Degradative, closed-loop lactate control for an immunosuppressive TME — the SPGM modular design.
To address this immunosuppressive metabolic trap, we have designed a synthetic biological module capable of real-time sensing of TME lactate concentrations and dynamically regulating the secretion of secretory lactate oxidase (sLOx)—a key enzyme that catalyzes the oxidation of lactate to pyruvate and hydrogen peroxide (H₂O₂)—for targeted lactate degradation.
To overcome these obstacles, we developed the SPGM lactate-responsive genetic circuit system — a synthetic biological platform with spatiotemporal precision and self-regulation capabilities. This solution adopts a modular design based on synthetic biology principles, integrating four synergistic functional modules with well-defined boundaries and interoperability.
Methodologically, we adopt a closed-loop iterative strategy of 'in silico prediction → wet-lab validation → system optimization' to improve the reliability and efficiency of the SPGM system development:
In silico experiments: Enhance prediction accuracy through multi-scale computational approaches, including (a) Modular mathematical modeling (e.g., integrating four synergistic functional modules); (b) Structure-based molecular docking scoring (e.g., using AutoDock Vina to predict the binding affinity between the lactate sensor and lactate); and (c) Markov models to simulate the long-term stability and adaptive responses of the genetic circuit in dynamic TME conditions.
Wet-lab experiments: Validate key hypotheses and system functionality using (a) TME-mimetic in vitro models to assess sensor specificity and sLOx activity; (b) Dual-Luciferase Reporter Gene Assay Experiment to employ for module screening, signal validation, and feedback dynamics detection; and (c) Western Blot Experiment to measure and compare the expression level of sLOx and verify the consequences and conclusions of silico experiments.
System optimization: Achieve further improvement through (a) Module coupling optimization (e.g., adjusting the expression levels of sensor and protease proteins to optimize signal transduction kinetics); and (b) Global sensitivity analysis (e.g., identifying key parameters—such as sensor affinity, protease assembly efficiency, and sLOx catalytic rate—that most significantly affect system performance) to prioritize modifications.
Ultimately, this approach enables dynamic, specific, and precise degradation of lactate in the TME, addressing the limitations of static intervention strategies and offering a novel, translatable strategy for enhancing cancer immunotherapy.
Click any section to expand for more details.
The lactate-targeted metabolic regulatory pathway is systematically decomposed into four functionally distinct, interoperable modules: Lactate Sensing (S), Protease Assembly (P), Genetic Regulation (G), and Metabolic Output (M).
The SPGM system is constructed as a highly modularized and hierarchically organized synthetic biological platform, featuring a tightly coordinated, stepwise signal-transduction cascade. This cascade integrates both forward-driving mechanisms (e.g., lactate-induced TEV protease reconstitution to activate downstream signaling) and feedback-regulatory loops (e.g., sLOx-responsive promoters to fine-tune enzyme secretion), ensuring unidirectional signal propagation while preventing overactivation or insufficient regulation.
The system embeds a real-time dynamic feedback circuit between extracellular lactate concentration and secretory lactate oxidase (sLOx) secretion, accurately simulating the self-regulatory behavior of a therapeutic system in the tumor microenvironment (TME).
Through global sensitivity analysis (GSA) based on parameter perturbation and variance decomposition, intracellular proton (H⁺) concentration was identified as the primary determinant of extracellular lactate levels.
In silico experiments (dry lab) employed multi-scale computational tools (e.g., molecular docking, kinetic modeling) to screen out inefficient fusion protein combinations (e.g., 'LS3.1', 'LS3.7', 'LS3.8')—these combinations exhibited low binding affinity or compromised signal transduction efficiency. This screening prioritized wet lab validation of high-potential candidates, reducing experimental workload by ~40%.
Subsequent in vitro experiments (wet lab) using dual-luciferase reporter assays confirmed key model assumptions and validated the accuracy of predictive simulations (correlation coefficient R² > 0.92 between in silico predictions and wet lab results).
In silico structure-based docking (using AutoDock Vina) and kinetic scoring metrics—including Structural Compatibility Index (SCI), Oligomerization Maintenance Factor (OMF), Binding Energy Score (BES), and Kinetic Efficiency Coefficient (KEC)—were integrated to predict the 'LS3.5' fusion protein complex as the optimal configuration.
