See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/259566573

Full-length RNA-seq from single cells using Smart-seq2

Article in Nature Protocols · January 2014

DOI: 10.1038/nprot.2014.006 · Source: PubMed

CITATIONS READS
3,479 21,130

6 authors, including:

Simone Picelli Omid R Faridani

Institute Ophthalmology Basel Karolinska Institutet
84 PUBLICATIONS 11,956 CITATIONS 41 PUBLICATIONS 9,825 CITATIONS

SEE PROFILE SEE PROFILE

Gosta Winberg Sven Sagasser

Karolinska Institutet Karolinska Institutet
68 PUBLICATIONS 10,600 CITATIONS 19 PUBLICATIONS 8,026 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Simone Picelli on 02 January 2014.

The user has requested enhancement of the downloaded file.

 protocol

Full-length RNA-seq from single cells using

Smart-seq2
Simone Picelli1, Omid R Faridani1, Åsa K Björklund1,2, Gösta Winberg1,2, Sven Sagasser1,2 & Rickard Sandberg1,2
1Ludwig Institute for Cancer Research, Stockholm, Sweden. 2Department of Cell and Molecular Biology, Karolinska Institutet, Stockholm, Sweden.
Correspondence should be addressed to R.S. ([email protected]).

Published online 2 January 2014; doi:10.1038/nprot.2014.006

Emerging methods for the accurate quantification of gene expression in individual cells hold promise for revealing the extent,
function and origins of cell-to-cell variability. Different high-throughput methods for single-cell RNA-seq have been introduced
that vary in coverage, sensitivity and multiplexing ability. We recently introduced Smart-seq for transcriptome analysis from single
cells, and we subsequently optimized the method for improved sensitivity, accuracy and full-length coverage across transcripts.
Here we present a detailed protocol for Smart-seq2 that allows the generation of full-length cDNA and sequencing libraries by
using standard reagents. The entire protocol takes ~2 d from cell picking to having a final library ready for sequencing; sequencing
will require an additional 1–3 d depending on the strategy and sequencer. The current limitations are the lack of strand specificity
and the inability to detect nonpolyadenylated (polyA−) RNA.
© 2013 Nature America, Inc. All rights reserved.

INTRODUCTION
It is becoming increasingly apparent that seemingly homogeneous with micro capillary pipettes is probably the best available option,
cell populations in vivo or cell cultures in vitro can show consider- although automated or semiautomated commercial solutions are
able heterogeneity in expression patterns1, owing to both intrinsic emerging. Tubes or plates containing single cells should be kept
stochastic processes and extrinsic factors, such as the surround- on ice or on an IsoFreeze PCR rack during collection and stored
ing microenvironment. Although high-throughput technologies, at −80 °C if not processed immediately.
such as microarray and, more recently, RNA-seq, have given us In this protocol we use a relatively mild (hypotonic) lysis
a better understanding of complex eukaryotic transcriptomes buffer8 capable of lysing cells without interfering with or inhib-
(e.g., see refs. 2–4), the RNA input requirements have limited iting the RT reaction. The lysis solution also contains free dNTPs
analyses to complex mixtures of often tens to hundreds of thou- and tailed oligo-dT oligonucleotides (30-nt poly-dT stretch and
sands of cells. As the results from expression analyses over tissues a 25-nt universal 5′ anchor sequence) that are responsible for the
or large cell populations reflect the cell population average, it priming of the RT reaction on polyadenylated RNA sequences.
does not inform on cellular variability nor does it allow for the The free dNTPs are added in the initial step to improve the yield
dis­covery of cellular subtypes. Instead, a single-cell resolution is of RT-PCRs12,14, probably through mechanisms that stabilize
needed to increase our understanding of cell-to-cell variability. RNA-primer hybridizations.
During the past few years, several groups have developed new The RT reaction is normally performed at 42 °C for 90 min when
sequencing-based methods for single-cell transcriptome ­analysis5–13. using commercial Smart-seq. However, some RNAs form second-
Our group has recently demonstrated that Smart-seq is able to ary structures (such as hairpins or loops) that might cause the
generate quantitative and reproducible data from both single cells enzyme to terminate chain elongation owing to steric hindrance.
and small amounts of purified RNA8. With Smart-seq2, we fur- This undesired effect15–17 can, to some extent, be overcome by the
ther refined reverse transcription (RT), template switching and addition of trehalose and betaine (N,N,N-trimethylglycine)18,19.
preamplification to obtain an increased cDNA yield from single Although some studies have shown that trehalose without other
cells, as well as higher sensitivity, fewer technical biases and less additives, or in combination with betaine, is capable of increasing
variability12. Here we present a detailed protocol for Smart-seq2 cDNA yield18,20, we obtained better results with betaine alone12.
(ref. 12) that entirely relies on off-the-shelf reagents. Therefore, our protocol uses betaine, a methyl group donor with
two important effects: it increases the thermal stability of pro-
Overview of the protocol teins and reduces or even eliminates the base pair composition
This protocol was developed for single cells but works equally dependence of DNA thermal melting transitions by destabiliz-
well on purified total RNA of a few picograms or more. A flow- ing the DNA helix21–23. It was also shown that betaine is more
chart of the protocol is provided in Figure 1. Cells should first be effective in destabilizing G-C pairs than A-T pairs23. Although its
disaggregated into a single-cell suspension, and you should aim beneficial effect on RT was established several years ago, it was not
for rapid processing at near-physiological conditions. A lengthy until recently that it was shown to improve the yield of full-length
procedure can lead to unwanted alterations in gene expression cDNA when using template-switching oligos (TSOs)20.
or, worse, to the death of a large fraction of the cell population. We also found that the increase in cDNA yield after adding
Therefore, cells should be picked promptly (in <30 min if pos- betaine (1 M) required an increased concentration of magne-
sible) and immediately placed in lysis buffer containing a ribo- sium chloride12. Mg2+ forms ion pairs with the carboxylate anion
nuclease inhibitor that blocks RNA degradation and stabilizes the of the betaine molecule, which under physiological conditions
RNA. When working with precious samples, manual cell picking is actually a stabilizing osmolyte that helps the cell respond to

