Fragment based search for small molecule inhibitors of HIV-1 Tat-TAR

Article abstract of DOI:10.1016/j.bmcl.2014.11.004

Basic molecular building blocks such as benzene rings, amidines, guanidines, and amino groups have been combined in a systematic way to generate ligand candidates for HIV-1 TAR RNA. Ranking of the resulting compounds was achieved in a fluorimetric Tat-TAR

Full text of DOI:10.1016/j.bmcl.2014.11.004

Bioorganic & Medicinal Chemistry Letters 24 (2014) 5576–5580
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Fragment based search for small molecule inhibitors of HIV-1
Tat-TAR
Mirco Zeiger, Sebastian Stark, Elisabeth Kalden, Bettina Ackermann, Jan Ferner, Ute Scheffer,
⇑
Fatemeh Shoja-Bazargani, Veysel Erdel, Harald Schwalbe, Michael W. Göbel
Goethe-Universität Frankfurt am Main, Institut für Organische und Chemische Biologie, D-60438 Frankfurt am Main, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
Basic molecular building blocks such as benzene rings, amidines, guanidines, and amino groups have
been combined in a systematic way to generate ligand candidates for HIV-1 TAR RNA. Ranking of the
resulting compounds was achieved in a fluorimetric Tat-TAR competition assay. Although simple
molecules such as phenylguanidine are inactive, few iteration steps led to a set of ligands with IC50 values
Received 19 August 2014
Revised 28 October 2014
Accepted 1 November 2014
Available online 6 November 2014
ranging from 40 to 150 lM. 1,7-Diaminoisoquinoline 17 and 2,4,6-triaminoquinazoline 22 have been fur-
ther characterized by NMR titrations with TAR RNA. Compound 22 is bound to TAR at two high affinity
sites and shows slow exchange between the free ligand and the RNA complex. These results encourage
investigations of dimeric ligands built from two copies of compound 22 or related heterocycles.
Ó 2014 Elsevier Ltd. All rights reserved.
Keywords:
Amidine
Guanidine
Ligand
NMR
RNA
TAR, the transactivation response element of HIV is a noncoding
RNA sequence important for viral replication.1,2 It is located at the
50 end of all viral transcripts. In the case of HIV-1, TAR forms a stem
loop structure of 59 nucleotides with an UCU bulge close to the
hexanucleotide loop (Fig. 1). Upon binding of Tat, a viral regulatory
protein expressed in the early stage of virus replication, the
protein–RNA complex recruits the human positive transcription
elongation factor (P-TEFb) which increases gene expression several
hundred-fold.3,4 This mechanism requires the formation of a ter-
nary complex of high affinity and specificity between two proteins
and the RNA. Random mutations of TAR, that are frequent due to
any structural information about TAR. Such fragments are: The
benzene ring able to stack and to undergo hydrophobic interactions
and amines, amidines, or guanidines as H-bond donors that can
be protonated under physiological conditions. All resulting com-
pounds were subjected to a fluorimetric competition assay using
the 31 mer TAR model and the dye labeled Tat peptide shown in
Figure 1.13 The dicationic arginine amide, the simplest TAR ligand
known to induce similar structural changes in TAR as longer Tat
peptides,14 shows an IC50 value in this assay of 1500
lM. Fragment
based search strategies for RNA ligands15 have been successfully
applied not only to TAR16 but also to other RNA targets such as
Hepatitis C virus IRES,17–19 ribosomal decoding site,20–22 Tetrahy-
the low accuracy of the reverse transcriptase, have
a high
probability to abolish its function. On this basis TAR has been an
attractive drug target for some time.5,6 However, almost two dec-
ades of research have introduced many types of TAR ligands but
none of them met the criteria to enter a drug development process
yet, despite the fact that the ternary complex of Tat, TAR, and
P-TEFb is still considered a promising site for antiviral intervention.7
Our previous work was focussed on peptides composed of non-
natural aromatic amino acids and lysine or arginine.8–12 Although
nanomolar target affinities could be achieved, the specificity for
TAR remained insufficient.
mena group I intron,23 tRNALys 24–26 riboswitches,27 and RNA
,
quadruplexes.28 All compounds shown in Figure 2 were either
purchased or synthesized as described in the Supplementary data.