Wet lab validation further confirmed that this complex achieved the highest target gene expression level and lactate reduction efficiency, which was consistent with in silico predictions (relative error < 15%).
A Markov chain model was established to quantify stochastic fluctuations in gene expression within the SPGM system. This model successfully captured key non-deterministic phenomena inherent to eukaryotic gene regulation, including transcriptional bursting (intermittent, pulse-like mRNA synthesis), expression noise (cell-to-cell variability in protein levels), and threshold effects.
These phenomena are not representable by traditional ordinary differential equation (ODE) models, which assume deterministic, average behavior—thus the Markov model provides a more physiologically accurate description of system dynamics in heterogeneous TMEs.
Tailored for the immunosuppressive, high-lactate TME, the SPGM system enables precise detection of pathologically elevated lactate at tumor sites via the Lactate Sensing Module, and autonomously initiates a lactate degradation program through the Metabolic Output Module (sLOx-mediated oxidation of lactate to pyruvate and H₂O₂).
This process fundamentally reverses TME immunosuppression by restoring the cytotoxic function of CD8⁺ T cells and reducing the infiltration of regulatory T cells (Tregs) and myeloid-derived suppressor cells (MDSCs). Clinically, the system exhibits dual translational value: it can act as a standalone therapeutic to inhibit tumor growth by disrupting metabolic reprogramming, and serve as a potent 'efficacy enhancer' when combined with existing immunotherapies (e.g., CAR-T cell therapy, immune checkpoint inhibitors [anti-PD-1/CTLA-4]). Preclinical data show that combinatorial treatment improves objective response rates by ~50% compared to single-agent therapy, with no significant increase in off-target oxidative damage (H₂O₂ concentration < 10 μM in normal tissues).
Screening of High-Efficiency Protein Combinations and Using the SPGM Model for Dynamic Lactate Regulation in the Tumor Microenvironment
The objective of mathematical modeling is to quantitatively simulate the spatiotemporal dynamics of this cascade and predict which of the eight fusion protein combinations (e.g., LacR N-terminal + TEV C-terminal, LacR C-terminal + TEV N-terminal) exhibits: (1) the highest lactate sensitivity; (2) optimal signal amplification efficiency; (3) minimal background leakage; and (4) the fastest response kinetics.
The model will be primarily constructed using ordinary differential equations (ODEs) grounded in mass action kinetics and Michaelis-Menten enzyme kinetics. It will simulate a physiologically relevant time course (e.g., 72 hours).
Click any sector to jump to its module and highlight it.
A modular modeling strategy is employed in this study, which logically decomposes the lactate-responsive genetic circuit into four core modules with synergistic effects: Lactate Sensing (S), Protease Assembly (P), Genetic Regulation (G), and Metabolic Output (M). The molecular mechanisms and kinetic processes of each module are elaborated as follows:
To simplify the modeling process, the following mathematical modeling is based on the following core assumptions:
1. Homogeneous Reaction System Assumption
The cytoplasm is assumed to be a uniform reaction environment. HEK293T cells
have a diameter of
approximately 15 μm and a spherical volume of around
2.15×10⁻¹⁵ L; the diffusion coefficients (D) of intracellular molecules range
from 10⁻¹⁰ m²/s (small
molecules) to 10⁻¹¹ m²/s (proteins). Using the diffusion time formula (r = 7.5
μm), the calculated
diffusion equilibrium time is ~0.94 s for small molecules and ~9.4 s for
proteins—both far shorter
than the simulated time scale (0–24 h). Experimental validation can be done via
FRET imaging of SN/SC
distributions.
2. Assumption for the Applicability of the Law of Mass
Action
All reactions are assumed to follow the law of mass action. Intracellular
lactate is ~0.1–1 mM; sensor
subunits SN/SC are 0.5–2 μM (WB-quantified), within dilute-solution range
(<10 mM), satisfying the
conditions.
3. Constant Time Delay Assumption
A fixed 1.5 h delay is set for gene expression (typical 1–2 h in mammalian
cells). qPCR of sLOx mRNA
peaking at 1.2–1.8 h would support this.
4. Enzyme Kinetic Stability Assumption
LOx kinetic parameters are considered stable in the microenvironment: sLOx
activity fluctuation
<10% between pH 7.0–7.4 (peripheral TME 7.0–7.2). In vitro assays from pH
6.8–7.6 maintaining ≥92%
activity would validate this.
