Reverse Hydroxamate Inhibitors of Bone Morphogenetic Protein 1

Article abstract of DOI:10.1021/acsmedchemlett.8b00173

Bone Morphogenetic Protein 1 (BMP1) inhibition is a potential method for treating fibrosis because BMP1, a member of the zinc metalloprotease family, is required to convert pro-collagen to collagen. A novel class of reverse hydroxamate BMP1 inhibitors was

Full text of DOI:10.1021/acsmedchemlett.8b00173

Letter
Cite This: ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX
Reverse Hydroxamate Inhibitors of Bone Morphogenetic Protein 1
Lara S. Kallander,*,† David Washburn,† Mark A. Hilfiker,† Hilary Schenck Eidam,†
Brian G. Lawhorn,† Joanne Prendergast,† Ryan Fox,† Sarah Dowdell,† Sharada Manns,† Tram Hoang,†
Steve Zhao,† Guosen Ye,† Marlys Hammond,† Dennis A. Holt,† Theresa Roethke,† Xuan Hong,‡
Robert A. Reid,‡ Robert Gampe,‡ Hong Zhang,‡ Elsie Diaz,‡ Alan R. Rendina,‡ Amy M. Quinn,‡
and Bob Willette†
‡
†Heart Failure Discovery Performance Unit, Metabolic Pathways and Cardiovascular Therapeutic Area, and Platform Technology
and Sciences, GlaxoSmithKline Pharmaceuticals, King of Prussia, Pennsylvania 19406, United States
S
* Supporting Information
ABSTRACT: Bone Morphogenetic Protein 1 (BMP1) inhibition is a potential method for treating fibrosis because BMP1, a
member of the zinc metalloprotease family, is required to convert pro-collagen to collagen. A novel class of reverse hydroxamate
BMP1 inhibitors was discovered, and cocrystal structures with BMP1 were obtained. The observed binding mode is unique in
that the small molecule occupies the nonprime side of the metalloprotease pocket providing an opportunity to build in
metalloprotease selectivity. Structure-guided modification of the initial hit led to the identification of an oral in vivo tool
compound with selectivity over other metalloproteases. Due to irreversible inhibition of cytochrome P450 3A4 for this chemical
class, the risk of potential drug−drug interactions was managed by optimizing the series for subcutaneous injection.
KEYWORDS: BMP-1, BMP1, PCP, TLL1, TLL2, MMP12, fibrosis
one Morphogenetic Protein 1 (BMP1, procollagen C-
proteinase, PCP) is part of a family of tolloid metal-
B
loproteases (including TLL1 and TLL2) that cleave the C-
terminal side of propeptides.1 This extracellular cleavage is
required as part of the cascade that generates mature collagen.
Collagen is important for normal physiological functioning;
however, excessive collagen or fibrosis occurs in certain disease
states of the heart, liver, kidney, lung, and skin. Other
researchers have reported that an antibody that inhibits BMP1
is advantageous for post myocardial infarction recovery in rats
and that this could theoretically be expanded to humans.2,3
Myocardial infarction often results in significant fibrotic
scarring of the heart with concomitant loss of cardiac function.
The potential to improve cardiac function in patients after a
myocardial infarction by small molecule inhibition is clearly
compelling from a treatment perspective.
Figure 1. Structure of published BMP1 inhibitors.
As an antifibrotic target, BMP1 has attracted interest for at
least two decades.4 The most potent small molecule BMP1
inhibitors, exemplified by compounds 1−3, have been reported
by Pfizer,5 Roche Biosciences,6 and Fibrogen,7 respectively, but
without reported advancement to clinical study (Figure 1).
These compounds share a common hydroxamic acid (C(O)-
NHOH) zinc-binding motif. However, many clinical trials of
hydroxamic acid-containing inhibitors have failed, aside from
Vorinostat,8 largely a result of poor selectivity for matrix
metalloproteinases (MMP) and/or poor metabolic stability of
the hydroxamate.9 We sought BMP1 inhibitors with an
alternative pharmacophore devoid of these liabilities. The
goal of this effort was to discover potent and selective BMP1
Received: April 13, 2018
Accepted: June 19, 2018
© XXXX American Chemical Society
A
DOI: 10.1021/acsmedchemlett.8b00173
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
a
inhibitors with suitable systemic exposure to test for the
potential treatment of fibrosis.