The fluorimetric competition assay afforded a fast identification
and ranking of potent ligands.29 Neither unsubstituted guanidinium
chloride 1 nor simple derivatives like phenylguanidine 2, benzyl-
guanidine 3, and its heterocyclic analogue 4 are able to displace
the reference ligand from TAR. Heteroaromatic guanidine 5 and
amidine 6 are inactive as well. In contrast, 2-aminoimidazole 7
looks promising at first glance. However, the IC50 value of this
compound improves steadily when aqueous solutions are kept
under air. This effect was also found with other compounds, for
example tetraaminoquinazoline 23. The electron rich heterocycles
in particular have the tendency to produce false positive results,
presumably by forming multiply charged oligomers. Such samples
The study presented here originated from the idea to define
elementary ligand fragments and to combine them unbiased from
⇑
Corresponding author. Tel.: +49 69 798 29222; fax: +49 69 798 29464.
E-mail address: M.Goebel@chemie.uni-frankfurt.de (M.W. Göbel).
http://dx.doi.org/10.1016/j.bmcl.2014.11.004
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

M. Zeiger et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5576–5580
5577
G
G
U
(a)
(b)
G
C
A
C
S
36
G
28
G
C
CONH2
HN
O
N
H
AAARKKRRQRRRAAAC
A
U26
38
U
S
G
C
C
HOOC
O
O
U
40
N
U
C
U
G
G
C
C
A
21
G
A
C
C
G
G
HO
O
42
43
44
COOH
N
O
N
17
16
+
5'
3'
Figure 1. Components of the fluorimetric competition assay. (a) 31 mer Model of HIV-1 TAR. (b) Structure of the dye labeled Tat mimetic. Conditions: 10 nM of peptide and of
RNA; 50 mM Tris–HCl, pH 7.4, 20 mM KCl, 0.01% Triton-X 100, 37 °C. The Kd of the complex is 0.8 0.3 nM.9
NH
NH2
NH
NH2
H
N
H
N
H
N
NH2
NH
N
H
H2N
N
1
2
6
3 (>20'000)
4 (>20'000)
(>20'000)
(>20'000)
N
N
N
NH2
NH2
N
H
N
H
N
NH2
N
NH2
5
7
8
(>11'000)
(>20'000)
(>20'000)
(>2'000 ?)
11
R
(50)
1 = R2 = H
O
NH
N
12 R1 = H R2 = CH3
NH
R1
N
NH2
N
H
N
(50)
R
(150)
N
NH2
1 = R2 = CH3
R2
13
9
10
11-13
(>10'000)
(>20'000)
NH2
N
NH2
N
NH2
N
H2N
N
HO
H2N
14
15
16
17
(8'000)
(2'000)
(750)
(150)
NH2
N
NH2
N
NH2
NH2
N
N
N
NH2
N
NH2
H2N
N
NH2
NH2
19
18
(400)
20
21
(700)
(2'000)
(110)
NH2
NH2 NH2
NH2
N
NH2
N
H
N
H2N
H2N
N
N
N
N
Bn
N
NH2
NH2
N
NH2
N
NH2
22
23
24
25
(50)
(40)
(50)
(45)
N
NH2
NH2
N
NH2
N
H2N
N
N
N
H2N
N
N
H2N
NH2
HN
N
N
N
NH2
N
N
NH2
26 (130)
27 (70)
28 (60)
29 (50)
Figure 2. Compounds tested in the Tat-TAR competition assay. IC50 values (
l
m) are shown in brackets. Caution is required to eliminate errors due to decomposition or
fluorescent emission of the ligands. See text and Figure 1 for details. Counterions: Clꢀ: 1, 3, 4, 9, 12, 21, 23, 24, 25, 27, 28; Brꢀ: 11; Iꢀ: 13; nitrate: 2; sulfate: 7; trifluoroacetate:
26; heterocycles 5, 6, 8, 10, 14–20, 22, and 29 were dissolved in buffer as neutral compounds.

5578
M. Zeiger et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5576–5580
must be carefully recrystallized and only freshly prepared
solutions can be used in the assay. Most compounds, however,
are either stable in solution or, after several days, show some
decomposition forming inactive products.
sives prazosin, terazosin, and doxazosin, or in dihydrofolate
reductase inhibitors such as methotrexate, and piritrexim. As in
the case of isoquinoline 14, the IC50 value of 19 improves by one
order of magnitude by adding NH2 into position 6 (22, 40 lM).