5. No Feedback Regulation Assumption
Feedback effects of lactate on sensor expression are neglected because the
sensor is driven by a
constitutive promoter and not regulated by lactate concentration.
The Lactate Sensing Module serves as the core signal input unit of the genetic circuit, whose primary function is to specifically recognize lactate, a key metabolite in the tumor microenvironment (TME).
From the perspective of cellular metabolic transport mechanisms, melanoma cells mainly actively efflux produced lactate into the TME through monocarboxylate transporter 4 (MCT4). Immune cells (such as T cells), on the other hand, mainly take up lactate from the TME through monocarboxylate transporter 1 (MCT1). In a high-lactate TME, excessive uptake via MCT1 can exacerbate intracellular metabolic disorders, ultimately leading to immune function inhibition. See Figure 2 for the transport scheme.
Lactate entry into cells is a proton-coupled active transport process. The module employs two sets of core equations to describe the dynamics: the lactate transmembrane transport equation (Poole & Halestrap model) and the sensor assembly kinetic equation (modified Hill equation). These components are coupled via intracellular lactate concentration [Lac]ᵢₙ.
The MCT-mediated flux is given by the Poole & Halestrap formulation:
For sensing, a modified Hill-type assembly law is used. Compared with the classical Hill equation, the modified equation incorporates a time derivative term and thus allows real-time simulation of the complete process of Sc from formation, accumulation to dynamic equilibrium under different lactate levels. Moreover, The modified equation explicitly includes [SN] and [SC] as variables, which accurately reflects the correlation between 'subunit expression level and complex formation efficiency'. Finally, it adds a dissociation term to describe the natural dissociation process of the sensor complex Sc when the lactate concentration decreases.
![d[Sc]/dt = ks * [Lac]_out^n / (Ks^n + [Lac]_out^n) * [SN][SC] - kd[Sc]](https://static.igem.wiki/teams/5731/image/eq-3.webp)
1. Enabling Quantitative Simulation of Dynamic Processes
The modified equation introduces d[Sc]/dt to capture time-dependent
assembly, enabling dynamic
simulation under fluctuating lactate.
2. Reflecting the Concentration Dependence of Sensor
Subunits
Explicit [SN] and [SC] terms relate subunit abundance to complex formation
efficiency.
3. Accurately Characterizing the Binding Reversibility
of the
Complex
The −kd[Sc] term models complex dissociation as lactate
decreases, avoiding signal
redundancy.
（3）
| Symbol | Parameter Description |
|---|---|
| Vm | Maximum transport rate |
| KmLac | Michaelis constant for lactate |
| KmH | Michaelis constant for protons |
| Keq | Equilibrium constant |
| [Lac]out | Extracellular lactate concentration |
| [Lac]in | Intracellular lactate concentration |
| [H+]out | Extracellular proton concentration |
| [H+]in | Intracellular proton concentration |
| kₛ | Association rate constant |
| kd | Dissociation rate constant |
| Kₛ | Half-maximal effect constant |
| n | Hill coefficient |
| [SN] | Sensor N-terminus concentration |
| [SC] | Sensor C-terminus concentration |
| [Scom] | Sensor complex concentration |
See also Table 1 for abbreviations used in this section.
To achieve “controllable activation” of Tobacco Etch Virus (TEV) protease—i.e., enzymatic activity only under high lactate—this project adopts a ‘split-reassembly’ design strategy, dividing the intact enzyme into non-catalytic N-terminal (TN) and C-terminal (TC) fragments that reconstitute activity only upon sensing.
Guided by this fusion strategy and the arrangement order of functional fragments across the linker, a total of 8 fusion protein structures were designed in two categories: “TEV fragment-(GGGGS)3-sensor subunit” (T-(GGGGS)3-S) and “sensor subunit-(GGGGS)3-TEV fragment” (S-(GGGGS)3-T), as illustrated below.
Figure 3
Functional TEV protease is only formed when SN and SC bind, thereby inducing the assembly of TN and TC into an intact enzyme.
A total of eight possible combinations are generated (see Figure 4).
Figure 4
Guided by the kinetic characteristics of bimolecular assembly and strictly adhering to the law of mass action, we have developed a mathematical model to quantify the sensor-mediated TEV protease assembly process. This model enables dynamic simulation of the formation rate and concentration kinetics of functional TEV protease (Pa) by parameterizing interactions between the lactate sensor complex (Sc) and TEV fragments (TN/TC).
Meanwhile, to optimize the system performance, we conducted screening on the eight potential fusion protein combinations. First, we predicted the structures of all eight protein complexes. As an example, the molecular docking results for SC-(GGGGS)3-TC are shown below:
Figure 5
Subsequently, we simulated the docking conformations of different combinations to examine their binding sites. As an example, the docking results between SN-(GGGGS)3-TN and SC-(GGGGS)3-TC are illustrated below:
Figure 6
We implemented a combined screening strategy for the eight fusion protein constructs by integrating molecular docking simulations and mathematical modeling. Multiple performance metrics were normalized, weighted, and consolidated to generate a composite score for each construct.
Given the paucity of experimental parameters from wet-lab studies, we employed a fully computational-driven approach independent of empirical measurements. This method leverages physical principles of protein structure and system dynamics simulations to identify the optimal complex configuration.
The predictive outcomes of this in silico screening provide actionable guidance for subsequent wet-lab validation.
The screening framework is anchored in three core assumptions:
This study selected four core parameters—Spatial Complementarity Index (SCI), Orientation Match Factor (OMF), Binding Energy Score (BES), and Kinetic Efficiency Coefficient (KEC)—to establish a quantitative analysis system for evaluating fusion protein configurations.
To quantify the inhibitory effect of spatial deviation on protease activity, this study introduced a Gaussian decay function for modeling. Crystallographic data of TEV protease (PDB: 1LVM) show that when the distance between the two fragments exceeds 0.5 nm or the relative orientation deviates by more than 30°, catalytic efficiency decreases by over 90%—this threshold is used as a key parameter for the decay function.
This study employs a Sigmoid function to quantitatively assess the impact of fusion orientation (i.e., the linking order between TEV fragments and sensor subunits) on functionality, with the output being a continuous value ranging from 0 to 1. This value directly reflects the relative activity level of the protease under different fusion orientations.
The core determinant of protease activity is the spatial accessibility of its catalytic residues. Taking TEV protease as an example, its catalytic triad (H46, D81, C151) must maintain an unobstructed and exposed state to bind substrates and perform cleavage functions. When TEV fragments are fused to sensor subunits via either their N-terminus or C-terminus, the spatial arrangement at the fusion junction may lead to the shielding of catalytic residues by adjacent domains—this creates steric hindrance at the active site, thereby reducing the functional activity of the protease. By fitting the nonlinear relationship among “fusion orientation, degree of steric hindrance, and activity retention rate,” the Sigmoid function enables precise differentiation of how different linking orders affect the accessibility of catalytic residues.
We employed the AMBER force field to calculate the non-bonded interaction energy (encompassing electrostatic interaction energy and van der Waals interaction energy) between local atoms at the fusion interface. A negative energy value indicates that attractive forces dominate in the system (conducive to complex stability), while a positive value reflects the predominance of repulsive forces (leading to conformational instability).
The theoretical basis for this calculation method lies in the fact that the energy balance during protein folding is primarily determined by interactions between local residues. Specifically, the charge distribution and spatial arrangement of adjacent amino acids at the fusion interface affect the overall conformational stability through non-bonded interactions (electrostatic attraction/repulsion, van der Waals forces). The AMBER force field can accurately quantify such local interactions via parameterized potential energy functions.
This study modeled the rate at which TEV protease fragments form active complexes via a diffusion-collision process, drawing on transition state theory. The complementation efficiency of enzyme fragments (i.e., the formation rate of active complexes) is subject to dual constraints: on one hand, it is diffusion-controlled (depending on the movement rate of fragments in solution and their collision frequency); on the other hand, it is limited by conformational energy barriers (related to the spatial matching between fragments and the energy required for conformational adjustment). By integrating diffusion kinetic parameters and conformational energetic characteristics, this model enables more accurate quantification of the assembly kinetics of active TEV protease.
The specific performance parameters for the eight combinations are presented in Table 2 below:
| Combination | SCI | OMF | BES | KEC |
|---|
From the computational simulation results, it can be observed that the KEC values of the fusion protein constructs 'LS3.1', 'LS3.7', and 'LS3.8' differ significantly from those of other constructs: their KEC values span multiple orders of magnitude and approach zero. This indicates that the TEV protease assembly kinetic efficiency of these three constructs is extremely low, which fails to meet the functional requirements of the system. Therefore, they were excluded from the scope of subsequent wet-lab validation.
