Structural basis for inhibition of mycobacterial and human adenosine kinase by 7-substituted 7-(het)aryl-7-deazaadenine ribonucleosides

Article abstract of DOI:10.1021/jm500497v

Adenosine kinase (ADK) from Mycobacterium tuberculosis (Mtb) was selected as a target for design of antimycobacterial nucleosides. Screening of 7-(het)aryl-7-deazaadenine ribonucleosides with Mtb and human (h) ADKs and testing with wild-type and drug-resistant Mtb strains identified specific inhibitors of Mtb ADK with micromolar antimycobacterial activity and low cytotoxicity. X-ray structures of complexes of Mtb and hADKs with 7-ethynyl-7-deazaadenosine showed differences in inhibitor interactions in the adenosine binding sites. 1D ¹H STD NMR experiments revealed that these inhibitors are readily accommodated into the ATP and adenosine binding sites of Mtb ADK, whereas they bind preferentially into the adenosine site of hADK. Occupation of the Mtb ADK ATP site with inhibitors and formation of catalytically less competent semiopen conformation of MtbADK after inhibitor binding in the adenosine site explain the lack of phosphorylation of 7-substituted-7-deazaadenosines. Semiempirical quantum mechanical analysis confirmed different affinity of nucleosides for the Mtb ADK adenosine and ATP sites.

Full text of DOI:10.1021/jm500497v

Subscriber access provided by George Mason University Libraries & VIVA (Virtual Library of Virginia)
Article
Structural basis for inhibition of mycobacterial and human adenosine
kinase by 7-substituted 7-(het)aryl-7-deazaadenine ribonucleosides
Jan Snasel, Petr Naus, Jiri Dostal, Ales Hnizda, Jindrich Fanfrlik, Jiri Brynda, Aurelie Bourderioux, Michal
Dusek, Hana Dvorakova, Jirina Stolarikova, Helena Zabranska, Radek Pohl, Petr Konecny, Petr Dzubak,
Ivan Votruba, Marian Hajduch, Pavlina Rezacova, Vaclav Veverka, Michal Hocek, and Iva Pichova
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/jm500497v • Publication Date (Web): 26 Sep 2014
Downloaded from http://pubs.acs.org on September 30, 2014
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal 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 44
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 44
1
1
2
Structural basis for inhibition of mycobacterial and human adenosine kinase by 7ꢀ
substituted 7ꢀ(het)arylꢀ7ꢀdeazaadenine ribonucleosides
3
4
5
6
7
8
9
Jan Snášel,†+ Petr Nauš,†+ Jiří Dostál†, Aleš Hnízda,† Jindřich Fanfrlík,† Jiří Brynda,† Aurelie
Bourderioux,† Michal Dušek°, Hana Dvořáková,§ Jiřina Stolaříková,ꢀ Helena Zábranská†,
Radek Pohl,† Petr Konečný,‡ Petr Džubák,‡ Ivan Votruba,†# Marián Hajdúch,‡ Pavlína
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Řezáčová,† Václav Veverka,† Michal Hocek†
*
and Iva Pichová†
*
†Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, Flemingovo nam.
2, 16610 Prague 6, Czech Republic;
°
Institute of Physics ASCR, v.v.i., Na Slovance 2, 182 21 Praha 8, Czech
§
Republic Laboratory of NMR Spectroscopy, Institute of Chemical Technology, Technicka 5, 16628 Prague 6,
Czech Republic; ꢀ Laboratory for Mycobacterial Diagnostics and Tuberculosis, Regional Institute of Public Health
in Ostrava, Partyzánské nám. 7, 70200 Ostrava, Czech Republic; ‡Institute of Molecular and Translational
Medicine, Palacky University and University Hospital in Olomouc, Faculty of Medicine and Dentistry,
Hněvotínská 5, 77515 Olomouc, Czech Republic.
# Deceased on May 25, 2014.
ACS Paragon Plus Environment

Page 3 of 44
Journal of Medicinal Chemistry
2
1
2
Abstract
3
4
5
6
7
8
Adenosine kinase (ADK) from Mycobacterium tuberculosis (Mtb) was selected as a target for
design of antimycobacterial nucleosides. Screening of 7ꢁ(het)arylꢁ7ꢁdeazaadenine
ribonucleosides with Mtb and human (h) ADKs and testing with wildꢁtype and drugꢁresistant
Mtb strains identified specific inhibitors of Mtb ADK with micromolar antimycobacterial
activity and low cytotoxicity. Xꢁray structures of complexes of Mtb and hADKs with 7ꢁ
ethynylꢁ7ꢁdeazaadenosine showed differences in inhibitor interactions in the adenosine
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
binding sites. 1D H STD NMR experiments revealed that these inhibitors are readily
accommodated into the ATP and adenosine binding sites of Mtb ADK, whereas they bind
preferentially into the adenosine site of hADK. Occupation of the Mtb ADK ATP site with
inhibitors and formation of catalytically less competent semiꢁopen conformation of MtbADK
after inhibitor binding in the adenosine site explain the lack of phosphorylation of 7ꢁ
substitutedꢁ7ꢁdeazaadenosines. Semiempirical quantum mechanical analysis confirmed
different affinity of nucleosides for the Mtb ADK adenosine and ATP sites.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 44
3
1
2
Introduction
3
4
5
6
7
8
Tuberculosis (TB) remains one of the leading public health problems worldwide and is
more prevalent in the world today than at any other time in human history. In 2012, an
estimated 8.6 million people developed TB, and 1.3 million died from the disease (WHO
tuberculosis report 2013). In recent decades, multidrugꢁresistant (MDR) and even more
ominous extensively drugꢁresistant (XDR) Mycobacterium tuberculosis (Mtb) strains have
emerged.1 At the same time, a high rate of HIV coꢁinfection, especially in developing
countries, has increased the susceptibility of the population to Mtb infection. Successful
treatment of TB requires long courses of several antibiotics. In HIVꢁpositive patients, TB
treatment is further complicated by undesirable drugꢁdrug interactions that decrease the
therapeutic concentration of antiretrovirals by induction of the hepatic cytochrome P450
oxidase system.2 Similar restrictions apply to coꢁadministration of diabetes and TB drugs.
Mtb can persist permanently in host lesions, evading immune surveillance. An estimated oneꢁ
third of the world’s population asymptomatically harbors a latent form of Mtb, with a lifeꢁ
long risk of disease activation and transmission. However, existing drugs do not target the
latent form of Mtb. Therefore, the discovery of new molecules with new mechanisms of
action is needed to address latent Mtb. New treatments may become critical as MDR and
XDR Mtb incidence increases.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Enzymes involved in the biosynthesis of purine nucleotides are potential targets for
development of new types of compounds active against Mtb. In most organisms, purine
nucleotides are formed from purine bases and phosphoribosylꢁ1ꢁpyrophosphate or by
phosphorylation of nucleosides in the purine salvage pathway, or they are synthesized de novo
in a multiꢁstep sequence. Mtb expresses enzymes from both pathways; however, the
interdependence and regulation of these processes remains unclear. Of the variety of possible
target enzymes from the Mtb purine salvage pathway,3,4 adenosine kinase (ADK, EC 2.7.1.20,
ACS Paragon Plus Environment

