Accelerating biopharmaceutical cell line selection with label-free multimodal nonlinear optical microscopy and...

Area of Science:

• Biotechnology
• Optical Microscopy
• Machine Learning

Background:

• Biopharmaceutical cell line selection is critical but inefficient.
• Current methods are time-consuming and labor-intensive.

Purpose of the Study:

• To develop a label-free, non-perturbative method for profiling biopharmaceutical cell lines.
• To expedite the selection of high-performing cell lines using intrinsic molecular contrast.

Main Methods:

• Utilized simultaneous label-free autofluorescence multiharmonic (SLAM) microscopy and fluorescence lifetime imaging microscopy (FLIM).
• Characterized Chinese hamster ovary (CHO) cell lines at early passages (0-2).
• Applied a machine learning (ML)-assisted analysis pipeline for single-cell classification.

Main Results:

• Achieved balanced accuracies exceeding 96.8% for monoclonal cell line classification at passage 2.
• Identified correlation features and FLIM modality as crucial for early classification.
• Demonstrated the effectiveness of multimodal nonlinear optical microscopy and ML.

Conclusions:

• Integrated optical bioimaging and ML offers a rapid, non-perturbative solution for cell line selection.
• This approach can accelerate the identification of superior biopharmaceutical cell lines.
• The techniques show potential for broader single-cell characterization in various biological fields.