Total Synthesis of Aspergillomarasmine A and Related Compounds: A Sulfamidate Approach Enables Exploration of Structure–Activity Relationships

Article abstract of DOI:10.1002/anie.201606657

The fungal secondary metabolite aspergillomarasmine A (AMA) has recently been identified as an inhibitor of metallo-β-lactamases NDM-1 and VIM-2. Described herein is an efficient and practical route to AMA and its related compounds by a sulfamidate approa

Full text of DOI:10.1002/anie.201606657

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201606657
German Edition: DOI: 10.1002/ange.201606657
Inhibitors
Total Synthesis of Aspergillomarasmine A and Related Compounds:
A Sulfamidate Approach Enables Exploration of Structure–Activity
Relationships
Silvia A. Albu, Kalinka Koteva, Andrew M. King, Salma Al-Karmi, Gerard D. Wright,* and
Alfredo Capretta*
Abstract: The fungal secondary metabolite aspergillomaras-
mine A (AMA) has recently been identified as an inhibitor of
metallo-b-lactamases NDM-1 and VIM-2. Described herein is
an efficient and practical route to AMA and its related
compounds by a sulfamidate approach. In addition, a series
of derivatives has been prepared and tested for biological
activity in an effort to explore preliminary structure activity
relationships. While it was determined that natural LLL isomer
of AMA remains the most effective inactivator of NDM-
1 enzyme activity both in vitro and in cells, the structure is
highly tolerant of the changes in the stereochemistry at
positions 3, 6, and 9.
NDM-1 and other MBLs therefore pose a pressing public
health threat, thus making the discovery of an MBL inhibitor
a high priority.
Recently we identified the fungal natural product asper-
gillomarasmine A (AMA) as a potent inhibitor of two key
types of MBLs, NDM-1 and VIM-2.[5] AMA inactivates
MBLs by sequestration of a catalytic zinc ion and successfully
rescued the activity of meropenem in a mouse infected with
NDM-1 expressing Klebsiella pneumoniae.[5]
AMA can be retrosynthetically deconstructed into three
units: Asp and two aminopropionic acid (APA) moieties
(Figure 1). Each unit has one chiral center at carbon atoms 3,
A
ntibiotic resistance is a growing health crisis.[1] Especially
concerning is the resistance to the most commonly prescribed
class of antibiotics, the b-lactams. Resistance to b-lactam
antibiotics is mediated primarily by enzymes which hydro-
lytically inactivate the b-lactam warhead characteristic of the
drugs by one of two mechanisms: serine b-lactamases (SBLs)
cleave the lactam ring by nucleophilic attack of a serine
residue; and metallo-b-lactamases (MBLs) act by using
a zinc-stabilized hydroxy anion. Compounds which inactivate
serine-b-lactamases are approved as co-drugs and are admin-
istered together with b-lactam antibiotics,[2] yet inhibitors of
MBLs are not yet clinically available. The importance of
discovering such molecules is even more urgent since the
recent identification of New Delhi metallo-b-lactamase-
1 (NDM-1), which is now widespread across the globe.[3]
Organisms capable of harboring the associated ndm gene
include most species of Enterobacteriaceae, Acinetobacter,
and Pseudomonas. Typically, these strains are simultaneously
resistant to all antibiotics except tigecycline and colistin.
Alarmingly, colistin resistance in such multidrug resistant
pathogens has recently been reported.[4] Bacteria-producing
Figure 1. AMA and naturally occurring analogues. AMA comprises
three units: A: Asp, B: APA1, and C: APA2.
6, and 9. In the original 1965 description of the discovery of
AMA,[6] the streochemical assignment was L-Asp, D-APA1,
and D-APA2 (LDD configuration). We[7] and others[8] have
recently shown, by total synthesis, that the configuration of
natural AMA is actually LLL (SSS). In 1979, the structure of
toxin A (AMA lacking unit C, the N-terminal APA2;
Figure 1), was assigned LD stereochemistry (at positions 3
and 6, respectively).[9] Later, in 1991, this assignment was
corrected to LL through chemical synthesis and isotope
feeding experiments.[10] Replacement of the terminal APA2
moiety with glycine gives the natural product aspergillomar-
asmine B (AMB, Figure 1) whose stereochemical assignment
has yet to be confirmed by total synthesis, but is expected to
that of mirror AMA.
