Cyclic peptide inhibitors of staphylococcal virulence prepared by Fmoc-based thiolactone peptide synthesis

Article abstract of DOI:10.1021/ja711126e

Virulence factor production in Staphylococcus aureus is largely under the control of the accessory gene regulator (agr) quorum sensing system. There are four agr groups, all of which exhibit bacterial interference: each agr type synthesizes a cyclic autoinducing peptide (AIP) with a distinct sequence that activates its cognate AgrC receptor and inhibits activation of others. To better understand inhibitory AIP-AgrC interactions, we aimed to identify the minimal molecular determinants required to inhibit both non-cognate and cognate receptors. This minimization of the AIP pharmacophore also may have therapeutic relevance as the use of native AIPs to block virulence of non-cognate agr strains can prevent the establishment of an infection in vivo. We synthesized and evaluated the inhibitory activities of 10 AIP derivatives based on a truncated AIP analogue that inhibits all four aortypes. To carry out the rapid, parallel synthesis of these peptides, we employed a new linker for Fmoc-based thioester peptide synthesis. Our results identify key structural elements that are necessary for AgrC inhibition and reveal key differences between non-cognate and cognate inhibitory requirements.

Full text of DOI:10.1021/ja711126e

Published on Web 03/12/2008
Cyclic Peptide Inhibitors of Staphylococcal Virulence
Prepared by Fmoc-Based Thiolactone Peptide Synthesis
Elizabeth A. George,† Richard P. Novick,‡ and Tom W. Muir*,†
The Laboratory of Synthetic Protein Chemistry, Rockefeller UniVersity, New York, New York
10065, and Tri-institutional Training Program in Chemical Biology, Skirball Institute,
Department of Microbiology, New York UniVersity Medical Center, New York, New York 10016
Received December 14, 2007; E-mail: muirt@rockefeller.edu
Abstract: Virulence factor production in Staphylococcus aureus is largely under the control of the accessory
gene regulator (agr) quorum sensing system. There are four agr groups, all of which exhibit bacterial
interference: each agr type synthesizes a cyclic autoinducing peptide (AIP) with a distinct sequence that
activates its cognate AgrC receptor and inhibits activation of others. To better understand inhibitory AIP-
AgrC interactions, we aimed to identify the minimal molecular determinants required to inhibit both non-
cognate and cognate receptors. This minimization of the AIP pharmacophore also may have therapeutic
relevance as the use of native AIPs to block virulence of non-cognate agr strains can prevent the
establishment of an infection in vivo. We synthesized and evaluated the inhibitory activities of 10 AIP
derivatives based on a truncated AIP analogue that inhibits all four agr types. To carry out the rapid, parallel
synthesis of these peptides, we employed a new linker for Fmoc-based thioester peptide synthesis. Our
results identify key structural elements that are necessary for AgrC inhibition and reveal key differences
between non-cognate and cognate inhibitory requirements.
Introduction
The induction of virulence in Staphylococcus aureus is largely
controlled by a quorum sensing system encoded by the accessory
gene regulator (agr) operon.1 AgrC, a receptor histidine kinase,
is activated upon binding to a cognate autoinducing peptide
(AIP), which facilitates the expression of downstream virulence
factors. There are many agr polymorphisms among staphylo-
cocci, including four within S. aureus, resulting in four agr
specificity groups with distinct AIP and AgrC sensor domain
sequences (Figure 1A). Remarkably, most cross-group AIP-
AgrC interactions are inhibitory, and although the physiological
relevance of this phenomenon is still under investigation, it is
a powerful tool used in the study of the agr system and has
therapeutic potential.2-4
The four S. aureus AIPs are seven to nine amino acids long
and all contain a thiolactone macrocycle, involving a conserved
cysteine sulfhydryl group and the R-carboxylate, and an
N-terminal tail region (Figure 1B). Previous structure-activity
relationship (SAR) studies identified the hydrophobic character
of the two C-terminal amino acids, which are highly conserved
across all staphylococcal AIPs, as an essential element for both
Figure 1. agr quorum sensing system. (A) AgrD is a propeptide that is
processed and secreted, in part, by AgrB to form the mature AIP. Once a
threshold AIP concentration is reached, the AIP molecules bind and activate
their target receptor histidine kinase, AgrC, triggering autophosphorylation
and phosphoryl transfer. AgrC transduces the signal to the response regulator
AgrA, which goes on to activate transcription of the agr locus and of the
effector molecule, RNAIII. (B) Structures and sequences of the AIPs of
each of the four S. aureus agr groups.
agonism and antagonism of AgrC.5 However, the AIPs also
contain activity determinants within either the tail region or the
macrocycle, depending on the group, that are required for full
activation of the receptor.3,6,7 While agonism is very sensitive
to changes in the AIP sequence, the antagonistic activities of
† Rockefeller University.
‡ New York University Medical Center.
(1) George, E. A.; Muir, T. W. ChemBioChem 2007, 8, 847.
(2) (a) Ji, G.; Beavis, R.; Novick, R. P. Science (Washington, DC, U.S.) 1997,
276, 2027. (b) Wright, J. W.; Jin, R.; Novick, R. P. Proc. Natl. Acad. Sci.
U.S.A. 2005, 102, 1691. (c) Fleming, V.; Feil, E.; Sewell, A. K.; Day, N.;
Buckling, A.; Massey, R. C. J. Bacteriol. 2006, 188, 7686.
(3) Mayville, P.; Ji, G.; Beavis, R.; Yang, H.; Goger, M.; Novick, R. P.; Muir,
T. W. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 1218.
(5) Wright, J. S., III; Lyon, G. J.; George, E. A.; Muir, T. W.; Novick, R. P.
Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 16168.
(6) Lyon, G. J.; Wright, J. S.; Muir, T. W.; Novick, R. P. Biochemistry 2002,
41, 10095.
(4) Lyon, G. J.; Mayville, P.; Muir, T. W.; Novick, R. P. Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 13330.
9
4914
J. AM. CHEM. SOC. 2008, 130, 4914-4924
10.1021/ja711126e CCC: $40.75 © 2008 American Chemical Society

Inhibition of Virulence
A R T I C L E S
Chart 1. Truncated AIP-II (trAIP-II) (1) and the Ten Analogues Synthesized and Tested in This Studya
a Changes in each analogue relative to 1 are in blue.
Experimental Procedures
the full-length, native AIPs seem to be fairly robust, being
broadly tolerant to substitutions and truncations.3-6,8 Further-
more, certain modifications convert an AIP to an antagonist of
the cognate AgrC while maintaining inhibition of all non-
cognate receptors. For example, and of relevance to the current
study, truncation of the four tail residues of AIP-II (to give
trAIP-II, 1) converts the peptide into a potent global inhibitor
of all four agr groups.4
To understand more fully the differences between agonism
and antagonism in the AIP-AgrC interaction, we sought to
discover the minimal determinants of AgrC inhibition. We
synthesized 10 AIP-II analogues (Chart 1) based on the trAIP-
II (1) scaffold to test the importance of the inhibition of the
peptide backbone, side chains, thiolactone linkage, and N-
terminus. To facilitate rapid, parallel synthesis of the AIPs, we
also developed a new Fmoc-based AIP synthesis that is generally
useful for the preparation of R-thioester peptides. The inhibitory
activities of the trAIP-II analogues were tested against both
AgrC-II and AgrC-I to compare trends for cognate and non-
cognate inhibition. This SAR analysis led to the identification
of a minimal pharmacophore structure with a reduced peptidic
character that retains the antagonism of AgrC.
Materials. Amino acids, HBTU (2-(1H-Benzotriazole-1-yl)-1,1,3,3-
tetramethyluronium hexafluorophosphate), HATU (2-(1H-7-Azaben-
zotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate), PyBop
(Benzotriazole-1-yl-oxy-tris-(dimethylamino)-phosphonium hexafluo-
rophosphate), and Rink, and MBHA resins were purchased from
Novabiochem (San Diego, CA) with the exceptions of Boc-N-methyl-
phenylalanine (Fluka, Buchs, Switzerland) and Fmoc-DAPA(Boc)-OH
(Neosystem Laboratoire, Strasbourg, France). All solvents were obtained
from Fisher (Pittsburgh, PA), and trifluoroacetic acid was obtained from
Halocarbon (River Edge, NJ). All other reagents were purchased from
Sigma Aldrich (Milwaukee, WI).
