Binding of peptides in solution by the Escherichia coli chaperone PapD as revealed using an inhibition ELISA and NMR spectroscopy

Article abstract of DOI:10.1016/S0968-0896(98)00162-X

PapD is the prototype member of a family of periplasmic chaperones which are required for assembly of virulence associated pili in pathogenic, gram-negative bacteria. In the present investigation, an ELISA has been developed for evaluation of compounds as inhibitors of PapD. Synthetic peptides, including an octamer, derived from the C-terminus of the pilus adhesin PapG were able to inhibit PapD in the ELISA. Evaluation of a panel of octapeptides in the ELISA, in combination with NMR studies, showed that the peptides were bound as extended β-strands by PapD in aqueous solution. The PapD-peptide complex was stabilized by backbone to backbone hydrogen bonds and interactions involving three hydrophobic peptide side chains. This structural information, together with previous crystal structure data, provides a starting point in efforts to design and synthesize compounds which bind to chaperones and interfere with pilus assembly in pathogenic bacteria. Copyright (C) 1997 Elsevier Science Ltd.

Full text of DOI:10.1016/S0968-0896(98)00162-X

BIOORGANIC &
MEDICINAL
CHEMISTRY
Bioorganic & Medicinal Chemistry 6 (1998) 2085±2101
Binding of Peptides in Solution by the Escherichia coli
Chaperone PapD as Revealed Using an Inhibition ELISA
and NMR Spectroscopy
Katarina Flemmer Karlsson,a Bjorn Walse,b Torbjorn Drakenberg,b
Sarbari Roy,a Karl-Erik Bergquist,a Jerome S. Pinkner,c Scott J. Hultgrenc
and Jan Kihlbergd,*
aOrganic Chemistry 2, Center for Chemistry and Chemical Engineering, Lund University, PO Box 124, S-221 00, Lund, Sweden
bPhysical Chemistry 2, Center for Chemistry and Chemical Engineering, Lund University, PO Box 124, S-221 00, Lund, Sweden
cDepartment of Molecular Microbiology, Washington University School of Medicine, Box 8230, St. Louis, MO 63110-1093, USA
d
Ê Ê
Organic Chemistry, Umea University, S-901 87 Umea, Sweden
Received 23 March 1998; accepted 18 June 1998
AbstractÐPapD is the prototype member of a family of periplasmic chaperones which are required for assembly of
virulence associated pili in pathogenic, gram-negative bacteria. In the present investigation, an ELISA has been
developed for evaluation of compounds as inhibitors of PapD. Synthetic peptides, including an octamer, derived from
the C-terminus of the pilus adhesin PapG were able to inhibit PapD in the ELISA. Evaluation of a panel of octapep-
tides in the ELISA, in combination with NMR studies, showed that the peptides were bound as extended b-strands by
PapD in aqueous solution. The PapD±peptide complex was stabilized by backbone to backbone hydrogen bonds and
interactions involving three hydrophobic peptide side chains. This structural information, together with previous
crystal structure data, provides a starting point in e€orts to design and synthesize compounds which bind to chaper-
ones and interfere with pilus assembly in pathogenic bacteria. # 1998 Elsevier Science Ltd. All rights reserved.
Introduction
®brillum joined to a thicker pilus rod.1 The mode of
binding of the PapG adhesin to the receptor-active dis-
Many gram-negative bacteria assemble extracellular
adhesive organelles termed pili on their surfaces that
allow them to colonize host tissue and give rise to
infections.1 This group of bacteria includes pathogens
that cause a variety of diseases, such as urinary tract
infection and the more severe pyelonephritis, meningitis,
whooping cough, pneumonia and diarrhoea.2 Pili are
rodlike, supramolecular protein ®bers that present
adhesins at their tip for speci®c attachment to receptors
found in the host.3,4 The P pilus of uropathogenic
Escherichia coli is the best characterized of all bacterial
pili and carries the adhesin PapG at the end of a thin tip
accharide,
a-d-galactopyranosyl-(1-4)-b-d-galactopyr-
anose [Gala(1-4)Galb], present in the globoseries of
glycolipids on kidney epithelium cells,5,6 has been
revealed in detail using synthetic receptor analogues.7±9
The P pilus tip ®brillar structure is composed of repeating
PapE subunits whereas the pilus rod consists of large
numbers of the major pilin subunit PapA. The adaptor
proteins PapF, PapK, and PapH join the di€erent parts
of the pilus to each other and anchor the pilus in the
outer cell membrane of the bacteria. All bacterial pilus
assembly requires the presence of periplasmic chaperones.
P pili are assembled by the chaperone PapD that binds
to and caps interactive surfaces on each pilus subunit thus
preventing premature aggregation during their secretion
into the periplasmic space.10±12 The crystal structure of
PapD has been solved at 2.5 A resolution and reveals
that PapD consists of two b-barrel domains oriented in
Key words: Pili; chaperone; ELISA; peptide; structure±activity;
NMR spectroscopy.
*Corresponding author. Tel.: 46 90 786 6890; fax: 46 90 13 88
85; e-mail: jan.kihlberg@chem.umu.se
0968-0896/98/$ - see front matter # 1998 Elsevier Science Ltd. All rights reserved
PII: S0968-0896(98)00162-X

2086
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
the shape of a boomerang.13 Furthermore, PapD has
been found to have signi®cant sequence homology
(25±56% identity) with chaperones that assemble other
types of pili suggesting that all chaperones have closely
related structures.2,14
(ELISA) in which compounds may be evaluated for
their ability to inhibit complex formation between PapD
and a fusion protein (MaltBP±PapG175-314)16 which
consists of the chaperone binding domain of the PapG
linked to the maltose binding protein. Evaluation of a
panel of synthetic peptides derived from the C-terminus
of P pilus proteins in the assay resulted in the identi®-
cation of a minimal inhibitory octapeptide. Investiga-
tion of alanine and serine scans of this octapeptide in
the ELISA, in combination with NMR studies of two
peptide±PapD complexes, revealed structural details
responsible for the formation of complexes between
PapD and peptides in aqueous solution.
The C-terminus of the P pilus proteins has been indi-
cated to constitute an essential part of the epitope on
each subunit that is recognized by the PapD chaper-
one.10,12,15,16 Thus, truncation of the C-terminal 13
residues in the adhesin PapG abolished complex forma-
tion with PapD.10 Moreover, synthetic 19-mer peptides
from the C-terminus of P-pilus proteins bound to PapD
and some of these peptides also inhibited chaperone
mediated folding of the adhesin PapG in an in vitro
assay.15 The crystal structure of PapD bound to the C-
terminal 19-mer peptide from PapG (PapG296-314),
determined at 3.0 A resolution, suggested a structural
basis for chaperone-subunit interactions.15 It was found
that the peptide was anchored by hydrogen and ionic
bonding of the C-terminal carboxylate group to residues
Arg8 and Lys112, which are found in the cleft of PapD
and are invariant in the chaperone superfamily. Seven
main-chain hydrogen bonds were also formed between
the peptide backbone and the G1 b-strand on the surface
of PapD resulting in a parallel b-strand interaction.
Results
Synthesis of pilus-related peptides and a reduced
analogue
To enable the present investigation, a number of pep-
tides derived from the C-terminus of the proteins which
make up the P pilus have been prepared (peptides are
denoted by their pilus protein origin and the fragment
they represent within this protein; sequences are shown
in Table 1). These synthetic peptides include the ®ve 19-
mer peptides (PapG296-314, PapF130-148, PapE131-
149, PapK139-157, and PapH155-173)18 as well as the
length series from PapG (ranging in length from C-
terminal 19- to 6-mer) and PapH (C-terminal 22- to
8-mer). The C-terminal octamer peptides from the six
pilus subunits, alanine- and serine-scan series of the
octapeptide PapG307-314, and several controls have
also been prepared. All peptides were prepared on solid
phase according to the Fmoc strategy using a fully
automatic peptide synthesizer19 constructed in our
laboratory. As reported previously, use of a polystyrene
resin grafted with polyethylene-glycol spacers (TentaGel
PEG PS resin) was important for the successful pre-
paration of the 19-mer pilus subunit derived peptides.18
After TFA-catalyzed cleavage from the solid phase, the
peptides were puri®ed to greater than 95% homogeneity
by reversed-phase HPLC and their structures were con-
®rmed by NMR spectroscopy (selected peptides only)
and mass spectrometry. Peptides corresponding to
sequences located internally in pilus proteins (i.e.
PapG299-306, PapG296-313, PapG306-313, and
PapH155-165) were prepared as C-terminal amides;
whereas all other peptides had a free C-terminal carboxyl
group. The N-terminal amino group of the peptides was
kept free in all cases, in order to improve solubility.
Furthermore, the hydrophobic side chains of Phe313
,
Leu311, Met309, and Met307 in the peptide made signi®cant
contact with PapD. Interestingly, these hydrophobic
residues are part of a conserved pattern of alternating
hydrophobic and hydrophilic residues present in the
C-terminus of pilus proteins.15 However, within the
crystal, the PapD-peptide b-sheet was extended even
further, since a second PapD-peptide complex was
placed adjacent to the ®rst, so that the two bound pep-
tide chains interacted as antiparallel b-strands. It can
not be ruled out that the structure of the complex was
signi®cantly in¯uenced by this dimerization.
In addition to providing information on how chaper-
ones recognize pilus proteins, studies of the interactions
between synthetic peptides and chaperones provide
structural information that may be used for design and
synthesis of ligands that bind to chaperones. By inhi-
biting pilus assembly, such compounds would interfere
with bacterial attachment to the host; therefore, they
constitute potential novel antibiotics. It was recently
pointed out that the development of compounds that
interfere with bacterial protein secretion (for instance, in
pilus assembly) constitutes an attractive approach to
overcome widespread bacterial resistance to existing
antibiotics.17 However, such e€orts have been hampered
by the lack of bioassays, which properly evaluate com-
pounds as inhibitors of the critical complex formation
between chaperone and pilus protein.15 Therefore, we
have developed an enzyme-linked immunosorbent assay
The peptide PapG307-314red (8), in which the a-car-
boxyl group of Pro314 has been reduced (Scheme 1),
was prepared in order to probe the importance of the
C-terminal carboxyl group for anchoring to PapD.

