|Home | About | Journals | Submit | Contact Us | Français|
Regulatory circuits controlling gene expression constantly rewire to adapt to environmental stimuli, differentiation cues, and disease. We review our current understanding of the temporal dynamics of gene expression in eukaryotes and prokaryotes and the molecular mechanisms that shape them. We delineate several prototypical temporal patterns, including ‘impulse’ (single-pulse) patterns in response to transient environmental stimuli, sustained (state transitioning) patterns in response to developmental cues, and oscillating patterns. We focus on impulse responses and their higher-order temporal organization in regulons and cascades, and describe how core protein circuits and cis-regulatory sequences in promoters integrate with chromatin architecture to generate these responses.
The transcriptional program that controls gene expression in cells and organisms is remarkably flexible, constantly reconfiguring itself to respond and adapt to perturbations. These changes are apparent across a broad range of time scales, from rapid responses to environmental signals (i.e., minutes to hours) to slower events during development and pathogenesis (i.e., hours to days) (Lopez-Maury et al., 2008).
Dissecting these dynamic changes, both functionally and mechanistically, is a fundamental challenge in biology and raises several key questions. What is the scope of temporal patterns of gene expression in biological systems? What functions do different patterns serve? What molecular mechanisms underlie the formation of each pattern, and what is their capacity to process the temporal signal into a specific change in expression over time? Finally, are any principles, either functional or mechanistic, shared among temporal responses in distinct time scales?
Recent parallel advances in genomics and cell biology provide an unprecedented opportunity to map dynamic gene expression and decipher its underlying mechanisms. At the same time, live cell imaging of fluorescent reporter proteins (Locke and Elowitz, 2009) allows us to study gene expression at fine temporal resolution suitable to the time scale of changes and at the single cell level. Such studies, when coupled with molecular manipulations and quantitative modeling, can identify basic mechanisms of temporal patterning. Further, genomic technologies provide global insights on regulation of gene expression by allowing us to measure and perturb many aspects of the regulatory system, such as mRNA levels, protein-promoter interactions (Badis et al., 2009; Lee et al., 2002), or chromatin modification states (Wei et al., 2009; Whitehouse et al., 2007). Finally, emerging methods in synthetic biology, robotics, and microfluidics (Szita et al., 2010) are poised to transform our ability to manipulate cellular inputs and components an unparalleled temporal resolution.
Here, we review recent advances in our understanding of transcriptional dynamics, including the prototypical patterns of temporal mRNA expression and their underlying molecular mechanisms. We identify a small number of prominent temporal patterns, such as single pulse-responses (‘impulses’), sustained state-transitioning patterns, and oscillations. Focusing on impulse responses, we then present the molecular circuits that generate these patterns, highlighting the prominent role that transcription factor localization, integration of multiple inputs through cis-regulatory elements and nucleosome occupancy plays in tuning the response to a given stimulus. Finally, we discuss the prospect for a unified view of regulatory dynamics across time scales and systems, and highlight critical directions for further research.
What capacity does a cell or organism have to generate temporal patterns of gene expression? Recent studies reveal several key classes of patterns (Figure 1). The first one, indefinite oscillators (Figure 1A), plays integral roles in homeostasis, such as the execution of the cell cycle or circadian rhythm. Other classes of temporal patterns follow an external stimulus. These include impulse (or single pulse) patterns in response to transient environmental stimuli (Figure 1B-D) and sustained (or state transitioning) patterns in response to transient developmental stimuli (Figure 1E). Each of these patterns serves a set of inter-related functional goals, including optimizing the investment of cellular responses, temporally compartmentalizing antagonistic processes, and imposing order on the biogenesis of complex biological systems. On a systems-wide scale, the regulation of individual genes is commonly organized at a higher-order into regulons in which a group of genes are controlled by the same transcription factors and, thus, share the same gene expression patterns. In addition, genes can be organized into transcriptional cascades and other patterns in which expression is ordered sequentially. Here, we focus on the impulse-like pattern, specifically its function and integration within transcriptional programs.
Changes in gene expression in response to perturbations of the surrounding environment, such asheat, salinity, or osmotic pressure, typically follow a characteristic ‘impulse’-like pattern (Chechik and Koller, 2009; Chechik et al., 2008). Transcript levels spike up or down abruptly following the environmental cue, sustain a new level for a certain period of time and then transition to a new steady state, often similar to the original levels (Figure 1B). Impulse patterns are prevalent in responses to environmental changes in all organisms, from bacteria to mammals (Braun and Brenner, 2004; Gasch et al., 2000; Litvak et al., 2009; Lopez-Maury et al., 2008; Murray et al., 2004).
