Literature Read and Summaries

Papers to Read

• Monk NA. Oscillatory expression of Hes1, p53, and NF-KB driven by transcriptional time delays
• Isaacs FJ et al. Prediction and measurement of an autoregulatory genetic module
• Atkinson MR et al. Development of genetic circuitry exhibiting toggle switch or oscillatory behavior in Escherichia coli
• Zhou TS, Chen LN, Aihara K. Molecular communication through stochastic synchronization induced by extracellular fluctuations
• McMillen D, Kopell N, Hasty J, Collins JJ. Synchronizing genetic relaxation oscillators by intercell signaling.

Summaries

Kaern M, Elston TC, Blake WJ, Collins JJ. (2005). Stochasticity in Gene expression: from theories to phenotypes. Nature Rev Genet 6: 451–64.

This review gives an introduction to the theoretical mechanisms of noise in gene expression levels in genotypically identical cells. In modeling stochastic behaviour, you need to focus on random formation and decay of single molecules and multi-component complexes. So a major part of research on noise illustrates new techniques developed to do single-cell analyses. It would be beneficial to think of ways that I could use multi-component complexes in order to enlarge the focus from a single molecule to a larger complex. For example, I think measuring something like hemoglobin could show more dramatic measurements in noise variation because it requires more than a single molecule to make a functional unit. Gene expression noise is defined as the relative deviation from the average, measuring as the standard deviation divided by the mean. Kaern covers some factors that influence the amount of noise in gene expression. The ‘finite-number effect’ is the most notable manifestation of noise – with a smaller number of molecules affecting protein abundance in a compartment, noise increases. Varying the rate of transcription causes a greater change in gene expression noise than varying the translational efficiency (the number of proteins per mRNA molecule). It’s predicted that the rate of transcription should be most sensitive to variation in the regulatory signal at intermediate induction levels. Mostly factors are addressed to learn how to minimize noise, but it could be interesting to see how to maximize noise and a potential application. In synthetic systems, longer cascades have more noise because there are more steps. Negative feedback loops provide a noise-reduction mechanism, and may minimize noise in downstream processes too. But Kaern also notes that negative feedback can also have a destabilizing effect if it involves a time delay (1) Positive feedback generally amplifies fluctuations and population heterogeneity, and possibly yielding bimodal population distributions. (2) I thought mechanisms causing distinct phenotypic consequences would be interesting by using noise to create two different and useful cell populations, perhaps that somehow function together. One oscillatory network design (3) consisted of an activator and repressor allowing dampened but synchronized oscillations; Kaern here notes that fluctuations in gene expression can in theory cause the emergence of oscillations that would not appear otherwise. Finally, the review concludes with biological significance of stochasticity. It is particularly beneficial to microbial cells that need to adapt efficiently to sudden environmental changes – a mechanism for “sampling” distinct physiological states without genetic mutation. I’d like to come up with some synthetic use and way to construct “sampling”.

Interesting / Useful Papers Cited: 1. Monk NA. Oscillatory expression of Hes1, p53, and NF-KB driven by transcriptional time delays 2. Isaacs FJ et al. Prediction and measurement of an autoregulatory genetic module 3. Atkinson MR et al. Development of genetic circuitry exhibiting toggle switch or oscillatory behavior in Escherichia coli

Isaacs FJ, Hasty J, Cantor CR, Collins JJ. (2003). Prediction and measurement of an autoregulatory genetic module. PNAS 100: 7714 – 19.

In this paper, the authors constructed a positive feedback module in E. coli using the bacteriophage lambda genetic switch mechanism. “Feedback loops…can include…the rapid switching between two or more outputs, and even the suppression or amplification of noise.” They simplified the lambda phage system down to the OR – right operator in the promoter – and the cI gene which codes for a temperature sensitive repressor protein. GFP was placed downstream of cI under the same promoter to be expressed together. They varied the temperature of the environment to tune the stability of the repressor, varying the degree of activation portion of the positive feedback loop. By destabilizing the repressor, they hoped to control the positive feedback loop. They found, however, that noise plays a significant role in the tuning of the positive feedback loop. As they increased the temperature to 39 degrees C, the monostable state split into two coexisting stable populations, and remained that way through 40 degrees. So stochasticity in the destabilization of the repressor either causes the cell to amplify the amount of GFP by a positive feedback loop, or causes a minimal amount of GFP – “trademark bistability of the positive feedback architecture”. In the discussion, the author also mentioned differing the plasmid copy numbers (since this noise is a finite number effect). So to make a noisier system, we could use a smaller abundance of repressor. So from this paper, I could explore making a similar system in which we utilize noise by destabilizing a repressor-like element in order to make bistable cell populations…for something.