Scheme 1. Synthesis of Reverse Hydroxamate Inhibitors
High-throughput screening for BMP1 inhibition revealed
compound 4,10 a reverse hydroxamate (NH(OH)C(O)),
having submicromolar activity against BMP1 but lacking
MMP selectivity. The X-ray crystal structure of the BMP1−4
complex (Figure 2) revealed that the reverse hydroxamate
a
Reagents and conditions: (a) 2.5 M n-BuLi in hexanes, THF, −78 to
−25 °C, 5 h (93%); (b) TiCl4, DIEA, DCM, NMP, ice/acetone bath,
1.5 h (87%); (c) BnONH2, 2 M Al(CH3)3 in toluene, THF, 0−5 °C,
1.5 h (70%); (d) MsCl, pyridine, 0−10 °C, 2 h (91%); (e) 40%
tetrabutylammonium hydroxide in water, 2-MeTHF, 50 °C, 2 h; (f)
CDI, formic acid, 2-MeTHF, 0−5 °C, 1 h (77% over two steps); (g)
EDC, HOBt, DIEA, DMF, THF, 50 °C, 1 h (40−70%); (h) 10% Pd/
C, H2, EtOH, RT, 3 h (70%).
stereochemistry of the R1 and R2 side chains were set using a
chiral Evans oxazolidinone. The reverse hydroxamate moiety
was introduced by displacing the chiral auxiliary with
benzyloxyamine, Mitsunobu ring closure to the beta-lactam,
and hydrolytic lactam opening followed by formylation to
provide the protected reverse hydroxamate carboxylic acid
intermediate. This was then coupled to an elaborated diamine
via standard amide bond forming reaction and the final
product was deprotected to reveal the reverse hydroxamate
diamine product while maintaining the chiral integrity.
Figure 2. (A) Compound 4. (B) Compound 4 and key interactions
with BMP1. Compound 4 is shown in green carbon. Key residues of
BMP1 are shown in cyan carbon. The zinc ion is in magenta. (C) X-
ray crystal structure of the BMP1−4 complex. Compound 4 and key
interactions with BMP1 with the protein shown in cyan ribbon and
light gray surface (RCSB 6BSM).
Initial explorations of the pentyl side chain SAR (Table 1)
yielded no significant enhancement of potency. Reduction in
Table 1. BMP1 Potency and MMP12 Selectivity
moiety engages in a bidentate zinc coordination, but unlike
most MMP inhibitors, compound 4 exclusively occupies the
nonprime side of the substrate binding pocket. Further,
inhibitor binding is accommodated by a significant movement
of a protein flap that contains a rare11 vicinal disulfide bond
formed between Cys65 and Cys66 and that closes over the S1
site in the published apo structure.12
a
a
BMP1 IC50 MMP12 IC50
(nM) (nM)
compd
R1
R2
The central methylene diamide of 4 is deeply buried and
engages in two direct H-bond interactions with Ser67. Between
this diamide and the reverse hydroxamate, the n-pentyl side
chain lies in an extended conformation, tucked underneath the
hydrophobic disulfide-containing flap. The benzofuran moiety
occupies a rather large shallow pocket, presumably the S3 site,
and makes a water-mediated interaction with the phenyl ring of
Tyr68.
4
5
6
7
8
9
10
R-CH2CH2CH2CH2CH3
R-CH2CH2CH3
H
R-CH2Ph
S-CH2CH2CH2CH2CH3
H
160 400
H
H
H
H
1250
6300
160
1250
>10,000
4000
4000
>10,000
126
>10,000
200
>10,000
R-CH2CH2CH2CH2CH3 S-Et
R-CH2CH2CH2CH2CH3 R-Et
Assay protocols are published.13 IC50 values are the average of at
least two determinations.
a
The reverse hydroxamate compounds were synthesized
according to the conditions shown in Scheme 1.13 The
B
DOI: 10.1021/acsmedchemlett.8b00173
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
the chain length resulted in a loss of activity with compounds 5
and 6, which is consistent with an efficient hydrophobic
interaction. Increasing the size and restricting the conforma-
tional freedom had no measurable benefit on inhibition (7). It
is clear from these compounds that the interaction of the
lipophilic group with the vicinal disulfide accounts for almost
two logs of potency at BMP1, which may be partially explained
by preorganization into the bioactive conformation in solution.
Matrix metalloprotease 12 (MMP12) was tested to represent
general metalloprotease selectivity, and the SAR for this
enzyme generally paralleled BMP1 (Table 1), including the
stereochemical preference at R1 (8). MMP9 activity (see
Supporting Information) was very similar to MMP12.
reduced BMP1 activity.17 To address this issue, we decided to
pursue considerably more potent and soluble molecules that
could be dosed subcutaneously and thus avoid the possibility
of inactivating gut 3A4 that could occur with oral delivery.