From the first set of compounds, a second series was generated
by a formal ring closure converting guanidines 2 and 3 into com-
pounds 8–10. The planarization of the structures leads to minor
gains in RNA affinity. 2-Aminobenzimidazole 8, due to its enlarged
stacking surface should bind more tightly to RNA than compound
7. The IC50 value of 7 is therefore surprising and should be consid-
ered with some care. By fusing guanidine 9 with a second benzene
ring, however, the first interesting RNA binder results: 2-aminop-
2,4,6-Triaminoquinazoline 22 has already been described as a
Tat-TAR inhibitor.30 The importance of the 6-amino groups is seen
from the fact that its replacement by methyl causes a 50 fold
reduction in affinity (20, 2000
observed when the amino group is shifted from position 6–7 (21,
110 M). Compound 21 shows faster decomposition than its iso-
lM). Decreased binding is also
l
mer 22 leading to high-affinity polycationic products. The problem
of low stability and false-positive assay data is most pronounced
with tetraaminoquinazoline 23. No significant change in IC50 val-
ues results when the 6-amino group of 22 is methylated or benzy-
lated. From the similarity of 22 and 24 to methotrexate one might
expect cytotoxic effects. Methylation of amino groups in the het-
erocyclic ring of methotrexate analogues, however, is known to
weaken their interaction with dihydrofolate reductase.31 We
therefore tested compound 26 which still has considerable affinity
to TAR. Replacement of the 6-amino group of 22 by aminomethyl
in quinazolines 27, 28, and pteridine 29 slightly impairs the IC50
value although these compounds will form dications at pH 7 more
readily. Interestingly, even the presence of multiple protonation
sites in compound 28 is noneffective.
For representative ligands from Figure 2, we tested the interac-
tion with RNA by 1H NMR titrations32 using the same 31 mer TAR
construct as in the fluorimetric assay (see Fig. 1). The imino region
from 12 to 14 ppm shows signals of guanosines and uridines
involved in stable base pairs: G16, G17, G21, G26, G28, G36, U38,
U42, G43, and G44. Due to the dynamic nature of base pair
A22-U40, no signal is visible for U40 in the absence of ligands.
Signal assignment was based on NOESY spectra and on comparison
with published data. 7-Aminoquinoline 17 was titrated in steps of
0.25 equiv into a 0.2 mM solution of the RNA up to a final concen-
tration of 0.4 mM. A continuous shift around 0.1 ppm for signals of
G21, G26, G28, and U38 was observed (see Supplementary data).
NMR signals shifted and line widths did not increase during the
erimidine 11 (50 lM). Stepwise methylation of the amino group
reduces the potential of such perimidines to form hydrogen bonds
and should heavily interfere with RNA binding. The dimethyl
derivative indeed shows a reduction in affinity. However, the effect
is unexpectedly small (150 lM) suggesting nonspecific intercala-
tion as the dominating binding mode of compounds 11–13.
1-Aminoisoquinolines represent another structural class we
have examined. The pKa values of these compounds are in the
range of 7–8, so they are approximately half protonated under
physiological conditions. The 1-aminoisoquinoline scaffold can be
found in a number of pharmaceutical agents and should have a
good bioavailability. Combining the structural elements of the
two moderately active isoquinolines 14 and 15 leads to 1,7-diami-
noisochinoline 17 (150 lM), a compound with TAR affinity
increased by a factor of 13 or 53, respectively. The position of the
amino groups is relevant: A fivefold lower IC50 value is seen with
1,5-diamino derivative 18 (700
lM). The replacement of the 7-
amino group by hydroxy (16, 750
lM) causes a similar change.
Both, NH2 and OH, can donate or receive hydrogen bonds but the
potential to become protonated might be important for the higher
affinity of compound 17.
Although isoquinoline 14 and quinazolines 9 and 10 are only
weak Tat-TAR inhibitors, a combination of their structural features
gives 2,4-diaminoquinazoline 19, a compound with significant
affinity (400 lM). The scaffold of 19 is present in the antihyperten-
Figure 3. 1D 1H NMR spectrum (600 MHz, 283 K) of the HIV-1 31 mer TAR RNA (0.2 mM RNA H2O/D2O 9:1, pH 6.2). The signals of imino protons are shown as a function of
added quinazoline 22.

