DNA contains the genetic information that is passed on from generation to generation (Avery et al., 1944). Despite the fact that every cell in our body harbours the same genetic information, they have to fulfil a multitude of different functions. This places regulation of gene expression at the root of cellular processes (Crick, 1970, Strasser, 2006). Importantly, transcription is not merely a binary process, by switching genes on or off, but rather a fine-tuned and regulated process achieved on several levels, including chromatin states, DNA methylation, transcription factor recruitment as well as post-transcriptional control of mRNA degradation (Vaquerizas et al., 2009; Jones, 2012; Voss and Hager, 2014; Fukaya et al., 2016; Gehring et al., 2017). Transcription was largely thought to occur in a continuous and deterministic fashion. However, research over the last two decades has shown the inherent discreteness and stochasticity of transcription (Raj and van Oudenaarden, 2008; Deng et al., 2014; Voss and Hager, 2014). Especially in eukaryotes, transcriptional noise and bursting are important contributors towards transcriptional variation and have decisive implications in the regulation of developmental and cellular responses. In this review, we will discuss sources of transcriptional noise and bursts, their influence on cellular heterogeneity and their impact on cellular fate. We will highlight how recent advances in sequencing technologies have helped to address some of these concepts and how we can leverage them to understand how ageing affects transcriptional processes and cellular heterogeneity.
Sources of transcriptional variability
Gene expression in a biological system is shaped by various stimuli and thus, noise can have cell-extrinsic or cell-intrinsic sources. Extrinsic noise is usually characterised by external signalling cues, such as cell-cell or cell-matrix interactions, and chemokines that diffuse in the extracellular environment (Figure 1). One representative example are cells in a developing organism that experience gradients of morphogens and thus will initiate different transcriptional responses leading to different cellular fates (Rogers and Schier, 2011). Inflammatory signalling is one more extrinsic source of noisy expression. For example, cells respond with oscillations in the production of NF-κB molecules and the downstream genes of the NF-kappaB (κB) pathway when triggered constantly by the cytokine TNF-α (Kellogg and Tay, 2015). NF-κB dynamics contribute to robust immunological responses by shaping the chromatin environment and stochastically regulating gene expression of specific cytokines (e.g. IFN-β) (Natoli et al., 2005; Apostolou and Thanos, 2008). Thus, extrinsic noise is caused by the variability of extracellular signals and molecules, which are released from individual cell. Therefore, it is tightly linked to intrinsic noise (Elowitz et al., 2002). Intrinsic noise stems from the inherent variability of intracellular and intranuclear fluctuations of molecules or alterations in the chromatin environment (Swain et al., 2002; Kaern et al., 2005). Elegant experiments by Elowitz and colleagues in Escherichia coli demonstrated transcriptional noise in this system and shed light on how intrinsic noise can be easily distinguished from extrinsic noise. Individual, genetically identical cells harbouring genomic integrations of lac-repressible YFP/GFP fusions grown under the same conditions showed cell-to-cell variability in fluorescence, arguing for the existence of intrinsic noise (Elowitz et al., 2002). Noise was greater in cells with lower levels of expression, making the level of intrinsic noise directly dependent on the availability of lac repressor, whereas extrinsic noise peaked at intermediate levels of transcription (Elowitz et al., 2002). Important to note is that the lac-operon is a system consisting of a low number of regulatory modules, highlighting the fact that even fairly simple transcription circuits are stochastic in nature. A similar approach was followed by the O’Shea lab in Saccharomyces cerevisiae, who used GFP/YFP fusions of the Pho5 promoter and measured cell-extrinsic and -intrinsic noise upon Pho5 induction (Raser and O’Shea, 2004). Using genome-wide GFP-tagging of open reading frames coupled to high-throughput flow cytometry, Newman and colleagues systematically characterised noise in S. cerevisiae under two environmental conditions (Newman et al., 2006). Intriguingly, proteins important for responses to environmental changes were very noisy, whereas proteins involved in protein synthesis were not, suggesting that noise might ease adaptations to new surroundings (Newman et al., 2006). Another well-studied example of cell-intrinsic noise are embryonic stem cells (ES cells). These cells contain variable levels of transcription factors that play a role in maintaining pluripotency, e.g. NANOG is not present in all OCT4-expressing undifferentiated ESCs (Hatano et al., 2005; Chambers et al., 2007). Undifferentiated ESCs cultured under specific conditions (serum/LIF) are highly dynamic and balance between self-renewal and pluripotency. Single-cell RNA sequencing analysis of serum/LIF cultured undifferentiated ESCs revealed that dynamic fluctuations of Polycomb factors and chromatin states (e.g. H3K27me3) are associated with heterogeneous gene expression of pluripotency factors and the stochastic expression of lineage regulators (Kumar et al., 2014). Suppression of heterogeneity by changing culture conditions to 2i – which induces the ground-state of pluripotency – or the epigenetic landscape (Kumar et al., 2014; Choi et al., 2017) can lead to genomic instability, imprinting errors and affect the differentiation potential of ESCs (Yagi et al., 2017; Guo et al., 2018). One potential benefit of stochasticity is the possibility of balancing cell fate decisions. Thus, cellular heterogeneity is employed by cells, tissues and organisms to adapt, compensate, respond, defend and/or regulate processes that are essential for their survival and biological function.
