4-Phenyl-7-azaindoles as potent and selective IKK2 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2009.03.034

The synthesis and SAR of a novel series of IKK2 inhibitors are described. Modification around the hinge binding region of the 7-azaindole led to a series of potent and selective inhibitors with good cellular activity.

Full text of DOI:10.1016/j.bmcl.2009.03.034

Bioorganic & Medicinal Chemistry Letters 19 (2009) 2504–2508
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
4-Phenyl-7-azaindoles as potent and selective IKK2 inhibitors
a
a
a
a
John Liddle a, , Paul Bamborough , Michael D. Barker , Sebastien Campos , Rick P. C. Cousins ,
Geoffrey J. Cutler b, Heather Hobbs a, Duncan S. Holmes a, Chris Ioannou a, Geoff W. Mellor b,
Mary A. Morse a, Jeremy J. Payne a, John M. Pritchard a, Kathryn J. Smith a, Daniel T. Tape a,
Caroline Whitworth a, Richard A. Williamson a
*
a GlaxoSmithKline R&D, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire, UK
b GlaxoSmithKline R&D, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, UK
a r t i c l e i n f o
a b s t r a c t
Article history:
The synthesis and SAR of a novel series of IKK2 inhibitors are described. Modification around the hinge
Received 26 January 2009
Revised 9 March 2009
Accepted 10 March 2009
Available online 14 March 2009
binding region of the 7-azaindole led to a series of potent and selective inhibitors with good cellular
activity.
Ó 2009 Elsevier Ltd. All rights reserved.
Keywords:
IKK2
IKK1
Azaindole
Kinase
IKK2 inhibitor
The transcription factor nuclear factor kappa B (NF-
j
B) is a
to be regulated by IKK1 and 2, respectively. Such effects, if con-
firmed, will need to be taken into account when considering clini-
cal utility of IKK inhibitors.
ubiquitously expressed, major regulator of inflammatory genes
including cytokines, chemokines and inflammatory enzymes.1–3
In unstimulated cells, NF-
complex with the IkB proteins. Upon cellular stimulation, I
teins are phosphorylated by I B kinase (IKK) leading to degrada-
tion of the complex and translocation of NF-
and the induction of gene transcription.4 The I
j
B resides in the cytoplasm as a silent
A series of 4-aryl-7-azaindole sulfonamides exemplified by 1
were identified as potent, ATP competitive, IKK2 inhibitors during
a kinase cross screening campaign (Table 1). Although 1 demon-
strated good IKK2 enzyme inhibition, it was less than 10-fold selec-
tive over IKK1 and offered poor selectivity over a broad kinase
panel. The high inhibition observed at ROCK1 was of particular
concern, given the involvement of this kinase in multiple biological
processes.13 Herein we describe our initial lead optimisation ef-
forts which led to potent, selective and cellular active IKK2
inhibitors.
3,5-Disubstituted 7-azaindoles have been shown to bind to ABL
and to CHK1 kinases with the N1 of the azaindole (pyrrole NH) lo-
cated towards the inside of the hinge region of the ATP site (‘N1-
in’), as has the unsubstituted 7-azaindole fragment in a PKA/PKB
chimera.15 However, other 7-azaindole kinase crystal structures
have shown that an alternative flipped ‘N1-out’ mode is also possi-
ble, in which the azaindole binds with a 180° flip, with the N1 atom
towards the outside of the hinge region (unpublished results).
Docking studies of 1 into an IKK2 homology model16 were ambig-
uous, but slightly favoured this N1-out binding mode, as shown in
Figure 1.
j
B pro-
j
jB to the nucleus
jB kinase complex
contains two catalytic subunits, IKK1 and IKK2, and a regulatory
subunit NF-
j
B essential modulator (NEMO).5
Although IKK1 and IKK2 are highly homologous serine-threo-
nine protein kinases (61% of amino acids are identical in their
aligned kinase domains) and contain similar structural domains,
studies in knockout mice and derived mouse embryonic fibroblasts
suggest that IKK2 is the predominant kinase involved in IkB phos-
phorylation and hence NF-
presented an unexpected phenotype including skeletal and skin
abnormalities.7 Given the attractiveness of NF-
B as a therapeutic
j
B activation.6,7 IKK-1 deficient mice
j
target in inflammatory and autoimmume disorders, the develop-
ment of selective IKK2 inhibitors has been an active area of phar-
maceutical research for several years.8,9 Recent publications10–12
have suggested that IKK family members may be involved in cell
cycle control through regulation of the mitotic kinase Aurora A.
