A fully automated, walkaway flow cytometry workflow for immune cell subset identification
Giulia Grazia, (Beckman Coulter Life Sciences, Cassina De' Pecchi [MI], Italy), Rita Bowers, Jason Maraman, (Beckman Coulter Life Sciences, Indianapolis, IN), Vashti Lacaille, (Beckman Coulter Life Sciences, Miami, FL)
In the fast-paced and highly competitive landscape of biopharma research, the adoption of a walkaway flow cytometry workflow offers many advantages, such as greater reproducibility, increased sample throughput, reduced sources of human error, improved data consistency, and less operator hands-on dependency, all contributing to more robust immunophenotyping studies for research applications.
In this application note, we present a fully automated, walkaway flow cytometry workflow solution that enables researchers to perform all the steps of a flow cytometry experiment using integrated automation systems that require little to no operator intervention.
The workflow diagram (Figure 1) illustrates the fully automated system for immune cell subset identification by flow cytometry from start to finish. The automated workflow leverages the Biomek i7 Multichannel liquid handler for automated sample preparation, integrated with a CytoFLEX LX flow cytometer for automated sample acquisition (Figure 2). The recorded data files from the CytoFLEX LX flow cytometer are then automatically uploaded into the Cytobank platform via the API (application programming interface), with subsequent automated data analysis performed using a pre-defined gating model applied to the newly created experiment in a programmatic way, achieving auto-generated results ready for review by the researcher.
Figure 1. Fully Automated Flow Cytometry Workflow: Diagram showing the five steps of the automated workflow using the Biomek i7 integrated workstation with a CytoFLEX LX flow cytometer, and API connectivity to the Cytobank platform.
(A)
(B)
(C)
Figure 2. A) Biomek i7 Multichannel workstation with direct access to the CytoFLEX LX flow cytometer. B) Deck layout of Biomek i7 Multichannel workstation. C) An enhanced view of automated loading of a 96-well plate into the CytoFLEX LX analyzer with the Biomek i7 workstation gripper.
Experiment design and execution of automated workflow
To demonstrate the feasibility of the fully automated workflow for immune cell subset identification, a DURAClone 10-color T cell subset dry reagent panel in 96-well plate format was used to characterize maturation stages of T cells present in PBMC samples from 16 healthy donors, using six replicate sample wells for each donor across the 96-well plate.
The hands-on steps required by the operator before starting the automated, walkaway workflow are:
- open a sample processing method on the Biomek i7 Multichannel workstation,
- open an experiment programmed for sample acquisition on the CytoFLEX LX flow cytometer,
- log in to the Cytobank user account at the Biomek i7 computer and
- load all required labware, samples and reagents onto the Biomek workstation deck.
After the automated workflow is started, the operator is free to walk away while the Biomek i7 Multichannel workstation performs the steps programmed in the method for automated sample processing as shown in Figure 3, including automated mixing by the on-deck orbital shaker (Beckman Coulter) and an automated wash step using an integrated microcentrifuge (Agilent Technologies).
Figure 3. Automated steps for staining donor PBMC samples with DURAClone T cell subset dry antibody cocktail in 96-well plate format with addition of a drop-in liquid DAPI viability dye on the Biomek i7 Multichannel workstation integrated with a CytoFLEX LX flow cytometer and microcentrifuge.
Sample acquisition
PBMC samples were acquired on the integrated CytoFLEX LX (6-laser/21-color) flow cytometer with UVviolet- blue-yellow green-red-infrared configuration using CytExpert software v2.5. The gating strategy shown in Figure 4 was used to gate cell populations and record data using acquisition stop settings of either 10,000 LIVE CD3+ T cells or acquisition time of 180 seconds for each sample well.
(A)
(B)
(C)
(D)
Figure 4. Gating Strategy in CytExpert Acquisition Software: (A) Cells were gated using FSC-A vs. SSC-A dot plot, (B) LIVE and DEAD cells were gated using DAPI viability dye vs. SSC-A dot plot with the Cells gate applied, (C) CD45+ cells were gated using CD45-Krome Orange vs. SSC-A dot plot with the LIVE Cells gate applied and (D) CD3+ cells were gated using CD3-APC-A750 vs. SSC-A dot plot with the LIVE CD45+ gate applied.