An important contributor to transcriptional noise is fluctuations in chromatin accessibility (Brown et al., 2013) and the limiting numbers of transcription factors. Single-cell chromatin accessibility measurements support the link between accessibility and transcriptional heterogeneity (Buenrostro et al., 2015), while mathematical modelling of the competition between genes for transcription factors revealed that this competition is sufficient to enhance mRNA copy number variability within an isogenic population (Das et al., 2017). Furthermore, pausing of RNA polymerase, a critical step during the initiation of transcription (Adelman and Lis, 2012) contributes to the stochasticity of gene expression. Genes showing high levels of paused RNA Pol II tend to be expressed with higher synchrony (Lagha et al., 2013). One major contributor to transcriptional noise in a given cell population is the cell cycle (Buettner et al., 2015). Another determinant that has been observed early on is DNA looping (Figure 2B). One of the first examples is the choice between lytic and lysogenic state of the bacteriophage λ, where DNA looping promotes the lysogenic state (Ptashne, 1986; Anderson and Yang, 2008). Mathematical modelling has shown that DNA looping ensures the stability and robustness of the lysogenic state, even when fluctuations in the number of repressor molecules occur (Morelli et al., 2009). Another example is the stochastic expression of the globin genes (de Krom et al., 2002). In erythroid cells, the α-globin genes reside in a different chromosome than the β-globin genes, which are positioned in a constitutively open chromatin domain. Studies of the murine β-globin locus showed that the distant locus control region (LCR) of the gene is in direct contact with the actively transcribing gene in vivo, with the intervening DNA looping out (Carter et al., 2002; Tolhuis et al., 2002; Schoenfelder et al., 2010). These studies were among the first that showed DNA looping to occur alongside actively transcribed, stochastically expressed genes in vivo.
Monoallelic gene expression is an additional mechanism that contributes to transcriptional noise and heterogeneous gene expression (Figure 3). This phenomenon was initially observed in the stochastic inactivation of one X chromosome in the somatic cells of female mammals (Lyon, 1961). In autosomal genes, the inactivation of one allele was observed firstly in olfactory receptors (Chess et al., 1994) and cytokine genes IL-2 (Holländer et al., 1998), IL-4 (Rivière et al., 1998) and IFNB1 (Tabbaa et al., 2013), among others. The mechanism of DNA looping, in the form of interchromosomal interactions, has also been shown to play a role in the regulation of stochastic monoallelic expression of the olfactory receptor genes in neurons (Lomvardas et al., 2006) and in response to viral infection for the IFNB1 gene (Apostolou and Thanos, 2008).
How can the low number of transcription factors and DNA looping confer the specificity required to transcribe the right gene, on one allele and in specific cells? One explanation is provided in the classic example of bacteriophage λ, where the repressor dimer binds cooperatively on different regulatory sequences in order to promote long-range interactions that regulate the decision between lytic or lysogenic state (Dodd et al., 2004). The mechanism of cooperative DNA binding enables a low number of transcription factor molecules to co-ordinately and precisely drive robust transcriptional outcomes in a heterogeneous cell population. Additionally, the cooperative DNA binding can provide the mechanistic explanation of chromatin loop formation that facilitates stochastic gene expression (Balaeff et al., 2004; Todeschini et al., 2014; Nikopoulou et al., 2018).
Gene expression in bursts
One contributor to transcriptional variability is transcriptional bursting, i.e. the stochastic activation and inactivation of transcription. These oscillations in transcription and translation are usually referred to as “bursts” because the production and degradation of molecules do not always follow the Poisson distribution, where the synthesis and degradation states exist in a stable rate. Rather molecules are produced in a pulse-like manner in randomly distributed time intervals. Live-cell imaging has been used extensively to define the transcriptional outcome by the combination of burst size and frequency (Suter et al., 2011; Bothma et al., 2014; Fukaya et al., 2016; Fritzsch et al., 2018), while RNA degradation plays an additional role (Figure 4). It is thought that regulation of bursting frequencies is a major regulator of specific gene expression levels (Dar et al., 2012). Furthermore, emerging evidence suggests that regulation of this change in frequency can be modulated by enhancer-promoter looping (Bartman et al., 2016). Several mathematical models have been used to explain gene expression bursts and we refer the interested reader to an excellent overview of these simulations (Rao et al., 2002). In each burst, a certain number of mRNA molecules is produced. Current evidence suggests that burst size could be fixed in a given local promoter structure (Suter et al., 2011; Fukaya et al., 2016) and that the acetylation status at promoter or enhancer directly influences bursting frequency (Nicolas et al., 2018).
One of the first descriptions of transcription bursts stems from electron micrographs observing transcription in Drosophila embryos. The fibre-like structures of transcriptional units exhibited regions without actively transcribing RNA polymerase. This phenomenon was attributed to interruptions in the initiation of transcription (McKnight and Miller, 1979). The oscillations in transcription, referred to as transcriptional bursts, are supported by recent live imaging experiments, which visualise the regulation of bursting frequency of zygotic patterning genes in the Drosophila embryo (Fukaya et al., 2016). Bursting of RNAs was observed employing gene editing in combination with the MS2-GFP system. The MS2-GFP system has proven to be a useful tool to study transcription in vivo and has been used to assess (i) endogenous actively transcribing PolII kinetics in yeast (Larson et al., 2011), where fluctuations in the activation of a promoter were measured, (ii) discontinuous transcription in a human cell line (Darzacq et al., 2007) and (iii) bursting in bacterial transcription (Golding et al., 2005).
From early on, microscopy was used to observe patterns of transcription in tissues. In situ hybridisation, β-galactosidase and CAT assays identified transcriptional bursts in nuclei from muscle tissues of transgenic mice (Newlands et al., 1998). An additional advantageous technique for monitoring mRNA production/degradation rates is single-molecule imaging, which allows the actual tracing and quantification of mRNA molecules in prokaryotes, eukaryotes and mammalian tissues. Single molecule in situ fluorescent hybridisation (smFISH), pioneered by the Singer lab, is one of the most commonly used techniques for the observation of the kinetics and the quantification of the produced nascent and mature mRNAs in a cell (Femino et al., 1998). Since its development, this technique has been used extensively to study the burst characteristics of gene expression in a variety of biological systems.
