• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of plosonePLoS OneView this ArticleSubmit to PLoSGet E-mail AlertsContact UsPublic Library of Science (PLoS)
PLoS One. 2011; 6(11): e27414.
Published online Nov 10, 2011. doi:  10.1371/journal.pone.0027414
PMCID: PMC3244268

A Simple Negative Interaction in the Positive Transcriptional Feedback of a Single Gene Is Sufficient to Produce Reliable Oscillations

Mark Isalan, Editor

Abstract

Negative and positive transcriptional feedback loops are present in natural and synthetic genetic oscillators. A single gene with negative transcriptional feedback needs a time delay and sufficiently strong nonlinearity in the transmission of the feedback signal in order to produce biochemical rhythms. A single gene with only positive transcriptional feedback does not produce oscillations. Here, we demonstrate that this single-gene network in conjunction with a simple negative interaction can also easily produce rhythms. We examine a model comprised of two well-differentiated parts. The first is a positive feedback created by a protein that binds to the promoter of its own gene and activates the transcription. The second is a negative interaction in which a repressor molecule prevents this protein from binding to its promoter. A stochastic study shows that the system is robust to noise. A deterministic study identifies that the dynamics of the oscillator are mainly driven by two types of biomolecules: the protein, and the complex formed by the repressor and this protein. The main conclusion of this paper is that a simple and usual negative interaction, such as degradation, sequestration or inhibition, acting on the positive transcriptional feedback of a single gene is a sufficient condition to produce reliable oscillations. One gene is enough and the positive transcriptional feedback signal does not need to activate a second repressor gene. This means that at the genetic level an explicit negative feedback loop is not necessary. The model needs neither cooperative binding reactions nor the formation of protein multimers. Therefore, our findings could help to clarify the design principles of cellular clocks and constitute a new efficient tool for engineering synthetic genetic oscillators.

Introduction

Cellular clocks control important functions of the cell, such as circadian (24-hour) rhythms, cell cycle, metabolism and signaling. Clock operation appears to involve the coupling of two different types of oscillators. The first are oscillators based on cytoplasmic reactions, such as phosphorylation [1] and oxidation [2], [3]. The second are genetic oscillators depending on gene expression regulation [4], [5]. In the last decade several synthetic genetic oscillators have been implemented in the laboratory [6][12]. The first mathematical model of a genetic oscillator was developed by Goodwin for periodic enzyme production [13]. This model was the groundwork for subsequent theoretical research on genetic oscillators in living systems, such as fungi and flies [14][19]. In these models, the rhythms are generated by a gene with a negative transcriptional feedback (NTF) (Fig. 1A). This NTF needs time delay and sufficiently strong nonlinearity in the transmission of the feedback signal for preventing the steady-state stabilization of the system [20], [21]. It has also been analyzed variants, involving two genes, of the model presented in the Fig. 1A [22].

Figure 1
Diagram of one-gene oscillators with negative and positive transcriptional feedbacks.

Positive transcriptional feedbacks (PTFs) are also present in many cellular clocks [23][25]. Models with two or more genes involving PTFs have been studied in genetic oscillators [26][34]. In these models the PTFs increase the expression of repressor genes. It has been shown how PTFs produce bistability [35], [36], increase the robustness of cellular clocks [37], [38] and could provide robust adaptation to environmental cycles [39]. Previously, it has been demonstrated that a single gene with only PTF does not produce oscillations [40]. Here we study a model with a simple condition to produce biochemical rhythms in a single gene with PTF (Fig. 1B). We chose a circadian period for the oscillator due to its relevance in biological systems. This model is based on two common features of genetic oscillators [4], [21], [26], [28], [38]. The first is a PTF created by a protein that activates the transcription of its own gene. The second is a negative interaction in which a repressor inhibits the activity of this protein. We performed stochastic and deterministic simulations that yielded similar results. The stochastic simulations show that the genetic oscillator is robust to noise. This noise is introduced in living cells by the stochasticity of gene expression [41], [42]. By means of a reduced deterministic model, we show that the oscillations exhibit limit-cycle behavior. This means that if a disturbance is applied to the system, the oscillations return to the original periodic solution [43], [44]. Also we show that this biological clock can be classified as a relaxation oscillator [28], [43], [44]. This type of clock is sometimes called hysteresis oscillator [26], [45] or amplified negative feedback oscillator [21], [25]. The relaxation oscillator comprises fast and slow oscillation creation stages. In our model these oscillations are characterized by sawtooth waveforms. Finally, we explain how the negative interaction works through a comparison with the dynamics of the typical enzymatic reaction. We show that the rate of the negative interaction is amplified by the PTF and has a saturation point.

Results

Model and simulations

The model is a simple one-gene network with two well-differentiated parts (Fig. 2). The first is a PTF created by a protein An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e001.jpg, which is a transcription factor of its own gene. When this protein binds to its promoter the transcription rate increases. The second part is a negative interaction in which a repressor molecule An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e002.jpg prevents An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e003.jpg from binding to its promoter. The molecule An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e004.jpg can be thought of as a protease, as a protein that sequesters An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e005.jpg, or as any other molecule that inhibits the function of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e006.jpg as shown in Fig. 2. A different version of the model can be formulated in which the negative interaction acts on the mRNA molecules instead of on protein An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e007.jpg.

