “is_a” and “part_of” relationships were set to 0.8 and 0.6, respectively. Sematic similarity was calculated for all tissue pairs, separately for up- and down-regulated, and excluding tissues with no or only one statistically significant GO term. Lists of GO terms were considered to be similar with similarity score >0.5.
 
Enrichment of KEGG terms was analysed with FunnMapOne (Scala et al., 2019), which was also used to visualise the enrichment of KEGG terms. For the enrichment, terms with p-value < 0.05 were considered significant. For the visualisation and clustering of tissues, default settings of FunMappOne were used.
 
3. Results
 
3.1. Ageing-associated changes in the expression of lncRNAs
 
Across 29 analysed tissues, we identified lncRNAs with ageing-associated expression patterns in 22 tissues. The number of lncRNAs with an ageing-associated expression pattern (age-lncRNAs) varied from 2 to 246 per tissue (median 38.5, 0.05 to 7.6% of all expressed lncRNAs, see Table 1). In total there were 612 age-lncRNAs up-regulated and 652 age-lncRNAs down-regulated across tissues. Of the up-regulated age-lncRNAs, 29 showed an ageing-associated expression pattern in three or more tissues and the great majority, 490 age-lncRNAs, were ageing-associated in only one tissue. For down-regulated, these numbers are 16 and 570, respectively. Age-lncRNAs, that showed an association with ageing in three or more tissues are hereafter referred as multi-tissue age-lncRNAs. In addition, there were 112 age-lncRNAs that showed a varied expression with ageing and that were up-regulated in one or more tissues and down-regulated in another tissue or vice versa. All age-lncRNAs are presented in Supplementary Table 1. There were 16 tissues with 10 or more age-lncRNAs, and these tissues are used in further analyses (adipose tissue, adrenal gland, blood vessel, brain, cervix uteri, colon, esophagus, heart, lung, muscle, nerve, ovary, prostate, salivary gland, stomach and uterus).
 
We compared the identified age-lncRNAs to lncRNAs previously reported to be ageing associated in individual tissues, in peripheral blood mononuclear cells, (PBMCs, Noren Hooten & Evans, 2019), tendon (Peffers et al., 2015) and brain subependymal zone (SEZ, Barry et al., 2015). GTEx does not contain expression data from PBMCs or tendon and comparing the ageing-associated lncRNAs identified in these tissues to all age-lncRNAs identified in the present study revealed only a modest overlap. There were 1938 ageing-associated lncRNAs in PBMCs, of which 117 were identified in the present study and 62 lncRNAs in tendon, of which 12 were identified in the present study. While GTEx contains several brain regions, SEZ is not included. Of the 6 ageing-associated lncRNAs reported in SEZ, none are age-lncRNAs in the brain, but 3 were identified as age-lncRNAs in other tissues.
 
 
 
 
.../...
 
 
 
 
 
 

  • like x 1





Also tagged with one or more of these keywords: aging, transcriptomics, lncrna, ncrna

1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users