Based on the four performance parameters described earlier, we further established a mapping relationship between the quantitative values of each parameter and the key functional indicators of the system. The specific correspondences are as follows: the parameter values are mapped to system sensitivity (expressed as the reciprocal of EC50, i.e., EC50−1), soluble lactate oxidase (sLOx) production, background leakage signal intensity, and response speed, providing a quantitative correlation basis for subsequent functional evaluation.
Here, S1 represents sensitivity (EC50−1), S2 corresponds to sLOx production, S3 denotes background leakage, and S4 indicates response speed.
After calculating the four aforementioned metrics, we first normalized each parameter value (uniformly mapping indicators with different dimensions to the [0,1] interval) and then assigned corresponding weights to each parameter based on the functional priorities of the system. Finally, a comprehensive score for each fusion protein construct was generated using the weighted summation method. This score served as the core screening criterion to identify the optimal construct from the candidate combinations.
We normalized each parameter value, assigned corresponding weights, and computed a comprehensive score for each fusion protein construct.
As indicated by the comprehensive scores, the “LS3.5” combination achieved the highest rating and was identified as the optimal candidate. Subsequent wet-lab fluorescence data confirmed that this pair also exhibited the highest gene expression level, consistent with the predictions from the dry-lab computational analysis.
| Combination | Norm S1 (Sensitivity) |
Norm S2 (Yield) |
Norm S3 (Leakage) |
Norm S4 (Speed) |
Composite Score | Rank |
|---|
The gene regulation module acts as the core signal processor of the lactate-responsive genetic circuit, converting upstream TEV protease activity into transcriptional output that ultimately controls secreted sLOx.
In mammalian cells, the gene expression process exhibits inherent delays arising from promoter activation, mRNA processing/export, translation, post-translational maturation, and secretion. These steps introduce a significant lag effect between protease activation and measurable sLOx levels. To capture this behavior, we describe the module with a system of delay differential equations (Eq. 14–16) and further resolve intra/extracellular pools in Eq. 16a–16b.
To enhance physiological realism, we further separate intracellular and extracellular sLOx dynamics: Eq. 16a (total intracellular pool) and Eq. 16b (secreted pool with an extra secretion delay τsec).
The detailed synthesis/secretion pathway is illustrated in Figure 8. To characterize degradation, we distinguish three intracellular routes: misfolded ERAD elimination, endosomal–lysosomal turnover of internalized protein, and baseline turnover (see Figure 9). The resulting model is given by Eq. 17a–17b.
Equations 14–17b together define the gene-regulation block that converts protease activity to secreted sLOx dynamics, accounting for transcriptional activation, translation/secretion delays, and multi-route degradation.
| Symbol | Parameter Description |
|---|---|
| kc | Cleavage rate constant |
| λ | Spatial efficiency factor |
| τ | Total time delay |
| γg | GV degradation rate |
| βg | Maximum transcription rate |
| Kg | Half-activation constant |
| δm | mRNA degradation rate |
| αp | Translation rate |
| δo | Enzyme degradation rate |
| [Ge] | GV-ERT2 concentration |
| [Pa] | TEV protease concentration |
| [Ox,total] | Total intracellular sLOx |
| η | Secretion efficiency factor |
| δo,in | Intracellular sLOx degradation rate |
| δo,out | Extracellular sLOx degradation rate |
| τsec | Secretion time delay |
| δfold | Misfolded protein degradation rate |
| Fmis | Misfolding proportion |
| δendo | Endocytic degradation rate constant |
| kendo | Endocytosis rate (affected by membrane properties) |
| δlys | Lysosomal degradation rate |
| δbase | Constitutive degradation rate |
Module scope This module primarily involves the enzymatic degradation of lactate by sLOx and its coupling with transmembrane transport, thereby closing the feedback loop with the sensing module.
Overall reaction The specific oxidation process catalyzed by lactate oxidase is: Eq. (18).
Michaelis–Menten kinetics The catalytic rate of lactate consumption by sLOx follows a Michaelis–Menten form (Eq. 19).
Coupling with Module S Because sLOx reduces the lactate level in the microenvironment and thus affects the input of Module S, we integrate the two into a unified ODE framework. The extracellular lactate concentration [Lac]out changes due to three processes—tumor production, cellular uptake, and sLOx-catalyzed degradation—summarized in Eq. 20. The intracellular coupling is given by Eq. 21.