Figure 2
Model of a one-gene oscillator with PTF.

Eleven biochemical reactions provide a full description of the model (see (3) in the section Methods: Biochemical reactions and rates). The system is assumed to have a uniform mixture of biomolecules. For this reason, we did not take into account diffusion processes. In this approach, the dynamics of the biochemical reactions (3) can be described by two different formalisms known as stochastic and deterministic approaches (see Methods: Deterministic and stochastic simulations for more details). These two approaches can lead to different behaviors. The stochastic dynamics of the reactions (3) were simulated using the Gillespie algorithm [46] and the deterministic dynamics using the following ordinary differential equations:

equation image
(1)

where the variables and rates are described in the section Methods: Biochemical reactions and rates. We used standard values within the diffusion limit for the rates [18], [38], [47].

The stochastic approach is more realistic than the deterministic simulation because it takes into account the randomness of the chemical reactions. This randomness produces fluctuations in the number of molecules. We fitted the reaction rates to obtain circadian oscillations in the stochastic simulation. Then, we compared the results with the deterministic simulation (Fig. 3). For both simulations the time evolution of the protein (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e027.jpg), repressor (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e028.jpg), protein-repressor complex (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e029.jpg) and mRNA (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e030.jpg) are very similar. The main difference is the appearance of fluctuations in the stochastic case around the number of molecules predicted by the deterministic approach. The fluctuations are more evident in the time evolution of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e031.jpg (Fig. 3G) than in the other biomolecules. This is because the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e032.jpg molecules oscillates in a lower range than An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e033.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e034.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e035.jpg. The oscillations in An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e036.jpg are characterized by sawtooth waveforms. On the other hand, there are differences between the stochastic and deterministic time evolution of the gene. There is a single gene in the model, which can be deactivated (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e037.jpg) or activated (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e038.jpg). Therefore, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e039.jpg molecule. The stochastic simulation shows realistic discrete transitions between 0 and 1 molecules (Figs. 3I and 3K). By contrast the deterministic simulation shows unrealistic continuum transitions (Figs. 3J and 3L). In both cases, however, the qualitative behavior is the same. Most of the time the gene is activated by An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e040.jpg, although it is deactivated for a short time when the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e041.jpg in the oscillations is low.

Figure 3
Stochastic and deterministic simulations of the model.

Model robustness to noise

The fluctuations in the stochastic simulation are the source of so-called intrinsic noise [41], [42]. In the genetic oscillator, this intrinsic noise generates variability in both the amplitude and period of the oscillations. The phase plane defined by An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e063.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e064.jpg illustrates this variability very clearly (Fig. 4A). The deterministic phase plane is a well-defined curve because the oscillations are identical (dashed line in Fig. 4A). In contrast, the stochastic phase plane is a curve that spreads around the deterministic curve due to intrinsic noise (solid line in Fig. 4A). We used the amplitude and period histograms, and the autocorrelation function to quantify the effect of this intrinsic noise on An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e065.jpg oscillations. The results are similar to circadian models with more chemical reactions [18], [26]. The amplitude histogram shows a mean of 6,723 molecules and a standard deviation of 858 molecules (Fig. 4B). The period histogram shows a mean of 24.3 hours and a standard deviation of 1.7 hours (Fig. 4C). In contrast, the absence of intrinsic noise in the deterministic simulation produces identical An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e066.jpg oscillations with lower amplitude and period equal to 6,164 molecules and 23.6 hours, respectively. On the other hand, the autocorrelation function shows a half-life time of about 120 hours (Fig. 4D).

Figure 4
Model robustness to noise.

The stochastic approach produces good oscillations in An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e096.jpg even when there are fewer than 30 molecules of M, R and C. (Figs. 4E–H). We changed the value of some rates to obtain this simulation as in [38] (see caption of Fig. 4). In the deterministic approach, where intrinsic noise is not present, these changes do not alter the dynamics of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e097.jpg significantly and produce a low number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e098.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e099.jpg, and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e100.jpg molecules. In particular, the amplitude and the period are slightly lower (Fig. S1). In the stochastic simulation the rate changes reduce the amplitude and period means to 6,166 molecules and 21.3 hours, respectively (Fig. S2A). The effects of intrinsic noise is now more pronounced because the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e101.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e102.jpg, and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e103.jpg molecules is low. This is reflected in an increase of the amplitude and period standard deviations to 2,132 molecules and 5.2 hours, respectively (Fig. S2B).

In cells, there are also fluctuations in the number (or activity) of molecules such as polymerases, ribosomes and degradation machinery. These fluctuations are the source of so-called extrinsic noise [41], [42]. We performed stochastic simulations varying the parameters in order to account for some aspect of extrinsic noise in the robustness study of the model. The results show that this oscillator is robust to small parameter variations (Fig. 4I) like more other complex models of genetic oscillators [27]. The largest amplitude and period changes occurred for variations in An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e104.jpg (see Table S1). The changes in the mean period and amplitude were always less than 15% and 31%, respectively. Particularly, variations in the rates An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e105.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e106.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e107.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e108.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e109.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e110.jpg produced changes of less than 3% and 8% in the mean period and amplitude, respectively. The changes in the standard deviation of the period and the amplitude were always less than 13% and 27%, respectively.

Reduced deterministic model

To identify the types of biomolecules mainly responsible for oscillations, it is useful to reduce the deterministic model by means of the quasi-steady-state assumption (QSSA) [43], [48]. This approximation differentiates between fast and slow variables. The greater the time-scale separation between the variables the more accurate the approximation is. In this approach it is assumed that fast variables quickly reach the equilibrium, i.e., their derivatives are zero. This assumption means that slow variables are responsible for the system dynamics. In this model, we assumed that the fast variables are An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e111.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e112.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e113.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e114.jpg, and the slow variables are An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e115.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e116.jpg. Then, the set of Eq. (1) can be simplified to

equation image
(2)

where An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e118.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e119.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e120.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e121.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e122.jpg. A good way to check if this approximation is correct is to compare the numerical solution of the complete and the reduced systems. Both numerical solutions agree except for quantitative differences in the period and the amplitude (Figs. 5A and 5B). These differences are due to the fact that the time-scale separation between fast and slow variables is not large enough for QSSA to be more accurate. Despite these differences, we can conclude that An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e123.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e124.jpg are mainly responsible for the system dynamics. The other types of biomolecules can be considered to be at equilibrium. The fluctuations in the fast variables do not significantly affect the system dynamics [38]. This explains the robustness of the model when the number of molecules is low (Figs. 4E–H). In fact, the system produces reliable oscillations even if the average of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e125.jpg is less than one molecule (Fig. 4H), and, surprisingly, even when the driven An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e126.jpg molecules oscillate in a range of less than An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e127.jpg molecules (Fig. 4F).

Figure 5
Model of the genetic oscillator reduced by QSSA.

The oscillations in the reduced deterministic model exhibit limit-cycle behavior (thin solid line in Fig. 5C). Therefore, if an external disturbance is applied to the oscillator, the system will go back to oscillating with the period and amplitude of its limit cycle. The unstable fixed point of the system is An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e143.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e144.jpg molecules (circle in Fig. 5C). For a bifurcation analysis of parameters An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e145.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e146.jpg indicating the range of values that produces limit-cycle oscillations, see Methods: Bifurcation diagram.

This genetic clock belongs to the so-called relaxation oscillators [28], [43], [44]. The mechanism responsible for the oscillations is represented by the nullclines An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e147.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e148.jpg (Fig. 5C). These nullclines are the solution of the equations An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e149.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e150.jpg, respectively. The nullcline An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e151.jpg is a straight line and the nullcline An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e152.jpg has the characteristic “Z” shape of relaxation oscillators [43][45]. The shape of the An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e153.jpg nullcline is the same as the hysteresis diagram obtained if An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e154.jpg is assumed constant (Fig. S6). Therefore, this genetic clock contains some features of hysteresis in its oscillatory mechanism. The An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e155.jpg nullcline has two branches that we can call “high” and “low” (Fig. 5C). These branches are steady states if the An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e156.jpg is a constant (Fig. S6). In each oscillation the system switches from one branch to the other using the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e157.jpg molecules as a transient signal. This process can be explained following the limit-cycle trajectory. When An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e158.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e159.jpg are about 1 and 200 molecules, respectively, their number increases until An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e160.jpg reaches its maximum of about 7,330 molecules and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e161.jpg reaches about 650 molecules. This is the transient from the low to the high branch. Then, the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e162.jpg molecules is reduced to about 0 molecules, whereas An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e163.jpg reaches its maximum of about 1,260 molecules. This is the transient from the high to the low branch. Finally, the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e164.jpg molecules is quickly reduced and the trajectory moves along the nullcline An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e165.jpg, returning to the starting point where a new cycle begins.

This genetic clock is characterized by containing fast and slow stages. The time evolution of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e166.jpg shows these two well-differentiated stages (Fig. 5D). In the slow stage An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e167.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e168.jpg, then the second differential equation in (2) can be approximated by An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e169.jpg. In this stage, therefore, the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e170.jpg molecules increases linearly according to equation An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e171.jpg. In the fast stage An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e172.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e173.jpg, then the second differential equation in (2) can be approximated by An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e174.jpg. In this stage, the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e175.jpg molecules decays exponentially according to equation An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e176.jpg. The two stages play different roles. The slow stage is characterized by the formation of a pulse of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e177.jpg molecules. On the other hand, the decay of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e178.jpg into An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e179.jpg in the fast stage provides the necessary conditions for a new pulse. These two stages produce oscillations in An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e180.jpg with sawtooth waveforms (solid line Fig. 5D).

How the negative interaction works

The negative interaction decreases the number of free An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e181.jpg molecules and takes the system back to the start of a new cycle. The detailed explanation of how this interaction works is related to the dynamics of the typical enzymatic reaction An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e182.jpg, where An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e183.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e184.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e185.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e186.jpg are the substrate, enzyme, complex substrate-enzyme and product, respectively. The total number of enzymes (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e187.jpg) is constant in the system. The rate of catalysis in this reaction is defined as An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e188.jpg. The value of this rate can be approximated by QSSA. The result of this approximation is the well-known Michaelis-Menten equation An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e189.jpg, where An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e190.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e191.jpg [43]. In this equation, the rate An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e192.jpgincreases asymptotically as a function of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e193.jpg. The rate An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e194.jpg reaches a maximum value (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e195.jpg) when the amount of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e196.jpg is large compared with the constant An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e197.jpg. In this situation, the enzymes are saturated because most are part of complex An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e198.jpg, and adding more An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e199.jpg does not increase the rate An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e200.jpg Therefore, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e201.jpg, and the rate of the catalysis An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e202.jpg reaches the constant value An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e203.jpg.