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2087
Table 1. Amino acid sequences for the C-terminus of the proteins PapG, F, E, K, A, and H that are assembled into the E. coli P pilus
Position from C-terminus
Peptidea
19 18
17 16 15 14 13 12
11 10
9
8
7
6
5
4
3
2
1
PapG296-314
PapF130-148
PapE131-149
PapK139-157
PapA145-163b
PapH155-173
Residue typec
G
G
Q
K
A
K
K
I
R
L
K
N
I
P
G
A
P
G
G
G
G
G
G
Ce
E
D
P
L
F
F
Y
F
Y
S
Q
S
G
T
S
T
T
T
V
V
M
A
A
A
A
L
T
S
M
M
L
V
S
L
M
A
L
S
I
F
Y
Y
Y
Y
Y
P
N
S
N
S
L
A
A
A
A
T
T
N
G
V
E
N
R
S
V
V
T
E
D
A
N
E
S
F
T
T
D
R
Q
E
A
K
V
E
A
F
L
L
Hd
F
F
V
Hd Pf
Pf Hd
Hd Pf Hd Pf
a The residues in the peptides are numbered as in the corresponding pilus subunits, which lack the signal sequence.
b Synthetic peptide not available due to diculties in puri®cation.
c Only positions where at least ®ve out of six residues belong to the same type have been included.
d H=Hydrophobic aromatic or aliphatic residue.
e C=Conserved glycine.
f P=Polar or charge residue.
prolinol (2), promoted by O-(benzotriazol-1-yl)-1,1,3,3-
tetramethyluronium tetra¯uoroborate (TBTU), to give
3. Substitution of the hydroxyl group in 3 was accom-
plished by treatment with iodine, triphenylphosphine,
and imidazole20 to give the iodide 4, which on catalytic
hydrogenation gave 5. Substantial cis/trans isomerism
(20±40% cis) was observed for the amide bond in com-
pounds 3±5, as revealed by the characteristic dad(i,i+1)
NOE observed between Phe-Ha and Pro-Hd for the
trans-form and the daa(i,i+1) NOE between the Phe
and Pro Ha protons of the cis-form. After cleavage of
the Boc-protective group, reduced dipeptide 5 was cou-
pled with Fmoc-protected fragment 6 using the highly
active coupling reagent, O-(7-azabenzotriazol-1-yl)-
1,1,3,3-tetramethyluronium
hexa¯uorophosphate
(HATU)21 in combination with 1-hydroxy-7-azabenzo-
triazole (HOAt)22 and collidine to give 7. Collidine has
been recommended for use in peptide-fragment coupling
reactions to avoid racemization of the activated, C-
terminal amino acid.23 Finally, cleavage of the Fmoc
group in 7 by treatment with piperidine gave 8.
Peptides as inhibitors of complex formation between
PapD and the PapG adhesin
The PapD±MaltBP±PapG175-314 inhibition ELISA. In
order to allow investigations of the ability of various
compounds, such as peptides derived from the subunits
of the E. coli P pilus, to inhibit complex formation
between the PapD chaperone and pilus proteins, an
enzyme-linked immunosorbent assay (ELISA) was
developed. This ELISA was based on the MaltBP±
PapG175-314 fusion protein16 that consists of the 140
C-terminal amino acids of the PapG adhesin linked to
the maltose binding protein. In contrast to PapG itself,
the MaltBP±PapG175-314 fusion protein can be puri®ed
Scheme 1. (a) TBTU, N-methylmorpholine, DMF (77%); (b)
I2, Ph3P, imidazole, toluene, 80 ꢀC (64%); (c) H2, Pd/C, Et3N,
EtOAc:EtOH 1:5 (93%); (d) HCl, HOAc; (e) HATU, HOAt,
trimethylpyridine, DMF, 0 ꢀC!rt (53%); (f) 20% piperidine in
DMF (ꢁ26%).
Synthesis of 8 was achieved by coupling of the reduced
Phe-Pro analogue 5 with hexapeptide fragment 6. The
dipeptide 5 was prepared by condensation of Na-tert-
butoxycarbonyl (Boc) protected phenylalanine (1) with

2088
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
in native form in the absence of PapD. Furthermore, the
complex between PapD and MaltBP±PapG175-314 has
been shown to be as stable as the native PapD±PapG
complex,16 which is the most stable of the complexes
formed between PapD and the proteins that make up
the P pilus. This argues that the ELISA is a relevant
model for studies of ligands that bind to the PapD
chaperone.
In the inhibition ELISA microtiter-plate-wells were
coated with the MaltBP±PapG175-314 fusion protein
and then blocked with BSA. Peptides to be investigated
as inhibitors were dissolved in DMSO and preincubated
with the PapD chaperone in PBS (DMSO was required
as a cosolvent due to the low solubility of most of the
peptides in aqueous solutions). The coated and blocked
microtiter-plate-wells were then incubated with the
peptide±PapD solution. PapD bound to MaltBP±
PapG175-314 in the wells was detected by addition of
anti-chaperone antibody (rabbit IgG) followed by goat
antiserum to rabbit IgG conjugated to alkaline phos-
phatase. As compared to
a
preliminary report24
describing the inhibition ELISA, all incubation times
and the protein-stabilizing agent used during incubation
of the coated wells with peptide±PapD solution were
optimized. This resulted in lower background readings
as well as higher and substantially more reproducible
levels of inhibition for the peptides.
Peptides from the C-terminus of the adhesin PapG and
the membrane anchor PapH inhibit complexation of
PapD to PapG
The C-terminal 19-mer peptides from the pilus adhesin
PapG and the membrane anchor protein PapH
(PapG296-314 and PapH155-173, cf. Table
1 for
sequences) were both potent inhibitors of complex
formation between PapD and MaltBP±PapG175-314
(Fig. 1A). The inhibitory power of PapG296-314
reached a maximum at a peptide/PapD ratio of ꢁ100
and inhibition then decreased as the ratio was increased.
This reduction in inhibitory power was assumed to be
due to limited solubility of PapG296-314 under the
conditions of the ELISA, aggregation of the peptide or
peptide±PapD complex, or combinations of these phe-
nomena. The 19-mer peptides PapE131-149 and
PapK139-157, which are derived from proteins that
make up the ¯exible pilus tip, did not show any sig-
ni®cant inhibitory power. PapF130±148, also derived
from a pilus tip protein, gave highly irreproducible
results in the ELISA and often induced increased binding
of PapD to MaltBP±PapG175-314 (data not shown). The
e€ect was tentatively explained as originating from for-
mation of peptide aggregates that promote complexation
of PapD and MaltBP±PapG175-314. The 19-mer pep-
tide from the C-terminus of PapA, the subunit making
Figure 1. Inhibition of binding of PapD to the MaltBP±
PapG175-314 fusion protein using (A) 19-mer peptides from
the C-terminus of P pilus proteins, (B) a length series of peptides
derived from the C-terminus of the pilus adhesin PapG, (C) a
length series of peptides derived from the C-terminus of the pilus
anchor PapH and (D) the octapeptide PapG307-314 and two
peptide controls (error bars represent 95% con®dence limits).

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2089
up the bulk of the pilus rod, could not be satisfactorily
puri®ed after synthesis and was therefore not available
for investigation in the ELISA.
a di€erent order, and ProtG16-23 (Thr-Leu-Lys-Gly-
Glu-Thr-Thr-Thr), derived from a part of protein G
that interacts as an antiparallel b-strand with immu-
noglobulin Fab fragments.25 Both these peptides lacked
inhibitory power (Figure 1D), further con®rming that
the side chains of peptides from PapG, in addition to
hydrogen bonds from the peptide backbone, form
important contacts with the PapD chaperone.
The minimum length of peptides from PapG and PapH
required for inhibition of binding of PapD to MaltBP±
PapG175-314 was investigated using two length series of
peptides, all members of which had the C-terminal resi-
due in the pilus protein as their last residue. The PapG
series consisted of the 19-, 16-, 11-, 8-, 7-, and a 6-mer;
whereas, the PapH series included the 22-, 19-, 16-, 11-,
and 8-mer. In the PapG series, the inhibitory power
increased successively as the peptide length increased
(Fig. 1B). Interestingly, the octapeptide PapG307-314
was signi®cantly more potent than the heptapeptide
PapG308-314; whereas, the di€erences in inhibitory
power between the octapeptide and PapG304-314 (11-
mer) and between the 11-mer and PapG299-314 (16-
mer) were minor and not signi®cant. However, the
inhibitory power of the 16-mer PapG299-314 was sig-
ni®cantly higher than that of octapeptide PapG307-314.
The peptide PapG299-306, representing an internal
PapG sequence, was not able to inhibit complex forma-
tion between PapD and MaltBP±PapG175-314 (data
not shown). Altogether, these results show that octa-
undecamer peptides derived from the very C-terminus of
PapG contain the essential structural features required
for binding to PapD and inhibition of complex forma-
tion with the PapG adhesin. In the PapH series, only the
19-mer PapH155-173, and the even longer 22-mer
PapH152-173, were able to inhibit complex formation
between PapD and MaltBP±PapG175-314 (Fig. 1C).
The shorter 8-mer (PapH166-173), 11-mer (PapH163-
173), and 16-mer (PapH158-173), as well as the 11-mer
PapH155-165 that represents an internal PapH
sequence (data not shown), were all devoid of activity.
Thus, in contrast to the octamer binding motif in
PapG, at least the 17 C-terminal amino acid residues
in PapH are required for inhibition of binding of PapD
to PapG.
Three hydrophobic side chains in the C-terminal octamer
peptide from PapG (PapG307-314) are important for
binding to PapD
The role of the side chains of the minimal inhibitory
peptide from PapG (PapG307-314) in binding to PapD
was investigated using an alanine and a serine scan ser-
ies of peptides. In the alanine scan series all residues of
PapG307-314 were replaced, one by one, by alanine,
whereas Ser312 and Thr308 were excluded in the analo-
gous serine scan series. Comparison of the inhibitory
power of the alanine and serine mutant at each position
in PapG307-314, with that of native PapG307-314,
revealed the role of that particular side chain for bind-
ing to PapD (Fig. 2). Some of the peptides containing
Ala mutations displayed a low solubility and/or a ten-
dency to aggregate which had to be taken into
account on evaluation of their inhibitory power. The
Ser mutants of PapG307-314 did not su€er from such
problems and results obtained with this peptide series
were therefore more readily interpreted.
The residues with even numbers in PapG307-314, i.e.
those not in contact with PapD in the crystalline com-
plex, do not appear to be important for binding to
PapD. Replacement of Pro314 with Ala or Ser did not
reduce the inhibitory power of PapG307-314 (Fig. 2A).
This was also found on substitution of Ser312 by Ala
(Fig. 2C). For Val310, the Ser mutant is equally potent
as the native peptide; whereas, the Ala mutant appears
to be inhibitory at low peptide/PapD ratios, but not at
higher ratios (Fig. 2E). As discussed for the 19-mer
PapG296-314, the decrease in inhibitory power of
the Val310!Ala substitution in PapG307-314 with
increasing concentration most likely originates from
low solubility and/or aggregation of the peptide in the
ELISA, rather than from a low intrinsic activity. The
role of Thr308 is less clear since the alanine mutant
peptide promotes binding of PapD and MaltBP±
The octamer C-terminal peptides from the remaining
four pilus subunits (PapF, PapE, PapK, and PapA)
were also analyzed for inhibitory activity in the ELISA.
However, with the possible exception of PapK150-157,
these peptides were not able to signi®cantly inhibit
complex formation between PapD and MaltBP±
PapG175-314 (data not shown). The lack of inhibitory
power displayed by the ®ve C-terminal octamer peptides
from the pilus proteins PapF, E, K, A, and H, as com-
pared to the corresponding octamer from PapG, indi-
cates the importance of speci®c interactions between
peptide side chains and PapD. This was further investi-
gated using the peptides PapG307-314random (Leu-
Met-Ser-Phe-Val-Met-Thr-Pro), which contains the
same amino acids as PapG307-314, but incorporated in
PapG175-314 (Fig. 2G).
A similar behavior was
observed for the 19-mer PapF130-148 and was tenta-
tively assumed to result from limited solubility or aggre-
gation of the peptide.
In contrast, three of the residues with odd numbers in
PapG307-314, which contact PapD in the crystalline
complex, were found to be important for interaction