To improve the potency, substitution on the aryl ring (Table
2, R3) was explored. Potency improvements were obtained
Table 2. BMP1 Potency with Substituents
Consideration of the X-ray crystal structures of related
inhibitors in MMPs along with the docking of compound 4
into the MMP12 crystal structure (Supporting Information)
suggested that 4 likely binds in the MMP12 prime site and that
the n-pentyl side chain occupies a deep hydrophobic S1
pocket. Furthermore, and quite significantly, it appeared that
the steric tolerance of a substituent adjacent to the reverse
hydroxamate nitrogen (vicinal to the n-pentyl side chain)
would have opposite stereochemical preferences by BMP1 and
MMPs. In particular, an R-substituent alpha to the reverse
hydroxamate might not be tolerated by MMP12 but was
predicted to be tolerated by BMP1. Similarly, an S-substituent
was predicted to have the opposite selectivity. Preparation of
alpha ethyl analogs 9 and 10 resulted in one of each analog
being preferred by each protease as predicted. Specifically,
compound 10 has good activity at BMP1 and no measurable
activity at MMP12.
BMP1 IC50
compd R2
R3
(nM)
12
13
14
15
16
17
18
19
20
21
H
H
H
H
H
H
25
8.0
20
1.0
1.0
3-CO2H
3-OCH3
3-OCH2CH3
4-CO2H
b
R-Et 3-OCH2CH3, 5-CO2H
R-Et 3-OCH2CH3, 4-CO2H
R-Et 3-OCH2CH3, 5-PO3H2
R-Et 3-OCH2CH3, 4-PO3H2
<1
b
<1
b
<1
b
<1
b
R-Et 3-OCH2CH3, 5−C(O)-S-NCH2(CO2H)
<1
(CH2CO2H)
b
22
R-Et 3-OCH2CH3, 4−C(O)-S-NCH2(CO2H)
<1
(CH2CO2H)
b
Lower limit of assay sensitivity is approximately 250 pM.
Exploration of SAR around the benzofuran moiety revealed
that replacement of the fused bicycle with the more elongated
phenyl furan led to a 6-fold boost in potency. When combined
with the R-ethyl group alpha to the reverse hydroxamate, this
led to the discovery of compound 11 (Figure 3), which
with both para and meta substitution (compounds 13−16),
although 3-ethoxy and 4-carboxy showed the most dramatic
improvements (>10-fold). The ethoxy group on 15 may be
making a positive van der Waals interaction with the upper
pocket defined by Ala78 and Gln77, whereas 16 likely makes a
positive interaction with the flexible Arg72.
An even larger boost in potency was obtained when two
substituents were added. An entire series of compounds was
identified, which had acidic groups to facilitate solubility and
subnanomolar potency (compounds 17−22).
Biochemical BMP1 potency was measured using small
peptide substrates that included the BMP1 cleavage sequences
of physiological protein substrates. The lower limit of the
BMP1 profiling assay in Table 2 was ca. 250 pM (half the
enzyme concentration), so tight-binding analysis was used for
more potent inhibitors 19 and 22, incorporating an enzyme−
inhibitor preincubation step to ensure equilibrium binding. As
would be predicted from the crystal structure of 22,
compounds 19 and 22 are substrate-competitive inhibitors.
Binding of 19 is single step, reversible with an on-rate constant
(kon = 3.6 μM−1 s−1) measured by stopped flow and an off-rate
constant koff = 0.93 h−1 (t1/2 = 45 min) measured by jump
dilution. Increased potency of 22 compared to 19 appears to
be driven by a faster on-rate (kon = 55 μM−1 s−1) accompanied
by a similar off-rate constant koff = 2.3 h−1 (t1/2 = 18 min).
Inhibitor binding affinities (Ki) measured by tight binding were
comparable to those calculated from stopped flow and jump
dilution rate constants (Ki = koff/kon). Both 19 and 22 have
BMP1 Ki values that are low picomolar (40 and 6.8 pM,
respectively). Inhibition constants (Ki) of TLL1 for 19 and 22
(240 and 31 pM, respectively) and TLL2 for 19 and 22 (260
pM, and 39 pM, respectively) were obtained using similar
methods.
Figure 3. Structure of oral BMP1 inhibitor 11.
exhibited excellent inhibitory activity against BMP1 (32 nM),
TLL1 (100 nM), and TLL2 (160 nM) and was selective
against 15 metalloproteases (>30-fold relative to BMP1).14
Additionally, 11 had good oral exposure in a rat15 (dose
normalized area under the curve (DNAUC) = 1.6 h·kg/L) and
thus represents a selective in vivo tool compound.
