Expanding the concept of housekeeping genes: Examining the influence of retroelements on gene expression stability during cellular differentiation

doi:10.21203/rs.3.rs-4484491/v1

Download PDF

Article

Expanding the concept of housekeeping genes: Examining the influence of retroelements on gene expression stability during cellular differentiation

https://doi.org/10.21203/rs.3.rs-4484491/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Housekeeping genes (HKGs), essential for cellular differentiation, are traditionally thought to be uniformly expressed. However, recent analyses show variable expression across tissues, challenging this view. Identifying HKGs based on stable expression necessitates precise, repeated experiments. The limited number of transcription factors during differentiation suggests HKGs stabilise phenotypes via compensatory gene regulation. We examined retroelements, comprising approximately 40% of the human genome, for their influence on gene expression stability near transcriptional start sites (TSSs). By mapping CpG islands and retroelements within 10 kb of TSSs, we analysed 56 RNA sequencing datasets. CpG-island genes near Alu elements exhibited low expression variability, especially when the Alu elements were closer to the TSSs. Conversely, non-CpG island genes near L1 and LTR elements showed high variability, indicating distinct regulatory roles. Despite the constant number of Alu-adjacent CpG-island genes across tissues, their expression was most highly expressed in embryonic stem cells and declined in most differentiated tissues, except for an increase in the ovary and testis. Alu elements near CpG islands help stabilise phenotypes through compensatory downregulation of HKGs. These results expand the HKG concept to include genes maintaining stable phenotypes during differentiation, aiding in differentiation stage identification.

In multicellular organisms, housekeeping genes (HKGs) are vital for maintaining fundamental cellular functions across diverse tissues^1–4. Traditionally, HKGs were regarded as ubiquitously expressed and contrasted with tissue-specific genes. However, recent advances in quantitative analysis, such as RNA sequencing free from background noise, have revealed that most genes display varying expression levels across tissues^1,5. This finding shifts the focus towards the stability and consistency of gene expression to accurately determine HKGs¹. Identifying HKGs based on stable expression necessitates sophisticated methods such as high-throughput data analysis and robust statistical models to account for inherent variability^6–8. Research has revealed key genetic features of HKGs, including the presence of CpG islands, distinctive retroelement patterns and nucleosome formation tendencies^9–15. These characteristics underscore the complex regulation of HKGs and highlight the challenges in their comprehensive classification. Furthermore, the regulation of HKGs by specific transcription factors remains poorly understood^16,17. Therefore, a refined HKG classification system is imperative to elucidate the epigenetic regulation and effects of these genes on phenotypic manifestations.

Retroelements, which constitute approximately 40% of the human genome, play crucial roles in shaping genomic architecture^18–20. This contrasts sharply with the mere 1% of the genome being composed of protein-coding sequences, underscoring the influence of non-coding sequences¹⁸. Known as parasitic genetic elements, retroelements such as Alu, L1 and LTR exhibit unique distributions within the genome. Alu elements, specific to primates, predominantly occupy gene-rich regions and are often associated with HKGs, whereas L1 elements, found in mammals, typically reside in gene-poor areas^21,22. LTR elements, present in various eukaryotes, integrate into diverse genomic locations²³. The specific locations of these elements suggest potential adaptive advantages affecting gene function and regulation. Although their influence on gene expression across evolution and development has been recognized^22,24,25, the mechanisms underlying these advantages remain unexplored. Understanding how retroelement distribution impacts gene regulation is key to revealing their roles in cellular functionality and developmental dynamics.

During stem cell differentiation, when cells acquire specific phenotypes essential for tissue specialisation, HKGs play a crucial role in ensuring the functionality of emerging cell types². During this phase, the limited reservoir of transcription factors within the nucleus regulates the activation of tissue-specific genes, necessitating a subtle downregulation of several HKGs to maintain a balanced transcript dose²⁶. This shift to tissue-specific transcripts is vital for preserving acquired phenotypes in progeny cells. DNA methylation, which changes with each cell division, plays a key role in epigenetic inheritance during cell differentiation^27,28. Although differentially methylated regions are associated with tissue differentiation, the specific genetic elements linked to these methylation shifts remain ambiguous^29–31. Recent studies have explored age-related methylation changes in several genes using small samples to address epigenetic diversity, revealing that these changes are influenced by the type and proximity of adjacent retroelements ^32–34. Given their proximity to several HKGs, Alu elements not only help manage the rapid downregulation necessary for phenotype acquisition but also significantly affect epigenetic modifications during differentiation^35–37. Investigating the positioning of retroelements relative to genes is expected to provide new insights into the regulatory mechanisms of gene expression.

