(725b) Analysis of High-Throughput Multiparametric Flow Cytometry Data to Identify Cellular Phenotypes Underlying Alcohol Mediated Aberrant Differentiation of Embryonic Stem Cells

Flow cytometry (FCM) is an established high-throughput research and clinical tool. Yet, analysis of data is often limited to extracting a small subset of information, which may not meet requirements such as re-engineering biological processes. The current work presents a pipeline to uncover biological information from single-cell multiplex data using available analysis tools, serving as a backdrop in stem cell research.

The motivation behind the present study is to determine the mechanisms of Fetal Alcohol Spectrum Disorder (FASD) using FCM. FASD is caused by alcohol consumption during pregnancy and is manifested as defects of the central nervous system in infants, and in craniofacial malformations, resulting in neurological disabilities and behavioral deficits. We have studied the effect of ethanol on mouse embryonic stem (ES) cells a to model embryogenesis in FASD.

ES cell differentiation gives rise to three primary cell lineages – i) endoderm, ii) mesoderm and iii) ectoderm. The experimental protocol was designed so as to drive ES cell differentiation towards neuroectoderm (NE) lineage by treating ES cells with retinoic acid (RA) for a period of 4 days. ES cells were exposed to ethanol (EtOH, 25-100mM in concentration) during RA treatment. The cells were stained with a cocktail of monoclonal antibodies to core transcription factors (TF) Sox2, Oct4 and Nanog, and analyzed by FCM over time; ES (Day 0), Day 2  and Day 4 of differentiation.

The three ES cell core transcription factors are linked with exit of ES cells from the pluripotency, self-renewal stage and early stages of ES differentiation cell fate decisions. The ratio of Oct4 to Sox2 is shown (Thomson et al., 2011) to determining the ES cell fate – overexpression of Oct4 led to triggering of mesoendodermal cells and overexpression of Sox2 led to neuroectodermal lineage. The expression of the three core TFs was collected along with cell morphology characteristics and cell viability. Raw FCM data was gated based on cell size/complexity using forward and side scattering characteristics, and viable cells (live) identification was based on viability staining). FCM data from 50,000 cells was transformed onto a Logicle scale (Parks et al., 2006) for best display. The transformed data were corrected for cell size/complexity by polynomial regression according to Knijnenburg et al. (2011). The corrected data was then visualized using a novel three-dimensional density plot based on bivariate kernel density. This helped to qualitatively uncover phenotypic information in the three TFs combined and revealed the presence of cell subpopulations. Data from each of the experimental conditions (EtOH dose and time of exposure) were clustered using a combination of spectral clustering (to estimate number of clusters) and Self-organization Maps or Kohonen Maps to identify these subpopulations. Further, minimum spanning tree technique was employed to match subpopulation progress across time points, along with estimation of correlative binary interactions between TFs using mutual information.

The consolidated data analysis approach strengthened the basis to understand the mechanism underlying FASD from our previous study (Vadigepalli et al., 2010). Current analysis indicates that i) the TFs get down-regulated during differentiation, ii) ethanol delays asymmetrically the down-regulation of TFs and thus the cell lineages, iv) ethanol affects the balance between Oct4 and Sox2 expressions, which favors mesoendoderm at the expense of neuroectoderm, iii) Oct4 expression remains up-regulated relative to Sox2 with ethanol exposure, and v) the distribution of cell subpopulations expressing the core TFs are altered by ethanol. The findings point to an early interference of ethanol in ES differentiation networks that alter cell commitment against normal development of brain and this may underlie a range of FASD defects.

1. Thomson M. et al. Pluripotency Factors in Embryonic Stem Cells Regulate Differentiation into Germ Layers, Cell 145, 2011.
2. Parks DR, Roederer M, Moore WA. A New ‘Logicle’ Display Method Avoids Deceptive Effects of Logarithmic Scaling for Low Signals and Compensated Data, Cytometry Part A, 2006.
3. Knijnenburg TA, Roda O, et al. A regression model approach to enable cell morphology correction in high-throughput flow cytometry, Molecular Systems Biology 7, 2011.
4. Vadigepalli, R., Ogony, J. and Anni, H. (2010) Alcohol toxicity on embryonic stem cell regulatory networks mediating neural fate decision. Society for Biological Engineering 2^nd International Conference on Stem Cell Engineering, Engineering Cell Fate 89:36-37
5. Vadigepalli, R., Ogony, J. and Anni, H. (2010) Ethanol effects on transcription factor network regulating stem cell differentiation. IEEE International Conference on Bioinformatics and Bioengineering, 298-299