Application of the PAMONO-Sensor for Quantification of Microvesicles and Determination of Nano-Particle Size Distribution

Victoria Shpacovitch, Irina Sidorenko, Jan Eric Lenssen, Vladimir Temchura, Frank Weichert, Heinrich Müller, Klaus Überla, Alexander Zybin, Alexander Schramm, Roland Hergenröder2017 · DOI

TL;DR

This paper introduces an enhanced PAMONO-sensor system that can now quantify microvesicles (MVs) and determine nanoparticle size distributions in real-time, without the need for labeling. This advancement significantly broadens the applicability of SPR-based sensors for analyzing complex biological samples, paving the way for more sophisticated diagnostic tools.

What Problem Does This Paper Address?

Accurately characterizing extracellular vesicles (EVs), particularly microvesicles (MVs), is crucial for understanding various pathologies. Existing methods like flow cytometry and nanoparticle tracking analysis (NTA) often require labeling, suffer from resolution limitations, or are time-consuming. While Surface Plasmon Resonance (SPR) sensors show promise, previous applications for EV detection were either complex to implement or lacked the ability to analyze particle size distributions within a sample. This paper aims to overcome these limitations by adapting and enhancing the PAMONO-sensor to enable label-free, real-time quantification and size distribution analysis of MVs.

Methods

The core of this work lies in the adaptation and enhancement of the PAMONO-sensor, which utilizes Surface Plasmon Resonance (SPR) imaging. Key methodological advancements include:

  • Protein A/G Functionalization: The sensor surface is functionalized with cysteine-conjugated protein A/G, which allows for oriented immobilization of anti-target antibodies. This is an improvement over previous methods that required biotin-conjugated antibodies and offered less control over antibody orientation.
  • Machine Learning for Size Distribution: A novel approach using Convolutional Neural Networks (CNNs) is employed to analyze the SPR images. These CNNs are trained to estimate particle size based on the SPR signal intensity, enabling real-time determination of nanoparticle size distributions in polydisperse samples. This moves beyond simple particle counting to a more detailed characterization.
  • Re-usable Sensor Chips: The protein A/G functionalization strategy also allows for the repeated use of the sensor chip after antibody elution, reducing costs and experimental time.
  • SPR Imaging: The PAMONO-sensor itself captures images of individual binding events of nanoparticles on the sensor surface by detecting local changes in reflectivity.

Datasets

  • Polystyrene Nanoparticles: Used for training and validating the machine learning models for particle size distribution analysis. Particles of 100 nm, 200 nm, and 300 nm were utilized.
  • Microvesicles (MVs) from SH-SY5Y Cells: These biological samples were used to demonstrate the sensor's ability to detect and quantify MVs.
  • Human Immunodeficiency Virus (HIV) Virus-Like Particles (VLPs): Used for testing the specificity of the sensor's detection capabilities.

Key Insights and Ideas

  • Label-Free Quantification of MVs: The PAMONO-sensor, with its protein A/G functionalization, successfully detects and quantifies MVs without the need for fluorescent or other labels, simplifying sample preparation and analysis.
  • Real-time Size Distribution Analysis: The integration of CNNs with SPR imaging is a significant breakthrough, enabling the estimation of particle size distributions in real-time. This provides a much richer characterization of samples than simple counting.
  • Sensor Reusability: The protein A/G functionalization strategy significantly enhances the practicality of SPR sensors by allowing for multiple cycles of antibody binding and elution, making the sensor chips reusable.
  • Comparative Analysis: The sensor allows for swift qualitative comparison of relative microvesicle concentrations between samples, which is valuable for screening and initial diagnostics.
  • Specificity is Achievable: The sensor demonstrates good specificity in distinguishing between particles with and without specific surface antigens (e.g., ovalbumin on VLPs).

Key Results

  • MV Detection: The PAMONO-sensor successfully detected microvesicles derived from SH-SY5Y cells, with measured sizes aligning with established literature values (around 117 ± 6 nm).
  • Relative Concentration Comparison: PAMONO-sensor measurements of counting rates showed a qualitative agreement with NTA measurements for comparing relative MV concentrations across different cell lines (SY5Y cont., SY5Y TrkB, SY5Y TrkA).
  • Specificity: For HIV-VLPs, the sensor showed that particles with ovalbumin on their surface generated approximately ten times more signals than particles without ovalbumin, indicating good antigen-based specificity.
  • Sensor Reusability: The gold sensor chip could be reused for antibody binding and VLP detection for at least three cycles, with a slight decrease in efficiency observed in the second cycle.
  • Particle Size Distribution: The CNN-based approach demonstrated the ability to estimate particle size distributions for polystyrene nanoparticles, showing reasonable agreement with reference measurements, although with more noise and a slight estimation error for larger particles.

Limitations and Future Work

  • Noise in Distributions: The particle size distributions obtained from the PAMONO-sensor showed more noise compared to reference methods, attributed to analyzing only a sub-sample of particles. Increasing the number of analyzed images can mitigate this.
  • Size Estimation Error: A 20-30 nm estimation error was observed for particles larger than 100 nm, stemming from the image intensity loss during CNN training data synthesis. This is identified as a calibration issue that can be adjusted.
  • Refractive Index Dependence: The CNN model trained on polystyrene particles might not be directly applicable to biological microvesicles due to differences in refractive indices. Further calibration with biological samples is necessary.
  • Future Work: The authors suggest adapting the PAMONO-sensor software for size distribution measurements of biological nano-vesicles, potentially using artificially prepared liposomes for initial training. They also highlight the potential for developing a diagnostic platform for liquid biopsy samples that can quantify MVs and analyze their size distribution patterns, which is crucial for understanding disease progression.

Summary updated

BibTeX
@misc{shpacovitch2017application,
  author = {Victoria Shpacovitch and Irina Sidorenko and Jan Eric Lenssen and Vladimir Temchura and Frank Weichert and Heinrich Müller and Klaus Überla and Alexander Zybin and Alexander Schramm and Roland Hergenröder},
  title = {{Application of the PAMONO-Sensor for Quantification of Microvesicles and Determination of Nano-Particle Size Distribution}},
  year = {2017},
  doi = {10.3390/s17020244},
}
Figure