Creating the Functional Single-Ring GroEL-GroES Chaperonin Systems via Modulating GroEL-GroES Interaction

Chaperonin and cochaperonin, represented by E. coli GroEL and GroES, are essential molecular chaperones for protein folding. The double-ring assembly of GroEL is required to function with GroES, and a single-ring GroEL variant GroELSR forms a stable complex with GroES, arresting the chaperoning reaction cycle. GroES I25 interacts with GroEL; however, mutations of I25 abolish GroES-GroEL interaction due to the seven-fold mutational amplification in heptameric GroES. To weaken GroELSR-GroES interaction in a controlled manner, we used groES7, a gene linking seven copies of groES, to incorporate I25 mutations in selected GroES modules in GroES7. We generated GroES7 variants with different numbers of GroESI25A or GroESI25D modules and different arrangements of the mutated modules, and biochemically characterized their interactions with GroELSR. GroES7 variants with two mutated modules participated in GroELSR–mediated protein folding in vitro. GroES7 variants with two or three mutated modules collaborated with GroELSR to perform chaperone function in vivo: three GroES7 variants functioned with GroELSR under both normal and heat-shock conditions. Our studies on functional single-ring bacterial chaperonin systems are informative to the single-ring human mitochondrial chaperonin mtHsp60-mtHsp10, and will provide insights into how the double-ring bacterial system has evolved to the single-ring mtHsp60-mtHsp10.

The above trans-ring allosteric effect of ATP binding/hydrolysis on GroES dissociation and substrate release is essential, and the two-ring assembly of GroEL is required for the GroEL-GroES chaperone function. Interestingly, human mitochondrial mtHsp60 exists as a single heptameric ring 12,13 . mtHsp60 interacts with its cochaperonin mtHsp10 only transiently 14 , and as such dissociation of mtHsp10 and release of folding substrate from mtHsp60 do not require the trans-ring allostery driven by the ATP binding/hydrolysis as seen in the double ring GroEL-GroES system (above). However, the model that mtHsp60-mtHsp10 functions as a single ring 14,15 has been challenged. It is proposed that although mtHsp60 exists as a single heptameric ring and interacts with heptameric mtHsp10, in the course of chaperone reaction cycle, two mtHsp60-mtHsp10 complexes associate via mtHsp60 equatorial domains to form a football shape (mtHsp60-mtHsp10) 2 16 . An mtHsp60 mutant bound with mtHsp10 was crystalized in the football conformation 17,18 , however, the structure does not explain why the two mtHsp60-mtHsp10 molecules associate into the (mtHsp60-mtHsp10) 2 football conformation. Thus, whether mtHsp60-mtHsp10, or broadly the chaperonin system, may operate via a single-ring mechanism is still not certain. The ability to function as single ring suggests an evolutionary adaptability of the chaperonin family.
To identify a functional single-ring chaperonin system, we set out to convert a nonfunctional single-ring GroEL variant, GroEL SR , by modifying its interaction with GroES. GroEL SR has four mutations (R451A/E461A/S463A/ V464A) to disrupt the inter-ring contact 19 . Although the GroEL SR -GroES cavity allows misfolded substrates to undergo folding to the native conformation [20][21][22][23] , it traps and does not release the substrates. GroEL SR -GroES has t 1/2 = 300 min −1 19 , considerable longer than the ~15 s lifetime of the GroEL-GroES complex 24,25 . Failure to release folding substrates accounts for the inability of GroEL SR -GroES to support cell growth 26 . Mutations in GroEL SR allow the single-ring GroEL SR -GroES to substitute the double ring GroEL-GroES in supporting cell growth under the normal condition 27,28 , and some mutations also support cell growth under the heat stress condition 29 . However, mechanistic understandings of these single-ring variants are limited as the mutational effects are most likely allosteric. Similarly, genetic analysis of GroES residues (G24/I25/V26/L27) on the GroEL-GroES interface has identified GroES mutants collaborate with GroEL SR at lower temperatures (18 °C and 30 °C); however, little biochemical characterization of the mutational effects is available 30 .