Although compound 11 is a suitable oral tool for in vivo
experimentation, unfortunately, it demonstrates cytochrome
P450 3A4 time-dependent inhibition (4.3-fold decrease in
enzyme activity over 15 min @ 33 μM). For a medication, this
is concerning because it could potentially cause drug−drug
interactions for patients on other medications. The reverse
hydroxamate functionality appears to be responsible for this
effect, which has also been reported previously.16 Unfortu-
nately, BMP1 activity is dramatically reduced without the
reverse hydroxamate metal chelator, similar to a published
report that replacements for a hydroxamic acid metal chelator
C
DOI: 10.1021/acsmedchemlett.8b00173
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
To help explain the exceptional potency of compound 22, an
X-ray crystal structure bound to BMP1 was solved (Figure 4).
In summary, high throughput screening identified a reverse
hydroxamate pharmacophore that binds BMP1. Cocrystal
structures revealed a binding mode on the nonprime side of
the BMP1 active site accommodated by a shift in the vicinal
disulfide loop. The chemical series was optimized for selectivity
and exposure to discover an oral in vivo tool, compound 11.
This class of compounds demonstrates irreversible cytochrome
P450 3A4 inhibition, so the series was further optimized to
picomolar potency for subcutaneous injection to avoid gut
3A4, which could potentially cause drug−drug interactions.
Two compounds were identified that were suitable for profiling
as precandidate compounds, 19 and 22. Pharmacology data for
these compounds will be reported in subsequent publications.
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acsmedchem-
lett.8b00173.
Synthetic procedures and characterization data for
preparing BMP1 inhibitors, selectivity information,
protein crystallization, binding kinetics, and a docking
file (PDF)
Figure 4. Cocrystal structure of 22 bound to BMP1. Compound 22 is
shown in green carbon, BMP1 in cyan carbon and a light gray surface,
and the zinc is in magenta (RCSB 6BSL).
AUTHOR INFORMATION
Corresponding Author
*E-mail: Lara.S.Kallander@gsk.com.
ORCID
Lara S. Kallander: 0000-0002-6369-7031
Hilary Schenck Eidam: 0000-0003-0939-298X
Brian G. Lawhorn: 0000-0003-4618-5314
■
Several advantageous interactions were noted compared to
compound 4 (Figure 2) that are consistent with the SAR.
Introduction of the phenylfuran as a benzofuran replacement
produced an edge-to-face π−π interaction with Tyr68, and
substitution on the phenylfuran with a 3-ethoxy group led to
improved hydrophobic contact in the upper pocket. Addition-
ally, there appear to be key interactions between both acidic
groups of the aspartic acid substituent and Arg61 and Arg72,
all of which appear to contribute to the >10,000-fold increase
in BMP1 activity of 22 over 4.
Further considerations for the quality of compounds 19 and
22 are metalloprotease selectivity and multiple species
exposure. Both compounds were found to be >100-fold
selective against 14 metalloproteases.14 Also, compounds 19
and 22 had reasonable exposure, low to moderate Cl, and a
MRT > 2 h after subcutaneous injection in both rat15 and
dog15 (Figure 5). Due to the overall profile, both 19 and 22
were progressed for further development consideration.
Author Contributions
All authors have given approval to the final version of the
manuscript, aside from Marlys Hammond, our friend and
former colleague who passed away in October 2017.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors would like to thank all members of the BMP1
team at GlaxoSmithKline for their contributions. We thank
Jaclyn Patterson, Lina Ding, Tatiana Randriambalohery, and
Pierrick Ondet for synthesizing analogs. We thank Monique M.
Thompson, Enoch Gao, Fan Fan, Tina Shah, and Nathan Gaul
for determining potency data and Gary Smith for leading the
HTS. We also value the strategic contributions to the program
from Pu Qin, Alan Olzinski, Erding Hu, Christine Schnack-
enberg, Saul Needle, Mark Harpel, and Victoria Ballard.
ABBREVIATIONS
■
DNAUC, dose normalized area under the curve; SAR,
structure−activity relationship; MMP12, matrix metalloprotei-
nase-12; SQ, subcutaneous
REFERENCES
■
(1) Hopkins, D. R.; Keles, S.; Greenspan, D. S. The bone
morphogenetic protein 1/Tolloid-like metalloproteinases. Matrix
Biol. 2007, 26, 508−523.
(2) Vukicevic, S.; Grgurevic, L.; Dumic-Cule, I. Methods and
compositions for treating and diagnosing acute myocardial infarction.
