CHEMMEDCHEM
FULL PAPERS
kind of modifications should be avoided in designing novel
JAK3 inhibitors.
Conclusions
In the current study, tofacitinib analogues with different bioiso-
steric hinge binding motifs were investigated, and several
novel synthetic approaches toward those compounds were de-
veloped. Our biological data impressively highlights that the
space for structural modifications on the hinge binding hetero-
cycle is very narrow within this substance class.
Experimental Section
Detailed experimental procedures and characterization of all com-
pounds are provided in the Supporting Information. A structure–
data file (SDF) including all compounds and the corresponding in-
hibition data for modeling purposes is available from the authors
upon request.
As revealed by our structure–activity relationship, bioiso-
steric replacement of arylÀCH groups by sp2 nitrogen atoms
results in minor losses in activity relative to that of the parent
compound tofacitinib. A notable exception is 1H-pyrazolo[3,4-
d]pyrimidine derivative 5, which is significantly less potent. In
general, substitution of the purine-derived heterocycles is
poorly tolerated within the presented compound set and re-
moval of the hydrogen bond donor function completely sup-
presses JAK inhibition. Introducing additional lipophilic bulk
does not counterbalance this loss. Truncating the core scaffold
while maintaining the three-dimensional pharmacophoric in-
teraction pattern leads to pyrimidine-derived inhibitors with
single-digit micromolar IC50 values. Removal of the exocyclic N-
methyl substituent of tofacitinib causes a severe drop in po-
tency, which can be compensated by attaching a nitro group
in the ortho position of the amino function. The key driver of
this increase in potency is the stabilization of the bioactive syn
conformation through an intramolecular hydrogen bond, as
evidenced by quantum chemical calculations and X-ray crystal
structure analysis. This strategy might be transferred to similar
heterocycles; however, it fails to restore the activity of other
scaffolds within our series. It remains to be tested whether re-
placement of the nitro group with stronger sp2-hybridized hy-
drogen bond acceptors such as carbonyl, carboxamide, or car-
boxylate groups could be used to further enhance this
“pseudo-rigidification” effect while simultaneously serving as
attachment points for additional substituents targeting the hy-
drophobic front region (“hydrophobic region II”) of the kinase.
Ultimately, tricyclic tofacitinib analogues proved to be extreme-
ly potent and a good starting point for further modification.
Substitution with residues targeting the hydrophobic front
region and the side chain of Cys909 exclusively found in JAK3
might be the next step toward inhibitors selectively inhibiting
this Janus kinase isoform. Another promising future approach
consists of targeting the hydrophobic back pocket by attach-
ing suitable substituents to the 8-position of rigidified tofaciti-
nib analogues such as 3 or 43. Given that most compounds in
our set were significantly less potent than tofacitinib, the selec-
tivity within the JAK family was not further assessed. Neverthe-
less, the current study is unique in providing a dataset contain-
ing a diverse array of bioisosteric compounds tested in a single
assay system, which delivers directly comparable results. We
observed that results from docking poorly correlated with the
outcome of biological testing within our library, although the
poses obtained were reasonable and mostly resembled the
binding mode of template 1. Thus, the data set represents
a valuable tool for the training and validation of computational
models. In contrast to many other studies presenting success-
ful efforts only, we also provide detailed evidence for which
Abbreviations
ATP: adenosine triphosphate; Boc: tert-butyl carbamate; dba: di-
benzylideneacetone; DBU: 1,8-diazabicyclo[5.4.0]undec-7-ene;
DCC: N,N’-dicyclohexylcarbodiimide; DIPEA: N,N-diisopropylethyla-
mine; DMF: N,N-dimethylformamide; ELISA: enzyme-linked immu-
nosorbent assay; HATU: 1-[bis(dimethylamino)methylene]-1H-1,2,3-
triazolo[4,5-b]pyridinium3-oxide hexafluorophosphate; IC50: half-
maximal inhibitory concentration; JAK: Janus kinase; LiHMDS: lithi-
um bis(trimethylsilyl)amide; MOM: methoxymethyl; QM: quantum
mechanical; PDB: protein data bank; p-TSA: p-toluenesulfonic acid;
RMSD: root-mean-square deviation; SE: standard error; SNAr: nucle-
ophilic aromatic substitution; STAT: signal transducer and activator
of transcription; TBHP: tert-butyl hydroperoxide; TBTU: O-(benzo-
triazol-1-yl)-N,N,N’,N’-tetramethyluronium tetrafluoroborate; tBu:
tert-butyl; THF: tetrahydrofuran; TMS: trimethylsilyl; Tosyl: p-tolue-
nesulfonyl; TYK2: tyrosine kinase 2.
Acknowledgements
The authors are grateful to Inte:Ligand Software–Entwicklungs
und Consulting GmbH (Maria Enzersdorf, Austria) for providing
the LigandScout software. The authors thank Daniela Mꢀller for
biological testing, Armin Schniers for support in chemical synthe-
sis, and Peter Keck for assistance in editing the manuscript and
fruitful scientific discussions.
Keywords: bioisosteres · cytokines · inflammation · Janus
kinase · structure–activity relationships
[1] M. Pesu, A. Laurence, N. Kishore, S. H. Zwillich, G. Chan, J. J. O’Shea, Im-
[6] Nat. Rev. Drug Discovery 2012, 11, 895.
[7] S. Dçker, M. Dewenter, A. El-Armouche, Dtsch. Med. Wochenschr. 2014,
139, 1003–1008.
[8] J. J. O’Shea, A. Kontzias, K. Yamaoka, Y. Tanaka, A. Laurence, Ann.
Rheum. Dis. 2013, 72, ii111–ii115.
[11] G. Thoma, F. Nuninger, R. Falchetto, E. Hermes, G. A. Tavares, E. Vangre-
[12] M. Soth, J. C. Hermann, C. Yee, M. Alam, J. W. Barnett, P. Berry, M. F.
Browner, K. Frank, S. Frauchiger, S. Harris, Y. He, M. Hekmat-Nejad, T.
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2014, 9, 2516 – 2527 2526