Discovery and structure-based design of 4,6-diaminonicotinamides as potent and selective IRAK4 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2017.09.029

The identification of small molecule inhibitors of IRAK4 for the treatment of autoimmune diseases has been an area of intense research. We discovered novel 4,6-diaminonicotinamides which potently inhibit IRAK4. Optimization efforts were aided by X-ray crystal structures of inhibitors bound to IRAK4. Structure activity relationship (SAR) studies led to the identification of compound 29 which exhibited sub-micromolar potency in a LTA stimulated cellular assay.

Full text of DOI:10.1016/j.bmcl.2017.09.029

Accepted Manuscript
Discovery and structure-based design of 4,6-diaminonicotinamides as potent
and selective IRAK4 inhibitors
Rajeev S. Bhide, Alec Keon, Carolyn Weigelt, John S. Sack, Robert J. Schmidt,
Shuqun Lin, Hai-Yun Xiao, Steven H. Spergel, James Kempson, William J.
Pitts, Julie Carman, Michael A. Poss
PII:
S0960-894X(17)30925-3
DOI:
http://dx.doi.org/10.1016/j.bmcl.2017.09.029
BMCL 25293
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
12 May 2017
12 September 2017
13 September 2017
Please cite this article as: Bhide, R.S., Keon, A., Weigelt, C., Sack, J.S., Schmidt, R.J., Lin, S., Xiao, H-Y., Spergel,
S.H., Kempson, J., Pitts, W.J., Carman, J., Poss, M.A., Discovery and structure-based design of 4,6-
diaminonicotinamides as potent and selective IRAK4 inhibitors, Bioorganic & Medicinal Chemistry Letters (2017),
doi: http://dx.doi.org/10.1016/j.bmcl.2017.09.029
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Bioorganic & Medicinal Chemistry Letters
journal homepage: www.els evier.com
Discovery and structure-based design of 4,6-diaminonicotinamides as potent and selective IRAK4
inhibitors
Rajeev S. Bhide∗ a, Alec Keona, Carolyn Weigeltb, John S. Sackb, Robert J. Schmidta, Shuqun Lina, Hai-
Yun Xiaoa, Steven H. Spergela, James Kempson, William J. Pittsa, Julie Carmanc, and Michael A. Possa
aDiscovery Chemistry, Research & Development, Bristol-Myers Squibb Company, Route 206 & Province Line Road, Princeton, NJ 08543
bMolecular Structure and Design,, Research & Development, Bristol-Myers Squibb Company, Route 206 & Province Line Road, Princeton, NJ 08543
cImmunoscience Biology, Research & Development, Bristol-Myers Squibb Company, Route 206 & Province Line Road, Princeton, NJ 08543
ARTICLE INFO
ABSTRACT
Article history:
Received
The identification of small molecule inhibitors of IRAK4 for the treatment of autoimmune
diseases has been an area of intense research. We discovered novel 4,6-diaminonicotinamides
which potently inhibit IRAK4. Optimization efforts were aided by x-ray crystal structures of
inhibitors bound to IRAK4. Structure activity relationship (SAR) studies led to the identification
of compound 29 which exhibited sub-micromolar potency in a LTA stimulated cellular assay.
Revised
Accepted
Available online
Keywords:
2009 Elsevier Ltd. All rights reserved
.
IRAK4
4,6-diaminonicotinamide
Rheumatoid arthritis
Autoimmune disease
Treatment of autoimmune diseases such as rheumatoid
arthritis (RA), inflammatory bowel disease (IBD) and Systematic
Lupus Erythematosus (SLE) has been an area of intense research
over the past two decades. Various mechanisms that impact
innate and/or adaptive immunity have been targeted for the
management of these autoimmune disorders.1
generally resulted in modest improvements in IRAK4 potency
(compare 1 with 4-6 and 8), with the notable exception of 7
which did not inhibit IRAK4 at the concentrations tested. The
ethanolamine substituents (2, 3) at C4 also improved the IRAK4
potency by about 2-fold, with 2 being notable due to its
significant decrease in JAK3 potency. However, the SAR was
not immediately apparent from these compounds due to changes
at both C2 and C4. While the observations discussed above could
have formed the basis of a chemistry SAR plan, we chose to first
examine JAK3 and IRAK4 inhibitor complexes to better guide
our optimization effort.