The activity10 and cellular level11 of Aurora A have been proposed
The IKK2 and IKK1 ATP binding sites differ by only one amino
acid residue (Gln100 of IKK2 is equivalent to Ser99 of IKK1). How-
ever, the sidechain of this residue points away from the ATP site
* Corresponding author. Tel.: +44 1438 763643.
E-mail address: john.2.liddle@gsk.com (J. Liddle).
0960-894X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2009.03.034

J. Liddle et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2504–2508
2505
Table 1
Table 2
Profile of lead azaindole 1
SAR of 4-phenyl-7-azaindoles
a
Kinase
pIC50
H
N
O
S
IKK2
IKK1
ROCK1
7.4
6.6
7.7
NH2
R
O
N
N
H
N
H
N
1
a
a
Compound
R
IKK2 pIC50
IKK1 pIC50
1
2
3
4
5
6
4-SO2NHCH2CH2NH2
4-SO2NH2
4-CONH2
7.4
6.5
6.2
5.5
7.5
6.7
6.6
6.8
6.4
5.4
6.5
6.1
pIC50 values are reported as mean of ꢁ2 experiments.
a
Enzyme inhibition assays using recombinant human protein.14
H
4-SO2NHCH2CH2NHCH3
4-SO2NHCH2CH2N(CH3)2
7
8
7.4
6.6
7.5
5.9
4 -
SO2NH
NH
4 -
SO2N
9
10
4-NHSO2CH3
3-SO2NH2
6.8
5.3
6.1
5.9
O
O
11
4 -
7.1
7.3
SO2NH
S
pIC50 values are reported as mean of ꢁ2 experiments.
a
Enzyme inhibition assays using recombinant human protein.14
logues of 1 were prepared, as exemplified by the azetidine 7, but
none offered any advantage in terms of potency or improved
IKK1 selectivity.
Figure 1. Predicted binding mode of compound 1 in IKK2.
The sulfonamide NH is not required for potency, evident from
the pyrollidine sulfonamide 8 retaining activity at IKK2. The re-
verse sulfonamide 9 maintained potency and gained marginal
selectivity over IKK1. Both the meta-sulfonamide 10 and the corre-
sponding meta-reverse sulfonamide were less potent at IKK2. The
sulfone 11 was one of the most potent sulfonamides found within
this sub-series but again lacked IKK1 selectivity.
and hence would have minimal effect on inhibitor binding. Any
influence from this substitution would be secondary, mediated
through conformational effects or an indirect influence on other
residues. Since there were no obvious structural differences be-
tween IKK1 and IKK2, it was felt that any selectivity for IKK2 over
IKK1 would have to be a result of cumulative effects on the ATP
site’s conformation arising from relatively remote sequence differ-
ences, and that this would be impossible to predict.
