Peptide biosensors for anticancer drugs: Design in silico to work in denaturizing environment

Filomena Guida, Anna Battisti, Ivan Gladich, Mauro Buzzo, Elena Marangon, Luciana Giodini, Giuseppe Toffoli, Alessandro Laio, Federico BertiBiosensors 2017 · DOI

TL;DR

Researchers have developed a computational method to design short peptides that can bind to anticancer drugs, even in harsh, denaturing environments like those found in blood samples. These custom-designed peptides show promise as highly selective biosensors for therapeutic drug monitoring, potentially leading to more personalized cancer treatments.

What Problem Does This Paper Address?

Monitoring the concentration of anticancer drugs in a patient's blood is crucial for optimizing treatment, minimizing side effects, and improving patient outcomes. This process, known as Therapeutic Drug Monitoring (TDM), is challenging because drugs can bind to proteins in the blood, and the conditions required to release them for analysis (like adding denaturants) can interfere with detection. Current methods like HPLC and LC-MS are expensive and require specialized labs. The goal is to develop a cheap, robust, and sensitive point-of-care device that can accurately measure drug levels in real-time, even in complex biological samples.

Methods

The core of this work is a computational approach to design short peptides that can act as biosensors. The researchers used a recently developed algorithm that searches the vast space of possible peptide sequences. This algorithm is unique because it explicitly accounts for the solvent environment, including denaturants, which is critical for designing peptides that work in biological samples.

Here's a breakdown of their approach:

  • In Silico Design: They employed a stochastic search algorithm that iteratively modifies peptide sequences through single point mutations. Molecular dynamics simulations are used at each step to assess the stability of the peptide-drug complex in the specified solvent. A Monte Carlo scheme then decides whether to accept or reject the mutation, aiming to improve binding affinity.
  • Explicit Solvent Modeling: Crucially, the simulations model the solvent explicitly, allowing them to capture the non-trivial effects of denaturants on the interactions between the peptide and the drug. This is a significant advantage over methods that ignore or simplify solvent effects.
  • Peptide Immobilization: For experimental validation, a short linker sequence (LysGlyGlyGly) was added to the designed peptides to facilitate their immobilization onto sensor surfaces (like SPR chips). Molecular dynamics simulations confirmed that this linker does not significantly alter the binding affinity.
  • Experimental Validation:
    • Surface Plasmon Resonance (SPR): Used to measure binding affinity in aqueous solutions. Peptides were immobilized on a sensor chip, and the drug (CPT-11) was flowed over it at varying concentrations.
  • Fluorescence Spectroscopy: Employed to measure binding affinity in both water and water-methanol mixtures. This technique was particularly useful for testing peptides in organic solvents relevant to TDM. The study used a "reversed" experiment, monitoring the quenching of the drug's fluorescence by the peptide.

Datasets

Not discussed in the paper. The study focuses on designing and experimentally validating peptides against a specific drug, CPT-11, and its metabolite SN-38.

Key Insights and Ideas

  • Designing for Denaturing Environments: The most significant insight is the successful design of peptides that bind their target drug specifically within a denaturing environment. This is a major step forward for biosensor development, as it addresses a key limitation of many existing methods.
  • Solvent-Specific Design: The computational approach's ability to account for specific solvent environments (like methanol) allows for the creation of peptides tailored to particular conditions, leading to higher affinity and selectivity.
  • Peptides as Viable Biosensing Elements: The study demonstrates that short, artificially designed peptides can be as effective, and in some cases more selective, than larger biomolecules like antibodies or aptamers for biosensing applications.
  • Computational Prediction Accuracy: The theoretical binding affinities predicted by molecular dynamics simulations showed good agreement with experimental results, validating the computational approach for selecting promising peptide candidates before costly experimental screening.
  • Selectivity for Drug vs. Metabolite: The designed peptides, particularly peptides 4 and 5, exhibited high selectivity for the parent drug (CPT-11) over its active metabolite (SN-38), which is crucial for accurate TDM.

Key Results

  • Binding Affinity: Peptides designed in vacuum showed moderate binding affinities (KD in the tens of micromolar range) in water. Peptides specifically designed in methanol demonstrated significantly higher affinities (KD in the low micromolar range).
  • Selectivity: Peptides 4 and 5 were highly selective for CPT-11 over SN-38, with peptide 5 showing micromolar affinity for CPT-11 and no binding to SN-38. Peptide 1, designed in vacuum, showed less selectivity.
  • Computational vs. Experimental Agreement: Theoretical binding energies from molecular dynamics simulations correlated well with experimental binding affinities, providing confidence in the predictive power of the computational method.
  • Environmental Specificity: The study clearly showed that peptides designed in methanol were more effective in methanol than those designed in vacuum, highlighting the importance of environment-specific design.

Limitations and Future Work

  • Limited Drug Targets: The study focused on a single anticancer drug, CPT-11, and its metabolite. Future work could expand this to other drugs and targets.
  • Complexity of Biological Matrices: While denaturants were considered, the full complexity of human plasma (with its diverse proteins and other molecules) was not fully replicated in all experiments. Further validation in more complex matrices would be beneficial.
  • Point-of-Care Device Development: The paper presents the design and validation of the sensing elements. The next step would be integrating these peptides into a functional point-of-care biosensor device.
  • Broader Environmental Conditions: While denaturants and organic solvents were addressed, the paper suggests that this approach could be extended to design peptides for extreme conditions of temperature, pressure, and pH, which would be valuable for various industrial and environmental sensing applications.

Summary updated

BibTeX
@inproceedings{guida2017peptide,
  author = {Filomena Guida and Anna Battisti and Ivan Gladich and Mauro Buzzo and Elena Marangon and Luciana Giodini and Giuseppe Toffoli and Alessandro Laio and Federico Berti},
  title = {{Peptide biosensors for anticancer drugs: Design in silico to work in denaturizing environment}},
  booktitle = {Biosensors},
  year = {2017},
  doi = {10.1016/j.bios.2017.09.012},
}
Figure