Microscopy has been proven to be an ideal tool to study transcriptional bursting as it allows monitoring the dynamics of this process. It is, however, limited to a few loci as experiments require genome editing approaches. Single-cell RNA-seq (scRNA-seq), on the other hand, offers a more static viewpoint as it usually just monitors one given time-point. However, this technology makes it possible to monitor changes genome-wide. Interestingly, recent studies have shown that highly sensitive scRNA-seq in combination with allele specific expression analysis can also be leveraged to infer transcriptional bursting kinetics (Reinius and Sandberg, 2015). The sequencing-based approach also gives insights into the encoding of bursting characteristics and its association with the stochasticity patterns (Kim and Marioni, 2013). Expression of genes in a given cell over time can be modelled as transient burstlike transcription (Paulsson, 2005; Raj and van Oudenaarden, 2008; Corrigan et al., 2016). A recently published study on transcriptional burst kinetics demonstrate that the burst frequency and bursting size are indeed encoded in core promoter elements and enhancers (Larsson et al., 2019). They also show that the burst kinetics play an important role in defining cell type specific expression patterns using allelic single-cell RNA sequencing. With the continued improvement of scRNA-seq technologies, it is now possible to precisely measure minute amount of mRNA at a single cell level. This makes it possible to measure genome-wide allelic patterns in a biological system for the first time and may enable the inference of global bursting kinetics.
Role of transcription in cellular identity
Transcriptional noise and bursting both contribute to heterogeneity amongst cells that are seemingly morphologically identical and occurs even in simple systems, such as bacteria, yeast or tissue culture cells. Importantly, cell-to-cell variability can have profound effects on cellular function. This is important to realise as bulk measurements of cells or even tissue might mask these features. Due to the convenience in the isolation and analysis of blood cells by fluorescence-based flow cytometry (FACS), this system has been widely used to investigate the impact of heterogeneity on cell fate. For instance, subpopulations of clonally derived hematopoietic progenitor cells with low or high expression of the stem cell marker Sca-1 are in very different transcriptional states and give rise to different blood cell lineages (Chang et al., 2008). The combination of FACS and genome-wide measurements of gene expression profiles is a powerful approach to determine the influence of heterogeneity on biological function.
The recent development of single-cell sequencing technology has dramatically increased the resolution and sensitivity of heterogeneity measurements, which can even be applied in mixed cell populations (Trapnell, 2015). This technology overcomes fundamental limitations inherent in bulk measurements, thus providing a high-resolution view of transitions between different cell states. However, the experimental approach as well as the computational analysis for scRNA-seq are still in progress (Kolodziejczyk et al., 2015; Stegle et al., 2015; Wagner et al., 2016; Ziegenhain et al., 2018). While opening exciting possibilities, the complexity of the data makes its analysis and interpretation difficult and confounding factors on the heterogeneity of gene expression, such as the cell cycle state, need to be taken into consideration (Buettner et al., 2015). Nevertheless, scRNA-seq has been successfully employed to identify yet unknown sub-populations of cells either during transition stages in differentiation (e.g. from naïve T cells to T helper cells; Buettner et al., 2015) or of rare, already differentiated cells in the lung (Plasschaert et al., 2018). The possibility to measure heterogeneity in a pool of cells and capture cells at transition points during development and differentiation is an inherent strength of scRNA-seq. Many exciting approaches have been recently developed to address these open questions, for example, RNA velocity (La Manno et al., 2018), which takes into consideration the abundance of spliced and unspliced mRNA in scRNA-seq data and uses this information to predict the future state of a cell.
Ageing and cellular heterogeneity
As discussed at several points during this review, flow cytometry is one of the most powerful tools for investigating cellular heterogeneity. It has been used very successfully to analyse specific parameters in single cells, especially for the analysis of the immune system. Novel approaches allow even simultaneous recording of up to 40 parameters per cell (Simoni et al., 2018). Next to the well-known heterogeneity in immune cells, also beta cells show a great variability, which is suggested to be important for their function (Benninger et al., 2018). An elegant, yet simple experiment, in which GFP was placed under the control of the insulin promoter revealed the existence of at least three different populations of pancreatic beta cells that showed difference in GFP brightness, cell size and granularity (Katsuta et al., 2012). This reporter was then used to investigate whether cells with different levels of GFP correspond to different chronological ages. Indeed, GFP levels were lower in young mice and increased with age. The existence of cells with different GFP levels in any given mouse suggests that the pancreas simultaneously harbours differently aged subpopulations (Aguayo-Mazzucato et al., 2017). The composition of GFP low to high cells changed during the lifespan of mice. Older mice have a higher percentage of older cells, which are also characterised by higher expression of the senescence marker p16Ink4a. At the same time these cells show down-regulation of beta cell signature genes and a general decline in beta cell function (Aguayo-Mazzucato et al., 2017).
A recent study used a very different approach to answer the question of increasing heterogeneity with age. Philipp and coworkers analysed various biophysical properties of cells including cell motility and mechanics, cellular traction strength as well as cell and nuclear morphology (Phillip et al., 2017). In addition, the authors measured biomolecular properties of cells, such as levels of ATP, the amount of cellular secretions, DNA damage response (DDR), nuclear organisation, and cytoskeletal content. The complete set of measurement was performed in 32 individual human fibroblasts spanning over nine decades using mainly microscopy approaches. The measured parameters indicated significant changes in cellular heterogeneity with age. However, not all parameters showed an increase in heterogeneity (e.g. cellular and nuclear size), but some also showed a decrease (e.g. cellular speed) (Phillip et al., 2017). As pointed out earlier in this review, cell cycle stages are a clear confounding factor of scRNA-seq studies (Buettner et al., 2015) and can lead to false assumptions with respect to heterogeneity, if not taken into consideration. The effect of the cell cycle state can be also observed in the biophysical properties of cells (Phillip et al., 2017). Taken together, heterogeneity changes during the lifespan of an organism and new technological developments are helping to unravel their contribution to the ageing process.
