Structural and initial biological analysis of synthetic arylomycin A 2

Article abstract of DOI:10.1021/ja073340u

The growing threat of untreatable bacterial infections has refocused efforts to identify new antibiotics, especially those acting by novel mechanisms. While the inhibition of pathogen proteases has proven to be a successful strategy for drug development,

Full text of DOI:10.1021/ja073340u

Published on Web 12/01/2007
Structural and Initial Biological Analysis of Synthetic
Arylomycin A2
Tucker C. Roberts,† Peter A. Smith,† Ryan T. Cirz,‡ and Floyd E. Romesberg*,†
Contribution from the Department of Chemistry, The Scripps Research Institute, 10550 N.
Torrey Pines Road, La Jolla, California, 92037, and Achaogen, Inc., 7000 Shoreline Court,
Suite 371, South San Francisco, California 94080
Received May 11, 2007; E-mail: floyd@scripps.edu
Abstract: The growing threat of untreatable bacterial infections has refocused efforts to identify new
antibiotics, especially those acting by novel mechanisms. While the inhibition of pathogen proteases has
proven to be a successful strategy for drug development, such inhibitors are often limited by toxicity due
to their promiscuous inhibition of homologous and mechanistically related human enzymes. Unlike many
protease inhibitors, inhibitors of the essential type I bacterial signal peptidase (SPase) may be more specific
and thus less toxic due to the enzyme’s unique structure and catalytic mechanism. Recently, the arylomycins
and related lipoglycopeptide natural products were isolated and shown to inhibit SPase. The core structure
of the arylomycins and lipoglycopeptides consists of a biaryl-linked, N-methylated peptide macrocycle
attached to a lipopeptide tail, and in the case of the lipoglycopeptides, a deoxymannose moiety. Herein,
we report the first total synthesis of a member of this group of antibiotics, arylomycin A2. The synthesis
relies on Suzuki-Miyaura-mediated biaryl coupling, which model studies suggested would be more efficient
than a lactamization-based route. Biological studies demonstrate that these compounds are promising
antibiotics, especially against Gram-positive pathogens, with activity against S. epidermidis that equals
that of the currently prescribed antibiotics. Structural and biological studies suggest that both N-methylation
and lipidation may contribute to antibiotic activity, whereas glycosylation appears to be generally less critical.
Thus, these studies help identify the determinants of the biological activity of arylomycin A2 and should aid
in the design of analogs to further explore and develop this novel class of antibiotic.
1. Introduction
However, SPase utilizes a unique Ser-Lys catalytic dyad and
appears to act via an unusual mechanism involving nucleophilic
attack on its substrate from the si-face as opposed to the re-
face attack characteristic of the more common Ser-His-Asp
catalytic triad serine proteases.4 Consistent with its unconven-
tional mechanism, SPase is not inhibited by standard serine
protease inhibitors.5 These observations suggest that the arylo-
mycins or related derivatives might be developed into a novel
class of antibiotics.
SPase is an essential enzyme because it is needed to process
proteins that are required for bacterial viability. However, SPase
is also required to process proteins involved in a variety of other
processes such as antibiotic resistance and virulence. For
example, secretion of the â-lactamases that confer resistance
to â-lactam antibiotics require SPase, and an arylomycin-like
lipoglycopeptide (see below) has been shown to inhibit the
secretion of these proteins in Staphylococcus aureus.6 In
addition, many components of bacterial secretion systems
involved in toxin delivery and adhesion require processing by
The emergence of bacteria that are resistant to all available
antibiotics threatens to end the ‘antibiotic era’ and has rein-
vigorated efforts to identify new antibiotics, especially those
acting by novel mechanisms. In 2002, a novel class of natural
products, the arylomycins, was isolated from Streptomyces strain
Tu 6075 and shown to inhibit the growth of several Gram-
positive bacteria.1 Subsequent studies demonstrated that the
antibiotic activity of the arylomycins results from their inhibition
of the bacterial type I signal peptidase (SPase), an essential
serine protease required to process cell surface bound pre-
proteins.2 Although protease inhibitors have been extensively
pursued as antimicrobials, they have been challenging to develop
as drugs due to low discrimination against human proteases.3
† The Scripps Research Institute.
‡ Achaogen, Inc.
(1) (a) Holtzel, A.; Schmid, D. G.; Nicholson, G. J.; Stevanovic, S.; Schimana,
J.; Gebhardt, K.; Fiedler, H. P.; Jung, G. J. Antibiot. (Tokyo) 2002, 55,
571-577. (b) Schimana, J.; Gebhardt, K.; Holtzel, A.; Schmid, D. G.;
Sussmuth, R.; Muller, J.; Pukall, R.; Fiedler, H. P. J. Antibiot. (Tokyo)
2002, 55, 565-570.
(4) Tschantz, W. R.; Sung, M.; Delgado-Partin, V. M.; Dalbey, R. E. J. Biol.
Chem. 1993, 268, 27349-27354.
(2) (a) Carlos, J. L.; Paetzel, M.; Klenotic, P. A.; Strynadka, N. C. J.; Dalbey,
R. E. The Enzymes, 3rd ed. 2001, 22, 27-55. (b) Paetzel, M.; Dalbey, R.
E.; Strynadka, N. C. Pharmacol. Ther. 2000, 87, 27-49. (c) Paetzel, M.;
Karla, A.; Strynadka, N. C. J.; Dalbey, R. E. Chem. ReV. 2002, 102, 4549-
4580.
(3) (a) Fear, G.; Komarnytsky, S.; Raskin, I. Pharmacol. Ther. 2007, 113, 354-
368. (b) Leung, D.; Abbenante, G.; Fairlie, D. P. J. Med. Chem. 2000, 43,
305-341.
(5) (a) Black, M. T.; Munn, J. G. R.; Allsop, A. E. Biochem. J. 1992, 282,
539-543. (b) Kim, Y. T.; Muramatsu, T.; Takahashi, K. J. Biochem. 1995,
117, 535-544. (c) Kuo, D. W.; Chan, H. K.; Wilson, C. J.; Griffin, P. R.;
Williams, H.; Knight, W. B. Arch. Biochem. Biophys. 1993, 303, 274-
280. (d) Zwizinski, C.; Date, T.; Wickner, W. J. Biol. Chem. 1981, 256,
3593-3597.
(6) Kulanthaivel, P., et al. J. Biol. Chem. 2004, 279, 36250-36258.