2090
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
Figure 2. Inhibition of binding of PapD to the MaltBP±PapG175-314 fusion protein using an alanine and a serine scan series of the
octapeptide PapG307-314 (error bars represent 95% con®dence limits).

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2091
with PapD. Replacement of Phe313 by Ala did not lead
to a reduction in inhibitory power, because the drop in
inhibition at the highest peptide/PapD ratio was again
considered to result from low solubility and/or aggre-
gation (Fig. 2B). However, replacement of Phe313 by Ser
abolished the peptides inhibitory power, signaling that
Phe313 is involved in a weak hydrophobic interaction
with PapD. This conclusion was further supported by
the observation that substitution of Phe313 for Tyr led to
a reduction in inhibitory power (data not shown). For
Leu311 and Met309, both the Ala and Ser mutants were,
at best, weak inhibitors revealing the importance of
these residues for binding to PapD (Fig. 2D and F).
Met307 does not appear to be essential for binding to
PapD since the Ser mutant is equally potent as native
PapG307-314, whereas the irregular behavior of the Ala
mutant presumably re¯ects low solubility or a tendency
to aggregate (Fig. 2H).
binds to PapD with these three residues in their respec-
tive binding sites on the chaperone. Consequently,
interactions between the peptide C-terminus and Arg8
and Lys112 in PapD are not possible.
1H NMR studies of 19-mer peptides from the C-terminus
of pilus proteins
1H NMR studies were undertaken to investigate if dif-
ferent conformational preferences between the 19-mer
peptides PapG296-314, PapE131-149, PapK139-157,
and PapH155-173 could explain the di€erences in their
inhibitory power. The latter three peptides were inso-
luble in aqueous solution at a pH close to 6 and the
NMR studies were therefore carried out in DMSO at
27 ꢀC. DMSO was chosen since all peptides employed in
the present investigation were evaluated in the ELISA in
mixtures of water and DMSO, where the DMSO con-
tent ranged from 1 to 50%. However, it should be
pointed out that the conformational preferences in
DMSO might be di€erent from those in aqueous
DMSO, making conclusions only tentatively valid under
the conditions used in the ELISA. In DMSO, sequence-
speci®c assignments of the residues in the peptides were
obtained based on COSY, TOCSY, ROESY, and
NOESY spectra according to standard procedures.26
Two sets of resonances were observed in the NMR
spectra of PapG296-314 and PapE131-149, which ori-
ginated from cis/trans isomerism about the amide
bond between Phe313 and Pro314 in PapG296-314 and
between Gly136 and Pro137 in PapE131-149, respectively.
The major trans forms displayed the characteristic
NOEs between Pro Hd and Ha of the preceding residue,
whereas the minor cis forms displayed the equally char-
acteristic NOEs between Pro Hd and Ha of the preced-
ing residue. Integration in the one-dimensional 1H
NMR spectra showed the cis content to be ꢁ20% for
PapG296-314 and 25±30% for PapE131-149. PapG296-
314 also contains a proline at position 300, whereas
PapK139-157 has a proline at position 143. The amide
bonds to both of these prolines existed predominantly in
the trans form.
Anchoring of the C-terminal carboxyl group of peptides
to invariant residues in PapD is not critical
The importance of `anchoring' the C-terminal carboxyl
group of an inhibitory peptide to Arg8 and Lys112 in the
binding site of PapD by hydrogen bonding and by a salt
bridge was probed in two di€erent ways. The peptide
PapG307-314red (cf. 8, Scheme 1), in which the C-
terminal carboxyl group has been reduced to a methyl
group, was found to be a slightly less powerful inhibitor
of complex formation between PapD and MaltBP±
PapG175-314 than PapG307-314 (Fig. 3). This suggests
only a minor role for anchoring to Arg8 and Lys112 in
peptide±PapD complexes. This conclusion was further
supported using the peptide PapG306-313, which lacks
Pro314, and was at least as powerful an inhibitor as the
native octamer PapG307-314. In view of the importance
of the side chains Phe313, Leu311, and Met309 for inter-
action with PapD, it was assumed that PapG306-313
Strong or medium intensity Hai/NHi+1 NOE con-
nectivities, as well as medium intensity NHi/NHi+1
NOEs, were observed along most or all of the peptide
chains of the four peptides. No medium or long range
NOEs could be detected. Since b-strands display strong
Hai/NHi+1 NOEs and lack NHi/NHi+1 NOEs, whereas
the opposite is found for a-helices, these NOEs indicate
that the four 19-mer peptides are conformationally
averaged between both the aR and the b region of the
f,É-space.26,27 When compared to random-coil chemi-
cal shift values determined for model peptides,28 the Ha
protons in the four 19-mer peptides were shifted slightly
up®eld (less than 0.15 ppm), except for Lys299 and
Figure 3. Inhibition of binding PapD to the MaltBP±PapG175-
314 fusion protein using peptides from the adhesin PapG which
lack the C-terminal Pro314 (PapG306-313) or in which the C-
terminal carboxyl group has been reduced to a methyl group
(PapG307-314red) (error bars represent 95% con®dence limits).

2092
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
Phe313 in PapG296-314 and Gly136 in PapE131-149
which display down®eld shifts. The latter observation
may, however, well be explained by the observation that
proline has been found to induce a ꢁ0.3 ppm down®eld
shift for the preceding residue.29 Thus, both NOEs and
Ha chemical shifts indicate that the four peptides are
conformationally averaged between the aR and the b
region of the f,É-space in DMSO. This conclusion was
also reached in a previous study of the conformation of
form to be detected in the free form.32 In contrast to the
NMR studies of the 19-mer peptides, DMSO was not
used as solvent in these studies of PapG307-314 and
PapG308-314.
Assignment of all the proton resonances in the free form
of heptapeptide PapG308-314 was obtained from
ROESY and TOCSY spectra (Table 2). Two forms of
the peptide were observed, originating from cis±trans
isomerization of the amide bond between Phe313 and
Pro314. The major form had a trans peptide bond and
integration of the cross peaks for Pro314 in the TOCSY
spectrum showed a cis-content of 40%. A similar cis±
trans ratio has also been observed for the longer 19-mer
PapG296-314 in aqueous solution.18 Strong-intensity
sequential Hai/NHi+1 NOE connectivities were
observed along the whole peptide chain but no sequen-
tial NHi/NHi+1 NOEs or medium- or long-range NOEs
could be detected. Only small deviations from the Ha
proton chemical shifts of short random-coil peptides29
were found for PapG308-314. Altogether, these obser-
vations indicate that the backbone dihedral angles of the
PapG308-314 are averaged, but predominantly in the b
(extended chain) region of f,É-space.26,27
1
PapG296-314 in aqueous solution using H NMR spec-
troscopy.18
NMR studies of complexes between PapD and peptides
from PapG
The complexes between PapD and the C-terminal octa-
and heptapeptide from the pilus adhesin PapG
(PapG307-314 and PapG308-314, respectively) were
studied by two di€erent NMR techniques. Isotope-edi-
ted NMR spectroscopy30 was used to study the complex
between PapD and PapG307-314, in which all residues
except Pro314 were 15N-labeled. Unfortunately, the low
solubility of PapG307-314 in aqueous solution at pH 6.0
did not allow studies of the unbound form of the octa-
peptide. However, on addition of PapD the peptide was
dissolved thereby allowing investigation of its structure
when bound by PapD. The heptapeptide PapG308-314
was more soluble in aqueous solution but did not bind
to PapD with sucient strength to permit studies of the
complex using 15N-edited NMR spectroscopy. Instead,
the weak complex between PapG308-314 and PapD was
studied by transferred nuclear Overhauser e€ect spec-
troscopy (TRNOE),31 which requires a fast exchange of
the ligand between the bound and free states thereby
allowing NOEs developed for the ligand in the bound
In the complex between PapD and 15N-labeled
PapG307-314, sequential Hai/NHi+1 NOE con-
nectivities were observed for the sequence from Met309
to Phe313. Taken together with the absence of both
intraresidue Hai/NHi and sequential NHi/NHi+1 NOEs
this suggested that the peptide was bound to PapD in an
extended b-strand conformation. This conclusion was
further con®rmed by the transferred NOEs obtained for
the complex between PapD and PapG308-314. The
NOESY spectrum of unbound PapG308-314 contains
Table 2. 1H NMR chemical shifts (d, ppm) for PapG308-314 in aqueous phosphate bu€er (pH 6.0) containing 10% D2O, obtained at
600 MHz and 27 ꢀC. When determined, chemical shifts for the cis form about the Phe-Pro amide bond are given on the second row for
a residue. Numbers in parentheses indicate the chemical shifts when PapD is present
Residue
NH
Ha
Hb
Hg
Others
a
a
Thr308
Met309
Val310
4.09
4.53
3.81
1.26
2.53b
1.98, 2.04 (2.01b)
1.97
2.00
2.08 (SMe)
8.32
8.33
4.04 (4.05)
4.07
0.85, 0.90
0.90b
1.56
Leu311
Ser312
Phe313
Pro314
8.32 (8.34)
8.37 (8.38)
7.99 (8.00c)
8.18
4.33 (4.34)
4.40
4.36 (4.37c)
4.43
1.43 (1.44), 1.56
1.54, 1.61
3.66b
0.83 and 0.90 (Hd,d0)
0.86 and 0.92 (Hd,d0)
1.61
3.76b
8.19
8.01
4.88
4.65
2.85, 3.20 (2.86, 3.21)
2.93b
7.28 and 7.33 (H-arom)
7.23, 7.35, 7.31 (H-arom)
3.66 (3.67) and 3.71 (Hd)
3.30 and 3.47 (Hd)
4.24 (4.25)
3.66
1.91, 2.20
1.80, 1.85
1.97b
1.67b
a
Not observed due to fast amide proton exchange.
b Degeneracy has been assumed.
c Broad resonance, chemical shift determination is uncertain.

