The transcription of DNA by RNA polymerase occurs as a discontinuous process described as transcriptional bursting. This bursting behavior is observed across species and has been quantified using various stochastic modeling approaches. There is a large body of evidence that suggests the bursts are actively modulated by transcriptional machinery and play a role in regulating developmental processes. Under a commonly used two-state model of transcription, various enhancer-, promoter- and chromatin microenvironment-associated features are found to differentially influence the size and frequency of bursting events – key parameters of the two-state model. Advancement of modeling and analysis tools has revealed that the simple two-state model and associated parameters may not sufficiently characterize the complex relationship between these features. The majority of experimental and modeling findings support the view of bursting as an evolutionarily conserved transcriptional control feature rather than an unintended byproduct of the transcription process. Stochastic transcriptional patterns contribute to enhanced cellular fitness and execution of proper development programs, which posit this mode of transcription as an important feature in developmental gene regulation. In this Review, we present compelling examples of the role of transcriptional bursting in development and explore the question of how stochastic transcription leads to deterministic organism development.

The transcription of DNA into RNA is a conserved and vital step in the process of gene expression across living organisms. To generate individual strands of RNA, RNA polymerases are loaded onto and traverse the DNA template. Development of key RNA assays has revealed transcription to be highly discontinuous in nature, exhibiting a behavior referred to as transcriptional bursting. Alternating periods of active and inactive promoter states lead to RNA production in this apparent bursting fashion. Early observations of discontinuous transcriptional activity were made in Drosophila melanogaster in 1979 and later in a macrophage cell line in 1994 (Miller and McKnight, 1979; Ross et al., 1994). These were later confirmed, both directly and indirectly, by the development of single-cell resolution RNA imaging assays (Blake et al., 2003; Chubb et al., 2006; Golding et al., 2005; Raj et al., 2006; Raser and O'Shea, 2004). The observation of transcriptional bursting across tissue types and species begs the question: what is the role of transcriptional bursting? Is it a feature of a complex gene expression framework, or simply a consequence of natural variabilities within biological systems? One physiologically relevant outcome of bursting is its contribution to heterogeneity of gene expression across a population of cells, especially for lowly expressed genes (Raj and van Oudenaarden, 2008; Rodriguez et al., 2018). Many have proposed that the resulting heterogeneity improves fitness of cell populations by making responses to varied external stimuli more robust (Carey et al., 2018; Cerulus et al., 2016; Levy et al., 2012; Thattai and van Oudenaarden, 2004; Yang et al., 2022). Others suggest that heterogeneity may aid in guiding cells down various differentiation pathways, thus enabling coordinated and appropriate development of organisms (Chang et al., 2008; Eldar and Elowitz, 2010; Fritzsch et al., 2018; Goolam et al., 2016; Hanna et al., 2009; Maamar et al., 2007). These developmental outcomes of heterogeneity – often a consequence of transcriptional bursting – suggest that bursting may be an evolutionarily selected process that encourages robust and coordinated organism development; however, the mechanisms and origins underlying these behaviors are not yet fully understood. In this Review, we aim to provide insights into the underlying causes and factors affecting transcriptional bursting, with an emphasis on how stochastic transcription leads to deterministic organism development.

Characterization of bursting

Until recently, the study of transcriptional regulation was largely focused on identifying and understanding the functions of key machinery necessary for the process; little emphasis was placed on analyzing transcriptional dynamics, in part, because of technological limitations. Observations of the discontinuous nature of transcription were made as early as 1979 when Miller chromatin spreads from Drosophila revealed periodic gaps in nascent transcripts along the DNA template (Miller and McKnight, 1979). These gaps were interpreted as periods of RNA polymerase II (Pol II) inactivity flanked by bursts of active transcription. Additional observations consistent with bursting were made in clonal macrophage cell lines, where it was found that only a small fraction of genetically identical cells actively expressed a GFP-tagged lacZ reporter. The active cell fraction could be modulated by external factors affecting promoter activity – and purportedly the probability of staying in active states – thus implying a bursting mode of active transcription interspersed with longer periods of inactivity (Ross et al., 1994). Although intriguing, direct study of this phenomenon required assays with single-cell resolution, which were not possible until the development of single-cell RNA techniques, such as single-molecule RNA fluorescence in situ hybridization (smFISH) and the MS2/MS2-coat protein (MCP) system (see Box 1) (Fig. 1).

Box 1. Single-cell RNA assays to measure transcriptional bursting

smFISH

smFISH is used to detect single copies of RNA in fixed cells and tissues via hybridization of fluorescent probes to target RNA molecules (Femino et al., 1998; Raj, 2013). Single-stranded probes containing complementary sequences to the target RNA are conjugated to fluorescent dyes and hybridized to the complementary target RNA. Treated samples are imaged under a fluorescence microscope and detected either directly or by signal amplification schemes (Fig. 1C). Observed signal counts and fluorescence intensities are correlated to the relative abundance of the targeted RNA species (Fig. 1D) (Raj, 2013).

MS2/MCP system

The bacteriophage-derived MS2/MCP system is a commonly used live imaging technique to visualize nascent transcripts by taking advantage of high-affinity binding between two components. The first is an RNA aptamer typically containing 12- to 128-repeat MS2 sequences that form secondary RNA hairpin structures upon transcription. The second is the uniformly expressed MCP fused with a fluorescent protein. Each MS2 stem loop recruits two copies of fluorescently labeled MCPs to the active transcription site. As Pol II traverses the template DNA and generates RNA strands, tens to hundreds of MCPs are localized to the stem loops at the active transcription site. Among the low-level fluorescence of diffused MCPs in the nucleus, these locally aggregated proteins form fluorescent puncta. As the nascent transcript is processed and exported to the cytoplasm, MCP-bound MS2 stem loops diffuse into the background. This enables real-time microscopy to capture stochastic transcriptional activity at the target locus with high temporal resolution (Fig. 1A,B) (Bertrand et al., 1998; Vera et al., 2019). Fluctuations in fluorescence intensity of the puncta localized to active sites can be interpreted as real-time relative transcriptional activity (Keller et al., 2020). Similarly functioning aptamers, such as PP7, have been concurrently implemented to enable simultaneous visualization of multiple genes (Larson et al., 2011).

Fig. 1.

Techniques for visualization of transcriptional bursting. (A) Schematic depicting the MS2/MCP-GFP system. MS2 stem loops are transcribed by RNA polymerase II (Pol II; blue) and bound by MCP-GFP fusion proteins, yielding a punctate fluorescent signal. (B) Annotated MS2 signal intensity trajectory from a Drosophila embryo and corresponding time series snapshots. Snapshot time points are indicated by red circles on the intensity trajectory. Transcriptionally active periods (shaded gray) are defined as those with positive slopes and inactive as those with negative slopes. Intensity trajectory is shown for the nucleus in the red dashed box. Adapted from Keller et al. (2020). Scale bar: 3 μm. (C) Schematic of single-molecule fluorescence in situ hybridization (smFISH) RNA probe recruitment to a target RNA sequence leading to concentrated fluorescent puncta. (D) smFISH image of Chinese hamster ovary cells with RNA probes targeting mRNA transcribed from the 7x-tetO transgene. Adapted from Raj et al. (2006). Scale bar: 5 μm. a.u., arbitrary units; GFP, green fluorescent protein; MCP, MS2 coat protein.

Fig. 1.

Techniques for visualization of transcriptional bursting. (A) Schematic depicting the MS2/MCP-GFP system. MS2 stem loops are transcribed by RNA polymerase II (Pol II; blue) and bound by MCP-GFP fusion proteins, yielding a punctate fluorescent signal. (B) Annotated MS2 signal intensity trajectory from a Drosophila embryo and corresponding time series snapshots. Snapshot time points are indicated by red circles on the intensity trajectory. Transcriptionally active periods (shaded gray) are defined as those with positive slopes and inactive as those with negative slopes. Intensity trajectory is shown for the nucleus in the red dashed box. Adapted from Keller et al. (2020). Scale bar: 3 μm. (C) Schematic of single-molecule fluorescence in situ hybridization (smFISH) RNA probe recruitment to a target RNA sequence leading to concentrated fluorescent puncta. (D) smFISH image of Chinese hamster ovary cells with RNA probes targeting mRNA transcribed from the 7x-tetO transgene. Adapted from Raj et al. (2006). Scale bar: 5 μm. a.u., arbitrary units; GFP, green fluorescent protein; MCP, MS2 coat protein.

