Flow Cytometry in Multi-Omics Analysis

Innovations in all types of cytometry—traditional flow, spectral flow, and mass cytometry—are driving its use for sorting, collecting, and analyzing cells for multi-omics (genomic, transcriptomic, and proteomic) research. Biocompare recently hosted a Bench Tips webinar where scientists actively involved in using cytometry discussed how they are exploiting technical advances for diverse research applications. They shared their experiences and best practices related to sample preparation, experimental set-up, and data analysis to help achieve accurate and reproducible results. Here are some key learnings from the webinar.

Flow cytometry is a technique used to sort and analyze the amount and type of cells in a given cell population. Conventional flow cytometry uses fluorescently labeled cells captured in a narrow spectrum of light, whereas full-spectrum flow cytometry, as the name suggests, captures fluorescence emitted across the entire spectrum and provides more information about the cell population.

Which cytometry technique to use?

With constant improvements and new cell sorting options, the question is always which instrument to use? The answer often boils down to the biological question and budgetary constraints. However, researchers are always curious to know how their choice will impact the design and running of the experiment, and the quality and completeness of the data obtained. The difference between conventional and spectral cytometry lies in how the emission is captured from the fluorophore. Rather than just looking at the peak emission, spectral cytometry captures the emission from all the lasers, for all the fluorophores across the entire spectrum. Hence, the information obtained is more comprehensive and complete. This also allows for fluorophores with similar emission profiles to be separated, which is not possible with a conventional cytometer. Spectral flow cytometry also reduces background emissions and improves resolution by eliminating autofluorescence from the unstained sample.

Mass cytometry is the next-generation flow cytometry. It is an adaptation of time-of-flight mass spectrometry (TOF-MS) where heavy metal tagged antibodies are used to label proteins on single cells. The heavier the metal tagged particle, the longer the time of flight and the elution time. Mass and conventional flow cytometry have the same workflow in terms of how the experiment is run. However, there is no spectral overlap in mass cytometry since it uses metal tags instead of fluorophores, and the data acquisition time is much slower.

Laura Ferrer Font, Ph.D., Senior Staff Scientist and High Dimensional Spectral Cytometry Specialist at the Malaghan Institute of Medical Research, conducted a study to see if the same population of cells can be identified using a full-spectrum cytometer versus mass cytometry. The findings using UMAP algorithms showed similar data maps with both techniques, which meant that the same populations of cells were being identified. The results validated that using a different algorithm gave the same results, which gave the team confidence that both spectral flow and mass cytometry yield similar results.

The four pillars of cytometry

Sample preparation, panel design, panel optimization, and data analysis are critical to any cytometry experiment, irrespective of the technique that is used.

Sample preparation

Sample preparation starts with having high-quality single-cell suspensions as a prerequisite for any cell sorting or single-cell sequencing experiment. Optimizing the digestion protocols and reagents is very important. “Digestion of the cells is important as it affects the cell number, cell viability, and cell integrity,” explained Ferrer Font. “If you have a lot of debris or if many cells are dead then the results will be unreliable.”

Panel design and optimization

Panel design and assigning the right fluorophores to the right markers is the next important step in cytometry. “Designing high-dimensional full-spectrum flow cytometry panel can be quite overwhelming,” noted Ferrer Font. “Designing the panel on paper is easy but sometimes that same design does not work in the cytometer.” Panel design involves knowing the cell biology, knowing the instrument and spectral signatures, and assigning the right fluorophore to each marker. During her talk, she shared some tips and tricks on how to design and optimize a full-spectrum flow cytometry panel and provided an overview of all the steps needed to go from having a set of markers on paper, to obtaining data that is ready-for-use in the analysis algorithms. (Panel design and optimization are also covered in a 2020 paper authored by Ferrer Font.)

“After designing your panel following the rules for optimal design, it’s important to optimize your panel using real-life samples,” added Ferrer-Font. (The spectral cytometry panel optimization guidance shared in the webinar can also be found in this 2021 paper.) Panel optimization involves titrating the antibodies, and evaluating reference controls, marker resolution, and data quality.

Antibody panel design and choice of metal tag are also important in determining the success of a mass cytometry experiment. Marie Lue Antony, Ph.D., Research Associate, Laboratory of Dr. Zohar Sachs, Department of Medicine, University of Minnesota, uses mass cytometry to measure the simultaneous expression of several markers, to identify signaling pathways in stem cells of acute myeloid leukemia.

