The crystal structure of NlpI
A prokaryotic tetratricopeptide repeat protein with a globular fold
Christopher G. M. Wilson
1
, Tommi Kajander
1
and Lynne Regan
1,2
1 Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT, USA
2 Department of Chemistry, Yale University, New Haven, CT, USA
Repeat proteins in general, and tetratricopeptide
repeats (TPRs) in particular, have recently attracted
interest from the perspectives of structure, function,
folding and design [1–6]. The TPR was first identified
during sequence analysis of proteins CDC23 and
nuc2+ from yeast [7,8], and has subsequently been
found in a wide variety of polypeptides from all gen-
era. It is a degenerate 34-residue motif, which adopts a
helix-turn-helix structure. The first helix is usually
termed the ‘A’ helix, while the second is referred to as
the ‘B’ helix [6]. The most common number of tandem
TPRs within a single protein is three, but as many as
16 have been predicted on the basis of sequence analy-
sis [2]. Natural and designed TPRs whose structures
have been determined share a common tertiary organ-
ization, which is dominated by interactions between A
helices and the preceding AB pair (Fig. 1A). Local
AB, BA¢ and nonlocal AA¢ helix packing generate an
extended superhelical array with right-handed twist.
The motif is often terminated by an additional A, or
‘capping helix’, whose exposed edge is hydrophilic in

character, thus promoting favourable interactions with
solvents. TPRs are therefore distinct from globular
proteins, because they do not possess a single hydro-
phobic core. Instead, stabilizing interactions are distri-
buted throughout the structure.
One significant consequence of TPR superhelicity is
the creation of concave (‘front-face’) and convex (‘back-
face’) surfaces. In the best-characterized examples to
date, the 3-TPR domains TPR1 and TPR2A of
human Hsp organizing protein (HOP), which bind the
C-terminal residues of Hsc70 and Hsp90, respectively,
ligand recognition occurs at the concave front-face
[9,10]. Each of the 3-TPR domains of HOP behave
as structurally and functionally independent domains
in vitro. In vivo, functioning independently as part of
Keywords
crystal structure; NlpI; lipoprotein;
tetratricopeptide; TPR
Correspondence
L. Regan, Yale University, PO Box 208114,
New Haven, CT 06520–8114, USA
Fax: +1 203432 5767
Tel: +1 203432 5566
E-mail:
(Received 7 September 2004, revised 19
September 2004, accepted 19 September
2004)
doi:10.1111/j.1432-1033.2004.04397.x
There are several different families of repeat proteins. In each, a distinct
structural motif is repeated in tandem to generate an elongated structure.

The nonglobular, extended structures that result are particularly well suited
to present a large surface area and to function as interaction domains.
Many repeat proteins have been demonstrated experimentally to fold and
function as independent domains. In tetratricopeptide (TPR) repeats, the
repeat unit is a helix-turn-helix motif. The majority of TPR motifs occur as
three to over 12 tandem repeats in different proteins. The majority of TPR
structures in the Protein Data Bank are of isolated domains. Here we pre-
sent the high-resolution structure of NlpI, the first structure of a complete
TPR-containing protein. We show that in this instance the TPR motifs do
not fold and function as an independent domain, but are fully integrated
into the three-dimensional structure of a globular protein. The NlpI struc-
ture is also the first TPR structure from a prokaryote. It is of particular
interest because it is a membrane-associated protein, and mutations in it
alter septation and virulence.
Abbreviations
HOP, Hsp organizing protein; Hsp, heat shock protein; SUPR, superhelical peptide repeat; TPR, tetratricopeptide repeat.
166 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS
the full-length HOP, they act to facilitate the assembly
of multichaperone regulatory complexes. The struc-
tural independence of these TPR domains, and the
presence of independent ligand-binding sites in each,
has been assumed to be characteristic properties of
TPR domains. Methods that identify motifs from
amino acid sequence (e.g. pFAM [11]) readily predict
TPRs, with the implication that they are discrete
domains. TPR domains, or even subsets of TPRs
within a domain [12], are often studied independently.
In the course of a wider effort toward understanding
TPR structure and function, a number of related
observations intrigued us. First, the only structures of

natural TPRs in the PDB (13 TPR-containing coordi-
nate sets) consist of nonglobular, extended arrays of
helices [9,10,13–21]. Second, the majority of these
(11 structures) are for isolated domains, taken from
larger parent proteins. Third, of those structures that
consist of more than the TPR sequence alone, the non-
TPR component can be unstructured, a C-terminal
capping helix [14], can bind back as an extended poly-
peptide to the concave face [16], or can assume a
completely independent domain organization [19,20].
Fourth, a single structure has been determined where
TPR motifs are seen to fold back on themselves: in the
seven-repeat peroxisomal targeting signal receptor
PEX5, repeats 6 and 7 pack against repeats 1 and 2
through loop motifs [13]. This is an unusual example,
however, because an alternative conformation for
PEX5 has been determined in which the structure is
extended [18]. This may represent a novel confor-
mational switching mechanism where dynamic, long-
range inter-TPR interactions are critical to function.
Alternatively, these two conformations could reflect
different crystallization conditions. Fifth, no structure
of a prokaryotic TPR has yet been determined. Here
we describe the first structure of a complete TPR-
containing protein from Escherichia coli.
New lipoprotein I (NlpI) is a 32 kDa, 276 residue
protein from E. coli K-12 [22]. The corresponding
chromosomal gene encodes a 294 amino acid polypep-
tide, whose N-terminal 18 residues comprise a periplas-
mic export sequence and ‘lipobox’ motif (Fig. 1B).

Following translocation across the inner plasma mem-
brane, the prosequence and lipobox cysteine are
recognized, enzymatically modified and proteolytically
processed by components of the lipoprotein biosyn-
thetic pathway. This yields an N-acyl-S-sn-1,2-diacyl-
glyceryl-cysteine (residue 19) as the N-terminus of the
mature, 276 residue, membrane-anchored protein [22].
The identity of the residue at the +2 position (Ser20)
in the mature protein suggests that NlpI is not retained
at the inner membrane, but is likely to be anchored at
the outer membrane [23–25]. The precise topological
location (periplasmic or extracellular face) is not
known. NlpI has been proposed to play a role in
bacterial septation, or regulation of cell wall degrada-
tion during cell division [22]. Disruption of the chro-
mosomal copy of the nlpI gene, or plasmid-mediated
overexpression of the protein, both lead to altered cell
morphology and to osmotic sensitivity.
NlpI is of potential clinical interest, because loss of
the nlpI gene affects the synthesis of pili and flagellae,
leading to changes in extracellular adhesion properties
which are correlated with an invasive, pathogenic
phenotype [26]. A BLAST search for similar sequences
Fig. 1. Canonical TPR structure and NlpI sequences. (A) Example of
TPR extended helical structure, from the consensus design
1NA0.pdb [6]. Three repeats are the most common number seen.
The AB and AA¢Wpacking angles are responsible for curvature and
superhelicity of the motif. (B) Amino acid sequence of NlpI from
the translation of the nlpI gene [22], including the signal
prosequence (underscored) and lipobox cysteine modification site