Application of these high-resolution techniques uncovered the intrinsic heterogeneity of cell populations previously masked by bulk averaged data (Levsky et al., 2002; Newman et al., 2006; Raj and van Oudenaarden, 2008). smFISH experiments revealed significant cell-to-cell variation in RNA counts within an identical clonal population, implicating transcriptional bursting as a contributor to population heterogeneity (Fig. 1D) (Raj et al., 2006; Zenklusen et al., 2008). Furthermore, MS2/MCP-based imaging enabled real-time visualization of the temporal dynamics of transcription at a single locus and provided additional evidence to support the prevalence of bursting as a mode of transcriptional activity (Fig. 1B) (see Box 1) (Blake et al., 2003; Chubb et al., 2006; Golding et al., 2005; Raj et al., 2006; Raser and O'Shea, 2004).

Bursting is now believed to be a general feature of transcription that involves intermittent recruitment and release of Pol II at the promoter. Depending on the frequency and amount of Pol II loaded on the promoter, resulting transcriptional activity may appear to be continuous (high) or fluctuating (low) over time. When a large number of Pol II are frequently recruited to the promoter, a ‘continuous’ transcriptional trajectory is observed, often resulting in homogeneous expression. When bursts are infrequent, cell-to-cell variabilities in gene expression give rise to heterogeneous cell populations. In fact, many genes have demonstrated distinct discontinuous transcriptional activities that result in high degrees of cellular heterogeneities (Raj and van Oudenaarden, 2008; Rodriguez et al., 2018). The discontinuous and stochastic nature of bursting has led to quantitative analyses of bursting parameters to help identify key regulators of transcriptional bursting. A major remaining challenge is accurately interpreting highly variable transcription data to understand the origins and regulators of bursting.

Stochastic modeling of bursting

Much work has been done to optimize extraction of key functional parameters from both single-cell and multicellular datasets. The most simplistic and widely used method is fitting to a two-state telegraph model, which posits the promoter alternating between transcriptionally active (ON) and quiescent (OFF) states (Berrocal et al., 2020; Ko, 1991; Munsky et al., 2012; Peccoud and Ycart, 1995; Singh et al., 2013). The rate of promoter switching from inactive to active (kon), and vice versa (koff), as well as polymerase loading rate (r) can be approximated from the model. Binarized promoter states, combined with Pol II elongation and loading rate, can be used to reconstruct experimentally observed transcription trajectories, validating these parameters as descriptors of bursting (Lammers et al., 2020). The two-state model is fairly generalizable and can fit various experimental datasets. A challenge of applying the two-state model, even when a dataset has good statistical fit, is that output parameters may not be sufficiently complex to allow for in-depth analyses (Wang et al., 2020). For other datasets, the two-state model fails to recapitulate the experimental observations accurately (Bothma et al., 2014; Senecal et al., 2014).

Several modifications have been proposed to adapt the widely used two-state model to extract detailed kinetic information from more complex datasets. Some have included a third ‘refractory’ promoter state, during which the promoter is resistant to activation yet distinguishable from the ‘inactive’ state, where it is primed for activation (Li et al., 2018; Pimmett et al., 2021; Suter et al., 2011; Zoller et al., 2015). Others have used multiple intermediate states with increasing probabilities of activation (Lammers et al., 2020; Zuin et al., 2022). Rather than assuming a discrete number of promoter states, Corrigan et al. (2016) proposed a continuum model of transcriptional bursting in which a gene's activity level is defined by its promoter initiation rate and exists on a spectrum of activity states (Corrigan et al., 2016). Lastly, some studies have highlighted the limitations of equilibrium models (such as the two-state model) in describing a system that operates under the regulatory control of many stochastic processes with different time scales, and have proposed non-equilibrium models to approximate bursting behaviors more closely (Culyba et al., 2018; Li et al., 2018). For methods to process transcriptional trajectories for various models, see Box 2.

Box 2. Parametrization of transcriptional trajectories for model inputs

In general, the input to two-state models is a binary ON/OFF state of the promoters over time, where transcriptional trajectories are converted into binary active/inactive states (Tunnacliffe et al., 2018; Wang et al., 2020). One approach uses the change in signal intensity (trajectory slope) to determine active and inactive states: the interval between two bursts and the duration of each burst can be quantified as 1/kon and 1/koff, respectively (Fig. 1B) (Bothma et al., 2014; Fukaya et al., 2016). However, this method does not account for the time delay between transcriptional initiation, MS2 signal detection and the Pol II elongation time, and can provide unreliable measurement for short durations. Hidden Markov models (HMMs) are often used to infer hidden promoter states from observed transcriptional trajectories (Berrocal et al., 2020; Zechner et al., 2014). In MS2-based live imaging data, accumulated fluorescence intensity reflects current and recent promoter states as elongating Pol II activities are detected across time. For more systematic analyses, classical HMMs have been adapted to account for this ‘memory’ and allow accurate inference of promoter states (Berrocal et al., 2020; Zechner et al., 2014). Other studies used autocorrelations between a trajectory at times t and t+dt to estimate the rate of transcriptional initiation and the transcript dwell time (Coulon and Larson, 2016; Desponds et al., 2016; Larson et al., 2011). Autocorrelation provides reliable estimates across various cell cycle durations, and can be used to extract parameters for continuum models (Coulon and Larson, 2016; Desponds et al., 2016). Lastly, a machine learning-based method has been presented to deconvolute the transcriptional trajectory and infer Pol II positions within a gene sequence (Pimmett et al., 2021; Tantale et al., 2021). This method does not require a priori determination of the number of promoter states, but instead infers them from Pol II initiation events.

Regardless of the method employed to process trajectory data, the analysis of bursting is susceptible to the limited temporal resolution of imaging data. As burst frequency increases, activity peaks may merge together, an observation that may be mechanistically relevant (i.e. longer burst duration) or may represent merged signal peaks owing to changes in bursting behavior that occur at time scales shorter than the time resolution of the assay (Fritzsch et al., 2018; Wang et al., 2020). For example, higher transcriptional activity when the enhancer-promoter distance is 1 kb resulted in a greater number of merged peaks compared with transcriptional activity driven by an enhancer located 7.5 kb away from the promoter (Fukaya et al., 2016). The fluorescence intensity of a highly expressed gene stays above the background level throughout the cell cycle, making it difficult to infer bursting parameters. Additionally, intrinsic noise from microscopes often induces fluctuations in activity trajectories, which can be interpreted as bursting events under certain binarization conditions.

Although many published models with higher degrees of complexity exist, the two-state model remains the choice model of many authors presenting analyses of transcriptional bursting in their own model systems. Although the parameters set forth by the two-state model may fall short in fully capturing the complexity of factors regulating transcription, this model is still widely used in the field. As such, many examples discussed throughout this Review employed the parameters from the two-state model.

The premise of using any of the described models is that the extracted parameters are representative of the interplay between many processes involved in transcriptional regulation. For example, larger burst sizes can be represented as an increase in promoter loading rate, r, or an increase in ON state duration, 1/koff, which can imply more Pol II being recruited to the promoters. Similarly, more frequent bursts can be attributed to a variety of factors, including more frequent enhancer–promoter interactions, increased transcription factor (TF) binding rates and higher rates of nucleosome turnover (Donovan et al., 2019). In seeking a generalizable regulator, it was shown that transcriptional dynamics of multiple Drosophila gap genes with distinct bursting behaviors can be characterized with a single parameter, the mean promoter occupancy. This suggests that intricate regulatory processes can be attributed to a general regulatory principle (Zoller et al., 2018). In subsequent sections, we will describe how various model parameters of bursting characteristics can instruct our mechanistic understanding of the process and the role of bursting in development.

The interactions between regulatory DNA elements, transcription factors, the Mediator complex, chromatin modifiers and other players can all contribute to bursting dynamics (Fig. 2) (Lammers et al., 2020). Characterization of the differential roles each of these plays on transcriptional regulation has revealed a trend whereby promoter-associated factors often, but not always, influence burst size whereas enhancer-associated ones impact frequency. These observations are often correlative, and counter examples are also provided to demonstrate the influences of enhancer- or promoter-associated factors on other bursting parameters. The departure of these examples from the identified trend implies that the widely used two-state model parameters may not be sufficient to describe the complex underlying mechanisms that govern bursting behavior.