9
15830
J. AM. CHEM. SOC. 2007, 129, 15830-15838
10.1021/ja073340u CCC: $37.00 © 2007 American Chemical Society

Synthesis and Analysis of Arylomycin A2
A R T I C L E S
macrocycles have been the focus of much recent effort due to
their diverse and remarkable biological activities.10 A key aspect
of these syntheses is macrocyclization, and multiple strategies
have been developed. For example, Evans et al.11 synthesized
the biaryl-linked AB ring macrocycle of vancomycin via biaryl
vanadium coupling, and both Nicolaou et al.12 and Boger et
al.13 assembled the same ring system by intermolecular Suzuki
cross-coupling followed by intramolecular macrolactamization.
Similarly, biphenomycin14 and the TMC-9515 family of mac-
rocycles were synthesized by first forming the biaryl-coupled
intermediate via Negishi,16 Stille,17 or Suzuki18 coupling reac-
tions, followed by derivatization, and finally ring closure by
macrolactamization. Derivatives of biphenomycin and TMC-
95A have also been synthesized by first assembling the
derivatized tripeptide and then cyclizing by intramolecular
Suzuki-Miyaura cross-coupling.19
Herein, we report the first total synthesis of an arylomycin
natural product, arylomycin A2. The synthesis employed an
intramolecular biaryl-bond forming macrocyclization for cyclic
tripeptide ring closure, which model studies predicted would
be more efficient than a macrolactamization-based route.
Atropisomerism of arylomycin A2 was observed and investigated
through the synthesis of derivatized core macrocycles. Deriva-
tives were also synthesized to address the contribution of
N-methylation, glycosylation, and fatty acid tail length to
antibiotic activity. Our data suggests that both N-methylation
and lipidation make important contributions to antibiotic activity;
however, the contributions are somewhat different for different
bacteria. Perhaps most notably, we report that this class of
natural product has very potent antibiotic activity against the
important human pathogen S. epidermidis. Synthetic access to
the arylomycins and related compounds should allow for further
study of their unique mechanism of action and for their
development as antibiotics.
Figure 1. Arylomycin A2 and lipoglycopeptide natural products.
SPase.7 Thus, the arylomycin family of natural products not
only have interesting antibacterial properties, but they may also
have the ability to modulate virulence and sensitize bacteria to
other antibiotics.8
There are two related series of arylomycins, the arylomycins
A and B. The core structure of both series of compounds consists
of a tripeptide macrocycle with a C-C biaryl linkage between
a hydroxyphenylglycine (MeHpg5) residue and a tyrosine (Tyr7)
residue (Figure 1). Attached to the core macrocycle is a
lipopeptide tail comprised of a C11-C15 fatty acid linked to the
N-terminal tripeptide D-MeSer2-D-Ala3-Gly4. In addition, the
peptide backbone is N-methylated at the second and fifth
residues (MeSer2 and MeHpg5). The two series of arylomycins
are differentiated by Tyr7 nitration in the B series.
The crystal structure of the Escherichia coli SPase-arylo-
mycin A2 complex has been reported.9 The structure reveals
that the C-terminal carboxylate of the natural product forms a
hydrogen-bond with both residues of the catalytic Ser-Lys dyad,
while the remainder of the peptide forms an extensive series of
hydrogen-bonding and packing interactions with the enzyme.
Sterically hindered biaryl-compounds, like the arylomycins, may
exhibit two rotational isomers, or atproisomers, due to slow
rotation about the interannular C-C bond. However, the crystal
structure suggests that E. coli SPase binds only the Sa atropi-
somer. This binding selectivity appears to result from packing
interactions between the protein side chains of Pro87 and Leu42
and the aryl rings of MeHpg5 and Tyr7.
Independently, in 2004 a group at Eli Lilly screened a library
of Streptomyces sp. natural products for SPase inhibitors and
identified a number of related lipoglycopeptides.6 Interestingly,
these lipoglycopeptides have the same core structural features
as the arylomycins (Figure 1), but their macrocycles are
modified by glycosylation and their conjugated fatty acid tails
are generally longer, by up to five carbons. Although lipidation,
glycosylation, and N-methylation are all common modifications
of nonribosomally synthesized peptides, it is unclear how these
modifications contribute to biological activity, in general, and
to the antibacterial properties of the arylomycins and related
lipoglycopeptides, in particular.
(10) Feliu, L.; Planas, M. Int. J. Pept. Res. Ther. 2005, 11, 53-97.
(11) (a) Evans, D. A.; Dinsmore, C. J. Tetrahedron Lett. 1993, 34, 6029-6032.
(b) Evans, D. A.; Dinsmore, C. J.; Evrard, D. A.; DeVries, K. M. J. Am.
Chem. Soc. 1993, 15, 6426-6427. (c) Evans, D. A.; Wood, M. R.; Trotter,
B. W.; Richardson, T. I.; Barrow, J. C.; Katz, J. L. Angew. Chem., Int. Ed.
1998, 37, 2700-2704.
(12) (a) Nicolaou, K. C.; Ramanjulu, J. M.; Natarajan, S.; Bra¨se, S.; Ru¨bsam,
F. Chem. Commun. 1997, 19, 1899-1900. (b) Nicolaou, K. C.; Jain, N.
F.; Natarajan, S.; Hughes, R.; Solomon, M. E.; Li, H.; Ramanjulu, J. M.;
Takayanagi, M.; Koumbis, A. E.; Bando, T. Angew. Chem., Int. Ed. 1998,
37, 2714-2716. (c) Nicolaou, K. C.; Natarajan, S.; Li, H.; Jain, N. F.;
Hughes, R.; Solomon, M. E.; Ramanjulu, J. M.; Boddy, C. N. C.;
Takayanagi, M. Angew. Chem., Int. Ed. 1998, 37, 2708-2714.
(13) (a) Boger, D. L.; Borzilleri, R. M.; Nukui, S. J. Org. Chem. 1996, 61,
3561-3565. (b) Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.;
Loiseleur, O.; Castle, S. L. J. Am. Chem. Soc 1999, 121, 3226-3227. (c)
Boger, D. L.; Miyazaki, S.; Kim, S. H.; Wu, J. H.; Castle, S. L.; Loiseleur,
O.; Jin, Q. J. Am. Chem. Soc. 1999, 121, 10004-10011.
(14) (a) Ezaki, M.; Iwami, M.; Yamashita, M.; Hashimoto, S.; Komori, T.;
Umehara, K.; Mine, Y.; Kohsaka, M.; Aoki, H.; Imanaka, H. J. Antibiot.
1985, 38, 1453-1461. (b) Uchida, I.; Shigematsu, N.; Ezaki, M.; Hash-
imoto, M.; Aoki, H.; Imanaka, H. J. Antibiot. 1985, 38, 1462-1468.