One of the most extensively studied impulse systems is the environmental stress response (ESR) program in yeast. The ESR consists of ~900 genes that exhibit short-term changes in transcription levels in response to various environmental stresses (Gasch et al., 2000). The transient impulse pattern of the ESR likely represents an adaptation phase, during which the cell optimizes its internal protein milieu before resuming growth (Gasch et al., 2000). Indeed, many of the down-regulated genes in the ESR are associated with protein synthesis, reflecting the characteristic transient suppression in translation initiation and growth (Gasch et al., 2000). The ESR is also associated with the brief induction of genes involved in specific response mechanisms, such as DNA-damage repair, carbohydrate metabolism, and metabolite transport (Capaldi et al., 2008; Gasch et al., 2000). A notable exception to the impulse-like stress response in yeast is the case of starvation, where the cells initiate more sustained programs, such as quiescence, filamentation, or sporulation (Lopez-Maury et al., 2008).
Transient impulse patterns are also prevalent in environmental responses in mammalian cells (Foster et al., 2007; Litvak et al., 2009; Murray et al., 2004), extending beyond environmental stimuli. For example, when innate immune cells, such as macrophages (Gilchrist et al., 2006; Ramsey et al., 2008) or dendritic cells (Amit et al., 2009), respond to pathogens, expression changes in individual genes follow a clear impulse pattern. These patterns, however, are often coupled to each other, forming multi-step transcriptional cascades, in which the products of genes that are induced early in a response affect the expression of downstream targets. These targets, in turn, may either exhibit an impulse pattern or a more sustained one, initiating a long-term change in the cell state (Amit et al., 2007a; Murray et al., 2004).
Impulse patterns can respond distinctly to the introduction of a signal versus its withdrawal. This differential response results in a ‘sign-sensitive delay’ (Figure 1C), in which the speed of the cell's response to one “sign-shift” (e.g., from the presence to the absence of a nutrient) is different from that of the complementary shift (e.g., from the absence to the presence a nutrient).
Sign-sensitive delays are common in responses of microorganisms to changes in nutrients. For example, consider the arabinose-utilization system of E. coli, where transcription from the LArabinose operon is induced in response to the cyclic adenosine monophosphate (cAMP). The transcriptional response to an increase in cAMP (i.e., “on” sign) is much slower than to a cAMP decrease (i.e., “off” sign) (Mangan et al., 2003). One possible reason for this asymmetry is that, at least inside a mammalian host, the “on” state is present most of the time while the “off” state is maintained only during short and rare pulses of glucose. Consequently, while the cell can halt the production of L-Arabinose genes soon after the introduction of glucose, it can tolerate slower commencement of their production when glucose levels decrease and cAMP is produced (Mangan et al., 2003). Alternatively, a sign-sensitive delay may reflect noise filtering; the cell refrains from activation of response pathways following spurious or transient signals. For the arabinose system, the “on” switch delay is approximately 20 minutes, comparable to the timescale of spurious pulses of cAMP in other natural settings (Alon, 2007).
Conversely, a delayed response to the “off” switch can prolong the effect of a transient stimulus. For example, the expression of flagella motor genes in E. coli persists for one hour after the biogenesis input signal is turned off, but no delay occurs during the “on” switch. Indeed, this delay time in shutting down is comparable to the time needed for the biogenesis of a complete flagella motor (Kalir et al., 2005).
Similar principles of signal processing in impulse responses have also been observed in mammalian systems. For instance, a small regulatory circuit that controls the expression of the gene encoding the pro-inflammatory cytokine interleukin-6 (IL-6) in mouse macrophages exhibits a delayed response to lipopolysaccharide (LPS) stimulation (the “on” switch) and discriminates between transient and persistent signals in the innate immune system (Litvak et al., 2009). Other “persistence detection” mechanisms have also been observed in responses to epidermal growth factor (EGF) (Amit et al., 2007a) and to extracellular-signal-regulated kinase (ERK) signaling (Murphy et al., 2002).
Most studies of environmental stimuli in the lab focus on one sustained signal at a time, but the natural environment to which cells are adapted is substantially more complex, noisy, and irregular (Lopez-Maury et al., 2008; Wilkinson, 2009). Impulse-like transcriptional programs reflect some strategies that cells employ to handle such temporally fluctuating environments.
Random fluctuations are optimally handled by sensing environmental changes and specifically responding by transcriptional changes in relevant genes, as described above (e.g., (Capaldi et al., 2008; Gasch et al., 2000)). In certain cases, a population of cells may respond stochastically; they activate different changes in gene expression in different cells of the same population, thus ‘hedging’ their adaptive bets (Lopez-Maury et al., 2008).