nature protocols | VOL.9 NO.1 | 2014 | 171

 protocol
Cell lysis (Steps 1–8) times to maximize cDNA yield. We chose ten cycles as suggested
by Carninci et al.18, even though that experiment used no TSOs.
Incidentally, between 10 and 15 cycles appeared to maximize the
Poly(A)+ RNA
yield with our setup as well (data not shown).
The template-switching reaction relies on 2–5 untemplated
Oligo(dT) primer nucleotides that are added to the cDNA 3′ end when the RT
Reverse transcription reaches the 5′ end of the RNA. In its original version, the reaction
and terminal transferase (Steps 9–11)
contained a TSO carrying three riboguanosines at its 3′ end15,26.
LNA-containing TSO The Moloney murine leukemia (M-MLV) reverse transcriptase
is thus able to switch template and synthesize a complementary
sequence to the TSO. Every full-length cDNA carries the entire
Template switching 5′ end of the transcript and an additional artificial sequence,
by reverse transcriptase (Steps 9–11)
ISPCR primers which in this case is the same as the one located at the 5′ end of
the oligo-dT primer (i.e., the 3′-end of the mRNA), making pos-
sible the subsequent PCR using a single primer.
ISPCR primers Smart-seq2 uses a TSO carrying two riboguanosines in the
PCR preamplification of cDNA (Steps 12–14) third- and second-last positions and a modified guanosine to
PCR cleanup (Steps 15–26)
produce a locked nucleic acid (LNA) as the last base at the 3′ end.
These locked nucleotides are characterized by an internal bond
© 2013 Nature America, Inc. All rights reserved.

between the O2′ and the C4′ of the furanose ring, linked by a
Tagmentation (Tn5) (Steps 28–31)
methylene group. The modification introduces a conformational
lock in the molecule that, however, still retains the physical prop-
erties of the native nucleic acids. Two interesting properties of
LNAs are likely to be advantageous for our protocol: the enhanced
thermal stability of the LNA monomers; and their ability to anneal
Gap repair, enrichment PCR
strongly to the untemplated 3′ extension of the cDNA27.
and PCR purification (Steps 32–36) After the first-strand reaction, the cDNA is amplified using a
P5 primer i5 index
limited number of cycles, usually 18 when working with single
cells or just as many as needed to get enough material for the fol-
lowing steps (one or few nanograms of cDNA are sufficient). To
i7 index P7 primer simplify the protocol, to improve sensitivity and to adapt for liquid
Sequencing (Steps 37–41) handling automation, we eliminated the bead (AMPure XP)-
mediated purification of first-strand cDNA before the PCR.
Read 2 seq
This was made possible by the use of the KAPA HiFi HotStart
i5 index seq Read 1 seq
Sequencing-ready fragment i7 index seq ReadyMix, which replaced the Advantage 2 polymerase mix used
in the SMARTer protocol. Incidentally, KAPA HiFi was recently
Figure 1 | Flowchart for Smart-seq2 library preparation. Outline of the
protocol and the corresponding procedure steps. The oligo-dT primer,
tested in the sequencing of bacterial genomes and found to intro-
TSO and ISPCR primer are described in the main text, whereas tagmentation duce the least bias when compared with unamplified samples28.
uses primers that are included in the Nextera XT sample preparation and We used tagmentation to quickly and efficiently construct
index kits. sequencing libraries from the amplified cDNA. The tagmentation
reaction takes advantage of a hyperactive derivative of the Tn5
transposase that catalyzes in vitro integration of predetermined
oligonucleotides into target DNA29. The main advantage of this
changes in osmotic pressure. A recent study showed that betaine approach is that DNA fragmentation and adapter ligation occur
becomes a DNA-destabilizing agent above 1 mM MgCl 2 (ref. 24). in a single step, and size selection is not necessary. A comparison
In agreement with that study, we found that a final concentration of sequence composition surrounding the fragmentation sites
of 9–12 mM MgCl2 was necessary for maximizing cDNA yield12. identified a bias signature that was stronger than that associated
Although it has been reported that too much MgCl2 has a negative with mechanical shearing methods29. Nonetheless, this was found
effect on the fidelity of the PCR25, we did not observe any negative to have little effect on the coverage for the human genome. Thus,
influence either for the Superscript II (Invitrogen) used in the RT we could reasonably assume that the fragments generated after
or the KAPA HiFi DNA polymerase used in the PCR. digestion of full-length cDNA evenly cover the entire length of
Through the addition of betaine, we could also exploit its pro- the transcript5.
tein thermal-stabilization properties to modify the temperature The fragments in the tagmented DNA library have an aver-
at which the RT is performed. After 90 min at 42 °C (the optimal age size that usually ranges from 200 to 600 bp and are ready
conditions for Superscript II, according to the manufacturer), we for enrichment PCR. We followed the protocol developed by
raised the temperature to 50 °C (the upper limit before inactiva- Illumina that allows the pooling of up to 96 samples through
tion) for 2 min to promote the unfolding of any RNA secondary the use of a dual-index strategy, referred to as index 1 (i7) and
structures. We then lowered it to 42 °C again for 2 min to allow index 2 (i5). Samples were pooled after enrichment PCR and
the completion of the RT and repeated this cycle a total of ten sequenced together on the same lane of the Illumina sequencers.

172 | VOL.9 NO.1 | 2014 | nature protocols

 protocol
Strengths and limitations of the protocol ­ latforms (such as the Agilent Bravo) or microfluidics solutions
p
This method has several advantages, briefly summarized below: (such as the Fluidigm C1).