(15) (a) Koguchi, Y.; Kohno, J.; Nishio, M.; Takahashi, K.; Okuda, T.; Ohnuki,
T.; Komatsubara, S. J. Antibiot. 2000, 53, 105-109. (b) Kohno, J.; Koguchi,
Y.; Niskio, M.; Nakao, K.; Kuroda, M.; Shimizu, R.; Ohnuki, T.;
Komatsubara, S. J. Org. Chem. 2000, 65, 990-995.
(16) Schmidt, U.; Meyer, R.; Leitenberger, V.; Lieberknecht, A.; Griesser, H.
J. Chem. Soc. Chem. Comm. 1991, 275-277.
(17) Albrecht, B. K.; Williams, R. M. Proc. Natl. Acad. Sci. U.S.A. 2004, 101,
11949-11954.
Both structure-activity relationship studies and the charac-
terization of the arylomycins and related lipoglycopeptides as
antibiotics or inhibitors of virulence would be greatly facilitated
by their total synthesis. The syntheses of similar biaryl-bridged
(18) (a) Inoue, M.; Sakazaki, H.; Furuyama, H.; Hirama, M. Angew. Chem.,
Int. Ed. 2003, 42, 2654-2657. (b) Kaiser, M.; Groll, M.; Renner, C.; Huber,
R.; Moroder, L. Angew. Chem., Int. Ed. 2002, 41, 780-783. (c) Lin, S.;
Danishefsky, S. J. Angew. Chem., Int. Ed. 2002, 41, 512-515. (d) Lin, S.;
Yang, Z. Q.; Kwok, B. H.; Koldobskiy, M.; Crews, C. M.; Danishefsky,
S. J. J. Am. Chem. Soc. 2004, 126, 6347-55.
(19) (a) Carbonnelle, A. C.; Zhu, J. Org. Lett. 2000, 2, 3477-3480. (b) Kaiser,
M.; Siciliano, C.; Assfalg-Machleidt, I.; Groll, M.; Milbradt, A. G.;
Moroder, L. Org. Lett. 2003, 5, 3435-3437. (c) Lepine, R.; Zhu, J. Org.
Lett. 2005, 7, 2981-2984.
(7) Cao, T. B.; Saier, M. H., Jr. Microbiol. 2001, 147, 3201-3214.
(8) Smith, P. A.; Romesberg, F. E. Nat. Chem. Biol. 2007, 3, 549-556.
(9) Paetzel, M.; Goodall, J. J.; Kania, M.; Dalbey, R. E.; Page, M. G. J. Biol.
Chem. 2004, 279, 30781-30790.
9
J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007 15831

A R T I C L E S
Roberts et al.
Scheme 1. Retrosynthesis of Arylomycin A2
Scheme 2. Iododipeptide Synthesisa
a Reagents and conditions: (a) Boc2O (1.0 equiv), NaHCO3 (1.5 equiv),
acetone:H2O (1:1), overnight; (b) Ala-OMe or Ala-OBn (1.0 equiv), EDC
(1.5 equiv), HOBt (1.0 equiv), TEA (1.3 equiv), DMF, overnight, 71% and
68% (2 steps) for 6 and 7, respectively; (c) MeI (10.0 equiv), K2CO3 (5.0
equiv), acetone, reflux, 17 h; (d) I2 (1.05 equiv), AgSO4 (1.05 equiv), MeOH,
30 min, 76% (2 steps).
Scheme 3. Tyrosine Boronic Ester Synthesisa
2. Results
a Reagents and conditions: (a) CbzCl (1.0 equiv), Na2CO3 (2.0 equiv),
acetone:H2O (1:1), overnight, 84%; (b) DMS (1.0 equiv), K2CO3 (2.0 equiv),
acetone, 2 h, 94%; (c) I2 (2.0 equiv), AgSO4 (2.0 equiv), MeOH, 2 h, 77%;
(d) bis(pinicolato)diboron (2.0 equiv), KOAc (12.0 equiv), PdCl2(dppf) (0.2
equiv), DMSO, overnight, 81%; (e) H2 (1 atm), 10% Pd/C, 95% EtOH.
For (a)-(d) see references 18b and 20a for experimental details.
Retrosynthetic analysis of arylomycin A2 suggested a logical
first disconnection between the macrocycle 1 and the lipopeptide
tail 2 (Scheme 1). Lipopeptide tail synthesis using peptide and
fatty acid couplings was expected to be straightforward, leaving
macrocyclization as the key step in the synthesis. As discussed
above, literature precedent suggested that the macrocycle might
be successfully assembled by intermolecular biaryl coupling,
followed by intramolecular macrolactamization, or by Suzuki-
Miyaura macrocyclization of an assembled tripeptide. We first
evaluated the potential of these approaches using model
substrates, 3 and 4.
2.1. Model Studies. Testing the macrolactamization and the
Suzuki-Miyaura macrocyclization routes required similar pre-
cursor substrates. The first precursors, phenylhydroxyglycine-
alanine iododipeptides 10 and 11, were synthesized as shown
in Scheme 2. The synthesis began with Boc protection of
commercially available 4-hydroxyphenylglycine 5. The pro-
tected hydroxyphenylglycine was then coupled to either Ala-
OMe or Ala-OBn with EDC/HOBt, yielding protected dipeptides
6 and 7. The phenols were protected as methyl ethers with MeI
and K2CO3 in acetone and selectively iodinated using Sy’s
conditions (I2/AgSO4).20 Iododipeptides 10 and 11 were obtained
in 54% and 42% yield, respectively, over four steps.
The pinacol boronate fragment 15 was synthesized following
the protocols of Danishefsky18d and Hutton,21 as shown in
Scheme 3. The commercially available tyrosine methyl ester
12 was Cbz protected and then methylated using dimethyl
sulfate, yielding 13. I2/AgSO4 was used to mono-iodonate the
aromatic ring ortho to the methoxy group, and the resulting
iodo tyrosine was converted to boronic ester 15 using Miyaura
boration conditions.22
2.1.1. Ring Closure Via Macrolactamization. The macrolac-
tamization precursor was assembled by coupling boronic ester
15 with iododipeptide 11 (Scheme 4). Exploratory Suzuki-
Miyaura coupling studies using PdCl2(dppf) as a catalyst showed
moderate yields of compound 17 (∼30 to 40%). Compound 17
proved difficult to characterize by NMR spectroscopy, as the
1H spectrum exhibited severely broadened and/or apparently
doubled resonances. This likely results from interconverting
atropisomers, which has been observed with similar biaryl
(21) Hunter, L.; Hutton, C. A. Aust. J. Chem. 2003, 56, 1095-1098.
(22) Ishiyama, T.; Murata, M.; Miyaura, N. J. Org. Chem. 1995, 60, 7508-
7510.