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2093
and the aromatic ring protons of Phe313. These medium
range TRNOEs are also in agreement with an exten-
ded, b-strand structure for PapG308-314 in which the
side chains of Val310 and Ser312 are on one side of the
b-strand and those of Leu311 and Phe313 on the other.
The possibilities that the medium range TRNOEs were
due to spin di€usion could be ruled out, as discussed
elsewhere in detail.34 No intermolecular TRNOEs,
between the ligand and the protein, were observed since
the PapD concentration used in the experiment was too
low to enable such observations.
TRNOEs may also originate from nonspeci®c binding
of a ligand to a protein or because of an increase in the
viscosity of the solution upon addition of the protein.31
As a control, a NOESY spectrum was acquired for a
solution of PapG308-314 and the protein calbinidin
D28k, which has a similar molecular weight as PapD,
using a peptide-protein ratio of 10:1. The spectrum did
not show any cross peaks indicating that the interaction
between PapG308-314 and PapD is speci®c. Spin di€u-
sion in large proteins, leading to intense cross peaks
even for protons that are not close in space, is a well-
known problem. The observed TRNOEs for PapG308-
314 could, therefore, be a€ected by spin di€usion
through protons of the protein. This type of indirect
transfer of magnetization can be alleviated by keeping
the fraction of bound peptide low35 and a 25:1 ratio
between PapG308-314 and PapD was therefore used.
Figure 4. Parts of the NOESY spectrum of the peptide
PapG308-314 by itself (A and C) and in the presence of PapD
(B and D). The NH/Ha (A and B) and NH/NH (C and D)
regions are shown. Strong sequential Hai/NHi+1 TRNOEs
were observed in the presence of PapD (marked in B with one
letter codes for the corresponding amino acids in the peptide),
whereas sequential NHi/NHi+1 TRNOEs were absent (D),
indicating that the peptide is bound in an extended, b-strand
conformation. The spectra were plotted at equal threshold.
Pulse ®eld gradients were used for water suppression, explain-
ing the presence of exchange peaks to water.
3
The JHNa coupling constants were determined for the
complex between PapD and PapG307-314 to obtain
further structural information on the complex. Owing to
the 15N labeling of the peptide the coupling constants
could be determined from the intensity of the 15N±Ha
cross peaks in a set of constant-time HMQC experi-
no cross peaks since the correlation time of the peptide
is such that otc&1, resulting in NOEs close to zero
(Fig. 4A and C).33 On addition of PapD (Mw&25 kDa
and thus otcꢂ1), to give a PapD-peptide ratio of 1:25,
strong TRNOE peaks built up in the PapD-peptide
complex were observed (Fig. 4B). TRNOEs were
observed for both the cis and trans forms of PapG308-
314 indicating the ability of PapD to bind both forms.
In general, the TRNOEs for the cis form were much
weaker, suggesting a weaker binding to PapD and con-
sequently the cis form was discarded from further ana-
lysis. In the trans form strong sequential Hai/NHi+1
TRNOEs were observed from Met309 through Phe313
(Fig. 4B). Sequential NHi/NHi+1 TRNOEs could not
be detected for PapG308-314 (Fig. 4D) and the sharp
NH resonances observed for the peptide in the presence
of PapD showed that this was not due to NH line
broadening. Furthermore, weak intraresidue Hai/NHi
TRNOEs could only be seen for residues Leu311 and
Phe313. Altogether, these TRNOEs indicate an extended
conformation for PapG308-314 when bound by
PapD.26,27 Medium range TRNOEs were observed
between the methyl groups of Val310 and the Hb protons
of Ser312, as well as between the methyl groups of Leu311
3
ments.36 JHNa values between 8.2 and 9.0 Hz were
obtained for residues Val310 (8.6‹0.13 Hz), Leu311
(8.2‹0.13 Hz), Ser312 (8.4‹0.16 Hz), and Phe313
(9.0‹0.11 Hz) in the PapG307-314. These coupling
constants correspond well with values found in b-strand
regions of proteins (8.5 Hz),37 clearly indicating that the
PapG307-314 is bound in an extended conformation by
PapD.
Further information on the mode of interaction of
PapG308-314 with PapD was obtained from line
broadening and di€erences in chemical shifts displayed
by some proton resonances. Broadening of resonances is
a
consequence of large chemical shift di€erences
between the free and bound form of the peptide. In the
presence of PapD the Hb resonances of Phe313 show a
signi®cant broadening as do the Hb protons of Leu311
and Met309, revealing these residues to contact PapD in
the complex (Fig. 5A and B). These Hb resonances
also display signi®cant di€erences in chemical shifts38 in

2094
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
and other chaperones in general. The ELISA has been
used to evaluate the inhibitory power of a set of syn-
thetic peptides related to P pilus proteins. It should also
prove useful for screening of chemical libraries prepared
by combinatorial chemistry,39±41 or for optimization of
lead compounds.
When evaluated in the inhibition ELISA, the C-terminal
19-mer peptides PapG296-314 and PapH155-173,
which are derived from the pilus adhesin PapG and
anchor PapH, respectively, were both found to be
potent inhibitors of formation of the PapD±PapG com-
plex. In contrast, the C-terminal 19-mer peptides from
PapF, E, and K, which make up the pilus tip, did not
show any signi®cant inhibitory power, even when eval-
uated at a 200-fold excess as compared to PapD. It is
possible that the superior inhibitory power of the PapG
and PapH peptides re¯ect the critical functions that
the corresponding proteins have for initiation and
termination of pilus assembly, respectively. The high
inhibition obtained for PapG296-314 agrees well with
the high anity displayed by PapD for this peptide
when it was coated in microtiter plate wells, or when
used to inhibit PapD mediated folding of denatured
PapG.15 However, PapH155-173 adsorbed to microtiter
plate wells did not bind the PapD chaperone,15 indicat-
ing the importance of using several bioassays when
screening ligands for binding to macromolecules. The
19-mer PapG296-314 could be N-terminally truncated
to the octapeptide PapG307-314 without a major loss
of inhibitory power in the ELISA, revealing that
PapG307-314 provides most of the interactions between
PapD and C-terminal peptides from PapG. These
results are in good overall agreement with the structure
of the crystalline complex between PapD and
PapG296-314, in which the peptide extends one of the
b-sheets in PapD by hydrogen bonding and side-chain
interactions involving the eight C-terminal residues of
the peptide.15 In the crystal, Met307, which is the ®rst
of these eight residues, contributes two of the nine
backbone-to-backbone hydrogen bonds to PapD. The
observations that truncation of Met307 from PapG307-
314 led to a signi®cantly reduced inhibitory power,
whereas replacement of this residue with Ala or Ser
had no such e€ect, con®rmed the importance of
backbone-to-backbone hydrogen bonding for stabili-
zation of PapD-peptide complexes also in solution.
Truncation of PapH155-173, either at the N-terminus
or at the C-terminus, abolished inhibition in the
ELISA. This suggests that conformational preferences
may be important for binding of PapH155-173 to
PapD, or that this peptide binds in a di€erent way, or
to a di€erent part of PapD than the PapG peptides.
Since limited solubility and tendencies to form aggre-
gates in¯uenced the inhibitory power of at least two of
the ®ve 19-mer peptides, the C-terminal octapeptides
Figure 5. Selected parts of the one-dimendional 1H NMR
spectrum of the peptide PapG308-314 by itself (upper) and in
the presence of PapD (lower). Resonances from Hb of Met309
,
Leu311 and Phe313 were broadened and shifted in the presence
of PapD and have been marked with the one letter codes for
the corresponding amino acid.
the presence and absence of PapD (Table 2). It should
be noted that the side chains of Phe313, Leu311, and
Met309 were also those found to make the most impor-
tant contacts with PapD in the inhibition ELISA. In
addition, broadening of backbone resonances, as well as
chemical shift di€erences (Table 2), were observed in the
one-dimensional proton spectrum and in the TOCSY
spectrum for several resonances from Leu311 and Ser312
when PapD was added. Most likely, the backbone in
the center of PapG308-314 is bound more rigidly by
PapD, as compared to the N- and C-terminus, explain-
ing these line broadening e€ects and chemical shift
di€erences.
Discussion
Periplasmic chaperones are required for assembly of pili
which mediate binding to host tissue in a large number
of pathogenic, gram-negative bacteria.1 Since inter-
ference with pilus assembly reduces bacterial virulence,
compounds that inhibit chaperone function may con-
stitute novel antibiotics. Chaperones involved in pilus
assembly show signi®cant homology to PapD,2,14 which
mediates pilus assembly in uropathogenic E. coli. This
suggests that all periplasmic chaperones have structures
closely related to that determined for PapD13 and, fur-
thermore, that chaperone-inhibitors could have a broad
spectrum. In the present investigation an ELISA in
which peptides, or other compounds, may be evaluated
as inhibitors of the binding of PapD to the MaltBP±
PapG175-314 fusion protein16 has been developed. The
complex between PapD and MaltBP±PapG175-314 has
previously been shown to be as stable as the native
PapD±PapG complex,16 arguing that the ELISA is a
relevant model for the function of PapD, in particular,

