Download Profiling DNA methylome landscapes of mammalian cells with

protocol
Profiling DNA methylome landscapes of mammalian
cells with single-cell reduced-representation
bisulfite sequencing
Hongshan Guo1,5, Ping Zhu1,2,5, Fan Guo1, Xianlong Li1, Xinglong Wu1,2, Xiaoying Fan1, Lu Wen1,3,4 &
Fuchou Tang1,3,4
1Biodynamic Optical Imaging Center, College of
Life Sciences, Peking University, Beijing, China. 2Peking-Tsinghua Center for Life Sciences, Peking University, Beijing,
China. 3Ministry of Education Key Laboratory of Cell Proliferation and Differentiation, College of Life Sciences, Peking University, Beijing, China. 4Center for Molecular
and Translational Medicine, Peking University Health Science Center, Beijing, China. 5These authors contributed equally to this work. Correspondence should be
addressed to L.W. ([email protected]) or F.T. ([email protected]).
© 2015 Nature America, Inc. All rights reserved.
Published online 2 April 2015; doi:10.1038/nprot.2015.039
The heterogeneity of DNA methylation within a population of cells necessitates DNA methylome profiling at single-cell resolution.
Recently, we developed a single-cell reduced-representation bisulfite sequencing (scRRBS) technique in which we modified
the original RRBS method by integrating all the experimental steps before PCR amplification into a single-tube reaction. These
modifications enable scRRBS to provide digitized methylation information on ~1 million CpG sites within an individual diploid
mouse or human cell at single-base resolution. Compared with the single-cell bisulfite sequencing (scBS) technique, scRRBS covers
fewer CpG sites, but it provides better coverage for CpG islands (CGIs), which are likely to be the most informative elements for
DNA methylation. The entire procedure takes ~3 weeks, and it requires strong molecular biology skills.
INTRODUCTION
DNA methylation of cytosine is a well-known epigenetic modification that is involved in gene expression regulation1. In mammals, DNA methylation is relatively stable in differentiated cells;
by contrast, global demethylation and remethylation occur
during early embryonic development and primordial germ cell
development2–17. Whole-genome bisulfite sequencing provides
a comprehensive view of the DNA methylome, but it is very
expensive owing to deep sequencing of the entire genome18–25.
An alternative and cost-efficient technique is RRBS26–28, in which
genomic DNA is first digested with a restriction endonuclease
(usually MspI) and then size-selected to enrich for CpG-dense
regions. RRBS provides comprehensive DNA methylation information about the genome. By sequencing ~10% of the mouse or
human genome, RRBS can reproducibly cover a large proportion
of the informative CpG sites in the genome, including >70% of
promoters, >80% of CGIs and a large number of CGI shores,
enhancers, exons, 3′ untranslated regions (UTRs) and repetitive
elements27,29. However, conventional RRBS techniques require
nanogram amounts of genomic DNA as starting material27,30,31,
and they are not applicable to single cells.
Development of the procedure
We recently developed scRRBS for profiling DNA methylation
at the single-cell level10,17. In a typical mammalian cell, there
are only two copies of DNA molecules, and each comes from
one of the two parents. If aiming to profile the DNA methylome of a single cell, it is crucial to avoid DNA loss as far as
possible, as all DNA fragments contain irreplaceable information. DNA loss is inevitable during purification steps before PCR
amplification, and there are five such steps in the standard RRBS
method: (i) genomic DNA purification; (ii) restriction enzyme
digestion; (iii) end repair and dA tailing (end repair/dA tailing);
(iv) adapter ligation; and (v) bisulfite conversion. A more recent
gel-free RRBS protocol reduces the number of purification steps,
but it still requires three purifications after the end repair/dA
tailing, the adapter ligation and the bisulfite conversion. In the
scRRBS method, we integrated all the five steps into a singletube reaction so that DNA purification does not occur until
the completion of the bisulfite conversion17 (Fig. 1). To achieve
this, the buffer system and the reaction volumes were modified
to preserve the different enzyme activities at each stage of the
one-tube reaction.
When starting with a single mouse diploid cell, scRRBS covers,
on average, 1 million CpG dinucleotides; this accounts for ~40%
of the CpG sites that can be recovered by standard RRBS using
thousands of cells (on average, 2.5 million CpG dinucleotides).
Importantly, ~70% of CGIs in the mouse genome can be captured. The coverage increases with the number of starting cells,
and it reaches a plateau at five cells that corresponds to 60% of
the coverage achieved using standard RRBS, suggesting that the
single-tube approach to some extent affects the coverage that can
be achieved from the standard RRBS.
Overview of the procedure
The scRRBS approach starts with cell picking and lysis in 5 µl
of lysis buffer, which contains protease to release the genomic
DNA. The next three steps, including the MspI digestion, the
end repair/dA tailing, and the adapter ligation, are accomplished
by adding the corresponding reaction components sequentially,
and the enzymes in the preceding reactions are inactivated by
heating; Tango buffer is used for all the reactions. The ligation
is performed by overnight incubation using the premethylated
sequencing adapters and highly concentrated T4 DNA ligase. The
ligated DNA fragments are directly processed until bisulfite conversion, and only after this step is the DNA purified in the presence of carrier tRNA. Next, the DNA is PCR-amplified and the
fragments between 200 and 700 bp are gel-selected and purified
as the final library for sequencing.
nature protocols | VOL.10 NO.5 | 2015 | 645
protocol
CCGG
GGCC
CCGG
GGCC
Picking a single cell
(Steps 1–9)
Mspl digestion
(Steps 10–12)
End-repair/dA-tailing
adapter ligation
(Steps 13–19)
U
Single-tube reaction
U
U
Bisulfite conversion
(Steps 20–23)
U
PCR amplification
(Steps 24–41)
© 2015 Nature America, Inc. All rights reserved.
A
T
G
High-throughput
sequencing
(Steps 42–44)
C
Figure 1 | Flowchart of the experimental procedures of the scRRBS
technique. Notably, we integrated cell lysis, MspI digestion, end repair/dA
tailing, adapter ligation and bisulfite treatment into a single-tube reaction
to avoid unnecessary DNA loss.
Comparison of scRRBS and single-cell bisulfite sequencing
Recently, Smallwood et al.32 reported an alternative scBS technique. The authors used the single-tube reaction strategy to
improve a previous protocol that was based on bisulfite conversion followed by random priming33. When starting with a single mouse diploid cell, scBS covers, on average, 3.7 million CpG
dinucleotides. Although the overall coverage of scBS is higher
than that of the scRRBS, it has two limits. First, although scBS
includes more CpG-sparse regions, its coverage of CGI is lower
than scRRBS. Second, scBS profiles the genome in a relatively random and less consistent manner, meaning that there is less overlap
between the individual CpG sites covered in different single cells.
Thus, the two approaches provide complementary information,
and the choice of which method to use should depend on the aim
of specific studies.
Applications of scRRBS
The advantage of the scRRBS method is its applicability to subnanogram levels of DNA as starting material, down to a single
cell. This is particularly useful when the starting materials are
very limited and precious, such as mammalian early embryos and
primordial germ cells. It also enables the heterogeneity of DNA
methylomes among individual cells to be studied, which may have
important roles in biological processes such as cell differentiation,
memory formation and oncogenesis. As examples, we have successfully applied scRRBS to individual mouse sperm cells, and we
detected each recovered CpG site as either fully unmethylated or
fully methylated, which is expected because sperm cells contain
only a single copy of the haploid genome17. We have also applied
this approach to individual human and mouse ­ pronuclei isolated from zygotes, and we observed global DNA ­demethylation
646 | VOL.10 NO.5 | 2015 | nature protocols
dynamics10. In addition, application of the technique to mouse
embryonic cells using as few as 20 cells faithfully captured the
DNA methylation status of imprinted genomic regions at a
methylation level of 50% (ref. 34).
Limitations of single-cell methylation profiling techniques
The first limitation of the scRRBS technique stems from the design
of the RRBS method; that is, although it captures a large portion
of CGIs and promoters, it provides representative, but lower,
coverage of CpG-sparse regions such as enhancers 27. It should
be noted that the scBS technique is also biased toward CpG-rich
genomic regions32, and there are currently no single-cell DNA
methylation profiling techniques that provide whole-genome
coverage. Second, both the scRRBS and the scBS techniques face
the issue of limited overlapping coverage between individual cells.
As mentioned above, the scBS approach suffers more from this
issue. Third, neither approach can discriminate between DNA
methylation (5mC) and hydroxymethylation (5hmC), and the
detected methylation information is in fact the sum of 5mC
and 5hmC. Fourth, neither technique has been adapted to highthroughput platforms such as the microfluidic chip. Fifth, the
mapping efficiencies of both scRRBS and scBS are relatively low
(~25% on average) when starting with a single cell. For scRRBS,
the mapping efficiency increases with the number of starting
cells, and it reaches a plateau of 50%, which is close to that of
standard RRBS17.