• It gives the possibility of analyzing hundreds of cells for a frac- Future applications
tion of the cost of currently available commercial kits. Compared Numerous applications are made feasible because of the reduction
with the widely used SMARTer kit, our protocol can generate in cost per sample and the increase in throughput, including (but
improved-quality libraries for ~12% of the cost. not limited to) the following:
• It allows the recovery of full-length cDNAs, owing to the prefer-
• Isolation of single cells from different regions of a primary tumor
ence of M-MLV reverse transcriptase for full-length over trun-
cated cDNAs as a substrate for its terminal transferase activity. to gain information on cellular heterogeneity. It has been shown
It is therefore possible to analyze all the exons of each transcript that solid tumors are often composed of multiple clonal sub-
and to detect the different splice variants, a big advantage over populations30–32, which confounds clinical analysis of bulk RNA
previous methods5–11,13. It also enables comprehensive SNP and or DNA samples from hundreds of thousands of cells. The gene
mutation analysis, widening its field of application. expression profiles from such mixed populations of tumor cells
• It allows a high degree of multiplexing; up to 96 samples can be may be strongly biased by the contributions from different cell
pooled and sequenced on a single lane of an Illumina sequencer. populations within the tumor, making it impossible to discern
the characteristics that determine the malignant phenotype33.
However, some limitations remain: Moreover, preliminary exome-sequencing studies at the single-
cell level have revealed that solid tumors and hematopoietic
• This method is selective for polyadenylated RNA, precluding tumors can show a very different clonal origin34,35. It is reason-
© 2013 Nature America, Inc. All rights reserved.

analyses of polyA− RNA. able to assume that larger numbers of single cells from different
• The reads do not reflect the strand-specific nature of mRNAs. tumor regions need to be analyzed to better understand tumor
• Despite high multiplexing at sequencing, the pooling of samples evolution and progression.
is performed after the adapter-ligated enrichment PCR step. The • The study of the different cell types that constitute a developing
relatively high manual labor compared with early multiplex- embryo, a tissue or a whole animal, with the goal of establish-
ing methods (e.g., CEL-seq9 or STRT7,10) can be decreased by ing the developmental hierarchy of stem cells, progenitors and
the adaptation of the protocol for automated liquid handling differentiated cells.

MATERIALS
REAGENTS • Nextera XT DNA sample preparation kit, 96 samples (Illumina,
• Dulbecco’s phosphate-buffered saline (DPBS), no calcium, no magnesium cat. no. FC-131-1096)
(Gibco, cat. no. 14190-144) • Nextera XT 24-index kit, 96 samples (Illumina, cat. no. FC-131-1001)
• 1× TrypLE Express enzyme, no phenol red (Gibco, cat. no. 12604-021) • Adapter oligos (See Reagent Setup). All oligos except the LNA-modified
• RNaseZap (Ambion, cat. no. AM9780) TSO were ordered from commercial vendor Biomers.net (http://www.
• DNA-OFF (Takara Bio, cat. no. 9036) biomers.net/) and were subjected to HPLC purification. LNA-modified
• Triton X-100 (Sigma-Aldrich, cat. no. T9284) ! CAUTION Triton X-100 is TSO was ordered from Exiqon (http://www.exiqon.com/).
harmful if swallowed. It causes serious eye damage. Handle it using • CASAVA 1.8.2 (Illumina) software, freely available for the Linux environment
appropriate safety equipment. EQUIPMENT
• dNTP mix (10 mM each; Fermentas, cat. no. R0192) • Center-well organ culture dish, 60 × 15 mm polystyrene dish
• First-strand buffer (5×; 250 mM Tris-HCl, pH 8.3, at room temperature (Corning, cat. no. 353037)
(25 °C); 375 mM KCl; 15 mM MgCl2; Invitrogen, cat. no. 18064-014) • Microcentrifuge tubes, polyallomer (Beckman Coulter, cat. no. 357448)
• DTT (Invitrogen, cat. no. 18064-014) ! CAUTION DTT is toxic when • Microcentrifuge Safe-Lock tubes, polypropylene (Eppendorf,
ingested. Avoid inhaling fumes or contact with the skin. Handle it using cat. no. 0030 120.086)
appropriate safety equipment. • Falcon polystyrene conical tube (50 ml, BD Biosciences, cat. no. 352095)
• Superscript II reverse transcriptase (Invitrogen, cat. no. 18064-014) • Mini vortexer (VWR, cat. no. 82019-170)
• Recombinant RNase inhibitor (Clontech, cat. no. 2313A) • Thermal cycler (S1000, Bio-Rad)
• Betaine (BioUltra ≥99.0%; Sigma-Aldrich, cat. no. 61962) • Magnetic stand 96 (Ambion, cat. no. AM10027)
• Magnesium chloride (MgCl2; anhydrous; Sigma-Aldrich, cat. no. M8266) • 8-strip, nuclease-free, 0.2-ml, thin-walled PCR tubes with caps (Sarstedt,
• Distilled water (Gibco, cat. no. 10977) cat. nos. 72.985.002 and 65.989.002)
• KAPA HiFi HotStart ReadyMix (2×; KAPA Biosystems, cat. no. KK2601) • V-bottom plates (96 well; VWR, cat. no. 47743-996)
 CRITICAL A HotStart DNA polymerase is necessary to minimize the • Filter tips: 10, 20, 100 and 200 µl (Gilson, cat. nos. F171203, F171303,
background amplification when working with single cells and is more F171403 and F171503)
practical when working with automated liquid-handling platforms. • Qubit assay tubes (Invitrogen, cat. no. Q32856)
• Agencourt Ampure XP beads (Beckman Coulter, cat. no. A 63881) • Qubit dsDNA high-sensitivity (HS) kit (Invitrogen, cat. no. Q32851)
• Ethanol 99.5% (vol/vol); Kemethyl, cat. no. SN366915-06) ! CAUTION It is • Qubit 2.0 fluorometer (Invitrogen, cat. no. Q32866)
flammable; handle it using appropriate safety equipment. • Agilent 2100 Bioanalyzer (Agilent Technologies, cat. no. G2938C)
• EB solution (10 mM Tris-Cl, pH 8.5; Qiagen, cat. no. 19086) • Agilent high-sensitivity DNA kit (Agilent Technologies, cat. no. 5067-4626)
• TruSeq dual-index sequencing primer kit for single-read runs (Illumina, • A compatible Illumina DNA sequencing instrument (instruments currently
cat. no. FC-121-1003) or paired-end runs (Illumina, cat. no. PE-121-1003) on the market are the MiSeq, HiSeq 2000, HiSeq 2500)

