50-color phenotyping of the human immune system with in-depth assessment of T cells and dendritic cells

Abstract

We report the development of an optimized 50-color spectral flow cytometry panel designed for the in-depth analysis of the immune system in human blood and tissues, with the goal of maximizing the amount of information that can be collected using currently available flow cytometry platforms. We established and tested this panel using peripheral blood mononuclear cells (PBMCs), but included CD45 to enable its use for the analysis of human tissue samples. The panel contains lineage markers for all major immune cell subsets, and an extensive set of phenotyping markers focused on the activation and differentiation status of the T cell and dendritic cell (DC) compartment.

We outline the biological insight that can be gained from the simultaneous measurement of such a large number of proteins and propose that this approach provides a unique opportunity for the comprehensive exploration of the immune status in tissue biopsies and other human samples with a limited number of cells. Of note, we tested the panel to be compatible with cell sorting for further downstream applications. Furthermore, to facilitate the wide-spread implementation of such a panel across different cohorts and samples, we established a trimmed-down 45-color version which can be used with different spectral cytometry platforms.

Finally, to generate this panel, we utilized not only existing panel design guidelines, but also developed new metrics to systematically identify the optimal combination of 50 fluorochromes and evaluate fluorochrome-specific resolution in the context of a 50-color unmixing matrix.

Background:

The immune system serves as the body's defense system against pathogens and is also essential for maintaining steady-state homeostasis in tissues (1) and preventing the development of malignant tumors (2). The composition and activation status of immune cells in the periphery and in tissues can be used to extrapolate immune cell differentiation and function. To facilitate data interpretation, an immune cell population is ideally analyzed in the context of other immune cell populations. Thus, to comprehensively study the functional state of the immune system, it is highly beneficial to capture as much information from as many different cell types as feasible. This is particularly relevant for assessing immune cell function in situ, for example in human tissue samples (3). Moreover, these human tissues are often limited in size and availability, e.g. tissue biopsies (4) or resected pieces of tumor tissue (5), which precludes parallel analysis with multiple panels or multiple applications. The development of an analysis approach that can provide broad and, for some cell subsets, in depth-phenotyping, paired with the ability to preserve cell populations of interest for downstream applications such as single-cell RNA-sequencing, is hence of importance.

The interactions between professional antigen-presenting cells (APCs) and different T cell subsets (6) are of particular interest in the context of studying anti-tumor immune responses (7). Dendritic cells (DCs) are highly specialized APCs, and generally divided into cross-presenting cDC1s and cDC2s (8), with CD163⁺ cDC3s being a more recently described subset during inflammatory conditions (9). Each of these DC populations appears to have a distinct function for steering an adaptive immune response. T cells consist of conventional CD4⁺ T cells, regulatory CD4⁺ T cells (Tregs), CD8⁺ T cells, γδ T cells and subsets of T cells with semi-invariant T cell receptors including mucosal-associated invariant T cells (MAIT cells) and invariant NK-T cells (10).

The panel presented here (Figure 1A) was designed to comprehensively capture the differentiation and activation status of T cells and APCs, while also measuring B cell, NK cell and innate lymphoid cells (ILCs) phenotypes (list of markers depicted in Figure 1B and Table 2). Optimization was done on human cryopreserved PBMCs, but the panel includes CD45 as a pan-hematopoietic marker and has been tested on human tissue-derived leukocytes (data not shown).

Figure 1: Overview gating of the 50-color panel on cryopreserved PBMCs.

PBMCs were obtained from commercial vendors, stained as described in the online section of the manuscript and acquired on a BD FACSDiscover^™ S8. The optical configuration of the instrument is described in Online Figures 1-2. Additional gating and staining controls are shown in Online Figures 9-11. Pre-gating of plots is annotated in the figure or indicated by dotted black arrows. For some plots different donors are shown for clarity. The gating strategy has been devised in such a way that the staining pattern for every marker in the panel can be shown at least once on a single A4 page. The raw data has been deposited on Flowrepository with the identifier FR-FCM-Z73V. Abbreviations: BUV: Brilliant Ultraviolet; BV: Brilliant Violet; BB: Brilliant Blue; RB: RealBlue; AF: AlexaFluor; PE: Phycoerythrin; APC: Allophycocyanin; Qdot: Quantum Dot; NIR: Near-Infrared;

(A) Gating strategy for CD45+ live cells, monocytes, B cells and γδ and αβ T cells.