In our model, the negative interaction is An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e204.jpg, where we assumed An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e205.jpg to simplify the model. We can think of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e206.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e207.jpg, and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e208.jpg as An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e209.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e210.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e211.jpg, respectively. Therefore, the rate of the negative interaction can be defined as An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e212.jpg (Fig. 6A). This rate represents the number of degraded An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e213.jpg molecules per hour. The negative interaction works as follows. The number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e214.jpg molecules increases quickly due to the positive feedback. This rise causes most of the An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e215.jpg molecules to bind to An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e216.jpg molecules forming the complex An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e217.jpg. At this point, the system reaches the saturation level (circle in Fig. 6A). The total number of repressor molecules in the system is An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e218.jpg. Therefore, at the saturation point, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e219.jpg and the rate An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e220.jpg reaches the value An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e221.jpg. The negative interaction is not fast enough to decrease the growth of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e222.jpg molecules immediately after the saturation point is reached. This is because the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e223.jpg molecules is low at this point. Nevertheless, new An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e224.jpg molecules enter the system at rate An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e225.jpg. Therefore, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e226.jpg increases linearly over time (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e227.jpg) compared with the enzymatic reaction in which An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e228.jpg is constant. This means that the rate of the negative interaction increases linearly according to equation An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e229.jpg. The value of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e230.jpg increases until the negative interaction is fast enough to reduce the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e231.jpg molecules and take the system back to the start of a new cycle. The maximum rate reached by the negative interaction is An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e232.jpg 3,180 molecules/hour (square in Fig. 6A).

Figure 6
Rate of the negative interaction.

In this model there is not an explicit negative feedback loop at the genetic level. It has been conjectured that all biochemical oscillators involve some sort of negative feedback loop [21]. In this genetic clock, an effective negative feedback loop appears in the reduced model (see the section Methods: The Jacobian matrix). Intuitively, this effective negative feedback loop can be explained as follows: when An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e252.jpg is rare, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e253.jpg is increased by the positive feedback. This rise in the production of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e254.jpg leads to the accumulation of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e255.jpg, which in turn increases An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e256.jpg. This accumulation of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e257.jpg increases until the negative interaction is fast enough to reduce the number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e258.jpg molecules. In this model, we assumed that An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e259.jpg is not degraded. If this complex is degraded according to the reaction An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e260.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e261.jpg increases at a slower rate, and its maximum value (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e262.jpg) is reduced (Figs. 6B and S5). The oscillations stop when An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e263.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e264.jpg (Fig. S5G), because not enough An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e265.jpg is accumulated in order to increase An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e266.jpg.

This genetic oscillator does not need cooperative binding reactions nor the formation of protein multimers, in contrast to the one-gene oscillator with TNF (Fig. 1A). It has been demonstrated that protein sequestration produces an effective high nonlinearity [49], [50]. But this high nonlinearity is not observed if the repressor molecule is recycled [49]. In our model the repressor An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e267.jpg can be used several times. Therefore, the negative interaction does not produce an effective high nonlinearity (see Supporting Information: Text S1).

Discussion

Genetic networks with NTFs and PTFs play an important role in cellular clocks. In this paper, we provided a simple model illustrating that a single gene with PTF has also the potential to produce reliable oscillations. The sufficient additional requirement is a simple and usual negative interaction of degradation, sequestration or inhibition acting on the positive feedback signal. The model presented in this article has a different oscillatory mechanism than the well-established NTF one-gene oscillator model. Our model can be classified as a relaxation oscillator. A two-gene model has been proposed as a different way of producing reliable circadian oscillations in cellular clocks [26], which also is a relaxation oscillator. This two-gene model is important because it is robust to noise [38]. The model introduced in this paper is a simpler way to produce relaxation oscillations than the previous two-gene oscillator. A comparison with our model reveals that the activation of the repressor gene is not a necessary condition to produce reliable circadian oscillations in the two-gene oscillator. We demonstrated that our model produces circadian oscillations that are just as robust to noise as the two-gene oscillator and other more complex models [18], [27]. Similarly to the two-gene oscillator, our model produces good oscillations when the average number of mRNA molecules is less than one. In fact, the number of proteins oscillates satisfactorily even when the other types of molecules involved in the clock are less than 30. Therefore, this model is a simpler genetic relaxation oscillator than the current two-gene clocks [25]. Our model does not need the activation of a second repressor gene by the PTF, cooperative binding reactions nor the formation of protein multimers.

A single gene with PTF and a negative interaction in the feedback signal is an alternative and simple way of generating reliable oscillations. Our study suggests that PTF, besides increasing robustness in cellular clocks, could be more directly and deeply involved in the production of oscillations than at first thought. Further research is necessary to elucidate the presence and the role of this genetic oscillator in natural cellular clocks. On the other hand, thanks to its simplicity, this model has the potential to be a new tool for engineering synthetic genetic oscillators. In this case the period and amplitude of the oscillations could be possibly controlled by externally manipulating the entry rate of the repressor molecules.

