Single Cell Analysis Using Droplet Microfluidics

Review
www.adv-biosys.com
Single-Cell Analysis Using Droplet Microfluidics

Kinga Matuła, Francesca Rivello, and Wilhelm T. S. Huck*
signaling dynamics.[3,4] Quantifying and

Droplet microfluidics has revolutionized the study of single cells. The ability understanding cellular heterogeneity,
and linking it to cellular function, is of
to compartmentalize cells within picoliter droplets in microfluidic devices has
great importance for many applications,
opened up a wide range of strategies to extract information at the genomic, for example, the discovery of rare cells
transcriptomic, proteomic, or metabolomic level from large numbers of (for example in the immune system),[5,6]
individual cells. Studying the different molecular landscapes at single-cell gaining a better understanding of the
resolution has provided the authors with a detailed picture of intracellular development of embryos,[7,8] studying the
heterogeneity and the resulting changes in cellular phenotypes. In addition, fundamental properties of genetic and
signaling networks,[9] and analyzing the
these technologies have aided in the discovery of rare cells in tumors or in
composition of solid tumors.[10,11] Isola-
the immune system, and left the authors with a deeper understanding of the tion and characterization of single cells
fundamental biological processes that determine cell fate. This review aims at the genome, transcriptome, proteome,
to provide a detailed overview of the various droplet microfluidic strategies or metabolome level, can be achieved in
reported in the literature, taking into account the sometimes subtle differ- a variety of different ways, ranging from
ences in workflow or reagents that enable or improve certain protocols. very high throughput (but limited number
of data points per cell) flow cytometry
Specifically, approaches to targeted- and whole-genome analysis, as well as experiments, to early studies showing
whole-transcriptome profiling techniques, are reviewed. In addition, an up-to- whole-genome or whole-transcriptome
date overview of new methods to characterize and quantify single-cell protein sequencing of a limited number of indi-
levels, and of developments to screen secreted molecules such as antibodies, vidual cells placed in microliter wells.[12]
cytokines, or metabolites at the single-cell level, is provided. We refer the interested reader to a number
of recent reviews that focus on single-cell
biology[1,2,6,11] or give an overview of the
different technologies used.[13,14]
1. Introduction Despite an exponential increase in the number of single-
cell studies reported, these experiments are by no means
Most of what we know about the fundamental rules underpin- trivial, with many technical challenges remaining, including
ning biological systems has been gathered by studying popula- the isolation of single cells (from solid samples), the handling
tions: populations of organisms, or populations of cells within of low quantities of biological materials, and laborious work-
organisms. However, the past decade has seen the develop- flows. In recent years, microfluidics, and in particular droplet
ment of a plethora of techniques that allow the study of large microfluidics has emerged as a technology of choice for an
numbers of individual cells, revealing cellular function in easily implementable, high throughput, and relatively cheap
intricate detail, but also showing a high level of cellular het- approach for a broad range of single-cell analysis studies
erogeneity.[1,2] This heterogeneity only becomes apparent at (Figure 1).
the single-cell level and is a result of cellular programming The purpose of this review is to compare and contrast the
networks, the inherent noise in gene expression, as well as available droplet microfluidics technologies for single-cell char-
acterization in order to allow potential users to rapidly imple-
ment the best tools.
K. Matuła, F. Rivello, Prof. W. T. S. Huck We have structured this review as follows: first, we will
Radboud University briefly introduce the field of droplet microfluidics and dis-
Institute for Molecules and Materials cuss the key characteristics that make it an ideal technology
Heyendaalseweg 135, 6525AJ Nijmegen, The Netherlands
E-mail: w.huck@science.ru.nl for single-cell analysis. Second, we will introduce different
The ORCID identification number(s) for the author(s) of this article
methods for single cells sample preparation. After this, we
can be found under https://doi.org/10.1002/adbi.201900188. will provide a state-of-the-art overview of key breakthroughs in
© 2019 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, single-cell genetic, epigenetic, transcriptomic, proteomic, and
Weinheim. This is an open access article under the terms of the Creative metabolic studies. Figure 2 provides a graphical overview of
Commons Attribution-NonCommercial-NoDerivs License, which per- the topics discussed in this review. Combined, the review will
mits use and distribution in any medium, provided the original work is be useful for both novice researchers in the field, but also for
properly cited, the use is non-commercial and no modifications or adap-
tations are made.
experienced scientists who wish to explore new combinations
of single-cell studies and hope to adapt existing technologies to
DOI: 10.1002/adbi.201900188 broaden their capabilities.
Adv. Biosys. 2020, 4, 1900188 1900188 (1 of 28) © 2019 The Authors. Published by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
23667478, 2020, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adbi.201900188 by CochraneChina, Wiley Online Library on [19/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.adv-biosys.com
2. Droplet Microfluidics
Kinga Matuła graduated
Microfluidics is the technology that controls the flow of liquids maxima cum laude in bio-
through micron-size channels. Forcing two immiscible liquids technology from the Rzeszow
through these channels leads to the formation of plugs of one University of Technology
fluid within a carrier fluid. This type of two-phase microfluidics (Poland). She studied at the
is typically known as droplet microfluidics and has become Otto-von-Guericke University
enormously popular in the last decade and a half. The pico- to Magdeburg and worked at
nanoliter droplets that are formed at frequencies up to several the Max Planck Institute
MHz essentially function as individual reaction flasks. Hence, for Dynamic and Complex
each droplet is equivalent to a well on a microliter plate but Technical Systems in
with a reaction volume a million times smaller, and cells encap- Magdeburg (Germany) where
sulated in droplets are physically and (bio)chemically isolated she prepared her master
from each other. thesis in chemical and process engineering. Afterwards,
Briefly, water-in-oil based droplet microfluidic technolo- she started her Ph.D. in soft condensed matter at the
gies use a water-based solution as the dispersed phase and oil Institute of Physical Chemistry of the Polish Academy of
with a surfactant as the continuous phase. The flows of these Sciences in Warsaw (Poland). In 2018, Kinga joined the
two immiscible phases are precisely controlled by micro group of Professor Wilhelm Huck where she is working on
fluidic pumps and by specially designed chips with molded or development of droplet microfluidic platforms for single
engraved microchannels. Droplet formation can be achieved cell analysis.
in a highly repeatable manner using device geometries such
as T-junctions, flow focusing, and co-flow (Figure 3). The most Francesca Rivello received
common geometry for droplet formation is flow-focusing[15] an M.Sc. cum laude in
where the injected dispersed phase (water phase) is sheared nanotechnology at the
by the continuous phase (oil with surfactant) pumped from University of Twente (The
two side channels positioned perpendicularly to the aqueous Netherlands) in 2015. She is
stream. As soon as fluids meet, water-in-oil droplets are formed currently a Ph.D. candidate
and stabilized by the surfactant dissolved in the oil carrier under the supervision of
phase. Droplets can be loaded with single cells by using a dilute Professor Wilhelm Huck in
cell suspension as the aqueous phase. The distribution of cells the Radboud University. Her
in droplets follows a Poisson distribution,[16,17] minimizing project focuses on the study
the risk of encapsulation of multiple cells in one droplet, but of single cells using droplet
yielding up to ≈90% of droplets containing no cells. Once microfluidic technologies with
formed, droplets with encapsulated cells can be manipulated in particular focus on their kinetic responses and on their
various ways: they can be merged, split, re-loaded in a second contribution to metastasis progression.
microfluidic device, incubated within the microfluidic chip,
detected, sorted, etc. (Figure 3). Technologies for droplet forma-
tion and manipulation have been reviewed extensively and will Wilhelm Huck is Professor of
not be further addressed here.[18–21] Physical Organic Chemistry.
Although throughput is not as high as in flow cytometry, After postdoctoral research
droplet generation and sorting are typically performed at speeds at Harvard University, he
of a few to tens of kHz. It is worth mentioning that droplets took up a position in the
can be produced in large amounts, circumventing limitations Department of Chemistry at
in the number of cells that can be studied when using well- the University of Cambridge,
based technologies. In the next paragraphs, we will outline how where he was promoted
droplet microfluidic technology can be used to extract valuable to Director of the Melville
information from large numbers of single cells. Laboratory for Polymer
Synthesis (2004) and Full
Professor of Macromolecular
3. Sample Preparation Chemistry (2007). In 2010, he moved to the Radboud
University Nijmegen, where he completely changed
Single-cell compartmentalization into pL to nL droplets research fields to focus on understanding how living
requires cells to be in suspension. Any sample prepara- cells function and where they come from. In this context,
tion method must aim to obtain a sufficient number of cells, his group is developing methods to determine reac-
without debris, contamination, or unwanted cells. Two main tion rates of (potentially all) chemical processes in
methods are used for cell isolation from solid tissues: standard individual cells.
digestion or laser-capture microdissection (LCM). Standard
digestion protocols can be applied to isolate cells from a tissue
Figure 1. Timeline showing the development of droplet microfluidic technologies for single-cell studies, organized by analyte and the availability of a
commercial platform. The various acronyms will be explained in the text below.
sample taking into consideration different parameters (e.g., As a point of a technical challenge, cells also tend to cluster
type of tissue, species of origin) to maximize the yield of viable and/or sediment in the syringe and in the inner surface of
cells.[22] The alternative to digestion treatment is LCM.[23,24] the tubing during loading onto the chip. To circumvent, or
However, this technique is low-throughput, the quality of minimize this problem researchers have either used density
LCM-harvested cells is relatively poor (especially in the case matching agents such as Iodixanol,[34] physically stirred the cell
of RNA), and often requires prior preservation of the material suspension with a magnet in the syringe,[35,36] or introduced
and more technical skills. conically tapered chip inlet regions.[37] Alternatively, flow rates
In the case of primary material (e.g., blood), cells of interest and channel geometries can be adjusted to achieve “super-
(e.g., white blood cells, circulating tumor cells) can be iso- Poisson” encapsulation of cells, leading to much higher frac-
lated by density gradient centrifugation (Percoll, Ficoll),[25,26] tions of droplets containing single cells.[38,39]
negative/positive depletion using antibodies conjugated After cell encapsulation, workflows differ significantly
with magnetic beads followed by separation on a magnetic depending on the type of analysis that is performed. There-
column,[27] by size-based microfluidic separation after prior fore, each technology, that is, profiling the (targeted- and
red blood cells (RBCs) selective lysis by osmotic shock and cen- whole-) genome, epigenome, (targeted- and whole-) transcrip-
trifugation, or by apheresis.[28,29] A wide range of microfluidic tome, proteome (secreted proteome or intracellular and mem-
tools that exploit physical (size, deformability, density, electric brane proteome), or metabolome, will be discussed separately.
charge) or biological (expression of certain proteins) proper-
ties can be applied to isolate cells of interest.[27,30,31] Fluores-
cence-activated cell sorting and magnetic-assisted cell sorting, 4. Single-Cell Genomics
based on the selection of cells expressing certain markers, are
widely used both as isolation and purification techniques.[32,33] DNA-sequencing of individual cells offers unique opportunities
Bacteria and other types of cells from soil can be dislodged by to investigate de novo assembly of the genome of rare or uncul-
sonication, and when particles settle, aliquots of supernatant tivable microbes,[40,41] heterogeneity of cells in a tumor (DNA
can be taken for further processing. It should be stressed that mutations)[42] or acquisition of resistance during treatment.[43,44]
the type of analysis influences if whole cells or only certain Extensive comparison of the genomic heterogeneity between the
organelles (e.g., only single nuclei are encapsulated in single- primary tumor and the metastatic tumor sites has led to a deeper
cell chromatin accessibility analysis) are used further in the understanding of metastasis formation,[43] and an understanding
protocol. of the genetic changes in the tumor environment leading to
Figure 2. In this review, we will provide an overview of droplet microfluidics approaches to gain information on all key processes in individual cells.
Technologies are categorized in sections as indicated in the figure.
drug resistance.[44] Technologies for single-cell genomic analysis flanking the sequence of interest in the genome that is ampli-
which do not use droplet microfluidics, but wells: single-cell ret- fied in the next step. In contrast, primers used for WGA bind to
rotransposon capture sequencing (scRC-seq),[45] single nucleus many places in the genome to start replication, thus the whole
exome sequencing (nuc-seq/SNES)[46] or valves: direct determin- DNA strand is amplified. In this section, droplet microfluidic
istic phasing (DDP),[47] will not be reviewed here. technologies for single-cell genome analysis are introduced and
A critical step in single-cell genome analysis is the amplifi- compared.
cation of pico- or nanogram amounts of DNA into microgram
amounts. Two strategies can be distinguished to amplify single-
cell derived genomic material: i) targeted-genome amplification 4.1. Targeted-Genome Analysis
(only targeted fragments are amplified), or ii) whole-genome
amplification (WGA). Three droplet microfluidic technologies allow single-cell targeted
Targeted-genome amplification utilizes a standard PCR genome analysis: single copy genetic amplification (SCGA), aga-
reaction. Designed primers bind only to the specific sequence rose-based PCR, and PCR-Activated Cell Sorting (PACS).
Figure 3. Microfluidic drop maker geometries and manipulation modules applied in single-cell workflows. Droplets can be generated using microfluidic
devices using three different geometries: T-junction, flow focusing, and co-flow. After droplet generation, microfluidic modules are available for droplet
manipulation: merging of droplets with different contents, splitting of bigger droplets into droplets with smaller volume, re-loading of the emulsion into
a new device, static on-chip incubation of the emulsion at different temperatures, detection of (fluorescence) signal in droplets, and sorting droplets
of interest (using e.g., electric field).
4.1.1. SCGA into droplets from very dilute solutions following Poisson sta-
tistics, leading to low efficiency in co-encapsulation: 1 in 100
Single copy genetic amplification (SCGA)[48] is based on the droplets containing both a single cell and a single bead.
generation of uniform nanoliter droplets to perform bead PCR.
Figure 4a shows the general workflow of SCGA. Briefly, single
cells are co-encapsulated in water-in-oil droplets with: an aga- 4.1.2. Agarose Droplet Microfluidic ePCR
rose bead covalently labeled with reverse primers, and PCR
mixture containing dye-labeled forward primers and enzymes. Yang’s group developed agarose-based emulsion PCR (ePCR).[50]
Subsequently, the droplets are collected off-chip, cells are Figure 4b shows the general workflow. An agarose solution is
lysed, and the emulsion undergoes a series of PCR thermocy- emulsified at 540 Hz with cells (2, 1, or 0.5 cells per droplet)
cles which generate dye-labeled double-stranded product on and PCR mix. Similarly to SCGA, the PCR mix contains fluo-
the bead surface. Following droplet PCR, de-emulsification is rescently labeled forward primers and enzymes. However, in
induced and the beads coated with the fluorescent amplicons this case, the reverse primers are covalently attached to the aga-
are recovered. Detection of the target DNA is performed by rose instead of being attached to a bead. After emulsification,
measuring single-bead fluorescence intensity by flow cytom- the droplets are collected for further PCR to amplify the target
etry. DNA fragments up to 1139 bp were obtained using SCGA DNA of interest. Subsequently, the agarose droplets are gelated
with 40% PCR efficiency, meaning that bead-attached products into solid beads with the fluorescently labeled PCR amplicons
can serve as a template for bulk sequencing. In addition, SCGA covalently attached to the agarose matrix. Finally, the beads are
constitutes a fast and cheap detection method, with only three, washed to remove the excess of fluorescently labeled forward
high throughput steps: emulsification, PCR, and the detection primers which have not been used during PCR reaction and are
of the fluorescence intensity of the beads using flow cytom- then analyzed using fluorescence microscopy or flow cytometry.
etry. A disadvantage of SCGA is the low droplet generation fre- Agarose droplet microfluidic ePCR offers multiple advan-
quency (below 6 Hz). To address this issue, Zeng et al. created tages compared to SCGA. First, it shows one order of magni-
a microfabricated emulsion generator array (MEGA) device tude increase in the efficiency of cells and primers co-encapsu-
for high-throughput generation of droplets.[49] The 96-channel lation since primers are part of the agarose liquid phase, and
MEGA device reaches a droplet generation rate of 940 Hz. A are therefore present in all droplets, while in SCGA the primers
second disadvantage of SCGA is that droplet PCR (dPCR) has are carried on solid beads (in 10% of the formed droplets).
lower efficiency compared to bulk PCR: 40% for dPCR com- Increase in co-encapsulation efficiency enables characteriza-
pared to bulk PCR but with equivalent template and bead con- tion of whole-cell populations, and decreases the duration and
centrations. Finally, SCGA encapsulates both beads and cells costs of the experiment. A further simplification is related to
Figure 4. Comparison of workflows for targeted genome analysis using droplet microfluidics. a) Single copy genetic amplification (SCGA). b) Agarose
droplet microfluidic ePCR.
the usage of agarose droplets that overcome the need of using encapsulated in droplets with proteases. After cell lysis in the
beads labeled with primers. droplets, the temperature is increased to inactivate the proteases.
Subsequently, the droplets containing the genome of single
cells are coupled by electrocoalescence with a second droplet
4.1.3. PACS containing PCR reagents and barcoding hydrogel beads (at 1:1
ratio). The barcoding hydrogel beads are labeled with oligonu-
PCR-Activated Cell Sorting (PACS) allows sorting of bacteria cleotides containing a cell-specific barcode and a gene-specific
based on the interrogation of small genomic regions (hun- primer sequence. After co-encapsulation of the hydrogel beads
dreds of bases).[51] The general workflow of PACS is shown in with the genome-containing droplets, the primers labeling the
Figure 5a. Bacteria and PCR reagents are co-encapsulated in hydrogel bead are photo-released by UV exposure and the target
droplets. Cells are lysed and if their DNA contains the sequence DNA sequences are cell-barcoded by PCR amplification. Finally,
of interest, the TaqMan probe (containing a reporter dye and the emulsion is broken and libraries are prepared in bulk, fol-
a quencher in close proximity) anneals to it. Hydrolysis of the lowed by sequencing. Pellegrino et al.[52] showed a proof-of-
probe by the exonuclease activity of the Taq polymerase allows concept experiment using 62 DNA targets to analyze the genetic
the release of the reporter dye from the probe and thus fluo- heterogeneity of individual acute myeloid leukemia cells.
rescence is increased due to the elimination of the quenching
effect. Upon PCR amplification, the intensity of the fluores-
cence signal increases in the droplet at every cycle. Subse- 4.2. Whole-Genome Analysis
quently, the fluorescent droplets are sorted for downstream
whole-genome analysis. Whole-genome analysis at the single-cell level is challenging
due to: i) difficulties surrounding the isolation of DNA from
individual cells, ii) efficiency and reliability of whole-genome
4.1.4. Commercial Developments amplification (WGA), iii) verification of sequences that can be
used for identification of variants, and iv) data analysis and
Mission Bio developed a platform for single-cell targeted- interpretation.[53]
genome analysis and commercialized it as Tapestri tech- In this section, we first introduce different whole-genome
nology.[52] Figure 5b presents the workflow. First, single cells are amplification techniques which have been one of the main
Figure 5. Schematic overview of the workflows for targeted genome analysis. a) PCR-Activated Cell Sorting (PACS). b) Tapestri technology. The
sequencing read-out is R1: read1. R2: read 2.
technical challenges in whole-genome analysis in the past dec- amplification (DOP-PCR) was introduced in 1992 by Telenius
ades. Then we describe two full pipelines for single-cell whole- et al.[54] DOP-PCR is based on the use of a single primer con-
genome analysis using droplet microfluidics: SiC-seq and CNV taining a random sequence (degenerate sequence). DOP-PCR
(10X Genomics). begins with a few cycles of pre-amplification at a low initial
annealing temperature in order to allow the random primers
to anneal. Subsequently, DNA amplicons are further ampli-
4.2.1. Whole-Genome Amplification Techniques fied by PCR.
DOP-PCR has two main advantages: it amplifies DNA at
DOP-PCR: The concept of degenerate oligonucleotide- multiple loci and it is species-independent. However, DOP-
primed polymerase chain reaction or pure PCR-based PCR provides low genome coverage so any differences in the
amplification are exponentially enlarged, resulting in overam- In the first step, SiC-seq requires preparation of a barcoding
plified and underamplified regions along the genome. droplet library. 15-base random oligonucleotides flanked by
MDA: Multiple displacement amplification (MDA), devel- constant sequences are co-encapsulated with: PCR reagents
oped by Dean and co-workers in 2001,[55] is based on isothermal and primers (complementary to constant regions of the bar-
amplification. Random hexamer primers are hybridized to codes and containing the Illumina P7 flow cell adapter). Sub-
the template, followed by strand displacement DNA syn- sequently, the content of droplets with all reagents is ampli-
thesis by the high fidelity φ29 polymerase. DNA amplifica- fied via droplet digital PCR (≈10 million barcoding droplets
tion using MDA results in much higher genome coverage are generated in a few hours). After the barcoding library is
compared to DOP-PCR.[56] However, MDA, similarly to DOP- prepared, single cells must be isolated. The cell suspension is
PCR, is based on an exponential amplification that leads to merged with a molten agarose stream using a co-flow droplet
sequence-dependent overamplification or underamplification maker. The agarose droplets are solidified by cooling and
along the whole genome that is not reproducible from cell to transferred from oil to aqueous carrier. The agarose beads are
cell. Different variations of MDA were developed to overcome permeable to enzymes, detergents, and small molecules, but
the amplification bias and improve the throughput, that is, sterically trap larger structures as genomic DNA, making them
MIDAS,[57] IMS-MIDAS,[58] SNES,[46] ddMDA.[59] MDA was suc- the ideal substrate to allow washing steps to be performed on
cessfully implemented in droplet microfluidics to increase its encapsulated cells. Subsequently, the cell wall is digested, and
efficiency.[60–62] Multiple displacement amplification in droplets a series of enzymatic and detergent treatments (solubilization
(sd-MDA) was also demonstrated on genomes derived from of lipids and digestion of proteins) are performed. Purified
single cells.[63] microbeads with genomic DNA are re-encapsulated in drop-
Hybrid Methods: Hybrid methods intend to overcome some lets containing tagmentation reagents in order to fragment the
of the shortcomings of the two aforementioned methods by genome and attach universal sequences to act as PCR handles.
combining some of their strengths. Multiple Annealing and After tagmentation, these droplets are in turn merged sequen-
Looping Based Amplification Cycles (MALBAC) and displace- tially with two other droplets: one droplet containing the PCR
ment DOP-PCR (PicoPLEX) are examples of hybrid methods reagents and one barcoding droplet. The obtained droplets are
that combine isothermal amplification followed by PCR ampli- thermo-cycled to allow amplification of the product and liga-
fication of generated amplicons. MALBAC is based on quasi- tion of the P5 and P7 adapter required for Illumina sequencing.
linear amplification that reduces the sequence-dependent bias, After library preparation, the droplets are pooled together
and the key improvement is not to generate copies of copies for sequencing. Sequencing data are filtered for quality and
but oppositely, amplify only the original genome sequence by grouped by barcode, providing a genomic sequence for all cells.
protecting the already amplified products. This is achieved SiC-seq has several advantages compared to the targeted-
by utilization of random primers that possess a sequence genome single-cell analysis tools described above. First, SiC-
(anchor) to promote looping of amplicons and prevent fur- seq allows an unbiased analysis of the single-cell genome. In
ther amplification before the second PCR. In contrast, Pico- addition, all steps can be performed using microfluidics, lim-
PLEX uses degenerate primers in the first reaction to add an iting manual handling steps and increasing reproducibility.
anchor sequence, followed by priming to the added sequence However, the microfluidic manipulation is more complex and
for the final PCR amplification. Yu et al. demonstrated an time-consuming compared to the targeted-genome analysis
integrated chamber microfluidic device designed for single- methods.
cell MALBAC reactions to identify copy number variations.[64]
Comparison of the different amplification methods have been
reviewed elsewhere and will not be further addressed in this 4.2.3. Commercial Developments
review.[65,66]
Single Cell Copy Number Variation (CNV) was introduced
to the market by 10X Genomics to study single-cell genomic
4.2.2. SiC-Seq heterogeneity and clonal evolution in a high throughput
manner (hundreds to thousands of cells),[68] the workflow of
Single-cell genome sequencing at ultrahigh-throughput CNV is shown in Figure 6b. First, single cells are encapsu-
(SiC-seq) developed by Lan et al.[67] was the first platform for lated at limiting dilution in a gel matrix with paramagnetic
single-cell genome analysis where most of the reactions are particles within a single-use microfluidic chip. Once droplets
performed in droplets. SiC-seq labels all the DNA material gelate, cells are trapped inside the beads and are subjected
from an individual cell with a barcode unique to this cell. to lysis followed by the removal of all nuclear proteins, while
The workflow comprises the following steps (Figure 6a): the genomic DNA remains trapped in the gel matrix. Subse-
1) generation of barcoding droplets, 2) cell encapsulation quently, purified genomic DNA inside the cell beads is co-
in agarose droplets, lysis, purification of genomic DNA, encapsulated in a second microfluidic device with barcoding
and gelation of the cell beads, 3) re-encapsulation of the puri- beads (10X Barcoded Gel Beads) and enzymes. Importantly,
fied cell beads with tagmentation reagents, 4) merging of both the cell beads and the barcoding beads are closely packed,
droplets bearing tagmented genome with droplet containing allowing high-efficiency co-encapsulation of one cell bead and
PCR reagents and barcoding droplets, 5) sequencing and data one barcoding bead in each droplet (≈80% of droplets con-
analysis. tain one cell bead and one barcoding bead). DNA inside the
Figure 6. Schematic overview of droplet microfluidic workflows used in whole-genome analysis. a) Single-cell genome sequencing at ultrahigh-
throughput (SiC-seq). b) Single Cell Copy Number Variation (CNV).
droplet is amplified to generate single-cell barcoded libraries It is worth mentioning that due to a recent loss of a patent
ready for sequencing and analysis. Importantly, all generated lawsuit brought by Bio-Rad regarding the use of non-fluor-
DNA from individual cells share a common 10X barcode. inated channels, 10X Genomics had to change their device
design. The new “Next GEM chip” appears to use step emulsifi- of the diversity of cells and tissues. In eukaryotes, chromatin
cation instead of flow focusing for droplet generation, although is formed by nucleosomes that are composed of DNA wrapped
we cannot verify this at present. around histone octamers. Histones can carry a range of cova-
lent modifications, but acetylation and methylation are the most
commonly studied since they play a crucial role in the regula-
5. Single-Cell Epigenomics tion of accessibility of DNA and gene transcription.[83] Tradi-
tional methods for detection of histone modifications include
Single-cell epigenetic analysis enables the investigation of herit- mass spectrometry, immunoblotting, and Coomassie staining,
able changes in phenotype that do not involve changes in the or chromatin immunoprecipitation (ChIP). Three variants of
DNA sequence. As epigenetic information comes in various ChIP have been reported: i) ChIP combined with microarrays
forms, for example, covalent modifications on DNA (methyla- (ChIP-on-chip or ChIP-chip) to provide extensive maps of his-
tion), chromatin accessibility and compaction, and higher-order tone modifications and their associated DNA, ii) ChIP followed
organization of chromosome domains, as well as post-trans- by qPCR if the target loci is known a priori (ChIP-qCPR), and
lational modifications of histones, it is clear that different iii) ChIP combined with whole DNA sequencing (ChIP-seq).
biochemical approaches are required to extract each layer of ChIP-seq, and many variations thereof, is now the primary
molecular information.[69] technology used to examine histone modifications and DNA-
protein interactions.[84,85] These methods allow mapping of his-
tone modifications at a population level but are insensitive to
5.1. DNA Methylation cell-to-cell variation. Single-cell analysis of histone modification
has proven to be challenging in particular due to the high level
DNA methylation is the most commonly investigated epigenetic of experimental noise during the immunoprecipitation step
DNA modification. Methylation represses gene expression, regu- when using small amounts of material.
lates various cellular processes, and plays a role in the develop-
ment of cancer.[70] In the mammalian genome, methylation only
takes place at cytosine bases that are directly followed by a guanine 5.2.1. Drop-ChIp (scChIp-seq)
base (CpG dinucleotide). CpG dinucleotides are underrepresented
in the genome but can be found in CpG islands (regions with a Droplet-based single-cell chromatin immune-precipitation
high frequency of CpG sites) which are normally hypomethylated. sequencing (Drop-ChIp), developed in 2015, is a microfluidic
In this context, methylation of the mammalian genome is mostly variation of the well-established ChIP-seq used to map, at a
studied in CpG islands and hypermethylation of CpG islands has single-cell level, posttranslational modifications of histones
been described in almost every type of tumor.[71,72] DNA meth- that can be associated with chromatin activity states.[86] Drop-
ylation can be studied at a single-cell level using different tools: ChIp overcomes the challenges associated with low input
single-cell hydroxymethylation sequencing (scAba-seq),[73] chem- starting material from single cells by labeling chromatin from
ical-labeling-enabled C-to-T conversion sequencing (CLEVER- single cells before immunoprecipitation and combining it for
seq),[74] single-nucleus methylcytosine sequencing (snmC-seq),[75] immunoprecipitation.
single-cell combinatorial indexing for methylation analysis (sci- An overview of the full workflow is presented in Figure 7.
MET),[76] improved version of single-nucleus methylcytosine First, a library of droplets with DNA barcodes is prepared by
sequencing (snmC-seq2),[77] methylated DNA immunoprecipita- emulsification of DNA from multi-well plates. Each barcoding
tion followed by next-generation sequencing (MeDIP-seq).[78] droplet contains multiple copies of the same double-stranded
Single-cell bisulfite sequencing (scBS-seq) is considered a DNA sequence. The DNA sequence is symmetrically designed
gold standard for studying DNA methylation. This method is allowing ligation on either side of the target DNA and com-
based on treating DNA with sodium bisulfite which deami- prises: a unique barcode sequence, an adapter, and a restric-
nates cytosine residues and converts them to uracil, while 5-mC tion site for selecting the desired product (PaCI site). Subse-
residues are left unaffected.[79,80] After bisulfite-conversion, quently, single cells are encapsulated in aqueous droplets
the DNA is PCR-amplified and sequenced. Information about with lysis buffer containing weak detergent and micrococcal
the methylation state can be retrieved from the sequencing nuclease (MNase). MNase preferentially cuts accessible linker
results as unmethylated cytosines are converted into thym- DNA and digests the chromatin in the droplets. Indexing of
ines, whereas methylated cytosines remain cytosines. Bisulfide cells is performed by merging chromatin-containing droplets
sequencing was combined with droplet microfluidics (micro- with a barcoding droplet and a droplet containing labeling
droplet PCR) to increase the efficiency of the technique.[81] buffer (with DNA ligase) within a third microfluidic chip.
Presently, there are no droplet microfluidic routes for high Barcode sequences are then ligated to both ends of the single-
throughput single-cell methylation analysis, even though groups cell derived DNA fragments, the emulsion is broken, and the
are working on single-cell bisulfite sequencing: Drop-BS.[82] combined chromatin from many single cells is immunopre-
cipitated in the presence of “carrier” chromatin (chromatin
extracted from other species). PaCI sites on the barcoded frag-
5.2. Histone Modification ments are digested and the resulting products are PCR ampli-
fied. Finally, after sequencing the obtained reads are clustered
Chromatin organization controls access to genetic information by their barcode sequences to extract single-cell chromatin
and regulatory elements and is therefore responsible for part profiles.
Figure 7. Overview of the Drop-ChIP workflow. R1: read 1. R2: read 2.
Although Drop-ChIp allows the profiling of chromatin struc- 5.3. Chromatin Accessibility
ture at a single-cell level, it still presents a number of limita-
tions. First, only ≈800 peaks are detected per single-cell (≈5% Chromatin accessibility can be defined as the degree to which
overall sensitivity for peak detection). The single-cell profiles macromolecules are able to physically contact packed DNA.
lack the sensitivity that is required in de novo peak calling. This highly dynamic property is determined by the topological
However, the detection of ≈800 true peaks with high specificity arrangement of nucleosomes and other chromatin-binding fac-
might be sufficient to classify or group individual cells with tors (non-histone macromolecules) that modify access to DNA
related chromatin profiles. in order to elicit transcription, differentiation, cell division, and
In addition, the pool of 1152 different barcodes can only DNA repair.[88,89] Methods that provide quantitative information
barcode a maximum of 100 cells to avoid repetition of a bar- on chromatin accessibility are primarily focused on measuring
code sequence (4% multi-barcoding error). This limitation is the susceptibility of chromatin to either enzymatic methyla-
overcome by adding a sample index sequence during the PCR tion or DNA cleavage. A number of methods have been devel-
amplification and pooling together more of these barcoded oped to profile chromatin accessibility at the single-cell level:
libraries before immunoprecipitation. Another method to over- i) assay for transposase-accessible chromatin using sequencing
come this limit would be to prepare a more complex barcoding (scATAC-seq),[90] and variations including combinatorial cellular
library in a similar way to SiC-seq.[67] Importantly, some of indexing applied in assay for transposase-accessible chromatin
these limitations were solved by the introduction of scChIp-seq sequencing (sci-ATAC-seq),[90] single-cell transposome hyper-
which allowed the detection of 10.000 unique loci per cell and sensitive site sequencing (scTHS-seq);[91] ii) method to detect
the increase in throughput to 5000 cells analyzable per run.[87] genome-wide DNase I hypersensitive sites (scDNase-seq);[92]
iii) single-cell micrococcal nuclease sequencing (scMNase- add a well-specific barcode to open chromatin. After well-bar-
seq);[93] and iv) nucleosome occupancy and methylome- coding, the nuclei are transposed, barcoded, and libraries are
sequencing (NOMe-seq).[94] prepared in the same way as in dscATAC-seq. The difference
is that in dsciATAC-seq individual cells are recognized by a
unique combination of the droplet-specific bead barcode and
5.3.1. dscATAC-seq the well-specific Tn5 barcode, consequently allowing the possi-
bility of overloading the cells in droplets.
In 2019, the Buenrostro lab developed a high throughput Single Cell ATAC (10X Genomics) constitutes another ver-
droplet-based single-cell assay for transposase-accessible chro- sion of dscATAC-seq with the main differences in the use of
matin using sequencing (dscATAC-seq) to unravel the single- hydrogel beads for barcoding, which dissolve in the droplets
cell chromatin profile by studying chromatin compaction and releasing the barcoding primers. Crucially, high cell barcoding
DNA-binding proteins regulating gene expression.[95] Figure 8 efficiency is reached as a result of closely packed beads in the
depicts the general workflow of dscATAC-seq. First, single channel.[96,97]
nuclei are isolated and treated in bulk by addition of transposase
enzyme (Tn5) in order to preferentially fragment the DNA in
open regions of the chromatin and to insert adapter sequences 6. Single-Cell Transcriptomics
to the ends of the DNA fragments. Then, single transposed
nuclei are encapsulated in nanoliter droplets together with bar- Single-cell transcriptomic sequencing was first reported by Tang
coding beads and PCR reagents. Barcoding beads are labeled et al. in 2009.[98] In recent years many technologies for single-
with DNA primers containing a cell barcode sequence and a cell transcriptomic sequencing emerged which do not use
PCR handle. Nuclei are heavily diluted prior to encapsulation droplet microfluidics, but: i) wells: massively parallel single-cell
in order to have one nucleus per droplet following Poisson sta- RNA-sequencing (MARS-Seq),[99] Seq-Well,[100] gene expression
tistics, while beads are loaded at higher concentrations. Drop- cytometry (CytoSeq),[101] cell expression by linear amplification
lets with multiple beads encapsulated can be identified during and sequencing (CEL-Seq),[102,103] Quartz-Seq,[104] Smart-seq,[105]
analysis thanks to a computational approach developed by the single-cell tagged reverse transcription (STRT),[106] split-pool
group. After emulsification, the droplets are thermocycled ligation-based transcriptome sequencing (SPLiT-seq),[107]
allowing single-nucleus barcoding of the accessible DNA frag- single-cell mRNA 3-prime end sequencing (SC3-seq),[108] single
ments, PCR amplification of the barcoded product and prepara- cell RNA barcoding and sequencing (SCRB-seq),[109] and ii)
tion of ready to sequence libraries. valve microfluidics: Hydro-Seq.[110] Although some of these
The throughput of dscATAC-seq was further improved by technologies show great promise, they are beyond the scope of
integrating it with combinatorial indexing (dsciATAC-seq) to this review since they do not make use of droplet microfluidics.
beat Poisson loading of cells in droplets (limited dilution of cells Targeted single-cell mRNA analysis in hydrogel beads was
in order to have one cell per droplet). Briefly, in dsciATACseq, developed in 2016 by Rakszewska et al.[111] using droplet micro-
the Tn5 transposase is loaded with barcoded DNA adapter to fluidics. The number of transcripts per cells was assessed by
Figure 8. Overview of the workflow of droplet-based single-cell assay for transposase-accessible chromatin using sequencing (dscATAC-seq). Single
Cell ATAC has the same workflow apart from the barcoding beads encapsulation. In dscATAC-seq beads are super-Poisson loaded in the droplets and
the presence of doublets beads is corrected computationally, while in Single Cell ATAC there are almost no bead doublets encapsulated in droplets
thanks to the high encapsulation efficiency reached using deformable hydrogel beads.
counting the number of fluorescent dots inside each gel bead. cDNA which is single-cell barcoded. After RT, the single-cell
However, this technology had several limitations: time-con- barcoded cDNA material is combined by breaking the emulsion
suming imaging and quantification of the fluorescent dots, and sequencing libraries are prepared following the CEL-seq
possibility to characterize only one target mRNA (which could protocol.[102,103] CEL-seq uses in vitro transcription (IVT) linear
possibly be extended to a few), and the limited range of mRNA amplification followed by RT-PCR to amplify the barcoded
strands that could be quantified before reaching overlap of fluo- cDNA product. After sequencing, the number of transcripts per
rescent spots. cell can be quantified thanks to the presence of the cell barcode
These limitations were solved by workflows using sequencing and of the UMI sequence. Figure 10 shows how the concept
as a detection method, allowing for whole-genome transcrip- of UMIs and cell barcodes can be used to retrieve quantitative
tome analysis. The field of single-cell transcriptomic profiling information (without amplification bias) about the mRNA con-
gained enormous popularity after the development of the high- tent of the cell from the mRNA libraries. Detection of ≈1*103
throughput droplet microfluidic technologies inDrop[112,113] and different genes per cell and 3*103 transcripts per cell has been
Drop-seq[114] in 2015. Even though the fundamental working reported using inDrop technology.[116]
principles of both technologies are the same, there are some
considerable differences as also discussed below.[115,116]
6.1.2. Drop-seq
6.1. Whole-Transcriptome Profiling Drop-seq[114] shares many features with inDrop, including
beads to barcode the transcriptomic material in nanoliter drop-
6.1.1. inDrop lets. Figure 9b shows the general Drop-seq workflow. Drop-seq
beads are labeled with ≈108 DNA primers comprising the PCR
inDrop (indexing droplets) uses droplet microfluidics to co- handle, the cell barcode, the UMI sequence, and a poly(dT)
encapsulate a highly diluted cell solution with barcoding tail. The key difference compared to inDrop is that Drop-seq
hydrogel beads (Figure 9a). The hydrogel beads are produced uses Poisson statistics for the encapsulation of barcoding beads
prior to the experiment, using a flow-focusing microfluidic made out of a hard resin. This allows easier emulsification but
device with as dispersed phase a solution of acrylamide:bis- reduces considerably the cell capture efficiency to 2–4% cells
acrylamide supplemented with acrydite-modified DNA barcoded compared to the 60–90% of inDrop. Second, while
primers. After the emulsion is gelated by thermal incubation, Klein et al. encapsulated the reverse transcription enzymes for
the hydrogel beads undergo two split-pool synthesis rounds to RT in nanoliter droplets (after photo-cleavage of the primers),
elongate all the primers labeling the beads with a unique cell Macosko et al. barcoded the mRNA material by RT in bulk after
barcode. 147 456 different barcodes are available allowing the de-emulsification. Importantly, the primers in Drop-seq remain
barcoding of ≈3000 cells with 99% unique labeling. attached to the solid bead during the RT reaction in bulk. Perfor-
An important improvement for the technologies using detec- mance of RT in bulk constitutes a potential advantage of Drop-
tion by sequencing such as inDrop and Drop-seq is the imple- seq compared to inDrop because it eliminates the risk of RT
mentation of a unique molecular identifier (UMI)[117,118] to inhibition in picoliter volumes and reduces the time in which
allow quantitative determination of mRNA transcripts avoiding the enzymes need to stay at room temperature before being
amplification bias. Briefly, UMIs are random sequences of used for RT. Finally, the library preparation is different. Both
4–12 bp which can be added to barcode the analyte of interest. protocols use Moloney murine leukemia virus (MMLV) reverse
After library preparation and sequencing, the amplification bias transcriptase which intrinsically adds a few C nucleotides at the
can be removed by enumerating the distinct number of UMIs 3’ end of the cDNA only when the reverse transcriptase reaches
aligned to each position. the end of the mRNA template but not when the reverse tran-
Each hydrogel bead used for inDrop contains ≈109 covalently scription is prematurely terminated. Drop-seq does make use
bound barcoding primers, each of them containing a photo- of the introduction of C nucleotides at the 3’ end of the cDNA
cleavable spacer, a T7 promoter region, PCR handle, cell bar- to add a PCR handle by using “SMART” technology.[105,120,121]
code, UMI and poly(dT) tail (Figure 11). Briefly, after RT of the mRNA in cDNA, a template switch oligo
Cell encapsulation follows Poisson encapsulation, while (TSO) primer with a PCR handle and a poly(dG) tail is used
hydrogel beads are loaded in a super Poisson fashion, yielding to hybridize to the untemplated C nucleotides on the cDNA
60–90% of droplets with exactly one barcoding bead encapsu- 3′ end, yielding a single-cell barcoded full-length cDNA with
lated and consequently 60–90% cell barcoding efficiency. The a PCR handle at the 3′. After reverse transcription with tem-
deformability of hydrogel beads allows them to closely pack plate switching, the barcoded cDNA material is amplified only
in the channel and order into a stream with regular spacing, by PCR. Importantly, the use of “SMART” technology consider-
providing a regular flow of beads, effectively beating Poisson ably increases the percentage of full-length clones compared to
encapsulation statistics.[119] standard barcoding protocols, since only the full-length clones
After co-encapsulation of single cells with barcoding beads, are PCR-amplified. Removal of the IVT step in Drop-seq has
reverse transcription (RT) and lysis mix in droplets, the cells the advantage to reduce the overall time required for library
are lysed and the primers labeling the hydrogel beads are preparation, although linear amplification has the potential
photo-released by UV exposure. Upon thermal incubation, the advantage of minimizing later PCR amplification bias.
poly(A) tail of the mRNA material hybridizes to the poly(dT) tail Detection of ≈2*103 different genes per cell and 7*103 tran-
of the barcoding beads and mRNA is reversely transcribed into scripts per cell has been reported using Drop-seq technology.[116]
Figure 9. Comparison of droplet microfluidic single-cell transcriptome workflows. a) inDrop, b) Drop-seq, c) Single Cell Gene Expression 3’ (10X
Genomics). R1: read 1. R2: read 2.
Interestingly, Drop-seq technology has also been adapted to 6.1.3. Hi-SCL

