Single-molecule Study on the Decay Process of the Football-shaped GroEL-GroES Complex Using Zero-mode Waveguides*

It has been widely believed that an asymmetric GroEL-GroES complex (termed the bullet-shaped complex) is formed solely throughout the chaperonin reaction cycle, whereas we have recently revealed that a symmetric GroEL-(GroES)2 complex (the football-shaped complex) can form in the presence of denatured proteins. However, the dynamics of the GroEL-GroES interaction, including the football-shaped complex, is unclear. We investigated the decay process of the football-shaped complex at a single-molecule level. Because submicromolar concentrations of fluorescent GroES are required in solution to form saturated amounts of the football-shaped complex, single-molecule fluorescence imaging was carried out using zero-mode waveguides. The single-molecule study revealed two insights into the GroEL-GroES reaction. First, the first GroES to interact with GroEL does not always dissociate from the football-shaped complex prior to the dissociation of a second GroES. Second, there are two cycles, the “football cycle ” and the “bullet cycle,” in the chaperonin reaction, and the lifetimes of the football-shaped and the bullet-shaped complexes were determined to be 3–5 s and about 6 s, respectively. These findings shed new light on the molecular mechanism of protein folding mediated by the GroEL-GroES chaperonin system.

An expression plasmid of tandem-fused GroES heptamer containing ES98C subunit at the C terminus (t-ES 98C ) was generated as previously described (15,16). ES98C is a GroES mutant with a cysteine residue added at the C terminus of each subunit (17). GroEL and GroES were expressed in E. coli and purified as previously described (18).
Fluorescence Labeling of GroEL and GroES-ES98C was labeled with Alexa Fluor 488 C 5 maleimide and maleimide PEO 2 -biotin at a ratio of 1:1:1 in HKM buffer (25 mM HEPES-KOH, pH 7.4, 100 mM KCl, and 5 mM MgCl 2 ). t-ES 98C was labeled with Cy3 maleimide in HKM buffer. Wild-type GroEL was labeled with Cy5 monofunctional NHS-ester in HKM buffer containing 20 mM sodium bicarbonate to raise the pH (ϳ8.5). Fluorescently labeled proteins were purified using NAP5 columns (GE Healthcare). The concentrations of the fluorescently labeled GroES mutants were determined using the Bradford method (Protein Assay; Bio-Rad). The concentration of Cy5-labeled GroEL (Cy5-EL) was determined by correcting for the 280 nm absorbance of the conjugated dye using the molar extinction coefficient of 130,480 M Ϫ1 cm Ϫ1 . The concentrations of GroEL and GroES were expressed as the molar concentrations of the tetradecamer and heptamer. The concentrations of the fluorescent dyes were determined using the following molar extinction coefficients: Alexa Fluor 488, 72,000 M Ϫ1 cm Ϫ1 at 494 nm; Cy3, 150,000 M Ϫ1 cm Ϫ1 at 552 nm; Cy5, 250,000 M Ϫ1 cm Ϫ1 at 649 nm. The molar ratios of Alexa Fluor 488 to the ES98C, Cy3 to t-ES 98C , and Cy5 to wild-type GroEL were 0.86, 1.0, and 1.4, respectively. Fluorescently labeled GroEL and GroES exhibited behaviors similar to those of the wild-type proteins (supplemental Fig. S1).
Sample Preparation for Microscopy-ZMWs were fabricated as previously described (19). To immobilize Alexa488-labeled ES98C (A488bio-ES) selectively at the bottom of ZMWs, polyvinylphosphonic acid was deposited onto an aluminum film as previously described (20). A flow cell was constructed from a glass slide and coverslip with ZMWs separated by two spaces of ϳ50-m thickness (Lumirror 50-S10; Toray Industries, Tokyo, Japan). A488bio-ES was immobilized on to the bottom of the ZMWs through biotinylated bovine serum albumin and streptavidin as previously described (21). Experimental results shown in this paper were obtained in HKM buffer containing 50 nM Cy5-EL, 300 nM Cy3-labeled t-ES 98C (Cy3-ES), 2 mM ATP, 10 M reduced LA (rLA), an ATP-regeneration system (5 mM phosphoenolpyruvate and 10 g/ml pyruvate kinase), an oxygen scavenging system (0.45% (w/v) D-glucose, 50 units/ml glucose oxidase, and 50 units/ml catalase), 7 mM dithiothreitol, and 2 mM Trolox. rLA (50 M) was prepared by incubating LA with 10 mM dithiothreitol for Ͼ10 min at room temperature prior to experiments. rLA is a favorable substrate protein for this experiment because rLA exists in a denatured state irrespective of the presence of GroEL (22). The solution contained an ATP regeneration system to keep the ATP concentration constant. This is because formation of the football-shaped complex is inhibited by ADP (14). Trolox was added to improve photophysical properties of Cy5 in solution (23). Because 300 nM Cy3-ES was required to form saturating amounts of the football-shaped complex (supplemental Fig. S2), experiments were performed in the presence of 300 nM Cy3-ES.