Methods

Biochemical reactions and rates

The biochemical reactions that fully describe the model in the Fig. 2 are as follows:

equation image
(3)

where An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e269.jpg denotes the gene without An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e270.jpg bound to its promoter, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e271.jpg denotes mRNA transcribed from An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e272.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e273.jpg denotes the activator protein translated from An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e274.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e275.jpg denotes the gene with An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e276.jpg bound to its promoter, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e277.jpg denotes the repressor and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e278.jpg denotes An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e279.jpg bound to An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e280.jpg. All the biochemical species are measured in molecules. The description of the rates is as follows: An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e281.jpg is the binding rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e282.jpg to the promoter of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e283.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e284.jpg is the unbinding rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e285.jpg from the promoter of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e286.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e287.jpg is the basal transcription rate, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e288.jpg is the activated transcription rate, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e289.jpg is the degradation rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e290.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e291.jpg is the translation rate, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e292.jpg is the degradation rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e293.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e294.jpg is the binding rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e295.jpg to An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e296.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e297.jpg is the decay rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e298.jpg into An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e299.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e300.jpg is the creation (or entry) rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e301.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e302.jpg is the degradation (or exit) rate of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e303.jpg.

We used standard values within the diffusion limit for the rates [18], [38], [47]. They are as follows: An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e304.jpg moleculesAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e305.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e306.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e307.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e308.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e309.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e310.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e311.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e312.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e313.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e314.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e315.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e316.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e317.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e318.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e319.jpg moleculesAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e320.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e321.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e322.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e323.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e324.jpg molecules hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e325.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e326.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e327.jpg. The cell has a single copy of the gene: An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e328.jpg molecule. The initial conditions are: An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e329.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e330.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e331.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e332.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e333.jpg, and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e334.jpg molecules. The initial conditions have been chosen to obtain a first cycle with an amplitude similar to the limit-cycle oscillations. Note that the rates An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e335.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e336.jpg include the volume of the system An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e337.jpg. Hence, these rates can be written as An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e338.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e339.jpg, where the rates An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e340.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e341.jpg are expressed in MAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e342.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e343.jpg. In order to generate circadian oscillations, first, we varied all the reaction rates, according to the values used in [18], [38] and [47], until we got oscillations with a period of around 24 hours in the stochastic simulation. Then we fine-tuned the oscillations varying rates An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e344.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e345.jpg until a period closer to 24 hours was achieved.

Deterministic and stochastic simulations

Models based on chemical reactions in a well stirred system are usually described by two different formalisms from a mathematical point of view:

Deterministic: this formalism is suitable for large numbers of molecules. It is described by a set of coupled ordinary differential equations that follow the law of mass action. These equations are called reaction rate equations and they can only be solved analytically for simple systems. For more complex systems numerical methods are necessary. In this approach the amount of each chemical species and the time are continuous. The velocity at which reactions occur is given by the reaction rate constants An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e346.jpg, or simply rate.

Stochastic: this formalism is suitable for small numbers of molecules because it takes into account the randomness of the chemical reactions. It is described by the so-called master equation, which is the time evolution of the probability that the system has a certain number of molecules of each chemical species at time An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e347.jpg. Few systems can be solved analytically with the master equation. It is possible, however, to simulate the stochastic behaviour with the Gillespie algorithm [46]. In this approach the amount of each chemical species and the time are discrete, and the rates An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e348.jpg turn into probabilities.

Bifurcation diagram

We calculated the bifurcation diagram for parameters An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e349.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e350.jpg. These are key parameters for two reasons. First, the rate of the negative interaction An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e351.jpg is proportional to An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e352.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e353.jpg when the saturation point is reached. Second, the fast and slow stages in the relaxation oscillations depend on An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e354.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e355.jpg, respectively. Specifically, we studied the range values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e356.jpg that produce stable oscillations through a bifurcation diagram. Then we studied how this range changes when the parameter An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e357.jpg varies.

The bifurcation diagram of the reduced model depending on An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e358.jpg shows two Hopf bifurcation points (Fig. S3A). The first Hopf bifurcation appears at An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e359.jpg molecules hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e360.jpg and the second at An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e361.jpg molecules hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e362.jpg. Most of the values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e363.jpg between these two points produce stable oscillations. Only for a short range of values around these points are the oscillations unstable (white circles in Fig. S3A). The oscillations have an amplitude of from 2,000 to 16,000 molecules, and a period of from 7 to 170 hours (Fig. S3B). The velocity of the reaction An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e364.jpg in (Eq. 3) does not depend on any biomolecule involved in the oscillator. Therefore, parameter An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e365.jpg can be interpreted as an external signal controlling the behaviour of the clock.

The variation of parameter An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e366.jpg changes the position of the two Hopf bifurcation points (white circles in Fig. S4). The different positions of these points define the regions with stable oscillations depending on the values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e367.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e368.jpg (regions I and II in Fig. S4). If parameter An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e369.jpg is increased, the range of values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e370.jpg that produces stable oscillations decreases. This range shrinks faster if An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e371.jpg is greater than 20 hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e372.jpg. We plotted an equivalent graph for the stochastic model because it is more realistic than the reduced graph (black circles in Fig. S4). In particular, we assumed that oscillations occurs in a region if the correlation in the first period is greater than 0.2. The stochastic model produces oscillations in the regions II and III (Fig. S4). The range of oscillations in the complete deterministic model is close to the region II.