Early SAR suggested that the ethylamine fragment of the sulfon-
amide 1 enhanced IKK2 potency whilst having little effect on IKK1
potency; the primary sulfonamide 2 was less potent and selective
(Table 2). While models of 2 docked into the IKK2 homology mod-
el16 did not position the sulfonamide oxygens within hydrogen-
bonding distance of the catalytic Lys44, it was felt that the distance
was close enough for the flexibility of the lysine sidechain to make
this likely. The sulfonamide was found to be the optimal linker; the
primary amide 3 was a marginally less potent at IKK2. This finding
is consistent with that recently reported for a series of aminopy-
rimidine IKK2 inhibitors which share the same pharmacophore
as the azaindoles.17 Interestingly, removal of the amide or sulfon-
amide led only to a modest drop in potency; 4-phenyl-7-azaindole
Since all the 4-phenyl-7-azaindoles lacked the desired selectiv-
ity over IKK1 and the broader kinase panel (data not presented
here) our attention turned to substitution around the azaindole
core to address this potential liability. It was proposed that 2-
substituted 7-azaindoles would be unable to bind in the ‘N1-in’
mode due to a lack of sufficient space at the back of the pocket
causing a steric clash, especially with the gatekeeper residue
Met96. Instead, in the ‘N1-out’ mode, the 2-substituent would
point towards solvent and hence lend itself well as a handle to
optimise the physicochemical properties of the molecule. AGC
family kinases including ROCK1 have a C-terminal extension to
the kinase domain, in which a hydrophobic motif is found. This
motif packs against the N-terminal lobe and is involved in dimer-
isation of ROCK1.18,19
Between the end of the kinase domain and
the hydrophobic motif, the peptide chain passes close to the sol-
vent exposed edge of the ATP site. In particular, the sidechain of
Phe368 (Fig. 2) would block access to the solvent side of the site
from the 2-position of the azaindole. Hence, it was hypothesised
that 2-substitution would improve selectivity for the series over
ROCK1.
Several 4-arylsulfonamide-7-azaindoles were chosen for subse-
quent substitution in the 2-position of the azaindole core. The data
for the aminoethyl sulfonamide is shown in Table 3. As expected,
4 retained some potency at IKK2 (IC50 ꢀ 30
lM).
Initial medicinal chemistry efforts focussed on modification of
the sulfonamide substituent with the aim of increasing IKK1 selec-
tivity. Mono-methylation of the terminal amino group of 1 was tol-
erated at both IKK2 and IKK1 whereas dimethylation had a
deleterious effect on potency at both isoforms as illustrated by
compounds 5 and 6, respectively. Several constrained amine ana-

2506
J. Liddle et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2504–2508
In comparison, the lead 7-azaindole 1 had a poor overall selec-
tivity profile against a panel of 31 kinases. 1 inhibited 4 kinases
within 10-fold of IKK2 (fold selectivity based on comparative IC50
values) whereas 13 inhibited only IKK1 within 30-fold of IKK2.
Such kinase selectivity may be rationalised by the inability of these
2-substituted azaindoles to adopt the ‘N1-in’ binding mode, which
may be the preferred mode for some kinases, because of a steric
clash with their gatekeeper residues. It is believed that the dra-
matic increase in selectivity over ROCK1 for 2-substituted azain-
doles compared to the 2-H azaindoles is
a result of the
hypothesis outlined earlier, and is due to a steric clash with
Phe368.
With the improved broad kinase selectivity of the 2-substituted
azaindoles in hand, we chose to re-optimise the sulfonamide moi-
ety, to deliver good cellular potency and greater IKK1 selectivity.
Since there was evidence that the azaindole 4-position SAR did
not necessarily parallel across varying 2-position substituents,
we decided to investigate the sulfonamide SAR with several 2-
substituted azaindoles. The data for the 2-methyl azaindole com-
pounds is shown in Table 4.
Figure 2. Predicted binding mode of compound
structure of ROCK1 (orange) is superimposed, showing the C-terminal extension
and Phe368 that occludes the front of the ATP site.
1 in IKK2 (cyan). The X-ray
polar substituents were tolerated in the 2-position. The key finding
was that the introduction of a 2-substituent had a dramatic effect
on the broad kinase selectivity profile of this series. The carboxam-
ide 12 retained good potency at IKK2 and had similar selectivity to
1 over IKK1. Importantly, 12 had a far superior selectivity profile
against a panel of 31 kinases, including ROCK1. However, 12 had
Table 4
SAR of 2-methyl-7-azaindoles
R
poor activity in the IKK2 NF-
jB mechanistic cellular assay (inhibi-
tion of TNF induced NF-
a
j
B reporter activation in A549 cells),14
with a drop of over a 100-fold in potency, a reflection of poor per-
meability caused by high polar surface area (PSA). Encouragingly,
the 2-methyl-7-azaindole 13 retained both the good enzyme po-
tency and broad kinase selectivity profile observed with carboxam-
ide 12. Moreover, due to a reduction in PSA, 13 had very good
cellular activity (pIC50 6.5) with approximately 10-fold drop-off
in potency between the enzyme and the cellular assays.
