Rational design and asymmetric synthesis of potent and neurotrophic ligands for FK506-binding proteins (FKBPs)

Article abstract of DOI:10.1002/anie.201408776

To create highly efficient inhibitors for FK506- binding proteins, a new asymmetric synthesis for pro-(S)-C5- branched [4.3.1] aza-amide bicycles was developed. The key step of the synthesis is an HF-driven N-acyliminium cyclization. Functionalization of the C5 moiety resulted in novel protein contacts with the psychiatric risk factor FKBP51, which led to a more than 280-fold enhancement in affinity. The most potent ligands facilitated the differentiation of N2a neuroblastoma cells with low nanomolar potency.

Full text of DOI:10.1002/anie.201408776

Angewandte
Chemie
DOI: 10.1002/anie.201408776
Drug Design
Rational Design and Asymmetric Synthesis of Potent and Neurotrophic
Ligands for FK506-Binding Proteins (FKBPs) **
Sebastian Pomplun, Yansong Wang, Alexander Kirschner, Christian Kozany, Andreas Bracher,
and Felix Hausch*
Abstract: To create highly efficient inhibitors for FK506-
binding proteins, a new asymmetric synthesis for pro-(S)-C5-
branched [4.3.1] aza-amide bicycles was developed. The key
step of the synthesis is an HF-driven N-acyliminium cycliza-
tion. Functionalization of the C5 moiety resulted in novel
protein contacts with the psychiatric risk factor FKBP51,
which led to a more than 280-fold enhancement in affinity. The
most potent ligands facilitated the differentiation of N2a
neuroblastoma cells with low nanomolar potency.
F
K506-binding proteins (FKBPs) are part of the immuno-
phillin family and best known for their immunosuppressive
activity as complexes with the natural products FK506 and
rapamycin.[1] In addition, FKBPs are prominently expressed
in the central nervous system and non-immunosuppressive
FKBP ligands have repeatedly displayed neuroprotective and
neurotrophic effects.[2] Brain-specific deletion of FKBP12 was
shown to enhance contextual memory, whereas FKBP52
controls neuronal growth cone guidance.[3] Among the human
FKBPs, FKBP51 has gained particular interest as a regulator
of stress-coping behavior and as a risk factor for stress-related
psychiatric disorders, thus suggesting FKBP51 inhibition as
a potential therapeutic approach for these indications.[4]
However, the development of potent drug-like inhibitors for
FKBPs in general and for FKBP51 in particular is challenging
owing to the shallow FK506 binding site.[5] Inspired by work
from Agouron,[6c] we previously identified a [4.3.1] bicyclic
Figure 1. a) Cocrystal structure of 1a in complex with the FK506-
binding domain of FKBP51 (PDB code: 4JFK, from Ref. [6a]). The free
space for a pro-(S)-C5 substituent is shown as green wire mesh.
Residues in close proximity to a potential pro-(S)-C5 substituent (Tyr57
and Asp68) are shown as sticks and wire mesh. b) Examples of C5-
unsubstituted [4.3.1] bicyclic FKBP ligands.[6] The pro-(S)-C5 position is
indicated in green.
scaffold as a preferred FKBP binding motif, which is
preorganized to mimic the active conformation of known
FKBP ligands (Figure 1).[6a,b] The [4.3.1] cyclization strategy
repeatedly enhanced ligand efficiency compared to previous
monocyclic scaffolds, although the actual affinities for the
large FKBPs remained moderate. We hypothesized that the
[4.3.1] bicycles might also serve as a rigid framework to attach
further substituents in a conformationally defined manner.
Herein, we report a synthesis for C5-branched [4.3.1] aza-
amide bicycles, which resulted in the first low nanomolar and
highly ligand-efficient FKBP inhibitors with neurotrophic
activity.
Analysis of the cocrystal structure of bicyclic [4.3.1] aza-
amide ligands[6a] with FKBP51 indicated that a substituent in
the pro-(S)-C5 position could fit into the binding pocket and
enable additional interactions with the protein (Figure 1a).
Since no examples of C5-derivatized 3,10-diazabicyclo-
[4.3.1]decan-2-one systems were found in the literature,[7] the
development of a new synthesis was required to investigate
this new class of FKBP ligands. Inspired by work on aza-
bridged bicyclic ring systems,[8] we envisioned a TMS-facili-
tated N-acyliminium cyclization for the construction of the
C5-branched bicyclic core fragment (Scheme 1).