Fig. 2.

Factors that regulate transcriptional bursting parameters. (A) Examples of enhancer- and promoter-associated modulations shown to influence bursting characteristics. (B) Examples of microenvironment-associated features shown to influence bursting behavior. TF, transcription factor; SNP, single nucleotide polymorphism.

Fig. 2.

Factors that regulate transcriptional bursting parameters. (A) Examples of enhancer- and promoter-associated modulations shown to influence bursting characteristics. (B) Examples of microenvironment-associated features shown to influence bursting behavior. TF, transcription factor; SNP, single nucleotide polymorphism.

Promoter-associated impacts on transcriptional bursting

The promoter is a core regulatory element required for the transcription of a gene. It is responsible for the recruitment of transcriptional machinery and plays an important role in initiation and transition to elongation of Pol II (Cazier and Blazeck, 2021; Goodrich and Tjian, 2010). Evidence of promoter control of burst size is extensive across multiple model organisms. In the amoeba Dictyostelium discoideum, 17 individual copies of an identical, replicated actin gene can be found throughout the genome – differing only in promoter sequence and genomic context (Tunnacliffe et al., 2018). Actin mRNA expression levels throughout development are strikingly similar, but display significantly different transcriptional bursting behaviors. Live-cell imaging revealed that burst size is instructed by the promoter associated with each gene, with only minor contribution from genomic context (Tunnacliffe et al., 2018). Promoter diversification of this actin gene family, which drives differential bursting behavior, is a proposed evolutionary mechanism that expands the range of external stimuli to which essential genes can respond (Tunnacliffe et al., 2018). However, a recent paper on human bronchial epithelial cells showed that all genes exhibit similarly small burst sizes of a few transcripts owing to stochastic splicing activity within introns (Wan et al., 2021). This implies that promoter-associated control on burst size can be constrained by other transcriptional machinery.

Specific promoter motifs can also regulate bursting behaviors. In primary mouse fibroblasts, core promoters containing TATA motifs drive significantly larger burst sizes than those without (Larsson et al., 2019). Similarly, TATA-containing promoters in Drosophila embryos were found to drive long active transcriptional states with high rates of promoter initiation, and thus large burst sizes (Pimmett et al., 2021). Independently, initiator (INR) promoter elements drove highly paused, stochastic bursting and spent time in an additional refractory promoter state (Pimmett et al., 2021). In both Drosophila and mouse fibroblasts, INR elements in the presence of TATA motifs led to synergistic cooperation and increases in burst size (Larsson et al., 2019; Pimmett et al., 2021). Modeling this behavior with the continuum model found that mutation of TATA motifs reduced the rate of TF binding to the promoters at the initiating state (Corrigan et al., 2016). In contrast to these findings, a study conducted in HeLa cells found evidence of TATA-dependent effects on both burst size and frequency (Tantale et al., 2016). In Saccharomyces cerevisiae, it has been shown that TATA motifs are largely associated with stress-response genes and enable cells to execute a rapid response to stress stimuli, providing an evolutionary advantage (Blake et al., 2006; Huisinga and Pugh, 2004; Zanton and Pugh, 2004). Conversely, housekeeping genes typically do not contain the TATA motif and thus have limited accessibility to high-activity states (Blake et al., 2006; Huisinga and Pugh, 2004; Zanton and Pugh, 2004).

Transcriptional activity can also be tuned by modulating specific TF binding events at the promoter (Fig. 2A). A study on galactose-responsive genes in budding yeast found that the binding affinity of the TF Gal4 to the upstream activating sequence (UAS) site on the GAL3 promoter is directly correlated with transcriptional burst size and not frequency (Donovan et al., 2019). By applying a cross-correlation analysis to Gal4 binding and GAL10 transcription dynamics, it was also shown that Gal4 dwell time was positively correlated with GAL10 burst size (Donovan et al., 2019). Given the discrepancy in time scale of TF binding that occurs in the order of seconds, compared with the time scale of transcriptional bursts in minutes to hours, the correlation between TF binding and target gene transcription was surprising (Lammers et al., 2020). A similar analysis in a murine mammary carcinoma cell line found that binding of glucocorticoid receptor at the mouse mammary tumor virus promoter was similarly temporally correlated with transcriptional bursts and that the TF dwell time directly affected burst size (Stavreva et al., 2019). However, modulating the galactose signaling level in yeast affected burst frequency, indicating that not all promoter-associated factors affect burst size (Donovan et al., 2019). Collectively, these findings support a view whereby the promoter element is able to influence bursting characteristics via recruitment of transcriptional machinery and other regulatory factors in different concentrations and frequencies.

Enhancer-associated impacts on transcriptional bursting

The enhancer element, as a regulator of transcriptional initiation, guides the spatiotemporal character of transcriptional bursts, determining the context and frequency of gene expression (Long et al., 2016). Enhancer-mediated control of burst dynamics has been probed through modulation of TF-binding sites, enhancer deletions/swaps and ectopic expression of relevant TFs (Fig. 2A) (Antolović et al., 2017; Falo-Sanjuan et al., 2019; Fukaya et al., 2016; Hoppe et al., 2020; Lee et al., 2019). For example, one study found that homologous genes in primary mouse fibroblasts had distinct burst frequencies, despite having shared lineages. These differentially expressed genes were significantly enriched for those with enhancers containing single nucleotide polymorphisms (SNPs), a phenomenon that extends to orthologous genes in human fibroblasts (Larsson et al., 2019). Similarly, deletion of the enhancer for the β-globin gene resulted in a 99.7% reduction in the fraction of cells expressing the gene (serving as a proxy for bursting frequency) with only minor differences in burst size (Bartman et al., 2016). Additionally, in Drosophila, the swapping of an enhancer for another that drives stronger or weaker mRNA expression resulted in varying frequencies for the same reporter gene (Fukaya et al., 2016). The effects of an enhancer's strength were also tested in mouse embryonic stem cells (mESCs) by comparing the transcriptional output of eGFP-tagged Sox2 under the control of both an endogenous and an altered Sox2 control region (SCR) enhancer with half the TF-binding sites removed. The altered SCR construct exhibited a 50% decrease in transcriptional output, suggesting that enhancer strength dictates the promoter ON rate (Zuin et al., 2022). By modulating the frequency of Pol II loading and affecting transcriptional kinetics, enhancers control mRNA production at a given stage.

Enhancers also modulate bursting by regulating TFs; TF-binding sites in the enhancer work in concert with endogenous protein gradients to steer the frequency and location of genes' bursting, modulating the ON rate of the promoter (kon) to control expression spatiotemporally (Keller et al., 2020). TF binding motifs in enhancers can determine a gene's sensitivity to activator protein levels, as observed in experimental modification of SPS and CSL motifs, as well as the TF Dorsal motifs, in developing Drosophila embryos (Falo-Sanjuan et al., 2019; Keller et al., 2020). High levels of morphogens such as Decapentaplegic (Dpp) or Notch in Drosophila lead to earlier onset times and a greater amount of actively transcribing Pol II at target nuclei, generating a gradient of bursting behavior that correlates with the Dpp and Notch concentration (Falo-Sanjuan et al., 2019; Hoppe et al., 2020). In Drosophila embryos, ectopic expression of Dpp altered the expression pattern of target genes, expanding the fraction of nuclei actively expressing the gene encoding the zinc-finger protein U-shaped (Ush), without affecting total expression and time of peak transcription. However, the enhancer's effect on burst frequency may be limited as the region of the Drosophila embryo with the highest endogenous Dpp was unaffected by the ectopic expression (Hoppe et al., 2020). Hence, TF concentrations and their binding affinity to regulatory regions seem to mainly affect bursting frequencies.

Recent studies have also demonstrated the role of histone modifications in regulating transcriptional bursting. Genes regulated by highly acetylated (and thus accessible) enhancers, as determined by H3K27ac chromatin immunoprecipitation followed by sequencing (ChIP-seq), demonstrate a much higher burst frequency and subsequently higher mRNA expression level than genes with less-accessible enhancers in mouse fibroblast cells (Larsson et al., 2019). Similarly, histone modifications have been shown to determine burst characteristics in mouse neuronal genes (Chen et al., 2019). For example, increased acetylation of the Npas4 enhancer results in an increased bursting frequency of the gene, reaffirming previous findings. Strikingly, greater acetylation of the Fos enhancer increased burst duration, and its de-acetylation reduced burst frequency. This interesting finding reveals a more nuanced interaction between enhancers and bursting dynamics than a simple analog dial that modulates frequency (Chen et al., 2019).