We designed this study to elucidate the complex interplay between retroelement distribution and the expression of HKGs. By utilising public quantitative RNA sequencing libraries, we sought to detect subtle variations in gene expression across different tissues. By analysing the presence of CpG islands and the retroelement distribution near transcriptional start sites (TSSs), we categorised genes to identify concurrent expression changes that contribute to phenotypic stabilisation. Investigating the influence of retroelements on the spatial organisation around HKGs, we expect to uncover specific genetic markers that predict gene expression patterns, thereby allowing us to infer the stage of tissue differentiation based on the degree of phenotypic stabilisation. The approach we propose here not only facilitates a comparison of gene expression across tissues but also helps establish a refined set of criteria for HKG classification.

CpG islands as key elements in HKG identification

We comprehensively analysed gene datasets from the hg38 human genome assembly. We extracted microRNA and noncoding RNA transcripts, selecting the longest transcript variants for genes undergoing alternative splicing to best represent gene function. Our dataset included only transcripts documented at least once across 54 different tissue libraries in the RNA sequencing expression library, yielding a final dataset of 20,869 genes. Among them, 12,590 genes (60.3%) had CpG islands within 200 bp of their TSSs, and 8,279 genes (39.7%) did not (Supplementary Data 1).

We assessed gene expression stability by calculating the tissue specificity index (TSI) and the coefficient of variation (CV)^6–8. A strong correlation between the TSI and CV (ρ = 0.977, P < 0.001) demonstrated their effectiveness as indicators of ubiquitous gene expression. CpG-island genes exhibited significantly lower TSI and CV values than non-CpG island genes (TSI median: 0.67 vs. 1.28, P < 0.001; CV median: 1.19 vs 3.18, P < 0.001), underscoring the potential role of CpG islands as indicators of ubiquitous gene expression (Fig. 1).

Influence of gene-adjacent retroelements on HKG expression

We analysed the influence of gene-adjacent retroelements on gene expression patterns and observed distinct distributions of Alu, L1 and LTR retroelements around the TSSs of HKGs. Alu elements were predominant in the upstream regions of TSSs (66.1% ), with L1 and LTR elements accounting for 18.3% and 15.6%, respectively. A similar pattern was noted in the downstream regions: Alu (66.6%), L1 (18.4%) and LTR (15%). Alu elements were associated with a high incidence of CpG islands upstream (71.5% vs. 57.8%) and downstream (72.0% vs. 58.4%) compared to genes without CpG islands (Supplementary Fig. 1). In contrast, L1 and LTR elements showed fewer CpG islands upstream (L1: 15.2% vs. 23.1%; LTR: 13.3% vs. 19.1%) and downstream (L1: 14.9% vs. 23.7%; LTR: 13.0% vs. 17.9%).

Employing the Kruskal‒Wallis test, significant variations were found in the TSI and CV among genes adjacent to the Alu, L1 and LTR elements in both upstream and downstream regions for the CpG-island genes and non-CpG island genes (P < 0.001 for all analyses, Supplementary Fig. 2). Post hoc analyses revealed that CpG-island genes adjacent to Alu elements had lower TSI and CV values than those adjacent to L1 and LTR elements, both upstream and downstream. This pattern was consistent for genes without CpG islands.

A detailed 1-kb interval analysis within a 10-kb radius from the TSSs revealed that 57.7% of the gene-adjacent retroelements were located within the 3-kb range (Fig. 2A, B). In the CpG-island genes, the nearest Alu, L1 and LTR elements were most frequently found within the 1–2-kb range, both upstream and downstream. For non-CpG island genes, these elements were located primarily within 1-kb of the TSSs. Statistical analysis revealed significant differences in the TSI and CV metrics across different distances for CpG-island genes near Alu, L1 and LTR elements (P < 0.001 for all; Fig. 2C, D, Supplementary Fig. 3A, B). The TSI and CV values decreased as the retroelements approached the TSSs. For non-CpG island genes, significant differences in the TSI and CV values (TSI, P = 0.012; CV, P = 0.009) were detected at various distances upstream of Alu elements. No significant differences were detected near L1 or LTR elements, except downstream of L1 (TSI, P = 0.007; CV, P = 0.015).

Combined effects of upstream and downstream retroelements on HKG expression

We systematically analysed the combined effects of retroelements in the regions immediately upstream and downstream of TSSs and identified nine combinations of Alu, L1 and LTR elements in both CpG-island and non-CpG island genes. The Alu-Alu combination was most prevalent, indicating its potentially significant role in genomic regulation. A comprehensive Kruskal‒Wallis test revealed significant differences in TSI and CV across these combinations in CpG-island and non-CpG island genes (Fig. 3, Supplementary Fig. 4). Post hoc analyses categorised these combinations into three groups: those with Alu elements both upstream and downstream (Alu-Alu), those with at least one Alu element (Alu mixed: Alu-L1, Alu-LTR, L1-Alu, LTR-Alu) and those with only L1 or LTR elements (L1/LTR only: L1-L1, L1-LTR, LTR-L1, LTR-LTR). TSI and CV values increased across the groups in that order.