[*] Dr. S. A. Albu, S. Al-Karmi, Prof. A. Capretta
Department of Chemistry and Chemical Biology
McMaster University
1280 Main St W, Hamilton, ON (Canada)
E-mail: capretta@mcmaster.ca
Dr. K. Koteva, Dr. A. M. King, Prof. G. D. Wright, Prof. A. Capretta
Michael G. DeGroote Institute for Infectious Disease Research
McMaster University
1280 Main St W, Hamilton, ON (Canada)
E-mail: wrightge@mcmaster.ca
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under:
http://dx.doi.org/10.1002/anie.201606657.
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
This report describes an improved,
efficient, and practical route to AMA
and its related compounds, and utilizes
a sulfamidate approach. Furthermore,
using this strategy and our previously
published aziridine method,[7] a series
of derivatives has been prepared and
tested for biological activity in an
effort to explore preliminary struc-
ture–activity relationships (SAR)
along with gaining insight into the
mechanism of action of AMA against
MBLs.
Scheme 2. Synthetic route to toxin A and aspergillomarasmine A. Reagents: a) CH3CN/THF (1:1);
b) CF3SO3H, anisole, DCM, 1 h at 08C!RT; c) H2, Pd/C. Cbz=benzyloxycarbonyl, DCM=di-
AMA can be derived from the
coupling of aspartic acid and two chloromethane.
tert-butyl ester with SOCl2 followed by oxidation with
ruthenium(III) chloride and sodium periodate allowed prep-
aration of L-1d. Treatment of the sulfamidate with the di-tert-
butyl ester of l-aspartic acid gave LL-3, which could be either
completely deprotected using Leiꢀs method to give toxin A
(LL-4) or hydrogenated in the presence of a palladium
catalyst to give the intermediate LL-5. Reaction with addi-
tional L-1d allowed introduction of a second APA moiety to
yield LLL-6. Deprotection using CF3SO3H and anisole in
dichloromethane gave LLL-AMA. The spectroscopic and
physical properties of synthesized material and natural AMA
were identical.
The versatility of the approach was demonstrated in the
synthesis of aspergillomarasmine B (LL-AMB). Two routes
were followed and are shown in Scheme 3. In the first, the Cbz
group of L-1d was hydrogenated to give L-1e, then alkylated
with tert-butyl bromoacetate to give L-7. Treatment of the
sulfamidate with the di-tert-butyl ester of l-aspartic acid gave
LL-8, which was deprotected to give LL-AMB. Alternatively,
alkylation of LL-5 with tert-butyl bromoacetate also yielded
LL-8 and subsequently LL-AMB.
Scheme 1. Aspartic acid ring opening of sulfamidates. Boc=tert-
butoxycarbonyl, THF=tetrahydrofuran.
activated serine fragments. While our original synthesis of
AMA took advantage of an aziridine approach,[7] the present
work utilizes a sulfamidate[11] as the active serine moiety. In
our initial exploratory experiments, we prepared a series of
sulfamidates from either a protected d-serine (to give D-1a
and D-1b; Scheme 1) or l-serine (to give L-1c) using
a modification of the procedure described by Chandrasekaran
and co-workers.[11f] Treatment of sulfamidates with the
dimethyl ester of l-aspartic acid in acetonitrile allowed
smooth coupling and the isolation of LD-2a, LD-2b, and
LL-2c in good yields. In our hands, we found the reaction
tolerated a number of protecting groups. Selective and total
deprotection, however, proved to be cumbersome with the
standard reaction conditions for methyl and benzyl ester
hydrolysis, as well as removal
The sulfamidate route above offers a number of advan-
tages over the aziridine approach previously developed.
Issues surrounding the regioselectivity of aziridine ring
opening are well documented[11f,13] and while these factors
of the Boc or benzyl amino
protecting
groups,
and
resulted in poor yields of the
isolated products.
In their synthesis of
AMA, Lei and co-workers[8]
utilized a deprotection proto-
col involving CF3SO3H[12] and
anisole in dichloromethane
for the one-pot removal of
both Boc and Cbz groups.
With this mind, we developed
a
sulfamidate route to
toxin A and AMA
(Scheme 2). Treatment of
commercially available N-
(benzyloxycarbonyl)-l-serine methylamine.
