Chemical synthesis, purification, and characterization of 3′-5′-linked canonical cyclic dinucleotides (CDNs)

Article abstract of DOI:10.1016/bs.mie.2019.04.022

So far, four cyclic dinucleotides (CDNs) have been discovered as important second messengers in nature, where three canonical CDNs of c-di-GMP, c-di-AMP and c-AMP-GMP were found in bacteria containing two 3′-5′ phosphodiester linkages and one non-canonica

Full text of DOI:10.1016/bs.mie.2019.04.022

ARTICLE IN PRESS
Chemical synthesis, purification,
and characterization of
30-50-linked canonical cyclic
dinucleotides (CDNs)
Changhao Wanga,*, Min Haoa, Qianqian Qia, Yashao Chena,
b,c,
€
Jorg S. Hartig
*
aKey Laboratory of Applied Surface and Colloid Chemistry, Ministry of Education, School of Chemistry
and Chemical Engineering, Shaanxi Normal University, Xi’an, China
bDepartment of Chemistry, University of Konstanz, Konstanz, Germany
cKonstanz Research School Chemical Biology (KoRS-CB), University of Konstanz, Konstanz, Germany
*Corresponding authors: e-mail address: changhaowang@snnu.edu.cn; joerg.hartig@uni-konstanz.de
Contents
1. Introduction
2. Synthesis of crude CDNs
2.1 Chemicals
2
6
6
2.2 Equipment
7
2.3 Synthesis procedure for crude c-AMP-CMP
3. Purification of crude CDNs
3.1 Preparation of HPLC buffers
3.2 Purification protocol for crude c-AMP-CMP
3.3 HPLC purification of crude c-AMP-CMP
4. Characterization of CDNs
4.1 Materials and equipment
4.2 MS analysis of c-AMP-CMP fractions
4.3 NMR analysis of c-AMP-CMP
5. Summary and prospects
Acknowledgments
7
10
10
10
12
12
12
13
14
17
17
17
References
Abstract
So far, four cyclic dinucleotides (CDNs) have been discovered as important second mes-
sengers in nature, where three canonical CDNs of c-di-GMP, c-di-AMP and c-AMP-GMP
were found in bacteria containing two 30-50 phosphodiester linkages and one non-
canonical CDN 2030-c-GMP-AMP was identified in mammals containing mixed 20-50
and 30-50 phosphodiester linkages. The CDNs are produced by specific cyclases and
degraded by phosphodiesterases (PDEs). All of the known CDNs could bind to the
#
Methods in Enzymology
ISSN 0076-6879
https://doi.org/10.1016/bs.mie.2019.04.022
2019 Elsevier Inc.
All rights reserved.
1

ARTICLE IN PRESS
2
Changhao Wang et al.
stimulator of interferon genes (STING) to induce type I interferon (IFN) responses and
the three bacterial CDNs are sensed by specific riboswitches to regulate gene expres-
sion. The emerging physiological functions of bacterial CDNs lead the motivation to
investigate other possible canonical CDNs. In recent years, many endeavors have been
devoted to develop fast, convenient and cheap strategies for chemically synthesizing
CDNs and their analogues. The phosphoramidite approach using commercial starting
materials has attracted much attention. Herein, we describe an adapted one-pot strat-
egy that enables fast synthesis of crude 30-50-linked canonical CDNs followed by puri-
fication of the obtained CDNs using reversed phase high-performance of liquid
chromatography (HPLC). Furthermore, we report the full characterization of CDNs by
mass spectrometry (MS) and nuclear magnetic resonance (NMR) techniques.
1. Introduction
Cyclic dinucleotides (CDNs) are natural cyclic RNA molecules bearing
a 12-membered ring scaffold and two ribonucleotides. To date, four CDNs
have been discovered in nature (Fig. 1), including three canonical CDNs of
bis-(3050)-cyclic dimeric guanosine monophosphate (c-di-GMP), bis-(3050)-
cyclic dimeric adenosine monophosphate (c-di-AMP) and bis-(3050)-cyclic
adenosine monophosphate-guanosine monophosphate (c-AMP-GMP), and
Fig. 1 The chemical structures of naturally occurring canonical CDNs with two 30-50 pho-
sphodiester linkages and non-canonical CDN with mixed 30-50 and 20-50 phosphodiester
linkages.