When fluctuations are stable and predictable, bacteria and yeast cells may use an anticipatory strategy for gene regulation (Mitchell et al., 2009; Tagkopoulos et al., 2008). For example, when exposed to heat shock, yeast induce an impulse response of genes needed for oxidative stress although these genes are not directly necessary for adaptation to heat shock. Interestingly, yeast do not induce heat shock genes in response to oxidative stress(Mitchell et al., 2009). This asymmetry (Figure 1D) may reflect the predictable order of the two stresses under natural circumstances: oxidative respiration and accumulation of oxidative radicals follow a temperature increase during fermentation.
Notably, this anticipation strategy differs from symmetrical cross protection (Kultz, 2005) through shared stress-response pathways (Gasch et al., 2000). Rather, it indicates that any optimization of transcriptional programs during evolution occurred in a complex adaptive landscape. Thus, a strategy that may appear ‘sub optimal’ when considering only one stimulus in the lab, may indeed be optimal in the presence of multiple simultaneous or sequential stimuli.
A functional temporal program of gene expression requires appropriate temporal coordination between genes (Figure 1F). Studies reveal two main classes of temporal coordination: regulatory modules and timing motifs.
A regulatory module consists of genes which are co-expressed with the same temporal pattern or amplitude (FANTOM consortium, 2009; Gasch et al., 2000; Spellman et al., 1998). Regulatory modules serve to coordinate the production of proteins that are needed to perform relevant cellular functions in the given response. Regulatory modules are a hallmark of all known transcriptional programs and all known temporal patterns (Figure 1), including oscillatory patterns (e.g. (Spellman et al., 1998)), sustained responses (e.g. (FANTOM consortium, 2009)), and impulse responses (e.g. (Chechik et al., 2008)).
Complementing the tight temporal coincidence within regulons, timing motifs reflect a particular order of transcriptional events among genes or modules, such as a linear cascade of genes with sequentially ordered expression (Alon, 2007; Chechik et al., 2008; Ihmels et al., 2004). In microorganisms, such ordering is commonly observed among genes encoding metabolic and biosynthetic enzymes, and therefore, it can play an important role in achieving metabolic efficiency or avoiding toxic intermediates (Chechik et al., 2008; Ihmels et al., 2004; Zaslaver et al., 2004). For example, following deprivation of amino acids, E. coli induces the expression of amino acid metabolic genes in the same order that their encoded enzymes are present in the relevant amino acid biosynthetic pathway (Ihmels et al., 2004). This ‘just-in-time’ pattern (Zaslaver et al., 2004), which may optimize resource utilization, has also been observed in other bacterial processes, most notably flagellar biogenesis (Kalir et al., 2005).
A broader range of ordered patterns of expression onset, typically in impulse responses, is found in metabolic enzymes in yeast (Chechik et al., 2008; Ihmels et al., 2004). These include timing motifs with gene expression in the same order as the metabolic pathway (i.e., a ‘just-in-time’ induction or shutoff of a pathway), as well as in the reverse order to the metabolic pathway. These reversed directions possibly contribute to the fast removal of an end metabolite that is either toxic or otherwise disruptive under the new condition (Chechik et al., 2008)). Coordinated timing motifs are also found at metabolic branch points (Chechik et al., 2008; Ihmels et al., 2004). For example, consider a metabolic funnel, where two enzymes (A, B) produce complementary metabolites that are together consumed by a third reaction (catalyzed by C). In the ‘funnel-same-time’ motif, the genes that encode the three enzymes (A, B, and C) are often expressed simultaneously, thus optimizing metabolite use by coordinating the production or consumption of metabolites along co-dependent branches. Similar temporal coordination was found for the genes encoding enzymes in ‘forks’, involving one enzyme producing two metabolites, which are then consumed by two separate reactions.
Cell fate decisions are typically associated with stable changes in gene expression that transition the regulatory system from one steady state to the next (Figure 1E). Such cell fate decisions are prevalent in development, (e.g. in cell differentiation (Basma et al., 2009; Nachman et al., 2007; Oliveri et al., 2008)); pathogenesis, ( e.g. in the generation of cancer stem-like cells (Iliopoulos et al., 2009)); and immune responses (e.g. in the response of innate immune cells to pathogens (Amit et al., 2007a; Amit et al., 2009; Ramsey et al., 2008; Wei et al., 2009).). State transitioning in cells involves sustained induction or repression of gene expression, stabilizing the cell on a new characteristic expression program, and disassociating it from its precursors. Nevertheless, processes that lead to such stable changes often involve a succession of impulse responses that promote transient effects necessary for achieving the transition.