Experimental design
Starting material. We have successfully applied this protocol to a
variety of mammalian cells, including human and mouse embryonic stem cells (mESCs), oocytes, sperm, early preimplantation
blastomeres and cancer cells. The method is likely to be applicable to most mouse and human cell types, regardless of whether
they are isolated from tissues or cell culture. However, it is recommended to select the healthiest-looking cells with good
morphology to avoid potential genomic DNA degradation
before cell lysis. We have not yet tested this method on any nonmammalian cells. Both single cells and a small number of cells
(several to hundreds of cells, such as cells isolated from a single
blastocyst) can serve as the starting material for scRRBS. If you
are aiming to profile the DNA methylation of a population with
a limited number of cells, it will be more efficient to pool the cells
for analysis and perform several biological replicates. This analysis provides information about the average methylation level for
each covered CpG site within the population, which is similar to
conventional RRBS. Alternatively, if you wish to study cell heterogeneity with respect to DNA methylation within a population, or
if there are only one or a few highly heterogeneous cells within the
biological sample, individual cells must be analyzed; in this case,
dozens (10–100) of single cells are required to obtain accurate
measurement of DNA methylation for a certain cell type. This
analysis detects the methylation status of covered CpG sites in
a digitized manner, as each individual mammalian cell contains
only two copies of DNA molecules, or one copy of DNA molecule
when a haploid germ cell is analyzed.
Controls. It is important to avoid any DNA contamination in
scRRBS experiments. All reagents and consumables used in this
protocol should be free of exogenous DNA, and ‘picking-bufferonly’ controls (i.e., only pick the PBS-BSA buffer but no cells
protocol
into the lysis buffer) are always essential for evaluating possible
contamination during the whole experimental process.
may be a consequence of the low sequence complexity of
scRRBS libraries.
Enzyme choice. The restriction endonuclease MspI (C↓CGG) is
commonly used in RRBS, and it is the focus of this scRRBS protocol. MspI digestion is strongly biased toward CGIs and promoters;
however, other restriction endonucleases such as TaqI (T↓CGA)
could be chosen to capture different genomic regions. In addition, digestion using two or more endonucleases may increase
the genomic coverage.
Data analysis. Bioinformatics analysis methods for DNA
methylation profiling data of whole-genome bisulfite sequencing
and RRBS have been reported previously, which are designed to
map bisulfite-converted sequence reads to a reference genome
and to determine cytosine methylation states at single-base
resolution10,17. Customized scripts written in Perl or Python
can be applied for discarding low-quality reads, trimming
adapter sequences and removing additional bases that are artificially introduced during the end-repair steps. Alternatively,
the Trim Galore tool can be used (http://www.bioinformatics.
babraham.ac.uk/projects/trim_galore/). Bioinformatic tools such
as Bismark35, BSMAP36 or BS Seeker37 can be applied for the
alignment of the bisulfite-converted sequences and downstream
analysis such as visualization of the data, comparison between
samples and identification of differentially methylated regions.
The Bismark program works well in our hands.
© 2015 Nature America, Inc. All rights reserved.
Sequencing. Different sequencing platforms can be used, including the Illumina HiSeq 2000 and HiSeq 2500 platforms we use.
We generally sequence each scRRBS library to ~10 million
100-bp paired-end reads. Increasing the sequencing depth to
20 million reads will only slightly raise the genomic coverage.
In our experience, a cluster density of 80% of the regular or
mixing scRRBS libraries with other non-bisulfite-sequencing
libraries will increase the yield and quality of the reads, which
MATERIALS
REAGENTS
 CRITICAL All reagents used in this experiment must be free of nucleases.
! CAUTION All samples from animals or humans should be processed
according to the guidelines of the local government or institution.
• Cell line or single cells of interest. In the PROCEDURE, we use as an example
mESCs maintained in conventional DMEM, containing 20% (vol/vol) FBS
in the presence of leukemia inhibitory factor (LIF)
• Tris-EDTA solution (100×; Sigma-Aldrich, cat. no. T9285-100ML)
• Triton X-100 (Sigma-Aldrich, cat. no. T8787-50ML) ! CAUTION Triton
X-100 is harmful if swallowed or inhaled, and it causes irritation to the skin,
the eyes and the respiratory tract. Handle it using appropriate equipment.
 CRITICAL Triton X-100 should be stored at room temperature (20–25 °C)
for at most 1 year.
• Potassium chloride solution (KCl; 1 M; Sigma-Aldrich,
cat. no. 60142-100ML-F)
• Glycerol (Sigma-Aldrich, cat. no. G5516-100ML)
• Protease (7.5 Anson units (AU); Qiagen, cat. no. 19155)
• Nuclease-free water (Ambion, cat. no. AM9932)
• MspI (10 U/µl; Thermo Scientific, cat. no. ER0541)
• Tango buffer (10×; Thermo Scientific, cat. no. BY5)
• λ-DNA (dam–, dcm–; 0.3 µg/µl, Thermo Scientific, cat. no. SD0021)
• Klenow fragment (exo–, 5 U/µl; Thermo Scientific, cat. no. EP0422)
• dATP (100 mM; New England BioLabs, cat. no. N0440S)
• dCTP (100 mM; New England BioLabs, cat. no. N0441S)
• dGTP (100 mM; New England BioLabs, cat. no. N0442S)
• T4 DNA ligase (highly concentrated; 30 Weiss U/µl; Thermo Scientific,
cat. no. EL0013)
• ATP (100 mM; Thermo Scientific, cat. no. R0441)
• Adapters, part of the Illumina TruSeq DNA sample preparation kits
(Illumina, cat. no. FC-121-2001)  CRITICAL Illumina TruSeq DNA
adapters are highly recommended. If they are not available, ensure that the
cytosines (Cs) of the ordered oligos are methyl group–modified (mCs) and
double HPLC-purified. The oligo sequences of these indexed adapters are
listed in Table 1.
• MethyCode bisulfite conversion kit (50 reactions; Thermo Scientific,
cat. no. MECOV-50)
• tRNA (Roche, cat. no. 10109517001)
• PfuTurbo Cx hotstart DNA polymerase (2.5 U/µl; Agilent Technologies,
cat. no. 600412)
• dNTP mix (10 mM each; Clontech, cat. no. D4030RA)
• Agencourt AMPure XP beads (Beckman Coulter; cat. no. A63881)
• Phusion high-fidelity (HF) PCR master mix with HF buffer (New England
BioLabs, cat. no. M0531S)
• Qubit dsDNA high-sensitivity (HS) assay kit (Thermo Scientific,
cat. no. Q32854)
• SYBR Gold nucleic acid gel stain (10,000×; Thermo Scientific,
cat. no. S-11494) ! CAUTION This reagent is harmful if swallowed or
inhaled, and it causes irritation to the skin, the eyes and the respiratory
tract. Handle it using appropriate equipment.
• Glycerol (Sigma-Aldrich, cat. no. G5516-100ML)
• QIAquick PCR purification kit (Qiagen, cat. no. 28106)
• Ethanol (Sigma-Aldrich, cat. no. E7023) ! CAUTION Ethanol is flammable.
Handle it using appropriate equipment.
• Primers for two rounds of PCR amplification (these oligos can be
synthesized and HPLC-purified from IDT): QP1: 5′-AATGATACGGCGAC
CACCGA-3′ and QP2: 5′-CAAGCAGAAGACGGCATACGA-3′
• High-sensitivity next-generation sequencing (NGS) fragment analysis kit
(Advanced Analytical Technologies, cat. no. DNF-486-0500)
• 40% Acryl/Bis (29:1; Amresco, cat. no. 0311-1L) ! CAUTION This compound
is harmful if swallowed or inhaled, and it causes irritation to the skin, the
eyes and the respiratory tract. Handle it using appropriate equipment.
• TBE buffer (5×; Amresco, cat. no. J885-1L)
• Ammonium persulfate (AP; Sigma-Aldrich, cat. no. A3678-100G)
! CAUTION AP is harmful if swallowed or inhaled, and it causes irritation
to the skin, the eyes and the respiratory tract. Handle it using appropriate
equipment.
• N,N,N′,N′-tetramethylethylenediamine (TEMED; Sigma-Aldrich,
cat. no. T9281-100ML) ! CAUTION TEMED is harmful if swallowed or
inhaled, and it causes irritation to the skin, the eyes and the respiratory
tract. Handle it using appropriate equipment.
• Ammonium acetate (Amresco, cat. no. 0103-500G)
• Magnesium acetate tetrahydrate (Amresco, cat. no. 0131-500G)
• SDS (Amresco, cat. no. S0227-500G)
• EDTA (Amresco, cat. no. 0322-1KG)
• PBS buffer (pH 7.2; 1×; Gibco, cat. no. 14249-95)
• Leukemia inhibitory factor (LIF; 10 million units/1 ml, Millipore,
cat. no. ESG1107)
• DMEM: nutrient mixture F-12 (DMEM/F-12; Gibco, cat. no. 11320-033)
• FBS (Gibco, cat. no. 12483-020)
• Trypsin-EDTA, 0.05% (wt/vol) (1×; Gibco, cat. no. 25300-062)
• Albumin, acetylated from bovine serum (Ac-BSA; Sigma-Aldrich,
cat. no. B8894)
nature protocols | VOL.10 NO.5 | 2015 | 647
protocol
Table 1 | Oligo sequences of adapters.
© 2015 Nature America, Inc. All rights reserved.
Oligo name
Sequence (all sequences read 5′–3′)
TruSeq universal adapter
AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATC*T
TruSeq adapter (index 1)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACATCACGATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 2)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACCGATGTATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 3)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACTTAGGCATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 4)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACTGACCAATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 5)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACACAGTGATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 6)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACGCCAATATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 7)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACCAGATCATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 8)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACACTTGAATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 9)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACGATCAGATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 10)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACTAGCTTATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 11)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACGGCTACATCTCGTATGCCGTCTTCTGCTTG
TruSeq adapter (index 12)
GATCGGAAGAGCACACGTCTGAACTCCAGTCACCTTGTAATCTCGTATGCCGTCTTCTGCTTG
Index sequences are six bases, underlined.
The linkage of the last two bases of the TruSeq universal adapter oligo should be phosphorothiated to avoid nuclease cleavage of the T overhang.
The TruSeq adapter oligo (indexes 1–12) should be 5′-terminal phosphorylated for ligation of the inserted DNA fragments.
• DNA loading buffer (6×; Clontech, cat. no. 9156)
• GeneRuler low-range DNA ladder (ready-to-use; Thermo Scientific,
cat. no. SM1193)
• DTT (Thermo Scientific, cat. no. R0861)
• KAPA HiFi HotStart ReadyMix (2×; Kapa Biosystems, cat. no. KK2602)
EQUIPMENT
• PCR tubes, 0.5-ml thin-walled with flat cap (Axygen, cat. no. PCR-05-C)
• PCR tubes, 0.2-ml thin-walled with flat cap (Axygen, cat. no. PCR-02-C)
• 1,000-µl universal fit filter tips (Axygen, cat. no. TF-1000-R-S)
• 200-µl universal fit filter tips (Axygen, cat. no. TF-200-R-S)
• 100-µl universal fit filter tips (Axygen, cat. no. TF-100-R-S)
• 10-µl universal fit filter tips (Axygen, cat. no. TF-300-R-S)
• DNA LoBind tubes, 1.5 ml (Eppendorf, cat. no. 022431021)
• DNA LoBind tubes, 2.0 ml (Eppendorf, cat. no. 022431048)
• Cell culture–treated dishes (60 mm; Thermo Scientific, cat. no. 130181)
• Falcon conical centrifuge tubes (15 ml; Corning Life Sciences,
cat. no. 14-959-49D)
• Magnetic rack (Diagenode, cat. no. kch-816-001)
• 10 µM filter spin (Zymo, cat. no. C1007-50)
• Dark Reader transilluminator (Clare Chemical Research, cat. no. DR88X)
• Flaming/Brown micropipette puller (Sutter Instrument, P-1000)
• Refrigerated microcentrifuge (5415R; Eppendorf, cat. no. 022621408)
• Nonrefrigerated microcentrifuge (Heraeus Pico17; Thermo Scientific,
cat. no. 75002410)
• Thermomixer (Eppendorf, cat. no. 5355 000.011)
• Qubit 2.0 fluorometer (Thermo Scientific, cat. no. Q32866)
• Borosilicate glass capillary (Sutter Instrument, cat. no. B100-58-10)
! CAUTION Be careful with the glass capillary; handle it with appropriate
protection.
• Aspirator tube assemblies for calibrated microcapillary pipettes
(Sigma-Aldrich, cat. no. A5177)
• HL-2000 HybriLinker (UVP, cat. no. 95-0031-02)
• Vortex (Scientific industries, cat. no. SI-0246)
• Thermocycler with a 96-well block (TProfessional Gradient 96;
Biometra, cat. no. 846-070-801)
• 16-tube magnetic rack (Diagenode, cat. no. B04000001 (kch-816-001))
648 | VOL.10 NO.5 | 2015 | nature protocols
• ChemiDoc XRS+ System with Image Lab Software (computer-connected,
Bio-Rad, cat. no. 170-8265)
• 4-gel vertical electrophoresis system (Bio-Rad, cat. no.165-8006)
• PowerPac basic power supply (Bio-Rad, cat. no. 164-5050)
• Transference decoloring shaker (Qilinbeier, cat. no. TS-8)
• Fragment Analyzer (Advanced Analytical Technologies, cat. no. FA12)
Software
• Perl. Download freely from http://www.perl.org/
• Trim_galore (version 0.3.7). Freely available at http://www.bioinformatics.
babraham.ac.uk/projects/trim_galore/
• Bowtie (version 1, ref. 38). Download from http://bowtie-bio.sourceforge.
net/index.shtml
• Bismark (version 0.7.6, ref. 35). Download from http://www.bioinformatics.
babraham.ac.uk/projects/bismark/. Other similar tools can also be used
• Picard toolkit. Download from http://broadinstitute.github.io/picard/.
Other similar tools can also be used
• Samtools39. Download from http://samtools.sourceforge.net/
• Java. Download from https://www.java.com/
• Custom scripts. An archive of all customized scripts used in this protocol is
available as Supplementary Data. Refer to Table 2 for details of each script
REAGENT SETUP
QP1, QP2 PCR primers Upon arrival, dissolve the forward (QP1) and
reverse (QP2) primers using nuclease-free water to 100 µM as stock solutions
and dilute them to 10 µM as working solutions. Both the stock and working
solutions are stable for 1 year if frozen at −80 °C.
Diffusion buffer Diffusion buffer is 500 mM ammonium acetate, 10 mM
magnesium acetate, 1 mM EDTA (pH 8.0) and 1% (wt/vol) SDS. Diffusion
buffer should be stored at room temperature for no longer than 6 months.
PBS-BSA buffer Dilute Ac-BSA stock solution (20 mg/ml) with 1×PBS to
1 mg/ml as working solution and prepare aliquots in 1.5-ml Eppendorf DNA
LoBind tubes. The working solution can be stored at −80 °C for at least
6 months. BSA is essential to eliminate the nonspecific binding of single cells
to the surface of the reaction tubes or glass capillaries.
Protease, 20 mg/ml Soak 20 mg of Qiagen lyophilized protease in 1 ml of
50% (vol/vol) glycerol and dissolve it for at least 30 min at 4 °C; next, prepare
protocol
Table 2 | Customized scripts for data analysis.
© 2015 Nature America, Inc. All rights reserved.
Script name
Functions
Used in steps
01. trimQC.sh
Quality control to filter out low-quality reads and to trim adapter sequences
45
02. Bismark_Genome_Preparation.sh
Index the reference genome
46
03. Alignment.sh
Read mapping using Bismark (version 0.7.6)
46
04. sort_pileup.sh
Sort mapped SAM files and generate pileup files
48
05. SingleC_ MetLevel.sh
DNA methylation level calculation of each covered cytosine
49
aliquots in 1.5-ml Eppendorf DNA LoBind tubes when fully dissolved. Store
the aliquots at −20 °C for no more than 6 months.
tRNA carrier Soak 10 mg of tRNA lyophilizate into 1 ml of nuclease-free
water to allow it to dissolve for at least 10 min, and then dilute it with
nuclease-free water to 10 ng/µl as the working concentration; divide the
solution into 1.5-ml Eppendorf DNA LoBind tubes and store the aliquots
at −80 °C for no more than 6 months.
End-repair dNTP mix Mix and dilute the dATP, dGTP and dCTP stock
solutions (100 mM each) to 1 mM, 0.1 mM and 0.1 mM, respectively. Divide
the solutions into aliquots in 1.5-ml Eppendorf DNA LoBind tubes and store
them at −20 °C for no more than 6 months.
Maintenance of mESCs mESCs are maintained without feeders in the presence
of LIF in DMEM containing 20% FBS for routine passage without modification.
Unmethylated λ-DNA spike-in Unmethylated λ-DNA (0.3 µg/µl) should
be diluted with nuclease-free water and quantified by a Qubit fluorometer
and the Qubit dsDNA HS assay kit to ~60 fg/µl. This spike-in stock solution
should be stored at a concentration >1 ng/µl at −20 °C for no longer than
3 months and quantified and diluted just before use.
CT conversion reagent Add 850 µl of nuclease-free water, 50 µl of
resuspension buffer (part of the MethylCode bisulfite conversion kit) and
300 µl of dilution buffer (part of the MethylCode bisulfite conversion kit)
directly to the CT conversion powder (also part of the same kit). Mix the
solution by brief intermittent vortexing until the solution becomes clear.
Always keep the dissolved CT conversion reagent in the dark, and then store
it for up to 2 weeks at −20 °C. Avoid repeated freeze/thaw cycles.