CH3
N
H
N
b
R
IKK2a pIC50 IKK1a pIC50 A549 cell pIC50 PSA Å2
H
N
O
13
14
15
16
17
18
19
20
7.7
7.8
7.7
7.4
7.7
7.4
7.2
6.5
6.9
6.9
6.5
5.5
7.9
6.0
5.4
5.1
6.5
5.7
6.3
6.1
7.3
6.3
6.0
5.7
101
101
96
S
NH2
NH2
O
Table 3
O
O
S
SAR of 2-position of 7-azaindole
HN
O O
H
N
S
R
O
N
S
OH
O
O
H
NH2
A549 cellb
CH3
N
R
IKK2a
pIC50
IKK1a
pIC50
ROCK1a
pIC50
PSA
Å2
O
86
S
OH
pIC50
H
N
O
O
S
S
109
100
75
1
7.4
7.7
6.6
6.8
6.2
7.7
5.0
101
144
O
O
S
S
N
N
H
CH3
N
O
O
O
NH2
O
O
12
5.4
6.5
N
H
N
N
O
O
S
HN
CH3
O
N
O
13
7.7
6.9
6.5
101
H3C
S
66
N
H
CH3
CH3
pIC50 values are reported as mean of ꢁ2 experiments.
pIC50 values are reported as mean of P2 experiments.
a
Enzyme inhibition assay using recombinant human protein.14
a
Enzyme inhibition assay using recombinant human protein.14
b
Inhibition of TNFa induced NF-j
B reporter activation in A549 cells.14
Inhibition of TNFa induced NF-j
B reporter activation in A549 cells.14
b

J. Liddle et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2504–2508
2507
Table 5
7.3) but 17 was essentially equipotent at IKK1. 18 however, was
threefold less potent at IKK2 and suffered a greater drop-off in cel-
lular potency but was approximately 25-fold selective over IKK1. It
is noteworthy that 17 has a relatively high PSA of 109 and this had
minimal effect on the drop-off in cellular potency. Importantly, the
poor selectivity of 17 and hence the high inhibition of IKK1 should
have no effect on the NF-kB cellular potency. The reasons for the
improved selectivity of the tertiary sulfonamides for IKK2 over
IKK1 are unclear. There are no amino acid differences in this region
of the ATP site. From the docked model, the sidechains of Asn150
and Asp166 are nearby. Whether the NH of the primary sulfon-
amide forms a hydrogen bond to these sidechains or not is ambig-
uous. However, the most likely explanation for the selectivity
difference between IKK1 and IKK2 is that these sidechains can
adopt different conformations in the two kinases.
Both the secondary and tertiary reverse sulfonamides 19 and 20
showed good selectivity over IKK1 of greater than 30-fold; again
the secondary sulfonamide was more potent at IKK2.
Compound 16 offered the preferred overall profile with good
IKK2 potency, excellent kinase selectivity including approximately
80-fold selectivity over IKK1, and cellular activity at micromolar
concentrations. The selectivity profile of 16 was determined
against a panel of 36 kinases. In this set only two kinases (IKK1
and AurB) were inhibited within 100-fold of IKK2 (fold selectivity
based on comparative IC50 values) (Table 5).
Kinase selectivity profile of 16
Kinase
pIC50
Fold selectivity
IKK2
IKK1
Aurora B
Aurora A
ROCK1
GSK-3b
7.4
5.5
5.8
5.1
<5
—
79
40
200
>250
>600
<4.6
The reverse sulfonamide 14 had similar IKK2 potency and IKK1
selectivity compared to 13 but had a far greater drop-off in cellular
potency. A number of reverse 2-ethylamino sulfonamides of this
type were prepared and all had a significant drop-off in cellular po-
tency of greater than a 100-fold. Such poor cellular permeability is
possibly due to the combination of the acidic reverse sulfonamide
(measured pKa ꢀ 7) and basic primary amine (measured pKa ꢀ 9)
and hence a high proportion (>95%) of the compound being ionised
at pH 7.4.