A direct mutation on groES impacts all seven GroES subunits in the GroES heptamer. To avoid this inherent mutational amplification and to incorporate mutations selectively into specific GroES subunits, we generated a concatenated gene groES 7 that links seven groES genes to express a continuous polypeptide GroES 7 with seven GroES modules 31 . We used groES 7 to incorporate mutations in specific GroES modules in GroES 7 to modify the GroEL-GroES interface in a controlled manner. We hypothesized that modifying the chaperonin/cochapernonin interaction would activate the single-ring GroEL SR -GroES. In our earlier study, we generated GroES 7 variants with reduced affinities for GroEL SR and identified active GroES 7 variants including GroES 7 I25D 1,4 for GroEL SR -mediated folding of malate dehydrogenase (MDH). Based on these previous findings, in the current study we designed and generated comprehensive GroES 7 variants, to systematically modulate binding of GroES 7 to GroEL SR . We characterized their interaction with GroEL SR , their activity in assisting in protein folding and their in vivo chaperone function. We found that three GroES 7 variants functioned with GroEL SR in supporting cell growth under both normal and heat shock conditions.

Results
We sought to create functional single-ring GroEL SR -GroES chaperonin systems that support cell growth under normal and heat shock conditions. GroEL SR -GroES has been shown to perform folding of substrate proteins, but its inability to release the folding substrate arrests the folding cycle, obstructing the chaperone function. To weaken GroEL SR -GroES interaction thereby to resume cycling of the folding reaction, here we systematically modified the GroEL SR -GroES interaction using a concatenated gene groES 7 we generated previously 31 . Mutations of GroES I25A and L27A have the same effect as I25D and L27D in abolishing GroEL-GroES interaction. The GroEL-GroES interaction can be characterized via three assays: the ATPase activity, since binding of GroES inhibits GroEL's ATPase activity by 50% 25,32,33 , the enzymatic activity of malate dehydrogenase (MDH) since efficient folding of MDH requires not only the formation but also dissociation of the GroEL-GroES folding cavity 34 , and measurement of dissociation constant (K d ). GroES interacts with GroEL via a tri-peptide I25/V26/L27 region 11 , and our previous study showed mutations of either I25D or L27D but not V26D in GroES abolish GroES's interaction with GroEL 31 . Specifically, we showed that both GroESI25D and GroESL27D mutants did not inhibit GroEL's ATPase activity, did not participate in GroEL-mediated MDH folding, and no stable GroEL-GroES complex could be isolated. We reasoned that a conserved mutation to Ala would have a less detrimental effect and would not completely abolish the hydrophobic GroEL-GroES interaction. As shown in Fig. 1A, both GroESI25A and GroESL27A did not inhibit GroEL's ATPase activity, suggesting that both Ala mutations abolished GroEL-GroES interaction. Additionally, no MDH activity was observed in either GroEL-GroESI25A or GroEL-GroESL27A (Fig. 1B), indicating that neither GroESI25A nor GroESL27A collaborated with GroEL in assisting folding of MDH. Finally, we measured GroEL-GroES interaction using microscale thermophoresis (MST). Both I25A and L27A mutations decreased GroES's binding affinity to GroEL by >1,000 fold from Kd's values of 3.83 (±0.93) nM to >5 uM (Supplementary Table S1).
We next evaluated the Ala mutational effect on GroES's interaction with the single-ring GroEL SR . Wild type GroES has a strong binding affinity for GroEL SR as shown in the three aspects: it inhibits the ATPase activity of GroEL SR by ~90% 19 , GroEL SR -GroES traps the refolding MDH resulting in lack of MDH activity 21 , and the GroEL SR -GroES complex is highly stable with a slow dissociation rate 19 . We have shown that either I25D or L27D mutations in GroES abolish GroEL SR -GroES interaction 31 . Figure 1A shows that like GroESI25D and GroESL27D, neither GroESI25A nor GroESL27A affected ATP hydrolysis of GroEL SR , suggesting that they did not interact with GroEL SR . Also similar to their Asp counterparts, neither GroESI25A nor GroESL27A collaborated with in GroEL SR in actively refolding MDH (Fig. 1B). Finally, like the Asp variants, the GroES Ala variants did not show binding affinity for GroEL SR based on MST (data not shown).
Together, the Ala mutations at I25 and L27 drastically abolished GroES's binding to both GroEL and GroEL SR , the same effect as observed with the Asp mutations. These findings suggest that the hydrophobic residue with extended side chain at positions 25 and 27 are important for productive GroES-GroEL interaction. Consistent with this finding, residues at these two positions in the GroES sequences from bacteria to human are mostly Ile and Leu and sometimes Met 1 .