Page 5 of 44
Journal of Medicinal Chemistry
4
1
2
3
4
5
6
7
8
9
Rv2202c)5 is considered a promising target for drug development. ADK catalyzes
phosphorylation of adenosine to adenosine monophosphate (AMP) in the phosphoryl transfer
reaction using adenosine 5′ꢁtriphosphate (ATP) as a substrate and releasing adenosine 5′ꢁ
diphosphate (ADP). ADK is present in most eukaryotes, fungi, plants, and parasites but is not
commonly found in bacteria. The activity of ADK has been confirmed in Mtb,6 and
biochemical characterization indicates that this enzyme shares low structural similarity with
and behaves very differently than human and other wellꢁcharacterized ADKs.3,7
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Nucleoside analogs are an important group of drugs used to treat viral infections and
cancer. Certain nucleoside inhibitors are active against Tripanosomagondii8 and other
parasites.9 Several adenosine analogs with various modifications have already been tested for
inhibition of Mtb ADK.3,6,10 These studies established the promise of these compounds for
drug development. In addition, diverse modified 7ꢁdeazapurine nucleosides have been
reported as ADK inhibitors.11ꢁ15 Recently, we discovered 6ꢁhetarylꢁ7ꢁdeazapurine
ribonucleosides with nanomolar cytostatic activities towards leukemia and cancer cell lines,16
and we later found that these nucleosides are excellent inhibitors of Mtb ADK but only weak
antimycobacterial agents.17 Related 7ꢁdeazaadenosines bearing small fiveꢁmembered
heterocycles at position 7 (1) were also found to be nanomolar cytostatics,18 whereas
compounds bearing bulky aromatic substituents were neither cytostatic nor cytotoxic. Here,
we report on the structureꢁactivity relationship of 7ꢁsubstituted 7ꢁ(het)arylꢁ7ꢁdeazaadenine
ribonucleosides with Mtb ADK and human ADK (hADK). We also describe screening of
these compounds for in vitro inhibition of Mtb ADK and hADK and for inhibition of two Mtb
strains.
Results and Discussion
Chemistry of 7ꢀsubstituted 7ꢀdeazaadenosines
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 44
5
1
2
3
4
5
6
7
8
9
Synthesis of 1aꢀs (Chart 1) was reported previously.18 Our previous study showed that
7ꢁdeazaadenosine derivatives bearing small fiveꢁmembered rings or acetylene exerted
significant cytostatic/cytotoxic effects. To identify nonꢁtoxic derivatives with potential
activity against Mtb ADK, we designed a series of nucleosides bearing bulky hydrophobic
aromatic substituents at position 7. These substituents included substituted benzenes (1tꢀaa),
polycyclic (hetero)aromatics (1abꢀaf), styrene (1ag), biaryls (1ah,ai), and arylꢁbenzylꢁether
(1aj). All compounds were prepared by a singleꢁstep procedure based on the aqueous Suzukiꢁ
Miyaura crossꢁcoupling reactions of 7ꢁiodoꢁ7ꢁdeazaadenosine (2) with the corresponding
arylboronic acids (Scheme 1). The reactions generally proceeded smoothly, providing the
desired 7ꢁarylꢁ7ꢁdeazaadenosines 1tꢀ1aj in good yields of 67ꢁ91% (with the exception of
compound 1ai, which was isolated in 53% yield).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Chart 1. Numbering of proton positions in modified nucleosides and structures of previously
reported 7ꢁsubstituted 7ꢁdeazaadenosines.
ACS Paragon Plus Environment

