Utilising mass cytometry with CD45 barcoding and standardised leucocyte phenotyping for immune trajectory assessment in critically ill patients

Editor—Critically ill patients have rapidly changing longitudinal immune responses (immune trajectory) to infection and tissue injury. If we can delineate this immune trajectory, then we can time treatments that stimulate or depress the immune system (immunomodulation) to mirror the dominant immune signal at distinct stages of critical illness. These rapidly changing longitudinal immune responses generate two major inferential challenges. First, the signal for illness-specific alterations must be delineated from the noise of inter-experimental variation within a study. Second, lack of standardisation of immune cell phenotyping assays between studies makes it difficult to discern whether observed differences between studies represent signal or noise.1Rubio I. Osuchowski M.F. Shankar-Hari M. et al.Current gaps in sepsis immunology: new opportunities for translational research.Lancet Infect Dis. 2019; 19: E422-E436Abstract Full Text Full Text PDF PubMed Scopus (64) Google Scholar These challenges need addressing in sepsis, as innate and adaptive immune systems are profoundly altered, immune phenotyping is not standardised, and rapidly changing sepsis-specific immune trajectory remains poorly characterised, for optimal immunomodulation.1Rubio I. Osuchowski M.F. Shankar-Hari M. et al.Current gaps in sepsis immunology: new opportunities for translational research.Lancet Infect Dis. 2019; 19: E422-E436Abstract Full Text Full Text PDF PubMed Scopus (64) Google Scholar,2Fish M. Arkless K. Jennings A. et al.Cellular and molecular mechanisms of IMMunE dysfunction and Recovery from SEpsis-related critical illness in adults: an observational cohort study (IMMERSE) protocol paper.J Intensive Care Soc. 2020; https://doi.org/10.1177/1751143720966286Crossref Google Scholar To address the challenge of inter-sample and inter-experiment variability and to enable simultaneous leucocyte phenotyping in longitudinal samples from the same patient, we developed a multiplexing approach using the stably expressed pan-leucocyte cluster of differentiation 45 (CD45) antigen as the target (referred to as barcoding) and cytometry by time of flight (CyTOF) as the phenotyping method.3Olsen L.R. Leipold M.D. Pedersen C.B. Maecker H.T. The anatomy of single cell mass cytometry data.Cytometry A. 2019; 95: 156-172Crossref PubMed Scopus (37) Google Scholar To address the non-standardised phenotyping challenge, we used markers proposed in the standardised human immunophenotyping panel4Maecker H.T. McCoy J.P. Nussenblatt R. Standardizing immunophenotyping for the human immunology project.Nat Rev Immunol. 2012; 12: 191-200Crossref PubMed Scopus (613) Google Scholar and included selected immune state markers that are treatment targets such as checkpoint molecules and the human leucocyte antigen DR isotype (HLA-DR).5van der Poll T. van de Veerdonk F.L. Scicluna B.P. Netea M.G. The immunopathology of sepsis and potential therapeutic targets.Nat Rev Immunol. 2017; 17: 407-420Crossref PubMed Scopus (568) Google Scholar This report summarises the development of CD45 barcoding, assessment of major cell populations with a standardised human immunophenotyping panel,4Maecker H.T. McCoy J.P. Nussenblatt R. Standardizing immunophenotyping for the human immunology project.Nat Rev Immunol. 2012; 12: 191-200Crossref PubMed Scopus (613) Google Scholar and comparison of individual to multiplexed samples. For all experiments, cryopreserved peripheral blood mononuclear cells (PBMC) were isolated using Ficoll-Hypaque (Sigma-Aldrich, St. Louis, MO, USA) density gradient centrifugation. For CD45 barcoding, commercially available CD45 89Y (Fluidigm Corp., South San Francisco, CA, USA) and in-house conjugated metal isotopes (115In, 159Tb, 209Bi) to the CD45 antibody (BioLegend®, BioLegend Inc., San Diego, CA, USA) were used. PBMCs were thawed in a 37°C water bath, washed in Roswell Park Memorial Institute (RPMI) 1640 medium supplemented with 10% fetal bovine serum and 5% penicillin–streptomycin, resuspended in 1 ml, and counted. Then 3×106 cells from five donors were transferred to 5-ml polystyrene tubes for barcoding. Barcoded samples were washed once in phosphate-buffered saline (PBS), resuspended in 98 μl PBS, stained (1:100) with a unique combination of CD45 antibodies for 20 min, washed twice in 4 ml PBS, and combined into a single barcoded sample. The CD45 barcoding rationale, purpose, and novelty are summarised in Figure 1a. For standardised leucocyte phenotyping, individual samples with 3×106 cells with combined barcoded samples were adjusted to 2×107 cells ml−1 in PBS, stained for viability in pre-warmed 5 μM cisplatin solution for 60 s, Fc receptors blocked (5 μl FC block [TruStain FcX BioLegend®; BioLegend] for 10 min) and incubated with an antibody cocktail containing 34 surface markers for 30 min, as per the Fluidigm staining protocol.6Available from: https://www.fluidigm.com/documents (accessed on 03-December-2020).Google Scholar Samples were then fixed in 1.6% formaldehyde for 10 min, centrifuged, and incubated in 125 μM Intercalator-Ir fix and perm buffer (Fluidigm) at 4°C until acquisition the following day using a Helios mass cytometer (Fluidigm). Samples were washed once in CSB, twice in cell acquisition solution (CAS), adjusted to 