[*] S. Pomplun,[+] Dr. Y. Wang,[$] [+] Dr. A. Kirschner, Dr. C. Kozany,
Dr. F. Hausch
Max Planck Institute of Psychiatry
Kraepelinstr. 2–10, 80804 Munich (Germany)
E-mail: hausch@mpipsykl.mpg.de
Dr. A. Bracher
Max Planck Institute of Biochemistry
82152 Martinsried (Germany)
[$] Current address: European Molecular Biology Laboratory
69117 Heidelberg (Germany)
[+] These authors contributed equally to this work.
[**] This work was supported by the M4 Award 2011 from the StMWIVT
(to FH). We are indebted to Mrs. E. Weyher and Dr. S. Uebel (MPI of
Biochemistry) for the HRMS measurement, to Mrs. C. Dubler and
Dr. D. Stephenson (Ludwig Maximilians University Munich) for the
NMR measurement and Mrs C.Sippel for the Ki determinations. The
diffraction data were collected at beamline ID29 of the European
Synchrotron Radiation Facility (ESRF) in Grenoble, France, and
beamline X10SA of the Swiss Light Source in Villigen, Switzerland.
The synthesis commenced with the alkylation of allyl-
amine with alkylbromide 2,[9] a precursor for a preferred
moiety at the N3-position of FKBP ligands (Scheme 2).[6a] For
Supporting information for this article (including experimental
details) is available on the WWW under http://dx.doi.org/10.1002/
anie.201408776.
Angew. Chem. Int. Ed. 2015, 54, 345 –348
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
345

.
Angewandte
Communications
(S)-6-oxo-2-piperidinecarboxylic acid furnished 7 in a good
yield of 72%. Subsequent Boc protection of the amide
activated the C6-carbonyl, thereby setting the stage for the
critical chemoselective reduction/N-acyliminium cyclization
sequence. Thus, reduction of 8 with DIBAL-H gave an
unstable hydroxy carbamate, which after aqueous work-up
was subjected to various cyclization conditions. At temper-
atures between ꢀ788C and 08C at different dilutions, a variety
of Brønsted or Lewis acids produced only dimers, trimers, or
decomposition products (Table S1 in the Supporting Infor-
mation). Analysis of the reaction mixtures revealed that
despite the strongly acidic conditions, the trimethylsilane
group was still present in the dimeric/trimeric products.
Therefore, we decided to explore simultaneous activation of
the hydroxy carbamate and the TMS leaving group. Gratify-
ingly, treatment of the reduction product of 8 with HF·pyr-
idine smoothly yielded the cyclization product 9 as the only
diastereomer in good yield (55% over 2 steps). Reaction with
3,5-dichlorosulfonylchloride as an exemplary N10-substituent
then afforded the key intermediate 10 with 9% overall yield
from 2. A cocrystal structure of compound 10 with FKBP51
validated the correct absolute stereochemistry of 10 and
showed that the desired binding mode is possible for C5-
substitued [4.3.1] bicycles (Figure S1 in the Supporting
Information). In a fluorescence polarization assay,[11] com-
pound 10 bound to FKBP12, FKBP51, and FKBP52 with Ki
values of 94 nm, 140 nm, and 194 nm, respectively, thus
providing a first proof of concept for C5-substituents as
affinity-enhancing modifications to FKBP ligands.
Scheme 1. Mechanism of the TMA-facilitated N-acyliminium cycliza-
tion. Boc=tert-butyloxycarbonyl.
An appealing aspect of the synthetic route chosen in
Scheme 2 was that the newly installed C5-vinyl group
represents a unique reactivity for subsequent derivatization.
In the hope of increasing the affinity even further, we decided
to introduce polar groups at the C5-vinyl substituent to
explore possible contacts with the polar Tyr57 or Asp68
residue (Figure 1a). Wacker oxidation of 10 provided an
equimolar mixture of methylketone 11 and aldehyde 12,
which after reduction with NaBH4 afforded the separable
alcohols 13, 14, and 15 (Scheme 3).[12] Likewise, addition of
MeMgBr to 11/12 yielded the secondary or tertiary alcohols
16, 17, and 18 (Scheme 3). Treatment of 10 with AD-Mix-a or
AD-Mix-b both resulted in inseparable, equimolar mixtures
of diastereomers (R)-19 and (S)-19. However, temporary bis-
TBDMS protection of the diol enabled the separation of the
two diastereomers by preparative HPLC to yield the two
diastereomerically pure diols (R)-19 and (S)-19 after depro-
tection with TBAF. Alcohols 13–18 and diols (R)-19 and (S)-
19 all bound to FKBP51 with low nanomolar affinities
(Scheme 3), up to five-fold more potently than the prototypic
FKBP ligand FK506. Like FK506 and most synthetic deriv-
atives thereof, the C5-derivatized bicycles are pan-selective
FKBP ligands with a slight preference for the small FKBP12.