Page 7 of 44
Journal of Medicinal Chemistry
6
1
2
3
Scheme 1. Synthesis of 7ꢁdeazaadenosines bearing bulky substituents at position 7
4
5
6
NH2
NH2
I
R
N
N
7
8
9
R-B(OH)2
Pd(OAc)2, TPPTS
N
N
N
O
N
O
HO
HO
Na2CO3, H2O/MeCN
100 °C
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
1t-1aj
OH OH
OH OH
O
R- =
CF3
1t
(82%)
1u
(86%)
1v
1w
1x
1y
(87%)
(86%) (81%) (80%)
O
O
O
O
O
1z
(81%)
1aa
(93%)
1ab
(84%)
1ac
(93%)
S
S
O
S
O
1af
85%)
1ad
(88%)
1ae
(91%)
O
1ag
(67%)
1ai
1aj
(70%)
1ah
(71%)
(53%)
Biological activity profiling of 7ꢀsubstitutedꢀ7ꢀdeazapurine derivatives
Compounds 1aꢀaj were tested for inhibition of human and Mtb ADKs (for cloning,
expression, and purification of these enzymes, see ref.19). Most compounds were also tested as
potential kinase substrates. The inhibition and substrate phosphorylation screening results
were correlated with the compounds’ in vitro cytotoxicity (MTT) against nonꢁmalignant BJ
and MRCꢁ5 human fibroblast cell lines (Table 1). As expected, the cytotoxicity strongly
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 44
7
1
2
3
depended on the bulkiness of the substituent at position 7, with all compounds bearing bulky
4
5
6
groups showing little to no cytotoxicity (with the exception of 1ai, which was toxic to BJ
7
8
9
cells). Most compounds did not significantly inhibit
substrates for this enzyme. Only the 7ꢁethynyl derivative (1a) was a potent (submicromolar)
ADK inhibitor, while a few additional derivatives (1l,m,u) were inhibitory at micromolar
hADK and were poorꢁtoꢁmoderate
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
h
concentrations. In contrast, most derivatives (e.g., 1f,g,j,k,m,s,t,u,y,af,ah) showed significant
inhibition of Mtb ADK with IC50 values in the micromolar range. Three bulky derivatives
(1ad,ae,ag) inhibited the Mtb enzyme at submicromolar concentrations but showed no
inhibitory activity toward
hADK. All compounds were also tested for activity against
Mycobacterium bovis and Mtb (strain My 331/88 and drugꢁresistant strain Praha 131). Most
compounds, except 1n,o,p,q,r, were active against Mtb with in vitro minimum inhibitory
concentration (MIC) ranging between 2 and 60 µM (see Table 1). The antimycobacterial
activities of 14 compounds from our series of 39 nucleoside analogs correlated well with Mtb
ADK inhibition. We found that 1d,i,x,aa,ab,ac,a were fairly active against Mtb and M. bovis
but did not inhibit Mtb and hADKs, perhaps suggesting that they can target other enzymes
from the purine salvage pathway. The MtbꢁADKꢁspecific derivative 1f (7ꢁ(2ꢁnaphthyl))
displayed the highest antimycobacterial activity with MIC of 2 ꢂM and 4 ꢂM for the wildꢁ
type and drugꢁresistant Mtb strains, respectively. The nonꢁtoxic, MtbꢁADKꢁspecific
dibenzofuran derivative (1ad) showed the best therapeutic index. This compound displayed
antimycobacterial activity with MIC of 4 ꢂM for both Mtb strains. Compounds 1f and 1ad are
promising lead structures for further development of antimycobacterial drugs. The lower
activities of 7ꢁdeazapurine derivatives against Mtb compared to M. bovis could be the result
of slightly different conditions used for cultivation and testing and/or the extent of compound
uptake through the cell wall. Therefore, our future focus will be on design of suitable
prodrugs and delivery of nucleosides.
ACS Paragon Plus Environment

Page 9 of 44
Journal of Medicinal Chemistry
8
1
2
3
Table 1. Cytotoxicity, human and Mtb ADK inhibition, and antimycobacterial activity
4
5
6
7
8
9
Compd.
BJa
MRCꢁ5
ADK
Phsb
ADK inhibition
IC50 (ꢂM)
M.
M.
bovisc
tuberculosisd
CC50
(ꢂM)
CC50
(ꢂM)
(%)
human
62
IC50
(ꢂM)
MIC (ꢂM)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
human
Mtb
My
Praha
331/88
131
43.1 ±3.6
50.9 ±6.4
21.9 ±1.4
18.0 ±3.2
90.6 ±5.8
13.9 ±1.9
18.3 ±2.1
13.4
±3.6
0.20
0.33 ± 0.02
0.13
1.34
2.08
0.36
1.87
0.39
17.4
8.31
4.73
3.93
7.27
8
8
8
1a
1b
1c
1d
1e
1f
±0.02
21.0
±3.2
ꢁ
ꢁ
>5
>5
>10
>10
8
16
8
61.4
±6.1
16
4
45.6
±3.5
ꢁ
>10
>10
>5
>10
94.4
±8.7
ꢁ
>10
32
2
32
4
13.6
±1.5
ꢁ
5.33 ± 0.65
2.10 ± 0.15
>10
9.49
8
9
2
2
6
>5
62.5
16
8
62.5
16
8
1g
1h
1i
±1.49
91.5
67.9
>5
±13.3
±14.9
65.4 ±2.6
12.9 ±0.9
>100
39.4
±4.5
>5
>10
42.5
±5.3
>5
6.60 ± 0.51
4.4± 0.30
8
8
1j
99.6
>20
62.5
62.5
1k
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 44
9
1
2
±1.0
3
4
5
6
7
8
10.1 ±1.5
>100
5.05
20
3
4.5 ±
0.31
>10
4.5 ± 0.20
>10
1.12
2.51
48.0
3.08
>100
>100
69.0
2.27
1.13
62.5
125
>250
125
>250
>250
>250
16
62.5
125
>250
125
>250
>250
>250
16
1l
±0.88
9
25.1
±5.1
2.3 ±
0.17
1m
1n
1o
1p
1q
1r
1s
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
9.51
0.18
7
>10
>10
13
±2.03
±0.13
48.9 ±5.3
71.2
±8.9
9
>5
0.12
0.33
47
19
37
ꢁ
>10
±0.03
±0.04
1.48
0.15
>20
>20
>10
>20
>3
±0.28
±0.06
0.08
0.10
±0.0
>10
±0.02
21.2 ±2.1
35.1
±3.8
1.80 ± 0.20
4.9 ± 0.40
57.3 ±1.4
37.9
±2.8
ꢁ
8
8
1t
>100
>100
ꢁ
ꢁ
3.3 ± 0.4 2.33 ± 0.25
6.75
5.29
32
16
32
16
1u
1v
63.9 ±4.4
52.9
±8.7
>20
>20
>10
>10
>10
>10
>100
96.3
±5.9
ꢁ
ꢁ
4.54
1.16
62.5
4
62.5
4
1w
1x
62.0 ±9.1
78.8
±12.0
ACS Paragon Plus Environment