Transcriptional variability during ageing
As described already, chromatin architecture directly influences transcriptional bursting and its maintenance is of critical importance to properly regulate transcription (Feser and Tyler, 2011). Work in yeast – that undergoes a specific form of senescence, termed replicative ageing (Longo et al., 2012) – demonstrated that transcriptional noise might be increased as histone expression decreases with higher age (Feser et al., 2010). The down-regulation of histone proteins leads to a dramatic loss of transcriptional regulation with age and a concomitant induction of all yeast genes due to an overall loss of promoter-localised nucleosomes (Hu et al., 2014). In addition, enhanced chromatin accessibility was proposed to be the underlying cause of the observed increase in DNA damage foci and the appearance of cryptic transcripts (Feser et al., 2010). Using a histone mutant library, the Berger lab identified the maintenance of H3K36 methylation as a key component for the suppression of cryptic transcription emanating from several genes in old yeast cells (Sen et al., 2015). Importantly, mutating either H3K36 itself or the enzymes placing (Set2p) or removing (Rph1p) tri-methylation of H3K36, modulated not only the levels of cryptic transcription, but also the lifespan of yeast, suggesting a link between these two phenomena. These studies demonstrated that chromatin modifications and local as well as global architecture change with age, accompanied by an increase in transcriptional noise. However, it is important to note that all of the studies were performed on a population level. A recent single-cell analysis that followed single yeast over their lifetime indicated that the situation might be more complex. By measuring transcriptional noise of the GAL1 gene, using its natural, but also synthetic promoters, the authors concluded that a single cell experiences a reduction in transcriptional noise during its lifetime (Liu et al., 2017). This reduction in noise occurs until the cell reaches its last four to five cell divisions, when noise increases dramatically. Mammalian cells entering senescence in culture show highly diverse phenotype even within a single culture. Single-cell transcriptome analysis revealed a high degree of variability in mRNA level from cell to cell (Wiley et al., 2017), indicating that one characteristic of senescent cells is higher transcriptional noise. At present it is not clear why single-cell and bulk measurements differ with regard to the extent of transcriptional noise with cellular ageing. One potential explanation is of technical nature, i.e. differences in cell-extrinsic vs. cell-intrinsic noise that is observed using these two different approaches.
Ageing and transcriptional variability
In the case of organismal ageing in mammalian cells, chromatin maintenance is failing, and chromatin rearrangements can be observed at all levels, from three-dimensional (3D) organisation (Florian et al., 2012; Koohy et al., 2018), down to changes in accessibility at specific sites (Bochkis et al., 2014; Moskowitz et al., 2017; Ucar et al., 2017; Koohy et al., 2018). In contrast to the data in yeast that demonstrated an overall increase in chromatin accessibility, the situation is more complex in mammalian cells. While in aged mouse pre-B cells ATAC-seq experiments revealed only a few changes in chromatin accessibility (Koohy et al., 2018), the same experimental setup showed large-scale chromatin rearrangements in PBMCs isolated from healthy human donors (Ucar et al., 2017). However, rather than an overall increase or decrease in chromatin accessibility, this study revealed that there are almost as many sites opening than closing with age. A similar phenomenon was also observed using MNase-seq to directly map nucleosome occupancy in young and old liver tissue (Bochkis et al., 2014), suggesting cell-type specific chromatin rearrangements. In purified CD8+ T cells, ageing leads to an overall decrease in chromatin accessibility, particularly at the promoter, while other regions in the genome becoming more accessible (Moskowitz et al., 2017). Importantly, all of the changes described were accompanied by changes in the gene expression programme and can also explain the phenotypic changes observed with age. Therefore, it will be interesting to compare different cell types, in particular those that show different characteristics, such as high/low proliferation rate, or cellular lifespan.
To the best of our knowledge, the first single cell RNA dataset of organismally aged cells was generated in cardiomyocytes in 2006. Fifteen genes were analysed by single-cell quantative polymerase chain reaction (qPCR) in around 15 isolated cells each from young (6 months) and old (27 months) mice (Bahar et al., 2006). Despite the limited data, it became clear that cell-to-cell variability was increased with higher age. This finding was corroborated by a study on around 2500 single pancreatic cells, in which donor age spanned six decades (Enge et al., 2017). Although the oldest age-span the authors examined was 38–54 years of age, noise clearly increased from young adult (21–22 years) to the mid-life span, suggesting that it would be even more profound at old age. This notion was also recently reported in lung, in which transcriptional noise and ageing were positively correlated in most of the 30 different cell types identified in the mouse lung (Angelidis et al., 2019). Interestingly, systematic gene expression alterations accompanying increases in transcriptional noise have been reported to be associated with stress response genes, such as FOSB, HSPA1A and JUND, suggesting that cellular stress might play a role in the observed variability with age (Enge et al., 2017).
An increase in transcriptional noise is not the only change in the transcriptional landscape during physiological ageing. Importantly, responses to extracellular cues are affected upon ageing. While this phenomenon has been known on a bulk or tissue level for decades (Riera et al., 2016), we have only recently started to understand this on a single-cell level. As discussed, transcriptional variability increases with age. While this is very interesting purely from a mechanistic perspective, it also has a profound influence on tissue function as it can impact cellular fate decisions and states (Enge et al., 2017), but also cellular function. This is of particular importance if the cell has to quickly respond to signalling events. In old CD4+ T-cells transcriptional variability includes genes that are important for the activation of T-cells and these are often transcribed in lower levels compared to younger cells (Martinez-Jimenez et al., 2017). Upon extracellular activation, these cells cannot activate the T-cell response fast enough. The failure of CD4+ T-cells to robustly up-regulate a core activation programme might contribute to the ageing-associated decrease of immune function. Finally, a recent study that aimed at cataloguing the different cell types of cells in the Drosophila brain throughout its life, found an exponential decline in the number of genes expressed during ageing. Importantly, the decrease of mRNA in the Drosophila brain did not affect cellular identity as cell clusters did not change during the lifetime of the fly (Davie et al., 2018). An interesting side note here is the idea to use single-cell transcriptomic data to identify the biological age of a given cell, i.e. the possibility to devise a cell-specific transcriptional ageing clock.