analyze single nuclei gene expression (single nuclei droplet-
based sequencing: snDrop-seq).[91] The only modification is High-Throughput Single-Cell Labeling (Hi-SCL) was devel-
related to the sample preparation since in snDrop-seq single- oped by Weitz lab in 2015 contemporarily to inDrop and Drop-
nuclei instead of single cells are encapsulated in droplets with seq.[123] Hi-SCL encapsulates single-cells in droplets with lysis
barcoding beads, allowing for the capture of nuclear-polyade- buffer. In parallel, barcoding water-in-oil droplets are prepared
nylated mRNA transcripts as well as pre-mRNAs. by emulsifying ≈109 barcodes per droplet, similarly to Drop-
In addition, Drop-seq has also been modified to allow simul- ChIp (Section 5.2.1). After cell lysis, the droplets containing the
taneous measurement of the transcriptome and a few targeted single-cell content are fused by electrocoalescence together with
RNA amplicons (droplet-assisted RNA targeting by single-cell two additional droplets: one containing the barcoding DNA
sequencing: DART-seq).[122] In this case, custom primers are primers and one containing the RT mix. Upon RT, single-cell
enzymatically attached to a subset of poly(dT) tailed primers on barcoded cDNA is produced by exploiting the annealing of the
the Drop-seq barcoding beads and subsequently, the Drop-seq poly(dT) tail of the barcoding primers with the poly(A) tail of
protocol can be used unaltered. the mRNA. The droplets are then de-emulsified, the cDNA is
Figure 10. General concept of using cell barcodes and UMIs to remove amplification bias and allow quantitative determination of mRNA transcripts.
Reads having the same cell barcode sequence are derived from the same single cell. Reads having the same cell barcode, the same transcript read, but
different UMIs, originate from different unique molecules. Reads with the same cell barcode, the same transcript read and the same UMI, result from
PCR amplification and originate from one single molecule.
purified and the library is prepared by PCR amplification, fol- when the reverse transcriptase reaches the end of the mRNA
lowed by sequencing and data analysis. template and not when the reverse transcription is prematurely
The major drawback of Hi-SCL is that the barcoding primers terminated. Subsequently, the poly(dG) sequence (on the bar-
do not have the UMI sequence, preventing quantification of coding primers) hybridizes to the untemplated C nucleotides
the transcripts. In addition, the number of available barcode on the cDNA 3′ end switching oligo template and leading to
sequences in Hi-SCL is considerably lower (≈103) compared to the transcript extension, that yields a single-cell barcoded full-
the other technologies presented above (≈105 inDrop, ≈7*105 length cDNA.
Single Cell Gene Expression System, ≈107 Drop-seq) limiting
the number of cells that can processed per run in order to avoid
repetition of the same barcode for two different cells. 6.1.5. Comparison of Single-Cell Transcriptomic Technologies
All technologies described above have advantages and dis-

6.1.4. Commercial Developments advantages due to the implementation of slightly different
protocols. The interested reader may refer to recent papers
The widely used 10X Genomics Chromium platform[124] comparing single-cell transcriptomic methods.[116,128] In Table 1
combines characteristics of inDrop and Drop-seq technolo- an overview of the detailed specifications of the aforementioned
gies with some modifications for 3′ gene expression analysis. droplet microfluidic technologies for transcriptome analysis
Figure 9c shows the workflow of Single Cell Gene expression 3′. is provided. Worth stressing is that the number of genes and
Co-encapsulation of barcoding bead and single cells using droplet UMIFM per cell are highly cell type dependent and they are
microfluidics follows the inDrop protocol, while the library provided here for comparison using the same cell type in
preparation consists only of reverse transcription with template inDrop, Drop-seq, and Single Cell Gene Expression.
switching and PCR amplification, similarly to Drop-seq. It should be stressed that cell hashing antibodies introduced
The key difference between Single Cell Gene expression 3′ by Stoeckius et al. in 2018[129] present a solution that can be
and inDrop or Drop-seq is the use of primer-carrying beads that incorporated in all technologies to enable sample multiplexing
dissolve upon co-encapsulation of cells. It might be that dis- (processing of multiple samples in parallel). Briefly, cells from
solution of the beads is a crucial factor in the higher number different samples are uniquely labeled by staining them with
of genes and transcripts identified per cell that have been oligo-tagged antibodies against ubiquitously expressed surface
reported: ≈3*103 different genes per cell and ≈17*103 tran- proteins. After staining, the cells derived from different sam-
scripts per cell.[116] ples can be pooled together and encapsulated using the micro-
All the above-described methods use cDNA construction fluidic system of choice. By sequencing the cell hashing tags
methods which allow for 3′ coverage of the mRNA since RT alongside the cellular transcriptome, each cell can be assigned
starts from the poly(A) tail, but lead to an underrepresentation to its original sample, considerably reducing batch effects. Cell
of the 5′ end of the mRNA. The Chromium platform allows for hashing not only helps to avoid technology-dependent batch
Single Cell Gene expression 5′ by including a TSO sequence effects but also makes it possible to load cells in droplets at
at the end of the barcoding primers instead of a poly(dT) tail higher concentrations compared to what is usually allowed by
and primers with a poly(dT) tail in the RT.[125–127] Briefly, in Poisson distribution.
the 5′assay, the poly(dT) primers bind to the poly(A) tail of the It is worth mentioning that one of the key differences
mRNA and the mRNA is reverse transcribed by MMLV which between the various single-cell transcriptomic technologies is
also adds a few C nucleotides at the 3′ end of the cDNA only the barcoding beads that are used. Their physical properties
Table 1. Comparison of specifications of droplet microfluidic technologies for single-cell whole-transcriptome profiling.
Technology Throughput (cells per run) Cell barcoding Approximate duration of an Genes per cell UMIFM per cell Type of profiling Commercial
efficiency experiment supplier
inDrop 100–3000 60–90% 4.5 h for cell barcoding ≈1*103 3*103 3’ expression 1CellBio
+ 2 days of library prep
Drop-seq 100–10.000 2–4% 2h for cell barcoding ≈2*103 7*103 3’ expression Dolomite Bio
+ 12 h of library prep
Single Cell Gene 100–10.000 per lane 50–65% 1.5 h for cell barcoding ≈3*103 ≈17 *103 3’ or 5’ expression 10X Genomics
Expression (eight lanes per chip) + 8h of library prep (user can decide)
Hi-SCL 20 to have 1% multi-bar- ≈50% 4.5 h for cells barcoding ≈1*103 Not applicable 3’ expression No
coding error, 100 used in[123]
leads to 4% multi-barcoding
error
Throughput: the maximum number of cells that can be barcoded per sequencing run in each technology is calculated based on two main considerations: i) the number of
different barcodes possibilities (1.47*105 for inDrop, 1.6*107 for Drop-seq, 7.34*105 for Single Cell Gene Expression, and 1.15*103 for Hi-SC) from which we calculated
the maximum number of cells that can be barcoded setting the multi-barcoding error to 1% and using the so-called “birthday problem” as explained in the supplemental
experimental procedures of Klein et al.[112]; ii) the time and reagents constraints which are particularly pronounced for techniques with low cell barcoding efficiency.
Approximate duration of an experiment: the cell barcoding step comprises the system setup, the microfluidic cell encapsulation and the reverse transcription when it is
done before de-emulsification. Genes per cells and UMIFM per cell (UMIFM stands for UMI filtered mapped) for inDrop, Drop-seq, and Single Cell Gene Expression were
taken from the comparative analysis of Zhang et al.[116] which used the GM1289 cell line for all three technologies, while for Hi-SCL data were taken directly from the
original paper.[123]
will dictate the encapsulation efficiency (e.g., hydrogel beads

which can deform and closely pack; dissolvable vs. undissolv-
able beads), and the DNA sequence labeling will influence the
library preparation (e.g., PCR handle inclusion, poly(dT) tail)
and the quantification of the results (e.g., UMI addition). A
comparative scheme of the different barcoding beads is pre-
sented in Figure 11.
6.2. Immune Profiling
Sequencing paired heavy- and light-chain variable regions

(VH and VL) of single B cells or alpha and beta chain variable
regions (Vα and Vβ) of single T cells is desirable to further our
understanding of the adaptive immune response, to study auto-
immunity, and for therapeutic applications. However, because
these variable regions are encoded by different mRNA tran-
scripts, they are usually analyzed separately. A low throughput
single-cell sequence analysis of VH:VL pairs can be performed
by sorting individual cells in wells and subsequently per-
forming single cell RT-PCR and sequencing.
Attempts to increase the throughput of the characteriza-
tion of VH:VL pairs were made by DeKosky et al.[130] In their
study, 104 single cells were compartmentalized in microwells
together with magnetic beads labeled with poly(dT) primers.
Subsequently, cells were lysed, and the mRNA annealed to
the magnetic beads was collected and emulsified with PCR
primers, reverse transcriptase and DNA polymerase. After
reverse transcription and linkage PCR, the resulting linked
VH:VL sequence was sequenced, enabling profiling of the
complementarity-determining regions. The throughput of
this method was later increased to 106 cells by single-cell
isolation using droplet microfluidics instead of compart- Figure 11. Comparison of the barcoding beads characteristics for the dif-
mentalization in micro wells.[131,132] In parallel, this tech- ferent single-cell transcriptomic technologies: DART-seq, inDrop, Drop-
nology has been developed commercially by GigaGen.[133,134] seq, 10X Genomics 3′, Hi-SCL, and 10X Genomics 5′.
AbVitro (later acquired by Juno Therapeutics) developed a emerged. Stoeckius et al. introduced cellular indexing of
similar approach for sequencing both BCR and TCR genes.[135] transcriptomes and epitopes by sequencing (CITE-seq) which
Droplet microfluidics was used to compartmentalize single allows simultaneous single-cell transcriptome and proteome
cells in droplets together with lysis/reaction mix containing quantification by staining cells with DNA labeled antibodies
a droplet barcode template molecule. After emulsification, prior to microfluidic processing.[141] In particular, CITE-seq
cells are lysed, mRNA is reverse-transcribed and tagged with a was shown to be compatible with Drop-seq protocol and with
molecule-specific and a droplet-specific barcode through PCR the commercial 10X Genomics Chromium platform. The
amplification. Finally, the resulting library is sequenced and general workflow of CITE-seq is shown in Figure 12a. Briefly,
the native receptor chain pairing can be reconstructed. The antibodies are conjugated by streptavidin-biotin binding to a
double-barcoding allows for the correction of the PCR amplifi- DNA sequence containing: an antibody (Ab) barcode, a PCR
cation bias and for the clustering the reads to their molecules handle, and a poly(dA) tail. Subsequently, cells are immu-
and cells of origin. nostained with DNA tagged Abs using traditional flow cytom-
Recently, 10X Genomics also commercialized a tool for etry staining protocols. Single-cells are co-encapsulated in
single-cell immune profiling.[136,137] The method uses the droplets with barcoding beads. After cell lysis and cleavage of
Chromium platform for Single Cell Gene expression 5′ (as the oligo sequence from the Abs by reduction of the disulfide
described in Section 6.1.4) but after reverse transcription and bond in droplets, the poly(A) tail of the cellular mRNA and
generation of the 5′ barcoded cDNA, the sample is divided the poly(dA) tail of the DNA strands labeling the Abs anneal
in two aliquots and separately PCR amplified with custom to the poly(dT) tail of the barcoding beads. During RT (which
primers. Finally, two separate 5′ mRNA and V(D)J enriched takes place in bulk in Drop-seq and in emulsion in the Chro-
libraries are generated. mium system) both the mRNA and the DNA Ab-labels are
single-cell barcoded by extension, as the reverse transcriptase
simultaneously extends the hybridized Ab DNA-tags and syn-
7. Single-Cell Proteomics thesizes complementary DNA from mRNA. Subsequently,
the transcriptomic and proteomic-derived material are sepa-
The culmination of genome expression is the proteome, the rated by size and PCR amplified separately, to ensure that
cell’s repertoire of activated and non-activated proteins, which the relative proportions of the libraries can be adjusted sep-
specifies the nature of the biochemical reactions that the cell is arately. Finally, the two libraries are sequenced, and during
able to carry out. Capturing proteomic information from single analysis the single-cell proteomic and transcriptomic data can
cells has proven to be a substantial technical challenge. Multi- be grouped. The proof of concept was demonstrated using 13
color (imaging) flow cytometry is an extremely powerful method monoclonal antibodies used for staining.
for high-throughput analysis of single-cell protein levels but is An interesting development of CITE-seq is ECCITE-seq
limited to ≈15–30 proteins mostly due to the spectral overlap of (Expanded CRISPR-compatible CITE-seq) allowing for multi-
the different fluorophores. By staining cells with isotope-labeled modal single-cell assays.[142]
antibodies instead of fluorescent labels, CyTOF (cytometry by
time of flight) allows simultaneous high-throughput (up to 1000
cells per second) quantification of up to 35 proteins (with the 7.1.2. REAP-seq
availability of more isotopes up to 100 markers could be meas-
ured).[138–140] Briefly, cells with bound antibody-isotope conju- RNA expression and protein sequencing assay (REAP-seq)
gates are nebulized and their atomic constituents are ionized was introduced by Peterson et al. in 2017,[143] almost at the
through an argon plasma. The resulting elements are analyzed same time as CITE-seq. The initial study used 82 antibodies
by time-of-flight mass spectrometry to determine the proper- for staining. The working principle of REAP-seq (Figure 12b)
ties of the cell. CyTOF has proven to be an extremely powerful is very similar to CITE-seq and also allows the simultaneous
technique; however, it is still limited in the number of pro- quantification of surface proteins with mRNA at a single-cell
teins that can be analyzed and only membrane proteins can be level making use of droplet microfluidics. One difference
characterized. between the two technologies is the conjugation chemistry of
In the following part of the review, technologies for high the oligonucleotides to the antibody. Instead of streptavidin-
throughput single-cell proteomic analysis making use of biotin binding, amine chemistry is used. Another difference
droplet microfluidics technology for membrane protein anal- is related to the (not)release of the antibody labels. In REAP-
ysis, as well as secreted proteins characterization are detailed. seq, after co-encapsulation in droplets of the stained cell
and the barcoding bead, the oligos labeling the antibodies
are not released into solution, and barcoding by RT is per-
7.1. Membrane Proteins formed with the antibody DNA label still conjugated to the
antibody.
7.1.1. CITE-seq Both technologies can be adapted to work with all the droplet
microfluidic technologies for single-cell transcriptomic analysis
After the development of single-cell transcriptomic analysis reviewed above. In particular, REAP-seq was shown to be com-
tools such as inDrop and Drop-seq, the possibility of using patible with the commercial 10X Genomics Chromium plat-
similar or combined technologies (single-cell transcrip- form. One common limitation of CITE-seq and REAP-seq is
tome analysis) platforms for single-cell proteomic analysis their ability to characterize only surface proteins.
Figure 12. Workflows of poly(dT)-based methods for membrane protein quantification. a) CITE-seq implemented with Drop-seq microfluidics and
library preparation, b) REAP-seq implemented on 10X Genomics Chromium platform. R1: read 1. R2: read 2.
7.1.3. Abseq is further PCR amplified to obtain clonal populations. Each

barcoding primer contains the cell barcode sequence and a
Abseq was developed by the Abate group, also in 2017.[144] capture sequence. The single cells droplets are obtained by
The basic approach used is the same as CITE-seq and REAP- emulsification of cells immunostained with DNA-tagged
seq: labeling antibodies with DNA tags, staining cells with Abs (containing a PCR handle, a UMI, Ab barcode, capture
such labeled antibodies before microfluidic encapsulation, sequence) with proteinase K (for cell lysis) following Poisson
sequencing of libraries, and data analysis (Figure 13a). How- distribution. A third microfluidic device is used to merge by
ever, there are a few differences compared to the preceding electrocoalescence one droplet from each emulsion with a
ones. Briefly, in Abseq the microfluidic workflow consists third droplet containing reagents for PCR amplification. After
of three devices instead of one for CITE-seq and REAP-seq. mixing of the content, each (very large, unstable) droplet
First, two emulsions are created independently: one with is split into four equal droplets, that are thermocycled and
single cells and the second one with the barcoding primers. PCR amplified. Abseq has the advantage that cells are lysed
The barcode emulsion is produced by encapsulation of one with proteinase K, the proteinase is heat-deactivated, and
DNA sequence, mixed with PCR reagents and primers that then enzymes for PCR reaction are added, which prevents
Figure 13. Workflows of non-poly(dT)-based methods for droplet microfluidic membrane protein quantification. a) Abseq, b) Feature Barcoding Tech-
nology. R1: read 1. R2: read 2.
the inhibition of PCR reaction by the cell lysate. However, 7.1.4. Commercial Developments
Abseq presents also some disadvantages compared to the
techniques already presented. First, using a capture sequence 10X Genomics commercialized, in collaboration with Bio-
different from a poly(dT) tail allows only for protein analysis legend, the characterization of both 3’gene expression and
and no mRNA. Second, as for Drop-seq, both the cells and surface protein expression at the single cells level as Feature
the barcodes are encapsulated following Poisson distribution, Barcoding Technology,[97] shown schematically in Figure 13b.
consequently, the cell barcoding efficiency is relatively low. This technology uses the Chromium System to encapsulate
Finally, the use of three microfluidic devices introduces sig- cells with barcoding beads as described above in Section 6.1.4.
nificant technological challenges. However, the hydrogel beads were redesigned (single-cell
3’v3.1 Gel Beads) to contain three different types of primers. the natural mRNA’s poly(A) tail. In contrast, Abseq and Fea-
Each primer contains a cell barcode, a UMI sequence, and a ture Barcoding Technology use a capture sequence homologous
poly(dT) sequence to bind poly-adenylated mRNA, or specific between the Ab-tag and the barcodes, specifically binding only
capture sequences for proteomic information. Cells are stained to the proteins tags and not to the mRNA.
with DNA tagged Abs. However, the DNA tags do not contain a DNA tagging of Abs is relatively straightforward, but oli-
poly(dA) tail, in contrary to CITE-seq and REAP-seq, but a cap- gonucleotide-conjugated antibodies compatible with CITE-
ture sequence. Tagged antibodies (with the capture sequence) seq and Feature Barcoding Technology are also commercially
compatible with the platform are available from their partner available.
Biolegend: TotalSeq-B. The use of different capture sequences
for mRNA and proteins has the advantage of independent bar-
coding of two analytes. mRNA and protein tags do not compete 7.2. Secreted Proteins
for the same barcoding primers.
Droplet microfluidic technologies offer the unique advantage
of confining individual cells with a stimulus and to ensure
7.1.5. Comparison of Technologies for Single-Cell Membrane entrapment of the molecules secreted by the cell. Technolo-
Proteins Analysis gies using droplet microfluidics to analyze secreted proteins
generally use fluorescence-based measurements instead of
Comparison of the benefits and drawbacks of the different strat- sequencing, making them low-cost, offering the possibility of
egies discussed above for single-cell membrane protein analysis kinetic studies, allowing phenotype visualization, and requiring
has to take into account differences related to the method itself, no bioinformatics skills. Importantly, technologies for single-
as well as to the microfluidic platform on which they are car- cell kinetic secretion measurements should have all reagents
ried out (e.g., Drop-seq, Chromium platform). A comparison of already present in the droplets to produce a fluorescent signal
the platforms is shown in Section 6.1.5. Importantly, to date, all and should be coupled with an imaging chamber keeping the
these technologies are limited to membrane protein analysis. droplets fixed in one location over time.
Recent work has demonstrated the possibility of extending this
approach to include intracellular proteins as well.[145,146] How-
ever, these technologies have only been implemented in well 7.2.1. Cytokine Secretion
plates with low cell throughput, and it would be desirable to use
them with high-throughput droplet microfluidics. The study of immune cells, and in particular of their cytokine
All the abovementioned techniques for single-cell prot- secretion at a single-cell level upon stimulation, is important
eomic analysis make use of DNA-tagged Abs (Figure 14). Mul- to understand immune cell responses and cell-to-cell response
tiple methods can be used to tag a DNA strand to the Abs: i) heterogeneity upon exposure to a stimulus. Droplet micro-
CITE-seq uses streptavidin-biotin binding, ii) REAP-seq: amine fluidics provides an ideal platform for the study of single-cell
chemistry using the commercially available Thunder-Link cytokine secretion by maintaining the cells under precisely
PLUS Oligo Conjugation System, and iii) Abseq: a bifunctional defined conditions.
crosslinker reactive toward thiol (via maleimide) and amine (via To date, a number of tools have been developed using
NHS) moieties. droplet microfluidics for single-cell cytokine secretion meas-
Moreover, CITE-seq and REAP-seq use a poly(dA) tail as a urement.[147–149] A comparison of the workflows for these
capture sequence, in this way the single-cell proteome study can techniques is depicted in Figure 15. In all cases, single cells
be easily integrated to the transcriptome study by mimicking are encapsulated in droplets at a limiting dilution with a
Figure 14. Comparison of the labeled antibodies used for single-cell proteome analysis protocols (CITE-seq, REAP-seq, Abseq, and Feature Barcoding
Technology, which uses TotalSeq-B antibodies). Different chemistries are adopted for the binding of the DNA sequence to the antibody: streptavidin-
biotin binding (CITE-seq), bifunctional crosslinker (Abseq), and amine chemistry (REAP-seq and TotalSeq-B). Different DNA sequences are used:
PCR handle, Ab barcode and poly(dA) tail for CITE-seq and REAP-seq, and PCR handle, Ab barcode, and capture sequence for Abseq and TotalSeq-B.
Figure 15. Comparison of droplet microfluidic technologies for single-cell cytokine secretion measurements. a) Membrane-anchored aptamer sensor,
b) Catch reagent construct bound to the cell surface for cytokine capture, c). Antibody-labeled beads for cytokine capture.
capture construct specifically binding to cytokines. Addition- encapsulated in the droplet.[149] After co-encapsulation of the
ally, a stimulus can be added in order to characterize the single cells with the capture construct, the emulsion is incu-
single-cell response. Once the cytokines bind to the capture bated to allow secretion of cytokines in droplets.
construct and are optionally stained, a fluorescent signal is Qiu et al. reported the use of aptamers directly anchored
emitted that can be visualized by flow cytometry or fluores- on the cell surface (membrane-anchored aptamer sensor),[147]
cence microscopy. which has the advantage that all the steps are performed
The main difference between the tools for cytokine secre- in droplets and in principle, the cytokine production can
tion analysis is in the engineering of the capture reagent which be monitored over time (Figure 15a). Briefly, aptamers are
contains two parts: one specifically binding to the cytokines linked to a cholesterol tail and anchored to the cell surface by
and the other binding to a solid support (e.g., bead, cell sur- hydrophobic interaction between the cholesterol and the cel-
face). Antibodies[148,149] as well as aptamers[147] have been used lular phospholipid layer. The two ends of the aptamers are
to specifically capture secreted cytokines. The capturing con- labeled with a fluorophore and a quencher. In the absence of
struct is directly bound to the cell surface[147,148] or to beads co- cytokines, the aptamers self-hybridize into a hairpin structure,
keeping the fluorophore and the quencher in close proximity, antibodies, and fluorescently labeled antigens. The droplets
resulting in quenched fluorescence. However, upon binding of are then immobilized in an observation chamber to form a 2D
the cytokines, the aptamers will switch into a specific tertiary droplet array and imaged using fluorescence microscopy. The
structure, separating the fluorophore and the quencher, thus secreted antibodies bind to the capture beads and are visualized
resulting in an increase in fluorescence signal. by the binding of a secondary fluorescently labeled antibody
Instead of aptamers, Abs can directly bind to the cell surface present in droplet. The application of a magnetic field induces
as shown by Wimmers et al. who use a bifunctional construct the formation of elongated, easily observable nanoparticle
labeled on the cell surface for cytokine capture.[148] In this case, aggregates in each droplet, termed beadlines. The single-cell
after emulsification, incubation, and binding of the cytokines antibody secretion rate is measured by fluorescence intensity
to the cell surface, the emulsion is broken (Figure 15b). There- of the detection antibody localized on the beadline and the anti-
after, a fluorescently labeled antibody is introduced to bind to bodies affinity for the antigen by fluorescence intensity meas-
the secreted cytokines bound at the cell surface. urement of the antigen localized to the beadline. DropMap was
Finally, Abs for the capture of cytokines have been immobi- commercialized by HiFiBio in 2019.
lized on polystyrene beads.[149] The workflow of this technique Finally, Debs et al. developed an assay to test secreted anti-
is shown in Figure 15c. Single cells are co-encapsulated with body functionality—defined as their capability of binding and
cytokine-capture beads in monodisperse agarose droplets. After inhibiting the activity of a target,[157] Figure 17. Single cells are
co-encapsulation, incubation, and binding of the cytokines to co-encapsulated with the Ab target in droplets. After Ab secre-
the capture beads, the agarose droplets are gelated in agarose tion, these droplets are merged with a smaller droplet con-
beads. Subsequently, the emulsion is broken and the obtained taining a FRET peptide which is the substrate for the target.
agarose beads are stained in bulk with fluorescently labeled A high fluorescent signal indicates the production of non
antibodies against the secreted cytokines. inhibiting antibodies.
7.2.2. Antibody Secretion 8. Single-Cell Metabolomics

The capability of B cells to secrete antibodies is a fundamental Traditionally, the metabolome is studied using four major
step in the immune response towards pathogens. Immuno- methods: mass spectrometry, fluorescence-based detection, fluo-
globulin G (IgG) is the most abundant type of antibody found rescence biosensors, and FRET biosensors. Measuring metabo-
in the blood.[150,151] The pool of antibody-secreting cells is heter- lites at the single-cell level is very difficult because of the very
ogeneous[152] and the study of the antibody secretion rate and of high diversity of compounds that encompass the metabolome,
the antibody specificity is of particular interest for many appli- and their large dynamic range (from few molecules per cell to
cations, for example, to monitor immune responses during 1010 molecules for major metabolites in larger cells), which is
disease and therapy, to optimize immunization and vaccina- highly variable (environmental cues vary its generation on a
tions protocol, to facilitate the generation of high-quality mono- time scale of seconds or even milliseconds). Droplet microflu-
clonal antibodies and for hybridoma selection.[153–155] Droplet idics could play a role in single-cell metabolomics studies as
microfluidics offers the unique capability of entrapping single they provide a controlled micro-environment. The main draw-
cells with their secreted antibodies and has been used to study backs of the technologies developed until now for single-cell
single-cell antibody secretion over time,[34] the secreted anti- metabolomics analysis using droplet microfluidics is that they
body affinity to specific antigens,[156] and the secreted antibody are limited to the study of only one or a few metabolites (out
functionality by binding and modulating (typically inhibiting) of the 114 100 metabolites present in the Human Metabolome
the activity of the target.[157] Database),[158] and that they are not easily tunable to the study of
Antibody secreting cells were sorted by Mazutis et al. by other metabolites than the one they were designed to work for.
means of a fluorescence signal localized at the surface of cap-
ture beads[34] as shown in Figure 16a. Briefly, individual cells
are encapsulated in picoliter droplets together with single 8.1. Detection of Cancer Cells (Lactate Secretion)
capture beads (coated with anti IgG antibodies) and fluores-
cently labeled Abs. After emulsification, droplets are incubated Single-cell lactate secretion has been measured by making
off-chip to allow the cells to secrete the antibodies of interest, use of droplet microfluidics for circulating cancer cells detec-
the antibodies to bind to the capture bead, and the fluores- tion due to their aberrant metabolism (Warburg effect),[159]
cent probes to localize at the surface of the capture bead by see Figure 18a. Single cells are emulsified in picoliter drop-
binding to the captured antibodies. Subsequently, the emulsion lets with glucose and a ratiometric dye. Subsequently, the
is re-injected in a microfluidic sorting device. The localized droplets are incubated to allow the cells to consume glucose
fluorescent signal on the capture beads is then used for high- and secrete lactate, consequently acidifying the droplets. The
throughput screening and sorting of antibody-secreting cells. single-cell metabolism measurement is performed indirectly
Secretion of antibodies binding to specific antigens by by measuring the pH of the cell-containing droplet as deter-
individual cells was studied over time using DropMap tech- mined by the fluorescence intensity at two different wave-
nology,[156] Figure 16b. Single cells are compartmentalized in lengths. Cancer cells, with higher lactate production compared
droplets together with ≈1300 magnetic nanoparticles pre-coated to white blood cells, are detected by the higher fluorescence
with a capturing molecule, fluorescently labeled detection ratio of the droplets in which they are encapsulated. The
Figure 16. Comparison of the different droplet microfluidic technologies for single-cell antibody secretion measurements. a) Sorter of secreting cells.
b) DropMap for the study of antibody secretion and affinity to specific antigens.
main advantage of this method is that droplet pH is meas- another microfluidic device which fuses, by electrocoalescence,
ured in real time and droplets of interest can be sorted for a second droplet that contains the assay enzyme solution.
further downstream processing. However, it allows only for Upon incubation, the enzymatic assay activates a fluorescent
one time-point measurement of the single-cell metabolism. substrate in the presence of the metabolite of interest. Finally,
Combination of this technology with the DropMap visualiza- droplets are reinjected in a third microfluidic device where the
tion chamber could be interesting for capturing the single-cell presence of the metabolite is measured indirectly by fluores-
metabolism over time. cence. This third device can also present a sorting unit to sort
bacteria expressing high levels of the metabolite of interest as
shown by Hammar et al.[161]
8.2. Screening of Bacterial Mutant Libraries (Ethanol or Lactate
Secretion)
9. Conclusions and Outlook
Droplet microfluidics has also been used to distinguish and
sort genetically engineered bacteria, in particular genetically It is clear that droplet microfluidics has revolutionized our
modified cyanobacteria from wild-type ethanol-producing[160] ability to perform quantitative measurements at the single-cell
and lactate-producing[161] strains, as shown in Figure 18b. In level. High-resolution data from individual cells provide com-
both cases, after cell encapsulation in droplets, the emulsion is pletely new insights into cellular processes. Although enor-
incubated to allow bacteria to secrete the metabolite of interest mous progress in single-cell analysis technologies has been
(ethanol or lactate). Subsequently, the emulsion is injected into made, there is still great potential for future technological
Figure 17. Droplet microfluidic technology for functional screening of single-cell secreted antibodies.
developments. Most current approaches are limited to a single cell level is a highly attractive target, and some of the platforms
output type of information (e.g., DNA, RNA, or proteins). Quan- described in this review (REAP-seq, CITE-seq, and Feature Bar-
titatively capturing multiple layers of information at the single- coding Technology) are already moving in this direction.
Figure 18. Workflows of the droplet microfluidic technologies for single-cell metabolomics measurements. a) Detection of cancer cells (lactate secre-
tion). b) Screening of bacterial mutant libraries (ethanol or lactate secretion).
A major challenge in all single-cell approaches is a lack of Conflict of Interest

information about dynamic processes (e.g., differentiation,
The authors declare no conflict of interest.
proliferation), a challenge further complicated by the differ-
ences in state heterogeneity of the cells present in a sample.
Experimental cell synchronization is challenging and can
alter cell dynamics. New developments in computational anal- Author Contributions
ysis, making use of so-called “pseudotime” algorithms, have K.M. and F.R. contributed equally to this work. All authors were involved
started to provide information about the changes in cellular in the preparation of the manuscript and approved the final version.
composition (primarily transcriptome landscape) over time
without the need of a priori knowledge of marker genes.[162,163]
In order to reconstruct time ordering of the data, the basic
assumption applied in such algorithms is that the transcrip-
Keywords
tomic changes are a continuous function with respect to time, droplet microfluidics, genomes, proteomes, single-cell, transcriptomes
allowing researchers to place cells within a sample on a vir-
tual timeline. Received: August 16, 2019
Revised: September 30, 2019
Another inherent characteristic of single-cell technologies
Published online: November 26, 2019
is that they require tissue-dissociation, prior to single-cell
encapsulation. This inevitably means that spatial information
regarding the original position of the cells in the tissue is lost.
Spatial transcriptomics aims to combine the cellular tran- [1] J. R. Heath, A. Ribas, P. S. Mischel, Nat. Rev. Drug Discovery 2016,
scriptome information with their spatial organization within a 15, 204.
tissue, adding an extra layer of information. Spatial transcrip- [2] G.-C. Yuan, L. Cai, M. Elowitz, T. Enver, G. Fan, G. Guo, R. Irizarry,
tomic information can be derived from whole-transcriptome P. Kharchenko, J. Kim, S. Orkin, J. Quackenbush, A. Saadatpour,
profiling data by using new computational techniques, rather T. Schroeder, R. Shivdasani, I. Tirosh, Genome Biol. 2017, 18, 84.
[3] A. Colman-Lerner, A. Gordon, E. Serra, T. Chin, O. Resnekov,
than experimental ones. An example is novoSpaRc[164] which
D. Endy, C. Gustavo Pesce, R. Brent, Nature 2005, 437, 699.
enables spatial reconstruction of single-cell gene expression [4] R. P. das Neves, N. S. Jones, L. Andreu, R. Gupta, T. Enver,
de novo by assuming, similarly to the “pseudotime” algo- F. J. Iborra, PLoS Biol. 2010, 8, e1000560.
rithms, that gene expression between spatially adjacent cells [5] R. Satija, A. K. Shalek, Trends Immunol. 2014, 35, 219.
is more similar than expression levels of cells separated by [6] E. Papalexi, R. Satija, Nat. Rev. Immunol. 2018, 18, 35.
larger distances. Additionally, the Seurat package enables the [7] M. Goolam, A. Scialdone, S. J. L. Graham, I. C. Macaulay,
addition of spatial information to the sc-RNA data by cre- A. Jedrusik, A. Hupalowska, T. Voet, J. C. Marioni, M. Zernicka-
ating a spatial reference map by in situ hybridization for a Goetz, Cell 2016, 165, 61.
subset of marker genes and by combining it with single-cell [8] Q. Chen, J. Shi, Y. Tao, M. Zernicka-Goetz, Nat. Commun. 2018, 9,
gene expression.[165] This approach was successfully applied 1819.
[9] S. J. Altschuler, L. F. Wu, Cell 2010, 141, 559.
to study a developing zebrafish embryo; however, it presents
[10] D. A. Lawson, K. Kessenbrock, R. T. Davis, N. Pervolarakis,
limitations for tissues where spatial patterning varies between Z. Werb, Nat. Cell Biol. 2018, 20, 1349.
samples (e.g., solid tumors) or where expression patterns [11] H. M. Levitin, J. Yuan, P. A. Sims, Trends Cancer 2018, 4, 264.
are similar in space. As already mentioned, there are sev- [12] X. Han, R. Wang, Y. Zhou, L. Fei, H. Sun, S. Lai, A. Saadatpour,
eral techniques designed for single-cell analysis that do not Z. Zhou, H. Chen, F. Ye, D. Huang, Y. Xu, W. Huang, M. Jiang,
involve droplet microfluidics. We expect that future develop- X. Jiang, J. Mao, Y. Chen, C. Lu, J. Xie, Q. Fang, Y. Wang, R. Yue,
ments will attempt to adapt highly sensitive single-molecule T. Li, H. Huang, S. H. Orkin, G.-C. Yuan, M. Chen, G. Guo, Cell
protocols (such as nanopore sequencing of DNA, RNA, and 2018, 172, 1091.
possibly proteins) for high-throughput experiments using [13] E. Shapiro, T. Biezuner, S. Linnarsson, Nat. Rev. Genet. 2013, 14, 618.
(droplet) microfluidics. [14] A. A. Kolodziejczyk, J. K. Kim, V. Svensson, J. C. Marioni,
S. A. Teichmann, Mol. Cell 2015, 58, 610.
In future developments, we see also promising applications
[15] S. L. Anna, N. Bontoux, H. A. Stone, Appl. Phys. Lett. 2003, 82,
for droplet microfluidic diagnostic techniques that exploit the 364.
“picoliter single-cell reaction flask” principle and screen single [16] R. Staszewski, Yale J. Biol. Med. 1984, 57, 865.
cells for secreted molecules (antibodies, cytokines, enzymes, [17] S. Moon, E. Ceyhan, U. A. Gurkan, U. Demirci, PLoS One 2011, 6,
metabolites), although we note that for large-scale, high- e21580.
throughput screening, droplet generation, and interrogation [18] S. Y. Teh, R. Lin, L. H. Hung, A. P. Lee, Lab Chip 2008, 8, 198.
speeds will need to be enhanced by 1–2 orders of magnitude. [19] A. Rakszewska, J. Tel, V. Chokkalingam, W. T.S. Huck, NPG Asia
Finally, further integration of analytical techniques (one could Mater. 2014, 6, e133.
think of chromatography and single cell mass spectrometry for [20] R. Seemann, M. Brinkmann, T. Pfohl, S. Herminghaus, Rep. Prog.
the detection and characterization of small molecules and pos- Phys. 2012, 75, 016601.
[21] D. J. Collins, A. Neild, A. deMello, A.-Q. Liu, Y. Ai, Lab Chip 2015,
sibly proteins) and the sequencing methods described in this
15, 3439.
review with droplet microfluidics should bring us closer to the [22] Worthington. Worthington Tissue Dissociation Guide 2019; Available
ultimate goal of measuring the complete molecular content of from: http://www.worthington-biochem.com/tissuedissociation/
single cells. In space. And time. default.html.
[23] M. R. Emmert-Buck, R. F. Bonner, P. D. Smith, R. F. Chuaqui, [50] Z. Zhu, W. Zhang, X. Leng, M. Zhang, Z. Guan, J. Lu, C. J. Yang,
Z. Zhuang, S. R. Goldstein, R. A. Weiss, L. A. Liotta, Science 1996, Lab Chip 2012, 12, 3907.
274, 998. [51] S. W. Lim, T. M. Tran, A. R. Abate, PLoS One 2015, 10, e0113549.
[24] J. B. Geigl, M. R. Speicher, Nat. Protoc. 2007, 2, 3173. [52] M. Pellegrino, A. Sciambi, S. Treusch, R. Durruthy-Durruthy,
[25] C. A. Sepulveda, S. J. Richman, I. Kaur, A. Shen, S. Khorana, K. Gokhale, J. Jacob, T. X. Chen, J. A. Geis, W. Oldham,
S. L. O’Connor, R. E. Champlin, E. J. Shpall, J. D. McMannis, Blood J. Matthews, H. Kantarjian, P. A. Futreal, K. Patel, K. W. Jones,
2007, 110, 1216. K. Takahashi, D. J. Eastburn, Genome Res. 2018, 28, 1345.
[26] A. W. Hamburger, F. E. Dunn, C. P. White, Br. J. Cancer 1985, 51, 253. [53] C. Gawad, W. Koh, S. R. Quake, Nat. Rev. Genet. 2016, 17, 175.
[27] C. Alix-Panabières, K. Pantel, Nat. Rev. Cancer 2014, 14, 623. [54] H. K. Telenius, N. P. Carter, C. E. Bebb, M. Nordenskjöld,
[28] M. B. Lambros, G. Seed, S. Sumanasuriya, V. Gil, M. Crespo, B. A. J. Ponder, A. Tunnacliffe, Genomics 1992, 13, 718.
M. Fontes, R. Chandler, N. Mehra, G. Fowler, B. Ebbs, P. Flohr, [55] F. B. Dean, J. R. Nelson, T. L. Giesler, R. S. Lasken, Genome Res.
S. Miranda, W. Yuan, A. Mackay, A. Ferreira, R. Pereira, C. Bertan, 2001, 11, 1095.
I. Figueiredo, R. Riisnaes, D. N. Rodrigues, A. Sharp, J. Goodall, [56] J. G. Paez, M. Lin, R. Beroukhim, J. C. Lee, X. Zhao, D. J. Richter,
G. Boysen, S. Carreira, D. Bianchini, P. Rescigno, Z. Zafeiriou, S. Gabriel, P. Herman, H. Sasaki, D. Altshuler, C. Li, M. Meyerson,
J. Hunt, D. Moloney, L. Hamilton et al., Clin. Cancer Res. 2018, 24, W. R. Sellers, Nucleic Acids Res. 2004, 32, e71.
5635. [57] J. Gole, A. Gore, A. Richards, Y.-J. Chiu, H.-L. Fung, D. Bushman,
[29] W. E. Janssen, D. Rahn, M. Hackett, D. Coyle, M. Tomblyn, H.-I. Chiang, J. Chun, Y.-H. Lo, K. Zhang, Nat. Biotechnol. 2013,
R. C. Smilee, C. Anasetti, H. F. Fernandez, Bone Marrow Trans- 31, 1126.
plant. 2012, 47, 1520. [58] H. M. B. Seth-Smith, S. R. Harris, P. Scott, S. Parmar, P. Marsh,
[30] J. Mong, M.-H. Tan, Trends Biotechnol. 2018, 36, 511. M. Unemo, I. N. Clarke, J. Parkhill, N. R. Thomson, Nat. Protoc.
[31] L. Yin, Y. Wu, Z. Yang, C. A. Tee, V. Denslin, Z. Lai, C. T. Lim, 2013, 8, 2404.
E. H. Lee, J. Han, Lab Chip 2018, 18, 878. [59] M. Rhee, Y. K. Light, R. J. Meagher, A. K. Singh, PLoS One 2016,
[32] Y. Xu, J. Xie, R. Chen, Y. Cao, Y. Ping, Q. Xu, W. Hu, D. Wu, L. Gu, 11, e0153699.
H. Zhou, X. Chen, Z. Zhao, J. Zhong, R. Li, Sci. Rep. 2016, 6, [60] M. Hammond, F. Homa, H. Andersson-Svahn, T. J. G. Ettema,
36515. H. N. Joensson, Microbiome 2016, 4, 52.
[33] A. Gross, J. Schoendube, S. Zimmermann, M. Steeb, R. Zengerle, [61] Y. Fu, C. Li, S. Lu, W. Zhou, F. Tang, X. S. Xie, Y. Huang, Proc. Natl.
P. Koltay, Int. J. Mol. Sci. 2015, 16, 16897. Acad. Sci. USA 2015, 112, 11923.
[34] L. Mazutis, J. Gilbert, W. L. Ung, D. A. Weitz, A. D. Griffiths, [62] A. M. Sidore, F. Lan, S. W. Lim, A. R. Abate, Nucleic Acids Res.
J. A. Heyman, Nat. Protoc. 2013, 8, 870. 2016, 44, e66.
[35] F. Burgoyne, A remote syringe for cells, beads and particle injection in [63] M. Hosokawa, Y. Nishikawa, M. Kogawa, H. Takeyama, Sci. Rep.
microfluidic channels. Chips & Tips (Lab on a Chip), 2009. 2017, 7, 5199.
[36] E. Brouzes, M. Medkova, N. Savenelli, D. Marran, M. Twardowski, [64] Z. Yu, S. Lu, Y. Huang, Anal. Chem. 2014, 86, 9386.
J. B. Hutchison, J. M. Rothberg, D. R. Link, N. Perrimon, [65] L. Huang, F. Ma, A. Chapman, S. Lu, X. S. Xie, Annu. Rev.
M. L. Samuels, Proc. Natl. Acad. Sci. USA 2009, 106, 14195. Genomics Hum. Genet. 2015, 16, 79.
[37] N. Sinha, N. Subedi, F. Wimmers, M. Soennichsen, J. Tel, J. Vis. [66] N. Estévez-Gómez, T. Prieto, A. Guillaumet-Adkins, H. Heyn,
Exp. 2019, e57848, https://doi.org/10.3791/57848. S. Prado-López, D. Posada, bioRxiv:443754, 2018.
[38] J. F. Edd, D. Di Carlo, K. J. Humphry, S. Köster, D. Irimia, [67] F. Lan, B. Demaree, N. Ahmed, A. R. Abate, Nat. Biotechnol. 2017,
D. A. Weitz, M. Toner, Lab Chip 2008, 8, 1262. 35, 640.
[39] E. W. M. Kemna, R. M. Schoeman, F. Wolbers, I. Vermes, [68] N. Andor, B. T. Lau, C. Catalanotti, V. Kumar, A. Sathe,
D. A. Weitz, A. van den Berg, Lab Chip 2012, 12, 2881. K. Belhocine, T. D. Wheeler, A. D. Price, M. Song, D. Stafford,
[40] Y. Marcy, C. Ouverney, E. M. Bik, T. Lösekann, N. Ivanova, Z. Bent, L. DeMare, L. Hepler, S. Jett, B. K. Lin, S. Maheshwari,
H. G. Martin, E. Szeto, D. Platt, P. Hugenholtz, D. A. Relman, A. J. Makarewicz, M. Rahimi, S. S. Sawhney, M. Sauzade, J. Shuga,
S. R. Quake, Proc. Natl. Acad. Sci. USA 2007, 104, 11889. K. Sullivan-Bibee, A. Weinstein, W. Yang, Y. Yin, M. A. Kubit,
[41] Y. Xu, F. Zhao, Protein Cell 2018, 9, 501. J. Chen, S. M. Grimes, C. J. Suarez, G. A. Poultsides et al., bioRxiv:
[42] Y. Wang, J. Waters, M. L. Leung, A. Unruh, W. Roh, X. Shi, K. Chen, 445932, 2018.
P. Scheet, S. Vattathil, H. Liang, A. Multani, H. Zhang, R. Zhao, [69] G. Kelsey, O. Stegle, W. Reik, Science 2017, 358, 69.
F. Michor, F. Meric-Bernstam, N. E. Navin, Nature 2014, 512, 155. [70] I. D. Karemaker, M. Vermeulen, Trends Biotechnol. 2018, 36, 952.
[43] M. L. Leung, A. Davis, R. Gao, A. Casasent, Y. Wang, E. Sei, [71] V. Greger, E. Passarge, W. Höpping, E. Messmer, B. Horsthemke,
E. Vilar, D. Maru, S. Kopetz, N. E. Navin, Genome Res. 2017, 27, Hum. Genet. 1989, 83, 155.
1287. [72] P. A. Jones, S. B. Baylin, Nat. Rev. Genet. 2002, 3, 415.
[44] A. Saadatpour, S. Lai, G. Guo, G.-C. Yuan, Trends Genet. 2015, 31, [73] D. Mooijman, S. S. Dey, J.-C. Boisset, N. Crosetto, A. van Oudenaa
576. rden, Nat. Biotechnol. 2016, 34, 852.
[45] K. R. Upton, D. J. Gerhardt, J. S. Jesuadian, S. R. Richardson, [74] C. Zhu, Y. Gao, H. Guo, B. Xia, J. Song, X. Wu, H. Zeng, K. Kee,
F. J. Sánchez-Luque, G. O. Bodea, A. D. Ewing, C. Salvador- F. Tang, C. Yi, Cell Stem Cell 2017, 20, 720.
Palomeque, M. S. van der Knaap, P. M. Brennan, A. Vanderver, [75] C. Luo, C. L. Keown, L. Kurihara, J. Zhou, Y. He, J. Li, R. Castanon,
G. J. Faulkner, Cell 2015, 161, 228. J. Lucero, J. R. Nery, J. P. Sandoval, B. Bui, T. J. Sejnowski,
[46] M. L. Leung, Y. Wang, J. Waters, N. E. Navin, Genome Biol. 2015, 16, T. T. Harkins, E. A. Mukamel, M. M. Behrens, J. R. Ecker, Science
55. 2017, 357, 600.
[47] H. C. Fan, J. Wang, A. Potanina, S. R. Quake, Nat. Biotechnol. [76] R. M. Mulqueen, D. Pokholok, S. J. Norberg, K. A. Torkenczy,
2011, 29, 51. A. J. Fields, D. Sun, J. R. Sinnamon, J. Shendure, C. Trapnell, B.
[48] P. Kumaresan, C. J. Yang, S. A. Cronier, R. G. Blazej, R. A. Mathies, J. O’Roak, Z. Xia, F. J. Steemers, A. C. Adey, Nat. Biotechnol. 2018,
Anal. Chem. 2008, 80, 3522. 36, 428.
[49] Y. Zeng, R. Novak, J. Shuga, M. T. Smith, R. A. Mathies, Anal. [77] C. Luo, A. Rivkin, J. Zhou, J. P. Sandoval, L. Kurihara, J. Lucero,
Chem. 2010, 82, 3183. R. Castanon, J. R. Nery, A. Pinto-Duarte, B. Bui, C. Fitzpatrick,
C. O’Connor, S. Ruga, M. E. Van Eden, D. A. Davis, D. C. Mash, [104] Y. Sasagawa, I. Nikaido, T. Hayashi, H. Danno, K. D. Uno, T. Imai,
M. M. Behrens, J. R. Ecker, Nat. Commun. 2018, 9, 3824. H. R. Ueda, Genome Biol. 2013, 14, 3097.
[78] Y. Zhu, Z. Cao, C. Lu, Analyst 2019, 144, 1904. [105] D. Ramsköld, S. Luo, Y.-C. Wang, R. Li, Q. Deng, O. R. Faridani,
[79] S. A. Smallwood, H. J. Lee, C. Angermueller, F. Krueger, H. Saadeh, G. A. Daniels, I. Khrebtukova, J. F. Loring, L. C. Laurent,
J. Peat, S. R. Andrews, O. Stegle, W. Reik, G. Kelsey, Nat. Methods G. P. Schroth, R. Sandberg, Nat. Biotechnol. 2012, 30, 777.
2014, 11, 817. [106] S. Islam, U. Kjällquist, A. Moliner, P. Zajac, J.-B. Fan,
[80] E.-A. Raiber, R. Hardisty, P. van Delft, S. Balasubramanian, Nat. P. Lönnerberg, S. Linnarsson, Genome Res. 2011, 21, 1160.
Rev. Chem. 2017, 1, 0069. [107] A. B. Rosenberg, C. M. Roco, R. A. Muscat, A. Kuchina, P. Sample,
[81] H. K. Komori, S. A. LaMere, A. Torkamani, G. T. Hart, Z. Yao, L. T. Graybuck, D. J. Peeler, S. Mukherjee, W. Chen,
S. Kotsopoulos, J. Warner, M. L. Samuels, J. Olson, S. R. Head, S. H. Pun, D. L. Sellers, B. Tasic, G. Seelig, Science 2018, 360, 176.
P. Ordoukhanian, P. L. Lee, D. R. Link, D. R. Salomon, Genome [108] T. Nakamura, Y. Yabuta, I. Okamoto, S. Aramaki, S. Yokobayashi,
Res. 2011, 21, 1738. K. Kurimoto, K. Sekiguchi, M. Nakagawa, T. Yamamoto, M. Saitou,
[82] C. Lu, Drop-BS: High-Throughput Single-Cell Bisulfite Sequencing on Nucleic Acids Res. 2015, 43, e60.
a Microfluidic Droplet Platform. Virginia Tech, Virginia 2018. [109] M. Soumillon, D. Cacchiarelli, S. Semrau, A. van Oudenaarden,
[83] V. W. Zhou, A. Goren, B. E. Bernstein, Nat. Rev. Genet. 2011, 12, 7. T. S. Mikkelsen, bioRxiv:003236, 2014.
[84] M. Esteller, Nat. Rev. Genet. 2007, 8, 286. [110] Y.-H. Cheng, Y.-C. Chen, E. Lin, R. Brien, S. Jung, Y.-T. Chen,
[85] P. J. Park, Nat. Rev. Genet. 2009, 10, 669. W. Lee, Z. Hao, S. Sahoo, H. Min Kang, J. Cong, M. Burness,
[86] A. Rotem, O. Ram, N. Shoresh, R. A. Sperling, A. Goren, S. Nagrath, M. S. Wicha, E. Yoon, Nat. Commun. 2019, 10, 2163.
D. A. Weitz, B. E. Bernstein, Nat. Biotechnol. 2015, 33, 1165. [111] A. Rakszewska, R. J. Stolper, A. B. Kolasa, A. Piruska, W. T. S. Huck,
[87] K. Grosselin, A. Durand, J. Marsolier, A. Poitou, E. Marangoni, Angew. Chem., Int. Ed. 2016, 55, 6698.
F. Nemati, A. Dahmani, S. Lameiras, F. Reyal, O. Frenoy, [112] A. M. Klein, L. Mazutis, I. Akartuna, N. Tallapragada, A. Veres,
Y. Pousse, M. Reichen, A. Woolfe, C. Brenan, A. D. Griffiths, V. Li, L. Peshkin, D. A. Weitz, M. W. Kirschner, Cell 2015, 161, 1187.
C. Vallot, A. Gérard, Nat. Genet. 2019, 51, 1060. [113] R. Zilionis, J. Nainys, A. Veres, V. Savova, D. Zemmour,
[88] E. I. Campos, D. Reinberg, Annu. Rev. Genet. 2009, 43, 559. A. M. Klein, L. Mazutis, Nat. Protoc. 2017, 12, 44.
[89] S. L. Klemm, Z. Shipony, W. J. Greenleaf, Nat. Rev. Genet. 2019, 20, [114] E. Z. Macosko, A. Basu, R. Satija, J. Nemesh, K. Shekhar,
207. M. Goldman, I. Tirosh, A. R. Bialas, N. Kamitaki, E. M. Martersteck,
[90] D. A. Cusanovich, R. Daza, A. Adey, H. A. Pliner, L. Christiansen, J. J. Trombetta, D. A. Weitz, J. R. Sanes, A. K. Shalek, A. Regev,
K. L. Gunderson, F. J. Steemers, C. Trapnell, J. Shendure, Science S. A. McCarroll, Cell 2015, 161, 1202.
2015, 348, 910. [115] J. P. Junker, A. van Oudenaarden, Mol. Cell 2015, 58, 563.
[91] B. B. Lake, S. Chen, B. C. Sos, J. Fan, G. E. Kaeser, Y. C. Yung, [116] X. Zhang, T. Li, F. Liu, Y. Chen, J. Yao, Z. Li, Y. Huang, J. Wang,
T. E. Duong, D. Gao, J. Chun, P. V. Kharchenko, K. Zhang, Nat. Mol. Cell 2019, 73, 130.e5.
Biotechnol. 2018, 36, 70. [117] S. Islam, A. Zeisel, S. Joost, G. La Manno, P. Zajac, M. Kasper,
[92] W. Jin, Q. Tang, M. Wan, K. Cui, Y. Zhang, G. Ren, B. Ni, J. Sklar, P. Lönnerberg, S. Linnarsson, Nat. Methods 2014, 11, 163.
T. M. Przytycka, R. Childs, D. Levens, K. Zhao, Nature 2015, 528, [118] T. Kivioja, A. Vähärautio, K. Karlsson, M. Bonke, M. Enge,
142. S. Linnarsson, J. Taipale, Nat. Methods 2012, 9, 72.
[93] B. Lai, W. Gao, K. Cui, W. Xie, Q. Tang, W. Jin, G. Hu, B. Ni, [119] A. R. Abate, C.-H. Chen, J. J. Agresti, D. A. Weitz, Lab Chip 2009,
K. Zhao, Nature 2018, 562, 281. 9, 2628.
[94] S. Pott, eLife 2017, 6, e23203. [120] Y. Y. Zhu, E. M. Machleder, A. Chenchik, R. Li, P. D. Siebert, Bio-
[95] C. A. Lareau, F. M. Duarte, J. G. Chew, V. K. Kartha, Z. D. Burkett, Techniques 2001, 30, 892.
A. S. Kohlway, D. Pokholok, M. J. Aryee, F. J. Steemers, R. Lebofsky, [121] R. Wellenreuther, I. Schupp, The German cDNA Consortium,
J. D. Buenrostro, Nat. Biotechnol. 2019, 37, 916. A. Poustka, S. Wiemann, BMC Genomics 2004, 5, 36.
[96] A. T. Satpathy, J. M. Granja, K. E. Yost, Y. Qi, F. Meschi, [122] M. Saikia, P. Burnham, S. H. Keshavjee, M. F. Z. Wang,
G. P. McDermott, B. N. Olsen, M. R. Mumbach, S. E. Pierce, M. Heyang, P. Moral-Lopez, M. M. Hinchman, C. G. Danko,
M. R. Corces, P. Shah, J. C. Bell, D. Jhutty, C. M. Nemec, J. Wang, J. S. L. Parker, I. De Vlaminck, Nat. Methods 2019, 16, 59.
L. Wang, Y. Yin, P. G. Giresi, A. L. S. Chang, G. X. Y. Zheng, [123] A. Rotem, O. Ram, N. Shoresh, R. A. Sperling, M. Schnall-Levin,
W. J. Greenleaf, H. Y. Chang, Nat. Biotechnol. 2019, 37, 925. H. Zhang, A. Basu, B. E. Bernstein, D. A. Weitz, PLoS One 2015,
[97] J. M. Granja, S. Klemm, L. M. McGinnis, A. S. Kathiria, A. Mezger, 10, e0116328.
B. Parks, E. Gars, M. Liedtke, G. X. Y. Zheng, H. Y. Chang, [124] G. X. Y. Zheng, J. M. Terry, P. Belgrader, P. Ryvkin, Z. W. Bent,
R. Majeti, W. J. Greenleaf, bioRxiv:696328, 2019. R. Wilson, S. B. Ziraldo, T. D. Wheeler, G. P. McDermott, J. Zhu,
[98] F. Tang, C. Barbacioru, Y. Wang, E. Nordman, C. Lee, N. Xu, M. T. Gregory, J. Shuga, L. Montesclaros, J. G. Underwood,
X. Wang, J. Bodeau, B. B. Tuch, A. Siddiqui, K. Lao, M. A. Surani, D. A. Masquelier, S. Y. Nishimura, M. Schnall-Levin, P. W. Wyatt,
Nat. Methods 2009, 6, 377. C. M. Hindson, R. Bharadwaj, A. Wong, K. D. Ness, L. W. Beppu,
[99] D. A. Jaitin, E. Kenigsberg, H. Keren-Shaul, N. Elefant, F. Paul, H. J. Deeg, C. McFarland, K. R. Loeb, W. J. Valente, N. G. Ericson,
I. Zaretsky, A. Mildner, N. Cohen, S. Jung, A. Tanay, I. Amit, Science E. A. Stevens, J. P. Radich et al., Nat. Commun. 2017, 8, 14049.
2014, 343, 776. [125] E. Azizi, A. J. Carr, G. Plitas, A. E. Cornish, C. Konopacki,
[100] T. M. Gierahn, M. H. Wadsworth Ii, T. K. Hughes, B. D. Bryson, S. Prabhakaran, J. Nainys, K. Wu, V. Kiseliovas, M. Setty, K. Choi,
A. Butler, R. Satija, S. Fortune, J. C. Love, A. K. Shalek, Nat. R. M. Fromme, P. Dao, P. T. McKenney, R. C. Wasti, K. Kadaveru,
Methods 2017, 14, 395. L. Mazutis, A. Y. Rudensky, D. Pe’er, Cell 2018, 174, 1293.
[101] H. C. Fan, G. K. Fu, S. P. A. Fodor, Science 2015, 347, 1258367. [126] K. G. Paulson, V. Voillet, M. S. McAfee, D. S. Hunter,
[102] T. Hashimshony, F. Wagner, N. Sher, I. Yanai, Cell Rep. 2012, 2, F. D. Wagener, M. Perdicchio, W. J. Valente, S. J. Koelle,
666. C. D. Church, N. Vandeven, H. Thomas, A. G. Colunga, J. G. Iyer,
[103] T. Hashimshony, N. Senderovich, G. Avital, A. Klochendler, C. Yee, R. Kulikauskas, D. M. Koelle, R. H. Pierce, J. H. Bielas,
Y. de Leeuw, L. Anavy, D. Gennert, S. Li, K. J. Livak, O. Rozenblatt- P. D. Greenberg, S. Bhatia, R. Gottardo, P. Nghiem, A. G. Chapuis,
Rosen, Y. Dor, A. Regev, I. Yanai, Genome Biol. 2016, 17, 77. Nat. Commun. 2018, 9, 3868.
[127] J. T. Neal, X. Li, J. Zhu, V. Giangarra, C. L. Grzeskowiak, J. Ju, I. [143] V. M. Peterson, K. X. Zhang, N. Kumar, J. Wong, L. X. Li,
H. Liu, S.-H. Chiou, A. A. Salahudeen, A. R. Smith, B. C. Deutsch, D. C. Wilson, R. Moore, T. K. McClanahan, S. Sadekova,
L. Liao, A. J. Zemek, F. Zhao, K. Karlsson, L. M. Schultz, J. A. Klappenbach, Nat. Biotechnol. 2017, 35, 936.
T. J. Metzner, L. D. Nadauld, Y.-Y. Tseng, S. Alkhairy, C. Oh, [144] P. Shahi, S. C. Kim, J. R. Haliburton, Z. J. Gartner, A. R. Abate, Sci.
P. Keskula, D. Mendoza-Villanueva, F. M. De La Vega, P. L. Kunz, Rep. 2017, 7, 44447.
J. C. Liao, J. T. Leppert, J. B. Sunwoo, C. Sabatti, J. S. Boehm et al., [145] J. P. Gerlach, J. A. G. van Buggenum, S. E. J. Tanis, M. Hogeweg, B.
Cell 2018, 175, 1972. M. H. Heuts, M. J. Muraro, L. Elze, F. Rivello, A. Rakszewska, A. v
[128] J. Ding, X. Adiconis, S. K. Simmons, M. S. Kowalczyk, an Oudenaarden, W. T. S. Huck, H. G. Stunnenberg, K. W. Mulder,
C. C. Hession, N. D. Marjanovic, T. K. Hughes, M. H. Wadsworth, bioRxiv:356329, 2018.
T. Burks, L. T. Nguyen, J. Y. H. Kwon, B. Barak, W. Ge, A. J. Kedaigle, [146] J. A. G. van Buggenum, J. P. Gerlach, S. E. J. Tanis, M. Hogeweg,
S. Carroll, S. Li, N. Hacohen, O. Rozenblatt-Rosen, A. K. Shalek, P. W. T. C. Jansen, J. Middelwijk, R. van der Steen, M. Vermeulen,
A.-C. Villani, A. Regev, J. Z. Levin, bioRxiv:632216, 2019. H. G. Stunnenberg, C. A. Albers, K. W. Mulder, Nat. Commun.
[129] M. Stoeckius, S. W. Zheng, B. Houck-Loomis, S. Hao, B. Z. Yeung, 2018, 9, 2384.
W. M. Mauck, P. Smibert, R. Satija, Genome Biol. 2018, 19, 12. [147] L. Qiu, F. Wimmers, J. Weiden, H. A. Heus, J. Tel, C. G. Figdor,
[130] B. J. DeKosky, G. C. Ippolito, R. P. Deschner, J. J. Lavinder, Chem. Commun. 2017, 53, 8066.
Y. Wine, B. M. Rawlings, N. Varadarajan, C. Giesecke, T. Dörner, [148] F. Wimmers, N. Subedi, N. van Buuringen, D. Heister, J. Vivié,
S. F. Andrews, P. C. Wilson, S. P. Hunicke-Smith, C. G. Willson, I. Beeren-Reinieren, R. Woestenenk, H. Dolstra, A. Piruska,
A. D. Ellington, G. Georgiou, Nat. Biotechnol. 2013, 31, 166. J. F. M. Jacobs, A. van Oudenaarden, C. G. Figdor, W. T. S. Huck,
[131] B. J. DeKosky, T. Kojima, A. Rodin, W. Charab, G. C. Ippolito, I. J. M. de Vries, J. Tel, Nat. Commun. 2018, 9, 3317.
A. D. Ellington, G. Georgiou, Nat. Med. 2015, 21, 86. [149] V. Chokkalingam, J. Tel, F. Wimmers, X. Liu, S. Semenov, J. Thiele,
[132] J. R. McDaniel, B. J. DeKosky, H. Tanno, A. D. Ellington, C. G. Figdor, W. T. S. Huck, Lab Chip 2013, 13, 4740.
G. Georgiou, Nat. Protoc. 2016, 11, 429. [150] K. Kwak, M. Akkaya, S. K. Pierce, Nat. Immunol. 2019, 20, 963.
[133] A. V. Medina-Cucurella, R. A. Mizrahi, M. A. Asensio, R. C. Edgar, [151] R. H. Painter, in Encyclopedia of Immunology, 2nd ed. (Ed: P. J.
J. Leong, R. Leong, Y. W. Lim, A. Nelson, A. R. Niedecken, Delves), Elsevier, Oxford 1998, pp. 1208–1211.
J. F. Simons, M. J. Spindler, K. Stadtmiller, N. Wayham, A. S. Adler, [152] T. Kurosaki, K. Kometani, W. Ise, Nat. Rev. Immunol. 2015, 15,
D. S. Johnson, Antibodies 2019, 8, 17. 149.
[134] A. S. Adler, R. A. Mizrahi, M. J. Spindler, M. S. Adams, [153] X.-Y. Wang, B. Wang, Y.-M. Wen, npj Vaccines 2019, 4, 2.
M. A. Asensio, R. C. Edgar, J. Leong, R. Leong, D. S. Johnson, [154] Z. Hu, P. A. Ott, C. J. Wu, Nat. Rev. Immunol. 2018, 18, 168.
mAbs 2017, 9, 1270. [155] A. L. Nelson, E. Dhimolea, J. M. Reichert, Nat. Rev. Drug Discovery
[135] A. W. Briggs, S. J. Goldfless, S. Timberlake, B. J. Belmont, 2010, 9, 767.
C. R. Clouser, D. Koppstein, D. Sok, J. V. A. Heiden, [156] K. Eyer, R. C. L. Doineau, C. E. Castrillon, L. Briseño-Roa,
M. V. Tamminen, S. H. Kleinstein, D. R. Burton, G. M. Church, V. Menrath, G. Mottet, P. England, A. Godina, E. Brient-Litzler,
F. Vigneault, bioRxiv:134841, 2017. C. Nizak, A. Jensen, A. D. Griffiths, J. Bibette, P. Bruhns, J. Baudry,
[136] L. D. Goldstein, Y.-J. J. Chen, J. Wu, S. Chaudhuri, Y.-C. Hsiao, Nat. Biotechnol. 2017, 35, 977.
K. Schneider, K. H. Hoi, Z. Lin, S. Guerrero, B. S. Jaiswal, [157] B. E. Debs, R. Utharala, I. V. Balyasnikova, A. D. Griffiths,
J. Stinson, A. Antony, K. B. Pahuja, D. Seshasayee, Z. Modrusan, C. A. Merten, Proc. Natl. Acad. Sci. USA 2012, 109, 11570.
I. Hötzel, S. Seshagiri, Commun. Biol. 2019, 2, 304. [158] D. S. Wishart, Y. D. Feunang, A. Marcu, A. C. Guo, K. Liang,
[137] K. E. Yost, A. T. Satpathy, D. K. Wells, Y. Qi, C. Wang, R. Kageyama, R. Vázquez-Fresno, T. Sajed, D. Johnson, C. Li, N. Karu,
K. L. McNamara, J. M. Granja, K. Y. Sarin, R. A. Brown, Z. Sayeeda, E. Lo, N. Assempour, M. Berjanskii, S. Singhal,
R. K. Gupta, C. Curtis, S. L. Bucktrout, M. M. Davis, A. L. S. Chang, D. Arndt, Y. Liang, H. Badran, J. Grant, A. Serra-Cayuela, Y. Liu,
H. Y. Chang, Nat. Med. 2019, 25, 1251. R. Mandal, V. Neveu, A. Pon, C. Knox, M. Wilson, C. Manach,
[138] S. C. Bendall, E. F. Simonds, P. Qiu, E.-a. D. Amir, P. O. Krutzik, A. Scalbert, Nucleic Acids Res. 2018, 46, D608.
R. Finck, R. V. Bruggner, R. Melamed, A. Trejo, O. I. Ornatsky, [159] F. Del Ben, M. Turetta, G. Celetti, A. Piruska, M. Bulfoni,
R. S. Balderas, S. K. Plevritis, K. Sachs, D. Pe’er, S. D. Tanner, D. Cesselli, W. T. S. Huck, G. Scoles, Angew. Chem., Int. Ed. 2016,
G. P. Nolan, Science 2011, 332, 687. 55, 8581.
[139] C. Giesen, H. A. O. Wang, D. Schapiro, N. Zivanovic, A. Jacobs, [160] S. Abalde-Cela, A. Gould, X. Liu, E. Kazamia, A. G. Smith, C. Abell,
B. Hattendorf, P. J. Schüffler, D. Grolimund, J. M. Buhmann, J. R. Soc., Interface 2015, 12, 20150216.
S. Brandt, Z. Varga, P. J. Wild, D. Günther, B. Bodenmiller, Nat. [161] P. Hammar, S. A. Angermayr, S. L. Sjostrom, J. van der Meer,
Methods 2014, 11, 417. K. J. Hellingwerf, E. P. Hudson, H. N. Joensson, Biotechnol. Bio-
[140] D. R. Bandura, V. I. Baranov, O. I. Ornatsky, A. Antonov, R. Kinach, fuels 2015, 8, 193.
X. Lou, S. Pavlov, S. Vorobiev, J. E. Dick, S. D. Tanner, Anal. Chem. [162] P. M. Magwene, P. Lizardi, J. Kim, Bioinformatics 2003, 19, 842.
2009, 81, 6813. [163] C. Trapnell, D. Cacchiarelli, J. Grimsby, P. Pokharel, S. Li,
[141] M. Stoeckius, C. Hafemeister, W. Stephenson, B. Houck-Loomis, M. Morse, N. J. Lennon, K. J. Livak, T. S. Mikkelsen, J. L. Rinn, Nat.
P. K. Chattopadhyay, H. Swerdlow, R. Satija, P. Smibert, Nat. Biotechnol. 2014, 32, 381.
Methods 2017, 14, 865. [164] M. Nitzan, N. Karaiskos, N. Friedman, N. Rajewsky,
[142] E. P. Mimitou, A. Cheng, A. Montalbano, S. Hao, M. Stoeckius, bioRxiv:456350, 2018.
M. Legut, T. Roush, A. Herrera, E. Papalexi, Z. Ouyang, R. Satija, N. [165] R. Satija, J. A. Farrell, D. Gennert, A. F. Schier, A. Regev, Nat. Bio-
E. Sanjana, S. B. Koralov, P. Smibert, Nat. Methods 2019, 16, 409. technol. 2015, 33, 495.

Single Cell Analysis Using Droplet Microfluidics

Uploaded by

Copyright:

Available Formats

Single Cell Analysis Using Droplet Microfluidics

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Single Cell Analysis Using Droplet Microfluidics

Uploaded by

Copyright:

Available Formats

Review

Single-Cell Analysis Using Droplet Microfluidics

signaling dynamics.[3,4] Quantifying and

Figure 7. Overview of the Drop-ChIP workflow. R1: read 1. R2: read 2.

Interestingly, Drop-seq technology has also been adapted to 6.1.3. Hi-SCL

All technologies described above have advantages and dis-

will dictate the encapsulation efficiency (e.g., hydrogel beads

6.2. Immune Profiling

Sequencing paired heavy- and light-chain variable regions

7.1.3. Abseq is further PCR amplified to obtain clonal populations. Each

7.2.2. Antibody Secretion 8. Single-Cell Metabolomics

A major challenge in all single-cell approaches is a lack of Conflict of Interest

You might also like