Reduced Stability and Enhanced Surface Hydrophobicity Drive the Binding of Apo-Aconitase with Groel During Chaperone Assisted Refolding

The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692

Contents lists available at ScienceDirect The International Journal of Biochemistry & Cell Biology journal homepage: www.elsevier.com/locate/biocel

Reduced stability and enhanced surface hydrophobicity drive the binding of apo-aconitase with GroEL during chaperone assisted refolding

Parul Gupta a,1, Saroj Mishra a, Tapan Kumar Chaudhuri a,b,∗ a Department of Biochemical Engineering and Biotechnology, Hauz Khas, New Delhi 110016, India b School of Biological Sciences, Indian Institute of Technology Delhi, Hauz Khas, New Delhi 110016, India article info abstract

Article history: Apo-aconitase, the Fe4S4 cluster free form of TCA cycle enzyme aconitase, binds with GroEL and dis- Received 2 September 2009 sociates itself upon maturation through insertion of the cluster. It is not clearly established as to why Received in revised form apo-protein binds with GroEL. In order to explore the possibility that stability is a factor responsible 21 December 2009 for the aggregation of apo-form at low ionic strengths and hence it associates with GroEL to avoid the Accepted 4 January 2010 unfavorable event, we carried out the unfolding studies with holo- and apo-aconitase. By probing the Available online 9 January 2010 unfolding process through the changes in secondary structural element, exposed surface hydrophobicity, and the microenvironment around tryptophan residues, we were able to establish the relevant changes Keywords: Apo-aconitase associated with the event. Apparent guanidine hydrochloride concentration required for unfolding of 50% Holo-aconitase of aconitase indicates that aconitase is destabilized in the absence of the Fe4S4 cluster. The destabilization Aconitase unfolding of the apo-aconitase was further reflected through its three times higher rate of unfolding as compared GroEL-assisted folding to the holo-protein. It was also observed that the apo-form has higher surface hydrophobicity than the Metallo protein biosynthesis holo-form. Hence, the lower ground state stability and higher solvent exposed hydrophobic surface of the apo-form makes it aggregation prone. Based on the present observation and earlier findings, we propose that binding of apo-aconitase to GroEL not only rescues it from the aggregation, but also assists in the final stage of maturation by orienting the cluster insertion site of GroEL bound apo-protein. This information sheds new light on the potential role of GroEL in the biosynthetic pathway of the metallo proteins. © 2010 Elsevier Ltd. All rights reserved.

1. Introduction around 30% of the proteins in the cell (Frausto da Silva and Williams, 1991), considerable effort has been expended in the last few years Cofactors play an important role in biological function, sta- in addressing the determinants of stability and folding properties bility and folding of various proteins (Wittung-Stafshede, 2002). of proteins, containing metal cofactors. Addressing the stability properties of proteins with bound pros- Aconitase (EC 4.2.1.3) is an 82 kDa protein that serves to cat- thetic groups may pose particular difficulties: (a) if the cofactor alyze the stereospecific conversion of citrate to iso-citrate through is covalently bound, it is likely that the unfolding peptide is able the intermediate cis-aconitate. It contains an iron–sulfur cluster, to refold spontaneously to its native state; in fact, there are sev- and the active enzyme requires the fully assembled [4Fe–4S] clus- eral available studies in which the working models are small ter. The crystal structure (Lauble et al., 1992; Robbins and Stout, monomeric proteins which contain a prosthetic group with a 1989) shows the protein to be composed of four structural domains. covalent linkage. (b) On the other hand, conformational stability The first, second and third domains form a relatively tightly packed studies on proteins containing non-covalently bound cofactors may structure connected to the fourth domain through a potentially be much more difficult to interpret, due to their complex large flexible hinge-linker peptide. This fourth domain, in the crystal multi-domain structures. In recognition of the important role of structure, is closely apposed to domains 1–3, forming a cleft that metal-containing proteins in the living systems, which make up serves as the active site of the enzyme. The iron–sulfur cluster is within the cleft and is intimately associated with amino acid residues that are involved in binding of the substrate and catalyz- ing the citrate–iso-citrate interconversion. Despite several studies Abbreviations: ANS, 1-anilino naphthalene-8-sulfonate; GdnHCl, guanidine on the biophysical and spectroscopic properties of Fe–S proteins hydrochloride; MG, molten globule; N, native; U, unfolded; PK, proteinase K. (Lippard and Berg, 1994), there are few folding and stability studies ∗ Corresponding author. Tel.: +91 11 2659 1012; fax: +91 11 2658 2282. of these proteins. There is only one study in which the thermody- E-mail addresses: [email protected] (P. Gupta), [email protected] namic stability of the apo- and holo-forms of a Fe–S protein has (S. Mishra), [email protected], [email protected] (T.K. Chaudhuri). 1 Present address: Characterization Services, Biologics Development centre, Dr. been directly compared. The study showed that when the cluster Reddy’s Laboratories Ltd., Bachupally, R R District, Andhra Pradesh, India. is removed from adrenodoxin, the transition temperature (Tm) and

