Network design principle for robust oscillatory behaviors with respect to biological noise

Lingxia Qiao, Zhibo Zhang, Wei Zhao, Ping Wei, Lei Zhang

TL;DR

Biological systems often rely on oscillations to encode information, but these oscillations are susceptible to noise. This paper systematically explores how different network structures in gene regulatory systems can achieve robust oscillations despite this noise. They discover that certain network motifs, particularly those involving a repressilator with positive autoregulation, are inherently more resilient, and the mechanisms for noise buffering depend on the type of noise: long periods help with parameter variability, while high amplitudes are key for buffering reaction stochasticity.

What Problem Does This Paper Address?

Many biological processes, from cell cycles to circadian rhythms, exhibit oscillatory behavior. These oscillations are crucial for coordinating cellular functions. However, biological systems are inherently noisy, with fluctuations arising from both external factors (parameter variability, or extrinsic noise) and the inherent randomness of molecular reactions (stochasticity, or intrinsic noise). This noise can easily disrupt the precise timing and amplitude of oscillations, leading to inaccurate signaling. The central question this paper addresses is: how can biological networks be designed to maintain accurate oscillations in the face of this pervasive noise? Previous work has identified some mechanisms for noise buffering in specific circuits, but a general, systematic understanding of network design principles for robust oscillation is lacking.

Methods

The authors employ a comprehensive, bottom-up approach to tackle this problem. They begin by defining a mathematical model for two- and three-node transcriptional regulatory networks, using ordinary differential equations (ODEs) to describe deterministic dynamics. To account for noise, they develop stochastic models:

  • Extrinsic Noise: Modeled as random fluctuations in kinetic parameters (e.g., production rates, degradation rates), with the noise magnitude proportional to the parameter value.
  • Intrinsic Noise: Modeled by simulating the discrete nature of chemical reactions using the Gillespie algorithm or chemical Langevin equations, focusing on molecular numbers rather than concentrations.

They then systematically screen a large number (1955) of possible two- and three-node network topologies. For each topology, they:

  1. Identify those capable of oscillation in the absence of noise using deterministic simulations.
  2. For the oscillatory topologies (474 identified), they sample numerous parameter sets and simulate their behavior under extrinsic and intrinsic noise conditions.
  3. Quantify oscillation accuracy using the dimensionless correlation time ($\tau/T$), where $\tau$ is the correlation time and $T$ is the period. A higher $\tau/T$ indicates more accurate oscillation.
  4. Assess robustness by looking at the 90th percentile of $\tau/T$ across different parameter sets for a given topology.

Finally, they analyze the relationship between network topology, oscillation period, amplitude, and robustness to different noise sources. They also experimentally validate their findings using a synthetic NF-κB signaling circuit.

Datasets

Not discussed in the paper. The study relies on simulated data generated from mathematical models.

Key Insights and Ideas

  • The "Repressilator with Positive Autoregulation" Motif is a Champion: This specific network structure consistently emerges as the most robust against both extrinsic and intrinsic noise, outperforming other common motifs like the activator-inhibitor oscillator.
  • Positive Autoregulation is a General Enhancer: Adding positive autoregulation to a network motif generally improves its robustness to noise, regardless of the specific noise source.
  • Noise Buffering Mechanisms are Noise-Specific: This is a crucial finding. The way a network filters noise is not one-size-fits-all.
  • Extrinsic Noise (Parameter Variability): Long oscillation periods are the key to buffering this type of noise. The system effectively "averages out" parameter fluctuations over longer timescales.
  • Intrinsic Noise (Reaction Stochasticity): High oscillation amplitudes are critical for filtering this noise. Larger amplitudes mean larger deviations from the mean, making the inherent randomness of individual reactions less impactful on the overall oscillation.
  • Topology Trumps Parameter Tuning (for Robustness): While tuning kinetic parameters can improve oscillation accuracy, changing the network topology to a more robust motif (like the repressilator with positive autoregulation) offers a more fundamental and effective way to achieve resilience.

Key Results

  • Out of 1955 two- and three-node network topologies, 474 were found to be capable of oscillation.
  • Topologies containing the repressilator with positive autoregulation (categories C1 and C2) showed significantly higher robustness to extrinsic noise compared to activator-inhibitor topologies (C3).
  • For intrinsic noise, the robustness ranking was similar, with C1 and C2 categories performing best.
  • A strong positive correlation was found between oscillation period and robustness to extrinsic noise (Spearman coefficient $\approx 0.94$).
  • A strong positive correlation was found between oscillation amplitude and robustness to intrinsic noise (Spearman coefficient $\approx 0.81$).
  • Experimental validation using synthetic NF-κB signaling circuits confirmed that adding a repressilator motif to an existing circuit improved oscillation accuracy and reduced period variability in the presence of noise.

Limitations and Future Work

  • Simplified Network Size: The study focused on two- and three-node networks. While these capture core motifs, real biological networks are often much larger and more complex.
  • Single Noise Source Assumption: The analysis primarily considered extrinsic or intrinsic noise, not their simultaneous presence. Real biological systems experience both concurrently.
  • Model Simplifications: The use of Hill functions for modeling transcriptional regulation, while computationally efficient, might not capture all the nuances of detailed elementary reaction steps. The authors acknowledge that a more detailed model might be necessary for certain scenarios.
  • Focus on Oscillation Accuracy: The study concentrated on the accuracy of oscillations (period and amplitude) and did not explore other noise-induced dynamic changes, such as the loss of multistability or the emergence of oscillations in previously non-oscillatory systems.

Future work could involve extending the analysis to larger networks, investigating the combined effects of multiple noise sources, and incorporating more detailed kinetic models. Exploring topologies that can oscillate only in the presence of noise is also highlighted as an interesting avenue.

Summary updated

Abstract

Oscillatory behaviors, which are ubiquitous in transcriptional regulatory networks, are often subject to inevitable biological noise. Thus, a natural question is how transcriptional regulatory networks can robustly achieve accurate oscillation in the presence of biological noise. Here, we search all two- and three-node transcriptional regulatory network topologies for those robustly capable of accurate oscillation against the parameter variability (extrinsic noise) or stochasticity of chemical reactions (intrinsic noise). We find that, no matter what source of the noise is applied, the topologies containing the repressilator with positive autoregulation show higher robustness of accurate oscillation than those containing the activator-inhibitor oscillator, and additional positive autoregulation enhances the robustness against noise. Nevertheless, the attenuation of different sources of noise is governed by distinct mechanisms: the parameter variability is buffered by the long period, while the stochasticity of chemical reactions is filtered by the high amplitude. Furthermore, we analyze the noise of a synthetic human nuclear factor κB (NF-κB) signaling network by varying three different topologies and verify that the addition of a repressilator to the activator-inhibitor oscillator, which leads to the emergence of high-robustness motif-the repressilator with positive autoregulation-improves the oscillation accuracy in comparison to the topology with only an activator-inhibitor oscillator. These design principles may be applicable to other oscillatory circuits.
BibTeX
@inproceedings{qiao2022network,
  author = {Lingxia Qiao and Zhibo Zhang and Wei Zhao and Ping Wei and Lei Zhang},
  title = {{Network design principle for robust oscillatory behaviors with respect to biological noise}},
  booktitle = {eLife},
  year = {2022},
  doi = {10.7554/elife.76188},
}
Figure