Compound 18, tested as a representative example, also tightly
bound FKBP12.6 (Ki = 6 nm) and FKBP13 (Ki = 70 nm).
Direct comparison with compound 1b,[7a] a direct analogue
of compounds 13–19 that lacks the C5 substituent, showed
that polar C5 substituents contributed more than 14 kJmolꢀ1
of additional binding energy, thereby leading to a more than
280-fold enhanced affinity.
Scheme 2. Reagents and conditions: a) Allylamine, CH2Cl2, 508C, 24 h,
85%. b) NsCl, TEA, CH2Cl2, RT, 1 h, 95%. c) AllylTMS, p-benzoqui-
none, Grubbs-Hoveyda II gen, CH2Cl2, 608C, 2 h, 95%. d) PhSH,
K2CO3, DMF, RT, 2 h, 65%. e) (S)-6-Oxo-2-piperidinecarboxylic acid,
HATU, DIPEA, DMF, RT, 2 h, 72%. f) Boc2O, TEA, DMAP, CH2Cl2, RT,
3 h, 86%. g) DIBAL-H, THF, ꢀ788C. h) HF·Pyridine, CH2Cl2, ꢀ78!
08C, 1 h, 55% (2 steps). i) 3,5-Dichlorobenzenesulfonylchloride,
DMAP, DIPEA, CH2Cl2, RT, 24 h, 55%. DIBAL-H=diisobutylaluminum
hydride, DIPEA=N,N-diisopropylethylamine, DMAP=4-dimethylami-
nopyridine, HATU=O-(7-azabenzotriazol-1-yl)-N,N,N’,N’-tetramethyl-
uronium hexafluorophosphate, NsCl=2-nitrobenzenesulfonyl chloride,
TEA=triethylamine, THF=tetrahydrofuran, TMS=trimethylsilane.
cross-metathesis of the resulting compound 3 with allyl
trimethylsilane, it was necessary to temporarily mask the
secondary amine to avoid inactivation of the ruthenium
catalysts. Nosyl protection was the protection group of choice
and allowed p-benzoquinone-assisted cross-metathesis[10] to
give 5 with excellent yields. After removal of the nosyl group,
amide bond formation between 6 and commercially available
346
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 345 –348

Angewandte
Chemie
Figure 2. a) Cocrystal structures of (R)-19 (PDB: 4W9O) and (S)-19
(PDB: 4W9P) in complex with the FK506-binding domain of FKBP51.
Hydrogen bonds are indicated by yellow dotted lines with distances in
ꢀ. The conserved water molecule is shown as a blue sphere and the
van der Waals surface of the interacting Asp68 is shown as wire mesh.
b) Stimulation of neurite outgrowth by (R)-19 and (S)-19. N2a cells
were transfected with a plasmid encoding a myristoylated yellow
fluorescent protein, deprived of serum, and treated with the com-
pounds for 24 h, followed by fluorescence imaging and morphological
analysis. FK506 at its most effective concentration (30 nm) was
included as a positive control. Data represent averages of more than
35 cells. Error bars represent ꢁstandard error. ***: p<0.001, **:
p<0.01, *: p<0.05, n.s.: not significant.
Scheme 3. Reagents and conditions: a) PdCl2, CuCl, O2, DMF/H2O, RT,
24 h. b) 1. NaBH4, EtOH/CH2Cl2, RT, 1 h; 13: 16%, 14: 9%; 15: 23%
(over 2 steps each). c) MeMgBr, THF, ꢀ788C, 1 h; 16: 12%; 17: 9%;
18: 34% (over 2 steps each). d) AD-Mix-b, tBuOH/H2O, RT, 24 h,
75%. e) TBDMSOTf, 2,6-Lutidine, CH2Cl2, 1 h, preparative HPLC.
f) TBAF, THF, RT, 30 min. Binding to FKBP12 or the FK506-binding
domain of FKBP51 or FKBP52 was determined by a competitive
fluorescence polarization assay with 40-fluorescein-Gly-rapamycin as
a tracer.[11] DMF=N,N-dimethylformamide, TBAF=tetra-n-butylammo-
nium fluoride, TBDMS=tert-butyldimethylsilyl, Tf=trifluoromethane-
sulfonyl.