1357-2725/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biocel.2010.01.002 684 P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692 the unfolding enthalpy change is considerably reduced; the holo- aconitase can be recruited to GroEL, it is not subject to misfolding ◦ ◦ form has a Tm of 51 C whereas the apo-form has a Tm of 37 C. and aggregation in its absence. In addition, the apo-protein is much more susceptible to protease One approach towards the understanding of cluster stabiliza- digestion than the holo-protein (Burova et al., 1995). tion and its involvement in protein stability in iron sulfur proteins In spite of the accumulation of a large number of experimen- is to investigate the involvement of the intermediate structures or tal studies, protein folding remains one of the most challenging to highlight the role of particular structural elements in cluster sta- subjects in structural biology. Characterization of folding inter- bilization. In order to highlight the structural role that the Fe–S mediates is considered an important strategy for the elucidation cluster is likely to play in holding the various domains of aconitase of the mechanism of protein folding. A common intermediate, core together and to investigate the mechanism of aconitase fold- the “molten globule” (MG) state, has been detected between the ing, the formation of intermediates at different stages of unfolding native (N) and the fully unfolded (U) states for many proteins (Fink, were characterized. The present study describes the measurements 1995; Ptitsyn, 1995). The MG state is characterized by pronounced of the GdnHCl-induced unfolding of holo- and the apo-forms of secondary structure, compact globularity, exposed hydrophobic the yeast mitochondrial aconitase, monitored by intrinsic fluo- surface, and the absence of rigid side-chain packing (Fink, 1995; rescence, far UV CD, and binding with the extrinsic fluorescence Kuwajima, 1989, 1996; Ptitsyn, 1987, 1995). However, most pro- probe, 1-anilino-naphthalene-8-sulfonate (ANS). The kinetics of teins for which a MG intermediate has been well characterized the unfolding process and the effect of Fe–S cluster on the rate are small, monomeric, single domain proteins (Arai and Kuwajima, of aconitase unfolding were further studied by stopped-flow tech- 1996; Kay and Baldwin, 1996; Kim and Baldwin, 1990; Matthews, niques. The role of co-chaperonin GroES during reconstitution of 1993; Ogasahara and Yutani, 1994). Multi-domain, oligomeric pro- holo-aconitase from apo-aconitase was investigated by partial proteins or metallo proteins remain relatively little explored (Jeanicke, teolytic digestion of apo aconitase-GroEL binary complex. 1991). Since the folding/unfolding of such proteins is accompa- nied by the association/dissociation of subunits or incorporation 2. Results of metal cluster and are aggregation prone, the processes are much more complicated than that of monomeric proteins. 2.1. Conformation and secondary structure of recombinant holo- The efficient folding for most proteins has been shown to occur and apo-aconitase inside the cavity of the cis ring of GroEL that houses the unfolded or partially folded protein, following encapsulation or capping of the The recombinant yeast mitochondrial aconitase was expressed cis ring by dome-shaped heptameric GroES in the presence of Mg- in E. coli as soluble, biologically active enzyme. The folded protein ATP (Weissman et al., 1995). The subsequent binding of Mg-ATP to was purified and its enzymatic activity estimated. The metal clus- the unoccupied or the trans ring of GroEL, results in the collapse ter free form was generated by stripping of the iron–sulfur cluster of the cis ring assembly with the concomitant release of GroES and from the holo-form, in order to obtain apo-aconitase. The pep- partially or completely folded protein from the cis cavity. It has been tide CD spectra of the recombinant holo-aconitase was measured generally accepted that the unfolded proteins of ∼57 kDa or smaller under native conditions (0 M GdnHCl at 25 ◦C), and compared with in size can fit inside the cavity encapsulated by GroES, as shown by the CD spectra of the apo-aconitase (Fig. 1A). There was a signif- studies both in vivo (Ewalt et al., 1997; Houry et al., 1991) and in icant difference in the CD spectra between the two forms of the vitro (Sakikawa et al., 1999). For certain proteins larger than this protein in the peptide regions, indicating differences in their sec- size, for example, 75 kDa methylmalonyl-CoA mutase (Weissman ondary structures. Removal of the Fe4S4 cluster, thus affects the et al., 1995) and 72 kDa phage P22 tailspike protein (Gordon et secondary structure of the protein, giving rise to the differences al., 1994) that is capable of binding to GroEL, chaperonin-assisted in the CD spectra of the holo- and apo-aconitase. This observation folding is independent of GroES. However, it was shown recently, is further supported by the intrinsic and the extrinsic fluorescence that both chaperonins GroEL/GroES are required for the productive spectroscopy of both holo- and apo-proteins. The tryptophan emis- folding of an 86 kDa maltose-binding fusion protein (Huang and sion spectra of holo- and apo-aconitases showed a broad band at Chuang, 1999) and 82 kDa mitochondrial aconitase (Chaudhuri et 345 nm, indicating that the average environment of the tryptophan al., 2001). Interestingly, for this group of large proteins, productive residues is relatively polar. The tryptophan emission intensity of folding is achieved through binding of GroES to the trans ring of apo-form was found to be ∼35% higher than the holo-form at the GroEL. same wavelength (Fig. 1B). The low emission intensity of holo- In one such study, the interaction of different aconitase forms protein compared to its apo-form suggests that there is an efficient with chaperone molecules was investigated on the isolated pro- energy transfer from the tryptophan residues to the iron–sulfur teins by using a non-mitochondrial chaperone system, namely cluster/other polar amino acid side chains surrounding the trypto- GroEL/GroES from Escherichia coli. Guanidine-unfolded yeast aconi- phans. This phenomenon of low tryptophan emission intensities tase was shown to bind to GroEL (that does not bind native has also been found in several heme proteins due to efficient holo-aconitase), and was released upon addition of the co- energy transfer from the tryptophan residues to the heme group chaperone GroES and ATP (Chaudhuri et al., 2001). Binding of (Hochstrasser and Negus, 1984). ANS, a probe for apolar binding apo-aconitase to GroEL and a refolding role of GroEL/ATP were sites, fluorescence is strongly dependent on the hydrophobicity also demonstrated in the study on the recovery of native struc- of the environment. Free ANS fluoresces weakly in aqueous solu- ture from apo-aconitase in systems consisting of an equimolar tion and the fluorescence increases very little in the presence of mixture of the apo-protein and GroEL. Thus, the role of accessory native holo-aconitase. The extrinsic fluorescence intensity of ANS proteins in the acquisition of the folded native structure when start- bound apo-aconitase was found to be ∼15 times higher than the ing from the apo-protein, i.e., under conditions mimicking what holo-form (Fig. 1C). The absence of the iron–sulfur cluster in apo- is most likely to occur under physiological conditions, was estab- aconitase thus leads to its altered surface properties, which leads lished. to increased ANS fluorescence in apo-protein as compared to the The observation that the apo-form of aconitase remains bound holo-protein. Hence, the study shows that the apo-form has signif- to GroEL was surprising, because apo-aconitase has a well defined icant differences as compared to the holo-aconitase, in terms of its: conformation (similar to holo-aconitase) and hence, in the absence lesser secondary structure, less compact tertiary structure and dif- of chaperonin, apo-aconitase produced by removing the Fe4S4 ferent surface properties in its native form. The similar max of the cluster from the holo-enzyme was stable in solution. While apo- holo- and apo-forms indicates that the average microenvironment P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692 685