Native polyacrylamide TBE gel, 12% (wt/vol) Add 7.5 ml of 30% (wt/vol)
acrylamide, 5 ml of 5× TBE buffer, 180 µl of 10% (wt/vol) AP, 16 µl of
TEMED and nuclease-free water up to 25 ml. Mix the solution well by
vortexing, pipette it into the gap between the glass gel-casting plates and
insert the ten-well combs. Let the gel stand at room temperature for at
least 1 h to ensure complete polymerization.  CRITICAL Gels can be stored
at 4 °C in sealed packs for up to 2 weeks.
PROCEDURE
Cell culture, single-cell isolation and lysis ● TIMING 4–5 d
 CRITICAL Incubations before bisulfite conversion (i.e., before Step 18) should be performed at a temperature no higher
than 80 °C; higher temperatures increase the risk of denaturing the double-stranded genomic DNA.
 CRITICAL The nuclease-free water and all of the tubes used in this protocol, including the 0.2-ml thin-walled PCR tubes
and the 1.5-ml Eppendorf DNA LoBind tubes, should be UV-sterilized for at least 5 min to prevent contamination.
 CRITICAL All steps performed with a PCR thermocycler should be carried out on an instrument with a heated lid to avoid
liquid evaporation from the reaction tubes.
1| Aspirate the medium from mESCs maintained in 60-mm cell culture dishes (at 80% confluency) and wash the entire
surface area with 1× PBS. Aspirate the PBS and add sufficient 0.05% (wt/vol) trypsin to cover the cell growth area.
Return the dishes to a 37 °C incubator and incubate them for 5 min.
 CRITICAL STEP One 60-mm dish at 80% confluency provides sufficient mESCs for scRRBS.
2| After the cells have detached, add 5 ml of DMEM with 20% (vol/vol) FBS to inactivate the trypsin, gently pipette the
suspension to obtain single cells and collect the cell suspension into a 15-ml Falcon conical tube.
3| Centrifuge the mixture at 300g for 5 min at room temperature and aspirate the supernatant.
4| Wash the cell pellet with 1 ml of PBS-BSA, centrifuge it at 300g for 5 min at room temperature and aspirate the
supernatant. Repeat this step once more, and finally suspend the pellet in 20–50 µl of PBS-BSA solution.
 CRITICAL STEP Always use PBS-BSA instead of PBS in this step to minimize nonspecific binding to the reaction tubes.
nature protocols | VOL.10 NO.5 | 2015 | 649
protocol
5| Prepare the cell lysis buffer as set out in the following table. Mix it well by vortexing, centrifuge the mixture at 9,000g
for 1 min at 4 °C, and then divide the mixture into 0.2-ml thin-walled PCR tubes (4.8 µl in each tube).
Component
Amount per sample (ml)
Final concentration
Tris-EDTA (1 M Tris, 0.1 M EDTA)
0.1
20 mM Tris, 2 mM EDTA
1 M KCl
0.1
20 mM
10% (vol/vol) Triton X-100
0.15
0.3% (vol/vol)
20 mg/ml protease
0.25
1 mg/ml
Nuclease-free water
up to 4.8
—
© 2015 Nature America, Inc. All rights reserved.
 CRITICAL STEP Always include at least one ‘picking-buffer-only’ control when preparing the lysis buffer.
 CRITICAL STEP Be sure to use protease instead of proteinase K when preparing lysis buffer, as protease can be
heat-inactivated at a much lower temperature than proteinase K.
 CRITICAL STEP If you are starting from single sperm cells, add 10 mM DTT to the lysis buffer to decondense the sperm nucleus.
6| Carefully transfer single cells from Step 4 into the tubes prepared in Step 5, taking care to avoid forming bubbles.
We have found mouth pipetting to be the most accurate way to achieve this40–42; however, alternative approaches include
fluorescence-activated cell sorting (FACS)43, laser-assisted microdissection44 or microfluidic platforms45,46.
 CRITICAL STEP The volume of carry-over PBS-BSA during mouth pipetting of single cells should be no >0.2 µl.
 CRITICAL STEP It is essential to include a ‘picking-buffer-only’ (cell lysis buffer without cells) control to evaluate possible
contamination during subsequent procedures.
! CAUTION Be careful with the glass capillary; handle it with appropriate protection.
! CAUTION Not all institutions permit mouth pipetting for manipulation of single cells, especially when working on certain
types of sample (e.g., primary human tissues or some infectious samples). Alternative approaches that comply with
institutional, local and governmental regulations should be adopted.
7| Centrifuge the tubes at 9,000g for at least 1 min at 4 °C to ensure that the cell is completely seeded into the lysis buffer.
 CRITICAL STEP Do not vortex the mixture, as the genomic DNA is easily fragmented.
? TROUBLESHOOTING
8| Lyse the cells at 50 °C for 3 h, and then incubate them at 75 °C for 30 min to inactivate the protease.
 CRITICAL STEP This step should be performed in a PCR thermocycler rather than in a heat block or a water bath because of
its capacity for stable temperature control.
9| Centrifuge the tubes at 9,000g for 1 min at 4 °C and immediately place the tubes on ice.
? TROUBLESHOOTING
 PAUSE POINT The lysate can be frozen at −80 °C for no longer than 3 months. However, we strongly recommend proceeding
to the next step immediately.
MspI digestion ● TIMING 3–4 h
10| Prepare the MspI digestion mixture, as described in the following table. Mix it well by vortexing, and then centrifuge the
mixture at 9,000g for 1 min at 4 °C.
Component
Amount per reaction (ml)
Final amount
10 U/µl MspI
0.9
9U
10× Tango buffer
1.8
1×
1
60 fg
up to 13
—
Unmethylated λ-DNA (60 fg/µl)
Nuclease-free water
 CRITICAL STEP Always ensure that the λ spike-in accounts for ~1% (mass/mass) of the total genomic DNA of every picked
single cell. For example, if you are starting with a single diploid mammalian cell, add 1 µl (~60 fg) of unmethylated λ-DNA to
the digestion mixture.
650 | VOL.10 NO.5 | 2015 | nature protocols
protocol
11| Add 13 µl of MspI digestion mixture from Step 10 to each tube from Step 9 so that the total volume of each sample is
18 µl: 4.8 µl of lysis buffer, 0.2 µl of carry-over PBS-BSA (single cell included) and 13 µl of MspI digestion mixture. Mix well
by vortexing the tubes, and centrifuge them at 9,000g for 1 min at 4 °C.
12| Incubate the reaction at 37 °C for 3 h, and then at 80 °C for 20 min to inactivate the restriction enzyme. Centrifuge the
tubes at 9,000g for 1 min at 4 °C, and place the tubes on ice immediately. Proceed to the next step immediately.
End-repair/dA-tailing reaction ● TIMING 1–2 h
13| Prepare the end-repair/dA-tailing mixture, as described in the following table. Mix well by vortexing, and then
centrifuge the mixture at 9,000g for 1 min at 4 °C.
Component
Amount per reaction (ml)
Final amount
1
5U
10× Tango buffer
0.2
1×
End-repair dNTP mix
0.8
40 µM dATP, 4 µM dCTP, 4 µM dGTP
© 2015 Nature America, Inc. All rights reserved.
5 U/µl Klenow Fragment,
exo–
14| Add 2 µl of end-repair/dA-tailing mixture from Step 13 to each tube from Step 12 so that the total volume of each
sample is 20 µl: 18 µl of MspI digested reaction and 2 µl of end-repair/dA-tailing mixture. Mix it well by vortexing the
tubes, and centrifuge the mixture at 9,000g for 1 min at 4 °C.
15| Incubate the reaction at 37 °C for 40 min, and then at 75 °C for 15 min to inactivate the enzyme. Centrifuge the tubes
at 9,000g for 1 min at 4 °C, and place the tubes on ice immediately.
Adapter ligation ● TIMING 9–10 h
16| Prepare the adapter ligation mixture as described in the following table. Mix it well by gentle inversion, and centrifuge
the mixture at 9,000g for 1 min at 4 °C.
Component
Amount per reaction (ml)
Final amount
1
30 Weiss U
10× Tango buffer
0.5
1×
100 mM ATP
0.25
1 mM
1:20 diluted adapter
1
—
Nuclease-free water
up to 5
—
30 Weiss U/µl T4 DNA ligase
 CRITICAL STEP Do not vortex the mixture, as vigorous vortexing may damage the Y-shaped adapters.
17| Add 5 µl of ligation mixture from Step 16 to each tube from Step 15 so that the total volume of each sample is 25 µl:
20 µl of end-repaired and dA-tailed reaction and 5 µl of ligation mixture. Mix the reaction well by gentle inversion, and
centrifuge the mixture at 9,000g for 1 min at 4 °C.
18| Incubate the reaction at 16 °C for 30 min, followed by 4 °C overnight (at least 8 h).
19| Heat-inactivate the enzyme reaction at 65 °C for 20 min. Next, centrifuge the tubes at 9,000g for 1 min at 4 °C and
immediately place the tubes on ice.
 PAUSE POINT The reaction may be stored at −80 °C for no longer than 3 months if necessary.