We expanded our study to explore the combined effects of upstream and downstream retroelement patterns at 1-kb intervals from TSSs via a heatmap (Fig. 4, Supplementary Fig. 5). Among the CpG-island genes, the Alu-Alu combination was associated with low TSI and CV values, particularly closer to the TSSs. Similarly, the Alu mixed group also showed decreasing TSI and CV values as the distance to the TSS decreased. The L1/LTR only group exhibited a similar trend of decreasing TSI and CV values with proximity to the TSSs, but the effect was less pronounced. Among the non-CpG island genes, all groups displayed high TSI and CV values regardless of their proximity to the TSSs.

Gene grouping based on the presence of CpG islands and the proximity of retroelements

To better understand variations in housekeeping properties based on retroelement combinations in CpG-island genes, we compared different distance intervals. Due to the low number of genes in each 1-kb interval, we conducted separate analyses for the upstream and downstream regions (Supplementary Fig. 6). A subsequent post hoc analysis of TSI values using the Kruskal‒Wallis test further classified the Alu-Alu group into three categories based on their proximity to the TSSs: within 2 kb upstream, within 2–7 kb upstream and more than 7 kb upstream. Similar distinctions were made for downstream distances: within 3 kb, within 3–5 kb and more than 5 kb. The Alu mixed group was segmented into three categories: within 2 kb upstream, within 2–5 kb upstream and more than 5 kb upstream, with downstream distances classified as within 2 kb, within 2–4 kb and more than 4 kb, respectively. Differences in gene expression stability in the L1/LTR only group were noted at distances within 2 kb upstream and more than 2 kb upstream, with downstream distances within 4 kb and more than 4 kb from the TSSs. Non-CpG island genes did not show significant differences based on distance in the three groups.

From these statistical results, we classified all genes into 11 distinct groups. For the CpG-island genes, we established three main groups based on their proximity to the TSSs: Alu-Alu, Alu mixed and L1/LTR only. The Alu-Alu group was further classified into three subgroups: within 2 kb upstream or 3 kb downstream, within 2–7 kb upstream or 3–5 kb downstream and more than 7 kb upstream or more than 5 kb downstream. The Alu mixed group was categorised into three subgroups: within 2 kb upstream or 2 kb downstream, within 2–5 kb upstream or 2–4 kb downstream and more than 5 kb upstream or more than 4 kb downstream. The L1/LTR only group included two subgroups: within 2 kb upstream or 4 kb downstream, and beyond 2 kb upstream or beyond 4 kb downstream. For non-CpG island genes, for which proximity to the TSSs was not a distinguishing factor, we designated a single category for each group.

Compensatory downregulation of Alu-adjacent CpG-island genes in differentiated tissues

To enhance our analysis of the initiation points for differentiation processes, we expanded our gene expression dataset to include human embryonic stem cells (both male and female) and analysed 56 expression libraries for the 11 defined gene groups, examining both the number of expressed genes and the aggregate of expressed transcripts (Supplementary Fig. 7). The number of expressed genes showed minimal fluctuation, averaging 17,938 genes (ranging from 16,730 to 20,074, a 16.7% variation), but the total number of transcripts varied substantially, averaging 545,679 and ranging from 263,394 to 797,908, a 67% difference. When comparing normalised proportion values across tissues, we observed smaller variations in the proportional gene expression counts than in the proportional transcript quantification (Fig. 5). The variations in the proportional gene expression counts among the 8 CpG-island gene groups ranged from 0.2 to 4.0%, and among the 3 non-CpG island gene groups from 2.2–4.0%. More pronounced differences were detected in the proportional transcript quantification, with CpG-island-associated genes varying from 2.9 to 29.4%, and non-CpG island genes exhibiting a much broader range from 18.7 to 50.3%.

We observed distinct patterns of differential tissue expression (Fig. 5). For instance, CpG-island genes consistently exhibited the highest transcript volumes in embryonic stem cells (both male and female), averaging 85.3%. This trend markedly decreased in differentiated tissues, dropping to 47.7% in the liver, 44.0% in whole blood and 24.5% in the pancreas. In embryonic stem cells, CpG-island genes in the Alu 2 kb-Alu 3 kb group displayed the highest expression, with an average of 1.5 transcripts per gene. However, the expression in that group decreased to 0.8, 1.0 and 0.5 transcripts per gene in the liver, whole blood and pancreas, respectively. Non-CpG island genes in the L1/LTR only group showed the lowest expression, with an average of 0.3 transcripts per gene in embryonic stem cells. The expression pattern of non-CpG island genes shifted in differentiated tissues, with increases noted. The highest average transcript levels were observed in the liver for non-CpG Alu-Alu genes at 2.0 transcripts per gene, in whole blood for non-CpG L1/LTR only genes at 6.8 transcripts per gene and in the pancreas for non-CpG Alu mixed genes at 4.7 transcripts per gene. These findings indicate that Alu-adjacent CpG-island genes were downregulated to compensate for the high expression of non-CpG island genes in differentiated tissues.

