(Received for publication, August 30, 1996, and in revised form, October 25, 1996)
From the Department of Biochemistry, University of
Texas Health Science Center at San Antonio, San Antonio, Texas
78284-7760 and § Genentech Inc.,
South San Francisco, California 94080
The molecular chaperone cpn60 binds many unfolded
proteins and facilitates their proper folding. Synthetic peptides have
been used to probe the question of how cpn60 might recognize such a diverse set of unfolded proteins. Three hybrid peptides were
synthesized encompassing portions of the bee venom peptide, apamin, and
the sequence KWLAESVRAGK from an amphipathic helix in the
NH2-terminal region of bovine rhodanese. Two disulfides
connecting cysteine residues hold the peptides in stable helical
conformations with unobstructed faces oriented away from the
disulfides. Peptides were designed to present either a hydrophobic or
hydrophilic face of the amphipathic helix that is similar to the one
near the amino terminus of rhodanese. Aggregation of these peptides was
detected by measuring 1,1-bis(4-anilino)napthalene-5,5
-disulfonic
acid (bisANS) fluorescence at increasing peptide concentrations, and aggregation was not apparent below 2 µM. Thus, all
experiments with the peptides were performed at a concentration of 1 µM. Reducing agents cause these helical peptides to form
random coils. Fluorescence anisotropy measurements of
fluorescein-labeled peptide with the exposed hydrophobic face yielded a
Kd = ~106 µM for binding to cpn60,
whereas there was no detectable binding of the reduced form. The
peptide with the exposed hydrophilic face did not bind to cpn60 in
either the oxidized or reduced states. Fluorescence experiments
utilizing bisANS as a probe showed that binding of the helical
hydrophobic peptide could induce the exposure of hydrophobic surfaces
on cpn60, whereas the same peptide in its random coil form had no
effect. Thus, binding to cpn60 is favored by a secondary structure that
organizes and exposes a hydrophobic surface, a feature found in
amphipathic helices. Further, the binding of a hydrophobic surface to
cpn60 can induce further exposure of complementary surfaces on cpn60
complexes, thus amplifying interactions available for target
proteins.
It has been accepted for some time that the amino acid sequence of a protein contains all the information to specify its tertiary conformation (1). However, in vitro refolding is often inefficient due to competing kinetic pathways that lead to aggregation or the formation of non-native conformations (2). It is believed that aggregation is a result of the association of hydrophobic surfaces that are normally hidden in the core of a folded protein but are exposed in partially folded states. The yields of properly folded protein in these in vitro experiments can be improved by carefully selecting conditions such as lower temperature (3), lower protein concentration (4), or the addition of "nondenaturing" detergents (5), all of which favor folding. It is of interest to consider how aggregation can be avoided in vivo, where temperatures remain relatively constant and high local concentrations of interactive nascent polypeptides are often present.
Molecular chaperones are a group of cellular proteins known to be mediators of in vivo folding. Molecular chaperone proteins are mainly found in one of three highly conserved groups, Hsp60, Hsp70, and Hsp90, ranging from prokaryotes to mammals. Some molecular chaperones are expressed during heat shock, whereas others are expressed constitutively and are necessary for cell survival at nonstressed physiological conditions (6). A characteristic property of molecular chaperones is that they can bind non-native forms of proteins and release them in a controlled manner.
Escherichia coli Cpn60 (GroEL) is homologous to the eukaryotic mitochondrial protein Hsp60, and it is one of the best studied molecular chaperones. Cpn60 is composed of fourteen 57-kDa monomers assembled into two stacked seven-subunit rings (7). Cpn60 assists in vitro refolding of a number of proteins, e.g. rhodanese (8), citrate synthase (9), and lactate dehydrogenase (10). Unfolded rhodanese forms a binary complex with Cpn60 (8) that requires the addition of the co-chaperonin cpn10 (groES), MgATP, and K+ for the most efficient release of folded, active rhodanese (11).