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
3
one non-canonical CDN of (2050),(3050)-cyclic guanosine monophosphate-
adenosine monophosphate (20,30-c-GMP-AMP). All of the four CDNs have
been identified as second messenger signaling molecules in regulation of many
cellular processes.
In 1987, c-di-GMP was initially identified as an allosteric activator of a
bacterial cellulose synthase in Gluconacetobacter xylinus (Ross et al., 1987).
C-di-GMP is naturally produced from two guanosine triphosphate (GTP)
molecules by diguanylate cyclases (DGCs) containing GGDEF domain,
while it could be degraded into 50-phosphoguanylyl-(3050)-guanosine (50-
pGpG) by phosphodiesterases (PDEs) containing EAL or HD-GYP domain
and 50-pGpG is further decomposed into two guanosine monophosphate
(GMP) molecules by PDE containing HD-GYP domain. In the past three
decades, c-di-GMP has been demonstrated as a regulator in many physio-
logical processes such as cell-cell communication, motility, biofilm forma-
tion, biosynthesis of exopolysaccharide, sessility and virulence (Hengge,
2009). In 2008, c-di-AMP was discovered (Witte, Hartung, Buttner, &
Hopfner, 2008) and 2 years later its bacterial production was found in Listeria
monocytogenes (Woodward, Iavarone, & Portnoy, 2010). It has been identi-
fied as a crucial intracellular signaling molecule for the regulation of cell wall
metabolism, DNA damage, osmotic stress response and bacterial sporulation
(Corrigan & Gruendling, 2013). DNA integrity scanning protein A (DisA)
that possessing diadenylate cyclase (DAC) activity can convert two adeno-
sine triphosphate (ATP) molecules to c-di-AMP, and the degradation of
c-di-AMP into 50-phosphoadenylyl-(3050)-adenosine (50-pApA) can be
conducted by PDEs containing DHH-DHHA or HD domain. In 2012,
the third bacterial CDN c-AMP-GMP was discovered in Vibrio cholerae
(Davies, Bogard, Young, & Mekalanos, 2012). C-AMP-GMP is synthesized
from ATP and GTP by a dinucleotide cyclase (DncV) from Vibrio cholerae,
while it could be degraded into an intermediate of 50-phosphoadenylyl-
(3050)-guanosine (50-pApG) by specific cyclic GMP-AMP PDEs (V-cGAP1,
V-cGAP2 or V-cGAP3) and further split into 50-adenylyl-(3050)-guanosine
(50-ApG) by V-cGAP1 (Gao et al., 2015). C-AMP-GMP has been demon-
strated to be a crucial signaling molecule for regulating bacterial virulence,
promoting intestinal colonization, downregulating chemotaxis and control-
ling exoelectrogenic process (Davies et al., 2012; Nelson et al., 2015).
In 2013, an isomer of c-AMP-GMP (20,30-c-GMP-AMP) with mixed
20-50 and 30-50 phosphodiester linkages was identified as a novel second
messenger in metazoan (Gao et al., 2013). 20,30-c-GMP-AMP is cyclized
from ATP and GTP by a cyclic GMP-AMP synthase (cGAS) upon binding

ARTICLE IN PRESS
4
Changhao Wang et al.
to double-stranded DNA in cytosol, and is degraded to an intermediate of 50-
pApG and fast decomposed into AMP and GMP by Ecto-nucleotide
pyrophosphatase phosphodiesterase 1 (ENPP1) (Kato et al., 2018). 20,30-
c-GMP-AMP has been demonstrated to bind and activate the stimulator of
interferon genes (STING) and thereafter induce the production of type
I interferons (IFNs) (Ablasser et al., 2013; Danilchanka & Mekalanos,
2013). Beside 20,30-c-GMP-AMP, all other three bacterial CDNs could also
activate STING to elicit the innate immune response in mammalian cells,
resulting in the expression of type I IFNs (Burdette et al., 2011; Gao et al.,
2013; Parvatiyar et al., 2012).
The CDN second messengers could be sensed by effectors made of pro-
tein or RNA, resulting in the production of specific phenotypic responses
for regulating the metabolic pathways. For protein effectors, the four
known CDNs could efficiently interact with their dinucleotide cyclases
and PDEs. Moreover, c-di-GMP has been identified to bind to PilZ domain
proteins, FleQ and PelD transcription factors, I site effectors (Hengge,
2009). C-di-AMP has been found to interact with transcription factor of
DarR and domain proteins of KtrA, CpaA, KdpD and PstA (Corrigan &
Gruendling, 2013). For RNA effectors, the three bacterial CDNs have
been identified to tightly bind to RNA molecules called riboswitches.
Riboswitches are noncoding RNA motifs located in the 50-untranslated
region (UTR) of messenger RNA (mRNA) molecules consisting of an
aptamer domain and an expression platform. After selectively binding to
small molecule ligands, the aptamer domain changes its conformation to
further rearrange the expression platform, leading to the regulation of gene
expression. C-di-GMP has been found to target two classes of riboswitches
of the genes related to environment, membranes, and motility (GEMM) with
high affinities (Lee, Baker, Weinberg, Sudarsan, & Breaker, 2010; Sudarsan
et al., 2008). In the in-line probing assays, c-di-GMP binds GEMM class
I and class II riboswitches with the apparent Kd at 10pM and 2nM, respec-
tively. C-di-AMP is recognized by the ydaO riboswitch with an apparent
Kd of about 700pM (Nelson et al., 2013). Very recently, c-AMP-GMP has
been discovered to be sensed by a subset of GEMM class Ib riboswitches in
Geobacter species, where GEMM class Ib is the mutant of c-di-GMP GEMM
class I riboswitch with mutation of G20 (Nelson et al., 2015). Bioinformatical
analysis of GEMM class I riboswitches revealed that several sequence varia-
tions are conserving with the mutation of either G20 or C92, which could
potentially sense other unknown CDNs (Nelson et al., 2015).

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
5
Besides the basic starting units GTP and ATP of known CDNs, the tri-
phosphates of cytidine triphosphate (CTP) and uracil triphosphate (UTP)
are existing in nature in high abundancy, leading us to speculate that poten-
tially novel CDNs with cytidine monophosphate (CMP) or uracil monop-
hosphate (UMP) moieties might also play similar roles in bacterial signaling.
Besides the known canonical CDNs, seven more 30-50-linked canonical
CDNs bearing natural RNA bases are depicted in Fig. 2. Since only the
known CDNs have their corresponding dinucleotide cyclases, the novel
unknown CDNs could be achieved by chemical synthesis. The chemical
approaches for the known CDNs are widely investigated and synthesis
strategies for 30-50-linked CDNs have been recently reviewed (Clivio,
Coantic-Castex, & Guillaume, 2013). Either for CDN synthesis or its
Fig. 2 The chemical structures of unknown 30-50-linked canonical CDNs. Reprinted with
permission from Wang, C., Sinn, M., Stifel, J., Heiler, A. C., Sommershof, A. & Hartig, J. S.
(2017). Synthesis of all possible canonical (30-50-linked) cyclic dinucleotides and evaluation
of riboswitch interactions and immune-stimulatory effects. Journal of the American
Chemical Society, 139, 16154–16160. Copyright (2017) American Chemical Society.

ARTICLE IN PRESS
6
Changhao Wang et al.
derivatives, the linear precursor of CDN is commonly using pho-
sphoramidite approach and the cyclization step is using phosphotriester
approach (Ching, Tan, Chua, & Lam, 2010; Hayakawa, Nagata, Hirata,
Hyodo, & Kawai, 2003; Hyodo, Sato, & Hayakawa, 2006; Shanahan,
Gaffney, Jones, & Strobel, 2011). Some reports described that c-di-GMP
could be produced by H-phosphonate approach (Yan & Aguilar, 2007;
Zhang, Gaffney, & Jones, 2004) and c-di-AMP, c-di-CMP and c-di-UMP
could be synthesized by phosphodiester approach (Coutsogeorgopoulos &
Khorana, 1964; Rammler, Lapidot, & Khorana, 1963; Sawai, Higa, &
Kuroda, 1992). An alternative base insertion approach was developed to
synthesize CDNs by introducing nucleobases at the last stage to a cyclic
sugar-phosphate backbone, yet this method is suitable to synthesize symmetric
CDNs other than asymmetric CDNs (Amiot, Heintz, & Giese, 2006).
Lam and coworkers employed nucleosides as the starting materials to
synthesize c-di-GMP, c-di-UMP and c-GMP-UMP using nearly 20 steps
and several times of purifications (Ching et al., 2010). Solid-phase synthesis
was applied to synthesize CDNs because they can be easily prepared by
automated synthesis using the starting materials with the support linking to
either the nucleobase or the 30-phosphate group (Kiburu, Shurer, Yan, &
Sintim, 2008; Shanahan et al., 2011). However, in most cases, the non-
commercial starting materials lead to complicated and time-consuming syn-
thesis routes. The most recent phosphoramidite approach is frequently used
because of commercial starting materials. In particular, the Jones group has
developed a one-pot strategy for synthesizing CDNs and their derivatives,
making the synthesis of CDNs become fast and convenient (Gaffney &
Jones, 2014; Gaffney, Stebbins, & Jones, 2013; Gaffney, Veliath, Zhao, &
Jones, 2010). Very recently, we modified the one-pot strategy to synthesize
all possible 30-50-linked canonical CDNs and achieved the CDN synthesis
within 1 day (Wang et al., 2017). In this chapter, we describe the adapted
one-pot strategy for synthesis of the 30-50-linked CDNs, purification proto-
cols, and the necessary characterization techniques.
2. Synthesis of crude CDNs
2.1 Chemicals
(a) RNAphosphoramidites:DMT-20O-TBDMS-rA(Bz)phosphoramidite,
DMT-20O-TBDMS-rC(Ac) phosphoramidite, DMT-20O-TBDMS-
rG(iB) phosphoramidite, and DMT-20O-TBDMS-rU phosphoramidite
(Sigma-Aldrich or Linktech)

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
7
(b) Pyridine trifluoroacetate
(c) tert-Butylamine (tBuNH2)
(d) Dichloroacetic acid (DCA)
(e) tert-Butyl hydroperoxide (tBuOOH, ꢀ5.5M in decane over molecu-
˚
lar sieve 4A)
(f ) 2-Chloro-5,5-dimethyl-1,3,2-dioxaphosphorinane
2-oxide (DMOCP)
(g) Acetonitrile, extra dry over molecular sieve (CH3CN, Acros)
(h) Pyridine, extra dry over molecular sieve (Acros)
(i) Sodium bisulfite (NaHSO3)
(j) Sodium bicarbonate (NH4HCO3)
(k) Methylamine, 33wt% in anhydrous ethanol
(l) Triethylamine (TEA)
(m) Triethylamine trihydrofluoride (TEA∙3HF)
(n) Dichloromethane (CH2Cl2)
(o) Ethyl acetate
(p) Diethyl ether
(q) Acetone
2.2 Equipment
(a) Rotary evaporator equipped with vacuum system
(b) Magnetic stirrer integrated with heating module
(c) 50μL Hamilton syringe
2.3 Synthesis procedure for crude c-AMP-CMP
As shown in Scheme 1, the synthesis of c-AMP-CMP is taken as an
example to describe the typical procedure as follows:
(a) Dissolve DMT-20O-TBDMS-rA(Bz) phosphoramidite (1, 0.46mmol,
1.0eq.) and pyridinium trifluoroacetate (107mg, 0.55mmol, 1.2eq.) in
dry CH3CN (4mL) with a 50mL round flask under argon atmosphere.
Then, add H2O (17 μL, 0.94mmol, 2eq.) to hydrolysis of 1 and stir for
1min to give 2.
(b) Add tBuNH2 (2.5mL) to remove the cyanoethyl group (CE) and stir
for 1min to give 3, concentrate the solution to foam under reduced
pressure using dry CH3CN for three times (5mLꢁ3).
(c) Dissolve the resulted foam in CH2Cl2 (5.5mL) under argon atmo-
sphere. Add DCA solution (6% in CH2Cl2, 5.5mL, 4.0mmol) to
remove the 4,40-dimethoxytriphenylmethyl group (DMT) and H2O

ARTICLE IN PRESS
8
Changhao Wang et al.
Scheme 1 Synthesis routes for c-AMP-CMP. Adapted with permission from Wang, C.,
Sinn, M., Stifel, J., Heiler, A. C., Sommershof, A. & Hartig, J. S. (2017). Synthesis of all possible
canonical (30-50-linked) cyclic dinucleotides and evaluation of riboswitch interactions and
immune-stimulatory effects. Journal of the American Chemical Society, 139,
16154–16160. Copyright (2017) American Chemical Society.
(83 μL, 4.6mmol, 10eq.) to suppress the retritylation, stir for 10min to
give 4. Then, add pyridine (0.66mL, 8.2mmol) to quench the excess
DCA, concentrate the solution to foam under reduced pressure using
dry CH3CN for three times (5mLꢁ3) and leave the last solution (A) at
about 1mL.
Tip: If there are some solids precipitated from the solution (A) after evapo-
ration, put the flask in a water bath (ca. 40°C) to re-dissolve the solids for the
next step.

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
9
(d) Dissolve DMT-20O-TBDMS-rC(Ac) phosphoramidite (0.6mmol,
1.3eq.) in dry CH3CN (5mL), concentrate the solution to foam under
reduced pressure doubly using dry CH3CN (5mLꢁ2) and leave the
last solution (B) at about 2mL. Then, mix the solution A from step
c and the solution B under argon atmosphere to proceed the coupling
reaction, stir for 2min to give 5.
(e) Add tBuOOH solution (5.5 M in decane, 0.25mL, 1.38mmol) to
oxide phosphite triester to phosphodiester bond and stir for 30min
to give 6. Then, put the flask into an ice bath, add NaHSO3 aqueous
solution (0.5M, 230 μL, 1.15mmol, 2.5eq.) to consume the excess
tBuOOH and stir for 10min. At last, remove the ice bath, stir for
5min at room temperature, and concentrate the solution to oil under
reduced pressure.
(f ) Dissolve the above oil in CH2Cl2 (7.4mL) under argon atmosphere. Add
DCA solution (6% in CH2Cl2, 7.4mL, 5.4mmol) and H2O (83 μL,
4.6mmol, 10eq.) to remove the 4,40-dimethoxytriphenylmethyl (DMT)
group, stir for 10min to give 7.
(g) Add pyridine (1.5mL, 18.6mmol) to quench the reaction, and concen-
trate the solution to oil. Next, add pyridine (15mL), concentrate the
solution under reduced pressure, and leave last solution at about
8mL. Then, add DMOCP (303mg, 1.6mmol, 3.5eq.) to promote
the cyclization and stir for 30min to give 8.
(h) Add H2O (294 μL, 16.3mmol) to quench the excess of DMOCP and
I2 (152mg, 0.6mmol) to transform H-phosphonate diester to pho-
sphodiester, stir for 5min to give 9. Next, pour the above mixture into
NaHSO3 aqueous solution (10mM, 88mL, 0.92mmol, 2.0eq.) to con-
sume the excess I2 and stir for 30min. Then, slowly add NaHCO3
(1.8g) to neutralize the excess acids and stir for 5min. At last, triply
use the mixture of ethyl acetate and diethyl ether (1:1, 100mLꢁ3)
to extract the aqueous solution, combine the organic layer and concen-
trate to oil under reduced pressure.
Tip: The compound 9 could be purified on silica gel column and the
compounds obtained after step (i) are hard to be purified using silica gel
column.
(i) Dissolve the above oil in CH3CN (2.5mL), add tBuNH2 (2.5mL) to
remove the 2-cyanoethyl (CE) group and stir for 10min to give 10.
Concentrate the solution to foam under reduced pressure using dry
CH3CN for three times (5mLꢁ3). Next, dissolve the foam with a
small volume of CH2Cl2. (Tip: Add CH3OH to promote the solubility

ARTICLE IN PRESS
10
Changhao Wang et al.
if the foam cannot be fully dissolved in CH2Cl2.) Then, slowly add
diethyl ether to precipitate the off-white solid, collect and dry the
precipitate.
(j) Dissolve the above crude precipitate (0.1mmol) in methylamine solu-
tion (33wt% in anhydrous ethanol, 10.0mL, 80mmol) under argon to
remove the benzoyl group (Bz) and acetyl group (Ac) and stir for
90min at room temperature to give 11. Then, concentrate the solution
to oil and dry the oil under reduced pressure for three times with a
mixture of pyridine (0.4mL) and TEA (0.2mL).
(k) Dissolve the above residual oil in pyridine (0.4mL) under argon and
place the reaction flask in an oil bath at 50°C. Then, spontaneously
add TEA (1.4mL, 10mmol) and Et3Nꢂ3HF (0.9mL, 5.5mmol) to
remove the tert-butyldimethylsilyl groups (TBDMS), and stir for 3h
to give 12.
(l) Remove the oil bath and cool to room temperature. Then, slowly add
acetone to the above solution to form the off-white precipitate and stir
for 10min. At last, collect the precipitate and dry it. The crude
c-AMP-CMP was obtained in triethylammonium salt form.
3. Purification of crude CDNs
3.1 Preparation of HPLC buffers
(1) Preparation of triethylammonium acetate buffer (TEAA, 50mM,
pH7.0): Dissolve 2.76mL glacial acetic acid into a 1.0L beaker con-
taining ꢀ970mL of milli-Q water. Next, slowly add 6.95mL TEA into
the above solution and mix the solution thoroughly. Then, adjust the
pH value to ꢀ7.0 with diluted acetic acid solution and add the total
volume to 1.0L with milli-Q water. Finally, filter the as-prepared
TEAA buffer through a 0.2μm HPLC-certified Nylon filter for HPLC.
(2) Preparation of ammonium bicarbonate buffer (ABC, 50mM, pH7.0):
Dissolve 3.95g ammonium bicarbonate into a 1.0L beaker containing
ꢀ970mL of milli-Q water. Next, adjust the pH value to ꢀ7.0 with
diluted formic acid solution and adjust the total volume to 1.0L with
milli-Q water. Finally, filter the as-prepared ABC buffer through a
0.2μm HPLC-certified Nylon filter for HPLC.
3.2 Purification protocol for crude c-AMP-CMP
For purification of CDN, a set of Shimadzu liquid chromatography (LC)
system is employed that equipped with LC 20AT pumps, SPD-M20A

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
11
PDA detectors, and FRC-10A fraction collector. Acetonitrile is used as
mobile phase A, while either TEAA buffer (50mM, pH7.0) or ABC
buffer (50mM, pH7.0) is used as mobile phase B. A 250/10 Nucleodur
C18 Pyramid, 5μm column (Macherey-Nagel) is used with the gradients
of acetonitrile (2–25%) in either TEAA buffer or ABC buffer. The high
performance liquid chromatography (HPLC) protocol for purification of
crude c-AMP-CMP is shown in Table 1.
Tips:
(1) Compared with TEAA buffer at 50mM, the peak of CDN on HPLC
trace becomes sharper when using TEAA buffer at 100mM. The sep-
aration efficiency of crude CDN is nearly the same either using 50mM
or 100mM TEAA buffer, but high concentration of TEAA needs lon-
ger time to remove the TEAA salt from the c-AMP-CMP fraction
through lyophilization.
Table 1 HPLC protocol for purification of crude c-AMP-CMP.
Sample preparation
HPLC instrument
30mg of crude c-AMP-CMP in 180μL H2O
Shimadzu systems equipped with LC 20AT
pumps, SPD-M20A PDA detectors, and
FRC-10A fraction collector
Column
VP 250/10 NUCLEODUR Pyramid, 5μM
Flow rate
4mL/min
260nm
35°C
Wavelength
Temperature
Mobile phase
A: TEAA buffer (50mM, pH 7.0)
B: CH3CN
Time (min)
CH3CN (B, %)
Time program
0
2
5
2
25
27.5
30
35
40
20
100
100
2
2

ARTICLE IN PRESS
12
Changhao Wang et al.
(2) Compared with TEAA buffer and ABC buffer at the same concentra-
tion, using the later one as mobile phase B will bring earlier elution time
of CDN and little higher separation efficiency. However, ABC buffer is
more difficult to remove than TEAA buffer in lyophilization under the
same condition.
(3) For different CDNs, the HPLC gradients of CH3CN is varied in the
range from 14 to 20% at the time point of 25min.
3.3 HPLC purification of crude c-AMP-CMP
The crude c-AMP-CMP sample solution is filtered in advance through a
polytetrafluoroethylene (PTFE) filter (25mmꢁ0.45μm) before injection.
First, a small volume of injection (1–5 μL) is injected to roughly check
the purity of the crude c-AMP-CMP, the separation efficiency and the sam-
ple concentration. When we inject 1 μL of crude c-AMP-CMP sample
(Fig. 3A), the main fraction 1 is at 17.6min with the amount ratio beyond
90%, which is the c-AMP-CMP evidenced by mass spectrometry (MS) and
nuclear magnetic resonance (NMR) in Section 4. We observed the fine sep-
aration efficiency between fraction 1 at 17.6min and fraction 2 at 18.1min.
Once 20 μL of the crude c-AMP-CMP sample is injected (Fig. 3B), we
found that the remaining time of fraction 1 is maintained although the inten-
sity of the fraction is out of the maximum range. However, the retention
time of the faction 2 shifts to 19.5min with injection volume of 20 μL.
The fractions at 17.5 and 19.5min are respectively collected for further
MS and NMR characterizations.
4. Characterization of CDNs
4.1 Materials and equipment
Liquid chromatography mass spectrometry (LC-MS) is conducted on a
Shimadzu LCMS-2020 single quadrupole mass spectrometer. High resolu-
tion mass spectra (HRMS) are recorded on a Bruker Daltronics microTOF
II mass spectrometer. All of the CDN fractions for NMR are dissolved in
deuterium water. One dimensional (1-D) and two dimensional (2-D)
NMR spectra are recorded at 300K on Bruker series instruments, e.g.
Bruker Avance III 400MHz spectrometers with a BBFO plus probe for
N to F/H or F, Bruker Avance III 600MHz spectrometer with a TCI-
H/C/N triple resonance cryoprobe and Bruker Avance III 600MHz spec-
trometer with QXI-H/C/N/P quadruple resonance probe. The data are
processed and analyzed by MestReNova software.

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
13
Fig. 3 HPLC traces for purification of crude c-AMP-CMP using protocol from Table 1.
(A) 1 μL and (B) 20 μL injection volume of crude c-AMP-CMP sample.
4.2 MS analysis of c-AMP-CMP fractions
The fractions from purification of crude c-AMP-CMP in Section 3.3 could
be directedly analyzed by LC-MS. The LC-MS data are listed in Table 2.
Compared to the calculated molecular weight of c-AMP-CMP (634.1),
the LC-MS data of fraction 1 corresponds to c-AMP-CMP under either
positive mode (M+1¼635.2) or negative mode (Mꢃ1¼633.1). For the

ARTICLE IN PRESS
14
Changhao Wang et al.
Table 2 LC-MS analysis for the c-AMP-CMP fractions.
LC-MS data
Fraction 1 (17.5min)
Fraction 2 (19.5min)
Positive mode (M+1)
Negative mode (Mꢃ1)
Possible M.W.
635.2
633.1
634.1
573.2
571.2
572.1
Possible structure
fraction 2, the experimental molecular weight is 572.1 from the LC-MS
results, indicating that fraction might be the non-cyclic dinucleotide
c-AMP-CMP (Table 2). The aim fraction 1 at 17.5min is further confirmed
by HRMS to ensure the mass analysis of c-AMP-CMP.
4.3 NMR analysis of c-AMP-CMP
After the fractions are analyzed by LC-MS, the aim fraction is lyophilized
for NMR characterization. If for an unknown CDN, the fine structure of
CDN should be identified by 1-D and 2-D NMR techniques. Protons
1
and carbons in c-AMP-CMP are assigned by H NMR, 13C NMR,
1H–13C heteronuclear single quantum correlation (HSQC), 1H–1H corre-
1
lation spectroscopy (COSY), and H–13C heteronuclear multiple-bond
correlation (HMBC), see Fig. 4 for details. 1H NMR and 13C NMR spectra
are first to confirm whether the number of protons or carbons is correct for
c-AMP-CMP (Fig. 4B and C).
Based on the chemical shifts, we can roughly assign the protons and car-
1
bons to the structure of c-AMP-CMP (Fig. 4A). H–1H COSY spectrum
reflects the correlations between J-coupled protons that can help to revise
1
the initial assignment of protons (Fig. 4D). In combination of H–13C
HSQC and 1H–13C HMBC (Fig. 4E and F), the precise assignment of pro-
tons and carbons could be confirmed, where 1H–13C HSQC reveals the cor-
relations between the proton and its bounding carbon and 1H–13C HMBC
reflects 1H–13C correlations over longer ranges of about 2–4 bonds. There
are two different phosphorus in c-AMP-CMP, that could be distinguished
by the spectra of 1H NMR, 31P NMR, and 1H–31P HMBC (Fig. 5). In view

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
15
Fig. 4 Assignment of protons and carbons in c-AMP-CMP. (A) Chemical structure of
1
1
c-AMP-CMP; NMR spectra of c-AMP-CMP in D2O: (B) H NMR, (C) 13C NMR, (D) H–1H
COSY, (E) 1H–13C HSQC, and (F) 1H–13C HMBC. Adapted with permission from
Wang, C., Sinn, M., Stifel, J., Heiler, A. C., Sommershof, A. & Hartig, J. S. (2017). Synthesis
of all possible canonical (30-50-linked) cyclic dinucleotides and evaluation of riboswitch
interactions and immune-stimulatory effects. Journal of the American Chemical Society,
139, 16154–16160. Copyright (2017) American Chemical Society.
1
of the assignment of protons in c-AMP-CMP (Fig. 5B), H–31P HMBC
spectrum provides the additional correlations information between phos-
phorus and the protons nearby, especially for H30, H3000, H50 and H5000
1
1
(Fig. 5D). From H NMR, 31P NMR and H–31P HMBC spectra, the
two phosphorus in c-AMP-CMP are accurately deciphered. Based on the
1-D and 2-D NMR spectra, we can precisely assign the protons, carbons
and phosphorus to an unknown CDN, which would be helpful to investi-
gate the interaction between CDN and receptors of either proteins or
riboswitches.

Fig. 5 Assignment of phosphorus in c-AMP-CMP. (A) Chemical structure of c-AMP-CMP; NMR spectra of c-AMP-CMP in D2O: (B) 1H NMR, (C) 31
P
NMR, and (D) 1H–31P HMBC. Adapted with permission from Wang, C., Sinn, M., Stifel, J., Heiler, A. C., Sommershof, A. & Hartig, J. S. (2017). Synthesis
of all possible canonical (30-50-linked) cyclic dinucleotides and evaluation of riboswitch interactions and immune-stimulatory effects. Journal of the
American Chemical Society, 139, 16154–16160. Copyright (2017) American Chemical Society.

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
17
5. Summary and prospects
In summary, we describe the detailed synthesis procedure for the
30-50-linked canonical CDNs using an adapted one-pot strategy and provide
helpful tips for some key steps. This typical procedure is suitable for synthesis
of CDNs with any combination of nucleobases and enables fast access to the
crude CDNs within 1 day. Furthermore, we present the detailed purifica-
tion protocol for crude CDNs and introduce the techniques for full charac-
terization of the purified CDNs by MS and NMR. This chapter provides a
fast and convenient way to obtain canonical CDNs and will promote the
investigations for biological functions of possible unknown CDNs.
Until now, four CDNs have been discovered in nature with the building
blocks GTP and ATP. All of the known CDNs are serving as second mes-
sengers for signal transduction and regulation of many physiological pro-
cesses. The possibility that all four canonical RNA building blocks could
be used to synthesize naturally occurring CDNs has motivated us to explore
whether further CDNs might play so-far unknown roles in bacterial
signaling. Utilizingthe describedCDNs asanalytical standardswehavealready
tried to detect novel CDNs in 12 different bacterial extracts by LC-MS,
although we so far did not find novel CDNs in these bacteria (Wang et al.,
2017). However, more sensitive detection methods could be applied and
extracts from more diverse species should be investigated in efforts to detect
additional CDNs in nature.
Acknowledgments
We are grateful for the financial supports of the Natural Science Foundation of China
(21703132), the Fundamental Research Funds for the Central Universities (GK201802001)
toC.W. and European Research Council Consolidator Grant (ERCCoG project“RiboDisc”)
to J.S.H.
References
Ablasser, A., Goldeck, M., Cavlar, T., Deimling, T., Witte, G., Roehl, I., et al. (2013). cGAS
produces a 20-50-linked cyclic dinucleotide second messenger that activates STING.
Nature, 498, 380–384.
Amiot, N., Heintz, K., & Giese, B. (2006). New approach for the synthesis of c-di-GMP and
its analogues. Synthesis, 4230–4236.
Burdette, D. L., Monroe, K. M., Sotelo-Troha, K., Iwig, J. S., Eckert, B., Hyodo, M., et al.
(2011). STING is a direct innate immune sensor of cyclic di-GMP. Nature, 478,
515–518.

ARTICLE IN PRESS
18
Changhao Wang et al.
Ching, S. M., Tan, W. J., Chua, K. L., & Lam, Y. (2010). Synthesis of cyclic di-nucleotidic
acids as potential inhibitors targeting diguanylate cyclase. Bioorganic & Medicinal Chemis-
try, 18, 6657–6665.
Clivio, P., Coantic-Castex, S., & Guillaume, D. (2013). (30-50)-Cyclic dinucleotides: Syn-
thetic strategies and biological potential. Chemical Reviews, 113, 7354–7401.
Corrigan, R. M., & Gruendling, A. (2013). Cyclic di-AMP: Another second messenger
enters the fray. Nature Reviews. Microbiology, 11, 513–524.
Coutsogeorgopoulos, C., & Khorana, H. (1964). Studies on polynucleotides. XXXI.
The specific synthesis of C30-C50-linked ribopolynucleotides (6). A further study of
the synthesis of uridine polynucleotides. Journal of the American Chemical Society, 86,
2926–2932.
Danilchanka, O., & Mekalanos, J. J. (2013). Cyclic dinucleotides and the innate immune
response. Cell, 154, 962–970.
Davies, B. W., Bogard, R. W., Young, T. S., & Mekalanos, J. J. (2012). Coordinated
regulation of accessory genetic elements produces cyclic di-nucleotides for V. cholerae
virulence. Cell, 149, 358–370.
Gaffney, B. L., & Jones, R. A. (2014). Synthesis of c-di-GMP analogs with thiourea, urea,
carbodiimide, and guanidinium linkages. Organic Letters, 16, 158–161.
Gaffney, B. L., Stebbins, N. D., & Jones, R. A. (2013). Synthesis of biotinylated
c-di-GMP and c-di-AMP using click conjugation. Nucleosides, Nucleotides & Nucleic
Acids, 32, 1–16.
Gaffney, B. L., Veliath, E., Zhao, J., & Jones, R. A. (2010). One-flask syntheses of
c-di-GMP and the [Rp,Rp] and [Rp,Sp] thiophosphate analogues. Organic Letters, 12,
3269–3271.
Gao, P., Ascano, M., Wu, Y., Barchet, W., Gaffney, B. L., Zillinger, T., et al. (2013).
Cyclic G(20,50)pA(30,50)p is the metazoan second messenger produced by DNA-activated
cyclic GMP-AMP synthase. Cell, 153, 1094–1107.
Gao, P., Ascano, M., Zillinger, T., Wang, W., Dai, P., Serganov, A. A., et al. (2013).
Structure-function analysis of STING activation by c G(20,50)pA(30,50)p and targeting
by antiviral DMXAA. Cell, 154, 748–762.
Gao, J., Tao, J., Liang, W., Zhao, M., Du, X., Cui, S., et al. (2015). Identification and
characterization of phosphodiesterases that specifically degrade 303’-cyclic GMP-AMP.
Cell Research, 25, 539–550.
Hayakawa, Y., Nagata, R., Hirata, A., Hyodo, M., & Kawai, R. (2003). A facile synthesis of
cyclic bis(30-50)diguanylic acid. Tetrahedron, 59, 6465–6471.
Hengge, R. (2009). Principles of c-di-GMP signalling in bacteria. Nature Reviews. Microbiol-
ogy, 7, 263–273.
Hyodo, M., Sato, Y., & Hayakawa, Y. (2006). Synthesis of cyclic bis(30-50)diguanylic acid
(c-di-GMP) analogs. Tetrahedron, 62, 3089–3094.
Kato, K., Nishimasu, H., Oikawa, D., Hirano, S., Hirano, H., Kasuya, G., et al. (2018).
Structural insights into cGAMP degradation by Ecto-nucleotide pyrophosphatase
phosphodiesterase 1. Nature Communications, 9, 4424.
Kiburu, I., Shurer, A., Yan, L., & Sintim, H. O. (2008). A simple solid-phase synthesis of
the ubiquitous bacterial signaling molecule, c-di-GMP and analogues. Molecular
BioSystems, 4, 518–520.
Lee, E. R., Baker, J. L., Weinberg, Z., Sudarsan, N., & Breaker, R. R. (2010). An allosteric
self-splicing ribozyme triggered by a bacterial second messenger. Science, 329, 845–848.
Nelson, J. W., Sudarsan, N., Furukawa, K., Weinberg, Z., Wang, J. X., & Breaker, R. R.
(2013). Riboswitches in eubacteria sense the second messenger c-di-AMP. Nature
Chemical Biology, 9, 834.

ARTICLE IN PRESS
30-50-Linked canonical cyclic dinucleotides
19
Nelson, J. W., Sudarsan, N., Phillips, G. E., Stav, S., Lunse, C. E., McCown, P. J., et al.
(2015). Control of bacterial exoelectrogenesis by c-AMP-GMP. Proceedings of the
National Academy of Sciences of the United States of America, 112, 5389–5394.
Parvatiyar, K., Zhang, Z., Teles, R. M., Ouyang, S., Jiang, Y., Iyer, S. S., et al. (2012).
The helicase DDX41 recognizes the bacterial secondary messengers cyclic di-GMP
and cyclic di-AMP to activate a type I interferon immune response. Nature Immunology,
13, 1155–1161.
Rammler, D., Lapidot, Y., & Khorana, H. (1963). Studies on polynucleotides. XIX. The
specific synthesis of C300-C500 inter-ribonucleotidic linkage. A new approach and its
use in the synthesis of C300-C500-linked uridine oligonucleotides. Journal of the American
Chemical Society, 85, 1989–1997.
Ross, P., Weinhouse, H., Aloni, Y., Michaeli, D., Weinbergerohana, P., Mayer, R., et al.
(1987). Regulation of cellulose synthesis in acetobacter-xylinum by cyclic diguanylic
acid. Nature, 325, 279–281.
Sawai, H., Higa, K., & Kuroda, K. (1992). Synthesis of cyclic and acyclic oligocytidylates by
uranyl ion catalyst in aqueous solution. Journal of the Chemical Society, Perkin Transactions,
1, 505–508.
Shanahan, C. A., Gaffney, B. L., Jones, R. A., & Strobel, S. A. (2011). Differential analogue
binding by two classes of c-di-GMP riboswitches. Journal of the American Chemical Society,
133, 15578–15592.
Sudarsan, N., Lee, E. R., Weinberg, Z., Moy, R. H., Kim, J. N., Link, K. H., et al. (2008).
Riboswitches in eubacteria sense the second messenger cyclic di-GMP. Science, 321,
411–413.
Wang, C., Sinn, M., Stifel, J., Heiler, A. C., Sommershof, A., & Hartig, J. S. (2017).
Synthesis of all possible canonical (30-50-linked) cyclic dinucleotides and evaluation of
riboswitch interactions and immune-stimulatory effects. Journal of the American Chemical
Society, 139, 16154–16160.
Witte, G., Hartung, S., Buttner, K., & Hopfner, K. P. (2008). Structural biochemistry
of a bacterial checkpoint protein reveals diadenylate cyclase activity regulated by
DNA recombination intermediates. Molecular Cell, 30, 167–178.
Woodward, J. J., Iavarone, A. T., & Portnoy, D. A. (2010). c-di-AMP secreted by intracel-
lular listeria monocytogenes activates a host type I interferon response. Science, 328,
1703–1705.
Yan, H., & Aguilar, A. L. (2007). Synthesis of 30,50-cyclic diguanylic acid (cdiGMP) using
1-(4-chlorophenyl)-4-ethoxypiperidin-4-yl as a protecting group for 20-hydroxy
functions of ribonucleosides. Nucleosides, Nucleotides & Nucleic Acids, 26, 189–204.
Zhang, Z. Y., Gaffney, B. L., & Jones, R. A. (2004). c-di-GMP displays a monovalent metal
ion-dependent polymorphism. Journal of the American Chemical Society, 126, 16700–16701.