nature protocols | VOL.9 NO.1 | 2014 | 173

 protocol
REAGENT SETUP Oligo-dT30VN (5′–AAGCAGTGGTATCAACGCAGAGTACT30VN-3′)
Mammalian cell isolation Single cells are obtained by trypsin treatment This anneals to all the RNAs containing a poly(A) tail. The 3′ end of this
of adherent cell cultures. To facilitate manual picking with a micro capillary oligo­nucleotide contains ‘VN’, where ‘N’ is any base and ‘V’ is either A, C or
pipette, a small volume of the cell suspension is added to a culture dish G. The two terminal nucleotides are necessary for anchoring the oligo-
containing DPBS and TrypLE Express enzyme in a ratio of 1:1. nucleotide to the beginning of the polyA tail and to avoid unnecessary
Cell lysis buffer Combine 0.2% (vol/vol) Triton X-100 and 2 U µl–1 RNase amplification of a long stretch of adenosines. Dissolve the oligonucleotide
inhibitor. This buffer can be stored at 4 °C for 6 months. in TE buffer, to a final concentration of 100 µM. Store this oligo at −20 °C
TSO (5′-AAGCAGTGGTATCAACGCAGAGTACATrGrG+G-3′) At the for 6 months.
5′ end, this TSO carries a common primer sequence, whereas, at the 3′ end, ISPCR oligo (5′-AAGCAGTGGTATCAACGCAGAGT-3′) This oligo acts as
there are two riboguanosines (rG) and one LNA-modified guanosine (+G) to PCR primer in the amplification step after RT. Dissolve the oligonucleotide
facilitate template switching. Dissolve TSO in TE buffer. TSO can be stored in in TE buffer to a final concentration of 100 µM. This oligo can be stored at
100 µM aliquots at −80 °C for 6 months. Avoid repeated freeze-thaw cycles. −20 °C for 6 months.

PROCEDURE
Preparing single-cell samples ● TIMING ~15 min
 CRITICAL Use a thermal cycler with a heated lid set to 105 °C for all incubations throughout this protocol.

1| Clean the hood with RNaseZap and DNA-OFF solutions before setting up the working plates. Spray pipettes with
© 2013 Nature America, Inc. All rights reserved.

RNaseZap.
 CRITICAL STEP All the experiments must be performed under a UV-sterilized hood with laminar flow, and all the surfaces
must be free from RNase to prevent degradation of RNA and from DNA to prevent cross-contamination from previous samples.
The hood must be used only for single-cell experiments up to (but excluding) the cDNA amplification step (Step 12). An ideal
scenario would be to place the hood in a separate room with a positive air pressure to prevent any contaminants from being
carried inside, where they might affect the experiments. The room should be equipped with a garmenting area in which the
user changes into a fresh disposable lab coat, hair net, dust mask, shoe covers and vinyl gloves (powder-free).

Single-cell lysis ● TIMING ~15 min (for eight-strip tubes)

2| Dilute the oligo-dT30VN primer to 10 µM by adding 10 µl of 100 µM oligo-dT primers and 90 µl of nuclease-free water to
a tube and mix well.

3| Prepare cell lysis buffer by adding 1 µl of RNase inhibitor to 19 µl of a 0.2% (vol/vol) Triton X-100 solution. If you are
working with purified RNA, this step can be omitted and a corresponding volume of water can be used instead.

4| Isolate single cells in the lowest possible volume (preferably ≤0.5 µl, possibly 0.3 µl) or pipet the appropriate amount
of RNA into a 0.2-ml thin-walled PCR tube. Single cells can be obtained either by using a micro capillary pipette or via FACS.

5| Place each single cell into a 0.2-ml thin-walled PCR tube containing 2 µl of cell lysis buffer, 1 µl of oligo-dT primer and
1 µl of dNTP mix.

6| Quickly vortex the tube to mix, and then spin down the solution (700g for 10 s at room temperature) and immediately
place it on ice.

7| Incubate the samples at 72 °C for 3 min and immediately put the tube back on ice.

8| Spin down the samples (700g for 10 s at room temperature) to collect the liquid at the bottom of the tubes, and then
put them immediately back on ice. The oligo-dT primer is now hybridized to the poly(A) tail of all the mRNA molecules.

Reverse transcription ● TIMING 3 h

9| Prepare the RT mix for all reactions plus one additional reaction by combining and mixing the reagents listed in the
table below.
 CRITICAL STEP Thaw all the reagents in advance and assemble the RT mix while performing denaturation (Step 7) to
minimize bias.
 CRITICAL STEP The addition of MnCl2 to improve the template-switching efficiency is not necessary26. It actually
decreases the final yield of cDNA and causes misincorporation during RT12.

174 | VOL.9 NO.1 | 2014 | nature protocols

 protocol

Component Volume (ml) Final concentration

SuperScript II reverse transcriptase 0.50 100 U

(200 U µl−1)

RNAse inhibitor (40 U µl−1) 0.25 10 U

Superscript II first-strand buffer (5×) 2.00 1×

DTT (100 mM) 0.50 5 mM

Betaine (5 M) 2.00 1M

MgCl2 (1 M) 0.06 6 mM

TSO (100 µM) 0.10 1 µM

Nuclease-free water 0.29 –

Total volume 5.70 –

© 2013 Nature America, Inc. All rights reserved.

10| Add 5.7 µl of the RT mix to the samples from Step 9 to obtain a final reaction volume of 10 µl. Mix the reaction by
gently pipetting up and down a few times without forming bubbles.

11| Spin down the samples (700g for 10 s at room temperature) to collect the liquid at the bottom of the tubes,
and incubate the reaction in a thermal cycler with a heated lid, as detailed below. Note that the ten cycles after the initial
incubation at 42 °C for 90 min are not crucial for a successful reaction, but they give a marginal increment in yield.

