Our Model: From a Mechanical Switch to a New Biosensor Strategy
Disclaimer: For better understanding of our methods and graphs take a look at our software page!
Highlights
- A PFOA-Induced Mechanical Switch: We discovered that PFOA binding induces a large-scale, 90-degree rotation of our GFP reporter domain, revealing an interesting mechanical action of our biosensor.
- Challenging Our Initial Hypothesis: The model showed that the GFP domain remains structurally rigid during its rotation, a negative result indicating that a change in intrinsic fluorescence is an unlikely signaling mechanism.
- Guiding a New FRET-Based Design: The predicted rotation provided the rationale for a more sophisticated approach: a FRET-based biosensor, where the large change in distance between domains would generate a clear signal.
- Uncovering a Binding Mechanism: Our simulations revealed that PFOA acts as an uncompetitive inhibitor, a prediction supported by our experimental MST binding data showing a strong PFOA-TYMS interaction (Kd = 217 µM).
1. The Central Question Driving Our Model
Our project began with a hypothesis for a PFOA biosensor: that PFOA binding to a TYMS-GFP fusion protein could cause a conformational change in the GFP domain and thus alter its fluorescence. We used molecular dynamics (MD) simulations to build a detailed, atom-level understanding of our system. We needed to see if a conformational change would happen, figure out its mechanics, and determine if that change could logically lead to a measurable signal.
2. A Multi-Simulation Approach: Building the Full Picture
We designed a series of simulations to investigate PFOA's effects at different levels, from our engineered sensor down to the native enzyme and its active site.
Experiment 1: The TYMS-GFP Sensor—A Mechanical Switch in Action
Our first simulation of the full TYMS-GFP fusion protein yielded a major breakthrough. We observed a massive 90-degree rotation of the GFP domain relative to the TYMS domain. A detailed analysis using a Dynamical Cross-Correlation Matrix (DCCM) pinpointed the mechanism to a molecular hinge in the linker, which was activated by numerous, transient PFOA interactions.
Figure 1: The 90-degree rotation of the GFP domain induced by PFOA binding, showing the molecular hinge mechanism in the linker region.
Experiment 2: A Critical Insight Challenges Our Hypothesis
This large-scale motion was exciting, but the next finding was even more important. We analyzed the internal structure of the GFP domain during this rotation by calculating its Root Mean Square Deviation (RMSD).
- Result: The RMSD of the GFP's core structure remained exceptionally stable (fluctuating between 1.0-1.75 Å).
- Interpretation and Impact: This finding, while challenging our initial hypothesis, was incredibly valuable. The stability of the GFP fold means its internal chromophore environment is not significantly disturbed. Therefore, a change in its intrinsic fluorescence (brightness) is highly unlikely.
Figure 2: RMSD analysis showing the structural stability of the GFP domain during rotation.
Experiment 3: Proposing a FRET-Based Sensor
Our model didn't just invalidate our initial idea; it pointed us directly to a more sophisticated solution. The simulation clearly showed a stable GFP domain undergoing a large-scale rotation that dramatically changed the distance and orientation between TYMS and GFP.
This led to our new, model-driven hypothesis: This large-scale rotation is a perfect candidate for a FRET-based biosensor.
Förster Resonance Energy Transfer (FRET) is a mechanism that relies on the close proximity of two different fluorophores (a donor and an acceptor). The large rotation our model predicts would dramatically change the distance between a donor fluorophore placed on one domain and an acceptor on the other, providing a clear, on/off signal. The model thus provided the foundational rationale for a next-generation design of our biosensor. All we would need to do is add a second fluorophore to the c-terminus of GFP, and we would have a FRET sensor, as we have two fluorophores, one donor and one acceptor.
To quantitatively test this new hypothesis, we redesigned our sensor in silico as a His-tag-GFP-Linker-TYMS-CFP construct and used our simulation data to predict its performance.
- 1. High-Precision Distance Analysis: We performed a distance analysis between the proposed donor site (the C-terminus of TYMS) and the acceptor site (the chromophore of GFP). We ran this analysis on two separate simulations: our full PFOA-bound trajectory (20mM concentration) ("ON" state) and a control simulation without PFOA ("OFF" state).
- 2. A Clear PFOA-Induced Separation: The results show a distinct and persistent increase in the donor-acceptor distance in the presence of PFOA. While both states are dynamic, the PFOA-bound trajectory consistently samples more extended conformations. On average, PFOA binding increases the distance between the FRET pair by 3.1 Å, with a maximum observed separation of over 16 Å.
