A critical motif for oligomerization and chaperone activity
of bacterial a-heat shock proteins
Sonja Studer, Markus Obrist, Nicolas Lentze and Franz Narberhaus
Institute of Microbiology, Eidgeno
¨
ssische Technische Hochschule, Zu
¨
rich, Switzerland
Oligomerization into multimeric complexes is a prerequisite
for the chaperone function of almost all a-crystallin type
heat shock proteins (a-Hsp), but the molecular details of
complex assembly are poorly understood. The a-Hsp pro-
teins from Bradyrhizobium japonicum are suitable bacterial
models for structure-function studies of these ubiquitous
stress proteins. They fall into two distinct classes, A and B,
display chaperone activity in vitro and form oligomers of
 24 subunits. We constructed 19 derivatives containing
truncations or point mutations within the N- and C-terminal
regions and analyzed them by gel filtration, citrate synthase
assay and coaffinity purification. Truncation of more than
the initial few amino acids of the N-terminal region led to the
formation of distinct dimeric to octameric structures devoid
of chaperone activity. In the C-terminal extension, integrity
of an isoleucine-X-isoleucine (I-X-I) motif was imperative
for a-Hsp functionality. This I-X-I motif is one of the
characteristic consensus motifs of the a-Hsp family, and here
we provide experimental evidence of its structural and
functional importance. a-Hsp proteins lacking the C-termi-
nal extension were inactive, but still able to form dimers.
Here, we demonstrate that the central a-crystallin domain
alone is not sufficient for dimerization. Additional residues

exchange [19]. Mammalian members of the a-Hsp family
are remarkably polydisperse [20,21], and some bacterial
proteins such as Escherichia coli IbpB also display pro-
nounced size heterogeneity [22]. Particularly rigid structures
are of prokaryotic origin, such as the 24-meric Hsp16.5
from Methanococcus jannaschii or nonameric Hsp16.3 from
Mycobacterium tuberculosis [11,14].
a-Hsp proteins are widely distributed but poorly con-
served. One major characteristic of this protein family is the
presence of a central a-crystallin domain, flanked by a
N-terminal region and a C-terminal extension [23]. The
highest degree of amino-acid similarity is found within
the a-crystallin domain, while the N-terminal region and the
C-terminal extension are variable in length and sequence.
Naturally occurring a-Hsp proteins lacking these flanking
regions acquire monomeric to tetrameric structures and are
poor or inactive chaperones [24,25]. In the last few years,
systematic a-Hsp structure–function studies have mainly
been focused on the mammalian representatives aA- and
aB-crystallin. These proteins proved remarkably resistant
against mutational alterations. For example, a wide variety
of truncations and point mutations within the N-terminal
region had no consequence on protein structure and
function [26–30]. However, some modifications in the
N-terminal region of eukaryotic a-Hsp proteins were
reported to affect chaperone activity [31] or complex
formation [26,32].
The C-terminal extension, being located on the outer
surface of the a-Hsp oligomers, is generally assumed to
contribute to complex solubility [33], but its further

that the soil bacterium Bradyrhizobium japonicum is a
suitable model organism for investigating prokaryotic
a-Hsp proteins [4,13,37]. It contains at least 10 a-Hsp
proteins, which are highly induced upon heat shock [4]. The
sequences of seven B. japonicum a-Hsp genes are available
and can be assigned to two distinct classes, A and B. These
two classes are not restricted to B. japonicum, but also occur
in other rhizobia [38]. Members of both classes have been
shown to prevent citrate synthase aggregation in vitro,to
form 400- to 500-kDa complexes and to interact with other
a-Hsp proteins of the same class [13].
The present study demonstrates that chaperone activity
of B. japonicum a-Hsp proteins is stringently coupled to
multimerization, and that both the N- and C-terminal
regions are required for the formation of chaperone-
competent complexes. In particular, we show that integrity
of the I-X-I motif is critical for assembly of functional a-Hsp
proteins, and that the isolated a-crystallin domain is unable
to dimerize.
EXPERIMENTAL PROCEDURES
Plasmid construction
Plasmids for the expression of B. japonicum hspF (pRJ5306)
and hspH (pRJ5307) provided with a C-terminal His
6
tag
have been described previously [13]. N- or C-terminal
truncations of hspF and hspH were constructed by PCR,
using pRJ5306 and pRJ5307 as templates. For subsequent
cloning, we took advantage of two unique restriction sites,
namely BsaIonhspF and SfiIonhspH. PCR products were