Reversed-Phase HPLC and Mass Spectrometry. Analytical and
semipreparative HPLC was performed using a Hewlett-Packard 1100
series instrument with diode array detection. A Vydac C18 column (5
µm, 4.6 mm × 150 mm) with a 1 mL/min flow rate was used for
analytical scale HPLC, and a Vydac C18 column (10 µm, 10 mm ×
250 mm) with a 4 mL/min flow rate was used for semipreparative
HPLC. A Waters 2795 Separations Module plus 996 photodiode array
detector with a Vydac C18 column (3 µm, 2.1 mm × 150 mm) and a
0.2 mL/min flow rate was used for LC-MS. In all cases, linear gradients
of 0.1% aqueous TFA (solvent A) versus 90% acetonitrile, 10% water,
and 0.1% TFA (solvent B) were utilized. ESI-MS was performed on a
PE Sciex API-100 single quadrupole electrospray mass spectrometer.
ESI-HRMS was performed on a Q-TOF Ultima hybrid quadrupole time-
of-flight electrospray mass spectrometer (University of Illinois at
Urbana-Champaign Mass Spectrometry Laboratory).
(7) McDowell, P.; Affas, Z.; Reynolds, C.; Holden, M. T.; Wood, S. J.; Saint,
S.; Cockayne, A.; Hill, P. J.; Dodd, C. E.; Bycroft, B. W.; Chan, W. C.;
Williams, P. Mol. Microbiol. 2001, 41, 503.
(8) (a) Otto, M.; Submuth, R.; Vuong, C.; Jung, G.; Gotz, F. FEBS Lett. 1999,
450, 257. (b) Scott, R. J.; Lian, L.; Muharram, S. H.; Cockayne, A.; Wood,
S. J.; Bycroft, B. W.; Williams, P.; Chan, W. C. Bioorg. Med. Chem. Lett.
2003, 13, 2449. (c) Lyon, G. J.; Wright, J. S.; Christopoulos, A.; Novick,
R. P.; Muir, T. W. J. Biol. Chem. 2002, 277, 6247.
(9) Schnolzer, M.; Alewood, P.; Jones, A.; Alewood, D.; Kent, S. B. Int. J.
Pept. Protein Res. 1992, 40, 180.
NMR Spectroscopy. Compounds 12-16 were dissolved in CD3-
1
OD or CDCl3, and H and 13C NMR spectra were recorded with a
Bruker 400 MHz spectrometer. Measurements were taken at 298 K,
and chemical shift values are relative to methanol (1H 3.31 ppm, 13C
49.00 ppm) or chloroform (1H 7.26 ppm, 13C 77.16 ppm). Peptides
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4915

A R T I C L E S
George et al.
1-11 were dissolved in DMSO-d6, and 1-D 1H, homonuclear 1H
TOCSY, and ROESY experiments were performed on a Bruker 400
MHz spectrometer, Bruker 600 MHz spectrometer with cryoprobe, or
Bruker Avance 700 MHz spectrometer with cryoprobe, respectively.
The mixing times were 77 and 200 ms for TOCSY and ROESY
experiments, respectively. In all cases, the sample temperature was 298
K, and the reported chemical shift values are relative to DMSO at 2.50
ppm. The chemical shift index (CSI) of each peptide was calculated
using the formula
added to a solution of 14 (800 mg, 2.6 mmol, 1.0 equiv) in 26 mL of
CH2Cl2 under N2. The reaction was stirred at room temperature for 2
h, then diluted with CH2Cl2 and ethyl acetate. The product was washed
with 0.1 N HCl, 1 N NaOH, and brine, dried over Na2SO4, and filtered,
and the solvents were removed in vacuo. Purification by flash
chromatography using 19:1 hexanes/ethyl acetate yielded 590 mg (73%)
of 15. 1H NMR (400 MHz, CDCL3) δ 9.64 (s, 1H), δ 4.22 (m, 1H), δ
2.99 (dd, 1H, J ) 13.4, 4.2 Hz), δ 2.80 (dd, 1H, J ) 13.4, 7.9 Hz), δ
1.31 (s, 9H), δ 0.91 (s, 9H), δ 0.12 (s, 3H), δ 0.10 (s, 3H); 13C NMR
(100 MHz, CDCl3) δ 202.17, 76.57, 48.22, 43.66, 29.97, 25.84, 18.30,
-4.55, -4.63, ESI-MS m/z calcd for C13H28O2S2Si 308.1, found 309.0
([M + H+]).
CSI ) Σ(|aδi - 1δi|)/n
2-(t-Butyl-dimethyl-silanyloxy)-3-t-butyldisulfanyl-propionic Acid
(16). Compound 15 (227 mg, 0.74 mmol, 1.0 equiv) was dissolved in
4 mL of t-butyl alcohol. 2-Methyl-2-butene (3.02 g, 42.9 mmol, 58
equiv) was added to the solution on ice, followed by a pre-dissolved
solution of monobasic sodium phosphate (225 mg, 1.63 mmol, 2.2
equiv) and 80% sodium chlorite (251 mg, 2.22 mmol, 3.0 equiv) in
water (4 mL). The reaction was stirred vigorously to ensure mixing of
the two phases at room temperature for 2 h. The reaction mixture was
diluted with ethyl acetate, washed with 0.1 N HCl and brine, dried
over Na2SO4, and filtered, and the solvents were removed in vacuo.
Purification by flash chromatography using 19:1 CH2Cl2/methanol
yielded 160 mg (66%) of 16 as a clear oil. 1H NMR (400 MHz, CDCL3)
δ 4.47 (m, 1H), δ 3.12 (dd, 1H, J ) 13.4, 3.6 Hz), δ 2.93 (dd, 1H, J
) 13.4, 7.9 Hz), δ 1.33 (s, 9H), δ 0.92 (s, 9H), δ 0.15 (s, 3H), δ 0.13
(s, 3H); 13C NMR (100 MHz, CDCl3) δ 176.60, 71.27, 48.23, 45.77,
30.00, 25.82, 18.29, -4.80, -4.86, ESI-HRMS m/z calcd for C13H28O3S2-
Si [M + H+] 325.1327, found 325.1329.
Fmoc AIP Synthesis. Peptides 1-7, 9, and 11 were manually
synthesized. Linker 16 (188 mg, 0.61 mmol, 1.1 equiv relative to resin
substitution) was added to a solution of HBTU (207 mg, 0.55 mmol,
1.0 equiv relative to resin loading) in DMF (1.2 mL), followed by DIEA
(diisopropylethylamine) (0.25 mL, 1.44 mmol, 2.6 equiv relative to
resin) to pre-activate the acid. After 3 min, the solution was added to
Rink amide resin and stirred by bubbling N2 through the vessel at room
temperature for 4 h. Remaining free amines were acetylated by two
treatments with a 1:1:8 solution of acetic anhydride/DIEA/DMF for
10 min. Deprotection of the TBS group was accomplished by the
addition of tetrabutyl ammonium fluoride (10 mL of a 1.0 M solution
in THF) overnight at room temperature. 4-Dimethylaminopyridine (0.1
equiv relative to phenylalanine) was used in addition to HBTU and
DIEA to double couple phenylalanine to the hydroxyl group of the
linker. Subsequent steps were completed with standard HBTU/DIEA
activation and piperidine deprotection protocols for Fmoc solid-phase
peptide chemistry with a few exceptions. N-Methylated amino acids
were double coupled, once with HBTU and once with HATU. Other
unnatural monomers (Fmoc-5-amino-pentanoic acid in 9 and 3-trityl-
sulfanyl-propionic acid in 11) were coupled with standard methods.
Finally, all peptides, with the exception of 11, were acetylated at the
N-terminus with a 1:1:8 solution of acetic anhydride/DIEA/DMF for
10 min. Each peptide was cleaved from the solid support by treatment
with 95:2.5:2.5 TFA/TIS/water for 3-4 h at room temperature, followed
by filtration and washing of the beads with the TFA cleavage cocktail.