The recorded images were analyzed using a homemade program on a Halcon image processor (MVTec Software GmbH, München, Germany) to obtain the time courses of fluorescence intensities of Cy3 and Cy5 signals. To characterize the formation processes of the football-shaped complexes, we picked up the events that signals from Cy3-ES were detected while Cy5-EL molecules stayed on the surface of ZMWs. The duration of GroEL-GroES binding was determined by marking the association and dissociation events of Cy3-ES and Cy5-EL molecules. To determine decay rate constants, histograms of lifetimes of the football-shaped and bullet-shaped complexes were fitted by the following equation that makes correction of photobleaching of Cy3 and Cy5: where N(t) is the number of the football-shaped or bulletshaped complexes at time t, t is the duration of the footballshaped or bullet-shaped complexes, and k is the decay rate constant, respectively. k b means the rate constant for photobleaching of Cy3 and Cy5 (0.034 and 0.019 s Ϫ1 ). Data fitting was carried out using the Kaleidagraph program (Synergy Software).

RESULTS
We employed a single-molecule assay to probe the GroEL-GroES interaction, including the football-shaped complex. To detect the football-shaped complexes at a single-molecule level, we designed the following experimental system (Fig. 1). A cysteine-introduced GroES variant (ES98C) (17) modified with Alexa Fluor 488 and biotin (termed A488bio-ES) was immobilized in ZMWs through a biotin-streptavidin linkage. We have previously confirmed that the immobilization of GroES does not perturb the binding to GroEL (21). Wild-type GroEL was fluorescently labeled with Cy5 (termed Cy5-EL). Additionally, a Cy3 fluorophore was specifically attached to the ES98C subunit of the tandem-fused GroES heptamer (15, 16) (termed Cy3-ES) in a 1:1 ratio. Surface-immobilized A488bio-ES was immersed in a solution containing 50 nM Cy5-EL, 300 nM Cy3-ES, 2 mM ATP, and excess amounts of LA, in which the disulfide bonds were reduced (i.e. rLA). 300 nM Cy3-ES was required to obtain saturating amounts of the football-shaped complexes (supplemental Fig. S2). In such conditions, conventional singlemolecule imaging by total internal reflection fluorescent microscopy was not suitable because fluorescence from a single molecule could not be detected due to the high background signal from the fluorescent molecules in the solution. To overcome this issue, we adapted a single-molecule assay with ZMWs. ZMWs comprise nanoscale holes in an aluminum film deposited on a fused silica coverslip and can reduce the observation volume by more than 3 orders of magnitude relative to total internal reflection fluorescence microscopy (24,25).
ZMWs allow single-molecule observations at micromolar concentrations of fluorescent molecules in solution (19,24,25). The ZMWs used in this study were ϳ80 nm in diameter and ϳ160 nm in depth (Fig. 1). The aluminum surface was coated with polyvinylphosphonic acid to prevent nonspecific protein adsorption (20).
The positions of single molecules of surface-immobilized A488bio-ES were determined by the fluorescence of Alexa Fluor 488. Freely diffusing Cy5-EL and Cy3-ES were observed to bind to and dissociate from A488bio-ES. Each binding event caused an increase in the fluorescence signal because Cy5-EL and Cy3-ES in solution cannot be observed due to rapid Brownian motion (26). As expected, fluorescences of Cy5-EL and Cy3-ES were detected simultaneously at the position of the immobilized A488bio-ES, indicating that the football-shaped complexes are formed in ZMWs. The repeated appearance and disappearance of Cy3-ES fluorescence were often observed, whereas Cy5-EL fluorescence continued over time (data not shown). The events imply that multiple rounds of association and dissociation of Cy3-ES to and from the bullet-shaped complex formed with Cy5-EL and A488bio-ES located on the surface. In contrast, the simultaneous appearance of Cy5-EL and Cy3-ES fluorescence was hardly observed without immobilization of A488bio-ES in ZMWs (data not shown).
According to the time course of fluorescence intensities of Cy5-EL and Cy3-ES, the formation process of the footballshaped complex could be classified into two types. In Type I, the bullet-shaped complex between Cy5-EL and Cy3-ES attached to A488bio-ES on the glass surface to form the football-shaped complex (Fig. 2, A and B). In Type II, Cy5-EL attached to A488bio-ES on the glass surface followed by attachment of Cy3-ES to form the football-shaped complex (Fig. 2C). Type I was classified further into subtypes, Type Ia and Type Ib, according to the decay process of the football-shaped complex. In Type Ia, the football-shaped complex dissociated to A488bio-ES and the bullet-shaped complex between Cy5-EL and Cy3-ES ( Fig. 2A). In Type Ib, the football-shaped complex dissociated to Cy3-ES and the bullet-shaped complex composed of Cy5-EL and A488bio-ES (Fig. 2B). The percentage of each type was determined as follows: Type Ia, 30%; Type Ib, 48%; Type II, 22% (n ϭ 926, Fig. 2D).