Fig. 1. (A) Far UV CD spectra of holo-aconitase, apo-aconitase and denatured aconitase measured in 10 mM sodium cacodylate, pH 7.4 at 25 ◦C. The continuous line denotes the holo-protein, the broken line denotes the apo-protein and dotted line denotes denatured protein. The protein concentration used was 2 ␮M. Denatured enzyme was prepared by 100 times dilution of holo-protein in 6 M GdnHCl. Respective buffer spectra were used for buffer subtraction. (B) Intrinsic tryptophan fluorescence emission spectra of holo-aconitase, apo-aconitase and denatured aconitase measured in 10 mM sodium cacodylate, pH 7.4 at 25 ◦C. The continuous line denotes the holo-protein, the broken line denotes the apo-protein and dotted line denotes denatured protein. The protein concentration used was 1 ␮M. Denatured enzyme was prepared by 100 times dilution of holo-protein in 6 M GdnHCl. Respective buffer spectra were used for buffer subtraction. (C) ANS bound extrinsic tryptophan fluorescence emission spectra of holo-aconitase, apo-aconitase and denatured aconitase measured in 10 mM sodium cacodylate, pH 7.4 at 25 ◦C. The continuous line denotes the holo-protein, the broken line denotes the apo-protein and dotted line denotes denatured protein. The protein concentration used was 1 ␮M with 0.1 mM ANS. Denatured enzyme was prepared by 100 times dilution of holo-protein in 6 M GdnHCl. Respective buffer spectra were used for buffer subtraction. of all the tryptophans in both the holo- and apo-forms is similar; of enzyme activity of holo-aconitase as a function of GdnHCl, however, these tryptophans have greater freedom of movement whereas, apo-aconitase is completely inactive in native form. and less proximity to polar quenchers in the apo-form, resulting in Aconitase activity progressively decreased as GdnHCl concentra- higher emission intensities. tion increased. It was found that the activity of the enzyme was completely lost at 0.5 M GdnHCl. This maybe due to the loss of 2.2. Unfolding transition of holo- and apo-aconitase the iron cluster or the destruction of the correct tertiary structure. Fig. 2B shows the unfolding transition curves of the two proteins The GdnHCl-induced unfolding transition of the folded recombi- measured by the CD ellipticity, at 222 nm, and the fraction of aconi- nant holo-aconitase was studied by monitoring the loss of enzyme tase denatured was plotted as a function of GdnHCl concentration. activity, changes in ellipticity in the far UV CD, changes in the Aconitase unfolding has been shown to be irreversible, as only intrinsic and extrinsic fluorescence spectroscopy. The results were 8–10% of denatured aconitase can refold back to yield native, compared with those of apo-aconitase. Fig. 2A shows the loss biologically active enzyme (Chaudhuri et al., 2001). Unfolding of 686 P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692

Fig. 2. (A) GdnHCl-induced loss of aconitase activity measured at 25 ◦C in 100 mM Tris–HCl, pH 7.4. The solid triangle denotes holo-aconitase and solid square denotes apo-aconitase. (B) GdnHCl-induced unfolding transition curves for holo- and apo-aconitase. The unfolding was carried out at 25 ◦C in 10 mM sodium cacodylate, pH 7.4, and the transitions were monitored by far UV CD measurements. Fraction denatured of unfolded species were plotted against the concentration of GdnHCl. Bold triangle denotes holo-aconitase and bold square denotes apo-aconitase. The protein concentration used was 2 ␮M. (C) GdnHCl-induced unfolding transition curves for holo- and apo-aconitase. The unfolding was carried out at 25 ◦C in 10 mM sodium cacodylate, pH 7.4, and the transitions were monitored by intrinsic fluorescence emission measurements. Relative fluorescence intensity of unfolded species was plotted against the concentration of GdnHCl. Bold triangle denotes holo-aconitase and bold square denotes apo-aconitase. The protein concentration used was 1 ␮M. (D) GdnHCl-induced unfolding transition curves for holo- and apo-aconitase. The unfolding was carried out at 25 ◦C in 10 mM sodium cacodylate, pH 7.4, and the transitions were monitored by ANS bound protein fluorescence emission measurements. Relative fluorescence intensity of unfolded species was plotted against the concentration of GdnHCl. Bold triangle denotes holo-aconitase and bold square denotes apo-aconitase. The protein concentration used was 1 ␮M and ANS concentration was 0.1 mM. aconitase becomes irreversible due to its large multi-domain con- mation of an intermediate during aconitase unfolding. The high formation, which makes it aggregation prone. Thus, in order to emission intensity of this intermediate reveals the presence of a estimate the difference in ground state stability of holo- and apo- conformation with increased flexibility. Fluorescence and CD data aconitase, we calculated the apparent value of Cm (Cmapp), from also indicates that the loss of aconitase activity at 0.5 M GdnHCl the aconitase unfolding profile. The Cmapp values, thus obtained are may be a direct consequence of large changes in the tertiary and summarized in Table 1. Results show that the unfolding transition secondary structure conformation in aconitase during GdnHCl- of the recombinant holo-protein occurs at a higher concentration of induced unfolding. GdnHCl (Cmapp 0.6 M) than the transition of apo-aconitase (Cmapp The protein bound ANS fluorescence is 15 times higher in the 0.25 M). The higher value of Cmapp for holo-form shows that apo- presence of native apo-aconitase than in the presence of holo- protein is considerably less stable than the holo-form. aconitase. A steep increase in ANS fluorescence emission takes The GdnHCl-induced unfolding transitions of recombinant holo- place on addition of GdnHCl, reaching a maximum at 0.5 M GdnHCl and apo-aconitase were also analyzed by intrinsic tryptophan for both holo- and apo-aconitase. With further increase of GdnHCl fluorescence emission spectroscopy. As shown in Fig. 2C, with concentration, the ANS fluorescence declines to a constant emis- increasing GdnHCl concentration, changes in the fluorescence sion intensity lower than the intensity in the absence of GdnHCl emission intensity occur in multiple stages. At GdnHCl concentra- (Fig. 2D). The change in the ANS intensity on intermediate for- tions lower than 0.25 M, the emission intensity remains almost mation from apo-aconitase was found to be 1.5 times, whereas constant. A marked increase in emission intensity takes place from holo-aconitase the change was about 6.2 times. Thus, it between 0.25 and 0.5 M GdnHCl. Further changes in the fluo- indicates that the apo-form is conformationaly much closer to rescence properties occur at GdnHCl concentrations between 0.5 the intermediate than the holo-form. It also reveals that the and 2.0 M, with a decrease in emission intensity. The maximum unfolding intermediate has a significant amount of exposed ANS fluorescence emission intensity at 0.5 M GdnHCl indicates the for- binding pockets. Both holo-aconitase and apo-aconitase show similar unfolding trends, however, apo-aconitase shows higher values of emission intensity for both tryptophan fluorescence and ANS Table 1 bound fluorescence. Concentration of GdnHCl at which half of aconitase remain folded. The above results show that the unfolding intermediate is characterized by increased flexibility and substantial secondary Protein Cm (M) structure (as indicated by far UV CD) and no enzymatic activ- Holo-aconitase 0.6023 ± 0.048 ity. Folding intermediates with such characteristics have been Apo-aconitase 0.2448 ± 0.013 observed for a few globular proteins and are called molten glob- P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692 687 ules (MG) (Ptitsyn, 1992). The fluorescence studies seem to indicate that the intermediate species might be accumulating during the 0.1005 unfolding pathway; however these non-native states were not dis- 0.0781 ) ± ± tinct in terms of their helical structure, as monitored by peptide CD 1 − analysis. Significant differences in intrinsic fluorescence emission (s obs pattern between the native and intermediate states of aconitase K clearly reveal the formation of the intermediate species during unfolding process. Thus, the unfolding studies show that the aconitase unfolding can be represented by a three state process, with formation of a 0.0034 1.0530 ± MG-like intermediate.

