Analysis of Mass Cytometry Data with viSNE Reveals

AAI 2014 #69.22
Analysis of Mass Cytometry Data with viSNE Reveals Phenotypic Heterogeneity within Human Tregs
Mark A.P. Konrad1, Erin F. Simonds2, Sean C. Bendall3, Tad C. George1, Tiffany J. Chen4, Pam Delucchi1
1Fluidigm
Sciences Inc., Sunnyvale, CA; 2Department of Neurology, University of California, San Francisco, San Francisco, CA; 3Department of Pathology, Stanford University, Palo Alto, CA; 4Cytobank Inc., Mountain View, CA
Results
Introduction
Regulatory T (Treg) cells play fundamental roles in suppressing the immune response and in the development and maintenance of
immunological tolerance to self-antigens [1]. Historically, Tregs have been characterized by constitutive expression of the transcription
factor Foxp3, in addition to a surface phenotype of CD4+CD25hiCD127lo. While this pattern of expression has proven highly useful for
studying Tregs, the use of additional markers should provide greater resolution of minor subsets within the Treg compartment.
Mass cytometry is a powerful new platform that couples flow cytometry with mass spectrometry and enables single-cell analysis of at
least 30 parameters simultaneously without the requirement for compensation [2]. To profile Treg phenotypes in human peripheral
blood, we employed a panel of 24 metal-conjugated antibodies against both surface and intracellular targets, covering various aspects
of Treg function and biology. A gating scheme is presented that uses nine of the markers in the panel and enables the identification of
naïve, effector, terminal effector, and Helios- Treg subsets. Recently it has been suggested that Helios- Tregs represent the
peripherally, rather than thymically derived population of Tregs [3]. Analysis of the remaining panel markers in both unstimulated Tregs
and Tregs TCR-stimulated in the presence of exogenous IL-2 reveals up-regulation of several markers in a manner consistent with
previous reports.
Figure 1. Identification of Human Treg Subsets Using Conventional Cytometry Gating
Figure 3. Analysis of Human Tregs with viSNE
Human total Tregs, naïve Tregs, effector Tregs, terminal effector Tregs, and Helios- Tregs were gated according to reported Treg
subset phenotypes [1]. The unstimulated sample from donor 1 is shown, and all contour plots are gated on total viable singlet cell
events. The number of cells in the CD4+ T cell gate is approximately 41K.
CD3+CD4+Foxp3+ cells were gated, as displayed in the contour plots, from three unstimulated samples, then combined (8,262 total
cells) and visualized in t-SNE space with viSNE software [4]. The viSNE map in the top row with red, blue and green dots shows the
donor of origin of each cell in the analysis. Cells in the remaining viSNE maps are colored according to intensity of the indicated
marker. The location of naïve, effector, terminal effector, and Helios- Treg populations are indicated with white dotted circles.
Naïve Tregs
CD25
Donor
Terminal effector Tregs
viSNE is a new high-dimensional cytometry analysis tool based on the t-Distributed Stochastic Neighbor Embedding (t-SNE) algorithm
[4]. viSNE plots cells in two dimensions based on marker expression, placing related cells closer than unrelated cells in a biaxial plot.
viSNE analysis of PBMCs stained with the Treg panel confirms the identification of Treg subsets identified with conventional cytometry
analysis and reveals additional heterogeneity of marker expression.
Methods and Materials
Expression
CD4+ T Cells
Total Tregs
CD45RO+CD45RA- Tregs
CD45RA
CD45RO
HLA-DR
Helios
Helios- Tregs
Preparation of PBMCs: PBMCs were isolated from three healthy donors by Ficoll®-Paque gradient density centrifugation.
Unstimulated samples were stained immediately for mass cytometry and stimulated samples were treated for 18 hours with platebound anti-human CD3, soluble anti-human CD28, and recombinant human IL-2.
T Cell Activation: 12-well plates were coated with anti-human CD3 (clone OKT3) at a final concentration of 10 µg/mL for two hours
at 37 ˚C, then washed with PBS to remove unbound antibody. PBMCs were resuspended in complete RMPI-1640 Medium with
10% FBS and plated into wells at a final concentration of 3X106 cells/mL. After the addition of anti-human CD28 (clone 28.2) at a
final concentration of 5 µg/mL and recombinant human IL-2 at a final concentration of 100 U/mL, PBMCs were cultured for 18 hours
in a humidified incubator at 37 ˚C.
Cell Staining: Both unstimulated and stimulated PBMCs were stained with the viable cell indicator Cell-ID™ Cisplatin for 5 minutes
at a final concentration of 5 µM. Approximately 6 million cells/sample were then stained with surface markers as indicated in Table 1
for 30 minutes at room temperature. Following surface staining, cells were fixed and permeabilized with the eBioscience Foxp3
Staining Buffer Set, then stained with intracellular targets as indicated in Table 1 for 30 minutes at room temperature. Following cell
staining, cells were resuspended in MaxPar® Fix and Perm Buffer and labeled overnight with Cell-ID Intercalator-Ir to enable
identification of nucleated cell events by mass cytometry.
Effector Tregs
Effector Tregs
Naïve Tregs
Effector Tregs
CD4+Foxp3- T Cells
Helios- Tregs
CTLA-4
GITR
Terminal effector
Tregs
CD39
Ki67
Mass Cytometry: Following DNA intercalation, samples were prepared for mass cytometry analysis by washing twice with MaxPar
Cell Staining Buffer then once with MaxPar Water. Immediately prior to sample acquisition on the CyTOF® 2 mass cytometer, cells
were resuspended in 4 mL of EQ™ Four Element Calibration Beads diluted to 0.1X in MaxPar Water. Samples were filtered
through cell-strainer cap tubes and injected into the mass cytometer for acquisition of approximately 750K events. All of the
channels indicated in Table 1 were collected in addition to Pt195 for Cell-ID Cisplatin viability stain, Iridium 191 and 193 for
nucleated cell identification, and Ce140, Eu151 and 153, and Ho165 for data normalization.
Data Analysis: FCS files were normalized to the EQ Four Element Calibration Beads using the CyTOF software. For conventional
cytometric analysis of Treg populations, FCS files were imported into DVS® Cytobank [5]. Manually gated CD3+CD4+FoxP3+
singlet events were exported for further analysis by t-Distributed Stochastic Neighbor Embedding (t-SNE). A total of 8,262 events
(2,600-3,000 events per donor) were pooled and run in a single t-SNE analysis using default parameters and clustering on all
markers except IdU and Ki-67.
Figure 2. Marker Expression on Human Treg Subsets
The expression of panel markers is shown in heat maps of either unstimulated or stimulated samples on the Treg subsets gated above. As a
control, the expression in Foxp3- non-Treg CD4+ T cells is shown in the bottom row of the heat maps. The data displayed are the medians of
each channel, and the values corresponding to the heat map are shown in the tables below.
Conclusions
•
From the 24-marker human Treg panel used in this study, a core set of nine markers (CD3, CD4, CD25, Foxp3, CD45RA,
CD45RO, CD95, HLA-DR, and Helios) is sufficient to identify four main populations of Tregs from PBMCs in multiple
donors: naïve, effector, and terminal effector Tregs, in addition to Helios- Tregs, which have recently been proposed to be
the peripherally derived rather than the thymically derived Treg subset.
•
Additional markers provide valuable insight into the activation status, effector function, and proliferative capacity of Treg
subsets. Markers such as CD39 and CTLA-4, known to identify the most highly suppressive Tregs, display higher
expression in unstimulated effector and terminal effector Tregs than in naïve Tregs. Markers including GARP, GITR, OX40, and ICOS are upregulated in Tregs TCR-stimulated for 18 hours in the presence of exogenous IL-2, with the greatest
up-regulation occurring in the effector terminal Treg population, consistent with previous findings.
•
viSNE analysis of human Tregs clearly identifies naïve, effector, terminal effector, and Helios- populations of Tregs and will
be of significant use for further deep phenotyping of human Treg subsets.
Table 1.
Mass Cytometry Antibody Panel for Human Treg Analysis
Channel
141
145
149
150
151
153
154
156
158
159
160
162
164
165
166
167
168
169
170
171
172
174
175
176
Staining
Isotope
Marker
Function of Marker
Method
Pr
CD49d
Treg identification
surface
Nd
CD4
Treg identification
surface
Sm
CCR4
homing and origin
surface
Nd
LAG-3
activation and memory
surface
Eu
ICOS
suppressive and effector Tregs
surface
Eu
CD45RA
naïve Treg identification
surface
Sm
CD3
Treg identification
surface
Gd
GARP
apoptosis and survival
surface
Gd
OX40
apoptosis and survival
surface
Tb
GITR
apoptosis and survival
surface
Gd
CD39
suppressive and effector Tregs
surface
Dy
Foxp3
Treg identification
intracellular
Dy
CD95
apoptosis and survival
surface
Ho
CD45RO
effector Treg identification
surface
Er
Helios
thymically derived Treg identification intracellular
Er
CD27
apoptosis and survival
surface
Er
Ki67
cell proliferation
intracellular
Tm
CD25
Treg identification
surface
Er
CTLA-4
suppressive and effector Tregs
intracellular
Yb
Granzyme B
suppressive and effector Tregs
intracellular
Yb
suppressive and effector Tregs
surface
LAP/TGFβ
Yb
HLA-DR
terminal effector Treg identification
surface
Lu
PD-1
apoptosis and survival
surface
Yb
CD127
activation and memory
surface
Naïve Tregs
Naïve Tregs
Effector Tregs
Effector Tregs
Terminal effector Tregs
Terminal effector Tregs
Helios- Tregs
Helios- Tregs
CD4+Foxp3- T cells
CD4+Foxp3- T cells
Unstimulated
18 hours anti-human CD3,
anti-human CD28, and IL-2
References
1. Schmetterer et al. “Naturally occurring regulatory T cells: markers, mechanisms, and manipulation”. FASEB J 6 (2012);
2253-76.
2. Ornatsky et al. “Highly multiparametric analysis by mass cytometry”. Immunol Methods 361:1-2 (2010); 1-20. Review.
3. Thornton et al. “Expression of Helios, an Ikaros transcription factor family member, differentiates thymic-derived from
peripherally induced Foxp3+ T regulatory cells”. J Immunol 184:7 (2010); 3433-41.
Naïve Tregs
Effector Tregs
Terminal effector Tregs
4. Amir et al. “viSNE enables visualization of high dimensional single-cell data and reveals phenotypic heterogeneity of
leukemia”. Nat Biotechnol 31:6 (2013); 545-52.
5. Kotecha et al. “Web-based Analysis and Publication of Flow Cytometry Experiments”. Current Protocols in Cytometry 2010
Jul, Chapter 10, Unit10.17. PMID: 20578106.
Helios- Tregs
CD4+Foxp3- T cells
Fluidigm Corporation
7000 Shoreline Court, Suite 100 • South San Francisco, CA 94080 Toll-free: 1.866.359.4354 www.fluidigm.com/singlecellgenomics
We thank all members of the mass cytometry reagent
development team at Fluidigm for their contributions to this
study.