Fluorescence was measured for each directly conjugated antibody in the 10-color T cell subset panel and DAPI viability dye staining using the appropriate detector matching the fluorochrome excitation and emission criteria.
Scripted step for automated data upload and analysis in the Cytobank platform
A scripted step was developed for Biomek Software using the Cytobank API to automate upload and analysis of data generated from the CytoFLEX LX flow cytometer via the Cytobank data analysis cloud, with no human intervention.
The scripted step automatically creates a new experiment in the Cytobank platform, adds the CytoFLEX LX-generated FCS files along with the sample tags containing metadata and donor ID from a user-supplied .tsv file, and applies scaling to the new experiment using a reference experiment in the platform (Figure 5).
(A)
(B)
Figure 5. A) Representative data showing the Scaling Editor result from an automatically uploaded FCS file in the Cytobank platform. B) Screen shot of automatically assigned sample tags in the Cytobank platform captured from local .tsv file listing individual donor IDs associated with the acquired sample well.
Automatic gating analysis—training a new model
Prior to the automated run, an automatic gating model identifying 30 different populations was trained on 24 files from the reference experiment with equal event sampling, four optimal clusters and a training magnification of 1, resulting in an average KPI for population identification of 92.
Please refer to our support articles to learn more about how to evaluate automatic gating performance in the Cytobank platform.
The gating strategy used in the reference experiment for identifying CD4+ and CD8+ T cells, dissecting naïve, effector, memory and terminal differentiation stages based on the expression of CD27, CD28, CD45RA and CD197 (CCR7); CD57+ and CD279+ (PD-1) is shown in Figure 6.
(A)
(B)
Figure 6. A) Representative data plots of gating strategy in Cytobank platform identifying the different T cell subset populations. B) Sunburst plot created in Cytobank platform displaying hierarchical relationship of manual gates from a representative sample.
Automated data analysis
Upon data upload of all the FCS files to the Cytobank platform, an inference run of the automatic gating model is initiated with no human intervention. During the inference run, target cell populations are automatically identified, thus removing the time-consuming and operator-dependent manual tasks of applying and adjusting gates for each sample by the operator. Upon completion of the inference run, the Cytobank platform sends an e-mail notification to the operator reporting that the analyzed results from the new experiment are ready for inspection and review in the Cytobank platform.
Review of automatic gating results
In the Gating Editor page of the Cytobank experiment, populations created by the inference run will appear marked by [auto] as shown in Figure 7A, and can be used to perform any advanced analysis within the Cytobank platform (e.g., dimensionality reduction with viSNE or UMAP). In addition, overlay dot plots (Figure 7B) can be easily created to provide the operator with a quick visual comparison of staining patterns from selected cell populations in the replicate samples from individual donors.
(A)
(B)
Figure 7. A) Screen shot of the Gating Editor menu in the Cytobank platform from one sample showing the different cell populations and gates automatically created by the inference run. B) Overlay dot plots of six replicate samples (rows) for T cell subsets and T cell memory populations from six individual donor PBMC samples (columns).
Please refer to our support articles to learn more about Automatic Gating in the Cytobank platform.
Exporting, graphing and statistical analysis of data
After gating has been reviewed, the export statistics option in the Cytobank platform enables the operator to quickly export selected statistics from the experiment into a user-defined table format. In addition, a time-saving template can be defined in the export statistics options for the operator to select and use in future experiments.
To save even more time and reduce manual data entry errors, the Cytobank platform enables operators to perform statistical tests on datasets directly in the Illustration Editor page, with no need to copy/ paste and/or manually enter data into a separate statistical analysis software program. Figure 8 shows an example of a statistical analysis test directly performed in the Cytobank platform comparing event count results for manually drawn CD4 gates to automatically created CD4 gates.