“When it comes to expanding panels, conventional cytometry is limited by the issue of spectral overlap and choosing fluorescent beads,” Antony explained. “With mass cytometry, there is no spectral overlap or loss of signal intensity, and we work with a panel of markers that can analyze 40 different parameters on a single cell.” Understanding the nature of the metal tags and the matrix is also very important, added Antony. The metal ion tags sometimes have some low level of impurities, which on ionization can cause spill over and oxidation state-related errors, commonly faced when working with a large panel of markers. During the webinar, Antony outlined experimental design considerations to help overcome challenges of spill-over and oxidation states and offered tips to expand an existing panel of antibodies to accommodate new markers without the hassle of adding multiple antibody panels.

Mass-tag barcoding is the next step for the simultaneous profiling of multiple samples, according to Antony. “It helps in adding replicates to your experiment, in the simultaneous profiling of multiple samples, minimizes amount of antibody used, limits technical variability, helps when there are limited number of cells, and reduces instrument measurement time.” The main challenge with mass cytometry is the destruction of the sample, which prevents its use for other downstream analysis. “It also has relatively slow throughput 200—1000 cells per second, compared to 50,000 cells per second with conventional flow cytometry.”

Data analysis

“Data analysis really starts at the very beginning when you are planning your experiment,” explained Todd Bartkowiak, Ph.D., Research Fellow in the Laboratory of Dr. Jonathan Irish and Dr. Rebecca Ihrie, in the Department of Cell and Developmental Biology at Vanderbilt University. In the webinar, he discussed challenges in planning and conducting experiments in mass cytometry, how to plan data collection, and what to do after the data has been acquired. “It’s important to think about what the experiment entails—whether it’s a longitudinal analysis or is it to compare treatments and outcomes. You have to think about the sample size, and if you have enough to make a statistical comparison. Do you have proper controls, and do you have enough sample left for validation?”

Manual gating protocols can be very labor intensive and prone to error when looking at high-dimensional datasets. High-dimensional algorithms can look at multiple parameters and markers at multiple timepoints, saving time and improving accuracy. There are different types of multi-dimensional analysis pipelines for dimensionality reduction, clustering, and high-level machine learning.

Dimensionality reduction using t-SNE (t-distributed stochastic neighbor embedding) or UMAP (uniform manifold approximation and projection) is used for data visualization. Multiple clustering algorithms, like SPADE, phonograph, or FlowSOM, can be used to group data based on certain similarities. Machine learning is a way to take the dimensionality reduction and clustering to distill complex data into knowledge. “Citrus can be used to correlate cell abundance to outcomes,” noted Bartkowiak. “Marker enrichment modeling identifies different cell populations in the sample. It’s great for phenotypes where you don’t know what to expect. RAPID (Risk Assessment Population Identification) correlates population abundance with outcomes.”

“Always perform a pilot experiment to go through your data analysis pipeline,” Bartkowiak added. “It is very important to validate the findings by having a high-dimensional data analysis pipeline and statistical approach in place. You should do manual gating, repeat analysis, orthogonal pipelines, randomization or drop data and, finally, set aside a validation cohort to repeat some of the experiments.”

What’s next for cytometry?

Daniel Schraivogel, Ph.D., Research Staff Scientist in the Genome Biology Unit at the European Molecular Biology Laboratory (EMBL), recently published a paper in Science that describes a novel, high-speed, image-enabled flow cytometric cell sorting, combined with multicolor fluorescence microscopy, to make high-speed sorting decisions based on spatial image-derived data.

“Combining microscopy and cell sorting is a difficult task as it involves blur-free imaging of fast flowing cells, and real-time image reconstruction and analysis,” said Schraivogel. His work describes combining conventional flow cytometry with BD CellView image technology to get cellular images in a microsecond time scale. “It’s easy to use, it works with all cell types that can be analyzed with a traditional sorter, and it is compatible with all types of fluorescent labeling.”

Conventional flow cytometry has not been able to sort cells during mitosis. During the webinar, Schraivogel talked about how image-enabled cell sorting can now be applied in functional genomics to perform genome-wide imaging-based screens in hours (as opposed to weeks and months) and how it can be used to identify and separate cells in different mitotic stages. There are some limitations with using this technique in terms of lower resolution and classification, which depends on its orientation relative to the imaging plane. Analysis, integration, and storing imaging data can also be challenging. But it opens up immense possibilities, which is likely to take cytometry further in its evolution and use.