Exceptionally, the expression of CpG-island genes in the Alu 2 kb-Alu 3 kb group increased to 1.6 transcripts per gene in ovarian and testicular tissues, indicating a germline-specific regulatory pattern. Additionally, in cultured fibroblasts, there were 2.1 transcripts per gene of the CpG-island genes in the Alu beyond 7 kb or 5 kb group.

In this study, we propose an expanded concept of HKGs that goes beyond their traditional definition of ubiquitous expression, emphasizing their vital role in maintaining phenotypes during tissue differentiation. Our findings indicate that the proximity of retroelements, especially Alu elements, to TSSs significantly influences gene expression stability. This finding underscores the importance of epigenetic mechanisms, such as DNA methylation and chromatin organisation, in stabilising HKG expression^32,37,38. The observed variability in transcript quantification, especially in CpG-island genes associated with Alu elements, highlights the complexity of their regulation and the significant impact of retroelement proximity on gene expression stability. By incorporating these elements into the HKG definition, we offer a more comprehensive understanding of their regulatory roles, emphasising their contribution to maintaining transcriptome balance and ensuring consistent phenotypic outcomes in progeny cells during differentiation ^26,37.

Our findings indicate that the architecture surrounding TSSs, particularly within CpG islands, is crucial for modulating HKG expression patterns. The central regions of CpG islands are protected from methylation-induced deamination, thus preserving the essential cellular functions of HKGs^39,40. Conversely, the margins of CpG islands are susceptible to methylation, largely due to the presence of densely methylated retroelements near genes^26,37,41. In embryonic stem cells, the presence of Alu elements adjacent to CpG islands may help maintain an open chromatin state and hypomethylation, facilitating high gene expression^15,42. During tissue differentiation, tissue-specific genes become strongly upregulated, leading to subtle downregulation of numerous HKGs, which is vital for maintaining transcriptome balance. Alu elements not only produce suppressive Alu RNAs that bind to downregulated genes but also influence nearby methylation processes^35,36,43,44. Given the cell-division-dependent nature of methylation changes, retroelement-associated methylation patterns may spread into nearby downregulated HKGs, stabilising phenotypic traits in daughter cells, preventing dedifferentiation and promoting robust tissue-specific terminal differentiation ^26,28,37,45. This strategic arrangement of retroelements is essential for regulating genome-wide methylation and facilitating advanced tissue differentiation in multicellular organisms.

The HKG identification system shows potential as a marker for distinguishing various stages of differentiation in primary tissues. Although many markers have been suggested for identifying stem cells in primary tissues, their application in actual tissue differentiation has proven challenging⁴⁶. The concentration of stem cell markers, which are produced early in differentiation, decreases due to asymmetric cell division, and stem cells exhibit various phenotypes^46,47. However, epigenetic factors are inherited by progeny cells, maintaining epigenotypes and synchronising gene expression during differentiation^27,48. Insufficient differentiation, often due to factors like inflammation or the recruitment of numerous stem cells, may result in an increased number of undifferentiated cells. In previous studies, cells with epigenetic instability were identified in the background mucosa of gastric cancer patients^33,34. Recognising these unstable cells can help predict disease occurrence.

Our results demonstrate that CpG-island genes exhibit distinct expression patterns across different tissues, with particularly high expression levels in embryonic stem cells and significant downregulation in differentiated tissues such as the liver, whole blood and pancreas. In fibroblast cell lines, a marked increase in the expression of a particular CpG-island gene group, including fibronectin, suggests that these cells require destabilisation of CpG-island genes for in vitro expansion and trait transformation. Furthermore, compared to embryonic stem cells, germline tissues such as the ovary and testis, which undergo genome-wide hypomethylation, showed increased expression of Alu-adjacent CpG-island genes. These findings imply that the HKG identification system can be used to infer the stages of tissue differentiation. Moreover, our study addresses the issues with recent HKG identification approaches by proposing a unified framework that incorporates gene architecture and epigenetic regulation¹. This approach minimises variations in interpretations among researchers and provides a deeper understanding of gene expression dynamics across different stages of cellular differentiation^6–8. These insights have potential implications for identifying accurate biomarkers and therapeutic targets in developmental biology and disease treatment.