(boxed). Proposed TPR motifs are shaded grey [27]. (C) Alignment
of the putative NlpI TPRs compared to the signature motif. Varia-
tions from the consensus are unshaded positions within the vertical
shaded bars. The fourth repeat contains the fewest matches to the
consensus.
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 167
finds highly conserved homologs from well-known
pathogenic species (see below).
Initial studies of NlpI primary structure predicted
three tandem TPR repeats [22]. A fourth repeat, imme-
diately following the third, and a fifth independent
TPR, have also been suggested [27], although the simi-
larity of the fourth repeat to the consensus is weak
(Fig. 1C). On the basis of these analyses, one might
anticipate the presence of an independent, extended
three-repeat. This could account for approximately
40% of the mature sequence, while the structural char-
acter of the remaining polypeptide is unknown.
Several properties make NlpI an attractive target
for structural characterization. The gene can be
obtained directly from laboratory strains of E. coli,
while its size and periplasmic localization suggest it
is likely to be a single domain, capable of independ-
ent folding. The single cysteine is responsible for
tethering the protein to a plasma membrane; the
absence of disulfide bridges or other structural cys-
teine, simplifies the protein chemistry by removing
the need for reducing agents during purification and
handling. We therefore cloned and expressed NlpI,

and determined the protein structure by X-ray crys-
tallography. The structure reveals a fold in which the
TPR is not an independent domain, but is an integ-
ral part of a globular protein.
Results and Discussion
Cloning and expression of NlpI
The gene for NlpI was obtained by direct PCR ampli-
fication from E. coli DH10B [28]. The sequence corres-
ponding to the mature polypeptide (residues 20–294,
lacking the signal prosequence and Cys20) overexpres-
ses exceptionally well in BL21 (DE3), with yields of
 100 mgÆL
)1
(Fig. 2A). Purified mature NlpI is sol-
uble to at least 200 mgÆmL
)1
in 10 mm Tris ⁄ HCl
pH 8.0, 10 mm NaCl.
To investigate the anticipated 3-TPR domain of
NlpI, we subcloned residues 62–197. This region also
expresses well, but in contrast to the mature polypep-
tide, the majority of protein is found in inclusion bod-
ies (Fig. 2A). Material purified from the lysate soluble
fraction precipitated after elution off Ni-nitrilotriacetic
acid agarose. Alternative expression, purification, solu-
bilization and refolding regimes were investigated, but
we were unable to obtain soluble 3-TPR. This result
was surprising, as we had anticipated the 3-repeat to
be an independent domain. The insolubility of this
region was the first indication that the 3-TPR might

be participating in a more complex structure. We
therefore pursued characterization of only the mature
polypeptide (20–294).
Analytical gel filtration chromatography indicated
that mature NlpI runs at nearly twice its anticipated
size ( 56 kDa vs. 31 kDa). This was not entirely
unexpected, because an extended array of helices occu-
pies a larger hydrodynamic radius than a globular
molecule of similar polypeptide length. However, cal-
culation of the Matthews coefficient from initial crys-
tallographic data indicated that a dimer was most
compatible with unit-cell dimensions (solvent content
 56%). This was further supported by in vivo formal-
dehyde crosslinking (data not shown) that trapped a
species corresponding to dimeric NlpI.
Structure of mature NlpI
We have solved the crystal structure of mature NlpI
to 2.0 A
˚
. Data collection and refinement statistics are
shown in Table 1. A region of the 2Fo-Fc electron
density map for residues 158–162 is shown in Fig. 3.
The protein crystallised in space group P2
1
2
1
2
1
with
two monomers in the asymmetric unit, related by a

twofold axis of noncrystallographic symmetry running
through the dimer interface. We conclude that the
contents of the asymmetric unit represent the biolo-
gically active protein. The two chains together form
an arrow-shaped structure, wider than it is deep
(Fig. 2B,C). N-termini of both molecules share a
common point of origin, a feature compatible with
membrane localization through N-terminal lipid
anchors on both chains. Table 2 shows the secondary
structure components, interhelix packing geometries,
and the angle of rotation between the AB helix pairs
present. With the exception of an extended, but not
unstructured region of polypeptide (30–37), NlpI is
composed of a-helix (64%) and turn motifs (23%).
NlpI monomers can be described generally as a
superhelical array of helix-turn-helix motifs, in which
the C-terminus is folded (rolled-up) inside the
N-terminus (Fig. 2D). A depression on one side of
each monomer contains a bound Tris molecule. This
cavity, formed by the curvature and packing of heli-
ces, is highly suggestive of a ligand binding pocket
and we speculate that it may represent the functional
site of the protein.
Helix packing interactions
TPRs
Many features of the distribution of side chain con-
tacts within NlpI are typical of a TPR protein. The
side chain contact map (Fig. 4A) is dominated by a
Crystal structure of NlpI C. G. M. Wilson et al.
168 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS

repetitive pattern of interactions parallel to the diago-
nal (i to i + 3, i to i + 4 within a continuous helix),
orthogonal to the diagonal (helix A interacting with
helix B), then parallel to the diagonal (helix A inter-
acting with helix A¢ of the next repeat), and finally
returning to the diagonal (helix B interacting with
A¢). This distribution is more or less continuous,
reflecting a progression of helix-turn-helix AB, AA¢,
BA¢ interactions through the structure. The exception
is helix 1, which interacts exclusively with helix 2
through hydrophobic packing of bulky groups (e.g.
Leu44 against Leu77, Met47 against the aliphatic
components of side chain Arg68). Of the six AB
helix-turn-helix pairs (Table 2), four closely resemble
TPRs: helices 2 and 3 (TPR1), 4 and 5 (TPR2), 6
and 7 (TPR3) and 12 and 13 (TPR4) contain the
characteristic pattern of signature residues that coin-
cides with helix-loop-helix lengths. Interhelix AB and
AA¢Wpacking angles fall within those typical for
TPRs [6]. These repeats correspond to the anticipated
tandem 3-TPR, and the isolated fifth TPR predicted
from the amino acid sequence (Fig. 1C) [22,27]. Heli-
ces 2 and 12 (A helices of TPRs 1 and 4, respect-
ively) contain additional helical residues preceding
the start of the TPR region. The final helix (14) does
not contain the solvating polar groups associated
with terminating ‘capping’ helices found in the other
TPR structures (e.g. PP5, TPR1 and TPR2A of
AB
DC

Fig. 2. Solubility of NlpI constructs and the structure of mature NlpI. (A) 10–20% gradient SDS ⁄ PAGE of NlpI expression products, showing
the insolubility of the 3-TPR construct vs. mature NlpI. BenchMark molecular mass markers (lanes 1 & 4); 3-TPR insoluble (lane 2, arrow-
head) and soluble (lane 3); mature NlpI insoluble (lane 5) and soluble (lane 6, arrowhead); mature NlpI following TEV protease cleavage, and
purification over Superdex 75 (lane 7, arrowhead, anticipated molecular mass of 31.8 kDa). (B) Side and (D) top views of the NlpI dimer.
Chains are coloured from N- (dark blue) to C-termini (orange). Axis of noncrystallographic rotational symmetry runs through the center ‘x’.
(C) Monomer of NlpI, showing the rolled-up array of helices with the C-terminus folding within the curvature of the N-terminus. Helix num-
bers are in brackets. Note that ‘A’ helices locate to the globular center, and the perpendicular arrangement of helices 10 and 11, against heli-
ces 8 and 9.
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 169
HOP), because the majority of these residues partici-
pate in the protein core.
NonTPR helix motifs
Packing interactions are more complex for the two
remaining pairs of helices (8 and 9, 10 and 11). These
are of particular significance since they are responsible
for the compact structure of NlpI. Helix pair 10–11
(Fig. 4B) is, at 27 residues (17 of which are helix) too
short to be termed a TPR. The interhelix AB W packing
angle is the highest (+ 172°), bringing them close to
parallel, and is also of the opposite sign to that which
characterizes a TPR. Interactions with the following
pair of helices (12 and 13) is distinguished by the only
negative AA¢ angle within the protein. Critically, this
combination of nonTPR packing angles imparts left-
handed superhelical character to the region. The pitch
of the overall right-handed superhelix is therefore
reduced, which brings the C-terminus up toward the
N-terminus.
Helices 8 and 9 correspond to the region of sequence