US Patent Application Publication No. 2015/0104455 A1, 2015.
Figure 5. Structure of subcutaneous (SQ) BMP1 inhibitors.
D
DOI: 10.1021/acsmedchemlett.8b00173
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX

ACS Medicinal Chemistry Letters
Letter
(3) Cvjeticanin, B.; Prutki, M.; Dumic-Cule, I.; Veir, Z.; Grgurevic,
L.; Vukicevic, S. Possible target for preventing fibrotic scar formation
following acute myocardial infarction. Med. Hypotheses 2014, 83,
656−658.
(4) Turtle, E. D.; Ho, W.-B. Inhibition of procollagen C-proteinase:
fibrosis and beyond. Expert Opin. Ther. Pat. 2004, 14, 1185−1197.
(5) Fish, P. V.; Allan, G. A.; Bailey, S.; Blagg, Ju.; Butt, R.; Collis, M.
G.; Greiling, D.; James, K.; Kendall, J.; McElroy, A.; McCleverty, D.;
Reed, C.; Webster, R.; Whitlock, G. A. J. Med. Chem. 2007, 50, 3442−
3456.
(6) Dankwardt, S. M.; Abbot, S. C.; Broka, C. A.; Martin, R. L.;
Chan, C. S.; Springman, E. B.; Van Wart, H. E.; Walker, K. A. M.
Potent and selective nonpeptidic inhibitors of procollagen C-
proteinase. Bioorg. Med. Chem. Lett. 2002, 12, 1233−1235.
(7) Ho, W.-B. N-Substituted-arylsulfonylamino_hydroxamic acids
useful as inhibitors of C-proteinase and for treating or preventing
disorders related to unregulated collagen production. WO 2000/
050390, 2000.
(8) Grant, S.; Easley, C.; Kirkpatrick, P. Vorinostat. Nat. Rev. Drug
Discovery 2007, 6, 21−22.
(9) Vandenbroucke, R. E.; Libert, C. Is there new hope for
therapeutic matrix metalloproteinase inhibition? Nat. Rev. Drug
Discovery 2014, 13, 904−927.
(10) Aubart, K. M.; Christensen, S. B., IV; Duraiswami, C.;
Karpinski, J. M. Peptide Deformylase Inhibitors. WO2005/005456,
2005.
(11) Hung, C. W.; Koudelka, T.; Anastasi, C.; Becker, A.; Moali, C.;
Tholey, A. Characterization of post-translational modifications in full-
length human BMP-1 confirms the presence of a rare vicinal disulfide
linkage in the catalytic domain and highlights novel features of the
EGF domain. J. Proteomics 2016, 138, 136−145.
(12) Mac Sweeney, A.; Gil-Parrado, S.; Vinzenz, D.; Bernardi, A.;
Hein, A.; Bodendorf, U.; Erbel, P.; Logel, C.; Gerhartz, B. Structural
basis for the substrate specificity of bone morphogenetic protein 1/
tolloid-like metalloproteases. J. Mol. Biol. 2008, 384, 228−239.
(13) Dowdell, S. E.; Eidam, H. S.; Elban, M.; Fox, R. M.; Hammond,
M.; Hilfiker, M. A.; Hoang, T. H.; Kallander, L. S.; Lawhorn, B. G.;
Manns, S.; Philp, J.; Washburn, D. G.; Ye, G. Hydroxy Formamide
Derivatives and Their Use. WO 2015/104684, 2015.
(14) Data collected at Reaction Biology Corporation; details are in
the Supporting Information.
(15) All studies were conducted in accordance with the GSK Policy
on the Care, Welfare and Treatment of Laboratory Animals and were
reviewed the Institutional Animal Care and Use Committee either at
GSK or by the ethical review process at the institution where the work
was performed.
(16) De Savi, C.; Waterson, D.; Pape, A.; Lamont, S.; Hadley, E.;
Mills, M.; Page, K. M.; Bowyer, J.; Maciewicz, R. A. Hydantoin based
inhibitors of MMP13discovery of AZD6605. Bioorg. Med. Chem.
Lett. 2013, 23, 4705−4712.
(17) Turtle, E.; Chow, N.; Yang, C.; Sosa, S.; Bauer, U.; Brenner,
M.; Solow-Cordero, D.; Ho, W.-B. Design and synthesis of
procollagen C-proteinase inhibitors. Bioorg. Med. Chem. Lett. 2012,
22, 7397−7401.
E
DOI: 10.1021/acsmedchemlett.8b00173
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX