Single Cell Solutions

Chromium™
Single Cell Solutions
Introduction
Advances in single cell RNA quantification techniques have
enabled comprehensive study of subpopulations of cells
within a heterogeneous population. We developed the
GemCodeTM Platform, which combines microfluidics with
molecular barcoding and custom bioinformatics software to
enable 3’ mRNA counting from thousands of single cells.
High-throughput, single cell expression
measurements enable discovery of
gene expression dynamics for profiling
individual cell types.
• Complete practical solution for single cell analysis
The GemCode™ Single Cell Platform
• Identify rare cell types in heterogeneous populations
Single cells, reagents and a single gel bead containing
barcoded oligonucleotides are encapsulated into nanolitersized GEMs using the GemCode Platform. Lysis and barcoded
reverse transcription of RNAs from single cells are performed
inside each GEM. High quality next generation sequencing
libraries are finished in a single bulk reaction. Finally, the
GemCode software suite is utilized for processing, analysis
and visualization of single cell gene expression data.
• More efficient than leading academic droplet systems
• Encapsulate up to 48,000 cells in 10 minutes
• Wide dynamic range
Figure 1
a.
Barcoded
cDNA
8-Channel Microfluidics Chip
x8
Oil
Cells + Reagents
Break
Amplify
Emusion cDNA
RT
Beads
Construct
Sequence
Library
GEM Outlet
b.
c.
10x
Barcodes
UMI (T)30VN
poly(A)
d.
Single Cell
GEMs
f.
g.
Median UMIs Per Cell (k)
e.
Oil
Median Genes Per Cell (k)
Cells
Reagents
10x
Barcodes
UMI (T)30VN
1010 cells recovered
Multiplet rate=1.5%
40
20
0
0
10x
i.
20
40
60
80
20
3
1
™
Human
Counts (k)
Genomics
| UMI
Chromium
Single
293T-only
10
2
25
50
75 100
Raw
Reads Per
Cell (k)
Cell
Solutions
Application
Note
j.
Jurkat-only
ts
Mouse UMI Counts (k)
Mouse-only
Figure 1. GemCode
single cell platform. (a) Formation of GEMs, RT takes place inside each GEM, whichMouse
is then (3T3)
pooled for cDNA
Human:Mouse
Mouse (3T3)
amplification and Human-only
library construction in bulk. (b) Formation of single-cell GEMs. (c) Barcoded oligonucleotides
Human contained
(293T) inside
Human (293T)
60
GEMs. (d) Final library molecules.
4
150
25
50
75
100
Raw Reads Per Cell (k)
293T-only
293T-only
cDNA
Insert
h.
Mean UMI Counts (log10)
Collect
Barcoded Primer
Gel Beads
cDNA
Sample
Index
3
2
r=0.96
1
0
0
1
2
3
4
ERCC molecules per GEM (log10)1
150
Jurkat-only
Technical Performance of GemCode
Platform for Single Cell Analysis
GemCode Platform Reveals
Major Subpopulations Among a
Heterogeneous PBMC Sample
We analyzed cell lines, peripheral blood mononuclear cells
(PBMCs), and bone marrow mononuclear cells (BMMCs) to
evaluate the technical performance of the GemCode platform
for single cell analysis. 1010 cells were captured, of which
483 were human and 535 were mouse, indicating a ~50% cell
capture rate with an inferred multiplet rate of 1.5%. ~4,500
genes and 27,000 transcripts were detected in each human
and
mouse cell, indicating comparable sensitivity to previous
a.
droplet platforms. cDNA conversion rate showed an efficiency
of ~10%. We tested the sensitivity of the system by mixing 2
a.
cell
types at various ratios. 1:1 mixed
detected an equal
b.
c. sample
10000
split of 293T and Jurkat cells. We were able to detect the rare
d.
1% of 293T cells when they were mixed
with Jurkat cells.
Clustering analysis was performed to dissect the heterogeneity
of PBMCs. Examination of the most variable genes in each
cluster revealed many well-characterized markers for specific
subpopulations of PBMCs. We scored ~68,000 PBMCs against
the average expression profile of 10 bead-enriched purified
PBMC subpopulations, and classified each cell based on its
similarity to a purified population. Cell classification was mostly
consistent with cell-marker based classification analysis.
Figure 1
Barcoded
cDNA
8-Channel Microfluidics Chip
x8
Oil
Cells + Reagents
Amplify
Break
Emusion cDNA
RT
Beads
Construct
Sequence
Library
Figure 2
GEM Outlet
10x
Barcodes
b.
a.
