Acta Crystallographica Section D
Editors: E. N. Baker and Z. Dauter
The F4 fimbrial chaperone FaeE is stable as a monomer that
does not require self-capping of its pilin-interactive surfaces
Inge Van Molle, Kristof Moonens, Lieven Buts, Abel Garcia-Pino, Santosh
Panjikar, Lode Wyns, Henri De Greve and Julie Bouckaert
Acta Cryst. (2009). D65, 411–420
Copyright c ? International Union of Crystallography
Author(s) of this paper may load this reprint on their own web site or institutional repository provided that
this cover page is retained. Republication of this article or its storage in electronic databases other than as
specified above is not permitted without prior permission in writing from the IUCr.
For further information see http://journals.iucr.org/services/authorrights.html
Acta Crystallographica Section D: Biological Crystallography welcomes thesubmissionof
papers covering any aspect of structural biology, with a particular emphasis on the struc-
tures of biological macromolecules and the methods used to determine them. Reports
on new protein structures are particularly encouraged, as are structure–function papers
that could include crystallographic binding studies, or structural analysis of mutants or
other modified forms of a known protein structure. The key criterion is that such papers
should present new insights into biology, chemistry or structure. Papers on crystallo-
graphic methods should be oriented towards biological crystallography, and may include
new approaches to any aspect of structure determination or analysis.
Crystallography Journals Online is available from journals.iucr.org
Acta Cryst. (2009). D65, 411–420Van Molle et al. · FaeE
Acta Cryst. (2009). D65, 411–420doi:10.1107/S0907444909005174
Acta Crystallographica Section D
The F4 fimbrial chaperone FaeE is stable as a
monomer that does not require self-capping of its
Inge Van Molle,a,b* Kristof
Henri De Grevea,band
aStructural Biology Brussels, Vrije Universiteit
Brussel, Pleinlaan 2, 1050 Brussels, Belgium,
bStructural Biology Brussels, Department of
Molecular and Cellular Interactions, VIB,
Pleinlaan 2, 1050 Brussels, Belgium, and
cEMBL Hamburg c/o DESY, Notkestrasse 85,
22603 Hamburg, Germany
Correspondence e-mail: firstname.lastname@example.org
# 2009 International Union of Crystallography
Printed in Singapore – all rights reserved
Many Gram-negative bacteria use the chaperone–usher
pathway to express adhesive surface structures, such as
fimbriae, in order to mediate attachment to host cells. Peri-
plasmic chaperones are required to shuttle fimbrial subunits or
pilins through the periplasmic space in an assembly-competent
form. The chaperones cap the hydrophobic surface of the
pilins through a donor-strand complementation mechanism.
FaeE is the periplasmic chaperone required for the assembly
of the F4 fimbriae of enterotoxigenic Escherichia coli. The
FaeE crystal structure shows a dimer formed by interaction
between the pilin-binding interfaces of the two monomers.
Dimerization and tetramerization have been observed pre-
viously in crystal structures of fimbrial chaperones and have
been suggested to serve as a self-capping mechanism that
protects the pilin-interactive surfaces in solution in the
absence of the pilins. However, thermodynamic and biochem-
ical data show that FaeE occurs as a stable monomer in
solution. Other lines of evidence indicate that self-capping of
the pilin-interactive interfaces is not a mechanism that is
conservedly applied by all periplasmic chaperones, but is
rather a case-specific solution to cap aggregation-prone
Received 11 December 2008
Accepted 12 February 2009
PDB References: FaeE, 3f65,
r3f65sf; 3f6i, r3f6isf; 3f6l,
The attachment of pathogenic bacteria to host cells is a key
event in their infection process and is typically mediated by
adhesins. These are often located on the bacterial surface in
polymeric proteinaceous appendages called fimbriae or pili
(Soto & Hultgren, 1999). It is a major challenge for Gram-
negative bacteria to generate these organelles on their surface,
a process that requires protein synthesis, folding, translocation
over two membranes and across the periplasmic space and
self-assembly into the final polymeric structure. Many Gram-
negative bacteria use the chaperone–usher pathway to tackle
this problem (Thanassi et al., 1998), employing two proteins
that will not be part of the final structure: a periplasmic
chaperone and an outer membrane gatekeeper known as the
usher. Once the fimbrial subunits have been translocated over
the inner membrane by the general secretory pathway, they
are bound by a cognate chaperone and guided to the usher,
the pilus-assembly platform in the outer membrane. Here,
they are assembled into the final adhesive structure and
transported across the outer membrane.
The crystal structures of the periplasmic chaperones PapD
(uropathogenic Escherichia coli P pili; Holmgren & Bra ¨nde ´n,
1989; Hung, Pinkner et al., 1999), FimC (uropathogenic E. coli
type 1 pili; Pellecchia et al., 1998), SfaE (uropathogenic E. coli
S pili; Knight et al., 2002), Caf1M (Yersinia pestis F1 antigen;
Zavialov et al., 2003) and SafB (Salmonella enterica Saf
fimbriae (Remaut et al., 2006) show two immunoglobulin (Ig)
domains oriented at a right angle in a boomerang shape.
Genetic studies have revealed several conserved residues that
occur throughout the family of PapD-like chaperones and are
important for their structure and function (Hung, Knight et al.,
1999). Those conserved residues can mainly be found in the
cleft between the two Ig domains.