We analyzed the decay process of the football-shaped complex to the bullet-shaped complex. The first GroES molecule to bind was sometimes the first GroES molecule to dissociate from the football-shaped complex (Type Ib), and the second GroES molecule to bind was sometimes the first GroES molecule to dissociate from the football-shaped complex (Type Ia). This observation indicates that the first GroES molecule to bind does not always dissociate from the football-shaped complex prior to the second GroES dissociation event. In other words, the dissociation of GroES molecules from the football-shaped complex can occur in a random order.
We also found that GroEL exited in three different states, that is, GroEL alone, the bullet-shaped complex, and the football-shaped complex. This finding indicated the existence of two reaction cycles in the GroEL-GroES interaction: bullet cycle (GroEL 7 bullet-shaped complex) and football cycle (bullet-shaped complex 7 football-shaped complex) (Fig. 3C). We The ZMWs used in this study were 80 nm in diameter. ZMWs were fabricated in a 100-nm-thick aluminum layer deposited on a quartz coverslip. The quartz at the entrance of the waveguides was etched (60 nm in depth) to improve the signal-to-noise ratio (19). The coverslip was illuminated from the bottom by a microscope in epi-illumination mode. The fluorophores were only excited and detected near the bottom of the ZMW. A488bio-ES was immobilized on the glass surface of the ZMWs via biotinylated bovine serum albumin (BSA), and streptavidin. The flow cell was filled with Cy5-EL, Cy3-ES, ATP, rLA, an ATP regeneration system, and oxygen scavenger enzymes plus reducing agents. Association and dissociation of Cy5-EL and Cy3-ES to and from A488bio-ES were simultaneously visualized. Experiment details are described under "Experimental Procedures." then examined the lifetimes of the football-shaped and the bullet-shaped complexes (Fig. 3). Time courses of fluorescence intensities of Cy5-EL and Cy3-ES under the Type Ia and Ib interactions were analyzed (Fig. 2, A and B). The type II inter-action was excluded from the analysis because in this interaction it was difficult to obtain adequate amounts of data for statistical purposes. The lifetime of the football-shaped complex was measured as the dwell time of both Cy5-EL and Cy3-ES on the surface (Fig. 3, A and B). The histograms of the lifetime of the football-shaped complexes in Type Ia and Ib interactions fitted well to single-exponential decay curves (with the coefficients of determination (R 2 ) of 0.943 and 0.979, respectively). The decay rate constants obtained were as follows: 0.21 Ϯ 0.024 s Ϫ1 for Type Ia interactions and 0.34 Ϯ 0.027 s Ϫ1 for Type Ib interactions (values are reported with the errors of the fits) (Fig.  3, A and B). The average lifetimes of the football-shaped complex determined from Type Ia and Ib interactions, the reciprocal of the decay rate constants, were ϳ4.8 s and ϳ2.9 s, respectively. The lifetime of the bullet-shaped complex was measured as the dwell time of Cy5-EL on the surface following the Cy3-ES dissociation in the Type Ib interaction (Fig. 3B). The histogram of the lifetime of the bullet-shaped complex was found to fit to a single-exponential decay (R 2 ϭ 0.950) with a rate constant of 0.18 Ϯ 0.010 s Ϫ1 (the value is reported with the fitting error) (Fig. 3B). The average lifetime of the bullet-shaped complex was estimated to be ϳ5.6 s.

DISCUSSION
We developed a single-molecule assay using ZMWs to probe the GroEL-GroES interaction in the presence of submicromolar concentrations of fluorescently labeled GroES. The assay allowed the direct observation and characterization of the football-shaped and the bullet-shaped GroEL-GroES complexes in real time. The decay process of the football-shaped complex to the bullet-shaped complex was analyzed. Here, the initially bound GroES does not always dissociate from the footballshaped complex prior to the dissociation of the second GroES molecule, i.e. dissociation of GroES molecules from the football-shaped complex can occur in a random order. The football-shaped complex appears to be formed when both rings of GroEL are occupied with ATP (11,14). Given that the ␥-phosphate of ATP stabilizes the GroEL-GroES complex (27,28), GroES can dissociate from the GroEL ring in which ATP hydrolysis occurs. The average lifetime of the football complex determined by the Type Ia interaction (ϳ4.8 s) was longer than that determined for the Type Ib interaction (ϳ2.9 s). The difference may be due to the GroEL ring with the initially bound GroES showing higher ATPase activity than the GroEL ring with a second GroES molecule bound.
We have also found the existence of two cycles in the GroEL-GroES interaction: bullet cycle and football cycle (Fig. 3C). GroEL can use both the bullet cycle and the football cycle, and the choice of cycle is situation-dependent. The bullet-shaped complex is considered to work as a platform for two cycles. This study was performed in the presence of excess amounts of denatured proteins. In this situation, the lifetime of the football-shaped complex, that is, the transition time from the football-shaped complex to the bullet-shaped complex, was estimated to be 3-5 s, whereas the lifetime of the bullet-shaped complex, the transition time from the bullet-shaped complex to GroEL, was ϳ6 s (Fig. 3C).
It is now clear that the functional cycle of GroEL is too complicated to be fully understood by the use of conventional ensemble techniques. Single-molecule assays will play increasingly important and indispensable roles in the study of the molecular mechanism of the chaperonin reaction cycle.