ATP Induces Non-identity of Two Rings in Chaperonin GroEL*

For its function, the Escherichia coli chaperonin GroEL requires the presence of ATP and co-chaperonin GroES. We have observed that ADP displays a two-step inhibition of GroEL-dependent ATP hydrolysis, wherein one-half of the GroEL ATPase sites is strongly inhibited by ADP while the other half is affected very mildly. It is suggested that interaction with ATP induces structural and functional differences between two initially identical rings in GroEL (inter-ring negative cooperativity) and that the subsequent binding of GroES occurs to the ring that is occupied first by ATP in a positively cooper- ative manner. re-ductive methylation procedure (22) by treatment of 40 p~ GroES (7- mer) with 5 mM formaldehyde and 0.5 mCi of NaBC3H1, in 95 pl of 0.2 M sodium borate (pH 9.0) for 25 min on ice followed by gel filtration through Sephadex G-25 superfine in buffer A. The specific radioactivity of the C3HIGroES preparation was about 1.8 x lo5 cpdpg of protein. The ATP Concentration Dependence of Interaction of GroES with GroEL-1.8 PM GroEL (14-mer) was incubated with 2.4 r3H1GroES in 40 p1 of buffer A containing different concentrations of ATP (4 p ~ - 8 mM) for 10 min at 25 “C. Then, GroES bound to GroEL was separated by centrifugation through 1.4 ml of 5-20% sucrose gradient in bufferA containing the same concentrations ofATP as in the samples (Beckman TLlOO centrifuge, TLS 55 rotor, 54,000 rpm for 150 min at 4 “C). 0.1-ml fractions were collected and counted.


For its function, the Escherichia coli
chaperonin GroEL requires the presence of ATP and co-chaperonin GroES. We have observed that ADP displays a two-step inhibition of GroEL-dependent ATP hydrolysis, wherein one-half of the GroEL ATPase sites is strongly inhibited by ADP while the other half is affected very mildly. It is suggested that interaction with ATP induces structural and functional differences between two initially identical rings in GroEL (inter-ring negative cooperativity) and that the subsequent binding of GroES occurs to the ring that is occupied first by ATP in a positively cooperative manner.
The Escherichia coli heat-shock protein GroEL is a molecular chaperone of the chaperonin family that interacts transiently a n d ATP dependently with various newly synthesized proteins before they fold into the native form (2) and mediates their folding, assembly, and transport (for reviews, see Refs. 3-71. The molecular mechanism for its action is still unclear, and a high resolution three-dimensional structure of it is not yet available. GroEL is an oligomeric protein consisting of 14 identical subunits (8,9), each with a molecular mass of 57.3 kDa (10). GroEL has 14 ATP-binding sites, presumably one site per subunit (111, and possesses a moderate ATPase activity (8). ATP bound to GroEL is in contact with the Cys-137 residue (12).
For successful folding and assembly of some proteins, GroEL requires the presence of the co-chaperonin GroES, which comprises a single ring of seven identical subunits (13) of a molecular mass of 10.4 kDa each (10). In the presence of Mg2' ions and ATP (13) and, also, ADP or nonhydrolyzable ATP analogues (ll), GroES 7-mer forms a 1:l stoichiometric complex with GroEL 14-mer and inhibits its ATPase activity (13)(14)(15). GroEL hydrolyzes ATP with a positive cooperativity, and GroES acts as an allosteric effector by increasing the cooperativity (11,15,16). It has been suggested that GroES acts as a "coupling" factor linking the hydrolysis of ATP by GroEL to the release of the target protein in a form that is committed to the native state (14). According to negatively stained electron micrfscopy (8,9,17) and x-ray crystallography at a resolution of 8.9 A (18), GroEL is a cylindrical structure (its height and diameter are both around 130 A) that possesses an unusual 7-fold symmetry a n d is comprised of two heptameric rings stacked "base-tobase." GroEL binds GroES asymmetrically as a cap to either of the two end surfaces of its double ring whereas the opposite end * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18  surface is markedly perturbed (17,191. The binding of only one GroES to GroEL is surprising since the base-to-base stacking of two identical rings in GroEL implies that its two end surfaces should also be identical. Hence, it has been proposed that ATPpromoted binding of GroES to either end surface of the GroEL cylinder induces allosterical inhibition of the GroES binding to the opposite surface of GroEL (15,19,20).
