Rational Re-engineering of a Transcriptional Silencing PreQ1 Riboswitch

Article abstract of DOI:10.1021/jacs.5b03405

Re-engineered riboswitches that no longer respond to cellular metabolites, but that instead can be controlled by synthetic molecules, are potentially useful gene regulatory tools for use in synthetic biology and biotechnology fields. Previously, extensive genetic selection and screening approaches were employed to re-engineer a natural adenine riboswitch to create orthogonal ON-switches, enabling translational control of target gene expression in response to synthetic ligands. Here, we describe how a rational targeted approach was used to re-engineer the PreQ₁ riboswitch from Bacillus subtilis into an orthogonal OFF-switch. In this case, the evaluation of just six synthetic compounds with seven riboswitch mutants led to the identification of an orthogonal riboswitch-ligand pairing that effectively repressed the transcription of selected genes in B. subtilis. The streamlining of the re-engineering approach, and its extension to a second class of riboswitches, provides a methodological platform for the creation of new orthogonal regulatory components for biotechnological applications including gene functional analysis and antimicrobial target validation and screening. (Graph Presented).

Full text of DOI:10.1021/jacs.5b03405

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)
Article
Rational Re-engineering of a Transcriptional Silencing PreQ1 Riboswitch
Ming-Cheng Wu, Phillip T. Lowe, Christopher J. Robinson, Helen
A. Vincent, Neil Dixon, James Leigh, and Jason Micklefield
J. Am. Chem. Soc., Just Accepted Manuscript • DOI: 10.1021/jacs.5b03405 • Publication Date (Web): 24 Jun 2015
Downloaded from http://pubs.acs.org on June 30, 2015
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of the American Chemical Society is published by the American Chemical
Society. 1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 9
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Rational Re-engineering of a Transcriptional Silencing PreQ1 Ri-
boswitch.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Ming-Cheng Wu‡, Phillip T. Lowe‡, Christopher J. Robinson, Helen A. Vincent, Neil Dixon,
James Leigh & Jason Micklefield*
School of Chemistry and Manchester Institute of Biotechnology, The University of Manchester, 131 Princess Street,
Manchester M1 7DN, United Kingdom.
ABSTRACT: Re-engineered riboswitches that no longer respond to cellular metabolites, but which instead can be con-
trolled by synthetic molecules, are potentially useful gene regulatory tools for use in synthetic biology and biotechnology
fields. Previously, extensive genetic selection and screening approaches were employed to re-engineer a natural adenine
riboswitch to create orthogonal ON-switches, enabling translational control of target gene expression in response to syn-
thetic ligands. Here we describe how a rational targeted approach was used to re-engineer the PreQ1 riboswitch from Ba-
cillus subtilis into an orthogonal OFF-switch. In this case, the evaluation of just six synthetic compounds with seven ri-
boswitch mutants, led to the identification of an orthogonal riboswitch-ligand pairing which effectively repressed the
transcription of selected genes in B. subtilis. The streamlining of the re-engineering approach, and its extension to a se-
cond class of riboswitches, provides a methodological platform for the creation of new orthogonal regulatory components
for biotechnological applications including gene functional analysis, antimicrobial target validation and screening.
have adverse effects, perturbing global gene expression
levels and cellular physiology.9
INTRODUCTION
Riboswitches are non-coding regions of mRNA that regu-
late gene expression through the selective binding of me-
tabolites.1,2 Most riboswitches are found within the 5ʹ un-
translated regions (5ʹ-UTRs) of bacterial mRNAs which
encode gene products required for the biosynthesis, ca-
tabolism or transport of the cognate metabolite.1,2 Ri-
boswitches exhibit a modular architecture consisting of
an aptamer domain which, upon ligand binding, induces
a conformational change in an overlapping expression
platform modulating gene expression, primarily through
transcription termination or translation initiation.1-3
Engineered riboswitches that respond to synthetic
non-natural ligands, rather than metabolites, are highly
desirable.4 The most common strategy used to create such
switches is to insert RNA aptamers that have been gener-
ated by in vitro selection into the 5ʹ-UTR of a target gene,
and use a genetic selection or screen to isolate function-
ing riboswitches.10-13 Recent studies have shown that hy-
brid riboswitches can also be rationally designed, through
the fusion of synthetic aptamers with expression plat-
forms derived from natural riboswitches, to create func-
tional chimeric switches.14,15 In theory, it should be possi-
ble to generate a riboswitch that responds to any ligand of
choice using these approaches. However, there is a lim-
ited availability of synthetic aptamers which have been
demonstrated to function as components of riboswitches,
with only theophylline and tetracycline aptamers finding
widespread use.16 This is because the development of new
aptamers in vitro can be laborious and, critically, does not
guarantee high specificity or functionality in vivo.4,17
RNA-based gene regulators, such as riboswitches, are
an attractive starting point for the development of new
gene expression tools for the synthetic biology and bio-
technology fields.4-8 RNA regulatory components have the
advantage over their protein counterparts in that they are
inherently more versatile, being relatively easy to design
or manipulate in a predictable manner and being trans-
ferable between organisms.4-6 Natural riboswitches pro-
vide a diverse array of ready-made components, with dif-
ferent aptamer domains identified for more than 20 spe-