Taken together, the data emerging from single cell RNA-seq data largely confirms the predictions made in bulk experiments and indicate that ageing results in an increased cell-to-cell transcriptional variability, which might stem from a general chromatin deregulation at older age (Booth and Brunet, 2016). Indeed, integrating the described, published scRNA-seq data on CD4+ cells (Martinez-Jimenez et al., 2017) with cell-to-cell epigenomic variability suggested that increased epigenomic noise is one mechanism that might lead to elevated transcriptional noise with age (Cheung et al., 2018). Interestingly, transcriptional noise was increased in particular on genes marked with H3K27me3 (Cheung et al., 2018). Future challenges include the development of robust single cell ChIP-seq protocols to study not only levels, but also localisation of histone modifications. Ideally, to properly correlate changes in the epigenome with transcript levels, both measurements would be performed in a single cell, similar to approaches that allow simultaneous measurements of DNA methylation and transcriptome (Angermueller et al., 2016; Hou et al., 2016; Hu et al., 2016; Pott, 2017; Clark et al., 2018). These exciting findings over the last few years demonstrated that ageing is associated with increase in cell-to-cell variability on an epigenomic and transcriptomic level and that these changes are – at least in part – responsible for the cellular heterogeneity observed at higher age.
The field of single cell biology has undergone enormous development over the last years through the development of high-throughput single-cell RNA-seq. While foundation of many of the theories was laid using microscopy and flow cytometry, novel single-cell technologies allow for the investigation of transcriptional processes and heterogeneity in an unprecedented depth and broadness. Importantly, the development of single cell epigenomic approaches has really just started. Single-cell ATAC-seq has been used successfully in several laboratories (Buenrostro et al., 2015; Cusanovich et al., 2015) and is now also commercially available. In addition, there are now the first reports of single-cell ChIP-seq type experiments that do not require elaborate technical solutions (Hainer et al., 2018). A more extensive overview can be found in a recent review focussing on single-cell epigenomics (Kelsey et al., 2017). Many of the studies using single cell approaches have so far addressed developmental processes and demonstrated the dynamics of a developing organism. The few studies addressing the dynamics of transcriptional noise and cellular heterogeneity during the ageing process highlight the many exciting questions to be answered.
We apologise to all researchers in the field whose research we were not able to cite due to space constraints. We would like to thank members of the Tessarz laboratory for comments on the manuscript. This work was supported by the Max Planck Society.
Aguayo-Mazzucato, C., van Haaren, M., Mruk, M., Lee Jr., T.B., Crawford, C., Hollister-Lock, J., Sullivan, B.A., Johnson, J.W., Ebrahimi, A., Dreyfuss, J.M., et al. (2017). β Cell aging markers have heterogeneous distribution and are induced by insulin resistance. Cell Metab. 25, 898–910. CrossrefPubMedGoogle Scholar
Angelidis, I., Simon, L.M., Fernandez, I.E., Strunz, M., Mayr, CH., Greiffo, F.R., Tsitsiridis, G., Graf, E., Strom, T.M., Nagendran, M., et al. (2019). An atlas of the aging lung mapped by single cell transcriptomics and deep tissue proteomics. Nat. Commun. 10, 963. PubMedCrossrefGoogle Scholar
Angermueller, C., Clark, S.J., Lee, H.J., Macaulay, I.C., Teng, M.J., Hu, T.X., Krueger, F., Smallwood, S., Ponting, C.P., Voet, T., et al. (2016). Parallel single-cell sequencing links transcriptional and epigenetic heterogeneity. Nat. Methods 13, 229–232. PubMedCrossrefGoogle Scholar
Avery, O.T., Macleod, C.M., and McCarty, M. (1944). Studies on the chemical nature of the substance inducing transformation of pneumococcal types: induction of transformation by a desoxyribonucleic acid fraction isolated from Pneumococcus type III. J. Exp. Med. 79, 137–158. PubMedCrossrefGoogle Scholar
Bartman, C.R., Hsu, S.C., Hsiung, C.C.-S., Raj, A., and Blobel, G.A. (2016). Enhancer regulation of transcriptional bursting parameters revealed by forced chromatin looping. Mol. Cell 62, 237–247. PubMedCrossrefGoogle Scholar
Bochkis, I.M., Przybylski, D., Chen, J., and Regev, A. (2014). Changes in nucleosome occupancy associated with metabolic alterations in aged mammalian liver. Cell Rep. 9, 996–1006. PubMedCrossrefGoogle Scholar