Cycle Temperature (°C) Time Purpose

1 42 90 min RT and template-switching

2–11 50 2 min Unfolding of RNA secondary structures

42 2 min Completion/continuation of RT and

template-switching

12 70 15 min Enzyme inactivation

13 4 Hold Safe storage

PCR preamplification ● TIMING 3 h

12| Prepare the PCR mix for all reactions plus one additional reaction by combining and mixing the following components:

Component Volume (ml) Final concentration

First-strand reaction 10 –

KAPA HiFi HotStart ReadyMix (2×) 12.50 1×

IS PCR primers (10 µM) 0.25 0.1 µM

Nuclease-free water 2.25 –

Total volume 25 –

13| Add 15 µl of PCR mix to each tube from Step 12, which contains the first-strand reaction. Vortex the tubes to mix,
and then spin them down (700g for 10 s at room temperature) to collect the liquid at the bottom of the tubes.

nature protocols | VOL.9 NO.1 | 2014 | 175

 protocol

14| Perform the PCR in a thermal cycler by using the following program:

Cycle Denature Anneal Extend Hold

1 98 °C, 3 min – – –

2–19 98 °C, 20 s 67 °C, 15 s 72 °C, 6 min –

20 – – 72 °C, 5 min –

21 – – – 4 °C

 CRITICAL STEP The number of PCR cycles depends on the input amount of RNA. We typically use 18 cycles for single
eukaryotic cells to obtain ~1–30 ng of amplified cDNA. The number of cycles can be increased for smaller cells (with less RNA
content) or lowered for large cells (with more RNA).
 CRITICAL STEP The KAPA HiFi HotStart ReadyMix buffer has a higher salt concentration compared with common PCR
buffers. The buffer composition affects DNA melting, and it is therefore important to perform the denaturation Step at 98 °C
and not at 95 °C. Similarly, the optimal annealing temperature for the ISPCR primers is different compared with other DNA
polymerases. In our hands, the KAPA HiFi has proven to be a very robust enzyme, giving satisfactory results in the range
64–68 °C for the annealing step, with no relevant differences in cDNA yield or length.
© 2013 Nature America, Inc. All rights reserved.

 PAUSE POINT PCR product can be stored at −20 or −80 °C for 6 months or longer.

PCR purification ● TIMING ~45 min

15| Before commencing the purification steps, equilibrate Ampure XP beads at room temperature for 15 min, and then vortex
well for several seconds.

16| Add 25 µl of Ampure XP beads (1:1 ratio) to each sample from Step 14 and mix by pipetting up and down ten times or
until the solution appears homogeneous. Transfer solutions to a 96-well plate with compatible magnet stand.
 CRITICAL STEP Do not increase the volume of beads in the purification step above the 1:1 ratio. A less-than-standard
amount of beads ensures that primer dimer carryover is kept to a minimum.

17| Incubate the mixture for 8 min at room temperature to let the DNA bind to the beads.

18| Place the 96-well plate on the magnetic stand for 5 min or until the solution is clear and the beads have been collected
at one corner of the well.

19| While samples are on the magnet, carefully remove the liquid without disturbing the beads.

20| Wash the beads with 200 µl of 80% (vol/vol) ethanol solution. Incubate the samples for 30 s and then remove the
ethanol.
 CRITICAL STEP It is important that the ethanol solution is freshly prepared every time, as ethanol absorbs moisture from
the environment, thus changing the final concentration.

21| Repeat Step 20 once more.

22| Remove any trace of ethanol and let the beads dry completely, leaving the plate at room temperature for 5 min or until
a small crack appears on the surface of the beads.
 CRITICAL STEP Avoid overdrying the beads because this will make their resuspension in the designated buffer more
difficult. As a precaution, cover the plate during this step or protect it from any possible source of contamination or air flows
that might disperse the beads around the well, thus leading to cross-contamination between adjacent wells (especially,
when the beads are overdried).

23| Add 17.5 µl of EB solution (or nuclease-free water). Mix ten times to resuspend the beads.

24| Incubate the plate off the magnet for 2 min.

25| Place the plate on the magnetic stand and leave it for 2 min or until the solution appears clear and beads have
accumulated in a corner of the well.

176 | VOL.9 NO.1 | 2014 | nature protocols

 protocol

26| Set the volume of the pipette to 15 µl, collect the supernatant without disturbing the beads and transfer it to a
fresh 0.2-ml thin-walled PCR tube.
 CRITICAL STEP Avoid aspirating the whole volume of EB solution or nuclease-free water. Leaving 2.5 µl in the well ensures
that bead carryover is kept to a minimum.

Quality check of the cDNA library ● TIMING 1 h

27| Check the size distribution on an Agilent high-sensitivity DNA chip. A good library should be free of short (<500 bp)
fragments and should show a peak at ~1.5–2 kb.
? TROUBLESHOOTING

Tagmentation reaction ● TIMING ~10 min

28| Tagmentation is carried out by using the Illumina Nextera XT DNA sample preparation kit. Set up the tagmentation
reaction on ice as follows and mix the solution carefully with a pipette:

Component Volume (ml) Final concentration

Tagment DNA buffer (TD, 2×) 10 1×

© 2013 Nature America, Inc. All rights reserved.

Amplicon tagment mix 5 –

DNA from PCR Variable –

Nuclease-free water Variable –

Total volume 20 –

 CRITICAL STEP When you are using the Nextera XT DNA sample preparation kit, the input should never exceed 1 ng.
An excess of DNA leads to incomplete cutting and libraries with longer average size.
 CRITICAL STEP We recommend assembling the reaction on ice. The Tn5 enzyme has optimal activity at 55 °C, but some
activity also occurs at room temperature.

29| Perform the tagmentation reaction in a thermal cycler by using the following program:

Cycle Temperature (°C) Time

1 55 5 min

2 4 Hold

Stripping Tn5 transposase off the tagmented DNA ● TIMING 5 min

30| To strip off the Tn5, add 5 µl of NT buffer (Nextera XT DNA sample preparation kit) to the 20-µl solution containing the
tagmented DNA (from Step 29). Pipet the solution up and down a few times to mix.

31| Incubate the mixture for 5 min at RT. DNA is now ready for the final enrichment PCR.

Amplification of adapter-ligated fragments ● TIMING 1 h

32| Prepare the enrichment PCR in a 0.2-ml tube using the reagents from the Nextera XT kit.