Interleukin-1 Receptor Associated Kinase 4 (IRAK4) is a
serine/threonine kinase that regulates signaling through various
TLRs including TLR2, 4, 5, and 7-9 which are involved in the
innate immune response.2 IRAK4 also regulates signaling
through members of the IL-1 family, and can therefore impact
adaptive immunity as well.3 Cells from IRAK4 deficient humans
fail to respond to agonists of the TLRs described above, and to
members of the IL-1 family.4 These patients are highly
susceptible to bacterial infections in childhood, but this risk
decreases with age.5 IRAK4 kinase-dead knock-in mice have
shown beneficial effects in mouse models of arthritis6 and
therefore, discovery of small molecule inhibitors of IRAK4 has
become an important subject of research.7,8 Recently, a potent
IRAK4 inhibitor, PF-06650833, has entered phase 2 clinical trials
(NCT02996500) for the treatment of RA9 and BAY 1834845 has
entered phase 1 clinical trials (NCT03054402).
Table 1. IRAK4 hit identification.
R1
O
HN
N
4
H2N
5
2
R2
N
H
A review of kinase enzyme inhibition data for compounds
from our JAK3 program led to the identification of a set of
analogs from a 4,6-nicotinamide series which showed moderate
inhibition of IRAK4 (Table 1). Analysis of this modest set of
compounds showed that substituents on the aryl group at C2

Additional structural information was then obtained from an
X-ray co-crystal structure of 4 in IRAK4 (Figure 2) which
revealed that 4 was bound to IRAK4 in a ‘flipped’ binding mode,
compared to the mode observed in the JAK3 crystal structure. In
this ‘flipped’ mode, hydrogen bonding interactions between the
NH at C2, the pyridyl nitrogen, and the NH of the amide of 4
with the hinge region of IRAK4 form a ‘classical’ triad hinge
binding interaction. Another feature of this binding mode is that
the C2 substituent of 4 is oriented towards the Tyr262 gatekeeper
in an apparent π stacking interaction. We were pleased by this
observation as the IRAK family of kinases are the only ones
which possess a tyrosine gatekeeper residue. We hoped that by
engaging this interaction, improvements in kinome selectivity
could be realized. In addition, the C2 substituent of 4 forms a
hydrogen bond between the indolone carbonyl and the amine of
Lys213. Ligand docking experiments of the compounds in Table
1 into this IRAK4 crystal structure revealed that 7 failed to
achieve a suitable pose, presumably due to the inability of the
gatekeeper pocket to accommodate the extended amide. In
contrast, compounds 1-3, 5, 6 and 8 could be docked in a manner
similar to 4. We also noticed that the primary amide group of 4
was oriented towards the solvent exposed region and the
Shellman loop, whereas in JAK3, the primary amide was
oriented towards the bulky gatekeeper, Met902. Based on these
observations, we reasoned that introduction of a substituent on
the amide group should further disfavor binding to JAK3, but
may be permissible in IRAK4, thereby possibly affording
selectivity for IRAK4 against JAK3.
IRAK4a JAK3a
Compound
R1
R2
IC50
(nM)
IC50
(nM)
1800
730
840
360
390
470
5
1
2
3
4
5
6
1000
8
3
9
7
>10000
2
7
180
3
8
aKinase activity was measured by Caliper assays, minimum in
duplicate, average IC50 reported.10
Initially, we were intrigued by extended amide 7, a potent
inhibitor of JAK3 that also displayed high selectivity against
IRAK4, and therefore obtained an X-ray co-crystal structure of 7
in JAK3. From the structure, it was noted that the pyridine
nitrogen and NH of the CONH2 group of 7 bind to the hinge
region of JAK3 kinase (Figure 1) while the C2 substituent
projects out towards solvent.11 In comparison, IRAK4 has several
unique structural features including a loop insertion in the N-
terminal region (Shellman loop) which is in close proximity to
the ATP binding site.12 One hypothesis for the poor IRAK4
potency of 7 was the possibility of an unfavorable steric
interaction between the bulky C2 substituent of 7 and the
Shellman loop region of IRAK4. Given the unique structural
features of IRAK4, an alternative hypothesis considered that
compounds 1-6, and 8 might be binding to IRAK4 in a different
binding mode.
Figure 2. Crystal structure of 4 in IRAK4 kinase domain.
Residues possibly involved in hydrogen bonds are labeled and
bonds are shown by dotted lines. (PDB code 5W84).
To test these assertions, we proceeded to expand the scope of
amide substitution at the C5 position. We chose the C4
substituent of 2 in an effort to disfavor JAK3 binding and the C2
substituent of 8, as that substituent was present in the most potent
IRAK4 inhibitor. Compounds were prepared by the route
presented in Scheme 1, with minor variations.10
Table 2. SAR at C5 amide.
Figure 1. Crystal structure of 7 in JAK3 kinase domain. Residues
possibly involved in hydrogen bonds are labeled, and bonds are
shown by dotted lines. (PDB code 5W86).