(20) (a) Sy, W. W. Tetrahedron Lett. 1993, 34, 6223-6224. (b) Sy, W. W.;
Lodge, B. A.; By, A. W. Syn. Comm. 1990, 20, 877-880.
9
15832 J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007

Synthesis and Analysis of Arylomycin A2
A R T I C L E S
Scheme 4. Macrolactamizationa
Scheme 5. Suzuki-Miyaura Macrocyclizationa
a Reagents and conditions: (a) 0.2 N LiOH (1.7 equiv), THF, 2.5 h; (b)
19 (1.0 equiv), 16 (1.1 equiv), HOBt (2.5 equiv), EDC (2.2 equiv), NaHCO3
(cat), AcCN:DMF (2.2:1), overnight. Atropisomer assumed for 20.
a Reagents and conditions: (a) 11 (1.0 equiv), 15 (1.2 equiv), PdCl2(dppf)
(0.2 equiv), K2CO3 (5.0 equiv), DMSO, 36 h, 36%; (b) H2 (1 atm), 10%
Pd/C (1/3 w/w), 95% MeOH, 3 h, quant.; (c) 3 (0.0005 M), EDC (4 equiv),
HOBt (4 equiv), DMF, 48 h, <10%. Atropisomer assumed for 3, 17, and
18.
Table 1. Cyclization Optimizationa
systems.23 Consistent with this conclusion, the reaction product
was a single spot by thin layer chromatography and the ESI-
MS determined mass matched precisely that predicted for the
coupled product 17. Simultaneous deprotection of the Cbz and
Bn groups with H2/Pd/C yielded the macrolactamization sub-
strate 3.
entry
Pd
ligand
solvent
base
yield of 20
With compound 3 in hand, we examined a variety of
macrolactamization conditions to produce macrocycle 18,
including HATU/DMAP, EDC/HOBt, FTAH/DIEA, and DPPA/
DIEA. EDC/HOBt proved optimal; however, the reaction
proceeded in low yield (∼25%) and produced a mixture of
compounds, with no single component present in greater than
10% yield. In addition, optimal yields of the product mixture
required low substrate concentrations (0.5 mM to 1 mM); at
higher concentrations (5 mM) only cyclic multimers were
observed. The low yields and prohibitively small scale on which
the reaction must be run render the macrolactamization route
impractical for arylomycin synthesis.
2.1.2. Ring Closure Via Intramolecular Suzuki-Miyaura
Coupling. Construction of model compound 4 to evaluate the
alternative intramolecular Suzuki-Miyaura coupling-based route
to 18 began with iododipeptide 10 (Scheme 5). Deprotection
of 10 with LiOH in THF/H2O yielded the free acid iododipeptide
19. This compound was then coupled with boronic ester 16,
obtained from H2/Pd/C reduction of 15, to yield the desired
cyclization substrate, tripeptide 4.
With tripeptide 4 in hand, we proceeded to test conditions
for the intramolecular Suzuki-Miyaura cross-coupling reaction.
Preliminary macrocyclization coupling experiments gave higher
yields than those observed for the macrolactamization route.
During optimization of reaction conditions (Table 1), we found
that solvent had the most significant effect on reaction efficiency,
with a 49% yield of the desired macrocycle 20 (2 steps) obtained
when the reaction was performed in acetonitrile. In contrast to
the macrolactamization route, the yield was independent of
1
2
3
4
5
6
7
8
9
Pd(II)Cl2
Pd(II)Cl2
Pd(II)Cl2
Pd(II)Cl2
Pd(II)Cl2
Pd(II)(OAc)2
Pd(II)Cl2
Pd(II)Cl2
Pd2(dba)3
Pd(II)(OAc)2
Pd2(dba)3
dppf
dppf
dppf
dppf
dppf
dppf
Sphos
bis-tpp
bis-tpp
PCy3
dppf
30:1 Tol:H2O
MeCN
DME
DMSO
MeCN
MeCN
MeCN
MeCN
MeCN
K2CO3
K2CO3
K2CO3
K2CO3
K3PO4
K2CO3
K2CO3
K2CO3
K2CO3
K2CO3
K2CO3
34%
49%
29%
10%
48%
44%
37%
49%
46%
34%
38%
10
11
MeCN
30:1 Tol:H2O
a All cyclization reactions were treated with 10 equiv of K2CO3, and
heated to 80 °C for 18-22 h under Ar. After workup, all reactions
were treated with a 4:1 mixture of CH2Cl2:TFA for approx 1.5 h and
purified. Yields are isolated yields. Abbreviations: dppf, 1,1′ bis(diphe-
nylphosphino)ferrocene; Sphos, 2-dicyclohexylphosphino-2′,6′-dimethoxy-
biphenyl; PCy3, tricyclohexylphosphine; bis-tpp, bis(triphenylphosphine);
dba, dibenzylideneacetone; Tol, toluene; MeCN, acetonitrile; DME, dime-
thoxyethane.
substrate concentration between 1 and 20 mM. These results
suggested that macrocyclization via Suzuki-Miyaura coupling
would provide a viable route to the arylomycin class of natural
products.
2.2. Synthesis of Arylomycin A2. Synthesis of the natural
product commenced with monomethylation of the amine
terminus of macrocycle 20. Monomethylation of 20 using
cesium hydroxide24 and iodomethane was unsuccessful, yielding
only overmethylated macrocycle. Monomethylation of 18 with
iodomethane and K2CO3 was also unsuccessful, yielding only
starting material. Attempts to use the formyl imine of the Boc-
deprotected macrocycle as a substrate for a Diels-Alder reaction
with cyclopentadiene (which would then be followed by a
tandem retro Diels-Alder imine reduction25) yielded no product.
(23) (a) Bois-Choussy, M.; Cristau, P.; Zhu, J. Angew. Chem., Int. Ed. 2003,
42, 4238-4241. (b) Deng, H.; Jung, J. K.; Liu, T.; Kuntz, K. W.; Snapper,
M. L.; Hoveyda, A. H. J. Am. Chem. Soc. 2003, 125, 9032-9034. (c) Lloyd-
Williams, P.; Giralt, E. Chem. Soc. ReV. 2001, 30, 145-157. (d) Shinohara,
T.; Deng, H.; Snapper, M. L.; Hoveyda, A. H. J. Am. Chem. Soc. 2005,
127, 7334-7336.
(24) Salvatore, R. N.; Nagle, A. S.; Jung, K. W. J. Org. Chem. 2002, 67, 674-
683.
9
J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007 15833

A R T I C L E S
Roberts et al.