The primary alcohol 15 maintained IKK2 potency and offered
10-fold selectivity over IKK1. Secondary sulfonamides were often
marginally more potent than tertiary sulfonamides at IKK2 due
to a postulated H-bond interaction between the sulfonamide NH
and the conserved Asp166 residue sidechain, or at least a close
contact with that residue that disfavours substitution. However,
tertiary sulfonamides offered greater selectivity over IKK1. For
example, sulfonamide 15 was marginally more potent than the
corresponding tertiary sulfonamide 16 but considerably less selec-
tive over IKK1. Both compounds showed approximately 10-fold
drop-off in cellular potency. A similar trend was observed with
the cyclic sulfones 17 and 18. 17 had excellent potency at IKK2
(pIC50 8.0) which translated into excellent cellular activity (pIC50
The 4-aryl-7-azaindoles described herein were prepared via a
number of alternative methods as previously reported.14 A versa-
tile route involved coupling the 4-bromo-7-azaindole 23 with the
corresponding tetramethyl-1,3,2-dioxaborolan-2-yl 26 species un-
der Suzuki cross coupling conditions (Scheme 1).
Modification of the 2-position was accomplished by lithiation of
the N-protected 4-bromo-7-azaindole 21 and subsequent reaction
Br
N
Br
N
Br
a
CH3
b
CH3
N
S
N
S
O
O
N
H
N
O
O
23
21
22
Me
N
O
O
O
O
O
S
B
OH
O
S
B
PFP
O
S
O
PFP
O
c
e
d
Br
O
O
O
24
25
26
Me
N
O
S
N
OH
O
PFP = Pentafluorophenyl
CH3
N
H
16
Scheme 1. Reagents and conditions: (a) LDA, THF, ꢂ35 °C 30 min then MeI, 2 h, 80%; (b) TBAF, THF, microwave 60 °C, 1 h, 89%; (c) pinacolato diboron, PdCl2(dppf), dppf,
NaOAc, 1,4-dioxane 105 °C, 16 h, 17%; (d) amine, Et3N, DMF, microwave 120 °C, 10 min, not isolated; (e) K3PO4, 20-(dimethylamino-biphenyl-palladium(II) chloride
dinorbornylphosphine complex, 1,4-dioxane, water, microwave 130 °C, 30 min, 56% for steps d and e.

2508
J. Liddle et al. / Bioorg. Med. Chem. Lett. 19 (2009) 2504–2508
9. Coish, P. D. G.; Wickens, P. L.; Lowinger, T. B. Expert Opin. Ther. Patents 2006, 16,
1.
10. Prajapati, S.; Tu, Z.; Yamamoto, Y.; Gaynor, R. B. Cell Cycle 2006, 5,
2371.
11. Irelan, J. T.; Murphy, T. J.; DeJesus, P. D.; Teo, H.; Xu, D.; Gomez-Ferreria, M. A.;
Zhou, Y.; Miraglia, L. J.; Rines, D. R.; Verma, I. M.; Sharp, D. J.; Tergaonkar, V.
Proc. Natl. Acad. Sci. 2007, 104, 16940.
12. Blazkova, H.; Von Schubert, C.; Mikule, K.; Schwab, R.; Angliker, N.; Schmuckli-
Maurer, J.; Fernandez, P. C.; Doxsey, S.; Dobbelaere, D. A. Cell Cycle 2007, 6,
2531.
with a suitable electrophile. The 2-methyl-7-azaindoles were pre-
pared by quenching with methyl iodide. Deprotection of the 7-aza-
indole 22 was carried out with TBAF in THF. The arylsulfonamide
boronate esters were prepared in two steps from pentafluorophe-
nyl 4-bromobenzenesulfonate 24.20 24 was treated with pinacolato
diboron in the presence of PdCl2(dppf) catalyst and sodium acetate
in refluxing 1,4-dioxane to yield the boronate ester 25. 25 was then
treated with the appropriate amine to yield the corresponding sul-
fonamide which was subsequently coupled with the azaindole core
23 under standard conditions to give the final compounds.
