Suspension MC Session

University of Lille, CNRS, Inserm, CHU Lille, Institut Pasteur de Lille, US 41 – UAR 2014 – PLBS, F-59000 Lille, France

Batch Effects Correction in Mass Cytometry Data
High-content cytometry allows a fine characterization of cell populations in a single sample thanks to the analysis of several dozen targets in a unique panel. Experimental solutions and detailed data processing pipelines were developed to reduce both the staining conditions variability between samples and the number of tubes to handle. This refinement and the possibility of multiplexing several samples (barcoding) have interested large-scale studies. However, an unavoidable variability appears between samples, barcodes, series and instruments (in multicenter studies) contributing to “batch effects” that must be properly controlled.

To correct these undesirable effects, several approaches (clustered or not) and methods (linear or quantile segmented) have been proposed, based or not on a reference sample present in each batch, but they all lack transpacency, intuition and user-friendlyness. 

We created a dedicated package named CytoBatchNorm, based on the CytofBatchAdjust script by Schuyler et al., which uses the unclustered events of the reference sample to calculate linear corrections. It provides a graphical interface allowing users to define a specific correction for each marker in a single run, with graphs visualization that guides users through quickly setting the parameters. It allows corrections to be previewed and inter-marker effects to be checked as the settings are updated. CytoBatchNorm will help the cytometry community to adequately scale data between batches, reliably reducing batch effects and improving subsequent dimension reduction and clustering.

Biosketch
Olivier Molendi-Coste (PhD) is a research engineer in the Scientific and Medical French Research Institute (INSERM). After obtaining a dietitian diploma (2001, Paris, France) and a PhD in neurosciences (2007, Lille, France), he performed a post-doctoral scholarship on HFD-induced insulin resistance in liver and adipose tissues (Brussels, Belgium), and oriented towards the study of the central roles of immune cells influence on tissue homeostasis in insulin resistance and cardiovascular diseases (INSERM U1011 “Nuclear receptors, cardiovacular diseases and Diabetes”, Lille, France). He drove the cytometry plateau from the European Genomic Institute for Diabetes (EGID) during 8 years before taking over the scientific management of the cytometry platform from Lille Pasteur Institute in 2022 (US41 PLBS). He is the coordinator of the Mass Cytometry Group of the French Association for Cytometry and participates to collectively improve unsupervised high-content data analysis and batch effects correction.

Pfizer-University of Granada-Junta de Andalucía Centre for Genomics and Oncological Research (GENYO), Granada, Spain

Center Harmonization for Multicenter CyTOF Immune Monitoring

3TR is a multicenter consortium aiming to identify shared molecular signatures across seven chronic immune-mediated diseases. As part of this effort, immunophenotyping data will be generated for more than 2,500 samples, requiring a harmonized and optimized multicenter CyTOF acquisition strategy.

Optimize the sample acquisition protocol for multicenter CyTOF studies and determine whether data acquired across different instruments can be aligned.

A pilot experiment was performed to optimize acquisition using 46-plex panel together with barcoded whole-blood samples from three donors and compensation beads. The samples were prepared, frozen, and later acquired on one three HELIOS, and one CYTOF XT instrument across three acquisition batches. Sample acquisition scheme varied by batch: Batch 1 used one long run per aliquot, while Batches 2–3 used two shorts runs. PCA was applied using median signal intensities (MSI) to evaluate the effect of acquisition time and aliquot used within and across centers. Main cell populations were manually gated and MSI, coefficients of variations (CV) were calculated. Center enrichment and batch effects were assessed.

The compensation matrix generated using the brightest cytometer (Center1), demonstrated the most effective spillover correction. Flow rate instability contributed to signal drifts in several markers, affecting background-related intensities and cell-associated signals The PCA analysis showed that restricting the analysis to the first 65 minutes of acquisition improved clustering for all centers. Bead-based normalization reduced center-associated variability, resulting in CV values of MSI and cell frequencies below 20% for most cell populations. Samples clustered according to the donors, but center effect was observed, thus requiring reference-based normalization.

Multicenter CyTOF experiments are feasible, however special sample acquisition protocol and additional reference-based normalization are required to align the data.

Department of Internal Medicine II, Medical Center – University of Freiburg, Freiburg, Germany

SceniTOF – Functional single-cell multiplexed metabolic profiling to map bioenergetics of heteroegeneous immune cells by mass cytometry in mice and men

Immune cell activation, differentiation, and effector function are tightly coupled to cellular metabolism, but many functional metabolic assays lack single-cell resolution. Single-cell metabolic regulome profiling (scMEP) by mass cytometry quantifies metabolic proteins alongside immunophenotypes, but protein levels do not necessarily reflect metabolic activity or flux. The method SCENITH (Single Cell ENergetIc metabolism by profilIng Translation inHibition), developed by flow cytometry, may overcome this limitation by quantifying pathway-specific effects on protein synthesis after brief ex vivo inhibitor treatment.

Here we established a mass cytometry-compatible version of the SCENITH assay using a secondary metal-coupled antibody to detect the translation inhibitor puromycin as an integral part of the SCENITH assay and combine it with the scMEP approach. Briefly, human or murine immune cells are treated ex vivo or after in vitro stimulation with inhibitors of glycolysis and oxidative phosphorylation, pulsed with puromycin, barcoded, and stained with a multiplexed CyTOF panel targeting metabolic regulators, lineage markers, and activation states. This combined SCENITH–CyTOF approach enables scalable mapping of energetic pathway dependence while concurrently resolving phenotypic and metabolic profiles. We demonstrate the applicability of the SceniTOF assay to identify the differences in metabolic state and flux across human T cell differentiation states and the murine B cell compartment.

Together, this integrated approach links multiplexed single-cell phenotyping with functional metabolic readouts, allowing SceniTOF to relate metabolic potential to actual metabolic flux by mass cytometry.

Pfizer-University of Granada-Junta de Andalucia Center for Genomics and Oncological Research (GENYO), Granada, Spain

What it takes to CyTOF-profile 2,500 clinical samples across four immune-mediated diseases

Background: 3TR is a multicenter consortium studying immune-mediated diseases to find common molecular signatures. Mass cytometry is employed to profile immune cell populations.

Objective: Establish a standardized staining and antibody-validation protocol to enable immune monitoring of >2,500 blood samples from 3TR clinical studies.

Methods: To ensure consistent recruitment, whole blood was fixed and stored at -80°C until staining. Antibody panel development included clone selection in fresh blood, antigen stability testing and titration in fixed/frozen samples from healthy donors and patients, using flow or mass cytometry. Large-scale metal conjugation was performed for higher yields. To improve technical robustness, two protocols were tested. Cells from 3 donors were thawed, lysed, aliquoted and either refrozen immediately or barcoded, pooled and refrozen at -80°C. Depending on the workflow, samples were subsequently barcoded and pooled or immediately stained with the frozen antibody cocktail.

Results: Of 12 antibody clones evaluated by flow cytometry, 8 met criteria for inclusion. Overall, 55 markers were titrated in CyTOF and 50 approved; 35 antibodies were conjugated in-house, 21 at large scale. The final panel includes 42 common and 4 disease-specific markers, comprising lineage and functional markers. Panel was tested following the described workflow. Several markers displayed shifts in signal intensity between 1-week and 1-month storage. Consequently, a maximum 10-day storage period before staining was established.

Conclusions: Standardized CyTOF protocol and 46-plex antibody panel enable robust and reproducible immune profiling of over 2,500 samples from patients, supporting biomarker discovery and personalized therapies.