Scheme 6. Formation of Methylated Macrocyclea
Scheme 8. Final Assembly of Macrocycle and Lipopeptide Taila
a Reagents and conditions: (a) NsCl (1.5 equiv), TEA (3 equiv), AcCN,
2 h; (b) MeI (10 equiv), K2CO3 (10 equiv), acetone, overnight, 37% (2
steps); (c) 2-MAA (3 equiv), DBU (5 equiv), AcCN, 30 min; (d) Boc-Gly-
OH (1.1 equiv), HOBt (3.3 equiv), EDC (3 equiv), CH2Cl2:DMF (3:1),
overnight, 74% (2 steps); (e) TFA: CH2Cl2 (4:1), 2 h, 96%. Atropisomer
assumed for 20, 21, and 22.
Scheme 7. Lipopeptide Tail Constructiona
a Reagents and conditions: (a) 1 (1.0 equiv), 2 (1.1 equiv), HOBt (3.3
equiv), EDC (3.0 equiv), NaHCO3 (cat), AcCN:DMF (2.2:1), 5 h, 63%;
(b) AlBr3 (25 equiv), EtSH, 4 h, 52%.
allowed for a one-pot coupling of 28 to isolauric acid chloride
(the use of 2-mercaptoacetic acid facilitates removal of the
reaction byproducts26). Methyl ester deprotection with lithium
hydroxide afforded the lipopeptide tail 2.
Compound 1 was coupled to lipopeptide 2 via amide
formation, yielding the protected arylomycin A2, 29, in 63%
yield (Scheme 8). Global deprotection with AlBr3/EtSH13c at
50 °C afforded arylomycin A2 in 52% yield. The authenticity
of the natural product was confirmed by comparison with the
NMR spectra and HPLC retention time of the authentic natural
product (kindly provided by Prof. Hans-Peter Fiedler, University
of Tubingen).
2.3. Atropisomerism. Restricted rotation, or atropisomerism,
about the interannular bond of biaryl systems commonly results
in conformational heterogeneity on the NMR time scale. As
previously reported,1 arylomycin A2 shows two sets of doubled
1H NMR resonances in DMSO. One doubling of the resonances
was shown to arise from cis-trans isomerism within the
lipopeptide tail.1 Interestingly, we found that several simple
macrocycle intermediates synthesized en route to the natural
product show doubled resonances, as well, suggesting that the
second resonance doubling in DMSO is due to atropisomerism
about the macrocycle biaryl bond. To examine this potential
atropisomerism, we synthesized several derivatives of the
glycine homologated core macrocycle (Figure 2). These model
macrocycles were chosen to focus on macrocycle dynamics in
the absence of the conformational heterogeneity associated with
the lipopeptide tail.
a Reagents and conditions: (a) NsCl (1.05 equiv), 1 N NaOH:THF (10:
1), overnight, 62%; (b) D-Ala-OMe (1.0 equiv), HOBt (3.3 equiv), EDC (3
equiv), NaHCO3 (1 equiv), CH2Cl2:DMF (3:1), overnight, 88%; (c) CH2N2
(12 equiv), CH2Cl2, 5 min, 90%; (d) 2-MAA (3 equiv), DBU (5 equiv),
AcCN, 15 min; (e) C12 (for 2) or C16 (for 33) acid chloride (1 equiv), 9%
NaHCO3, CH2Cl2, 5 h; (f) 0.2 N LiOH (1.15 equiv), THF, 3 h, 19% and
43% for 2 and 33, respectively (3 steps).
Attempted methylation of the 4-nitrobenzenesulfonyl (nosyl)-
protected macrocycle 21 with diazomethane also yielded only
starting material. However, treating compound 21 with 10 equiv
of both iodomethane and K2CO3 in refluxing acetone yielded
the desired monomethylated product 22 in 37% yield (2 steps)
after column chromatography (Scheme 6). Nosyl deprotection
of 22 with 2-mercaptoacetic acid and DBU gave the monom-
ethylated macrocycle, which was coupled to Boc-protected
glycine with EDC/HOBt, yielding compound 23 in 74%
yield (2 steps). Boc deprotection of macrocycle 23 yielded
compound 1.
The N-terminal peptide residues and the fatty acid chain were
assembled in the absence of the macrocycle due to the low
solubility of the full peptide in organic solvent (Scheme 7).
Benzyl ether protected D-serine 24 was first nosylated under
aqueous conditions, yielding compound 25. Coupling of 25 to
D-Ala-OMe using EDC/HOBt yielded 26, and subsequent
sulfonamide methylation with diazomethane yielded compound
27. Nosyl deprotection using 2-mercaptoacetic acid and DBU
We first examined the protected macrocycle analog 23. The
1D 1H spectrum of compound 23 showed two sets of resonances
(26) Di Gioia, M. L.; Leggio, A.; Le Pera, A.; Liguori, A.; Napoli, A.; Siciliano,
(25) Grieco, P. A.; Bahsas, A. J. Org. Chem. 1987, 52, 5746-5749.
C.; Sindona, G. J. Org. Chem. 2003, 68, 7416-7421.
9
15834 J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007

Synthesis and Analysis of Arylomycin A2
A R T I C L E S
Table 2. MICs (µM) of Selected Compounds
strain
arylomycin A2
compd 37
compd 39
E. coli MG1655
E. coli MG1655 w/PMBn
S. aureus 8325
>128
128
>128
1
n.d.
n.d.
n.d.
n.d.
>128
16
>128
0.5
>128
>128
>128
1
n.d.
n.d.
S. epidermidis ATCC 35984
B. anthracis Sterne
32
E. faecium AEFA001a
E. faecalis ATCC 29212
E. faecalis ATCC 51299
>64
>64
>64
n.d.
n.d.
Figure 2. Model macrocycles synthesized to evaluate atropisomerism.
a Part of the Achaogen, Inc. strain collection.
H39a and H39b show similar and relatively strong NOEs with
H45, but either no NOE or only a weak NOE with H41.
Furthermore, in 31, H25 shows a similarly intense NOE with
H31 and H27, in contrast with 23 and 30, where H25 shows a
much stronger NOE with H31. The structure of 31 most
consistent with the data positions the aryl rings in the same
plane (Figure 3C). This structure is unlikely to represent a stable
conformation and is more likely to result from the presence of
rapidly interconverting Sa and Ra atropisomers, which also
explains why only a single set of NMR signals are observed
for this macrocycle. The absence of the methyl group appears
to allow the hydroxyphenylglycine ring to rapidly interconvert
between two conformations, one in which H25 is proximal to
H31 and one in which H25 is proximal to H27. This, in turn,
appears to cause changes in the structure of the coupled Tyr7
aromatic ring and a puckering of the associated methylene unit,
which positions H45 closer to H39b than to H39a. Thus, in the
absence of the N-methyl group, the macrocycle becomes more
flexible. We conclude that N-methylation appears to rigidify
the macrocycle and localize it preferentially to a conformation
that is most appropriate for binding E. coli SPase.