Native(N) → Intermediate(I) → Unfolded(U).

Also, the differences in the unfolding proﬁles of apo- and holo- aconitase show that the apo-aconitase in its native form has much lower stability (shown by low value Cm ) and large exposed 0.0011 0.0183

app ± surface area (as shown by extrinsic ANS fluorescence) than the holo-aconitase. The lack of stability of apo-aconitase was further substantiated by kinetic unfolding of holo- and apo-aconitase by stopped-flow fluorescence spectroscopy.

2.3. Kinetics of aconitase unfolding 0.0253 0.3159 The results from unfolding studies indicate that the loss of the 0.0138 0.0351 ± ± iron–sulfur cluster from holo-aconitase leads to an increased flexibility of structure of apo-aconitase resulting in decreased relative stability of apo-aconitase when compared to holo-aconitase. This loss in stability of apo-aconitase with respect to holo-aconitase was further investigated by estimating the rate of unfolding of holo- and apo-aconitase by time resolved fluorescence emission spectroscopy. The observations from unfolding kinetic measurement 0.0061 0.0821 indicate that substantial part of the unfolding of aconitase occurs 0.0034 0.1664 ± within the dead time of the stopped-flow instrument (20 ms). This ± burst phase, during unfolding, may have been caused by rapid ini- tial unzipping of tertiary structure of the protein. However, the time range under which these tertiary structures unfold is exactly not known. The kinetic progress curves for unfolding of the holo- protein was well fitted to a two exponential equation (Fig. 3A). Whereas, the kinetic progress curves for unfolding for the apo- 0.1005 0.0125 protein was well fitted to a three exponential equation (Fig. 3B), and 0.0781 0.0759 ± the apparent rate constants and the amplitudes for the two proteins ± are presented in Table 2. The unfolding reaction of apo-aconitase was found to be 3 times faster than that of the holo-protein, indicating reduced stability towards unfolding. Thus, it appears that the anchoring of the fourth domain of aconitase by the iron–sulfur cluster is essential for the stability of this protein (see Section 3). 0057 0.3159 2.4. Native 82 kDa apo-aconitase in GroEL cavity cannot be 0.043 1.0530 ± capped by GroES ±

We have established so far, that the apo-form of aconitase is less stable than the holo-aconitase and has much larger surface hydrophobicity than the holo-form. Hydrophobicity is an important parameter for substrate recognition by GroEL (Chaudhuri and Gupta, 2005). Denatured aconitase binds to GroEL and requires 0.0002 0.207 GroES binding in trans for its correct folding. Apo-aconitase has also 0.0021 0.0995 ± been shown to bind stably with GroEL (Chaudhuri et al., 2001). Can ± the GroEL bound apo-aconitase be encapsulated within cis cavity by GroES, owing to its partially folded conformation? In order to determine whether this near MW, native form of aconitase can be another example of protein substrate, >80 kDa that can be encapsulated within the cis cavity of GroEL, limited proteolytic digestion of GroEL-apo complex was carried out in presence of 2 M excess of GroES. Digestion mixtures were analyzed by SDS-PAGE (Fig. 4). The ProteinHolo- A0 0.9513 A1 K1 A2 K2 A3 K3 Apo- 0.7788

apo-form was shown to be digested completely when treated with Table 2 Kinetic unfolding parameters of aconitase. 688 P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692

GroEL, which are stained with Coomassie Blue and migrate slower than GroES, are proteolytic products of GroEL. Fig. 4 shows that the apo-aconitase bound to GroEL was digested completely, even when both GroES and Mg-ADP were present (lane 8). These conditions facilitate the formation of a stable GroEL–GroES-ADP7 complex. The data indicate that native apo-aconitase, despite its similar size to the ␣␤ heterodimer (Song et al., 2003), cannot be enclosed by GroES inside GroEL cavities. As positive control, the unfolded aconitase bound to GroEL was also digested completely (lane 4) when GroES and Mg-ADP were present.