Becskei A, Seraphin B, Serrano L. (2001). Positive feedback in eukaryotic gene networks: cell differentiation by graded to binary response conversion. EMBO 20: 2528 – 35.

In this paper, the authors worked with S. cerevisiae. The introduction of this paper gives a good clear background for positive feedback loops / bistability – might be useful in the future for a write-up. The point of this paper was to see if positive feedback generates a binary response in eukaryotic circuits like it does in prokaryotic. In this construct, they used a promoter (?) CYC1 to regulate GFP and an activator (?) rtTA, which can be induced with doxycycline (rtTA is being induced I believe, not the promoter). rtTA binds to CYC1 which creates a positive feedback loop (don’t understand the mechanism so much here). This circuit was used in a plasmid and in the chromosome, and both locations caused a bimodal GFP distribution. Both the induction level and the gene copy number contributed to the activation level. Cells randomly turned on GFP production, switching from off to on (but not switching the other way). So this paper also observed bistability with a positive feedback loop, but the system does not stochastically switch both ways (not a toggle switch). The switching time itself is the random part. So from this paper, I learned that positive feedback loops seem inherently bistable in their output, and the output could be a direct product of noise. Copy number and induction level is important. We could make a device to perhaps make two different populations, as some useful cell differentiation.

Vilar JMG, Kueh HY, Barkai N, Leibler S. (2002). Mechanisms of noise-resistance in genetic oscillators. PNAS 5988 – 92.

This paper was really cool, but very mathematical (could figure some out if refreshed on Mathematical Methods for Scientists info) – quite hard to read, I skimmed most. Gist: Authors created a simulation of a genetic oscillator using mathematical models. They made a deterministic model and a stochastic model modifying the deterministic model. Their point in making the deterministic model was to study strategies of minimizing noise (compared to the stochastic model). But interestingly they found that parameter values to produce a stable state in the deterministic model continue to produce oscillation cycles in the stochastic model. So theoretically, the presence of noise can actually initiate a new cycle and drive an oscillatory circuit. Noise here was due to finite number effect, with respect to repressor degradation rate and the activator-repressor complex. Figure 6. So this paper is exciting because it models a way in which one can take advantage of cellular noise to perform functions not possible by deterministic means. So constructing a noise-driven oscillator is an idea. On a population scale though, the application would need to take into account that all cells would not be oscillating together, and stop oscillating at different times.

Wang J, Zhang J, Yuan Z, Zhou T. (2007). Noise-induced switches in network systems of the genetic toggle switch. BMC Sys Biol 1: 50.

I didn’t like this paper much; it was very difficult to read because of the degree of math and the grammar – it may have been translated. But it was another article about bistability and switches. Here, they elucidate the single and coupled genetic toggle switch systems in E. coli through mathematical modeling of the Gardner single toggle switch (2 repressors facing opposite ways) and a multicellular toggle switch with quorum sensing. In the Gardner single toggle switch, the degradation rate of the repressors can induce successive switching between two stable steady states. This is the design of the Gardner switch, so the noise is governing the timing I believe, based on Figure 2A. In the multicellular gene regulatory network, the system acts as a toggle switch and uses an intercellular signaling mechanism to couple between cells. The signalizing system is Lux operon, in which the autoinducer (AI) made by LuxI diffuses across the cell membrane. One notable thing I did get out of this part was that noise in such a signaling system can make noise more notable, and lead to a robust collective rhythm. So noise in quorum sensing may be a very useful factor to consider, and perhaps using the Lux operon for signaling or positive feedback.

Articles Cited: 1. Zhou TS, Chen LN, Aihara K. Molecular communication through stochastic synchronization induced by extracellular fluctuations 2. McMillen D, Kopell N, Hasty J, Collins JJ. Synchronizing genetic relaxation oscillators by intercell signaling.