5×105 per ml in CAS with 1:10 EQ™ four element calibration beads (Fluidigm), and data were acquired at 300 events s−1 (Fig. 1b). For comparisons between individual and multiplexed samples, we normalised flow cytometry standard (FCS) files with EQ beads. Barcoded samples recorded in multiple recordings were concatenated and debarcoded using the Catalyst package.7Crowell H, Zanotelli V, Chevrier S, Robinson M. CATALYST: Cytometry dATa anALYSis Tools. R package version 1.14.0, 2020. Available from: https://github.com/HelenaLC/CATALYST. Bioconductor version: Release (3.12) (accessed on 17 11 2020).Google Scholar All samples were manually gated (https://www.cytobank.org). Proportions of live cells and median fluorescent intensity (MFI) of all markers on all populations were exported as comma separated values (CSV) files for analyses. We compared proportions of cell populations, the MFI of all 34 markers between staining conditions, and conducted unsupervised analyses using principal component analyses (PCA). We also conducted unsupervised analysis of single cell data using self-organising map8Van Gassen S. Callebaut B. Van Helden M.J. et al.FlowSOM: using self-organizing maps for visualization and interpretation of cytometry data.Cytometry A. 2015; 87: 636-645Crossref PubMed Scopus (518) Google Scholar (FlowSOM) projected onto uniform manifold approximation and projection (UMAP).9McInnes L. Healy J. Saul N. Großberger L. UMAP: uniform manifold approximation and projection.J Open Source Softw. 2018; 3: 861Crossref Google Scholar The stain intensity of each CD45 metal isotope was compared on debarcoded and individual samples as a surrogate for its binding characteristics. Histogram peaks of in-house conjugated metal isotopes for CD45 were similar to commercially available CD45 89Y. There was an expected loss of 89Y signal in barcoded samples owing to binding competition between two antibodies. Clear positive and negative signals reflect the key for debarcoding (Fig. 1c). The 34-surface marker panel consisted of 30 class and four state markers. This enabled identification of 29 major leucocyte populations, including lymphocyte, monocyte, natural killer, and dendritic cell subsets (Fig. 1d). These also included cellular subsets of therapeutic relevance in sepsis such as T-helper subsets (Th-1, Th-2, Th-17, regulatory T cells), expression of checkpoint molecules on T cells, and expression of HLA-DR on monocytes. High correlation was observed between individually stained and barcoded multiplexed samples in proportions of 29 leucocyte subsets (R=0.99; P<0.001; Fig. 1e) and MFI of 34 markers (R=0.92, P<0.001; Fig. 1f). In the PCA, the barcoded and individual samples clustered together for each donor (Fig. 1g). Furthermore, barcoded and individually stained leucocyte population projections were similar with FlowSOM projected on UMAP (Fig. 1h). Overall, CD45 barcoding facilitates longitudinal patient sample measurements to derive immune trajectories, while maintaining standardised immunophenotyping to identify lymphocyte subsets in sepsis patients in whom lymphopaenia is common, as described in our study protocol.2Fish M. Arkless K. Jennings A. et al.Cellular and molecular mechanisms of IMMunE dysfunction and Recovery from SEpsis-related critical illness in adults: an observational cohort study (IMMERSE) protocol paper.J Intensive Care Soc. 2020; https://doi.org/10.1177/1751143720966286Crossref Google Scholar This formed the rationale for using PBMCs instead of whole blood for phenotyping and limited barcoding to maximise detection of lymphocyte subsets that are treatment targets (such as terminally exhausted T cells) in sepsis patients. Our use of lanthanide metals is a low-cost, flexible panel design method, whereas barcoding on CD45 potentially enables our methods to be used in whole-blood phenotyping and ex vivo stimulation experiments. Thus, our approach is a useful advance on a recent proof-of-concept report of mass cytometry in five sepsis patients.10Gossez M. Rimmele T. Andrieu T. et al.Proof of concept study of mass cytometry in septic shock patients reveals novel immune alterations.Sci Rep. 2018; 8: 17296Crossref PubMed Scopus (17) Google Scholar Although variability between barcoded experiments has not been addressed, this could be facilitated by using the same technical control in each experiment for batch normalisation. Discriminant biomarkers for trajectory identified using our approach to standardised immunophenotyping can then be applied to near patient testing using flow cytometry. In summary, we report a feasible barcoding approach for standardised immunophenotyping to derive immune trajectory in critically ill patients. Our method would highlight leucocyte immune trajectory in sepsis patients and inform development of near patient immunophenotyping tests to provide targeted and timely immunomodulation in sepsis. The authors thank past and future patients, their families and medical professionals who have been involved in the IMMERSE study.2Fish M. Arkless K. Jennings A. et al.Cellular and molecular mechanisms of IMMunE dysfunction and Recovery from SEpsis-related critical illness in adults: an observational cohort study (IMMERSE) protocol paper.J Intensive Care Soc. 2020; https://doi.org/10.1177/1751143720966286Crossref Google Scholar We thank the flow cytometry core at the NIHR Biomedical Research Centre for logistical support.