3. Discussion

The present results show that the iron–sulfur cluster of the recombinant holo-aconitase expressed in E. coli remarkably sta- bilizes the native structure and decreases the fluorescent ANS binding pockets of the native state. The crystal structure shows, the protein to be composed of four structural domains (Lauble et al., 1992; Robbins and Stout, 1989). The fourth domain requires the assistance of GroEL for folding (Dubaquie et al., 1998) and is closely apposed to the tightly conformed domains 1–3. Our study aims to present key structural differences that take place in holo- aconitase on removal of the iron–sulfur cluster and its consequence on the binding affinity of the resultant apo-aconitase to GroEL. The Fig. 3. (A) Kinetics of holo-aconitase unfolding as monitored by fluorescence emis- apo-form showed lesser secondary structure as compared to the sion intensity changes at 345 nm. The final enzyme concentration was 1 ␮M. Bold holo-form and the changes in the secondary and tertiary structure line represents the theoretical curve for a two phase exponential function accord- of holo-aconitase on the removal of iron–sulfur cluster results in ing to Eq. (1). Inset shows the residual between experimental points and theoretical curve for parameter values given in Table 2. (B) Kinetics of apo-aconitase unfold- increased tryptophan emission intensity. This alteration in the local ing as monitored by fluorescence emission intensity changes at 345 nm. The final geometry leads to a large increase in the fluorescent ANS accessible enzyme concentration was 1 ␮M. Bold line represents the theoretical curve for a surface of apo-aconitase. The removal of iron–sulfur cluster from three phase exponential function according to Eq. (1). Inset shows the residual the active site may result in a free hinge movement of the linker between experimental points and theoretical curve for parameter values given in sequence, making the fourth domain free from the other three. The Table 2. reduced stability of the apo-aconitase can be attributed to the loss of this compact tertiary structure conformation. Increase in the solvent accessible surface area, due to the dislodging of the fourth domain from the other three domains and loss of compact tertiary structure in apo-aconitase results in an extremely large ANS bound fluorescence intensity. In the current study, apo-aconitase has been shown to be conformationaly less stable than the holo-form with more flexible structure (high tryptophan fluorescence emission) and faster unfolding rates on denaturation with GdnHCl (kobs for unfolding). The lower GdnHCl concentration required to reach the midpoint of unfolding in case of apo-aconitase lends further sup- port to the conclusion. Apparently, GroEL-bound proteins exist in collapsed, loosely packed conformations, with varying degrees of native secondary structure (Fenton and Horwich, 2003). Kuwajima and coworkers have reported that the protein state recognized by GroEL is more unfolded and expanded than the typical molten glob- Fig. 4. Limited proteolytic digestion of GroEL–apo-aconitase complex. GroEL–apo- aconitase complex at 1 ␮M was digested by 80 ␮g/mlofPKat30◦C for 15 min. ule state of alpha-lactalbumin (Okazaki et al., 1994). Presumably, Samples were analyzed by SDS-PAGE followed by Coomassie Blue staining. Lane partial unfolding of holo-aconitase, due to the loss of iron–sulfur 1, 1:1 holo-aconitase and GroEL (untreated); lane 2, 1:1 holo-aconitase and GroEL cluster leads to the exposure of hydrophobic surfaces of the helices (treated); lane 3, 1:1 denatured aconitase-GroEL complex with 2 M excess of and the buried ␤-sheets which can be bound by GroEL (Houry et al., GroES in presence of Mg-ADP (untreated); lane 4, 1:1 denatured aconitase–GroEL 1991). Thus, we have shown that apo-aconitase is structurally less complex with 2 M excess of GroES in presence of Mg-ADP (treated); lane 5, 1 ␮M apo-aconitase (untreated); lane 6, 1 ␮M apo-aconitase (treated); lane 7, 1:1 stable than holo-aconitase and binds stably to GroEL due to the apo aconitase–GroEL complex with 2 M excess of GroES in presence of Mg-ADP presence of large exposed surface area (primarily hydrophobic), (untreated); lane 8, 1:1 apo aconitase–GroEL complex with 2 M excess of GroES in so as to avoid aggregation through intermolecular association. It presence of Mg-ADP (treated); lane 9, 1:1 apo aconitase–GroEL complex without was observed that unlike holo-aconitase, the apo-aconitase aggre- GroES and Mg-ADP (untreated); lane 10, 1:1 apo aconitase–GroEL complex without ◦ GroES and Mg-ADP (treated). gates even at 4 C, when dialyzed against a low ionic strength buffer (Gupta and Chaudhuri, unpublished results). Perhaps the driving force behind the binding of apo-aconitase with GroEL is to avoid PK (lane 6), without the presence of GroEL, GroES or nucleotides, its aggregation in the highly crowded macromolecular cytosolic while holo-aconitase was completely resistant to proteolytic diges- environment. tion by PK (lane 2). In the absence of GroES and nucleotide, the Further, the interaction of GroES with GroEL–apo aconitase apo-aconitase bound to GroEL was digested completely after incu- complex was studied to ascertain the possibility of encapsulation bation of the complex with PK (lane 10). The faint bands underneath of apo-aconitase within the GroEL cis cavity. The crystal structure P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692 689 of GroEL–GroES-ADP7 complex (Xu et al., 1997) shows a 2-fold We conclude here that apo-aconitase is destabilized signifi- enlargement of the cis cavity over the trans upto a volume of cantly as compared to holo-form and the reduced stability is due 175,000 Å3. Theoretically, this volume is capable of accommodat- to its loosely structured conformation. The reduction of stability ing a globular protein of ∼142 kDa underneath GroES, assuming in the apo-form has been reflected through its faster unfolding a perfect fit to the actual folded protein volume (Xu et al., 1997). rate. Because of its higher exposed hydrophobic surface as well as Nonetheless, the upper limit for an unfolded protein to be encapsu- lower stability, apo-aconitase is susceptible to aggregation. Binding lated inside the GroEL cavity by GroES has been shown to be 57 kDa of apo-aconitase with GroEL reduces the possibility of aggrega- in both in vitro (Sakikawa et al., 1999) and in vivo (Ewalt et al., 1997; tion of the former, and hence may be considered as the driving Houry et al., 1991) studies. This size constraint of 57 kDa for an force behind their association. Based on our observations and ear- unfolded protein may reflect that an unfolded polypeptide is more lier reports, we propose a possible mechanism for the biosynthesis extended than the fully folded protein of similar size. It has been of holo-aconitase (Fig. 5). Followed by co-translational folding and shown that the acid-denatured 82 kDa yeast mitochondrial aconi- post-translational interactions with E. coli folding machinery like tase cannot be capped by GroES inside GroEL cavity (Chaudhuri trigger factor, DnaK/DnaJ/GrpE, non-native aconitase finds itself et al., 2001), which is consistent with the above size limit for bound to GroEL. Through the action of GroES and ATP, aconitase unfolded proteins. However, the 86 kDa native-like heterodimeric is released in the solution as partially folded apo-form. Because (␣␤) intermediate in the BCKD assembly pathway represents the of its reduced stability and higher exposed (hydrophobic) surface, largest protein substrate known to fit inside the GroEL cis cav- the apo-form is susceptible to aggregation and the Fe4S4 cluster ity underneath GroES, which significantly exceeds the current size binding site may not be properly constructed. The apo-protein thus limit of 57 kDa established for unfolded proteins (Song et al., 2003). binds to GroEL to avoid aggregation and to enable the proper con- Volume changes upon protein unfolding may also explain the struction and orientation of the cluster insertion site. Although, striking difference between the folded and the unfolded protein the majority part of the model in Fig. 5 has been representing structure with respect to their ability to be capped by GroES. our earlier observation (Chaudhuri et al., 2001), the last part of According to Chalikian and Bresiauer (1996) the space larger than the model has been put in the box area which is representing the the actual protein volume is needed for an unfolded protein inside observations from the present study. Hence, binding of Fe4S4 clus- the cis cavity to be encapsulated by GroES (Chalikian and Bresiauer, ter to the GroEL-bound apo-aconitase becomes a favorable process 1996). Hence, we hypothesize that, the large exposed solvent and the cluster bound protein starts rearranging itself to its folded accessible surface of apo-aconitase and the loosely structured con- form. As the folding occurs, contacts between aconitase and GroEL formation of apo-aconitase leads to large intramolecular voids and become weaker due to the burial of exposed hydrophobic sur- increased possibility of molecular vibrations between the unfolded face of aconitase (Badoc et al., 1991; Fisher, 1992). Thus, aconitase protein and the solvent, resulting in large partial specific volume is released into the solution as folded form without the assis- of apo-aconitase. Hence, apo-aconitase failed to get encapsulated tance of GroES binding and ATP hydrolysis. Further studies with with the cis cavity of GroEL by GroES inspite of forming a stable GroEL–apo-aconitase complex are needed to verify those possibil- complex with GroEL. ities.