postulated by some to be a fourth TPR, following on
immediately after the tandem 3-repeat (Fig. 1C [27]).
However, the interactions taking place within this pair
indicate that it is not a TPR. Helix 8 contains the
hydrophobic sequence LWLYL(168–171), and these
groups are involved in long-range interactions (dis-
cussed below). Helix 8 participates more in these than
in packing against its partner, helix 9. The signature
glycine residue of the A helix is missing, a space occu-
pied in a TPR by a bulky hydrophobic or aromatic ring
from the partner B helix (knobs-in-holes complement-
arity). Instead, the glycine position is occupied by ala-
nine, with the remaining space filled by a tyrosine
(Tyr171) from the same (A) helix. The complementarity
Table 1. SeMet-NlpI Data processing, MAD phasing and refinement statistics. FOM, figure of merit, value from DM is at 2.0 A
˚
, whereas
SOLVE ⁄ RESOLVE values are at 2.5 A
˚
.
Space-group P2
1
2
1
2
1
Unit-cell (A
˚
)a¼ 64.35 b ¼ 81.65 c ¼ 136.66
Wavelength (A

˚
) Peak (0.9785) Inflection (0.9795) Remote (0.9500)
Resolution (A
˚
) 30–2.05 30–2.0 30–2.0
Total number of reflections 408241 506497 282547
Number of unique reflections
a
85218 91502 91856
Completeness (%)
a
96.1 (84.0) 96.0 (83.6) 94.9 (81.2)
I ⁄ Sigma
b
33.8 (4.7) 40.4 (4.8) 29.3 (2.7)
R
merge
b
(%) 5.4 (30.7) 4.9 (35.5) 5.3 (46.6)
Redundancy
a
4.8 5.5 3.1
FOM after
SOLVE 0.58
FOM after
RESOLVE 0.72
FOM after
DM 0.64
R ⁄ R
free

(%) 17.5 ⁄ 20.6
Number of all atoms 4615
Number of water molecules 436
Avearge B-factors (A
˚
2
)
Monomer A (main chain ⁄ side chains) 20.4 ⁄ 23.6
Monomer B (main chain ⁄ side chains) 21.6 ⁄ 23.8
Water molecules 39.3
Ramachandaran plot (%) (most favoured ⁄ allowed ⁄ disallowed) 94.0 ⁄ 6.0 ⁄ 0.0
rmsd bond lengths (A
˚
) 0.026
rmsd angles (°) 1.96
a
Values are given with Friedel-pairs (hkl and -h-k-l) kept separate.
b
Value for the high resolution shell is given in parentheses.
Fig. 3. Sample 2Fo-Fc density map for NlpI (sigma ¼ 1.0). Residues
shown are Gln158-Asn162 (QDDPN) for chain A, which corres-
ponds to the turn region between helices 7 and 8. Image was pro-
duced with
BOBSCRIPT [57] and RASTER3D [58].
Crystal structure of NlpI C. G. M. Wilson et al.
170 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS
between helix 8 and 9 is therefore less marked, and they
are less tightly packed together compared to AB pairs
before or after (Fig. 4B). The apparent lack of contacts
between helices 8 and 9 is, however, compensated for

by an unusual association with the next pair of helices,
10 and 11. These pack against 8 and 9 at an angle of
96° (visible in Fig. 2D), which is the highest inter-
repeat rotation angle within the structure. A unique,
nonTPR interaction takes place where the indole ring
of Trp200 (helix 10) inserts between helices 8 and 9,
against Phe190 and the amide backbone of Phe165.
This locks the pairs together (Fig. 4D). The abrupt
increase in helical array curvature is the second factor
responsible for bringing distal regions of sequence back
toward proximal ones.
Long-range interactions & globularity
The presence of long-range contacts within NlpI is
revealed by clusters in the contact map far from the
diagonal (Fig. 4A). These interactions take place only
between A helices, which dominate the inside of the
NlpI helical roll (Fig. 2C). The clusters can be consid-
ered as four overlapping groups (Fig. 4E). Cluster 1
involves helices 12 and 14, packing against the
N-terminal region of NlpI. These constitute the most
distant interactions between elements of primary
structure, and include a hydrogen bond between the
backbone carbonyl of Asn263 and the backbone amide
of Leu34. The positive AA¢Wangle between helix 12
and helix 10 is in part responsible for this. Cluster 2
consists of loop against loop interactions between
TPR1 and TPR2, with helix 14 (including a Ca back-
bone contact between Gly76 and His266). From func-
tional perspectives this is perhaps significant, as this
first TPR is more open than any other, forming an

exposed ‘lip’ on the NlpI monomer, and therefore most
closely resembles classical TPR front-face environments
in providing an interaction surface. Cluster 3 contains
solvent inaccessible hydrophobic groups (Leu134,
Ile138, Val269) and hydrogen bond interactions
(Tyr142-Lys242) between the third and fourth TPR,
with helix 14. Cluster 4, which overlaps with clusters 2
and 3, forms the core of NlpI long-range interactions.
These include bulky hydrophobic, aromatic and surface
solvent exposed groups from helices 6, 8, 12 and 14.
For example, Trp169 (helix 8) has hydrophobic con-
tacts with Leu134, Ile138, Tyr142 (helix 6), Phe238
(helix 12) and Val269 (helix 14). Trp169 and Tyr273
(helix 14) are on opposite sides of the protein core, and
do not interact directly, but they are bridged by alipha-
tic and aromatic groups (Ile138, Tyr142, Phe238 and
Val269) from helices 6 and 12. Aromatic ring inter-
actions, including ‘T’ face-edge (Tyr131-Phe156,
Table 2. Primary, secondary and tertiary structure statistics for mature NlpI. Excluding the N-terminal 3
10
helix (H0), NlpI contains 14 a-heli-
ces. W Packing angles characteristic of TPRs range from )160° to )174° (AB), 11° to 32° (AA¢) and 40° to 53° (AB-A’B¢ repeat rotation) [6].
AB pairs 4 and 5 (helices 8–9, 10–11) do not display true TPR characteristics, but are responsible for the sharp curvature of the array and
reduced superhelical pitch, leading to the formation of a globular structure.
Helix
a
Residues Deviation
b,c
A
˚