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2095
from the six pilus proteins were also evaluated in the
ELISA. Apart from PapG307-314, and possibly
PapK150-157, none of the octapeptides were able to
inhibit binding of PapD to MaltBP±PapG175-314. It
should, however, be pointed out that amino acid resi-
dues in the pilus proteins PapA and PapE, that were
located within the peptides employed in the present
investigation, have been found to be essential for sub-
unit binding to PapD.12
one of its components can then be expected42 to have a
drastic in¯uence on the stability of the complex. In
contrast, in a PapD±peptide complex, the residues are
located on the surface of the complex allowing interac-
tions with the aqueous environment to compensate for
loss of the salt bridge.
1H NMR spectroscopy was used in an attempt to
establish if di€erent conformational propensities
played a role in determining the ability of the pilus
derived 19-mer peptides to inhibit binding of PapD to
the MaltBP±PapG175-314 fusion protein. Due to the
low solubility of the peptides in aqueous solution, the
NMR studies were performed in DMSO, which was
also used as the cosolvent in the inhibition ELISA.
The importance of individual side chains in the minimal
inhibitory peptide from PapG (PapG307-314) for bind-
ing to PapD in solution was investigated using both an
alanine and a serine scan peptide series. It was found
that substitution with serine, in addition to alanine,
substantially facilitated interpretation of the results
obtained in the inhibition ELISA. Comparison of the
inhibitory powers of the members of the two series with
that of PapG307-314 revealed that the side chains of
Met309, Leu311, and Phe313 provided important contacts
to PapD. The remaining ®ve side chains could all be
replaced without any major in¯uence on the inhibitory
power. Furthermore, line broadening and chemical shift
3
Inspection of NOEs, JNHa coupling constants and che-
mical shifts indicated that the 19-mer peptides behaved
as random coils in DMSO solution. This conclusion
was also reached in
a previous NMR study of
PapG296-314 performed in aqueous solution.18 How-
ever, conformational studies of PapG296-314 carried
out in di€erent solutions using CD spectroscopy have
revealed18 that the peptide does have a signi®cant pro-
pensity to adopt a b-strand structure, i.e. the con-
formation adopted in the crystalline complex with
PapD. In contrast, the conformations of PapF130-148,
PapE131-149, and PapK139-157 appeared to be in¯u-
enced by the surrounding environment to a much larger
extent.18 The preference for adopting a b-strand struc-
ture may therefore provide part of the explanation for
the high ability of PapG296-314 to bind to and inhibit
PapD.
di€erences were observed for the b-protons of Met309
,
Leu311, and Phe313 in the 1H NMR spectrum of
PapG308-314 on addition of PapD, con®rming that
these residues contact PapD in aqueous solution. In the
crystalline complex between PapD and PapG296-314
the side chains of these three residues, as well as that of
Met307, were found to contribute 20% of the total bur-
ied surface area between peptide and protein.15 As
revealed by the inhibition studies the side chain of
Met307 is less important; an observation which agrees
well with the fact that four out of the six pilus proteins
have an alanine at this position (cf. Table 1). In con-
trast, all P pilus proteins have large hydrophobic side
The conformation of C-terminal hepta- and octapep-
tides from PapG, when bound by PapD, was investi-
gated in solution using both transferred nuclear
Overhauser e€ect spectroscopy (TRNOE) and isotope
edited NMR spectroscopy. The observation of strong
sequential Hai/NHi+1 TRNOEs along the backbone of
the peptides and the absence of sequential NHi/NHi+1
TRNOEs indicated that the peptides were bound by
PapD in an extended, b-strand like conformation. This
conclusion was further supported by TRNOEs revealing
that the side chains of Val310 and Ser312 are on one side
of the b-strand and those of Leu311 and Phe313 on the
chains at the positions corresponding to Met309, Leu311
and Phe313
,
.
Inspection of the crystalline complex between PapD and
the peptide PapG296-314 suggested that formation of a
salt bridge between the C-terminal carboxyl group of
the peptide and the invariant residues Arg8 and Lys112
in PapD was essential for complex formation.15 How-
ever, in solution, the ability of peptides to form this salt
bridge was not critical for inhibition of PapD±PapG
complex formation. This conclusion was based on the
fact that peptides from PapG in which Pro314 had been
deleted, or in which the a-carboxyl group had been
reduced to a methyl group, retained all or most of their
inhibitory power. At ®rst this appears to be in con¯ict
with previous results showing that site-speci®c mutation
of Arg8 or Lys112 in PapD abolished the ability of PapD
to bind to pilus proteins and mediate pilus assembly in
vivo.15 However, in pilus protein±PapD complexes, the
salt bridge is most likely deeply buried and deletion of
3
other. In addition, the JNHa coupling constants
obtained for 15N-labeled PapG307-314 were between
8.2 and 9.0 Hz, i.e. close to those found in b-strand
regions of proteins.37 NMR spectroscopy, therefore,
revealed that peptides from PapG are bound in the same
conformation by PapD as in the crystal. It should,
however, be pointed out that the peptides used in the
NMR studies are too short to allow formation of the
dimers found in the crystal and that dimerization was
previously excluded34 using asymmetrical ®eld-¯ow
fractionation.

2096
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
The information obtained by evaluation of peptides as
inhibitors of PapD±PapG complex formation, together
with that from NMR studies of peptide±PapD complexes,
thus provided a coherent picture of how PapD binds
peptides from the adhesin PapG in solution. With only
two exceptions, this picture is in good agreement with
the structure in the crystalline complex of PapD and the
longer peptide PapG296-314.15 Importantly, the present
investigation reveals that the structure of the complex
was not signi®cantly in¯uenced by the dimerization pre-
sent in the crystal. The crystal structure, in combination
with the present solution studies, may therefore be used
as a starting point in e€orts to design compounds that
inhibit pilus assembly in pathogenic bacteria. Evaluation
of such chaperone inhibitors should be facilitated by
using the inhibition ELISA developed by us.
sulfonate; COSY: correlated spectroscopy; DIPSI:
decoupling in the presence of scalar interactions;
ELISA: enzyme linked immunosorbent assay; HATU:
O-(7-azabenzotriazol-1-yl)-1,1,3,3-tetramethyluronium
hexa¯uorophosphate; HETCOR: heteronuclear correla-
tion spectroscopy; HMQC: heteronuclear multiple
quantum coherence spectroscopy; HOAt: 1-hydroxy-7-
azabenzotriazole; HSQC: heteronuclear single quantum
coherence spectroscopy; MaltBP: maltose binding pro-
tein; NMM: N-methylmorpholine; NOESY: nuclear
Overhauser e€ect spectroscopy; Pap: pyelonephritis-
associated pilus; ROESY: rotating frame nuclear Over-
hauser e€ect spectroscopy; TBTU: O-(benzotriazol-1-yl)-
1,1,3,3-tetramethyluronium tetra¯uoroborate; TOCSY:
total correlation spectroscopy; TRNOE: transferred
nuclear Overhauser e€ect.
General methods and materials
Conclusions
TLC was performed on Silica Gel 60 F254 (Merck, Ger-
many) with detection by UV light. Flash column chro-
matography was performed on silica gel (Matrex, 60 A,
35±70 mm, Grace Amicon, U.S.A.) with distilled solvents.
DMF was dried by distillation and kept over molecular
sieves (4 A). Organic solutions were dried over Na2SO4
before being concentrated. 1H and 13C NMR spectra
were recorded at 400 and 100 MHz, respectively, at
300 K. Residual CHCl3 (dH 7.25 ppm) and CDCl3 (dC
77.0 ppm) were used as internal standards. First-order
chemical shifts and coupling constants were obtained
from one-dimensional spectra. Assignments of reso-
nances were based on COSY and HETCOR experi-
ments. Ions for positive fast-atom bombardment mass
spectra were produced by a beam of Xenon atoms
(6 keV) from a matrix of glycerol and thioglycerol.
Analytical reversed-phase HPLC was performed on a
Kromasil C-8 column (100 A, 5 mm, 4.6Â250 mm,
Hichrom Ltd., U.K.) using a linear gradient of 0!80%
B in A over 60 min with a ¯ow rate of 1.5 mL/min and
detection at 214 nm (solvent systems A: 0.1% aqueous
tri¯uoroacetic acid and B: 0.1% tri¯uoroacetic acid in
CH3CN). Preparative reversed-phase HPLC was per-
formed on a Kromasil C-8 column (100 A, 5 mm,
20Â250 mm, Hichrom Ltd., U.K.) using the same eluent
and a ¯ow rate of 10 mL/min.
Bacterial periplasmic chaperones, such as PapD in E.
coli, bind the protein subunits that make up adhesive
pili and incorporate them into the functional pilus. An
ELISA has now been developed which allows evalua-
tion of compounds as inhibitors of complex formation
between PapD and the pilus adhesin PapG. Evaluation
of peptides from the C-terminus of the six proteins that
make up the E. coli P pilus, revealed that peptides from
the adhesin PapG were the most potent PapD inhibitors
and that most of the inhibitory activity was retained
in the C-terminal octapeptide PapG307-314. NMR
spectroscopy of the complexes between PapD and
PapG307-314, or PapG308-314, revealed that the pep-
tides were bound as extended b-strands by PapD in
aqueous solution. Furthermore, evaluation of a panel of
analogues of PapG307-314 in the ELISA, as well as
further NMR studies, showed that the complex was
stabilized by backbone-to-backbone hydrogen bonds
and interactions involving the hydrophobic side chains
of Met309, Leu311, and Phe313 of the peptide. Most
likely, a preference for adopting a b-strand structure
also improved peptide binding to PapD. The structure
deduced for the complex between PapD and the
PapG307-314 peptide in aqueous solution is, with some
exceptions, in good agreement with the structure deter-
mined for the crystalline complex of PapD bound to the
19-mer peptide PapG296-314. This structural informa-
tion may be used in e€orts to design and synthesize
compounds that inhibit periplasmic chaperones in
pathogenic bacteria.
Na-tert-Butoxycarbonyl-L-phenylalanine (2S)-hydroxy-
methylpyrrolidinyl amide (3). l-Prolinol (2, 137 mg,
1.36 mmol) and N-methylmorpholine (NMM) (343 mg,
3.39 mmol) were added to a solution of Na-tert-butoxy-
carbonyl-l-phenylalanine (1, 300 mg, 1.13 mmol) and
2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium tetra-
¯uoroborate (TBTU) (436 mg, 1.36 mmol) in DMF
(6 mL) and the mixture was stirred at room temperature
for 5 h under nitrogen. The mixture was diluted with
EtOAc (20 mL) and washed with saturated aqueous
Experimental
Abbreviations used: Boc: tert-butoxycarbonyl; CHAPS:
3-[(3-cholamidopropyl)dimethylammonio]-1-propane-