In this study, many of 56 tissues, ranging from embryonic stem cells to differentiated tissues, showed a gradual decrease in CpG-island gene expression, with variations observed based on the distribution of Alu retroelements. However, there are inherent limitations in directly analysing these patterns for tissue classification. Tissues contain a mixture of cell types with differing levels of differentiation. Additionally, DNA methylation changes after the formation of active chromatin under the influence of the tissue microenvironment⁴⁵. This means that even CpG-island genes in active chromatin may exhibit strong gene expression and be less influenced by methylation^49,50. These factors must be considered when analysing tissue-specific HKG stability. Future studies should employ refined experiments that consider tissue regions, single-cell levels, and differentiation stages to accurately infer HKG stability patterns in tissues.

While numerous studies have documented tissue-type-dependent methylation patterns, the specific effects of retroelements adjacent to genes, particularly at the margins of CpG island, remains underexplored^29,31,51. The variability at these margins, often compounded by low GC content, poses significant obstacles to experimental reproducibility^33,34,52. Furthermore, the epigenetic heterogeneity across different cells necessitates the use of small tissue samples to accurately detect subtle methylation changes^53,54. Building on prior research, which involved a meticulous selection of methylation sites and stringent experimental conditions to reduce PCR amplification bias, foundational insights into genome-wide methylation changes were gained^13,53–55. Notably, previous analyses revealed that age-related concurrent methylation changes in CpG-island genes are closely correlated with the type and proximity of adjacent retroelements ^32–34. This work provides a deeper understanding of the complex interplay between retroelements and gene expression. The current analysis further confirmed the significant influence of retroelement distribution on expression patterns and highlighted its implications for understanding the complexities of genomic regulation across different tissue types.

We observed weak associations between L1 and LTR elements and the properties of HKGs. Notably, L1 elements, predominantly found in genes lacking CpG islands, appear to play a specialised role in modulating gene expression, similar to their involvement in X-chromosome dosage compensation, where they contribute to the repression of neighbouring gene expression⁵⁶. During early cell differentiation, tissue-specific genes without CpG islands exhibited strong expression, facilitated by chromatin remodelling, and they were unaffected by L1-related methylation due to their low CpG density⁵⁷. Additionally, L1 methylation may downregulate other genes weakly expressed within heterochromatin regions. LTR elements, which are prevalent throughout the genome and especially common in facultative heterochromatin³⁷, display distinct differences in activity between species; they remain largely fossilised and inactive in humans but are actively expressed in mice, suggesting that LTRs have evolved divergent roles in genomic regulation across different mammalian lineages^18,37,58. Previous studies demonstrated that LTRs are associated with slow methylation changes in CpG-island genes^32–34, emphasising their pivotal role in long-term gene expression regulation.

Our investigation faced several constraints, notably the presence of alternative splicing variants¹, which introduce variability into TSSs and affect associated retroelement distributions. The RNA sequencing libraries we used lacked the detailed information needed to effectively differentiate these variants, complicating our analysis of gene expression patterns. Furthermore, we did not extensively explore the sequence characteristics at the TSSs used to identify HKGs in this study. HKGs are reported to exhibit a broader range of sequence variations than previously understood, including variations in some promoter binding sites¹⁶. Due to the complexity of the data and the multitude of variables present, we prioritised retroelements as our primary focus. Preliminary observations suggested that shorter CpG islands might correlate with HKG attributes, but we did not thoroughly analyse that relationship. The proximity of retroelement may reduce CpG-island length, potentially skewing results. Future research should focus on enhancing the resolution of variant data in RNA sequencing libraries and expanding the analysis of sequence variations and CpG island characteristics for HKG identification.

Our research extends beyond the traditional view of retroelements as merely parasitic DNA, underscoring their pivotal role in the evolution of eukaryotic genomes. Whereas prior studies focused on their roles in gene silencing and as alternative promoters^59–61, our findings highlight the profound influence of these elements, particularly Alu, during cellular differentiation. Comparative genomics between humans and mice reveals remarkable differences: Alu elements constitute approximately 10.7% of the human genome and form suppressive RNAs that bind to RNA polymerase, whereas in mice, SINE B2 elements constitute only 2.4% of the genome^37,58. In humans, Alu elements are densely positioned around the TSSs of CpG-island genes, enhancing their regulatory effects on gene expression^13,24. The higher prevalence of Alu elements in humans may confer evolutionary advantages, facilitating adaptability during cell differentiation, crucial for effective stem cell replacement and longevity^37,62,63. These insights highlight the specialised regulatory functions of Alu elements, distinguishing the human genomic architecture from that of other species and reinforcing their essential role in promoting longevity through sustained cellular differentiation.