isopropyl thio-b-
D
-galactoside to a final concentration of
0.5 m
M
, and grown for a further 2–3 h. After harvesting,
cells were resuspended in binding buffer (500 m
M
KCl,
20 m
M
Tris/HCl, 5 m
M
imidazole, pH 7.9) containing
1m
M
phenylmethanesulfonyl fluoride and 10 lgÆmL
)1
DNaseI. Lysis was performed in a French pressure cell at
1000 p.s.i, and soluble crude extracts were prepared by
centrifugation at 12 000 g for 30 min at 4 °C.
Proteins were purified by Ni-nitrilotriacetic acid affinity
chromatography (Ni-nitrilotriacetic acid resin from Qiagen)
under native conditions essentially as described previously
[13]. The column was pre-equilibrated with binding buffer
and then washed with washing buffer (500 m
M
KCl, 20 m
M
Tris/HCl, pH 7.9) containing increasing imidazole concen-

Biotech). Prior to use, CS (Sigma) was dialyzed against Tris/
EDTA buffer (10 m
M
Tris/HCl, 1 m
M
EDTA, pH 8.0).
For each protein, chaperone activity assays were performed
at least twice with preparations from independent purifica-
tions.
Gel filtration
Analytical size exclusion chromatography of purified pro-
teins was performed at room temperature on a Superdex
200 HR 30/10 column (Amersham Pharmacia Biotech)
using a BioCAD perfusion chromatography system
(PerSeptive Biosystems). After equilibrating the column
with elution buffer (500 m
M
KCl, 20 m
M
Tris/HCl, 250 m
M
imidazole, pH 7.9), 200-lL protein samples were injected
and separated at a flow rate of 0.6 mLÆmin
)1
. Absorbance
was recorded at 280 nm. The following standards were used
Ó FEBS 2002 A cricital motif for a-Hsp oligomerization (Eur. J. Biochem. 269) 3579
for calibration: thyroglobulin (669 kDa), ferritin (440 kDa),
aldolase (158 kDa), albumin (67 kDa), and ribonuclease A
(13.7 kDa), all from Amersham Pharmacia Biotech. Gel

and HspF(D40N), approximately half of the N-terminal
region was eliminated, including a proline and an arginine
residue conserved in all seven B. japonicum a-Hsp proteins
(Fig. 1). To examine the importance of these two particular
residues, we constructed two HspH derivatives in which the
proline (P8A) or the arginine (R18A) was replaced by an
alanine.
The characteristic I-X-I motif in the C-terminal extension
is present in all B. japonicum and E. coli a-Hsp proteins. To
investigate its role in oligomerization, we constructed two
a-Hsp derivatives lacking one [HspF(D5C)] or both
[HspH(D20C)] isoleucine residues. Two further constructs,
HspH(D5C) and HspH(D15C), contained the entire I-X-I
motif. To assess the role of the conserved motif in the
context of the full-length protein, three HspH variants were
constructed, in which the isoleucines were replaced by
alanine, either individually, resulting in HspH(I133A) and
HspH(I135A), or simultaneously, leading to
HspH(II133,135AA).
Finally, we searched for the minimal fragment required
for dimer formation. For this purpose HspH variants
lacking entire subregions were constructed. Three proteins,
Fig. 1. Amino-acid alignment of B. japonicum and E. coli a-Hsp proteins. The alignment includes the seven known sequences of B. japonicum a-Hsp
proteins, belonging to class A [HspH (accession number O86110), HspA (P70917), HspB (P70918), HspD (O69241) and HspE (O69242)] and B
[HspC (AAC44757) and HspF (CAA05837)]. IbpA (P29209) and IbpB (G65170) from E. coli (both class A) were added for comparison. The
alignment was created with
CLUSTAL W
[48]. Amino acids that are identical in all analyzed proteins are shown in white letters shaded in black. White
letters and black letters shaded in grey indicate amino acids that are identical in at least 80 or 60% of all proteins, respectively. Arrows mark
truncations introduced to HspH and HspF, and asterisks indicate alanine exchange mutations in HspH. Note that four N-terminal residues were