AB pair W°
c
AB W°
c
AA¢
Rotation
d
(AB)(A¢B¢)°
Helix pair sequence
(signature TPR residues underscored)
0 27–29 – – – – – consensus motif
1 38–51 6.4 – – – – W LG Y A F A P
2 58–74 7.6 1 )160.5 +31.3 57.6 DDERAQ
LLYERGVLYDSLGLRALARNDFSQALAIRPDM
3 78–91 5.9 (TPR1) (pair 1–2)
4 96–108 5.9 2 )165.5 +16.7 59.0 PEV
FNYLGIYLTQAGNFDAAYEAFDSVLELDPTY
5 112–125 5.1 (TPR2) (pair 2–3)
6 131–142 8.9 3 )153.0 +26.5 16.2
YAHLNRGIALYYGGRDKLAQDDLLAFYQDDPND
7 146–159 12.2 (TPR3) (pair 3–4)
8 164–177 14.6 4 +163.3 +26.9 96.2 PFRSLWLYLAEQKLDEKQAKEVLKQHFEKSDKEQW
9 179–192 3.9 (pair 4–5)
10 199–206 12.9 5 +172.6 )26.0 40.4 GWNIVEFYLGNISEQTLMERLKADATD
11 212–222 7.4 (pair 5–6)
12 226–246 19.4 6 )158.3 +38.2 261
NTSLAEHLSETNFYLGKYYLSLGDLDSATALFKLAVANNVHNF
13 250–261 8.6 (TPR4) (pair 1–6)
14 269–283 7.9 – – – –
a

Defined by by PROCHECK [51].
b
From ideal helix geometry.
c
Calculated by PROMOTIF [52].
d
Obtained by transforming one helical pair onto
another with lsqman [53].
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 171
Tyr141-Tyr142, Phe205-Tyr243) and offset pi-stacking
(Phe85-Tyr101, Trp298-His232) are evident, with the
spaces between these moieties filled by bulky aliphatics
(Fig. 4F).
Long-range interactions are a characteristic property
of globular protein structures. By virtue of their exten-
ded helical organization, TPR and other repeat pro-
teins typically lack this feature, but rather have
stabilizing apolar contacts distributed throughout the
molecule, both within and between repeats. In contrast,
NlpI contains a central hydrophobic core, composed of
distant motifs from TPR and nonTPR helices. TPRs
are therefore compatible with globular structures, but
they do not appear to be capable of defining it. The
increased curvature and reduced superhelical pitch
required to form a compact, tertiary structure are
derived from nonTPR elements within the fold.
Quaternary organization
NlpI is a homodimer in solution, and the crystal
structure reveals monomers to be related by twofold

axis of symmetry. The dimer interface consists of the
extended N-terminal region, helix 1 and TPR helices
2, 3, 11, 12, 13 and 14 (Table 3, Figs 5 and 6B). The
values obtained for interface surface area, interaction
type (two-thirds hydrophobic, but also hydrogen
bonds and salt-bridges), gap volume index and planar-
ity (which relate to the complementarity of the inter-
face surfaces) fall within the ranges associated with
known homodimeric states [29]. Three aspects are
especially noteworthy. First, rotational symmetry
places the N-termini of both monomers spatially close
to each other. A lipid-modified dimer will therefore be
anchored to a plasma membrane in a specific orienta-
tion (N-termini ‘face down’ toward the membrane).
This is significant, because the potential ligand binding
AB
C
D
EF
Fig. 4. Contact map of mature NlpI and packing interactions. (A) Backbone (upper left from diagonal) and side chain (lower right from diago-
nal) contacts within 5 A
˚
. Long-range contact clusters are boxed. (B) Packing interactions between nonTPR helices 10 (red) and 11 (blue), and
(C) helices 8 (red) and 9 (blue). Space-filling atoms shown are large and small hydrophobic residues (F, Y, W, I, L, V, A and G). Bulky groups
of helix 8 point toward the protein core. (D) View of NlpI helices 8 and 9 (with helix 7 removed), showing diminished association between
the pair, and the insertion of Trp200 from helix 10. Right-handed superhelical curvature imparted by the first three TPRs appears to cease,
allowing the subsequent structure to roll-up. (C) Location of long-range packing clusters from (A), which define the core of NlpI. (F) Aromatic
and bulky side chains surrounding Trp169 (orange).
Table 3. NlpI dimer interface statistics. Values were obtained with
SURFNET [29,55]. SA ¼ surface area.

Interface statistics Region Residues
Monomer buried SA (A
˚
2
) 1585.5 N-terminal 23–26, 30–36
% Monomer SA 12.6 Helix 1 38, 41, 44, 45, 48,
49, 51
% Nonpolar atoms 36.6 Helix 2 68, 76, 77
% Polar atoms 63.4 Helix 3 78, 79, 80, 83
Planarity 2.5 Helix 11 212, 216
Hydrogen bonds 16 Helix 12 237, 240, 244
Salt bridges 2 Helix 13 255, 258, 259,261,
262, 264, 266
Buried H
2
O 16 Helix 14 271, 275, 278
Gap volume index 2.45 Total 38
Crystal structure of NlpI C. G. M. Wilson et al.
172 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS
cleft (discussed below) of each monomer would then
be exposed, and oriented roughly perpendicular to the
plane of the membrane. NlpI localized in this manner
could then serve as a tether to which other functional
components would bind. Second, the dimer interface
is made up of distant regions of monomer sequence.
That is, N-terminal and C-terminal portions of mono-
mer polypeptide come together, forming the molecular
surface. Quaternary structure is therefore dependent
on tertiary structure, and their formation may be
interdependent (cooperative). Third, it was previously

noted that the first TPR (helices 2 and 3) participates
in long-range interactions within a monomer through
loop residues (Asp73, Ser74, Leu75, Arg78), while the
majority of the inner front-face assumes an open ‘lip’
conformation (Fig. 4D). In contrast, seven residues of
the outer back-face participate in the dimer interface,
packing against C-terminal portions of polypeptide
from the partner molecule. Consideration of mono-
meric NlpI alone gives the impression that these heli-
ces make few molecular contacts when in fact they
make many, albeit with a separate polypeptide chain.
The insolubility of NlpI 3-TPR (fragment 62–197) is
therefore understandable, in terms of the failure of an
isolated motif to form critical intra- and intermole-
cular contacts. These observations demonstrate the
capacity, and on occasion the necessity, of TPRs to
participate at all levels of structure organization, and
suggest that the fold is more versatile than was previ-
ously thought.
We now know the structure of NlpI, and observe
that the TPRs in this protein do not form an inde-
pendent domain. One could therefore ask if there
are any features of the TPR sequences that hint at
differences between these TPRs, and those that fold in-
dependently. Unfortunately, with the limited sequence–
structure data available at this stage, there are no
correlations strong enough to allow us to predict, or
subclassify, which TPR sequences will form an exten-
ded array and which will adopt globular structures.
Implications for function: a putative binding cleft