Page 11 of 44
Journal of Medicinal Chemistry
10
1
2
59.9 ±2.7
15.5 ±0.7
50.2
±8.8
ꢁ
ꢁ
>10
>10
13.1 ± 1.5
>10
3.07
1.38
32
8
32ꢁ
8
3
4
5
6
7
8
9
1y
1z
46.0
±8.1
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
>100
23.1 ±2.5
>100
>100
>100
ꢁ
ꢁ
ꢁ
>10
>10
>10
>10
>10
>10
7.05
1.74
1.90
8
8
1aa
1ab
1ac
16
32
16
32
66.0
±9.2
>100
>100
100
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
ꢁ
>10
>20
>10
>10
>20
>20
>20
0.6 ± 0.05
0.3 ± 0.02
0.19
4.53
4
16
62.5
8
4
16
125
8
1ad
1ae
1af
1ag
1ah
1ai
±0.08
95.1
±7.7
>100
68.5
±6.3
2.06 ± 0.14 44.32
19.6 ±1.7
40.6 ±7.9
48.2
±9.6
0.9 ± 0.12
7.67 ± 0.81
>10
1.30
1.68
93.7
±6.5
4
4
0.12
64.2
74.42
6.39
62.5
16
62.5
32
±0.03
±12.1
>100
92.1
>10
1aj
±12.5
aCytotoxicity (MTT test) in BJ and MRCꢁ5 fibroblasts
bPhs, ADK substrate activity (phosphorylation), conversion to 5′ꢁphosphate (%)
c 50% growth inhibitory concentration of Mycobacterium bovis BCG cultivated in vitro
dIncubation time = 14 d. MIC of isoniazid as reference compound was 0.5 ꢂM for the My
331/88 strain.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 44
11
1
2
3
4
5
6
7
8
We observed differences in phosphorylation of 7ꢁsubstitutedꢁ7ꢁdeazapurine
nucleosides by human and Mtb ADK. While no synthesized analog was phosphorylated by
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Mtb ADK, several compounds served as substrates for
hADK (see Table 1). This may
indicate differences in the structures of these enzymes and/or different binding modes of
nucleoside analogs into the adenosine and ATP binding sites.
To gain insight into the mechanism of action of 7ꢁsubstitutedꢁ7ꢁdeazapurine
nucleosides on
h
ADK and Mtb ADK, the kinetics of compound 1a, which inhibits both ADKs
ADK, were examined. For both enzymes, the Lineweaverꢁ
but is phosphorylated only by
h
Burke double reciprocal plots of initial velocities against increasing concentrations of
adenosine indicated mixed competitive inhibition (Figure 1A, B) with 1a, suggesting
inhibitor binding either into the adenosine or ATP binding sites (or simultaneously to both
sites) with almost comparable affinities. The plots from competition of 1a with ATP in Mtb
ADK also showed mixed competitive inhibition (Fig. 1C) but for hADK nearly parallel lines
for increasing concentration of ATP were detected (Fig. 1D). This type of inhibition can be
interpreted as a mixed uncompetitive inhibition, which suggests preferential binding of
inhibitor into the complex formed between the enzyme and substrate (ATP in this case)
however, does not exclude binding of 1a into the ATP site.
ACS Paragon Plus Environment

