WHITE PAPER
A Complete Workflow for Single Cell Gene Expression Profiling
A Complete Workflow for Single Cell Gene Expression Profiling
Lucas Dennis a, Janelle Crane b, Xiaoyang Wang b
a
NanoString Technologies, Seattle WA 98109 | bBecton Dickinson Biosciences, San Jose CA 95131
Introduction
Certain populations of cells, once thought homogeneous, have recently
revealed a startling amount of gene expression variation. New understanding
has been driven largely by recent advances in the sensitivity of gene
expression detection technologies, which are now capable of analysis at
the individual cell level. These new tools enable researchers to characterize
cell heterogeneity by elucidating previously hidden relationships between
individual cells within a population.
To capitalize on the wealth of expression information contained within a single
cell, a robust process must be employed to identify, isolate and profile an
individual cell. Such a process would ideally ensure cell viability during sorting
and be fast, scalable, precise, and flexible, as well as allow for expression
information to be accurately obtained from a large number of genes or
gene pathways. Here, we describe a complete workflow for single cell gene
expression using indexed fluorescence-activated cell sorting (FACS) on the
BD FACSJazz™ cell sorter (BD Biosciences) in combination with nCounter®
Single Cell Gene Expression Assays (NanoString Technologies, Inc.) that
meets all of these criteria (FIGURE 1).
FACS is a powerful technology capable of delivering individual cells for
downstream molecular analyses quickly and accurately. It can identify,
characterize, and isolate individual cells based on cell light scattering
information and specific cell surface or intracellular marker fluorescence
staining intensity information1. The BD FACSJazz cell sorter is a benchtop
instrument designed for individual laboratories that are new to flow cytometry.
With up to six-color/eight-parameter detection, and index sorting, the BD
FACSJazz offers robust performance for genomics researchers. The BD
FACSJazz also offers the flexibility to deposit cells quickly and precisely into
a variety of collection devices, including 96- and 384-well plates, tubes, or
custom-made devices. By tracking and documenting sorted cell information
as positions in a collection device, index sorting enables researchers to
correlate cellular phenotypes with subsequent molecular analyses, such as
gene expression profiles.
The nCounter Single Cell Gene Expression Assay utilizes NanoString’s
optical-barcode technology to enable gene expression analysis of up to 800
genes with a simple and highly automated workflow. The ability to analyze
up to 800 genes in a single tube eliminates the need for sample partitioning,
minimizes the number of amplification cycles required, and allows researchers
to assay entire molecular pathways or groups of pathways. The digital readout provides highly reproducible results and allows for easy comparison with
other digital technologies such as next-generation sequencing.
To demonstrate the capability and complementarity of the BD FACSJazz and
nCounter Analysis System, we performed a blind experiment by sorting single
cells from a mixed cell population followed by nCounter gene expression
analysis. Cellular identity was determined independently by FACS and gene
expression analysis using the BD FACSJazz and nCounter, respectively. Our
results showed a 100% correlation in cell identity based on gene expression
data and on flow cytometry phenotyping. These results described below
validate the combined BD FACSJazz and nCounter Analysis system workflow
as a powerful method to obtain gene expression profiles from rare cells within
mixed cell populations.
Complete Workflow
* TaqMan® Preamp Master Mix + Pooled MTE Primers
** Reporter CodeSet, Buffer, Capture ProbeSet
FIGURE 1: Complete single cell workflow using the BD FACSJazz and nCounter Analysis System. A population of cells is analyzed and sorted based on size and cell surface markers
to obtain specific individual cells. Each selected cell is placed into a well of a 96-well plate and lysed. After cDNA conversion and target-specific enrichment, each sample is hybridized
with a CodeSet containing target-specific Capture and Reporter probes. After a simple overnight hybridization, cDNA/target probe complexes are purified and counted by the nCounter
Analysis System.
Molecules That Count®
Translational Research – Gene Expression –
Single Cell
–
CNV
1
WHITE PAPER
A Complete Workflow for Single Cell Gene Expression Profiling
Cell Staining and Individual Cell Sorting on the BD FACSJazz
Before sorting, the BD FACSJazz was cleaned with ethanol and DEPC treated phosphate buffered
saline.x1000
The fluidics system settings were optimized for sorting
x1000one cell into the center of each well
of a 96-well
plate
(FIGURE
2).
To
verify
the
precise
positioning
60in the center of well, one fluorescent
60
bead (Rainbow Fluorescent Particles, Spherotech) was sorted into each well of a 96 well plate and
50
50
imaged on an imaging system (BD Pathway™ 855) prior to sorting the cells2.
x1000
60
50
40
40
40
10
Single cells
were selected (“gated”) based on parameters 10
that describe their relative size and
structure,0 forward scatter (FSC) and side scatter (SSC) (FIGURES
3C–E). Cells from each cell line
0
10
10
10
10
10 unique
10 markers (FIGURES
10
were subsequently
gated
based
on their
3F–H)10so that10identity could
be 10
CD3
FITC
1
CD19
APC
tracked by the index sorting feature of BD FACS™ Sortware software .
0
1
2
3
0
4
1
2
3
Raji
SSC
SSC
SSC
Three cell lines, Raji, Jurkat, and HEL were separately stained with their specific cell surface
markers:30CD19 APC for Raji, CD3 FITC for Jurkat , and CD235a30PE for HEL, respectively [all reagents
were from BD Biosciences] (FIGURES 3A). After staining, cells were mixed in a ratio of 80:15:5
20
20
(Raji:Jurkat:HEL).
The mixed cells were analyzed and sorted on a BD FACSJazz sorter (FIGURES 3B).
20
10
4
FIGURE 2: Fluidics system settings optimized for sorting one cell
into the 0center of each well of a 96-well plate. Prior to sorting
101
102
103
104
100
beads (red arrow) are sorted in order to verify positioning of
CD235a PE
subsequently sorted cells. A representative plate for well position
accuracy is shown. In the experiment described here, a 96-well PCR
plate was used.
CD19+
80%
CD235–
Sorting Statistics
CD3–
Populations
Jurkat
Events
% Total
96
100.00%
78
81.25%
SSC Singlets
CD19–
15%
CD235+
CD3–
HEL
CD19–
5%
CD235–
CD3+
x1000
x1000
60
20
Raji
Jurkat
8
8.33%
HEL
5
5.21%
50
10000
FSC Width
30
20
SSC Width
15
40
SSC
30
10
5000
5
10
0
0
0
10
20
30
40
50
FSC
60
x1000
0
10
20
30
40
50
FSC
60
x1000
0
50
50
50
40
40
40
SSC
60
SSC
x1000
60
SSC
x1000
60
30
20
20
10
10
10
0
100
101
102
CD19 APC
103
104
30
40
50
60
x1000
30
20
0
20
SSC
x1000
30
10
0
100
101
102
CD3 FITC
103
104
100
101
102
3 4
1010
CD235a PE
FIGURE 3: FACS analysis of a mixed population. A) Raji, Jurkat, and HEL cells are mixed at 80:15:5 ratios. B) Gating hierarchy showing how each cell population is derived and proportion
of each population after sorting. C) Live cells were gated via forward and side scatter. D–E) Individual FSC and SSC gates were used to eliminate doublets and ensure that only single
cells were collected. F) Raji cells were gated based on positive CD19 staining. G) Jurkat cells were gated based on positive CD3 staining. H) HEL cells were gated based on positive
CD235a staining.
2
NanoString® Technologies, Inc.
WHITE PAPER
NanoString® Technologies, Inc.
Prior to sorting, each well of a 96-well polypropylene PCR plate (GeneMate,
BioExpress) was pre-loaded with 3 µL of a chemical lysis solution (Cells-to-Ct™
lysis buffer, Life Technologies). A BD FACSJazz sorter was used to deposit
single cells from the mixed population (1 cell per well from the SSC Singlets
gate, FIGURE 3E) directly into the 96-well plate. Cells were transported at
low pressure at the speed of 500 events per second through a large nozzle
(100 micron diameter), minimizing damage to live cells and helping to ensure
cell viability3. Dispensing 96 single cells took less than two minutes. FIGURE
4 shows each sorted cell within the population of analyzed cells and the
location of these cells on the 96-well plate. The 96-well plates containing the
single cells were immediately frozen at -80°C, and then shipped to NanoString
on dry ice.
60
50
50
50
40
40
40
30
SSC
x1000
60
SSC
x1000
60
SSC
x1000
30
30
20
20
20
10
10
10
0
0
100
101
102
CD19 APC
103
104
0
100
101
102
103
104
100
CD3 FITC
101
102
103
104
CD235a PE
FIGURE Raji
4: Sorting of individual cells into a 96-well plate on the BD FACSJazz. Indexed sorting enables tracking of cellular identity and position on a collection device after a sort.
CD19+
A) Sorted Raji cells
[orange]. B) Sorted HEL cells [green]. C) Sorted Jurkat cells [blue]. D) Location of index-sorted single cells on a 96-well plate.
80%
CD235–
CD3–
Jurkat Transcription and Multiplexed Target Enrichment
Reverse
CD19–
CD235+
15%sample identities
The
were blinded and cells were prepared for nCounter gene expression analysis via Multiplexed Target Enrichment (MTE) following steps
CD3–
outlined in the nCounter Single Cell Expression Assay Protocol4. MTE is a two-step process in which input RNA is converted to cDNA and then amplified with targetspecificHEL
primers. MTE primers were pooled at a final concentration of 500 nM per oligo in Tris-EDTA (pH 7.5) to ensure a working concentration of 50 nM per oligo
CD19–
in 5%
the MTE. The
entire cell lysate (3 μL) was used as input for reverse transcription with the SuperScript® VILO™ Kit. After cDNA conversion, a highly multiplexed
CD235–
CD3+
enrichment of target sequences was performed using TaqMan® PreAmp Master Mix and the MTE primer pool using 18 cycles of amplification, resulting in 11 μL of
amplified cDNA per cell.
nCounter CodeSet Hybridization and Analysis
Amplified cDNA was hybridized with an nCounter CodeSet
containing a panel of probes for 526 genes. After an overnight
incubation at 65°C, cDNA/probe complexes were purified and
deposited on an imaging surface by the automated nCounter
Prep Station. Individual cDNA/probe complexes were counted
and tabulated by the nCounter Digital Analyzer, resulting in a
digital read-out of expression quantity for each target gene.
Raw data was imported into nSolver™ analysis software,
normalized for hybridization efficiency using internal positive
controls, and exported for downstream analysis.
Unsupervised clustering of gene expression counts was used
to identify subsets of cells based on gene expression patterns.
To achieve this, normalized data was filtered by excluding
genes and samples with fewer than 50 counts for greater than
95% of probes; resulting in the exclusion of genes that were
unexpressed and samples with insufficient amplification. This
resulted in a final data set of 272 probes across 92 queried
samples. Count data was then log converted and clustered,
identifying three distinct cell populations (groups A, B, and C)
with 76 cells comprising group A, 9 cells comprising group B,
and 7 cells comprising group C (FIGURE 5). Interestingly, within
the largest group of cells, A, two distinct subpopulations were
observed, highlighting the value of examining expression at
the single cell level.
A Complete Workflow for Single Cell Gene Expression Profiling
FIGURE 5: Cluster analysis of the 272 gene expression profiles (y-axis) for each cell (x-axis) identified
three distinct clusters. Expression is shown on a blue-yellow axis with blue representing low expression (log
transformed counts) and yellow high expression (log transformed counts).
3
NanoString® Technologies, Inc.
FACS and Gene Expression Data Comparison
FACS
When well positions associated with cell identity of the three index sorted
cell lines were compared with well positions associated with the three cell
populations defined by gene expression, there was complete agreement
between the FACS analysis and the gene expression profile (see TABLE 1).
The number of cells from each cell line observed after sorting was within the
expected range (see TABLE 2) based on sampling statistics. In this experiment,
both FACS and gene expression failed to identify a small percentage of cells.
However, when information from both types of analysis was considered, 100%
of cells could be identified, demonstrating the complementarity of examining
expression at both the protein/cell surface marker and RNA levels using the
BD FACSJazz and nCounter Analysis System.
nCounter
A Complete Workflow for Single Cell Gene Expression Profiling
A
B
C
No Call
Raji
Jurkat
HEL
No Call
76
0
0
2
0
7
0
1
0
0
4
1
0
2
3
0
TABLE 1: Comparison of cells observed by FACS and gene expression analysis.
Observed
Expected
Raji
Jurkat
HEL
78
77 ±8
10
14 ±7
8
5 ±4
TABLE 2: Comparison of cells observed vs. expected based on sampling from the
population mixture (80% Raji cells, 15% Jurkat cells, and 5% HEL cells).
A Complete Single Cell Workflow
This complete end-to-end workflow allows for the rapid isolation of individual cells from a heterogeneous cell population and permits quantification of expression
for up to 800 transcripts from each cell, making it ideally suited to support the study of pathway-based biology at the single cell level. FACS captures cells quickly
and gently, so that highly accurate gene expression profiles of healthy living cells are obtained. The speed of single cell sorting by FACS also means that a large
population of cells can be processed easily and efficiently. In cases where cell surface or internal markers can be used to discriminate cells of interest, rapid
segmentation of cell types using FACS permits experimental design to be more targeted, reducing the cost and complexity of experiments and helping to simplify
interpretation of the data. We show that individual cells that are 5% of the overall cellular population can be readily distinguished by the methods described here,
providing a valuable tool for researchers interesting in characterizing rare cell populations.
The results also demonstrate the ability of gene expression data to be used to identify markers of individual cell subpopulations, which can be used to further refine
sorting parameters for subsequent experiments. Finally, the data indicate the potential for identifying subpopulations with high resolution from within seemingly
homogenous populations of cells using highly multiplexed gene expression analysis. With the ability to measure up to 800 genes in an individual cell, this complete
workflow permits researchers to assay expression of entire molecular pathways or groups of pathways within a single cell allowing for an unprecedented view of
biology.
References
1. Technical Bulletin: BD Flow Cytometry. 23-10236-00, Nov 2008.
2.Single cell sorting and cloning. F. L. Battye et al., Journal of Immunological Methods, 243:25-32, 2000.
3.Chapter 4: Isolation of Breast Cancer Stem Cells by Single-Cell Sorting, in Biomedical Tissue Culture. P. L. Pham et al., pp. 59-72, 2012.
4.nCounter Single Cell Expression Assay Protocol. MAN-C0027.
NanoString Technologies, Inc.
530 Fairview Ave N
Suite 2000
Seattle, WA 98109 USA
Toll-free: 1 888 358 6266 | Fax: 206 378 6288
www.nanostring.com | [email protected]
LEARN MORE
To learn more about the nCounter® Single Cell Gene Expression
Assay, visit us at www.nanostring.com/singlecell or contact
your local NanoString Technologies representative.
Cells-to-CT and VILO are trademarks, and SuperScript and TaqMan are registered trademarks, of Life Technologies
Corporation.
BD, BD Logo and all other trademarks are property of Becton, Dickinson and Company.
© 2014 NanoString Technologies, Inc. All rights reserved. NanoString®, NanoString Technologies®, nCounter®, Molecules
That Count®, and nDesign™ are registered trademarks or trademarks of NanoString Technologies, Inc., (“NanoString”) in the
United States and/or other countries. BD FACSJazz is a trademark of BD Biosciences.
FOR RESEARCH USE ONLY. Not for use in diagnostic procedures.
WP-SC020-01
February 2014