Cox CD, Peterson GD, Allen MS, Lancaster JM, McCollum JM, Austin D, Yan L, Sayler GS, Simpson ML. (2003). Analysis of noise in quorum sensing. OMICS 7: 317 – 34.

It’s shown especially by the bacteriophage lambda system that noise is pivotal in gene circuitry (pathway depends on the chance early and strong production of cI). Like the lambda switch, bacterial quorum sensing systems operate in a population and contain a bistable switching element, so it’s likely noise plays a pivotal role in QS too. For their simulation, they used the Lux regulon. The threshold in the Lux system is the concentration of the LuxR-AI (autoinducer) complex required to initiate the state transition (not induced → induced). The noise in the LuxR-AI complex increases as the threshold is approached. Once the threshold is reached, the production of LuxI increases rapidly with a small AI increase. Some cells would reach the threshold earlier than others because of the noise here. These early switching cells reinforce their behavior throughout the population by production of more AI (helping other cross the threshold). The authors hypothesize that although this does not seem like a dynamic noise system, there is a benefit in distributing this additional positive feedback throughout the population, versus in individual cells. By doing so, the noise acting as positive feedback distributed may act to ensure the operation of the QS circuit even if other organisms or processes are actively interfering with the signal system. This last point (underlined) sounds very promising for building a device. Quorum sensing may be very beneficial here. Nicely worded intro sentence: “The noise sources in genetic circuits are most often found in pairs at the point of molecular synthesis (or polymerization or complex formation) and the associated point of decay due to random timing and discrete nature of these events.”

Zhou T, Chen L, Wang R, Aihara K. (2004). Intercellular communication induced by random fluctuations. Genome Informatics 15: 223 – 33.

Nice sentence: “In particular, gene regulation is an inherently noisy process, from transcriptional control, alternative splicing, translation, diffusion to chemical modification reactions of transcriptional factors, which all involve stochastic fluctuations owing to low copy numbers of man species per cell.” This paper implements a multi-cell synthetic system model to show theoretically that noise can be exploited to facilitate mutual communication. Their model uses LuxI and LuxR with a pLacLux0 promoter. AI is distributed into the extracellular environment (Figure 1) and freely diffuses through cell membranes. I believe coupling here refers to synchronized behavior, but may just be the communication through AI signaling between two cells (sending to reception). Noise exists in the transcription and translation, AI synthesis, degradation reactions, diffusion, (and cell coupling). Without noise, the system converges into a stable equilibrium. With strong noise, the multi-cell system is synchronously oscillated. This paper altogether supports that noise enhances temporal regularity of a dynamic system. So this paper gives a lot of mathematical insight on noise enhancing an oscillatory system by quorum sensing. It is more in-depth than the Cox et al paper; it helps to identify the noisy factors that we can manipulate. Can speak w/ Dr. Heyer to understand it better.

Acar M, Becskei A, van Ourdenaarden A. (2005). Enhancement of cellular memory by reducing stochastic transitions. Nature 435: 228 – 32.

This paper was mentioned in the Kaern 2005 review. It explores the key parameters that determine the stability of cellular memory using the yeast galactose-signaling network as a model system. The system has three loops – 2 positive, 1 negative. Since there are 2 positive feedback loops, it’s possible to have multistability. They studied these feedback loops by doing loop knockouts. They put YFP under control of Gal1, so it’s expressed by the positive feedback loops (Figure 1). The positive mediated by Gal3p generates two stable expression states with persistent memory of previous galactose consumption states. The other positive loop amplifies the difference between the two states. The negative feedback loop reduces the strength of the first positive loop. (Figure 2) When they knocked out the negative feedback loop (Gal80), they found that the negative feedback loop weakens the effect of the positive feedback loop, and the system does not display a memory of the initial galactose consumption state. So the signaling pathway can store information on previous galactose exposures for generations. By opening the negative feedback loop, the memory persistence can be “tuned from hours to months” By further understanding this system, we could use positive feedback loops (without a negative feedback loop) to differentiate cells into stable populations. But we can also introduce a similar negative feedback loop to allow switching between states. Need to re-read this article more – don’t quite understand the memory component.