Scheme 3. Synthetic route to aspergillomarasmine B. Reagents: a) CsCO3 then tert-butyl bromoacetate;
b) CH3CN/THF (1:1); c) TEA in EtOH/DMF (1:1); d) CF3SO3H, anisole, DCM, 1 h at 08C!RT. TEA=tri-
2
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
These are not the final page numbers!

Angewandte
Communications
Chemie
Table 1: Concentration response and FICI values for select AMA
played a very minor role in our original synthesis (with yields
compounds.[a]
of the unwanted regioisomer never greater than 5% and
easily separable by chromatography), the sulfamidate
approach completely avoids these. Furthermore, the stream-
lining of protecting groups and the reduction in the number of
overall steps allows a much higher yielding synthesis with an
overall yield of LLL-AMA of 19%, starting from the
protected serine.
A series of compounds related to AMA were synthesized
(by routes either described previously[7] or as presented in the
Supporting Information) and the biological activity of these
analogues was explored through dose-response assays using
purified, recombinant NDM-1 as well as cell-based assays to
assess in vivo synergy with meropenem in an NDM-1 express-
ing strain of E. coli (Table 1). We focused in particular on
units A and C (Figure 1) in this study as these are sites of
known natural variants (unit C) or are readily sampled using
our synthetic strategies.
Compound
Structure
IC50 [mm] RC [mm][b]
LLL-AMA
LDL-AMA
LLD-AMA
LDD-AMA
DLL-AMA
LL-AMB
9.3ꢀ0.3 12.5
12.5ꢀ0.2 50
11.9ꢀ1.5 50
10.2ꢀ0.4 50
11.6ꢀ0.1 25
18.3ꢀ1.3 50
As we reported previously,[7] altering the stereochemistry
at positions 6 (LDL-AMA, LDD-AMA) and 9 (LDD-AMA,
LLD-AMA) had minimal effect on in vitro NDM-1 inactiva-
tion and isomers retained synergy with meropenem, though
with reduced efficiency. The stereochemistry of the a-carbon
atom of Asp at position 3 (DLL-AMA) is also flexible, with
no significant influence on activity. Therefore there is
tolerance in the stereochemistry at positions 3, 6, and 9,
thus reflecting significant plasticity in biological activity.
NDM-1 IC50 curves for the known natural products LL-
AMB (AMA structure in which APA2 is replaced by Gly)
and lycomarasmine (AMA where APA2 is replaced by Gly-
amide, LL-26) exhibited complete enzyme inactivation,
although the latter had only marginal activity in cells
expressing NDM-1. The N-terminal amine of unit C is
therefore dispensable but retention of this group confers
optimal activity with pure enzyme and in cells. Protection of
the amine with the bulky nosyl group (LLL-14) significantly
weakens enzyme inhibition activity and abrogates in-cell
activity. Alternate spacing of the C-terminal carboxy group
(C2 of AMB vs. C3 AMA) is accommodated with retention of
activity, nevertheless, a free carboxy rather than an amide is
preferred for activity in the cell, perhaps reflecting access and
transport to the periplasm.
Complete removal of APA2 (toxin A; LL-4) has a signifi-
cant impact on both in vitro enzyme activity and synergy with
meropenem in cells. Unlike compounds with either a Gly or
APA equivalent unit in the APA2 position, toxin A showed
unusual behavior in enzyme inhibition studies. Rather than
precipitous dose dependence from full activity to inactivation,
toxin A IC50 curves showed a shallow inhibition gradient and
did not result in complete loss of enzyme action (see
Figure S90 in the Supporting Information), thus indicating
complex, and possibly nonspecific inhibition. Furthermore,
toxin A had no ability to rescue meropenem activity in cell
potentiation assays. The APA2 unit is therefore essential for
AMA activity, but is tolerant of some modification. Replace-
ment of the N-terminal Asp of toxin A with Glu (LL-17)
abolished both enzyme inhibition and in-cell activity.