The detailed mechanism for the binding and release of target proteins by cpn60 is incompletely understood. The fact that cpn60 can bind a wide variety of non-native polypeptides suggests that the molecular recognition is not specific for the sequence of the target. The 2.8 Å crystal structure and mutational analysis of cpn60 suggest that the polypeptide binding sites contain essential hydrophobic residues located in the apical domains of each cpn60 monomer (12, 13). The common feature shared by the proteins that bind cpn60 has been suggested to be the presence of hydrophobic surfaces on folding intermediates. It was proposed by Martin et al., for example, that dihydrofolate reductase and rhodanese are bound to cpn60 as "molten globule" intermediates (14).
It has been proposed that the secondary structure of regions of target
proteins may be important for binding to cpn60 (15). For example,
transferred nuclear Overhauser enhancement NMR experiments were used to
study the binding of a peptide based on the NH2-terminal sequence of rhodanese. The binding was extremely weak
(Kd = ~1000 µM), but the peptide,
which was a random coil in solution, appeared when bound to adopt a
helical conformation. It was conjectured that the sequence that could
form an amphipathic helix might bind along the hydrophobic face. A
second example, the capture of an all -sheet Fab fragment by cpn60
(16), indicates that the specific secondary structure is not the
determinant for the recognition by cpn60.
To investigate the structural and chemical nature of cpn60-substrate
interactions, we chose to study constrained peptides. Apamin is a
peptide component of bee venom that contains 18 residues with two
disulfide bonds bridging Cys1 to Cys11 and
Cys3 to Cys15. These disulfides lock the
COOH-terminal half of apamin (residues 9-16) into a helical
conformation (18, 35). Various sequences can be substituted into the
COOH terminus of apamin and forced into a helical conformation by the
presence of these disulfide bridges (19).1
The apamin framework can be used to present either the hydrophobic or
hydrophilic face of an amphipathic helix (17), and thus it serves as a
useful tool for examining the structural requirements for binding to
cpn60. Hybrid apamin peptides were constructed containing the
NH2-terminal sequence of rhodanese (KWLAESVRAGK), which has
been shown to bind in a helical conformation to cpn60. These peptides
were then tested for their ability to bind cpn60. The peptides have the
following sequences:
C1NC3K(FITC)APETALC11WLAC15SVRAGK-NH2
(APA09),
C1NC3KAPETALC11WLAC15SVRAGK-NH2 (APA14), and
C1NC3K(FITC)APETKWC11AESC15RAGK-NH2
(APA08), with bold letters representing the rhodanese amino
acids.2 The acronyms APA09 and APA14 are
used in describing the hydrophobic peptides, and APA08 is used in
describing the hydrophilic peptide. APA09 and APA08 were both
synthesized with a FITC-labeled3 lysine at
residue 4. Using these peptides, we demonstrate that an amphipathic
-helix with hydrophobic residues on its exposed helical face can
bind to cpn60, and this binding can lead to exposure of hydrophobic
surfaces buried within cpn60.
All reagents were analytical grade.
1,1-Bis(4-anilino)napthalene-5,5
-disulfonic acid (bisANS) was
obtained from Molecular Probes (Junction City, OR). The chaperonin,
cpn60, was purified from lysates of cells containing the multicopy
plasmid pGroESL (20). Following purification, cpn60 was dialyzed
against 50 mM Tris-HCl, pH 7.5, containing 1 mM
DTT. Then glycerol was added to 10% (v/v), and aliquots were frozen in
liquid nitrogen and stored at
80 °C. The monomer concentration of
cpn60 was measured at 280 nm with an extinction coefficient of
1.22 × 107 M
1
cm
1 as determined by quantitative amino acid analysis,
assuming a molar mass of 60 kDa (21).
The peptides were synthesized on
para-methyl benzhydrylamine resin utilizing t-butoxycarbonyl
chemistry with rapid manual cycles as described previously (22). The
peptides were prepared with t-butoxycarbonyl-Lys ( Fmoc)
in the lysine 4 position. The Fmoc was removed with 20% piperidine in
DMA after stepwise synthesis and prior to hydrogen fluoride cleavage.
Fluorescein was incorporated into APA08 and APA09 by the addition of
4-fold excess FITC and diisopropylethylamine in DMA. The final
t-butoxycarbonyl group was removed, the peptides were
cleaved from the resin, and the side chain protecting groups were
removed by treatment with anhydrous liquid hydrogen fluoride at 0 °C
for 1 h (10 ml of hydrogen fluoride, 1.0 ml of anisole, 0.2 ml of
ethyl methylsulfide, 0.1 ml of thioanisole, 0.1 g of
para-thiocresol/g of resin). The resultant peptide/resin mixture was
washed with diethyl ether and extracted with 20% acetic acid. The
crude linear peptide was lyophilized and purified by C18 reverse phase
HPLC, eluting with a 23-38% acetonitrile gradient. The purified
linear peptide was cyclized by dissolving it in water at a
concentration of 0.1 mg/ml, adjusting the pH to 8.0-8.5 with ammonium
hydroxide and stirring at 20 °C for 3 days. The pH was then adjusted
to 3.0 with acetic acid, and the solution was either lyophilized or
maintained as the 0.1 mg/ml acidified solution and submitted for HPLC
purification. The identity of each peptide was characterized by amino
acid sequencing and mass spectroscopy.
The hydrophilic peptide APA08 has the sequence CNCK(FITC)APETKWCAESCRAGK-NH2. The hydrophobic peptides APA09 and APA14 have the sequences CNCK(FITC)APETALCWLACSVRAGK-NH2 and CNCKAPETALCWLACSVRAGK-NH2, respectively. The bold characters in the above sequences represent the NH2-terminal portions of bovine rhodanese inserted into the apamin peptide. Disulfide bonds form between cysteine residues 1 and 11 and between cysteine residues 3 and 15 in the oxidized form of each peptide. The cyclized (oxidized) peptides are designated as APA08c, APA09c, and APA14c, with the lowercase c denoting the peptides in the cyclized form. The linear (reduced) peptides are designated as APA08r, APA09r and APA14r, with the lowercase r denoting the peptides in the reduced form.
Circular Dichroism MeasurementsEach peptide was
reconstituted in 20 mM sodium acetate, pH 5.5, and placed
in a 0.05-cm pathlength quartz cell. All samples were filtered. The
final concentrations were determined by absorbance measurements. The
concentrations of each peptide were as follows: 0.165 mg/ml for APA08
cyclized, 0.020 and 0.031 mg/ml for APA09 linear and cyclized,
respectively, and 0.30 mg/ml for APA14 linear and cyclized. The samples
were analyzed with an Aviv Circular Dichroism Spectrometer (model
DS/60, Aviv Associates, New York, NY) using a constant bandwidth of 1.0 nm. Spectra were taken from 190 to 250 nm at 0.2-nm intervals with an
averaging time of 3.0 s/data point. All spectra were acquired in
triplicate and averaged. The spectrum of an appropriate buffer control
sample was then subtracted from each of the sample spectra. The final
spectral data were converted to mean residue weight ellipticities for
determination of secondary structure content (mean residue weight
factors: APA08, 129.9; APA09, 124.1; and APA14, 105.6). The mean
residue weight ellipticities were used to calculate the relative
contribution of secondary structural components (-helix and
-sheet) to the overall spectra. The spectral deconvolution method
utilized for these calculations has been described previously (23).
The peptide APA14 (2.0 mg) was dissolved in 440 µl of
95% H2O/5% D2O, and the pH was adjusted to
4.6 by microliter additions of 1 M NaOH; 50 µl of
CD3CN (final concentration, 1.9 M) was added to
improve solubility. All spectra were acquired at 25 °C on a Bruker
AMX-500 spectrometer. Two-dimensional phase sensitive correlation spectroscopy (24), NOE spectroscopy (25-27), jump-return NOE spectroscopy (28), and total correlation spectroscopy (29, 30) spectra
were acquired with low power coherent irradiation of the water
resonance for 1.5 s prior to the pulse sequence and during the NOE
spectroscopy mixing time. Total correlation spectroscopy mixing was
achieved with a clean DIPSI-2rc sequence applied for 90 ms (31), and
mixing times of 300 ms were used in both NOE spectroscopy experiments.
The spectra were processed and analyzed using Felix (Biosym
Technologies, San Diego, CA), and assigned using established techniques
(32). 3JHN-H were obtained by line fitting
2 slices from the correlation spectroscopy spectrum
processed with very high digital resolution. A second set of NMR
spectra were acquired after the addition of 10 µl of 2 M
DTT to allow the uncyclized (reduced) peptide to be studied. A rotating
frame Overhauser spectroscopy spectrum (33) of the uncyclized peptide
failed to identify any additional dipolar coupling interaction.
Fluorescence of bisANS was measured using an excitation wavelength of 397 nm and an emission wavelength of 500 nm. Anisotropy measurements on APA08 and APA09 peptides, each of which had Lys4 labeled with fluorescein, were performed with an excitation wavelength of 492 nm and emission wavelength at 519 nm. Fluorescence studies were carried out on a SLM model 500C fluorometer with an excitation slit width of 2.5 nm and emission slit width of 10 nm.
Peptide Binding MeasurementsBinding experiments were carried out on both APA08 and APA09 peptides in the oxidized and reduced states as follows. To a cuvette containing 50 mM Tris-HCl, pH 7.8, a given peptide was added to a concentration of 1 µM, and the anisotropy values were recorded at 20 °C. To this solution, cpn60 that had been dialyzed against 50 mM Tris-HCl, pH 7.8, was added to the cuvette to a final concentration of 10 µM, and the anisotropy value was measured.
A dissociation constant was estimated for oxidized APA09, which was the only fluorescently labeled peptide that displayed detectable binding to cpn60. The approach used was to titrate a 1 µM solution of the oxidized APA09 peptide with increasing amounts of cpn60. Titrations using increasing peptide concentrations were not possible because of strong self-association, as expected with amphipathic structures and demonstrated with these peptides (see "Results"). The dissociation constant was estimated as follows:
![]() |
(Eq. 1) |
![]() |
(Eq. 2) |
In Equation 2, robs is the measured anisotropy at a given protein concentration, fB and fF are the fractions of fluorophore bound and free, respectively, and rB and rF are the anisotropy values when the fluorophore is all bound or all free, respectively. Because rfree is small, Equation 2 can be simplified to:
![]() |
(Eq. 3) |
![]() |
(Eq. 4) |
![]() |
(Eq. 5) |
![]() |
(Eq. 6) |
![]() |
(Eq. 7) |
![]() |
(Eq. 8) |
To follow hydrophobic exposure on cpn60, 10 µM bisANS was added to a cuvette containing 1 µM cpn60, and the emission spectrum was recorded. Subsequently, 1 µM oxidized APA14 was added to the cuvette, and the bisANS emission was recorded. BisANS fluorescence was insignificant in the presence of 1 µM oxidized APA14 alone (data not shown). The increase in bisANS fluorescence in the presence of cpn60 was attributed to hydrophobic exposure induced on the chaperonin by the oxidized APA14 peptide. To follow loss of hydrophobic sites following reduction of the peptide, 2 mM DTT was added to the samples above. Emission spectra were collected from 0 to 40 min.
The CD
spectra of the oxidized peptides, APA09c (Figs.
1A), APA14c (Fig. 1B), and APA08c
(Fig. 1C), indicate a high degree of helical secondary
structure, whereas the spectra of reduced peptides, APA09r (Fig.
1A) and APA14r (Fig. 1B), indicate there is
little or no regular helical structure. The secondary structure calculated for each peptide in either the oxidized or reduced form is
shown in Table I. The fluorescein-labeled peptides APA08 and APA09 were not soluble at concentrations required for NMR. However,
analysis of oxidized APA14c by 1H NMR spectroscopy
identified intense HN-HN sequential NOEs and
medium range NOEs characteristic of a helix between residues 9 and 16 (Fig. 2); the presence of a helical conformation is
confirmed by the small values of 3JHN-H for
these residues. The data for residues 17-19 are not indicative of a
well formed helix, suggesting that they form a frayed terminus to the
helix or adopt a turn conformation. These findings are in line with the
structure modeled for APA14 and are consistent with the
three-dimensional structures determined for other apamin peptides
(35).1 Many of the chemical shifts of oxidized APA14c are
found to be very different from random coil values, suggesting that the
peptide does adopt a well defined structure. The addition of
dithiothreitol to the peptide resulted in a loss of the characteristic
helical NOEs, an increase in the 3JHN-H
scalar coupling constants (Fig. 2, lower panel), and movement of the resonances to approximately random coil positions. Helical interactions were also absent from rotating frame Overhauser spectroscopy spectra, indicating that the loss of NOE intensity was not
the result of a change in rotational correlation time. Thus, the
disulfide bonds are required to maintain APA14 in a helical
conformation. The predicted structure of these peptides is shown in
Fig. 3, and the surface hydrophobicity of each peptide is also presented.
|
Low Peptide Concentrations Are Critical to Avoid Aggregation
Oxidized APA09c and APA14c are amphipathic helical
peptides with exposed hydrophobic faces. Thus, the question of
solubility and formation of aggregates was addressed. BisANS
fluorescence was used to detect aggregation of oxidized APA14c (Fig.
4) and reduced APA14r (data not shown). A cuvette
containing 10 µM bisANS was titrated with increasing
amounts of either oxidized APA14c or reduced APA14r. There was
relatively little increase in bisANS fluorescence in the range of
peptide concentrations from 0 to 2.0 µM, but there was a
substantial increase that occurred above 2 µM APA14. As
shown in Fig. 4B, the fluorescence response is reminiscent
of similar titrations that are used to define the critical micelle
concentrations of amphiphilic detergents (36). Thus, significant
self-association of the peptide occurs when its concentration is higher
than 2-3 µM. To avoid artifacts and interpretive
complications from aggregation, all experiments with APA08, APA09 (the
fluorescein-labeled form of APA14), and APA14 were done at a
concentration of 1 µM.
Binding of Oxidized and Reduced APA09 to cpn60 Can Be Measured by Fluorescence Anisotropy
Fluorescence anisotropy binding studies were done using the fluorescein-labeled APA09 peptides (same sequence as APA14). Initial experiments were done to determine any differences in the binding affinity of oxidized APA09c and reduced APA09r.
The anisotropy of 1 µM APA09c free in solution had a
value of 0.061 ± 0.004. The anisotropy of 1 µM
APA09c in the presence of 10 µM cpn60 was 0.084 ± 0.003. On the other hand, there was virtually no change in the
anisotropy of free APA09r (0.055 ± 0.008) as compared with APA09r
in the presence of cpn60 (0.050 ± 0.006). The dissociation
constant for oxidized APA09c was determined as described under
"Experimental Procedures." Cpn60 was titrated into a solution of 1 µM oxidized APA09c over a concentration range of 1-65
µM cpn60. The inverse of the change in anisotropy was plotted as a function of the inverse of the cpn60 monomer
concentration, and the data were fit with a straight line (Fig.
5B). The anisotropy of bound peptide was
determined from the y-intercept of the straight line as explained under
"Experimental Procedures." The concentration of cpn60/APA09c
complex formed was then plotted against the concentration of free cpn60
multiplied by the concentration of free APA09c (Fig. 5). The data were
fit to a straight line, and by taking the inverse of the slope, a
Kd was estimated to be approximately 106 (± 6)
µM. Independent of the assumptions used for the analysis of these data, it is clear that the oxidized peptide shows considerable binding, whereas the reduced version of the same peptide shows no
detectable interaction. The oxidized APA08c peptide was subjected to
the same fluorescence anisotropy experimental conditions, and no change
in anisotropy was detected. Thus, the oxidized APA08c hydrophilic
peptide did not bind to cpn60 as observed by fluorescence anisotropy
measurements.
Oxidized APA14 Induces Hydrophobic Exposure on cpn60
Changes
in bisANS fluorescence were used to assess changes in hydrophobic
exposure in solutions containing cpn60 together with APA14c. When 10 µM bisANS was mixed with 1 µM cpn60, the maximum fluorescence intensity was observed at 502 nm. When 1 µM oxidized APA14c was added, the maximum fluorescence
intensity increased by a factor of 1.7 with no change in the wavelength maximum. Fig. 6 shows the time course of the bisANS
fluorescence intensity after the addition of 2 mM DTT to
the APA14c-cpn60 mixture. The decrease in bisANS fluorescence that
followed the reduction of the peptide could be modeled by a single
exponential with a decay rate constant = 0.065/min (Fig. 6). A
control sample showed that the fluorescence intensity of 10 µM bisANS did not change upon the addition of 2 mM DTT. These results indicate that the complex of cpn60
with APA14c displays considerably more hydrophobic exposure than the
sum of the individual components, and the result requires the presence
of the oxidized and not the reduced peptide.
Cpn60 can bind a wide variety of proteins in non-native states,
implying that the specific primary sequence does not supply the
information required for recognition (9, 11, 14, 37). Because cpn60
does not bind proteins in their native folded state, the recognition
does not require formation of native tertiary structure (14). The
binding of various proteins must occur by a more general mechanism of
molecular recognition. One common feature of partially folded proteins
is the solvent exposure of hydrophobic residues that would be buried in
the native state. As proteins fold, local regions can form secondary
structures such as helices that can align residues to produce
hydrophobic surfaces. These surfaces could provide the necessary
binding motif for recognition by the cpn60 hydrophobic binding
sites.
The oxidized peptides APA09c, APA14c, and APA08c used in this study address the issue of secondary structural requirements for binding to cpn60 and suggest a possible mechanism of cpn60 binding to unfolded proteins via induced hydrophobic exposure on cpn60 as a result of an initial amphipathic peptide binding.
The small size of APA09 compared with that of the cpn60 tetradecamer (798 kDa) allowed the binding of the peptide to cpn60 to be followed by fluorescence anisotropy. Oxidized APA09c was determined to bind weakly to cpn60 with a Kd = ~106 µM, assuming one binding site on cpn60. No binding of reduced APA09r was detected under the same experimental conditions. This result shows the effect of organizing the hydrophobic surface onto a helical motif. Although the binding interactions of these small peptides are weak, they should be viewed as a part of the many weak interactions that would occur with a partially folded protein containing numerous elements of local secondary structure. This implied multiple binding of a protein to the oligomeric cpn60 is supported by the previous observation that cpn60 quaternary structure is stabilized by interactions with partially folded proteins (38). That the hydrophilic peptide, APA08, shows no binding to cpn60 supports the suggestion that hydrophobic interactions are important in forming the cpn60-protein complex. These results are complementary to the data from Fenton et al. (13), where polypeptide binding studies were performed on various mutants of cpn60 to demonstrate the requirement for hydrophobic residues on cpn60 to bind partially folded proteins.
The present studies suggest that initial binding of amphipathic elements to cpn60 can induce the formation of increased hydrophobic exposure. When the interactions with the peptides are eliminated by reduction, the induced hydrophobic surfaces are lost. This may be related to the effects seen with proteins whose exposed, interactive surfaces are buried as they fold with two consequences: (a) they are bound less tightly to cpn60, and (b) the interactive surfaces on cpn60 are hidden and protected against nonproductive interactions.
The 2.8 Å crystal structure of cpn60 (12) as well as mutational analysis (13) have suggested that the polypeptide binding site on cpn60 resides in hydrophobic sequence within the apical domain of each monomer. The crystal structure in this region has poor resolution, presumably due to flexibility or malleability in this area of the apical domain (12). It is tempting to speculate that interactions with amphipathic structural elements could modify the exposure in this part of the structure and interactions with nucleotides or the co-chaperonins could modulate the interactions. Thus, organized amphipathic structures on unfolded proteins may not only be a part of the general recognition motif for cpn60 binding, but they may also serve as a means of amplifying the hydrophobic surface exposed by cpn60. Finally, the observed strong net binding to cpn60 of partially folded proteins may result from multiple weak interactions stemming from the symmetrical association of many subunits. This would allow some flexibility of the bound protein by dissociation at some of the sites without permitting its release from the complex.