 |
INTRODUCTION |
"Oh wad some Power the giftie gie us
To see oursels as ithers
see us!
It wad frae monie a blunder free us."
Robert Burns,
1786
The history of the
chaperonins involves two diverse, seemingly unrelated observations
dating back to the 1970s
the genetics of bacterio-phage morphogenesis
and the synthesis of Rubisco during the biogenesis of chloroplasts.
They coalesce in the late 1980s with the demonstration by the groups of
Georgopoulos and Ellis that GroEL and the Rubisco-binding protein are
homologous to one another (9) and with our demonstration in 1989 of
what it was the chaperonin proteins really do (7).
| |
GENETICS OF BACTERIOPHAGE MORPHOGENESIS |
In the early 1970s Costa Georgopoulos isolated
temperature-sensitive (ts), mutant strains of
Escherichia coli that were unable to support the growth of
phage 
and a number of other bacteriophages (6 and references
therein). The E. coli gene responsible for this phenotype
was eventually given the name groE. By dint of some further
genetics Costa, together with Barbara Hohn, identified the
groE gene product, a protein with a subunit
Mr of approximately 65 that we now
know as GroEL (6). This enabled the groups of Roger Hendrix and Barbara
Hohn to purify and record the first, now familiar, electron micrographs
of GroEL. It was a 14-mer consisting of two heptameric rings stacked
back to back. It also hydrolyzed ATP. Quite what GroEL did in the cell
and how it was involved in phage morphogenesis was not known at this
stage. Nevertheless, Hohn et al. (10) wrote "In groE mutants,
bacteriophage T4 capsid protein aggregates in lumps at the cell
membrane instead of forming normal capsids; bacteriophage head
protein aggregates to form polyheads, spirals and inactive prehead-like
particles and bacteriophage T5 tail assembly is affected. These
phenomena could be explained by assuming an assembly aiding function
present in normal cells but missing or malfunctional in the mutant."
In retrospect you can see that they were describing a malfunction in
protein folding that leads to aggregation instead of the native state.
But recall also that these observations were made immediately after
Anfinsen's seminal work (1) that provided the dominating intellectual
framework in the 1970s and 1980s in all matters related to protein
folding. By then the prevailing dogma was that protein folding was a
spontaneous event and suggesting that the folding of some proteins was
assisted by other proteins was tantamount to heresy.
With his bank of groE mutants Costa next designed a screen
to look for suppressor mutations. This led to the discovery of GroES
(6) and to evidence that the two groE gene products, GroEL and GroES, interacted with one another in vivo. Next Costa's group purified GroES, showed that in vitro it formed a 1:1 complex with GroEL
in the presence of ATP, that GroES inhibits the ATPase activity of
GroEL, and that it had a ring-like oligomeric structure. Still the
function of GroEL and GroES remained elusive.
Costa also reported two further genetic observations relating to GroEL
and GroES the significance of which was not then immediately obvious
but which, with the benefit of hindsight, would later become more
significant. The first was that a temperature-sensitive mutation could
be suppressed by over expressing GroEL and GroES (4). It is now
apparent that many (not all) ts mutations are the result of
a malfunction in protein folding at the restrictive temperature. Later,
when we knew what GroEL and GroES really did, my colleagues at
DuPont demonstrated that one could suppress
ts-mutations in a wide variety of structurally unrelated
proteins simply by over-expressing both GroEL and GroES in the cells
harboring the ts-mutant genes (15).
The second important observation relates to the indispensable
nature of the chaperonins. By the late 1980s it had become apparent that GroEL and GroES were an important part of the heat shock response.
Using a genetic approach Costa's group demonstrated that both GroEL
and GroES were indispensable proteins for bacterial growth at all
temperatures between 20°C and 40°C (5). Again, the full
significance of this result does not really become apparent until one
knows the function of GroEL and GroES within the cell, assisting other
proteins to fold. There are few events more central, more fundamental
to the cell.
| |
BIOSYNTHESIS OF RUBISCO |
In the late 1970s, John Ellis and his colleagues found that one
could study protein synthesis in chloroplasts by feeding radiolabeled amino acids to intact, isolated chloroplasts. The major product of such
a synthesis is the large subunit of Rubisco. After incubating intact,
isolated chloroplasts with [35S]Met for varying
lengths of time, Barraclough and Ellis (2) quenched the reaction by
osmotically lysing the chloroplasts and removed the green membranous
material. They first subjected the soluble material to non-denaturing
PAGE. After electrophoresis they either stained the gel with Coomassie
or subjected it to autoradiography. They observed that the
radioactivity was initially associated with a slower migrating band
than the holoenyzme of Rubisco. Only later did radioactivity appear
together with the holo-enzyme, implying a precursor-product
relationship. When the slower migrating radioactive band was excised
and subjected to SDS-PAGE, the radioactive band comigrated with the
large subunit of Rubisco, whereas the Coomassie stain was associated
with a protein with a subunit size of about 60 kD. They concluded that
the synthesis of the holo-enzyme of Rubisco involved the transient
formation of a complex with another large, oligomeric protein, the
Rubisco subunit binding protein (RsuBP). Because at that time there was
no intellectual framework into which the observation could be fitted,
so pervasive was the Anfinsen view (1), this observation had little
impact beyond those interested in chloroplast biogenesis or Rubisco. Ellis's group subsequently purified RsuBP and noted that it was an
oligomer of approximately 750 kD with a subunit mass of approximately 60 kD and that it bound ATP (11).
About then Harry Roy's group did a series of pulse-chase experiments
following the fate of the Rubisco large subunits (13). They showed that
the "hot" large subunits first sedimented as a 29S complex; this
was the binary RsuBP
large subunit complex seen in the Ellis's
non-denaturing gels. Next, in the chase the "hot" large subunits
sedimented as a broad 7S species. With the benefit of hindsight and the
crystal structure, I believe that these 7S species were a mixture of
L2 dimers and
(L2)2 tetramers en route to
forming the (L2)4 octomeric
core. They also showed that ATP was needed to chase the material out of
the 29S complex into the 7S species. Finally, the 7S species was chased
into the holoenzyme, which sediments at 18S. In retrospect these papers were very insightful, although at the time they were unappreciated given the prevailing dogma.
In 1987 at a NATO meeting John Ellis first used the term "molecular
chaperone" in referring to RsuBP. Later (3) he speculated on the
existence of a class of proteins (molecular chaperones) "whose
function is to ensure that the folding of certain other polypeptide
chains and their assembly into oligomeric structures occur
correctly." He went on to suggest that molecular chaperones "do not
form part of the final structure nor do they necessarily possess steric
information specifying assembly."
Shortly after that Tony Gatenby and I decided to take a serious look at
the RsuBP. By "serious" I meant mechanistic studies with purified
materials. So I purified some RsuBP while Tony developed an E. coli-based in vitro transcription-translation system for making
radiolabeled large subunits. However, our first experiment yielded a
surprise. While the autoradiogram of the non-denaturing-PAGE gel
clearly showed the presence of RsuBP-large subunit binary complex, the
control, to which no RsuBP had been added, also showed the very same
complex! I rudely suggested to Tony that he must have added RsuBP to
both samples. Tony was not amused! But we did not wait long for an
explanation, for within a short time John Ellis informed us that RsuBP
was homologous to GroEL, which obviously (in hindsight) was present in
the E. coli transcription-translation system. Our response
was "GroEL? What's that?"
CONNECTING PHAGE MORPHOGENESIS AND RUBISCO SYNTHESIS
A protein in plants similar to GroEL was first observed by Tsuprun
and Pushkin in 1981 (12). They reported electron micrographs and other
properties of a protein from pea leaves indistinguishable from GroEL.
However, in the absence of a function for this protein the significance
of these results was not obvious or appreciated. A further 6 years were
to pass before the connection between phage morphogenesis and Rubisco
synthesis was established (9). This came about by the cloning and
sequencing of RsuBP and of the GroE operon. RsuBP
and GroEL were clearly homologous proteins. The gene for GroES was
located immediately upstream of GroEL. The authors begin the discussion
as follows: "We have described a ubiquitous, conserved, abundant
protein that is associated with the post-translational assembly of at
least two structurally distinct oligomeric protein complexes. We
conclude that the role of this protein is to assist other polypeptides
to maintain or assume conformations which permit their correct assembly
into oligomeric structures."
The function of RsuBP in assisting the assembly of Rubisco in the
chloroplast is immediately obvious. Less obvious was the function of
GroEL and GroES in E. coli. It is clearly not there to
assist the folding of Rubisco, a protein not normally found in E. coli; nor can the sole function of GroEL and GroES be the assembly
of new phage particles. It has to be doing something of benefit to
E. coli. So we (Pierre Goloubinoff, Tony Gatenby, and I)
reasoned that if the synthesis of Rubisco in chloroplasts involves
RsuBP, the chloroplast version of GroEL, then the synthesis of
recombinant Rubisco in E. coli would also involve GroEL and perhaps also GroES. We had two plasmids, one encoding the
cyanobacterial Rubisco operon for both large and small subunits and the
other encoding the dimeric L2 Rubisco. While
expression of these proteins in E. coli did yield some
soluble, biologically active Rubisco, most of the Rubisco was an
insoluble, inactive inclusion body. Pierre engineered a plasmid
encoding the GroE operon so that one could over-express
GroEL and GroES and Rubisco in the same E. coli cell. The
total amount of Rubisco protein expressed by E. coli was
approximately the same whether or not one also over-expressed GroE. What was dramatically different, however, was the fate
of Rubisco. In the cells that over-expressed GroE nearly all
of the Rubisco was soluble and biologically active. For this to occur required both GroEL and GroES to be over-expressed. This was true for
both L2 and
S4 · (L2)4 · S4
forms of Rubisco. We thus concluded that the formation of active
Rubisco in E. coli involved GroEL and GroES
post-translationally at some point between the formation of the nascent
polypeptide and the formation of the L2 dimer (8).
At this point I felt we had a great opportunity to figure out exactly
what GroEL and GroES did. We knew the identity of one of its in vivo
substrates, the large subunit of Rubisco. We knew how to assay
Rubisco. We had also narrowed the involvement of GroE to two
possibilities: either the folding of the monomer and/or the association
of the folded monomers to give the biologically active dimer. Some have
described our next step as "very bold." We purified GroEL and GroES
to homogeneity and set out to do some neo-Anfinsen experiments using
chemically denatured L2 Rubisco as the substrate.
The results of the very first such experiment sent Pierre over the
moon. "It works, it works! We've got to write a paper for Nature!" I was less impressed. It was true that no active
Rubisco was formed at all if one omitted any of the components (GroEL, GroES, Mg2+, or ATP). But compared with the
control with native Rubisco, the recovery was miserable, approximately
1% as I recall. However, I spotted a potential flaw in the
experimental protocol. Thinking that GroEL/ES would act as an enzyme,
Pierre had used substrate (unfolded Rubisco) in molar excess. In short
there was too much Rubisco in the pot. So Pierre was sent back to the
bench with instructions to use equimolar amounts of unfolded Rubisco
and GroEL 14-mers. Wow! Now we recovered >80%. It was almost too good to be true and so the very next day I repeated (successfully) the
experiment with my own hands. Now I was over the moon! Now we could do
the kind of experiments that enzymologists like to do with purified
components instead of with messy, ill-defined mitochondrial or
chloroplast extracts. It did not take us very long to work out some of
the essential features, and we reported this in a paper to
Nature (7).
The Nature paper contains many mechanistic insights. We
begin by defining three species of Rubisco on the basis of
circular dichroism: native Rubisco N, unfolded (in urea or
GdnHCl) Rubisco U, and acid denatured Rubisco A. Rubisco A had some
secondary structure, and we suggested that it might correspond to a
molten globule, a term that was then fashionable but which I no longer consider very useful. We were lucky on two counts. First, we
arbitrarily chose conditions for refolding that were non-permissive;
i.e. no reconstitution of enzymatic activity occurred in the absence of
the complete system (GroEL, GroES, and MgATP). Second, there was a
fourth component needed that we luckily included, a monovalent cation
K+ or
NH4+. Our preparations of GroEL
and GroES were stored as ammonium sulfate precipitates, and enough
NH4+ was carried over into the
reaction mix to satisfy that particular requirement.
We showed that ATP hydrolysis was required by quenching the reaction
with hexokinase plus Glc (to convert the ATP to ADP). The yield of
refolded Rubisco depended on whether one started with Rubisco U or
Rubisco A, but the rate constant was the same regardless. We concluded
that "reconstitution involves a common intermediate, the formation of
which preceeds the same chaperonin-dependent rate-determining step. We
propose that on removal of the denaturant by dilution, a rapid
chaperonin-independent folding of Rubisco U to a state rich in
secondary structure and resembling Rubisco A (the Rubisco I state), is
followed by reaction with GroEL, GroES and Mg-ATP."
We noted that one required a slight molar excess of GroEL oligomers
over Rubisco monomers. We attributed this to the instability of the
Rubisco-I state and established that, absent a molar equivalent of
GroEL oligomers, Rubisco-I underwent irreversible aggregation. We wrote
"when either Rubisco U or Rubisco A was diluted into solutions
containing the chaperonins, two competitive and mutually exclusive
processes occurred. The unfolded or partly folded protomers either
formed biologically unproductive aggregates or they formed a binary
complex with GroEL, which prevented aggregation and directed the
protomers along a biologically productive pathway." Later, we
amplified this in the discussion "It is clear that the concept of a
stable substrate `patiently' awaiting interaction with
substoichiometric quantities of a catalyst does not apply here.
Instead, there is an urgent need to stabilize this fickle intermediate
by formation of a binary complex with GroEL."
We next demonstrated the existence of this binary complex by
electrophoretic and immunological methods commenting that "the need
to sequester unstable folding intermediates evidently exists in vivo as
well as in vitro. The formation of a binary complex between the Rubisco
large subunit and chloroplast GroEL was demonstrated long ago." Here
we refer to the Barraclough-Ellis experiment of a decade before (2).
We next showed that the discharge of this binary GroEL-Rubisco-I
complex depended on GroES and MgATP. We concluded that "the chaperonin-dependent reconstitution of Rubisco involves a strictly ordered set of reactionsfirst, the formation of a stable binary complex of GroELRubisco-I, which at least in vitro can be
demonstrated as an MgATP- and GroES-independent event, followed by the
MgATP- and GroES-dependent discharge of folded and stable, but
catalytically inactive monomers, which subsequently assemble into
active dimers."
In our concluding paragraph we also noted that "the E. coli chaperonins clearly did not evolve to facilitate the folding
or assembly, or both, of Rubisco, a protein not normally found in E. coli. It follows that the interaction between GroEL and
the partly folded Rubisco cannot be specific for Rubisco alone.
Instead, the specificity of the interaction must lie in some so far
unidentified structural element of the partly folded protein. Because
the native protein shows no detectable interaction with GroEL, this
structural element must be missing from, or inaccessible in, the native protein."
It was not very long before purified, chemically denatured proteins
were being thrown at every molecular chaperone and heat shock protein
known to man. Gone were the messy experiments with organelles and other
variants of chicken soup.
In the ensuing 10 years much progress has been made unraveling
the mechanistic and structural details of the chaperonin nano-machine (for reviews, see 16 and 14). Paradoxically, however, the problem that
Tony Gatenby and I set out to investigate 12 years ago, how to refold
hexadecameric, higher plant Rubisco
S4 · (L2)4 · S4 from its denatured subunits, remains unsolved. To date, despite several
forlorn attempts both with or without chaperonins, not so much as a
whiff of Rubisco activity has been reconstituted. This only goes to
show that the grand old protein of plant biology still has a few
secrets to divulge.
TOP
INTRODUCTION
GENETICS OF BACTERIOPHAGE...