Page 13 of 44
Journal of Medicinal Chemistry
12
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1: Kinetics of competition of 1a with adenosine and ATP in Mtb ADK and hADK
binding sites. The panels show the LineweaverꢁBurk plots: (A) Inhibition of Mtb ADK in the
presence of increasing concentrations of adenosine (Ado, a ꢁ 0 µM; b ꢁ 0.5 µM; c ꢁ 1 µM)
and 5 mM ATP; (B) Inhibition of hADK in the presence of increasing concentrations of Ado
(a ꢁ 0, b ꢁ 20 nM, c ꢁ 30nM) and 120 µM ATP; (C) Inhibition of Mtb ADK in the presence of
increasing concentrations of ATP (a ꢁ 0, b ꢁ0 .25 µM, c ꢁ 0.5 µM) and 5 µM Ado; (D)
Inhibition of hADK in the presence of increasing concentrations of ATP (a ꢁ 0, b ꢁ 15 nM, c ꢁ
30 nM, d ꢁ 60 nM) and and 5 µM Ado.
Crystal structures of hADKꢀ1a and Mtb ADKꢀ1a complexes
To explore the binding of 7ꢁsubstitutedꢁ7ꢁdeazapurine nucleosides to human and
mycobacterial ADKs on a structural level, we determined crystal structures of hADK and Mtb
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 44
13
1
2
3
4
5
ADK in complex with 1a, a compound that efficiently inhibits both enzymes and is
specifically phosphorylated by hADK.
6
7
8
9
The hADKꢁ1a complex crystallized in the P21212 space group with two protein
molecules in the asymmetric unit. The crystallographic model was refined to 2.5 Å resolution.
Three molecules of 1a were modeled into a wellꢁdefined electron density map (Figure 1S). In
both hADK molecules in the asymmetric unit, 1a binds into the substrate binding site located
at the interface between a large core domain and a small lid domain. The position of 1a
mimics that of adenosine found in the crystal structure of the hADKꢁadenosine complex (PDB
code 1BX4).20 The overall structure of the hADKꢁ1a complex is very similar to that of the
adenosine bound enzyme. The protein adopts a closed conformation in which the small lid
domain closes over the active site. Compound 1a is completely buried in a hydrophobic
pocket formed by amino acid residues Leu16, Leu40, Leu134, Ala 136, Leu138, and Phe170
(Figure 2). The position of the base ring moiety is stabilized by a stacking interaction with the
Phe170 side chain. The ethynyl moiety of 1a is buried deep in the hydrophobic pocket and
takes the place of a water molecule observed in the structure of hADK with bound adenosine.
Similar replacement of a water molecule by an iodine atom was also observed for 5ꢁiodoꢁ5´ꢁ
deoxytubericidin (PDB code 2I6A).7 The ribose ring of 1a forms numerous direct hydrogen
bonds with residues Asn14, Asp18, Gly64, Asn68, Asn296, and Asp300 (Figure 2A).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 15 of 44
Journal of Medicinal Chemistry
14
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2: Binding of 1a to the adenosine binding site (A) and ATP binding site (B) of hADK
and to Mtb ADK (C). Residues interacting with 1a are shown as sticks. Polar interactions are
shown by magenta dashed lines. Adenosine positions superposed from crystal structures of
hADKꢁadenosine (PDB code 1BX4)20 or Mtb ADKꢁadenosine (PDB code 2PKM)21
complexes are represented as black lines. Panel D shows superposition of the semiꢁopen
conformation of the Mtb ADKꢁ1a complex (magenta) with the open conformation of Mtb
ADK apo form (yellow, PDB code 2PKF) and the closed conformation of the Mtb ADKꢁ
adenosine complex (green, PDB code 2PKM).
In one of the hADK molecules in the asymmetric unit (protein chain A), an additional
molecule of 1a is bound in the ATP binding site located in the core domain about 15 Å from
the adenosine binding site. Partial occupancy within the asymmetric unit points to a lower
affinity of this site for 1a. The following amino acid residues form polar contacts with 1a:
Thr265, Gly267, Arg268, Thr271, Asp286, Gln288, and His324 (Figure 1B). The pose of 1a
closely resembles the binding of adenosine molecule into ATP binding site as found in the
structure of the hADKꢁadenosine complex (PDB code 1BX4).20 The position of the sugar
moieties of 1a and adenosine are very similar, but the positions of the bases are different.
Interestingly, a loop surrounding the ATP binding site (and connecting β14 to α11) is partially
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 44
15
1
2
3
disordered in the hADKꢁ1a structure, and amino acid residues 286ꢁ292 could not be modeled
4
5
6
into a continuous electron density map.
7
8
9
The Mtb ADKꢁ1a complex crystallized in the P31 space group with two protein
molecule per asymmetric unit representing the biologically relevant Mtb ADK dimerThe final
crystallographic model was refined to 2.5 Å resolution with two molecules per asymmetric
unit. The quality of electron density map for residues 9ꢁ48 and 99ꢁ120 is limited suggesting
that the lid domain region is partially disordered. One molecule of 1a bound to the enzyme
active site, however, was modeled into a wellꢁdefined electron density map in both molecules
in the asymmetric unit (Figure 1S). A single molecule of 1a binds into the adenosine binding
pocket located at the interface of the core and lid domains; the ATP binding site remains
unoccupied. Compound 1a is deeply buried in the enzyme’s active site, and its position is
similar to that of adenosine (Figure 2C). The base ring moiety forms a stacking interaction
with Phe131, and N1 engages in a polar interaction with the side chain of Ser19. The ethynyl
moiety of 1a makes a small number of van der Waals interactions with the hydrophobic side
chains of Phe115 and Phe131. The ribose ring of 1a forms numerous direct hydrogen bonds
with the side chains of Asp23, Asn65, Gln187, and Asp274 and the main chain of Gly60
(Figure 2C). The lid domain (residues 9ꢁ48 and 99ꢁ120) adopts a unique conformation in the
Mtb ADKꢁ1a complex. While the conformation of the lid domain is open in the free or ATPꢁ
bound enzyme, it adopts a closed conformation upon binding of adenosine in hADK20 and 2ꢁ
fluor adenosine, the only inhibitor for which the MtbADK complex structure is available
(PDB code 2PKK).21 The conformation of the lid domain in the Mtb ADKꢁ1a complex
structure can be described as semiꢁopen, because its position is in between the open and
closed states (Figure 2D). When the closed conformation of the lid is superposed on the Mtb
ADKꢁ1a complex structure, steric clashes occur between the ethynyl moiety of 1a and
residues Ser130, Phe131, and Phe115. We can therefore conclude that binding of 1a prevents
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 17 of 44
Journal of Medicinal Chemistry
16
1
2
3
catalytically competent arrangement of the closed enzyme. This can serve as a structural
4
5
6
explanation for the observation that 1a is not phosphorylated by Mtb ADK.
7
8
9
Competition of 7ꢀsubstituted 7ꢀdeazapurine derivatives with adenosine and ATP in
Mtb and human adenosine kinases
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
We used 1D H STD NMR experiments to establish the mode of competition of 7ꢁ
hetarylꢁ7ꢁdeazapurine derivatives with adenosine and ATPγS for their respective binding sites
and also to define the binding epitopes of the newly developed inhibitors.
First, we examined the binding behavior and mutual competition of adenosine and
ATPγS to hADK and Mtb ADK. For adenosine in the presence of hADK, the strongest
enhancement was observed for the signal from (Hꢁ1’) and weaker enhancements for two
aromatic proton signals (Hꢁ8) and (Hꢁ2) of ~20% and ~8% relative to (Hꢁ1’) (Table 2),
which suggests that protons (Hꢁ1′) and (Hꢁ8) are in closer contact with the protein than the
rest of the substrate molecule. An analogous experiment for ATPγS yielded much weaker
absolute STD enhancements, which were limited to (Hꢁ1′) and (NH2). In addition, the STD
enhancements induced by binding to hADK of both substrates were not significantly
affected in spectra obtained from their equimolar mixture, which indicates that adenosine
and ATP
γ
S do not compete and bind into their original binding sites.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 44
17
1
2
3
4
5
6
7
8
Table 2: The relative STD enhancements of substrate (adenosine or ATPγS) and inhibitor
signals in the presence of hADK or Mtb ADK. Concentration of all nucleosides was kept at
the 500 µM.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
hADK
Mtb ADK
Ado ATP
Ado ATP
1a 1m 1u 1ae 1g
1a 1m 1u 1ae 1g
(%)
(%)
Hꢁ1' 100 100 100 100 100 100 100 100 100 100 100 100 100 100
Hꢁ2
8
<1
<1
33
17
57 72 33 64 44
41 72 39 36 58
52 52 60 74
57
24
63
54
37
33
34
45
42 50 43
42 68 29
37 83 38
31
Hꢁ8 20
Ar
Ar'
78
67
78
Ar''
39
The STD spectra for adenosine in the presence of Mtb ADK yielded maximum
enhancement for (Hꢁ1′) and moderate enhancement for (Hꢁ8) and (Hꢁ2) of ~58% and ~44%
(Figure 3A; Table 2), suggesting a similar contact epitope to that observed for hADK. The
overall STD enhancement for ATPγS in the presence of Mtb ADK was comparable to that of
adenosine, again with (Hꢁ1′) being the most affected proton (Figure 3B; Table 2). However,
certain differences were found for the remaining ATPγS protons in comparison with their
binding to hADK. There was a very weak enhancement of the (NH2) signal (<5%), and the
(Hꢁ2) STD signal (~57%) was stronger than the (Hꢁ8) signal (~24%). Importantly,
simultaneous addition of adenosine and ATPγS to MtbADK led to a dramatic reduction in
the ATPγS enhancements, suggesting direct competition between binding of adenosine and
ATPγS, as illustrated in Figure 3C. Intracellular ATP levels are generally in the low
millimolar range,22ꢁ26 and adenosine is present at micromolar concentrations.27 Therefore, we
ACS Paragon Plus Environment