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2097
NH4Cl (2Â20 mL) and saturated aqueous NaHCO3
(2Â20 mL). The organic phase was dried, ®ltered, and
concentrated. Flash column chromatography (heptane/
EtOAc, 1/1) of the residue gave 3 (303 mg, 77%): [a]d25
0 ꢀ (c 1.40, CHCl3); 1H NMR (CDCl3), amide bond
rotamer ratio 4:1, d 7.33±7.20 (m, 5 H, Ph), 5.38 (bd,
J=9.1 Hz, 0.8 H, trans-Phe NH), 5.35 (bd, J=10.3 Hz,
0.2 H, cis-Phe NH), 4.81 (dt, J=8.8, 6.0 Hz, 0.2 H, cis-
Phe Ha), 4.66 (dt, J=9.0, 5.5 Hz, 0.8 H, trans-Phe Ha),
4.35 (bs, 0.8 H, trans-pyr OH), 4.20±4.12 (m, 0.8 H,
trans-pyr Ha), 3.92 (bs, 0.2 H, cis-pyr OH), 3.62±3.51
(m, 1.6 H, trans-Phe Hd, trans-pyr CH2O), 3.51±3.34
(m, 1.6 H, trans-Phe CH2O, cis-pyr Hd,d, cis-pyr CH2O),
3.34±3.27 (m, 0.2 H, cis-pyr Ha), 3.09±2.88 (m, 2 H, Phe
Hb), 2.63 (dt, J=10.0, 7.2 Hz, 0.8 H, trans-pyr Hd),
1.99±1.89 (m, 0.8 H, trans-pyr Hb), 1.76±1.57 (m, 2 H,
pyr Hg), 1.50±1.37 (m, 9.8 H, Phe NBoc, trans-pyr Hb),
1.29±1.24 (m, Phe NBoc rotamer), 1.20±1.06 (m, 0.4 H,
cis-pyr Hb,b); 13C NMR (CDCl3) d 173.0, 155.5, 136.6,
130.0, 129.8, 129.0, 128.9, 127.5, 80.3, 67.1, 61.8, 54.0, 48.3,
40.7, 28.8, 28.7, 28.3, 24.7, 22.2; HRMS±FAB: (M+H+)
calcd for C19H28O4N2, 349.2127; found, 349.2137.
through Celite, which was washed with CH2Cl2 and
EtOH. The combined ®ltrates were concentrated and
the residue was dissolved in CH2Cl2 (20 mL). The solu-
tion was washed with H2O (10 mL), dried, ®ltered, and
concentrated. Flash column chromatography (heptane/
acetone, 6/1) of the residue gave 5 (57 mg, 93%): [a]d25
+6 ꢀ (c 1.30, CHCl3); H NMR (CDCl3), amide bond
1
rotamer ratio 5:3, d 7.30±7.19 (m, 5 H, Phe Ph), 5.43
(bd, J=7.7 Hz, 1 H, Phe NH), 4.66 (dt, J=9.6, 5.1 Hz,
0.4 H, cis-Phe Ha), 4.57 (dt, J=8.7, 5.7 Hz, 0.6 H, trans-
Phe Ha), 4.21±4.11 (m, 0.6 H, trans-pyr Ha), 3.46 (ddd,
J=11.7, 9.4, 2.5 Hz, 0.4 H, cis-pyr Hd), 3.42±3.33 (m,
0.6 H, trans-pyr Hd), 3.30±3.18 (m, 0.8 H, cis-pyr Ha,
cis-pyr Hd), 3.07±2.88 (m, 2 H, Phe Hb), 2.75±2.65 (m,
0.6 H, trans-Phe Hd), 1.94±1.83 (m, 0.6 H, trans-Phe
Hb), 1.74±1.64 (m, 1 H, pyr Hg), 1.45±1.37 (m, 10.4 H,
Phe NBoc, trans-pyr Hb, cis-pyr Hb), 1.31±1.21 (m, Phe
NBoc rotamer), 1.09 (2 d, each J=3.1 Hz, 3 H, pyr Me);
13C NMR (CDCl3) d 170.0, 155.5, 136.9, 130.1, 129.8,
128.8, 128.7, 127.2, 80.0, 53.8, 53.5, 53.3, 46.9, 45.5,
41.5, 40.7, 32.8, 32.1, 28.8, 28.8, 24.3, 21.9, 21.5, 19.6;
HRMS±FAB: (M+H+) calcd for C19H28O3N2,
333.2178; found, 333.2282.
Na-tert-Butoxycarbonyl-L-phenylalanine (2S)-iodomethyl-
pyrrolidinyl amide (4). Compound
3
(200 mg,
Na-9-Fluorenylmethoxycarbonyl-L-methionyl-L-threonyl-
L-methionyl-L-valyl-L-leucyl-L-serine (6). Peptide frag-
ment 6 was synthesized on PEG-PS resin (TentaGel S
PHB Ser(t-Bu) Fmoc resin, 0.21 mmol/g, 300 mg) as
described below in the general section on peptide
synthesis. The crude product (33 mg) was puri®ed by
reversed-phase preparative HPLC to give 6 (7 mg, 12%);
MS±FAB: (M+H+) calcd, 903; found, 903.
0.57 mmol), iodine (364 mg, 1.44 mmol), triphenylphos-
phine (376 mg, 1.44 mmol), and imidazole (137 mg,
2.01 mmol) were stirred in toluene (20 mL) for 1.5 h at
80 ꢀC. The mixture was diluted with toluene (25 mL)
and the solid residue was dissolved in acetone (3 mL).
The combined organic phase was washed with saturated
aqueous NaHCO3 (2Â15 mL) and H2O (2Â15 mL),
dried, ®ltered, and concentrated. Flash column chro-
matography (heptane/EtOAc, 4/1) of the residue gave 4
(168 mg, 64%): [a]2d5 +18 ꢀ (c 1.21, CHCl3); 1H NMR
(CDCl3), amide bond rotamer ratio 5:2, d 7.34±7.20 (m,
5 H, Phe Ph), 5.32 (bd, J=8.2 Hz, 1 H, Phe NH), 4.63
(dt, J=7.7, 7.7 Hz, 0.7 H, trans-Phe Ha), 4.58±4.51 (m,
0.3 H, cis-Phe Ha), 4.20±4.13 (m, 0.7 H, trans-pyr Ha),
3.62±3.52 (m, 1H, pyr Hd), 3.47 (dd, J=9.5, 2.8 Hz, 1
H, pyr CH2I), 3.36±3.25 (m, 0.3 H, cis-pyr Ha), 3.07±
2.88 (m, 4 H, Phe Hb,b, pyr Hd, pyr CH2I), 2.08±1.95
(m, 1 H, pyr Hb), 1.89±1.74 (m, 3 H, pyr Hb, pyr Hg,g),
1.45 (s, 2.7 H, cis-Phe NBoc), 1.41 (s, 6.3 H, trans-Phe
NBoc), 1.27 (s, Phe NBoc rotamer); 13C NMR (CDCl3)
d 171.1, 155.5, 136.8, 130.0, 129.8, 129.0, 128.8, 127.5,
127.4, 80.2, 58.5, 53.7, 48.0, 46.4, 40.4, 30.8, 28.8, 24.3,
9.2; HRMS±FAB: (M+H+) calcd for C19H27O3N2I,
459.1145; found, 459.1163.
Na-9-Fluorenylmethoxycarbonyl-L-methionyl-L-threonyl-
L-methionyl-L-valyl-L-leucyl-L-seryl-L-phenylalanine (2R)-
methylpyrrolidinyl amide (7). The Na-tert-butoxy-
carbonyl group was removed from 5 (4 mg, 12 mmol) by
treatment with a solution of HCl in acetic acid (1 M,
400 mL) for 1 h. Portions of CH2Cl2 (5Â1 mL) were
added to the solution and then removed by concentra-
tion. The residue was dissolved in DMF (150 mL), then
peptide fragment 6 (5.2 mg, 6 mmol), HATU (46 mL of a
0.26 M solution in DMF, 12 mmol) and HOAt (32 mL of
a 0.37 M solution in DMF, 12 mmol) were added. The
solution was cooled to 0 ꢀC and trimethylpyridine
(TMP, 80 mL of a 0.45 M solution in DMF, 36 mmol)
was added. The coupling was monitored by analytical
reversed-phase HPLC. After 1 h at 0 ꢀC and 2 h at room
temperature, puri®cation (without prior work up) was
performed by preparative reversed-phase HPLC to give
7 (3.38 mg, 53%); MS±FAB: (M+H+) calcd, 1117;
found, 1117.
Na-tert-Butoxycarbonyl-L-phenylalanine
(2R)-methyl-
pyrrolidinyl amide (5). Compound 4 (84 mg, 0.18 mmol)
and triethylamine (32 mL, 0.23 mmol) were dissolved in
EtOAc (1 mL) and EtOH (5 mL) and 10% Pd/C (50 mg)
was then added under argon. The reaction mixture was
hydrogenated at 1 atm for 1 h and 10 min, then ®ltered
L-Methionyl-L-threonyl-L-methionyl-L-valyl-L-leucyl-L-
seryl-L-phenylalanine (2R)-methylpyrrolidinyl amide (8).
Peptide 7 (2.3 mg, 2 mmol) was dissolved in a solution