Maintaining phenotypic stability is crucial for multicellular organisms, and our study expands the concept of HKGs beyond their traditional definition of ubiquitous expression. We demonstrate that Alu elements, which are dominant in the human genome, play an essential role in regulating HKG expression associated with CpG islands during cellular differentiation. Notably, retroelement distributions that exceed critical thresholds at specific gene locations can induce significant changes in gene expression. This suggests that repetitive sequences, often seen as parasitic DNA, actually function as symbiotic components essential for precise tissue differentiation. By re-evaluating retroelements as key regulatory mechanisms, our findings highlight their pivotal role in maintaining transcriptome balance, influencing phenotypic stability and advancing our understanding of genomic architecture in evolutionary biology and development.

Analysis of CpG islands and retroelement distribution in the human genome

We systematically examined the distribution of CpG islands and retroelements around the TSSs of individual genes in the human genome. Genomic data for human genes, retroelements, and CpG islands were obtained from the UCSC Genome Browser human genomes build GRCh38 (www.genome.ucsc.edu/). Information on CpG islands and retroelements was compiled for regions within 10 kb upstream and downstream of the TSSs for each RNA transcript. Genes were classified as CpG-island genes if the CpG islands were located within 200 bp of the TSSs. Retroelements were categorised as Alu, L1, or LTR, and the type and distance between the nearest retroelement and the TSS were determined, with distances segmented into 1-kb intervals. For genes with alternative splicing variants, the longest variant was selected as the standard for analysis.

Comprehensive acquisition of tissue-specific and embryonic stem cell RNA sequencing data

RNA sequencing data for 54 different tissues were obtained from Genotype-Tissue Expression (GTEx) version 8 (https://gtexportal.org/)⁴. To ensure consistency and comparability across tissue samples, we used the normalised transcripts per million data provided by GTEx. Additionally, RNA sequencing data for male and female human embryonic stem cells were acquired from previously published sources, identified by Sequence Read Archive database experiment accessions SRX1261741 and SRX1261751⁶⁴. The embryonic stem cell RNA sequencing data were processed and normalised using a pipeline adapted from the standardised RNA-seq methodology provided by GTEx, ensuring comprehensive coverage and uniform data treatment across all samples.

Analysis of HKG indicators

To assess the housekeeping gene expression data, TSI and CV values were computed across 54 different tissues using data from the GTEx portal. The TSI quantifies the tissue specificity of gene expression by assessing the standard deviation of expression across tissues relative to the mean expression across tissues, and the CV measures expression variability by calculating the ratio of the standard deviation of expression to the mean expression. These calculations were used to systematically evaluate the variability and tissue-specificity of gene expression, providing critical insights into the expression stability of HKGs across tissues.

Comparison of tissue expression data

To ensure accurate comparisons of gene expression levels across different tissue types, we implemented a specific normalisation procedure tailored to the gene groups. We calculated the total number of expressed genes and the sum of expressed transcripts within each gene group. To quantify these findings, we introduced two key metrics: proportional gene expression counts, which we defined as the proportion of expressed genes relative to the total number of expressed genes within each gene group, and proportional transcript quantification, which we defined as the proportion of expressed transcripts relative to all transcripts from expressed genes in each group. Additionally, we calculated the average number of transcripts per gene within each gene group by dividing the sum of the expressed transcripts by the total number of expressed genes. These metrics, particularly designed to minimise biases arising from variations in gene or transcript abundance, allow for a precise and unbiased comparison of gene expression levels across various tissue types.

Statistical analysis

Given that nonnormal distributions are characterised by skewness, the median was used as a measure of central tendency for analysing the TSI and CV values. Correlations between the TSI and CV were assessed using Spearman rank correlation analyses. Differences in the TSI and CV between CpG-island and non-CpG island genes were evaluated with the Mann‒Whitney U test. Differences in the TSI and CV based on the type and proximity of the retroelements were examined using the Kruskal‒Wallis test. Post-hoc pairwise comparisons were conducted to identify specific differences between groups after significant Kruskal‒Wallis tests were performed. Statistical significance was established at a two-tailed P value of 0.05. All the statistical analyses were conducted using SPSS version 20 and R statistics, ensuring comprehensive examination and validation of the results.

Competing interests

The authors declare no competing interests.