(B) Overview of the 50 targets analyzed with this experiment. Some of the markers can be used for phenotyping multiple immune cell lineages.

(C) Representative plots for the main phenotyping markers in the B cell lineage (IgG, IgD, IgM and CD24).

(D) Gating strategy to delineate invariant NKT cells, MAIT cells, CD4⁺ and CD8⁺ T cells, as well as CD4⁺ regulatory T cells (Tregs).

(E) Representative plots for CD69, CD103, CD57 and PD-1 expression on non-naïve CD8⁺ cytotoxic T cells.

(F) Expression pattern for CD39, CXCR3, CCR4, CD45RO and ICOS on the CD4⁺ Treg population.

(G) Histogram overlays for the expression pattern of BTLA, CD27, CD28, CD38, TIGIT and KLRG1 on NK cells (grey), MAIT cells (orange), CD4⁺ Tregs (red), CD4⁺ non Tregs (purple), CD8⁺ naïve T cells (green) and CD8⁺ non-naïve T cells (blue). Dotted red lines indicate positivity cut-offs.

(H) Gating strategy for NK cells and NK cell subsets based on CD56, CD161, CD16 and Nkp46.

(I) Gating strategy for Basophils (CD123⁺ FcER1⁺ HLA-DR⁻), plasmacytoid DCs (CD303⁺ HLA-DR⁺), pan conventional DCs (CD11c⁺ HLA-DR⁺), and the cDC1 (CD141⁺) and cDC2 (FcER1⁺) subsets.

(J) Histogram overlays for the expression pattern of CD86, CD40, CD11b, CD1c and CD163 on B cells (grey), CD8⁺ naïve T cells (green, negative control), CD14⁺ monocytes (purple), CD16⁺ monocytes (yellow), CD141⁺ cDC1s (red) and FcER1⁺ cDC2s (blue). Dotted red lines indicate positivity cut-offs.

(K) Gating strategy for Lin⁻ CD2⁺ CD127⁺ innate lymphoid cells (ILCs).

(L) A selection of fluorescence-minus-one (FMO) controls for the indicated markers: PD-1 and BTLA on CD8⁺ T cells, ICOS and CCR4 on Tregs, and CD86 and CD1c on Lin-cells as indicated. Dotted red lines indicate positivity cut-offs. Note that there is no or only negligible spreading error (SE) present. Additional FMO and gating controls are shown in Online Figure 9.

Table 2.

