News & Views

Gene regulation: Expression feels two pulses

Single-cell analyses reveal that combinatorial changes in the intracellular locations of transcription factors can tune the expression of the factors' target genes in response to environmental stimuli. See Article p.54

Most transcription factors exert their action continuously, but some act in pulses by moving rapidly in and out of the nucleus. Transmitting cellular signalling-pathway information in pulses or oscillations has several advantages over continuous signalling. For example, information can be encoded in the frequency or amplitude of pulsing, boosting the amount of information transmitted. Investigating this phenomenon has proved difficult, however, because the behaviour of pulsatile transcription factors varies greatly from cell to cell1. In this issue, Lin et al.2 (page 54) overcome this hurdle and demonstrate that combinations of transcription-factor pulses that change in response to environmental stimuli can regulate gene expression.

The Msn2 protein, which is expressed in the budding yeast Saccharomyces cerevisiae, was the first transcription factor to be identified as pulsatile, moving to the nucleus to activate transcription when cells are exposed to light3. Although such pulsatile signalling patterns can be advantageous, they are also prone to disruption, because the message transmitted varies as time passes. In electronics, such problems are typically solved by ensuring that more than one component can perform the same task. Theoretically, the same principles apply to cell signalling — propagating signalling pulses through multiple pathways that are then reintegrated is predicted to improve reliability4. Indeed, Msn2 is known5 to act with the pulsatile transcriptional repressor protein Mig1 to control gene expression in response to various stresses.

Lin et al. analysed the dynamics of Msn2 and Mig1 pulses by generating strains of S. cerevisiae in which the two transcription factors were tagged by different fluorescent proteins, allowing their intracellular locations to be tracked. The authors attached these cells to a microfluidic device through which cell-growth media were passed, and monitored transcription-factor movements as well as any subsequent changes in the transcription of genes whose expression is regulated by both factors.

Depleting glucose in the cell media triggered the export of Mig1 from the nucleus and the import of Msn2, increasing the expression of target genes. If Msn2 and Mig1 acted according to a simple continuous regulatory scheme, or if they were pulsatile but the timing of pulses was completely random, then glucose depletion would gradually alter the average level of each transcription factor in the nucleus across the population of cells, and the expression of target genes would gradually increase to a new steady-state level. Instead, however, the authors observed a 'transient phase' immediately after glucose depletion, during which the average nuclear levels of Msn2 and Mig1 were higher and lower, respectively, than when they subsequently reached steady-state levels (Fig. 1a). An overshoot such as this is often observed when systems adapt to change6, and it can decrease the time it takes for target-gene expression levels to reach the new steady state. Indeed, a kinetically similar response is known to occur7 when glucose concentration increases: target-gene expression is repressed by Mig1, lowering levels of the corresponding RNA transcript, but a transient destabilization of the transcript helps to speed up the process by promoting transcript degradation.

Figure 1: Interpreting transcription-factor pulses.
Figure 1

The transcription factors Msn2 and Mig1 enter the nucleus in pulses to respectively activate and repress transcription of the same target genes. a, Lin et al.2 report changes in transcription-factor pulsing in response to environmental stimuli. A decrease in glucose concentration in the medium around the cell causes a large transient decrease in the level of Mig1 in the nucleus, and a rapid increase in Msn2. This 'overshoot' allows cells to adapt to change by promptly increasing gene expression. After this transient phase, nuclear levels of each factor, and hence gene expression, remain steady on a population-wide level. b, The authors show that it is only in the transient phase that all cells display synchronized non-overlapping pulses. After this phase, levels of each factor pulse randomly in single cells. When pulses of both factors overlap in the nucleus, gene expression falls. However, when pulses of Msn2 do not overlap with Mig1, expression increases.

Lin and colleagues hypothesized that the overshoot they observed was not just a transient event, but might persist in steady-state conditions in the form of pulses that could not be observed on a population-wide level because their effects averaged out. When analysing single cells, however, the authors found that the pulsing of each transcription factor was sporadic and irregular under steady-state environmental conditions, making it hard to define individual pulses. In principle, many criteria could be used to define such pulses, but most would be of little practical relevance. The authors developed an interesting and pragmatic approach to detecting individual pulses, based on a neuroscience technique called spike-triggered averaging8. In their adapted version, which the authors dubbed pulse-triggered averaging, pulses were measured as averages that were based on the dynamics of Mig1 and Msn2 over a set time period around peaks in nuclear Msn2 levels.

Justifying their approach, the authors demonstrated that a target gene responded to changes in Msn2 or Mig1 that the technique registered as pulses. For instance, Msn2 pulses were followed by elevated gene expression if there was no overlap with a Mig1 pulse (that is, if Mig1 was not in the nucleus at the same time). Conversely, there was a decrease in target-gene expression if the Msn2 pulse was counteracted by an overlapping Mig1 pulse. These observations confirm that the transcription-factor pulses do indeed persist under steady-state conditions (Fig. 1b).

At elevated steady-state glucose concentrations, the percentage of overlapping pulses increased to a level beyond that expected by chance, enhancing the efficiency with which target-gene expression was repressed. By contrast, during the transient phase, Lin and colleagues detected only non-overlapping pulses, suggesting that during this period the timing of pulses is modulated. In this way, pulses are synchronized between cells, resulting in an overshoot at the population-wide level.

Pulse-triggered averaging could now become a powerful tool for analysing other regular oscillating reactions in cells. So far, most studies have focused on average cell behaviour, but cell-cycle checkpoints, for instance, elicit single-cell responses with considerable cell-to-cell variability9. Pulse-triggered averaging may help to disentangle the underlying regulatory interactions.

Moreover, the authors' approach makes it possible to analyse gene regulation without understanding all of a system's parameters. For instance, the current study showed not only that a fully overlapping repressor pulse can neutralize an activating pulse, but also that the same neutralization can occur when the two pulses are separated by a few minutes, without needing to understand the root causes. It is important to note that not all Msn2 pulses elicited target-gene expression, even when Mig1 activity was low. This provides a reminder that mass-action kinetics, stochastic modelling and identification of reaction mechanisms must be included in complete models of gene regulation. Research into these topics has undergone marked development in recent years, and may soon converge, making it possible to understand the dynamics of signalling pathways in detail.

Notes

References

  1. 1.

    & Curr. Biol. 24, 205–211 (2014).

  2. 2.

    , , , & Nature 527, 54–58 (2015).

  3. 3.

    , , , & J. Cell Biol. 161, 497–505 (2003).

  4. 4.

    & eLife 4, e06559 (2015).

  5. 5.

    , , , & EMBO J. 24, 4115–4123 (2005).

  6. 6.

    , & J. Phys. Chem. B 112, 16752–16758 (2008).

  7. 7.

    et al. Nature Commun. 3, 682 (2012).

  8. 8.

    Neural Netw. 14, 599–610 (2001).

  9. 9.

    et al. Nature Commun. 5, 4048 (2014).

Download references

Author information

Affiliations

  1. Antoine Baudrimont and Attila Becskei are in the Biozentrum, University of Basel, CH-4056 Basel, Switzerland.

    • Antoine Baudrimont
    •  & Attila Becskei

Authors

  1. Search for Antoine Baudrimont in:

  2. Search for Attila Becskei in:

Corresponding author

Correspondence to Attila Becskei.

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.