Additional factors affecting bursting parameters

The complex nature of bursting is also regulated by factors beyond those dictating enhancer or promoter strength and accessibility (Fig. 2B). Several studies have linked the regulation of chromatin topology to changes in bursting character, although the degree to which it affects gene expression remains to be elucidated. The spatial dynamics of chromatin are hypothesized to be regulated by several families of proteins (e.g. boundary/insulator elements and remodelers, etc.) that participate in a complex series of feedback loops to demarcate regions of activity and inactivity known as topologically associating domains (TADs) (Luppino et al., 2020; Nichols and Corces, 2015; Ren et al., 2017). The role of TADs and 3D chromatin in determining bursting dynamics has been investigated through perturbation of chromatin structure by manipulating insulator binding and using paralogous models (wild-type and balancer flies with inverted or rearranged chromosomes). For example, a study of random transposition of the SCR found that transcriptionally active regions are enriched for enhancer–promoter contacts, and upon translocation of the SCR enhancer outside of its cognate TAD, the burst frequency of the target gene reduces to the same level as that of a gene with its enhancer removed completely (Zuin et al., 2022). Parameterization of the bursting data across the varied enhancer–TAD scenarios yielded a sigmoidal distribution of gene activation probability. This supports the conclusion that the TAD structure may be important for facilitating more frequent transient enhancer–promoter contacts, thus contributing to longer lasting regulatory states that govern gene activation (Zuin et al., 2022).

Other studies have found conflicting effects on bursting upon structural perturbation. For example, one study combined Hi-C, which detects chromatin interactions, and RNA sequencing to examine the interplay between chromatin conformation and transcriptional activity in Drosophila embryos. Here, changes to TAD structures significantly altered inter- and intra-TAD interactions but catalyzed few changes in gene expression (Ghavi-Helm et al., 2019). A similar study in Drosophila embryos found that chromatin structure and interactions were similar across tissue types (dorsal ectoderm, neural ectoderm and mesoderm), regardless of differential gene expression profiles (Ing-Simmons et al., 2021). This effect was especially interesting upon examination of the differential expression of Doc1, a gene that is exclusively expressed in dorsal ectoderm tissues, yet no change in chromatin insulation was observed in non-Doc1-expressing tissues.

Despite the lack of clear correlation between structure and genome-wide transcriptional activity, exceptions were found wherein some genes, and often pairs of genes, were found to rely on structure to determine their bursting profiles. Altered transcriptional bursting is mostly observed in a locus-dependent manner and provides evidence that complicates the perceived importance of chromatin structure. Using reporter constructs in Drosophila embryos, the effects of boundary formation on transcriptional bursting have also been reported (Yokoshi et al., 2020); when an enhancer–promoter pair is separated by about 9 kb, homie/nhomie insulator-mediated interactions increase both burst size and frequency. However, deletion of a homie element does not completely abolish bursting, suggesting that boundary elements augment intra-TAD enhancer–promoter interactions (Yokoshi et al., 2020). The presence of paired genes subject to structure-mediated transcription dynamics has also been observed in a study of paralogous genes in Drosophila embryos (Levo et al., 2022). Genes implicated in developmental patterning and segmentation, such as knirps-like and knirps, were found to share enhancers and interact across distances >100 kb. Chromatin interactions, facilitated by tethering elements, induced co-transcriptional bursting, which was abolished upon deletion of a tethering element-binding site. This interaction was observed across several other gene pairs and from up to 250 kb away, a distance on par with that of mammalian regulatory distances, providing evidence for the selective mediative effects of 3D organization (Levo et al., 2022). Specificity is conferred to chromatin looping by direction-specific interactions between insulators and boundary elements and seems to vary in importance from gene to gene.

The probabilistic nature of transcription has led to studies analyzing the effect of enhancer–promoter interactions and bursting activities. Live-imaging studies on enhancer–promoter dynamics suggest that, at least in the case of the pluripotency gene Sox2 and its distal enhancer SCR, there is little correlation between proximity and burst frequency (Alexander et al., 2019). By contrast, another study used smFISH on mESCs with randomly integrated SCR elements and determined that as enhancer–promoter distance increased – decreasing the probability of enhancer–promoter contacts – total transcription levels decreased, although the relationship between mRNA output and contact probability was non-linear (Zuin et al., 2022). The interplay between regulatory regions across space, time and genomic context is complex and highly dynamic, making it challenging to extend correlations to downstream bursting behaviors.

Recently, it was suggested that clustering of proteins bound to the enhancer or promoter region can also modulate bursting behavior (Fig. 2B). RNA Pol II was shown to form clusters at active transcription loci (Cho et al., 2016; Cisse et al., 2013), and later studies demonstrated that Pol II molecules form clusters with mediators and specific TFs to regulate target gene expression (Boija et al., 2018; Cho et al., 2018; Chong et al., 2018). Recent studies have used single-molecule nanoscopy to visualize distinct TF clusters at enhancers and their frequent interactions with the target promoters to form ‘superclusters’ that incorporate TFs and Pol II (Li et al., 2019; Li et al., 2020). As more studies report the role of TF clusters in gene regulation, many suggest that clustering also affects bursting characteristics.

In mouse embryonic fibroblasts, RNA Pol II cluster dwell time is correlated with burst size, and perturbation of Pol II clustering suppresses bursting, and removal of a clustering inhibitor re-induces transcriptional bursting (Cho et al., 2016). Using optogenetics-mediated induction of clustering, one study showed that interactions among the intrinsically disordered domain of p300 (EP300), a transcriptional co-activator, and TFs led to stabilization of the transcriptional complex. Here, co-aggregation between TFs and p300 resulted in a twofold increase in transcriptional bursting duration (Ma et al., 2021). Additionally, clustering of p300 protein facilitated the transactivation of its catalytic domain, which recruits histone acetylases and other co-activators such as bromodomain-containing protein 4 (Brd4) to the promoter, resulting in longer bursts (Ma et al., 2021). The impact of clustering was also observed at the physiological level of TF clustering; in mESCs, modulation of Brd4 cluster size using a small molecule inhibitor revealed a positive correlation between the cluster size and burst frequency (Li et al., 2019). Moreover, these Brd4 clusters drove coordinated transcriptional bursting from sister chromatids of Nanog, indicating that the clustered microenvironments can mediate coordinated transcription from two proximal promoters (Li et al., 2020). These results suggest that clustering of proteins, not only cis regulatory elements such as enhancers and promoters, can regulate bursting characteristics.

Despite the substantial research into regulators of transcriptional bursting, the question remains: is bursting a feature of gene and developmental regulation or simply a consequence of a stochastic process – or both under different conditions? The ubiquity of stochastic transcriptional programs across species and developmental times suggests the phenomenon is indeed a feature, not only a consequence (Larsson et al., 2019). Yet, there are cases in which heterogeneity of gene expression may lead to detrimental effects (Lehner, 2008). One way to reconcile the apparent dichotomy of heterogeneity as beneficial versus detrimental is by recognizing that various cellular functions may benefit differentially from expression variability. Bursting characteristics of transcription can be adjusted to induce more homogeneous or heterogeneous gene expression in a context-dependent manner, a crucial control for proper development (Fig. 3A). Although most published studies to date do not reveal direct links between bursting and development, we highlight the most compelling results that connect the two.

Fig. 3.

Implications of transcriptional bursting on development. (A) Highly discontinuous bursting trajectories result in high heterogeneity of gene expression across cell populations. (B) Example illustrating improved survival outcomes of cell populations with high expression heterogeneity upon heat stress. The schematic was generated based on findings from Levy et al. (2012). (C) Example illustrating that expression heterogeneity of Sox21 contributes to modulation of pluripotency versus differentiation character of cell lineages in the early mouse embryo. The schematic was generated based on findings from Goolam et al. (2016). ICM, inner cell mass.

Fig. 3.

