In industrial fermentation scenarios, the stressful environments faced by yeast exhibit complexity and dynamism, with their core characteristics summarized in three aspects:
Firstly, intermittent fluctuations—parameters such as temperature, substrate concentration, and dissolved oxygen in industrial bioreactors often undergo non-sustained changes along with production processes . For instance, there are periodic temperature peaks caused by microbial metabolic heat release in the mid-to-late fermentation stage, and osmotic pressure fluctuations before substrate exhaustion. This "stress occurrence-relief-reoccurrence" intermittency traps the traditional strategy of "continuously activating stress-resistance programs" in a dilemma: it wastes energy when there is no stress and responds sluggishly when stress suddenly emerges.
Secondly, synergistic superposition of multiple stresses—single stress rarely exists in isolation. High temperature is often accompanied by oxidative stress; in high-osmolarity environments, cells intensify their metabolic burden to maintain turgor pressure, which in turn induces elevated ROS levels; some industrial substrates even trigger both osmotic stress and carbon source metabolic pressure simultaneously. This "1+1>2" synergistic damage poses a severe challenge to stress-resistance systems that only target a single type of stress.
Thirdly, significant individual variability—even within the same bioreactor, different yeast cells exhibit a differentiated state of "some cells being tolerant while others being on the brink of apoptosis" under the same stress, due to microenvironmental heterogeneity and stochasticity in gene expression. The traditional "uniform regulation" model either leads to the death of sensitive cells due to insufficient regulation or causes metabolic waste in tolerant cells due to excessive regulation.
These characteristics determine that the yeast stress-resistance system must break through the traditional design framework of "static, single-dimensional, and uniform"—we need to endow yeast with intelligent capabilities similar to "real-time monitoring-dynamic judgment-precise response," namely, building a "cellular pulse" sensing and regulation system. Through programmable multi-dimensional processing logic, cells can flexibly adjust their stress-resistance strategies according to the actual state of stress (type, intensity, and duration).
To this end, we have developed two core technical modules in a targeted manner: On one hand, we established a multi-component yeast stress-resistance gene library
. By screening and validating sensor promoters and stress-resistance genes that cover three major stress types—high temperature, osmotic pressure regulation, and ROS (reactive oxygen species)—we laid the material foundation for multi-dimensional regulation. On the other hand, we designed a three-node repressor-based genetic oscillator, which utilizes three orthogonal repressor systems (TetR, LacI, and λcI) to construct a closed-loop negative feedback. This circuit converts the intermittent signals of industrial stresses into rhythmic outputs with controllable "pulse frequency and amplitude": When mild intermittent stress occurs, the oscillator outputs low-frequency pulses, activating only a small number of stress-resistance genes to conserve energy; when multiple stresses overlap, the pulse amplitude strengthens and frequency increases, synchronously activating corresponding groups of stress-resistance genes in the library to form a synergistic response. Meanwhile, this circuit features programmability—by adjusting the degradation rate of repressor proteins and the sensitivity of promoters, it can adapt to the individual differences of different yeast cells. Ultimately, this achieves precise matching of "stress signal → cellular pulse → stress-resistance response," enabling yeast to avoid both the metabolic loss caused by "excessive defense" and the survival crisis due to "insufficient defense" in complex industrial environments.
We combine anti-stress regulatory genes for three types of environmental stresses with a classic three-node oscillator. In other words, each of the three anti-stress genes is coupled to a gene encoding one of the oscillator’s three proteins, so that whenever an oscillator protein is expressed, the anti-stress protein coupled to it is co-expressed. Specifically, lacI^ts is coupled to the hyperosmotic-response module, λ cI to the heat-response module, and tetR to the ROS-response module. As illustrated below:
In the absence of stress, mutual repression among the three oscillator proteins prevents any of the coupled anti-stress proteins from being overexpressed. The concentrations of anti-stress proteins remain at low levels and vary periodically in a staggered manner, thereby keeping the metabolic burden on yeast within a modest, acceptable range.
When a certain stress occurs, the inhibitory effect of corresponding repressor is relieved. The design is shown below.
