The chaperonin GroEL consists of two supramolecular rings that are stacked to form a double-decker architecture (Fig. 1a)12. (Each ring consists of seven protein subunits; each subunit has a molecular mass of 60 kDa.) GroEL possesses a cylindrical cavity with a diameter of 4.5 nm, and a wall thickness of 4.6 nm (ref. 13). The chaperonin cpn originates from the thermophilic bacterium, Thermus thermophilus HB814,15, and consists of a GroEL-like tetradecamer (840 kDa) of a protein subunit called cpn60 (60 kDa), which is hybridized, with a capping protein assembly [cpn10]7 (70 kDa) on either side of its cylindrical cavity (Fig. 1a)16,17. Transmission electron microscopy (TEM) showed that the cavity size and wall thickness of cpn ( 5 nm) are both comparable to those of GroEL.

Figure 1: Chaperonin proteins as ATP-responsive barrels for inclusion of nanoparticles.
figure 1

a, Schematic illustrations (top and side views) of GroEL (crystal structures, right)13 and cpn. b, Schematic representation of the formation of GroEL–CdS nanoparticle complexes by inclusion of CdS nanoparticles into the cylindrical cavity of GroEL, and its ATP-triggered guest release.

We prepared CdS nanoparticles (2–4 nm) according to a method reported in ref. 18. For the complexation with chaperonins, a dimethylformamide (DMF) solution of CdS nanoparticles was added to a Tris–HCl buffer solution of GroEL or cpn. In the presence of the chaperonin proteins, the photoluminescence of CdS nanoparticles at 430–720 nm (excitation wavelength λext = 370 nm) lasted for an unusually long period of time without decay (for example, 400 days with GroEL). In the absence of the chaperonins, the characteristic fluorescence disappeared within 2 hours. The tubular structure of the chaperonin proteins was essential for the stabilization of CdS nanoparticles. Fluorescence was not enhanced by the use of GroES, a bowl-shaped, [cpn10]7-like GroEL-capping protein, in place of GroEL.

We isolated complexes of cpn or GroEL with CdS nanoparticles by size-exclusion chromatography (SEC). By means of inductively coupled plasma mass spectrometry (ICP-MS), we confirmed that the complexes both contained Cd2+. The cpn–nanoparticle complex showed circular dichroism bands, whose intensities were identical to those of intact cpn. An analytical SEC trace of this solution with an ultraviolet (UV)/fluorescence dual detector showed single, sharp elution peaks, which were superimposable with one another at nearly the same elution volume as intact cpn (Fig. 2a). As intact cpn is barely fluorescent, the above results strongly indicate that the CdS nanoparticles co-localized with cpn to form an inclusion complex. The photoluminescence of the complexes at 430–720 nm upon excitation at 370 nm is characteristic of CdS nanoparticles (Supplementary Fig. S1)19,20. We also conducted analytical SEC of the cpn–CdS inclusion complexes using a refractive index (RI)/multi-angle light scattering dual detector, where the molecular mass of the complex was evaluated to be 915 ± 4 kDa, which is nearly identical to that of intact cpn (910 kDa) (Supplementary Fig. S2). These results demonstrate that cpn within the protein–nanoparticle complex preserves its own structural identity, without formation of higher aggregates or dissociation into the protein subunits.

Figure 2: SEC analysis of cpn–CdS the nanoparticle inclusion complex and its specific response to ATP in KCl-containing Tris–HCl buffer.
figure 2

a, Analytical SEC trace of cpn–CdS inclusion complexes monitored by an UV (upper chart)/fluorescence (FL) (lower chart) dual detector, and b, that of cpn–CdS inclusion complexes after addition of ATP and MgCl2. The peak at 12.5–15.0 ml is derived from ATP and its hydrolysed products. c, Changes in fluorescence intensity of cpn–CdS complexes in Tris–HCl buffer: 1, ATP + MgCl2; 2, ATP; 3, MgCl2; 4, ADP + MgCl2; 5, ADP.

We also succeeded in imaging these complexes by TEM. Figure 3a shows a TEM picture of the cpn–CdS inclusion complex, isolated by SEC and stained with uranyl acetate, where the bullet-like side view and the doughnut-like end-on view of cpn protein are clearly observed, as reported previously16,17. The TEM picture of the cpn–CdS inclusion complex in the end-on orientation shows a darker central region when compared to intact cpn (Fig. 3c, Supplementary Fig. S4), owing to the presence of a CdS nanoparticle within the protein cavity (Fig. 3b). No CdS nanoparticles were observed on the exterior of the protein surface. When a single inclusion complex was irradiated for 1 minute by a focused electron beam, the protein was partially destroyed, so that the CdS nanoparticle included within the chaperonin cavity was more clearly imaged by TEM (Supplementary Fig. S3). From the TEM image under such partially destructive conditions, only a single nanoparticle appeared to be present within the chaperonin cavity. As the CdS nanoparticles are positively charged owing to Cd2+ ions on the surface, it is reasonable to conclude that a second nanoparticle will not be included in the chaperonin cavity owing to electrostatic repulsion. For statistical analysis of the TEM picture, we randomly selected 793 images of end-on cpn–CdS nanoparticle complexes, and found that 75% showed a dark central cavity due to the presence of a CdS nanoparticle.

Figure 3: Transmission electron micrographs of cpn–CdS nanoparticle complexes and intact cpn.
figure 3

a, b, cpn–CdS nanoparticle complexes isolated by SEC. c, Intact cpn. The models on the right sides of the images in b and c are schematic representations of the end-view of the complex and intact cpn, respectively.

The CdS nanoparticles within the chaperonin proteins are electronically insulated. The photoexcited singlet state of CdS nanoparticles is known to be quenched by an electron acceptor such as methyl viologen (MV2+) via photoinduced electron transfer from the former to the latter21,22. Fluorescence titration of CdS nanoparticles in DMF with MV2+ at 25 °C gave Stern–Volmer plots (Fig. 4), from which the Stern–Volmer constant was evaluated to be 3.3. In sharp contrast, the fluorescence quenching of the inclusion complex with GroEL took place much less efficiently even at a high concentration of MV2+, where the observed Stern-Volmer constant (0.20) was more than 15 times smaller than that for intact CdS nanoparticles. We also found an even smaller Stern–Volmer constant for the cpn–nanoparticle complex (0.09). These observations are quite reasonable, considering that the thickness of the protein envelope around the CdS nanoparticle (4.6 nm for GroEL13) exceeds an upper limit of the electron-transferable distance (3 nm) in ordinary systems.

Figure 4: Shielding effects of chaperonins on electron transfer from CdS nanoparticles to methyl viologen (MV2+).
figure 4

Stern–Volmer plots for fluorescence titration with MV2+ at 25 °C of a DMF solution of CdS nanoparticle (filled circles), a Tris–HCl buffer solution containing GroEL–CdS nanoparticle inclusion complexes (open squares), and a Tris–HCl buffer solution of cpn–CdS nanoparticle inclusion complexes (open circles). I and I0 represent the intensities of fluorescence and its initial value, respectively.

Chaperonins belong to the family of heat-shock proteins23, and cpn, originating from a thermophilic bacterium, has been reported to be stable against thermal denaturation up to 80 °C (ref. 14). We found that the cpn–CdS nanoparticle complex is also thermally stable, and maintains its characteristic photoluminescence activity up to 80 °C. In heating–cooling cycles in a temperature range of 4–80 °C (Fig. 5a), the photoluminescence of the inclusion complex in Tris–HCl buffer was quenched upon heating, but recovered the original intensity upon cooling down to 4 °C. However, when heated at 90 °C for 10 min, the protein component underwent partial denaturation as observed by circular dichroism, and the complex irreversibly lost 18% of its photoluminescence activity (Fig. 5a, arrow). Although intact CdS nanoparticles in DMF also showed a similar temperature-dependent photoluminescence profile, the response became irreversible upon heating at only 30 °C, forming a colloidal precipitate of CdS. A GroEL–CdS nanoparticle complex showed thermoresponsive photoluminescence activity (Fig. 5b), but became partially irreversible upon heating for 20 min at 60 °C (Fig. 5b, arrow), just below the denaturation temperature of intact GroEL24. Thus, the thermal stability of the included nanoparticle is totally dependent on that of the chaperonin protein.

Figure 5: Thermal stabilities and relative fluorescence intensities of GroEL–CdS and cpn–CdS inclusion complexes.
figure 5

Top, temperature-dependent photoluminescence profiles of a, cpn–CdS and b, GroEL–CdS nanoparticle inclusion complexes in Tris–HCl buffer. Fluorescence intensities at 4 °C were used as the bases for relative fluorescence intensities (bottom). Numbers of plateau in top panels denote annealing temperatures.

Chaperonin proteins, upon binding with ATP in the presence of Mg2+ and K+, undergo a large conformational change, which results in the release of refolded guest proteins from the cavity25,26. We found that cpn–CdS nanoparticle complexes also respond to ATP, and release included nanoparticles from the cavities under similar conditions (Fig. 1b). When a Tris–HCl buffer solution of ATP containing MgCl2 was added to a buffer solution of cpn–CdS nanoparticle complexes containing KCl, the mixture turned slightly cloudy within seconds to give colloidal substances, where the supernatant solution after centrifugation was no longer fluorescent (Fig. 2c). The release of CdS nanoparticles from cpn complexes by the action of ATP was clearly demonstrated by analytical SEC with a UV/fluorescence dual detector (Fig. 2b). After the addition of ATP, the UV response of the SEC trace of the complexes showed a sharp elution peak assignable to cpn and an additional broad peak in the lower-molecular-mass region due to ATP and its hydrolysed products, while no fluorescence responses were observed for these two peaks. The fraction corresponding to cpn, isolated by SEC from the reaction mixture, lost most (nearly 90%) of its Cd2+, as determined by ICP-MS. On the other hand, addition of ATP or MgCl2 alone to the KCl-containing buffer solution of the inclusion complexes neither resulted in the generation of colloidal substances nor fluorescence quenching (Fig. 2c)—in contrast, a DMF solution of intact CdS nanoparticles immediately lost its luminescence activity upon addition of MgCl2.

Thus, cpn makes the included CdS nanoparticle thermally stable, tolerant to electrolytes, and electronically dormant, but can readily release the guest nanoparticle from its cylindrical cavity upon binding with ATP. As expected, the cpn–CdS nanoparticle complexes hardly responded to ADP (adenosine-5′-diphosphate)—there was no substantial change in the fluorescence intensity, regardless of the presence or absence of Mg2+ (Fig. 2c). We also found that GroEL–CdS nanoparticle complexes behave similarly upon binding with ATP.


Sample preparation

To a N2-purged DMF solution of Cd(OAc)2·2H2O (30 ml, 5.56 mM), with vigorous stirring, was added dropwise a MeOH solution of Na2S·9H2O (3 ml, 5.56 mM) at -40 °C. The resulting mixture was stirred for 1.5 h to afford a yellow solution, which showed absorption and emission spectra (λext = 370 nm, emission wavelength λem = 530 nm) characteristic of nanosized CdS particles. TEM analysis showed that the resulting CdS nanoparticles have a size distribution in which the majority were in the diameter range 2–4 nm. To a 1.9-ml Tris–HCl buffer (25 mM, pH 7.5 with 100 mM KCl) solution of cpn (0.5 µM) was added dropwise a 100-µl DMF solution of CdS nanoparticles (5.56 mM based on Cd2+) at 4 °C. The mixture was allowed to stand at 4 °C until the initial spectral change subsided (100 h), and then subjected to SEC using Tris–HCl buffer (25 mM, pH 7.5 with 100 mM KCl) as eluent on Sephacryl S-300 HR or S-400 HR with an ÄKTA-prime low-pressure chromatography system (Amersham Pharmacia Biotech). SEC traces at 280 nm allowed the isolation of a protein-containing fraction, which was concentrated by ultrafiltration with an USY-5 ultrafilter unit (Advantec) at 4 °C, affording 2 ml of a Tris–HCl buffer solution of the cpn–CdS nanoparticle complex (0.1 µM). Preparation and isolation of GroEL–CdS nanoparticle inclusion complexes were performed in a similar fashion.


Analytical SEC traces (TSK gel G4000SWXL (Tosoh); eluent, Tris–HCl buffer 25 mM, pH 7.5 with 100 mM KCl) of cpn–CdS nanoparticle inclusion complexes were followed by UV (observed at 280 nm) and fluorescence (excitation at 370 nm, observed at 530 nm) responses by UV970 and FP2020-plus spectrophotometers (JASCO), respectively, at a flow rate of 0.5 ml min-1. Analytical SEC (Tosoh TSK gel G4000SWXL or Shodex Protein KW-804) of the complexes was performed using an UV/fluorescence dual detector and an RI/multi-angle light scattering dual detector, at a flow rate of 0.5 ml min-1.

TEM imaging

Samples were applied to an electron microscope specimen grid covered with a thin carbon support film that had been hydrophilized by ion bombardment. After drying with a pre-water-soaked filter paper, the grid was negatively stained with 1% uranyl acetate. TEM micrographs were then recorded on a JEM-2010 TEM (JEOL) operating at an anode voltage of 120 kV.

Methyl viologen titration

To a DMF solution of CdS nanoparticles (2 ml, 700 nM based on Cd2+) were added 10-µl aliquots of a 100-µM Tris–HCl buffer (25 mM, pH 7.5 with 100 mM KCl) solution of MV2+. Likewise, Tris–HCl buffer (25 mM, pH 7.5 with 100 mM KCl) solutions of GroEL–CdS complex and cpn–CdS complex (2 ml, 700 nm based on Cd2+) were titrated. Emission (λext = 370 nm) spectra were recorded on a FP-777W spectrophotometer (JASCO).

Thermal stability

Fluorescence spectra (λext = 370 nm, wavelength of observed fluorescence λobsd = 530 nm) of GroEL–CdS complexes and cpn–CdS complexes were recorded at designated temperatures on a FP-777W spectrophotometer (JASCO), where the fluorescence intensities at 4 °C were used as the bases for relative fluorescence intensities. The temperature was directly controlled by a ECT271 Peltier thermometric apparatus (JASCO; 40 °C min-1 on heating and 25 °C min-1 on cooling).

ATP response

To a 2-ml Tris–HCl buffer (25 mM, pH 7.5 with 100 mM KCl) solution of cpn–CdS complexes (0.5 µM based on cpn) were added aqueous solutions of ATP (100 mM) and MgCl2 (1 M) ([ATP] = 20 µM, [Mg2+] = 25 mM after mixing), and the mixture was incubated at 70 °C for 10 min. The supernatant solution was subjected to fluorescence spectroscopy and analytical SEC with an UV/fluorescence dual detector.