Implications of transcriptional bursting on development. (A) Highly discontinuous bursting trajectories result in high heterogeneity of gene expression across cell populations. (B) Example illustrating improved survival outcomes of cell populations with high expression heterogeneity upon heat stress. The schematic was generated based on findings from Levy et al. (2012). (C) Example illustrating that expression heterogeneity of Sox21 contributes to modulation of pluripotency versus differentiation character of cell lineages in the early mouse embryo. The schematic was generated based on findings from Goolam et al. (2016). ICM, inner cell mass.

Environmental response

The importance of cellular heterogeneity in healthy organism development has been well established. An assembly of single cells at any stage of development has been shown to have improved fitness and robust response to environmental stimuli with increasing degrees of heterogeneity (Carey et al., 2018; Cerulus et al., 2016; Thattai and van Oudenaarden, 2004; Yang et al., 2022). For example, in S. cerevisiae, increased heterogeneity of Tsl1 expression resulted in improved survival outcomes upon heat stress pressures (Fig. 3B) (Levy et al., 2012). Additionally, synthetically generated Escherichia coli promoters resulted in significantly lower transcriptional noise levels compared with native ones (Wolf et al., 2015). This suggests that evolutionary selection pressures drove increased promoter-driven expression variability as a fitness-endowing feature (Wolf et al., 2015). In studying the phenomenon of bursting transcription, there is abundant evidence demonstrating how bursting contributes to diversification of cell populations. For example, the 17 replicated actin genes in Dictyostelium are activated under varying environmental conditions, throughout different points in developmental time and, notably, each has unique bursting features. This careful switching is proposed to provide a wider range of stimuli to which the organism can safely respond (Tunnacliffe et al., 2018). In Drosophila mesectoderm cells, genes respond differentially to Notch signals; some display synchronized and sustained transcription under some cellular conditions, whereas others exhibit stochastic bursting profiles (Falo-Sanjuan et al., 2019). The ability to toggle between sustained and stochastic responses is thought to confer fitness by priming enhancers to respond in a synchronous manner to Notch signaling during key developmental stages. Conversely, a lower and hence more stochastic expression at other stages buffers signal fluctuations arising from changing environmental conditions (Falo-Sanjuan et al., 2019). Similar effects have been reported in various species for genes driven by TATA motif-containing promoters, which provide enhanced response speeds to a range of stress conditions. Colonies of wild-type S. cerevisiae with a TATA-containing promoter driving the zeocin resistance gene exhibit faster stress responses and outcompete colonies with a mutated TATA motif. The intact TATA motif confers a more stochastic transcriptional profile and an ability to thrive under increased stress compared with its mutated counterpart (Blake et al., 2006). In all, robust responses to acute stress or naturally varying conditions, and the resulting expression heterogeneity, seem to be a beneficial feature of transcriptional bursting, helping to support precise and coordinated gene expression throughout organism development.

Differentiation

Transcriptional bursting has been implicated in cell decision making and differentiation across cell types. A study in Dictyostelium revealed that genes crucial for differentiation exhibited significantly more variable expression in the time preceding cell fate decisions (Antolović et al., 2017). This observation extended to other genes involved in acute stress responses and led to the hypothesis that expression variability may direct cells toward varying differentiation pathways to produce subpopulations that are better equipped to respond to the changing cellular environment (Antolović et al., 2017). The up- and downregulation of particular genes at developmental junctures also exists in more complex multicellular organisms. In mouse embryos, Sox21 mRNA expression [a target of Oct4 (Pou5f1) and Sox2] is highly variable at the four-cell state and this heterogeneity is implicated in the initiation of cell fate decisions by modulating the balance of pluripotency and differentiation of the downstream cell lineages (Fig. 3C) (Goolam et al., 2016). Coordinated bursting has also been observed between alleles in daughter cells up to 6 h after division, suggesting a level of inheritability of transcriptional bursting programs that may play a role in cell fate decisions (Fritzsch et al., 2018).

Evidence supporting the beneficial effects of heterogeneous gene expression is abundant; however, there are instances in which increased heterogeneity is linked to detrimental or undesirable effects at the organism level. One such example is the increased expression heterogeneity of circadian-related genes with aging in mice (Wolff et al., 2023). Although the number of rhythmically expressed genes (REGs) decreases with age, the expression variability of the remaining REGs increases significantly in the hypothalamus, lung, kidney and muscle tissues. This dysregulation is associated with the aging physiology of the affected tissues (Wolff et al., 2023). Similar bias away from heterogeneity was observed in S. cerevisiae, in which expression noise analysis implied that dosage-sensitive genes have been evolutionarily tuned to minimize expression variability (Lehner, 2008). These genes are often found to be highly and continuously expressed – a proposed ‘simple mechanism’ to reduce noise (Lehner, 2008). In these cases, increased heterogeneity produced undesirable outcomes, which is in contrast to the evidence of heterogeneous expression as a fitness-endowing feature. We reconcile these opposing effects by acknowledging that bursting behaviors can be adjusted to drive strong gene expression (resulting in more homogeneous expression) that is beneficial for these specific cellular functions.

Transcriptional analysis of key genes across developmental stages provides additional evidence for the role of bursting heterogeneity in differentiation. A study of transcriptional bursting of an α-hemoglobin gene in mouse erythroblasts undergoing erythropoiesis found extensive heterogeneity over the course of differentiation (Jeziorska et al., 2022). Cells were partitioned into groups of varying degrees of bursting intensity and frequency. Within one such group, with an assumption that there is no buffering of the initial transcriptional noise by post-transcriptional mechanisms, the initiation and termination of erythropoiesis were characterized by more frequent bursting and expression variability. Meanwhile, the intermediate stage was marked by higher and more continuous levels of transcription with more homogeneous expression profiles (Jeziorska et al., 2022). In all, these reports support a view whereby transcriptional bursting modulation and the resulting expression is a tightly regulated biological process. Expression heterogeneity, or lack thereof, has been shown to play important roles in maintaining proper development programs and contributing to the robustness of cellular populations and organisms to varied stimuli.

In this Review, we have shown that various enhancer-, promoter- and microenvironment- associated factors differentially influence the characteristics of transcriptional bursting. We have provided evidence that suggests this bursting behavior is an evolutionarily conserved and important feature of gene regulation. Additional evidence has demonstrated that bursting is implicated in regulating gene expression heterogeneity (or homogeneity) across cell populations and tissues over the course of developmental timelines. The analyses included here are presented in the context of the parameters and variables set forth by the widely used two-state model of transcription; however, the complex contributions of various factors affecting bursting may not be well recapitulated. The development of more advanced models has begun to shed light on the nuanced features of transcription regulation, although there still remain many unanswered questions on the exact mechanisms of transcriptional bursting. Moreover, there is increasing evidence supporting an important role for transcription in maintaining proper development and differentiation programs. As the field continues to advance, further experimentation utilizing advanced techniques such as genome editing, optogenetics and targeted genomics, as well as more advanced modeling, will be necessary to elucidate the mechanisms underlying transcriptional bursting regulation.

We thank Lim lab members for helpful discussions and feedback on the manuscript.

Funding

E.A.L.P. is partially funded through the University of Pennsylvania Fontaine Society. This work is supported by the National Institutes of Health (R35GM133425 awarded to B.L.). Deposited in PMC for release after 12 months.