Based on the length of the loop between the F1 and G1
strands, periplasmic chaperones have been classified as FGL
(long loop between Ig-fold ?-strands F and G) or FGS (short
FG loop) chaperones (Hung et al., 1996). Historically, FGS
chaperones were believed to be involved in the biosynthesis of
fimbriae with a complex subunit organization, presenting one
adhesin at the tip, while FGL chaperones assisted in the
assembly of polyadhesive structures consisting of mainly one
subunit. Following this classification, Zavialov and coworkers
assigned the F4 and the related F5 fimbriae as an intermediate
family of adhesive fibres: FGS chaperone-assembled poly-
adhesive fibres (Zavialov et al., 2007). The assembly of these
fimbriae is assisted by FGS chaperones, but although the
fimbrial operon contains ten genes the fimbrial structure is
dominantly composed of one major subunit that also deter-
mines the adhesive properties of the fimbriae. More recent
phylogenetic analysis based on usher ancestry revealed that
only the FGL organelles form a monophyletic group, while the
FGS structures can be divided into different clades (Nuccio &
Baumler, 2007). Following this classification, F4 and F5
fimbriae belong to the family of ? fimbriae.
The crystal structures of chaperone–pilin complexes show
that pilins have an incomplete Ig fold that lacks the seventh
?-strand. Upon translocation across the inner membrane, the
C-terminal carboxylateof the fimbrial subunits is bound by the
conserved Arg8 and Lys112 residues in the cleft of the
chaperone. Binding by the chaperone facilitates the release of
the pilins from the inner membrane and provides them with a
folding template. The pilins fold directly on the surface of the
chaperone through a ?-zippering mechanism along the G1
strand of the chaperone. By ?-strand pairing of the G1 strand
of the chaperone with the F strand of the pilin and the
insertion of a conserved pattern of hydrophobic residues into
the groove on the surface of the pilin, the chaperone com-
plements the fold of the pilin (Jones et al., 1997). In the case of
the FGL chaperones, in addition to the G1 strand of the
chaperone, the A1 strand is also involved in the donor-strand
complementation mechanism. The A1 strand interacts with
the A strand of the pilin (Zavialov et al., 2003). The chaperone
thus protects the pilins from aggregation, misfolding and
proteolytic degradation. Periplasmic chaperones are essential
for pilus assembly since they catalyze the folding of the pilins
(Vetsch et al., 2004) and keep them in an assembly-competent
conformation (Sauer et al., 2002; Zavialov et al., 2003).
In pilin-free chaperones, the hydrophobic residues of the
pilin-binding motif are surface-exposed. As several crystal
structures of periplasmic chaperones have revealed oligomers
with capped pilin-binding interfaces,a self-capping mechanism
by which the chaperones also protect their pilin-binding
surfaces in solution has been suggested (Hung, Pinkner et al.,
1999; Knight et al., 2002; Zavialov & Knight, 2007). The
Y. pestis Caf1M chaperone indeed forms tetramers in solution
that are created by tight packing of the subunit binding
surfaces into a hetero-sandwich. This self-capping mechanism
protects the chaperone from proteolytic degradation and
aggregation (Zavialov & Knight, 2007). Also, expression of a
Q108C mutant of PapD without the P-pilus subunits resulted
in the presence of up to 30% disulfide-linked PapDQ108C
dimers in the periplasm, showing dimer formation in vivo
(Hung, Pinkner et al., 1999).
Here, we describe the crystal structure of the F4 fimbrial
chaperone FaeE and apply differential scanning calorimetry,
chemical cross-linking and dynamic light scattering to explore
the self-capping mechanism of FaeE in solution. Additionally,
we try to answer the question whether self-capping is a general
mechanism applied by periplasmic chaperones or is a case-
specific solution to protect aggregation-prone surfaces.
2. Material and methods
2.1. Protein expression, purification and crystallization
The faeE or faeE-faeG genes (plasmid pHD163 or pHD147,
respectively) from the F4ad+E. coli strain C1360-79 were
expressed in E. coli C43 (DE3) under the T7 promoter. FaeE
and FaeE–FaeG were purified by ion-exchange chromato-
graphy in 20 mM TES pH 7.0 and elution with an NaCl
gradient in the same buffer and were finally used for crystal-
lization in 20 mM TES pH 7.0, 150 mM NaCl, all as described
previously (Van Molle et al., 2005). FaeE crystals were grown
in 0.1 M Tris pH 7.5, 50%(v/v) 2-methyl-2,4-pentanediol
(MPD), 0.2 M NH4H2PO4. FaeE crystallized in three crystal
forms, all of which belonged to space group C2. Crystallization
drops set up with FaeE at 20 mg ml?1were found to contain
crystals of both form 1 (unit-cell parameters a = 195.7, b= 78.5,
c = 184.6 A˚, ? = 102.2?) and form 2 (unit-cell parameters
a = 136.4, b= 75.7, c= 69.4 A˚, ?= 92.8?). FaeESeMetcrystals (set
up at 25 mg ml?1) belonged to crystal form 2. Crystals of form
3 (unit-cell parameters a = 109.7, b= 78.6, c= 87.8 A˚, ?= 96.4?)
were found in drops originally set up with the FaeE–FaeG
complex at 15 mg ml?1. In these drops only FaeE crystallized,
while FaeG polymerized and precipitated (Van Molle et al.,
2.2. X-ray data collection and structure determination
X-ray data were collected on the ESRF beamline ID14-1
(Grenoble, France) for crystal form 1 and the DESY/EMBL
beamline BW7A (Hamburg, Germany) for crystal forms 2 and
3 (Van Molle et al., 2005). An initial partial FaeE model was
obtained by phase determination using a three-wavelength
MAD data set from the FaeESeMetcrystal form 2 (Van Molle et
Van Molle et al.
Acta Cryst. (2009). D65, 411–420
al., 2005) using the Auto-Rickshaw software pipeline (Panjikar
et al., 2005). Substructure determination, phase calculation,
solvent flattening and model building were performed auto-
matically using SHELXD (Sheldrick, 2008), MLPHARE
(Collaborative Computational Project, Number 4, 1994), DM
(Cowtan & Zhang, 1999) and ARP/wARP (Perrakis et al.,
2001), respectively, within Auto-Rickshaw (Panjikar et al.,
2005). This initial model was further completed by manual
model building in Coot (Emsley & Cowtan, 2004) and was
refined against the absorption-edge data to 2.7 A˚resolution
(Table 1) using REFMAC5 (Murshudov et al., 1997) from the
CCP4 suite (Collaborative Computational Project, Number 4,
1994) and phenix.refine (Adams et al., 2002) using TLS
restraints. Chain A of this model was used to search for
molecular-replacement solutions of the FaeE structures for the
data sets of crystal forms 1 and 3 using Phaser from the CCP4
suite. Molecular-replacement solutions containing eight or two
monomers in the asymmetric unit were searched for using a
3.5 A˚resolution cutoff for crystal forms 1 and 3, respectively.
The models obtained were improved by iterative manual
model building in Coot and refinement using REFMAC5 and
phenix.refine using TLS restraints to 2.3 and 2.8 A˚resolution
for crystal forms 1 and 3, respectively.
2.3. Analysis of packing interfaces in the crystal structures of
FaeE, PapD, SfaE and Caf1M
The FaeE structures were validated using MolProbity
(Table 1; Davis et al., 2007) and the coordinates for the
structures of FaeE in crystal forms 1, 2 and 3 were submitted to
the Protein Data Bank (PDB codes
3f65, 3f6i and 3f6l, respectively). The
root-mean-square deviations (r.m.s.d.s)
between the FaeE structures in crystal
forms 1, 2 and 3 were calculated using
the LSQKAB program from the CCP4
suite (Kabsch, 1976). The interfaces
present in the crystal structures of FaeE
(PDB codes 3f65, 3f6i and 3f6l), PapD
(1qpp), SfaE (1l4i) and Caf1M (2os7)
were analyzed using the PISA program
from the EMBL–EBI website (Krissinel
& Henrick, 2007).
2.4. Differential scanning calorimetry
All calorimetric measurements were
performed on a MicroCal VP-DSC
differential scanning microcalorimeter
with a 0.515 ml cell and data were
analyzed using the program MicroCal
Origin DSC 7.0. All samples were
filtered using 0.45 mm Minisart filters
(Sartorius) and degassed for 10 min
using a vacuum pump. DSC scans
followed heating of the samples from
293 to 353 K. Scan rates of 90, 75, 60, 30
and 15 K h?1were applied to assess the scan-rate dependence
of the transition temperature (Tm). To evaluate the pH-
dependence of Tm, DSC scans were recorded in 20 mM
sodium acetate pH 5.2, 20 mM phosphate pH 7.5 and 20 mM
N-cyclohexyl-3-aminopropanesulfonic acid (CAPS) pH 10.5
DSC thermograms of the denaturation of FaeE were
analyzed assuming a two-state reversible transition,
ð1 ? ?Þ¼ exp
where K is the equilibrium denaturation constant, R is the
universal gas constant and ? is the fraction of protein in the
The model function for the DSC thermograms (Cp? Cp,N
can be obtained from the first partial derivative of the
enthalpy of the denaturation function,
pþ?ð1 ? ?Þ
pðT ? T1=2Þ?2;
where ?Hvis the van’t Hoff enthalpy of denaturation at the
transition temperature (T1/2), ? is the fraction of denatured
protein and ?Cp
which is assumed to be independent of the temperature. The
thermodynamic parameters that characterize the melting
curves (?Hv, ?Cp
function for the DSC endotherms to the experimental
temperature profiles using the MicroCal Origin DSC 7.0
?is the corresponding heat-capacity change,
?and T1/2) were obtained by fitting the model
Acta Cryst. (2009). D65, 411–420Van Molle et al.
Data-collection and refinement statistics.
Values in parentheses are for the highest resolution shell.
Crystal form 1Crystal form 2Crystal form 3
Resolution range (A˚)
No. of unique reflections
Rfree(5% test set)
No. of protein atoms
No. of solvent atoms
Average B factor (A˚2)
Residues in allowed regions (%)
Residues in disallowed regions (%)
ijIiðhklÞ ? hIðhklÞij=P
iIiðhklÞ, where Ii(hkl) is the observed intensity and hI(hkl)i is the average
intensity for symmetry-related reflections.
2.5. Chemical cross-linking
FaeE at 0.75 mg ml?1concentration was incubated with
different concentrations of glutaraldehyde in 20 mM phos-
phate buffer pH 7.5 for 30 min at room temperature. The total
reaction volume was 100 ml. The reaction was stopped by
adding 10 ml 1 M Tris–HCl pH 8.0. Cross-linking products
were analyzed on SDS–PAGE.
