Kallikrein 5 inhibitors identified through structure based drug design in search for a treatment for Netherton Syndrome

Article abstract of DOI:10.1016/j.bmcl.2019.01.020

Netherton syndrome (NS) is a rare and debilitating severe autosomal recessive genetic skin disease with high mortality rates particularly in neonates. NS is caused by loss-of-function SPINK5 mutations leading to unregulated kallikrein 5 (KLK5) and kallikrein 7 (KLK7) activity. Furthermore, KLK5 inhibition has been proposed as a potential therapeutic treatment for NS. Identification of potent and selective KLK5 inhibitors would enable further exploration of the disease biology and could ultimately lead to a treatment for NS. This publication describes how fragmentation of known trypsin-like serine protease (TLSP) inhibitors resulted in the identification of a series of phenolic amidine-based KLK5 inhibitors 1. X-ray crystallography was used to find alternatives to the phenol interaction leading to identification of carbonyl analogues such as lactam 13 and benzimidazole 15. These reversible inhibitors, with selectivity over KLK1 (10–100 fold), provided novel starting points for the guided growth towards suitable tool molecules for the exploration of KLK5 biology.

Full text of DOI:10.1016/j.bmcl.2019.01.020

Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Kallikrein 5 inhibitors identified through structure based drug design in
search for a treatment for Netherton Syndrome
⁎
Gemma V. Whitea, , Emma V. Edgara, Duncan S. Holmesa, Xiao Qing Lewella, John Liddlea,
Oxana Polyakovaa, Kathrine J. Smitha, James H. Thorpea, Ann L. Walkera, Yichen Wangb,
Robert J. Younga, Alain Hovnanianb
a GlaxoSmithKline R&D, Medicines Research Centre, Gunnels Wood Road, Stevenage, Hertfordshire SG1 2NY, UK
b INSERM UMR1163 Laboratory of Genetic Skin Diseases, Imagine Institute and Université Paris Descartes-Sorbonne Paris Cité, Paris, France
A R T I C L E I N F O
A B S T R A C T
Keywords:
FBDD
Netherton syndrome (NS) is a rare and debilitating severe autosomal recessive genetic skin disease with high
mortality rates particularly in neonates. NS is caused by loss-of-function SPINK5 mutations leading to un-
regulated kallikrein 5 (KLK5) and kallikrein 7 (KLK7) activity. Furthermore, KLK5 inhibition has been proposed
as a potential therapeutic treatment for NS. Identification of potent and selective KLK5 inhibitors would enable
further exploration of the disease biology and could ultimately lead to a treatment for NS. This publication
describes how fragmentation of known trypsin-like serine protease (TLSP) inhibitors resulted in the identifica-
tion of a series of phenolic amidine-based KLK5 inhibitors 1. X-ray crystallography was used to find alternatives
to the phenol interaction leading to identification of carbonyl analogues such as lactam 13 and benzimidazole
15. These reversible inhibitors, with selectivity over KLK1 (10–100 fold), provided novel starting points for the
guided growth towards suitable tool molecules for the exploration of KLK5 biology.
KLK1
KLK5
LEKTI
Netherton syndrome
SPINK5
TLSP
Netherton syndrome (NS) is a rare genetic skin disorder char-
acterised by ichthyosis, constant allergy and severe itch.1 It can affect
all ages, but is most severe in neonates where dehydration and infection
are major causes of morbidity and mortality.2 The condition is caused
by loss of function mutations in the serine protease inhibitor kazal-type
5 gene (SPINK5), resulting in partial or complete loss of expression of
the protein lympho-epithelial kazal-type-related inhibitor (LEKTI),3 the
endogenous inhibitor of numerous serine proteases, including kallikrein
5 (KLK5).4 It has been shown that in the absence of LEKTI, unopposed
KLK5 activity activates KLK7 and both KLK5 and KLK7 degrade des-
moglein-1 (Dsg1),5 leading to stratum corneum detachment from the
granular layer, resulting in a severe defective skin barrier.6–8 Potent and
selective inhibitors of KLK5, a trypsin-like serine protease (TLSP),
would be invaluable probes to deconvolute the associated biology,
however none are currently available. Since we started this work
structures of a number of KLK5 inhibitors have been published, most
are weak inhibitors of KLK5 with modest selectivities over other related
proteases (Fig. 1).9–12 This paper describes progress towards the
discovery of potent and selective, small molecule KLK5 inhibitors for
use as tool molecules in the search for a topical medicine to treat NS.
