Analysis of primitive genetic interactions for the design of a genetic signal differentiator

Wolfgang Halter, Richard M. Murray, Frank AllgöwerSynthetic 2019 · DOI

TL;DR

This paper investigates how basic biological processes like transcription and translation can be combined to build a genetic "differentiator" that mimics a mathematical derivative. Using cell-free experiments and modeling, they found that while these processes can approximate the desired functions, their performance is highly dependent on operating conditions and inherent biological nonlinearities. This work provides a more realistic understanding of building complex genetic circuits.

What Problem Does This Paper Address?

The field of synthetic biology aims to engineer biological systems to perform specific functions, much like designing electronic circuits. A key challenge is creating complex genetic circuits that reliably implement desired behaviors. One such behavior is a "differentiator," which outputs a signal proportional to the rate of change of its input. This is analogous to how a derivative in calculus describes the instantaneous rate of change of a function.

Previous work proposed a theoretical framework for building genetic differentiators by combining fundamental genetic interactions (like transcription, translation, and promoter binding) to mimic basic control system components (gain, integrator, difference). However, the practical realization of these genetic components and their limitations, especially due to biological nonlinearities and resource constraints, remained unclear. This paper addresses this gap by experimentally verifying and refining the requirements for these primitive genetic interactions and analyzing their impact on the performance of a genetic signal differentiator.

Methods

The researchers employed a multi-pronged approach:

  • Cell-Free Experimental System (TX-TL): They utilized a transcription-translation (TX-TL) system derived from E. coli. This system allows for rapid prototyping and testing of genetic circuits outside of living cells, offering a controlled environment.
  • Experimental Probing: They designed experiments where simple genetic constructs were exposed to various input signals (e.g., step changes in DNA concentration) and their outputs (mRNA and protein levels) were measured over time.
  • Semi-Mechanistic Modeling: They developed mathematical models (ODEs) that describe the biochemical reactions involved in gene expression. These models were fitted to the experimental data to estimate kinetic parameters.
  • Linearization and Transfer Function Analysis: They analyzed the linearized versions of their models to derive transfer functions, which are standard tools in control theory for characterizing system behavior across different frequencies.
  • Describing Function Approach: To account for the nonlinearities of the biological system, they used a describing function approach. This method analyzes the system's response to sinusoidal inputs by looking at the fundamental frequency component of the output, allowing for an assessment of nonlinear behavior.

Datasets

The paper does not explicitly mention training or evaluation datasets in the traditional machine learning sense. Instead, they generated their own experimental data using the TX-TL system. This data consists of time-series measurements of mRNA and protein concentrations in response to controlled input changes (e.g., varying DNA concentrations). This experimental data serves as the benchmark for fitting their models and validating their findings.

Key Insights and Ideas

  • Transcription is a PT1 element, not just a gain: Contrary to initial assumptions, transcription isn't a simple instantaneous gain. The finite rate of mRNA degradation means it behaves more like a first-order system (PT1 element) with a delay.
  • Biological processes are inherently nonlinear and resource-limited: Transcription and translation are not perfectly linear. They are subject to saturation effects due to finite cellular resources (like RNAP and ribosomes), which significantly impacts their performance, especially at higher input levels.
  • Combinatorial promoters offer only approximate signal differences: While combinatorial promoters can be engineered to respond to the difference between two signals, this realization is not exact. The ability to approximate a true difference is limited and highly dependent on the relative concentrations of the interacting molecules and their binding affinities.
  • Cell-free systems are valuable for characterizing fundamental biological modules: The TX-TL system proved effective for isolating and characterizing the behavior of basic genetic components, providing crucial data for building more complex synthetic circuits.
  • Operating conditions are critical for genetic circuit performance: The performance of a genetic differentiator is highly sensitive to the specific operating regime, including the amplitude and offset of input signals, and the relative concentrations of molecular components.

Key Results

  • Requirement Verification:
    • Requirement 4 (mRNA degradation much faster than protein degradation) was verified.
  • Requirements 1 and 2 (linearity of production and degradation terms) were verified with approximately 95% accuracy within specific operating ranges.
  • Linearized Model: The linearized model of the genetic differentiator approximates a system with a relative degree one, indicating a delayed gain.
  • Nonlinear Model Performance:
    • The nonlinear model's performance as a differentiator is highly dependent on the offset of the input signal.
  • An optimal offset value (around 10) was identified where the nonlinear model closely matched the linearized model's performance and approximated an ideal differentiator for frequencies up to approximately 0.02 rad/min.
  • Time-domain analysis showed that at higher input offsets, the output signal becomes distorted, indicating nonlinear effects.
  • Combinatorial Promoter Limitations: Experiments demonstrated that combinatorial promoters can only approximate signal differences to a limited extent, with performance being sensitive to dissociation constants and DNA concentrations.

Limitations and Future Work

  • Limitations:
    • The TX-TL system's resource degradation limits experimental observation to a finite time frame (around 200 minutes), potentially biasing parameter estimates.
  • The exact realization of a signal difference using combinatorial promoters is not possible; approximations are necessary.
  • The describing function approach relies on the assumption that higher harmonics of the output signal can be neglected, which may not hold under all conditions.
  • Future Work:
    • The authors suggest exploring sequestration-based mechanisms as a potentially more robust alternative for realizing signal differences in future work.
  • Further investigation into the impact of nonlinearities and resource limitations on the performance of more complex genetic circuits is warranted.
  • The study of protein degradation's effect on the transfer function, which could introduce a zero in the right-half plane and a sign change, is mentioned as a potential avenue for future analysis.

Summary updated

Abstract

We study the dynamic and static input-output behavior of several primitive genetic interactions and their effect on the performance of a genetic signal differentiator. In a simplified design, several requirements for the linearity and time-scales of processes like transcription, translation and competitive promoter binding were introduced. By experimentally probing simple genetic constructs in a cell-free experimental environment and fitting semi-mechanistic models to these data, we show that some of these requirements can be verified, while others are only met with reservations in certain operational regimes. Analyzing the linearized model of the resulting genetic network, we conclude that it approximates a differentiator with relative degree one. Taking also the discovered nonlinearities into account and using a describing function approach, we further determine the particular frequency and amplitude ranges where the genetic differentiator can be expected to behave as such.
BibTeX
@inproceedings{halter2019analysis,
  author = {Wolfgang Halter and Richard M. Murray and Frank Allgöwer},
  title = {{Analysis of primitive genetic interactions for the design of a genetic signal differentiator}},
  booktitle = {Synthetic},
  year = {2019},
  doi = {10.1093/synbio/ysz015},
}
Figure