We explored the role of the unit A Asp by change to Glu
(LLL-23). This analogue retained good in vitro enzyme
LL-4
(toxin A)
NA[c]
NA
NA
NT
NT
NT
NA
LL-10
LL-11
NA
LLL-12
LLL-13
LLL-14
LL-17
NA[c]
NA[c]
36.4ꢀ1.0 NA
NA[c]
NA[c]
NA
NT
LLL-21
LLL-22
LLL-23
17.0ꢀ0.2 NT
12.1ꢀ0.4 NA
13.5ꢀ0.6 100
LL-26
(lycomarasmine)
[a] IC50 measurements required <20% residual activity. [b] RC=Rescue
concentration is the concentration of compound needed to fully regain
the activity of meropenem at the clinical breakpoint (2.5 mm).
[c] Incomplete and non-ideal inhibition (see figure in Supporting
Information). NA=not active and indicates that either no inhibition or
synergy observed at given concentrations; NT=not tested (see
Supporting Information for full data sets); Ns=2-nitrobenzenesulfonyl.
inhibition activity, but lacked the ability to potentiate
meropenem in cells.
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
Protection of the carboxylates of toxin A (LL-10 and LL-
Keywords: antibiotics · enzymes · inhibitors · lactams ·
total synthesis
11), AMA (LLL-12, LLL-13), and the Glu analogue (LLL-
21, LLL-22) consistently resulted in loss of enzyme and in-cell
activities.
Taken together, these results demonstrate the importance
of free carboxylates on enzyme inactivation, and is consistent
with the proposed mode of action of AMA through seques-
tration of a catalytic Zn2+ ion.[5] The mode of action of AMA
is complex and not completely resolved, but is unlikely to be
simple nonspecific zinc chelation as evidenced by the
importance of the unit C APA2. This complexity is demon-
strated by the activity of toxin A (LL-4). This compound, and
derivatives, for example LL-17, wherein the unit A Asp is
changed to Glu lack comparable in vitro and in-cell activities
despite retaining a free carboxylate. In contrast, lycomara-
sime (LL-26), which also lacks a unit C carboxylate, does
retain in vitro enzyme inhibition activity, but has lost signifi-
cant in-cell activity. Furthermore, modification of the unit C
APA free amine (LL-14) diminishes enzyme inhibition
3.5 fold, yet all four carboxylates are available for metal
binding. These results argue for a more nuanced mechanism
of action of MBL inactivation by AMA rather than simple
zinc chelation.
[1] E. D. Brown, G. D. Wright, Nature 2016, 529, 336 – 343.
[2] K. M. Papp-Wallace, R. A. Bonomo, Infect. Dis. Clin. N. Am.
2016, 30, 441 – 464.
[3] D. Yong, M. A. Toleman, C. G. Giske, H. S. Cho, K. Sundman, K.
Lee, T. R. Walsh, Antimicrob. Agents Chemother. 2009, 53,
5046 – 5054.
[4] a) X. Yao, Y. Doi, L. Zeng, L. Lv, J.-H. Liu, Lancet Infect. Dis.
2016, 16, 288 – 289; b) H. Yu, F. Qu, B. Shan, B. Huang, W. Jia, C.
Chen, A. Li, M. Miao, X. Zhang, C. Bao, Y. Xu, K. D. Chavda,
Y.-W. Tang, B. N. Kreiswirth, H. Du, L. Chen, Antimicrob.
Agents Chemother. 2016, 0, 0.
[5] A. M. King, S. A. Reid-Yu, W. Wang, D. T. King, G. De Pascale,
N. C. Strynadka, T. R. Walsh, B. K. Coombes, G. D. Wright,
Nature 2014, 510, 503 – 506.
[6] A. L. Haenni, M. Robert, W. Vetter, L. Roux, M. Barbier, E.
Lederer, Helv. Chim. Acta 1965, 48, 729 – 750.
[7] K. Koteva, A. M. King, A. Capretta, G. D. Wright, Angew.
Chem. Int. Ed. 2016, 55, 2210 – 2212; Angew. Chem. 2016, 128,
2250 – 2252.
[8] D. Liao, S. Yang, J. Wang, J. Zhang, B. Hong, F. Wu, X. Lei,
Angew. Chem. Int. Ed. 2016, 55, 4291 – 4295; Angew. Chem.
2016, 128, 4363 – 4367.