2.6. Dynamic light scattering (DLS)
Prior to DLS measurements, FaeE samples (20 mg ml?1)
were cleared from dust and air by centrifugation for 1 h at
277 K and 14 000 rev min?1in a microcentrifuge. DLS
measurements were conducted on a RiNA Laser-Spectro-
scatter 210 (Netzwerk RNA-Technologien GmbH, Germany).
To assess the particle-size distribution in the sample, ten
measurements of 10 s each were performed.
2.7. Prediction of b-aggregation-prone regions
The Caf1M (UniProtKB entry P26926), FaeE (P25401),
PapD (P15319), FimC (P31697), SfaE (Q9EXJ6) and SafB
(Q8ZRK3) amino-acid sequences were analyzed using the
Escamilla et al., 2004), with input parameters set at physio-
logical conditions (pH 7, ionic strength 0.02 M). Protein
stability was set at ?42 kJ mol?1.
3. Results and discussion
3.1. Structure of the chaperone FaeE
Although the structure of FaeE was solved in the three
crystal forms previously reported (Van Molle et al., 2005), we
will mainly discuss the FaeE structure from crystal form 1,
which was determined at the highest resolution. Details of
data-collection and refinement statistics are listed in Table 1.
Crystal form 1 contains eight molecules in the asymmetric
unit. The asymmetric units of crystal forms 2 and 3 each
contained two molecules. The FaeE structure from crystal
form 1 has an r.m.s.d. with the FaeE structures of crystal forms
2 and 3 of 1.04 and 0.88 A˚, respectively (199 and 204 C?atoms
superposed, respectively). FaeE shows the typical boomerang-
shaped structure containing two immunoglobulin domains
that has been observed for all periplasmic chaperones. In the
first domain, strands A10, A100, B1, C1, D10, D100, E1, F1 and
G1 are organized in two sheets, one containing strands A10,
B1, D10and E1 and the other containing strands C1, D100, F1
and G1. The second domain also comprises two sheets: one
containing strands A2, B2 and D2 and one containing strands
C2, E2 and F2. A ?-helix connects strands C2 and D2 (Fig. 1).
PapD and FaeE are longer than the other FGS chaperones. In
the structure of PapD, this results in an additional ?-strand in
the second domain, H2 (Holmgren et al., 1992). In the FaeE
structures, there is no clear electron density for residues 205–
224 corresponding to strand G2 and
The structure of the FGS chaperones
and particularly the relative orientation
of the two domains is stabilized by salt
bridges (Hung, Knight et al., 1999). A
salt-bridge network is also present in
FaeE and is formed between Glu80,
Arg112 and Asp193 (Fig. 1). The side
chain of Glu80 forms an additional
hydrogen bond to the main-chain N
atom of Ile146.
As in the other FGS chaperones, the
conserved residues Arg8 and Lys108
(Arg8 and Lys112 in the PapD nomen-
clature) are located at the bottom of the
pilin-binding anchoring cleft. The Arg8
side chain points away from the cleft
and reorients upon interaction with the
pilin, allowing anchoring of the pilin by
of the pilin (Kuehn et al., 1993). In the
free FaeE chaperone, Arg8 makes a
hydrogen bond to the main-chain O
atom of Asp193 (Fig. 1).
The eight molecules in the asym-
metric unit of crystal form 1 show a high
diversity in loop visibility. Monomers A,
B, C and G are complete up to residue
206, but the electron density is missing
for a major part of the C-terminal
Van Molle et al.
Acta Cryst. (2009). D65, 411–420
Ribbon diagram of the FaeE monomer (left), as present in all three crystal forms, highlighting its
pilin-anchoring cleft (right). The conserved salt bridges between residues Glu80, Arg112 and
Asp193 (Glu83, Arg116 and Asp196 in PapD nomenclature) and the hydrogen bonds formed
between Glu80 and Ile146 and between Arg8 and Asp193 are indicated. Arg8 and Lys108 are the
conserved subunit-anchoring residues. The chaperone G1 strand extends beyond its antiparallel
?-strand interactions with the F1 strand because of crystal-packing interactions in the chaperone
dimer (Fig. 2).
Acta Cryst. (2009). D65, 411–420Van Molle et al.
(a) Detailed view of the interface between the FaeE monomers in FaeE dimer 1 (left). The dimer is formed by extensive hydrogen bonding between the
G1 strands of both monomers. An additional hydrogen bond is formed between Lys106 in the G1 strand of one monomer and Leu94 in the F1–G1 loop
of the other monomer. The F1 and G1 strands and the F1–G1 loops of both monomers are indicated in the inset on the left; the hydrogen-bonding
residues of the G1 strands and F1–G1 loops of both monomers are indicated in the enlarged view. (b) The PapD dimer (PDB code 1qpp) is formed
around a noncrystallographic twofold axis, orienting the two monomers at an almost right angle. Hydrogen bonds are found between the Gln108 residues
and the Thr109 and Ala106 residues of both monomers. (c) The SfaE dimer (PDB code 1l4i) is maintained by hydrophobic packing of the N-terminal
domain of one monomer against the subunit-binding interface of the other monomer. The Phe105 side chain of one monomer (green) intercalates
between theA10andG1 strands of the othermonomer (cyan).Main-chain hydrogen bonds between Ala3 (cyan)and Ala103 (green) and between Ser109
(cyan) and Phe105 (green) are also indicated.
domain of monomers D, F and H. Monomer E has an inter-
mediate completeness. Specifically, the loops between strands
A2 and B2 and between D2 and E2 (Fig. 1) show no or poor
electron density in those monomers that make fewer packing
contacts.In monomers A, B, C and G these loops are stabilized
by packing with symmetry-related molecules.