inhibit neurite outgrowth of N2a neuroblastoma cells,[14]
a model for the early steps of neuronal differentiation. In
these cells, (R)-19 and (S)-19 both increased neurite out-
growth at low nanomolar concentrations (Figure 2b and
Figure S3). However, we also observed that both compounds
lost their neurite-outgrowth-enhancing activity at higher
concentrations. A similar bell-shaped dose–response curve
has previously been observed with FK506 and close analogues
thereof in SH-SY5Y neuroblastoma cells[15]. Notably, the
immunosuppressive FKBP ligand rapamycin potently and
completely blocked neurite outgrowth (Figure S4), likely
through inhibition of mTOR, which is the main effect of
rapamycin.[16] The fact that structurally different ligands like
(R)-19 and (S)-19 fully replicate the bell-shaped neurite
outgrowth effect of FK506 strongly suggests that inhibition of
an FKBP is responsible for the initial neurite outgrowth
stimulation and for the reduction at higher concentrations.
Considering the reported anti- and pro-neuritotrophic func-
tions of FKBP51 and FKBP52, respectively,[14] we suggest that
inhibition of FKBP51, and at higher concentrations, of
FKBP52, are responsible for the rising and declining arms
of the bell-shaped dose–response curve.
To investigate the structural basis for the observed affinity
gain, we solved the cocrystal structures of (R)-19 and (S)-19
with the FK506-binding domain of FKBP51.[13] Both ligands
form a water-mediated hydrogen bond between the secon-
dary hydroxyl group and Asp68, which directs two different
orientations of the terminal hydroxymethylene for (R)-19 and
(S)-19 (Figure 2a). For (S)-19, the C5-diol can also adopt
a different conformer to form a direct hydrogen bond to
Asp68 (Figure S2).
The potency of the novel ligands allowed their application
in more advanced biological systems. FKBP51 is known to
Angew. Chem. Int. Ed. 2015, 54, 345 –348
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
347

.
Angewandte
Communications
C5-branched
3,10-diazabicyclo-
[5] a) X. J. Wang, F. A. Etzkorn, Biopolymers 2006, 84, 125; b) E. A.
Blackburn, M. D. Walkinshaw, Curr. Opin. Pharmacol. 2011, 11,
365.
[6] a) Y. Wang, A. Kirschner, A. K. Fabian, R. Gopalakrishnan, C.
Kress, B. Hoogeland, U. Koch, C. Kozany, A. Bracher, F. Hausch,
J. Med. Chem. 2013, 56, 3922; b) Y. Wang, F. Hausch (Max-
Planck-Gesellschaft zur Fçrderung der Wissenschaften e.V.),
WO2014015993, 2014; c) S. Katoh, H. Kawakami, M. Linton, V.
Kalish, J. Tatlock, E. Villafranca (Agouron Pharmaceuticals),
WO0004020, 2000.
Taken
together,
[4.3.1]decan-2-ones are a highly efficient class of FKBP
ligands that are suitable for probing the cellular role of
FKBPs. Polar substituents in the C5 position robustly increase
affinity independently of their exact position and orientation.
C5 substitution within the [4.3.1] bicyclic scaffold is thus
a general strategy to enhance affinity for FKBPs, while
leaving flexibility for modifications to fine-tune additional
properties such as solubility, lipophilicity, or selectivity.
Structure–activity relationship studies to address this will be
reported in due course.
[7] The closest structures are C5-unsubstituted 3,10-diazabicyclo-
[4.3.1]decan-2-ones describe in Wang et al and Katoh et al
(Ref [6]) and 3,10-diazabicyclo [4.3.1] decanes lacking the
crucial C2-carbonyl: a) H. Audrain, D. Bender, J. Scheel-
Krꢀger, E. Ø. Nielsen, G. M. Olsen, D. Peters, P. Cumming, J.
Labelled Compd. Radiopharm. 2003, 46, 873; b) T. Sasaki, S.
Eguchi, T. Kiriyama, J. Org. Chem. 1971, 36, 2061.
Received: September 3, 2014
Published online: November 20, 2014
[8] a) P. M. Esch, H. Hiemstra, W. J. Klaver, W. N. Speckamp,
Heterocycles 1987, 26, 75; b) W. J. Klaver, H. Hiemstra, W. N.
Speckamp, Tetrahedron 1988, 44, 6729; c) J. Sisko, J. R. Henry,
S. M. Weinreb, J. Org. Chem. 1993, 58, 4945; d) J. Sisko, S. M.
Weinreb, J. Org. Chem. 1991, 56, 3210; e) P. Somfai, J. ꢁhman,
Tetrahedron Lett. 1992, 33, 3791.