Component Volume (ml) Final concentration

DNA 25 –

Nextera PCR master mix 15 –

Index 1 primers (N7xx) 5 –

Index 2 primers (N5xx) 5 –

Total volume 50 –

nature protocols | VOL.9 NO.1 | 2014 | 177

 protocol

33| Next, perform the PCR by using the following program:

Cycle Denature Anneal Extend Hold

1 – – 72 °C, 3 min –
2 95 °C, 30 s – – –
3–14 95 °C, 10 s 55 °C, 30 s 72 °C, 30 s –
15 – – 72 °C, 5 min –
16 – – – 4 °C

Amount (pg) No. of cycles

500–1,000 6–8
100–500 10–12
50–100 12–14
10–50 14–16
© 2013 Nature America, Inc. All rights reserved.

 CRITICAL STEP The number of cycles depends on the amount of DNA used for tagmentation. If we are starting from 100 pg
of amplified cDNA, we usually perform 12 PCR cycles. The optimal number of cycles depends on the sample and the experi-
ment. It may be helpful to run a range of cycles to determine the best conditions. Above are cycling guidelines on the basis
of the input DNA used for tagmentation.

PCR purification ● TIMING ~45 min

34| Repeat Steps 15–24, but use only 30 µl of AMPure XP beads in a ratio of 0.6:1. This will minimize the carryover of
primer dimers.

35| Place the tube on the magnetic stand to capture the beads for 2 min or until the solution is clear. Set the volume of the
pipette to 15 µl, collect the supernatant and transfer it into a new 1.5-ml polyallomer tube to minimize DNA absorption to
the tube during long-term storage.

Quality check of the final cDNA library ● TIMING 1 h

36| Measure the concentration of each library by Qubit according to the manufacturer’s instructions. Check the size
distribution on an Agilent high-sensitivity chip. Size distribution varies according to the amount of DNA used in the
tagmentation. As a rule of thumb, a broad peak with an average size of 300–800 bp will be observed.
? TROUBLESHOOTING
 PAUSE POINT The sample can be stored after this step at −20 °C for 3 months.

Library pooling ● TIMING ~30 min

37| Perform a dilution of the library, aiming to get a 2 nM solution. Use the concentration (ng µl–1) obtained with the Qubit
and the average size obtained on the Bioanalyzer to calculate the molarity of the final library.

38| Measure the diluted library by Qubit and adjust the concentration to get 2 nM, if necessary.

39| Pool equal nanomoles of each sample, ensuring that none of them has the same combination of N5xx and N7xx adapters.
If you are pooling just a few samples (especially fewer than six), make sure that you maintain the color balance for each
base of the index read, or else the index read sequencing might fail because of registration failure. Check the Nextera DNA
sample preparation guide (available at http://www.illumina.com) to determine the compatibility of each index.

40| Measure once again the concentration of the pooled libraries by Qubit, and then adjust the concentration to 2 nM, if necessary.

DNA sequencing ● TIMING ~1–2 d

41| Perform single-end or paired-end sequencing of the libraries according to the manufacturer’s protocol and by using the
TruSeq dual-index sequencing primers on either a HiSeq 2000, 2500 or MiSeq instrument; a 50-bp single-end sequencing on a
HiSeq 2000 takes ~2 d, whereas that on a MiSeq takes less than 1 d. The degree of multiplexing per lane depends on the read
depth obtained on each instrument. Multiplex sequencing of 50 libraries on a single HiSeq 2000, from which 200 million reads
are regularly obtained, would give an average sequence depth of 4 million reads per cell. If the experiment is designed to identi-
fy SNPs, mutations or RNA splice variants, one might require a higher sequence depth per cell by pooling fewer libraries per lane.

178 | VOL.9 NO.1 | 2014 | nature protocols

 protocol
Data analysis ● TIMING ~1 d
42| Export raw sequences and quality scores in FastQ format for all clusters that passed filtering (i.e., pass filter clusters)
using CASAVA software.

43| Use the index reads to demultiplex the sequenced reads into correct sample origin (i.e., cells).

44| Align the reads to the genome and transcriptome by using an aligner developed for RNA-seq data.

45| Estimate gene expression levels for each gene or transcripts by following dedicated protocols, e.g., ref. 36.

? TROUBLESHOOTING
Troubleshooting advice can be found in Table 1.

Table 1 | Troubleshooting table.

Step Problem Possible reason Solution

27 No product after cDNA Cell was dead or damaged at the Keep the cells in conditions that are as close as
© 2013 Nature America, Inc. All rights reserved.

preamplification moment of picking possible to those present in the tissue/culture of

origin. Check the quality of cells (i.e., by staining
for dead cells) and avoid leaving them in DPBS or
cell medium for too long before picking

Noisy amplification with a tail Large excess of TSO in relation to the Decrease the amount of TSO used (a titration may
of short fragments (see Fig. 2c, mRNA is present in the cell (appears be necessary) or use a 5′-blocked TSO that does not
‘hedgehog pattern’) more frequently with cells that have a form concatamers
low transcriptional activity)

Overall yield is low or lower Sample quality was poor Keep cells in conditions that are as physiological as
than expected possible before and during the picking. Work fast,
immediately put the cells in lysis buffer containing
RNase inhibitor. Keep the cells that have already
been picked on ice until you start the RT

Amount of cDNA is much larger Contamination occurred during the Perform all the experiments in a dedicated area that
than usual first steps has been thoroughly cleaned and decontaminated
from RNases and DNases
Use a dust mask, gloves, hair net, dedicated lab
coat and change gloves often

Overamplification of cDNA after RT The number of PCR cycles can vary from sample to
sample. Start with 18 cycles and adjust it later,
if necessary

36 Tagmented DNA shows a wide Too much DNA was used in the Use less DNA and never exceed 1 ng for optimal
distribution with a tail of tagmentation reaction results
longer fragments

Partial inhibition of Tn5 transposase Elution from AMPure XP beads in the previous
step was done using a solution containing a
detergent or a buffer with a too high salinity.
Even a small amount of detergent completely
inhibits the enzyme

Enough DNA is used as input Too few cycles of PCR Increase the number of PCR cycles
for tagmentation but no or very
low peaks after enrichment PCR
is observed