3.2. Dimer interfaces in the crystal structures of FaeE
The three crystal forms reveal a dimeric FaeE structure
formed through antiparallel ?-strand hydrogen bonding
between the G1 strands of two FaeE monomers (FaeE dimer
1, Fig. 2a). Additionally, hydrogen bonds are formed between
Lys106 N?in one monomer and the main-chain O atom of
Leu94 in the F1–G1 loop of the other monomer. The G1
strand and the F1–G1 loop are also involved in pilin binding
by the chaperone. The AB dimer of crystal form 1 was chosen
for the study of FaeE dimerization because it had been
determined with the highest completeness. It has r.m.s.d.
deviations of 2.07 and 1.09 A˚with the dimers present in crystal
forms 2 and 3, respectively (383 and 408 C?atoms superposed,
respectively). The initial complexation of FaeE with FaeG in
the solution used to obtain crystals of form 3 does not influ-
ence the dimerization of FaeE in the crystal.
A second type of dimer (FaeE dimer 2), created by crystal
packing, can be observed in the asymmetric unit of crystal
form 1 between monomers A and D and between monomers E
and H (Fig. 3). This dimer is also present in crystal forms 2 and
3 and is formed throughcrystal packing with symmetry-related
molecules. FaeE dimer 2 is formed around a noncrystallo-
graphic twofold-symmetry axis, allowing the C1–D10loop of
one monomer to contact the C1 strand of the other monomer.
Hydrogen bonds are formed between Thr45 and Asp38, Thr45
and Trp36, Arg46 and Gln34 and between Pro47 and Gln34. In
addition, Asp44 forms two coordinated salt bridges to Arg104.
The contact area between the two monomers in FaeE dimer
2 is larger than the interface area in dimer 1 (698.5 versus
615.2 A˚2). Despite this fact, the AB dimer is the more relevant
one to study, because this dimer has also been observed for
other fimbrial chaperones such as PapD and SfaE (Hung,
Pinkner et al., 1999; Knight et al., 2002). In all observed cases
the subunit binding G1 strands and F1-G1 loops are impli-
cated in this dimer.
Although crystal packing shields the subunit-binding
sequences by dimer formation in every case, superposition of
FaeE dimer 1 and the SfaE and PapD dimers shows that the
dimer interfaces are very different (Fig. 2). Ten hydrogen
bonds are formed between main-chain residues of the G1
strands of the FaeE monomers in
FaeE dimer 1 (Fig. 2a). Compared
with the well aligned G1 strands
of the FaeE monomers, the G1
strands of the PapD monomers
are slightly twisted (Fig. 2b).
Therefore, only a limited number
of hydrogen bonds are formed
between the PapD G1 strands
(Hung, Pinkner et al., 1999). In
the SfaE dimer, the N-terminal
domain of one monomer (mono-
mer A; green in Fig. 2c) is packed
of the other (monomer B; cyan in
Fig. 2c; Knight et al., 2002). The
G1 strand of the first monomer
crosses the A1 strand of the
second monomer at nearly right
angles. Phe105, which is part of
the conserved pilin-binding patch
of the G1 strand of monomer A, is
intercalated between the A10and
G1 strands of monomer B.
Dimerization of PapD occurs
through interaction of the F1–G1
and C2–D2 loops of both mono-
mers and significant conforma-
tional changes in those loops can
be seen when comparing mono-
meric and dimeric PapD (Hung,
Pinkner et al., 1999). There is also
a difference in the F1–G1 loop
Van Molle et al.
Acta Cryst. (2009). D65, 411–420
Superposition of both dimer interfaces formed by monomer A in crystal form 1. FaeE dimer 1 is formed
between monomers A (green) and B (cyan), while FaeE dimer 2 is formed between monomers A and D
(yellow). The inset shows a detailed view of the hydrogen bonds between the C1–D10loop of chain A and
the C1 strand of chain D, as well as the salt bridge formed between Asp44A and Arg104D. The
intramolecular hydrogen bond between Asp38D O?2and Trp36D N?2is also indicated. Here again the G1
strand of both monomers is elongated beyond the contacts it makes with the F1 strand because of the
contacts it makes within the FaeE dimer 1 interface, as shown for the FaeE dimer 1 formed between
monomers A and B. This figure also demonstrates that the C-terminal domain of monomer D is
conformation when comparing dimerized and subunit-bound
PapD. The F1–G1 loop of PapD is in a more elongated
conformation in the PapD–PapK complex and is buried in the
groove of PapK. This conformation cannot be adopted in the
PapD dimer because of steric hindrance (Hung, Pinkner et al.,
1999). In FaeE dimer 1, the F1–G1 and C2–D2 loops of both
monomers make no contacts (Fig.2). As mentioned above, the
G1 strand extends beyond the contacts it makes with the F1
strand because of contacts in the dimer interface. It can thus
be presumed that the G1 strand is shorter in monomeric FaeE.
In the SfaE dimer structure, the F1–G1 loop is invisible and no
structural information is available about monomeric or
subunit-bound SfaE (Knight et al., 2002).