Fig. 5. Proposed aconitase biosynthetic pathway. 690 P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692

It is true that there is no in vivo evidence yet to prove the notion were harvested, disintegrated and fractionated. The clear lysate that apo-aconitase requires GroEL for cofactor induced maturation. obtained from cell fractionation after DNAse treatment, was dia- However, the work from Chaudhuri et al. (2001), clearly demon- lyzed overnight against 50 mM NaOAc buffer, pH 4.6 containing strated that functional aconitase was released into the solution only 1 mM DTT. The clear lysate obtained after dialysis was loaded on to when Fe4S4 cluster was added to the GroEL–apo-aconitase complex SP Sepharose column for final purification. Elution of GroES was in a GroES and ATP independent manner. This information may be done using a gradient of 0.5 M NaCl. Pure GroES was eluted at considered as an in vitro evidence for the requirement of GroEL in ∼0.2 M NaCl. the maturation process of apo-aconitase. There should be experiments to be carried out in the future to demonstrate the in vivo 5.2. Preparation of apo-aconitase evidence for GroEL-assisted maturation of holo- aconitase or other metallo proteins. This would certainly be beneficial to understand Apo-aconitase was obtained by chemical stripping of the the steps in the metallo protein biosynthesis. iron–sulfur cluster from holo-aconitase (Kennedy and Beinert, Hence, the present study throws some light on the possi- 1988). All procedures were conducted at 0–4 ◦C using 0.1 M HEPES, ble biosynthetic pathways for the metallo-enzymes and forms an pH 7.5. EDTA and potassium ferricyanide were added to the enzyme important basis for further studies on the role of GroEL in the in solution in the buffer in the given order, in a molar ratio of enzyme vivo and in vitro folding of metallo proteins. to EDTA to ferricyanide, 1:50:20. After 5–7 min, when the loss of color from the cluster was extensive, the solution was desalted by centrifugation on G-50 spin columns equilibrated with the buffer. 4. Materials The eluted apo-protein was diluted to the desired concentration and stored at −80 ◦C. Strains, plasmids and culture conditions – E. coli M15 strain was used for the expression of pQE60Aco and pACYCELS. The gene 5.3. Aconitase assay for yeast mitochondrial aconitase (aco), cloned in the pQE60 vector from Qiagen, was obtained from Dr. Sabine Rospert, Germany. Aconitase activity was quantitated using a coupled enzyme M15 strain contains multiple copies of the plasmid, pREP4, contain- assay. Aconitase catalyzes conversion of citrate to iso-citrate, ing the gene coding for LacI and kanamycin resistance and helps which in turn is converted to ␣-keto glutarate in the presence in regulating the expression of aconitase in pQE60. This recom- of iso-citrate dehydrogenase along with the formation of NADPH binant strain was further used for purification of aconitase and from NADP (Morrison, 1954). The assay was performed by taking GroEL. BL21(DE3) E. coli strain was transformed with pET22dES 20–50 ␮g of protein in a 1 ml reaction volume (0.1 M Tris–Cl, pH 8, (for over-expression of GroES) and was used for GroES purification. 0.66 mM sodium citrate, 0.66 mM MnSO , 0.5 mg/ml ␤-NADP and The pACYCELS containing groEL and groES; and pET22dES contain- 4 0.17 mg/ml iso-citrate dehydrogenase). The formation of NADPH ing groES gene were generous gifts from Prof. Arthur L. Horwich, was monitored at 340 nm using kinetics application on Beckman USA. Coulter DU 800 (USA). Enzyme concentrations were determined Sodium citrate for aconitase activity measurement was obtained using Bradford dye binding assay (Bradford, 1976). from Merck; the guanidine hydrochloride, iso-citrate dehydrogenase and cis-aconitate were Sigma products. The concentration of 5.4. Guanidine hydrochloride-induced denaturation of aconitase GdnHCl was determined by the refractive index measurement at 589 nm using an Atago 3T refractometer (Pace, 1986). Q Sepharose, Concentrated solution of 100 ␮M aconitase was diluted 50 times CM Sepharose, SP Sepharose and Superdex 75 were Pharmacia with buffers of different GdnHCl concentrations. The denatured products. All other reagents were local products of analytical grade. aconitase solution (final aconitase concentration 2 ␮M) was incubated at 25 ◦C for 30 min. Denaturation of aconitase was monitored 5. Methods by observing the loss of enzymatic activity, change in far UV CD ellipticity and intrinsic tryptophan fluorescence intensity. 5.1. Purification of various recombinant proteins 5.5. Change in secondary structure on denaturation of aconitase, Purification of aconitase was carried out according to the proto- monitored by far UV CD spectroscopy col given by Horwich and coworkers (Chaudhuri et al., 2001) with necessary modifications. Aconitase over-expression was carried CD spectra were taken on a Jasco J-810 spectropolarimeter using out under the conditions described by Chaudhuri and coworkers an optical cuvette with a path length of 2.00 mm for measurements (Gupta et al., 2006). Post-induction culture was harvested, cells in the peptide region. The CD spectra of the protein were measured were lysed after DNAse treatment and the suspension was cen- in 10 mM sodium cacodylate buffer, pH 7.4. The protein concentra- trifuged to get clear lysate. Soluble fraction from the lysate was tion for the CD measurement was 2 ␮M. Change in the ellipticity applied on