UMI (T)30VN
poly(A)
6
a.
cDNA
2
3
10000
7
b.
Collect
100
50
150
293T:Jurkat
(1:1)
10000
CD14+
Monocytes
50
0
0
50
100
150
293T:Jurkat (1:99)
Clusters 2 1 3 4 5 9 8 6 7
LYPD2
b.
CD14
Dendritic
100
LGALS2
50
B and Dendritic
50
CLEC4C
MZB1
0
2000
Megakaryocytes
0
100
3396 cells recovered
Multiplet rate=3.1%
0
Principal Component 1
50
100
150
JurkatCytotoxic
SNV Counts
1000
0
50
100
150
Jurkat SNV Counts
-1
NKG2
NK
GZMK
LYPD2
550
CD14
CD14+ Monocytes
Dendritic
0.1
B and Dendritic
10.0
0.1
1.0
Fresh PBMC UMI Counts (log10)
Megakaryocytes
LGALS2
500
CLEC4C
MZB1
PF4
Cytotoxic
CD8A
NK
NKG2
e.
Frozen PBMC UMI Counts (log10)
Median Genes Per Cell (10k reads/cell)
B
T
1.0
CD16+ Monocytes
r=0.97
10.0
1.0
CD34+
CD14+
CLEC4C
4 Monocytes
MZB1
2.5
PF4 T Reg
CD4+/CD25+
CD4+/CD45RO+ T Memory
CD4+ TCD8A
Helper2
tSNE1
NKG2
8
g.
5
tSNE1
100
CD34+
1300
-1
Normalized
Expression
f. 700
g.
CD8+
Cytotoxic T 650
CD56+r=0.97
NK
1500 Dendritic
10.0
Fresh
Frozen
CD45 RA+
Naive T
50
CD14+
600
Monocytes
25
1.0
0
CD19+
550
B Cells
1700
1
1500
1300
2.5
T
CD34+
B
CD14+
0.1
Dendritic 500 Monocytes
Fresh Frozen CD45
RA+ Fresh Frozen
Monocytes
Naive
10.0
0.1T
1.0
Fresh PBMC UMI Counts (log10)
Fresh Frozen
CD4+/CD25+ T Reg
CD4+/CD45RO+ T Memory
CD4+ T Helper2
-1
Normalized
Expression
600
550
0.1 Cell Solutions Application
500
10x Genomics | Chromium™ Single
Note
10.0
1.0
Fresh PBMC UMI Counts (log10)
LGALS2
NK
d.
c.
Figure 2
LYPD2CD19+
B Cells
7
CD14
3
Cytotoxic
e.
Dendritic
Fresh Frozen
tSNE1
f. 700
g.
Figure 3. Analysis of 68,000 fresh PBMCs. (a) tSNE Fresh
plot of 68,000 PBMCs. (b) Top variable
1700
Frozen
genes from each of 9 clusters are normalized and presented
in a heat map. Representative
100
650
markers from each cluster are shown on the right, and the putative cluster ID is shown on
the left. (c) tSNE plot of 68,000 PBMCs, with each cell colored by the cell type of purified
PBMCs. 20,000 reads/cell in this experiment.
Median Genes Per Cell (10k reads/cell)
Frozen PBMC UMI Counts (log10)
CD79A
2
Dendritic
9 RA+
CD45
B Naive
and Dendritic
T
1
Megakaryocytes
1700
CD56+ NK
CD14+
CD79A
Monocytes
GZMK
CD14+ Monocytes
Normalized
Figure 2. Technical performance of 10x Genomics GemCode single cell platform.
Expression
(a) Scatter plot of human and mouse counts detected in a mixture of 293T and 3T3 cells.
Median genes (b) and counts (c) detected per celle.
in a mixture of 293T (cyan) and 3T3 (red)
f. 700
cells at different raw reads per cell. (d) Mean observed count
0 for each ERCC molecule is
Genes
compared to expected number of ERCC molecules per GEM. (e) Principal
component (PC) UMIs
analysis on normalized scRNA-seq data of Jurkat and 293T cells mixed atr=0.97
4 different ratios
650
(100% 293Ts, 100% Jurkat, 1:1 293T:Jurkat and c.
1:99 293T and Jurkat). (f) Scatterplot of 293T
10.0
and Jurkat SNV counts observed in each cell for Jurkat and 293T cells mixed at four ratios.