3.3. The oligomerization state of FaeE in solution
Subunit binding by periplasmic chaperones occurs through
the complementation of the hydrophobic core of the subunit
by the G1 strand of the chaperone (plus the A1 strand for the
FGL chaperones), which consists of a pattern of alternating
hydrophobic residues (Choudhury et al., 1999; Sauer et al.,
1999; Zavialov et al., 2003). Upon subunit binding, the often
bulky side chains of these residues are inserted into hydro-
phobic pockets on the surface of the pilin subunit. When the
chaperone is not in complex with a pilin, these hydrophobic
side chains are exposed to the solvent. Chaperone oligomer-
ization could therefore serve as a self-capping mechanism that
serves to protect the chaperone from aggregation or degra-
dation when not in complex with a pilin (Hung, Pinkner et al.,
1999; Knight et al., 2002; Zavialov & Knight, 2007). Such a
self-capping mechanism would be confirmed by chaperone
oligomerization in solution. It has been shown that the Caf1M
F1-antigen chaperone from Y. pestis tetramerizes (Zavialov &
Knight, 2007) in solution and that PapD dimerizes (Hung,
Pinkner et al., 1999). In order to investigate FaeE dimerization
in solution, we conducted differential scanning calorimetry
(DSC), glutaraldehyde cross-linking and dynamic light-
scattering (DLS) measurements.
Equilibrium thermodynamics could be used to analyze the
DSC data for FaeE at all pH values when samples were heated
to 353 K, the point at which the unfolding of FaeE is com-
pleted (Fig. 4). The folding–unfolding transition is either
almost completely (96% at pH 5.2) or partially reversible
(60% at pH 7.5 and 77% at pH 10.5). Moreover, the transition
temperature is independent of the scan rate, indicating that all
irreversible processes occur after the unfolding transition. This
allowed equilibrium thermodynamics to be applied even in
those cases where the transition is not completely reversible.
All thermodynamic parameters (?CP
calculated from the fitting of the model function to the
experimental DSC data (Table 2).
DSC measurements are a valuable tool to investigate the
oligomerization state of a protein in solution. In one approach,
the dependence of the DSC profile on the protein concen-
tration can be analyzed. The DSC transition of an oligomeric
protein depends on the protein concentration because the
energy required for the disruption of the intermolecular
interactions depends on the protein concentration. DSC
measurements of FaeE at different concentrations show that
the DSC profile is independent of the protein concentration
(Fig. 4 and Table 2) and thus suggest that FaeE is a monomer.
To assess the influence of pH on the DSC transitions, the
measurements were repeated at pH 7.5 and 10.5. These results
confirmed the independence of the T1/2 of the protein
concentration (Table 2) and thus the monomeric nature of
FaeE. Using a second approach, the ratio of the calorimetric
and van’t Hoff enthalpies determined from the DSC profiles
(?Hcand ?Hv, respectively) gives an estimate of the oligo-
merization state of a given protein. ?Hc expresses the
enthalpy of denaturation per mole and ?Hvthe enthalpy of
denaturation per cooperative unit. Consequently, the ratio
between ?Hcand ?Hvcan be seen as the number of co-
?, T1/2and ?Hv) can be
Acta Cryst. (2009). D65, 411–420Van Molle et al.
Thermal denaturation of FaeE in 20 mM acetate buffer pH 5.2. All scans
were recorded at 90 K h?1.
Thermodynamic parameters of the thermal denaturation of the
chaperone FaeE at different pH values and concentrations.
The error on T1/2is 0.5 K and the error on ?Hvis 13 kJ mol?1. For pH 5.2 and
pH 10.5, ?Cp
constant in the fitting of the DSC profiles for the other protein concentrations.
For pH 7.5, ?Cp
kept constant in the fitting of the DSC profile at 0.06 mg ml?1. For pH 5.2 +
5% PEG 6000, ?Cp
fitting of the DSC profiles for the other protein concentrations.
?was determined for the highest protein concentration and kept
?was determined at 0.33 mg ml?1and at 0.13 mg ml?1and
?was determined at 0.2 mg ml?1and kept constant in the
5.2 + 5%
operative units per mole. This ratio is always about 1.0 for
FaeE (Table 2). This also points out that both the FaeE
N-terminal and C-terminal domains unfold simultaneously as
one cooperative unit. In conclusion, both DSC approaches
identify FaeE as a monomer in solution.
Chemical cross-linking and dynamic light-scattering (DLS)
experiments add further evidence for the stable nature of
monomeric FaeE. SDS–PAGE analysis of FaeE samples
incubated with glutaraldehyde showed the formation of FaeE
oligomers with increasing glutaraldehyde concentration,
without the accumulation of a particular oligomeric species
(Fig. 5). This is in contrast to the Caf1M chaperone, for which
chemical cross-linking clearly showed accumulation of the
tetrameric species without the formation of higher oligomers
(Zavialov et al., 2007), and the PapD chaperone, which
demonstrated the accumulation of PapD dimers (Hung,
Pinkner et al., 1999). Thus, the absence of the accumulation of
a dimeric FaeE species in the cross-linking experiment
provides additional evidence for the monomeric nature of
FaeE. Finally, the monomeric character of FaeE was confirmed
by comparing the Stokes radius of FaeE determined by DLS
measurements prior to crystallization (2.5 nm; Fig. 5) with the
radius of the FaeE monomer in its crystal structures (3.3 nm
for a monomer, 5 nm for a dimer).