The peptides were either purified by semipreparative HPLC or taken
on to the cyclization step without purification.
where a is the peptide of interest, 1 is peptide 1, i is the proton type
(e.g., Cys1 NH), and n is the total number of protons included in the
calculation. Protons directly bonded to the macrocyclic scaffold plus
the exocyclic N-terminal amide proton were included in each CSI (Cys1
NH, RH, âH; Ser2 NH, RH; Ser3 NH, RH; Leu NH, RH; and Phe NH,
RH). The protons of the modified residue(s) were excluded.
Synthesis. 3-t-Butyldisulfanyl-propane-1,2-diol (12). A solution
containing 1-mercapto-ethane-1,2-diol (3.0 g, 27.7 mmol, 1.0 equiv),
2-methyl-propane-2-thiol (25 g, 277 mmol, 10.0 equiv), and triethy-
lamine (7.0 g, 69.3 mmol, 2.5 equiv) dissolved in methanol (55 mL)
was stirred at room temperature with bubbling O2 overnight (>10 h).
Methanol and excess 2-methyl-propane-2-thiol were removed in vacuo,
and the crude product was purified by flash chromatography using 3:2
hexanes/ethyl acetate to give 5.4 g of 12 (99%) as a clear oil. 1H NMR
(400 MHz, CD3OD) δ 3.83 (m, 1H), δ 3.58 (dd, 1H, J ) 11.3, 4.8
Hz), δ 3.52 (dd, 1H, J ) 11.3, 5.8 Hz), δ 2.90 (dd, 1H, J ) 13.3, 5.3
Hz), δ 2.78 (dd, 1H, J ) 13.3, 7.3 Hz), δ 1.34 (s, 9H); 13C NMR (100
MHz, CD3OD) δ 72.33, δ 65.90, δ 48.45, δ 45.64, δ 30.25, ESI-MS
m/z calcd for C7H16O2S2 196.3, found 393.0 ([2M + H+]).
1,2-Bis-(t-butyl-dimethyl-silanyloxy)-3-t-butyldisulfanyl-pro-
pane (13). t-Butyl-chloro-dimethyl-silane (11 g, 72 mmol, 3.0 equiv),
imidazole (9.8 g, 144 mmol, 6.0 equiv), and 4-dimethylaminopyridine
(170 mg, 1.4 mmol, 0.06 equiv) were added on ice to a solution of 12
(4.7 g, 24 mmol, 1.0 equiv) dissolved in 240 mL of anhydrous DMF.
The reaction was stirred under N2 at room temperature overnight at
which point 0.5 N NaOH was added to quench the reaction, and the
product was extracted with ethyl acetate, washed with 1 N HCl, dried
over Na2SO4, filtered, and dried in vacuo. Excess DMF was removed
by an azeotrope with hexanes to yield 10 g (99%) of 13, which was
1
taken on to the next step without further purification. H NMR (400
MHz, CDCl3) δ 3.89 (m, 1H), δ 3.61 (dd, 1H, J ) 10.1, 5.0 Hz), δ
3.53 (dd, 1H, J ) 10.1, 6.3 Hz), δ 2.99 (dd, 1H, J ) 13.1, 4.9 Hz), δ
2.73 (dd, 1H, J ) 13.1, 6.7 Hz), δ 1.33 (s, 9H), δ 0.90 (s, 18H), δ
0.08 (m, 12H); 13C NMR (100 MHz, CDCl3) δ 72.70, 66.40, 47.81,
46.00, 30.06, 26.10, 26.04, 25.86, 25.82, 18.50, 18.30, -2.79, -3.41,
-4.39, -4.20, -5.20, -5.18, ESI-MS m/z calcd for C19H44O2S2Si2
424.2, found 423.0 ([M-]).
2-(t-Butyl-dimethyl-silanyloxy)-3-t-butyldisulfanyl-propan-1-ol (14).
A cold 1:1 mixture of TFA and water (18 mL) was added dropwise to
a solution of 13 (4.4 g, 10 mmol, 1.0 equiv) in 45 mL of THF. The
reaction was stirred at 0 °C for 4.5 h, and then it was slowly poured
into a separation funnel containing an ice cold solution of NaHCO3 to
bring the pH above 7. The product was extracted with ethyl acetate,
washed with water and brine, dried over Na2SO4, and filtered, and the
solvents were removed in vacuo. Purification by flash chromatography
The cyclization buffer contained 20% acetonitrile and 80% 6 M
guanidinium chloride in 0.1 M sodium phosphate buffer pH 6.5. TCEP
(tris-(2-carboxy)ethyl phosphine) (70 mM) was added to this solution,
and the pH was brought back up to 6.6-6.8 with 4 M NaOH. Each
linear peptide was dissolved in this solution to a final concentration of
100 µM and rocked at room temperature for 2-24 h, and the reaction
was monitored by analytical HPLC. The desired AIP product was then
purified by semipreparative HPLC.
1
with 15:1 hexanes/ethyl acetate yielded 2.0 g (65%) of 14. H NMR
(400 MHz, CD3Cl) δ 3.97 (m, 1H), δ 3.70 (dd, 1H, J ) 11.3, 3.5 Hz),
δ 3.62 (dd, 1H, J ) 11.3, 4.0 Hz), δ 2.87 (dd, 1H, J ) 13.4, 7.2 Hz),
δ 2.79 (dd, 1H, J ) 13.4, 5.4 Hz), δ 1.34 (s, 9H), δ 0.91 (s, 9H), δ
0.13 (s, 3H), δ 0.10 (s, 3H); 13C NMR (100 MHz, CDCl3) δ 72.14,
65.01, 48.06, 44.05, 29.98, 25.93, 18.19, -4.29, -4.48, -4.62, ESI-
MS m/z calcd for C13H30O2S2Si2 310.2, found 309.0 ([M-]).
2-(t-Butyl-dimethyl-silanyloxy)-3-t-butyldisulfanyl-propionalde-
hyde (15). Dess-Martin periodinane (1.3 g, 3.1 mmol, 1.2 equiv) was
Data for Peptide 1. 1H NMR (400 MHz, DMSO-d6) δ 8.68 (d, 1H,
J ) 8.8 Hz), δ 8.46 (d, 1H, J ) 8.2 Hz), δ 8.12 (m, 2H), δ 7.60 (d,
9
4916 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

Inhibition of Virulence
A R T I C L E S
1H, J ) 8.7 Hz), δ 7.25 (m, 2H), δ 7.21 (m, 3H), δ 5.12 (t, 1H, J )
5.4 Hz), δ 5.03 (t, 1H, J ) 5.4 Hz), δ 4.56 (m, 1H), δ 4.39 (m, 1H),
δ 4.28 (m, 1H), δ 4.22 (m, 1H), δ 3.92 (m, 1H), δ 3.64 (m, 1H), δ
3.58 (m, 2H), δ 3.52 (m, 1H), δ 3.31 (obsc 1H), δ 3.16 (m, 1H), δ
2.84 (m, 2H), δ 1.86 (s, 3H), δ 1.38 (m, 1H), δ 1.30 (m, 1H), δ 1.22
(m, 1H), δ 0.74 (d, 3H, J ) 6.3 Hz), δ 0.71 (d, 3H, J ) 6.3 Hz),
ESI-MS m/z calcd for C26H37N5O8S 579.2, found 580.0 ([M + H+]).
Data for Peptide 2. 1H NMR (400 MHz, DMSO-d6) δ 8.67 (d, 1H,
J ) 9.0 Hz) δ 8.44 (d, 1H, J ) 7.9 Hz), δ 8.21 (m, 1H), δ 7.99 (d, 1H,
J ) 7.7 Hz), δ 7.59 (d, 1H, J ) 8.3 Hz), δ 7.26 (m, 2H), δ 7.20 (m,
3H), δ 5.21 (t, 1H, J ) 5.5 Hz), δ 4.59 (m, 1H), δ 4.32 (m, 1H), δ
4.21 (m, 1H), δ 3.98 (m, 1H), δ 3.89 (m, 1H), δ 3.62 (m, 1H), δ 3.52
(m, 1H), δ 3.32 (obsc, 1H), δ 3.31 (obsc, 1H), δ 3.16 (m, 1H), δ 2.84
(m, 2H), δ 1.86 (s, 3H), δ 1.31 (m, 2H), δ 1.18 (m, 1H), δ 0.74 (d,
3H, J ) 6.2 Hz), δ 0.69 (d, 3H, J ) 6.2 Hz), ESI-MS m/z calcd for
C25H35N5O7S 549.2, found 550.0 ([M + H+]).
3.21 (obsc, 1H), δ 2.90 (m, 1H), δ 2.76 (m, 2H), δ 2.11 (m, 1H), δ
2.01 (m, 1H), δ 1.84 (s, 3H), δ 1.50 (m, 3H), δ 1.32 (m, 1H), δ 1.20
(m, 2H), δ 0.97 (m, 1H), δ 0.75 (d, 3H, J ) 6.5 Hz), δ 0.68 (d, 3H,
J ) 6.5 Hz); ESI-MS m/z calcd for C25H36N4O5S 504.2, found 505.0
([M + H+]).
1
Data for Peptide 11. H NMR (400 MHz, DMSO-d6) δ 8.76 (d,
1H, J ) 8.4 Hz), δ 8.18 (d, 1H, J ) 7.5 Hz), δ 8.06 (d, 1H, J ) 7.8
Hz), δ 7.34 (d, 1H, J ) 8.7 Hz), δ 7.26 (m, 3H), δ 7.20 (m, 2H), δ
5.09 (t, 1H, J ) 5.0 Hz), δ 4.98 (t, 1H, J ) 5.3 Hz), δ 4.54 (m, 1H),
δ 4.26 (m, 1H), δ 4.18 (m, 1H), δ 3.88 (m, 1H), δ 3.64 (m, 4H), δ
3.33 (obsc, 1H), δ 3.21 (m, 1H), δ 2.85 (m, 2H), δ 2.62 (obsc, 1H), δ
2.34 (m, 1H), δ 1.42 (m, 1H), δ 1.24 (m, 2H), δ 0.74 (d, 3H, J ) 6.3
Hz), δ 0.70 (d, 3H, J ) 6.3 Hz); ESI-MS m/z calcd for C24H34N4O7S
522.2, found 523.0 ([M + H+]).
trAIP-II Lactam (10) Synthesis. The peptide was synthesized
similarly to the full-length AIP-II lactam.3 Chain elongation was
completed using Fmoc SPPS protocols. Fmoc-serine(benzyl)-OH was
used in residue positions two and three. Fmoc-DAPA(Boc)-OH was
used in place of cysteine. Once cleaved from the resin with TFA, the
partially protected linear peptide was cyclized with PyBop and DIEA,
purified by semipreparative HPLC, and treated with 25:1 HF/4-methyl-
phenol (p-cresol) for 1 h at 0 °C to remove the benzyl groups. The
final product was purified with semipreparative HPLC. 1H NMR (400
MHz, DMSO-d6) δ 8.67 (m, 1H), δ 8.11 (m, 1H), δ 8.04 (m, 1H), δ
7.72 (m, 1H), δ 7.59 (m, 1H), δ 7.49 (m, 1H), δ 7.20 (m, 5H), δ 4.38
(m, 2H), δ 4.31 (m, 1H), δ 3.94 (m, 1H), δ 3.84 (m, 1H), δ 3.79 (m,
1H), δ 3.71 (m, 1H), δ 3.62 (m, 3H), δ 3.21 (obsc, 1H), δ 3.00 (m,
1H), δ 2.85 (m, 1H), δ 1.84 (s, 3H), δ 1.37 (m, 1H), δ 1.25 (m, 1H),
δ 1.10 (m, 1H), δ 0.76 (d, 3H, J ) 6.5 Hz), δ 0.69 (d, 3H, J ) 6.2
Hz); ESI-MS m/z calcd for C26H38N6O8S 562.3, found 563.5 ([M +
H+]).
1
Data for Peptide 3. H NMR (400 MHz, DMSO-d6) δ 8.46 (m,
1H), δ 8.36 (d, 1H, J ) 7.4 Hz), δ 8.30 (d, 1H, J ) 8.0 Hz), δ 8.02
(m, 2H), δ 7.24 (m, 5H), δ 4.64 (m, 1H), δ 4.41 (m, 2H), δ 3.99 (m,
1H), δ 3.79 (dd, 1H, J ) 14.0, 3.6 Hz), δ 3.57 (m, 3H), δ 3.35 (obsc,
1H), δ 3.13 (m, 1H), δ 2.91 (m, 1H), δ 2.62 (m, 1H), δ 1.86 (s, 3H),
δ 1.49 (m, 1H), δ 1.12 (m, 2H), δ 0.78 (d, 3H, J ) 6.5 Hz), δ 0.71 (d,
3H, J ) 6.5 Hz); ESI-MS m/z calcd for C25H35N5O7S 549.2, found
550.0 ([M + H+]).
1
Data for Peptide 4. H NMR (400 MHz, DMSO-d6) δ 8.39 (m,
1H) δ 8.29 (m, 2H), δ 8.15 (d, 1H, J ) 9.4 Hz), δ 8.08 (m, 1H), δ
7.23 (m, 5H), δ 4.67 (m, 1H), δ 4.42 (m, 1H), δ 4.10 (m, 1H), δ 3.98
(m, 1H), δ 3.79 (m, 1H), δ 3.58 (dd, 1H, J ) 14.2, 6.4 Hz), δ 3.28
(obsc, 1H), δ 3.18 (m, 1H), δ 2.87 (m, 1H), δ 2.72 (m, 1H), δ 1.86 (s,
3H), δ 1.43 (m, 2H), δ 1.10 (m, 1H), δ 0.78 (d, 3H, J ) 6.5 Hz), δ
0.71 (d, 3H, J ) 6.5 Hz); ESI-MS m/z calcd for C24H33N5O6S 519.2,
found 520.3 ([M + H+]).
trAIP-II MeF (8) Synthesis. Manual solid-phase peptide synthesis
with standard in situ neutralization/HBTU activation protocol for Boc
chemistry9 was used for chain elongation on a mercaptopropionamide
MBHA resin. Leucine was double coupled to N-methylated phenyla-
lanine with HBTU, then HATU. The peptide was cleaved from the
resin by treatment with 25:1 HF/p-cresol for 1 h at 0 °C, precipitated,
washed with diethyl ether, and purified by semipreparative HPLC. After
lyophilization, the linear peptide was dissolved again in MeCN, water,
and 0.1% TFA and cyclized in solution by the addition of a 4× volume
of 0.1 M sodium phosphate buffer, pH 7, for 2 h at room temperature.
The final AIP product was purified by semipreparative HPLC. 1H NMR
(400 MHz, DMSO-d6) δ 8.50 (d, 1H, J ) 8.3 Hz), δ 8.25 (d, 1H, J )
8.6 Hz), δ 7.86 (d, 1H, J ) 9.2 Hz), δ 7.80 (d, 1H, J ) 9.1 Hz), δ 7.25
(m, 3H), δ 7.14 (m, 2H), δ 5.16 (m, 1H), δ 5.00 (m, 1H), δ 4.63 (m,
1H), δ 4.39 (m, 1H), δ 4.28 (m, 1H), δ 4.25 (m, 1H), δ 4.20 (m, 1H),
δ 3.63 (m, 1H), δ 3.53 (m, 3H), δ 3.21 (obsc, 2H), δ 3.19 (obsc, 1H),
δ 2.89 (m, 1H), δ 2.64 (s, 3H), δ 1.88 (s, 3H), δ 1.52 (m, 2H),
δ 1.32 (m, 1H), δ 0.86 (d, 3H, J ) 6.0 Hz), δ 0.81 (d, 1H, J )
6.1 Hz); ESI-MS m/z calcd for C27H39N5O8S 593.3, found 595.0
([M + H+]).
Data for Peptide 5. 1H NMR (400 MHz, DMSO-d6) δ 8.73 (d, 1H,
J ) 8.3 Hz), δ 8.56 (d, 1H J ) 9.0 Hz), δ 7.98 (d, 1H J ) 7.4 Hz), δ
7.49 (d, 1H, J ) 8.1 Hz), δ 7.25 (m, 2H), δ 7.18 (m, 3H), δ 5.25 (m,
1H), δ 4.95 (m, 3H), δ 4.38 (m, 1H), δ 4.16 (m, 1H), δ 4.08 (m, 1H),
δ 3.80 (m, 1H), δ 3.73 (m, 2H), δ 3.33 (obsc, 1H), δ 3.12 (m, 1H), δ
3.00 (s, 3H), δ 2.90 (m, 1H), δ 2.83 (m, 1H), δ 1.87 (s, 3H), δ 1.26
(m, 3H), δ 0.73 (m, 6H); ESI-MS m/z calcd for C27H39N5O8S 593.3,
found 595.0 ([M + H+]).
