Analysis of aptamer discovery and technology

REVIEWS
Analysis of aptamer discovery

and technology
Matthew R. Dunn, Randi M. Jimenez and John C. Chaput
Abstract | Aptamers are nucleic acid molecules that mimic antibodies by folding into complex 3D
shapes that bind to specific targets. Although some aptamers exist naturally as the ligand-binding
elements of riboswitches, most are generated in vitro and can be tailored for a specific target.
Relative to antibodies, aptamers benefit from their ease of generation, low production cost, low
batch‑to‑batch variability, reversible folding properties and low immunogenicity. However, the
true value of aptamers lies in the simplicity by which these molecules can be engineered into
sensors, actuators and other devices that are often central to emerging technologies. This Review
examines changing trends in aptamer technology by analysing the first quarter century of
aptamer data that is available in the scientific literature (1990–2015). We highlight specific
examples that showcase the use of aptamers in key applications, discuss challenges that have
impeded the success of aptamers in practical applications, provide suggestions for choosing
chemical modifications that can lead to enhanced activity or stability, and propose standards for
the characterization of aptamers in the scientific literature.
Phenotypes Nucleic acids (DNA and RNA) are commonly thought (for example, temperature and salt concentration).
Observable characteristics, of as the genetic blueprint of life because they carry the Molecules bound to the target are separated from the
such as binding or catalytic instructions on how an organism can grow, develop unbound pool and amplified to generate a new popula-
activity, that are encoded in and replicate1,2. However, these molecules can also fold tion of molecules that is enriched in members that share
the sequence of a nucleic acid
or protein.
into complex 3D structures (known as ribozymes) that a common functional property14. Nucleic acids are ideally
catalyse reactions3, control gene expression4, communi- suited for this purpose because they can fold into shapes
Genotypes cate cellular responses5 and mediate protein synthesis6. with a defined function (phenotypes) and their sequences
Genetic sequences that encode Although the biological importance of folded RNA struc- (genotypes) can be replicated in vitro to produce progeny
phenotypes.
tures has long been recognized7, the in vitro evolution molecules with similar characteristics to the parent
Antibodies of nucleic acid molecules with non-biological functions sequence10. The ability to amplify individual molecules
Protein affinity reagents was achieved only when it became possible to gener- with desired phenotypes and optimize their functions
produced by the immune ate large populations of degenerate oligonucleotides by by directed evolution is a distinguishing feature that
system that can be made to solid-phase synthesis and amplify individual members separates nucleic acids from other organic molecules,
recognize a wide range of
targets called antigens.
using the polymerase chain reaction (PCR)8,9. These tech- most of which cannot replicate because they lack a
nologies enable the isolation of functional nucleic acid genotype–phenotype connection15,16.
molecules that can bind to a specific target or catalyse Aptamers are often compared with antibodies, as both
a chemical reaction10. Affinity reagents based on either molecules function as affinity reagents17. However, unlike
DNA or RNA are referred to as aptamers, the Latin root antibodies and other protein-based affinity reagents
of which means ‘to fit’ (REF. 11) (BOX 1). Aptamers can (including single-chain variable fragment antibodies,
be synthesized to bind to a wide range of chemical and affibodies and designed ankyrin repeat proteins)18–22,
Departments of
Pharmaceutical biological targets from small molecules to whole cells. aptamers have unique advantages that make them powerful
Sciences, Chemistry, and The process used to isolate aptamers from large ran- tools in the arsenal of affinity reagents. At the time of
Molecular Biology and dom-sequence libraries is called in vitro selection, which writing, aptamers can be produced on larger scales
Biochemistry, University of is also termed SELEX (systematic evolution of ligands by than antibodies, and the retained genetically encoded
California, Irvine, California
92697, USA.
exponential enrichment)11–13. Similar to natural selection, sequence of aptamers can be expressed in vivo in cultured
SELEX is an iterative process of selection and amplifi- cells23. As aptamer production is a chemical process
Correspondence to J.C.C.
jchaput@uci.edu cation in which large pools of nucleic acid molecules rather than a biological process, it avoids the problem
doi:10.1038/s41570017-0076 (typically >1 trillion distinct sequences) are challenged to of viral or bacterial contamination that can occur
Published online 4 Oct 2017 bind to a desired target under a defined set of conditions during antibody manufacturing and reduces the potential
NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 1

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
for batch‑to‑batch variability that plagues the antibody been designed to base pair with the binding domain of
market and frustrates researchers seeking to reproduce the folded structure27,28. Aptamers may unfold during
data24. As therapeutics, aptamers generally exhibit less of prolonged storage at ambient temperature; however, this
an immune response than do proteins, and their small does not affect their function, as aptamers can refold
size (<30 kDa versus ~150 kDa for a full-size antibody) into a functional state using a simple annealing proce-
increases their chances of availability to biological areas dure of heating and cooling in an appropriate buffer.
that are inaccessible to antibodies25,26. In addition, their Overcoming the cold-chain problem will lower the cost
ease of chemical modification allows extended control of shipping aptamers relative to antibodies. The ability
over their renal clearance and half-life27. Therapeutic to engineer aptamers into ligand-responsive devices that
aptamers can also be rapidly deactivated with antisense activate only in the presence of a desired analyte has led
oligonucleotides (referred to as antidotes) that have to a surge of research activity aimed at designing sensors
and other genetically controlled elements for diagnostic
applications29. Finally, because aptamers are oligo
Box 1 | Introduction to aptamers nucleotides, they can be used as reagents that can be
Aptamers are single-stranded nucleic acid molecules that can fold into discrete 3D seamlessly integrated with other technologies involving
structures with ligand-binding sites that are complementary in shape and charge to a nucleic acid-based systems, such as amplification systems,
desired target. Aptamers are generated by a process known as in vitro selection or DNA nanotechnology or DNA computing30–32. Additional
SELEX (systematic evolution of ligands by exponential enrichment), which is a comparisons between aptamers and antibodies may be
Darwinian evolution process in which iterative rounds of selection and amplification are found in previous reviews17,25.
used to enrich for sequences that can perform a desired function, such as target binding In this Review, we examine the changing trends in
or catalysis. A schematic representation of a typical SELEX cycle used to generate aptamer technology by analysing the first quarter century
aptamers with affinity for an arbitrary target is shown in part a of the figure. To apply of aptamer data that is available in the scientific literature
the principles of Darwinian evolution, a large population (library) of nucleic acid
(1990–2015). In particular, we highlight recent examples
sequences (typically up to 1015 unique molecules) is incubated with a target. Molecules
that showcase the use of aptamers in key applications, dis-
that bind to the target are separated from the pool of non-functional sequences
through the use of physical methods, such as gel electrophoresis or affinity cuss challenges that have impeded the success of aptamers
chromatography. Functional sequences are recovered and amplified to create a new in practical applications, provide suggestions for choos-
population of molecules that is now enriched in sequences that can perform the chosen ing chemical modifications that can lead to enhanced
function. The process of selection and amplification is repeated until the pool is function or stability, and propose recommendations for
dominated by sequences that exhibit high affinity to the desired target. Aptamers aptamer characterization. Certain specialized topics are
isolated by SELEX are analysed for possible secondary structural motifs using beyond the scope of this Review; accordingly, we direct
computational programs that identify regions of the sequence that are capable of readers interested in learning more about aptamers to
Watson–Crick base-pairing. Aptamers can form several types of secondary structure; several excellent reviews that cover a wide range of topics
one example is shown in part b — the predicted secondary structure of a DNA aptamer
from chemical biology to therapeutics27,33–36. It is our hope
binding to human α-thrombin. The structure of an α-thrombin–aptamer complex
that this Review will stimulate chemists to think about
(shown in part c), which includes the protein and the complementary aptamer, was
solved by solution NMR (Protein Data Bank identifier: 3DD2)129. The NMR structure new ways in which aptamers can be used to push the
shows that nucleic acid sequences can fold into compact tertiary structures with boundaries of science beyond what is currently known
discrete ligand-binding sites. Residues at the protein–aptamer interface (highlighted in or to solve major problems that affect human health and
orange, part d) form strong intermolecular interactions that enable the aptamer to the environment.
recognize a particular site on the protein surface. These interactions are similar to the
types of interactions found at the interface of a protein–protein quaternary structure. The primary literature (1990–2015)
Images in part c and d courtesy of W. Van Horn, Arizona State University, USA. The number of citations per year is a common metric
that is used to provide evidence for growth in a field of
a b
C U GA science. However, this measure of scientific productivity
A A G A uses databases that can inflate the scientific literature by
C A
A
U C A UG including duplicate citations, non-peer-reviewed articles
Library Target A U
G C A and commentaries. Instead, a more accurate reflection
G C of scientific productivity would be the number of non-
5′ G C 3′
redundant peer-reviewed publications generated per year,
Amplify Incubate Secondary structure which can be compared with other scientific disciplines
c as well as the normal growth pattern of the scientific lit-
erature. For example, in the case of aptamers, interro-
In vitro selection (SELEX) gating Web of Science, Google Scholar and PubMed for
the term ‘nucleic acid aptamer’ with the year range set
α-Thrombin – aptamer
complex to 1990–2015 returned >45,000 entries (FIG. 1a), most of
which are duplicates or non-primary literature (defined
d here as original peer-reviewed publications, excluding
Recover
commentaries and books). The total number of publi-
Select cations is reduced by one order of magnitude if only the
non-redundant publications are considered.
We have downloaded and catalogued the 4,795
α-Thrombin Aptamer
non-redundant articles into a manually curated database
Nature Reviews | Chemistry

2 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem
©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
a Aptamer publications Figure 1 | Trends in aptamer publications.

a | Non-redundant articles published in the aptamer field
Starting query: Filter: Example: Article type:
Nucleic acid aptamer Domain Science technology Research Review
between 1990 and 2015. b | The annual frequency of
Databases: Language English articles articles aptamer articles is shown on the left. The growth of the
Web of Science, Chemistry Nucleic acid aptamer literature relative to the scientific literature is
PubMed, Google Scholar Misc. Peer-reviewed shown on the right. The number of the aptamer articles
Year range: Replicates in search has been normalized taking into account the total number
1990 – 2015 of annual scientific publications. These data were
obtained from the PubMed database, which had the least
Number of publications number of duplicate and non-primary literature hits for
searches performed using the term ‘nucleic acid aptamer’
4,000 – 47,300 4,795 3,995 800 (7,114 hits compared with 4,795 observed publications;
Subclassification: Supplementary information S1 (table)). c | Distribution of
Application 3,152 aptamer publications per year organized by backbone
Discovery 843
chain. d | Bar chart showing the distribution of aptamer
targets and backbone structures reported in the set of 843
b Publication frequency Application Discovery Total aptamer articles discussing systematic evolution of
700 ligands by exponential enrichment (SELEX). e | Breakdown
200 of the top three target categories based on a total of
No. of publications
600
Relative increase
500 150
1,003 independent aptamer selections. The difference
400 between aptamer publications and selections (843 versus
300 100 1,003) is due to a small number of cases in which multiple
Global SELEX experiments are described in the same article.
200 growth
25
100 Misc., miscellaneous.
0 0
1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 (FIG. 1a),and a summary of the information obtained
Year Year from the database along with a complete list of references
is provided in the Supplementary information S1–S21
c Backbone category d Target category (tables). Of the non-redundant articles, 843 are research
articles reporting the discovery of new aptamers (dis-
DNA (428) DNA RNA Unnatural
covery articles), 3,152 are research articles describing the
No. of publications
550 application of aptamers (application articles) and 800 are

review articles (FIG. 1a). A plot of the number of articles
RNA (288)
published per year (FIG. 1b) reveals an exponential growth
500
pattern that began in 2005 and continued through to
Unnatural (127) 2015. When normalized against the number of publica-
450 tions recorded annually in the PubMed database (chosen
1990 1995 2000 2005 2010 2015
because of its low redundancy), it becomes clear that the
number of aptamer publications is increasing at a faster
400
e Target breakdown rate than the normal growth pattern observed for the
Other
scientific literature.
57 350
Trends in aptamer discovery
No. of publications
Infectious 149
agents 378 Mammalian The set of aptamer discovery articles was queried for
300
specific information about backbone chemistry, target
selection, characterization and downstream application.
Proteins (584) 250 To simplify our analysis, the selection process for one
target was counted as a single entry. Articles describing
Fluorophores Food and selections for multiple targets were given one entry per
and synthetics environmental 200
33 toxins selection. Missing information was listed as unreported.
45
52
104
Biological 150 Backbone chains. From the 843 discovery articles, we
molecules identified 1,003 in vitro selection experiments performed
Pharmaceuticals against 705 unique targets that ranged in complexity
100
Small molecules (234) from small molecules to whole cells. It should be noted
that these values do not include aptamers that were gen-
Non-cancer
mammalian cells 50 erated at private companies. Although the dominant
30 backbone chain used for aptamer discovery is DNA-
58 Pathogenic based (>50%) — chosen presumably for its increased
microbes 0 chemical and biological stability relative to RNA, com-
53
ns
ses
Nu s
ac ic
ule l
ll
mercial availability and ease of handling — the past few

lec al
Pro s
s
Ce
tei
cle
Cancer cells
id
mo Sm
u
Vir
years have witnessed a rise in unnatural nucleic acids as

Cells (141) polymers for the development of new aptamers (FIG. 1c,d).

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
The category of unnatural polymers consists primarily of >75% of those reporting aptamer application do not
nucleic acids with chemically modified nucleobase and disclose the sequence of the aptamer discovered or used
sugar moieties, which have been synthesized with the in the study. Furthermore, of the aptamer discovery
aim of enhancing the function or nuclease stability of articles that do report a sequence, ~20% report only a
aptamers for clinical applications. This important area single aptamer sequence, which is startling as selection
of aptamer development is addressed in later sections. processes generally produce many aptamer sequences.
As sequence information is essential for reproducing
Target distribution. Our analysis of 1,003 in vitro exper- previously published data, this information along with
iments reveals new insights into the targets chosen for the binding conditions should be considered minimum
aptamer selections. Proteins constitute by far the larg- requirements for publication (see below for more on
est target category with 584 entries, followed by small suggested standards). Despite an unexpected number of
molecules and cells with 234 and 141 entries, respec- poorly characterized aptamers in the scientific literature,
tively (FIG. 1d,e). In addition, a few aptamer species that we have identified several excellent publications in which
have been generated against viruses and nucleic acid in vitro-selected aptamers were characterized in great
molecules were also identified (22 entries each). The detail. One straightforward example comes from Szostak
top five targets chosen for aptamer development are and co-workers on the isolation of an ATP-binding
human α‑thrombin, streptavidin, vascular endothelial RNA aptamer in which the natural uridine residues
growth factor (VEGF), influenza haemagglutinin and found in RNA were replaced with the unnatural residue
adenosine‑5ʹ‑triphosphate (ATP). 5-(3‑aminopropyl)uridine (UNH2)39. UNH2 expands the
Although the inherent properties of an aptamer can chemical diversity of RNA by offering a primary amine
be influenced by the selection process and backbone that exists in the ammonium form (pKa ~9.5) at physio-
chemistry, it is not always possible to determine whether logical pH, enabling favourable electrostatic interactions
an aptamer can be generated to bind to a given target 10. with the negatively charged phosphates on the ATP
Consequently, studies that focus on methodology devel- ligand. The UNH2 residue satisfies all of the conditions
opments tend to favour targets that have been used suc- for in vitro selection, namely: the modified nucleotide
cessfully and that are considered to be highly aptagenic. triphosphate serves as a substrate for RNA polymerase,
Common examples of these targets include human and reverse transcriptase is able to copy UNH2-modified
α‑thrombin, VEGF and influenza haemagglutinin, RNA back into complementary DNA ( cDNA ) for
which have clear implications for aptamer therapeutics25, amplification by PCR.
whereas ATP and streptavidin benefit from selection pro- After ten cycles of selection and amplification, RNA
tocols that are straightforward to implement and known molecules that remained in the pool were cloned and
to produce aptamers after just a few cycles of selective sequenced39. Members of representative families were
amplification37,38. screened for ATP binding to identify clones that were
dependent on the presence of the amino modification for
Characterization. One of the most important properties target binding. One clone, 10N23, showed 40‑fold higher
Aptagenic
of an aptamer is how well it binds to its designated target binding to ATP when transcribed with UNH2 compared
A target that is known to
produce an aptamer by in vitro and distinguishes it from others that may be present in with uridine triphosphate (UTP), and was 5ʹ and 3ʹ
selection. a biological mixture. The key parameters to consider are truncated by end-mapping deletion analysis to identify
the binding affinity and specificity. The binding affinity the boundaries of the core binding domain. Secondary-
Dissociation constant is typically reported as a solution equilibrium dissociation structure analysis indicated a certain stem–loop with a
The equilibrium constant for
constant (Kd), with a low value of Kd corresponding to bulge to be a region possibly responsible for this high
the dissociation of the target–
aptamer complex (Kd = [target] a high binding affinity. The specificity is quantitatively affinity. To identify the nucleotides that were crucial for
[aptamer]/[target–aptamer]). measured as the ratio of Kd for the cognate target versus binding, a doped library was constructed such that a por-
Kd for a non-cognate target. Specificity is often reported tion of the sequence was partially randomized and carried
Lysate
for just a few off-target proteins that either have broad through additional rounds of in vitro selection. UNH2 res-
The contents of a cell produced
by cell lysis. affinities for nucleic acids in general or represent homo- idues essential for binding were identified by individually
logues of the target protein. It would be impractical to replacing each UNH2 residue with cytidine and screening
cDNA measure the affinity constants for many off-target mol- for binding to ATP. Finally, affinity and specificity values
The complementary DNA ecules. In general, Kd values for most protein aptamers were obtained by measuring the Kd of the core binding
sequence that results when
RNA (or xeno-nucleic acid
are in the low- to sub-nanomolar regime, whereas Kd domain for ATP and a panel of ATP analogues. This
(XNA)) is reverse transcribed values for small-molecule aptamers are in the low- to study demonstrates many of the key features of a well-
into DNA. sub-micromolar regime. characterized aptamer, which includes sequence informa-
Our analysis has led to several surprising revelations tion, mutational analysis, secondary-structure prediction,
Doped library
regarding the level of aptamer characterization in published as well as qualitative and quantitative binding assays
A nucleic acid library that
contains variants of a single scientific articles. We find that ~20% of the analysed articles under well-defined binding conditions.
sequence, which feature a do not include a binding affinity value of any kind and
small fraction of non-wild type <25% report a specificity test. Among the articles that do Post-SELEX modifications. Several strategies have been
residues at each position. For report a specificity test, only ~10% use a complex biological developed to improve the function of in vitro-selected
example, if the wild-type
residue is A, then the library
mixture, such as human serum or total Escherichia coli aptamers. This approach, which is generally referred
would contain mostly A with a lysate, to evaluate off-target binding. We also find that to as post-SELEX optimization, involves modifying
mixture of C with T and G. ~10% of the articles reporting aptamer discovery and in vitro-selected aptamers with functional groups that

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Recombinant protein were not present in the original selection40. RNA aptam- affinity have been achieved by bivalent affinity reagents,
therapies ers developed for diagnostic and therapeutic applications, which combine two ligands that can independently
Therapies that are produced for example, are routinely modified with nuclease- recognize different sites on a protein surface47.
through recombinant DNA resistant analogues that protect the 2ʹ‑hydroxyl groups In contrast to the gains in binding affinity, small
technology, which involves
expressing and purifying a
from hydrolytic attack41. Pegaptanib sodium (Macugen; structural changes to an aptamer sequence rarely result
protein from bacterial or Pfizer/Eyetech) (FIG. 2a), an RNA aptamer selected against in substantial gains in functional activity. However, in
mammalian cells. Many VEGF165, is a prime example of a SELEX product mod- two separate examples (α‑thrombin and VEGF), Yang
biologics, such as monoclonal ified for enhanced biological stability 42. Macugen was and co-workers showed that a simple PS2‑walking
antibodies, are recombinant
selected from a library of 2ʹ‑fluoropyrimidines and mod- experiment, in which each phosphodiester linkage along
protein therapies.
ified post-SELEX by substituting nearly all of the purine the oligonucleotide backbone was individually replaced
Phosphodiester linkage residues with 2ʹ‑O‑methyl analogues, inverting the 3ʹ by a PS2 linkage, could reveal the location of the PS2
The chemical linkage, terminal nucleotide and adding a 40 kDa poly(ethylene substitution responsible for transforming a modest affin-
connecting the monomeric glycol) moiety to the 5ʹ end43. Together, these substitu- ity aptamer (Kd ~1–2 nM) into an ultrahigh affinity binder
units in a nucleic acid polymer.
In general, the linkage takes
tions enabled Macugen to become an effective treatment (Kd ~1–2 pM)46. In the case of the α‑thrombin aptamer,
the form RO(O)P(O−)ORʹ. for neovascular age-related wet macular degeneration analysis of the binding interactions by X‑ray crystall
(AMD)42. Although Macugen has since been replaced by ography (FIG. 2b) revealed that the increase in binding
Induced-fit more effective monoclonal antibodies44 and recombinant affinity was caused by an induced-fit rearrangement of the
A biochemistry model in which
protein therapies45, its distinction as the first therapeu- aptamer at the protein–RNA interface46. Given the simplic-
the initial interactions of an
enzyme–substrate complex (or
tic aptamer to be approved by the US Food and Drug ity of PS2-walking experiments, this technique represents
antibody–target complex) are Administration (FDA) remains an important milestone a powerful approach for improving the binding affinity of
strengthened by in aptamer technology. in vitro-selected aptamers.
conformational changes that Most post-SELEX modifications are designed to
increase the strength of the
intermolecular interactions
enhance the biological stability of in vitro-selected Trends in aptamer applications
involved in substrate or target aptamers40, as chemical modifications that lead to Aptamers have had a notable effect on the development
recognition. enhanced functional activity are harder to predict. Yang of new chemical tools with practical applications across
and co‑workers have challenged this trend by demon- a wide range of scientific disciplines48. To help quantify
strating that a single phosphorodithioate linkage (PS2, these efforts, we have categorized each of the 3,152 articles
FIG. 2b) inserted into the phosphodiester backbone can that discuss aptamer applications into the five techno-
lead to a ~1,000‑fold increase in the binding affinity of logical groups that broadly define the application areas
known RNA aptamers46. Similar enhancements in binding (FIG. 3a). More than 95% of the analysed articles belong to
a b O
P
U
A G O –O O
C B
A P O
G U
U O O
– B
U
G A O O OH
A U
C A
U C O OH
PO2 linkage
A A
A U RNA
G C S
O O
G C P
P P 3.5
5′ PEG C G 3′ – 3′ –S O B
–O O –O O B
Linked dT B O
O O 3.5
Natural ribonucleotide 5.4 O OH
O F O OMe 3.7
2′-fluoropyrimidines Native
2′-methoxypurines PS2 Modified PS2 linkage
2′- F- RNA 2′-OCH3-RNA
Figure 2 | Post-SELEX modifications. a | Macugen (pegaptanib sodium) is a therapeutic RNA aptamer,

Nature the| sequence
Reviews Chemistryand
secondary structure of which are presented on the left. The molecule has a stem–loop structure with a bulge, with the 5ʹ
terminus bearing a PEG group and the 3ʹ terminus having an inverted nucleotide. These modifications, as well as the
derivatization of most of the 2ʹ‑hydroxyl sites, are performed post-SELEX and confer stability towards nucleases, such that
is was the first FDA-approved aptamer42. Macugen comprises natural ribonucleotides (shown in black on the right) as well
as unnatural nucleotides, with the latter having their 2ʹ‑hydroxyls replaced with fluoro (blue) or methoxy groups (red).
b | Comparing the X-ray structures of human α‑thrombin (green ribbon) complexed to either an RNA aptamer or its
PS2-modified analogue (orange ribbon) enables the structural features that underlies the tighter affinity binding of the
latter to be identified46. The zoomed region shows a superimposition of the native (green sticks) and PS2-modified (blue
sticks) aptamer–thrombin complexes, highlighting that the PS2 linkage at position 18 (sulfur in yellow; Protein Data Bank
identifier: 5DO4) can approach Phe232 on the thrombin surface more closely than can the native PO2 linkage (oxygen in
red; PDB ID: 3DD2); the bond lengths of the intermolecular interactions are in angstroms. The chemical structures of the
PO2 and PS2 linkages are given on the right. dT, 2′-deoxythymidine; FDA, US Food and Drug Administration; PEG,
poly(ethylene glycol); SELEX, systematic evolution of ligands by exponential enrichment.

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Fluorescent reporter three of the five technological categories: scientific tools attached to conformationally flexible regions of an
Molecules that elicit a (comprising gene regulatory elements, nanotechnology, aptamer can transduce ligand-binding events into
fluorescent signal. affinity chromatography and non-clinical sensors), clin- an optical signal34 (FIG. 4). Strategies that rely on more
ical reagents (comprising therapeutics, diagnostics, drug than one fluorescent reporter have become particularly
Fluorescence resonance
energy transfer
delivery systems and clinical biosensors) and environ- valuable, as they enable signal transduction based on
A photophysical effect in which mental sensors (comprising reagents for food and water fluorescence resonance energy transfer (FRET). Stojanovic
energy is transferred between analysis). The remaining publications fall into the techno- and co‑workers have developed a mix-and-measure
fluorescent or light-sensitive logical groups of informatics and biophysical discovery. assay that responds with high sensitivity and selec-
moieties.
Whether these areas will experience continued tivity to small-molecule targets that are chelated by
growth is unclear, as >50% of the articles describe the in situ derivatizing agents49. Chelated targets were
same set of well-characterized aptamer–target pairs. found to have high-affinity binding interactions
The most common targets for application development with aptamers, whereas the unchelated molecules
are α-thrombin, adenosine nucleosides and nucleo- proved to be resistant to selection owing to their confor-
tides (ATP, ADP, AMP and adenosine), VEGF, platelet- mational heterogeneity (linear and cyclic forms of the
derived growth factor BB (PDGF‑BB, which features sugar). The selection used a structure-switching approach
two B subunits), cocaine and theophylline (FIG. 3b). In that involved the release of active library members from
addition to historical reasons, these targets were cho- an immobilized complementary strand upon ligand
sen because the aptamers that were generated to bind to recognition49. Owing to the structure-switching design,
these targets fold into a privileged stem–loop motif that aptamers isolated from the selection could be readily
can be easily engineered into ligand-responsive elements converted into optical sensors by adding a quencher to the
that produce signal changes that are compatible with a 3ʹ end of the complementary strand and a fluorophore to
wide range of detection modalities (for example, optical the 5ʹ end of the aptamer (FIG. 4c,d). In this configuration,
or electrochemical)34. Based on this analysis, it appears aptamers undergo conformational change in the presence
that the application field would benefit from the study of the small-molecule analyte, which produces a dose-
of other examples of aptamer–target pairs. Furthermore, dependent fluorescence signal as the quencher strand is
the development of robust multiplexed platforms that released from the aptamer. The generality of this approach
use combinations of aptamers to simultaneously detect opens the door to highly sensitive detection systems for
many analytes would help to expand the application of clinical and environmental applications. For example, the
aptamers in sensor-based technology. real-time monitoring of analytes in whole blood50.
Sensors. One area of aptamer-based research that has Environmental screening. The development of aptamer-
received substantial attention is the field of optical sensors34. based sensors for environmental screening is one of
Numerous studies have shown that fluorescent dyes the newest and fastest growing areas of aptamer-based
a Application breakdown b Top ten most applied targets

Target Count
α-Thrombin 665
Clinical reagents
Adenosine nucleoside and/or 374
nucleotide
No. of publications
Vascular endothelial 144

growth factor (VEGF)
Informatics Platelet derived growth 86
factor BB (PDGF-BB)
Cocaine 77
Environmental sensors
Theophylline 71
Scientific tools Lysozyme 65
Biophysical discovery Nucleolin 65
Immunoglobulin E (IgE) 60
1990 1995 2000 2005 2010 2015
Ochratoxin A (OTA) 60
Year
Figure 3 | Distribution of aptamer applications and targets. a | The annual number of articles discussing possible
aptamer applications is shown as a violin plot illustrating the growth across five broadly definedNature Reviews | Chemistry
subcategories:
biophysical discovery, which includes structural and thermodynamic analyses; clinical reagents, including therapeutics,
drug conjugates, diagnostic agents and clinically tailored biosensors; informatics, which accounts for in silico modelling
and selections, machine learning and software development; scientific tools, including chromatography, non-clinical
sensors, gene regulation and nanotechnology; and environmental sensors, accounting for food and water sample analysis.
b | Table listing the top ten targets for application development. The complete target list is provided in Supplementary
information S16 (table). BB denotes that two B domains are present.

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
research. When normalized for sample size, this category prevents VEGF from stimulating blood vessel growth
contains the most diverse set of aptamer–target pairs with and eventual vision loss by inhibiting the binding
a strong emphasis on reagents that can detect pathogens, of VEGF165 (the major pathological VEGF isoform)
toxins, antibiotics and pesticides in food, water and soil to its cognate receptor 42. Another interesting example is
samples51,52. Investment by companies such as NeoVentures the aptamer-based anticoagulation system, REG1 (RB006
Biotechnologies has resulted in commercial kits for plus RB007; Regado Biosciences), which is a drug–
detecting ochratoxin A and aflatoxins in food samples. antidote pair that acts on coagulation factor IXa
Given the need to safeguard domestic food stocks and (REFS 53,54). RB006 is an aptamer that prevents blood clots
the environment, this area of aptamer-based application by inhibiting factor IXa-mediated coagulation, and RB007
could grow rapidly in the coming years. is a complementary antidote sequence that reverses the
effects of RB006 in a dose-dependent manner. Together,
Therapeutics. To date, the FDA has approved one RB006 and RB007 prevent arterial thrombosis by provid-
aptamer for the treatment of AMD, and ten aptamers ing a rapid and tightly regulated system for controlling
have undergone clinical trials for the treatment of dis- the coagulation cascade mechanism55,56. A third example
eases such as AMD, coagulation disorders, cancer and is NOX‑A12 (NOXXON Pharma), an l‑RNA aptamer
inflammation27. All of the aptamers that have entered that prevents tumour proliferation by inhibiting CXC
clinical trials thus far fall into the general category of chemokine ligand 12 (CXCL12)57,58. l‑RNA and l‑DNA
antagonists (BOX 2) because they act by disrupting the aptamers represent a general class of aptamers known as
function of a pathological target protein. For example, spiegelmers, which are mirror-image aptamers that are
Macugen, the only FDA approved aptamer therapeutic, not recognized by nucleases present in human blood and
serum59, as nucleases are specific for d‑sugar moieties.
Spiegelmers are created using the principle of mirror-
a Bimodal sensor designs b Split-aptamer design image symmetry, in which natural d‑RNA aptamers are
generated against the mirror image of a desired target
F Q
F Q and then, after the selection is complete, the resulting
RNA sequences are chemically synthesized with l‑RNA
Q F Q F
phosphoramidites to produce the desired l‑RNA sequence
(FIG. 5), which recognizes the natural target 60. In clinical
Signal off trials, NOX‑A12 showed favourable activity against a range
FA FD FA of solid tumours and multiple myeloma. Joyce and col-
leagues have shown that selected l‑RNA aptamers can also
Q Q F be used to inhibit the processing of specific microRNAs61.
F As l‑RNA cannot base pair with d‑RNA, these aptamers
FD
function through highly specific tertiary structures
Signal on Signal off Signal on
rather than Watson–Crick base-pairing.
c Structure-switching SELEX d Structure-switching sensors
Solid support Drug delivery. In addition to serving as drugs, aptamers
Q F F
have been developed to selectively deliver therapeutic
agents to the surface or cytoplasm of human cells62. In
this case, aptamers function as targeted drug delivery
vehicles that can be used to increase the efficacy of a drug
N30 Q
and reduce the side effects of traditional non-targeted
approaches such as chemotherapy and radiotherapy,
which are commonly used to treat patients with cancer.
F Fluorophore Q Quencher
Human prostate-specific membrane antigen (PSMA), a
transmembrane protein associated with prostate cancer
Analyte Chelating agent
that is overexpressed on the surface of solid tumours
Figure 4 | Aptamer-based optical sensors. a | In the bimodalNaturedesign,Reviews
unimolecular and constitutively internalized into the cell, was the
| Chemistry
aptamer sensors either emit or quench a fluorescent signal based on a conformational first model system developed for aptamer-based drug
change induced by an analyte. Binding to a target may result in a fluorophore and delivery 63. In 2002, Coffey and co-workers showed that
quencher coming into proximity, or may involve a fluorescent donor–acceptor pair aptamers selected to bind to the extracellular domain of
(FD and FA) separating to produce a fluorescent signal34. b | In the split-aptamer design, PSMA could be internalized by a clathrin-dependent
aptamer sequences are separated into two strands that come together in the presence pathway 64. In numerous subsequent studies, PSMA-
of an analyte to either emit or quench a fluorescent signal. c | Structure-switching SELEX specific aptamers have been used to deliver small-
promotes the discovery of aptamers that can be used as optical sensors with minimal molecule therapeutics and small interfering RNAs
sequence engineering. In this example, functional aptamers are released from a solid
(siRNAs) that are covalently or non-covalently bound to
support when they bind to a desired analyte, which, in turn, is bound to a chelating agent.
The N30 random region is shown in blue. d | Aptamers discovered through struc- the aptamer27. The efficacy of this approach has improved,
ture-switching SELEX can be converted into sensors by replacing the solid support at the and, for example, we now have access to a PSMA-specific
end of the DNA capture probe with a quencher and adding a fluorescent dye to the end of aptamer drug conjugate that facilitates tumour regression
the aptamer strand49. SELEX, systematic evolution of ligands by exponential enrichment. following systemic administration65. This general concept
Parts c and d are adapted with permission from REF. 49, Macmillan Publishers Limited. of aptamer-based drug delivery has been extended to

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Box 2 | Therapeutic aptamers function as antagonists of target-specific aptamers according to the fold enrich-
ment of individual clones. This analysis provides a use-
Aptamers developed for therapeutic purposes function as antagonists of the binding ful method for distinguishing target-specific aptamers
interaction of a protein with its cognate ligand. Although non-competitive interactions from highly abundant members that function with non
are possible, most aptamers function by directly competing with the natural ligand for specific activity 72. Selectivity of the aptamers for specific
the ligand-binding site on the protein surface.
β 2AR conformations was demonstrated using bio
chemical, pharmacological and biophysical approaches.
The aptamers A1, A2 and A13 exhibited strong con-
formational selectivity for the high-affinity agonist
(BI167107)-bound active β2AR conformation, whereas
Protein aptamer A16 showed conformational selectivity for
Cognate ligand the inactive β2AR conformation of the inverse-agonist
Aptamer ICI‑118,551. The discovery that aptamers can serve as
allosteric GPCR modulators adds to the diversity of
Antagonist binding Cognate ligand
(competitive inhibition) binding ligands available to study the structural and func-
tional regulation of GPCRs and represents a basis for
Nature Reviews | Chemistry the design of GPCR ligands with improved safety and
nanoparticles; in this approach, nanoparticles encapsu- enhanced therapeutic efficacy.
lating therapeutic agents are coated with aptamer recog-
nition elements for aptamer-mediated internalization Natural product synthesis. Natural product synthesis is
and drug release66. an important aspect of drug discovery because many FDA
Another important application of aptamer-based drug approved drugs are analogues of small-molecule natural
delivery involves developing strategies to fight HIV. In products74. In pursuit of new drugs, chemists have taken
patients with HIV, viral entry into immune cells occurs ‘top-down’ approaches by modifying natural products
when envelope glycoprotein gp120 — a glycoprotein isolated from nature as well as ‘bottom-up’ approaches
located on the exposed surface of the viral particle — that require the total synthesis of a desired com-
interacts with the CD4 receptor and CC-chemokine pound from simple commercially available precursors.
receptor 5 (CCR5), both of which are located outside Regardless of the approach taken, functional group
the host cell67. As this interaction is crucial for viral selectivity invariably becomes a problem, as many nat-
invasion, individuals that carry certain CCR5 mutations ural products are structurally complex. Herrmann and
can be immune to some HIV strains68. However, these co-workers have used aptamers as protecting groups to
mutations are rare; therefore, drugs that interfere with facilitate the highly chemo- and regioselective derivat-
the binding process could provide a possible route for ization (>99%) of natural antibiotics in a single, high
therapeutic intervention. Pioneering work by Rossi and yielding (83%) synthetic step75. This technique relies
co-workers has led to the development of gp120 aptamers on the aptamers to fold into well-defined shapes with
that function with dual activity: first, to inhibit the inter- discrete ligand-binding sites that recognize and shield
action between gp120 and the CD4 receptor and, second, certain regions of a molecule from chemical derivati-
as a cell-type specific delivery agent for siRNA69,70. The zation. The possibility of using aptamers to distinguish
gp120 aptamer was reported to deliver anti-HIV siRNA equivalent functional groups provides a simple and cost-
into HIV infected cells and inhibit HIV activity in vitro. effective strategy for selectively modifying complex
Subsequent work performed in a humanized mouse natural products.
model showed that systemic administration of gp120
aptamer–drug conjugates suppressed HIV replication Barriers to commercial success
by several orders of magnitude 71, indicating that The promise of rapid and low-cost production of reagents
aptamer–drug conjugates offer a promising treatment for that are compatible with any target imaginable has led
HIV infection. researchers to embrace aptamers as a chemical alternative
Allosteric modulation. The affinity and selectivity to antibodies. Once selected, these molecules provide a
with which SELEX-derived aptamers bind to a given renewable source of affinity reagents based on genetically
target are governed by the stringency of the selection encoded sequences that exhibit high ligand-binding
process10. Lefkowitz and colleagues have developed an affinity and low batch‑to‑batch variability. However, like
iterative approach combining in vitro selection with most new technologies, aptamers took years to develop
next-generation sequencing to discover RNA aptamers and followed a Gartner hype cycle, which is defined by
that stabilize several functionally distinct conformations a ‘peak of inflated expectations’, followed by a ‘trough of
of a model G‑protein-coupled receptor 72 (GPCR, FIG. 6a). disillusionment’, a subsequent ‘slope of enlightenment’
A highly diverse library of 2ʹ‑fluoropyrimidine-modified and, finally, a ‘plateau of productivity’. After nearly three
oligonucleotides afforded a series of RNA aptamers that decades of intense research, aptamers now seem ready
stabilize the active, inactive and ligand-specific receptor to enter the long-awaited period of strong productivity.
conformations of β2-adrenoceptor (β2AR), a protein The past few years alone have witnessed tremendous
that is a well-characterized member of the GPCR family73. growth, with new technologies overcoming many of
Of particular utility in this study were comparative the problems that have limited the use of aptamers in
bioinformatic analyses, which helped in the identification practical applications.

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Mirror-image symmetry Chemical diversity. The limited chemical functionality

Symmetry with respect to a of natural genetic polymers (DNA and RNA) generally
reflection or plane of explains the dominance of proteins over nucleic acids
symmetry. Mirror-image as scaffolds for biological receptors and catalysts10. For
molecules are non-
superimposable. For example,
example, when compared with antibodies, aptamers
your right hand cannot be have fewer monomer units that can be used to mediate
superimposed on top of your target recognition events (4 bases versus 20 amino Step 1
left hand. acids). This weakness is compensated for by aptamer Synthesize
mirror-image
libraries being very large (≥1015 unique sequences), target
von Willebrand factor
A glycoprotein found in the
with each molecule being a water-soluble species with Step 2
blood that is involved in the potential to fold into a tertiary structure with a well- Perform SELEX
on mirror target
haemostasis (blood clotting). defined ligand-binding site. Despite this, many targets with D-RNA
have proved to be resistant to aptamer selections. The
Click chemistry
A general class of high-yielding
synthesis of nucleic acid libraries that carry a wider range Step 3
of chemical groups seems to be a solution to this problem. Synthesize
cycloaddition reactions that L-RNA
occur between azide and One of the first approaches to increasing the func- aptamer
alkyne functional groups. Click tional diversity of nucleic acid libraries involves expand-
chemistry is often used to add
ing the genetic alphabet to include additional bases that
new functional groups to
biomolecules.
achieve Watson–Crick base-pairing through alternative
arrangements of hydrogen-bond donor and acceptor
groups. Benner and colleagues, for example, developed
an artificially expanded genetic information system,
termed AEGIS, that was used to isolate aptamers with
low nanomolar affinity for a human adenocarcinoma Figure 5 | l‑RNA aptamers or spiegelmers. A cartoon
breast cell line (MDA‑MB‑231)76. In addition to the showing the general strategy used to isolate l‑RNA
standard nucleobases of A, C, G and T, entries in the aptamers (spiegelmers) by in vitro selection59. In step 1,
AEGIS library also contain the nonstandard nucleotides the mirror image of the desired target is chemically
Z (6‑amino‑5‑nitro‑2(1H)-pyridone) and P (2‑amino- synthesized. In the case of proteins, targets are generated
imidazo[1,2‑a]-1,3,5‑triazin‑4(8H)one), which were by solid-phase synthesis using unnatural d‑amino acids.
found to be essential for aptamer binding (FIG. 6b). For small molecules, the target is generated by chemical
Although the information capacity of the AEGIS library synthesis. In step 2, standard in vitro selection is performed
using a natural RNA (or DNA) library to isolate aptamers
is higher than a standard nucleic acid library, the func-
that recognize the mirror-image version of the desired
tional similarity of AEGIS bases to natural bases suggests target. In step 3, l‑RNA (or DNA) aptamers are generated
that the AEGIS system explores a relatively small region by solid-phase synthesis using l‑phosphoramidites to
of chemical space. construct the mirror-image version of a sequence isolated
In an effort to search larger regions of chemical space, from the selection. Based on the rules of mirror-image
several groups have developed new strategies for intro- symmetry, the unnatural l‑RNA aptamer will bind the
ducing functional groups at specific positions in the natural l‑protein with an affinity constant that is identical
sequence. In this regard, Hirao and co-workers have to the binding interaction observed between the natural
constructed a library in which an unnatural hydrophobic d‑aptamer and unnatural d‑protein. SELEX, systematic
base was positioned at several predefined locations in the evolution of ligands by exponential enrichment. Adapted
with permission from from REF. 59, Elsevier.
random region of the DNA library77 (FIG. 7a). The genetically
expanded library was then used to isolate aptamers
with affinity for the human proteins VEGF165 and For each round of selection, the alkyne-modified DNA
interferon‑γ. Despite the requirement for the modi- pool, prepared by substituting thymidine triphosphate
fied base to occupy predefined positions, the resulting (TTP) with C5‑ethynyl‑2ʹ‑deoxyuridine (EdU), is further
aptamers exhibit substantially stronger (>100‑fold modified with an azide-containing compound using a
increase) binding relative to known aptamers isolated copper-mediated alkyne–azide cycloaddition reaction
from natural libraries77. The improved binding affinity (FIG. 7c), commonly referred to as click chemistry. Following
was attributed to the increased hydrophobic character of several rounds of selection and amplification, the
the unnatural base. This approach was extended using click-SELEX strategy generated an indole-modified
genetically expanded bases that were not constrained to aptamer with low-nanomolar affinity for the green
specific library positions to create high-affinity aptamers fluorescent protein (GFP) — a target known to have low
against von Willebrand factor78. As expected, the chemically aptagenic activity 80. GFP binding activity was shown to
modified aptamers showed higher affinity than previous be dependent on the presence of the indole modification,
von Willebrand factor aptamers isolated from unmodified demonstrating that in addition to increased binding affinity,
DNA libraries. chemical modifications can also improve the overall likeli-
Mayer and co-workers established a conceptually sim- hood of isolating aptamers for difficult targets79. The com-
pler approach, termed click-SELEX, in which functional patibility of this approach with other functional groups
groups are added to specific nucleotides in a DNA library provides a powerful method for exploring new regions of
after the library has been enzymatically synthesized79. chemical space.

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
a OH b
H
N
H
Inactive β2AR Active β2AR
N N H O CH3
HO O
HN N
High-affinity agonist N H N
R
O (BI167107) N N
O R
A T
H
N O H N
Perform Perform N N H N
SELEX SELEX R
N N
on inactive on active
N H O R
β2AR β2AR
G H C
H
O H N NO2
N
N N H N
R
N
N H O R
P H Z

Figure 6 | Examples of aptamer applications and chemical modifications. a | The G-protein-coupled receptor
β2-adrenoceptor (β2AR) exists in inactive (shown on the left; Protein Data Bank identifier: 2RH1) and active conformations
(shown on the right; PDB ID: 3SN6)72. The inactive form is converted to the active form on treatment with the agonist
BI167107 (shown in red). The two forms can be separately subjected to SELEX, and in this case the inactive conformation
is stabilized by three different aptamers, denoted A1, A2 and A13, whereas the active conformation was stabilized by
another aptamer, A16. The possible sites at which the aptamers bind to β2AR are indicated by orange ovals. This
stabilization enabled structural analysis of this model membrane receptor by cryo-electron microscopy. b | Artificially
expanded genetic information systems (AEGIS). Based on the premise of complementary hydrogen bonding, new
Watson–Crick base pairs, such as P•Z, can be generated by creating new combinations of hydrogen-bond donor and
acceptor groups that recognize each other through new arrangements of hydrogen-bond complementarity76.
R = 1ʹ-(5ʹ‑phosphatylribose); SELEX, systematic evolution of ligands by exponential enrichment. Adapted with permission
from REF. 72, Macmillan Publishers Limited.
The use of slow off-rate modified aptamers According to SomaLogic, SOMAmers have been gener-
(SOMAmers), developed by SomaLogic, provides access ated against >3,000 proteins and these reagents are cur-
to aptamers with still greater chemical diversity 81. In rently used as a point‑of‑care technology for monitoring
the original strategy, diversity-enhancing functional protein levels in human serum86.
groups were added to the 5 position of deoxyuridine res- Methods for evolving DNA aptamers with modi-
idues, which were subsequently converted to nucleotide fied bases are currently limited to one or two types of
triphosphates82 (FIG. 7b). Polymerase-mediated synthesis chemical modifications per library. Liu and co-workers
produces a combinatorial library of DNA molecules in overcame this problem using a ligase-mediated transla-
which each thymidine residue is uniformly replaced by tion system that makes it possible to synthesize nucleic
a specific type of functional group. Among the types of acid polymers that carry a rich assortment of different
chemical moieties tested, functional groups that resemble functional groups in the same library 87. In this approach,
the side chains of hydrophobic amino acid residues have T4 DNA ligase is used to mediate the DNA-templated
proved to be successful against a wide range of protein polymerization of short 5ʹ‑phosphorylated oligo
targets82,83. Results from several protein–SOMAmer nucleotides containing a wide variety of functional
cocrystal structures indicate that these functional groups groups (FIG. 7d). The resulting library of single-stranded
have a role in target binding and aptamer folding 84. This nucleic acid polymers is a suitable template for primer
strategy has substantially improved the percentage of extension using an archaean thermophilic B-family
effective in vitro-selected aptamers, as exemplified by polymerase, which enables regeneration of the starting
Primer extension the generation of aptamer–target pairs that were pre- template by copying functionally rich sequences back into
A DNA (or xeno-nucleic acid viously difficult to observe. The technology platform their encoding cDNA87. Hili and co-workers applied the
(XNA)) replication step in which of SomaLogic was expanded by the inclusion of DNA same approach to synthesize DNA libraries that were mod-
a nucleic acid primer is libraries that carry two types of modified bases85. Shorter ified with peptides and a wide range of small molecules
extended by annealing the
primer to a template and
random-sequence libraries of doubly modified aptamers with diverse functionality 88,89. Despite the fact that
copying the template with a have higher binding affinity for a target than traditional this methodology has not been used to generate an
polymerase. libraries (30-mer versus 40‑mer random regions)85. affinity reagent by in vitro selection, the high level of

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
technical optimization performed thus far suggests that of in vitro selection, it should come as no surprise that
sequence-defined polymers generated by template- aptamers selected to function in one environment may
directed ligation offer a rich source of chemical diversity not function well in other environments. This observation
for generating high-quality aptamer-like reagents. is not unique to aptamers; for example, antibodies gen-
erated against whole proteins tend to function better in
Specificity. The long-held adage of ‘you get what you pull-down assays, whereas those produced against short
select for’ has been one of the great truisms of the peptide epitopes are better suited for western blots91,92. The
aptamer world90. Consequently, after nearly three decades obvious solution to this problem is to develop selection
strategies that more closely resemble the desired down-
a b O O
stream applications. However, isolating aptamers that
OH R′
exhibit both high affinity and high specificity can be dif-
HN N
H
ficult to achieve using conventional selection methods,
OH O N
as these approaches either enrich for affinity (positive
R
H
selection) or enrich for specificity (negative selection),
R′ =
S
N but rarely do both simultaneously.
N H
Efforts to overcome this problem have led to inno-
H vative strategies for generating aptamers that retain
N N
N
–
O
+
N
high target binding affinity in the presence of complex
H biological environments. One approach involves the
O
in vivo selection of nuclease-resistant aptamers in a live
Ds Px N
animal model developed to target a specific disease or
HN condition. Sullenger and co-workers demonstrated that
N
tumour-specific aptamers could be generated by admin-
istering a 2ʹ‑fluoropyrimidine RNA library intravenously
c N
N to the animal, harvesting the organ of pathological inter-
O
R′′ N O
H est and extracting the library members that bound the
N
NH R-N3 NH R′′ = organ93. This technique of in vivo selection was used to
O N
Cu(I)
O
isolate nuclease-resistant aptamers for an animal model
O N
O
O O of colorectal cancer metastasis. More recently, the in vivo
selection approach has been used to generate aptamers
O
EdU O that can penetrate the parenchyma tissue of the brain in
a wild-type mouse model94. A second approach for gen-
d R′′′ erating aptamers with high specificity involves exploiting
CF 3 N the capacity of fluorescence-activated cell sorting (FACS)
NH to simultaneously screen for affinity and specificity.
R′′′ = Using a strategy called multiparameter particle display
F 3C NH
BzHN (MPPD), Soh and co-workers developed several
examples of DNA aptamers that function with low- to
S NH O NH O NH O NH
subnanomolar affinity in human serum95. This level of
binding activity was achieved using droplet PCR to create
‘monoclonal’ aptamer particles, a process whereby
O OH H 2N streptavidin-coated beads are decorated with many copies
of the same aptamer sequence96. The aptamer particles
N
are then incubated with a complex mixture comprising
the desired target labelled with AlexaFluor 488 (a green
O NH O NH O NH
dye) and competing serum proteins labelled with Alexa
Figure 7 | Chemical modifications used in aptamer selections. Numerous types of Fluor 647 (a red dye). Following incubation, FACS is used
chemical modifications have been developed to increase theNature Reviews
functional | Chemistry
diversity of DNA to isolate aptamers that exhibit both high target binding
aptamers. a | One example includes expanding the genetic alphabet with hydrophobic affinity and specificity, based on a dual-colour gating
base analogues that recognize one another through shape complementarity, such as the parameter. The resulting DNA aptamers were shown to
7-(2‑thienyl)imidazo[4,5‑b]pyridine (Ds) and diol-modified 2‑nitro‑4‑propynylpyrrole (Px) outperform monoclonal antibodies in a standard enzyme-
base pair77. b | SOMAmers, or slow off-rate modified aptamers, contain hydrophobic side linked immunosorbent assay (ELISA)95. Although the
chains at the C5 position of pyrimidine nucleotides that can enhance the functional MPPD strategy is limited to nucleic acid modifications
properties of DNA aptamers82. c | Click-SELEX uses the base analogue that can be amplified using PCR, the ability to isolate
C5‑ethynyl‑2ʹ‑deoxyuridine (EdU) as a substrate for DNA polymerases and modifies the aptamers with high affinity and specificity outside an animal
library post-synthesis using click chemistry to endow the library with enhanced chemical
model represents an important advancement in the
functionality79. d | In an effort to create functionally rich DNA molecules bearing many
different side chains, template-directed ligation was developed to synthesize DNA development of high-quality aptamers.
molecules from shorter segments of sequence-defined members87. The red ribbon
represents a short DNA oligonucleotide that functions as a synthetic codon for the Stability. In vitro selection has produced several aptamers
assembly of highly functionalized unnatural nucleic acids. R = 1ʹ-(2ʹ‑deoxy that can bind to a wide range of targets, from small
-5ʹ‑phosphatylribose); SELEX, systematic evolution of ligands by exponential enrichment. molecules to whole cells10. However, aptamers composed

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Fluorescence-activated cell of natural DNA and RNA are poor candidates for diag- only those XNA molecules that bind to a desired target.
sorting nostic and therapeutic applications owing to their lim- The selected XNA molecules are then copied back into
A development of flow ited stability in biological environments. For example, an DNA for amplification by PCR113. As with regular in vitro
cytometry that enables sorting unmodified DNA aptamer developed as an α-thrombin selection, the process is iterated until the pool becomes
of a mixture of cells into two or
more fractions based on the
inhibitor exhibited an in vivo half-life of <2 minutes when dominated by sequences with affinity for the target.
light scattering and assayed in a primate animal model97. Introducing chemical To date, three different XNA polymers (1,5‑anhydro
fluorescence signal of each cell. modifications that stabilize the nucleic acid structure hexitol nucleic acid (HNA) 114, 2ʹ‑deoxy‑2ʹ‑fluoro
against nucleic acid degrading enzymes can enhance the arabino nucleic acid (FANA)115 and α-l‑threose nucleic
Enzyme-linked
utility of aptamers in practical applications. In particular, acid (TNA)116,117) have been replicated in a Darwinian
immunosorbent assay
A diagnostic test that uses
substitution of the 2ʹ‑hydroxyl group of ribonucleotides evolution system to produce aptamers against a target
affinity reagents, typically — fundamental for phosphodiester bond cleavage by of interest (FIG. 8). Of these, TNA has a backbone repeat
antibodies, to detect a nucleases — with amino, fluoro or methoxy groups can unit that is one atom shorter than the repeat unit of DNA
substance. confer resistance98. Many 2ʹ‑modified aptamers have or RNA118,119, making it one of the more difficult XNAs
been produced by in vitro selection43,69,99–102; however, to replicate in vitro but also one of the most biologically
these modified sites are still prone to attack by nucleases, stable genetic polymers developed to date that is capable
demonstrating that the biological stability of an aptamer of Darwinian evolution120. Although XNA aptamers are
can range from being slightly more stable than natural still in their infancy, this technology platform is poised
DNA and RNA to biologically inert103,104. to witness substantial growth in the coming years121, as
Efforts to create aptamers that are immune to nuclease therapeutic aptamers require extreme biological stability.
digestion have focused on the development of replication Indeed, XNA polymers are now being developed with
systems that can facilitate the synthesis of artificial genetic expanded chemical functionality, which should increase
polymers (commonly referred to as xeno-nucleic acids or their binding affinity and targetability 122.
XNA) with backbone structures that are distinct from
natural DNA and RNA105. l‑RNA and l‑DNA aptamers, Throughput. Recent advances in aptamer selection
also known as spiegelmers (discussed above), were one of techniques have increased the quality and throughput
the first types of alternative genetic polymers developed of in vitro-selected aptamers. In conventional methods,
for aptamer production60. Although spiegelmers represent nonspecific binding and PCR amplification problems
a powerful approach for generating biologically stable can negatively affect the efficiency of aptamer selections
aptamers, this strategy is limited to the subset of targets by limiting the enrichment of functional molecules. As
that can be generated by chemical synthesis. For example, expected, an assay with a 100‑fold partitioning efficiency
this number is <1% of all human proteins, as most human (meaning 100‑fold enrichment of functional molecules
proteins are not amenable to solid-phase synthesis106. over non-functional library members) will require
However, this situation appears to be changing with many more rounds of selection than will an assay that
the development of mirror-image polymerases that can can achieve a partitioning efficiency of 100,000‑fold
synthesize l‑DNA and l-RNA107,108. Such enzymes have per round of selection. Unfortunately, most bead-based
the potential to expand the number of l‑RNA aptamers aptamer selections have partitioning efficiencies in the
that can be used in clinical trials by overcoming the target ~500‑fold range, which is why traditional processes
generation problem. often require ten or more rounds of selection. New solu-
The advent of new polymerase engineering tech- tion-based selection approaches have been developed to
niques109–112 have made possible the development of XNA avoid the problem of nonspecific aptamer binding to a
polymerases that can copy genetic information back and solid support matrix 123. For selections performed on a
forth between DNA and XNA. Using these enzymes, XNA solid surface, such as an ELISA plate, surface passivation
aptamers have been produced through an in vitro selection techniques that include surfactants have shown great
strategy that involves copying degenerate DNA libraries promise for reducing nonspecific binding of the library
into XNA and applying in vitro selection methods to isolate to the plastic surface124. These techniques are often
NH2 O
CH3
N N HN
O O
P N N O N
O – P
–O O O B B
B O O
P –O O A T
O O
F O
O NH2
O
O
N NH N
FANA HNA TNA N O N

N NH2
G C
Figure 8 | Xenonucleic acids. Recent advances in polymerase engineering have enabled the coding and decoding
of genetic information in artificial genetic polymers commonly referred to as xenonucleic acids (XNAs). Like DNA,
these polymers are formed from A, T, G and C. Examples of XNA aptamers generated by in vitro selection include
2ʹ‑fluoroarabino nucleic acid (FANA), 1,5‑anhydrohexitol nucleic acid (HNA) and α-l‑threose nucleic acid (TNA)114–117.

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Emulsion PCR coupled with quantitative real-time PCR and emulsion PCR As discussed earlier in this Review, many aptamers
A variant of the polymerase techniques that minimize the accumulation of PCR described in the literature are poorly characterized in
chain reaction (PCR) that is artefacts125. Finally, because of the recursive nature of terms of their sequence, binding properties, structure
performed in water‑in‑oil in vitro selection, protocols have been developed that and stability. On the basis of these observations, we
droplets.
accelerate aptamer production using liquid handling suggest that authors, editors and reviewers consider the
robots that can perform incubation, washing and following factors when preparing and reviewing aptamer
chromatography steps126. The use of automated proto- manuscripts.
cols minimizes human error and maximizes throughput, 1. Sequence information. Does the manuscript contain
accuracy and reproducibility. the sequence of each nucleic acid aptamer described
in the study? Are the sequences clearly annotated with
Choosing the right chemical modification information about the primer binding sites, random
The preceding section described new approaches that region boundaries and chemical modifications? If
have been taken to overcome some of the barriers that chemical modifications are used, is it clear which
have limited the success of aptamers in commercial and modifications were used (vendor and catalogue
clinical applications. As discussed, the most challenging number) and where the modifications are located
barriers to aptamer development include limited chemical in the sequence? If the aptamers were truncated or
diversity, poor selectivity, low biological stability and low otherwise modified, are the resulting sequence variants
throughput. Fortunately, recent advances have produced clearly defined?
new aptamer technologies that are vastly superior to 2. Characterization. Does the manuscript include char-
traditional approaches. Many of the newer technologies acterization of the binding interaction of the in vitro-
are based on chemical modifications and polymerases selected aptamer(s) to the cognate target? Were affinity
that are commercially available. However, other technol- measurements determined with relevant controls or
ogies require extensive knowledge of organic synthesis validated using independent techniques? Are the data
and polymerase engineering, as these building blocks believable or could it be the result of avidity rather than
and catalysts are not commercially available. In turn, affinity? Are the binding conditions (for example, salt
many engineered polymerases require expression and concentration, pH and temperature) reported? If nec-
purification from E. coli. essary, could the binding assay be repeated by an inde-
Choosing the right chemical modification depends on pendent laboratory? Are comparisons made between
several factors, including the desired downstream applica- the affinity of the final aptamer and the starting library,
tion, commercial availability of the chemical modification which are required to show gain‑of‑function activity?
and polymerase, activity of the polymerase for the modi- 3. Structure. Do the authors predict the secondary struc
fied unit when present in the template or as a nucleoside ture of the in vitro-selected aptamer? Was the
triphosphate (xNTP), and the chemistry and molecular structure validated with a doped library selection or
biology experience of the user. For users interested in were key residues tested by mutagenesis? Was struc-
developing chemically modified aptamers, we have out- ture probing performed to elucidate structure–activity
lined the strengths and weaknesses of the chemical mod- relationships?
ifications described in this Review (TABLE 1). Although we 4. Specificity. Did the authors measure aptamer specific-
anticipate that most researchers would prefer to use chem- ity against non-cognate targets? Affinity measurements
ical modifications and polymerases that are commercially made to homologues, analogues, chemical mimics
available, others may wish to pursue more exotic types of or common biological molecules present in complex
chemical modifications that require chemical synthesis mixtures can provide a strong indication of the poten-
and use polymerases that have been developed in spe- tial for off-target binding. These assays are particularly
cialized laboratories. In general, applications that require important for aptamers that are developed for practi-
extreme biological stability would benefit most from XNA cal applications that require the aptamer to function
systems, including l‑RNA and l‑DNA, that are resistant in a complex biological medium. A strong indicator
to nuclease digestion. Applications that require specific of aptamer specificity can be a pull-down assay per-
target binding would benefit instead from modifications formed in a relevant biological environment, such as
that increase the chemical diversity of the library reper- total E. coli lysate or human serum.
toire. If high biological stability and high target binding 5. Stability. Did the authors measure the biological,
affinity and selectivity are required, then it may be nec- chemical or thermal stability of the aptamer? Biological
essary to explore techniques that expand the chemical stability assays are particularly important for aptamers
functionality of XNA libraries. developed for diagnostic or therapeutic applications.
In such cases, the authors should consider measuring
Suggested standards the stability of the aptamer in relevant environments,
The growing demand for aptamers as high-quality such as concentrated human liver microsomes128
affinity reagents warrants a discussion on steps that can (which contain a high abundance of both endo-
be taken to increase the quality, reproducibility and trans- and exo-nucleases) or 50% human serum in media.
parency of studies describing in vitro-selected aptamers. If degradation occurs, it may be worthwhile for the
Analogous efforts in the antibody community are aimed authors to determine the identity of the metabolites
at preparing high-quality protein capture reagents produced by enzymatic digestion so that more stable
with known sequences and dependable functions127. chemical derivatives can be constructed.

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
Table 1 | Chemical modifications and their contributions to aptamer technologies

Type Targets Characteristics Availability
2′-F RNA Numerous proteins t1/2 ~45 min in HLM 113
NTP analogues available
t1/2 ~45 min in 50% HS113 Natural or engineered polymerases
2′-OCH3 RNA Numerous proteins t1/2 ~24 h in SVPE97 NTP analogues available
Natural or engineered polymerases
Click-SELEX Green fluorescent protein Side chains increase the Reagents available
functional diversity
Natural polymerases
SOMAmers About 3,000 protein targets Hydrophobic groups improve NTPs require synthesis
aptamer binding and targetability
Natural polymerases
AEGIS Breast cancer cell line AEGIS increases the sequence ZTP and PTP available
diversity
(MDA‑MB‑231) Natural polymerases
Spiegelmers Numerous proteins and Refractory to nucleases 55
Reagents available
some small molecules
Require targets that can be Natural polymerases
synthesized
FANA HIV reverse transcriptase t1/2 ~45 min in HLM113 fNTPs are available
t1/2 ~4 h in 50% HS113 Requires engineered polymerases
Fluorine increases the
hydrophobicity of DNA
HNA HIV TAR RNA motif Refractory to nucleases107 hNTPs are not available
Hen egg lysozyme Requires specialized knowledge Requires engineered polymerases
TNA Human α‑thrombin Refractory to nucleases 113
tNTPs are not available
Requires specialized knowledge Requires engineered polymerases
2′‑ F, 2′‑ deoxy‑2′‑ fluoro; 2′‑OCH3, 2′‑ deoxy‑2′‑ methoxy; AEGIS, artificially expanded genetic information system;
FANA, 2′‑deoxy‑2′‑ fluoroarabino nucleic acid; fNTPs, FANA nucleotide triphosphates; HLM, human liver microsomes;
HNA, 1,5‑anhydrohexitol nucleic acid; hNTPs, HNA nucleotide triphosphates; HS, human serum; NTPs, nucleotide triphosphates;
PTP, (2-aminoimidazo[1,2-a]-1,3,5-triazin-4(8H)one) triphosphate; SOMAmers, slow off-rate modified aptamers; SVPE, snake venom
phosphodiesterase; TAR, trans-activation response element; TNA, α‑l threose nucleic acid; tNTPS, TNA nucleotide triphosphates;
ZTP, (6-amino-5-nitro-2(1H)-pyridone) triphosphate.
Outlook functions that are not currently accessible to antibodies

The past quarter century has witnessed tremendous or that antibodies have difficulty performing, such as
growth in the discovery and application of in vitro- the recognition of small-molecule targets or targets that
selected aptamers. The specific quantifiable metrics dis- are toxic to cells, as well as the use of next-generation
cussed in this Review provide an opportunity to reflect sequencing techniques that can accelerate the search for
on the accomplishments of aptamers as a powerful class high-quality aptamers. Given the importance of chemical
of synthetic affinity reagents. However, relative to anti- modifications to next-generation aptamers, the field
bodies, aptamers remain an early-stage technology that as a whole would benefit from innovations in nucleic
will require further development to ensure sustained acid chemistry that make chemically modified nucleo-
growth over the next 25 years. Key areas of development tides more accessible. In particular, new advances that
that will help close the gap between aptamers and anti- increase the scale and purity of modified nucleotide
bodies include new selection techniques that enable triphosphates (xNTPs) would have an important and
aptamers to function in their desired downstream appli- lasting effect on the field. By focusing on these areas
cation, biologically stable scaffolds that are refractory to and learning from past results, we are confident that the
nuclease digestion and new sensors that can be tailored promise of creating an inexpensive renewable source of
to any given target. In addition, when considering new high-quality affinity reagents is finally within our grasp
aptamer projects, greater emphasis should be placed on and could be realized in the next few years.
1. Venter, J. C. et al. The sequence of the human genome. 5. Zovoilis, A., Cifuentes-Rojas, C., Chu, H. P., 9. Saiki, R. et al. Primer-directed enzymatic
Science 291, 1304–1351 (2001). Hernandez, A. J. & Lee, J. T. Destabilization of B2 RNA amplification of DNA with a thermostable DNA
2. Lander, E. S. et al. Initial sequencing and analysis by EZH2 activates the stress response. Cell 167, polymerase. Science 239, 487–491 (1988).
of the human genome. Nature 409, 860–921 1788–1802 (2016). 10. Wilson, D. S. & Szostak, J. W. In vitro selection of
(2001). 6. Moore, P. B. & Steitz, T. A. The involvement of RNA functional nucleic acids. Annu. Rev. Biochem. 68,
3. Doudna, J. A. & Cech, T. R. The chemical repertoire in ribosome function. Nature 418, 229–235 611–647 (1999).
of natural ribozymes. Nature 418, 222–228 (2002). 11. Ellington, A. D. & Szostak, J. W. In vitro selection
(2002). 7. Onoa, B. & Tinoco, I. Jr. RNA folding and unfolding. of RNA molecules that bind specific ligands. Nature
4. Sonenberg, N. & Hinnebusch, A. G. Regulation of Curr. Opin. Struct. Biol. 14, 374–379 (2004). 346, 818–822 (1990).
translation initiation in eukaryotes: mechanisms 8. Caruthers, M. H. Gene synthesis machines: DNA 12. Tuerk, C. & Gold, L. Systematic evolution of ligands by
and biological targets. Cell 136, 731–745 chemistry and its uses. Science 230, 281–285 exponential enrichment: RNA ligands to bacteriophage
(2009). (1985). T4 DNA polymerase. Science 249, 505–510 (1990).

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
13. Robertson, D. L. & Joyce, G. F. Selection in vitro of an 42. Ng, E. W. M. et al. Pegaptanib, a targeted anti-VEGF 66. Liang, C. et al. Aptamer-functionalized lipid
RNA enzyme that specifically cleaves single-stranded aptamer for ocular vascular disease. Nat. Rev. Drug nanoparticles targeting osteoblasts as a novel RNA
DNA. Nature 344, 467–468 (1990). Discov. 5, 123–132 (2006). interference-based bone anabolic strategy. Nat. Med.
14. Levine, H. A. & Nilsen-Hamilton, M. A mathematical 43. Ruckman, J. et al. 2ʹ‑Fluoropyrimidine RNA-based 21, 288–294 (2015).
analysis of SELEX. Comput. Biol. Chem. 31, 11–35 aptamers to the 165‑amino acid form of vascular 67. Douek, D. C., Roederer, M. & Koup, R. A. Emerging
(2007). endothelial growth factor (VEGF165). Inhibition of concepts in the immunopathogenesis of AIDS. Annu.
15. Szostak, J. W. In vitro genetics. Trends Biochem. Sci. receptor binding and VEGF-induced vascular Rev. Med. 60, 471–484 (2009).
17, 89–93 (1992). permeability through interactions requiring the exon 68. Tang, J. & Kaslow, R. A. The impact of host genetics on
16. Joyce, G. F. In vitro evolution of nucleic acids. Curr. 7‑encoded domain. J. Biol. Chem. 273, 20556–20567 HIV infection and disease progression in the era of
Opin. Struct. Biol. 4, 331–336 (1994). (1998). highly active antiretroviral therapy. AIDS 17 (Suppl.
17. Jayasena, S. D. Aptamers: an emerging class of 44. Group, C. R. et al. Ranibizumab and bevacizumab for 4), 51–60 (2003).
molecules that rival antibodies in diagnostics. Clin. neovascular age-related macular degeneration. 69. Zhou, J. et al. Selection, characterization and
Chem. 45, 1628–1650 (1999). N. Engl. J. Med. 364, 1897–1908 (2011). application of new RNA HIV gp 120 aptamers for
18. Ruigrok, V. J., Levisson, M., Eppink, M. H. M., 45. Heier, J. S. et al. Intravitreal aflibercept (VEGF trap- facile delivery of Dicer substrate siRNAs into HIV
Smidt, H. & van der Oost, J. Alternative affinity tools: eye) in wet age-related macular degeneration. infected cells. Nucleic Acids Res. 37, 3094–3109
more attractive than antibodies. Biochem. J. 436, Ophthalmology 119, 2537–2548 (2012). (2009).
1–13 (2011). 46. Abeydeera, N. D. et al. Evoking picomolar binding in 70. Zhou, J., Li, H., Li, S., Zaia, J. & Rossi, J. J. Novel dual
19. Binz, H. K. et al. High-affinity binders selected from RNA by a single phosphorodithioate linkage. Nucleic inhibitory function aptamer–siRNA delivery system for
designed ankyrin repeat protein libraries. Nat. Acids Res. 44, 8052–8064 (2016). HIV‑1 therapy. Mol. Ther. 16, 1481–1489 (2008).
Biotechnol. 22, 575–582 (2004). 47. Williams, B. A. R. et al. Creating protein affinity 71. Zhou, J. et al. Functional in vivo delivery of multiplexed
20. Hey, T., Fiedler, E., Rudolph, R. & Fiedler, M. Artificial, reagents by combining peptide ligands on synthetic anti-HIV‑1 siRNAs via a chemically synthesized aptamer
non-antibody binding proteins for pharmaceutical and DNA scaffolds. J. Am. Chem. Soc. 131, 17233–17241 with a sticky bridge. Mol. Ther 21, 192–200 (2013).
industrial applications. Trends Biotechnol. 23, (2009). 72. Kahsai, A. W. et al. Conformationally selective RNA
514–522 (2005). 48. Bittker, J. A., Phillips, K. J. & Liu, D. R. Recent aptamers allosterically modulate the β2‑adrenoceptor.
21. Hoogenboom, H. R. Selecting and screening advances in the in vitro evolution of nucleic acids. Curr. Nat. Chem. Biol. 12, 709–716 (2016).
recombinant antibody libraries. Nat. Biotechnol. 23, Opin. Chem. Biol. 6, 367–374 (2002). 73. Lefkowitz, R. J. A brief history of G‑protein coupled
1105–1116 (2005). 49. Yang, K. A. et al. Recognition and sensing of low- receptors (Nobel Lecture). Angew. Chem. Int. Ed. 52,
22. Sidhu, S. S. & Fellouse, F. A. Synthetic therapeutic epitope targets via ternary complexes with 6366–6378 (2013).
antibodies. Nat. Chem. Biol. 2, 682–688 (2006). oligonucleotides and synthetic receptors. Nat. Chem. 74. Koehn, F. E. & Carter, G. T. The evolving role of natural
23. Carothers, J. M., Goler, J. A., Juminaga, D. & 6, 1003–1008 (2014). products in drug discovery. Nat. Rev. Drug Discov. 4,
Keasling, J. D. Model-driven engineering of RNA 50. Ferguson, B. S. et al. Real-time, aptamer-based tracking 206–220 (2005).
devices to quantitatively program gene expression. of circulating therapeutic agents in living animals. Sci. 75. Bastian, A. A., Marcozzi, A. & Herrmann, A. Selective
Science 334, 1716–1719 (2011). Transl Med. 5, 213ra165 (2013). transformations of complex molecules are enabled by
24. Marx, V. Finding the right antibody for the job. Nat. 51. Amaya-Gonzalez, S., de‑los-Santos-Alvarez, N., aptameric protective groups. Nat. Chem. 4, 789–793
Methods 10, 703–707 (2013). Miranda-Ordieres, A. J. & Lobo-Castanon, M. J. (2012).
25. Keefe, A. D., Pai, S. & Ellington, A. D. Aptamers as Aptamer-based analysis: a promising alternative for 76. Sefah, K. et al. In vitro selection with artificial
therapeutics. Nat. Rev. Drug Discov. 9, 537–550 food safety control. Sensors 13, 16292–16311 expanded genetic information systems. Proc. Natl
(2010). (2013). Acad. Sci. USA 111, 1449–1454 (2014).
26. Nimjee, S. M., Rusconi, C. P. & Sullenger, B. A. 52. Hayat, A. & Marty, J. L. Aptamer based 77. Kimoto, M., Yamashige, R., Matsunaga, K. I.,
Aptamers: an emerging class of therapeutics. Annu. electrochemical sensors for emerging environmental Yokoyama, S. & Hirao, I. Generation of high-affinity
Rev. Med. 56, 555–583 (2005). pollutants. Front. Chem. 2, 41 (2014). DNA aptamers using an expanded genetic alphabet.
27. Zhou, J. & Rossi, J. Aptamers as targeted 53. Smiley, D. A. & Becker, R. C. Factor IXa as a target for Nat. Biotechnol. 31, 453–457 (2013).
therapeutics: current potential and challenges. Nat. anticoagulation in thrombotic disorders and conditions. 78. Matsunaga, K. I., Kimoto, M. & Hirao, I. High-affinity
Rev. Drug Discov. 16, 181–202 (2017). Drug Discov. Today 19, 1445–1453 (2014). DNA aptamer generation targeting von Willebrand
28. Rusconi, C. P. et al. Antidote-mediated control of an 54. Rusconi, C. P. et al. RNA aptamers as reversible factor A1‑domain by genetic alphabet expansion for
anticoagulant aptamer in vivo. Nat. Biotechnol 22, antagonists of coagulation factor IXa. Nature 419, systematic evolution of ligands by exponential
1423–1428 (2004). 90–94 (2002). enrichment using two types of libraries composed of
29. Tan, W., Donovan, M. J. & Jiang, J. Aptamers from 55. Povsic, T. J. et al. Use of the REG1 anticoagulation five different bases. J. Am. Chem. Soc. 139, 324–334
cell-based selection for bioanalytical applications. system in patients with acute coronary syndromes (2017).
Chem. Rev. 113, 2842–2862 (2013). undergoing percutaneous coronary intervention: 79. Tolle, F., Brändle, G. M., Natzner, D. & Mayer, G.
30. Sismour, A. M. et al. PCR amplification of DNA results from the phase II RADAR-PCI study. A versatile approach towards nucleobase-modified
containing non-standard base pairs by variants of EuroIntervention 10, 431–438 (2014). aptamers. Angew. Chem. Int. Ed. 54, 10971–10974
reverse transcriptase from Human Immunodeficiency 56. Povsic, T. J. et al. A phase 2, randomized, partially (2015).
Virus‑1. Nucleic Acids Res. 32, 728–735 (2004). blinded, active-controlled study assessing the efficacy 80. Shui, B. et al. RNA aptamers that functionally interact
31. Pinheiro, A. V., Han, D., Shih, W. M. & Yan, H. and safety of variable anticoagulation reversal using with green fluorescent protein and its derivatives.
Challenges and opportunities for structural DNA the REG1 system in patients with acute coronary Nucleic Acids Res. 40, e39 (2012).
nanotechnology. Nat. Nanotechnol. 6, 763–772 syndromes: results of the RADAR trial. Eur. Heart J. 81. Gold, L. et al. Aptamer-based multiplexed proteomic
(2011). 34, 2481–2489 (2013). technology for biomarker discovery. PLoS One 5,
32. Lund, K. et al. Molecular robots guided by 57. Hoellenriegel, J. et al. The Spiegelmer NOX‑A12, a e15004 (2010).
prescriptive landscapes. Nature 465, 206–210 novel CXCL12 inhibitor, interferes with chronic 82. Vaught, J. D. et al. Expanding the chemistry of DNA for
(2010). lymphocytic leukemia cell motility and causes in vitro selection. J. Am. Chem. Soc. 132, 4141–4151
33. Mayer, G. The chemical biology of aptamers. Angew. chemosensitization. Blood 123, 1032–1039 (2010).
Chem. Int. Ed. 48, 2672–2689 (2009). (2014). 83. Gupta, S. et al. Chemically modified DNA aptamers
34. Cho, E. J., Lee, J.‑W. & Ellington, A. D. Applications of 58. Marasca, R. & Maffei, R. NOX‑A12: mobilizing CLL bind interleukin‑6 with high affinity and inhibit
aptamers as sensors. Annu. Rev. Anal. Chem. 2, away from home. Blood 123, 952–953 (2014). signaling by blocking its interaction with interleukin‑6
241–264 (2009). 59. Vater, A. & Klussmann, S. Turning mirror-image receptor. J. Biol. Chem. 289, 8706–8719 (2014).
35. Ozer, A., Pagano, J. M. & Lis, J. T. New technologies oligonucleotides into drugs: the evolution of 84. Gelinas, A. D., Davies, D. R. & Janjic, N. Embracing
provide quantum changes in the scale, speed, and Spiegelmer therapeutics. Drug Discov. Today 20, proteins: structural themes in aptamer–protein
success of SELEX methods and aptamer 147–155 (2015). complexes. Curr. Opin. Struct. Biol. 36, 122–132
characterization. Mol. Ther. Nucleic Acids 3, e183 60. Klussmann, S., Nolte, A., Bald, R., Erdmann, V. A. & (2016).
(2014). Furste, J. P. Mirror-image RNA that binds d‑adenosine. 85. Gawande, B. N. et al. Selection of DNA aptamers with
36. McKeague, M. et al. Analysis of in vitro aptamer Nat. Biotechnol. 14, 1112–1115 (1996). two modified bases. Proc. Natl Acad. Sci. USA 114,
selection parameters. J. Mol. Evol. 81, 150–161 61. Sczepanski, J. T. & Joyce, G. F. Specific inhibition of 2898–2903 (2017).
(2015). microRNA processing using l‑RNA aptamers. J. Am. 86. Ostroff, R. M. et al. Unlocking biomarker discovery:
37. Sassanfar, M. & Szostak, J. W. An RNA motif that Chem. Soc. 137, 16032–16037 (2015). large scale application of aptamer proteomic
binds ATP. Nature 364, 550–553 (1993). 62. Kruspe, S., Mittelberger, F., Szameit, K. & Hahn, U. technology for early detection of lung cancer. PLoS One
38. Wilson, D. S., Keefe, A. D. & Szostak, J. W. The use of Aptamers as drug delivery vehicles. ChemMedChem 9, 5, e15003 (2010).
mRNA display to select high-affinity protein-binding 1998–2011 (2014). 87. Hili, R., Niu, J. & Liu, D. R. DNA ligase-mediated
peptides. Proc. Natl Acad. Sci. USA 98, 3750–3755 63. Olson, W. C., Heston, W. D. & Rajasekaran, A. K. translation of DNA into densely functionalized nucleic
(2001). Clinical trials of cancer therapies targeting prostate- acid polymers. J. Am. Chem. Soc. 135, 98–101
39. Vaish, N. K., Larralde, R., Fraley, A. W., Szostak, J. W. specific membrane antigen. Rev. Recent Clin. Trials 2, (2013).
& McLaughlin, L. W. A novel, modification-dependent 182–190 (2007). 88. Guo, C., Watkins, C. P. & Hili, R. Sequence-defined
ATP-binding aptamer selected from an RNA library 64. Lupold, S. E., Hicke, B. J., Lin, Y. & Coffey, D. S. scaffolding of peptides on nucleic acid polymers. J. Am.
incorporating a cationic functionality. Biochemistry Identification and characterization of nuclease- Chem. Soc. 137, 11191–11196 (2015).
42, 8842–8851 (2003). stabilized RNA molecules that bind human prostate 89. Kong, D., Lei, Y., Yeung, W. & Hili, R. Enzymatic
40. Eaton, B. E. et al. Post-SELEX combinatorial cancer cells via the prostate-specific membrane synthesis of sequence-defined synthetic nucleic acid
optimization of aptamers. Bioorg. Med. Chem. 5, antigen. Cancer Res. 62, 4029–4033 (2002). polymers with diverse functional groups. Angew. Chem.
1087–1096 (1997). 65. Pastor, F., Kolonias, D., Giangrande, P. H. & Gilboa, E. Int. Ed. 55, 13164–13168 (2016).
41. Khvorova, A. & Watts, J. K. The chemical evolution of Induction of tumour immunity by targeted inhibition of 90. Zhao, H. & Arnold, F. H. Combinatorial protein design:
oligonucleotide therapies of clinical utility. Nat. nonsense-mediated mRNA decay. Nature 465, strategies for screening protein libraries. Curr. Opin.
Biotechnol. 35, 238–248 (2017). 227–230 (2010). Struct. Biol. 7, 480–485 (1997).

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.
REVIEWS
91. Bordeaux, J. et al. Antibody validation. Biotechniques replication and transcription. Nat. Chem. 8, 698–704 123. Mendonsa, S. D. & Bowser, M. T. In vitro evolution of
48, 197–209 (2010). (2016). functional DNA using capillary electrophoresis. J. Am.
92. Marx, V. Calling the next generation of affinity reagents. 108. Pech, A. et al. A thermostable d‑polymerase for mirror- Chem. Soc. 126, 20–21 (2004).
Nat. Methods 10, 829–833 (2013). image PCR. Nucleic Acids Res. 45, 3997–4005 124. Ouellet, E., Foley, J. H., Conway, E. M. & Haynes, C. Hi‑Fi
93. Mi, J. et al. In vivo selection of tumor-targeting RNA (2017). SELEX: a high-fidelity digital-PCR based therapeutic
motifs. Nat. Chem. Biol. 6, 22–24 (2010). 109. Larsen, A. C. et al. A general strategy for expanding aptamer discovery platform. Biotechnol. Bioeng. 112,
94. Cheng, C., Chen, Y. H., Lennox, K. A., Behlke, M. A. & polymerase function by droplet microfluidics. Nat. 1506–1522 (2015).
Davidson, B. L. In vivo SELEX for identification of brain- Commun. 7, 11235 (2016). 125. Williams, R. et al. Amplification of complex gene libraries
penetrating aptamers. Mol. Ther. Nucleic Acids 2, e67 110. Ghadessy, F. J., Ong, J. L. & Holliger, P. Directed by emulsion PCR. Nat. Methods 3, 545–550 (2006).
(2013). evolution of polymerase function by compartmentalized 126. Cox, J. C. et al. Automated selection of aptamers
95. Wang, J. et al. Multiparameter particle display (MPPD): self-replication. Proc. Natl Acad. Sci. USA 98, against protein targets translated in vitro: from gene to
a quantitative screening method for the discovery of 4552–4557 (2001). aptamer. Nucleic Acids Res. 30, e108 (2002).
highly specific aptamers. Angew. Chem. Int. Ed. 56, 111. Houlihan, G., Arangundy-Franklin, S. & Holliger, P. 127. Bradbury, A. & Pluckthun, A. Reproducibility:
744–747 (2017). Exploring the chemistry of genetic information storage standardize antibodies used in research. Nature 518,
96. Wang, J. et al. Particle display: a quantitative screening and propagation through polymerase engineering. Acc. 27–29 (2015).
method for generating high-affinity aptamers. Angew. Chem. Res. 50, 1079–1087 (2017). 128. Barrett, S. E. et al. An in vivo evaluation of amphiphilic,
Chem. Int. Ed. 126, 4896–4901 (2014). 112. Chen, T. & Romesberg, F. E. Directed polymerase biodegradable peptide copolymers as siRNA delivery
97. Griffin, L. C., Tidmarsh, G. F., Bock, L. C., Toole, J. J. & evolution. FEBS Lett. 588, 219–229 (2014). agents. Int. J. Pharm. 466, 58–67 (2014).
Leung, L. L. In vivo anticoagulant properties of a novel 113. Taylor, A. I. & Holliger, P. Directed evolution of artificial 129. Long, S. B., Long, M. B., White, R. R. & Sullenger, B. A.
nucleotide-based thrombin inhibitor and enzymes (XNAzymes) from diverse repertoires of Crystal structure of an RNA aptamer bound to
demonstration of regional anticoagulation in synthetic genetic polymers. Nat. Protoc. 10, thrombin. RNA 14, 2504–2512 (2008).
extracorporeal circuits. Blood 81, 3271–3276 (1993). 1625–1642 (2015).
98. Keefe, A. D. & Cload, S. T. SELEX with modified 114. Pinheiro, V. B. et al. Synthetic genetic polymers capable Acknowledgements
nucleotides. Curr. Opin. Chem. Biol. 12, 448–456 of heredity and evolution. Science 336, The authors thank members of the Chaput laboratory for
(2008). 341–344 (2012). helpful comments and suggestions, and N. Chim for assis-
99. Lin, Y., Qiu, Q., Gill, S. C. & Jayasena, S. D. Modified 115. Alves Ferreira-Bravo, I., Cozens, C., Holliger, P. & tance with the figures. This work was supported by the
RNA sequence pools for in vitro selection. Nucleic Acids DeStefano, J. J. Selection of Defense Advanced Research Projects Agency (DARPA) Folded
Res. 22, 5229–5234 (1994). 2ʹ‑deoxy‑2ʹ‑fluoroarabinonucleotide (FANA) aptamers Non-Natural Polymers with Biological Function (Fold Fx)
100. Burmeister, P. E. et al. Direct in vitro selection of a that bind HIV‑1 reverse transcriptase with picomolar Program (Award No.N66001‑16‑2‑4061).
2ʹ‑O‑methyl aptamer to VEGF. Chem. Biol. 12, 25–33 affinity. Nucleic Acids Res. 43, 9587–9599 (2015).
(2005). 116. Yu, H., Zhang, S. & Chaput, J. C. Darwinian evolution of Competing interests statement
101. Burmeister, P. E. et al. 2ʹ‑Deoxy purine, 2ʹ‑O‑methyl an alternative genetic system provides support for TNA The authors declare no competing interests.
pyrimidine (dRmY) aptamers as candidate as an RNA progenitor. Nat. Chem. 4, 183–187 (2012).
therapeutics. Oligonucleotides 16, 337–351 (2006). 117. Dunn, M. R. & Chaput, J. C. Reverse transcription of Publisher’s note
102. Thirunavukarasu, D., Chen, T., Liu, Z., Hongdilokkul, N. threose nucleic acid by a naturally occurring DNA Springer Nature remains neutral with regard to jurisdictional
& Romesberg, F. E. Selection of 2ʹ‑fluoro-modified polymerase. ChemBioChem 17, 1804–1808 (2016). claims in published maps and institutional affiliations.
aptamers with optimized properties. J. Am. Chem. Soc. 118. Schöning, K. U. et al. Chemical etiology of nucleic acid
139, 2892–2895 (2017). structure: the α-threofuranosyl-(3ʹ→2′) oligonucleotide How to cite this article
103. Cummins, L. L. et al. Characterization of fully system. Science 290, 1347–1351 (2000). Dunn, M. R., Jimenez, R. M. & Chaput, J. C. Analysis of
2ʹ‑modified oligoribonucleotide hetero- and 119. Ebert, M. O., Mang, C., Krishnamurthy, R., aptamer discovery and technology. Nat. Rev. Chem. 1, 0076
homoduplex hybridization and nuclease sensitivity. Eschenmoser, A. & Jaun, B. The structure of a TNA– (2017).
Nucleic Acids Res. 23, 2019–2024 (1995). TNA complex in solution: NMR study of the octamer
104. Noronha, A. M. et al. Synthesis and biophysical duplex derived from α-(l)-threofuranosyl-(3ʹ‑2ʹ)-
properties of arabinonucleic acids (ANA): circular CGAATTCG. J. Am. Chem. Soc. 130, 15105–15115
DATABASES
dichroic spectra, melting temperatures, and (2008).
RCSB Protein Data Bank: http://www.rcsb.org/pdb/home/
ribonuclease H susceptibility of ANA•RNA hybrid 120. Culbertson, M. C. et al. Evaluating TNA stability under
home.do
duplexes. Biochemistry 39, 7050–7062 (2000). simulated physiological conditions. Bioorg. Med. Chem.
105. Joyce, G. F. Toward an alternative biology. Science 336, Lett. 26, 2418–2421 (2016). FURTHER INFORMATION
307–308 (2012). 121. Tizei, P. A., Csibra, E., Torres, L. & Pinheiro, V. B. NeoVentures Biotechnologies: http://www.neoventures.ca
106. Legrain, P. et al. The human proteome project: current Selection platforms for directed evolution in synthetic
state and future direction. Mol. Cell. Proteomics 10, biology. Biochem. Soc. Trans. 44, 1165–1175 (2016). SUPPLEMENTARY INFORMATION
M111.009993 (2011). 122. Mei, H. et al. Synthesis and polymerase activity of a See online article: S1–S21 (tables)
107. Wang, Z., Xu, W., Liu, L. & Zhu, T. F. A synthetic fluorescent cytidine TNA triphosphate analogue. Nucleic
ALL LINKS ARE ACTIVE IN THE ONLINE PDF
molecular system capable of mirror-image genetic Acids Res. 45, 5629–5638 (2017).

©
2
0
1
7
M
a
c
m
i
l
l
a
n
P
u
b
l
i
s
h
e
r
s
L
i
m
i
t
e
d
,
p
a
r
t
o
f
S
p
r
i
n
g
e
r
N
a
t
u
r
e
.
A
l
l
r
i
g
h
t
s
r
e
s
e
r
v
e
d
.

Analysis of aptamer discovery and technology

Uploaded by

Copyright:

Available Formats

Analysis of aptamer discovery and technology

Uploaded by

Document Information

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

Analysis of aptamer discovery and technology

Uploaded by

Copyright:

Available Formats

REVIEWS

Analysis of aptamer discovery

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 1

Nature Reviews | Chemistry

a Aptamer publications Figure 1 | Trends in aptamer publications.

550 application of aptamers (application articles) and 800 are

mercial availability and ease of handling — the past few

years have witnessed a rise in unnatural nucleic acids as

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 3

4 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

Figure 2 | Post-SELEX modifications. a | Macugen (pegaptanib sodium) is a therapeutic RNA aptamer,

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 5

a Application breakdown b Top ten most applied targets

Vascular endothelial 144

Biophysical discovery Nucleolin 65

6 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 7

8 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

Mirror-image symmetry Chemical diversity. The limited chemical functionality

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 9

Nature Reviews | Chemistry

10 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 11

FANA HNA TNA N O N

12 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 13

Table 1 | Chemical modifications and their contributions to aptamer technologies

Outlook functions that are not currently accessible to antibodies

14 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

NATURE REVIEWS | CHEMISTRY VOLUME 1 | ARTICLE NUMBER 0076 | 15

16 | ARTICLE NUMBER 0076 | VOLUME 1 www.nature.com/natrevchem

You might also like