Overall, our results demonstrate that FaeE can occur as a
monomer in solution and does not require self-capping for
stabilization and that FaeE dimerization is merely a result of
crystallization effects. At the high concentrations used, crys-
tallization of FaeE might only be favoured upon aggregation
when the hydrophobic surfaces are capped, in this case by
dimerization. The complexation significance score (CSS)
calculated by the PISA algorithm (0.0) indeed indicates that
the dimers formed inthe crystal structure are aresult of crystal
packing only. The same is true for the dimers present in the
crystal structures of PapD and SfaE (CSS of 0.0 and 0.1,
respectively). In contrast, the CSS for the Caf1M tetramer is
1.0, indicating that the tetramer present in the crystal structure
is representative of oligomerization in solution. As during
crystallization, in vivo crowding effects in the periplasm might
urge the chaperone to shield its hydrophobic surfaces. In order
to investigate the effect of molecular crowding on the dimer-
ization of FaeE, the DSC experiments were repeated in the
presence of 5% PEG 6000. This technique was previously
applied by Zavialov & Knight (2007) to show the effect of
crowding on the tetramerization of the Caf1M chaperone. This
concentration of PEG 6000 clearly stimulated Caf1M tetra-
merization. However, the DSC profile of FaeE in 5% PEG
6000 is independent of the protein concentration (Table 2),
supporting the fact that FaeE dimerization is not required for
the stability of the chaperone when it is not in complex with a
The oligomerization of the Caf1M chaperone seems to be a
result of the high ?-aggregation propensity of its G1 strand
(Fig. 6a; Zavialov & Knight, 2007). The sequence-based
statistical algorithm TANGO (Fernandez-Escamilla et al.,
2004) does not predict any ?-aggregation propensity for the
G1 strands of PapD, FimC and FaeE, supporting the notion
that they could occur as stable monomers without the need for
capping of their interactive surfaces. Direct evidence for the
monomeric nature of the type 1 fimbrial chaperone FimC in
solution comes from its NMR structure (Pellecchia et al.,
1998). The fast amide-proton exchange for the F1–G1 loop
confirmed that it is solvent-exposed. The fact that 70% of the
PapDQ108C was still present as monomers and that no dimers
were found for wild-type PapD (Hung, Pinkner et al., 1999)
indicates that dimer formation is not strictly required for
PapD stability. Additionally, the PapD chaperone was also
found as a monomer in its earliest crystal structure (Holmgren
& Bra ¨nde ´n, 1989), as was the SafB chaperone (H. Remaut,
personal communication). Surprisingly, the TANGO predic-
tions for PapD and FaeE are very similar and both indicate the
same stretch of residues as prone to ?-aggregation (residues
149–156 for PapD and 145–153 for FaeE, in both cases
belonging to the B2–C2 loop and the C2 strand; Figs. 6b and
6c). These residues are not involved in any crystal-packing
Van Molle et al.
Acta Cryst. (2009). D65, 411–420
(a) Chemical cross-linking of FaeE at 0.75 mg ml?1concentration
incubated with varying concentrations of glutaraldehyde as indicated at
the top [%(w/v)]. Lane M, protein molecular-weight markers; molecular
weights are indicated in kDa on the left. (b) Histogram and distribution
profile (inset) of the DLS measurement on FaeE at 20 mg ml?1, clearly
showing a monodisperse distribution of particles with a Stokes radius of
interfaces in the crystal structures. For SfaE, TANGO predicts
a high ?-aggregation propensity for the dimerization interface
in the crystal structure, which could support SfaE dimerization
in solution. However, no evidence is present for this.
4. Concluding remarks
The pilin-free form of the F4 fimbrial chaperone FaeE is
predominantly present as a monomer in solution. In its crystal
structure, FaeE dimerizes by means of antiparallel ?-strand
pairing of its pilin-interactive G1 strands (FaeE dimer 1). This
form of oligomerization has also been observed in the crystal
structures of PapD and SfaE and tetramers have been
observed both in crystals and in solution for the Caf1M
chaperone. However, we find that oligomerization or self-
capping of fimbrial chaperones, which has been suggested to
protect the chaperone from degradation, is not an absolute
requirement for FaeE stability in solution, nor is it a general
mechanism to shield the pilin-interactive motif in free peri-
plasmic chaperones. Oligomerization of the Caf1M chaperone
seems to be the result of the high ?-aggregation propensity of
its G1 strand. Understandably, the presence of an excess of
free periplasmic chaperone, conditional for its crystallization,
would be rather unusual during pilus biogenesis, hence the
need for self-capping and thus protection of free chaperone
would be minor.
Acta Cryst. (2009). D65, 411–420 Van Molle et al.
TANGO ?-aggregation propensity prediction for Caf1M (a), FaeE (b), PapD (c), FimC (d), SafB (e) and SfaE(f). Regions with the highest ?-aggregation
propensity are coloured red. For PapD, a second region with ?50% ?-aggregation propensity is coloured orange. A cutoff value of 30% ?-aggregation
propensity was used. The Caf1M, FaeE, PapD, FimC, SafB and SfaE structures were deduced from PDB entries 1p5v, 3f65, 1pdk, 1qun, 2co7 and 1l4i,
respectively. FGL chaperones are coloured green and FGS chaperones are coloured cyan. The vertical axis in the insets indicates the percentage of
?-aggregation propensity, while the horizontal axis gives the residue number.