M. Zeiger et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5576–5580
5579
Figure 4. 1H, 1H NOESY spectrum (700 MHz, 283 K, 150 ms mixing time) of the imino protons of HIV-1 31 mer TAR RNA (0.5 mM RNA, H2O/D2O 9:1, pH 6.2) in the presence of
2 equiv of compound 22. Through space connectivities for the lower stem (G17-G44-G43-U42-G21) are highlighted by a dotted line. The dashed line shows the correlation of
spins belonging to the upper stem (G26-U38-G28-G36).
titration. Accordingly, free compound 17 is in fast exchange with
the TAR-17 complex. From analysis of the chemical shifts, we
predict the binding site to be close to G26, similar to the complex
TAR forms with arginine amide.14 This assumption is consistent
with the effects on G26, G28, and U38 and also the absence of shifts
for G36. Binding at G26 is expected to stabilize the A22-U40 base
pair, which also affects the resonance of G21.
In the same way, a titration with quinazoline 22 was performed.
Above 0.4 mM, 22 was added in steps of 1.3 equiv up to a final
ligand-RNA ratio of 9.8:1. Significant changes occur only in the
range from 0 to 2 equiv of 22. Interestingly, most signals do not
shift continuously. They gradually fade away and reappear in their
new position: such response during ligand titration is characteris-
tic for slow exchange of free compound 22 and the TAR-22
complex (Fig. 3). From NMR, triaminoquinazoline 22 shows higher
affinity than compound 17, in agreement with fluorimetric assay
data.
in NMR. It also has two high affinity binding sites in TAR. Com-
pound 22 and its tetraamino analog 23 both have been identified
previously as potent Tat-TAR inhibitors in an industrial drug
screening program.30,34 From chemical probing with diethyl pyro-
carbonate, a binding site of 23 in the loop region close to A35 was
assumed. Compound 23 also protects G36 against methylation by
dimethyl sulfate.34 Given the similarity between compounds 22
and 23, this observation coincides with the strong influence
compound 22 exerts on the signal of G36. The second binding site
of 22 must be in the lower stem around G21. This observation
opens a new strategy for the development of drug-like heterocyclic
Tat-TAR inhibitors by combining two units of compounds 17 and
22 with appropriate linkers.
Acknowledgments
Financial support by the Deutsche Forschungsgemeinschaft
(SFB 902 Molecular Principles of RNA-based Regulation) is gratefully
acknowledged. H.S. is member of the cluster of excellence
Macromolecular Complexes. Work at the Center for Biomolecular
Magnetic Resonance (BMRZ) is supported by the state of Hesse.
Resonance assignments for the newly arising signals were
obtained from NOESY spectra in the presence of 2 equiv of 22
(Fig. 4). With the exception of G16 G17, and G44, all other signals
are significantly displaced. Effects are most pronounced for G36,
G28, U42, and G21.
TAR not only binds the arginine rich domain of Tat but also a
number of arginine-derived ligands. Arginine amide14 and other
guanidinium compounds33 have been shown by NMR to form
complexes with G26 resembling a G-C Hoogsteen base pair. Based
on these findings, we tested whether combining guanidines,
amines and aromatic rings would yield high affinity TAR ligands.
In fact, this hypothesis proved to be effective. Just a few iteration
steps led from inactive molecules to several candidates with IC50
Supplementary data
Supplementary data (Synthetic procedures and characterization
data for all new compounds or new synthetic pathways. NMR
titration of the 31 mer TAR RNA with compound 17. 1H and 13C
NMR spectra of compounds depicted in Fig. 2.) associated with this
article can be found, in the online version, at http://dx.doi.org/
10.1016/j.bmcl.2014.11.004.
values below 200 lM. From our NMR data we propose that
heterocycles containing the substructure of amidines and guani-
dines may behave like arginine amide and interact with G26.
Indeed, the NMR titration of TAR with isoquinoline 17 is consis-
tent with such interpretation. The best ligand found in this study
is triaminoquinazoline 22. Both structures have in common the
distance between the amidine part and the isolated amino group.