Reagents used for OMIP-XX

Specificity	Alternative Name	Clone	Fluorochrome	Purpose
CD45RA	Isoform of CD45	HI100	Spark UV 387	Phenotyping of T cells, naive vs memory
CD40	TNFRSF5	5C3	BUV395	Phenotyping of B cells and DCs, activation
CD3	Part of the TCR complex	UCHT1	BUV496	Lineage marker of pan T cells
CD56	Neural cell adhesion molecule 1, NCAM1	NCAM16.1	BUV563	Lineage marker of pan NK cells
CD2	LFA-2	S5.5	Qdot 605	Identification of ILCs, NK cell phenotyping
CD141	BDCA-3, or Thrombomodulin	1A4	BUV615	Lineage marker of cDC1s
Nkp46	CD335, Natural cytotoxicity triggering	9E2	Qdot 625	Phenotyping of NK cells
CD303	Clec4c	V24-785	BUV661	Lineage marker of pDCs
CD86	B7-2	FUN-1	BUV737	Phenotyping of DCs and monocytes, activation
CD45	Protein tyrosine phosphatase, receptor	HI30	BUV805	Pan-Haematopoietic marker
CD161	Killer cell lectin-like receptor subfamily B 1,	DX12	BV421	Phenotyping of T cells and NK cells, MAIT marker
IgM	Immunoglobulin M	SA-DA4	Super Bright 436	Phenotyping of B cells and DCs
CD1c	N.A	L161	Pacific Blue	Phenotyping of DCs and monocytes
CD28	N.A	CD28.2	BV480	Phenotyping of T cells
CD19	N.A	SJ25C1	AmCyan	Lineage marker of B cells
CD11c	Integrin alpha X, ITGAX	S-HCL-3	BV510	Lineage marker for conventional DCs
CD45RO	Isoform of CD45	UCHL1	BV570	Phenotyping of T cells, naive vs memory
CD197	CCR7, chemokine receptor 7	G043H7	BV605	Phenotyping of T cells, naive vs memory
CD69	N.A	FN50	BV650	Phenotyping of T cells, tissue residency marker
FcER1	high-affinity IgE receptor	AER-37	BV711	Lineage marker for cDC2s and basophils
CD103	Integrin alpha E, ITGAE	Ber-ACT8	BV750	Phenotyping of T cells, tissue residency marker
CD194	CCR4, chemokine receptor 4	1G1	BV786	Phenotyping of T cells, migration
CD25	Interleukin-2 receptor alpha chain, IL2RA	BC96	BB515	Phenotyping of T cells, marker for Tregs
CD14	Lipopolysaccharide receptor	M5E2	RB545	Lineage marker for monocytes
CD8	N.A	OKT8	NovaFluor Blue 555	Lineage marker for CD8+ T cells
CD4	N.A	SK3	NovaFluor Blue 585	Lineage marker for CD4+ T cells
CD272	B- and T-lymphocyte attenuator, BTLA	J168-540	RB613	Phenotyping of T cells, activation
CD27	TNFRSF7	M-T271	BB660	Phenotyping of T cells, activation
CD11b	Integrin alpha M, ITGAM	M1/70	PerCP	Phenotyping of APCs
IgG	Immunoglobulin G	G18-145	BB700	Phenotyping of B cells, differentiation
TCRgd	Gamma delta T cell receptor	B1	RB705	Lineage marker of gd T cells
CD127	Interleukin-7 receptor subunit alpha, IL7RA	HIL7R-	RB744	Phenotyping of T cells, Treg identification
TIGIT	T cell immunoreceptor with Ig and ITIM	TgMab-2	RB780	Phenotyping of T cells and NK cells
IgD	Immunoglobulin D	W18340F	PerCP-Fire 806	Phenotyping of B cells, differentiation
CD57	Human natural killer-1, HNK-1	NK-1	PE	Phenotyping of T cells and NK cells
CD20	N.A	2H7	Spark YG 593	Lineage marker of B cells
CD24	Signal transducer CD24	SN3	PE-Alexa Fluor 610	Phenotyping of B cells, differentiation
CD183	CXCR3, CX chemokine receptor 3	G025H7	PE-Fire 640	Phenotyping of T cells, migration
CD123	Interleukin-3 receptor, IL3RA	9F5	PE-Cy5	Lineage marker of Basophils and pDCs
CD278	Inducible T-cell costimulator, ICOS	ISA-3	PE-Cy5.5	Phenotyping of T cells, activation
CD16	Fc gamma receptor, FcγRIII	3G8	PE-Alexa Fluor 700	Lineage marker of monocytes, phenotyping of NK
CD279	PD-1, Programmed Death 1	EH12.1	PE-Cy7	Phenotyping of T cells, activation and exhaustion
HLA-DR	MHC class II	L243	PE-Fire 810	MHC class II
Va24-JA18	V alpha 24 – J alpha 18 TCR chain	6B11	APC	Marker for invariant NKT cells
Va7.2	V alpha 7.2 TCR chain	3C10	Alexa Fluor 647	Marker for MAIT cells
KLRG1	Killer cell lectin-like receptor subfamily G	SA231A2	Spark NIR 685	Phenotyping of T cells and NK cells, activation
CD39	Ectonucleoside triphosphate	A1	R718	Phenotyping of T cells and NK cells, activation
Live/Dead	N.A	Amine	Zombie-NIR	Live/Dead cell discrimination
CD163	Scavenger receptor for hemoglobin	GHI/61	APC-Cy7	DC phenotyping marker, marker for DC3s
CD38	cyclic ADP ribose hydrolase	HB-7	APC-Fire 810	Phenotyping of T cells and NK cells, activation

B cells are identified by the lineage markers CD19 and CD20 (Figure 1A). Basic B cell differentiation status can be assessed using expression of the immunoglobulin subclasses IgM, IgD, IgG and the sialoglycoprotein CD24 (Figure 1C), as well as CD27 and CD38 (11).