Alexander
,
J. M.
,
Guan
,
J.
,
Li
,
B.
,
Maliskova
,
L.
,
Song
,
M.
,
Shen
,
Y.
,
Huang
,
B.
,
Lomvardas
,
S.
and
Weiner
,
O. D.
(
2019
).
Live-cell imaging reveals enhancer-dependent Sox2 transcription in the absence of enhancer proximity
.
eLife
8
,
e41769
.
Antolović
,
V.
,
Miermont
,
A.
,
Corrigan
,
A. M.
and
Chubb
,
J. R.
(
2017
).
Generation of single-cell transcript variability by repression
.
Curr. Biol.
27
,
1811
-
1817.e3
.
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
.
Berrocal
,
A.
,
Lammers
,
N. C.
,
Garcia
,
H. G.
and
Eisen
,
M. B.
(
2020
).
Kinetic sculpting of the seven stripes of the Drosophila even-skipped gene
.
eLife
9
,
e61635
.
Bertrand
,
E.
,
Chartrand
,
P.
,
Schaefer
,
M.
,
Shenoy
,
S. M.
,
Singer
,
R. H.
and
Long
,
R. M.
(
1998
).
Localization of ASH1 mRNA particles in living yeast
.
Mol. Cell
2
,
437
-
445
.
Blake
,
W. J.
,
Kærn
,
M.
,
Cantor
,
C. R.
and
Collins
,
J. J.
(
2003
).
Noise in eukaryotic gene expression
.
Nature
422
,
633
-
637
.
Blake
,
W. J.
,
Balázsi
,
G.
,
Kohanski
,
M. A.
,
Isaacs
,
F. J.
,
Murphy
,
K. F.
,
Kuang
,
Y.
,
Cantor
,
C. R.
,
Walt
,
D. R.
and
Collins
,
J. J.
(
2006
).
Phenotypic consequences of promoter-mediated transcriptional noise
.
Mol. Cell
24
,
853
-
865
.
Boija
,
A.
,
Klein
,
I. A.
,
Sabari
,
B. R.
,
Dall'Agnese
,
A.
,
Coffey
,
E. L.
,
Zamudio
,
A. V.
,
Li
,
C. H.
,
Shrinivas
,
K.
,
Manteiga
,
J. C.
,
Hannett
,
N. M.
et al. 
(
2018
).
Transcription factors activate genes through the phase-separation capacity of their activation domains
.
Cell
175
,
1842
-
1855.e16
.
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
.
Carey
,
J. N.
,
Mettert
,
E. L.
,
Roggiani
,
M.
,
Myers
,
K. S.
,
Kiley
,
P. J.
and
Goulian
,
M.
(
2018
).
Regulated stochasticity in a bacterial signaling network permits tolerance to a rapid environmental change
.
Cell
173
,
196
-
207.e14
.
Cazier
,
A. P.
and
Blazeck
,
J.
(
2021
).
Advances in promoter engineering: novel applications and predefined transcriptional control
.
Biotechnol. J.
16
,
2100239
.
Cerulus
,
B.
,
New
,
A. M.
,
Pougach
,
K.
and
Verstrepen
,
K. J.
(
2016
).
Noise and epigenetic inheritance of single-cell division times influence population fitness
.
Curr. Biol.
26
,
1138
-
1147
.
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
.
Chen
,
L.-F.
,
Lin
,
Y. T.
,
Gallegos
,
D. A.
,
Hazlett
,
M. F.
,
Gómez-Schiavon
,
M.
,
Yang
,
M. G.
,
Kalmeta
,
B.
,
Zhou
,
A. S.
,
Holtzman
,
L.
,
Gersbach
,
C. A.
et al. 
(
2019
).
Enhancer histone acetylation modulates transcriptional bursting dynamics of neuronal activity-inducible genes
.
Cell Rep.
26
,
1174
-
1188.e5
.
Cho
,
W.-K.
,
Jayanth
,
N.
,
English
,
B. P.
,
Inoue
,
T.
,
Andrews
,
J. O.
,
Conway
,
W.
,
Grimm
,
J. B.
,
Spille
,
J.-H.
,
Lavis
,
L. D.
,
Lionnet
,
T.
et al. 
(
2016
).
RNA Polymerase II cluster dynamics predict mRNA output in living cells
.
eLife
5
,
e13617
.
Cho
,
W.-K.
,
Spille
,
J.-H.
,
Hecht
,
M.
,
Lee
,
C.
,
Li
,
C.
,
Grube
,
V.
and
Cisse
,
I. I.
(
2018
).
Mediator and RNA polymerase II clusters associate in transcription-dependent condensates
.
Science
361
,
412
-
415
.
Chong
,
S.
,
Dugast-Darzacq
,
C.
,
Liu
,
Z.
,
Dong
,
P.
,
Dailey
,
G. M.
,
Cattoglio
,
C.
,
Heckert
,
A.
,
Banala
,
S.
,
Lavis
,
L.
,
Darzacq
,
X.
et al. 
(
2018
).
Imaging dynamic and selective low-complexity domain interactions that control gene transcription
.
Science
361
,
eaar2555
.
Chubb
,
J. R.
,
Trcek
,
T.
,
Shenoy
,
S. M.
and
Singer
,
R. H.
(
2006
).
Transcriptional pulsing of a developmental gene
.
Curr. Biol.
16
,
1018
-
1025
.
Cisse
,
I. I.
,
Izeddin
,
I.
,
Causse
,
S. Z.
,
Boudarene
,
L.
,
Senecal
,
A.
,
Muresan
,
L.
,
Dugast-Darzacq
,
C.
,
Hajj
,
B.
,
Dahan
,
M.
and
Darzacq
,
X.
(
2013
).
Real-time dynamics of RNA polymerase II clustering in live human cells
.
Science
341
,
664
-
667
.
Corrigan
,
A. M.
,
Tunnacliffe
,
E.
,
Cannon
,
D.
and
Chubb
,
J. R.
(
2016
).
A continuum model of transcriptional bursting
.
eLife
5
,
e13051
.
Coulon
,
A.
and
Larson
,
D. R.
(
2016
).
Chapter seven - Fluctuation analysis: dissecting transcriptional kinetics with signal theory
. In
Methods in Enzymology
(ed.
G. S.
Filonov
and
S. R.
Jaffrey
), pp.
159
-
191
.
Academic Press
.
Culyba
,
M. J.
,
Kubiak
,
J. M.
,
Mo
,
C. Y.
,
Goulian
,
M.
and
Kohli
,
R. M.
(
2018
).
Non-equilibrium repressor binding kinetics link DNA damage dose to transcriptional timing within the SOS gene network
.
PLoS Genet.
14
,
e1007405
.
Desponds
,
J.
,
Tran
,
H.
,
Ferraro
,
T.
,
Lucas
,
T.
,
Romero
,
C. P.
,
Guillou
,
A.
,
Fradin
,
C.
,
Coppey
,
M.
,
Dostatni
,
N.
and
Walczak
,
A. M.
(
2016
).
Precision of readout at the hunchback gene: analyzing short transcription time traces in living fly embryos
.
PLOS Comput. Biol.
12
,
e1005256
.
Donovan
,
B. T.
,
Huynh
,
A.
,
Ball
,
D. A.
,
Patel
,
H. P.
,
Poirier
,
M. G.
,
Larson
,
D. R.
,
Ferguson
,
M. L.
and
Lenstra
,
T. L.
(
2019
).
Live-cell imaging reveals the interplay between transcription factors, nucleosomes, and bursting
.
EMBO J.
38
,
e100809
.
Eldar
,
A.
and
Elowitz
,
M. B.
(
2010
).
Functional roles for noise in genetic circuits
.
Nature
467
,
167
-
173
.
Falo-Sanjuan
,
J.
,
Lammers
,
N. C.
,
Garcia
,
H. G.
and
Bray
,
S. J.
(
2019
).
Enhancer priming enables fast and sustained transcriptional responses to Notch signaling
.
Dev. Cell
50
,
411
-
425.e8
.
Femino
,
A. M.
,
Fay
,
F. S.
,
Fogarty
,
K.
and
Singer
,
R. H.
(
1998
).
Visualization of single RNA transcripts in Situ
.
Science
280
,
585
-
590
.
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
.
Fukaya
,
T.
,
Lim
,
B.
and
Levine
,
M.
(
2016
).
Enhancer control of transcriptional bursting
.
Cell
166
,
358
-
368
.
Ghavi-Helm
,
Y.
,
Jankowski
,
A.
,
Meiers
,
S.
,
Viales
,
R. R.
,
Korbel
,
J. O.
and
Furlong
,
E. E. M.
(
2019
).
Highly rearranged chromosomes reveal uncoupling between genome topology and gene expression
.
Nat. Genet.
51
,
1272
-
1282
.