Cookson, W., Liang, L., Abecasis, G., Moffatt, M. & Lathrop, M. Mapping complex disease traits with global gene expression. Nat Rev Genet 10, 184-94 (2009).
Eisenberg, E. & Levanon, E.Y. Human housekeeping genes, revisited. Trends Genet 29, 569-74 (2013).
Consortium, G.T. et al. Genetic effects on gene expression across human tissues. Nature 550, 204-213 (2017).
Ong, S.L.M., Baelde, H.J., van, I.D.G.P., Bovee, J. & Szuhai, K. Identification of stable housekeeping genes for induced pluripotent stem cells and -derived endothelial cells for drug testing. Sci Rep 12, 16160 (2022).
Kapranov, P., Willingham, A.T. & Gingeras, T.R. Genome-wide transcription and the implications for genomic organization. Nat Rev Genet 8, 413-23 (2007).
McIntyre, L.M. et al. RNA-seq: technical variability and sampling. BMC Genomics 12, 293 (2011).
McCarthy, D.J., Chen, Y. & Smyth, G.K. Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. Nucleic Acids Res 40, 4288-97 (2012).
Kryuchkova-Mostacci, N. & Robinson-Rechavi, M. A benchmark of gene expression tissue-specificity metrics. Brief Bioinform 18, 205-214 (2017).
Wolf, S.F. & Migeon, B.R. Clusters of CpG dinucleotides implicated by nuclease hypersensitivity as control elements of housekeeping genes. Nature 314, 467-9 (1985).
Versteeg, R. et al. The human transcriptome map reveals extremes in gene density, intron length, GC content, and repeat pattern for domains of highly and weakly expressed genes. Genome Res 13, 1998-2004 (2003).
Ganapathi, M. et al. Comparative analysis of chromatin landscape in regulatory regions of human housekeeping and tissue specific genes. BMC Bioinformatics 6, 126 (2005).
Vinogradov, A.E. Dualism of gene GC content and CpG pattern in regard to expression in the human genome: magnitude versus breadth. Trends Genet 21, 639-43 (2005).
Anonymous. Study on methylation changes at CpG island margins. Journal Name, Volume, Pages.
Eller, C.D. et al. Repetitive sequence environment distinguishes housekeeping genes. Gene 390, 153-65 (2007).
Farre, D., Bellora, N., Mularoni, L., Messeguer, X. & Alba, M.M. Housekeeping genes tend to show reduced upstream sequence conservation. Genome Biol 8, R140 (2007).
Lawson, M.J. & Zhang, L. Housekeeping and tissue-specific genes differ in simple sequence repeats in the 5'-UTR region. Gene 407, 54-62 (2008).
Lander, E.S. et al. Initial sequencing and analysis of the human genome. Nature 409, 860-921 (2001).
Venter, J.C. et al. The sequence of the human genome. Science 291, 1304-51 (2001).
Bannert, N. & Kurth, R. Retroelements and the human genome: new perspectives on an old relation. Proc Natl Acad Sci U S A 101 Suppl 2, 14572-9 (2004).
Kazazian, H.H., Jr. Mobile elements: drivers of genome evolution. Science 303, 1626-32 (2004).
Bohne, A., Brunet, F., Galiana-Arnoux, D., Schultheis, C. & Volff, J.N. Transposable elements as drivers of genomic and biological diversity in vertebrates. Chromosome Res 16, 203-15 (2008).
Benit, L., Lallemand, J.B., Casella, J.F., Philippe, H. & Heidmann, T. ERV-L elements: a family of endogenous retrovirus-like elements active throughout the evolution of mammals. J Virol 73, 3301-8 (1999).
Estecio, M.R. et al. Genome architecture marked by retrotransposons modulates predisposition to DNA methylation in cancer. Genome Res 20, 1369-82 (2010).
Sundaram, V. et al. Widespread contribution of transposable elements to the innovation of gene regulatory networks. Genome Res 24, 1963-76 (2014).
Anonymous. Review of the epigenetic implication of gene-adjacent retroelements. Journal Name, Volume, Pages.
Yatabe, Y., Tavare, S. & Shibata, D. Investigating stem cells in human colon by using methylation patterns. Proc Natl Acad Sci U S A 98, 10839-44 (2001).
Easwaran, H.P., Schermelleh, L., Leonhardt, H. & Cardoso, M.C. Replication-independent chromatin loading of Dnmt1 during G2 and M phases. EMBO Rep 5, 1181-6 (2004).
Doi, A. et al. Differential methylation of tissue- and cancer-specific CpG island shores distinguishes human induced pluripotent stem cells, embryonic stem cells and fibroblasts. Nat Genet 41, 1350-3 (2009).
Ji, H. et al. Comprehensive methylome map of lineage commitment from haematopoietic progenitors. Nature 467, 338-42 (2010).
Hodges, E. et al. Directional DNA methylation changes and complex intermediate states accompany lineage specificity in the adult hematopoietic compartment. Mol Cell 44, 17-28 (2011).
Anonymous. Study on age-related methylation changes in primary tissue. Journal Name, Volume, Pages.
Anonymous. Study on methylation changes in gastrointestinal cancer. Journal Name, Volume, Pages.