Here we suggest that asymmetric ATP-promoted binding of GroES to GroEL is a consequence of non-identity of two stacked rings in GroEL caused by its interaction with ATP.
EXPERIMENTAL PROCEDURES Proteins, Chemicals, and Buffers-GroEL and GroES were purified from E. coli TG2 cells (11,21). Effect of ADP and ATPyS' on Initial Rate of ATP Hydrolysis by GroEGThe reaction was started by mixing 10 pl of 0.8 m M [y3'P1ATP with 30 pl of 0.2 p~ GroEL (14-mer) in buffer A without or in the presence of 0.3 p~ GroES (7-mer) and various concentrations of ADP or ATPyS. Before mixing, both solutions were incubated at 25 "C for 5 min.
The reaction was carried out at the same temperature and terminated at different time intervals from 1 to 6 min after the start by removal of 8-p1 aliquots to 1.5-ml Eppendorf tubes containing 70 p1 of 1 M perchloric acid and 1 m M sodium phosphate, cooled on ice. The subsequent quantitative analysis of the [32PlPi released was carried out as described previously (21). Control reactions were run under all conditions in the absence of GroEL, and extracted orthophosphate was subtracted as background. (The [32PlPi background was constant and equals about 1.5 x lo3 cpm, whereas incubation in the presence of GroEL for 2 or 4 min gave about 2.5 or 5 x lo4 cpm of the extracted orthophosphate, respectively.) Note that initial rates were measured when ~1 0 % of [y-32PlATP was hydrolyzed.
3H Labeling of GroES-The labeling was carried out using the reductive methylation procedure (22) by treatment of 40 p~ GroES (7mer) with 5 m M formaldehyde and 0.5 mCi of NaBC3H1, in 95 pl of 0.2 M sodium borate (pH 9.0) for 25 min on ice followed by gel filtration through Sephadex G-25 superfine in buffer A. The specific radioactivity of the C3HIGroES preparation was about 1.8 x lo5 cpdpg of protein.
The ATP Concentration Dependence of Interaction of GroES with GroEL-1.8 PM GroEL (14-mer) was incubated with 2.4 PM r3H1GroES (7-mer) in 40 p1 of buffer A containing different concentrations of ATP (4 p~-8 mM) for 10 min at 25 "C. Then, GroES bound to GroEL was separated by centrifugation through 1.4 ml of 5-20% sucrose gradient in bufferA containing the same concentrations ofATP as in the samples (Beckman TLlOO centrifuge, TLS 55 rotor, 54,000 rpm for 150 min at 4 "C). 0.1-ml fractions were collected and counted.

RESULTS AND DISCUSSION
As we have shown previously (ll), in the presence of GroES only half of the 14 ATP-binding sites of GroEL have a strong affinity for ADP and can bind and hydrolyze ATP with a positive cooperativity. It has been suggested that two initially identical rings in GroEL become functionally non-identical within the GroEL-GroES complex, and only one of them is able to interact with ATP (or ADP) in a positively cooperative manner (11). This suggestion has been further supported by the observation that hydrolysis of ATP by the GroEL-GroES complex occurs i n an asymmetric manner in which one of the rings of GroEL is completely inhibited, while the other hydrolyzes ATP with significantly altered kinetic properties (15). All of these and other (19) data have led to the generally accepted conclusion that functional non-identity of two rings in GroEL is The abbreviation used is: ATPyS, adenosine 5'-3-O-(thio)triphosphate. caused by its interaction with GroES. In particular, it has been proposed that GroEIJGroES interaction imposes half-of-thesites reactivity upon the double-ring structure of GroEL (15). Meantime, the requirement for this interaction of the continuous presence of adenine nucleotides, which are suggested to bind first to GroEL (11,13, E ) , raises the question whether it is the interaction of adenine nucleotides with GroEL that makes two rings of GroEL non-identical.