One-, two-and three I25A or I25D mutated GroES modules in GroES 7 gradually decreased GroEL-GroES 7 interaction. The drastic mutational effect on abolishing GroEL-GroES interaction can be explained by the amplification effect that one mutation in groES affects all seven subunits in GroES. To control GroES's affinity for GroEL in a systematic manner, we created a gene groES 7 in our previous study 31 . groES 7 links seven copies of groES to express a continuous polypeptide GroES 7 with seven GroES modules, allowing us to mutate specific residue(s) at desired GroES module(s) in GroES 7 to create combinations of the mutated and wild type GroES modules. We have shown that mutations of either I25D or L27D in one (1 st ), two (1 st and 4 th ) and three (1 st , 4 th and 7 th ) GroES modules in GroES 7 gradually decrease GroEL-GroES 7 interaction and steadily relieve the strong inhibition on ATPase activity of GroEL SR . Since I25D mutation displays greater mutational effect than L27D mutation 31 , we focused the current study on investigating the I25 mutational effects in GroES 7 . We generated extensive GroES 7 variants with two-or three-I25D or I25A GroES modules. There are three unique ways to place two mutated GroES modules, so we had all six two-mutated variants, GroESI25A 1,2 , GroESI25A 1,3 , GroESI25A 1,4 , GroESI25D 1,2 , GroESI25D 1,3 and GroESI25D 1,4 . We generated four variants with three mutated GroES modules: GroESI25A 1,4,6 , GroESI25A 1,4,7 , GroESI25D 1,4,6 and GroESI25D 1,4,7 .
As the number of either the I25A or I25D modules increased in GroES 7 , the GroEL-GroES 7 interactions decreased. For the I25A series, one mutated module, GroESI25A 1 , inhibited ATPase of GroEL to a level (53.2%) higher than that of GroES (42.5%). Two-mutated modules, GroESI25A 1,2 , GroESI25A 1,3 and GroESI25A 1,4 , had markedly reduced inhibitions with the remaining ATPase activities of 58.5-62.8%. Three-mutated modules, GroESI25A 1,4 , 6 and GroESI25A 1,4 , 7 , further relived the inhibition with the remaining ATPase activity of 79.0-82.9% ( Fig. 2A and Supplementary Table S1). As expected, binding affinity of GroES 7 for GroEL decreased as the number of the mutated module increased, with one-mutated module only moderately affecting affinity, two-mutated modules reducing the affinity by two folds, and three-mutated modules by more than 25 folds ( Fig. 3A and Supplementary Table S1).

Mutational effects on in vivo chaperone function.
We reasoned that the MDH-folding active chaperonin systems should have chaperone function, and examined whether the single-ring GroEL SR -GroES 7 systems were able to substitute the canonical double ring GroEL-GroES in supporting growth via a conditional lethal E. coli strain MGM100 35 . Interestingly, the ability to refold MDH is not correlated with the in vivo chaperone function. For example, of the six GroES 7 variants with two-mutated modules, only GroES 7 I25A 1,3 and GroES 7 I25D 1,4 were able to function with GroEL SR at both the optimal temperature of 37 °C and under heat shock temperature of 42 °C (Fig. 5). The three GroES 7 variants with two-mutated modules, GroES 7 I25A 1,2 , GroES 7 I25A 1,4 and GroES 7 I25D 1,2 , might partially function with GroEL SR at 37 °C, but they did not function with GroEL SR under heat shock. One GroES 7 variants with two-mutated modules, GroES 7 I25D 1,3 , did not function even at 37 °C. In addition, all four GroES 7 with three-mutated modules, despite their little activity in MDH folding, functioned with GroEL SR at 37 °C; moreover, one of them, GroES 7 I25A 1,4,7 -GroEL SR , was functional also at 42 °C. Finally, GroESI25A, with all seven mutated subunits, was functional with GroEL SR at 37 °C, despite its inability to interact with GroEL SR based on ATPase and MST assays and to refold MDH. The reason for the lack of correlation between MDH folding activity and in vivo chaperone function is not clear, however, it is noted that MDH is not the authentic cellular substrate for GroEL-GroES although MDH folding assay is commonly used in the chaperone field. Nevertheless, we identified three GroES 7 variants, GroES 7 I25A 1,3 , GroES 7 I25D 1,4 and GroES 7 I25A 1,4,7 , to function with the single-ring GroEL SR in supporting cell growth under both the optimal and heat shock conditions. These GroES 7 variants have mutations on the interface with GroEL that directly weaken the GroEL SR -GroES interaction, providing the molecular basis for functional single-ring chaperonin system.