Notably, the mechanisms by which the three stresses relieve repression are not identical. Heat directly inactivates the temperature-sensitive LacI^ts protein, partially or completely. In contrast, ROS/hyperosmotic stress first activates their responsive promoter pSOD/pGPD1, which drives expression of the effector proteins RecA/tTA. RecA promotes autocatalytic cleavage of λ CI, whereas tTA competes with TetR for binding to the tetO operator, thereby releasing repression of genes downstream of tetO.
Next, we will use truth tables, gene circuit schematics, and selected deterministic model simulations to illustrate how protein expression patterns change as different numbers of stresses arrive. The figure below provides an overview. In the truth table, 10 denotes definite high expression of the protein, 1 denotes at least normal-level expression, and 0 denotes suppressed expression.
The simulation outputs shown are limited to conceptual illustration; construction and explanatory details of the model are provided in the ‘Dry Lab’ section. By the way, it should be clarified that the stress types depicted in the figures below are merely examples. The three stresses are equivalent, and 1 or 2 stresses can refer to any one or any two among the three.
First, as noted above, when no stress is present, the three protein pairs exhibit sustained oscillations, and all anti-stress proteins remain at low concentrations with periodic fluctuations, as shown below.
With exactly one stress present, taking heat as an example, repression of the λ cI and its coupled anti-heat gene by LacI^ts is relieved, leading to high-level production of the heat-resistance protein and enabling yeast to resist elevated temperatures. This appears in the Time-Protein Concentration plots as an upward shift of the equilibrium position (the “midline” of oscillation) for the “λ CI and heat-resistance” pair relative to the other two pairs, as shown in the figure below.
It is worth noting that the simulation results shown on the left were calculated under an extreme scenario: high temperature greatly suppresses the action of LacI^ts, then the massive production of λ CI leads to nearly no expression of TetR, and little TetR is insufficient to block LacI expression, resulting in the LacI level becoming high, comparable to that of λ CI. However, by adjusting the expression parameters of each protein in the model, one can achieve a configuration where the orange curve representing λ CI is far higher than the red curve representing LacI, and the latter is far higher than the blue curve representing TetR, as illustrated on the right. This is also why we use 10 and 1 to represent these two protein expression states respectively in the truth table. These adjustments correspond to promoters with different sensitivities to repressor proteins, stresses of varying severity, and even proteins whose stress sensitivities differ due to mutations. In our project design, our engineered yeast is programmable, and the choice and combination of all these parameters are intended to serve the user’s real production needs.
For clarity, subsequent simulation figures in this section are likewise generated under extreme scenarios.
Additionally, the figure on the left also shows that the oscillation amplitudes gradually decrease over time, reflecting an overall weakening of mutual repression within the oscillator, which diminishes the oscillator’s effective existence.
When heat stress leaves, the previously accumulated LacI^ts becomes functional again and reestablishes repression of λ cI, causing expression of the heat- resistance protein to decrease and, after a certain period of time, settle into periodic oscillations.
With exactly two stresses present, taking heat and hyperosmotic stress as an example, repression of both λ cIand lacI^ts is released. The two protein pairs (λ CI with the heat-resistance protein, and LacI^ts with the hyperosmotic- resistance protein) are expressed at comparatively higher levels, equipping yeast to face both stressors simultaneously. In the plots, the midlines of these two pairs shift upward.
Moreover, as shown below, when heat stress is stronger than hyperosmotic stress for the yeast, the midline for the heat-associated pair is higher than that for the hyperosmotic-associated pair, consistent with the principle that more severe stress, the stronger resistance demands.
Then, if one of the two stresses ends first, the system transitions to the “exactly one stress” case.
If both stresses end simultaneously, the accumulated λ CI/LacI^ts rapidly reasserts repression on lacI^ts/λ cI, reducing expression of the hyperosmotic/heat-resistance protein and, after a certain period of time, restoring periodic oscillations.
When all three stresses are present, all three gene pairs are derepressed, similar to the scenarios above. The figure below shows the case where the magnitudes of the three stresses are identical, so the three stress-resistance proteins are nearly equivalent in status. The oscillation midlines coincide and the amplitudes gradually decay over time. Again, this is due to weakening of mutual repression within the oscillator, reducing its effective existence and driving the protein concentrations toward constant amounts and the system towards a stable state.