a Q Sepharose column attached in tandem with a CM values in the range of 200–250 nm was measured for each sample Sepharose column. Aconitase was eluted from the CM Sepharose at a scan speed of 50 nm/min. Spectra were averaged over three column using 0.5 mM cis-aconitate. GroEL was purified by elu- scans for each sample. Respective buffers with GdnHCl (without tion from the Q Sepharose column, using a gradient of 1 M NaCl. aconitase) were used as blanks. GroEL fractions were obtained at 0.5–0.6 M NaCl. The pooled GroEL Fraction denatured was calculated at 222 nm and plotted against fraction was further purified according to the published protocol concentration of GdnHCl. This unfolding curve was used to obtain (Weissman et al., 1995) with some modifications. The non-native the Cmapp values for both holo- and apo-forms of aconitase. protein substrates bound within the hydrophobic cavity of GroEL were removed by affigel blue treatment. The sample was then 5.6. Denaturation of aconitase monitored by fluorescence applied to a Superdex 75 HR 10/30 pre-packed column, in order spectroscopy to obtain pure GroEL. GroES was purified according to the published protocol Fluorescence measurements were performed on a LS55 spec- (Weissman et al., 1995) with some modifications. GroES was over- trofluorimeter (Perkin–Elmer, USA) using a 10 mm path-length expressed in the BL21DE3 E. coli cells from pET22d vector. Cells cuvette and a protein concentration of 1 ␮M. Tryptophan emission P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692 691 intensity in the range of 300–400 nm was measured after excitation Human Resource and Development (MHRD), Govt. of India, Depart- at 295 nm for aconitase at different concentrations of GdnHCl. Tryp- ment of Science and Technology (DST), Govt. of India, and Industrial tophan emission spectra were obtained using an excitation slit of Research and Development division (IRD), IIT Delhi, India. 5 nm and emission slits of 2.5 nm and a scan rate of 100 nm min−1. Respective buffers with GdnHCl (without aconitase) were used as References blanks. ANS fluorescence spectra for the samples were recorded after excitation of the probe at 370 nm and emission between 400 Arai M, Kuwajima K. Rapid formation of a molten globule intermediate in refolding and 600 nm, using excitation and emission slits of 5 nm and a scan of ␣-lactalbumin. Fold Des 1996;1:275–87. rate of 100 nm min−1. Final ANS concentration was optimized for Badoc IG, Smith CJ, Wood S, Halsall DJ, Holbrook JJ, Lund P, et al. Binding of a maximum emission intensity to 0.1 mM. chaperonin to the folding intermediates of lactate dehydrogenase. Biochemistry 1991;30:9195–200. Bradford M. A rapid and sensitive method for quantitation of microgram quan- 5.7. Kinetic measurements tities of protein utilizing the principle of protein–dye binding. Anal Biochem 1976;72:248–54. Burova TV, Bernhardt R, Pfeil W. Conformational stability of bovine holo and apo Unfolding reactions of the holo- and apo-proteins (10 ␮M) were adrenodoxin – a scanning calorimetric study. Protein Sci 1995;4:909–16. induced by GdnHCl concentration jump, from 0 to 4 M, performed Chalikian TV, Bresiauer KJ. On volume changes accompanying conformational tran- by a stopped-flow apparatus RX 2000 (Applied Photophysics, Sur- sitions of biopolymers. Biopolymers 1996;39:619–26. Chaudhuri TK, Gupta P. Factors governing the substrate recognition by GroEL rey, UK) installed in the cell compartment of the Perkin Elmer LS55 chaperone: a sequence correlation approach. Cell Stress Chaperones 2005;10: spectrofluorimeter (Perkin–Elmer, USA). All kinetic measurements 24–36. were done in the presence of 10 mM sodium cacodylate, pH 7.4 Chaudhuri TK, Arai M, Terada TP, Ikura T, Kuwajima K. Equilibrium and ◦ kinetic studies on folding of the authentic and recombinant forms of and 25 C. The dead time of the stopped-flow CD apparatus was human alpha-lactalbumin by circular dichroism spectroscopy. Biochemistry 20 ms. The two solutions were mixed with a mixing ratio of 1:10. 2000;39:15643–51. The reactions were monitored by the tryptophan emission intensity Chaudhuri TK, Farr GW, Fenton WA, Rospert S, Horwich AL. GroEL/GroES-mediated folding of a protein too large to be encapsulated. Cell 2001;107:235–46. at 345 nm after excitation at 295 nm. The kinetic progress curves Dubaquie Y, Looser R, Funfschilling U, Jeno P, Rospert S. Identification of in vivo for unfolding were fitted by the non-linear least-squares method substrates of the yeast mitochondrial chaperonins reveals overlapping but non- (Chaudhuri et al., 2000)toEq.(1): identical requirement for hsp60 and hsp10. EMBO J 1998;17:5868–76. Ewalt KL, Hendrick JP, Houry WA, Hartl FU. In vivo observation of polypeptide flux A t = A ∞ + A ˛ −k t through the bacterial chaperonin system. Cell 1997;90:491–500. ( ) ( ) obs iexp( i ) (1) Fenton WA, Horwich AL. Chaperonin-mediated protein folding: fate of substrate i polypeptide. Q Rev Biophys 2003;36:229–56. Fink AL. Compact intermediate states in protein folding. In: Proteins: structure, func- where A(t) and A(∞) are the observed values of the fluorescence tion, and engineering. In: Biswas BB, Roy S, editors. Subcellular biochemistry, vol. intensity at time t and infinite time, respectively, A is the 24. New York: Plenum Press; 1995. p. 27–53. obs Fisher MT. Promotion of the in vitro renaturation of dodecameric glutamine syn- observed total amplitude [A(0) − A(∞)], and ki and ˛i are