Golding
,
I.
,
Paulsson
,
J.
,
Zawilski
,
S. M.
and
Cox
,
E. C.
(
2005
).
Real-time kinetics of gene activity in individual bacteria
.
Cell
123
,
1025
-
1036
.
Goodrich
,
J. A.
and
Tjian
,
R.
(
2010
).
Unexpected roles for core promoter recognition factors in cell-type-specific transcription and gene regulation
.
Nat. Rev. Genet.
11
,
549
-
558
.
Goolam
,
M.
,
Scialdone
,
A.
,
Graham
,
S. J. L.
,
Macaulay
,
I. C.
,
Jedrusik
,
A.
,
Hupalowska
,
A.
,
Voet
,
T.
,
Marioni
,
J. C.
and
Zernicka-Goetz
,
M.
(
2016
).
Heterogeneity in Oct4 and Sox2 targets biases cell fate in 4-cell mouse embryos
.
Cell
165
,
61
-
74
.
Hanna
,
J.
,
Saha
,
K.
,
Pando
,
B.
,
Van Zon
,
J.
,
Lengner
,
C. J.
,
Creyghton
,
M. P.
,
Van Oudenaarden
,
A.
and
Jaenisch
,
R.
(
2009
).
Direct cell reprogramming is a stochastic process amenable to acceleration
.
Nature
462
,
595
-
601
.
Hoppe
,
C.
,
Bowles
,
J. R.
,
Minchington
,
T. G.
,
Sutcliffe
,
C.
,
Upadhyai
,
P.
,
Rattray
,
M.
and
Ashe
,
H. L.
(
2020
).
Modulation of the promoter activation rate dictates the transcriptional response to graded BMP signaling levels in the Drosophila embryo
.
Dev. Cell
54
,
727
-
741.e7
.
Huisinga
,
K. L.
and
Pugh
,
B. F.
(
2004
).
A genome-wide housekeeping role for TFIID and a highly regulated stress-related role for SAGA in Saccharomyces cerevisiae
.
Mol. Cell
13
,
573
-
585
.
Ing-Simmons
,
E.
,
Vaid
,
R.
,
Bing
,
X. Y.
,
Levine
,
M.
,
Mannervik
,
M.
and
Vaquerizas
,
J. M.
(
2021
).
Independence of chromatin conformation and gene regulation during Drosophila dorsoventral patterning
.
Nat. Genet.
53
,
487
-
499
.
Jeziorska
,
D. M.
,
Tunnacliffe
,
E. A. J.
,
Brown
,
J. M.
,
Ayyub
,
H.
,
Sloane-Stanley
,
J.
,
Sharpe
,
J. A.
,
Lagerholm
,
B. C.
,
Babbs
,
C.
,
Smith
,
A. J. H.
,
Buckle
,
V. J.
et al. 
(
2022
).
On-microscope staging of live cells reveals changes in the dynamics of transcriptional bursting during differentiation
.
Nat. Commun.
13
,
6641
.
Keller
,
S. H.
,
Jena
,
S. G.
,
Yamazaki
,
Y.
and
Lim
,
B.
(
2020
).
Regulation of spatiotemporal limits of developmental gene expression via enhancer grammar
.
Proc. Natl. Acad. Sci. USA
117
,
15096
-
15103
.
Ko
,
M. S. H.
(
1991
).
A stochastic model for gene induction
.
J. Theor. Biol.
153
,
181
-
194
.
Lammers
,
N. C.
,
Kim
,
Y. J.
,
Zhao
,
J.
and
Garcia
,
H. G.
(
2020
).
A matter of time: using dynamics and theory to uncover mechanisms of transcriptional bursting
.
Curr. Opin. Cell Biol.
67
,
147
-
157
.
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
.
Larsson
,
A. J. M.
,
Johnsson
,
P.
,
Hagemann-Jensen
,
M.
,
Hartmanis
,
L.
,
Faridani
,
O. R.
,
Reinius
,
B.
,
Segerstolpe
,
Å.
,
Rivera
,
C. M.
,
Ren
,
B.
and
Sandberg
,
R.
(
2019
).
Genomic encoding of transcriptional burst kinetics
.
Nature
565
,
251
-
254
.
Lee
,
C.
,
Shin
,
H.
and
Kimble
,
J.
(
2019
).
Dynamics of Notch-dependent transcriptional bursting in its native context
.
Dev. Cell
50
,
426
-
435.e4
.
Lehner
,
B.
(
2008
).
Selection to minimise noise in living systems and its implications for the evolution of gene expression
.
Mol. Syst. Biol.
4
,
170
.
Levo
,
M.
,
Raimundo
,
J.
,
Bing
,
X. Y.
,
Sisco
,
Z.
,
Batut
,
P. J.
,
Ryabichko
,
S.
,
Gregor
,
T.
and
Levine
,
M. S.
(
2022
).
Transcriptional coupling of distant regulatory genes in living embryos
.
Nature
605
,
754
-
760
.
Levsky
,
J. M.
,
Shenoy
,
S. M.
,
Pezo
,
R. C.
and
Singer
,
R. H.
(
2002
).
Single-Cell Gene Expression Profiling
.
Science
297
,
836
-
840
.
Levy
,
S. F.
,
Ziv
,
N.
and
Siegal
,
M. L.
(
2012
).
Bet hedging in yeast by heterogeneous, age-correlated expression of a stress protectant
.
PLoS Biol.
10
,
e1001325
.
Li
,
C.
,
Cesbron
,
F.
,
Oehler
,
M.
,
Brunner
,
M.
and
Höfer
,
T.
(
2018
).
Frequency modulation of transcriptional bursting enables sensitive and rapid gene regulation
.
Cell Syst.
6
,
409
-
423
.e11.
Li
,
J.
,
Dong
,
A.
,
Saydaminova
,
K.
,
Chang
,
H.
,
Wang
,
G.
,
Ochiai
,
H.
,
Yamamoto
,
T.
and
Pertsinidis
,
A.
(
2019
).
Single-molecule nanoscopy elucidates RNA polymerase II transcription at single genes in live cells
.
Cell
178
,
491
-
506.e28
.
Li
,
J.
,
Hsu
,
A.
,
Hua
,
Y.
,
Wang
,
G.
,
Cheng
,
L.
,
Ochiai
,
H.
,
Yamamoto
,
T.
and
Pertsinidis
,
A.
(
2020
).
Single-gene imaging links genome topology, promoter–enhancer communication and transcription control
.
Nat. Struct. Mol. Biol.
27
,
1032
-
1040
.
Long
,
H. K.
,
Prescott
,
S. L.
and
Wysocka
,
J.
(
2016
).
Ever-changing landscapes: Transcriptional enhancers in development and evolution
.
Cell
167
,
1170
-
1187
.
Luppino
,
J. M.
,
Park
,
D. S.
,
Nguyen
,
S. C.
,
Lan
,
Y.
,
Xu
,
Z.
,
Yunker
,
R.
and
Joyce
,
E. F.
(
2020
).
Cohesin promotes stochastic domain intermingling to ensure proper regulation of boundary-proximal genes
.
Nat. Genet.
52
,
840
-
848
.
Ma
,
L.
,
Gao
,
Z.
,
Wu
,
J.
,
Zhong
,
B.
,
Xie
,
Y.
,
Huang
,
W.
and
Lin
,
Y.
(
2021
).
Co-condensation between transcription factor and coactivator p300 modulates transcriptional bursting kinetics
.
Mol. Cell
81
,
1682
-
1697.e7
.
Maamar
,
H.
,
Raj
,
A.
and
Dubnau
,
D.
(
2007
).
Noise in gene expression determines cell fate in bacillus subtilis
.
Science
317
,
526
-
529
.
Miller
,
O. L.
and
Mcknight
,
S. L.
(
1979
).
Post-replicative nonribosomal transcription units in D. melanogaster embryos
.
Cell
17
,
551
-
563
.
Munsky
,
B.
,
Neuert
,
G.
and
Van Oudenaarden
,
A.
(
2012
).
Using gene expression noise to understand gene regulation
.
Science
336
,
183
-
187
.
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
.
Nichols
,
M. H.
and
Corces
,
V. G.
(
2015
).
A CTCF code for 3D genome architecture
.
Cell
162
,
703
-
705
.
Peccoud
,
J.
and
Ycart
,
B.
(
1995
).
Markovian modeling of gene-product synthesis
.
Theor. Popul. Biol.
48
,
222
-
234
.
Pimmett
,
V. L.
,
Dejean
,
M.
,
Fernandez
,
C.
,
Trullo
,
A.
,
Bertrand
,
E.
,
Radulescu
,
O.
and
Lagha
,
M.
(
2021
).