Figure 3. (A) NOEs suggesting that macrocycles 23 and 30 exist
predominantly as the Sa atropisomer. (B) The conformation of arylomycin
A2 (truncated for clarity) bound in the active site of Spase (PDB ID: 1T7D9).
(C) NOEs suggesting that macrocycle 31 exists as rapidly interconverting
Sa and Ra atropisomers. (D) NOEs suggesting that macrocycles 32 exists
predominantly as the Ra atropisomer.
in DMSO, with a ratio of approximately 3:1. The deprotected
1
To further study the effect of N-methylation, we examined
macrocycle 32, where the phenol moieties are protected but the
exocyclic nitrogen is unmethylated. In the 1D 1H spectrum we
observed two sets of resonances in a ratio of approximately 4:1.
Interestingly, the ROESY spectrum of 32, suggests that it
predominantly adopts the Ra atropisomer (Figure 3D). This
conclusion is based on the strong NOEs between H25 and H27
(but not between H25 and H31), between H39a and H41, and
between H41 and H34 (but not between H27 and H34). The
chemical shifts and NOEs of the minor species of 32 suggest
that it corresponds to the Sa atropisomer. This data further
reinforces the conclusion that N-methylation pre-organizes the
macrocycle for SPase binding by favoring the Sa atropisomer.
macrocycle 30 showed a similar 1D H spectrum in DMSO,
with resonances present in a ratio of approximately 4:1. Thus,
the observation of two stable conformations is not an artifact
of macrocycle protection, and deprotection favors the major
atropisomer. ROESY spectra were next acquired for both model
macrocycles in DMSO. Exchange cross-peaks between the
different sets of resonances confirmed that they correspond to
equilibrating isomers. For the major isomers of both 23 and
30, NOEs were observed between H45 and both H39a and H39b
;
between NMe24 and both H31 and H27; between H34 and H27;
and between H38 and H41 (Figure 3A). In addition, a stronger
NOE was observed between H25 and H31 than between H25 and
H27. We also examined compound 30 in D2O and MeOD and
the results were virtually identical. These NOEs are uniquely
consistent with the proton-proton distances of the Sa atropi-
somer, which is the atropisomer observed in the structure of
arylomycin A2 bound to SPase9 (Figure 3B).
To examine the contribution of methylation at MeHpg5 to
macrocycle structure and dynamics, we synthesized macrocycle
31. Although the resonances of 31 are significantly overlapped
in DMSO, it was apparent that, in contrast to 23 and 30, 31
shows a single set of resonances and a distinctly different pattern
of NOEs. To confirm this conclusion and to facilitate interpreta-
tion of the NOEs, spectra were acquired in MeOD; these spectra
were similar to those acquired in DMSO, but were better
resolved. In the ROESY spectrum of 31 in MeOD, H39a shows
an NOE only with H41, while H39b shows a strong NOE only
with H45 (Figure 3C). This contrasts with 23 and 30 where both
2.4. Biological Activity of Arylomycin A2. With the
synthetic natural product in hand, we tested its antibiotic activity
against the major human pathogens E. coli (strain MG1655)
and S. aureus (strain 8325). We found that high concentrations
of arylomycin A2 inhibit growth of polymyxin B nonapeptide
permeabilized E. coli (minimum inhibitory concentration (MIC)
of 128 µM), and as reported,1b have no effect on the growth of
wild-type, non-permeabilized E. coli (Table 2). We also did not
observe any growth inhibition of S. aureus with the natural
product at concentrations up to 128 µM.
Because several lipoglycopeptides are known to inhibit the
growth of S. aureus,6 we were interested in deconvoluting the
contributions of the fatty acid tail length and the phenylhydroxy-
glycine deoxymannose to antibacterial activity. Toward this goal,
iso-C16 arylomycin 37, possessing the longer iso-C16 fatty acid
9
J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007 15835

A R T I C L E S
Roberts et al.
infections. We found that vancomycin, linezolid, and dapto-
mycin kill S. epidermidis with MICs of 1.0, 0.5, and 1.0 µM,
respectively. Thus, both arylomycin A2 and 37 are as potent
against S. epidermidis as the antibiotics that are currently
prescribed for its treatment.
Finally, to test the prediction that exocyclic methylation pre-
organizes the macrocycle for biological activity, we synthesized
the derivative of 37 where the methyl group is removed.
Compound 37 was selected, instead of arylomycin A2, due to
its increased potency against E. coli. The des-N-methyl analog
39 was synthesized by Boc-deprotection of macrocycle 32 and
coupling to dipeptide 35, followed by global deprotection
(Scheme 7). Remarkably, removing the methyl group dramati-
cally reduced the ability of the compound to inhibit the growth
of permeabilized E. coli (Table 2). In contrast, removing the
methyl group did not have a significant effect on growth
inhibition with either of the Gram-positive bacteria, S. aureus
or S. epidermidis.
Figure 4. Inhibition of S. aureus growth. Bacterial cultures were subject
to compound 37 or 39 in DMSO at the concentrations indicated or to DMSO
alone. Colony forming units (cfu) were determined as described in the
Supporting Information.
3. Discussion
Nature is replete with biaryl- or biaryl ether-linked cyclic
peptides that possess remarkable biological activities.17 Modi-
fication by N-methylation, glycosylation, and/or lipidation can
impart cyclic peptides with potent antibacterial activity, as
exemplified by vancomycin27 and teicoplanin.28 Interestingly,
the arylomycins and related lipoglycopeptides bear all three
modifications; however, the specific contributions of these
modifications to biological function are unknown. The ability
of these natural products to inhibit SPase, a structurally and
mechanistically unique bacterial Ser-Lys dyad protease required
for protein export, makes them promising candidates for
development as antibiotics. Synthetic access to these compounds
and their derivatives was expected not only to allow for their
evaluation as potential antibiotics, but also to provide a unique
opportunity to study the contribution of N-methylation, lipida-
tion, and glycosylation to antibacterial activity.