Primer dimers are present on Overamplification or an excessive Reduce either the number of PCR cycles or the
the Bioanalyzer trace amount of adaptors relative to DNA amount of adaptors. Use a lower ratio of AMPure
is used for tagmentation beads:DNA (never exceed a 1:1 ratio, preferably
use a 0.6:1 ratio)

nature protocols | VOL.9 NO.1 | 2014 | 179

 protocol

● TIMING
Day 1
Steps 1–8, preparing single-cell samples, picking and lysis: ~30 min (for eight-strip tubes)
Steps 9–11, reverse transcription: 3 h
Steps 12–14, PCR preamplification: 3 h
Day 2
Steps 15–26, PCR purification: ~45 min
Step 27, quality check of the cDNA library: 1 h
Steps 28 and 29, tagmentation reaction: ~10 min
Steps 30 and 31, purification of tagmented DNA: 5 min
Steps 32 and 33, amplification of adapter-ligated fragments: 1 h
Steps 34 and 35, PCR purification: ~45 min
Step 36, quality check of the final cDNA library: 1 h
Steps 37–40, library pooling: ~30 min
Days 3–5
Step 41, DNA sequencing: 1–2 d
Steps 42–45, data analysis: ~1 d
© 2013 Nature America, Inc. All rights reserved.

ANTICIPATED RESULTS
Step 27
After cDNA pre-amplification and purification, one has the first opportunity to determine the overall quality of the initial
cell. This is normally performed with an Agilent Bioanalyzer. A good library should have an average size of 1.5–2 kb and a
small amount of short fragments (Fig. 2a). In the case of suboptimal preamplification, the peak of primer dimers will be
higher than the cDNA library (Fig. 2b) or, in extreme cases, mostly amplification of primer concatamers will be observed
(Fig. 2c). A library prepared with degraded RNA is characterized by a shift toward short fragments, even though some
amplification of longer fragments could take place (Fig. 2d).

Step 36
After this step, the final cDNA library is ready for sequencing. The cDNA has been tagmented and amplified with index
primers that specifically enrich the final library for fragments carrying adapters on both ends. The expected size of the library
is ~300–800 bp, and the amount of adapter dimers is low (Fig. 3).

Step 46 (end result)

a 350
b 400
Sequence reads from each individual cell are normally in the
(fluorescence units)

(fluorescence units)

300 350 range of 1–20 million, depending on the level of multiplexing

DNA amount

DNA amount

250 300
200 250 in the sequencing. When sequencing 50-bp single-end reads,
200
150
100
150 we find that normally 60% of reads map uniquely to the
100
50 50 genome (20% multimapping and 20% with no match); of the
0 0
35 150 300 500 10,380 35 150 300 500 10,380 uniquely mapping reads, >60% of the reads map to annotated
c DNA size (bp)
d DNA size (bp)
Ref Seq exons, 20% intronic and 20% intergenic, but these
300 250 values depend on the completeness of the gene annotations.
(fluorescence units)
(fluorescence units)

250 200
The read coverage across transcripts should be even. Represen­
DNA amount
DNA amount

200 150
150
100
tative values for single-cell libraries prepared from HEK293T
100
50 50 and DG-75 cells are listed in Supplementary Table 1.
0 0
35 150 300 500 10,380 35 150 300 500 10,380 700
DNA size (bp) DNA size (bp) 600
(fluorescence units)

500
DNA amount

Figure 2 | Bioanalyzer electropherograms of pre-amplified cDNA libraries. 400

(a) Representative example of the cDNA size distribution obtained from 300
the successful cDNA preamplification from a single MEF cell (mouse). 200
The profile has a peak ~1.5–2 kb with a small number of fragments below 100
500 bp and a small amount of primer dimers. (b) Representative example 0
35 150 300 500 10,380
of the primer dimer–dominated cDNA profile from a single HEK293T cell DNA size (bp)
(human). Amplification of full-length cDNA is visible but a large peak of
primer dimers that were not removed by bead purification dominates the Figure 3 | Bioanalyzer electropherogram of a sequencing library after
profile. (c) Data from a single T cell (human) with a ‘hedgehog’ pattern, tagmentation. Representative profile of a library derived from a single
most probably due to the formation of TSO concatamers. (d) Data from a HEK293T cell (human) prepared according to the Nextera XT protocol.
single interneuron (mouse), with a wide distribution of fragments indicative Amplification (12 cycles) was carried out according to manufacturer’s
of RNA degradation before library preparation. instructions, as described in Step 33.

180 | VOL.9 NO.1 | 2014 | nature protocols

 protocol
Note: Any Supplementary Information and Source Data files are available in the of message diversity by RT-PCR. Biotechniques 19, 806–812, 814–805
online version of the paper. (1995).
17. Kuo, K.W., Leung, M.F. & Leung, W.C. Intrinsic secondary structure of
Acknowledgments This work was supported by the European Research Council human TNFR-I mRNA influences the determination of gene expression by
Starting grant no. 243066 (R.S.), the Swedish Foundation for Strategic Research RT-PCR. Mol. Cell Biochem. 177, 1–6 (1997).
FFL4 (R.S.), the Åke Wiberg Foundation (R.S.) and the Swedish Research Council 18. Carninci, P. et al. Thermostabilization and thermoactivation of
grant nos. 2008-4562 (R.S.) and 2010-6844 (Å.K.B.). thermolabile enzymes by trehalose and its application for the synthesis of
full length cDNA. Proc. Natl. Acad. Sci. USA 95, 520–524 (1998).
19. Spiess, A.N. & Ivell, R. A highly efficient method for long-chain cDNA
AUTHOR CONTRIBUTIONS S.P. developed and wrote the protocol; O.R.F.
synthesis using trehalose and betaine. Anal. Biochem. 301, 168–174
conceived and designed LNA-based oligos; Å.K.B. performed computational
(2002).
analyses; S.S. picked cells; G.W. contributed to protocol development;
20. Pinto, F.L. & Lindblad, P. A guide for in-house design of template-switch-
R.S. designed the study and wrote the manuscript.
based 5′ rapid amplification of cDNA ends systems. Anal. Biochem. 397,
227–232 (2010).
COMPETING FINANCIAL INTERESTS The authors declare competing financial 21. Santoro, M.M., Liu, Y., Khan, S.M., Hou, L.X. & Bolen, D.W. Increased
interests: details are available in the online version of the paper. thermal stability of proteins in the presence of naturally occurring
osmolytes. Biochemistry 31, 5278–5283 (1992).
Reprints and permissions information is available online at http://www.nature. 22. Rajendrakumar, C.S., Suryanarayana, T. & Reddy, A.R. DNA helix
com/reprints/index.html. destabilization by proline and betaine: possible role in the salinity
tolerance process. FEBS Lett. 410, 201–205 (1997).
1. Wilkinson, D.J. Stochastic modelling for quantitative description of 23. Rees, W.A., Yager, T.D., Korte, J. & von Hippel, P.H. Betaine can eliminate
heterogeneous biological systems. Nat. Rev. Genet. 10, 122–133 (2009). the base pair composition dependence of DNA melting. Biochemistry 32,
2. Wilhelm, B.T. et al. Dynamic repertoire of a eukaryotic transcriptome 137–144 (1993).
surveyed at single-nucleotide resolution. Nature 453, 1239–1243 (2008). 24. Lambert, D. & Draper, D.E. Effects of osmolytes on RNA secondary and
© 2013 Nature America, Inc. All rights reserved.