[9] a) R. Gopalakrishnan, C. Kozany, S. Gaali, C. Kress, B. Hooge-
land, A. Bracher, F. Hausch, J. Med. Chem. 2012, 55, 4114; b) R.
Gopalakrishnan, C. Kozany, Y. Wang, S. Schneider, B. Hooge-
land, A. Bracher, F. Hausch, J. Med. Chem. 2012, 55, 4123.
[10] S. H. Hong, D. P. Sanders, C. W. Lee, R. H. Grubbs, J. Am.
Chem. Soc. 2005, 127, 17160.
Keywords: drug discovery · FKBP ligands · cyclization ·
neurotrophic agents · rational design
.
[1] a) S. L. Schreiber, Science 1991, 251, 283; b) F. Edlich, G. Fischer,
Handb. Exp. Pharmacol. 2006, 172, 359; c) S. Gaali, R. Gopa-
lakrishnan, Y. Wang, C. Kozany, F. Hausch, Curr. Med. Chem.
2011, 18, 5355.
[2] a) R. E. Babine, J. E. Villafranca, B. G. Gold, Expert Opin. Ther.
Pat. 2005, 15, 555; b) M. Gerard, A. Deleersnijder, J. Demeu-
lemeester, Z. Debyser, V. Baekelandt, Mol. Neurobiol. 2011, 1;
c) C. B. Kang, Y. Hong, S. Dhe-Paganon, H. S. Yoon, Neuro-
signals 2008, 16, 318; d) S. H. Snyder, D. M. Sabatini, M. M. Lai,
J. P. Steiner, G. S. Hamilton, P. D. Suzdak, Trends Pharmacol.
Sci. 1998, 19, 21; e) C. Christner, T. Herdegen, G. Fischer, Mini-
Rev. Med. Chem. 2001, 1, 377; f) J. Xu, M. H. Lacoske, E. A.
Theodorakis, Angew. Chem. Int. Ed. 2014, 53, 956; Angew.
Chem. 2014, 126, 972.
[3] a) C. A. Hoeffer, W. Tang, H. Wong, A. Santillan, R. J.
Patterson, L. A. Martinez, M. V. Tejada-Simon, R. Paylor, S. L.
Hamilton, E. Klann, Neuron 2008, 60, 832; b) S. Shim, J. P. Yuan,
J. Y. Kim, W. Zeng, G. Huang, A. Milshteyn, D. Kern, S.
Muallem, G.-l. Ming, P. F. Worley, Neuron 2009, 64, 471.
[4] a) D. L. Cioffi, T. R. Hubler, J. G. Scammell, Curr. Opin.
Pharmacol. 2011, 11, 308; b) N. M. Galigniana, L. T. Ballmer, J.
Toneatto, A. G. Erlejman, M. Lagadari, M. D. Galigniana, J.
Neurochem. 2012, 122, 4; c) M. V. Schmidt, M. Paez-Pereda, F.
Holsboer, F. Hausch, ChemMedChem 2012, 7, 1351; d) C. L.
Storer, C. A. Dickey, M. D. Galigniana, T. Rein, M. B. Cox,
Trends Endocrinol. Metab. 2011, 22, 481.
[11] C. Kozany, A. Mꢂrz, C. Kress, F. Hausch, ChemBioChem 2009,
10, 1402.
[12] No attempts were made at this point to control the regio- or
stereochemistry. We rather opted to use the lack thereof to our
advantage to rapidly generate diversity at the C5 position.
[13] A. Bracher, C. Kozany, A. K. Thost, F. Hausch, Acta Crystallogr.
Sect. D 2011, 67, 549.
[14] H. R. Quintꢃ, D. Maschi, C. Gomez-Sanchez, G. Piwien-Pilipuk,
M. D. Galigniana, J. Neurochem. 2010, 115, 716.
[15] a) R. D. Price, T. Yamaji, N. Matsuoka, Br. J. Pharmacol. 2003,
140, 825; b) W. P. Revill, J. Voda, C. R. Reeves, L. Chung, A.
Schirmer, G. Ashley, J. R. Carney, M. Fardis, C. W. Carreras, Y.
Zhou, L. Feng, E. Tucker, D. Robinson, B. G. Gold, J.
Pharmacol. Exp. Ther. 2002, 302, 1278.
[16] a) J. Heitman, N. R. Movva, M. N. Hall, Science 1991, 253, 905;
b) A. M. Marz, A.-K. Fabian, C. Kozany, A. Bracher, F. Hausch,
Mol. Cell. Biol. 2013, 33, 1357.
348
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 345 –348