2098
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
of piperidine in DMF (20%, 300 mL) at room tempera-
ture and the reaction was monitored by analytical
reversed-phase HPLC. After 1 h, puri®cation (without
prior work up) was performed by preparative reversed-
phase HPLC to give 8 (0.68 mg, 26%); MS±FAB:
(M+H+) calcd, 895; found, 895.
containing 0.05% Tween1 20 (polyoxyethylene sorbi-
tanmonolaurate, Fluka, Switzerland) and 50 mL of the
resulting solution was added, in triplicate, to the coated
wells and incubated at 25 ꢀC for 2 h. The plates were
washed three times with PBS and the wells were incu-
bated with a solution of rabbit antiserum to PapD in
PBS containing 3% BSA (1:1000 dilution, 50 mL) for 2 h
at 25 ꢀC. The plates were again washed three times with
PBS and the wells were incubated with a solution of
goat antiserum to rabbit IgG coupled to alkaline phos-
phatase (Sigma, U.S.A.) in PBS containing 3% BSA
(1:1000 dilution, 50 mL) for 2 h at 25 ꢀC. After three
washings with PBS and three washings with developing
bu€er (10 mM diethanolamine, 0.5 mM MgCl2, pH
9.5), ®ltered p-nitrophenyl phosphate (Sigma, U.S.A.) in
developing bu€er (10 mg/mL, 50 mL) was added to each
well and incubated for 30 min in the dark at 25 ꢀC. The
absorbance at 405 nm was read and the per cent inhibi-
tion was calculated by dividing the amount of PapD
binding to the coated MaltBP±PapG175-314 fusion
protein in the presence of peptide with the amount of
PapD binding in the absence of peptide, and sub-
tracting the ratio from 1. The per cent inhibition
(Inhibitory Power) represents the average of three
experiments, each performed in triplicate. From these
the standard deviation (SD) was calculated and 95%
con®dence intervals were obtained as:
Peptide synthesis
Synthesis, puri®cation, and characterization of the pep-
tides PapG296-314, PapF130-148, PapE131-149,
PapK139-157, and PapH155-173 have been described
elsewhere.18 All other peptides used in this study were
prepared in the same manner by the 9-¯uorenylmethoxy-
carbonyl (Fmoc) solid-phase synthesis strategy using a
fully automated continuous ¯ow peptide synthesizer,
constructed in our laboratory essentially as described
previously by others.19 Synthesis was performed on a
polystyrene resin grafted with polyethylene-glycol
spacers (TentaGel PEG-PS resin, Rapp Polymere, Ger-
many) using commercially available Na-Fmoc amino
acids carrying standard side-chain protective groups
(Bachem, Switzerland). All but two peptides (PapG299-
306 and PapH155-165) were synthesized as C-terminal
carboxylic acids on resins functionalized with a 4-
alkoxybenzyl alcohol linker and the appropriate C-
terminal amino acid or a 2-chlorotrityl linker43 and
Fmoc-proline. The peptides PapG299-306 and
PapH155-165 were instead prepared as C-terminal
amides using the Rink linker, p-[a-(¯uoren-9-ylmethoxy-
formamido)-2,4-dimethoxybenzyl]phenoxyacetic acid.44,45
15N-Labelled amino acids (15N, 95±99%, CIL, U.S.A.)
were protected with an Na-Fmoc group prior to their
use in synthesis. After TFA-catalyzed cleavage from the
resin, the peptides were puri®ed to greater than 95%
homogeneity by reversed-phase HPLC and molecular
weights were con®rmed by FABMS.
p
Inhibitory Power Æ SDÁ1:96
n
s
n
X
1
2
ꢀ
where SD ˆ
ꢀx xi†
n
1
iˆ1
and where n ˆ 3 and xꢀ is the mean of the values for the
Inhibitory Power ꢃxi†.
In the ELISA maximum binding of PapD to the
MaltBP±PapG175-314 coated wells was obtained in the
absence of peptide (but with DMSO still present). This
maximum binding value was used in calculation of the
Inhibitory Power as described above. Since ¯uctuation
was observed for the Inhibitory Powers of the pep-
tides, the ¯uctuation of the maximum binding value was
also determined. This was done from ®ve randomly
chosen microtiter plates, each of which gave four values
for the maximum binding value. It was found that the
average of the SD of the maximum binding value divi-
ded by the maximum binding value was 8.5%, with a
SD of 4.84%.
Peptide inhibition ELISA
Microtiter plate wells (Nunc-Immuno Plate Maxisorp,
Denmark) were each coated with a solution of the
MaltBP±PapG175-314 fusion protein (50 mL at
10 pmol/50 mL) in PBS (138 mM NaCl, 2.7 mM KCl,
8.2 mM phosphate, pH 7.4) for 24 h at 4 ꢀC. The solu-
tions in the wells were discarded. The wells were each
blocked with a solution of 3% BSA (Sigma, U.S.A.) in
PBS (200 mL) for 2±3 h at 25 ꢀC: the plates were then
washed vigorously three times with PBS. The peptide to
be evaluated was dissolved to six di€erent concentra-
tions (600, 300, 150, 38, 9, and 2 mM) in DMSO (Fluka,
Switzerland). Each peptide solution (10 mL) was pre-
incubated with a solution of PapD (10 mL at 30 pmol/
10 mL) in KMES bu€er (20 mM potassium 2-(N-mor-
pholino)-ethanesulfonate, pH 6.5) for 2 h at 25 ꢀC. The
PapD±peptide solution was diluted 50 times with PBS
NMR spectroscopy of PapG296-314, PapE131-149,
PapK139-157, and PapH155-173
PapG296±314, PapE131-149, PapK139-157, and
PapH155-173 were dissolved in DMSO-d6 (99.9%,
Glaser AG, Switzerland) at a concentration of 3.0, 2.1,

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2099
0.6, and 0.5 mM, respectively. Phase sensitive
COSY,46,47 TOCSY,48±50 NOESY,51±53 and ROESY54,55
spectra were acquired at 500 MHz and at 27 ꢀC. The
DMSO-d6 resonance at 2.50 ppm was used as internal
shift reference. TOCSY experiments were performed
using the modi®cation suggested by Rance56 and the
dipsi-2 mixing sequence57 using a spin-lock period of
100 ms. Mixing times of 100±200 ms were used in the
ROESY and NOESY (PapH155-173) experiments. All
spectra were recorded in the phase sensitive mode58 with
512 t1 increments and 4096 complex data points in t2.
After zero ®lling, the ®nal size of the data matrices was
1024Â2048 points. Data were processed on a Sun Sparc
workstation (PapH155-173) using the FELIX software
(Biosym Technologies, U.S.A.) or on a Silicon Graphics
INDY workstation (PapG296-314, PapE131-149, and
PapK139±157) using the XWIN-NMR software
workstation using the FELIX software (Biosym
Technologies, U.S.A.). One-dimensional spectra were
acquired with 512 scans, 16 k complex data points, and
presaturation.
3D-NOESY-HSQC study of the complex between PapD
and 15N-labelled PapG307-314
A stock solution of PapD13 in PBS (ꢁ0.3 mM, pH 7.4)
was diluted in PBS (pH 6.0) and concentrated to a ®nal
concentration of 0.6 mM using a Centricon 10 micro
concentrator (cut-o€=10 kDa, Amicon, U.S.A.).
PapG307-314 was dissolved in DMSO-d6 (20 mL) and
added to the PapD solution to give a PapG307-314±
PapD ratio of 1:1. A sensitivity enhanced 2-D 15N-
HSQC spectrum with gradient selection,62 a 2-D 15N-
HSQC±TOCSY spectrum62 with a DIPSI-2 relaxation
compensated isotropic mixing sequence,63 and a 3-D
15N-NOESY-HSQC spectrum62 were acquired at
600 MHz and 27 ꢀC. Water-¯ip-back pulses were used to
suppress the water signal. Mixing times of 80 and
100 ms were used in the heteronuclear TOCSY and
NOESY experiments, respectively. A spectral width of
7490 Hz was used in the 1H dimension of all experi-
ments. The HSQC experiment was recorded with a
spectral width of 1600 Hz in the 15N dimension and with
128 scans, 128 t1 increments, and 2048 complex data
points in t2. The HSQC-TOCSY and NOESY-HSQC
experiments were acquired with a spectral width of
1000 Hz in the 15N dimension. The HSQC-TOCSY
experiment was recorded with 256 scans, 128 t1 incre-
ments, and 1024 complex data points in t2. In the
NOESY-HSQC experiment 128 t1 increments, 32 t2
increments and 1024 complex data points in t3 were
acquired collecting 12 scans for each t1 and t2 point.
3JHNa coupling constants were determined from six
constant-time 2-D HMQC experiments36 recorded with
45, 60, 80, 100, 120, and 140 ms dephasing times. The
spectra were recorded with 1024 complex data points in
t2 and 64 scans for each of the 64 t1 points using the
same spectral widths as used in the HSQC-TOCSY
experiment. Linear prediction was applied to all spectra
in order to increase the spectral resolution. Cross peak
intensities at various dephasing times were ®tted against
the following equation
3
(Bruker, Germany). Determination of JHNa coupling
constants (PapH155-173) were made in a resolution-
enhanced COSY spectrum, zero-®lled to give a 0.7 Hz/
point digital resolution in o2.
TRNOE study of the PapD±PapG308-314 complex
Three identical samples were prepared for NMR spec-
troscopy by dissolving PapG308-314 to a concentration
of 0.5 mM in 50 mM phosphate bu€er, pH 6.0, con-
taining 10% DO and 0.02% NaN3 (the low solubility of
PapG308-314 prevented use of a higher concentration).
A stock solution of PapD13 in PBS (ꢁ70 mM, pH 7.4)
was concentrated to 200 mM by centrifugation in an
Ultrafree micro concentrator (cut-o€=5 kDa, Milli-
pore, U.S.A.) prior to dilution into one of the PapG308-
314 samples. The resulting PapD concentration was
20 mM giving a PapG308-314±PapD ratio of 25:1. A
control sample was prepared by dissolving lyophilized
59
Calbindin D28k into another of the PapG308-314
samples to give a PapG308-314±Calbindin D28k ratio of
10:1. No additions were made to the third PapG308-314
solution. Phase sensitive NOESY,51,52 ROESY,54,55 and
TOCSY48 spectra were acquired at 600 MHz and 27 ꢀC
using the water resonance at 4.75 ppm as internal shift
reference. Presaturation was used to attenuate the water
signal except in the NOESY experiments where pulse
®eld gradients were used for water suppression. The
WATERGATE-NOESY pulse sequence60 was modi®ed
as described.61 NOESY spectra were acquired with
mixing times of 20, 200, and 500 ms and ROESY spectra
were acquired with a continuous wave spin-lock mixing
time of 200 ms. The TOCSY experiments, modi®ed as
described above for the 19-mer peptides, were recorded
with a spin-lock period of 90 ms. All spectra were
recorded using the hypercomplex method58 with 400 t1
increments and 2048 complex data points in t2. After
zero ®lling, the ®nal size of the data matrixes were
1024Â2048. Data were processed on a Sun Sparc
Iyꢀt† ˆ Iyꢀ0†N1 exp‰ t=T2 t=ꢀ2T1†Š
È
É
cosꢀpJrt† ‡ sinꢀpJrt†=ꢀ2pJrt†
q
with Jr ˆ J2
1=ꢃ2pT1†2, where N1 is the number
HH
of t1 increments in the constant-time dimension, t the
dephasing time, T2 the transverse relaxation rate, T1 the
selective longitudinal relaxation rate of the Ha protons
3
and JHH the JHNa coupling constant. A uniform value
of T1 for the Ha protons was assumed to be 0.2 s. The

2100
K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
value of JHH was optimized together with T2. Data were
processed on a Sun Sparc workstation using the FELIX
software (Biosym Technologies, U.S.A.).
15. Kuehn, M. J.; Ogg, D. J.; Kihlberg, J.; Slonim, L. N.;
Flemmer, K.; Bergfors, T.; Hultgren, S. J. Science 1993, 262,
1234.
16. Xu, Z.; Jones, C. H.; Haslam, D.; Pinkner, J. S.; Dodson,
K.; Kihlberg, J.; Hultgren, S. J. Mol. Microbiol. 1995, 16, 1011.
17. Stephens, C.; Shapiro, L. Chem. Biol. 1997, 4, 637.
Acknowledgements
18. Flemmer Karlsson, K.; Walse, B.; Drakenberg, T.; Kihl-
berg, J. Lett. Peptide Sci. 1996, 3, 143.
The authors are grateful to C. Pongratz for statistical
analysis of the ELISA data for the pilus subunit derived
19. Cameron, L. R.; Holder, J. L.; Meldal, M.; Sheppard, R. C.
J. Chem. Soc. Perkin Trans. 1. 1988, 2895.
19-mer peptides, S. Linse for the gift of Calbindin D28k
,
G. Carlstrom for pulse sequence programming
(TRNOE) and G. Carlstrom and J. Evenas for help with
the 3-D-NOESY-HSQC NMR experiments. This work
was funded by the Swedish National Board for Indus-
trial and Technical Development and by the Swedish
Natural Science Research Council.
20. Garegg, P. J.; Samuelsson, B. J. Chem. Soc. Perkin Trans.
1. 1980, 2866.
21. Carpino, L. A.; El-Faham, A.; Minor, C. A.; Albericio, F.
J. Chem. Soc., Chem. Commun. 1994, 201.
22. Carpino, L. A. J. Am. Chem. Soc. 1993, 115, 4397.
23. Carpino, L. A.; El-Faham, A. J. Org. Chem. 1994, 59, 695.
24. Flemmer, K.; Xu, Z.; Pinkner, J. S.; Hultgren, S. J.; Kihl-
berg, J. Bioorg. Med. Chem. Lett. 1995, 5, 927.
Supplementary material
25. Derrick, J. P.; Wigley, D. B. Nature 1992, 359, 752.
26. Wuthrich, K. NMR of Proteins and Nucleic Acids; New
York: Wiley & Sons, 1986.
400 MHz 1H NMR spectra and 100 MHz 13C NMR
spectra for compounds 3±5. Tabulated 500 MHz 1H
NMR data for the peptides PapG296-314, PapE131-
149, PapK139-157, and PapH155-173.
27. Dyson, H. J.; Wright, P. E. Annu. Rev. Biophys. Biophys.
Chem. 1991, 20, 519.
28. Bundi, A.; Grathwohl, C.; Hochmann, J.; Keller, R. M.;
Wagner, G.; Wuthrich, K. J. Magn. Reson. 1975, 18, 191.
29. Wishart, D. S.; Bigham, C. G.; Holm, A.; Hodges, R. S.;
Sykes, B. D. J. Biomol. NMR. 1995, 5, 67.
References and Notes
1. Hultgren, S. J.; Abraham, S.; Caparon, M.; Falk, P.; St.
Geme, J. W., III; Normark, S. Cell 1993, 73, 887.
30. Gronenborn, A. M.; Clore, G. M. Crit. Rev. Biochem. Mol.
Biol. 1995, 30, 351.
2. Hung, D. L.; Knight, S. D.; Woods, R. M.; Pinkner, J. S.;
Hultgren, S. J. EMBO J. 1996, 15, 3792.
31. Ni, F. Prog. Nucl. Magn. Reson. Spectrosc. 1994, 26, 517.
32. The o€-rate for dissociation of the heptapeptide PapG308-
314 from the complex with PapD was estimated in the follow-
ing way. The rate of peptide association with PapD was
3. Kuehn, M. J.; Heuser, J.; Normark, S.; Hultgren, S. J. Nat-
ure 1992, 356, 252.
1
assumed to be di€usion limited (108 M
s
1). A dissociation
4. Bullitt, E.; Makowski, L. Nature 1995, 373, 164.
constant of ꢁ10 6 M has been determined for a modi®ed
5. Kallenius, G.; Mollby, R.; Svenson, S. B.; Winberg, J.;
Lundblad, A.; Svensson, S.; Cedergren, B. FEMS Lett. 1980, 7,
297.
PapG307-314 peptide (G. Soto, personal communication),
1
resulting in an o€-rate of ꢁ100 s for this octapeptide. Since
PapG308-314 is a weaker inhibitor than the octapeptide
PapG307-314, the o€-rate of PapG308-314 should be even
higher, thus corresponding to the fast exchange regime.
6. Le‚er, H.; Svanborg Eden, C. FEMS Lett. 1980, 8, 127.
7. Kihlberg, J.; Hultgren, S. J.; Normark, S.; Magnusson, G. J.
Am. Chem. Soc. 1989, 111, 6364.
33. Neuhaus, D.; Williamson, M. P. The nuclear Overhauser
e€ect in structural and conformational analysis; New York:
VCH Publishers, 1989.
8. Striker, R.; Nilsson, U.; Stonecipher, A.; Magnusson, G.;
Hultgren, S. J. Mol. Microbiol. 1995, 16, 1021.
9. Nilsson, U.; Striker, R. T.; Hultgren, S. J.; Magnusson, G.
Bioorg. Med. Chem. 1996, 4, 1809.
34. Walse, B.; Kihlberg, J.; Flemmer Karlsson, K.; Nilsson,
M.; Wahlund, K.-G.; Pinkner, J. S.; Hultgren, S. J.; Draken-
berg, T. FEBS Lett. 1997, 412, 115.
10. Hultgren, S. J.; Lindberg, F.; Magnusson, G.; Kihlberg, J.;
Tennent, J. M.; Normark, S. Proc. Natl. Acad. Sci. U.S.A.
1989, 86, 4357.
35. Campbell, A. P.; Sykes, B. D. Annu. Rev. Biophys. Chem.
Biomol. Struct. 1993, 22, 99.
11. Kuehn, M. J.; Normark, S.; Hultgren, S. J. Proc. Natl.
Acad. Sci. U.S.A. 1991, 88, 10586.
36. Kuboniwa, H.; Grzesiek, S.; Delaglio, F.; Bax, A. J. Bio-
mol. NMR. 1994, 4, 871.
12. Bullitt, E.; Jones, C. H.; Striker, R.; Soto, G.; Jacob-
Dubuisson, F.; Pinkner, J.; Wick, M. J.; Makowski, L.; Hultg-
ren, S. J. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 12890.
37. Smith, L. J.; Bolin, K. A.; Schwalbe, H.; MacArthur, M.
W.; Thornton, J. M.; Dobson, C. M. J. Mol. Biol. 1996, 255,
494.
13. Holmgren, A.; Branden, C.-I. Nature 1989, 342, 248.
38. The di€erences in chemical shift observed for some of the
residues in PapG308-314 on addition of PapD may appear to
be small (0.01 ppm). However, the observed chemical shift is an
14. Holmgren, A.; Kuehn, M. J.; Branden, C.-I.; Hultgren, S. J.
EMBO J. 1992, 11, 1617.

K. Flemmer Karlsson et al./Bioorg. Med. Chem. 6 (1998) 2085±2101
2101
average of the shift for free and bound peptide. Since a pepti-
49. Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355.
de:PapD ratio of 25:1 was used in the NMR studies the ratio
between free and bound peptide is >25:1. Thus, the actual
chemical shift di€erence between the free and bound form
could be estimated to be ꢁ0.3 ppm, or even greater.
50. Davis, D. G.; Bax, A. J. Am. Chem. Soc. 1985, 107, 2820.
51. Jeener, J.; Meier, B. H.; Bachmann, P.; Ernst, R. R. J.
Chem. Phys. 1979, 71, 4546.
52. Kumar, A.; Ernst, R. R.; Wuthrich, K. Biochem. Biophys.
Res. Comm. 1980, 95, 1.
39. Gallop, M. A.; Barrett, R. W.; Dower, W. J.; Fodor, S. P. A.;
Gordon, E. M. J. Med. Chem. 1994, 37, 1233.
53. Macura, S.; Ernst, R. R. Mol. Phys. 1980, 41, 95.
40. Gordon, E. M.; Barrett, R. W.; Dower, W. J.; Fodor, S. P. A.;
Gallop, M. A. J. Med. Chem. 1994, 37, 1385.
54. Bothner-By, A. A.; Stephens, R. L.; Lee, J.-M.; Warren,
C. D.; Jeanloz, R. W. J. Am. Chem. Soc. 1984, 106, 811.
41. Thompson, L. A.; Ellman, J. A. Chem. Rev. 1996, 96, 555.
55. Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 63, 207.
56. Rance, M. J. Magn. Reson. 1987, 74, 557.
42. Fersht, A. R.; Shi, J. P.; Knill-Jones, J.; Lowe, D. M.;
Wilkinson, A. J.; Blow, D. M.; Brick, P.; Carter, P.; Waye,
M. M. Y.; Winter, G. Nature 1985, 314, 235.
57. Shaka, A. J.; Lee, C. J.; Pines, A. J. Magn. Reson. 1988, 77,
274.
43. Barlos, K.; Chatzi, O.; Gatos, D.; Stavropoulos, G. Int. J.
Peptide Protein Res. 1991, 37, 513.
58. States, D. J.; Haberkorn, R. A.; Ruben, D. J. J. Magn.
Reson. 1982, 48, 286.
44. Rink, H. Tetrahedron Lett. 1987, 28, 3787.
59. Leathers, V. L.; Linse, S.; Forsen, S.; Norman, A. W. J.
Biol. Chem. 1990, 265, 9838.
45. Bernatowicz, M. S.; Daniels, S. B.; Koster, H. Tetrahedron
Lett. 1989, 30, 4645.
60. Piotto, M.; Saudek, V.; Sklenar, V. J. Biomol. NMR. 1992,
661.
46. Aue, W. P.; Bartholdi, E.; Ernst, R. R. J. Chem. Phys.
1976, 64, 2229.
61. Hwang, T.-L.; Shaka, A. J. J. Magn. Reson. A. 1995, 112,
275.
47. Rance, M.; Sorensen, O. W.; Bodenhausen, G.; Wagner,
G.; Ernst, R. R.; Wuthrich, K. Biochem. Biophys. Res. Com-
mun. 1983, 117, 479.
62. Zhang, O.; Kay, L. E.; Olivier, J. P.; Forman-Kay, J. D. J.
Biomol. NMR. 1994, 4, 845.
48. Braunschweiler, L.; Ernst, R. R. J. Magn. Reson. 1983, 53,
521.
63. Cavanagh, J.; Rance, M. J. Magn. Reson. 1992, 96, 670.