Page 19 of 44
Journal of Medicinal Chemistry
18
1
2
3
4
5
can speculate that ATP binds into Mtb ADK first, and conformational changes in ADK
allow adenosine binding and phosphorylation in bacteria.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3: 1D 1H STD NMR substrate/inhibitor competition experiments for Mtb ADK. Panel
(A) shows a region of the STD spectrum obtained for Mtb ADK in the presence of adenosine,
while panel (B) shows the spectrum for the enzyme in the presence of ATPγS. Panel (C)
shows the spectrum for the mixture of substrates, with intensity of the ATPγS specific signals
significantly reduced. Panels (D) and (E) show competition of 1a with adenosine and ATPγS
illustrated by the clear drops in the STD enhancement of signals from both substrates.
We then selected compounds containing either smaller (1a, 1g, 1m) or bulkier (1u,
1ae) substituents in position 7. They displayed different (1g, 1ae) or comparable (1a, 1m,
1u) inhibition activities against human and mycobacterial ADK. We tested their competition
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 44
19
1
2
3
4
with adenosine and ATP
γ
S in the presence of either hADK or Mtb ADK. All selected 7ꢁ
5
6
7
8
deazaadenosines selectively competed with adenosine, as their STD enhancements were
significantly reduced, but did not affect ATP S signals in the presence of hADK. This
suggests significantly lower affinity of tested compounds to ATP site relatively to ATP S.
γ
9
γ
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The ATP site was occupied by 1a in our crystal structure of hADK but this was due to high
inhibitor excess in the crystallization experiment. However, the STD data in the presence of
Mtb ADK indicated strong competition of inhibitors with both substrates (Figure 3D/E),
which confirmed that these compounds can also bind into the ATP site. The STD
enhancements for the substrates and inhibitors normalized to the (Hꢁ1′) signal enhancement
are shown in Table 2. We observed that (Hꢁ1′) is subjected to the strongest STD
enhancement for all inhibitors tested, independent of the protein present in sample solution.
This suggests that the sugar moiety is in direct contact with the protein in all complexes. The
substantial contributions from the purine protons (Hꢁ2) and (Hꢁ8) are found in all complexes,
with approximately 50% enhancement relative to (Hꢁ1′). Proteinꢁdependent differences were
found in the STD enhancements of the signals from substituents at position 7. The relative
enhancements for aromatic substituent signals were higher than (Hꢁ2) and (Hꢁ8)
enhancements in hADK complexes, while they remained comparable in Mtb ADK
complexes, suggesting that the substituents maintain much more extensive contacts with
hADK than Mtb ADK. The potential binding of 7ꢁdeazanucleosides into the ATP binding
site contributes to the loss of their phosphorylation.
Binding of selected 7ꢀhetarylꢀ7ꢀdeazapurine derivatives to human ADK by proteinꢀ
detected NMR
We used twoꢁdimensional NMR spectroscopy to study the detailed mechanism of
inhibitor binding to hADK. The low stability of Mtb ADK at the high concentration (100 ꢀM)
ACS Paragon Plus Environment

Page 21 of 44
Journal of Medicinal Chemistry
20
1
2
3
4
5
required for relatively long NMR experiments did not allow us to prepare a sample of
sufficient quality for binding experiments. Initially, we analyzed 2D 15N/1H HSQC spectra of
6
reference samples of 15Nꢁlabeled hADK in the absence and presence of adenosine or ATP
γS.
7
8
9
The 2D 15N/1H HSQC spectrum of free hADK consisted of 270 unique signals from backbone
and side chain amide groups, some subjected to extensive spectral overlaps, which
corresponds to the size of the analyzed protein. Both ligands induced a number of specific
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
15
changes in the positions of signals in 2D N/1H HSQC spectra, indicating that this approach
is suitable for studying binding modes of hADK inhibitors.
Preliminary NMR experiments using 15N labeled material revealed a relatively poor
long term stability of hADK at temperatures above 25 °C. The 3D experiments required for
backbone resonance assignment were acquired at 25 °C to minimize protein precipitation
and overall material consumption. The exchange of nonꢁlabile protons for deuterons enabled
sequenceꢁspecific backbone resonance assignments; however, a certain proportion of amide
groups unexpectedly did not backꢁexchange during protein purification in an H2O
environment and remained deuterated, as suggested by comparison of the 2D 15N/1H HSQC
15
2
spectra obtained from N and H/13C/15Nꢁlabeled hADK. The high quality of NMR spectra
allowed us to assign essentially all 175 detectable signals in perdeuterated hADK using a
previously described approach.29,30 Detailed analysis revealed that the amide groups buried
in the protein core and contributing as hydrogen bond donors were often not successfully
protonated during purification in an H2O environment. The distribution of successfully
assigned backbone amide groups is illustrated in Figure S2.
15
As a next step, we acquired and analyzed 2D N/1H HSQC spectra of hADK bound
to the set of compounds (1a,g,m,u,ae) investigated by NMR 1D 1H STD experiments. These
data were compared with the reference spectra; we selected 17 wellꢁresolved representative
backbone amide signals in the spectrum of free hADK (Table 3 and Figure 4). All
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 44
21
1
2
3
4
5
6
7
8
9
compounds studied, with the exception of 1ae, which contains a large 4ꢁphenoxathiinyl
group, induced changes in 2D 15N/1H HSQC spectra that closely correspond to those
observed for adenosine binding, suggesting direct competition of inhibitors for the adenosine
binding pocket. The spectrum of hADK in the presence of 1ae exhibited signal changes that
did not unambiguously correspond to any reference spectrum, indicating that the binding
mode of this inhibitor might be atypical or mixed. Additional analysis using backbone
resonance assignments revealed that this compound induced specific differences compared
to the reference spectrum acquired for hADK in the presence of adenosine. Similar data were
obtained for the isopropylphenyl derivative (1u) with a relatively large substituent in
position 7.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4: Binding of inhibitors to hADK analyzed using 2D 15N/1H HSQC. Each row
represents a different backbone amide group crossꢁpeak (the residue number is given in the
first column). Columns illustrate changes in the spectra obtained for a selected inhibitor.
Contours are colored as follows: blue, free hADK; green, hADK in the presence of ATPγS;
red, hADK in the presence of adenosine; magenta, hADK in the presence of inhibitor. .
Concentration of all nucleosides was kept at the 500 µM.
ACS Paragon Plus Environment