Pan T cells are identified by expression of CD3 (Figure 1A), followed by subsetting γδ T cells from the larger fraction of αβ T cells. MAIT cells are gated using CD161 and the invariant TCR alpha chain, TCR Vα7.2 (Figure 1D). Invariant NK-T cells can be identified using an antibody against the Vα24 and Jα18 of the TCR alpha locus. For conventional αβ T cells, after gating CD4⁺ and CD8⁺ T cells, Tregs can be identified by high expression of the IL-2 receptor alpha chain (CD25) and low expression of the IL-7 receptor (CD127) (Figure 1D). In the CD4⁺ Th cell and CD8⁺ cytotoxic T cell fraction, naïve and memory subsets can be identified by the differential expression of CD45RA, CD45RO and the chemokine receptor CCR7 (CD197) (12). Furthermore, tissue-resident memory T cells (TRMs), can be identified by the expression of CD69 and the integrin CD103 (13) (Figure 1E). CD69 also functions as a marker of recent T cell activation.

For detailed assessment of Treg function and phenotype, the ectonucleoside triphosphate diphosphohydrolase CD39 and inducible costimulator of T cells (ICOS) can be used, together with the chemokine receptors CXCR3 and CCR4 (Figure 1F) (14).

Both CD4⁺ helper T cells and CD8⁺ cytotoxic T cells can differentiate into different effector and memory lineages, but they can also enter an exhausted state during chronic infections or cancer development (15). In our panel, several phenotyping markers allow us to assess the functional state of T cells in depth: the exhaustion and activation marker Programmed Death 1 (PD-1, CD279), the senescence marker CD57, the co-inhibitory receptor BTLA (CD274), the co-receptors CD27 and CD28, the cyclic ADP ribose hydrolase CD38, TIGIT (T cell immunoreceptor with Ig and ITIM domains) and killer cell lectin-like receptor subfamily G member 1 (KLRG1). Figure 1G depicts well-defined separation for all these markers across multiple T cell subsets, and NK cells as a reference population. Furthermore, for CD4⁺ Th cells the two main effector lineages can be distinguished using the expression of the chemokine receptors CXCR3 (mostly expressed on Th1 cells) and CCR4 (mostly expressed on Th2 cells). These chemokine receptors can also be used to study different functional capacity and homing properties of CD8⁺ cytotoxic T cells and Tregs (Figure 1F).

NK cells are gated using the lineage-defining molecules CD56 (NCAM) and CD161 (KLRB1) and can be subsetted by the expression of CD16 (Figure 1H). Nkp46 functions as an additional NK cell marker that is suitable for studying tissue-derived NK cells (16).

In the myeloid cell compartment, CD14 (the LPS receptor) and CD16 (FcγRIII) are used traditionally to distinguish classical (CD14⁺ CD16⁺), intermediate (CD14⁺ CD16^dim) and non-classical (CD14⁻CD16⁺) monocytes (17) (Figure 1A). FcER1 together with the IL-3 receptor alpha chain (CD123) are commonly used markers for basophils. Within the Lin-(CD3⁻CD19⁻CD56⁻CD14⁻CD16⁻) HLA-DR⁺ compartment, our panel identifies plasmacytoid dendritic cells (pDCs) by the expression of CD303, while classical DCs (cDCs) are marked as CD11c⁺ HLA-DR⁺. The cross-presenting cDC1 and the cDC2 subsets are identified by CD141 (also known as BDCA-3) and CD1c (BDCA-1) expression (18), respectively, with FcER1 functioning as an additional and more distinct marker for the cDC2 lineage (9) (Figure 1I). DC activation status can be assessed by the expression of the co-receptors CD40, CD86 and the Integrin alpha M (CD11b). Furthermore, CD163 allows the separation of the recently defined DC3 subset (8,19) and to phenotype monocytes (Figure 1J). Finally, ILCs can be identified by co-expression of CD127 and CD2 (Figure 1K).

A representative gating tree with all the above-mentioned subsets is shown in Figure 1, including some fluorescence-minus-one (FMO) controls: on CD8⁺ T cells for the molecules PD-1, BTLA; on CD25⁺ CD127⁻ Tregs for ICOS, CCR4; and on pan CD11c⁺ MHCII⁺ cDCs for CD86 and CD1c (Figure 1L). All of these FMOs highlight that by using systematic panel design there is negligible spreading error (SE) (20) for these populations and markers of high interest.