IRAK4a
IC50
(nM)
JAK3a
IC50 (nM)
R
Figure 3: Crystal structure of 10 in IRAK4 kinase domain.
Residues possibly involved in hydrogen bonds are labeled and
bonds are shown by dotted lines. (PDB code 5W85).
Compound
-NH2
19
10
3,500
1,600
3,400
5,400
3,600
8,600
29,000
9
With these positive initial results in hand, we turned our
attention toward the exploration of the SAR at C4 while
maintaining the benzothiazole group at C2 and the methyl amide
at C5.
10
11
12
13
14
15
-NHCH3
64
Table 3. SAR studies at C4.
50
70
IRAK4
IC50
(nM)
JAK3/
IRAK4
ratio
R
R1
Compound
50
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
Me
H
730
13
>50
430
300
200
100
86
16
-NH(CH3)2
14,600
a Kinase activity was measured by Caliper assays, minimum in
duplicate, average IC50 reported.10
17
18
19
20
21
22
23
24
25
26
27
28
29
30
From this set of analogs, we were gratified to find that our
design of compounds 9-14 (Table 2) resulted in a significant
improvement in IRAK4 potency over the initial leads (2, 8, Table
1). As we had hypothesized amide 9 was a potent inhibitor of
IRAK4. It appears that the C-2 substituent contributed
significantly to maintaining selectivity over JAK3. Amide
substitutions (10-14) were well tolerated in IRAK4, although
increasing the size of the alkyl group on the amide (11-14)
resulted in a modest decrease in IRAK4 potency compared to 10.
Finally, as anticipated from our examination of the IRAK4
crystal structure (Figure 1), the dimethyl analog 15 showed a
significant loss in IRAK4 potency, presumably due to the loss of
a hydrogen bond interaction with the amide carbonyl of Met265,
or alternatively due to steric clash with the IRAK4 hinge region
or a combination of both effects.
20
47
130
82
170
40
200
200
200
488
800
12
25
We were able to obtain an X-ray crystal structure of 10 bound
to IRAK4 (Figure 3) which confirmed the binding mode
observed in Figure 2. Comparing the crystal structures of 4 and 9
suggested that the bicyclic rings at C2 may be involved in a π
stacking interaction with the gatekeeper residue (Tyr262).
Additionally, a water molecule was observed that appeared to
interact with the ligand as well as with Lys213, Glu233 or
Asp329. Furthermore, it was noted that the hydroxyl on the
phenylpropanol moiety at C4 may be involved in a hydrogen
bond with Asp272.
11
21
2,000
50
230
1400
1
Gatekeeper
Tyr262
Val263
5
Glu233
Met265
Lys213
Asp329
20
a Kinase activity was measured by Caliper assays, minimum in
duplicate, average reported.10
As indicated in Table 3, incorporation of a simple methyl
group at C4 (16) resulted in a significant decrease in potency. In
contrast, cyclopentyl and cyclohexyl groups (17, 18) improved
potency over benzyl substitution (17, 18 vs 8). We had
Ala315
Asp272

anticipated that introduction of an alcohol group on the
cyclohexane would improve both potency and selectivity by
engaging in an interaction with Asp272. However, among the
various stereo- and regio- isomeric cyclohexyl alcohols (19-21,
23-25) or simple alkyl alcohols (22, 28) that were prepared, no
significant improvement in potency was observed, however
introduction of a diol group on the benzyl amine moiety (29) did
provide a modest improvement in potency. It is interesting to
note the significant decrease in JAK selectivity between the
alcohol 9 and diol 30. This highlights the contribution of a
secondary amide at this position to improved selectivity over
JAK3.
membrane permeability assay (PAMPA), but exhibited poor
cellular potency (>1500 nM). The cause for this apparent
disconnect is unknown. To probe this anomaly and as part of an
effort to identify compounds with improved cellular activity and
acceptable metabolic stability, SAR studies were conducted at C2
in combination with the methyl amide at C5 and the optimized
diol group at C4. Groups at C2 were chosen such that we could
study the effect of varying cLogP, number of hydrogen bond
donors, groups known to block metabolic hotspots and improve
solubility.
With the exception of 29, all of the compounds in Table 3
showed acceptable permeability (>150) in the parallel artificial
Table 4. SAR at C2.
LTA_IL6
PBMCb
IC50 (nM)
PAMPA
(nm/sec) @ pH
7.4
IRAK4a
IC50 (nM)
HLM %
remainingc
MsLM %
remainingc
R
Compound
5
300
1800
690
19
6
90
45
91
59
72
98
83
65
80
21
99
69
42
15
73
36
41
20
29
5
31
32
33
10
4
6
1600
5000
4100
1300
1600
3000
6.5
11
51
111
70
63
13
5
34
35
36
37
10
10
13
38
aKinase activity was measured by Caliper assay. bLTA-IL6 assay.10 cHuman and mouse liver microsome assay. All assays were done at a
minimum in duplicate, average reported.