Data for Peptides 6a and 6b. (The cis and trans amide isomers
were ∼50% each.) 1H NMR (400 MHz, DMSO-d6) δ 8.66 (d, 1H, J )
7.4 Hz), δ 8.47 (d, 1H, J ) 7.6 Hz), δ 8.24 (d, 1H, J ) 8.6 Hz), δ 8.10
(m, 3H), δ 7.48 (m, 2H), δ 7.25 (m, 5H), δ 7.14 (m, 5H), δ 4.90 (m,
1H), δ 4.83 (m, 1H), δ 4.59 (m, 1H), δ 4.51 (m, 1H), δ 4.45 (m, 1H),
δ 4.35 (m, 1H), δ 4.26 (m, 1H), δ 4.12 (m, 2H), δ 3.99 (m, 1H), δ
3.88 (m, 1H), δ 3.77 (m, 1H), δ 3.63 (m, 1H), δ 3.56 (m, 3H), δ 3.34
(obsc, 1H), δ 3.29 (obsc, 1H), δ 3.26 (obsc, 1H), δ 3.24 (obsc, 1H), δ
3.14 (m, 2H), δ 3.02 (m, 1H), δ 2.97 (m, 2H), δ 2.66 (s, 6H), δ 2.49
(obsc, 1H), δ 1.86 (s, 6H), δ 1.39 (m, 5H), δ 1.29 (m, 1H), δ 0.77 (m,
6H), δ 0.73 (m, 6H); ESI-MS m/z calcd for C27H39N5O8S 593.3, found
595.0 ([M + H+]).
Data for Peptide 7. 1H NMR (400 MHz, DMSO-d6) δ 8.45 (d, 1H,
J ) 8.2 Hz), δ 8.25 (d, 1H, J ) 9.9 Hz), δ 8.05 (d, 1H, J ) 8.5 Hz),
δ 7.76 (d, 1H, J ) 9.1 Hz), δ 7.25 (m, 3H), δ 7.21 (m, 2H), δ 5.15 (m,
1H), δ 4.82 (m, 1H), δ 4.75 (m, 1H), δ 4.67 (m, 1H), δ 4.44 (m, 1H),
δ 4.22 (m, 1H), δ 3.63 (m, 2H), δ 3.50 (m, 1H), δ 3.45 (m, 1H), δ
3.37 (obsc, 1H), δ 3.36 (obsc, 1H), δ 3.11 (m, 1H), δ 3.01 (s, 3H), δ
2.89 (m, 1H), δ 2.70 (m, 1H), δ 1.90 (m, 1H), δ 1.85 (s, 3H), δ 1.37
(m, 1H), δ 1.25 (m, 1H), δ 0.75 (d, 3H, J ) 6.5 Hz), δ 0.69 (d, 3H,
J ) 6.4 Hz); ESI-MS m/z calcd for C27H39N5O8S 593.3, found 595.0
([M + H+]).
Data for Peptide 9. 1H NMR (400 MHz, DMSO-d6) δ 8.24 (d, 1H,
J ) 7.8 Hz), δ 8.09 (m, 1H), δ 7.94 (d, 1H, J ) 8.9 Hz), δ 7.74 (d,
1H, J ) 9.2 Hz), δ 7.26 (m, 3H), δ 7.20 (m, 2H), δ 4.67 (m, 1H), δ
4.34 (m, 1H), δ 4.20 (m, 1H), δ 3.43 (m, 1H), δ 3.28 (obsc, 1H), δ
Inhibition Assays. The inhibitory activities of peptides 1-11 for
agr activation were analyzed using a previously described assay with
a â-lactamase reporter gene read-out.6 Briefly, cell cultures of agr-
null cell lines RN9222 (CA1-I) and RN9372 (CA2-II) with plasmids
containing agrCA and agr-P3::blaZ were grown in CYGP to early
exponential phase growth (OD650 ) 0.15-0.30). The 80 µL aliquots
were treated with 125 nM agonist AIP and varying concentrations of
peptides 1-11 or buffer for 1 h with shaking at 37 °C in a
THERMOmax microplate reader (Molecular Devices). Cell density was
monitored by OD650 readings taken every 1 min. Immediately following
Nitrocefin addition, hydrolysis was monitored by OD490 readings taken
every 20 s over 20 min. All peptide stock solution concentrations were
determined by amino acid analysis (AAA) at the Keck AAA and Protein
Sequencing Lab (Yale University, New Haven, CT).
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4917

A R T I C L E S
George et al.
Scheme 1. General Routes for Synthesis of AIPs with Truncated AIP-II (1) Shown as an Examplea
a (A) AIP synthesis using Boc chemistry.6 The resin is modified with a thiol handle via a previously described method.11 (B) Proposed Fmoc chemistry-
based route to AIPs involving a latent thioester linker.12
Data Analysis. Assay data were plotted as initial â-lactamase
hydrolysis velocity versus log peptide concentration. Error bars represent
the standard deviation from the mean. The values obtained for initial
velocity were normalized to cell density. The normalized data were
analyzed using nonlinear regression and fit to a sigmoidal dose response
curve using the PRISM 4.0 package (GraphPad Software, San Diego,
CA). IC50 values were extracted from the Hill equation used for
nonlinear regression analysis
II is a very poor agonist but potent antagonist of the agr
response.3 Thus, we designed a lactam (10) analogue to
investigate as to whether the thiolactone linkage is necessary
for inhibition in the context of truncated AIP-II. Last, to
complete the structure-activity analysis, we designed peptide
11 in which the acetylated R-amino group of the cysteine was
removed, allowing us to determine as to whether this vestige
of the native AIP tail plays any role in the inhibitory activity
of the peptide.
The standard AIP synthesis used in our laboratory involved
a combined solid-phase, solution-phase route in which a linear
unprotected peptide R-thioester, prepared by t-butoxy carbonyl
(Boc) solid-phase peptide synthesis (SPPS), underwent cycliza-
tion through transthioesterification in aqueous solution (Scheme
1A).6 Although this is an efficient synthetic route, the reliance
on Boc-SPPS to generate the R-thioester precursor does create
a bottleneck in the process, due to the need to use anhydrous
HF in the cleavage step. In principle, the generation of peptide
R-thioester precursors using Fmoc-SPPS would make feasible
the parallel synthesis of many AIP analogues, thereby allowing
more rapid structure-activity analysis of this system. However,
the synthesis of R-thioester peptides via Fmoc-SPPS is a long-
standing problem in peptide chemistry; and although many
methods have been developed, none has gained widespread use
owing to a variety of drawbacks, mainly a lack of generality or
low yields (summarized in ref 10a,b).
Emax - basal
E ) basal +
50 - log[A])nH
)
1 + 10((logIC
in which E is the effect, [A] is the antagonist concentration, nH is the
midpoint slope, and Emax and basal are the upper and lower asymptotes,
respectively.
Results
Design and Synthesis of trAIP-II Derivatives. To elucidate
the minimal determinants of inhibition by trAIP-II, we set out
to systematically perturb each structural aspect of the molecule
(Chart 1). Previously, alanine scanning of full-length AIP-II
showed that the two serine hydroxyl groups are not necessary
for biological activity.3 Guided by this observation, we designed
trAIP-II analogues 2-4, in which the serines were replaced with
glycines either individually or in combination. Given that the
hydrophobicity of the C-terminal residues is known to be
required for biological activity,5 the leucine and phenylalanine
side chains were not perturbed in this study. Next, each amide
in the macrocycle was N-methylated (5-8) to probe interactions
involving the AIP backbone, which had yet to be evaluated in
any AIP. To test as to whether these side-chain and backbone
modifications could be combined, we completely removed the
serines in peptide 9 and replaced them with a methylene linker
derived from 5-aminopentanoic acid.