Clusters 2 1 3 4 5 9 8 6 7
~100,000 reads/cell for human/mouse, 250,000 for ERCC, and 30,000 for 293T and Jurkat.
600
0.1
CD16+ Monocytes
6
2.5
CD8A
B
T
CD34+
PF4
d.
Dendritic
GZMK
150
5
UMIs
CD8+
Cytotoxic T
CD56+ NK
c.
b.
CD79A
Genes
Proportion of Cells
100
tSNE1
0
d.
Median UMIs Per Cell (10k reads/cell)
Jurkat-only
tSNE2
0
tSNE2
8
Clusters 2 1 3 4 5 9 8 6 7
0
1
5
0
1
2
3
4
Genes
ERCC molecules per GEM (log10UMIs
)
150
B
T
Counts
tSNE2
Mean Counts
UMI Counts
Mean
(log10(log
) 10)
Counts
Jurkat-only
50
150
a.
-1
CD3D
100
1000
0
293T-only
293T:Jurkat
9
Median Genes Per Cell (10k reads/cell)
293T:Jurkat (1:99)
c.
CD16+ Monocytes
0
293T:Jurkat (1:1)
293T-only
150
1
1
2000
Proportion of Cells
1
f.j.
2
2
4
Fresh Frozen
Proportion of Cells
Jurkat-only
25
50 0 75
100
Raw Reads Per Cell (k)
r=0.96
tSNE2
25
50
75 100
Raw Reads Per Cell (k)
3
tSNE2
80
1
Human
(k)(k)
Human
UMICounts
Counts
293T-only
Principal Component 3
60
10
2
6
9
Sample
Index
Median UMIs Per Cell (10k reads/cell)
40
h.
d.
1000
20
3
cDNA
Insert
tSNE2
Frozen PBMC UMI Counts (log10)
e.i.
20
UMI (T)30VN
Median UMIs Per Cell (10k reads/cell)
0
10x
Barcodes
2000
Mouse (3T3)
Human (293T)
Counts
0
4
Median
MedianCounts
UMIs Per Cell
Cell(k)
(k)
20
g.
c.
Mouse (3T3)
Human (293T)
293T SNV Counts
1010 cells recovered
Multiplet rate=1.5%
40
Single Cell
GEMs
f.
b.
Mouse-only
Human:Mouse
Human-only
60
Oil
293T SNV Counts
Mouse Counts
UMI Counts
Mouse
(k) (k)
e.
a.
Cells
Reagents
Median Genes Per Cell (k)
Barcoded Primer
Gel Beads
1500
1300
50
25
2
0
Fresh Frozen
CD34+
Dendritic
B
Monocytes
T
Comparison of Specific
Subpopulations in AML PBMCs
CLL and AML BMMCs Show Expansion
of Distinct Populations
Single cell profiling enables comparison of specific subpopulations in frozen PBMC samples from healthy controls
and patients with acute myeloid leukemia (AML). This analysis
revealed misregulation of the FLT3 pathway that would have
been missed by bulk RNA-seq.
We performed single cell analysis of frozen BMMCs from
healthy controls, chronic lymphocytic leukemia (CLL) and
AML patients. We observed a proliferation of B cells in the
CLL sample, and a proliferation of myeloid progenitors in the
AML sample, which is consistent with the disease pathology.
Healthy individual
CLL patient
AML patient
Significant gene sets
14
Figure 5. Single cell profiling from healthy and malignant tumor cell samples.
Single cell profiling of BMMCs from healthy, CLL and AML patients. ~30,000 reads/cell
in this experiment.
Figure 4. Single cell profiling from healthy and malignant tumor cell samples. Selection of
myeloid populations in normal and AML PBMCs. Bottom graphs show an overlap of
significant gene sets between bulk RNA-seq and myeloid-cell specific comparisons.
Whereas the bulk comparison revealed expected pathways, such as upregulation of stem
cell genes, the myeloid-specific comparison revealed upregulation of the FLT3 pathway.
Conclusion
We performed high-throughput gene expression profiling
of mRNAs in single cells using the Chromium Single Cell
Solution. Our scalable approach enables detection of rare
cells in a heterogeneous tumor population. Moreover, efficient
cell capture enables analysis of clinically relevant sample
types with limited cell input.
Literature Cited: Zheng et al. 2016. Haplotyping germline and cancer genomes with
high-throughput linked-read sequencing. Nat Biotechnol 34(3):303-11.
Learn more at 10xGenomics.com | Email us at info@10xgenomics.com
10x Genomics | Chromium™ Single Cell Solutions Application Note