In spite of this similarity, the interaction with TAR is quite differ-
ent: 22 not only is a much better ligand, showing slow exchange
References and notes
1. Dingwall, C.; Ernberg, I.; Gait, M. J.; Green, S. M.; Heaphy, S.; Karn, J.; Lowe, A.
D.; Singh, M.; Skinner, M. A.; Valerio, R. Proc. Natl. Acad. Sci. U.S.A. 1989, 86,
6925.
2. Weeks, K. M.; Crothers, D. M. Cell 1991, 66, 577.
3. Wei, P.; Garber, M. E.; Fang, S. M.; Fischer, W. H.; Jones, K. A. Cell 1998, 92, 451.
4. Tahirov, T. H.; Babayeva, N. D.; Varzavand, K.; Cooper, J. J.; Sedore, S.; Price, D.
H. Nature 2010, 465, 747.
5. Bannwarth, S.; Gatignol, A. Curr. HIV Res. 2005, 3, 61.

5580
M. Zeiger et al. / Bioorg. Med. Chem. Lett. 24 (2014) 5576–5580
6. Baba, M. Antiviral Res. 2006, 71, 301.
28. Garavís, M.; López-Méndez, B.; Somoza, A.; Oyarzabal, J.; Dalvit, C.; Villasante,
A.; Campos-Olivas, R.; González, C. ACS Chem. Biol. 2014, 9, 1559.
7. Massari, S.; Sabatini, S.; Tabarrini, O. Curr. Pharm. Des. 1860, 2013, 19.
8. Krebs, A.; Ludwig, V.; Pfizer, J.; Dürner, G.; Göbel, M. W. Chem. Eur. J. 2004, 10,
544.
9. Ludwig, V.; Krebs, A.; Stoll, M.; Dietrich, U.; Ferner, J.; Schwalbe, H.; Scheffer,
U.; Dürner, G.; Göbel, M. W. ChemBioChem 2007, 1850.
10. Suhartono, M.; Weidlich, M.; Stein, T.; Karas, M.; Dürner, G.; Göbel, M. W. Eur. J.
Org. Chem. 2008, 1608.
11. Ferner, J.; Suhartono, M.; Breitung, S.; Jonker, H. R. A.; Henning, M.; Wöhnert, J.;
Göbel, M.; Schwalbe, H. ChemBioChem 2009, 10, 1490.
12. Suhartono, M.; Schneider, A. E.; Dürner, G.; Göbel, M. W. Synthesis 2010, 293.
13. Matsumoto, C.; Hamasaki, K.; Mihara, H.; Ueno, A. Bioorg. Med. Chem. Lett.
2000, 10, 1857.
14. Aboul-ela, F.; Karn, J.; Varani, G. J. Mol. Biol. 1995, 253, 313.
15. Moumné, R.; Catala, M.; Larue, V.; Micouin, L.; Tisné, C. Biochimie 2012, 94,
1607.
16. Davidson, A.; Begley, D. W.; Lau, C.; Varani, G. J. Mol. Biol. 2011, 410, 984.
17. Swayze, E. E.; Jefferson, E. A.; Sannes-Lowery, K. A.; Blyn, L. B.; Risen, L. M.;
Arakawa, S.; Osgood, S. A.; Hofstadler, S. A.; Griffey, R. H. J. Med. Chem. 2002, 45,
3816.
18. Seth, P. P.; Miyaji, A.; Jefferson, E. A.; Sannes-Lowery, K. A.; Osgood, S. A.; Propp,
S. S.; Ranken, R.; Massire, C.; Sampath, R.; Ecker, D. J.; Swayze, E. E.; Griffey, R.
H. J. Med. Chem. 2005, 48, 7099.
19. Parsons, J.; Castaldi, M. P.; Dutta, S.; Dibrov, S. M.; Wyles, D. L.; Hermann, T.
Nature Chem. Biol. 2009, 5, 823.
20. Yu, L.; Oost, T. K.; Schkeryantz, J. M.; Yang, J.; Janowick, D.; Fesik, S. W. J. Am.
Chem. Soc. 2003, 125, 4444.