cific metabolite ligands,1,2 and expression platforms that
exhibit considerable mechanistic diversity.1,2 The modu-
larity of these regulatory components is also demonstrat-
ed by the natural occurrence of tandem riboswitches
which can function as two-input Boolean logic gates or
provide more binary switching responses.2 Despite these
advantages, most riboswitches respond to essential me-
tabolites that are present in cells at variable levels, and
the exogenous addition of metabolites at the concentra-
tions required to modulate riboswitch function could
To address the need for new riboswitch-based gene
expression tools that respond to non-natural ligands, we
have previously reported a strategy for re-engineering
natural riboswitches to generate orthogonal aptamer do-
mains that are selective for artificial ligands rather than
the cognate natural metabolite.18-20 The ligand binding
site of the add A-riboswitch from Vibrio vulnificus was
subjected to saturation mutagenesis, then an exhaustive
chemical genetic selection strategy was used to identify
functional orthogonal riboswitch-ligand combinations.18
Through this approach two novel mutant aptamer do-
ACS Paragon Plus Environment

Journal of the American Chemical Society
mains were generated that respond to triazine-based lig-
Page 2 of 9
A
ands.18 Subsequent rounds of structure-guided screening
identified superior second-generation pyrimidopyrimi-
dine ligands.20 In principle, this strategy is applicable to
any riboswitch. However, the extensive selection and
screening process adopted was laborious, with more than
100 potential ligands being evaluated against the panel of
add riboswitch mutants.18,20 With hindsight, the add A-
aptamers could have been re-engineered through a more
rational approach. Indeed, a common feature of the or-
thogonal aptamer-ligand pairings that we have identified
is that the direction of a few key hydrogen bonding inter-
actions between the aptamer and ligand had been re-
versed, through conservative uracil to cytosine mutations,
with complementary changes to the structure of the syn-
thetic ligand. Consequently, rational re-design of natural
aptamer domains should be achievable through muta-
tions that reverse hydrogen bond directionality, disrupt-
ing interactions between the aptamer and the natural
ligand, whilst avoiding mutations that might adversely
affect the global structure of the aptamer domain. In turn,
a priori knowledge of the structure of the re-engineered
ligand-binding pocket, allows predictions to be made
about the types of ligand that are likely to bind, allowing
more rapid and effective screening against a smaller tar-
geted synthetic compound-library.
1
2
3
4
5
6
7
8
9
B
C
H
N
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
G11
O
H
H
H
+
O
N
N
N
N
H
H
N
O
C17
N
H
N
NH
N
N
H
N
H
N
NH2
NH
PreQ1
H
N
G5
N
O
H
N
U6
N
O
A30
N
Figure 1. Secondary structure of the PreQ1-I aptamer and H-
bond ligand contacts. A) The stems P1 (green) and P2 (blue)
are connected through three loop regions (L1-3). B) Stem P2
is formed through a tertiary interaction, which is stabilised
by ligand binding, resulting in a H-type pseudoknot. C) The
PreQ1 ligand (blue) H-bonds with one residue from each of
the three loops, U6 (L1), C17 (L2) and A30 (L3). This ligand-
binding core is stacked between the two stems of the ap-
tamer, with the aminomethyl group of PreQ1 pointing out
into the major groove where it interacts with G5 (P1) and the
phosphate of G11 (P2). (See figure S1 for more details)
nition.22,23 This canonical base pair interaction, along with
a trans sugar edge/Watson-Crick base pairing with A30,
and additional hydrogen bonding to G5 and U6, allows
for complete recognition of the PreQ1 ligand by the ri-
boswitch aptamer (Figure 1B,C). Based on these structural
insights, residues G5, U6 and C17, along with residue C18
which Watson-Crick pairs with G5 in the P1 stem, were
subjected to site-directed mutagenesis to alter the bind-
ing specificity of the aptamer towards new non-native
ligands. A30 was not considered for mutagenesis as it H-
bonds to the C2-amino group of the ligand. The C2-amino
group was shown to be essential for ligand binding by the
PreQ1 riboswitch,21 and would be required for H-bonding
to either C17 or U17. To avoid unnecessarily perturbing
riboswitch folding, whilst keeping the dimensions of the
ligand-binding pocket constant, the transversion of py-
rimidines to purines or vice versa was avoided, resulting
in a library of just seven mutants: the M1 (C17U), M2
(U6C) and M3 (G5A) single mutants; M4 (G5A, C18U) and
M5 (U6C, C17U) double mutants; M6 (G5A, C17U, C18U)
triple mutant; and M7 (G5A, U6C, C17U, C18U) quadruple
mutant.
In this report we demonstrate how a rational approach
can be used to re-engineer the PreQ1 class I riboswitch
from Bacillus subtilis.21-23 PreQ1 class I riboswitches, com-
monly found in Firmicutes and Proteobacteria,21,24 regu-
late genes required for the biosynthesis of the hypermodi-
fied nucleotide queuosine, present at the wobble position
of GUN tRNAs.21,25 The minimal PreQ1 aptamer domain is
the smallest known riboswitch aptamer at just 34 nucleo-
tides in length (Figure 1A), and is selective for the queu-
osine precursors 7-aminomethyl-7-deazaguanine (PreQ1)
and 7-cyano-7-deazaguanine (PreQ0).21 Ligand-binding
stabilises a H-type pseudoknot in the aptamer domain
(Figure 1B),22,23 which leads to the formation of a down-
stream transcriptional terminator stem required to re-
press gene expression (Figure S1).21 We introduced muta-
tions at three key sites in the ligand-binding pocket to
disrupt hydrogen-bonding to the natural queuosine pre-
cursors, and these candidate aptamers were evaluated in
vivo against a small rationally designed library of synthet-
ic compounds. From this approach, a functional mutant
riboswitch-ligand pairing was identified which was highly
effective in repressing gene expression in B. subtilis.
RESULTS AND DISCUSSION
Targeted mutagenesis of the PreQ1 riboswitch. The
PreQ1 class I riboswitch from B. subtilis was considered a
good candidate for re-engineering, as it is well character-
ised and analogues of the natural ligands are both syn-
thetically accessible and likely to be cell permeable. From
published structures of the PreQ1 aptamer, it is apparent
that Watson-Crick base pairing between the conserved
cytidine C17 residue and PreQ1 is essential in ligand recog-
It was anticipated that riboswitches possessing the
C17U mutation might recognise synthetic ‘diamino-faced’
compounds (Figure 2, 1-6). Similarly, U6C was envisaged
to potentially alter riboswitch specificity from the native
deazapurine ligands to furopyrimidines (3-6), whilst G5A
2
ACS Paragon Plus Environment

Page 3 of 9
Journal of the American Chemical Society
A
1
2
3
4
5
6
7
8
C
N
N
O
NH2
NH2
O
N
H2N
1
NH
5
7
N
N
Riboswitch
Mutations
N
H
N
9 N
NH2
N
NH2
N
NH2
H
H
3
DPQ0 (2)
PreQ0
1
WT
M1
M2
M3
M4
M5
M6
M7
C17U
O
O
U6C
NH2
N
H2N
NH2
NH2
NH2
HO
HO
O
G5A
G5A
9
C18U
N
N
N
U6C C17U
C17U C18U
G5A U6C C17U C18U
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
O
N
NH2
G5A
O
O
O
N
NH2
N
NH2
N
NH2
3
5
4
6
B
1000
750
500
250
0
D
400
300
200
100
0
DMSO
1
2
3
4
5
6
WT
M1
M2
M3
M4
M5
M6
M7
Riboswitch construct
Compounds
Figure 2. Screening PreQ1 riboswitch mutants for control of LacZ expression in B. subtilis. A) Schematic of the chromosomally-
integrated LacZ expression construct. The lacZ gene (blue box arrow) is transcribed from the B. subtilis queCDEF constitutive
promoter (black line arrow). A wild-type or mutant PreQ1 riboswitch (grey box) terminates transcription upon ligand binding.
Riboswitch mutations tested in this study are tabulated. B) β-galactosidase activity of wild-type and mutant riboswitch con-
structs assayed in the absence of exogenous ligand. C) Structures of synthetic compounds screened for activity with mutant ri-
boswitch constructs. D) β-galactosidase activity of the M1 (C17U) riboswitch construct assayed in the presence of 1 mM of the
compounds above. Data represent the mean of three repeats with error bars indicating standard deviation.
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
might permit interactions with compounds possessing
different C7 side chains (1, 4-6). Given that the G5A muta-
tion was predicted to disrupt formation of the P1 aptamer
stem, the complementary C18U was also included along-
side G5A (see Figure S1 for further information). The se-
lected mutations were introduced into a lacZ expression
vector, in which the promoter and entire 5´-UTR of the
queCDEF operon, encompassing the PreQ1 riboswitch
aptamer domain and expression platform, controlled ex-
pression of the lacZ gene (Figures 2A & S2A).21 The con-
structs were then integrated into the chromosome of B.
subtilis, to generate reporter strains enabling riboswitch-
ligand interactions to be monitored in vivo using a modi-
fied Miller assay.26
lost the ability to function as transcriptional terminating
elements. Subsequently, transformants were screened for
their ability to control LacZ expression in response to the
exogenous addition of PreQ0. The LacZ activity of all mu-
tant riboswitch transformants was unaffected by the addi-
tion of exogenous PreQ0 (at 2.5 mM), demonstrating that
these mutants no longer respond to PreQ0 (Figure S3A).
The Breaker group previously reported structural modula-
tion of a C17U mutant of the PreQ1 riboswitch in the pres-
ence of 2,6-diaminopurine (DAP).21 Therefore, we also
tested DAP in our assay, but no significant repression of
LacZ activity was observed for any of our mutant PreQ1
riboswitches in the presence of 2.5 mM DAP (Figure S3B).
This demonstrates that whilst DAP may bind to the M1
(C17U) aptamer in vitro, it is not capable of controlling
gene expression in vivo through the M1 riboswitch, and so
is not a suitable orthogonal riboswitch ligand.
In vivo functionality of PreQ1 riboswitch mutants.
Levels of β-galactosidase (LacZ) activity were determined
initially in the absence of ligand (Figure 2B). The strain
possessing the wild-type riboswitch produced very low
basal levels of LacZ activity, suggesting that cellular levels
of PreQ1 and PreQ0 were sufficient to bind the wild-type
riboswitch and repress LacZ expression; this demon-
strates how the parent riboswitch is of little utility as a
regulatory tool in cells that produce the natural ligands.
In contrast, all the mutant riboswitch strains exhibited
elevated levels of LacZ activity; presumably because they
no longer bind the native ligands, or because they have
The mutant strains fell into two distinct groups: those
that exhibited high levels of LacZ expression in the ab-
sence of ligand (M1, M2, M5 and M7), and those that ex-
hibited moderate levels (M3, M4 and M6). It was noted
that the moderate LacZ expressors all contained the G5A
mutation, which is situated underneath the ligand-
binding pocket at the top of the P1 stem. It is possible that
this mutation leads to alternative folding of the ri-
boswitch, interfering with downstream transcription (re-
3
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 9
ducing LacZ expression). The compensatory C18U muta- tures stabilise the ligand-bound pseudoknot and/or alter
1
2
3
4
5
6
7
8
tion was designed to restore P1 stem integrity by allowing
base-pairing to the G5A mutation; however, mutants M4
and M6 produced only intermediate levels of LacZ despite
having both mutations present. These data suggest that
the G5A mutation is unsuitable for the development of
orthogonally selective PreQ1 riboswitches. In light of this,
the single mutant M1 (C17U) and double mutant M5
(U6C, C17U) were considered the best candidates for the
creation of orthogonally selective riboswitches, and were
therefore prioritised for further study.
kinetics of transcription to allow the riboswitch to func-
tion. Surprisingly, 3 was found to repress LacZ expression
under the control of the M1 riboswitch to an even greater
extent (59 % at 2 mM), despite a predicted lone pair clash
between U6 and O9 of the ligand 3. These observations
suggest that there may be some flexibility in the position-
ing of U6 to accommodate mismatches in H-bonding po-
tential which, when combined with the structural disrup-
tion resulting from mutation of this residue, indicate that
U6 is not a suitable target residue for creating robustly
functioning orthogonal riboswitch mutants.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In vivo evaluation of candidate ligands. Six potential
ligand compounds 1-6 (Figures 2C) were designed and
synthesised, focusing on 'diamino-faced' analogues of
PreQ1, to target the C17U mutation present in both the M1
and M5 riboswitches. In addition, four of these com-
pounds were synthesised with an oxygen at position 9, to
accommodate the U6C mutation in M5. This targeted
compound library was screened for the ability to repress
LacZ activity controlled by the mutant riboswitches M1
(Figure 2D) and M5 (Figure S4A). This strategy identified
a functional riboswitch-ligand pairing between the M1
mutant and ligand 2, the diamino analogue of PreQ0
(DPQ0). The M1 riboswitch exhibited selectivity, mediat-
ing in vivo repression only in the presence of DPQ0 (2),
despite all library compounds having a 'diamino-face'.
Furthermore, when tested against other riboswitch mu-
tants that contain the C17U mutation (M1, M5, M6 and
M7), DPQ0 was active only against M1 (Figure S3C). The
M1-DPQ0 pairing was capable of dose-dependent control
of lacZ expression, with an IC50 value of 498 µM, and
~90% repression observed at DPQ0 concentrations of 2
mM or higher (Figure 3A). The proposed binding model
of M1 bound to DPQ0 (Figure 3B) shows how the C17U
mutation alters the base pairing complementarity from
the ‘guanine-face’ of PreQ0 to the ‘diamino-face’ of its
analogue DPQ0. Since PreQ0 cannot repress the M1 ri-
boswitch (Figure S3A), this mutant riboswitch is orthogo-
nal with respect to the wild-type system in terms of in
vivo functionality. Growth rate studies (Figure S5) also
reveal that DPQ0 is not toxic to B. subtilis at concentra-
tions required to modulate M1 riboswitch activity.
A
[DPQ0]'(mM)'
'
O
N
B
H
H
N
N
N
N
H
O
C17U
H
N
N
N
H
DPQ0
H
H
N
O
N
N
H
NH
A30
N
O
N
U6
N
Figure 3. Dose-dependent repression of gene expression
using the orthogonal M1 riboswitch. A) B. subtilis cells with
lacZ under transcriptional control of the M1 riboswitch,
where assayed for β-galactosidase activity with DPQ0 (6.25
µM - 4 mM). Data represent the mean of three repeats with
error bars indicating standard deviation. The data were fit
with a four-parameter logistic function to derive an IC50 of
498 µM. B) Proposed H-bonding model for DPQ0 (blue)
binding to the M1 riboswitch. The mutated C17U residue is
shown in green. H-bonds are depicted as red dashed lines.
The U6C mutation was rationalised to switch specifici-
ty of the riboswitch, from deazapurines, which possess an
N9 hydrogen-bond donor, to synthetic furopyrimidines
with an O9 H-bond acceptor. Accordingly, several ‘dia-
mino-faced’ ligands were synthesised with an oxygen at
position 9 (3-6); however, none were capable of repress-
ing the M5 (U6C, C17U) riboswitch under the conditions
of our LacZ assay (Figure S4A). The U6 residue also plays
a role in PreQ1 riboswitch folding, as it base pairs with the
Hoogsteen edge of A29.22,23 This base pair will be lost with
the U6C mutation, which may disrupt aptamer folding
and ligand-binding kinetics. We therefore investigated
whether the M5 riboswitch could repress LacZ expression
when cells were grown at 16 °C, rather than at 37 °C (Fig-
ure S4B). Under these conditions ligand 3 was capable of
modestly repressing LacZ expression through the M5 ri-
boswitch (28 % at 2 mM), suggesting that lower tempera-
In vitro characterisation of DPQ0 binding to the or-
thogonal M1 riboswitch aptamer. PreQ0 is converted
directly into PreQ1 by the NADPH-dependent enzyme
QueF,27 which reduces the nitrile group to an aminome-
thyl moiety. Since DPQ0 is a ‘diamino-faced’ analogue of
PreQ0, the possibility that QueF might reduce the nitrile
group of DPQ0 in vivo was investigated. The nitrile reduc-
tase QueF from B. subtilis was overproduced in Escherich-
ia coli by standard protocols. Whilst recombinant QueF
was shown to reduce PreQ0 in vitro (Figure S6) it did not
accept DPQ0 as an alternative substrate. This suggests
4
ACS Paragon Plus Environment

Page 5 of 9
Journal of the American Chemical Society
that DPQ0, and not its reduced aminomethyl derivative, ligand complex as a result of the mutation of a key resi-
1
2
3
4
5
6
7
8
acts directly on the M1 riboswitch to control gene expres-
sion in Bacillus.
due (C17U) and its binding to a synthetic ligand analogue.
Importantly, in vitro orthogonality was observed between
the two aptamer-ligand pairings, as PreQ0 was shown to
have no affinity for the mutant M1 aptamer, and similarly
DPQ0 had no affinity for the wild-type PreQ1 aptamer.
To further confirm that DPQ0 is the in vivo ligand for
the M1 riboswitch, we used isothermal titration calorime-
try (ITC) to study the binding of DPQ0 and PreQ0 to in
vitro transcribed M1 and wild-type aptamer domains. The
native PreQ0 ligand was found to bind to the wild-type 34
nt minimal aptamer domain with an apparent equilibri-
um dissociation constant (KD) of 0.57 ± 0.02 µM (Figure
4A,C). This value was larger than the 0.1 µM KD reported
from in-line probing analysis of PreQ0 binding to a 52 nt
fragment of the B. subtilis PreQ1 riboswitch.21 However,
this fragment included an additional stem loop structure
(P0) at the 5′-end of the minimal aptamer domain, which
was reported to increase affinity of the aptamer for the
natural ligand.21 Therefore, the apparent KD we report
here for PreQ0 binding to the minimal 34 nt aptamer is
consistent with the modest decrease in binding affinity
observed when the P0 stem is absent. The synthetic lig-
and DPQ0 was found to bind the mutant M1 aptamer with
an apparent KD of 14.6 ± 1.46 µM (Figure 4B,C), 26-fold
higher than that of the wild-type aptamer-PreQo pairing.
This reduction in binding affinity might be attributed to
perturbation of the structure or folding of the aptamer-
It was apparent from the ITC analysis that the predict-
ed stoichiometry of PreQ0 binding to the wild-type ap-
tamer was too low (Figure 4C). An n value of 0.27 suggests
that only around one quarter of the aptamer sample was
ligand-binding competent. We suspected that this was a
consequence of a previously reported PreQ1 aptamer di-
merization artefact.28,29 We prepared aptamer samples
incorporating the C12U mutation, to prevent dimerization
without disrupting important interactions in the aptamer-
ligand complex.22,23 Whilst the C12U mutation prevented
aptamer dimerisation, it reduced the affinity of the ‘wild-
type’ aptamer for PreQ0 (KD of 1.92 ± 0.08 µM), and led to
a lack of detectable binding between DPQ0 and the C12U
mutant of M1 (Figure S7). It should be noted that dimeri-
sation of aptamers affects only the n parameter during
ITC analysis, all other parameter calculations are depend-
ent solely on an accurate determination of the ligand con-
centration in the syringe.30 Therefore, apparent KD esti-
mates presented here are nonetheless reliable (Figure 4C).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Regulation of MreB expression demonstrates M1 util-
ity. MreB is an actin homolog which plays a crucial role in
maintaining the distinctive rod-shaped morphology of B.
subtilis and many other bacterial species, and is a promis-
ing target for antimicrobial drug screening.31 To demon-
strate the potential applications of our M1-DPQ0 ri-
boswitch-ligand pairing in controlling native gene expres-
sion, we created a strain of B. subtilis in which the expres-
sion of MreB was placed under the conditional control of
the full-length M1 riboswitch (Figure 5A).
A
B
Time!(Min)!
Time!(Min)!
Wild-type B. subtilis, which can freely express MreB,
adopts a rod-shaped morphology (Figure 5B-I).32 An mreB
deletion strain of B. subtilis (ΔmreB) exhibits compro-
mised growth, with cells round and swollen in shape
(Figure 5B-II). A linear M1-mreB construct was chromo-
somally integrated into ΔmreB cells at the amyE locus, to
produce the repressible M1-mreB strain. In the absence of
ligand, MreB is expressed by the M1-mreB strain to restore
the rod-shaped morphology (Figure 5B-III), demonstrat-
ing that the integrated construct successfully comple-
ments the ΔmreB phenotype. Following supplementation
of the cell media with DPQ0, M1 repressed expression of
MreB in the M1-mreB strain (Figure 5B-IV). In vivo or-
thogonality was demonstrated, as the same concentration
of PreQ0 had no morphological effect on M1-mreB cells
(Figure 5B-VI). Similarly the morphology of the wild-type
strain was unaffected by the addition of DPQ0 (Figure 5B-
V), confirming that the effects of this ligand on M1-mreB
cells are mediated through the M1 riboswitch.
PreQ0&vs.&WT&&&
PreQ0&vs.&M1&&&
DPQ0!vs.&WT!
DPQ0!vs.&M1!
Molar!Ra/o!
Molar!Ra/o!
C
aptamer
ligand
n
ΔG
ΔH
-TΔS
KD
(µM)
(Kcal/mol) (Kcal/mol) (Kcal/mol)
WT
M1
PreQ0
DPQ0
0.27
1.08
-8.52
-6.60
-15.36
-6.98
6.84
0.38
0.57
14.63
Figure 4. ITC analysis of aptamer−ligand interactions. A)
Upper panel: wild-type PreQ1 aptamer titrated with PreQ0.
Lower panel: binding isotherms for 18.7 µM of wild-type
(black circles) or mutant M1 (white diamonds) aptamers ti-
trated with 150 µM of PreQ0. B) Upper panel: mutant M1 ap-
tamer titrated with DPQ0. Lower panel: binding isotherms
for 18.7 µM of mutant M1 (black diamonds) or wild-type
(white circles) aptamers titrated with 225 µM of DPQ0. Ex-
periments were conducted at 25 °C in 50 mM HEPES (pH 7.5)
buffer, containing 100 mM KCl and 20 mM MgCl2. C) Curves
were fitted in Origin, using the supplied software (MicroCal),
to calculate the thermodynamic parameters presented.
Measurement of B. subtilis cell dimensions (Figure S9)
revealed that whilst the addition DPQ0 had a significant
effect (P <0.001) on the M1-mreB strain, reducing the
mean length/width ratio of cells from 3.64 to 1.74, it had
5
ACS Paragon Plus Environment

Journal of the American Chemical Society
little effect on wild-type cells (5.39 and 5.19 in the absence
Page 6 of 9
volved extensive in vitro selection4,17 or laborious screen-
ing procedures.18,20 In this manuscript we demonstrate
how a more rational targeted approach can be used to
streamline the process of creating orthogonal riboswitch
tools.
1
2
3
4
5
6
7
8
and presence of DPQ0 respectively). These data indicate
that DPQ0 affects the morphology of M1-mreB cells by
repressing mreB expression through its specific interac-
tion with the M1 riboswitch. Furthermore, western blot-
ting revealed that MreB protein levels were significantly
reduced in the M1-mreB strain following incubation with
DPQ0 (Figure S10). These studies shows how the orthogo-
nal M1-DPQ0 riboswitch-ligand pairing can be used to
regulate the expression of native genes in B. subtilis, to
complement knockout phenotypes, for gene functional
analysis and other purposes.
Based on previous biochemical and structural stud-
ies,21-23 just seven variants of the PreQ1-I riboswitch and
six synthetic ligand candidates were designed and tested
for their ability to repress gene expression. A single mu-
tant (C17U) M1, which does not respond to the natural
ligand PreQ0, was shown to provide effective dose de-
pendant repression of lacZ expression in B. subtilis upon
addition of a diamino analogue DPQ0. ITC experiments
revealed that M1 binds DPQ0, but has no affinity for
PreQ0, and conversely that the wild-type aptamer binds
PreQ0, but not DPQ0. Interestingly, 2,6-diaminopurine
(DAP), which had previously been shown to bind to M1 in
vitro,21 was unable to modulate M1-lacZ expression in vivo.
The in vivo selectivity for DPQ0 over DAP may be due to
differences in the rates of ligand-riboswitch association.21
Similar kinetic selectivity is suggested to account for how
the PreQ1 riboswitch discriminates between PreQ1 and
guanine in vivo, despite binding both ligands in vitro.21 In
addition, M1 and a double mutant (U6C, C17U) M5, de-
signed to recognise furopyrimidine analogues, afforded
repression of lacZ expression with the furopyrimidine 3,
but only when cells were grown at lower temperature (16
°C). The riboswitch-furopyrimidine complexes may be
unstable at higher temperatures, or alternatively the ki-
netics of riboswitch ligand-binding and transcription may
be less favourable.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
A
B
To demonstrate the general applicability of the or-
thogonal M1-DPQ0 riboswitch, we created conditional
mutants of the bacterial actin homolog MreB in B. sub-
tilis, enabling cell morphology to be controlled through
the exogenous supply of DPQ0. MreB is essential for the
normal growth of rod-shaped bacteria and has been iden-
Figure 5. Controlling cell morphology using the orthogonal
M1 riboswitch. A) The MreB expression construct. The mreB
gene (green arrow) is transcribed from the B. subtilis
queCDEF constitutive promoter (black arrow). The M1 ri-
boswitch (grey box) terminates transcription upon binding
DPQ0. B) Conditional gene repression affects B. subtilis
morphology. The MreB expression construct was chromoso-
mally-integrated into an mreB knockout strain of B. subtilis
(ΔmreB), to create the repressible M1-mreB strain. Cells were
imaged at 60× magnification: (I) wild-type; (II) ΔmreB cells;
(III) M1-mreB cells; (IV) M1-mreB cells with 2 mM DPQ0; (V)
wild-type cells with 2 mM DPQ0; (VI) M1-mreB cells with 2
mM PreQ0. For further images see Figure S8 in Supporting
Information.
tified as
a
potential antibiotic target.31 The dose-
dependent control of mreB expression levels could there-
fore provide a useful tool in creating sensitive strains for
antibiotic screening. Establishing tunable conditional
mutants of essential genes in pathogenic strains is im-
portant in antimicrobial target validation and whole cell
screening;33-35 by lowering the levels of essential gene
products in the cell it is possible to develop more sensi-
tive screens for compounds that selectively inhibit their
function. Many important pathogenic and antibiotic re-
sistant bacteria contain native PreQ1 riboswitches, includ-
ing Firmicutes (e.g. Staphylococcus, Clostridium, Listeria,
Enterococcus) and Proteobacteria (e.g. Campylobacter,
Salmonella, Neisseria).24 M1 variants of the native PreQ1
riboswitches might therefore be deployed as host-
optimised expression tools for antibiotic screening and
target validation purposes in these pathogenic strains.
Such tools would be timely given the problems of antimi-
crobial resistance and the lack of alternative tunable ex-
pression systems for interrogating many microbial patho-
gens.
CONCLUSION
Natural riboswitch aptamer domains and expression plat-
forms represent a diverse toolbox of modular gene regula-
tory components. However, if this resource is to be fully
exploited in the creation of new genetic circuits, there is a
need to develop riboswitch aptamers which are orthogo-
nal to one another and which respond to synthetic or
non-native ligands. Previous efforts to develop synthetic
aptamers or to re-engineer natural aptamers have in-
6
ACS Paragon Plus Environment

Page 7 of 9
Journal of the American Chemical Society
ton X-100, for 4 hr at 37 °C. Transcription reactions were
MATERIALS & METHODS
1
resolved by 12% PAGE under denaturing conditions (8 M
urea), with bands excised and RNA recovered by the
‘crush-and-soak’ method. Aptamers were then dialysed
overnight at 4 °C into ITC buffer: 50 mM HEPES (pH 7.5),
100 mM KCl, 20 mM MgCl2. Ligand solutions were made
up in the same buffer following dialysis. Samples were
degassed and ITC experiments were performed at 25 °C
using a VP-ITC microcalorimeter (Microcal Inc). Experi-
ments were conducted with the RNA aptamer sample in
the cell, titrated with one 2 µl injection then twenty-four
12 µl injections of ligand solution.
Ligand synthesis. Pyrrolo[2,3-d]pyrimidines (PreQ0, 1 &
2) and furo[2,3-d]pyrimidines (3-6) were prepared in one
or two synthetic steps from the inexpensive precursors
2,4,6-triaminopyrimidine or 2,4-diamino-6-hydroxypyr-
imidine (Sigma Aldrich) following established procedures
(see Supporting Information for full details).
2
3
4
5
6
7
8
9
Construction of the riboswitch-controlled LacZ ex-
pression constructs. The pDG1661 vector36 containing
the native promoter and the entire PreQ1 riboswitch, am-
plified from the upstream region of B. subtilis queCDEF
operon (Figure S2A), was kindly supplied by the Breaker
lab.21 To generate riboswitch mutants, two flanking mas-
ter primers (MC4 and MC5), and two internal mutagenic
primers (MC6 to MC19), were designed to amplify 5′- and
3′-fragments of the PreQ1 riboswitch (Table S1). Overlap-
ping PCR was then employed to join both fragments to-
gether, generating the complete mutant PreQ1 riboswitch
sequences, which were cloned into the EcoRI and BamHI
sites of pDG1661, upstream of the lacZ reporter gene. The
resulting plasmids were verified by sequencing and trans-
formed into B. subtilis 1A40, where they integrated into
the amyE locus of the chromosome by homologous re-
combination.21
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Construction of the M1 riboswitch-controlled mreB
expression construct. The mreB gene was amplified
from B. subtilis 168 by primers MC20 and MC21 (Table S1),
and cloned into the EcoRI and BamHI sites of pDG1662. A
sequence comprised of the amyE-front fragment and
mreB gene was amplified using the primers MC22, MC23,
MC24 and MC25. Separately, another sequence comprised
of the M1-riboswitch mutant, the cat (chloramphenicol
resistance) gene and the amyE-back fragment was ampli-
fied from the corresponding LacZ expression construct in
pDG1661 (see above) using the primers MC26, MC27,
MC28 and MC29. The two sequences were spliced togeth-
er through an identical 50 bp linker region by overlapping
PCR, to generate a linear construct with flanking amyE-
front and amyE-back sequences for homologous recombi-
nation (Figure S2B). This linear DNA construct was trans-
formed into B. subtilis 3725 cells and integrated into the
amyE locus. Cell Morphology (MreB Expression) assays
were carried out as described previously.20 (also see sup-
porting information for further details).
β-galactosidase (LacZ Expression) Assays. A high-
throughput microplate method for quantifying β-
galactosidase activity in B. subtilis was modified from that
described in Lynch et al.26 Overnight cultures were dilut-
ed 50-fold into 190 μl Nutrient Broth, containing 5 μg/ml
chloramphenicol and either 10 μl of ligand in DMSO or
with DMSO alone, then left to grow in a 96-well plate at
37 °C. A microplate reader (Anthos Zenyth 3100) was used
to monitor OD595 until it reached the 0.085-0.14 range
(equivalent to an OD595 of 0.3-0.5 in a 1 cm pathlength
cuvette), cells were then lysed by the addition of 20 µl
PopCulture reagent (Novagen), containing 40 U/ml lyso-
zyme. Lysed cultures were diluted 10-fold into 150 µl of Z
buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1
mM MgSO4, 50 mM β-mercaptoethanol, pH 7.0). The β-
galactosidase reaction was started by adding 30 μl of O-
nitrophenyl-β-D-galactoside (4 mg/ml in Z buffer), left at
30 °C until a yellow colour developed, then quenched by
adding 75 μl of 1 M Na2CO3. The length of time (Time)
between substrate addition and quenching, and the OD405
was recorded. The Miller units were calculated using the
ASSOCIATED CONTENT
Supporting Information. Further experimental infor-
mation, details of ligand synthesis, additional data, support-
ing figures (Figures S1−S10) and a table (Table S1) are pre-
sented. This material is available free of charge via the Inter-
net at http://pubs.acs.org.
AUTHOR INFORMATION
* Corresponding Author
jason.micklefield@manchester.ac.uk.
‡ These authors contributed equally.
following formula: Miller units = 1000 x {OD405 / (OD595
Time x [volume of cell lysate/ total volume])}
x
Present addresses: Institute of Molecular Biology, Academia
Sinica, Taiwan (M.-C.W); School of Chemistry, University of
St Andrews (P.T.L); Faculty of Life Sciences, University of
Manchester (N.D.).
Preparation of RNA aptamers and Isothermal Titra-
tion Calorimetry. Riboswitch aptamers were prepared
by in vitro transcription following a protocol modified
from Gurevitch.37 Briefly, dsDNA templates (comprising
the aptamer sequence downstream of the T7 RNA poly-
merase promoter) were created by annealing complemen-
tary primers purchased from Sigma Aldrich (Table S1).
Transcription reactions were performed by incubating T7
RNA polymerase with dsDNA templates and 6 mM each
of ATP, UTP, GTP, CTP, in 30 mM Tris buffer (pH 8.0), 28
mM MgCl2, 20 mM DTT, 3.4 mM Spermidine, 0.01% Tri-
ACKNOWLEDGMENT
This work was supported by the BBSRC grant BB/I012648/1,
the Manchester Centre for Synthetic Biology (SynBioChem)
BB/M017702/1, and a BBSRC PhD studentship (for J.L.). The
EPSRC provided a PhD studentship (for P.T.L.). We thank
Prof. Jeffery Errington for providing the B. subtilis 3725
strain, Prof. Ronald R. Breaker for the vector pDG1661 and
Dr. Majid Al Nakeeb for technical support.
7
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 9
(20) Robinson, C. J.; Vincent, H. A.; Wu, M.-C.; Lowe, P. T.;
Dunstan, M. S.; Leys, D.; Micklefield, J. J. Am. Chem. Soc. 2014,
136, 10615–10624.
(21) Roth, A.; Winkler, W. C.; Regulski, E. E.; Lee, B. W.; Lim.
J.; Jona, I.; Barrick, J. E.; Ritwik, A.; Kim, J. N.; Welz, R.; Iwata-
Reuyl, D.; Breaker, R. R. Nat. Struct. Mol. Biol. 2007, 14, 308-317.
(22) Klein, D. J., Edwards, T. E.; Ferré-D'Amaré A. R. Nat.
Struct. Mol. Biol. 2009, 16, 343-344.
REFERENCES
1
2
3
4
5
6
7
8
(1) Serganov, A.; Nudler, E. Cell 2013, 152, 17-24.
(2) Breaker, R. R. Cold Spring Harbor Perspect. Biol. 2012, 4,
No. a003566.
(3) Garst, A. D.; Edwards, A. L. Batey, R. T. Cold Spring Harbor
Perspect. Biol. 2011, 3, No. a03533.
(4) Link, K. H.; Breaker, R. R. Gene Ther. 2009, 16, 1189-1201.
(5) Chappell, J.; Takahashi, M. K.; Meyer, S.; Loughrey, D.;
Watters, K. E.; Lucks, J. B. Biotechnol. J. 2013, 8, 1379-1395.
(6) Vazquez-Anderson, J.; Contreras, L. M. RNA Biol. 2013, 10,
1778-1797.
(23) Kang, M.; Peterson, R.; Feigon, J. Mol Cell. 2009, 33, 784-
790.
(24) McCown, P. J.; Liang, J. J.; Weinberg, Z.; Breaker, R. R.
9
Chem. Biol. 2014, 21, 880-889.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(25) Iwata-Reuyl, D. Bioorg. Chem. 2003, 31, 24-43.
(26) Lynch, S. A.; Desai, S. K.; Sajja, H. K.; Gallivan, J. P. Chem.
Biol. 2007, 14, 173-184.
(27) Reader, J. S.; Metzgar, D.; Schimmel, P.; de Crécy-Lagard,
V. J. Biol. Chem. 2004, 279, 6280-6285.
(28) Rieder, U.; Kreutz, C.; Micura R. Proc. Natl. Acad. Sci.
USA 2010, 107, 10804–10809.
(29) Eichhorn, C.D.; Feng, J.; Suddala, K.C.; Walter, N. G. Nu-
(7) Berens, C.; Groher, F.; Suess, B. Biotechnol. J. 2015, 10, 246-
257.
(8) Berens, C.; Suess, B. Curr. Opin. Biotechnol. 2015, 31, 10-15.
(9) Cho, B. K.; Federowicz, S. A.; Embree, M.; Park, Y. S.; Kim,
D.; Palsson, B.Ø. Nucleic Acids Res. 2011, 39, 6456-6464.
(10) Weigand, J. E.; Sanchez, M.; Gunnesch, E. B.; Zeiher, S.;
Schroeder, R.; Suess, B. RNA 2008, 14, 89-97.
(11) Sharma, V.; Nomura, Y.; Yokobayashi, Y. J. Am. Chem. Soc.
2008, 130, 16310-16315.
(12) Fowler, C. C.; Brown, E. D.; Li, Y. Chembiochem 2008, 9,
1906-1911.
(13) Lynch, S. A; Gallivan, J. P. Nucleic Acids Res. 2009, 37,
184–192.
(14) Ceres, P.; Garst, A. D.; Marcano-Velázquez, J. G.; Batey, R.
T. ACS Synth. Biol. 2013, 2, 463-472.
(15) Ceres, P.; Trausch, J. J.; Batey, R. T. Nucleic Acids Res.
2013, 41, 10449-10461.
(16) Groher F.; Suess, B. Biochim. Biophys. Acta 2014, 1839,
964–973
(17) Wittmann, A.; Suess B. FEBS Letters. 2012, 586, 2076–
2083.
(18) Dixon, N.; Duncan, J. N.; Geerlings, T.; Dunstan, M. S.;
McCarthy, J. E.; Leys, D.; Micklefield, J. Proc. Natl. Acad. Sci. USA
2010, 107, 2830 – 2835.
(19) Dixon, N.; Robinson, C. J.; Geerlings, T.; Duncan, J. N.;
Drummond, S. P.; Micklefield, J. Angew. Chem. Int. Ed. 2012, 51,
3620-3624.
cleic Acids Res. 2011, 40, 1345-1355.
(30) Tellinghuisen, J. Anal. Biochem. 2011, 414, 297–299.
(31) Bean, G. J.; Flickinger, S. T.; Westler, W. M.; McCully, M.
E.; Sept, D.; Weibel, D. B.; Amann, K. J. Biochemistry, 2009, 48,
4852-4857.
(32) Formstone, A.; Errington J. Mol Microbiol 2005, 55, 1646-
1657.
(33) Abrahams, G. L.; Kumar, A.; Savvi, S.; Hung, A. W.; Wen,
S.; Abell, C.; Barry, C. E., III; Sherman, D. R.; Boshoff, H. I.; Miz-
rahi, V. Chem. Biol. 2012, 19, 844-854.
(34) Singh, S. B.; Phillips, J. W.; Wang, J. Curr. Opin. Drug Dis-
covery Dev. 2007, 10, 160-166.
(35) Payne, D. J.; Gwynn, M. N.; Holmes D. J.; Pompliano, D. L.
Nat. Rev. Drug Discov. 2007, 6, 29-40.
(36) Guérot-Fleury, A. M.; Frandsen, N.; Straiger, P. Gene 1996,
180, 57-61
(37) Gurevitch, V.V. Methods Enzymol. 1996, 275, 382-397.
8
ACS Paragon Plus Environment

Page 9 of 9
Journal of the American Chemical Society
Table of Contents artwork
1
2
3
4
5
6
7
8
9
3'
PreQ1-M1
queC
lacZ
P
A
A
U
C
C
U
A
G
N
NH2
N
DPQ0%
U
30
A
N
H
U
C
N
NH2
A
A
A
A
A
A
U
A
U
6
U
C
G
C17
C
A
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
G
A
G
A
U
C
U
5'
[DPQ0]%(mM)%
%
9
ACS Paragon Plus Environment