Bothma, J.P., Garcia, H.G., Esposito, E., Schlissel, G., Gregor, T., and Levine, M. (2014). Dynamic regulation of eve stripe 2 expression reveals transcriptional bursts in living Drosophila embryos. Proc. Natl. Acad. Sci. USA 111, 10598–10603. CrossrefGoogle Scholar
Brown, C.R., Mao, C., Falkovskaia, E., Jurica, M.S., and Boeger, H. (2013). Linking stochastic fluctuations in chromatin structure and gene expression. PLoS Biol. 11, e1001621. PubMedCrossrefGoogle Scholar
Buenrostro, J.D., Wu, B., Litzenburger, U.M., Ruff, D., Gonzales, M.L., Snyder, M.P., Chang, H.Y., and Greenleaf, W.J. (2015). Single-cell chromatin accessibility reveals principles of regulatory variation. Nature 523, 486–490. PubMedCrossrefGoogle Scholar
Buettner, F., Natarajan, K.N., Casale, F.P., Proserpio, V., Scialdone, A., Theis, F.J., Teichmann, S.A., Marioni, J.C., and Stegle, O. (2015). Computational analysis of cell-to-cell heterogeneity in single-cell RNA-sequencing data reveals hidden subpopulations of cells. Nat. Biotechnol. 33, 155. PubMedCrossrefGoogle Scholar
Chambers, I., Silva, J., Colby, D., Nichols, J., Nijmeijer, B., Robertson, M., Vrana, J., Jones, K., Grotewold, L., and Smith, A. (2007). Nanog safeguards pluripotency and mediates germline development. Nature 450, 1230–1234. CrossrefPubMedGoogle Scholar
Chang, H.H., Hemberg, M., Barahona, M., Ingber, D.E., and Huang, S. (2008). Transcriptome-wide noise controls lineage choice in mammalian progenitor cells. Nature 453, 544–547. PubMedCrossrefGoogle Scholar
Cheung, P., Vallania, F., Warsinske, H.C., Donato, M., Schaffert, S., Chang, S.E., Dvorak, M., Dekker, C.L., Davis, M.M., Utz, P.J., et al. (2018). Single-cell chromatin modification profiling reveals increased epigenetic variations with aging. Cell 173, 1385–1397. CrossrefPubMedGoogle Scholar
Choi, J., Huebner, A.J., Clement, K., Walsh, R.M., Savol, A., Lin, K., Gu, H., Di Stefano, B., Brumbaugh, J., Kim, S.Y., et al. (2017). Prolonged Mek1/2 suppression impairs the developmental potential of embryonic stem cells. Nature 548, 219–223. PubMedCrossrefGoogle Scholar
Clark, S.J., Argelaguet, R., Kapourani, C.A., Stubbs, T.M., Lee, H.J., Alda-Catalinas, C., Krueger, F., Sanguinetti, G., Kelsey, G., Marioni, J.C., et al. (2018). scNMT-seq enables joint profiling of chromatin accessibility DNA methylation and transcription in single cells. Nat. Commun. 9, 781. CrossrefPubMedGoogle Scholar
Cusanovich, D.A., Daza, R., Adey, A., Pliner, H.A., Christiansen, L., Gunderson, K.L., Steemers, F.J., Trapnell, C., and Shendure, J. (2015). Multiplex single cell profiling of chromatin accessibility by combinatorial cellular indexing. Science 348, 910–914. PubMedCrossrefGoogle Scholar
Dar, R.D., Razooky, B.S., Singh, A., Trimeloni, T.V., McCollum, J.M., Cox, C.D., Simpson, M.L., and Weinberger, L.S. (2012). Transcriptional burst frequency and burst size are equally modulated across the human genome. Proc. Natl. Acad. Sci. USA 109, 17454–17459. CrossrefGoogle Scholar
Darzacq, X., Shav-Tal, Y., de Turris, V., Brody, Y., Shenoy, S.M., Phair, R.D., and Singer, R.H. (2007). In vivo dynamics of RNA polymerase II transcription. Nat. Struct. Mol. Biol. 14, 796–806. CrossrefPubMedGoogle Scholar
Davie, K., Janssens, J., Koldere, D., De Waegeneer, M., Pech, U., Kreft, K., Albar, S., Makhzami, S., Christiaens, V., Bravo González-Blas, C., et al. (2018). A single-cell transcriptome atlas of the aging Drosophila brain. Cell 174, 982–998. PubMedCrossrefGoogle Scholar
de Krom, M., van de Corput, M., von Lindern, M., Grosveld, F., and Strouboulis, J. (2002). Stochastic patterns in globin gene expression are established prior to transcriptional activation and are clonally inherited. Mol. Cell 9, 1319–1326. CrossrefPubMedGoogle Scholar
Deng, Q., Ramsköld, D., Reinius, B., and Sandberg, R. (2014). Single-cell RNA-seq reveals dynamic, random monoallelic gene expression in mammalian cells. Science 343, 193–196. CrossrefPubMedGoogle Scholar
Dodd, I.B., Shearwin, K.E., Perkins, A.J., Burr, T., Hochschild, A., and Egan, J.B. (2004). Cooperativity in long-range gene regulation by the lambda CI repressor. Genes Dev. 18, 344–354. CrossrefPubMedGoogle Scholar
Enge, M., Arda, H.E., Mignardi, M., Beausang, J., Bottino, R., Kim, S.K., and Quake, S.R. (2017). Single-cell analysis of human pancreas reveals transcriptional signatures of aging and somatic mutation patterns. Cell 171, 321–330.e14. Google Scholar
Feser, J., Truong, D., Das, C., Carson, J.J., Kieft, J., Harkness, T., and Tyler, J.K. (2010). Elevated histone expression promotes life span extension. Mol. Cell 39, 724–735. PubMedCrossrefGoogle Scholar