Figure 3: Distance between the proposed FRET donor and acceptor sites over time. The control simulation (red) shows a compact "OFF" state. In the presence of PFOA (blue), the sensor shifts to a more extended "ON" state. The difference (green) is consistently positive, confirming a PFOA-induced separation.
- 3. A Quantitative Prediction of the FRET Signal: Using the mean values from our simulations and the known Förster Radius for the GFP-CFP FRET pair, (R₀ ≈ 48 Å), we calculated the theoretical FRET efficiency for both states. Most FRET pairs have a Förster Radius of 4.8 nm or 48 Å.
| State | Average Distance (R) | Predicted FRET Efficiency |
|---|---|---|
| On (PFOA) | 42.750 ± 5.877 Å | 0.6513 ± 0.0011 |
| Off (Control) | 39.667 ± 3.645 Å | 0.7492 ± 0.0007 |
Results:
Absolute ΔE = -0.0979 (~9.8 ppt)
Relative ΔE = -13.07%
Calculation Details:
- The FRET Efficiency for each frame was calculated using the FRET efficiency formula:
- E = R₀⁶ / (R₀⁶ + r⁶):
- E: FRET efficiency (the fraction of donor energy transferred to the acceptor).
- R₀: The Förster radius, a characteristic distance where energy transfer efficiency is 50%.
- r: The actual distance between the donor and acceptor molecules.
- This formula shows that efficiency decreases as the distance (r) increases, falling off with the sixth power.
- Then the mean was taken, as the relationship of the data (6th power) is non linear.
Interpretation and Impact:
Our simulation predicts that PFOA binding will cause a clear and measurable drop in FRET efficiency from 74.92% to 65.13%. This ~13% relative change in FRET efficiency represents a strong signal that can be detected with standard laboratory equipment.
3. A Second Discovery: PFOA's Impact on the Active Site
Beyond the linker dynamics, our simulations also suggested PFOA could directly impact the TYMS active site.
An Uncompetitive Inhibition Mechanism
We simulated PFOA with a TYMS-GFP monomer both with and without its natural cofactor, mTHF. When PFOA was alone with the enzyme, it quickly drifted out of the active site. However, when the cofactor mTHF was present, PFOA remained stably bound to the TYMS-mTHF complex. This is the classic signature of uncompetitive inhibition, where the inhibitor only binds to the enzyme-substrate complex.
Figure 3: Schematic showing the uncompetitive inhibition mechanism where PFOA binds to the TYMS-mTHF complex.
Experimental Validation: Confirming PFOA Binding with MST
While our molecular dynamics simulations suggested that PFOA binding is stabilized when TYMS is in its cofactor-bound conformation, they did not exclude the possibility of weaker interactions with the apo enzyme. To test whether PFOA can interact with TYMS in vitro, we next measured the binding affinity using MicroScale Thermophoresis (MST). Our MST experiments revealed that PFOA does indeed bind to TYMS even in the absence of mTHF, with a dissociation constant (Kd) of 217 µM. This indicates that while cofactor binding may enhance the stability of the interaction in the simulated environment, PFOA is still capable of associating directly with the enzyme itself.
MST Binding Results
MicroScale Thermophoresis experiments confirm PFOA binding to TYMS with high affinity.
PFOA-TYMS Binding
Strong Binding
Control: mTHF Only
No Binding
Positive Control: dUMP
Strong Binding
4. Conclusion
We began with a hypothesis, and our simulations revealed its flaws, guiding us toward a more robust, FRET-based biosensor design. The model uncovered a detailed mechanical rotation, pinpointed its cause to a molecular hinge, and identified a proposed uncompetitive inhibition mechanism that was then supported with experimental binding data. Furthermore, we calculated predicted FRET Efficiencies for our TYMS-GFP protein and observed a 13% relative change in efficiency which provides a strong future starting point in our final goal of developing a PFOA biosensor.
5. Our Methods & Commitment to Open Science
We believe in documenting the entire scientific process. For any teams interested in our approach, we've provided our full protocol and made all of our data and scripts publicly available.
- Detailed Protocol: Our complete, step-by-step MD protocol is available on our Software Page. It details our use of AMBER, AlphaFold3, and NeuralPlexer, along with the ff19SB, GAFF2, and OPC force fields.
- Code and Data Availability: All input scripts, trajectory data, and analysis notebooks used to generate the results on this page are available at our public GitLab repository.