3. Wang, E.T. et al. Alternative isoform regulation in human tissue tertiary structure stabilities and RNA-Mg2+ interactions. J. Mol. Biol. 370,
transcriptomes. Nature 456, 470–476 (2008). 993–1005 (2007).
4. Mortazavi, A., Williams, B.A., McCue, K., Schaeffer, L. & Wold, B. Mapping 25. Eckert, K.A. & Kunkel, T.A. High fidelity DNA synthesis by the
and quantifying mammalian transcriptomes by RNA-seq. Nat. Methods 5, Thermus aquaticus DNA polymerase. Nucleic Acids Res. 18, 3739–3744
621–628 (2008). (1990).
5. Tang, F. et al. mRNA-seq whole-transcriptome analysis of a single cell. 26. Zhu, Y.Y., Machleder, E.M., Chenchik, A., Li, R. & Siebert, P.D. Reverse
Nat. Methods 6, 377–382 (2009). transcriptase template switching: a SMART approach for full-length cDNA
6. Tang, F. et al. RNA-seq analysis to capture the transcriptome landscape of library construction. Biotechniques 30, 892–897 (2001).
a single cell. Nat. Protoc. 5, 516–535 (2010). 27. Petersen, M. & Wengel, J. LNA: a versatile tool for therapeutics and
7. Islam, S. et al. Characterization of the single-cell transcriptional landscape genomics. Trends Biotechnol. 21, 74–81 (2003).
by highly multiplex RNA-seq. Genome Res. 21, 1160–1167 (2011). 28. Quail, M.A. et al. Optimal enzymes for amplifying sequencing libraries.
8. Ramskold, D. et al. Full-length mRNA-seq from single-cell levels of RNA and Nat. Methods 9, 10–11 (2012).
individual circulating tumor cells. Nat. Biotechnol. 30, 777–782 (2012). 29. Adey, A. et al. Rapid, low-input, low-bias construction of shotgun
9. Hashimshony, T., Wagner, F., Sher, N. & Yanai, I. CEL-Seq: single-cell fragment libraries by high-density in vitro transposition. Genome Biol. 11,
RNA-seq by multiplexed linear amplification. Cell Rep. 2, 666–673 (2012). R119 (2010).
10. Islam, S. et al. Highly multiplexed and strand-specific single-cell RNA 5′ 30. Navin, N. et al. Inferring tumor progression from genomic heterogeneity.
end sequencing. Nat. Protoc. 7, 813–828 (2012). Genome Res. 20, 68–80 (2010).
11. Sasagawa, Y. et al. Quartz-seq: a highly reproducible and sensitive 31. Torres, L. et al. Intratumor genomic heterogeneity in breast cancer with
single-cell RNA-seq reveals non-genetic gene expression heterogeneity. clonal divergence between primary carcinomas and lymph node
Genome Biol. 14, R31 (2013). metastases. Breast Cancer Res. Treat 102, 143–155 (2007).
12. Picelli, S. et al. Smart-seq2 for sensitive full-length transcriptome profiling 32. Shipitsin, M. et al. Molecular definition of breast tumor heterogeneity.
in single cells. Nat. Methods 10, 1096–1098 (2013). Cancer Cell 11, 259–273 (2007).
13. Yan, L. et al. Single-cell RNA-seq profiling of human preimplantation embryos 33. Navin, N. & Hicks, J. Future medical applications of single-cell sequencing
and embryonic stem cells. Nat. Struct. Mol. Biol. 20, 1131–1139 (2013). in cancer. Genome Med. 3, 31 (2011).
14. Huang, L., Lee, J., Sitaraman, K. & Gallego, A. A new highly sensitive 34. Xu, X. et al. Single-cell exome sequencing reveals single-nucleotide
two-step RT-PCR system. Focus 22, 6–7 (2000). mutation characteristics of a kidney tumor. Cell 148, 886–895 (2012).
15. Abbotts, J., Bebenek, K., Kunkel, T.A. & Wilson, S.H. Mechanism of HIV-1 35. Hou, Y. et al. Single-cell exome sequencing and monoclonal evolution
reverse transcriptase. Termination of processive synthesis on a natural of a JAK2-negative myeloproliferative neoplasm. Cell 148, 873–885
DNA template is influenced by the sequence of the template-primer stem. (2012).
J. Biol. Chem. 268, 10312–10323 (1993). 36. Trapnell, C. et al. Differential gene and transcript expression analysis of
16. Brooks, E.M., Sheflin, L.G. & Spaulding, S.W. Secondary structure in the 3′ RNA-seq experiments with TopHat and Cufflinks. Nat. Protoc. 7, 562–578
UTR of EGF and the choice of reverse transcriptases affect the detection (2012).

nature protocols | VOL.9 NO.1 | 2014 | 181

View publication stats