In thinking about this problem, we were drawn to the latent
thioester-linker work of Botti et al.12 and Danishefsky et al.13
In these methods, C-terminal thioesters were generated following
an intramolecular O to S acyl shift. We imagined that the Botti
(10) (a) Muralidharan, V.; Muir, T. W. Nat. Methods 2006, 3, 429. (b) Camarero,
J. A.; Mitchell, A. R. Protein Pept. Lett. 2005, 12, 723.
(11) Camarero, J. A.; Cotton, G. J.; Adeva, A.; Muir, T. W. J. Pept. Res. 1998,
51, 303.
(12) Botti, P.; Villain, M.; Manganiello, S.; Gaertner, H. Org. Lett. 2004, 6,
4861.
(13) Warren, J. D.; Miller, J. S.; Keding, S. J.; Danishefsky, S. J. J. Am. Chem.
Soc. 2004, 126, 6576.
Previous SAR studies on full-length AIP-II revealed that the
thiolactone linkage is a critical determinant in receptor activa-
tion, but, interestingly, not inhibition; a lactam analogue of AIP-
9
4918 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

Inhibition of Virulence
A R T I C L E S
Scheme 2. (A) Synthesis of Linker 16a and (B) Fmoc AIP Synthesisb
a Conditions: (a) t-butylthiol, TEA, O2, MeOH, 99%; (b) TBSCl, imidazole, cat. DMAP, DMF, 99%; (c) TFA/H2O 1:5, THF, 65%; (d) DMP, DCM,
73%; (e) NaOCl, 2-methyl-2-butene, sodium phosphate buffer, 66%. bConditions: (f) 16, HBTU, DIEA, DMF; (g) i: 1 M TBAF, THF, ii: Fmoc-SPPS, and
iii: TFA, TIS, H2O; (h) TCEP, buffer, pH 6.6-6.8.
methodology in particular could be directly adapted for use in
an AIP synthesis (Scheme 1B). In this approach, Fmoc-cysteine
(t-butylthio)-OH was first coupled to PEGA resin, followed by
R-amino diazotization and subsequent hydrolysis under aqueous
conditions to afford an R-hydroxy cysteine resin. After chain
elongation and cleavage from the resin, reduction of the cysteine
side-chain disulfide led to an O to S acyl shift, resulting in
R-thioester formation. Although we were able to make AIPs
with this method, the isolated yields were very low (data not
shown). This result highlights the crucial drawbacks of this
method: the linker is synthesized on resin, and the resin options
are consequently limited to those that have good swelling
properties in water. To overcome these limitations, we designed
a solution-phase synthesis of 2-(t-butyl-dimethyl-silanyloxy)-
3-t-butyldisulfanyl-propionic acid (16), a molecule that acts as
a masked thioester and that can be coupled onto any resin using
standard approaches.
Initial efforts to generate R-hydroxy-cysteine-based linkers
in solution from t-butylthiol protected L-cysteine proved to be
unsatisfactory, primarily due to the poor yields associated with
the diazotization/hydrolysis step. Therefore, we adopted an
alternate route in which linker 16 was generated in five steps
from commercially available thioglycerol, with an overall yield
of 31%. As shown in Scheme 2, the thiol group was first
protected as the t-butyl disulfide to give 1,2-diol 12. Attempts
to selectively oxidize the primary alcohol of 12 directly to the
carboxylic acid using TEMPO14 were unsuccessful. Instead, we
protected both alcohols with t-butyl-dimethyl-silanyloxy (TBS)
groups, followed by selective removal of the primary TBS group
to give 14. Oxidation to acid 16 was then accomplished in two
steps, using Dess-Martin periodinane, followed by treatment
with sodium chlorite.
Racemic linker 16 was coupled onto Rink-amide polystyrene
resin with HBTU and DIEA. Following a capping procedure
and on-resin removal of the TBS group with TBAF, the
C-terminal residue of each peptide was attached to the resin
linker using HBTU in the presence of catalytic DMAP (a double
coupling procedure was typically employed). LC-MS analysis
of the cleaved amino-acylated resin indicated that this loading
procedure was highly efficient (i.e., no free linker was observed
in the crude mixture; data not shown). Peptide-chain assembly
was then performed using standard Fmoc-SPPS protocols with
slight modifications for the N-methylated residues, and the
peptides were cleaved from the resin with TFA (see Experi-
mental Procedures). In all cases, LC-MS analysis of the crude
cleavage mixtures revealed the presence of two closely eluting
peaks, both with the mass expected for desired linear peptide-
(14) (a) Davis, N. J.; Flitsch, S. L. Tetrahedron Lett. 1993, 34, 1181. (b)
Wovkulich, P. M.; Shankaran, J. K.; Uskokovic, M. R. J. Org. Chem. 1993,
58, 832.
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4919

A R T I C L E S
George et al.
the desired thiolactones (data not shown). Analysis of the
reaction mixture after 5 min revealed the presence of several
intermediates with a mass consistent with the fully reduced
version of peptide 1′ (Figure 2B). Presumably, one of these
intermediates corresponds to the linear R-thioester generated
in situ following the O to S acyl shift, although this is likely a
minor component due to the greater stability of oxyesters over
thioesters.15 Given that acyl transfer from a hydroxyl group to
a thiol group is highly unfavorable, we favor the mechanism
shown in Scheme 2B, in which the linear R-thioester peptide is
an obligatory intermediate leading to the final transthioesteri-
fication step, which acts as a thermodynamic sink in the process.
Consistent with this mechanism, efficient cyclization of the
linear peptides was observed only when both thiol groups were
reduced. For example, the linear precursor to peptide 11, which
has an unprotected N-terminal thiol group and protected linker
thiol group, could be cyclized only in the presence of TCEP.
Thus, the N-terminal thiol attacks only the thioester that is
generated by deprotection of the linker and subsequent acyl shift
(see Supporting Information).
Using the general procedure shown in Scheme 1B, we were
able to generate all but one of the desired thiolactone peptides
in good yield and purity (Table 1 and Supporting Information).
The sole exception was peptide 8, which contains N-methyl-
phenylalanine at its C-terminus. Attempts to synthesize the linear
precursor to this peptide by Fmoc-SPPS on linker 16 led to
low yields, and the crude mixture contained significant amounts
of deletion products lacking the last two residues, Leu and
N-Me-Phe. We attribute these problems to diketopiperazine
formation resulting from the presence of the tertiary amide at
N-Me-Phe, a problem that is well-known for C-terminal proline-
containing peptides prepared by Fmoc-SPPS16 and that has been
reported as a side reaction in the synthesis of N-methylated
peptides.17 Consequently, the R-thioester precursor to peptide
8 was generated using a Boc-SPPS strategy employing in situ
neutralization chemistry9 during the chain assembly to suppress
DKP formation. Last, the lactam (10) analogue was prepared
by cyclizing a partially protected linear peptide in solution
followed by global deprotection.3
Structure-Activity Relationships of AIPs. A cell-based
reporter assay was used to determine the inhibitory activity of
each peptide against the receptor histidine kinase, AgrC.6
Briefly, early exponential phase S. aureus reporter cultures of
the indicated group were treated with AIP analogues 1-11 in
the presence of a group-specific AIP agonist at a concentration
(125 nM) above the EC90,3,6 and the extent of agr activation
was determined using a â-lactamase reporter with a colorimetric
readout. To test both cognate and non-cognate agr inhibition
by the trAIP-II analogues, group II and I reporter cell lines
RN9372 and RN9222 were used, respectively. These strains lack
the ability to produce AIPs but contain the AgrC-AgrA two
component signaling cassette as well as â-lactamase driven by
the agr-dependent P3 promoter. Representative dose-response
curves for peptide 1 are shown in Figure 3, and the IC50 values
for all the peptides against cognate (group II) and non-cognate
(group I) strains are listed in Table 1. Importantly, the measured
Figure 2. Cyclization reaction of crude truncated AIP-II (1′). (A) HPLC
chromatograms of the reaction mixture at 0, 5, 80, and 230 min time points,
from top to bottom. The gradient used was 15-90% B (see Experimental
Procedures) in all four runs. The starting material appears as two peaks,
corresponding to two diastereomers of the linker. At least three of the
reaction intermediates correspond to the fully reduced linear peptide,
indicating that the linear thioester is present. The linker is cleaved during
the cyclization to give one product peak. (B) Mass spectra of the peaks
indicated by the corresponding symbols or numbers. 877 Da corresponds
to the mass of the linear starting material 1′, 701 Da corresponds to the
fully reduced peptide, and 580 Da is the mass of the cyclic product, 1.
linker products. The top panel in Figure 2A shows the crude
cleavage mixture obtained for the linear precursor to peptide 1,
which we will refer to as 1′. We reasoned that the two products
likely corresponded to the two expected diastereomers that result
from use of the racemic linker. No attempts were made to purify
these putative stereoisomers since the linker would eventually
be cleaved off during the cyclization reaction. Indeed, incubation
of crude peptide 1′ in an aqueous buffer containing the reducing
agent TCEP resulted in the collapse of the two peaks into a
single new peak with a mass consistent with the desired
1
thiolactone peptide (Figure 2 and Table 1). Homonuclear H
2-D NMR analysis confirmed that the reaction product was
peptide 1 (full chemical shift assignments of all peptides
prepared in this study are given in the Supporting Information).