was only partially compromised, whereas chaperone activity
was completely abolished. Note that in many gel filtration
runs [e.g. HspH or HspH(D3N)], a considerable portion of
the protein eluted after 12–13 min in the void volume of the
column, reflecting the tendency of all B. japonicum a-Hsp
proteins to form large aggregates. Extended truncations in
the N-terminal region of HspF caused similar, but not
identical effects as in HspH. When 30 or more amino-acid
residues were removed from the N-terminus of HspF,
chaperone activity was completely lost (Fig. 2F,G) and
oligomer formation was impaired even more drastically
than in the HspH derivatives. Gel filtration analysis of
HspF(D30N) and HspF(D40N) suggested that both pro-
teins were only dimers (Fig. 3F,G), whereas HspH deri-
vatives devoid of the first half of the N-terminal region
[HspH(D20N)] appeared as octameric complexes. In either
case, it is evident that a largely intact N-terminal region
is strictly required for the formation of active a-Hsp
oligomers.
In an attempt to narrow down the critical determinants to
the amino-acid level, we tested the effect of two single
amino-acid exchanges (P8A and R18A) on chaperone
activity and oligomeric state of HspH. Although these
residues are conserved throughout class A and class B
proteins of B. japonicum (Fig. 1), neither exchange inhibited
chaperone activity and oligomerization (data not shown).
Fig. 2. Effect of truncations in the N-terminal
region on chaperone activity of HspH and
HspF. Thermally induced aggregation of
citrate synthase at 43 °Cisdepictedasa

When assessing the impacts of C-terminal truncations on
oligomerization and chaperone function, we focused our
attention on the conserved I-X-I motif. First, the C-terminal
extension of HspH was shortened by five and 15 amino
acids [HspH(D5C) and HspH(D15C), respectively], leaving
the I-X-I motif intact. These alterations had no influence on
chaperone activity. Both proteins efficiently protected CS
from thermally-induced aggregation (Fig. 4A,B) and as-
sembled into large complexes (Fig. 5A,B). But while
HspH(D5C) appeared similar to full-length HspH in terms
of oligomer formation, HspH(D15C) formed larger aggre-
gates with an apparent molecular mass exceeding 2 MDa.
All C-terminally truncated a-Hsp derivatives exhibited an
increased tendency to precipitate.
Elimination of the entire I-X-I motif in HspH(D20C)
led to a complete loss of chaperone activity (Fig. 4C) and
a protein that was unable to multimerize. The only species
encountered in the gel filtration profile was a dimer of
 29 kDa (Fig. 5C). Removal of the C-terminal isoleucine
from the I-X-I motif in HspF(D5C) was sufficient to
completely abolish chaperone activity (Fig. 4D) and
Fig. 3. Oligomerization of N-terminally truncated HspH and HspF derivatives. The oligomeric state of purified a-Hsp variants was determined by gel
filtration over a Superdex 200 column at a flow rate of 0.6 mLÆmin
)1
. N-Terminally truncated derivatives of HspH (A–D) and HspF (E–G) were
analyzed and compared with native HspH (H) and HspF (I). The minor peak at 12–13 min that is observed in all gel filtration profiles represents
large protein aggregates eluting in the column’s void volume. J, calibration curve for the gel filtration profiles in (A–I). K
av
¼ (V
e