Quantitative imaging of transcription in living Drosophila embryos reveals the impact of core promoter motifs on promoter state dynamics
.
Nat. Commun.
12
,
4504
.
Raj
,
A.
(
2013
).
Single-Molecule RNA FISH
. In
Encyclopedia of Biophysics
(ed.
G. C. K.
Roberts
), pp.
2340
-
2343
.
Berlin, Heidelberg
:
Springer
.
Raj
,
A.
and
Van Oudenaarden
,
A.
(
2008
).
Nature, nurture, or chance: stochastic gene expression and its consequences
.
Cell
135
,
216
-
226
.
Raj
,
A.
,
Peskin
,
C. S.
,
Tranchina
,
D.
,
Vargas
,
D. Y.
and
Tyagi
,
S.
(
2006
).
Stochastic mRNA synthesis in mammalian cells
.
PLoS Biol.
4
,
e309
.
Raser
,
J. M.
and
O'Shea
,
E. K.
(
2004
).
Control of stochasticity in eukaryotic gene expression
.
Science
304
,
1811
-
1814
.
Ren
,
G.
,
Jin
,
W.
,
Cui
,
K.
,
Rodrigez
,
J.
,
Hu
,
G.
,
Zhang
,
Z.
,
Larson
,
D. R.
and
Zhao
,
K.
(
2017
).
CTCF-mediated enhancer-promoter interaction is a critical regulator of cell-to-cell variation of gene expression
.
Mol. Cell
67
,
1049
-
1058.e6
.
Rodriguez
,
J.
,
Ren
,
G.
,
Day
,
C. R.
,
Zhao
,
K.
,
Chow
,
C. C.
and
Larson
,
D. R.
(
2018
).
Intrinsic dynamics of a human gene reveal the basis of expression heterogeneity
.
Cell
176
,
213
-
226.e18
.
Ross
,
I. L.
,
Browne
,
C. M.
and
Hume
,
D. A.
(
1994
).
Transcription of individual genes in eukaryotic cells occurs randomly and infrequently
.
Immunol. Cell Biol.
72
,
177
-
185
.
Senecal
,
A.
,
Munsky
,
B.
,
Proux
,
F.
,
Ly
,
N.
,
Braye
,
F. E.
,
Zimmer
,
C.
,
Mueller
,
F.
and
Darzacq
,
X.
(
2014
).
Transcription factors modulate c-Fos transcriptional bursts
.
Cell Rep.
8
,
75
-
83
.
Singh
,
A.
,
Vargas
,
C. A.
and
Karmakar
,
R.
(
2013
).
Stochastic analysis and inference of a two-state genetic promoter model
. In
2013 American Control Conference
, pp,
4563
-
4568
.
Stavreva
,
D. A.
,
Garcia
,
D. A.
,
Fettweis
,
G.
,
Gudla
,
P. R.
,
Zaki
,
G. F.
,
Soni
,
V.
,
Mcgowan
,
A.
,
Williams
,
G.
,
Huynh
,
A.
,
Palangat
,
M.
et al. 
(
2019
).
Transcriptional bursting and co-bursting regulation by steroid hormone release pattern and transcription factor mobility
.
Mol. Cell
75
,
1161
-
1177.e11
.
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
.
Tantale
,
K.
,
Mueller
,
F.
,
Kozulic-Pirher
,
A.
,
Lesne
,
A.
,
Victor
,
J.-M.
,
Robert
,
M.-C.
,
Capozi
,
S.
,
Chouaib
,
R.
,
Bäcker
,
V.
,
Mateos-Langerak
,
J.
et al. 
(
2016
).
A single-molecule view of transcription reveals convoys of RNA polymerases and multi-scale bursting
.
Nat. Commun.
7
,
12248
.
Tantale
,
K.
,
Garcia-Oliver
,
E.
,
Robert
,
M.-C.
,
L'Hostis
,
A.
,
Yang
,
Y.
,
Tsanov
,
N.
,
Topno
,
R.
,
Gostan
,
T.
,
Kozulic-Pirher
,
A.
,
Basu-Shrivastava
,
M.
et al. 
(
2021
).
Stochastic pausing at latent HIV-1 promoters generates transcriptional bursting
.
Nat. Commun.
12
,
4503
.
Thattai
,
M.
and
Van Oudenaarden
,
A.
(
2004
).
Stochastic gene expression in fluctuating environments
.
Genetics
167
,
523
-
530
.
Tunnacliffe
,
E.
,
Corrigan
,
A. M.
and
Chubb
,
J. R.
(
2018
).
Promoter-mediated diversification of transcriptional bursting dynamics following gene duplication
.
Proc. Natl. Acad. Sci. USA
115
,
8364
-
8369
.
Vera
,
M.
,
Tutucci
,
E.
and
Singer
,
R. H.
(
2019
).
Imaging single mRNA molecules in mammalian cells using an optimized MS2-MCP system
. In
Imaging Gene Expression: Methods and Protocols
(ed.
Y.
Shav-Tal
), pp.
3
-
20
.
New York, NY
:
Springer
.
Wan
,
Y.
,
Anastasakis
,
D. G.
,
Rodriguez
,
J.
,
Palangat
,
M.
,
Gudla
,
P.
,
Zaki
,
G.
,
Tandon
,
M.
,
Pegoraro
,
G.
,
Chow
,
C. C.
,
Hafner
,
M.
et al. 
(
2021
).
Dynamic imaging of nascent RNA reveals general principles of transcription dynamics and stochastic splice site selection
.
Cell
184
,
2878
-
2895.e20
.
Wang
,
Y.
,
Qi
,
J.
,
Shao
,
J.
and
Tang
,
X.-Q.
(
2020
).
Signaling mechanism of transcriptional bursting: a technical resolution-independent study
.
Biology
9
,
339
.
Wolf
,
L.
,
Silander
,
O. K.
and
Van Nimwegen
,
E.
(
2015
).
Expression noise facilitates the evolution of gene regulation
.
eLife
4
,
e05856
.
Wolff
,
C. A.
,
Gutierrez-Monreal
,
M. A.
,
Meng
,
L.
,
Zhang
,
X.
,
Douma
,
L. G.
,
Costello
,
H. M.
,
Douglas
,
C. M.
,
Ebrahimi
,
E.
,
Pham
,
A.
,
Oliveira
,
A. C.
et al. 
(
2023
).
Defining the age-dependent and tissue-specific circadian transcriptome in male mice
.
Cell Rep.
42
,
111982
.
Yang
,
X.
,
Luo
,
S.
,
Zhang
,
Z.
,
Wang
,
Z.
,
Zhou
,
T.
and
Zhang
,
J.
(
2022
).
Silent transcription intervals and translational bursting lead to diverse phenotypic switching
.
Phys. Chem. Chem. Phys.
24
,
26600
-
26608
.
Yokoshi
,
M.
,
Segawa
,
K.
and
Fukaya
,
T.
(
2020
).
Visualizing the role of boundary elements in enhancer-promoter communication
.
Mol. Cell
78
,
224
-
235.e5
.
Zanton
,
S. J.
and
Pugh
,
B. F.
(
2004
).
Changes in genomewide occupancy of core transcriptional regulators during heat stress
.
Proc. Natl. Acad. Sci. USA
101
,
16843
-
16848
.
Zechner
,
C.
,
Unger
,
M.
,
Pelet
,
S.
,
Peter
,
M.
and
Koeppl
,
H.
(
2014
).
Scalable inference of heterogeneous reaction kinetics from pooled single-cell recordings
.
Nat. Methods
11
,
197
-
202
.
Zenklusen
,
D.
,
Larson
,
D. R.
and
Singer
,
R. H.
(
2008
).
Single-RNA counting reveals alternative modes of gene expression in yeast
.
Nat. Struct. Mol. Biol.
15
,
1263
-
1271
.
Zoller
,
B.
,
Nicolas
,
D.
,
Molina
,
N.
and
Naef
,
F.
(
2015
).
Structure of silent transcription intervals and noise characteristics of mammalian genes
.
Mol. Syst. Biol.
11
,
823
.
Zoller
,
B.
,
Little
,
S. C.
and
Gregor
,
T.
(
2018
).
Diverse spatial expression patterns emerge from unified kinetics of transcriptional bursting
.
Cell
175
,
835
-
847.e25
.
Zuin
,
J.
,
Roth
,
G.
,
Zhan
,
Y.
,
Cramard
,
J.
,
Redolfi
,
J.
,
Piskadlo
,
E.
,
Mach
,
P.
,
Kryzhanovska
,
M.
,
Tihanyi
,
G.
,
Kohler
,
H.
et al. 
(
2022
).
Nonlinear control of transcription through enhancer–promoter interactions
.
Nature
604
,
571
-
577
.

Competing interests

The authors declare no competing or financial interests.