Such a combination of transient and stable changes in transcription was observed during phorbol myristate acetate (PMA)-induced differentiation of myelomonocytic leukemia cells (THP-1) cells (FANTOM consortium, 2009). Sustained responses included repression of genes required for cell cycle progression and DNA synthesis, which is consistent with the growth arrest associated with PMA-induced differentiation. In addition, genes that characterize the differentiated phenotype (e.g. immune response) were persistently induced. Conversely, transient, impulse-like, changes were associated with various transcription factors that play an important role early in the transition, promoting the differentiation program prior to repression of the factors that maintain the undifferentiated state. A similar pattern, specifically immediate early impulse responses of key regulators followed by stable changes of downstream genes, has been observed in many other mammalian systems, including responses to growth (Amit et al., 2007a), pathogens (Amit et al., 2009), and stress (Murray et al., 2004) signals.
Impulse responses are not limited to the immediate wave of transcription at the beginning of the state transition. Rather, a succession of impulses, forming a series of transcriptional “waves,” has been observed in various state transitioning responses (Amit et al., 2007a; Amit et al., 2009; Ramsey et al., 2008; Shapira et al., 2009). For instance, the response of immune dendritic cells to pathogens involves several waves of induction in which co-regulated genes follow a simple impulse profile with distinct onset and offset times (Amit et al., 2007a). As in PMA-induced differentiation, the first wave is an immediate-early response enriched for genes that encode proteins with roles in transcriptional regulation. Then a subsequent transcriptional wave is enriched for genes that are required for extracellular signaling (e.g. interferon-beta 1 or IFNB1) and motility (e.g. chemokine ligand 3 or CCL 3) during that time interval (~2-4 hours post stimulus) in the in vivo innate immune response. This temporal organization allows innate immune cells to activate the CCL3 ligand at the appropriate time, favoring the migration of activated cells to the draining lymph node to activate the adaptive immune response.
Long transcriptional cascades of ordered sequential regulation are also at the basis of many complex developmental processes (Davidson, 2010). For instance, in the sea urchin embryo, the transcriptional program of skeletogenic cell development in endomesoderm specification includes several layers of regulation that correspond to developmental phases (Oliveri et al., 2008). Progression through the phases is facilitated by a regulatory cascade in which transcriptional factors that are active during one phase (e.g. early micromere specification) activate genes in the next phase (e.g. late specification). Notably, transcriptional changes in genes encoding regulatory factors can also feedback and regulate the expression of their temporal ‘predecessors’ (Amit et al., 2007a). Such mechanisms are used to shape both impulse responses and sustained responses, as we discuss below.
What is the cell's capacity to ‘compute’ a temporal pattern of mRNA expression? Are there canonical molecular mechanisms that underlie distinct types of patterns? Can a single mechanistic unit generate more than one pattern depending on the incoming signal or its downstream target? In this section, we focus on the molecular mechanisms that generate impulse responses at single genes, gene modules, and temporal motifs.
Regulatory systems that control gene expression are often represented as networks (i.e., directed graphs) with the nodes corresponding to regulatory proteins (e.g., transcription factors ) and the edges linking a DNA binding protein to proteins encoded by genes it binds to and regulates (e.g., (Hu et al., 2007; Lachmann et al., 2010; Shen-Orr et al., 2002)). Such graphs have been assembled from many small scale studies on regulation of individual genes and operons (Shen-Orr et al., 2002) or by systematic chromatin immunoprecipitation (ChIP), in vitro assays, and computational analysis of cis-regulatory sequence elements (Badis et al., 2009; Harbison et al., 2004; Hu et al., 2007; Lachmann et al., 2010; Lee et al., 2002).
Although network graphs appear highly complex, they can be effectively decomposed to putative functional units based on recurring topological patterns (Figure 2). These ‘network motifs’ (Shen-Orr et al., 2002) are small sub networks consisting of only a few nodes and edges with a topological pattern that is significantly over-represented in the transcriptional graph.
Although the patterns themselves are static, they can be associated, analytically (Bolouri and Davidson, 2003; Goentoro et al., 2009; Kittisopikul and Suel, 2010; Mangan et al., 2003; Shen-Orr et al., 2002; Tyson et al., 2003) or experimentally (Basu et al., 2004; Cantone et al., 2009; Kaplan et al., 2008; Mangan et al., 2006; Rosenfeld et al., 2002), with different dynamic interpretations, thus relating the architecture of these network components with a functional capacity for generating temporal responses (Figure 2). These responses include rapid or slowed responses (Figure 2A,B), feedback control (Figure 2C), sign sensitive delays (Figure 2D), temporal ordering (Figure 2E), and temporal coordination in modules (Figure 2F).