Note, the 2-D 1H- TOSCY spectrum of purified 1 was identical
to an authentic sample of trAIP-II prepared by our previous
Boc-SPPS route (see Supporting Information).
The macrocyclization reaction was found to be the most
efficient between pH 6.6-6.8, taking ∼1-2 h to reach comple-
tion. If the pH was lowered, we observed mostly a reduced
version of peptide 1′ with decreasing amounts of product
formation, presumably due to slower O to S acyl shift, while at
higher pH values, hydrolysis to give the free R-carboxy-peptide
became increasingly problematic. The use of TCEP as the
reducing agent was also critical since it did not react with the
thioester-containing product to an appreciable degree. In
contrast, use of thiol-based reducing agents led to the generation
of the corresponding linear R-thioester peptides in addition to
(15) Jencks, W. P. Catalysis in Chemistry and Enzymology; McGraw-Hill: New
York, 1969.
(16) Pedroso, E.; Grandas, A.; de las Heras, X.; Eritja, R.; Giralt, E. Tetrahedron
Lett. 1986, 27, 743.
(17) Khosla, M. C.; Smeby, R. R.; Bumpus, F. M. J. Am. Chem. Soc. 1972, 94,
4721.
9
4920 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

Inhibition of Virulence
A R T I C L E S
Table 1. Expected and Observed Masses of All Peptides Used in This Study as Well as Their Inhibitory Activities against Both
Non-cognate (AgrC-I) and Cognate (AgrC-II) Receptors
exptl mass
obsd mass (M
+
H+)
IC50 (nM), AgrC-I (95% CI)
IC50 (nM), AgrC-II (95% CI)
1
1*a
2
3
4
579.2
579.2
549.2
549.2
519.2
593.3
593.3
593.3
593.3
504.2
562.3
522.2
580.0
579.0
550.7
550.0
520.7
595.0
595.0
595.0
595.0
505.0
564.0
524.0
90, (78-104)
260, (95-695)
189, (151-237)
40, (33-48)
293, (245-350)
230, (190-270)
1160, (855-1560)
194, (167-225)
77, (66-89)
238, (170-334)
5
6
639, (534-765)
319, (284-357)
7880, (4100-1.52 × 104)
1050, (795-1380)
7b
8b
9
1141, (799-1630)
1390, (1250-1560)
1680, (1310-2170)
5.07 × 105, (5.41 × 104 to 4.76 × 106)
1.67 × 105, (9.64 × 104 to 2.88 × 105)
4.73 × 104, (3.37 × 104 to 6.63 × 104)
10
11
a 1* are data obtained for peptide 1 made by Boc chemistry. b IC50 values for peptides 7 and 8 could not be determined due to lack of inhibition up to
the highest concentrations tested (>30 µM).
methylation of the serine residues (5 and 6) was for the most
part tolerated, although cognate inhibition was somewhat
sensitive to N-methylation of serine 2. In contrast, N-methylation
of Leu or Phe residues (7 and 8) resulted in a complete loss of
activity, further highlighting the functional importance of this
region of the molecule.
The replacement of the sulfur with an NH, to give lactam
10, had a similar effect on function as a replacement of both
serines with a methylene linker to give 9 (Table 1). Thus,
although the thiolactone linkage is not necessary for inhibition
by full-length AIP-II, the lactam substitution is not tolerated
well in the context of the truncated AIP. The relatively low
activity of peptide 9 cannot be explained by the loss of the serine
side chains in accordance with the activity data for peptide 4
(Table 1). It is more likely that the complete loss of one amide
bond or the increased flexibility introduced by the methylene
linker is responsible for the reduced potency.
Figure 3. Inhibition curves for truncated AIP-II (1) in the presence of 125
nM agonist. Vmax is the initial rate of nitrocefin cleavage by the â-lactamase
reporter measured by absorbance of the hydrolyzed lactam. The error bars
show the standard deviation from the mean for the two Vmax values obtained
for each concentration point.
Last, the removal of the N-terminal amide group caused a
dramatic loss of inhibitory activity against both cognate (∼250-
fold) and non-cognate (∼55-fold) strains. This result shows that
the N-terminal vestige of the macrocyclic ring is in fact
important for inhibition.
IC50 values for peptide 1 prepared by our new Fmoc-SPPS route
were in good agreement with those previously determined for
this peptide.4
There are several interesting trends within the activity data
summarized in Table 1. First, the overall rank order of the
peptides, in terms of their inhibitory activities, is remarkably
similar for the cognate and non-cognate agr groups, although
there are nuances to this that we comment on next. Second, all
of the peptides that had a measurable activity are more potent
antagonists of the non-cognate AgrC receptor than of the cognate
receptor. Furthermore, this difference in inhibitory activities
against the two receptors increases significantly for the less
potent analogues. This trend is more clearly seen in Figure 4,
which shows a log-log plot of IC50 values. Thus, antagonism
of the non-cognate agr group is much more tolerant to changes
in the AIP structure than is antagonism of the cognate group.
For instance, replacement of the thiolactone in 1 with a lactam
(10) leads to a modest decrease in the non-cognate IC50 (∼15-
fold) relative to the effect of this change on the cognate activity
(∼500-fold). Similarly, replacement of the two serine residues
in 1 with a methylene linker (9) resulted in a ∼13-fold versus
1700-fold reduction in non-cognate and cognate activities,
respectively.
We next explored as to whether the differential activity of
our analogues in any way correlated with the solution structure
of the peptides, as measured by 1H NMR chemical shifts around
the macrocycle. Figure 5A shows a plot of the global CSI for
each analogue 2-11 as compared to the parent peptide 1, in
order of peptide potency. While this simple analysis does not
report on the receptor bound conformation of the peptides, we
nonetheless saw no clear trend between potency and overall
chemical shift perturbations. For instance, N-methylated peptides
7 and 8, which have no measurable activity, have CSI values
comparable to peptides 3 and 4, which are the most potent AIP
analogues. Similarly, the chemical shifts around the macrocycle
in peptide 11 are largely unperturbed as compared to 1, yet this
analogue is one of the least potent antagonists. One interpretation
of the previous information is that the exocyclic and two
C-terminal amides of trAIP-II, which are perturbed in peptides
11 and 7-8, respectively, make specific contacts with the AgrC
receptor. However, we cannot rule out the possibility that these
modifications prevent the AIP from assuming a bioactive
conformation required for receptor binding. The largest CSI
observed is for the lactam analogue 10. In this case, chemical
shift changes propagate around the entire macrocycle (Figure
5 and Supporting Information), indicating that this substitution
Modifications to the serine residues in peptide 1 were well-
tolerated. Both the single and the double glycine mutants (2-
4) resulted in a minimal loss of activity (<5-fold) and in some
cases actually increased potency (Table 1). Moreover, N-
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4921

A R T I C L E S
George et al.
Figure 4. Log vs log plot of inhibitory activities of trAIP-II and derivatives against AgrC-II and AgrC-I. All peptides fall below the dotted black line (y
) x), indicating that all are more active against AgrC-I than --II. The difference in activity against the two receptors increases as the inhibitory activity
decreases. The error bars represent 95% confidence intervals for each IC50 value.
induces far-reaching changes in the solution structure of the
peptide. A similar effect was previously observed upon substitu-
tion of the thiolactone to a lactam in the context of full-length
AIP.3,6 The full-length lactam remained a potent cross-group
agr antagonist but was an extremely weak agonist of the cognate
agr group. Thus, whether it is agonism in the case of the full-
length AIP lactam or antagonism in the case of truncated lactam
10, the structural perturbation caused by the lactam substitution
appears to significantly reduce binding to the cognate AgrC. In
contrast, the presence or absence of the tail residues appears to
greatly modulate the affect of the structural perturbation induced
by the lactam on binding to the non-cognate receptor, AgrC-I.
bond in these peptides; however, we were unable to make full
assignments for the minor, presumably cis, isomers of 7 and 8
owing to their low abundance.
Discussion
In this study, we developed an efficient synthesis of S. aureus
AIPs by employing an Fmoc-SPPS route to peptide R-thioesters.
This methodology is based on the handle design of Botti et al.12
but extends this latent thioester linker concept to any resin type.
We have shown that the thiolactone-containing AIP structure
can be obtained by simply incubating the corresponding
protected peptide R-oxyester precursor in aqueous buffer in the
presence of a non-nucleophilic reducing agent TCEP (Figure
2). Collectively, our data support the cascade-type mechanism
shown in Scheme 1B, in which the reducing agent first triggers
the O-to-S acyl transfer at the C-terminus, thereby setting up a
transthioesterification reaction between the nascent R-thioester
and the cysteine sulfhydryl to give the thiolactone, with
concomitant expulsion of the linker. Although not the focus of
the current study, we expect that linker 16 will be generally
useful for the preparation of peptide R-thioesters for use in native
chemical ligation19 and expressed protein ligation20 studies.
Indeed, in other work, we have successfully used this linker in
the Fmoc-SPPS synthesis of phosphopeptide R-thioesters for
use in protein semisynthesis (T. W. Muir, unpublished results).
1
The H NMR spectra of N-methylated peptide 6 contained
twice the number of resonances expected for a peptide of this
size. Indeed, using standard 2-D sequential assignment methods,
we found that the sample contained two species, which were
present as a 1:1 mixture in DMSO-d6 at 298 K (Figure 5B).
There was no evidence for chemical heterogeneity in this peptide
by LC-MS analysis (see Supporting Information). Likewise, we
ruled out the possibility of these signals resulting from epimer-
ization (i.e., C-terminal epimers) of the peptide since the same
batch of amino acylated linker resins was used for the synthesis
of other peptides in the study, and in these cases, no extra NMR
resonances were observed, for example, peptides 3 and 4.
Further analysis of the ROESY spectrum of 6 revealed the
presence of a strong NOE cross-peak between the two serine
alpha protons in one of the species but not the other (Figure
5C). This cross-peak is consistent with the presence of a cis-
amide bond between Ser2 and N-Me-Ser3. Analogous to the
prolyl-tertiary amide, N-methylation is known to increase the
propensity for trans-cis isomerization of an amide bond.18 Two
of the other three N-methylated peptides, 7 and 8, were also
present as two conformers as indicated by NMR, with the minor
conformer comprising only 10-30% of the total sample
population. By analogy with peptide 6, these conformers almost
certainly stem from cis-trans isomerization of the tertiary amide
The staphylococcal agr response is responsible for integrating
growth-dependent signals (i.e., quorum sensing) with a complex
genetic network to coordinate the expression of a large number
of virulence factors.1 Previously, we have shown that cross-
group antagonism of the agr response by non-cognate AIPs can
be used to attenuate the spread of an infection in animal models.3
Thus, understanding the minimal determinants of inhibition by
AIP analogues is of some interest.21 The ability to prepare AIPs
(19) Dawson, P. E.; Muir, T. W.; Clark-Lewis, I.; Kent, S. B. Science
(Washington, DC, U.S.) 1994, 266, 776.
(20) Muir, T. W.; Sondhi, D.; Cole, P. A. Proc. Natl. Acad. Sci. U.S.A. 1998,
95, 6705.
(18) LaPlanche, L. A.; Rogers, M. T. J. Am. Chem. Soc. 1964, 86, 337.
(21) Gorske, B. C.; Blackwell, H. E. Org. Biomol. Chem. 2006, 4, 1441.
9
4922 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008

Inhibition of Virulence
A R T I C L E S
Figure 5. (A) CSI for all peptides, which are depicted in order of decreasing activity from left to right (6 trans and cis are bracketed to indicate that they
do not have two separate activities, as they are two conformers of one molecule). CSI is the average change in chemical shift relative to the NMR spectrum
for 1 (see Experimental Procedures). The chemical shifts of the modified residue are excluded; thus, CSI values represent overall changes in the macrocycle
beyond the site of modification. (B) Amide region of the ROESY spectrum of 6, color-coded to identify the spin systems that correspond to each isomer.
The blue and red peaks correspond to the trans and cis isomers, respectively. (C) Alpha region of the ROESY spectrum of 6, color-coded as in panel B.
via the Fmoc-SPPS route described previously allowed us to
perform a systematic SAR analysis of truncated AIP-II (1),
which was previously found to inhibit S. aureus virulence of
all four agr groups.4
Several interesting trends emerged from this analysis. First,
the overall rank order of the analogues for cognate (AgrC-II)
and non-cognate (AgrC-I) inhibition were almost identical,
despite the fact that the sensor domains of AgrC-II and AgrC-I
share only 31% sequence identity. However, non-cognate
antagonism was more resistant to changes in AIP structure than
was cognate antagonism; the former values varied by no more
than ∼50-fold, whereas the latter values varied by over 3 orders
Figure 6. Peptide 1, color-coded to indicate which portions of the structure
are critical for inhibition of cognate (AgrC-II) and/or non-cognate (AgrC-
I) receptors.
of magnitude (Figure 4). This observation is broadly in line
with previous SAR studies on full-length AIPs, which revealed
that cross-group antagonism is more tolerant to changes in
peptide structure than intra-group agonism.6 Pharmacological
studies of the trAIP-II analogue have suggested that both cognate
and non-cognate antagonisms operate through a simple competi-
tive binding mechanism in which trAIP-II competes with the
group-specific AIP agonist for a common binding site on the
sensor domain of AgrC.8c The present SAR data are certainly
consistent with this mechanism, although the details of the
cognate and non-cognate interactions must be subtly different
since certain modifications have a more acute effect on the
former interaction than the latter (e.g., peptides 9-11).
9
J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008 4923

A R T I C L E S
George et al.
This study has allowed the principle pharmacophore for agr
antagonism by trAIP-II, against both cognate and non-cognate
AgrC receptors, to be identified. As summarized in Figure 6,
one-half of the molecule that is defined by the cysteine and the
two C-terminal hydrophobic residues is critical for activity. In
contrast, the remainder of the molecule appears to be far less
important for this activity and, in the case of non-cognate agr
inhibition, can even be replaced by an alkyl linker without a
dramatic loss of activity. This information will no doubt be
useful in guiding future medicinal chemistry initiatives on this
system.
York State Office of Science, Technology, and Academic
Research. The Q-TOF Ultima mass spectrometer (University
of Illinois at Urbana-Champaign) was purchased in part with
a grant from the National Science Foundation, Division of
Biological Infrastructure (DBI-0100085). This work was sup-
ported by the U.S. National Institutes of Health (Grant R 501
AI42783).
Supporting Information Available: Plots of inhibitory activi-
ties of all peptides. Experimental evidence for proposed mech-
anism of thiolactone formation. HPLC and MS characterization
Acknowledgment. We are very grateful to M. Goger for
assistance with the acquisition of ROESY and TOCSY NMR
spectra. We also thank Y. Wei for help with TOCSY NMR.
Thanks to M. Vila-Perello, J. Cisar, and M. S. Jensen for helpful
discussions. T.W.M. is a member of the New York Structural
Biology Center, which is a STAR center supported by the New
1
and NMR chemical shift assignments of all peptides. H and
13C NMR spectra of linker 16. This material is available free
of charge via the Internet at http://pubs.acs.org.
JA711126E
9
4924 J. AM. CHEM. SOC. VOL. 130, NO. 14, 2008