A modular strategy was implemented to synthesize arylo-
mycin A2 by independently constructing the C-terminal mac-
rocycle and an N-terminal lipopeptide tail before joining them
to form the natural product. The most significant obstacles were
closure of the tripeptide macrocycle and subsequent methylation
of the exocyclic amine. Macrolactamization, an efficient route
to macrocyclization of similar compounds,12b,13c,16-18 proceeded
only at low substrate concentrations and with low yields. It is
possible that formation of the 14-membered ring of the
arylomycins is uniquely disfavored relative to the smaller and
larger rings of the other biaryl-linked systems for which
macrolactamization has been reported (12 atoms for the van-
comycin AB ring system12b,13c and 17 atoms for the complestatin
and TMC-95A systems17,18,29). The specific biphenyl linkage
topology of the arylomycins may also prevent efficient mac-
rolactamization. In contrast, macrocyclization by intramolecular
Suzuki-Miyaura coupling proved more successful, possibly due
to preorganization of the peptide backbone,29 and accordingly,
this strategy was used in the synthesis of the natural product.
tail of the lipoglycopeptides and the sugar free biaryl linkage
of arylomycin A2, was synthesized by modification of the above-
described procedure (Scheme 7). Isopalmitic acid chloride was
coupled to nosyl deprotected D-Ser(OBn)-D-Ala-OMe, and the
product was deprotected with LiOH yielding 35 in 43% yield
(3 steps). This lipopeptide was then coupled to macrocycle 1
and globally deprotected, as described for the natural product
(Scheme 8).
The antibiotic activity of the iso-C16 arylomycin 37 was
examined with E. coli and S. aureus. Compound 37 inhibited
the growth of permeabilized E. coli with an MIC of 16 µM.
Exposure of S. aureus to up to 128 µM of 37 did not completely
abolish growth; however, by visual inspection dramatically
reduced cell densities were apparent at concentrations as low
as 16 µM. To quantify these results, we determined the viable
colony forming units of S. aureus as a function of time after
exposure to 37 (Figure 4). In the absence of the inhibitor, S.
aureus outgrew ∼400-fold during an 8 h period, while in the
presence of 32 µM of 37, they outgrew only ∼4-fold. Thus,
like the analogous lipoglycopeptide (Shen-Bin, P., personal
communication), 37 significantly inhibits the growth of S.
aureus.
Encouraged by the results with S. aureus, we examined the
antibacterial activity of compound 37 against a variety of other
Gram-positive pathogens, including Enterococcus faecalis (ATCC
29212 and 51299), Enterococcus faecium (Achaogen Inc. strain
AEFA001), Bacillus anthracis (Sterne), and Staphylococcus
epidermidis (ATCC 35984). Compound 37 does not inhibit the
growth of the enterococci E. faecalis or E. faecium. However,
it does inhibit the growth of B. anthracis, with an MIC of 32
µM. Moreover, compound 37 is remarkably potent against S.
epidermidis, inhibiting its growth with an MIC of 0.5 µM. To
examine the contribution of fatty acid tail length to S. epider-
midis toxicity, we determined the MIC of arylomycin A2.
Surprisingly, the natural product showed virtually identical
inhibition of S. epidermidis (MIC ) 1.0 µM). To better gauge
the potency of arylomycin A2 and 37, we also determined the
potency of vancomycin, linezolid, and daptomycin, which are
currently the standard treatment options for S. epidermidis
(27) McCormick, M. H.; Stark, W. M.; Pittenger, G. E.; Pittenger, R. C.;
McGuire, J. M. Antibiot. Ann. 1955-1956, 606-611.
(28) (a) Bardone, M. R.; Paternoster, M.; Coronelli, C. J. Antibiot. 1978, 31,
170-177. (b) Parenti, F.; Beretta, G.; Berti, M.; Arioli, V. J. Antibiot. 1978,
31, 276-283.
(29) Elder, A. M.; Rich, D. H. Org. Lett. 1999, 1, 1443-1446.
9
15836 J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007

Synthesis and Analysis of Arylomycin A2
A R T I C L E S
Monomethylation of the macrocycle’s exocyclic amine proved
surprisingly challenging. Although this step was eventually
accomplished using nosyl protection followed by MeI/K2CO3,
the failures suggest that the amine environment is sterically
encumbered. The steric congestion about the exocyclic amine
is likely the reason why modifications at this site can influence
the conformation of the proximal biaryl moiety (see below).
cell membrane may limit activity. Indeed, arylomycin A2 inhibits
growth of an outer membrane permeabilized strain of E. coli
with an MIC of 128 µM. Further supporting the suggestion that
the activity of the natural product is limited by the outer
membrane of Gram-negative bacteria, arylomycin A2 has
previously been shown to inhibit the growth of several Gram-
positive aerobic soil bacteria, including Arthrobacter pascens,
Rhodococcus erythropolis, and Streptomyces Viridochromoge-
nes.1 However, we found that the natural product did not inhibit
the growth of S. aureus, suggesting that additional factors must
also play a role in the compound’s biological activity.
In addition to the peptide itself, there are four structural
aspects of the arylomycins and related lipoglycopeptides that
may contribute to biological activity, including, atropisomerism
about the C-C biaryl bond, N-methylation, lipidation, and
glycosylation. To characterize atropisomerism in these com-
pounds, we examined macrocycles, 23 and 30. These model
macrocycles bear a glycine residue in lieu of the full lipopeptide
tail, and thus allow for the characterization of atropisomerism
dynamics of the macrocycle, free from the complexities associ-
ated with the known cis-trans isomerism of the lipopeptide tail.1
The observation of doubled resonances and exchange peaks in
the ROESY spectra of both macrocycles in DMSO confirmed
that these simple macrocycles exist as two exchanging confor-
mations, and that this heterogeneity does not result from the
aryl protecting groups. On the basis of several key sets of NOEs
To help understand the contribution of fatty acid tail length
and macrocycle glycosylation to antibiotic activity, we synthe-
sized the hybrid iso-C16 arylomycin 37, which possesses the
longer iso-C16 fatty acid tail of the lipoglycopeptides and the
unglycosylated macrocycle of the arylomycins. Interestingly,
elongation of the fatty acid tail produced a significantly more
potent inhibitor of permeabilized E. coli, inhibiting growth with
an MIC of 16 µM. This MIC is not significantly different from
that observed with the lipoglycopeptides,6 suggesting that tail
length is the dominant factor differentiating the antibiotic activity
of arylomycins and the lipoglycopeptides against permeablized
Gram-negative bacteria. The more hydrophobic tail likely favors
insertion into the hydrophobic inner membrane, possibly
facilitating access to SPase. Unlike arylomycin A2, compound
37 also significantly inhibits the growth of S. aureus, at
concentrations as low as 16 µM, however, it fails to display an
MIC due to a minimal but measurable growth at concentrations
up to 128 µM (Figure 4). The behavior is qualitatively similar
to that observed with the corresponding lipoglycopeptide
(Sheng-Bin, P., personal communication). Thus, as with Gram-
negative bacteria, increasing the length of the fatty-acid tail
increases potency against S. aureus and glycosylation does not
appear to make a significant contribution.
(including H34 to H27 and/or to H41; H41 to H39a and/or H39b
;
H45 to H39a and/or H39b; and H25 to H27 and/or H31), the major
conformation observed in both cases was the Sa atropisomer,
which is the same as that observed in the crystal structure of
the natural product bound to E. coli SPase (Figure 3B).
Methylation is a conspicuous modification of the natural
product. It has already been reported that N-methylation at
MeSer2 contributes to cis-trans isomerization within the li-
popeptide tail,1 an observation consistent with the effects of
N-methylation on peptide structure in organic solvents.30
Structural studies demonstrate that the MeHpg5 methyl group
is oriented into solvent and does not mediate any interactions
with SPase, nor does it cause isomerization about the amide
bond. However, N-methylation is known to impact other aspects
of peptide structure,31 and as discussed above, the challenges
associated with installation of the methyl group suggested that
it resides in a well-packed environment and thus might influence
biaryl ring structure. To examine the effect of N-methylation
on arylomycin atropisomerism, we synthesized and characterized
macrocycle 31. The data suggests that 31 exists as an average
of rapidly interconverting Sa and Ra atropisomers. Characteriza-
tion of 32 revealed that the effect of methylation is even more
pronounced when the interactions between the aryl hydroxy
groups are exacerbated with methyl protecting groups; the
phenol-protected macrocycle of 32 adopts predominantly the
Ra atropisomer. Thus, N-methylation appears to favor the Sa
atropisomer, the same atropisomer observed in the structure of
the arylomycin A2 complex with E. coli SPase.
To help understand the contribution of methylation and
atropisomerism to antibiotic activity, and to test the hypothesis
that N-methylation favors the Sa atropisomer that binds E. coli
SPase, we synthesized 39, the des-N-methyl analog of com-
pound 37. Interestingly, whereas 37 inhibits the growth of
permeablized E. coli with an MIC of 16 µM, compound 39
showed no MIC, up to 128 µM. Decreased inhibition of SPase
by 39 is unlikely to result from decreased cell penetrance as
SPase resides on the outer surface of the inner membrane and
the outermembrane is essentially removed as a significant barrier
by permeabilization. Thus, on the basis of these results, along
with the structural studies described above, we tentatively
conclude that N-methylation favors the Sa atropisomer and pre-
organizes the natural product for binding to E. coli SPase. In
contrast, 37 and 39 showed identical inhibition of S. aureus.
Whether the dissimilar consequences of methylation result from
differences between the different SPase proteins, or other
differences between the biology of these Gram-negative and
Gram-positive pathogens, remains to be determined.
With access to synthetic arylomycin A2, we tested its
antibacterial activity against the major human pathogens E. coli
and S. aureus, representatives of Gram-negative and Gram-
positive bacteria. As reported with the isolated natural product,1
arylomycin A2 had no effect on wild-type E. coli, despite binding
E. coli SPase in Vitro with a dissociation constant of ap-
proximately 800 nM.19 This suggests that penetration of the outer
Having identified fatty acid chain length and methylation as
important determinants of antibacterial activity, we proceeded
with compound 37 to examine the inhibition of other Gram-
positive pathogens. We found no antibacterial activity against
the enterococci E. faecalis and E. faecium, but 37 shows
promising activity against both B. anthracis and S. epidermidis.
Most notably, 37 is as potent against S. epidermidis as
(30) Vitoux, B.; Aubry, A.; Cung, M. T.; Marraud, M. Int. J. Pept. Protein
Res. 1986, 27, 617-632.
(31) Manavalan, P.; Momany, F. A. Biopolymers 1980, 19, 1943-1973.
9
J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007 15837

A R T I C L E S
Roberts et al.
vancomycin, linezolid, and daptomycin, the antibiotics that are
currently prescribed for the treatment of S. epidermidis infec-
tions. Interestingly, in contrast with E. coli and S. aureus, fatty
acid chain length is not critical for activity against S. epider-
midis, as arylomycin A2 inhibited growth with a similar MIC.
Also in contrast to E. coli, but similar to S. aureus, removal of
the methyl group from 37 did not significantly reduce the
potency against S. epidermidis. The potency of these compounds
against S. epidermidis is particularly interesting, as this bacteria
is known to be the primary cause of infections associated with
indwelling medical devices, suggesting that these compounds
may have well-defined and important therapeutic applications.
pre-organize the macrocycle for binding E. coli SPase, where
fatty acid tail length, but not glycosylation, is also important.
With regard to S. aureus, fatty acid tail length, but not
N-methylation or glycosylation, appears to be important. Finally,
with S. epidermidis, neither increasing the length of the fatty
acid chain, nor removal of the N-methyl group alters growth
inhibition. Clearly, the impacts of methylation, lipidation, and
potentially glycosylation on antibiotic activity depend on the
target bacteria. Future studies aimed at increasing the potency
and spectrum of these compounds will focus on understanding
the bacteria-specific effects as well as on the optimization of
the molecule, itself. The reported modular synthetic route to
these natural products should facilitate these studies, and should
also help further define the potential of this class of natural
products as antibiotics and/or inhibitors of bacterial virulence.
4. Conclusion
The identification of antibiotics that act via unique mecha-
nisms is critical for combating bacteria, which are rapidly
evolving resistance to all currently available antibiotics. The
bacterial Ser-Lys dyad protease SPase is an attractive target for
drug design because it is surface exposed, essential for bacterial
viability, and it has a unique structure and catalytic mechanism.
In addition, it is required for processing proteins involved in
resistance to other antibiotics and virulence. We have reported
the first total synthesis of the SPase inhibitor arylomycin A2,
as well as several of its derivatives. Suzuki-Miyaura coupling,
as opposed to macrolactamization, was found to provide the
most efficient route to the natural product. Synthetic access to
arylomycin A2 and several of its derivatives allowed us to
determine that exocyclic N-methylation at MeHpg5 appears to
Acknowledgment. We thank D.L. Boger and P.S. Baran for
helpful conversations, and H.-P. Fiedler for kindly providing a
sample of arylomycin A2. Funding for this work was provided
by the Office of Naval Research (Grant No. N00014-03-1-0126)
and Achaogen, Inc.
Supporting Information Available: Experimental procedures,
characterization of new compounds, 1H, 13C, and ROESY NMR
spectra, and complete ref 6. This material is available free of
charge via the Internet at http://pubs.acs.org.
JA073340U
9
15838 J. AM. CHEM. SOC. VOL. 129, NO. 51, 2007