Notably, the relation between the topology of a motif and its induced temporal pattern is far from unique, and depends on the characteristics of the incoming signal and of the interacting molecules (Macia et al., 2009). For instance, protein production rate, protein degradation rate, or activation thresholds of regulators can each alter the dynamic transcriptional pattern generated by the motif (Lahav et al., 2004). Moreover, different motifs or combinations of motifs (Geva-Zatorsky et al., 2006) can induce similar behaviors. For a more thorough discussion of network motifs, we refer the reader to other extensive reviews ((Alon, 2007; Davidson, 2009, 2010; Tyson et al., 2003).
The feed forward loop (Figure 2D) is a major building block of combinatorial regulation (Amit et al., 2009). A feed forward loop has a unidirectional structure consisting of three nodes: an upstream regulator X that regulates a downstream regulator Y, which in turn regulates a downstream target Z (which is not necessarily a regulator). An additional edge is directed from Y to Z, thus closing a unidirectional “loop”. Each interaction can be suppressing or activating, resulting in eight unique feed forward loop structures.
One commonly found structure in transcriptional networks (Alon, 2007) is the Type-1 coherent feed forward loop, in which all of the interactions are activating. This feed forward loop can generate a sign- sensitive time delay. The length of the delay and whether it occurs during the “off” or the “on” switch depends on the specific molecular parameters of the loop. The particular logic mediated by the loop largely depends on the organization of cis-regulatory elements in the promoter of the target gene (‘Z’). For instance, when the two transcription factors in a coherent feed forward loop exhibit an “or” logic at the promoter of the downstream gene (i.e., only one transcription factor suffices to activate the gene), the resulting dynamics is usually a sign-sensitive delay with faster response to the “on” switch, and a prolonged transcriptional response, as in flagellar biogenesis (Kalir et al., 2005). Conversely, an ‘and’ logic for the two transcription factors (i.e., both factors are needed to activate the gene) is associated with a faster response to the “off” switch, as in the L-arabinose operon. This feed-forward loop structure facilitates persistence detection (Mangan et al., 2003).
Another prevalent form of the feed-forward loop is the incoherent variant (Figure 2D), in which Y acts as a repressor rather than an activator. Depending on its parameters, this motif can induce pulse-like responses (Basu et al., 2004), lead to a rapid (Mangan et al., 2006) or non-monotone (Kaplan et al., 2008) response of the downstream target Z, or provide a mechanism for detecting fold-change in the level of the input component X (Goentoro et al., 2009).
The single input module (Figure 2F) motif occurs when a single regulator has multiple targets (Alon, 2007; Lee et al., 2002). This architecture, often associated with regulatory hubs (“master regulators”), can facilitate a temporally coordinated response of multiple genes in a module.
However, the activation of the downstream genes in single input module is not necessarily concurrent, and differences in their promoter properties can lead to ordered activation (Figure 3). For example, a transcription factor's affinity for a specific cis-regulatory sequence affects the fraction of time that it occupies a binding site (Bruce et al., 2009; Tanay, 2006). The stronger the binding affinity, the higher the probability that the transcription factor remains bound to a site and recruits the transcriptional machinery (Hager et al., 2009). Differential recruitment at different promoters results in a range of induction thresholds, allowing a single transcription factor with a temporally fluctuating level to generate an ordering of its target genes.
This principle was demonstrated in a recent study using a series of genetically modified promoters of the Pho5 gene during the response to phosphate starvation in yeast (Lam et al., 2008)). In this system (Figure 3), promoters with high affinity sites for the transcription factor Pho4 that are “open” (i.e., not occluded by nucleosomes) responded to weaker signals of slight phosphate deprivation (Figure 3B) and had a shorter response time (Figure 3A) to phosphate starvation compared to those with lower-affinity sites. Similar behavior was observed for synthetic promoter variants and for different targets of Pho4 that had similar promoter architecture.
Thus, graded binding affinities complement the single input module motif in which a single transcription factor induces temporal ordering among its targets through differential binding affinity (Figure 3A). In the phosphate starvation responses, this results in tuning of the responding genes to the severity and duration of phosphate depletion. At intermediate phosphate levels (with intermediate levels of nuclear Pho4), first-response genes with exposed high-affinity sites like PHO84 and PHM4 allow the cell to take up environmental phosphate and mobilize internal reserves. Under starvation conditions, this initial response is followed by a second-order response such as up-regulation of PHO5 and other phosphate- scavenging components (Springer et al., 2003).
More generally, such graded affinities may explain the ordered timing of an impulse-like response of genes within metabolic pathways, in timing motifs such as ‘just-in-time’. In yeast, the timing of ordered activation in a timing motif was found to correlate with the affinity of the respective gene with its regulating transcription factor (Chechik and Koller, 2009; Chechik et al., 2008). Similar principles were also observed in E. coli (Zaslaver et al., 2004).
The position of nucleosomes in a gene promoter impacts the accessibility of transcription factors for their DNA binding sites. Therefore, nucleosome positioning also affects the order of activations across several genes regulated by the same transcription factor. This effect was convincingly demonstrated in the Pho5 system (Lam et al., 2008). Most Pho4 binding sites are occluded under nucleosomes in normal conditions, but they become exposed when chromatin is dynamically remodeled in response to phosphate starvation (Figure 3B). The threshold of response, and hence a gene's onset time, is thus also affected by the chromatin architecture of the repressed state. Conversely, the dynamic range of the response is determined by the active state's architecture. Maximum transcriptional outputs of the Pho5 variants differed by up to 7-fold, and correlated with the number, affinity, and placement of Pho4 sites, irrespective of their accessibility in the initial (pre-starvation) chromatin state. These results suggest a mechanism by which the cell decouples the determinants of promoter activation timing (e.g., site affinity and nucleosome positions) from the determinants of expression capacity (e.g., site affinity alone). Global studies on changes in nucleosome positions in response to environmental signals (Deal et al., 2010) support the generality of the Pho model, at least in yeast (Shivaswamy et al., 2008).
Recent studies suggest that oscillations in the localization or activity of trans regulators that control single input modules play a substantial role in governing (non-oscillating) impulse transcriptional patterns. Most notably, coordinated impulse patterns across a regulon may often stem from limited oscillations in the nuclear localization of a regulatory factor controlling the target genes (Ashall et al., 2009; Cai et al., 2008). This has been suggested for the transcription factor Crz1 in yeast, which employs a ‘pulsing’ mechanism to encode information about extracellular calcium levels (Figure 4; (Cai et al., 2008)). When extracellular calcium increases, Crz1 is dephosphorylated and exhibits short bursts of translocation to the nucleus. At higher levels of calcium, the cells respond, not by increasing the amount of nuclear Crz1 in each translocation burst, but rather by increasing the frequency of the bursts (Figure 4A).
Such ‘frequency modulation’ may be important because of the non-linearity (Yuh et al., 2001) and diversity (Kim et al., 2009) of the input functions associated with different target promoters. Since distinct Crz1 target promoters (Figure 4A) probably respond differently to changing levels of Crz1, amplitude modulation of Crz1 would not maintain their relative ratios (Figure 4B, inset). In contrast, modulation of the frequency of Crz1's nuclear localization can control the expression of multiple target genes in a more proportional manner and thus, maintain more stable ratios of gene expression, regardless of the shapes of their input functions (Figure 4B, main graph). This behavior is probably because the strong non-linear component (i.e., dependence on Crz1 magnitude) is now kept relatively constant for different calcium levels, and the variable part is the amount of time the promoters are exposed to a fixed amount of nuclear Crz1.
Oscillations in the level or localization of transcription factors have been observed in diverse environmental responses, such as those involving NFKB (nuclear factor K-light-chain-enhancer of activated B cells) (Ashall et al., 2009; Covert et al., 2005; Friedrichsen et al., 2006; Nelson et al., 2004; Tay et al., 2010), and p53 (Geva-Zatorsky et al., 2006; Loewer et al., 2010) (Ashall et al., 2009; Friedrichsen et al., 2006; Nelson et al., 2004; Tay et al., 2010) in mammals and the SOS response to DNA damage in bacteria (Friedman et al., 2005). In the latter two systems, monitoring with high temporal resolution showed tightly-regulated oscillations in the nuclear levels of the key regulators (e.g., p53) with variable amplitude but more precise timing. These results raise the possibility that frequency-modulated regulation may represent a general principle by which cells coordinate their impulse responses to signals.
(Friedman et al., 2005; Tay et al., 2010) Protein oscillators driven by external stimuli often lead to non-oscillatory, impulse transcriptional patterns. For example, the expression of p21, a p53-target gene is induced in a non-oscillatory manner during DNA damage (Loewer et al., 2010), and oscillatory NFKB following TNFα stimulation is coupled to impulse-like patterns in a host of early response genes such as the NFKB inhibitor IkBα , even when assessed at the single-cell level (Tay et al., 2010).
Thus, protein oscillations in environmental response systems may play a general mechanistic role in regulating downstream impulse transcriptional changes. First, oscillation of transcription factor levels can maintain a steady response as long as the damage signal is present and constitutive supply of the downstream gene products is needed (as in the p53 response). Second, oscillations in transcription factor localization can underlie the induction of proportional responses through frequency modulation (as with Crz1). Finally, combinations of oscillators can generate various kinetic patterns. For instance, activation of NFKB in mouse embryo fibroblasts treated with LPS depends on two pathways, MyD88 dependent, and MyD88 independent (Covert et al., 2005). Perturbing each of the pathways and leaving the other intact leads, in both cases, to an oscillatory NFKB activity, However, in the wild type setting, both oscillators act upon LPS stimulation but with a ~30 minutes relative phase shift, resulting in a stable, non-oscillatory pattern of NFKB activity. It is likely that other combinations, as well as modulation of both amplitude and frequency, will play a role at encoding other complex patterns of transcriptional regulation at single genes and gene modules.
Impulse patterns can be attenuated and ordered in more complex programs, and through more elaborate regulatory architectures, most notably within developmental programs. In particular, in the cascade motif (Figure 2E), regulators are ordered in layers, and proteins from one layer control ones in subsequent layers (Hooshangi et al., 2005; Rappaport et al., 2005). This pattern was observed in transcriptional networks in sea urchin development (Bolouri and Davidson, 2003; Davidson, 2009, 2010; Oliveri et al., 2008), state-transitioning systems in microorganisms (Chu et al., 1998), and environmental responses in mammalian cells (Amit et al., 2007a; Amit et al., 2009; Ramsey et al., 2008; Shapira et al., 2009). A cascade-like network topology entails an inherent temporal order of regulation events (Hooshangi et al., 2005). It was postulated to enable context-specific responses (Davidson, 2009) and to provide robustness both to spurious input signals (Hooshangi et al., 2005) and to noise in the rates of protein production (Rappaport et al., 2005).
Regulatory interactions between different layers in a cascade can form multi-component loops in which genes in a late transcriptional wave regulate genes from earlier waves (Figure 2C). The ensuing feedback effect can contribute to the ultimate attenuation of impulse responses, even under a sustained signal (Amit et al., 2007b). For example, stimulation of human cell lines with EGF induces several ordered impulse responses (Amit et al., 2007a)(Amit et al., 2007a), including the induction of ‘delayed early’ genes. ‘Delayed early’ genes are primarily induced by transcription factors that were themselves induced as ‘immediate early’ genes. ‘Delayed early’ genes encode a large number of signaling protein and RNA binding proteins that attenuate RNA levels and protein activity of the initial response pathways. Such negative transcriptional feedback mediated through a transcriptional cascade is common in environmental responses in yeast as well (Segal et al., 2003).
A more basic form of feedback is the auto-regulatory loop by which a transcription factor regulates its own gene. The negative form of this motif (Figure 2A) facilitates a rapid transcriptional response of the auto-regulating gene. It has been associated with the induction of a rapid impulse response to EGF stimulation in human cells (Amit et al., 2007a), and to DNA damage in E. coli (Camas et al., 2006). (Amit et al., 2007a)(Camas et al., 2006) The complementary positive auto-regulatory loop (Figure 2B) is associated with the opposite effect, for it results in a slow response time (Alon, 2007). Positive loops, with either one or more components (Figure 2B,C respectively)), can also result in substantial variation between isogenic cells, due to stochastic effects, and play an important role in maintaining stability after state transitioning (Davidson, 2009; Kim et al., 2008; Macarthur et al., 2009; Oliveri et al., 2008).
Diverse mechanisms drive impulse-like changes in mRNA levels, which can occur on a broad range of time scales, from rapid environmental stress responses to slower and more elaborate developmental processes. What can we learn by comparing these processes across time scales? The emerging picture supports a few basic principles. Just-in-time responses and sign-sensitivity optimize process efficiency, whereas the organization of the impulse responses in functional waves and cascades provides temporal compartmentalization and order to gene expression.
Although in this Review, we have made convenient distinctions between impulse responses, state transitions, and oscillators, most biological systems intertwine these temporal patterns. For example, oscillations in protein levels or localization can also lead to impulse responses, and ordered impulses are important in generating sustained responses through cascades. Furthermore, many of the underlying molecular mechanisms driving these temporal patterns can be intimately linked. For example, a gene may be poised for transcription with a pre-initiation complex in anticipation of both a developmental and environmental stimuli. (Davidson, 2009; Hooshangi et al., 2005; Kim et al., 2008; Lee et al., 2002; Li et al., 2007; Macarthur et al., 2009; Oliveri et al., 2008; Rappaport et al., 2005)
Similarly, the mechanistic regulatory building blocks surveyed here are typically embedded within a wider network context. First, many responses, especially in metazoans, involve a large number of inputs into a single promoter during both environmental and developmental responses (Amit et al., 2009; FANTOM consortium, 2009). In addition, transcriptional cascades are often combined with other motifs, such as negative feedbacks (Amit et al., 2007a), feed forward loops (Basu et al., 2004; Shen-Orr et al., 2002), and single input modules (Shen-Orr et al., 2002). Furthermore, multiple cis-regulatory elements and sequences affecting nucleosome positions are integrated within more complex cis-regulatory functions in both yeasts (Gertz et al., 2009; Raveh-Sadka et al., 2009) and metazoans (Kaplan et al., 2009; Yuh et al., 2001; Zinzen et al., 2009). Both computational studies and synthetic molecular circuits (Cantone et al., 2009) have provided additional insights into the cross talk between motifs (Ishihara et al., 2005; Ma et al., 2004) and into the dynamics of complex networks (Walczak et al., 2010) that integrate multiple motifs. Nevertheless, the correspondence between simple sub-networks and motifs and the observed temporal patterns of mRNA levels (Alon, 2007; Davidson, 2009, 2010) suggests a substantial degree of modularity in the operation of regulatory systems.