Discussion
The chaperonin system is essential for cellular viability by mediating folding of cellular proteins. The double-ring assembly of bacterial GroEL is required for the chaperone function, because the trans-ring allostery is required  Supplementary Tables S1 and S2. to dissociate the stably formed GroEL-GroES complex and to release the enclosed folding substrate protein. The human mitochondrial mtHsp60 may adopt a distinct single-ring mechanism because mtHsp60 exists as a single heptameric ring and has a lower affinity for mtHsp10. A recent model for mtHsp60-mtHsp10 suggests, however, that during the mtHsp60-mtHsp10 reaction cycle two mtHsp60-mtHsp10 complexes associate to form a football shape (mtHsp60-mtHsp10) 2 , suggesting that mtHsp60-mtHsp10 may not truly function in a single-ring mechanism. We sought to show that the chaperonin system may rely solely on the single-ring mechanism to execute the chaperone function, by activating a single-ring form of GroEL, GroEL SR .
GroEL SR is not functional with GroES because without the allostery from the absent second ring the tight GroEL SR -GroES interaction traps folding protein substrates and arrests the chaperone reaction cycle. To obtain functional single-ring GroEL SR -GroES by selectively weakening GroEL SR -GroES interaction in a systematic manner, we utilized a novel reagent groES 7 , that links seven groES to express GroES 7 with seven genetically independent GroES modules. We created extensive GroES 7 variants with one, two and three modules of either GroESI25A or GroESI25D mutations. We systematically characterized mutational effect on various activities of GroEL and GroEL SR . We found that as the number of the mutated modules increased the inhibition on ATPase activity, the binding affinity and MDH folding activity of GroEL steadily decreased, suggesting that gradual decrease in GroEL-GroES 7 interaction. Decreases in inhibiting ATPase activity of and in binding affinity for GroEL SR were greater than as seen in GroEL, and suggested that GroES 7 variants with mutated modules resumed a recyclable reaction with the single ring GroEL SR . Notably in mediating MDH folding, GroES 7 variants with two mutated modules were active with GroEL SR with both the folding yield and kinetics comparable to the canonical  Supplementary Tables S1 and S2. double ring GroEL-GroES. Importantly, we found three GroES 7 variants, GroES 7 I25A 1,3 , GroES 7 I25D 1,4 and GroES 7 I25A 1,4,7 , were functional with GroEL SR under both normal and heat shock temperatures.
The chaperonin-cochaperonin interaction is central for chaperonin to function as single ring. Early genetic screens isolated GroEL SR variants that are functional with GroES at 37 °C 27 , and the chaperonin-cochaperonin  A and B) and GroEL SR -mediated MDH folding (C and D). The enzymatic activity of native MDH is set to 100%. Experiments were repeated more than three times, and representative data from individual runs were shown. The MDH yields are summarized in Supplementary Tables S1 and S2. interaction in these functional GroEL SR -GroES systems is much weaker compared to GroEL-GroES 27,29,36 . Since these mutated GroEL SR residues are not located in the GroEL-GroES interface, the mutational effects on GroEL SR -GroES interaction are presumably allosteric and molecular basis for the allosteric effect remains unclear. Direct mutations on the GroEL SR -GroES interface, G24/I25/V26/L27, in the GroES mobile loop, identified GroES variants GroESI25F and GroESI25L that appear functional with GroEL SR at 37 °C 30 . Both variants decreased inhibition on the ATPase activity of GroEL SR , suggesting their reduced interaction with GroEL SR ; however, no further characterizations on the GroEL SR -GroES interaction have been reported. Our abilities to directly modulate the chaperonin-cochaperonin interface, shown in this and previous 31 studies, confirm that reduced chaperonin-cochaperonin interaction is key to create functional single ring. We found that modifying two or three of the seven individual GroEL-GroES interactive surfaces is effective in rendering single ring GroEL SR -GroES functional in vivo. Positions of the modified individual interfaces, 1,2, 1,3 and 1,4 or 1,4,6 and 1,4,7, have different effects on functionality of GroEL SR -GroES. These findings support the structural observations that each of the GroEL-GroES interfaces, including conformations of both the GroES mobile loop and the GroEL Helix H and I, is unique 37 . In terms of interaction strength, we found that the working chaperonin-cochaperonin interaction for a functional single ring GroEL-GroES-based system follows the Goldilocks principle: interaction must not be too loose or too tight. Our studies provide the first step for future mechanistic investigations on the Goldilocks chaperonin-cochaperonin interaction of the single-ring chaperonin system.