29. Fluorimetric binding assay: Experiments were performed as described
previously9 in 96-well microplates by titration of compounds 1–29 into a
mixture of 31 mer TAR RNA (10 nM; BioSpring, Frankfurt, Germany) and Tat
peptide (10 nM; Thermo Electron Corporation, Ulm, Germany) in TK buffer
(50 mM Tris–HCl, 20 mM KCl, 0.01% Triton-X 100, pH 7.4, 37 °C). Different to
previous reports, the excitation wavelength was optimized here (Reader:
Tecan safire2; kex = 489 nm, kem = 590 nm). Contact quenching of the two dyes,
not FRET, is the dominant effect explaining the fluorescence change in this
assay. An independent titration into buffer free of RNA and Tat peptide was
conducted for each compound. Fluorescent emission of ligands thus could be
subtracted from the emission value of the Tat peptide. Titrations were repeated
with ligand solutions kept under air for 1–3 days. In most cases no change or a
minor decline of RNA affinities was observed. In contrast, higher affinities were
found for compounds 7 and 23. We attribute these false positives to the
formation of oligomeric decomposition products of the heterocycles.
30. Czarnik, A. W.; Mack, D. P.; Mei, H.-Y.; Moreland, D. W. PCT Int. Appl. WO
9925327 A2 19990527, 1999.
31. Roth, B.; Smith, J. M.; Hultquist, M. E., Jr. J. Am. Chem. Soc. 1950, 72, 1914.
32. NMR titration experiments: Titrations were carried out in an aqueous phosphate
buffer (25 mM potassium phosphate, 50 mM potassium chloride) with 10% (v/
v) D2O at pH = 6.2. RNA concentration was 0.2 mM at the beginning of the
titration. As a reference standard 1 mM 3-(trimethylsilyl)-1-propanesulfonic
acid (TMPSA) was added. The 1D 1H NMR spectra were recorded on a Bruker
600 MHz spectrometer equipped with a cryogenic probehead at 283 K using
the jump-return-echo method for water suppression and the maximum of the
sinusoidal excitation profile placed on the center of the RNA imino proton
resonances. Assignment of the imino protons at an RNA concentration of
0.5 mM in the presence of two equivalents of 22 was performed by a NOESY
spectrum with 150 ms mixing time (Fig. 4). Spectra were recorded on a Bruker
21. Foloppe, N.; Chen, I.-J.; Davis, B.; Hold, A.; Morley, D.; Howes, R. Bioorg. Med.
Chem. 2004, 12, 935.
22. Bodoor, K.; Boyapati, V.; Gopu, V.; Boisdore, M.; Allam, K.; Miller, J.; Treleaven,
W. D.; Weldeghiorghis, T.; Aboul-ela, F. J. Med. Chem. 2009, 52, 3753.
23. Johnson, E. C.; Feher, V. A.; Peng, J. W.; Moore, J. M.; Williamson, J. R. J. Am.
Chem. Soc. 2003, 125, 15724.
24. Chung, F.; Tisné, C.; Lecourt, T.; Dardel, F.; Micouin, L. Angew. Chem., Int. Ed.
2007, 46, 4489.
25. Chung, F.; Tisné, C.; Lecourt, T.; Seijo, B.; Dardel, F.; Micouin, L. Chem. Eur. J.
2009, 15, 7109.
700 MHz spectrometer equipped with
a cryogenic probehead at 283 K.
Recording and processing of the NMR spectra were performed with the
software Topspin 2.0 (Bruker, Germany).
33. Davis, B.; Afshar, M.; Varani, G.; Murchie, A. I. H.; Karn, J.; Lentzen, G.; Drysdale,
M.; Bower, J.; Potter, A. J.; Starkey, I. D.; Swarbrick, T.; Aboul-ela, F. J. Mol. Biol.
2004, 336, 343.
26. Moumné, R.; Larue, V.; Seijo, B.; Lecourt, T.; Micouin, L.; Tisné, C. Org. Biomol.
Chem. 2010, 8, 1154.
34. Mei, H.-Y.; Cui, M.; Heldsinger, A.; Lemrow, S. M.; Loo, J. A.; Sannes-Lowery, K.
A.; Sharmeen, L.; Czarnik, A. W. Biochemistry 1998, 37, 14204.
27. Chen, L.; Cressina, E.; Leeper, F. J.; Smith, A. G.; Abell, C. ACS Chem. Biol. 2010, 5,
355.