Page 23 of 44
Journal of Medicinal Chemistry
22
1
2
3
4
Table 3: Analysis of hADK 2D 15N/1H HSQC spectra in the presence of inhibitors. We
selected a set of wellꢁresolved backbone amide signals and followed their changes upon
addition of selected inhibitors. Free hADK and the hADK in the presence of adenosine or
ATPγS were used as reference samples. The signals were sorted into four distinct categories
according to their behavior in the presence of an inhibitor.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Number of signals corresponding to the reference of
Free
Inhibitor hADK
hADK+adenosine hADK+ATPγS Specific signal
5
3
3
5
6
13
14
13
11
6
1
1
5
3
8
0
0
1
2
2
1a
1g
1m
1u
1ae
Quantum mechanical analysis of binding of selected 7ꢀhetarylꢀ7ꢀdeazapurine derivatives
to human and Mtb ADK
Complexes of hADK and Mtb ADK with the compounds used for NMR competition
experiments (1a,g,m,u,ae) were further analyzed using semiempirical quantum mechanical
(SQM) based scoring function.31,32 To build in silico models of hADK complexes, we used the
Xꢁray structure of hADK in complex with two adenosine molecules (PDB code 1BX4).20 In
this structure, one adenosine molecule is located in the adenosine binding site and the second
in the ATP binding site. The inhibitors were built and scored in both binding sites. The
complex of hADK with 1ae could not be built due to numerous steric clashes between the
inhibitor and protein. The calculated scores for the remaining inhibitors are summarized in
Table 4. The inhibitors have higher affinity to the adenosine binding site (scores from ꢁ30 to ꢁ
65 kcal/mol) than to the ATP binding site (scores from ꢁ12 to ꢁ23 kcal/mol). This is due to the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 44
23
1
2
3
4
5
6
7
8
9
more favorable interactions in the adenosine binding site (ꢀEint more negative by about 60%).
These results are in good agreement with our experimental NMR finding that the inhibitors
compete with higher affinity with adenosine than with ATP in the presence of hADK.
Compounds 1a, 1g, and 1m fit well into the adenosine binding site. The 7ꢁdeazaderivative 1u
is too big to fit into the conformation of hADK that was used for modeling, and its binding
requires changes in the protein conformation (see Figure S3). The conformational changes
significantly increased the penalty for the change of the conformational “free” energies
(ꢀG'wint) of both the protein and ligand (by about 100 and 40%, respectively). This supports
the NMR results showing that compounds with large substituents (1u and 1ae) induce
different structural changes in hADK than adenosine or compounds with small substituents in
position 7.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 4. Inhibition constant (IC50) and score of the studied inhibitors to hADK calculated as a
sum of the gasꢁphase interaction energy (ꢀEint), the interaction solvation/desolvation free
energy (ꢀꢀGsolv), the change of the conformational “free” energies of the protein and ligand
(ꢀG'wconf(P,L)), and the entropy change upon binding (ꢁTꢀSint).
Human ADK
Inhibitor
IC50
Score
∆Eint
∆∆Gsolv ∆G'wconf(L) ∆G'wconf(P) ꢁT∆Sint
(µM)
(kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol) (kcal/mo
l)
Adenosine binding site
0.2 ± 0.015
ꢁ59.7
ꢁ149.6
59.3
7.6
23
0
1a
ACS Paragon Plus Environment

Page 25 of 44
Journal of Medicinal Chemistry
24
1
2
3
4
5
6
7
8
9
>5
ꢁ65.0
ꢁ56.6
ꢁ30.6
ꢁ158.4
ꢁ157.2
ꢁ163.9
61.3
61.3
68.1
8.6
9.1
23.5
30.1
49.7
0
0
4
1g
1m
1u
2.3 ± 0.17
3.3 ± 0.4
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
11.5
ATP binding site
0.2 ± 0.015
>5
ꢁ22.6
ꢁ22.4
ꢁ22.9
ꢁ11.7
ꢁ93.5
ꢁ97.1
ꢁ95.8
ꢁ101.5
43.8
47.3
49.4
52.7
2.4
2.5
2.2
4.6
24.7
24.8
21.3
28.5
0
0
0
4
1a
1g
2.3 ± 0.17
3.3 ± 0.4
1m
1u
Mtb Adk complexes with the selected inhibitors were also studied using the SQM
scoring function. Previous structural analyses of Mtb ADK show that binding of ATP or its
analog does not change the conformation of this enzyme. However, the Xꢁray structure of
Mtb ADK crystallized with 1a (PDB code 4PVV) shows that the binding of adenosine analog
results in movement of lid domain and formation of a semiꢁclosed conformation. This
suggests that inhibitors compete with adenosine in an open conformation. Subsequently, the
binding affinity of the adenosine analogs to the adenosine site might increase when the
conformation of Mtb ADK is changed from open to semiꢁopen. To support this hypothesis,
we used both the Xꢁray structure of Mtb ADK dimer in open conformation in complex with
an ATP analog (PDB code 2PKN)21 and the X ray structure of semiꢁopen conformation of
Mtb ADK crystallized with 1a to build our in silico models of Mtb ADK complexes. The
closed conformation of Mtb ADK (Mtb ADK dimer crystallized with adenosine, PDB code
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 44
25
1
2
3
4
5
6
7
8
9
2PKM) did not allow us to build models. The calculated scores are summarized in Table 5,
and the SQM optimized complexes of 1a and 1ae are shown in Figure 5. Interestingly, the
inhibitors have comparable affinity to both the adenosine and ATP binding sites in the open
conformation of Mtb ADK (scores ranges from ꢁ20 to ꢁ25 and from ꢁ11 to ꢁ17 kcal/mol to the
ATP and adenosine binding sites, respectively). However, the affinity of 7ꢁsubstitutedꢁ7ꢁ
deazapurine nucleosides to the adenosine binding site increases significantly in the semiꢁopen
conformation of Mtb ADK with bound nucleosides (score from ꢁ50 to ꢁ59 kcal/mol) and is
comparable to those calculated for these inhibitors bound in hADK adenosine site.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 5. Inhibition constant (IC50) and score of inhibitors of Mtb ADK calculated as a sum of
gasꢁphase interaction energy (ꢀEint), the interaction solvation/desolvation free energy
(ꢀꢀGsolv), the change in conformational “free” energies of the protein and ligand
(ꢀG'wconf(P,L)), and the entropy change upon binding (ꢁTꢀSint).
Mtb AdK
Inhibitor
IC50
Score
∆Eint
∆∆Gsolv ∆G'wconf(L) ∆G'wconf(P ꢁT∆Sint
)
(µM)
(kcal/mo (kcal/mol) (kcal/mol) (kcal/mol)
l)
(kcal/mol)
(kcal/mol)
Adenosine binding site in semiꢁopen conformation
0.33 ± 0.02 ꢁ57.0
ꢁ131.3
ꢁ140.7
ꢁ141.5
ꢁ140.8
63.3
66.9
68.0
69.5
2.9
3.4
4.9
3.8
8.1
0
0
0
4
1a
1g
2.1 ± 0.15
4.5 ± 0.2
ꢁ59.2
ꢁ57.2
ꢁ51.4
11.3
11.4
12.1
1m
1u
2.33 ±
0.25
0.3 ± 0.02 ꢁ49.8
ꢁ142.9
72.0
5.2
15.9
0
1ae
ACS Paragon Plus Environment