Screening of targeted fragments of known TLSP inhibitors was a key
element of our screening approaches, encompassing both neutral and
basic motifs that respectively bind to key Tyr and Asp residues at the
base of the S1 pocket. Topical administration offered a potential route
for delivery of basic compounds, avoiding a restraint on many potent
basic TLSP inhibitors that demonstrated limited or no oral exposure.13
Ensuing fragments demonstrating KLK5 inhibition at a single con-
centration (10 µM or 1 mM), were screened in a full concentration re-
sponse.14 Salicylamide 1 was a hit with good ligand efficiency (LE) and
lipophilic ligand efficiency (LLE(AT)),15 which gratifyingly yielded a li-
ganded structure using back-soaking in a surrogate KLK6 mutant
system16 (Fig. 2).
Salicylamide 1 was anchored through a network of hydrogen
bonding interactions from both the amidine and phenol motifs. The
phenol formed hydrogen bonding interactions with the catalytic serine
(Ser195) and histidine (His57) residues (Fig. 3), whilst the benzamidine
Abbreviations: Dsg1, desmoglein-1; FBDD, fragment based drug discovery; HTS, high throughput screening; KLK1, kallikrein 1; KLK5, kallikrein 5; KLK7, kallikrein
7; KLK14, kallikrein 14; LEKTI, lympho-epithelial kazal-type-related inhibitor; NS, Netherton syndrome; SPINK5, serine protease inhibitor kazal-type 5 gene; TLSP,
trypsin-like serine protease
⁎ Corresponding author.
E-mail address: gemma.v.white@gsk.com (G.V. White).
https://doi.org/10.1016/j.bmcl.2019.01.020
Received 11 December 2018; Received in revised form 16 January 2019; Accepted 17 January 2019
0960-894X/©2019ElsevierLtd.Allrightsreserved.
Pleasecitethisarticleas:White,G.V.,Bioorganic&MedicinalChemistryLetters,https://doi.org/10.1016/j.bmcl.2019.01.020

G.V. White et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
motif formed a tight network of hydrogen binding interactions, in-
cluding the characteristic TLSP salt bridge interaction with Asp189 in
the S1 binding pocket. Additional contacts were made with Ser190 and
Asn217 and through water to Trp215. The amide linker between the
benzamidine and phenol holds the compound in a rigid planar con-
formation.
Salicylamide 1 offered good vectors to explore growth into the S1′
and S3 pockets of the protein (Fig. 2). Targeting the S1′ and S3 pockets
was attractive due to differences in the residues between KLK5 and the
key liability target, KLK1. Selectivity over KLK1, also known as tissue
kallikrein, was sought due to its critical role in the regulation of blood
pressure via the cleavage of kininogen to release the vasoactive bra-
dykinin.17
Selectivity over KLK1 was also believed to be a good indicator of the
likely broader selectivity profile, given its level of sequence homology;
limitation of potential systemic exposure may obviate this particular
requirement, but nonetheless it gave a useful baseline of likely TLSP
selectivity. Between KLK5 and KLK1, the protein backbone tertiary
structure is largely conserved for the S1 and S3 pockets but shows a
greater deviation in the S1′ and S2 pockets. Key residues of interest in
S1′ were TYR41 (a GLN in KLK1) and LYS60 (a SER in KLK1). In the S3
pocket, Tyr218 in KLK5 is Val in KLK1, creating differences in the size
and electrostatic preference of molecules that occupy this region. The
S1′ and S3 pockets were explored in turn.