In conclusion, we have described the SAR around a series of
ATP-competitive 4-aryl-7-azaindoles which represent a novel class
of IKK2 inhibitors. The key discovery was that substitution close to
the azaindole hinge binding region led to dramatic improvements
in broad kinase selectivity. Further optimisation of the sulfonamide
moiety, with a conscious effort to maintain moderate PSA, im-
proved IKK1 selectivity and led to compounds with good cellular
activity. Subsequent efforts are focussed on optimising the phar-
macokinetic profile of this series and these studies will be reported
in due course.
13. Olson, M. F. Curr. Opin. Cell Biol. 2008, 20, 242.
14. The IKK2 enzyme assay, NF-jB cellular mechanistic assay and azaindole
syntheses are described in the following reference: Bamborough, P. A.; Barker,
M. D.; Campos, S. A.; Cousins, R. P. C.; Faulder, P.; Hobbs, H.; Holmes, D. S.;
Johnston, M. J.; Liddle, J.; Payne, J. J.; Pritchard, J. M.; Whitworth, C. WO2008/
034860.
15. Kinase complexes of 3,5-disubsituted 7-azaindoles have been solved with c-
Abl (PDB entry 2qoh: Zhou, T.; Parillon, L.; Li, F.; Wang, Y.; Keats, J.; Lamore, S.;
Xu, Q.; Shakespeare, W.; Dalgarno, D.; Zhu, X. Chem. Biol. Drug Des. 2007, 30,
171) and CHK1 (PDB entry 1zys: Stavenger, R.; Zhao, B.; Zhou, B.; Brown, M.;
Lee, D.; Holt, D. Unpublished results). The structure of a PKA-PKB chimera with
7-azaindole has also been solved (PDB entry 2uvx: Donald, A.; McHardy, T.;
Rowlands, M.; Hunter, L.; Davies, T.; Berdini, V.; Boyle, R.; Aherne, G.; Garrett,
M.; Collins, I. J. Med. Chem. 2007, 50, 2289.
16. A homology model of IKK2 was built as previously described: Christopher, J. A.;
Avitabile, B. G.; Bamborough, P.; Champigny, A. C.; Cutler, G. J.; Dyos, S. L.;
Grace, K. G.; Kerns, J. K.; Kitson, J. D.; Mellor, G. W.; Morey, J. V.; Morse, M. A.;
O’Malley, C. F.; Patel, C. B.; Probst, N.; Rumsey, W.; Smith, C. A.; Wilson, M. J.
Bioorg. Med. Chem. Lett. 2007, 17, 3972. A model of IKK1 was also built using the
same template and refinement procedure.
References and notes
1. Barnes, P. J.; Karin, M. N. Eng. J. Med. 1997, 336, 1066.
2. Li, Q.; Verma, I. M. Nat. Rev. Immunol. 2002, 2, 725.
17. Bingham, A. H.; Davenport, R. J.; Fosbeary, R.; Gower, L.; Knight, R. L.;
Lowe, C.; Owen, D. A.; Parry, D. M.; Pitt, W. R. Bioorg. Med. Chem. Lett.
2008, 18, 3622.
18. Doran, J.; Liu, X.; Taslimi, P.; Saadat, A.; Fox, T. Biochem. J. 2004, 384, 255–262.
19. Jacobs, M.; Hayakawa, K.; Swenson, L.; Bellon, S.; Fleming, M.; Taslimi, P.;
Doran, J. J. Biol. Chem. 2006, 281, 260.
3. Karin, M.; Yamamoto, Y.; Wang, Q. M. Nat. Rev. Drug Disc. 2004, 3, 17.
4. Hayden, M. S.; Ghosh, S. Genes Dev. 2004, 18, 2195.
5. Yamamoto, Y.; Gaynor, R. B. Trends Biochem. Sci. 2004, 29, 72.
6. Li, J.; Peet, G. W.; Pullen, S. S. J. Biol. Chem. 1998, 273, 30736.
7. Hu, Y.; Baud, V.; Delhase, M. Science 1999, 284, 316.
8. Strnad, J.; Burke, J. R. Trends Pharm. Sci. 2007, 28, 142.
20. Avitabile, B. G.; Smith, C. A.; Judd, D. B. Org. Lett. 2005, 7, 843.