Bisulfite conversion ● TIMING 3–4 h
20| Use all of the 25-µl ligation mixture from Step 19 to perform the bisulfite treatment. We use the MethylCode bisulfite conversion
kit, according to the manufacturer’s instructions, as summarized in Steps 20–23: add 125 µl of well-dissolved CT conversion reagent
(Reagent Setup) to the 25-µl ligation mixture, mix well by vortexing, and then centrifuge the mixture at 9,000g for 1 min at 4 °C.
21| Perform the bisulfite conversion under the following conditions on the thermocycler: first at 98 °C for 10 min, then at
64 °C for 2.5 h and finally at 4 °C for temporary storage.
nature protocols | VOL.10 NO.5 | 2015 | 651
protocol
22| Purify the converted DNA by using the MethylCode bisulfite conversion kit according to the manufacturer’s instructions.
1 µl of tRNA (10 ng/µl) should be added to the supplied DNA binding buffer to act as a protective carrier.
23| Elute the converted DNA with 30 µl of preheated (50 °C) elution buffer (included in the MethylCode bisulfite conversion
kit) according to the kit instructions.
 PAUSE POINT The eluted DNA may be stored at −80 °C for at least 6 months.
First-round PCR enrichment ● TIMING 3–4 h
24| Prepare the first-round PCR mixture as described in the following table. Mix it well by vortexing, and centrifuge the
mixture at 9,000g for 1 min at 4 °C.
Component
Amount per reaction (ml)
Final amount
5
1×
0.4
1U
1
200 µM (each)
10 µM QP1 primer
1.5
300 nM
10 µM QP2 primer
1.5
300 nM
up to 20
—
10× PCR Pfu Cx Buffer
2.5 U/µl Pfu Turbo Cx hotstart DNA polymerase
© 2015 Nature America, Inc. All rights reserved.
dNTP mix (10 mM each)
Nuclease-free water
 CRITICAL STEP PfuTurbo Cx hotstart DNA polymerase must be used in this step because this enzyme is resistant to uracil
stalling, thus ensuring that both methylated and unmethylated DNA fragments will be amplified with similar efficiencies.
? TROUBLESHOOTING
25| Add 20 µl of the first-round PCR mixture from Step 24 to the tube from Step 23 so that the total volume of the PCR mix
is 50 µl: 30 µl of eluted converted DNA and 20 µl of first-round PCR mixture. Mix it well by vortexing, and then centrifuge
the mixture at 9,000g for 1 min at 4 °C.
26| Perform 25 cycles of PCR on the PCR thermocycler and run the following program:
Cycle number
1
2–26
Denature
Anneal
Extend
Hold
60 °C, 30 s
72 °C, 1 min
95 °C, 2 min
95 °C, 20 s
27
72 °C, 5 min
28
4 °C
27| After the PCR cycles, centrifuge the tubes at 9,000g for 1 min at 4 °C. Next, purify the PCRs twice via 1:1-fold AMPure
XP beads, as described in Box 1, and finally elute in 20 µl of nuclease-free water.
Second-round PCR enrichment ● TIMING 3–4 h
28| Prepare the second-round PCR mixture as described in the following table. Mix well by vortexing, and then centrifuge
the mixture at 9,000g for 1 min at 4 °C.
Component
Amount per reaction (ml)
Final amount
Phusion HF PCR master mix
with HF buffer (2×)
25
1×
10 µM QP1 primer
2.5
500 nM
10 µM QP2 primer
2.5
500 nM
652 | VOL.10 NO.5 | 2015 | nature protocols
protocol
Box 1 | DNA purification using AMPure XP beads ● TIMING 0.5–1 h
© 2015 Nature America, Inc. All rights reserved.
1. Remove the AMPure XP bead suspension from storage and vortex the AMPure XP beads until they are well dispersed. Let the bead
suspension stand for at least 30 min to bring the beads to room temperature.
2. Add an equal volume of well-dispersed AMPure XP bead suspension to the reaction tubes (e.g., 50 µl of suspension to 50 µl of the
reaction solution) and mix well by repeated pipetting.
3. Incubate the mixture at room temperature for 15 min, and then place the tubes on the magnetic rack for at least 5 min until the
solution becomes clear.
4. Remove and discard the supernatant carefully to avoid disturbing the beads.
5. Add 200 µl of freshly prepared 80% (vol/vol) ethanol and invert the tubes several times with a magnetic rack; carefully discard the
supernatant.
6. Repeat the wash step once more as described in Step 5.
7. Let the tubes stand at room temperature for at least 10 min with the lids open until the beads are dry.
8. Resuspend the dried beads with an appropriate volume (usually 20–50 µl) of nuclease-free water, mix well by repeated pipetting and
incubate the mixture at room temperature for at least 2 min.
9. Place the resuspended solution on the magnetic rack again, and then let the tubes stand for at least 5 min until the solution
becomes clear.
10. Transfer the supernatant to new nuclease-free tubes without disturbing the beads.
 CRITICAL STEP KAPA HiFi HotStart ReadyMix can be used as an alternative to Phusion HF PCR master mix with HF buffer,
if available.
? TROUBLESHOOTING
29| Add 30 µl of the second-round PCR mixture from Step 28 to the tube from Step 27 so that the total volume of the PCR
mixture is 50 µl: 20 µl of eluted DNA and 30 µl of second-round PCR mixture. Mix well by vortexing and centrifuge the
mixture again at 9,000g for 1 min at 4 °C.
30| Perform another 25 PCR cycles with the following thermocycler program:
Cycle number
1
2–26
27
28
Denature
Anneal
Extend
60 °C, 30 s
72 °C, 1 min
Hold
98 °C, 2 min
98 °C, 10 s
72 °C, 5 min
4 °C
31| After the PCR cycles, centrifuge the tubes at 9,000g for 1 min at 4 °C. Next, purify the PCRs once using 1:1-fold AMPure
XP beads, as described in Box 1; finally, elute in 20 µl of nuclease-free water.
Size selection of the amplified DNA fragments ● TIMING 9–10 h
32| Prepare the 12% (wt/vol) native polyacrylamide TBE gel (Reagent Setup) using the Bio-Rad electrophoresis system.
 CRITICAL STEP As more primer dimers are generated in the scRRBS protocol than in regular RRBS protocols, we recommend
the use of a gel-based approach to thoroughly remove the primer dimers.
33| Add 4 µl of 6× DNA loading buffer to the sample, mix it well by vortexing and spin down the mixture briefly.
34| Pour freshly prepared 1× TBE running buffer into the electrophoresis apparatus, and place the gel cassette into the
apparatus. Then, pull out the combs carefully to avoid destroying the wells.
35| Load 24 µl of well-mixed samples and 2 µl of low-range DNA ladder into different wells. Run the gels at 150 V for ~1 h,
stopping before the bromophenol blue loading dye runs out of the gel.
36| After electrophoresis, remove the gel cassette from the electrophoresis apparatus, peel the gel off the glass and stain it
with a 1:10,000 dilution of SYBR Gold nucleic acid gel stain in 1× TBE buffer for 10 min at room temperature in darkness on
a shaker.
nature protocols | VOL.10 NO.5 | 2015 | 653
protocol
37| Wash the gel twice with 1× TBE buffer to remove the stain dye, and then visualize the gel with a Dark Reader
transilluminator in darkness (Fig. 2a).
38| Excise the gel slices containing 200–700-bp DNA fragments using a clean disposable blade, and then transfer the gel
slices into 0.5-ml thin-walled PCR tubes.
! CAUTION Handle the blade with appropriate protection.
39| Recover DNA fragments from the gels using an appropriate method. We have found that we achieve maximum recovery
using the ‘crush and soak’ method (Box 2).
41| Purify the eluate twice using 1:1-fold AMPure XP beads, as described in Box 1, and finally elute in 20 µl of
nuclease-free water.
 CRITICAL STEP There will be some residual primer dimers even after the gel purification, necessitating this AMPure XP
beads purification step.
? TROUBLESHOOTING
 PAUSE POINT The final eluate could be stored at −20 °C for no longer than 3 months.
a
Marker
b
#1
#2
#3
9,760
Marker
9,725
Single mESC RRBS
LM
6,000
UM
9,625
3,000
2,000
1,500
1,200
1,000
900
800
700
9,600
600
9,700
700 bp
9,675
9,575
500
211
RFU
100 bp
225
9,650
200 bp
9,550
400
9,525
300
296
9,500
9,475
200
9,450
9,425
100
9,400
35
6,000
700
800
900
1,000
1,200
1,500
2,000
3,000
600
500
400
300
200
100
35
9,378
Size (bp)
c
9,820
9,800
Single mouse pronucleus RRBS
LM
3,000
2,000
1,500
1,200
1,000
900
800
700
9,600
263
227
9,700
9,650
600
213
500
9,550
400
300
372
9,500
9,450
200
9,400
100
35
Size (bp)
6,000
3,000
700
800
900
1,000
1,200
1,500
2,000
600
500
400
300
200
100
9,344
654 | VOL.10 NO.5 | 2015 | nature protocols
6,000
UM
9,750
35
Figure 2 | The size distribution of
the typical scRRBS libraries. (a) The
polyacrylamide TBE gel results of
three scRRBS libraries (before
gel-based size selection); the DNA is
smeared from 200 bp to ~5 kb, with
some detectable adapter dimers at
~120 bp. (b,c) The Fragment Analyzer
results of two scRRBS libraries
established from a single mESC (b)
and a single mouse pronucleus (c).
Typical DNA size distribution in
scRRBS libraries ranges from 160
to 350 bp, with visible peaks
corresponding to MspI fragments
for some repetitive elements.
RFU
© 2015 Nature America, Inc. All rights reserved.
40| Purify the eluate from Step 39 using the QIAquick PCR purification kit (or any equivalent kit) according to the
manufacturer’s protocol. Elute the DNA from the column using 50 µl of preheated EB buffer (part of the QIAquick PCR
purification kit).
protocol
Box 2 | ‘Crush and soak’ method ● TIMING 12–14 h
© 2015 Nature America, Inc. All rights reserved.
1. Use a needle from a 1-ml syringe to make several holes in the bottom of 0.5-ml thin-walled PCR tubes.
 CRITICAL STEP Handle the needle with appropriate protection.
2. Transfer the gel slices into 0.5-ml thin-walled PCR tubes, place the tubes into 1.5-ml Eppendorf DNA LoBind tubes and centrifuge
them at 13,000g for at least 1 min at room temperature until the gel slice collects in the bottom of the 1.5-ml tubes.
3. Add diffusion buffer (~350 µl, Reagent Setup) to the 1.5-ml tubes until all the gel debris is covered with buffer.
4. Incubate the slurry on a Thermomixer by shaking for 2–12 h at 50 °C. Longer incubations give better recoveries.
 CRITICAL STEP The slurry should be incubated for no less than 2 h.
5. Transfer the eluate and the gel debris to the top of a 10-µm filter spin, and then centrifuge the filter at 3,000g for 0.5–1 min at
room temperature to ensure that all of the eluate passes through the filter and flows to the bottom of the collection tubes.
6. Transfer the flow-through to new 1.5-ml Eppendorf DNA LoBind tubes.
 CRITICAL STEP The flow-through should not be discarded.
Quality control and high-throughput DNA sequencing ● TIMING 10–18 d
42| Quantify the final single-cell RRBS library from Step 41 with a Qubit fluorometer and the Qubit dsDNA HS assay kit.
Use the qPCR assay to determine the concentration of each single-cell RRBS sample.
 CRITICAL STEP The typical yield of the scRRBS libraries is ~20–30 ng (using the Qubit fluorometer for quantification) after
gel-based size selection and AMPure XP beads purification, with <1 ng in the pick-buffer-only negative controls.
 CRITICAL STEP The standard curve–based qPCR assay is a standard quantification assessment for the Illumina libraries
before deep sequencing. First, prepare serial tenfold dilutions of the standard Illumina libraries to generate the standard
curve; on the basis of this curve the number of adapter-insert-adapter molecules of the scRRBS libraries can be determined.
43| Assess the final libraries using a Fragment Analyzer (Advanced Analytical Technologies) to check the size distributions
(Fig. 2b).
 CRITICAL STEP If available, an Agilent Bioanalyzer 2100 can be used as an alternative to the Fragment Analyzer to
evaluate the quality and the size distributions of the final libraries.
 CRITICAL STEP Typical DNA size distribution in scRRBS libraries ranges from 160 to 350 bp, with visible peaks
corresponding to MspI fragments for some repetitive elements (Fig. 2b,c). If the Fragment Analyzer results show that
there are still primer dimers present, perform another clean-up step with 1:1-fold AMPure XP beads as described in Box 1.
44| Sequence the libraries using HiSeq 2000/2500 sequencers
with cluster densities at 75–85% of that used in regular bulk
DNA or RNA sequencing.
Data analysis for single-cell RRBS data ● TIMING 2–3 d
 CRITICAL An overview of the major procedures involved
in scRRBS data analysis is summarized in Figure 3. An
archive containing custom scripts used in this protocol is
available as Supplementary Data. The overview of these
custom scripts, including the script names, functions and
the corresponding protocol steps in which they should be
used, is listed in Table 2.
Raw data
Step 1: quality control (Step 45)
Clean data
Step 2: alignment (Step 46)
SAM file
Step 3: sort and pileup (Steps 47 and 48)
Pileup file
Figure 3 | Schematic of the bioinformatic analysis procedures of the scRRBS
data. The raw sequencing reads generated from the HiSeq 2000 sequencer
are trimmed to remove reads containing low-quality bases or adapter
contaminants. Next, the cleaned reads are used to align to the reference
genome (bisulfite converted in silico) using the default parameters. SAM
files are obtained after bismark mapping, sorted into BAM files and used to
generate the standard pileup files. Customized Perl scripts are applied to
calculate the DNA methylation levels of each covered cytosine based on the
reported C (methylated reads) divided by the total number of reported C and
T (total reads) at the same genomic positions.
Step 4: DNA methylation
level calculation (Step 49)
DNA methylation profiles
Downstream analyses
nature protocols | VOL.10 NO.5 | 2015 | 655
protocol
45| Perform quality control to filter low-quality reads and reads containing adapter sequences, as well as to remove
additional bases that contain cytosines, which were artificially introduced during library preparation. This step can be
conducted using the trim_galore tool with the following command line:
trim_galore --quality 20 --phred33 --stringency 3 --gzip --length 36 --rrbs --paired
--trim1 --output_dir <output_dir> <read1.fastq.gz> <read2.fastq.gz>
This command will generate two files suffixed with ‘_val_1.fq.gz’ and ‘_val_2.fq.gz’ for paired-end sequencing data. We call
these two files ‘clean data’ here.
 CRITICAL STEP The parameter ‘--rrbs’ used here will remove the artificial bases introduced in the end-repair step from the
3′ ends of the reads.
46| When clean data have been obtained, they can be mapped to the reference genome using the Bismark alignment tool35.
For the first-time use of a reference genome and before alignment, ensure that the genome is bisulfite-converted in silico and
indexed well using bismark_genome_preparation, which is attached with the Bismark software package. Example command
line:
© 2015 Nature America, Inc. All rights reserved.
bismark_genome_preparation --verbose –path_to_bowtie bowtie-1.0.0 <genome_folder>
This command will create a transformed reference genome index for bisulfite data alignment.
bismark –path_to_bowtie bowtie-1.0.0 –quiet -o <output_dir> --temp_dir <tmp_dir>
<reference_folder> -1 <read1_val_1.fq.gz> -2 <read2_val_2.fq.gz>
This command will generate a SAM file named ‘read1_val_1.fq.gz_bismark_pe.sam’ containing all the mapping outputs.
A final alignment report will be generated upon completion of the Bismark alignment. In this report, the number of total
sequencing reads, the number of uniquely mapped reads and the mapping ratio of the sample are included. This report will
show the average DNA methylation levels of cytosines in the CpG, CHG and CHH contexts (where H stands for adenine, cytosine or thymine nucleotide), respectively.
Because the samples were spiked with trace amounts of unmethylated λ-DNA, the cytosines in a non-CG (CHH and CHG)
context in the λ-DNA genome are definitely unmethylated. The nonconversion rate of scRRBS is calculated as the number of
sequenced cytosines in non-CG contexts divided by all the covered cytosines in non-CG contexts in the λ-DNA genome.
47| Once the alignment is complete, use SortSam.jar in the Picard toolkit to sort the mapping result in coordinate order to
prepare for the next step.
Example command:
java -Djava.io.tmpdir=<TMP> SortSam.jar I=<read1_val_1.fq.gz_bismark_pe.sam>
O=<read1_val_1.sort.bam> SORT_ORDER=coordinate VALIDATION_STRINGENCY=LENIENT
VERBOSITY=ERROR TMP_DIR=<TMP>
This command will generate a sorted BAM file.
48| Create a pileup file of mapped data prepared for DNA methylation–level calculation. This step will give a pileup result for
each covered locus in the genome. Example command:
samtools mpileup -f <genome.fa> <read1_val_1.sort.bam> > <read1_val_1.pileup>
49| Calculate the DNA methylation level for each covered CpG and non-CpG site. The DNA methylation level of each covered
cytosine is calculated as the number of reported C divided by the total number of reported C and T at the same genome
position. Example command:
Perl SingleC_MetLevel.pl <genome.fa> <read1_val_1.pileup> > <read1_val_1.SingleCmet>
The output file is tab-delimited, and it contains the chromosome, base position, reference genome, chain, total coverage
depth, number of methylated reads, number of unmethylated reads, methylation level and reference context (CpG, CHH or
CHG). This output file contains well-processed data that can be used to calculate the average DNA methylation of the
samples or the DNA methylation levels of any interesting annotated regions.
656 | VOL.10 NO.5 | 2015 | nature protocols
protocol
? TROUBLESHOOTING
Troubleshooting advice can be found in Table 3.
© 2015 Nature America, Inc. All rights reserved.
Table 3 | Troubleshooting table.
Step
Problem
Possible reason
Solution
7
Cell transfer failure
Single cells may stick to the inside
wall of the mouth pipette
First, suck a small volume of PBA-BSA into the pipette, and then
pick the single cell into it. Ensure that the cell is already in the
pipette but near the tip of the pipette. Push all of the carryover
PBA-BSA out of the pipette, together with the cell
9
Incomplete cell lysis
A single cell is not loaded into the
lysis buffer, or the cell is not intact
before transferring it by mouth
pipette
During cell transfer, ensure that the cell is seeded into the
bottom of the tube. Centrifugation is crucial, and it cannot be
omitted. In addition, ensure that the cells are of good
morphology, and if possible always pick the healthiest cells.
The use of 5 µl of lysis buffer and 3 h of incubation is sufficient
to completely lyse the single cells
24,
28
Low PCR amplification
efficiency
Too much DNA loss before PCR
amplification
Poor quality of PCR reagents
Avoid excessive pipetting, and use LoBind tubes during the whole
procedure
Ensure that the reagents for PCR have not expired. Divide them
into small batches to avoid unnecessary freeze-thaw cycles,
especially for the primers
41
Excessive primer-dimer Too many PCR cycles or excessive
contamination
use of adapter or PCR primers
Perform another round of AMPure XP bead purification to remove
contaminants. In addition, ensure that the concentrations of PCR
primers and PCR polymerase are appropriate
● TIMING
Steps 1–9, cell culture, single-cell isolation and cell lysis: 4–5 d
Steps 10–12, MspI digestion: 3–4 h
Steps 13–15, end-repair/dA-tailing reaction: 1–2 h
Steps 16–19, adapter ligation: 9–10 h
Steps 20–23, bisulfite conversion: 3–4 h
Steps 24–27, first-round PCR amplification: 3–4 h (plus purification time; see Box 1)
Steps 28–31, second-round PCR amplification: 3–4 h (plus purification time; see Box 1)
Steps 32–41, size selection of the amplified DNA fragments: 9–10 h (plus purification time; see Box 1)
Steps 42–44, quality control and high-throughput DNA sequencing: 10–18 d
Steps 45–49, data analysis for single-cell RRBS data: 2–3 d
Box 1, DNA purification using AMPure XP beads: 0.5–1 h
Box 2, ‘Crush and soak’ method: 12–14 h
ANTICIPATED RESULTS
The yields of scRRBS libraries from different sample types (haploid cells or diploid cells) do not vary substantially. The
typical yield is ~20–30 ng (using a Qubit fluorometer for quantification) after gel-based size selection and AMPure XP beads
purification, with <1 ng in the ‘picking-buffer-only’ negative controls (using Qubit a fluorometer for quantification), and the
DNA fragment size in scRRBS libraries ranges from 160 to 350 bp, with visible peaks corresponding to the MspI fragments for
certain repetitive elements (Fig. 2b).
We have performed scRRBS on individual mouse and human metaphase II oocytes, sperm, male and female pronuclei of
zygotes, as well as individual mESCs10,17. The average mapping ratio of scRRBS is ~25%, which is lower than that observed
with standard RRBS, which ranges from 50 to 70%. This low mapping ratio may be due to the higher number of PCR amplification cycles required for scRRBS (Fig. 4).
For a mammalian individual diploid cell, scRRBS is expected to cover ~40% of the CpG sites (~1 million CpG sites in a
mouse and human diploid cell) that can be recovered by standard RRBS using thousands of cells10. Coverage lower than this
may be due to degradation of genomic DNA before cell lysis.
In our studies, only scRRBS samples with a high bisulfite conversion rate (>98%) were used for further analysis. We recovered between 0.2 and 1.5 million CpG sites from each individual haploid or diploid cell (Fig. 4 and Supplementary Fig. 1).
nature protocols | VOL.10 NO.5 | 2015 | 657
protocol
Bulk mESCs
Single male pronucleus
Single female pronucleus
Single sperm
Single MII oocyte
Single mESC
Single male pronucleus
Single female pronucleus
Single sperm
Single MII oocyte
Bulk mESCs
Negative controls
Single mESC
Pooled 5–20 mESCs
Single male pronucleus
Single female pronucleus
Single sperm
Single MII oocyte
Bulk hESCs
Pooled 20–200 cells
Single male pronucleus
Single female pronucleus
Single sperm
Figure 5 | The methylation status of a representative locus of sperm-specific
differentially methylated regions (DMRs). The methylation levels of most
of the CpG sites at this locus in the four single human sperm cells are fully
methylated (black filled circles), and most of the CpG sites at this locus
in the three human metaphase II oocytes are unmethylated (white open
circles). The circles in the bulk hESC track indicate the CpG sites covered in
the bulk hESC RRBS sample, with DNA methylation levels ranging from 0% to
100% (color key: white to black, respectively). The filled brown circles in the
bottom track represent all of the CpG sites at this genomic locus.
chr1: 1,098,913–1,100,078 (1,166 bp)
MII oocyte-#1
MII oocyte-#2
MII oocyte-#3
Sperm-#1
Sperm-#2
Sperm-#3
Sperm-#4
The majority of the covered CpG sites should show digitized
Genomic track
DNA methylation; i.e., they are either fully methylated or
unmethylated (Fig. 5; Supplementary Figs. 2 and 3).
Plotting the scRRBS data of individual cells across genes shows that methylation levels are high on gene bodies compared
with neighboring genomic regions, and that there is an expected hypomethylation valley around the transcriptional
start sites (TSSs) (Fig. 6). Moreover, methylation levels gradually increase from the 5′ end (TSS side) to the 3′ end
(transcriptional end site (TES) side) of the gene body. The scRRBS technique is able to reveal global demethylation of the
maternal and paternal genomes during zygotic development (Fig. 6; Supplementary Figs. 4 and 5), in which the paternal
genome is demethylated much faster than the maternal genome in human zygotes, a finding consistent with previous
immunofluorescence analysis10.
Bulk_hESC
a
b
MII oocyte (n = 3)
PN 9–11 h after ICSI (n = 3)
PN 14–15 h after ICSI (n = 3)
PN 18–22 h after ICSI (n = 3)
PN 25–28 h after ICSI (n = 5)
60
40
20
0
Gene body
Down
–15 kb
0%
TSS
20%
40%
60%
Up
80%
Sperm (n = 4)
PN 9–11 h after ICSI (n = 2)
PN 14–15 h after ICSI (n = 3)
PN 18–22 h after ICSI (n = 3)
PN 25–28 h after ICSI (n = 5)
80
DNA methylation level (%)
80
DNA methylation level (%)
© 2015 Nature America, Inc. All rights reserved.
Single MII oocyte
Mapping ratio (%)
5
Total unique CpG sites (× 10 )
Figure 4 | Box plots of the mapping efficiencies
a
b 25
and the total unique CpG sites covered in our
80
scRRBS data. (a) Box plot of the mapping
20
efficiencies of some scRRBS libraries. The six
60
items on the left in a indicate different types of
human cells, including single human sperm cells
15
(n = 4), single metaphase II oocytes (n = 2),
40
single female pronuclei (n = 11), single male
10
pronuclei (n = 11), 20–200 pooled blastomeres of
20
human preimplantation embryos (n = 6) and bulk
5
hESCs (human embryonic stem cells; n = 2). The
seven items (excluding the negative controls) on
0
0
the right in a represent different types of
mouse cells, including single mouse sperm cells
(n = 4), single metaphase II oocytes (n = 2),
single female pronuclei (n = 4), single male
pronuclei (n = 4), single mESCs (n = 8), 5–20
pooled mESCs (n = 3) and bulk mESCs (n = 2) as
control. (b) Box plot of the total unique CpG sites
covered in our scRRBS data. The four items on the
left in b represent four different types of human
Human cells
Mouse cells
Human cells
Mouse cells
cells, including single sperm cells (haploid,
n = 4), single metaphase II oocytes (with polar bodies removed, diploid, n = 2), single female pronuclei (haploid, n = 11) and single male pronuclei (haploid,
n = 11). The six items on the right in b indicate different types of mouse cells, including single mESCs (diploid, n = 8), single sperm cells (haploid, n = 4),
single metaphase II oocytes (with polar bodies removed, diploid, n = 2), single female pronuclei (haploid, n = 4), single male pronuclei (haploid, n = 4)
and bulk mESCs (n = 2), respectively. Middle lines in the box indicate the median values, edges and whiskers of the box indicate the 25th/75th percentiles,
and the 2.5th/97.5th percentiles, respectively. Some extreme values outside of the whisker boundaries are considered outliers.
15 kb
100%
TES
658 | VOL.10 NO.5 | 2015 | nature protocols
60
40
20
0
Gene body
Down
–15 kb
0%
TSS
20%
40%
60%
Up
80%
15 kb
100%
TES
Figure 6 | The average DNA methylation levels
across gene bodies and the flanking intergenic
regions. (a,b) Average DNA methylation levels
along the transcripts and 15 kb upstream
and downstream of the TSSs and the TESs of
all RefSeq genes in the scRRBS data set of
human single metaphase II oocytes and single
female pronuclei at different time points after
intracytoplasmic sperm injection (ICSI) (a),
as well as in the scRRBS data set of human
single sperm cells and single male pronuclei at
different time points after ICSI (b). This shows
global demethylation patterns in the male and
female pronuclei on gene bodies and neighboring
intergenic regions.
protocol
Note: Any Supplementary Information and Source Data files are available in the
online version of the paper.