the appar- thetase from Escherichia coli in the presence of GroEL (Chaperonin-60) and ATP. ent rate constant and fractional amplitude, respectively, of the ith Biochemistry 1992;31:3955–63. kinetic phase. Frausto da Silva JJR, Williams RJP. The biological chemistry of the elements. Oxford: Clarendon Press; 1991. Gordon CL, Sather SK, Casjens S, King J. Selective in vivo rescue by GroEL/ES of ther- 5.8. Formation of GroEL–aconitase binary complex molabile folding intermediates to phage P22 structural proteins. J Biol Chem 1994;269:27941–51. Gupta P, Aggarwal N, Batra P, Mishra S, Chaudhuri TK. Co-expression of chaper- Denatured aconitase was prepared by 10-fold dilution of onin GroEL/GroES enhances in vivo folding of yeast mitochondrial aconitase 100 ␮M holo-enzyme into 25 mM glycine phosphate, pH 2, and and alters the growth characteristics of Escherichia coli. Int J Biochem Cell Biol incubating at 25 ◦C for 30 min. GroEL–aconitase binary complex 2006;38:1975–85. Hochstrasser RM, Negus DK. Picosecond fluorescence decay of tryptophans in myo- was formed by a 10-fold dilution of the denatured aconitase or globin. Proc Natl Acad Sci USA 1984;81:4399–403. apo-aconitase into refolding buffer (50 mM HEPES pH 7.4, 10 mM Houry WA, Frishman D, Eckerskorn C, Lottspeich F, Hartl FU. Identification of in vivo KCl, and 10 mM MgCl ) containing 1 ␮M wild type GroEL. Binary substrates of the chaperonin GroEL. Nature 1991;402:147–54. 2 Huang YS, Chuang DT. Mechanisms for GroEL/GroES-mediated folding of a large complexes were purified by gel filtration using a Superdex 75 HR 86 kDa fusion polypeptide in vitro. J Biol Chem 1999;274:10405–12. 10/30. Isolated complex was concentrated using a Centricon-30 Jeanicke R. Protein folding: local structures, domains, subunits, and assemblies. Bio- concentrator with a 30 kDa cut off membrane (Millipore) at 25 ◦C, chemistry 1991;30:3147–61. Kay MS, Baldwin RL. Packing interactions in the apomyoglobin folding intermediate. and its concentration was measured by Bradford assay (Bradford, Nat Struct Biol 1996;3:439–45. 1976). Kennedy MC, Beinert H. The state of cluster SH and S2− of aconitase during cluster interconversions and removal. J Biol Chem 1988;263:8194–8. Kim PS, Baldwin RL. Intermediates in the folding reactions of small proteins. Annu 5.9. Proteinase K digestion for GroES encapsulation Rev Biochem 1990;9:631–60. Kuwajima K. The molten globule state as a clue for understanding the folding and ␮ ␮ cooperativity of globular-protein structure. Proteins 1989;6:87–103. Binary complex between 1 M GroEL and 1 M aconitase were ␣ ␮ Kuwajima K. The molten globule state of -lactalbumin. FASEB J 1996;10:102–9. incubated with 2 M GroES in the presence of ADP for 10 min at Lauble H, Kennedy MC, Beinert H, Stout CD. Crystal structures of aconitase with ◦ 25 C. Proteolytic digestion was carried out at optimized conditions, iso-citrate and nitroiso-citrate bound. Biochemistry 1992;31:2735–48. with PK at a final concentration of 80 ␮g/ml for 15 min at 30 ◦C. Lippard SJ, Berg JM. Principles of bioinorganic chemistry. California: University Sci- ence Books; 1994. PMSF to a final concentration of 1 mM was added in order to stop Matthews CR. Pathways of protein foldings. Annu Rev Biochem 1993;62:653–83. further proteolysis reaction. Samples obtained after PK treatment Morrison JF. The activation of aconitase by ferrous ions and reducing agents. Biochem were analyzed on 15% SDS-PAGE. J 1954;58:685–92. Ogasahara K, Yutani K. Unfolding–refolding kinetics of the tryptophan synthase ␣- subunit by CD and fluorescence measurements. J Mol Biol 1994;236:1227–40. Acknowledgements Okazaki A, Ikura T, Nikaido K, Kuwajima K. The chaperonin GroEL does not rec- ognize apo-alpha-lactalbumin in the molten globule state. Nat Struct Biol 1994;1:439–46. The authors acknowledge the generous gifts of pQE60Aco Pace CN. Determination and analysis of urea and guanidine hydrochloride denatu- plasmid from Prof. Sabine Rospert and pACYCEL, pACYCELS and ration curves. Methods Enzymol 1986;131:266–80. pET22dES from Prof. A.L. Horwich. Miss. Parul Gupta is a recip- Ptitsyn OB. Protein folding: hypotheses and experiments. J Protein Chem 1987;6:273–93. ient of senior research fellowship from CSIR, Govt. of India. The Ptitsyn OB. The molten globule state. In: Creighton TE, editor. Protein folding. New work has also been supported by the funding from Ministry of York: W.H. Freeman and Company; 1992. p. 243–300. 692 P. Gupta et al. / The International Journal of Biochemistry & Cell Biology 42 (2010) 683–692

Ptitsyn OB. Molten globule and protein folding. Adv Protein Chem 1995;47:83– Weissman JS, Hohl CM, Kovalenko O, Kashi Y, Chen S, Braig K, et al. Mechanism 229. of GroEL action: productive release of polypeptide from a sequestered position Robbins AH, Stout CD. The structure of aconitase. Proteins 1989;5:289–312. under GroES. Cell 1995;83:577–87. Sakikawa C, Taguchi H, Makino Y, Yoshida M. On the maximum size of pro- Wittung-Stafshede P. Role of cofactors in protein folding. Acc Chem Res teins to stay and fold in the cavity of GroEL underneath GroES. J Biol Chem 2002;35:201–8. 1999;274:21251–6. Xu Z, Horwich AL, Sigler PB. The crystal structure of the asymmetric GroEL–GroES- Song JL, Li J, Huang SY, Chuang DT. Encapsulation of an 86 kda assembly intermediate (ADP)7 chaperonin complex. Nature 1997;388:741–50. inside the cavities of GroEL and its single-ring variant SR1 by GroES. J Biol Chem 2003;278:2515–21.