Fig. 1. Structures of published KLK5 inhibitors.
Substituents at the 3-position, ortho to the phenol, appeared to
enable growth towards the S1′ pocket. Introduction of a methyl group
(2) at the 3-position of 1 had no effect on the KLK5 potency (Table 1),
but ∼3 fold increase in KLK1 inhibitory activity was observed. 3-chloro
(3) increased both the KLK5 and KLK1 potency. Di-phenol 4 was
equipotent against both proteases, but the methoxy analogue 5, which
protruded further into the S1′ pocket, resulted in a compound which
was ∼10–30 fold more potent against both KLK5 and KLK1 compared
to 1. Disappointingly, growth of the methoxy to ethoxy compound 6
resulted in an increase in KLK1 potency, but was detrimental to KLK5
activity. Overlay of x-ray crystal structures for the methoxy 5 and
ethoxy 6 analogues show close alignment, maintaining the hydrogen
bond network with both the benzamidine and phenol to the protein
(Fig. 4). The reduction in KLK5 potency can be rationalised by the
Fig. 2. a) Salicylamide 1 structure and data; b) Contoured structure of the
active site and S1 pocket with 1, crystallised in hK6 WT, 12 superposed onto
KLK5 residues. Key to pockets: S1: Orange; S3: Yellow; S1′: Blue; Oxyanion
hole: Grey; Catalytic Residues: Pink. All residue numbering corresponds to the
targeted KLK5 enzyme.
Table 1
Substitution of the 3 position on the salicylamide ring.
Cmpd no. R3
KLK5
Mean
pIC50
KLK5 LE/
KLK1
mean
pIC50
Phenol pKa
(measured)18
d
LLE(AT)
1a
2
H
5.2
5.1
5.6
5.6
0.37/0.35
0.35/0.30
0.38/0.33
0.38/0.40
0.41/0.41
0.33/0.31
4.9
5.5
6.2
5.4
6.5
6.9
6.8
6.8
5.0
–
Me
Cl
3b
4b
5
OH
OMe 6.3
6.6b
–
6c
OEt
5.3
n numbers ≥2 unless otherwise referenced.19
a
b
c
d
Formic acid salt.
TFA salt
Fig. 3. Active site and S1 pocket residues with 1, crystallised in hK6 WT, il-
lustrating the network of hydrogen bonding interactions that anchor the
binding.
HCl salt.
LE = pIC50 * 1.37/nheavy and LLE(AT) = 0.111 + 1.36 * (pIC50-clogP)/
nheavy.
2

G.V. White et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Fig. 5. Overlay of x-ray crystal structures of salicylamide 1 in grey with the
racemic lactam 13 in pink.
Fig. 4. Overlay of hK6 WT x-ray crystal structures soaked with phenol 1 in
white, methoxy 5 in blue and ethoxy 6 in orange showing close alignment of the
ligands.
due to the hypothesis that o-substitution would disrupt the planarity of
the amide linker with the benzamidine ring. Due to 2′- and 3′- sub-
stitution providing poor growth opportunities, focus returned to the
salicylamide, with attempts to replace the phenol, in a quest to improve
potency for KLK5 with selectivity versus KLK1 (Fig. 5.).
forced unfavourable conformation of the ethoxy chain occupying an
orthogonal position in KLK5 (hK6 WT) (Fig. 4). Substitution at the 3-
position, also had a significant effect on the pKa of the phenol (Table 1),
however, no clear correlation of pKa with KLK5 and/or KLK1 potency
was observed.18 The bound structure suggested that the substituents
from the 4-position would grow towards a solvent exposed region and
the few simple substituents introduced showed no significant im-
provement to the KLK5 potency (data not shown).
Inspection of the bound structure and the observation that the pKa
of the phenol did not correlate with KLK5 potency (Table 1), prompted
consideration of carbonyl mimics of the phenol. The carbonyl oxygen
lone pairs offered potential to form hydrogen bonding interactions to
the catalytic serine and histidine, or occupy the oxyanion hole forming
hydrogen bond interactions to the backbone NH of Gly193 and Ser195.
Methylated pyridone 12 (Table 3) showed reduced KLK5 and KLK1
potency compared to its corresponding 3-methyl phenol analogue 2.
Saturation of the aromatic ring in lactam 13 and cyclic urea 14 pro-
vided the first analogues with selectivity for KLK5 over KLK1 (∼100-
fold), despite modest KLK5 potency. A binding-based explanation of the
selectivity was not obvious from x-ray crystallography, however,
In the optimisation of the benzamidine containing ring, introduction
of small substituents in the 2′- or 3′- position probed binding interac-
tions in this region of the protein, and to test if substituents, particularly
at the 3′- position, were tolerated for potential growth into S3.
Introduction of F at the 3′- position (7 and 10) was tolerated, however,
the larger Cl (11) was detrimental to KLK5 potency. It was probable
that the reduction in KLK5 potency was due to the Cl forcing the
amidine out of the plane of the phenyl ring, preventing the optimal
binding in the pocket (Table 2).
Table 3
Substitution at the 2′- position was tolerated with the introduction
of F (9), but methyl analogue 8 had no effect on KLK5 potency.
Accessing the S3 pocket from the 2′- position was deemed unsuitable
Phenol replacements. aformic acid salt. n numbers ≥2 unless otherwise refer-
enced.19
Table 2
a
Substitution at R2′ and R3′ on the amidine motif. formic acid salt. n numbers
≥2 unless otherwise referenced.14
Cmpd no. KLK5 Mean pIC50 R-Group
KLK5 LE/LLE(AT) KLK1 Mean pIC50
12a
3.8
5.6
5.0
5.0
4.9
0.26/0.41
0.38/0.33
0.36/0.52
0.36/0.43
0.31/0.31
3.3
6.2
3.3
3.2
3.7
2
Cmpd no. R3
R3′ R2′ KLK5 Mean
pIC50
KLK5 LE/
LLE(AT)
KLK1 mean
pIC50
13
14
15
1a
H
H
F
H
H
5.2
4.7
0.37/0.35
0.32/0.28
0.34/0.33
0.38/0.38
0.41/0.41
0.37/0.36
0.31/0.26
4.9
4.9
5.3
5.5
6.5
6.6
5.9
7a
H
8a
H
H
H
H
F
Me 4.9
9a
H
F
5.6
6.3
6.0
5.0
5
OMe
OMe
H
H
H
10a
11a
OMe Cl
3

G.V. White et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
Scheme 1. (i) 4-aminobenzonitrile, PCl3, chlor-
obenzene, 130 °C, sealed tube, 21%; (ii) MeOH, HCl
(g), then 2 N Methanolic NH3, RT, 72%; (iii) 2-
Hydroxyacetic acid, 4 N HCl(aq), 130 °C, sealed tube,
96%; (iv) Dess-Martin periodiane, DCM, 0 °C – RT,
33%; (v) 4-Aminobenzonitrile, NaOCl, Bu4NHSO4,
PEG, 120 °C, 7.5%; (vi) MeOH, 0 °C, HCl(g), then 4 N
Methanolic NH3, RT – 50 °C, 97%.21.
overlay of x-ray crystal structures of phenol 1 with lactam 13 showed a
clear change in binding mode with the carbonyl oxygen of 13 occu-
pying the oxyanion hole forming hydrogen bonding interactions with
the NH of the backbone of Gly193 (2.86 Å) and Ser195 (2.92 Å). The
lactam 13 and urea 14 were down prioritised as they proved synthe-
tically challenging, the former racemised readily and the urea was
poorly soluble.
Acknowledgement
The authors thank Pam Nassau for preparation of the KLK5 protein.
This work was funded by GlaxoSmithKline.
Appendix A. Supplementary data
In addition to these carbonyl analogues, a series of unsaturated
heterocyclic derivatives were prepared to target introduction of a water
mediated hydrogen-bond with the oxyanion hole as a replacement for
the direct H-bond formed by the phenolic derivatives. Of these benzi-
midazole 15 was the most promising, showing moderate KLK5 potency,
albeit reduced selectivity for KLK1 (10-fold) compared to the carbonyl
analogues 13 and 14. X-ray crystallography showed that the catalytic
serine had a split distribution, indicating hydrogen bonding interactions
to the both the benzimidazole nitrogen and the NH of the amide linker.
Interestingly, the rather linear structure of 15 was quite distinct for the
bent conformation that the thrombin-inhibiting drug dabigatran, itself a
benzimidazole, adopts; this gave confidence that selectivity could be
achieved over this TLSP subfamily. Together, the compounds in Table 3
indicated that the fragment growth strategy could provide compounds
with enhanced potency and selectivity; the synthetic tractability and
stability of 15 offered opportunities for further growth and optimisa-
tion.
Supplementary data to this article can be found online at https://
doi.org/10.1016/j.bmcl.2019.01.020.
References
1. Hovnanian A. Netherton syndrome: skin inflammation and allergy by loss of protease
inhibition. Cell Tissue Res. 2013;351:289–300.
2. Craiglow BG. Ichthyosis in the newborn. Semin Perinatol. 2013;37:26–31.
3. Mägert H-J, Ständker L, Kreutzmann P, et al. LEKTI, a novel 15-domain type of
human serine proteinase inhibitor. J Biol Chem. 1999;274:21499–21502.
4. Furio L, Pampalakis G, Michael IP, Nagy A, Sotiropoulou G, Hovnanian A. KLK5
inactivation reverses cutaneous hallmarks of Netherton syndrome. PLoS Genet.
2015;11:e1005389.
5. Descargues P, Deraison C, Bonnart C, et al. Spink5-deficient mice mimic Netherton
syndrome through degradation of desmoglein 1 by epidermal protease hyperactivity.
Nat Genet. 2004;37:56–65.
6. Hovnanian A. Netherton syndrome: new advances in the clinic, disease mechanism
and treatment. Expert Rev Dermatol. 2012;7:81–92.
7. Furio L, Hovnanian A. Netherton syndrome: defective kallikrein inhibition in the skin
leads to skin inflammation and allergy. Biol Chem. 2014;395:945–958.
8. Van Smeden J, Janssens M, Boiten WA, et al. Intercellular skin barrier lipid com-
position and organization in Netherton syndrome patients. J Invest Dermatol.
2014;134:1238–1245.
The preparation of the most potent phenol analogue, methoxy 5,
and benzimidazole 15 are shown in Scheme 1.20
9. Teixeira TSP, Freitas RF, Abrahão Jr O, et al. Biological evaluation and docking
studies of natural isocoumarins as inhibitors for human kallikrein 5 and 7. Bioorg Med
Chem Lett. 2011;21:6112–6115.
In conclusion, through targeted fragment screening and structure
guided drug design, novel KLK5 inhibitors with encouraging potency
and efficiency, with 10–100 fold selectivity over the key liability target
KLK1 have been identified. The addition of substituents to the hit sal-
icylamide 1 did not provide a significant increase in potency or show
evidence of potential to achieve selectivity over liability target KLK1.
However, phenol replacements led to 100-fold selectivity for carbonyl
analogues lactam 13 and cyclic urea 14. Synthetic tractability and
stability issues led to benzimidazole 15, with a novel binding interac-
tion with the catalytic serine. Benzimidazole 15 is ligand efficient,
shows some selectivity and possessed discernible growth vectors to-
wards the S1′ and S3 pockets. Together, these achieved a significant
milestone towards the goal of accessing tool molecules for exploration
of KLK5 biology.
10. Reboud RM, El Amri-Yerroum C, Tan X. Use of coumarin derivatives for the pre-
paration of drugs for treating skin diseases. PCT WO2013010963A1, 24 Jan 2013.
11. Oliveira JPC, Freitas RF, Silva de Melo L, et al. Isomannide-based peptidomimetics as
inhibitors for human tissue kallikreins 5 and 7. ACS Med Chem Lett. 2014;5:128–132.
12. Waagberg F, Leonardsson G. Benzoxazinone derivatives for treatment of skin dis-
eases. PCT WO2015112081A1, 30 Jul 2015.
13. Young RJ. The successful quest for oral factor Xa inhibitors; learnings for all of
medicinal chemistry? Bioorg Med Chem Lett. 2011;21:6228–6235.
14. Two functional protease assays (KLK1 and KLK5), utilising fluorogenic tri-peptide
substrates, were run in a kinetic format, at substrate kM measuring inhibitor potency
(pIC50). IC50 determinations performed as a minimum in duplicate against re-
combinant human KLK5 using an 11 point dose response curve. IC50 values are re-
ported as an average of two or more experiments, from hard values where available.
Full assay details are available in the Supplementary data.
4

G.V. White et al.
Bioorganic&MedicinalChemistryLettersxxx(xxxx)xxx–xxx
15. Mortenson PN, Murray CW. Assessing the lipophilicity of fragments and early hits. J
Comput Aided Mol Des. 2011;25:663–667.
stated. Compound 2: 91% by LCMS, ≥95% by NMR; Compound 3: 92% by LCMS,
≥95% by NMR; Compound 4: 83% by LCMS, ≥90% by NMR; Compound 14: 92% by
LCMS, ≥95% by NMR.
16. Crystallography methods: Protein production, crystallisation, crystal soaking ex-
periments and structure determinations were carried out using the methods de-
scribed in Thorpe et al. (manuscript in preparation). X-ray datasets were collected at
the European Synchrotron Radiation Facility (ESRF) or Diamond. The coordinates
and structure factors have been deposited in the Protein Data Bank under the ac-
cession codes (to be inserted once manuscript accepted).
20. General experimental details: All commercial reagents and solvents were obtained
from commercial sources and used without further purification. 1H NMR spectra,
chemical shifts are given in ppm (δ) relative to tetramethylsilane (TMS) as an in-
ternal standard. Purity of the final compounds was of > 95% using interpretation of
a combination of LCMS and NMR data unless stated otherwise.
17. Madeddu P, Emanueli C, El-Dahr S. Mechanisms of disease: The tissue kallikrein-
kinin system in hypertension and vascular remodeling. Nature Clin Pract. 2007;3:208.
18. The pKa was determined on a Sirius T3 (Sirius Analytical Inc, UK) instrument by UV
spectroscopy from acid-base titration of the compounds in the presence of various
concentrations of methanol co-solvent. The aqueous pKa is calculated from the
Yasuda-Shedlovsky extrapolation.
21. Characterisation for phenol 5: 1H NMR (400 MHz, DMSO-d6) δ = 15.58 (br s, 1H),
10.43–8.16 (m, 4H), 7.89 (d, J = 8.8 Hz, 2H), 7.81 (d, J = 8.8 Hz, 2H), 7.38 (dd,
J = 1.8, 7.8 Hz, 1H), 6.70 (dd, J = 1.8, 7.8 Hz, 1H), 6.16 (t, J = 7.8 Hz, 1H), 3.71 (s,
3H). LCMS m/z = 286 (M+H)+. Characterisation for benzimidazole 15: 1H NMR
(400 MHz, DMSO-d6) δ = 11.75–9.00 (br s, assumed 3H), 8.48 (br s, 1H), 8.13 (d,
J = 8.7 Hz, 2H), 7.88–7.81 (m, 3H), 7.75 (d, J = 8.2 Hz, 1H), 7.49–7.43 (m, 1H),
19. Compound purity: Compounds are > 95% purity by LCMS and NMR unless otherwise
7.42–7.35 (m, 1H), 4.19 (s, 3H); LCMS m/z = 294 (M+H)+
.
5