Page 27 of 44
Journal of Medicinal Chemistry
26
1
2
3
4
Adenosine binding site in open conformation
5
6
7
8
0.33 ± 0.02 ꢁ17.4
ꢁ72.5
ꢁ77.8
ꢁ72.3
ꢁ77.5
44.4
47.3
48.8
44.7
4.4
5.3
2
6.3
8.9
0
0
0
4
1a
1g
2.1 ± 0.15
4.5 ± 0.2
ꢁ16.4
ꢁ12.9
ꢁ13.3
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
8.6
1m
1u
2.33 ±
0.25
5.4
10.1
0.3 ± 0.02 ꢁ11.4
ꢁ75.1
47.2
5.3
11.2
0
1ae
ATP binding site in open conformation
0.33 ± 0.02 ꢁ23.8
ꢁ76.2
ꢁ78.8
ꢁ75.4
ꢁ84
43.6
45.5
44.6
46.1
47.2
0.9
1.1
1.3
2.9
2.1
8
0
0
0
4
0
1a
1g
2.1 ± 0.15
4.5 ± 0.2
ꢁ24.8
ꢁ20.1
7.4
9.4
1m
1u
2.33 ± 0.25 ꢁ20.3
0.3 ± 0.02 ꢁ24.3
10.7
10.3
ꢁ83.8
1ae
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 44
27
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Model of 1a bound to adenosine binding site in semiꢁopen conformation (A), ATP
binding site in open conformation (C) and adenosine binding site in open conformation (E)
of Mtb ADK and 1ae bound to adenosine binding site in semiꢁopen conformation (B), ATP
binding site in open conformation (D) and adenosine binding site in open conformation (F) of
Mtb ADK.
ACS Paragon Plus Environment

Page 29 of 44
Journal of Medicinal Chemistry
28
1
2
Conclusions
3
4
5
6
7
8
A new series of 7ꢁ(het)arylꢁ7ꢁdeazaadenine ribonucleosides bearing small and bulky
substituents in position 7 was synthesized and evaluated as potential new antimycobacterial
agents. Several compounds bearing smaller substituents (e.g., 1a) were found to inhibit both
human and Mtb ADK and were cytotoxic, whereas several bulky derivatives (e.g., 1ad, 1ae,
1ag) were specific inhibitors of the Mtb enzyme and were not cytotoxic.7ꢁ(2ꢁNaphthyl)ꢁ7ꢁ
deazaadenine (1f) was the most active compound against Mtb strain My331/88 and drugꢁ
resistant strain Praha 131 in vitro (MIC = 2 or 4 ꢂM, respectively), but its cytotoxicity
disfavors its further development. The dibenzofuran derivative (1ad) had the best therapeutic
index. This compound was a submicromolar MtbꢁADKꢁspecific inhibitor and was active
against Mtb strains with a MIC of 4 µmol/L. This compound is a promising lead structure for
further drug development. Following binding of the 7ꢁ(het)arylꢁ7ꢁdeazaadenine
ribonucleosides into the adenosine site, Mtb ADK adopts a unique semiꢁopen conformation of
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
the lid domain and hADK the close conformation. Enzyme kinetics and 1D H STD NMR
analysis of the inhibitors’ competition with adenosine and ATPγS showed that 7ꢁ(het)arylꢁ7ꢁ
deazaadenine ribonucleosides bind preferentially into the adenosine site in hADK and are
readily accommodated in both, to the adenosine and ATP sites in Mtb ADK. Quantum
mechanical analysis indicated that these compounds have higher affinity for the ATP site of
Mtb ADK in the open conformation but movement of the lid domain, increases binding
affinity of these inhibitors in the adenosine site. Both the preferential binding of 7ꢁ(het)arylꢁ7ꢁ
deazaadenine ribonucleosides in the Mtb ADK ATP binding site and the unique semiꢁopen
conformation of Mtb ADK during inhibitor binding in the adenosine site explain the
observation that 7ꢁ(het)arylꢁ7ꢁdeazaadenine ribonucleosides are not phosphorylated by ADK.
Similar mechanisms may also occur for deazaadenine ribonucleoside derivatives modified at
other positions.
ACS Paragon Plus Environment