As depicted in Table 4, all of the compounds (29-38) were
potent inhibitors of IRAK4, although none exhibited improved
cellular potency as compared to 29. All of the compounds also
exhibited poor permeability (PAMPA < 150). Despite the poor
permeability for compound 29, upon further testing, it showed
acceptable potency in a lipoteichoic acid (LTA) stimulated
cellular assay in which inhibition of IL6 cytokine generation was
measured (30-38 were all > 1µM).10 In a human whole blood
(LTA-IL6) assay10 29 also exhibited good potency (IC50 = 300
nM). Furthermore, 29 was greater than 50-fold selective when
tested against an in house set of kinases using a caliper assay,
with the exception of LCK and MAP4K1 (HIPK1). In addition to
LCK and HPK1, kinome wide screening (KINOMEscan®)
revealed potent binding (< 10% of control) against DDR1,
DDR2, FLT3, FLT1, MUSK, MER, RIOK2, TRKA, TRKB, and
TIE1. In general, most of the compounds in this series showed
acceptable stability in a human liver microsomal assay. However,
due to poor murine metabolic stability (31% remaining after 10
min incubation), compound 29 was not progressed into mouse
PK studies.
Synthesis of 29 is shown in Scheme 1. Methyl 4,6-
dichloronicotinate (39) was hydrolyzed with LiOH in a 2:1
mixture of ethanol/water to afford acid 40.10 Acid 40 was
converted to the corresponding acid chloride using oxalyl

chloride in the presence of catalytic DMF, and this mixture was
then treated with methylamine to obtain methyl amide 4,6-
dichloro-N-methylnicotinamide (41). Next, a solution of 41 was
heated in the presence of (2S,3S)-3-amino-3-phenylpropane-1,2-
diol and diisopropylethylamine (DIPEA) at 120 °C for 3 hours to
obtain diol 42. Finally, 6-aminobenzothiazole (43) and
compound 42 were heated together in a sealed tube for 2 hours to
obtain 29.
In conclusion, we have discovered a novel series of potent and
selective IRAK4 inhibitors. This effort was aided by X-ray
crystallography studies which revealed that the compounds bind
to IRAK4 in a ‘flipped’ binding mode relative to the binding
mode observed in JAK3 crystal structures for compounds from
the same series. Key interactions included a classic hinge binding
motif, and an apparent π-π stacking interaction of the C2
substituents with the unique tyrosine gatekeeper in IRAK4.
Attempts to improve binding by targeting a hydrogen bond to the
carboxylic acid of Asp272 were less productive. Further studies
describing optimization of the cellular potency and ADME
properties of compounds in this series will be the subject of
future disclosures.
Acknowledgement
We would like to thank John Hynes for reviewing the manuscript.
Also, we wish to thank the MDT team at Bristol-Myers Squibb,
Dianlin Xie for the baculovirus expression of IRAK4 protein, Susan
Kiefer for the IRAK4 protein purification, and Daniel Camac for the
co-crystallization of 4, 10. Lastly, we would like to thank Proteros
Biosciences GmBH for the co-crystal structure of 7 with JAK3.
Scheme 1. a) LiOH, EtOH/H2O (2:1), 22 oC, 4 h; b) Oxalyl chloride, DMF
followed by methyl amine, 54% yield; c) (2S,3S)-3-amino-3-phenylpropane-
o
1,2-diol, DIPEA, 120 oC, 3 h; d) 6-aminobenzothiazole, 1.2 equiv., 150 C, 2
h, 47% yield.
Saiah, E; Shin, J; Soutter, H; Strohbach, J; Symanowicz, P;
Thaisrivongs, S; Thomason, J; Trzupek, J; Vargas, R;
Vincent, F; Wang, X; Winkler, A; Wright, S; Yan, J; Zapf,
C; MEDI-261 Abstracts of Papers, 251st ACS National
Meeting & Exposition, San Diego, CA, U.S., March 13-17,
2016. b)Lee, K. L.;Ambler, C. M.; Boscoe, B. P.; Bree, A.
G.; Bodfuehrer.: Cvhang, J. S.; Choi, C.; Chung, S.;
Curran, K. J.; Day, J. E.; Dehnhardt, C. M.; Dower, K.;
Drozda, S. E.; Frisbie, R. K.; Gavrin, L. K.; Goldberg, J.