Acknowledgments We thank J. Qiao and L. Yan for their great help.
The project was supported by the National Science Foundation of China
(31322037 and 31271543) and the National Basic Research Program of China
(2012CB966704 and 2011CB966303). This work is supported by a collaborative
grant from the Center for Molecular and Translational Medicine.
AUTHOR CONTRIBUTIONS L.W. and F.T. conceived the experiments and
supervised the project. H.G., F.G., X.L., X.W. and X.F. carried out all of the
experiments. P.Z. conducted the bioinformatic analyses. H.G., P.Z., L.W. and
F.T. wrote the manuscript with contributions from all of the authors.
COMPETING FINANCIAL INTERESTS The authors declare no competing financial
interests.
© 2015 Nature America, Inc. All rights reserved.
Reprints and permissions information is available online at http://www.nature.
com/reprints/index.html.
1. Bird, A. DNA methylation patterns and epigenetic memory. Genes Dev. 16,
6–21 (2002).
2. Smith, Z.D. et al. A unique regulatory phase of DNA methylation in the
early mammalian embryo. Nature 484, 339–344 (2012).
3. Smallwood, S.A. et al. Dynamic CpG island methylation landscape in
oocytes and preimplantation embryos. Nat. Genet. 43, 811–814 (2011).
4. Gu, T.P. et al. The role of Tet3 DNA dioxygenase in epigenetic
reprogramming by oocytes. Nature 477, 606–610 (2011).
5. Hackett, J.A. et al. Germline DNA demethylation dynamics and imprint
erasure through 5-hydroxymethylcytosine. Science 339, 448–452 (2013).
6. Iqbal, K., Jin, S.G., Pfeifer, G.P. & Szabo, P.E. Reprogramming of the
paternal genome upon fertilization involves genome-wide oxidation of
5-methylcytosine. Proc. Natl. Acad. Sci. USA 108, 3642–3647 (2011).
7. Mayer, W., Niveleau, A., Walter, J., Fundele, R. & Haaf, T. Demethylation
of the zygotic paternal genome. Nature 403, 501–502 (2000).
8. Wossidlo, M. et al. 5-Hydroxymethylcytosine in the mammalian zygote is
linked with epigenetic reprogramming. Nat. Commun. 2, 241 (2011).
9. Okada, Y., Yamagata, K., Hong, K., Wakayama, T. & Zhang, Y. A role for
the elongator complex in zygotic paternal genome demethylation. Nature
463, 554–558 (2010).
10. Guo, H. et al. The DNA methylation landscape of human early embryos.
Nature 511, 606–610 (2014).
11. Smith, Z.D. et al. DNA methylation dynamics of the human
preimplantation embryo. Nature 511, 611–615 (2014).
12. Hackett, J.A. & Surani, M.A. DNA methylation dynamics during the
mammalian life cycle. Philos. Trans. R. Soc. Lond. B Biol. Sci. 368,
20110328 (2013).
13. Seisenberger, S. et al. The dynamics of genome-wide DNA methylation
reprogramming in mouse primordial germ cells. Mol. Cell 48, 849–862 (2012).
14. Xie, W. et al. Epigenomic analysis of multilineage differentiation of human
embryonic stem cells. Cell 153, 1134–1148 (2013).
15. Reik, W. & Kelsey, G. Epigenetics: cellular memory erased in human
embryos. Nature 511, 540–541 (2014).
16. Burgess, D.J. Stem cells: epigenome reprogramming—of mice and men.
Nat. Rev. Genet. 15, 571 (2014).
17. Guo, H. et al. Single-cell methylome landscapes of mouse embryonic stem
cells and early embryos analyzed using reduced representation bisulfite
sequencing. Genome Res. 23, 2126–2135 (2013).
18. Lister, R. et al. Human DNA methylomes at base resolution show
widespread epigenomic differences. Nature 462, 315–322 (2009).
19. Shirane, K. et al. Mouse oocyte methylomes at base resolution reveal
genome-wide accumulation of non-CpG methylation and role of DNA
methyltransferases. PLoS Genet. 9, e1003439 (2013).
20. Molaro, A. et al. Sperm methylation profiles reveal features of epigenetic
inheritance and evolution in primates. Cell 146, 1029–1041 (2011).
21. Gifford, C.A. et al. Transcriptional and epigenetic dynamics during
specification of human embryonic stem cells. Cell 153, 1149–1163 (2013).
22. Ziller, M.J. et al. Charting a dynamic DNA methylation landscape of the
human genome. Nature 500, 477–481 (2013).
23. Lister, R. et al. Global epigenomic reconfiguration during mammalian brain
development. Science 341, 1237905 (2013).
24. Wen, L. et al. Whole-genome analysis of 5-hydroxymethylcytosine and
5-methylcytosine at base resolution in the human brain. Genome Biol. 15,
R49 (2014).
25. Meissner, A. et al. Genome-scale DNA methylation maps of pluripotent and
differentiated cells. Nature 454, 766–770 (2008).
26. Meissner, A. et al. Reduced representation bisulfite sequencing for
comparative high-resolution DNA methylation analysis. Nucleic Acids Res.
33, 5868–5877 (2005).
27. Gu, H. et al. Preparation of reduced representation bisulfite sequencing
libraries for genome-scale DNA methylation profiling. Nat. Protoc. 6,
468–481 (2011).
28. Smallwood, S.A. & Kelsey, G. Genome-wide analysis of DNA methylation in
low cell numbers by reduced representation bisulfite sequencing. Methods
Mol. Biol. 925, 187–197 (2012).
29. Smith, Z.D., Gu, H., Bock, C., Gnirke, A. & Meissner, A. High-throughput
bisulfite sequencing in mammalian genomes. Methods 48, 226–232
(2009).
30. Gu, H. et al. Genome-scale DNA methylation mapping of clinical samples
at single-nucleotide resolution. Nat. Methods 7, 133–136 (2010).
31. Boyle, P. et al. Gel-free multiplexed reduced representation bisulfite
sequencing for large-scale DNA methylation profiling. Genome Biol. 13,
R92 (2012).
32. Smallwood, S.A. et al. Single-cell genome-wide bisulfite sequencing for
assessing epigenetic heterogeneity. Nat. Methods 11, 817–820 (2014).
33. Miura, F., Enomoto, Y., Dairiki, R. & Ito, T. Amplification-free wholegenome bisulfite sequencing by post-bisulfite adaptor tagging.
Nucleic Acids Res. 40, e136 (2012).
34. Guo, F. et al. Active and passive demethylation of male and female
pronuclear DNA in the mammalian zygote. Cell Stem Cell 15, 447–458
(2014).
35. Krueger, F. & Andrews, S.R. Bismark: a flexible aligner and methylation
caller for bisulfite-seq applications. Bioinformatics 27, 1571–1572
(2011).
36. Xi, Y. & Li, W. BSMAP: whole-genome bisulfite sequence mapping program.
BMC Bioinformatics 10, 232 (2009).
37. Chen, P.Y., Cokus, S.J. & Pellegrini, M. BS Seeker: precise mapping for
bisulfite sequencing. BMC Bioinformatics 11, 203 (2010).
38. Langmead, B., Trapnell, C., Pop, M. & Salzberg, S.L. Ultrafast and memoryefficient alignment of short DNA sequences to the human genome. Genome
Biol. 10, R25 (2009).
39. Li, H. et al. The Sequence Alignment/Map format and SAMtools.
Bioinformatics 25, 2078–2079 (2009).
40. Tang, F. et al. RNA-seq analysis to capture the transcriptome landscape of
a single cell. Nat. Protoc. 5, 516–535 (2010).
41. Tang, F. et al. mRNA-seq whole-transcriptome analysis of a single cell.
Nat. Methods 6, 377–382 (2009).
42. Yan, L. et al. Single-cell RNA-seq profiling of human preimplantation
embryos and embryonic stem cells. Nat. Struct. Mol. Biol. 20, 1131–1139
(2013).
43. Evrony, G.D. et al. Single-neuron sequencing analysis of L1
retrotransposition and somatic mutation in the human brain. Cell 151,
483–496 (2012).
44. Frumkin, D. et al. Amplification of multiple genomic loci from single cells
isolated by laser micro-dissection of tissues. BMC Biotechnol. 8, 17
(2008).
45. Ma, C. et al. A clinical microchip for evaluation of single immune cells
reveals high functional heterogeneity in phenotypically similar T cells.
Nat. Med. 17, 738–743 (2011).
46. Fan, H.C., Wang, J., Potanina, A. & Quake, S.R. Whole-genome molecular
haplotyping of single cells. Nat. Biotechnol. 29, 51–57 (2011).
nature protocols | VOL.10 NO.5 | 2015 | 659