Figure 8. Box plot of event count for the CD4 population comparing the manually identified population versus the one identified by the automatic gating algorithm. A Student t-test was run with a Bonferroni post-test and p-values for all comparisons are shown.
Operators can also create graphs of the experiment data within the Cytobank platform by selecting the Plots menu in Illustration Editor. As an example, summary box plots (Figure 9) were created for CD3, CD4 and CD8 populations of T cells to show the reproducibility of cell populations in the six replicates for each donor sample, as well as show the biological differences in the different T cell population percentages among individual donors.
Figure 9. Summary box plots created in the Cytobank Illustration Editor showing percentages of CD3+, CD3+/CD4+, and CD3+/CD8+ T cell populations from six replicate samples in six individual donor PBMC samples. The dots represent independent samples.
Summary
In this application note, we have described the feasibility of implementing a fully automated flow cytometry workflow solution for immune cell subset identification. We demonstrated how automation and advanced technologies can be used by researchers to eliminate tedious and repetitive manual steps in daily lab work and free up time to focus on other tasks and responsibilities.
We leveraged some of the different instrumentation, reagents and software readily available from Beckman Coulter Life Sciences to obtain robust and reproducible data for immune T cell subset identification from 16 different PBMC samples in an automated manner. For this study, we used a Biomek i7 Multichannel liquid handler integrated with a CytoFLEX LX flow cytometer, together with a DURAClone dry antibody cocktail in a 96-well plate format and drop-in liquid DAPI viability dye. To further automate the workflow, we developed a custom script to automatically upload data files to the Cytobank platform, automatically analyze the data using a trained gating model, and notify the operator when results are ready for review.
To accelerate the data review and reporting process, we described several additional built-in features available within the Cytobank platform that enable researchers to perform statistical tests, export statistics, create overlay plots, and graph data without having to copy/paste data into different software programs.
In summary, by embracing automation and advanced technologies, scientists can accelerate their research efforts and ultimately bring innovative therapies to patients more efficiently and reliably.
Materials and methods
PBMC Samples: Cryopreserved PBMCs from 16 normal donors were quickly thawed in a 37°C water bath, washed two times with warm RPMI-1640 media, resuspended in 1 mL of staining buffer (PBS + 2% BSA) and filtered through a 70 μm cell strainer. PBMC samples were assessed for cell count and viability with DURAClone IM Count Tubes using the vendor-recommended staining protocol and acquired on a CytoFLEX series flow cytometer.
Dry Reagent Panels (Beckman Coulter Life Sciences)
*Product - Part Number | Staining Panel |
DURAClone IM Count Tube – C00162 | CD45-FITC, Counting Beads, 7-AAD |
Lucid IM T Cell Subsets (96-well shallow plate format) – custom product | CD57-PB, CD45-KrO, CD45RA-FITC, CCR7-PE, CD28-ECD, PD-1-PC5.5, CD27-PC7, CD4-APC, CD8-A700, CD3-APC-A750 |
Liquid reagents
*Reagent | Supplier | Part number |
RPMI-1640 media | HyClone | SH30255.01 |
Bovine Serum Albumin (BSA) | Sigma-Aldrich | A3059 |
PBS (1X) Buffer | Gibco | 10010-023 |
DAPI Viability Dye | Beckman Coulter Life Sciences | B30437 |
Biomek i7 Consumables
Labware | Supplier | Part number |
Biomek i-Series, 90 μL pipette tip, non-sterile | Beckman Coulter Life Sciences | B85881 |
Biomek i-Series, 230 μL pipette tip, non-sterile | Beckman Coulter Life Sciences | B85903 |
Biomek i-Series, 1025 μL pipette tip, sterile, filtered | Beckman Coulter Life Sciences | B85955 |
96-Well U-bottom plate, polystyrene | Greiner Bio-One | 650101 |
Reservoir, single cavity, 300mL | Agilent Technologies | 204017-100 |
Quarter modular reservoir, divided by width, sterile | Beckman Coulter Life Sciences | 372792 |
Cytobank Data Analysis Platform* (Beckman Coulter Life Sciences)
*For Research Use Only. Not for use in diagnostic procedures.