Our results that the chaperonin system may rely on the single-ring mechanism are informative to the human mitochondrial chaperonin mtHsp60-mtHsp10. mtHsp60 exists predominately as single heptameric ring 12 in equilibrium with the monomeric form 16 . The lack of the double ring conformation is consistent with its absence of the two conserved salt bridges (K105-D435 and E461-R452; residue naming according to GroEL) that are important to stabilize the inter-ring interaction 38 . In addition, compared to the stable GroEL-GroES complex (K d of 0.1-26 nM 20, 33, 34, 39 , or 3.83 ± 0.93 nM of this study, in the presence of ADP), the reduced mtHsp60-mtHsp10 interaction 14 supports the dispensable role of a second ring in the chaperoning reaction cycle. Further support for mtHsp60-mtHsp10 functioning in a single ring mechanism comes from the functional single ring GroEL SR /mtHsp60 chimera 14,15 . Interestingly, in the presence of both ATP and mtHsp10 two mtHsp60 heptameric rings appear to associate, forming the football (mtHsp60-mtHsp10) 2 conformation 16 . Investigations on whether mtHsp60 undergoes an association to form a double ring conformation in the mtHsp60-mtHsp10 reaction cycle are hindered by the dynamic nature of mtHsp60 quaternary assembly and mtHsp60-mtHsp10 interaction. Genetic screens identified a mutant mtHsp60 E321K with high affinity for mtHsp10, forming stable mtHsp60 E321K -mtHsp10 and arresting the chaperone cycle 40 , reminiscent of GroEL SR arresting GroEL-GroES cycle. mtHsp60 E321K -mtHsp10 crystalized in the football conformation 17,18 , that is, two heptameric mtHs-p60 E321K -mtHsp10 complexes associate via mtHsp60 E321K . The two mtHsp60 E321K heptameric rings interface via the equatorial domains as seen in GroEL, and as expected no charge-charge interactions in the place of the two conserved inter-ring salt bridges (K105-D435 and E461-R452) are observed. Strikingly, the inter-ring interface in mtHsp60 E321K is twice as that in the naturally occurring double-ring GroEL. Such extensive inter-ring interface suggests a stable, GroEL-like double ring conformation, which is in direct contrast to the observed, single-ring conformation. Such extensive inter-ring interface may suggest cross-ring communication and regulation, justifying the assembly of the double ring conformation for biochemical activities. For example, the ATP-induced cross-ring allostery manifests in various aspects in GroEL-GroES. Notably, binding of ATP to one GroEL ring prevents ATP binding to the opposite ring 41 , and ATP binding in one ring initiates GroES dissociation from the opposite GroEL ring 21 . For mtHsp60-mtHsp10, the negative ATP binding cooperativity has not been reported, and mtHsp10 dissociates readily from mtHsp60 due to the weak interaction. Besides the lack of biochemical support, structure of (mtHsp60 E321K -mtHsp10) 2 does not offer structural insights into either cross-ring communication or the double ring assembly of the football conformation important for the mtHsp60-mtHsp10 reaction cycle. Thus, the mechanistic significance for association of two mtHsp60-mtHsp10 to form a football conformation of (mtHsp60-mtHsp10) 2 is not clear, and whether the football conformation is the productive intermediate in the chaperone cycle is unknown. However, considering the complex cellular conditions, it is probable that two heptameric mtHsp60-mtHsp10 (mtHsp60) molecules might associate to form the double ring assembly as seen in structure of (mtHsp60 E321K -mtHsp10) 2 . The cellular conditions favorable for molecular association include the abundance of cellular chaperonin (2.6 μM for GorEL 42 ), the high concentration of cellular macromolecules (300-400 mg/ml in E. coli 43 ) and the macromolecular crowding effect 43 that results in increasing the effective concentration of mtHsp60. While investigations on these important mechanistic aspects of mtHsp60-mtHsp10 continue, here we, in conjunction with previous studies 14,15,[27][28][29] , show that the chapreonin can rely on the single-ring mechanism to function. Our results demonstrate the mechanistic adaptability of the chaperonin system, and our functional single ring GroEL SR -GroES 7 variants will provide valuable tools to study the molecular evolution of this ancient protein family from bacterial double-ring to human mitochondrial single-ring conformations.