[9] E. Bach, S. Christensen, L. Dalgaard, P. O. Larsen, C. E. Olsen,
V. Smedegꢁrd-Petersen, Physiol. Plant Pathol. 1979, 14, 41 – 46.
[10] P. Friis, C. E. Olsen, B. L. Møller, J. Biol. Chem. 1991, 266,
13329 – 13335.
Certainly, alteration of the AMA units A and C and
stereocenters 3, 6, and 9 can have substantial effects on in-cell
activity despite the ability to inhibit the enzyme in vitro. This
alteration could reflect transport issues or differences in
interaction with the outer membrane.
On the surface, the AMA structure is deceptively simple.
Composed of three distinct units, Asp-APA1-APA2 with
three chiral centers, it presents four carboxyl groups, two
secondary amines, and an N-terminal primary amine. The
natural LLL isomer of AMA remains the most effective
inactivator of NDM-1 enzyme activity both in vitro and in
cells. The structure nevertheless is highly tolerant of the
changes in the stereochemistry at positions 3, 6, and 9 though
with impact on activity of AMA in cells. Full activity requires
an Asp in unit A and a free carboxy group in the APA2
position, though APA2 itself can by truncated to Gly (AMB)
with no significant impact on activity. Modification of AMA
carboxy groups correlates with loss of activity, and is
consistent with the proposed mechanism of action of zinc
sequestration.
[11] a) J. E. Baldwin, A. C. Spivey, C. J. Schofield, Tetrahedron:
Asymmetry 1990, 1, 881 – 884; b) S. L. Cobb, J. C. Vederas,
Org. Biomol. Chem. 2007, 5, 1031 – 1038; c) S. B. Cohen, R. L.
Halcomb, J. Am. Chem. Soc. 2002, 124, 2534 – 2543; d) B. M.
Kim, S. M. So, Tetrahedron Lett. 1998, 39, 5381 – 5384; e) R. E.
Melꢂndez, W. D. Lubell, Tetrahedron 2003, 59, 2581 – 2616;
f) N. B. Rashid Baig, R. N. Chandrakala, V. S. Sudhir, S. Chan-
drasekaran, J. Org. Chem. 2010, 75, 2910 – 2921; g) A. Saha, R. B.
Nasir Baig, J. Leazer, R. S. Varma, Chem. Commun. 2012, 48,
8889 – 8891; h) V. Sai Sudhir, N. Y. Phani Kumar, S. Chandrase-
karan, Tetrahedron 2010, 66, 1327 – 1334; i) L. T. Boulton, H.
Thijs Stock, J. Raphy, D. C. Horwell, J. Chem. Soc. Perkin Trans.
1 1999, 1421 – 1430; j) C. M. Taylor, S. T. De Silva, J. Org. Chem.
2011, 76, 5703 – 5708; k) L. Wei, W. D. Lubell, Can. J. Chem.
2001, 79, 94 – 104.
[12] H. Yajima, N. Fujii, H. Ogawa, H. Kawatani, Chem. Commun.
1974, 107 – 108.
[13] K. OꢀBrien, K. ꢃ. Proinsias, F. Kelleher, Tetrahedron 2014, 70,
5082 – 5092.
Acknowledgments
This research was funded by a Collaborative Health Research
Project Grant (446433-13), a Canadian Institutes of Health
Research Grant (MT-13536), and by a Canada Research
Chair in Antibiotic Biochemistry (to G.D.W.).
Received: July 9, 2016
Revised: August 13, 2016
Published online: && &&, &&&&
4
www.angewandte.org
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Inhibitors
S. A. Albu, K. Koteva, A. M. King,
S. Al-Karmi, G. D. Wright,*
A. Capretta*
&&&& — &&&&
The total synthesis of aspergillomaras-
mine A and related natural products
using a sulfamidate approach is de-
scribed, and preliminary structure–activ-
ity relationships are explored. While the
natural LLL isomer of AMA remains the
most effective inactivator of NDM-
Total Synthesis of Aspergillomarasmine A
and Related Compounds: A Sulfamidate
Approach Enables Exploration of
1 enzyme activity both in vitro and in
cells, the structure is highly tolerant of
changes in the stereochemistry at posi-
tions 3, 6, and 9. Cbz=benzyloxycar-
bonyl.
Structure–Activity Relationships
Angew. Chem. Int. Ed. 2016, 55, 1 – 5
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5
These are not the final page numbers!