Multicellular Feedback Control Strategies in Synthetic Microbial Consortia: From Embedded to Distributed Control

TL;DR

This article reviews recent advances in multicellular feedback control, where sensing, computation, and actuation are distributed across different cell populations within synthetic microbial consortia, giving rise to biological multiagent control systems governed by molecular communication.

Abstract

Living organisms rely on endogenous feedback mechanisms to maintain homeostasis in the presence of uncertainty and environmental fluctuations. An emerging challenge at the interface of control systems engineering and synthetic biology is the design of reliable feedback strategies to regulate cellular behavior and collective biological functions. In this article, we review recent advances in multicellular feedback control, where sensing, computation, and actuation are distributed across different cell populations within synthetic microbial consortia, giving rise to biological multiagent control systems governed by molecular communication. From a control-theoretic perspective, these consortia can be interpreted as distributed biomolecular control systems, where coordination among populations replace embedded regulation. We survey theoretical frameworks, control architectures, and modeling approaches, ranging from aggregate population-level dynamics to spatially aware agent-based simulations, and discuss experimental demonstrations in engineered \textit{Escherichia coli} consortia. We highlight how distributing control functions across populations can reduce metabolic burden, mitigate retroactivity, improve robustness to uncertainty, and enable modular reuse of control components. Beyond regulation of gene expression, we discuss the emerging problem of population composition control, where coordination among growing and competing cell populations becomes an integral part of the control objective. Finally, we outline key open challenges that must be addressed before multicellular control strategies can be deployed in real-world applications such as biomanufacturing, environmental remediation, and therapeutic systems. These challenges span modeling and simulation, experimental platform development, coordination and composition control, and long-term evolutionary stability.

Figures (6)

• Figure 1: Multicellular control conceptual framework. (A) Key ingredients of a cellular consortium. (B) Two-population architecture in which a Controller cell population regulates the expression of a gene within a Target population through quorum sensing (QS)--mediated communication, forming a closed feedback loop (reproduced from martinelli2022multicellular). (C) Four-population architecture implementing a distributed PID control, with separate cell populations responsible for proportional, integral, and derivative actions whose combined signals regulate the target population, enabling robust and tunable control of collective behavior (reproduced from martinelli2025multicellular).
• Figure 2: In Silico validation of the multicellular feedback control strategy using aggregate and agent-based models. (a) Aggregate population model: output of the Target population under set-point and time-varying reference signals, showing accurate regulation, limited overshoot, and stable steady-state behavior despite nonlinear dynamics and diffusive communication. (b) Agent-based simulations in a microfluidic-like spatial domain: population-averaged Target output tracking trapezoidal and sinusoidal reference signals with negligible phase delay and low cell-to-cell variability. (c) Robustness analysis: effect of increasing spatial separation between Controller and Target populations and of parameter perturbations ($\pm 20\%$) on regulation performance, showing preserved stability and reduced but nonvanishing dynamic range under severe communication attenuation. (d) Composition independence and heterogeneity: regulation performance under varying Controller-to-Target population ratios and single-cell parameter variability, demonstrating that accurate control is maintained even when Controllers represent a small fraction of the consortium and when biological parameters vary across cells. Together, the results indicate that the proposed distributed feedback architecture provides reliable regulation across spatial scales, biological noise, and population composition. All panels are reproduced from fiore2017silico.
• Figure 3: Validation of two-population feedback control. (A) Schematic of two-population architecture with bidirectional QS communication and molecular titration for error computation. (B) Coefficient of variation across biological replicates for closed-loop (blue) versus open-loop (orange) control at different IPTG concentrations, showing sixfold reduction at 3 $\mu$M. (C) Normalized target fluorescence versus consortium composition (target percentage), demonstrating composition-independent output for closed-loop (flat, p$>$0.05) versus composition-dependent open-loop (negative slope, p$<$0.01). (D) Input-output relationships showing improved linearity for closed-loop ($R^2=0.91$) versus open-loop ($R^2=0.67$) control. All panels are reproduced from salzano2025vivo.
• Figure 4: Approaches to ratiometric control. (A) Schematic of a composition control scheme (reproduced from salzano2022ratiometric). (B) Representation of a dual chamber bioreactor architecture for composition control (left), and experimental data of biomass and composition regulation with a switching and learning-based controller (reproduced from brancato2025bioreactor). (C) Composition control in microfluidics exploiting a reversible memory machanism. On the right panel the microfluidics platform is represented. On the right panel a representative example of regulation with a switching controller is shown (reproduced from salzano2022ratiometric).
• Figure 5: a Agent-based simulation explicitly models intracellular dynamics, together with cell growth, division, mechanical interactions, and extracellular signaling. b Comparison of representative agent-based simulators for microbial populations. Both panels are reproduced from matyjaszkiewicz2017bsim.