models may be required following recent
demonstrations that GroEL can help fold
some proteins that are too large to Ft into
the cavity of a single ring.
The mechanism of action of the Group
II proteins is much less well understood.
Actin and tubulin have both been shown to
interact with speciFc subunits of the CCT
complex (which are always present in the
same order in the complex), and there is
evidence that the substrates remain bound
throughout the folding cycle, unlike the
case in GroEL in which they are discharged
into the cavity. It may be that the large
domain movements of the complex that
take place during the folding cycle act to
more compact folded form.
Structure and Function of the Hsp60
It is the remarkable structure of the Hsp60
proteins that has been in part responsible
for the degree of interest that they have
raised. Members of both the Group I
and Group II families have been studied
by different structural techniques, notably
microscopy, and more recently in the
case of GroEL by NMR. Hsp60 proteins
assemble into double-ring structures with
several subunits in each ring; in the case of
CCT from eukaryotes, the relative position
of each of these subunits is unique, so
that each always has the same neighbors.
The individual subunits of proteins from
both families consist of three distinct
domains: an equatorial domain, which
is responsible for the contacts between
the two rings and for ATP-binding; a
ﬂexible intermediate domain; and an
apical domain, which contains the region
in which protein binding initially occurs.
Moreover, both structures undergo large
through the different stages of the protein
folding cycle. Again, it is the GroEL protein
that is best understood.
The sequence of events that is thought
to take place during the folding cycle of
the GroEL protein is depicted in ±ig. 3.
Partially folded protein binds to one ring
via a series of hydrophobic contacts with
residues at the top of the apical domain.
GroES binds to some of the same residues,
and when it binds (which requires the
presence of bound nucleotide) it displaces
the bound peptide into the center of the
cavity and simultaneously causes a very
large conformational change in the ring to
which it binds, the effect of which is to
nearly double the size of the cavity. This
structure slowly turns over the ATP, and
until this reaction is complete, no protein
or nucleotide can bind to the opposite ring.
Once the ATP has been turned over to
ADP, unfolded protein can bind to the
the net effect of which is to displace the
bound substrate, GroES, and ADP, from
the opposite ring to which binding Frst
occurred. Thus, both rings are active in
binding unfolded proteins, but do not do
so at the same time as each other.
The models for action of GroEL dis-
cussed above can now be seen in the light
of this mechanistic cycle. The AnFnsen
cage and assisted folding models predict
that the key event in the cycle is the encapsi-
dation of the unfolded protein, which may
lead to either the passive or active involve-
ment of the cavity in promoting protein
folding. The iterated annealing model sees
the main event as being the large confor-
mational change that takes place in the
apical domains of GroEL as they move to
bind GroES. It is proposed that these act
to pull apart misfolded protein, before re-
leasing it again at the start of the protein