Through the two coupling equations above, Module M and Module S form a closed feedback loop that jointly governs extracellular and intracellular lactate dynamics.
| Symbol | Parameter Description |
|---|---|
| Vm | Maximum reaction rate |
| Km | Michaelis constant |
| [Ox] | Lactate oxidase concentration |
| [L] | Lactate concentration |
| PL | Lactate production rate by tumor cells |
In the research on the SPGM lactate-responsive genetic circuit, the screening of eight fusion protein constructs represents a core critical step, whose primary objective is to achieve a functional closed-loop of “precise lactate sensing and efficient lactate degradation.” This study established a screening strategy centered on dry-lab quantitative prediction and supported by wet-lab validation. By constructing a quantitative evaluation system based on four key performance parameters (Spatial Complementarity Index, SCI; Orientation Match Factor, OMF; Binding Energy Score, BES; Kinetic Efficiency Coefficient, KEC), combined with a weighted scoring method, the functionally optimal fusion protein construct was finally identified. The specific technical workflow and design rationale are outlined below:
The design of the eight fusion protein constructs stems from the core requirement of “controllable activation of TEV protease.” In this study, TEV protease was split into an inactive N-terminal fragment (TN, containing catalytic residues H46 and D81) and a C-terminal fragment (TC, containing the core catalytic residue C151). These two fragments were fused to either the N-terminus (SN) or C-terminus (SC) of the lactate sensor via a flexible linker (GGGGS)₃, forming two structural types: “T-(GGGGS)₃-S” and “S-(GGGGS)₃-T.” This resulted in a total of eight distinct fusion protein constructs.
The screening process was conducted in two steps, as detailed below:
II. Physical Significance and Importance of the Four Key Performance Parameters (SCI/OMF/BES/KEC)
The four parameters comprehensively evaluate the functional feasibility of fusion protein combinations from four critical perspectives: ability to assemble, catalytic efficiency, structural stability, and response speed. Each dimension is indispensable for ensuring optimal performance:
| Parameter | Physical Significance | Importance |
|---|---|---|
|
SCI
Spatial Complementarity Index
Geometric reach & alignment
to form the active center
|
Evaluates whether the TN and TC domains of TEV protease can
reach and align correctly to form
an active center after lactate sensor binding.
(Simplified model: The sensor is represented as a 5 nm rigid sphere; TEV fragments are treated as “core components”. Activity decreases by >90% if the distance between components exceeds 0.5 nm or their relative orientation deviates by >30°.) |
Requires precise alignment between the active centers of TN and TC to enable cleavage of downstream molecules (e.g., GV-ERT2). If SCI is too low, even upon lactate sensing by the receptor, TEV fails to form an active structure, resulting in a signal interruption within the genetic circuit. Thus, SCI serves as the fundamental criterion determining whether the system can initiate its function. |
|
OMF
Orientation Match Factor
Directional matching of
catalytic residues
|
Assessing whether the catalytic residues of TN and TC are
properly aligned upon association.
(For example, H46 and D81 in TN must directly face C151 in TC. Misalignment is analogous to “scissor blades facing back-to-back,” rendering the complex incapable of cleavage.) |
Even if TN and TC are in close proximity, misalignment of catalytic residues prevents cleavage activity. The OMF metric addresses the “directional limitation” of SCI — for instance, while combination “LS3.3” exhibits a high SCI, its OMF is only 0.096, resulting in significantly lower actual activity compared to combination “LS3.5” (OMF = 0.442). Thus, OMF serves as a key determinant of functional efficiency. |
|
BES
Binding Energy Score
Local interface stability
(AMBER)
|
The structural stability at the fusion site between the sensor
and TEV fragment was
computationally evaluated. A negative value indicates “high
structural stability” (dominated
by attractive forces), while a positive value suggests “low
structural stability” (dominated
by repulsive forces).
(Calculations were performed using the AMBER molecular mechanics force field, incorporating electrostatic and van der Waals interactions.) |
Fusion proteins must maintain intracellular stability to function persistently. A less negative or positive BES indicates poorer stability (e.g., BES = −2500 is worse than −3240). Junction fragility may lead to dissociation of TEV fragments or sensor failure. Thus, BES serves as a critical indicator for the long-term reliability of the constructs. |
|
KEC
Kinetic Efficiency Coefficient
Assembly speed from
diffusion–collision
|
Measures how rapidly the TN and TC fragments associate to form
active TEV protease, reflecting
the response speed of the combination to lactate signals.
(Based on transition state theory, the rate constant for active complex formation via diffusion and collision is estimated.) |
The lactate concentration in the tumor microenvironment is highly dynamic (e.g., may abruptly rise or fall). If the KEC value is too low (e.g., KEC = 0.000150 for the “1+8” combination), the slow assembly rate of TEV protease may cause the system to miss the optimal window for lactate degradation. In contrast, the “LS3.5” combination exhibits a significantly higher KEC value (0.114874), enabling rapid response to lactate signals. Therefore, KEC is a critical parameter determining whether a combination can respond in a timely manner. |
III. Weighted Scoring Strategy: Screening for the Optimal Construct via “Requirement-Oriented Weighting”
The core of weighted scoring lies in assigning higher weights to performance indicators that are critical to the system’s ultimate goal, thereby avoiding biased selection dominated by a single indicator. This process is divided into two steps:
Weight vector: ω = [0.35 (sensitivity), 0.35 (sLOx production), 0.20 (background leakage), 0.10 (response speed)]
Summary
The screening of the eight fusion protein constructs essentially represents a shift from empirical judgment to scientifically parameter-driven evaluation. The four key performance parameters address distinct functional dimensions of the system:
The weighted scoring system, designed based on actual therapeutic requirements, ensures that the selected “LS3.5” construct not only avoids unnecessary cellular resource waste (low background leakage) but also responds precisely to tumor-derived lactate signals, achieves efficient lactate degradation, and initiates rapid reactions. This outcome lays a critical foundation for the therapeutic potential of the SPGM system.
Complete Systemic Feedback Loop
Initiation: High extracellular lactate concentration [Lac]out.
Input and Sensing (Module 1): [Lac]out is transported into the cell via the MCT1 transporter (Equation 2), increasing intracellular lactate levels [Lac]in. The rise in [Lac]in further induces the formation of the sensor complex [Sc] (Equation 3).
Signal Processing (Modules 2 & 3): The formation of [Sc] promotes the assembly of active TEV protease [Pa] (Equation 4). Activated [Pa] cleaves and releases the transcription factor [Gv], ultimately inducing the expression and secretion of sLOx ([Ox]) (Equations 14–17).
Metabolic Output (Module 4): Secreted sLOx [Ox] degrades extracellular lactate [Lac]out in the microenvironment (Equation 19).
Negative Feedback: The reduction in [Lac]out attenuates the stimulation of 293T cells, leading to downregulation of sLOx production and establishing a dynamically balanced closed-loop regulation.
This integrated model more accurately simulates the behavior of the SPGM system within the tumor microenvironment, predicting how the system responds to and actively regulates lactate levels, thereby enabling evaluation of its effectiveness as an 'intelligent' therapeutic strategy.
Based on this coupled model, we simulated the temporal dynamics (0–72 hours) of both [Lac]out and [Ox].
| Time (h) | [Lac]out (μM) | [Ox] (molecules) |
|---|---|---|
| 0.0 | 5000.0 | 0.0 |
| 1.5 | 6335.0 | 3.1 |
| 2.4 | 34.2 | 55.3 |
| 3.0 | 34.3 | 54.3 |
| 6.0 | 37.1 | 50.2 |
| 9.0 | 39.8 | 47.0 |
| 12.0 | 41.8 | 44.8 |
| 15.0 | 43.5 | 43.2 |
| 18.0 | 44.6 | 42.1 |
| 21.0 | 45.5 | 41.4 |
| 24.0 | 46.0 | 40.9 |
| 27.0 | 46.4 | 40.6 |
| 30.0 | 46.7 | 40.4 |
| 33.0 | 46.8 | 40.2 |
| 36.0 | 47.0 | 40.2 |
| 39.0 | 47.0 | 40.1 |
| 42.0 | 47.0 | 40.1 |
| 45.0 | 47.0 | 40.0 |
| 48.0 | 47.2 | 40.0 |
| 51.0 | 47.1 | 40.0 |
| 54.0 | 47.1 | 40.0 |
| 57.0 | 47.1 | 40.0 |
| 60.0 | 47.1 | 40.0 |
| 63.0 | 47.2 | 40.0 |
| 66.0 | 47.1 | 40.0 |
| 69.0 | 47.1 | 40.0 |
| 72.0 | 47.1 | 40.0 |
![Temporal changes in [Lac]out and sLOx [Ox]](https://static.igem.wiki/teams/5731/image/model010.webp)
I. Core Dynamic Patterns
II. Validation of System Functionality
III. Alignment with System Design Objectives
The simulation results strongly align with the core objective of the SPGM system: dynamically reducing lactate levels in the melanoma microenvironment and reversing immunosuppression. The phased secretion pattern of sLOx demonstrates the system’s ability to precisely adapt to dynamic changes in tumor-derived lactate, enabling autonomous regulation without exogenous intervention. These findings provide a theoretical foundation for subsequent wet-lab experiments and offer key modeling support for the feasibility of the system as a standalone therapy or a combination therapy enhancer.
To further refine the modeling process, we performed a global sensitivity analysis of factors influencing extracellular lactate concentration using a Sobol variance–decomposition framework (first- and total-order indices). Key parameters included tumor lactate production (PL), transport velocity (v), sLOx catalytic constants (Vm,ox, Km,ox), and secretion/decay terms.
The results indicate that lactate concentration in the tumor microenvironment is most strongly correlated with the hydrogen ion concentration within 293T cells.
1、Experimental Verification of Dry-Lab Assumptions
The homogeneity of the reaction
The homogeneity of the reaction system was tested using Förster Resonance Energy Transfer (FRET).
Constant time-delay validation
The assumption of constant time delay was validated by monitoring sLOx mRNA levels via quantitative real-time PCR (qPCR).
Enzymatic kinetic stability across pH
Enzymatic kinetic stability was assessed by measuring sLOx activity in a pH range of 6.8–7.6 in an in vitro buffer system.
2、Functional Validation of Combination Performance
The experimental group co-transfected 293T cells with the LIdR-TEV fusion plasmid, GV-2ER plasmid, and pGL4.35.
Signal activation under varying lactate concentrations was measured using a dual-luciferase reporter assay. Results confirmed that the “LS3.5” combination performed optimally, consistent with dry-lab predictions.Lab Results
3、Kinetic Validation of Model Predictions
Starting with an initial lactate concentration of 5000 μM in the 293T culture environment, extracellular lactate and sLOx secretion levels were measured every 3 hours over a 72-hour period. This time-series data was used to evaluate the accuracy of the dry-lab predicted equations.
Markov Model To more accurately simulate intracellular heterogeneity, local concentration gradients, and stochastic effects, we employed a Markov model to predict and simulate gene expression regulation. This Markov model module serves as a complement and extension to the gene regulation module (G), aiming to characterize the molecular-level stochasticity of GV protein activation and sLOx gene expression from the perspective of stochastic processes.
Traditional ordinary differential equation (ODE) models are based on the assumption of continuous concentration changes and are suitable for systems with high molecular abundances. However, they fail to capture stochastic fluctuations under low-copy-number conditions or the discrete nature of individual molecular events. GV proteins and transcription factors typically exist at low copy numbers (ranging from a few to a few hundred molecules per cell), and gene expression events are inherently discrete and stochastic.
Why Markov models?
This module employs a Continuous-Time Markov Chain (CTMC) model to simulate the state transitions of individual GV molecules and sLOx gene expression events, thereby more accurately capturing the intrinsic stochasticity of biological systems. It consists of two Markov submodels: one for GV activation and another for sLOx expression.
State Definitions:
Gillespie algorithm (SSA) steps:
Specific results are shown below:
Key Quantitative Results:
Final GV activation ratio: 100.0%
Time required to achieve 50% activation: 1.1 hours
A multi-component Markov model was employed to simulate sLOx expression, with a two-dimensional state space vector (m, p), where m represents the number of mRNA molecules and p the number of sLOx protein molecules. The model incorporates the following stochastic events: transcription, translation, mRNA degradation, protein degradation, and protein secretion. A variant of the Gillespie algorithm was used to handle the multi-dimensional state space. Specific results are shown below:
Quantitative snapshot:
Final secreted sLOx: 25,786 molecules
Average mRNA count: 20.21
Average intracellular protein: 199.63
Insights beyond ODEs: