Variable-Temperature NMR Spectroscopy, Conformational Analysis, and Thermodynamic Parameters of Cyclic Adenosine 5′-Diphosphate Ribose Agonists and Antagonists

Article abstract of DOI:10.1021/acs.joc.7b02749

Cyclic adenosine 5′-diphosphate ribose (cADPR) is a ubiquitous Ca²⁺-releasing second messenger. Knowledge of its conformational landscape is an essential tool for unraveling the structure-activity relationship (SAR) in cADPR. Variable-temperature ¹H NMR spectroscopy, in conjunction with PSEUROT and population analyses, allowed us to determine the conformations and thermodynamic parameters of the furanose rings, γ-bonds (C4′-C5′), and β-bonds (C5′-O5′) in the cADPR analogues 2′-deoxy-cADPR, 7-deaza-cADPR, and 8-bromo-cADPR. A significant finding was that, although the analogues are similar to each other and to cADPR itself in terms of overall conformation and population (ΔG°), there were subtle yet important differences in some of thermodynamic properties (ΔH°, ΔS°) associated with each of the conformational equilibria. These differences prompted us to propose a model for cADPR in which the interactions between the A2′-N3, A5″-N3, and H2-R5′ atoms serve to fine-tune the N-glycosidic torsion angles (χ).

Full text of DOI:10.1021/acs.joc.7b02749

Subscriber access provided by READING UNIV
Article
Variable-Temperature NMR Spectroscopy, Conformational Analysis,
and Thermodynamic Parameters of cADPR Agonists and Antagonists
Sarah-Marie Saatori, Tanner J. Perez, and Steven McGrath Graham
J. Org. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.joc.7b02749 • Publication Date (Web): 24 Jan 2018
Downloaded from http://pubs.acs.org on January 25, 2018
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.
The Journal of Organic Chemistry 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 38
The Journal of Organic Chemistry
1
2
3
Variable-Temperature NMR Spectroscopy, Conformational Analysis, and
Thermodynamic Parameters of cADPR Agonists and Antagonists
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
*
SarahꢀMarie Saatori, Tanner J. Perez, and Steven M. Graham
Department of Chemistry, St. John's University, 8000 Utopia Parkway, Queens, NY 11439, USA.
ABSTRACT: cyclic Adenosine 5'ꢀdiphosphate ribose (cADPR) is a
2+
ubiquitous Ca ꢀreleasing second messenger.
Knowledge of its
conformational landscape is an essential tool in unraveling the structureꢀ
1
activity relationship (SAR) in cADPR. Variableꢀtemperature HꢀNMR
spectroscopy, in conjunction with PSEUROT and population analyses
allowed us to determine the conformations and thermodynamic parameters
of the furanose rings, γ–bonds (C4'–C5'), and β–bonds (C5'–O5') in the
cADPR analogs 2'ꢀdeoxyꢀcADPR, 7ꢀdeazaꢀcADPR, and 8ꢀbromoꢀcADPR.
A significant finding was that while the analogs are similar to each other
and to cADPR itself in terms of overall conformation and population
°
(
ꢁG ), there were subtle, yet important differences in some of
°
°
thermodynamic properties (ꢁH , ꢁS ) associated with each of the
conformational equilibria. These differences prompted us to propose a
model for cADPR in which the interactions between the A2'–N3, A5"–N3,
and H2–R5' atoms serve to fine tune the Nꢀglycosidic torsion angles (χ).
keywords: agonists; cADPR analogs; calcium release; nucleotides; signal messengers; synthesis.
1
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 2 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ꢀ
INTRODUCTION
+
Beyond its role in redox metabolism, nicotinamide adenine dinucleotide (NAD ) has emerged of late as
1
,2
+
a regulator of numerous other biological pathways. NAD acts as an acyl acceptor in the decacylation
of histone proteins in reactions catalyzed by silent information regulatorꢀ2 (Sir2) proteins (sirtuins), as
the donor in the monoꢀ and polyꢀadenosine diphosphoribosylation of proteins as catalyzed by monoꢀ
ADPꢀribosyl transferases and polyꢀADPꢀribosyl polymerases (PARPs), as a 5'ꢀcap to some prokaryotic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
RNAs, and, relevant here, as a substrate for cyclic adenosine 5'ꢀdiphosphate ribose (cADPR, Figure 1)
+
synthases. At the phenotypic level, a recent mouse study suggested enhancing NAD levels can prolong
4
+
lifespan. NAD levels can be depleted by the multifunctional transmembrane glycoprotein CD38 (and
+
5,6
CD157), which is an NAD hydrolase, cADPR synthase, and a cADPR transporter. Whereas CD38
+
deficient mice have greatly enhanced NAD levels compared to wild type, overexpression of CD38 in
human kidney cells led to the downꢀregulation of over 100 proteins involved in an array of cellular
7
+
processes. Clearly NAD is a major player in the regulation of cellular metabolism and homeostasis,
2
+
but we wish to focus on one aspect in particular: the role of its Ca ꢀreleasing metabolite cADPR.
8
2+
First discovered in 1987, cADPR quickly emerged as Ca ꢀreleasing second messenger signaling
9
2+
molecule. Changes in intracellular Ca levels regulate a host of cellular processes that have
1
0,11
physiological significance, notably heart muscle contraction, fertilization, and insulin secretion.
The
2
+
end result of the formation of cADPR is Ca ꢀefflux from the ryanodine receptor (RyR), a large (~2
mDa total mass) tetrameric protein located in the sarco(endo)plasmic reticulum membrane. It is
generally believed that cAPDR does not interact with the RyR directly, but rather through the agency of
some other protein. RyR activity is modulated by multiple protein ligands, including kinases,
1
2,13
phosphatases, calmodulin, and the FKꢀ506 binding protein FKBP12.6 (calstabin).
Which of these
ligands – if any – is the cognate protein binding partner for cADPR remains to be established, although a
recent photoaffinity labelling study utilizing a cADPR analogue identified glyceraldehyde 3ꢀ
1
4
phosphate dehydrogenase (GAPDH) as a low affinity (18 ꢀM) cADPR binding protein.
The SAR in cADPR. The structureꢀactivity relationship (SAR) in cADPR continues to evolve. Over
1
00 cADPR analogs are known, and while the SAR is far from settled, certain patterns have emerged.
2
+
15,16
The parent cAPDR (4a, Figure 1 and Figure 2) has an EC for Ca release of 30–90 nM
in the sea
6
5
0
1
urchin homogenate (SUH) assay. Three early analogs – 2'ꢀdeoxyꢀcADPR (2'ꢀdA cADPR, 4b) ,
2
ACS Paragon Plus Environment

Page 3 of 38
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
+
Figure 1. Synthesis of cADPR from NAD and breakdown of cADPR to ADPR. ADP ribosyl cylase (ADPRC)
from the mollusk Aplysia californica is a soluble, commercially available enzyme. CD38 and CD 157 are
transmembrane glycoproteins. cADPR can be hydrolyzed to ADPR in the continued presence of these enzymes;
o
it can also undergoes spontaneous hydrolysis, with a halfꢀlife of 2.5 days at pH 7 at 37 C
17
15
7
ꢀdeazaꢀcADPR (4c) , and 8ꢀbromoꢀcADPR (8ꢀBrꢀcADPR, 4d) (Figure 2) – are, respectively, a
2
+
potent agonist (EC 58 nM), a partial agonist (90 nM, with only 66% of the total Ca released), and an
5
0
antagonist (IC ~1000 nM). We wished to perform detailed conformational analyses on cADPR
5
0
analogs, and we chose these three analogs for this variable temperature NMR study because (1) they
span the range of pharmacological activity, (2) they are easily synthesized using the 'chemoenzymatic'
+
approach (chemical synthesis of the appropriate NAD analog and enzymatic cyclization with the
commercially available Aplysia californica ADP ribosyl cyclase, Figure 1 and Scheme 1), and (3) they
are all based on a deoxyribofuranose or ribofuranoses scaffold and as such do not require substantial
1
8
modification to the default parameters of PSEUROT (see Results and Discussion), the program used to
1
1
convert 3ꢀbond vicinal Hꢀ H coupling constants to (deoxy)ribofuranose conformations. Compounds
1
9
4
b, 4c, and 4d, along with 3ꢀdeazaꢀcADPR (9a, EC ~1 nM), and cyclic aristeromycin diphosphate
50
20
ribose (cArisDPR, 10, EC 80 nM) typify one class of cADPR analogs, the 'Aꢀring' modified analogs
50
2
1
(
see the legend to Figure 2). A common feature of deaza analogs 4c and 9a, carbocyclic 10, and
cATPR (8) is an improved resistance to hydrolysis, which has implications for NMR studies (vide infra).
A-ring and Adenine modified Analogs. Generally, substitution at the 8ꢀposition results in an
15
antagonist. 8ꢀaminoꢀcADPR, like 8ꢀBrꢀcADPR 4d, is an antagonist, in both the SUH assay and in
2
2
Jurkat T cells (JTC). The 7ꢀdeaza alteration alters activity inconsistently; whereas 7ꢀdeazaꢀcADPR 4c
2
+
is less effective at Ca release than cADPR, 8ꢀbromoꢀ7ꢀdeazaꢀcADPR is a more effective antagonist
23
than 8ꢀbromoꢀcADPR 4d. Removal of the Aꢀring 2'ꢀOH from the antagonist 8ꢀNH ꢀcADPR
2
3
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 4 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Structures of cADPR analogs. Based on its name – cADPR, for 'cyclic adenosine diphosphate
ribose', we prefer to call the ribofuranose ring derived from adenosine (the 'A' in cADPR) the 'Aꢀring' and
the other ribofuranose ring (the 'R' in cADPR') the 'Rꢀring'. Others have called these rings the 'southern' and
'northern' ribose rings, respectively, but we fear this can lead to confusion in the subsequent discussion
where the terms 'north' and 'south' refer to conformational descriptors of furanose ring geometries
led to a 30ꢀfold drop in potency, whereas the same deletion in the antagonist 8ꢀBr cADPR 4d had no
2
4
effect on activity. In terms of modifications to the Aꢀring furanose, the Potter group found the Aꢀring
2
'ꢀOH could be deleted (2'ꢀdA cADPR, 4b) with essentially no loss in activity, but deletion of the Aꢀring
'ꢀOH resulted in a ~100ꢀfold drop in agonistic activity and replacing the 3'ꢀOH with a methoxy group
3
16
resulted in a weak antagonist. Additionally, analog potencies cannot be compared across different
assays and organisms – agonists in for example a SUH assay may not show the same behavior in a JTC
22
assay.
cIDPR Analogs. A second class of cADPR analogs (Figure 2), the N1ꢀcyclic inosine diphosphate
2
5,26
ribose (N1ꢀcIDPR, 5) analogs, has been studied extensively by the Potter group.
The requisite
+
dinucleotide precursor, nicotinamide hypoxanthine dinucleotide (NHD ), is a substrate for ADPRC but
cyclizes at N7 rather than N1. Recognizing the cyclase's preference for a synꢀoriented base, and
+
knowing that bulky C8ꢀsubstituents favor a syn orientation, they rationalized that 8ꢀbromoꢀNHD might
adopt a conformation more suitable for ADPRCꢀcatalyzed cyclization at N1. The initially formed
8
8
ꢀbromoꢀN1ꢀcIDPR was subsequently converted to N1ꢀcIDPR (5), as well as to 8ꢀphenylꢀ, 8ꢀazidoꢀ, and
ꢀaminoꢀcIDPR analogs. The cIDPR analogs show greatly enhanced hydrolysis resistance, are
membrane permeable, and 5 behaved almost identically to cADPR in a JTC assay. cIDPR 5 is also an
2
7
inhibitor (IC = 276 ꢀM) of the cADPR hydrolase activity of soluble human CD38.
50
4
ACS Paragon Plus Environment

Page 5 of 38
The Journal of Organic Chemistry
1
2
3
4
5
R-ring and Adenine modified Analogs. The analogs discussed above were all prepared via the
chemoenzymatic approach: chemical synthesis of a nicotinamide dinucleotide precursor and subsequent
cyclization using ADPRC. While some Rꢀring modified cADPR analogs have been prepared using this
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
8
strategy (requiring the preparation of a nicotinamide mononucleotide (NMN) analog), generally total
chemical synthesis is required. In the Rꢀring modified series (Figure 2), cyclic adenosine diphosphate
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
9,30
carbocyclic ribose (cADPcR, 6, EC 15 nM versus 50 nM for cADPR),
and cyclic adenosine
50
3
1
diphosphate thioribose (cADPtR, 7, EC 39 nM versus 210 nM for cADPR), prepared by the Shuto
50
group, are illustrative. The carbocyclic 6 series again showed hydrolysis resistance; both series showed
interestingly that the 8ꢀamino analogs were agonists.
Pyrophosphate Modified Analogs. Comparatively speaking, pyrophosphateꢀmodified cADPR
analogs are rare (Figure 2). Phosphorothioate 13 (see Ring Deleted Analogs below and Figure 3) is one
3
2
example; cyclic adenosine triphosphate ribose (cATPR, 8, 20 times more potent than cADPR), and
33
two analogs where the pyrophosphate is replaced by methylenebisphosphonate, are others. The latter
analogs, cADPR[CH ] (9b, EC 111 nM versus 59 nM for cADPR) and 3ꢀdeazaꢀcADPR[CH ] (9c,
2
50
2
EC 302 nM versus ~1 nM for 3ꢀdeazaꢀcADPR 9a), as well as cATPR (8), were prepared via the
50
chemoenzymatic approach.
Ring Deleted Analogs. Some cADPR analogs (Figure 3) have been prepared that entirely lack and
Aꢀring, an Rꢀring, or both. Members of this class of analogs are the cyclic inosine diphosphate ribose
3
4
ethers (cIDPRE, 11, and their 8ꢀsubstituted congeners 12), phosphorothioate 13 (one of the
35
diastereomers is an agonist and the other an antagonist), cyclic triazole diphosphate ribose ether
3
6
(
cTDPRE, 14) , and cyclic inosine diphosphate diethers (cIDPDE, 15). These 'Rꢀring deleted' analogs,
Figure 3. Structures of ring deleted cADPR analogs.
5
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 6 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
prepared by the Zhang group, often retain agonistic activity. Especially remarkable is that the 'no ring'
analog 15, is only 3ꢀfold less potent an agonist than cADPR. The 'no ring' 15 notwithstanding, deletion
of the Aꢀring is deleterious to activity. Aꢀring deleted N9ꢀbutylꢀcIDPRE (16), like cIDPR 5 is an
27
inhibitor of the cADPR hydrolase activity of soluble human CD38 . The cAcyloDPR analog 17,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
prepared by the Potter group, was found to be inactive in the SUH system.
Rationale. Despite this wealth of pharmacological data, a key question in the SAR of the cADPR
system remains mostly unanswered: in making cADPR analogs, do the alterations to the cADPR
scaffold change the structure – specifically, its conformation – or is the structure largely preserved and
consequently what changes is its interaction with the receptor? If a new conformation is adopted, has an
analog become better or worse suited to the receptor binding site? Or is cADPR merely a scaffold upon
which groups can be deleted from or appended to, allowing for better (or worse) contacts in the as yet
22
unknown receptor binding site. In an extensive study by Moreau, a strong correlation was found
2
1
between activity and the conformation of the 'southern' ribose ring (the 'Aꢀring' in our terminology), at
3
7
least in the sea urchin system. Kudoh found cADPR (1a), cADPcR (6), and a 6ꢀanalog to be similar in
Aꢀring structure in an NOEꢀrestrained MD simulation. Less is known about the conformation of the
3
8,39
40
Rꢀring, and other than our earlier
and recent work on cADPR and its Aꢀring 2'ꢀOMe and 3'ꢀOMe
analogs, even less is known regarding the conformation about the pyrophosphate backbone and the
glycosidic bonds. Detailed structural information of the conformational landscape is lacking for the
o
o
o
entire cADPR family, and missing completely is any thermodynamic data (ꢁG , ꢁH , ꢁS ) on the
conformational equilibria – data could begin to explain why certain conformations are favored. Do
conformational changes in one part of cADPR drive changes in another, that is, are the molecular
motions coupled? This work, a variable temperature NMR spectroscopy investigation, aims to fill part
of that gap by conducting a detailed conformational analysis and thermodynamic study on the known
cADPR analogs 2'ꢀdAꢀcADPR (4b), 7ꢀdeazaꢀcADPR (4c), and 8ꢀBrꢀcADPR (4d).
ꢀ
RESULTS AND DISCUSSION
Synthesis. The known cADPR analogs 4b, 4c, and 4d were synthesized using a chemoenzymatic
approach consisting of 5'ꢀphosphorylation of an adenosine analog (1) to give an AMP analog (2),
formation of a pyrophosphate bond by activation of the 5'ꢀphosphate of 2 to nucleophilic attack by
+
NMN, thus providing the requisite NAD analog (3), and cyclization using the commercially available
Aplysia ADPRC (Scheme 1) to yield a cADPR analog (4). Starting from 7ꢀdeazaadenosine (1c) or
6
ACS Paragon Plus Environment

Page 7 of 38
The Journal of Organic Chemistry
1
2
3
4
5
8
ꢀbromoadenosine (1d), 5'ꢀselective phosphorylation was performed by the Yoshikawa method (TEP, 2
4
1
equiv. POCl , 0 °C; water was omitted). Upon completion of the reaction (2–4 hours) the reaction
3
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
mixture was slowly dripped via cannula into iceꢀcold, stirred, dry ether. After a brief centrifugation, the
solid pellet was stirred in iceꢀcold water. Ionꢀexchange chromatography (formic acid gradient) afforded
the AMP analogs as their free acids. In our hands the reaction was almost completely selective for
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
1
5'ꢀphosphorylation and only trace amounts of inorganic phosphate could be detected by P NMR. The
synthesis and purification of 7ꢀdeazaꢀ5'ꢀAMP (2c) proceeded uneventfully, but the phosphorylation of
8
–bromoadenosine (1d) to the corresponding 5'ꢀphosphate 2d proved problematic, as described below.
Monitoring the conversion of 1d to 2d by reversedꢀphase (RP) HPLC analysis indicated the
formation of two products with significantly different retention times (4.1 and 4.8 min with baseline
separation) by RP HPLC but rather similar UV spectra (λmax 262 and 264 nm, respectively). Partial
separation was achieved by ionꢀexchange chromatography (Sepharose Q resin eluted with formic acid).
The early eluting fractions (earlyꢀ2d) were sufficiently pure to be pooled, as were the later fractions
1
(laterꢀ2d), as judged by RP HPLC analysis of the individual fractions. The H NMR spectra of earlyꢀ2d
and laterꢀ2d were remarkably similar, with H2 signals at 8.36 ppm and 8.34ppm, H1' signals at 6.14 (d,
J = 5.6 Hz) and 6.13 ppm (d, J = 5.6 Hz), and H2' signals at 5.21 ppm (t, J = 5.5 Hz) and 5.25 ppm (t,
J = 5.5 Hz). Ultimately we determined the acidic conditions of the Yoshikawa phosphorylation
promoted an exchange of the 8ꢀbromo substituent for chlorine (a phenomenon we were unaware of at
4
2
13
the time) which had been previously noted in the literature. Comparison of the C NMR chemical
43
+
+
shifts for C8, and, ultimately LCMS data (earlyꢀ2d, 140 ppm, m/z MH 382; laterꢀ2d, 129 ppm, MH
4
4
m/z 426), showed that earlyꢀ2d was 8ꢀClꢀAMP 2e and laterꢀ2d was the desired 8ꢀBrꢀAMP 2d.
With the desired monophosphates in hand, we then focused on the coupling step – joining the
+
adenosine 5'ꢀmonophosphates 2bꢀ2d to NMN via pyrophosphate bond formation to form NAD analogs
3
9
16
3
bꢀ3d. As in our previous synthesis of cADPR analogs, we chose, as had others, to use the
4
5
Michelson procedure in which diphenyl phosphoryl chloride (DPPC, (PhO) POCl) is used to activate
2
1
the 5'ꢀphosphate of a nucleotide towards nucleophilic attack via the formation of a P ꢀnucleosideꢀ5'ꢀ
7
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 8 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 1. Synthesis of cADPR analogs
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Reagents and conditions: (a) (i) PO(OEt) , POCl , 0 °C, 2–4 h, (ii) H O, 0 °C (iii) triꢀnꢀoctylamine, CH OH; (b)
(PhO) POCl (diphenyl phosphoryl chloride, DPPC), triꢀnꢀbutylamine (TBA), dioxane, DMF; (c) (i) Ac NMN,
DMF, TBA, pyridine, (ii) NH , CH OH, 0 °C, 15 min; (d) ADPRC, HEPES buffer, pH 7.0, 2–3 h. adenosine
'ꢀmonophosphate (AMP, 2a), 2'ꢀdeoxyadenosine 5'ꢀmonophosphate (dAMP, 2b), and NAD (3a), each as their
3
3
2
3
2
2
3
3
+
5
39
free acids, were obtained from commercial sources. Ac NMN was prepared as described.
2
2
P ꢀdiphenyl pyrophosphate (mixed anhydride, Scheme 1). Nucleophilic attack by the 5'ꢀphosphate
group of a second nucleotide displaces the weak base leaving group diphenyl phosphate from the mixed
anhydride. In principle, either nucleotide could be the activated nucleotide and the other the attacking
nucleophile; in practice we chose to activate the AMP analogs 2bꢀ2d and use 2',3'ꢀdiꢀOꢀacetyl NMN
(
Ac NMN; the acetyl groups enhance the solubility of NMN in organic solvents) as the attacking
2
nucleophile, so as to minimize the exposure time of the somewhat fragile NMN glycosidic bond to the
reaction conditions. The free acids of AMP analogs 2bꢀ2d were thusly converted to their triꢀnꢀ
octylammonium salts to improve their solubility and reacted with DPPC. The mixed anhydride was
precipitated into ether, the supernatant decanted, dried briefly under vacuum, and a solution of Ac NMN
2
+
then added. Ionꢀexchange chromatography as for 5'ꢀAMP analogs 2c and 2d afforded NAD analogs
3bꢀ3d in modest yield.
+
Interestingly, the least challenging aspect of the synthesis was the enzymatic cyclization of NAD
analogs 3bꢀ3d to cADPR analogs 4bꢀ4d. Cyclization reactions were typically run on a 50 ꢀmol
(
30ꢀ40 mg) scale in ~ 6 mM HEPES buffer at room temperature with a substrate concentration of
4
6
+
~0.85 mM. All three NAD analogs were readily converted to the corresponding cADPR analogs over
the course of several hours, in conversions exceeding 60% as monitored by RP HPLC. Ionꢀexchange
chromatography on the SepharoseꢀQ column utilizing either TEAB or TEAA buffers gave the cADPR
analogs 4bꢀ4d, as their triethylammonium salts, in good yield.
8
ACS Paragon Plus Environment

Page 9 of 38
The Journal of Organic Chemistry
1
2
3
1
NMR Spectroscopy. H NMR (400 MHz) spectra for the triethylammonium salts of each of the cADPR
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
analogs 4bꢀ4d were recorded at 277, 298, 318, 338, and 353 K, though the lability of the R ring C1'ꢀN1
1
bond in 4b and 4d limited data collection at the higher temperatures. The H NMR spectra at 298 K of
cADPR analogs 4bꢀ4d, as well as the 600 MHz spectrum (298 K) of cADPR (4a) for reference, are
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
shown in Figure 4. (See Pages S23, S27, S28, and S34 for the spectra of each analog at each of the
temperatures.) Samples were 10–20 mM in D O containing TMSP as an internal standard. The spectra
2
were all firstꢀorder (confirmed by spectral simulation, not shown), allowing the threeꢀbond vicinal
3
coupling constants ( J) needed for the conformational analysis (see Population Analysis) using
PSEUROT to be extracted directly from the spectra.
The information needed to perform conformational analysis on cADPR/analogs resides in eight
signals (ten for 2'ꢀdA analog 4b) : the Aꢀring H1' (henceforth A1') and R1' (d, J ), A3' and R3' (dd, J ,
1'2
2'3'
and J ), A5' and R5' (dt, J , J , J ), A5" and R5" (ddd and dt, J , J , J ), plus A2" (J and
3
'4'
5'5" 4'5' 5'P
5'5" 4'5" 5"P
1'2"
J ) for deoxyꢀ4b. The 1'2', 2'3', and 3'4' (and 1'2" and 2"3' for 4b) couplings are used to determine the
2
"3'
furanose conformations via PSEUROT, and the 5'/5" couplings, via 'sum rules', to determine the γ– and
β–bond conformations (see Population Analysis). The H4'ꢀsignals (broad multiplets due to coupling
with H3', H5', H5", and sometimes phosphorus) and H2' signals (apparent triplet due to the near
1
equivalence of J J ) are less than ideal. A combination of routine 1D Hꢀ and phosphorusꢀdecoupled
1
'2 2'3'
1
1
31
H ( H{ P}) NMR were used to extract the coupling constants, as generally the needed signals were
3
8ꢀ40
well enough resolved. As we had done previously,
additional information was obtained through the
4
7
use of the 1D TOCSY experiment, useful especially at the higher temperatures as cADPR/analogs
began to decompose.
2
2
Two notable features of the 298 K spectra were apparent upon inspections. First, as has been noted,
was the unusually high chemical shift for the A2' signal in the spectra of 4a, 4c, and 4d (5.4–5.5 ppm
48
versus 4.6–4.8 for adenosine and AMP), characteristic of a synꢀoriented base. The second feature was
the consistency in the chemical shifts and coupling patterns across the analog series 4aꢀ4d. The R1' and
all 5'/5" signals, and to a slightly lesser extent the R2', R3', and R4' signals, were quite similar (See
Tables 1 and 2 for coupling constants), suggesting comparable conformations and environments. (The
differences in the Aꢀring chemical shifts can be attributed to the analog modifications, for example
9
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 10 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
D
A3
R3
A4
A1 R1
A2
R2 R4
A5”/R5’
R5” A5’
8-Br cADPR (4d)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
C
B
R4 R2
R5'
R1
A1
A2
A3
A5" R3 A4
R5” A5’
7-deaza cADPR (4c)
R4
A2'
A2"
*
* *
2'-dA cADPR (4b)
A1
R1
A3
R2
A5"/R3 R5' A4
R5” A5’
HOD
~
A
R5'/A4
A1 R1
A2
R2R4/A3
A5”/R3'
R5” A5’
cADPR (4a)
1
Figure 4. The 400 MHz 1D H NMR spectra of the cADPR analogs at 298 K. Panel A: cADPR (4a) reference
spectrum (600 MHz). Panel B: 2'ꢀdAꢀcADPR (4b). The peaks in the A2' signal marked with stars are from the
C satellite peak of the triethylammonium signal. Panel C: 7ꢀdeazaꢀcADPR (4c). Panel D: 8ꢀBrꢀcADPR (4d).
Primes in the 1', 2', 3', and 4' signals omitted for clarity. Note the generally overlapped A3'/R2'/R4' region in each
analog, the overlapped A5"/R3' signals of 2'ꢀdA cADPR (4b), and the overlapped A5"/R5'/A4' of 8ꢀBr cADPR
(4d). The 8ꢀbromo sample was contaminated with a small amount of its hydrolysis product, 8ꢀBr AppRib.
4
0
13
Samples were 10–20 mM in D O containing TMSP as an internal standard.
2
2'ꢀdeoxyribo versus ribo.) The R5' and R5" pair was consistently a doublet of triplets (dt), indicating
JR4'R5', JR4'5", JR5'P, and J , are similar in magnitude (~2 – 3 Hz). The A5' (dt) signal was similar in
5"P
appearance to R5' and R5", but A5" is a ddd, indicating at a minimum a conformation about the A ring
4
9
C4'ꢀC5' bond different than that of the R ring C4'ꢀC5' bond. These patterns remained largely intact
throughout the variable temperature study, although the magnitude of the coupling constants changed.
1
31
Accurate values for the J4'5' and J4'5" couplings were generally extracted from a H{ P} experiment (See
pages S22, S23, S26, S31, S33); the J and J couplings were extracted by subtracting the appropriate
5
'P
5"P
5
0
J4'5' or J4'5" and J5'5" coupling from their corresponding signals.
10
ACS Paragon Plus Environment

Page 11 of 38
The Journal of Organic Chemistry
1
2
3
4
As mentioned earlier, cADPR analogs 5–8, 9a, and, significantly here 7ꢀdeaza analog 4c, show
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
increased resistance to hydrolysis of the R1'ꢀN1 glycosidic bond (compare to cADPR; t1/2 = 2.5 d, pH 7,
1
3
7 °C). Pages S23, S27, S28, and S34 in the SI show the H NMR spectra of 4b, 4c, and 4d at the
highest temperature from which coupling data could be extracted. The 2'ꢀdA analog 4b and 8ꢀBr analog
d have completely broken down to their respective ADPꢀriboses (confirmed by HPLC; data not
shown), whereas 7ꢀdeaza 4c is largely intact.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
51
Conformational Analysis Using PSEUROT. First proposed by Kilpatrick et al. in 1947, the
pseudorotation concept is the basis for the conformational analysis of fiveꢀmembered rings such as the
furanose ring found in nucleosides and nucleotides. Fiveꢀmembered rings avoid planarity by displacing
3
2
ꢂ
ꢁ
one atom (envelope conformation, E, E, E, E) or two atoms (twist conformation, ꢀ, ꢀ) from the
2
3
ꢁ
ꢂ
ring plane defined by the remaining atoms, with atoms below ('exo') the ring plane indicated by
subscripts and those above ('endo') by superscripts. (Figure 5a). Nucleos(t)ide furanose rings, lacking
symmetry, allow for twenty possible 'ideal' or 'canonical' conformations – ten possible envelope forms
Figure 5. (a) Fiveꢀmembered ring furanose geometries and their descriptors. In Sꢀtype conformers, note how H1'
and H are anti while H and H are perpendicular, whereas in Nꢀtype conformers the opposite is true. (b)
2'
3'
4'
Endocyclic bond definitions. (c) The three canonical orientations of the γ– and β–bonds. (d) Atom, bond, and
furanose definitions for cADPR. See ref. 53 for details.
11
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 12 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and ten possible twist forms – which alternate between E and T forms as a furanose traverses a full
5
2
pseudorotational itinerary. The AltonaꢀSundaralingam (AS) formalism assigns a numerical value, the
phase angle of pseudorotation' ('phase angle', P) to each of the twenty conformers. A second parameter
the maximum puckering amplitude, ψm, – describes by how much the ring deviates from planarity. (A
'
–
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
o
o
planar ring would have ψm = 0 and idealized cyclohexane 60°; typical ψm's in nucleos(t)ides are 25 –
o
o
o
o
3
4
5 .) The allowed values of P are 0 – 360 ; by analogy to a compass rose P = 0 ( T ; C2'ꢀexoꢀC3'ꢀ
2
o
2
endo) is 'north' and P = 180 ( T ; C2'ꢀendoꢀC3'ꢀexo;) is 'south'. These relationships are shown in
3
Figure 5a. It should be stressed that there is no highꢀenergy barrier between adjacent 'conformers' on the
pseudorotational wheel and that the notion of twenty 'stops' in pseudorotational space is merely a
descriptive convenience. A nucleos(t)ide can adopt any value of P.
5
3
The geometry of a furanose ring can also be based on the values of the endocyclic torsion angles
ν –ν (Figure 5b); indeed, this is the basis of AS formalism. The relationships are:
0
4
ꢄꢅ ꢇꢅ ꢉꢊꢄꢅ ꢇꢅ ꢉ
ꢆ
ꢈ
ꢋ
ꢌ
tan ꢃ = ꢁ
(1)
(2)
(3)
ꢅꢍꢄꢎꢏꢐ ꢂꢑ°ꢇꢎꢏꢐ ꢒꢁ°ꢉ
ꢕꢍ
ꢓ =
ꢔ
ꢖꢗꢎ ꢘ
ꢛꢜꢙ
νꢙ = ꢓ cos ꢚꢃ + ꢞ ꢟꢟꢟꢟꢟꢟꢟꢟꢟꢟꢟꢟwhereꢟꢠ = 0, 1, 2, 3, 4
ꢔ
ꢝ
The elegance of the AS formalism resides in eq 1, as it reduces five endocyclic torsion angles to a single
parameter, P, that simply describes which atom(s) are puckered out of the ring plane. Any individual
endocyclic torsion angles can be reꢀcreated with eq 3.
5
4
For reasons both steric and stereoelectronic not all of the 20 canonical forms are populated. The
furanose ring of a nucleos(t)ides tend to localize to either a 'northꢀtype' (N) or a 'southꢀtype' (S)
conformation. Under the generally true case of a twoꢀstate N ꢀ S equilibrium, five parameters are
needed to describe the system: the P of each conformer (P and P ), the maximum puckering amplitudes
N
S
ꢟN
ꢟS
of each (ꢓ and ꢓ ), and the mole fraction (of either conformer). For cADPR and its analogs, a
m
m
complete description of the conformational landscape requires knowledge of the conformation of both
furanose rings as well as the Aꢀring and Rꢀring γ– and β–bond conformations (Figure 5c and 5d). These
χ
χ
and R,
six conformationally flexible subunits, plus the conformation about the glycosidic bonds
A
2
2
(
Figure 5d) drive the conformation. The aforementioned study of cADPR analogs examined only the
Aꢀring conformation, and only to the extent of determining if the Aꢀring furanose was Nꢀtype or Sꢀtype.
12
ACS Paragon Plus Environment

Page 13 of 38
The Journal of Organic Chemistry
1
2
3
ꢟN
ꢟS
The equilibrium ratios and precise values for P , P , ꢓ , ꢓ and the γ– and β–bond conformations of
N
S
m
m
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the cADPR analogs 4bꢀd can be obtained through a more rigorous analysis of coupling constants, and to
our knowledge this is the first such study to do so in significant detail.
The furanose ring coupling constants for cADPR analogs 4bꢀd are summarized in Tables 1 and 2.
Also included therein are the pseudorotation parameters for 4bꢀd as determined by PSEUROT, as well
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
0
as the pseudorotation parameters for cADPR itself (4a), determined previously. The coupling
1
8
constants from Tables 1 and 2 were the input for version 6.3 of the PSEUROT program. In addition to
coupling constants related to the furanose conformation (Aꢀ and Rꢀring J , J , and J , for analogs 4c
1
'2' 2'3'
3'4'
and 4d, plus for analog 4b J , and J ), the user provides as input to PSEUROT substituent
1
'2"
2"3'
electronegativities, an initial guess as to the values of the five parameters that describe a twoꢀstate NꢀS
N
S
equilibrium (P , P , ꢓ , P , ꢓ , and the S mole fraction), and a set of parameters to correlate the
N
S
m
S
m
1
1
endocyclic ring atom torsions to the exocyclic Hꢀ H torsions (the soꢀcalled 'A' and 'B' parameters).
While the conformation of a furanose ring is described (Figure 5b and eq 1) by the endocyclic ring
1
1
1
1
torsions, the observed Hꢀ H coupling constants (J ) depend on the exocyclic Hꢀ H torsion angles
obs
(
Figure 5a). Combined with the initial guess of P and P , PSEUROT uses the A and B parameters to
N S
create exocyclic torsion angles (e.g., H1'ꢀC1'-C2'ꢀH2') by 'building in' hydrogen atoms onto the
corresponding endocyclic torsions (e.g., ν = O4'ꢀC1'-C2'ꢀC3') specified by P. With the exocyclic
4
torsion angles thus specified, a Karplus equation within PSEUROT calculates a predicted coupling
constant, J , which it compares to J . One of the reasons we chose these analogs to study first is that
calc
obs
they contain only riboꢀ and deoxyribofuranose rings, for which the A and B values were specifically
5
5
parametrized. (Furanoses different than riboꢀ and deoxyriboꢀ generally require adjustment of the
default A and B values or values specifically derived for the molecule under investigation.) PSEUROT
ꢟN
then calculates the expected coupling constants for such a mixture and iteratively adjusts P , P , ꢓ ,
N
S
m
ꢟS
ꢓ , and the S mole fraction so as to produce the best fit between the calculated and observed coupling
m
constants.
13
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 14 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
1
1
31
a–c
Table 1. 400 MHz Hꢀ H and Hꢀ P Coupling Constants (J, Hz) of the cADPR Aꢀring for 4bꢀd and PSEUROT
6
.3 output (degrees) for cADPR analogs 4bꢀd
Aꢀring T (K) J1’2’ J1’2" J2’2"
J2’3’ J2"3’ J3’4’
J4’5’
J4’5” J5’5” J5’P
J5” P
2
2
77
98
6.85 6.96 14.20 6.30 3.58 2.83 2.84 7.45 10.81 3.14 3.45
6.68 7.01 14.19 6.31 3.78 2.97 2.97 7.21 10.89 3.37 3.51
6.51 7.07 14.22 6.40 3.96 3.29 3.12 7.00 10.89 3.54 3.66
6.19 7.22 14.32 6.44 4.22 3.42 3.37 6.72 10.81 4.37 4.07
nd
4
2
b
'ꢀdA
318
338
353
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
d
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
e
e
2
2
318
77
98
6.38
6.28
6.13
5.97
5.84
5.06
5.12
5.18
5.28
5.28
2.65 2.64 7.23 10.9 nd
nd
f
2.81 2.52 7.00 11.0 3.57 3.48
2.95 2.56 6.74 11.0 3.74 3.46
3.09 2.71 6.69 11.0 3.78 3.51
3.20 2.81 6.47 11.0 3.98 3.58
4
c
7ꢀdeaza 338
3
53
g
2
2
318
338
77
98
5.41
5.37
5.27
5.15
nd
5.13
5.20
5.24
5.28
nd
3.44 2.34 6.17 11.14 2.93 4.47
3.50 2.29 6.10 11.10 3.31 4.34g
4d
3.53 2.28 5.92 11.00 3.61 4.58
8
ꢀBr
3.57 nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
3
53
nd
h
t
N
S
PS 6.3
PN
22.5
359.1
352.2
20.3
ψ
35
38
35
35
PS
ꢟꢓ
m
RMS
m
j
k
4a
4b
177.1 31.5
170.7 27.3
159.1 30.7
173.7 28.8
0.099
0.128
0.062
0.045
4
4
c
d
a
b
c
Values ± 0.04 Hz. Only J , J , and J are used in PSEUROT. Furanose ring values are from the normal 1D
1
'2' 2'3'
3'4'
1
1
31
H spectrum, J4’5’, J4’5”, and J5’5” values from the H{ P} spectrum. J determined by subtraction of
5’P
Σ(J + J5’5”) from Σ(J4’5’ + J5’5” + J5’5P), the latter being the width of 5'ꢀsignal. J determined analogously.
nd = not determined due to sample decomposition. Value from 1D TOCSY with irradiation at A . Value from
A . Value determined directly from A5". Errors in PS6.3 output: cADPR (4a) P ± 4.3 , P ± 2.6 ,ꢟꢓ ꢟ ± 0.7 ;
4
’5’
5"P
d
e
f
2
'
g
h
o
o
ꢟS
o
4
'
N
o
S
o
m
o
o
ꢟS
o
ꢟS
o
2
'ꢀdA cADPR (4b): P ± 4.2 , P ± 2.3 ,ꢟꢓ ꢟ ± 0.6 ; 7ꢀdeaza cADPR (4c): P ± 2.1 , P ± 1.6 ,ꢟꢓ ꢟ ± 0.4 ; 8ꢀBr
m m
o o o t
N
S
N
S
ꢟS
cADPR (4d): P ± 2.4 , P ± 1.6 ,ꢟꢓ ꢟ ± 0.4 . Rootꢀmean square error of the observed vs. PSEUROTꢀcalculated
N
j
S
m
k
J values in Hz. Data for cADPR (4a) from ref. 40. Underlined values restrained in PSEUROT 6.3. Note the
irregularity in the J5"P values for 4d (boxed).
14
ACS Paragon Plus Environment

Page 15 of 38
The Journal of Organic Chemistry
1
2
3
4
1
1
1
31
a–c
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 2. 400 MHz Hꢀ H and Hꢀ P Coupling Constants (J, Hz) of the cADPR Rꢀring for 4bꢀd and PSEUROT
b
6.3 output (degrees) for 4bꢀd
Rꢀring
T (K) J1’2’
J2’3’
5.07
5.05
5.08
5.13
nd
J3’4’
2.11
2.17
2.25
2.36
nd
J4’5’
2.10
2.13
J4’5”
J5’5” JP5’
JP5”
3.42
3.70
3.89
nd
277
98
318
338
4.00
4.04
4.06
4.02
2.05 11.90 2.18
2.43 11.96 2.14
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
4
b
2.16d 2.52 12.02 2.39
d
2
'ꢀdA
~2.39 ~2.39 nd
nd
nd
e
3
53
nd
nd
nd
nd
nd
2
2
318
77
98
4.00
4.04
4.06
4.07
4.06
5.00
5.02
5.03
5.05
5.04
2.45
2.54
2.61
2.67
2.74
nd
nd
nd
nd
nd
2.03
2.12
2.15
2.08
2.49 12.0
2.59 12.0
2.70 12.1
2.74 12.1
2.39
2.53
2.60
2.73
3.42
3.45
3.58
3.73
4
c
7
ꢀdeaza 338
353
2
2
318
338
3
77
98
3.98
4.03
4.07
4.10
nd
5.06
5.08
5.10
5.09
nd
2.23
2.23
2.28
2.28
nd
2.12
2.05
2.16
nd
2.20 12.0
2.23 12.0
2.25 12.1
nd
nd
2.12
2.59
2.39
nd
3.46
3.65
4.00
nd
f
4
d
8ꢀBr
nd
nd
53
nd
nd
nd
g
N
h
S
PS 6.3
PN
ꢓ
m
PS
ꢓ
RMS
m
i
j
4
a
340.0 40
337.4 40
343.2 40
337.2 40
215.8 36.6
216.2 36.6
212.2 34.5
215.4 36.2
0.031
0.041
0.046
0.030
4b
4
4
c
d
a
b
c
Values ± 0.04 Hz. Only J , J , and J are used in PSEUROT. Furanose ring values are from the normal 1D
H spectrum, J4’5’, J4’5”, and J5’5” values from the H{ P} spectrum. J determined by subtraction of
1
'2' 2'3'
3'4'
1
1
31
5’P
Σ(J + J5’5”) from Σ(J + J5’5” + J5’5P), the latter being the width of 5'ꢀsignal. J determined analogously.
4’5’
4’5’
5"P
d
e
f
Value from R . nd = not determined due to sample decomposition. Value from 1D TOCSY with irradiation at
4
'
g
o
o
ꢟS
o
o
R /R . Errors in PS6.3 output: cADPR (4a) P ± 0.8 , P ± 0.3 ,ꢟꢓ ꢟ ± 0.3 ; 2'ꢀdA cADPR (4b) P ± 1.2 , P ±
2
'
4'
N
o
S
o
m
N
o
S
o
ꢟS
o
ꢟS
o
0
0
.5 , ꢓ ꢟ ± 0.5 ; 7ꢀdeaza cADPR (4c): P ± 1.1 , P ± 0.5 ,ꢟꢓ ꢟꢟ ± 0.4 ; 8ꢀBr cADPR (4d): P ± 0.9 , P ±
m
N
S
m
N
S
o
ꢟS
o
h
i
.4 ,ꢟꢓ ꢟꢟ ± 0.3 . Rootꢀmean square error of the observed vs. calculated J values in Hz. Data for cADPR (4a)
m
j
from ref. 40. Underlined values restrained in PSEUROT 6.3. Note the irregularity in the J4 and J5'P values for
'5'
4d (boxed).
15
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 16 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
To determine the values of five parameters requires five observables (coupling constants);
ribofuranoses, with only J , J , and J are thus said to be underdetermined, meaning their N ꢀ S
1'2' 2'3'
3'4'
equilibria cannot be completely described. Generally being interested in the major conformer, one could
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
56
restrain the P and ꢓ of the minor conformer. It has been shown that the coupling constants are much
m
less sensitive to changes in ꢓ than to changes in P. If puckering amplitude information is available
m
ꢟN
from other sources (e.g. a crystal or QM structure), such data can be used to restrain either or both ꢓ ,
m
ꢟS
and ꢓ . Regrettably, reliable crystal structure data is not available for either cADPR or analogs 4b–4d.
m
The only crystal structure of cADPR in the Cambridge Structural Database (CSD Entry YINYID) is of
the free acid and was deposited without coordinates. There was enough information in the text of the
5
7a
o
o
associated paper to estimate a P ~175 for the Aꢀring and a P ~208 for the Rꢀring, but the free acid
S
S
raises concerns regarding the γ– and β–bond conformations in the crystal. Another study examined the
crystal structure of cADPR bound to CD38, but retrieval of the cADPR ligand from the protein structure
5
7b
(
Protein Data Bank ID: 2O3Q) returns an Aꢀring with the xyloꢀconfiguration rather than ribo.
A third
5
7c
2
report, this time with cADPR bound to the Aplysia cyclase, has a cADPR structure with Aꢀring E
+
1
t
(
P ~162°)/γ ꢀbond conformations and Rꢀring E (P ~306°)/γ ꢀbond conformations; the opposite
γꢀconformations were found here
The option we ultimately chose was to collect NMR spectra over a range of temperatures sufficient to
3
8–40
obtain significant changes in the coupling constants. Based on our earlier work
and reluctant to use
ꢟN
the crystal structure data, we restrained the ꢓ of the Aꢀring and Rꢀring minor conformers to 35° (38°
m
for 2b) and 40°, respectively. All other parameters were freely optimized. Assuming the identities of
the two conformers involved in the NꢀS equilibrium are temperature independent, a change in the
coupling constants reflects a change in their equilibrium ratio (see Thermodynamic Parameters). Each
temperature represents not only an additional PSEUROT observable but also the opportunity to perform
o
o
van't Hoff analyses, from which ꢁH and ꢁS can be determined.
Population Analysis. The PSEUROT output is summarized in Tables 1 and 2, and the changes in the
north/south ratio as a function of temperature shown in Table 3. Also included in Tables 1 and 2 are
couplings of each 5'ꢀ and 5"ꢀproton to their respective 4'ꢀprotons (related to the γ–bond conformations)
16
ACS Paragon Plus Environment

Page 17 of 38
The Journal of Organic Chemistry
1
2
3
4
a
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 3. Equilibrium populations for the furanose rings, γ–bonds, and β–bonds in cADPR (4a) and analogs 4bꢀd
Aꢀring
γ ꢀγ
Rꢀring
γ ꢀγ
4:96
5:95
6:94
7:93
nd
t
+
–
t
t
+
+
t
T (K)
277
298
NꢀS
29:71
32:68
35:65
35:65
36:64
0.92
β ꢀβ
NꢀS
26:74
27:73
27:73
26:74
26:74
0.04
β ꢀβ
58:42
57:43
56:44
54:46
53:47
0.97
9:91
11:89
12:88
13:87
nd
2:98
4:96
7:93
8:92
nd
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
a
318
b
cADPR 338
53
3
2
c
R
0.99
0.98
0.94
277
98
28:72
30:70
33:67
36:64
nd
69:31
68:32
67:33
67:33
nd
8:92
9:91
11:89
17:83
nd
24:76
24:76
25:75
26:74
nd
6:94
10:90
11:89
12:88
nd
3:97
4:96
6:94
nd
nd
0.99
2
4
b
318
338
2
'ꢀdA
3
53
2
R
0.98
0.98
0.88
0.85
0.88
277
98
318
ꢀdeaza 338
53
25:75
26:74
28:72
30:70
32:68
0.96
65:35
61:39
59:41
60:40
59:41
0.80
nd
28:72
29:71
29:71
30:70
30:70
0.92
nd
nd
2
10:90
11:89
11:89
13:87
0.92
10:90
11:89
13:87
13:87
0.88
4:96
5:95
6:94
7:93
0.98
4
c
7
3
2
R
277
98
33:67
34:66
34:66
35:65
nd
51:49
49:51
47:53
nd
nd
0.97
12:88
13:87
16:84
nd
nd
0.95
26:74
25:75
26:74
25:75
nd
8:92
7:93
8:92
nd
nd
0.29
3:97
6:94
7:93
nd
nd
0.89
2
4
d
318
338
8
ꢀBr
3
53
2
R
0.90
0.21
a
58a
N:S populations from PSEUROT 6.3. Aꢀring γ–bond population: f = [13.3 – (J + J )]/9.7 (eq 4). Rꢀring
γ+
4'5'
4'5"
5
8a
58a
γ–bond: f = [13.3 – (J + J )]/9.4 (eq 5). Aꢀring β–bond: f = [26.4 – (J + J )]/21.4 (eq 6). Rꢀring β–
γ+
4'5'
4'5"
βt
5'P
5"P
5
8b
bond: f = [25.5 – (J + J )]/20.5 (eq 7). Equations 4 and 6 were used as found in the associated references.
βt
5'P
5"P
Equations 5 and 7 were modified slightly (see the Supporting Information) to improve the fit between the
5
8
observed and backꢀcalculated coupling constants; the backꢀcalculated coupling constants also allowed for
b
c
d
identification of the minor conformer. Data for cADPR from ref. 40. Value not determined. Correlation
coefficient from the van't Hoff plots (Figure 6).
17
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 18 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and adjacent phosphorus atom (related to the β–bond conformations). The γ– and β–bond
[58
conformations and populations were elucidated using the 'sum rules' shown in Table 3. In some cases
these sum rules were slightly modified (see the Supporting Information) so as to provide a better fit
between the backꢀcalculated and observed coupling constants.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
We were concerned that the macrocyclic nature of the cADPR analogs would restrict their
conformational freedom, but gratifyingly the majority of the coupling constants (Tables 1 and 2) showed
a regular, albeit small, dependence on temperature. The exception was the Rꢀring furanose (Table 2),
which showed either essentially no variation (J1'2' and J2'3' in 4b/4c/4d, and J3'4' in 4d) or very minor
variation (J3'4' in 4b and 4c) in the couplings. All Aꢀring furanoses in 4b, 4c, and 4d favored south
ꢟS
conformations, generally by a factor of 2:1, with P = 166° ± 7°, and ꢓ ꢟ = 29° ± 2° (somewhat flatter
S
m
than their free nucleoside counterparts). As we found for cADPR, the Aꢀring γ–bonds favored the trans
t
conformation (percent γ at 298 K: 4a, 57%, 4b, 68%, 4c, 61%, but in 4d only 49%) and all β–bonds
t
were heavily biased (83–92%) towards the β conformer. For the Rꢀring furanoses, also similar to that
ꢟS
of cADPR, the favored conformers were all southꢀtype (N:S ~ 1:3), with P = 214° ± 2°, ꢓ ꢟ =
S
m
t
3
5° ± 1°, a highly populated (~90% or more) β state, but now an almost exclusively (>87–98%)
+
59
populated the γ state. The favored conformers of the triethylammonium salts s of 4a–d all show a
t
remarkable degree of similarity in both identity and population, with an S-type furanose, γ bond, and
t
+
t
β bond for the A-ring and an S-type furanose, γ bond, and β bond for the R-ring. Many of the solution
5
7a
conformations determined herein compare quite well to those of the published crystal structure of the
free acid of cADPR (Aꢀring P ~ 175°, β = –138°, Rꢀring P ~ 208°, β = +160°), though both γ–bonds in
S
S
4
0
the crystal structure were trans, and to our recent preliminary MD simulation of cADPR. Despite
these surface similarities in cADPR 4a and cADPR analogs 4bꢀd, analysis of the thermodynamic
parameters reveals some important distinctions relevant to the SAR of the cADPR system (next section).
Thermodynamic Parameters. Data for cADPR 4a is from reference 40. Inspired in part by the work of
the Chattopadhyaya group, who have performed extensive conformational analysis and thermodynamic
5
4a,60
studies on monomeric nucleosides and nucleotides, as well as DNA and RNA,
we wished to
establish if the same methodology could be applied to cADPR and its analogs. Having determined the
conformer populations in 4b–4d we next sought to determine the thermodynamics (ꢁG°, ꢁH°, and ꢁS°)
18
ACS Paragon Plus Environment

Page 19 of 38
The Journal of Organic Chemistry
1
2
3
for each of the conformational equilibria. Qualitatively, the expression ꢁG° = ꢁH° – TꢁS° predicts the
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
conformer whose population increases with increasing temperature is the more disordered conformer.
4
0
As was found for cADPR (1a) itself, it was always the minor conformer whose population increased
+
–
+
with temperature, thus the minor conformers (Nꢀtype furanoses, β or β bonds, Aꢀring γ bond, Rꢀring
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
t
–
40
γ or γ ) are more disordered. It thus follows that, as was the case in cADPR, the major conformers are
favored because of enthalpy. The exception to this trend was the Rꢀring furanoses in 4a, 4b, 4c, and 4d,
whose N:S ratios were either essentially unchanged, or changed marginally, with increasing
temperature. The van't Hoff equation (vide infra) predicts a temperatureꢀinsensitive population such as
the Rꢀring furanoses in 4a–4d to have a zero or small ꢁH° for their NꢀS equilibrium and thus that the
Rꢀring major conformer (here, Sꢀtype) is favored by entropy.
The population data from Table 3 formed the basis of a more thorough analysis using van't Hoff plots
(ln K as a function of 1000/T, Figure 6), revealing the precise values of ꢁH°, and ꢁS° for each
equilibria; this data is shown in Table 4. The majority (15 of 18) of the plots in Figure 6 showed
2
excellent linearity as reflected in correlation coefficients (R ) of approximately 0.90 or higher (Table 3).
2
The moderately good fit (R = 0.80) for the A–ring γ–bond in 7–deaza analog 4c can likely be attributed
the sum rule used to derive the populations, as the coupling constants changed with temperature in a
consistent manner. The R–ring furanose and γ–bond in 8–bromo analog 4d displayed the smallest (and
somewhat inconsistent) temperatureꢀdependent variations in coupling constants, resulting in essentially
2
unchanging populations and flat van't Hoff plot of poorer fit (R = 0.21 and 0.29, respectively). Thus,
other than (1) the exceptional Rꢀring N ꢀ S equilibria, which is indeed driven largely (in 4a and 4d) or
t
+
significantly (in 4b and 4c) by an entropy term favoring the Sꢀconformer and (2) the Aꢀring γ ꢀ γ
t
+
transition, where γ is only slightly favored over γ and ꢁH° and ꢁS° are roughly comparable, the
remaining conformational equilibria are generally characterized by a negative ꢁH° term that outweighs
the TꢁS° component by a factor of approximately two. This confirms that in the majority of cases the
higher populated conformer is favored for reasons of enthalpy. Exceptions are noted in the subsequent
discussion.
19
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 20 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1.8
.6
1.4
.2
1.0
3.8
3.4
3.0
3
.6
1
3.1
2.6
2.1
1.6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2.6
2.2
1.8
1.4
0.8
0.6
0.4
0.2
0.0
1.1
.6
0
0.1
-0.4
-0.9
1.0
1000/T (K)
A, furanoses
1000/T (K)
B, β-bonds
C, γ-bonds
1000/T (K)
Figure 6. van't Hoff plots (ln χ /χ , χ /χ , and χ /χβ+/–, as a function of 1000/T) for 2'ꢀdA cADPR (4b, •, ),
ꢀdeaza cADPR (4c, ꢀ, ꢁ), and 8ꢀBr cADPR (4d, ꢁ, ꢂ). Filled symbols represent the Aꢀring portion of each
cADPR and open symbols the Rꢀring portion. The slope of each line is –ꢁH /R and the yꢀintercept is ꢁS /R. For
each conformation the slope, intercept, and correlation coefficient, R , respectively, are: Panel A: furanose
Aꢀrings 4b (•) 0.57, –1.09, 0.98; 4c (ꢀ) 0.44, –0.48, 0.96; 4d (ꢁ) 0.12, 0.26, 0.90. Furanose Rꢀrings: 4b ( ) 0.17,
0.56, 0.85; 4c (
c (ꢀ) 0.44, 0.72, 0.92; 4d (ꢁ) 0.70, –0.48, 0.95. Rꢀring βꢀbond: 4b ( ) 1.72, –2.68, 0.99; 4c (
.98; 4d (
S
N
γ+ γt
βt
7
o
o
2

ꢁ
) –0.13, 0.49, 0.92; 4d (
ꢂ
) –0.05, 1.23, 0.21. Panel B: Aꢀring βꢀbond: 4b (•) 1.26, –1.99, 0.88;

4
0
4
0
ꢁ
) 1.17, –0.72,
ꢂ
) 1.99, –3.75, 0.89. Panel C: Aꢀring γꢀbond: 4b (•) –0.13, –0.32, 0.98; 4c (ꢀ) –0.30, 0.54, 0.80;

d (ꢁ) –0.27, 0.95, 0.97. Rꢀring γꢀbond: 4b ( ) 1.25, –1.83, 0.88; 4c (
.29. Underlined values indicate a questionable fit.
ꢁ
) 0.65, 0.02, 0.88; 4d (
ꢂ
) 0.24, 1.71,
20
ACS Paragon Plus Environment

Page 21 of 38
The Journal of Organic Chemistry
1
2
3
A-Ring Furanoses. The N ꢀ S equilibria in the in the Aꢀring furanoses of cADPR (4a), 2'ꢀdeoxy 4b,
and 7ꢀdeaza 4c are characterized by a larger negative ꢁH° value (–0.97 ± 0.16 kcal/mol) and a smaller
negative TꢁS° value (–0.46 ± 0.18 kcal/mol), somewhat smaller than that of adenosine (Ado; ꢁH°, TꢁS°
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
6
0a
–
1.05, –0.62 kcal/mol). The Chattopadhyaya group found that Ado, over the same temperature range
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
as our study, experienced a broader change in equilibrium composition (10% decrease in Sꢀtype) with
increasing temperature compared to Aꢀring of cADPR (7% decrease in Sꢀtype). One might expect that
cADPR, being cyclic, has less conformational freedom than its acyclic adenosine counterpart. One
stereoelectronic effect that drives the NꢀS equilibrium is the anomeric effect (AE). The AE (donation
of an O4' lone pair into the C1'ꢀN9 antibonding σ* orbital) increases with increasing electron demand of
the nucleobase, or with protonation of the nucleobase. The AE is generally more favorable in Nꢀtype
6
0a–d
60b,c
conformations.
In two studies
of the effect of pH on the NꢀS equilibrium, the same group
found a dramatic decrease in the amount of Sꢀtype (or increase in Nꢀtype) conformer at pH values at or
+
below the pK of the adenine N1 atom. The NꢀS ꢁH° for protonated adenosine (Ado
•
H ) had dropped
a
to a mere –0.05 kcal/mol, which they attributed to differing strengths of the AE in the neutral (weaker)
and protonated (stronger) forms. The N6ꢀamino group of cADPR/analogs is protonated at physiological
2
8,61
pH,
and yet their Aꢀring NꢀS ꢁH° values are hardly diminished relative to neutral Ado. It is
curious that the Aꢀring furanoses in 4a, 4b, and 4c 'resist' the AE and seem thermodynamically to be
more similar to neutral Ado.
The ꢁH° for the Aꢀring NꢀS equilibria in 8ꢀBr 4d, on the other hand, is significantly diminished
(ꢀ0.24 kcal/mol) compared to 4a, 4b, and 4c, suggesting perhaps a steric interaction between the
8
ꢀbromo substituent and the Aꢀring furanose in the Sꢀform. Interestingly, the 8ꢀBr 4d NꢀS equilibria is
the only Aꢀring furanose with a positive TꢁS° term, so while the – ꢁH° contribution to ꢁG° has been
reduced, apparently a more disordered Sꢀtype conformer compensates.
R-Ring Furanoses. The N ꢀ S equilibria in the Rꢀrings of 4a, 4c, and 4d are characterized by small
values of ꢁH° and positive TꢁS° terms that favor Sꢀtype conformers, demonstrating that indeed the
Rꢀring N ꢀ S equilibria have a significant entropy component. These R-rings are inherently more
disordered than their A-ring counterparts. This is somewhat surprising; in a comprehensive study of
6
0a
nine dideoxyꢀ, deoxyꢀ, and ribo nucleosides Plavec et al. found only one (deoxycytidine) whose
21
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 22 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Table 4. Thermodynamic parameters at 298 K for the furanose rings, γ–bonds, and β–bonds in cADPR
A Ring
R Ring
ꢁS° TꢁS°
b
b
ꢁH°
ꢁS° TꢁS° ꢁG° Pop.
ꢁH°
ꢁG° Pop.
4a
4b
4c
4d
NꢀS
NꢀS
NꢀS
NꢀS
–0.81 –1.20 –0.36 –0.45 32:68
–1.13 –2.17 –0.65 –0.49 31:69
–0.88 –0.95 –0.28 –0.60 27:73
–0.24 +0.52 +0.15 –0.40 34:66
+0.04 +2.15 +0.64
–0.33 +1.12 +0.33
–0.25 +0.97 +0.29
+0.10 +2.45 +0.73
–0.61 26:74
–0.67 24:76
–0.54 29:71
–0.63 26:74
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
c
Ado
NꢀS
NꢀS
–1.05
–0.05
–0.62 –0.43 33:67
+0.10 –0.15 44:56
+
c
Ado H
•
t
+
4
4b
4c
a
γ ꢀ γ
+0.52 +1.23 +0.37 +0.16 57:43
+0.25 –0.64 –0.19 +0.44 68:32
+0.60 +1.07 +0.32 +0.29 62:38
+0.54 +1.89 +0.56 –0.02 51:49
–1.81 –0.24 –0.07
–2.49 –3.64 –1.08
–1.30 +0.04 +0.01d –1.31 10:90
–0.47 +3.39 +1.01
ꢀ1.74
–1.41 8:92
5:95
t
+
+
+
γ ꢀ γ
t
γ ꢀ γ
t
4
d
γ ꢀ γ
–1.48 8:92
ꢀ/+
t
4
a
β ꢀ β –1.03 +0.80 +0.24 –1.27 10:90
β ꢀ β –2.50 –3.96 –1.18 –1.32 10:90
–4.51 –8.72 –2.60
–3.42 –5.32 –1.58
–2.33 +1.43 –0.43
–3.96 –7.46 –2.22
–1.91 4:96
–1.84 4:96
–1.91 4:96
–1.74 5:95
o
ꢀ/+
t
4
b
ꢀ/+ t
4
c
β ꢀ β –0.88 +1.44 +0.43 –1.31 10:90
ꢀ/+ t
4
d
β ꢀ β –1.39 –0.96 –0.29 –1.10 13:87
a
b
o
o
o
ꢁ
H , TꢁS , and ꢁG in kcal/mol, ꢁS in cal/mol K. Underlined values denote the major contributor to ꢁG .
Populations recalculated based on the calculated ꢁG ; compare to Table 3. Values in bold are discussed in the
A Model for cADPR section. Data from ref. 60b,c. Ado = adenosine, Ado
Poor fit in the van't Hoff plot; R = 0.29.
o
c
+
•
H = N1ꢀprotonated adenosine.
d
2
o
o
+
ꢁ
H value did not make a significant contribution to its ꢁG . But now the comparison to Ado
•
H is
+
more appropriate. The positive charge in Ado H led to a 0.72 kcal/mol diminishment of a TꢁS° term
•
that originally favored Nꢀtype neutral Ado (TꢁS° = –0.62 kcal/mol in Ado and +0.10 kcal/mol in
+
Ado
•
H ). The positive charge in cADPR/analogs is closer to the Rꢀring than to the Aꢀring and thus
reasonably exerts more influence on the Rꢀring TꢁS° term. Whether the ꢁH° value for the N ꢀ S
equilibria in the 2'ꢀdeoxyꢀ4b and 7ꢀdeazaꢀ4c is truly negative, though, will require further study on
additional analogs over greater temperature ranges. A van't Hoff plot lacking significant slope is
extremely sensitive to additional data points, but it is safe to say, based on the general lack of effect of
temperature on the Rꢀring N:S ratios that the ꢁH° term is small and that the TꢁS° term dominates the
+
N ꢀ S equilibria. While thermodynamically the Rꢀrings resemble Ado
•
H (comparable ꢁH° and TꢁS°
terms), neither the Aꢀring nor Rꢀring populations shift much from their approximate 1:3 N:S ratios,
suggesting again some resistance to conformational change.
22
ACS Paragon Plus Environment

Page 23 of 38
The Journal of Organic Chemistry
1
2
3
γ–Bonds. In contrast to all the other conformationally mobile rotatable bonds, where a distinct
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
t
+
preference one conformer is observed, the Aꢀring γ–bonds exists as a mixture of the γ and γ rotamers,
with a slight preference for the transꢀrotamer. The ꢁH° and TꢁS° terms are more evenly matched, with
cADPR (1a), 7ꢀdeaza analog 4c, and 8ꢀbromo analog 4d possessing positive ꢁH° values that disfavor
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
+
the γ rotamer but positive TꢁS° values that favor it. The TꢁS° term, however, for 4b is negative, and
+
the γ state is disfavored both enthalpically and entropically. Of all the analogs, 4b has the highest
t
populated Aꢀγ state, ~69%. The situation is different for the Rꢀring γ–bonds, most of which show a
+
strong (>90%) ꢁH°ꢀdriven preference for the γ rotamer. (While the Rꢀring γ–bond in cADPR analog
+
2
4
d exists almost exclusively as the γ rotamer, the poor fit in the van't Hoff plot (R = 0.29) makes it
impossible to conclude that it is too is favored for enthalpic reasons.) The crystal structure, of the free
40
acid of cADPR, showed both γ–bonds as trans. Our recent MD simulation of cADPR found both γ–
+
bonds to be γ and a northꢀfavoring Aꢀring, though once the A–γ bond was restrained to trans a south A–
ring was favored. We suspect this may be due to the bsc0 refinement of the AMBER force field which
t
62
+
was designed to minimize the γ state. Thus three different experiments – this NMR study (R–γ and a
t
+
+
+
mixture of A–γ and A–γ ), our recent MD simulation (R–γ and A–γ when unrestrained), and the
t
t
crystal structure (of the free acid; R–γ and A–γ ) – give three different outcomes for the γ–bonds.
Clearly the A–γ bond, finely balanced between its two observed orientations, is sensitive to its
environment. Analog 4b is also unusual in that its Rꢀγ bond shows much larger favorable –ꢁH° and
+
much larger unfavorable –TꢁS° values for the γ state compared to the other three compounds, though
the ꢁG° values are comparable. The removal of the A-ring 2'-OH somehow greatly enthalpically favors
+
the remote R-
γ
state while simultaneously penalizing it entropically.
It is to be expected that there will be differences in the conformational preferences of the γ–bonds in
6
0a
cADPR compared to those of the free nucleosides. In the Chattopadhyaya group study they noted that
+
without exception the population of the γ state decreased with increasing temperature, regardless of
+
whether the γ state was the favored rotamer or not. The conclusion to be drawn is that in free
t
nucleosides the γ state is the more disordered rotamer. Only in the Rꢀring γ–bonds of 4a–4d was this
+
trend observed; the Aꢀγ state became more populated with increasing temperature. Generally in 4a–4d
+
t
we can say the more disordered rotamers are the Aꢀγ and the Rꢀγ . The adenine base is necessarily in
the syn orientation in cADPR and the analogs. Given that in free nucleosides a syn conformation
23
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 24 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
+
48,58a
+
generally decreases the γ population
it is remarkable the Aꢀγ state in cADPR is populated to the
extent that it is.
t
β–Bonds. The β is favored for enthalpy reasons and is highly populated in both rings of cADPR 4a and
analogs 4b, 4c, and 4d. The TꢁS° term is smaller than the ꢁH° term by an approximate factor of two,
and the favorable ꢁH° for the Rꢀring β–bonds is greater than that of the Aꢀring β–bonds by the same
factor. The TꢁS° term for the Aꢀring β–bonds is somewhat variable in sign, but the Rꢀring β–bond is
always negative.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
A Model for cADPR. Based on the above discussion, we would propose the following model (Figure 7)
for cADPR that accommodates our data, potentially explains some of the SAR for cADPR, and provides
direction for future efforts.
What parts of the story should the model include? It is clear from the discussion above that the
conformations of cADPR 4a and analogs 4b, 4c, and 4d are overall quite similar, meaning activity
cannot be explained on basis of divergent structures. Changes to the preferred conformations of the
Aꢀring β–bond, (Aꢀβ), Rꢀβ, and Rꢀγ bonds will incur a substantial enthalpy penalty. In a sense, these
subunits could be considered to be thermodynamically 'restrained'. Establishing that these conformers
are kinetically restrained due to high barriers to rotation would require either lineshape analysis or
substantial modeling efforts well beyond the scope of the present study. The furanose rings are certainly
flexible, but there is no NMRꢀbased evidence that the structural modifications (2'ꢀdeoxyꢀ, 8ꢀbromo,
7
ꢀdeaza) had any impact on the preferred furanose conformations (Aꢀring P ~ 168°; Rꢀring P ~ 215°).
S
S
If cADPR is indeed 'conformationally agnostic' as to the orientation of its furanose rings, this would
seem to leave the Aꢀring γ–bond and the two Nꢀglycosidic bonds, Aꢀχ and Rꢀχ (defined in the legend to
Figure 7) as the only sites capable of variability.
In our model (Figure 7), the nucleobase is positioned so that Aꢀχ is ~60° and Rꢀχ is ~10°,
5
7a
40
approximating the values from the crystal structure and our MD simulation , with the adenine
C2ꢀN3ꢀC4 edge pointing towards the area between A2' and A5" hydrogens. The phosphate backbone is
behind the plane of the page. One attractive feature of this model is that it immediately provides a
1
rationale for the 'unusually high' H chemical shift of A2', as it is edge on to the adenine and deshielded
by the aromatic ring current of the nucleobase. Analogously, the higher chemical shifts of A5" and R5'
t
+
relative to their geminal partners A5' and R5" are explained. Two models of cADPR (Aꢀγ and Aꢀγ )
were built in HyperChem and a simple vacuum geometry optimization (AMBER force field) was
24
ACS Paragon Plus Environment

Page 25 of 38
The Journal of Organic Chemistry
1
2
3
performed. The starting structures were built using the conformations listed in Tables 1 and 2: Aꢀring
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
t
t
+
P ~177°, Rꢀring P ~216°, Aꢀβ , Rꢀβ , and Rꢀγ . Based on the very similar crystal structure/MD values,
S
S
Aꢀχ was set to +60° and Rꢀχ to +10°. The remaining pyrophosphate backbone torsions and partial
atomic charges were taken from their respective MD simulations. No attempt was made to determine if
the structures were the global minimum structures, nor were alternate pyrophosphate backbone
conformers considered. The resulting structures are shown in Figure 7. The close contacts and key
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
t
torsion angles in the Aꢀγ structure (Figure 7a,b) are: H –H8 2.6ꢀ2.7 Å, H –N3 2.5ꢀ2.8 Å, H ꢀN3
A1'
A2'
A5"
2
.5ꢀ3.0 Å, H –H2 2.5ꢀ3.0 Å, χ ~+60°, χ ~+5°, and H and H8 eclipsed.
R5'
A
R
A1'
t
Consider the case of an 8ꢀbromo substituent (4d), which has the least populated (51%) Aꢀring γ ꢀstate
of the series. The ꢁH° for the Aꢀring NꢀS equilibria in 4d is 0.57 kcal/mol less favorable than it was
for 4a (ꢁꢁH°Aꢀring = ꢁH° – ꢁH° = +0.57 kcal/mol), compensated for by a more favorable entropy
4d
4a
change (TꢁꢁS° = +0.51 kcal/mol. What is the origin of the enthalpy penalty? To avoid an 8ꢀBr/HA1'
eclipsing interaction, the Aꢀχ bond could rotate towards the Aꢀring C2', decreasing the distance between
the Aꢀring C2' and the 8ꢀbromo substituent, making Aꢀχ less positive than +60°. Simultaneously this
would decrease the H –N3 distance, increasing the steric pressure and destabilizing the Aꢀring
A5"
t
furanose Sꢀstate and γ states. The 7ꢀdeaza analog 4c retains its agonist properties perhaps because the
exchange of N7 for a methine does not perturb χ. Why it should be that 2'ꢀdeoxy cADPR 4b, which has
t
the most populated (68%) Aꢀring γ ꢀstate of the series, is more difficult to explain. We note that our
3
9
earlier study of 3'ꢀOMe ꢀcADPR clearly showed an altered Aꢀγ bond, also difficult to rationalize. It
A
would seem Aꢀring substituents in the cADPR system can have unpredictable effects on the
conformation of other remote groups.
This model – the notion that (1) agonistic/antagonistic potency is derived in part from an HA5"–N3
t
interaction that impacts the Aꢀγ population and subtly alters the Aꢀχ torsion and (2) that the Rꢀring
Sꢀtype conformation is more disordered – can be tested. A hydrogen bond between the Aꢀring 2ꢀOH
and the amino group in the potent antagonist 8ꢀamino cADPR would 'lock in' a reduced Aꢀχ value,
t
which should reduce the Aꢀγ population more significantly than was the case for 8ꢀbromo analog 4d.
2
4
The 8ꢀaminoꢀ2'ꢀdeoxy cADPR analog, a much less potent antagonist, lacks such an interaction, and
would have a more relaxed Aꢀχ torsion; it also accounts why a similar deletion in 4d does not lead to a
reduction is antagonistic potency. Conversely, the 3ꢀdeaza analog 9a is the most potent analog known to
date, and the increased steric bulk of the new methine group should make Aꢀχ more positive than +60°
25
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 26 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
as well as impact the Aꢀγ conformation. To the second point, it could be that the Rꢀring is just a spacer
between the adenineꢀN1 and the phosphate backbone, and in its entropically favored Sꢀconformer allows
local solvent to become more disordered. Certainly solvent is less organized about the cIDPRE analogs
1
1 and 12, where the Rꢀring has been replaced with an ether linkage. In contrast to the antagonistic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
activity displayed by most 8ꢀsubstituted analogs, 8ꢀNH ꢀcIDPRE and 8ꢀN ꢀcIDPRE (12) are agonists,
2
3
equipotent with cADPR, perhaps because the nucleobase, free of its Rꢀring, is not forced to adopt the χ
angles that promote antagonism.
Figure 7. A model for cADPR. (a) Schematic representation of 4a–d based on our NMR data and the cADPR
crystal structure/MD simulation values for the glycosidic bonds Aꢀχ and Rꢀχ. (b) and (c) HyperChem (HC)
t
+
minimized structures (Amber99). The Aꢀring is in the γ (b) and γ (c) conformation. The Aꢀχ bond was defined
using standard accepted rules, but the Rꢀχ bond defined as O4' –C1'–N1–C2 so that it will have the same
R
magnitude and sign as the Aꢀring.
Of course, the conformational behavior of a ligand in solution, even if correctly modeled, is never
guaranteed to translate to the ligand in the bound state. The penalty cADPR or an analog would have to
+
pay to bind in, for example, the (solution state) less favored Aꢀγ conformation (< 0.4 kcal/mol) could
easily be compensated for by a single favorable hydrogen bond contact to its binding protein. The point
26
ACS Paragon Plus Environment

Page 27 of 38
The Journal of Organic Chemistry
1
2
3
we would make is that this thermodynamic study and the model generated therein provide new insights
and testable ideas for the study of the conformational behavior and SAR of cADPR. Additional detailed
thermodynamic and structural studies of these and other cADPR analogs will be investigated in due
course.
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ꢀ
CONCLUSION
A series of previously known cADPR analogs were synthesized and their solution state behavior studied
1
utilizing variable temperature H NMR spectroscopy. Analysis of the coupling constants using
PSEUROT and other methodologies allowed us to determine the conformations about majority of the
rotatable bonds in this 20ꢀmembered macrocycle. Population and van't Hoff analyses provided us with
t
+
+/–
t
the thermodynamics of the N ꢀ S, γ ꢀ γ , and β ꢀ β equilibria in both the Aꢀring and Rꢀring of
cADPR. By combining the identities of the conformers with knowledge of ꢁH°, ꢁS°, and ꢁG° for each
of their conformational equilibria, we were able to propose a model that explains how small variations in
analog structure can drive the associated equilibria in unexpected ways.
ꢀ EXPERIMENTAL SECTION
General Information. Moistureꢀsensitive reactions were performed under an atmosphere of dry
nitrogen using syringe techniques. Phosphoryl chloride, triethyl phosphate (TEP), diphenyl phosphoryl
chloride (DPPC), triꢀnꢀoctylamine (TOA), triꢀnꢀbutylamine (TBA), and anhydrous DMF, stored in a
desiccator box, were used as obtained from commercial suppliers. Pyridine and triethylamine were
distilled from KOH; 1,4ꢀdioxane and diethyl ether were filtered through activated alumina prior to use.
Deionized water was 18 MOhm/cm quality. Nucleoside/nucleotide starting materials were obtained
from commercial sources and not dried prior to use. 2',3'ꢀdiꢀOꢀacetyl NMN was prepared as described
(Graham, S.M; Macaya, D.J.; Sengupta, R.N.; Turner, K.B. Org. Lett. 2003, 6, 233–236). ADP
ribosyl cyclase was from SigmaꢀAldrich. Analytical HPLC utilized a 150 × 4.6 mm C18 column and
UVꢀdiode array detection. Eluant A was 20 mM pH 3 potassium phosphate buffer (KPB), eluant B a
9
1
0:10 mixture of eluant A:2ꢀpropanol, and eluant D 40:60 CH CN:H O. The flow rate was generally
3 2
mL/min. Preparative ionꢀexchange chromatography was done using a 10 × 1.6 cm Sepharose Q
27
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 28 of 38
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
column (GE Healthcare/Pharmacia HiPrep Q HP 16/10, flow rate 5 mL/min) and water:0.25 M formic
acid (FA), ammonium formate (AF), triethylammonium bicarbonate (TEAB), or triethylammonium
acetate (TEAA) gradients. 'Lyophilized' signifies a frozen aqueous sample lyophilized three time from
water; 'evaporation' signifies removal of water or other highꢀboiling solvents on a rotary evaporator
connect to a mechanical oil pump.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
NMR Spectroscopy. NMR spectra (400 MHz) were acquired on a spectrometer outfitted with a
1
1
31
1
variable temperature (VT) unit which had been recently calibrated. All H spectra (routine H, { P} H,
1D TOCSY) were collected with a spectral width of 8000 Hz, 64K time domain data points, and zeroꢀ
filled to 64K points (spectral resolution of 0.12 Hz/pt). VT spectra were collected similarly but zeroꢀ
filled to 256K points (spectral resolution of 0.03 Hz/pt). The 1D TOCSY mixing time was generally
1
00 ms. Resolution enhancement was routinely applied (lb ~–1 Hz and gb ~0.08ꢀ0.12) to all spectra
prior to transform. The J values in Table 1 were obtained with more aggressive processing (lb ꢀ2, gb 0.2
for ring signals and up to lb ꢀ4 for 5'/5" signals. Samples were prepared by lyophilizing once from
~
1 mL of 99.9% D O and subsequent dissolution in 99.96% D O. The internal standard used was
2 2
3ꢀ(trimethylsilyl)ꢀ2,2,3,3ꢀtetradeuteropropionic acid (TMSPꢀd ), and spectra were generally shimmed so
4
that the TMSP halfꢀheight linewidth was 0.7 Hz or less. All cADPR spectra were approximately 10ꢀ20
mM concentration. When given, assignments are based on either a COSY or a 1D TOCSY experiment.
PSEUROT Calculations. Version 6.3 of the PSEUROT program was used. PSEUROT no longer runs
on 64 bit PCꢀoperating systems, but will run on a Windows XP 'virtual machine'. Version 6.3 utilizes
the Donders form of the generalized Karplus equation (GKE). Substituent electronegativities were as
follows: heterocyclic base (0.56), O4' (1.26), 2'ꢀOH (1.26), C3' (0.62), C1' (0.62), C4' (0.62), C2' (0.62),
3
'ꢀOH (1.26), C5' (0.68). PSEUROT uses 'A' and 'B' parameters to convert the endocyclic ring torsion
angles to the exocyclic HꢀCꢀCꢀH needed by the GKE to calculate expected J values. Standard values
appropriate for ribofuranoses (4c, 4d, and 4b Rꢀring) were (reported as A,B): 1'ꢀ2' bond (1.102, 123.3),
2'ꢀ3' bond (1.090, 0.2), 3'ꢀ4' bond (1.095, –124.9). For the 4b Aꢀring deoxyribofuranose the values
were: 1'ꢀ2' bond (1.030, 121.4), 1'ꢀ2: bond (1.020, 0.9), 2'ꢀ3' bond (1.060, 2.4), 2"ꢀ3' bond (1.060, 122.9),
3
3
'ꢀ4' bond (1.090, –124.0). The puckering amplitude of the minor north conformers were restrained to
5° (Aꢀring) and 40° (Rꢀring). Full details and sample PSEUROT input files can be found in the
Supporting Information. The program and an updated (2013) pdf version of the PSEUROT 6.3 manual
is available from the corresponding author upon request.
28
ACS Paragon Plus Environment

Page 29 of 38
The Journal of Organic Chemistry
1
2
3
4
5
General Procedure for 5'-Phosphorylation (2). A suspension of adenosine analog 1c or 1d
(
~399 ꢀmol) was suspended in 1400 ꢀL of TEP, cooled to 0 °C in an iceꢀwater bath, and POCl (77 ꢀL,
3
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
824 ꢀmol) added via syringe. The suspension clarified within approximately 15 minutes and was
allowed to stir at 0°C. Reaction progress was monitored by HPLC (95:5 A:B, isocratic), following the
disappearance of the starting material (t ~11 min) and the appearance of product (t ~4 min). After 1 h,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
R
R
the reaction was transferred via cannula to 50 mL of iceꢀcold rapidly stirred dry ether. The resulting
precipitate was briefly centrifuged, supernatant decanted, and the process repeated with 2 × 2 mL
portions of ether. After briefly drying under a stream of nitrogen, the precipitate was dissolved in water
(40 mL) and stirred at room temperature for 1 h. The pH of the crude product solution was adjusted to
7
.6 by the addition of KOH, and the reaction mixture purified on the Sepharose Q column (formate
form) utilizing a gradient of 4% ꢀ 25% FA over 1000 mL. Product eluted at approximately 50 mM FA.
Product fractions were evaporated to dryness, evaporated twice more from water, three times from
methanol, and dried under high vacuum overnight. To improve solubility in the coupling with NMN,
AMP analogs were converted their trioctylammonium salts. Thus, to a suspension of the free acids of
AMP analogs 2b, 2c, or 2d (~200 ꢀmol) in 5 mL of HPLC grade methanol was added one equivalent
(~83 ꢀL, 200 ꢀmol) of TOA. After stirring at room temperature for 15 minutes, the resulting solution
was evaporated to dryness and briefly dried under high vacuum. Final traces of water were removed by
evaporation from dry dioxane. The resulting analog AMP•TOA salt was used without purification in
the next step.
1
7
7
-Deazaadenosine 5'-monophosphate tri-n-octyl ammonium salt (2c). From 106.3 mg (399 ꢀmol)
of 1c was obtained 63.3 mg (183 ꢀmol, 46%) of the free acid of 2c as a clear glass, which was converted
1
to its trioctylammonium salt. H NMR (400 MHz, CD OD): δ 8.066 (s, 1H, H2), 7.576 (d, 1H, J = 3.80,
3
H8), 6.629 (d, 1H, J = 3.76, H7), 6.255 (d, 1H, J = 5.88, H1'), 4.490 (t, 1H, J = 5.5, H2'), 4.351 (dd,
1
H, J = 5.16, 3.40, H3'), 4.179 (m, 1H, H4'), 4.090 (m, 2H, H5', H5"), 3.045 (m. 6H, N–(CH ) , 1.668
2 3
31
(
m, 6H, N–(CH CH ) , 1.30 (m, 30H), 0.886 (t, 9H, J = 6.9 Hz, 3 × –CH ). P NMR (162 MHz, D O):
2
2 3
3
2
δ 0.956.
8
-Bromo-adenosine 5'-monophosphate free acid (2d) and 8-chloroadenosine 5'-monophosphate free
acid (2e). 8ꢀBromoadenosine (1d, 69.3 mg, 200.2 ꢀmol) was reacted as described above. HPLC
analysis after 15 min showed two product peaks with retention times of 4.1 min (UV λmax = 262 nm) and
4.8 minutes (UV λmax = 264 nm) in an approximate 26:74 ratio. After 30 min, the reaction was worked
29
ACS Paragon Plus Environment