The Jacobian matrix

The Jacobian matrix of the reduced system (2) is:

equation image
(4)

where the element An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e374.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e375.jpg are always negative, the element An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e376.jpg is always positive and the element An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e377.jpg can be positive or negative depending on the values of the rates. With the rates given in the section Methods: Biochemical reactions and rates and the fixed point of the reduced system (Fig. 5C) the sign pattern for the Jacobian matrix is:

equation image
(5)

A two-component negative feedback loop is created in the reduced model because An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e379.jpg (see Chapter 9 of the reference [48]). The Jacobian matrix (5) has a tipically sign pattern that produces Hopf bifurcation in chemical systems with two variables [43], [48]. The two-component systems with this sign pattern in the Jacobian matrix are called activator-inhibitor models [48].

Software

Code for stochastic and deterministic simulations was written in FORTRAN and XPPAUT (http://www.math.pitt.edu/bard/xpp/xpp.html), respectively. Simulations have been contrasted using CAIN software (http://cain.sourceforge.net/). The stability analysis to determine steady states and limit cycles was performed with XPPAUT. The histograms and autocorrelation function were plotted using FORTRAN and GNU Octave (http://www.gnu.org/software/octave/). The code for complete and reduced deterministic simulations in XPPAUT is available in File S1 and File S2. The code for stochastic and deterministic simulations in CAIN is available in File S3.

Supporting Information

Figure S1

Time evolution of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e380.jpg with and without a low number molecules. Comparison between deterministic simulation of the time evolution of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e381.jpg with (dashed line) and without (solid line) a low number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e382.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e383.jpg, and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e384.jpg molecules. (Solid line graph: the values of the parameters are as in the section Methods: Biochemical reactions and rates. Dashed line graph: the changed rates are An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e385.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e386.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e387.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e388.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e389.jpg moleculesAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e390.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e391.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e392.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e393.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e394.jpg molecules hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e395.jpg.)

(PDF)

Figure S2

Amplitude and period histograms of the stochastic simulation of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e396.jpg. A, B. Amplitude and period histograms of the stochastic simulation of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e397.jpg, respectively. The values of the parameters are as in the section Methods: Biochemical reactions and rates but now we set An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e398.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e399.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e400.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e401.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e402.jpg moleculesAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e403.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e404.jpg, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e405.jpg hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e406.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e407.jpg molecules hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e408.jpg. (A and B were calculated for 1,000 successive cycles. We assumed that a cycle occurs if the number of proteins An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e409.jpg increases to 1,000 molecules and then decreases to 700 molecules. The amplitude was calculated as the greatest number of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e410.jpg molecules in each cycle. The period was calculated as the time interval that it takes the numbers of proteins An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e411.jpg to reach 1,000 molecules for the first time in two successive cycles.)

(PDF)

Figure S3

Bifurcation diagram of the reduced model. A. Bifurcation diagram depending on An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e412.jpg. The solid/dashed line represents stable/unstable fixed points. Black/white circles are the maximum and minimum values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e413.jpg during unstable/stable oscillations. HB denotes a Hopf Bifurcation point. HBAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e414.jpg and HBAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e415.jpg appear when the value of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e416.jpg is 4.78 and 217.6 molecules hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e417.jpg, respectively. B. Period of the stable oscillations in A.

(PDF)

Figure S4

Oscillatory regions in the reduced and stochastic models depending on An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e418.jpg and An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e419.jpg. Region I. Oscillations in reduced model. Region II. Oscillations in both reduced and stochastic model. Region III. Oscillations in the stochastic model. Region IV. No oscillations in any model. White circles represent the locus of Hopf bifurcations in the reduced model (data are presented in Table S2). Black circles represent locus of oscillations in the stochastic simulation (data are presented in Table S3). We assumed in the stochastic case that oscillations occur in a region if the correlation in the first period is greater than 0.2. (The lines connecting circles are designed to clearly single out the different regions.)

(PDF)

Figure S5

Rate of the negative interaction for different values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e420.jpg. Rate of the negative interaction (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e421.jpg) for different values of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e422.jpg, where An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e423.jpg is the rate of reaction An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e424.jpg. Deterministic simulations A, B, C, D, E, F and G correspond to An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e425.jpg equals 0.0, 0.1, 0.2, 0.3, 0.4, 0.5 and 0.6 hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e426.jpg, respectively. The values of the other parameters are as in the section Methods: Biochemical reactions and rates. The oscillations stop when An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e427.jpg 0.6 hourAn external file that holds a picture, illustration, etc.
Object name is pone.0027414.e428.jpg (G). If An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e429.jpg is increased, An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e430.jpg increases slower, and its maximum value (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e431.jpg) is lower. The value of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e432.jpg corresponds to the peak of the oscillations (An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e433.jpg is the value of the steady state in G).

(PDF)

Figure S6

Hysteresis diagram. Hysteresis diagram depending on An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e434.jpg. The curve is the solution of the equation An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e435.jpg, where An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e436.jpg is assumed constant. The two solid lines in the diagram are the two stable steady states “high” and “low” as a function of An external file that holds a picture, illustration, etc.
Object name is pone.0027414.e437.jpg. The dashed line represents the unstable points in the diagram.

(PDF)

TableS1

Data points of Fig. 4I.

(PDF)

Table S2

Data points of locus Hopf bifurcation in reduced model (Fig. S4).