If stresses end sequentially, the system transitions to the “exactly one/two stresses” cases. If all three end simultaneously, the accumulated oscillator proteins rapidly cross-repress each other. After a transient with fluctuations in expression, the system returns to periodic oscillations.
In summary, we have demonstrated the yeast response under different stress conditions. Generally speaking, when a stress is present, expression of its corresponding stress-resistance protein becomes higher than the others. When the stress leaves, the proteins that repress it, previously accumulated during the stress, quickly reduce its expression.
See the 'Dry Lab' section for details of the deterministic model.
Coupling stress-resistance proteins to a three-node oscillator offers multiple advantages:
First and foremost, as discussed above, in the absence of stress, expression of stress-resistance proteins remains low, limiting the metabolic burden on yeast and avoiding affecting target product production caused by overexpression of unnecessary proteins.
Secondly, upon stress onset, repression of the stress-response module is lifted, enabling rapid and strong expression of protective proteins. Moreover, the extent of de-repression scales with stress intensity, aligning stress-resistance protein output with the severity of the challenge. This supports survival of yeasts and their production of target products under stress.
Third, after stress removal, the elevated levels of specific oscillator proteins accumulated during stress rapidly repress the specific anti-stress modules, swiftly returning the system to low-concentration periodic oscillations. In other words, the post-stress delay before downregulation of stress-resistance proteins is short.
Last but not least, by choosing promoters of different strengths or protein mutants with different sensitivities, one can tune the sensitivity and activation thresholds to different stresses. This programmability allows users to select configurations tailored to their manufacturing requirements.
In a nutshell, these advantages not only enable yeast to counter three stresses simultaneously while maintaining production of target products, but also improve performance relative to a conventional “responsive promoter – stress-resistance protein” circuit (as illustrated below). Without stress, the presence of the oscillator keeps repressor protein levels lower. After stress removal, the oscillator proteins also rapidly shut down the stress-response modules. Together, our project provides a programmable and customizable strategy to reduce cellular burden and conserve energy in yeasts.
To address the issue that traditional single-bacterium systems suffer from excessive burden due to superimposed "sensing-response" functions, which may impair production efficiency, we have designed and constructed a "sensing bacterium-producing bacterium" dual-task division system. By decoupling the signal capture function from the stress-resistant execution function and relying on the precise collaboration between the two types of bacteria, we ultimately achieve an efficient balance between stress resistance performance and production efficiency.
The sensor strain serves as a "monitor and encoder of stress signals": Equipped with orthogonal quorum sensing systems of the AHL/LuxR family, it activates corresponding sensitive promoters when exposed to stresses (e.g., high temperature, ROS, high osmolarity), driving the expression of specific AHL synthases. Specifically, mild single stress triggers secretion of low-concentration single AHL, while multiple stresses induce high-concentration mixed AHLs—enabling precise encoding of "stress type → signal type" and "stress intensity → signal concentration". Focused solely on signal-related functions, it significantly reduces metabolic consumption.
The producer strain acts as an "executor and regulator of stress responses": It constitutively expresses matching LuxR receptors (via constitutive promoters) to decode AHL signals in real time. Combined with the three-node repressor oscillator, it initiates graded stress responses based on signal intensity—low-concentration signals trigger low-frequency pulses and expression of "lightweight" stress-resistance genes; high-concentration mixed signals activate high-frequency pulses and "heavyweight" stress-resistance gene combinations. Without allocating resources to sensing, more metabolic flux is directed to product synthesis, significantly improving efficiency.
The core advantages of this model lie in specialized functions and robust collaboration: Single strains are not required to undertake multiple functions, which greatly reduces their metabolic burden and results in a higher biomass retention rate of the microbial community during continuous fermentation; meanwhile, real-time synchronization of stress signals is achieved through quorum sensing. Ultimately, this significantly improves the system’s stable product production rate in fluctuating environments, meeting the needs of large-scale biomanufacturing.