Florian, M.C., Dörr, K., Niebel, A., Daria, D., Schrezenmeier, H., Rojewski, M., Filippi, M.D., Hasenberg, A., Gunzer, M, Scharffetter-Kochanek, K., et al. (2012). Cdc42 activity regulates hematopoietic stem cell aging and rejuvenation. Cell Stem Cell 10, 520–530. CrossrefPubMedGoogle Scholar
Fritzsch, C., Baumgärtner, S., Kuban, M., Steinshorn, D., Reid, G., and Legewie, S. (2018). Estrogen-dependent control and cell-to-cell variability of transcriptional bursting. Mol. Syst. Biol. 14, e7678. CrossrefPubMedGoogle Scholar
Gehring, N.H., Wahle, E., and Fischer, U. (2017). Deciphering the mRNP code: RNA-bound determinants of post-transcriptional gene regulation. Trends Biochem. Sci. 42, 369–382. CrossrefPubMedGoogle Scholar
Guo, R., Ye, X., Yang, J., Zhou, Z., Tian, C., Wang, H., Wang, H., Fu, H., Liu, C., Zeng, M., et al. (2018). Feeders facilitate telomere maintenance and chromosomal stability of embryonic stem cells. Nat. Commun. 9, 2620. PubMedCrossrefGoogle Scholar
Hainer, S.J., Boskovic, A., Rando, O.J., and Fazzio, T.G. (2018). Profiling of pluripotency factors in individual stem cells and early embryos. bioRxiv, 286351. Google Scholar
Hatano, S.-Y., Tada, M., Kimura, H., Yamaguchi, S., Kono, T., Nakano, T., Suemori, H., Nakatsuji, N., and Tada, T. (2005). Pluripotential competence of cells associated with Nanog activity. Mech. Dev. 122, 67–79. PubMedCrossrefGoogle Scholar
Holländer, G.A., Zuklys, S., Morel, C., Mizoguchi, E., Mobisson, K., Simpson, S., Terhorst, C., Wishart, W., Golan, D.E., Bhan, A.K., et al. (1998). Monoallelic expression of the interleukin-2 locus. Science 279, 2118–2121. CrossrefPubMedGoogle Scholar
Hou, Y., Guo, H., Cao, C., Li, X., Hu, B., Zhu, P., Wu, X., Wen, L., Tang, F., Huang, Y., et al. (2016). Single-cell triple omics sequencing reveals genetic, epigenetic, and transcriptomic heterogeneity in hepatocellular carcinomas. Cell Res. 26, 304–319. CrossrefPubMedGoogle Scholar
Hu, Z., Chen, K., Xia, Z., Chavez, M., Pal, S., Seol, J.-H., Chen, C.-C., Li, W., and Tyler, J.K. (2014). Nucleosome loss leads to global transcriptional up-regulation and genomic instability during yeast aging. Genes Dev. 28, 396–408. PubMedCrossrefGoogle Scholar
Hu, Y., Huang, K., An, Q., Du, G., Hu, G., Xue, J., Zhu, X., Wang, C.Y., Xue, Z., Fan, G. (2016). Simultaneous profiling of transcriptome and DNA methylome from a single cell. Genome Biol. 17, 88. CrossrefPubMedGoogle Scholar
Katsuta, H., Aguayo-Mazzucato, C., Katsuta, R., Akashi, T., Hollister-Lock, J., Sharma, A.J., Bonner-Weir, S., and Weir, G.C. (2012). Subpopulations of GFP-marked mouse pancreatic β-cells differ in size, granularity, and insulin secretion. Endocrinology 153, 5180–5187. PubMedCrossrefGoogle Scholar
Koohy, H., Bolland, D.J., Matheson, L.S., Schoenfelder, S., Stellat, C., Dimond, A., Varnai, C., Chovanec, P., Chessa, T., Denizot, J., et al. (2018). Genome organization and chromatin analysis identify transcriptional downregulation of insulin-like growth factor signalling as a hallmark of aging in developing B cells. Genome Biol. 19, 126. CrossrefPubMedGoogle Scholar
Kumar, R.M., Cahan, P., Shalek, A.K., Satija, R., DaleyKeyser, A., Li, H., Zhang, J., Pardee, K., Gennert, D., Trombetta, J.J., et al. (2014). Deconstructing transcriptional heterogeneity in pluripotent stem cells. Nature 516, 56–61. CrossrefPubMedGoogle Scholar
Lagha, M., Bothma, J.P., Esposito, E., Ng, S., Stefanik, L., Tsui, C., Johnston, J., Chen, K., Gilmour, D.S., Zeitlinger, J., et al. (2013). Paused Pol II coordinates tissue morphogenesis in the Drosophila embryo. Cell 153, 976–987. PubMedCrossrefGoogle Scholar
La Manno, G., Soldatov, R., Zeisel, A., Braun, E., Hochgerner, H., Petukhov, V., Lidschreiber, K., Kastriti, M.E., Lönnerberg, P., Furlan, A., et al. (2018). RNA velocity of single cells. Nature 560, 494–498. CrossrefPubMedGoogle Scholar
Larson, D.R., Zenklusen, D., Wu, B., Chao, J.A., and Singer, R.H. (2011). Real-time observation of transcription initiation and elongation on an endogenous yeast gene. Science 332, 475–478. CrossrefPubMedGoogle Scholar
Larsson, A.J.M., Johnsson, P., Hagemann-Jensen, M., Hartmanis, L., Faridani, O.R., Reinius, B., Segerstolpe, A., Rivera, C.M., Ren, B., Sandberg, R., et al. (2019). Genomic encoding of transcriptional burst kinetics. Nature 565, 251–254. PubMedCrossrefGoogle Scholar
Lomvardas, S., Barnea, G., Pisapia, D.J., Mendelsohn, M., Kirkland, J., and Axel, R. (2006). Interchromosomal interactions and olfactory receptor choice. Cell 126, 403–413. CrossrefPubMedGoogle Scholar
Martinez-Jimenez, C.P., Eling, N., Chen, H.C., Vallejos, C.A., Kolodziejczyk, A.A., Connor, F., Stojic, L., Rayner, T.F., Stubbington, M.J.T., Teichmann, S.A., et al. (2017). Aging increases cell-to-cell transcriptional variability upon immune stimulation. Science 355, 1433–1436. CrossrefPubMedGoogle Scholar