A.; Han, S.; Hegen, M.; Helpworth, D.; Hope, H. R.;
Kamtekar, S.; Kilty, I. C.; Lee, A.; Lin, L-L.; Lovering, F.
E.; Lowe, M. D.; Mathias, J. P.; Morgan, H. M.; Murphy,
E. A.; Papaioannaou, N..; Patny, A.; Pierce, B. S.; Rao, V.
R.; Saiah, E.; Samardjiev, I. J.; Samas, B. M.; Shen, M.W.
H.; Shin, J. H.; Soutter, H. H.; Strobbach, J. W.;
Symanowicz, P. T.; Thomason, J. R.; Trzupek, J. D.;
Vargas, R.; Vincent, F.; Yan, J.; Zapf, C. W.; Wright, S.
W. J. Med Chem. 2017, 60, 5521-5542.
References and notes
1. Navegantes, K. C.; de Souza Gomes, R.; Aparecida Tartari
Pereira, P.; Czailkoski, P. G.; Heitmann Mares Azevedo,
C.; Chagas Monteiro, M. J. Transl. Med. 2017, 15, 36.
2. Kawai, T.; Akira, S.; Nat. Immunol, 2010, 11, 373.
3. Sims, J. E.; Smith, D. E.; Nat. Rev. Immunol, 2010, 10, 89.
4. Ku, C-L.; Bernuth, H.; Picard, C.; Zhang, S-Y.; Chang, H-
H.; Yang, K.; Chrabieh, M.; Issekutz, A. C.; Cunningham,
C.K.; Gallin, J.; Holland, S.M.; Roifman, C.; Ehl, S.;
Smart, J.; Tang, M.; Barrat, F. J.; Levy, O.; McDonald, D.;
Day-Good, N.K.; Miller, R.; Takada, H.; Hara, H.; Al-
Hajjar, S.; Al-Ghonaium, A.; Speert, D.; Sanlaville, D.; Li,
X.; Geissmann, F.; Vivier, E.; Maródi, L.; Garty, B-Z.;
Chapel, H.; Rodriguez-Gallego, C.; Bossuyt, X.; Abel, L.;
Puel, A.; Casanova. J-L.; J. Exp. Med. 2007, 204, 2407.
5. Picard, C.; Casanova, J-L.; Puel, A. Clin. Microbiol. Rev,
2011, 24, 490.
6. Seganish, W. M.; Ex. Op. Ther. Patents 2016, 26, 917.
7. Chaudhary, D.; Robinson, S.; Romero, D. L.; J. Med.
Chem., 2015, 58, 96
8. Koziczak-Holbro, M.; Littlewood-Evans, A.; Pollinger, B.;
Kovarik, J.; Dawson, J.; Zenke, G.; Burkhart C.; Müller,
M.; Gram, H.; Arthr. Rheumat., 2009, 60, 661.
9. a) Lee, K; Allais, C; Ambler, C; Anderson, D; Boscoe, B;
Bree, A; Brodfuehrer, J; Bunnage, M; Choi, C; Chung, S;
Curran, K; Day, J; Dehnhardt, C; Dermenci, A; Drozda, S;
Frisbie,; Gavrin, L; Goldberg, J; Han, S; Hegen, M;
Hepworth, D; Jacobson, B; Kilty, I; Kortum, S; Lee, A;
Lovering, F; Lowe, M; Mathias, J; Murphy, E;
Papaioannou, N; Patny, A; Pierce, B; Ramsey, S; Rao, V;
10. Bhide, R. S.; Duncia, J. V.; Hynes, J.; Nair, S. K.; Pitts, W.
J.; Srikantha, R. K.; Gardner, D.S.; Murugesan, N.; Paidi,
V, R.; Santella, J. B.; Sistla, R.; Wu, H.; US 2015/0274696
A1
11. Compounds 4, 10, and 7 have been assigned PDB IDs
5W84, 5W85, and 5W86, respectively.
12. Kuglstatter, A.; Villasenor, A. G.; Shaw, D.; Lee, S. W.;
Tsing, S.; Niu, L.; Song, K. W.; Barnett, J. W.; Browner,
M. F. J. Immunol., 2007, 178, 2641.

Graphical Abstract
Leave this area blank for abstract info.
Discovery and structure-based design of 4,6-
diaminonicotinamides as potent and selective
IRAK4 inhibitors
Rajeev S. Bhide, Alec Keon, Carolyn Weigelt, John S. Sack, Robert J. Schmidt, Shuqun Lin, Hai-Yun
Xiao, Steven H. Spergel, James Kempson, William J. Pitts, Julie Carman and Michael A. Poss