(PDF)

Table S3

Data points of locus of oscillations with less than 20% of correlation in the first period in the stochastic model (Fig. S4).

(PDF)

Text S1

The negative interaction does not produce an effective high nonlinearity.

(PDF)

File S1

Complete deterministic model (XPPAUT software).

(ODE)

File S2

Reduced deterministic model (XPPAUT software).

(ODE)

File S3

Stochastic and deterministic model (CAIN software).

(XML)

Acknowledgments

Part of this work was carried out by JMMB during his stay at the Novel Computation Group led by Dr. Martyn Amos at MMU. We would also like to thank Rachel Elliott and Niall Murphy for polishing the English, and an anonymous reviewer for useful comments on the manuscript.

Footnotes

Competing Interests: The authors have declared that no competing interests exist.

Funding: This research has been partially funded by Spanish Ministry of Science and Innovation (MICINN) grant BES-2007-16220 and project TIN2009-14421, by the UPM and Madrid Regional Government and by research project BACTOCOM funded by European Commission under FP7, FET proactive program. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1. Rust MJ, Markson JS, Lane WS, Fisher DS, O'Shea EK. Ordered phosphorylation governs oscillation of a three-protein circadian clock. Science. 2007;318:809–812. [PMC free article] [PubMed]
2. O'Neill JS, Reddy AB. Circadian clocks in human red blood cells. Nature. 2011;469:498–503. [PMC free article] [PubMed]
3. O'Neill JS, van Ooijen G, Dixon LE, Troein C, Corellou F, et al. Circadian rhythms persist without transcription in a eukaryote. Nature. 2011;469:554–558. [PMC free article] [PubMed]
4. Dunlap JC. Molecular bases for circadian clocks. Cell. 1999;96:271–290. [PubMed]
5. Young MW, Kay SA. Time zones: a comparative genetics of circadian clocks. Nat Rev Genet. 2001;2:702–715. [PubMed]
6. Elowitz MB, Leibler S. A synthetic oscillatory network of transcriptional regulators. Nature. 2000;403:335–338. [PubMed]
7. Atkinson MR, Savageau MA, Myers JT, Ninfa AJ. Development of genetic circuitry exhibiting toggle switch or oscillatory behavior in Escherichia Coli. Cell. 2003;113:597–607. [PubMed]
8. Fung E, Wong WW, Suen JK, Bulter T, Lee S, et al. A synthetic gene-metabolic oscillator. Nature. 2005;435:118–122. [PubMed]
9. Stricker J, Cookson S, Bennett MR, Mather WH, Tsimring LS, et al. A fast, robust and tunable synthetic gene oscillator. Nature. 2008;456:516–519. [PubMed]
10. Tigges M, Marquez-Lago TT, Stelling J, Fussenegger M. A tunable synthetic mammalian oscillator. Nature. 2009;457:309–312. [PubMed]
11. Toettcher JE, Mock C, Batchelor E, Loewer A, Lahav G. A synthetic-natural hybrid oscillator in human cells. Proc Natl Acad Sci USA. 2010;107:17047–17052. [PMC free article] [PubMed]
12. Danino T, Mondragon-Palomino O, Tsimring L, Hasty J. A synchronized quorum of genetic clocks. Nature. 2010;463:326–330. [PMC free article] [PubMed]
13. Goodwin BC. Oscillatory behavior in enzymatic control processes. Adv Enzyme Regul. 1965;3:425–438. [PubMed]
14. Goldbeter A. A model for circadian oscillations in the Drosophila period protein (PER). Proc R Soc London Ser B. 1995;261:319–324. [PubMed]
15. Ruoff P, Rensing L. The temperature-compensated Goodwin model simulates many circadian clock properties. J Theor Biol. 1996;179:275–285.
16. Leloup JC, Gonze D, Goldbeter A. Limit cycle models for circadian rhythms based on transcriptional regulation in Drosophila and Neurospora. J Biol Rhythms. 1999;14:433–448. [PubMed]
17. Ruoff P, Vinsjevik M, Monnerjahn C, Rensing L. The Goodwin model: simulating the effect of light pulses on the circadian sporulation rhythm of Neurospora crassa. J Biol Rhythms. 2001;209:29–42. [PubMed]
18. Gonze D, Halloy J, Goldbeter A. Robustness of circadian rhythms with respect to molecular noise. Proc Natl Acad Sci USA. 2002;99:673–678. [PMC free article] [PubMed]
19. Bratsun D, Volfson D, Tsimring LS, Hasty J. Delay-induced stochastic oscillations in gene regulation. Proc Natl Acad Sci USA. 2005;102:14593–14598. [PMC free article] [PubMed]
20. Griffith JS. Mathematics of cellular control processes I. Negative feedback to one gene. J Theor Biol. 1968;20:202–208. [PubMed]
21. Novák B, Tyson JJ. Design principles of biochemical oscillators. Nat Rev Mol Cell Biol. 2008;9:981–991. [PMC free article] [PubMed]
22. Widder S, Schicho J, Schuster P. Dynamic patterns of gene regulation I: Simple two-gene systems. J Theor Biol. 2007;246:395–419. [PubMed]
23. Reppert SM, Weaver DR. Coordination of circadian timing in mammals. Nature. 2002;418:935–941. [PubMed]