To answer this question, the inhibitory effects ofADP and the nonhydrolyzable ATP analogue ATPyS on ATP hydrolysis by GroEL were tested. As seen in Fig. 1, in contrast to ATPyS, ADP shows a two-step inhibition. About half of the ATPase activity of GroEL is strongly inhibited by ADP so that the initial rate of ATP hydrolysis is halved at about an equimolar ratio ofADP to [Y-~'P]ATP (similar to the effect ofATPyS), while the other half of the activity is affected very mildly by the increasing concentration of ADP. In other words, only one-half of the GroEL ATPase sites (corresponding to one of the two rings) shows a strong affinity for ADP, while the other half (other ring) practically does not. This result is consistent with the ability ofADP to bind to only one of the two GroEL rings, revealed in the presence of GroES (11). The ADP concentration dependence of ATP hydrolysis by GroEL was also tested in the presence of GroES (Fig. 1). As expected (13-151, the initial rate of ATP hydrolysis is halved in the presence of GroES. The remaining half of the ATPase activity is fully blocked at about the equimolar ratio of the ADP added to [y-32PlATP. Comparison of the inhibition effects of ADP on the initial rates of ATP hydrolysis by GroEL in the absence and presence of GroES suggests that GroES inhibits specifically the ADP-independent half of the GroELATPases, whereas the other half, with a high affinity for ADP, is not affected by GroES. A sigmoidal dependence of the yield of GroES-GroEL complex upon the ATP concentration (Fig. 2) indicates that interaction of GroES with GroEL occurs with a positive cooperativity with respect to ATP. In addition, the Hill plot (see inset to Fig. 2) shows the value of the Hill coefficient (nH) of about 3.1, which coincides with that for ATP hydrolysis by GroEL in the presence of GroES (nH = 2.8 or 3.0, see Refs. 11 or 16, respectively). This evidence confirms a positive cooperativity in the binding and hydrolysis ofATP by GroEL (11,16) and correlates with the idea that binding of GroES to GroEL depends on the cooperative interaction of ATP with GroEL. In addition, we would like to emphasize the fact that, for the concentration of GroEL used (1.8 VM of 14-mer), its equimolar binding to GroES is reached at about 25 VM ATP (Fig. 2). As we have shown previously (111, when the same concentrations of GroEL and ATP were used in the presence of GroES, only half of the ATPbinding sites of GroEL (one of its two rings) were occupied with ATP in a positively cooperative manner. This correlation provides additional evidence that cooperative binding of ATP to only one of the two rings in GroEL is quite enough for equimolar binding of GroES.
Summarizing, we suggest that (i) in the absence of GroES, two rings in GroEL are strongly different with respect to their affinity for ADP; (ii) one of the rings, the ATPase sites of which have been occupied first with ATP in a positively cooperative manner, binds GroES; and (iii) the ATPases of this ring possess a high affinity for ADP and function within the GroEL-GroES complex (see also Ref. 15). There are two alternative opinions on the origin of the functional non-identity of two rings in GroEL. Either two rings just appear identical, while being actually different in some functionally important regions (ATPbinding sites), or two rings are indeed initially identical and become different upon the interaction with ATP. We prefer the second suggestion. First of all, the cooperative mechanism of binding and hydrolysis of ATP by GroEL (11,15,16,23) indicates, by definition (24,25), that some structural rearrangements of the intersubunit interactions in GroEL 14-mer should be induced by binding of substrate molecules. Moreover, these ATP-induced rearrangements in GroEL were detected experimentally (17,21,26). Hence it seems likely that a positively cooperative binding and/or hydrolysis of ATP by either of the two rings in GroEL induces some changes in intersubunit interactions that are transmitted to the other ring and diminish its initially equivalent affinity for ADP (so-called negative cooperativity or half-of-the-sites reactivity (27, 28)). In other words, because of the interaction with ATP two initially identical rings in GroEL become structurally and functionally different. CONCLUSION The following scenario can be suggested, which describes how two identical rings of GroEL become non-identical. An initial cause of this change is the interaction of GroEL with ATP, which occurs with a positive cooperativity within the rings and with a negative cooperativity between them. The binding and/or hydrolysis of ATP by GroEL induces some structural changes making the rings different. Either of them, which are occupied first with ATP in a positively cooperative manner, binds GroES. Due to the inter-ring negative cooperativity, the ATP-binding sites of the other ring show the depressed affinity for adenine nucleotides and GroES. Thus, an asymmetric binding of GroES to GroEL can be a consequence of non-identity of two stacked rings in GroEL, which arises from its interaction with ATP.