The authors acknowledge the use of the EMBL beamline Download full-text
BW7A at the DESY synchrotron (Hamburg, Germay) and of
ESRF beamline ID14-1 (Grenoble, France). The Fonds voor
Wetenschappelijk Onderzoek-Vlaanderen is thanked for
doctoral and postdoctoral grants for KM and LB, and for grant
G.0389.07. The authors thank Dr Han Remaut for carefully
reading the manuscript.
Adams, P. D., Grosse-Kunstleve, R. W., Hung, L.-W., Ioerger, T. R.,
McCoy, A. J., Moriarty, N. W., Read, R. J., Sacchettini, J. C., Sauter,
N. K. & Terwilliger, T. C. (2002). Acta Cryst. D58, 1948–1954.
Choudhury, D., Thompson, A., Stojanoff, V., Langermann, S.,
Pinkner, J., Hultgren, S. J. & Knight, S. D. (1999). Science, 285,
Collaborative Computational Project, Number 4 (1994). Acta Cryst.
Cowtan, K. D. & Zhang, K. Y. (1999). Prog. Biophys. Mol. Biol. 72,
Davis, I. W., Leaver-Fay, A., Chen, V. B., Block, J. N., Kapral, G. J.,
Wang, X., Murray, L. W., Arendall, W. B. III, Snoeyink, J.,
Richardson, J. S. & Richardson, D. C. (2007). Nucleic Acids Res. 35,
Emsley, P. & Cowtan, K. (2004). Acta Cryst. D60, 2126–2132.
Fernandez-Escamilla, A. M., Rousseau, F., Schymkowitz, J. &
Serrano, L. (2004). Nature Biotechnol. 22, 1302–1306.
Holmgren, A. & Bra ¨nde ´n, C. I. (1989). Nature (London), 342,
Holmgren, A., Kuehn, M. J., Bra ¨nde ´n, C. I. & Hultgren, S. J. (1992).
EMBO J. 11, 1617–1622.
Hung, D. L., Knight, S. D. & Hultgren, S. J. (1999). Mol. Microbiol. 31,
Hung, D. L., Knight, S. D., Woods, R. M., Pinkner, J. S. & Hultgren,
S. J. (1996). EMBO J. 15, 3792–3805.
Hung,D. L., Pinkner, J. S., Knight, S. D. & Hultgren, S. J. (1999). Proc.
Natl Acad. Sci. USA, 96, 8178–8183.
Jones, C. H., Danese, P. N., Pinkner, J. S., Silhavy, T. J. & Hultgren, S. J.
(1997). EMBO J. 16, 6394–6406.
Kabsch, W. (1976). Acta Cryst. A32, 922–923.
Knight, S. D., Choudhury, D., Hultgren, S., Pinkner, J., Stojanoff, V. &
Thompson, A. (2002). Acta Cryst. D58, 1016–1022.
Krissinel, E. & Henrick, K. (2007). J. Mol. Biol. 372, 774–797.
Kuehn, M. J., Ogg, D. J., Kihlberg, J., Slonim, L. N., Flemmer, K.,
Bergfors, T. & Hultgren, S. J. (1993). Science, 262, 1234–1241.
Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Acta Cryst.
Nuccio, S. P. & Baumler, A. J. (2007). Microbiol. Mol. Biol. Rev. 71,
Panjikar, S., Parthasarathy, V., Lamzin, V. S., Weiss, M. S. & Tucker,
P. A. (2005). Acta Cryst. D61, 449–457.
Pellecchia, M., Gu ¨ntert, P., Glockshuber, R. & Wu ¨thrich, K. (1998).
Nature Struct. Biol. 5, 885–890.
Perrakis, A., Harkiolaki, M., Wilson, K. S. & Lamzin, V. S. (2001).
Acta Cryst. D57, 1445–1450.
Remaut, H., Rose, R. J., Hannan, T. J., Hultgren, S. J., Radford, S. E.,
Ashcroft, A. E. & Waksman, G. (2006). Mol. Cell, 22, 831–842.
Sauer, F. G., Fu ¨tterer, K., Pinkner, J. S., Dodson, K. W., Hultgren, S. J.
& Waksman, G. (1999). Science, 285, 1058–1061.
Sauer, F. G., Pinkner, J. S., Waksman, G. & Hultgren, S. J. (2002). Cell,
Sheldrick, G. M. (2008). Acta Cryst. A64, 112–122.
Soto, G. E. & Hultgren, S. J. (1999). J. Bacteriol. 181, 1059–1071.
Thanassi, D. G., Saulino, E. T. & Hultgren, S. J. (1998). Curr. Opin.
Microbiol. 1, 223–231.
Van Molle, I., Buts, L., Coppens, F., Qiang, L., Wyns, L., Loris, R.,
Bouckaert, J. & De Greve, H. (2005). Acta Cryst. F61, 427–431.
Vetsch, M., Puorger, C., Spirig, T., Grauschopf, U., Weber-Ban, E. U.
& Glockshuber, R. (2004). Nature (London), 431, 329–333.
Zavialov, A. V., Berglund, J., Pudney, A. F., Fooks, L. J., Ibrahim,
T. M., MacIntyre, S. & Knight, S. D. (2003). Cell, 113, 587–596.
Zavialov, A. V. & Knight, S. D. (2007). Mol. Microbiol. 64, 153–
Zavialov, A., Zav’yalova, G., Korpela, T. & Zav’yalov, V. (2007).
FEMS Microbiol. Rev. 31, 478–514.
Van Molle et al.
Acta Cryst. (2009). D65, 411–420