Article Text

Download PDFPDF

125 Characterization and label-free isolation of phenotypically defined human immune cells using ghost cytometry, a novel AI powered flow cytometry technology
  1. Keisuke Wagatsuma1,
  2. Hiroko Nomaru1,
  3. Kazuki Teranishi1,
  4. Satoru Akai2,
  5. Yuri An1,
  6. Kaoru Komoriya2,
  7. Yoko Kawamura1,
  8. Asako Tsubouchi2,
  9. Andy Wu1 and
  10. Sadao Ota1
  1. 1ThinkCyte Inc, Tokyo, Japan
  2. 2ThinkCyte Inc., San Carlos, CA, USA

Abstract

Background Background: Identification and isolation of specific human immune cell populations with technologies that are minimally disruptive to native biological states are a critical requirement in modern cellular immunotherapy approaches. For both early-stage cell therapy research and the development of autologous and allogenic cellular therapeutics, technologies that can isolate specific immune cell subsets while preserving their native biological and functional properties are needed to improve the quality and efficacy of cell therapy drug products.

Methods Here we applied Ghost Cytometry, a novel method of analyzing and sorting cells using high dimensional image information, to isolate human immune cell subsets with minimal external perturbation. Ghost Cytometry leverages a combination of proprietary optics and artificial intelligence to capture digital-optical profiles (Ghost Motion Imaging (GMI) signals) from individual immune cells. The approach generates a novel, multi-dimensional data set for individual cells, which can be leveraged to enable label-free sorting of defined phenotypic cell populations for downstream R&D

Results We characterized human T-cells and generated ‘ground truth’ functional profiles for 1) glycolysis level (by depletion of a glucose analog), 2) exhaustion state (by PD-1 and LAG3 expression), 3) activation or resting profiles (by CD25 expression) and 4) viability (by propidium iodide exclusion and Annexin V expression) based on their unique GMI signals. From this, we developed a set of machine-learning derived classifiers for identification of unlabeled T-cell subsets. The classifiers showed area under the curve (AUC) performance ranges between 0.923 and 0.995, demonstrating high classification accuracy.

Conclusions Here we report on a novel method to characterize and isolate human immune cell subsets with enhanced therapeutic potential. This approach enables isolation of target T-cell subsets with defined phenotypic profiles in an unperturbed state, without the requirement of using external fluorescent markers. The results described here have practical applications for investigators and drug developers in cell therapy research and development.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.