24. Gallego M, Virshup DM. Post-translationalmodifications regulate the ticking of the circadian clock. Nat Rev Mol Cell Biol. 2007;8:139–148. [PubMed]
25. Purcell O, Savery NJ, Grierson CS, di Bernardo M. A comparative analysis of synthetic genetic oscillators. J R Soc Interface. 2010;7:1503–1524. [PMC free article] [PubMed]
26. Barkai N, Leibler S. Circadian clocks limited by noise. Nature. 2000;403:267–268. [PubMed]
27. Smolen P, Baxter DA, Byrne JH. Modeling circadian oscillations with interlocking positive and negative feedback loops. J Neurosci. 2001;21:6644–6656. [PubMed]
28. Hasty J, Isaacs F, Dolnik M, McMillen D, Collins JJ. Designer gene networks: Towards fundamental cellular control. Chaos. 2001;11:207–220. [PubMed]
29. Leloup JC, Goldbeter A. Toward a detailed computational model for the mammalian circadian clock. Proc Natl Acad Sci USA. 2003;100:7051–7056. [PMC free article] [PubMed]
30. François P. A model for the Neurospora circadian clock. Biophys J. 2005;88:2369–2383. [PMC free article] [PubMed]
31. Guantes R, Poyatos JF. Dynamical principles of two-component genetic oscillators. PLoS Comput Biol. 2006;2:e30. [PMC free article] [PubMed]
32. Hong CI, Jolma IW, Loros JJ, Dunlap JC, Ruoff P. Simulating dark expressions and interactions of frq and wc-1 in the Neurospora circadian clock. Biophys J. 2008;94:1221–1232. [PMC free article] [PubMed]
33. Conrad E, Mayo AE, Ninfa AJ, Forger DB. Rate constants rather than biochemical mechanism determine behaviour of genetic clocks. J R Soc Interface. 2008;5:S9–S15. [PMC free article] [PubMed]
34. Munteanu A, Constante M, Isalan M, Sole R. Avoiding transcription factor competition at promoter level increases the chances of obtaining oscillation. BMC Syst Biol. 2010;4:66. [PMC free article] [PubMed]
35. Becskei A, Seraphin B, Serrano L. Positive feedback in eukaryotic gene networks: cell differentiation by graded to binary response conversion. EMBO J. 2001;20:2528–2535. [PMC free article] [PubMed]
36. Ferrell JE. Self-perpetuating states in signal transduction: positive feedback, double-negative feedback and bistability. Curr Opin Cell Biol. 2002;14:140–148. [PubMed]
37. Tsai TY, Choi YS, Ma W, Pomerening JR, Tang C, et al. Robust, tunable biological oscillations from interlinked positive and negative feedback loops. Science. 2008;321:126–129. [PMC free article] [PubMed]
38. Vilar JMG, Kueh HY, Barkai N, Leibler S. Mechanisms of noise-resistance in genetic oscillators. Proc Natl Acad Sci USA. 2002;99:5988–5992. [PMC free article] [PubMed]
39. Mondragón-Palomino O, Danino T, Selimkhanov J, Tsimring L, Hasty J. Entrainment of a population of synthetic genetic oscillators. Science. 2011;333:1315–1319. [PubMed]
40. Griffith JS. Mathematics of cellular control processes II. Positive feedback to one gene. J Theor Biol. 1968;20:209–216. [PubMed]
41. Elowitz MB, Levine AJ, Siggia ED, Swain PS. Stochastic gene expression in a single cell. Science. 2002;297:1183–1186. [PubMed]
42. Swain PS, Elowitz MB, Siggia ED. Intrinsic and extrinsic contributions to stochasticity in gene expression. Proc Natl Acad Sci USA. 2002;99:12795–12800. [PMC free article] [PubMed]
43. Murray JD. Mathematical biology I: An introduction. New York: Springer. 2002;551
44. Strogatz SH. Nonlinear dynamics and chaos: with applications to physics, biology, chemistry, and engineering. Addison-Wesley, Reading, MA. 1994;498
45. Tyson JJ, Chen KC, Novák B. Sniffers, buzzers, toggles and blinkers: dynamics of regulatory and signaling pathways in the cell. Curr Opin Cell Biol. 2003;15:221–231. [PubMed]
46. Gillespie DT. Exact stochastic simulation of coupled chemical reactions. J Phys Chem. 1977;81:2340–2361.
47. Dublanche Y, Michalodimitrakis K, Kümmerer N, Foglierini M, Serrano L. Noise in transcription negative feedback loops: simulation and experimental analysis. Mol Syst Biol. 2006;2:41. [PMC free article] [PubMed]
48. Fall CP, Marland ES, Wagner JM, Tyson JJ. Computational Cell Biology. Springer, Berlin. 2002;468
49. Buchler NE, Louis M. Molecular titration and ultrasensitivity in regulatory networks. J Mol Biol. 2008;384:1106–1119. [PubMed]
50. Buchler NE, Cross FR. Protein sequestration generates a flexible ultrasensitive response in a genetic network. Mol Syst Biol. 2009;5:272. [PMC free article] [PubMed]

Articles from PLoS ONE are provided here courtesy of Public Library of Science

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

  • MedGen
    MedGen
    Related information in MedGen
  • PubMed
    PubMed
    PubMed citations for these articles
  • Substance
    Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...