Our panel development strategy was based on established best practices (21-23) and multiple novel approaches. First, we utilized the similarity index (Online Figure 3), the fluorochrome brightness (Online Figure 4) and a newly developed automated algorithm for systematic fluorochrome selection. While previously described metrics such as the complexity index (24) assign an overall “score” to a given set of fluorochromes, our strategy allowed us to identify the best feasible combinations of fluorochromes to move beyond 40 colors. Second, we developed a new metric to evaluate unmixing-dependent spreading error that occurs in highly complex spectral flow cytometry panels and affects all events in the measurement (Online Figure 5, and Mage and Mair, manuscript in preparation). Finally, we utilized the instrument-specific spillover-spreading matrix (SSM) (20) and the total spread matrix (TSM) (Corselli et al, manuscript in preparation) for the optimal assignment of fluorochromes based on the biological co-expression of markers (Online Figure 6).

All steps of this panel design process, including the novel approaches, are described in detail in the online material of this manuscript, including additional staining and gating controls (Online Figure 9). Of note, this panel was developed on two full spectrum cytometers in parallel: a 7-laser instrument with a total of 186 detectors (commercially available from Sony Biotechnology as the Sony ID7000^™) and a 5-laser instrument with a total of 78 detectors (commercially available from BD Biosciences as the BD FACSDiscover^™ S8). The final and fully optimized panel as shown in Figure 1 was acquired on the BD FACSDiscover^™ S8 (instrument configuration and setup details are listed in Online Figure 1 and 2), together with the FMO control samples. This instrument also allowed cell sorting, highlighting that 50-parameter sorting is feasible to allow very fine-grained isolation of any immune population of interest (sorting strategy and purity of the sorted populations shown in Online Figure 14). Furthermore, we established a trimmed-down panel version of 45 colors that is cross-platform compatible on the BD FACSDiscover^™ S8 and Sony ID7000^™ (Online Figure 12). To the best of our knowledge, this is the first report of a high-dimensional 40-color+ panel that is usable across multiple independent spectral cytometry platforms.

Manual analysis is critical to assess data quality and separation of markers in any flow cytometry panel (25), but in many cases additional insights can be gleaned from high-dimensional panels by using exploratory computational tools such as dimensionality reduction and clustering (26-28). In Online Figure 15 we show a UMAP plot with overlaid FlowSOM (29) clusters, heatmaps of the annotated clusters and histogram overlays of the raw data for some markers, highlighting the fine-grained cellular heterogeneity that can be revealed using our 50-color panel.

Overall, our data shows that this panel can serve as a widely usable and powerful immunophenotyping resource for comprehensive analysis of human immune cells in peripheral blood and non-lymphoid tissues, including human tissues of small size such as biopsies or some resected tissues (3,5) .

The opportunity to reliably analyze 50 different target molecules (with the option to perform parallel cell sorting) in a high-throughput fashion is likely to enable previously impossible avenues to study the human immune system (30). Finally, development of such a comprehensive panel may enable the consistent use of a single panel suitable for multiple different studies. Together with deposition of these data into publicly accessible databases, such a consistent use would facilitate subsequent cross-study analyses with machine learning approaches such as FAUST (31,32) or other suitable computational techniques.

Similarity to published OMIPs:

The most similar OMIP to our manuscript is OMIP-069 (the first 40-color OMIP to be reported, (24)) and OMIP-044 (the first 28-color OMIP reported (33)). There is some overlap with published 28-color OMIPs focusing on T cell phenotyping (e.g. OMIP-050 and OMIP-058 (34)) and several other lower dimensional OMIPs focused on T cells, but up to date there is no OMIP that reports the use of 50 different fluorochromes allowing such in-depth phenotyping of T cells and antigen-presenting cells (APCs).

Table 1.

Summary table for application of OMIP XX

Purpose	50-color phenotyping of antigen-presenting cells and T cells
Species	Human
Cell Types	PBMCs and human tissue
Cross-References	OMIP-069, OMIP-044, OMIP-050, OMIP-058 and others