We examined the conservation of NlpI structure
through a sequence alignment of the 12 most similar
sequences identified through a BLAST search (Fig. 6A).
When conserved positions are mapped onto the struc-
ture of NlpI, they correlate to three distinct locations
within the protein. Two of these are clearly structural in
nature: the globular core (discussed previously) and the
dimerization interface (also predominantly hydropho-
bic, Fig. 6B), suggest that NlpI homologs share a com-
mon tertiary and quaternary organization. The third
conserved region corresponds to the depression on one
side of each NlpI monomer (Fig. 6C). The cleft is lined
with polar (Asn267), acidic (Asp163, Glu235, Glu231
and Glu270), aromatic (Tyr131, Phe165, Trp198,
Phe268) and hydrophobic groups (Ile104, Val269). Visu-
ally, the shape of the depression is highly suggestive of a
binding site. The presence of four invariant acidic
groups (one aspartate and three glutamate) implicates
electrostatic interactions, possibly with a basic motif, in
the putative binding event. The high degree of sequence
conservation in the cleft suggests all homologs of NlpI
bind the same ligand.
In addition, our attention was drawn to this cavity
during the final stages of model building, because it
contained a patch of 2Fo-Fc density that could not be
accounted for by the polypeptide, water molecules,
Mg
2+
or Cl
–

ions. The structure of Tris, also a compo-
nent of the crystallization mother liquor, was found to
fit the density envelope, making hydrogen bonds with
carboxylate groups of Glu235 and Glu270, and with
the back bone amide of Val269. Phe165 and Phe268
face each other, flanking the two carboxylates and Tris
(Fig. 6D).
NlpI is thought to play a role in the regulation of the
cell wall and extracellular surface, but its exact function
is not known, and no ligand interactions have yet been
described. There has been some suggestion that the
C-terminus may associate with the periplasmic protease
Tsp, and it has been proposed that removal of residues
beyond Gly282 serves to activate the protein [27]. How-
ever, the C-terminus of NlpI does not contain a motif
that resembles the canonical ‘WVAAA’ associated with
Fig. 5. NlpI dimer interface. Chain B has been translated and rota-
ted to expose the surface in contact (yellow). The interface is com-
posed of remote regions of sequence from the N- and C-termini.
Val32 is indicated to illustrate the rotational symmetry of the inter-
face. Contacts were obtained with
SURFNET [55] and CONTACT from
the CCP4 [42,43] suite of programs.
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 173
Tsp recognition [30]. Our structure suggests the func-
tionality of NlpI in fact lies within the cleft associated
with globular body of the fold.
Structural homologs
A DALI search for homologous structures finds two

PDB entries with significant similarity to NlpI. The
first, p67
phox
from human (Z-score ¼ 12.4, rmsd of
2.5 A
˚
over 152 residues) consists of four TPR motifs,
and an extended C-terminus that packs against the
concave front-face groove through hydrophobic
interactions. The intracellular ligand, Rac-GTP, is
required for the assembly of the multiprotein
NADPH oxidase complex, and binds to the surface
formed by TPR connecting loops and the C-terminal
polypeptide [16]. The ligand-binding mode is there-
fore distinct from that of TPR1 and TPR2A of
HOP. The structural similarity between NlpI and
p67
phox
is illustrated in Fig. 7A, and reveals a close
match between the first four AB repeats of each
protein.
The second structure, domain III from E. coli malt-
ose transcriptional regulator MalT (Z-score ¼ 12.3,
A
BC D
Fig. 6. Homologs of NlpI, structural conservation the putative binding cleft. (A) CLUSTALX alignment [48] of the 12 sequences most similar to
E. coli NlpI identified by BLAST [49]. Positions are colored as follows: red, identical (*); yellow, similar (:). 1, Escherichia coli; 2, Salmonella
typhinurium;3,Yersinia pestis;4,Yersinia enterocolitica;5,Vibro haemolyticus;6,Vibro vulnificus;7,Vibro cholerae;8,Photorhabdus lumi-
nescens;9,Photobacterium profundum; 10, Haemophilus influenzae;11,Haemophilus ducreyi;12,Shewanella oneidensis. E. coli residues
underscored locate to the hydrophobic core. (B) Molecular surface revealing conserved positions within the dimerization interface (mostly

hydrophobic). (C) The surface of an NlpI monomer, showing the putative ligand binding cleft and bound Tris molecule. (D) Tris molecule,
conserved acidic and aromatic side chains within the cleft. Orange dashes indicate hydrogen bonds between Tris and the amide
backbone of Val269, and conserved side chain carboxylates of Glu235 and Glu270.
Crystal structure of NlpI C. G. M. Wilson et al.
174 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS
rmsd of 4.3 A
˚
over 212 residues), consists of eight
superhelical peptide repeat (SUPR) motifs that assume
a superhelical fold [31]. SUPRs resemble TPRs, but
their helices are slightly longer (16–18 residues) and
the sequence consensus is more degenerate. The
N-terminal region of MalT is responsible for func-
tional dimerization, while the C-terminus is thought to
contain a maltotriose binding site, formed by the con-
cave surface of four SUPR repeats. Close structural
similarity can be seen between the first three AB
repeats of NlpI and MalT (Fig. 7B). However, by the
sixth repeat the reduction in NlpI superhelical pitch
has folded the protein back onto itself, while MalT
continues in a more regular superhelix (and in conse-
quence lacks hydrophobic core interactions).
In terms of their biological roles, p67
phox
and MalT
both mediate intermolecular interactions, and are
responsible for the assembly of multiprotein com-
plexes. It is therefore interesting to speculate, on the
basis of structural identity and the presence of a con-
served surface cleft, whether NlpI participates in ana-

logous multiprotein assemblies in E. coli.
Conclusion
We have determined the first structure of a complete
TPR-containing globular protein, which is also the first
TPR from a prokaryotic organism. The structure
reveals an intimate association between the TPR motifs
and the rest of the protein, showing how the TPR par-
ticipates in the overall fold. Until now, many of the
TPR-containing regions of proteins have behaved as
separate domains: they fold and function completely
independently of the rest of the protein. Here we show
an alternate arrangement, in which the TPR is insepar-
ably part of the whole structure. Nothing about the
sequence, or any a priori considerations, suggested that
this structure would be different from the TPRs from
independent domains. The structure provides a strong
hint at the location of the active site, though as yet the
ligand bound by NlpI has not been identified. Its
involvement in bacterial virulence, and likely presence
of identical interactions in many pathogenic species,
makes NlpI a potential target for new antibiotics.
Experimental procedures
Cloning NlpI constructs
DNA encoding NlpI sequences was obtained by PCR
amplification from a single colony of E. coli DH10B [28]
(Invitrogen, Carlsbad, CA, USA), grown on Luria–Bertani
agar overnight. All oligonucleotides were chemically syn-
thesized by the W. M. Keck Core Facility (Yale University,
New Haven, CT, USA). Primers to amplify mature NlpI
(residues 20–294) were 5¢-aataatccatggggagtaatacttcctggcgta

aaagtgaagtcc-3¢ and 5¢-attattggatccctattgctggtccgattctgccag-3¢.
3-TPR NlpI primers (residues 62–197) were 5¢-aataatccatgg
gggcacagcttttatatgagcgcggag-3¢ and 5¢-aataatggatcctcactgttc
cttatccgatttttcgaagtgc-3¢. PCR products were doubly diges-
ted with NcoI and BamHI (New England Biolabs, Beverly,
MA, USA), and purified by agarose gel electrophoresis
onto dialysis membrane, prior to ligation into doubly diges-
ted, dephosphorylated expression vector pET11a-HT. This
vector was assembled in-house from vectors pProEX-HTa
(Invitrogen) and pET11a [32] (Novagen, San Diego, CA,
USA), and places cloned sequences under T7 promoter con-
trol. Expression in an E. coli DE3 bacterial host produces
an N-terminal hexahistidine-tagged protein, cleavable with
TEV protease. Ligation products were transformed into
electrocompetent E. coli DH10B (Invitrogen), and trans-
formants sequenced by the W. M. Keck Facility.
Expression and purification
Plasmids, verified by DNA sequencing, were transformed
into E. coli BL21 (DE3) Gold (Stratagene), and grown in
Luria–Bertani medium supplemented with 100 lgÆmL
)1
car-
benicillin at 37 °C until cell culture absorbance at 600 nm
was 0.5. The temperature was reduced to 25 ° C before
induction with 100 lm isopropyl thio-b-d-galactoside. Cells
were harvested by centrifugation (6000 g, 20 min) after 4 h
further growth, and stored at )80 °C. Selenomethione
(SeMet)-labelled NlpI was expressed in E. coli methionine
auxotroph B834 (DE3), grown in M9 medium [33] supple-
mented with 50 mgÆL

)1
l-methionine and 100 mgÆL
)1
thiamine. At a cell density of D
600
¼ 0.4, bacteria were har-
vested and resuspended in fresh M9 supplemented with
Fig. 7. Structural homologues of NlpI. Structure alignment of NlpI
(blue) with (A) p67
phox
and (B) domain III of MalT (yellow). Super-
imposition of coordinates was performed with
LSQ_EXPLICIT and
LSQ_IMPROVE [45].
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 175
50 mgÆL
)1
l-selenomethione (Pierce, Rockford, IL, USA).
Growth was continued for 20 min at 25 °C, before induc-
tion with 100 lm isopropyl thio-b-d-galactoside overnight.
Cell pellets were resuspended in buffer A [50 mm Tris ⁄ HCl
pH 8.0, 300 mm NaCl, 0.1% (v ⁄ v) Triton X-100] plus lyso-
zyme to a final concentration of 0.1 mgÆmL
)1
, and incuba-
ted for 30 min on ice prior to sonication (20 second pulses
at maximum power, for a total process time of 3 min, with
20 s cooling between pulses). Insoluble material was
removed by centrifugation (20 min, 26 000 g,4°C). Affin-

ity purification was performed with slow rocking at 4 °C
overnight, with one-fifth volume of Ni-charged nitrilotri-
acetic acid agarose slurry (Qiagen). Washing steps were
performed at room temperature in a disposable standing
column (Bio-Rad, Hercules, CA, USA), with 5 bed vol-
umes buffer A and 5 bed volumes buffer B (50 mm
Tris ⁄ HCl pH 8.0, 300 mm NaCl, 5 mm imidazole). Bound
protein was eluted in buffer C (50 mm Tris ⁄ HCl pH 8.0,
150 mm NaCl, 300 mm imidazole). Hexahistidine tags were
removed by treatment with 10 units of AcTEV protease
(Invitrogen) overnight at room temperature, followed by
dialysis against buffer D (50 mm Tris ⁄ HCl pH 8.0, 150 mm
NaCl). TEV protease and uncleaved fusion were removed
by gravity flow through a 1 mL bed of fresh Ni-nitrilotri-
acetic acid agarose. Protein was loaded onto a Superdex
S200 16 ⁄ 60 prep grade column (Amersham Biosciences),
equilibrated in buffer D. Fractions containing NlpI were
pooled, dialysed against buffer E (10 mm Tris ⁄ HCl pH 8.0,
10 mm NaCl) and concentrated with a Centriprep YM10
spin concentrator (Millipore, Billerica, MA, USA). Mature
NlpI mass was verified by MALDI mass spectrometry.
Protein concentration was estimated by SDS ⁄ PAGE
against BenchMark Protein Ladder (Invitrogen), and by
absorbance at 280 nm assuming a calculated e
mature NlpI
¼
43 240 m
)1
Æcm
)1

[34].
Size exclusion chromatography
Analytical sizing runs were performed with a Superdex
200 10 ⁄ 300 GL Tricorn column (Amersham Biosciences)
equilibrated in buffer D. Flow-rate was 0.25 mLÆmin
)1
and detection was by UV absorbance at 280 nm. Calibra-
tion runs were performed with Gel Filtration Standard
mix (Bio-Rad). Purified NlpI was diluted in buffer D to
give 100 lL at concentrations between 15 lm and 1.5 mm
(0.5 and 50 mgÆmL
)1
). The apparent molecular mass was
estimated from proteins standards, assuming an inverse
relationship between retention time and log molecular
mass.
Crystallization of mature NlpI
Purified NlpI was subjected to Crystal Screen I and II
(Hampton Research, Aliso Viejo, CA, USA) as hanging
drops, at a protein concentration of 150 mgÆmL
)1
. Crystals
formed overnight in 100 mm Hepes pH 7.5, 200 mm MgCl
2
,
30% (v ⁄ v) PEG 400 at 22 °C. Final conditions were
50 mgÆmL
)1
NlpI mixed 1 : 1 with 100 mm Tris ⁄ HCl
pH 8.5, 360 mm MgCl

2
, 27% (v ⁄ v) PEG 400, 0.8% (v ⁄ v)
n-butanol, at 22 ° C. SeMet-labelled NlpI crystallized iso-
morphously, under identical conditions. Rectangular crys-
tals grew over 3 days at 22 °C to 0.3 · 0.5 · 1.0 mm.
Data collection & phasing
Crystals were flash-frozen from mother liquor in a nitrogen
gas cryo-stream. In-house data collection used a Mar345
image plate detector (MAR Research), coupled to a Rigaku
CuKa rotating anode source. NlpI crystallized in the space-
group P2
1
2
1
2
1
, with unit-cell dimensions a ¼ 64.35 A
˚
,b¼
81.65 A
˚
,c¼ 136.66 A
˚
, with two monomers in the asym-
metric unit and a Matthews coefficient of 2.82 A
˚
3
ÆDa
)1
(solvent content ¼ 56.36%) [35,36]. A three-wavelength

MAD data set was collected from a SeMet–NlpI crystal at
beamline X12C, of the National Synchrotron Light Source,
Brookhaven National Laboratory. Data was collected for
three wavelengths (Table 1), in two oscillation ranges from
/ and / +180° to ensure both Friedel-mates were collected
for each reflection and to obtain accurate, redundant anom-
alous data. However, less data was collected for the remote
wavelength because of limited beam time. Anomalous data
was nevertheless complete for the remote wavelength
(Table 1). Data from 30 A
˚
to 2.0 A
˚
were indexed and
scaled with the hkl2000 software package [37] (Table 1).
The search for Se sites was performed with solve [38] at
2.5 A
˚
, assuming a dimer containing eight Se atoms within
the asymmetric unit-cell (including an N-terminal SeMet).
Six Se sites were found (N-terminal SeMet was disordered).
The initial phases from solve were subjected to solvent flat-
tening and twofold noncrystallographic symmetry (NCS)
averaging. The initial model, containing most of the secon-
dary structure elements, was obtained with resolve [39] at
2.5 A
˚
. However, phase extension to 2.0 A
˚
did not work as

well as expected. Solvent flattening, NCS averaging with an
operator (obtained from the resolve dimer model, and
refined with imp [40]), and phase extension, were repeated
with dm [41] in 100 steps, using the CCP4 graphical user
interface [42,43] and the phases from solve. This resulted
in significant improvement in map quality.
Model refinement
Several schemes were tested for refinement. The best results
were obtained using cns [44] for rigid body refinement at
2.0 A
˚
, and conjugate gradient minimization at 2.5 A
˚
with
maximum likelihood target, resulting in R-factors of
R ⁄ R
free
¼ 30.0 ⁄ 30.9%. The model was then manually
improved and completed in o [45]. Thereafter the structure
was refined with refmac 5.0 [46] through the ccp4 graphi-
cal user interface [42,43], with overall anisotropic B-factor
Crystal structure of NlpI C. G. M. Wilson et al.
176 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS
refinement of data and bulk solvent to 2.0 A
˚
resolution.
The first round of refinement converged to R-factors of
R ⁄ R
free
¼ 21.1 ⁄ 25.2%. The model was once more inspected

and refined further with refmac 5.0. In final rounds of
refinement water molecules were added with arp ⁄ warp
[47], and manual inspection with o. Initially, strict NCS
restraints were used, but these were released in the final sta-
ges of refinement (‘loose’ in refmac ⁄ CCP4). The final
model contained residues 26–284 for chain A, residues
23–284 for chain B, 436 water molecules in asymmetric unit
and one bound Tris molecule per monomer, with R-factors
R ⁄ R
free
¼ 17.5 ⁄ 20.6% and excellent geometry (Table 1).
Sequence and structure analysis
Sequence alignments were performed with clustalx [48].
NlpI sequences were obtained from, and BLAST searches
[49] performed through, the ExPASy database server
( [50]. Stereochemical quality of the
model was assessed with procheck [51]. Helix geometry
and packing angles were calculated with promotif [52],
and the CCP4 utility lsqman [53]. Main chain and side
contacts were determined using contact within CCP4
[42,43]. Contact map figures were generated with molmol
[54]. The dimer interface was assessed surfnet [55]
through the Protein–Protein Interaction Server (http://
www.biochem.ucl.ac.uk/bsm/PP/server/) at UCL, London,
UK [29]. Structural homologs were identified through the
DALI server ( at the European
Bioinformatics Institute, Hinxton, UK [56]. Electron den-
sity map images were created with bobscript [57], and ren-
dered with raster3d [58]. Structure representations were
prepared with pymol (DeLano Scientific, http://pymol.

sourceforge.net).
Model coordinates
Coordinates and structure factors have been submitted to
the RCSB Protein Databank, under accession code
1XNF.pdb.
Acknowledgements
This work was supported in part by NIH grants
GM62413 and GM57265 (L.R.). C.G.M.W. is a James
Hudson Brown–Alexander Brown Coxe Postdoctoral
Fellow (Yale University School of Medicine). T.K. is
supported by a postdoctoral fellowship from Helsingin
Sanomat Centennial Foundation (Finland). Data for
this study were measured at beamline X12C of the
National Synchrotron Light Source. We are grateful
for the assistance of Dr Anand Saxena in the use of
this beamline. The National Synchrotron Light Source,
Brookhaven National Laboratory, is supported by the
US Department of Energy, Division of Materials
Sciences and Division of Chemical Sciences, under
Contract No. DE-AC02–98CH10886. X12C is suppor-
ted principally by the Offices of Biological and Envi-
ronmental Research and of Basic Energy Sciences of
the US Department of Energy, and from the National
Center for Research Resources of the National Insti-
tutes of Health.
References
1 Cortajarena AL, Kajander T, Pan W, Cocco MJ &
Regan L (2004) Protein design to understand peptide
ligand recognition by tetratricopeptide repeat proteins.
Protein Eng Des Sel 17, 399–409.

2 Main ER, Jackson SE & Regan L (2003) The folding
and design of repeat proteins: reaching a consensus.
Curr Opin Struct Biol 13, 482–489.
3 Zweifel ME, Leahy DJ, Hughson FM & Barrick D
(2003) Structure and stability of the ankyrin domain of
the Drosophila Notch receptor. Protein Sci 12, 2622–
2232.
4 Bradley CM & Barrick DM (2002) Limits of cooperativ-
ity in a structurally modular protein: response of the
Notch ankyrin domain to analogous alanine substitu-
tions in each repeat. J Mol Biol 324, 373–386.
5 D’Andrea L & Regan L (2003) TPR proteins: the
versatile helix. Trends Biochem Sci 28, 655–662.
6 Main ER, Xiong Y, Cocco MJ, D’Andrea L & Regan
L (2003) Design of stable alpha-helical arrays from an
idealized TPR motif. Structure 11, 497–508.
7 Sikorski RS, Boguski MS, Goebl M & Hieter P (1990)
A repeating amino acid motif in CDC23 defines a
family of proteins and a new relationship among genes
required for mitosis and RNA synthesis. Cell 60, 307–
317.
8 Hirano T, Kinoshita N, Morikawa K & Tanagida M
(1990) Snap helix with knob and hole: essential
repeats in S. pombe nuclear protein nuc2+. Cell 60,
319–328.
9 Scheufler C, Brinker A, Bourenkov G, Pegoraro S,
Moroder L, Bartunik H, Hartl FU & Moarefi I (2000)
Structure of TPR domain-peptide complexes: critical
elements in the assembly of the Hsp70-Hsp90 multicha-
perone machine. Cell 101, 199–210.

10 Brinker A, Scheufler C, Von Der Mulbe F, Fleckenstein
B, Herrmann C, Jung G, Moarefi I & Hartl FU (2002)
Ligand discrimination by TPR domains. Relevance and
selectivity of EEVD-recognition in Hsp70 · Hop ·
Hsp90 complexes. J Biol Chem 277, 19265–19275.
11 Bateman A, Birney E, Cerruti L, Durbin R, Etwiller L,
Eddy SR, Griffiths-Jones S, Howe KL, Marshall M &
Sonnhammer EL (2001) The Pfam protein families
database. Nucleic Acids Res 30, 276–280.
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 177
12 Davie JK, Edmondson DG, Coco CB & Dent SY
(2003) Tup1-Ssn6 interacts with multiple class I histone
deacetylases in vivo. J Biol Chem 278, 50158–50162.
13 Gatto GJ Jr, Maynard EL, Guerrerio AL, Geisbrecht
BV, Gould SJ & Berg JM (2003) Correlating structure
and affinity for PEX5: PTS1 complexes. Biochemistry
42, 1660–1666.
14 Das AK, Cohen PW & Barford D (1998) The structure
of the tetratricopeptide repeats of protein phosphatase
5: implications for TPR-mediated protein–protein inter-
actions. EMBO J 17, 1192–11199.
15 Abe Y, Shodai T, Muto T, Mihara K, Torii H,
Nishikawa S, Endo T & Kohda D (2000) Structural
basis of presequence recognition by the mitochondrial
protein import receptor Tom20. Cell 100, 551–560.
16 Lapouge K, Smtih SJM, Walker PA, Gamblin SJ,
Smerdon SJ & Rittinger K (2000) Structure of the TPR
domain of p67phox in complex with Rac.GTP. Mol Cell
6, 899–907.

17 Gatto GJ Jr, Geisbrecht BV, Gould SJ & Berg JM
(2000) Peroxisomal targeting signal-1 recognition by
the TPR domains of human Pex5. Nat Struct Biol 7,
1091–1095.
18 Kumar A, Roach C, Hirsh IS, Turley S, Dewalque S,
Michels PAM & Hol WGJ (2001) An unexpected
extended conformation for the third Tpr motif of the
peroxin Pex5 from Trypanosoma brucei. J Mol Biol 307,
271–282.
19 Taylor P, Dornan J, Carrello A, Minchin RF,
Ratajczak T & Walkinshaw MD (2001) Two structures
of cyclophilin 40: folding and fidelity in the TPR
domains. Structure 9, 431–438.
20 Sinars CR, Cheung-Flynn J, Rimerman RA, Scammell
JG, Smith DF & Clardy JC (2003) Structure of the
large Fk506-binding protein Fkbp51, an Hsp90-binding
protein and a component of steroid receptor complexes.
Proc Natl Acad Sci USA 100, 868–873.
21 Dohm JA, Lee SJ, Hardwick JM, Hill RB & Gittis AG
(2004) Cytosolic domain of the human mitchondrial
fission protein FIS1 adopts a TPR fold. Proteins 54,
153–156.
22 Ohara M, Wu HC, Sankaran K & Rick PD (1999)
Identification and characterization of a new lipoprotein,
NlpI, in Escherichia coli K-12. J Bacteriol 181, 4318–
4325.
23 Seydel A, Gounon P & Pugsley AP (1999) Testing the
‘+2 rule’ for lipoprotein sorting in the Escherichia coli
cell envelope with a new genetic selection. Mol Micro-
biol 34, 810–821.