Methods
Protein expression and purification. groEL and groEL SR (GroEL R452A/E461A/S463A/V464A) were in pTrc vector, groES was in pET3b, and groES 7 and the groES 7 variants were in a modified pET28b 31 . E. coli BL21(DE3) cells were used to express the proteins. Conditions for cell growth, induction of protein expression, and protein purification are described in ref. 31. To remove the residual proteins bound to GroEL or GroEL SR , the chaperonins (1 mg/ml) were dialyzed against 50 mM TrisCl pH 7.5, 1 mM EDTA and 30% methanol, loaded onto a FastQ column (GE Healthcare), and eluted with 0-1 M NaCl gradient. The chaperonin-containing fractions were combined, dialyzed with TEA buffer (50 mM triethanolamine 7.5, 50 mM KCl and 20 mM MgCl 2 ) and 0.1% NaN 3 at 4 °C overnight. The purified chaperonins were verified with minimal Trp fluorescence.
ATPase activity assays via Malachite green. Chaperonins and cochaperonins were dialyzed into TEA reaction buffer containing 50 mM KCl and 20 mM MgCl 2 , to 0.125 μM tetradecameric chaperonins, and 0.3 μM heptameric cochaperonins. ATPase activity was measured via malachite green as described in ref. 31 at room temperature (22 °C) with 2 mM ATP as the starting concentration. Absorption at 660 nm (A 660 ) was measured, and the final A 660 values were averaged over three readings. The amount of hydrolyzed free phosphate was derived from a standard curve, and the hydrolysis rate was normalized to GroEL monomer. At least three independent experiments were performed. MDH refolding assay. Chaperonins and cochaperonins were dialyzed into TEA reaction buffer. Malate dehydrogenase (Roche) was unfolded in TEA buffer including 3 M GdmHCl to a final concentration of 36.7 μM (monomeric MDH) for 60 minutes prior to the experiments. MDH refolding assay via monitoring the enzymatic activity of the refolded MDH at A 340 , was described in ref. 31. The final protein concentrations were 1 μM of GroEL or 2 μM GroEL SR , 4 μM of cochaperonin, and 0.7 μM of monomeric MDH. The enzymatic activity of native MDH was set to 100%, and at least three independent experiments were performed.
Chaperonin-cochaperonin binding via microscale thermophoresis (MST) assay. GroES, GroES 7 and GroES 7 variants were fluorescently labeled with DyLight TM 650 NHS Ester Amine Reactive Dye (ThermoScientific) according to manufacturer's protocol. The labeled chaperonin was separated from the free dye using MidiTrap (GE Healthcare) followed by dialysis (to 50 mM TrisCl pH 7.5, 100 mM KCl, 10 mM MgCl 2 , and 1 mM EDTA), and its concentration was measured using the Bradford assay. For each unlabeled proteins (GroEL or GroEL SR ), a serial dilution of 15 samples were prepared in the binding buffer (50 mM TrisCl pH 7.5, 100 mM KCl, 10 mM MgCl 2 , 1 mM EDTA, 2 mM ADP, and 0.5 mg/mL BSA). 10 ul of the unlabeled protein was incubated with 10 ul of the labeled cochaperonin for 30 min, and the solution was loaded into a glass capillary (NanoTemper Technologies) for MST measurements. The thermophoresis measurements were carried out using NanoTemper Monolith NT115 (NanoTemper Technologies) with 80% LED power and 40% IR-Laser power. At least three independent experiments were performed. Initial MST data were processed using Monolith NT115, and dissociation constant (K d ) was determined using KalidaGraph by fitting the following equation: where m1 is the thermophoresis reading of the labeled cochaperonin in the absence of the unlabeled titrating protein, m2 is the thermophoresis reading when all the labeled cochaperonin was bound with the unlabeled titrating protein, and m3 is the K d .
In vivo complementation assay. The MGM100 E. coli cell strain (kanamycin resistant, Kan R ) was obtained from the E. coli Genetic Stock Center at Yale University. pTrc is a lac promoter-based expression vector; the lac-based vector pBbE5c 44 was used to express GroES, GroES 7 and GroES 7 variants. CaCl 2 competent MGM100 cells were co-transformed with both plasmids and plated onto LB agar containing 50 μg/mL kanamycin, 100 μg/ mL ampicillin, 50 μg/mL chloramphenicol, and 0.2% w/v arabinose. Conditions for cell growth and titration are described in ref. 29.
Data availability statement. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.