McKnight, S.L. and Miller Jr, O.L. (1979). Post-replicative nonribosomal transcription units in D. melanogaster embryos. Cell 17, 551–563. Google Scholar
Morelli, M.J., Ten Wolde, P.R., and Allen, R.J. (2009). DNA looping provides stability and robustness to the bacteriophage lambda switch. Proc. Natl. Acad. Sci. USA 106, 8101–8106. CrossrefGoogle Scholar
Moskowitz, D.M., Zhang, D.W., Hu, B., Le Saux, S., Yanes, R.E., Ye,Z., Buenrostro, J.D., Weyand, C.M., Greenleaf, W.J., and Goronzy, J.J. (2017). Epigenomics of human CD8 T cell differentiation and aging. Sci. Immunol. 2. doi: 10.1126/sciimmunol.aag0192. PubMedGoogle Scholar
Natoli, G., Saccani, S., Bosisio, D., and Marazzi, I. (2005). Interactions of NF-kB with chromatin: the art of being at the right place at the right time. Nat. Immunol. 6, 439–445. CrossrefGoogle Scholar
Newlands, S., Levitt, L.K., Robinson, C.S., Karpf, A.B., Hodgson, V.R., Wade, R.P., and Hardeman, E.C. (1998). Transcription occurs in pulses in muscle fibers. Genes Dev. 12, 2748–2758. CrossrefPubMedGoogle Scholar
Newman, J.R.S., Ghaemmaghami, S., Ihmels, J., Breslow, D.K., Noble, M., DeRisi, J.L., and Weissman, J.S. (2006). Single-cell proteomic analysis of S. cerevisiae reveals the architecture of biological noise. Nature 441, 840–846. CrossrefPubMedGoogle Scholar
Nikopoulou, C., Panagopoulos, G., Sianidis, G., Psarra, E., Ford, E., and Thanos, D. (2018). The transcription factor ThPOK orchestrates stochastic interchromosomal interactions required for IFNB1 virus-inducible gene expression. Mol. Cell 71, 352–361.e5. PubMedCrossrefGoogle Scholar
Phillip, J.M., Wu, P.H., Gilkes, D.M., Williams, W., McGovern, S., Daya, J., Chen, J., Aifuwa, I., Lee, J.S., Fan, R., et al. (2017). Biophysical and biomolecular determination of cellular age in humans. Nat. Biomed. Eng.1, 0093. CrossrefGoogle Scholar
Plasschaert, L.W., Žilionis, R., Choo-Wing, R., Savova, V., Knehr, J., Roma, G., Klein, A.M., and Jaffe, A.B. (2018). A single-cell atlas of the airway epithelium reveals the CFTR-rich pulmonary ionocyte. Nature 560, 377–381. CrossrefPubMedGoogle Scholar
Schoenfelder, S., Sexton, T., Chakalova, L., Cope, N.F., Horton, A., Andrews, S., Kurukuti, S., Mitchell, J.A., Umlauf, D., Dimitrova, D.S., et al. (2010). Preferential associations between co-regulated genes reveal a transcriptional interactome in erythroid cells. Nat. Genet. 42, 53–61. CrossrefPubMedGoogle Scholar
Sen, P., Dang, W., Donahue, G., Dai, J., Dorsey, J., Cao, X., Liu, W., Cao, K., Perry, R., Lee, Y.P., et al. (2015). H3K36 methylation promotes longevity by enhancing transcriptional fidelity. Genes Dev. 29, 1362–1376. PubMedCrossrefGoogle Scholar
Strasser, B.J. (2006). A world in one dimension: Linus Pauling, Francis Crick and the central dogma of molecular biology. Hist. Philos. Life Sci. 28, 491–512. Google Scholar
Suter, D.M., Molina, N., Gatfield, D., Schneider, K., Schibler, U., and Naef, F. (2011). Mammalian genes are transcribed with widely different bursting kinetics. Science 332, 472–474. CrossrefPubMedGoogle Scholar
Tabbaa, O.P., Nudelman, G., Sealfon, S.C., Hayot, F., and Jayaprakash, C. (2013). Noise propagation through extracellular signaling leads to fluctuations in gene expression. BMC Syst. Biol. 7, 94. CrossrefPubMedGoogle Scholar
Tolhuis, B., Palstra, R.J., Splinter, E., Grosveld, F., and de Laat, W. (2002). Looping and interaction between hypersensitive sites in the active b-globin locus. Mol. Cell 10, 1453–1465. CrossrefGoogle Scholar
Ucar, D., Marquez, E.J., Chung, C.H., Marches, R., Rossi, R.J., Uyar,A., Wu, T.C., George, J., Stitzel, M.L., Palucka, A.K., et al. (2017). The chromatin accessibility signature of human immune aging stems from CD8+ T cells. J. Exp. Med. 214, 3123–3144. PubMedCrossrefGoogle Scholar
Vaquerizas, J.M., Kummerfeld, S.K., Teichmann, S.A., and Luscombe, N.M. (2009). A census of human transcription factors: function, expression and evolution. Nat. Rev. Genet. 10, 252–263. PubMedCrossrefGoogle Scholar
Wiley, C.D., Flynn, J.M., Morrissey, C., Lebofsky, R., Shuga, J., Dong, X., Unger, M.A., Vijg, J., Melov, S., Campisi, J. (2017). Analysis of individual cells identifies cell-to-cell variability following induction of cellular senescence. Aging Cell 16, 1043–1050. CrossrefPubMedGoogle Scholar
Yagi, M., Kishigami, S., Tanaka, A., Semi, K., Mizutani, E., Wakayama, S., Wakayama, T., Yamamoto, T., and Yamada, Y. (2017). Derivation of ground-state female ES cells maintaining gamete-derived DNA methylation. Nature 548, 224–227. PubMedCrossrefGoogle Scholar
About the article
Published Online: 2019-04-22
Published in Print: 2019-06-26