Anonymous. Study on inflammation-associated methylation changes in gastrointestinal cancer. Journal Name, Volume, Pages.
Ponicsan, S.L., Kugel, J.F. & Goodrich, J.A. Genomic gems: SINE RNAs regulate mRNA production. Curr Opin Genet Dev 20, 149-55 (2010).
Zhang, Y. et al. Repetitive elements and enforced transcriptional repression co-operate to enhance DNA methylation spreading into a promoter CpG-island. Nucleic Acids Res 40, 7257-68 (2012).
Anonymous. Review of species-specific retroelements in humans and mice. Journal Name, Volume, Pages.
Gonzalez-Rico, F.J. et al. Alu retrotransposons modulate Nanog expression through dynamic changes in regional chromatin conformation via aryl hydrocarbon receptor. Epigenetics Chromatin 13, 15 (2020).
Gardiner-Garden, M. & Frommer, M. CpG islands in vertebrate genomes. J Mol Biol 196, 261-82 (1987).
Simmen, M.W. Genome-scale relationships between cytosine methylation and dinucleotide abundances in animals. Genomics 92, 33-40 (2008).
Arnaud, P., Goubely, C., Pelissier, T. & Deragon, J.M. SINE retroposons can be used in vivo as nucleation centers for de novo methylation. Mol Cell Biol 20, 3434-41 (2000).
Turker, M.S. Gene silencing in mammalian cells and the spread of DNA methylation. Oncogene 21, 5388-93 (2002).
Wang, X. et al. Spreading of Alu methylation to the promoter of the MLH1 gene in gastrointestinal cancer. PLoS One 6, e25913 (2011).
Polo, J.M. et al. A molecular roadmap of reprogramming somatic cells into iPS cells. Cell 151, 1617-32 (2012).
Lin, C.S., Xin, Z.C., Dai, J. & Lue, T.F. Commonly used mesenchymal stem cell markers and tracking labels: Limitations and challenges. Histol Histopathol 28, 1109-16 (2013).
Beckmann, J., Scheitza, S., Wernet, P., Fischer, J.C. & Giebel, B. Asymmetric cell division within the human hematopoietic stem and progenitor cell compartment: identification of asymmetrically segregating proteins. Blood 109, 5494-501 (2007).
Lee, H.J., Hore, T.A. & Reik, W. Reprogramming the methylome: erasing memory and creating diversity. Cell Stem Cell 14, 710-9 (2014).
Vavouri, T. & Lehner, B. Human genes with CpG island promoters have a distinct transcription-associated chromatin organization. Genome Biol 13, R110 (2012).
Siggens, L., Cordeddu, L., Ronnerblad, M., Lennartsson, A. & Ekwall, K. Transcription-coupled recruitment of human CHD1 and CHD2 influences chromatin accessibility and histone H3 and H3.3 occupancy at active chromatin regions. Epigenetics Chromatin 8, 4 (2015).
Eckhardt, F. et al. DNA methylation profiling of human chromosomes 6, 20 and 22. Nat Genet 38, 1378-85 (2006).
Peng, D. et al. Location-specific epigenetic regulation of the metallothionein 3 gene in esophageal adenocarcinomas. PLoS One 6, e22009 (2011).
Anonymous. Study on methylation changes associated with chromosomal losses in cancer. Journal Name, Volume, Pages.
Anonymous. Study on methylation changes in adult stem cells and primary tissues. Journal Name, Volume, Pages.
Anonymous. Study on methylation changes at CpG-island margins in cancer patients. Journal Name, Volume, Pages.
Chow, J.C. et al. LINE-1 activity in facultative heterochromatin formation during X chromosome inactivation. Cell 141, 956-69 (2010).
Weber, M. et al. Distribution, silencing potential and evolutionary impact of promoter DNA methylation in the human genome. Nat Genet 39, 457-66 (2007).
Waterston, R.H. et al. Initial sequencing and comparative analysis of the mouse genome. Nature 420, 520-62 (2002).
Burns, K.H. Transposable elements in cancer. Nat Rev Cancer 17, 415-424 (2017).
Payer, L.M. & Burns, K.H. Transposable elements in human genetic disease. Nat Rev Genet 20, 760-772 (2019).
Grillo, G. et al. Transposable Elements Are Co-opted as Oncogenic Regulatory Elements by Lineage-Specific Transcription Factors in Prostate Cancer. Cancer Discov 13, 2470-2487 (2023).
Shibata, D. Inferring human stem cell behaviour from epigenetic drift. J Pathol 217, 199-205 (2009).
Anonymous. Study on periodic fluctuations in cancer incidence. Journal Name, Volume, Pages.
Choi, J. et al. A comparison of genetically matched cell lines reveals the equivalence of human iPSCs and ESCs. Nat Biotechnol 33, 1173-81 (2015).

There is NO Competing Interest.

HKGSupplementarymaterials20240527.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Expanding the concept of housekeeping genes: Examining the influence of retroelements on gene expression stability during cellular differentiation

Status:

Version 1

Abstract

Figures

Introduction

Results

Discussion

Materials and Methods

Statistical analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1