Most of mechanisms driving mRNA concentrations described in this Review, and that have been deciphered in detail so far, are transcriptional, but other pathways also affect mRNA levels, including mRNA processing, transport and degradation. Although recent studies (Shalem et al., 2008) suggest that such mechanisms can play a substantial role in shaping temporal profiles of mRNA levels, these mechanisms are still far less understood than transcription regulation. Indeed, the scarcity of experimental methods to monitor these processes has hampered progress in this area. However, we anticipate that recent advances in massively parallel cDNA sequencing (RNA-Seq (Mortazavi et al., 2008)) will help advance this front.
More generally, deciphering circuitry, that is, understanding the capacity of molecular mechanisms to encode complex signals and decode them into specific responses, will require tight integration between experiments, analysis, and computation, in particular for temporal responses. First, there is a substantial need for direct manipulation of both signals, for example using microfluidic devices, and of individual components, by either manipulation of trans components (Amit et al., 2009; Costanzo et al., 2010; FANTOM consortium, 2009) or cis sequences (Gertz et al., 2009; Patwardhan et al., 2009)). Monitoring temporal responses in segregating populations (Eng et al., 2010) can provide a complementary means for testing the effect of many simultaneous genetic perturbations. Analytical methods and computational models can guide the design of these perturbations to a search space that is maximally informative and biologically relevant, for example by employing sequence models of gene regulation (Gertz et al., 2009; Raveh-Sadka et al., 2009) to choose relevant variants to test, or provisional models of trans regulation (Amit et al., 2009) to choose targets for gene silencing or disruption.
Improving the ability to monitor a larger number of circuit components over time in living cells is important for broadening the scope of many single-cell studies and for deepening our understanding of population-level phenomena observed with genomics profiling technologies. Recent advances in simultaneously monitoring in vivo multiple types of RNA (Kern et al., 1996; Muzzey and van Oudenaarden, 2009) or proteins (Bandura et al., 2009) are promising. Notably, although the difference between a single cell and population view is a recurring theme of recent studies, reconciling the two is important for a functional understanding of a response, especially in multi-cellular organisms (Simon et al., 2005). For example, a recent study of the NFKB response to TNFα stimulation showed that the observed cellular heterogeneity may be optimal for achieving a functional population (or mean) response for paracrine cytokine signaling (Paszek et al., 2010).
Computational analysis of time course data presents several unique and challenging problems, where the most common tasks are identifying differentially expressed genes, grouping them into clusters of similar temporal patterns, and inferring their regulatory interactions. Recent studies have shown that a useful algorithmic starting point is to derive a continuous representation of transcriptional profiles by fitting to a particular mathematical function (Chechik and Koller, 2009; Storey et al., 2005). Specifically, impulse responses fit well to a certain class of sigmoid “impulse-like” functions, that have a small number of biologically interpretable parameters (e.g. onset time) (Chechik and Koller, 2009; Chechik et al., 2008). The fitted continuous representations can then be used in conjunction with the original expression values, aiming to provide a more robust analysis, particularly for differential expression (Storey et al., 2005) and clustering (Chechik and Koller, 2009). (Chechik and Koller, 2009; Chechik et al., 2008)Despite these advances and the vast amount of research on the more advanced task of regulatory network inference (Bansal et al., 2007; Karlebach and Shamir, 2008), there is still much to be done. The emerging complexity of regulatory mechanisms and the expected availability of more diverse and refined temporal data leave substantial room for developing more refined mechanistic models of gene regulation that account for both cis and trans elements and their integration in time.
Finally, advances in synthetic biology promise not only the ability to manipulate biological entities but also to design systems to aid the development and interpretation of analytical models with increasing complexity. This would be particularly critical to decipher the complex web of interactions and the multiplicity of inputs that determines temporal changes in gene regulation in living cells.