24 Wu HC (1996) Biosynthesis of lipoproteins. In Escheri-
chia coli and Salmonella: Cellular and Molecular Biology,
Vol. 1 (Neidhardt F C, ed.), pp 1005–1014. American
Society for Microbiology Press, Washington DC, USA.
25 Pugsley AP, Kornacker MG & Ryter A (1990) Analy-
sis of the subcellular location of pullulanase produced
by Escherichia coli carrying the pulA gene from
Klebsiella pneumoniae strain UNF5023. Mol Microbiol
4, 59–72.
26 Barnich N, Bringer MA, Claret L & Darfeuille-Miehaud
A (2004) Involvement of lipoprotein NlpI in the virulence
of adherent invasive Escherichia coli strain LF82 isolated
from a patient with Crohn’s disease. Infect Immun 72,
2484–2493.
27 Tadokoro A, Hayashi H, Kishimoto T, Makino Y,
Fujisaki S & Nishimura Y (2004) Interaction of the
Escherichia coli lipoprotein NlpI with periplasmic Prc
(Tsp) protease. J Biochem 135, 185–191.
28 Grant S, Jessee J, Bloom F & Hanahan D (1990)
Differential plasmid rescue from transgenic mouse
DNAs into Escherichia coli methylation-restriction
mutants. Proc Natl Acad Sci USA 87, 4645–4649.
29 Jones S & Thornton JM (1996) Principles of protein–
protein interactions derived from structural studies.
Proc Natl Acad Sci USA 93, 13–20.
30 Keiler KC & Sauer RT (1996) Sequence determinants
of C-terminal substrate recognition by the Tsp protease.
J Biol Chem 271, 2589–2593.
31 Steegborn C, Danot O, Huber R. & Clausen T (2001)
Crystal structure of transcription factor MalT domain

III: a novel helix repeat fold implicated in regulated oli-
gomerization. Structure 9 , 1051–1060.
32 Studier FW, Rosenberg AH, Dunn JJ & Dubendorff
JW (1990) Use of T7 RNA polymerase to direct expres-
sion of cloned genes. Methods Enzymol 185, 60–89.
33 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular
Cloning: A Laboratory Manual, 2nd edn. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, NY,
USA.
34 Pace CN, Vajdos F, Fee L, Grimsley G & Gray T
(1995) How to measure and predict the molar absorp-
tion coefficient of a protein. Protein Sci 4, 2411–2423.
35 Kantardjieff KA & Rupp B (2003) Matthews coefficient
probabilities: Improved estimates for unit cell contents
of proteins, DNA and protein-nucleic acid complex
crystals. Protein Sci 12, 1865–1871.
36 Matthews BW (1968) Solvent content of protein crys-
tals. J Mol Biol 33, 491–497.
37 Otwinowski Z. & Minor W (1997) Processing X-ray dif-
fraction data collected in oscillation mode. Methods
Enzymol 276, 307–326.
38 Terwilliger TC & Berendzen J (1999) Automated MAD
and MIR structure solution. Acta Cryst D 55, 849–861.
39 Terwilliger TC (1999) Reciprocal-space solvent flatten-
ing. Acta Cryst D 55, 1863–1871.
40 Kleywegt GA & Jones TA (1994) Halloween masks and
bones. In From First Map to Final Model (Bailey, S,
Hubbard, R. & Waller, D, eds), pp 59–66. SERC Dares-
bury Laboratory, Warrington, UK.
41 Cowtan K (1994) ‘dm’: An automated procedure for

phase improvement by density modification. Joint
Crystal structure of NlpI C. G. M. Wilson et al.
178 FEBS Journal 272 (2005) 166–179 ª 2004 FEBS
CCP4 ESF-EACBM Newsletter Protein Crystallogr 31,
34–38.
42 Collaborative Computational Project Number 4. (1994)
The CCP4 Suite: programs for protein crystallography.
Acta Cryst D 50, 760–763.
43 Potterton E, Briggs P, Turkenberg M & Dodson EJ
(2003) A graphical user interface to the CCP4 program
suite. Acta Cryst D 59, 1131–1137.
44 Brunger AT, Adams PD, Clore GM, DeLano WL, Gros
P, Grosse-Kunstleve RW, Jiang J-S, Kuszewski J,
Nilges N, Pannu NS, et al. (1998) Crystallography and
NMR system (CNS): a new software system for macro-
molecular structure determination. Acta Cryst D 54,
905–921.
45 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in elec-
tron density maps and the location of errors in these
models. Acta Cryst A 47, 110–119.
46 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the
maximum-likelihood method. Acta Cryst D 53,
240–255.
47 Perrakis A, Sixma TK, Wilson KS & Lamzin VS (1997)
wARP: improvement and extension of crystallo-
graphic phases by weighted averaging of multiple
refined dummy atomic models. Acta Cryst D 53, 448–
455.

48 Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F
& Higgins DG (1997) The CLUSTAL–windows inter-
face: flexible strategies for multiple sequence alignment
aided by quality analysis tools. Nucleic Acids Res 25,
4876–4882.
49 Altschul SF, Gish W, Miller W, Myers EW & Lipman
DJ (1990) Basic local alignment search tool. J Mol Biol
215, 403–410.
50 Gasteiger E, Gattiker A, Hoogland C, Ivanyi I, Appel
RD & Bairoch A (2003) ExPASy: the proteomics server
for in-depth protein knowledge and analysis. Nucleic
Acids Res 31, 3784–3788.
51 Laskowski RA, MacArthur MW, Moss DS & Thornton
JM (1993) PROCHECK: a program to check the stereo-
chemical quality of protein structures. J Appl Cryst 26,
283–291.
52 Hutchinson EG & Thornton JM (1996) PROMOTIF –
a program to identify and analyze structural motifs in
proteins. Protein Sci 5, 212–220.
53 Kleywegt GJ & Jones TA (1997) Detecting folding
motifs and similarities in protein structures. Methods
Enzymol 277, 525–545.
54 Koradi R., Billeter M & Wuthrich K (1996) MOLMOL:
a program for display and analysis of macromolecular
structures. J Mol Graphics 14, 51–55.
55 Laskowski RA (1995) SURFNET: a program for visu-
alizing molecular surfaces, cavities and intermolecular
interactions. J Mol Graphics 13, 323–330.
56 Holm L & Sander C (1993) Protein structure compar-
ison by alignment of distance matrices. J Mol Biol 233,

123–138.
57 Esnouf RM (1997) An extensively modified version of
molscript that includes enchanced colouring capabilities.
J Mol Graphics 15, 132–134.
58 Merritt EA & Murphy MEP (1994) Raster3d, Version
2.0. A program for photorealistic molecular graphics.
Acta Cryst D 50, 869–873.
C. G. M. Wilson et al. Crystal structure of NlpI
FEBS Journal 272 (2005) 166–179 ª 2004 FEBS 179