M
(m), 600 n
M
(·)and1.2l
M
(d); as well as the absorbance
of a-Hsp proteins in the absence of CS (S).
3582 S. Studer et al. (Eur. J. Biochem. 269) Ó FEBS 2002
severely impaired oligomer formation (Fig. 5D). The
apparent molecular mass of 44–48 kDa probably repre-
sents a dimer.
As both truncations affecting the I-X-I motif impaired
chaperone activity and oligomer formation, we replaced the
two isoleucine residues in HspH by alanines and analyzed
whether these mutations had a similar effect. In fact,
HspH(I133A), HspH(I135A) and HspH(II133,135AA)
were devoid of chaperone activity (Fig. 4E–G) and
appeared as small, dimeric structures of 43–50 kDa
(Fig. 5E–G). Summarizing the gel filtration data, the
calibration curve in Fig. 5H illustrates that all modifications
affecting the I-X-I motif resulted in the formation of small,
presumably dimeric complexes. Taken together, these
results demonstrate that the isoleucine motif plays a crucial
structural and functional role in the assembly of functional
HspH or HspF oligomers.
The a-crystallin domain alone is not sufficient
for dimer formation
All HspF and HspH variants investigated so far acquired
dimeric structures, indicating that the dimerization motif
remained untouched. In order to determine the a-Hsp

column [13].
Untagged HspH did not coelute with the His
6
-tagged
a-crystallin domain (Fig. 6A), suggesting that the isolated
a-crystallin domain was unable to dimerize. The a-crys-
tallin domain carrying the C-terminal extension appeared
to enable weak interactions with native HspH, because
after copurification of HspH(aC)–His
6
with HspH, a faint
band corresponding to HspH was visible on SDS gels. In
contrast, HspH(Na)–His
6
, an HspH derivative consisting
of the a-crystallin domain plus the N-terminal region,
strongly interacted with native HspH. After co-affinity
purification, both proteins were detected on SDS gels in a
1 : 1 ratio. A similarly efficient copurification of HspH
was observed with HspH(D20N)–His
6
. This suggests that
the a–Hsp interaction is in part mediated by a portion of
the N-terminal region in vicinity of the a-crystallin
domain.
To ensure that efficient dimerization requires solely
a-crystallin domain and N-terminal region and not the
C-terminal extension, HspH(Na)–His
6
was also subjected to

Some prokaryotic a-Hsp proteins possess remarkably rigid
structures [11,14] that render them more susceptible to point
mutations and truncations. We therefore chose two dis-
tantly-related bacterial a-Hsp proteins from B. japonicum in
order to study, in detail, the contribution of their subregions
to chaperone activity and oligomerization.
Tight coupling of chaperone activity
and complex assembly
B. japonicum HspH and HspF oligomers contain  24
subunits, as shown previously with crude extracts [13] and
confirmed with purified proteins in the present study.
Transmission electron microscopical analysis of both pro-
teins revealed roughly spherical structures (data not shown)
similar to those observed for other members of the a-Hsp
family [10,12,20,21].
We have demonstrated that assembly of this native
structure is severely impaired by truncations in the
N-terminal region and the C-terminal extension. Interest-
ingly, all of the resulting low molecular mass complexes
were devoid of chaperone activity. Although we cannot rule
out that the truncated proteins had lost substrate binding
regions in addition to oligomerization sites, this finding
illustrates that chaperone activity and oligomerization are
tightly coupled. The ability to bind unfolded proteins and
prevent them from aggregation clearly requires fully
assembled multimeric a-Hsp complexes. Even complexes
that exceed the size of the wild-type protein, e.g.
HspHD15C, retained chaperone activity. Much like our
a-Hsp variants, naturally occurring multimerization-incom-
petent a-Hsp proteins have poor or lacking chaperone

to the deduced a-crystallin domain of HspH (Fig. 1) in
order to avoid disruption of a predicted b strand was not
sufficient for dimer formation. This argues that more
residues towards the N-terminus are involved in dimeriza-
tion.
Importance of the N-terminal region
Additional portions of the N- and C-terminal regions are
required for assembly into functional multimeric complexes.
Interestingly, class A (HspH) and class B (HspF) proteins
exhibited similar, yet not identical assembly properties.
Truncations in the N-terminal region of HspF interfered
Fig. 6. Interaction of truncated HspH variants with full-length and
truncated HspH demonstrated by coaffinity purification. Crude extracts
with overexpressed native HspH were mixed with extracts containing
truncated His
6
-tagged HspH variants according to the schematic
representation on the right half of the figure. After denaturation and
renaturation, the extracts were applied to Ni-nitrilotriacetic acid
affinity columns, and eluates were analyzed by tricine SDS/PAGE [49].
Coomassie-stained eluates on 13% polyacrylamide gels are shown on
the left half of (A) and (B). The expected positions of relevant proteins
are indicated by arrows. (A) Copurification of untagged HspH with
His
6
-tagged HspH derivatives. Lane 1, purification of the His
6
-tagged
HspH variant; lane 2, His
6

disordered [14,34]. Possibly, the N-terminal regions of
B. japonicum a-Hsp proteins are also disordered, which
might explain why subunit contact sites in the N-terminal
region could not yet be narrowed down to individual
conserved residues.
A crucial motif in the C-terminal extension
Wheat Hsp16.9 and M. jannaschii Hsp16.5 are distantly
related proteins with distinct quaternary structures. Never-
theless, the role of their C-terminal extensions in oligomeric
assembly is strikingly similar. The C-termini reach out to
neighbouring subunits, where the I-X-I motif undergoes
intramolecular hydrophobic interactions with a b strand in
the a-crystallin domain [14,34]. Our mutational studies
strongly suggest that the C-terminal extension and the
conserved I/V-X-I/V motif in particular plays a similar
structural role in both classes of bacterial a-Hsp proteins.
Partial or entire removal of the two conserved isoleucines
led to a dramatic reduction in complex size and solubility
and a complete loss of chaperone activity. All modifications
touching the motif resulted in dimers. On the other hand,
truncations that left the motif intact did not inhibit
multimerization and chaperone activity. Similarly, alter-
ations outside of the I/V-X-I/V motif did not impair
oligomer formation of vertebrate a-Hsp proteins [31,33],
though some of them lowered chaperone activity [31,35]. So
far, very few mutational studies have included the isoleucine
motif. C-Terminally truncated C. elegans Hsp16–2 lacking
the I-X-I motif still formed high molecular mass complexes
and retained full chaperone activity [47]. In aA-crystallin,
C-terminal truncations touching the I/V-X-I/V motif did

4. Mu
¨
nchbach,M.,Nocker,A.&Narberhaus,F.(1999)Multiple
small heat shock proteins in rhizobia. J. Bacteriol. 181, 83–90.
5. de Jong, W.W., Caspers, G.J. & Leunissen, J.A.M. (1998) Gen-
ealogy of the a-crystallin – small heat-shock protein superfamily.
Int. J. Biol. Macromol. 22, 151–162.
6. Lee, G.J., Roseman, A.M., Saibil, H.R. & Vierling, E. (1997) A
small heat shock protein stably binds heat-denatured model sub-
strates and can maintain a substrate in a folding-competent state.
EMBO J. 16, 659–671.
7. Ehrnsperger, M., Graber, S., Gaestel, M. & Buchner, J. (1997)
Binding of non-native protein to Hsp25 during heat shock creates
a reservoir of folding intermediates for reactivation. EMBO J. 16,
221–229.
8. Veinger, L., Diamant, S., Buchner, J. & Goloubinoff, P. (1998)
The small heat-shock protein IbpB from Escherichia coli stabilizes
stress-denatured proteins for subsequent refolding by a multi-
chaperone network. J. Biol. Chem. 273, 11032–11037.
9. Horwitz, J. (1992) a-crystallin can function as a molecular chap-
erone. Proc. Natl Acad. Sci. U.S.A. 89, 10449–10453.
10. Kim, R., Kim, K.K., Yokota, H. & Kim, S.H. (1998) Small heat
shock protein of Methanococcus jannaschii, a hyperthermophile.
Proc. Natl Acad. Sci. U.S.A. 95, 9129–9133.
11. Chang, Z.Y., Primm, T.P., Jakana, J., Lee, I.H., Serysheva, I.,
Chiu, W., Gilbert, H.F. & Quiocho, F.A. (1996) Mycobacterium
tuberculosis 16-kDa antigen (Hsp16.3) functions as an oligomeric
structure in vitro to suppress thermal aggregation. J. Biol. Chem.
271, 7218–7223.
12. Lee,G.J.,Pokala,N.&Vierling,E.(1995)Structureandin vitro

J. Mol. Biol. 277, 27–35.
21. Haley, D.A., Bova, M.P., Huang, Q.L., Mchaourab, H.S. &
Stewart, P.L. (2000) Small heat-shock protein structures reveal a
continuum from symmetric to variable assemblies. J. Mol. Biol.
298, 261–272.
22. Shearstone, J.R. & Baneyx, F. (1999) Biochemical characteriza-
tion of the small heat shock protein IbpB from Escherichia coli.
J. Biol. Chem. 274, 9937–9945.
23. Caspers, G.J., Leunissen, J.A. & de Jong, W.W. (1995) The
expanding small heat-shock protein family, and structure predic-
tions of the conserved Ôa-crystallin domainÕ. J. Mol. Evol. 40,238–
248.
24. Kokke, B.P.A., Leroux, M.R., Candido, E.P.M., Boelens, W.C. &
de Jong, W.W. (1998) Caenorhabditis elegans small heat-shock
proteins Hsp12.2 and Hsp12.3 form tetramers and have no cha-
perone-like activity. FEBS Lett. 433, 228–232.
25. Leroux, M.R., Ma, B.J., Batelier, G., Melki, R., Peter, E. &
Candido, M. (1997) Unique structural features of a novel class of
small heat shock proteins. J. Biol. Chem. 272, 12847–12853.
26. Bova, M.P., Mchaourab, H.S., Han, Y. & Fung, B.K.K. (2000)
Subunit exchange of small heat shock proteins – analysis of
oligomer formation of aA-crystallin and Hsp27 by fluorescence
resonance energy transfer and site-directed truncations. J. Biol.
Chem. 275, 1035–1042.
27. Andley, U.P., Mathur, S., Griest, T.A. & Petrash, J.M. (1996)
Cloning, expression and chaperone-like activity of human
aA-crystallin. J. Biol. Chem. 271, 31973–31980.
28. Derham, B.K., van Boekel, M.A., Muchowski, P.J., Clark, J.I.,
Horwitz, J., Hepburne-Scott, H.W., de Jong, W.W., Crabbe, M.J.
& Harding, J.J. (2001) Chaperone function of mutant versions of

tion of endogenous and heterologous small heat-stress proteins in
tobacco protoplasts. Plant J. 24, 397–411.
37. Mu
¨
nchbach, M., Dainese, P., Staudenmann, W., Narberhaus, F. &
James, P. (1999) Proteome analysis of heat shock protein expres-
sion in Bradyrhizobium japonicum. Eur. J. Biochem. 264, 39–48.
38. Nocker, A., Krstulovic, N P., Perret, X. & Narberhaus, F. (2001)
ROSE elements occur in disparate rhizobia and are functionally
interchangeable between species. Arch. Microbiol. 176, 44–51.
39. Koteiche, H.A. & Mchaourab, H.S. (1999) Folding pattern of the
a-crystallin domain in aA-crystallin determined by site-directed
spin labeling. J. Mol. Biol. 294, 561–577.
40. Kappe
´
, G., Leunissen, J.A.M. & de Jong, W.W. (2002) Evolution
and diversity of prokaryotic small heat shock proteins. Prog. Mol.
Subcell. Biol. 28, 1–17.
41. Narberhaus, F. (2002) a-Crystallin-type heat shock proteins:
Socializing minichaperones in the context of a multichaperone
network. Microbiol. Mol. Biol. Rev. 66, 64–93.
42. van de Klundert, F., Smulders, R.H., Gijsen, M.L.J., Lindner,
R.A., Jaenicke, R., Carver, J.A. & de Jong, W.W. (1998) The
mammalian small heat-shock protein Hsp20 forms dimers and is a
poor chaperone. Eur. J. Biochem. 258, 1014–1021.
43. Feil, I.K., Malfois, M., Hendle, J., van der Zandt, H. & Svergun,
D.I. (2001) A novel quaternary structure of the dimeric a-crystallin
domain with chaperone-like activity. J. Biol. Chem. 276, 12024–
12029.
44. Haslbeck,M.,Walke,S.,Stromer,T.,Ehrnsperger,M.,White,