The Molecular Chaperone Hsc70 Assists the in VitroFolding of the N-terminal Nucleotide-binding Domain of the Cystic Fibrosis Transmembrane Conductance Regulator*

The most common disease-causing mutation in the cystic fibrosis transmembrane conductance regulator is a single amino acid deletion (ΔF508) in the N-terminal cytosolic nucleotide-binding domain (NBD1). This mutation has previously been shown to be a temperature-sensitive folding mutation that alters the folding pathway but not the native state stability of the isolated domain (Qu, B.-H., and Thomas, P. J. (1996) J. Biol. Chem. 271, 7261–7264). Here we provide evidence that the molecular chaperone Hsc70 productively interacts with NBD1 to increase the folding yield of the domain and inhibit off-pathway associations leading to the formation of high molecular weight aggregates. Furthermore, we have sublocalized a region within NBD1 where Hsc70 binds. Notably, inhibition of NBD1 aggregation is not dependent upon the presence of Hsc70 in the early stages of folding, indicating that the chaperone may act on a folding intermediate. In the presence of K+ and Mg2+-ATP, conditions where Hsp70 binds substrate rapidly and can release it, Hsc70 is less effective at inhibiting NBD1 aggregation. Thus, the rate of release of unfolded substrate is an important factor in preventing aggregation and promoting folding of the domain. These results demonstrate that Hsc70 promotes the otherwise inefficient folding of ΔF-NBD1 and provide insight into the mechanisms by which molecular chaperones assist proteins in folding.

Correct protein folding is an essential biological process. When proteins do not obtain correct native structure, their normal function is either impaired or absent. Not surprisingly, mutations that result in misfolded proteins have been implicated in a growing number of diseases (1). To assure their correct folding in the cellular milieu many proteins interact with molecular chaperones during the process of folding (2).
The Hsp70 class of molecular chaperones is thought to bind early in the folding process to extended conformations of polypeptide chains (3) with a preference for hydrophobic sequences (4) and to maintain the polypeptides in a soluble conformation competent for folding upon release. The cycle of ATP binding and hydrolysis on Hsp70 is coupled to substrate binding and release by conformational changes in the chaperone that represent interdomain communication between the substrate-binding and ATPase domains of Hsp70 (5)(6)(7). The ADPbound form of Hsp70 has a tight affinity for substrate and a slow rate of substrate release (8). K ϩ and Mg 2ϩ -ATP are required for release of substrate from Hsc70 (9,10). In effect, substrate turnover occurs in the presence of K ϩ and Mg 2ϩ -ATP with likely consequences for the mechanism by which Hsp70s promote protein folding. Upon release from the sequestration of the chaperone, polypeptides may either fold to the native conformation or enter the degradation machinery of the cell. Chaperones may play an active role in directing misfolded or mutant proteins to proteolysis. Hsp70, for example, has recently been shown to be required for ubiquitination of some proteins (11), the first step toward degradation by the proteasome (12).
Defective folding and maturation of the most common mutant form of CFTR, 1 deletion of phenylalanine 508 (⌬F508), underlie the pathogenesis of most cases of cystic fibrosis (CF) (13). Deletion of phenylalanine 508 in the 22-kDa N-terminal nucleotide-binding domain (NBD1) alters the folding of the domain in vitro (14,15). Thus, altered folding is the basis of incorrect glycosylation of CFTR, retention in the endoplasmic reticulum, lack of transit to the plasma membrane, and degradation in a proteasome-dependent manner observed in the cell (16 -18). The constitutively expressed Hsp70 isoform, Hsc70, has been shown to interact with immature wild type and ⌬F508 CFTR in vivo; however, the mutant protein has an extended interaction with Hsc70 (19). Understanding how Hsc70 influences the partitioning of wild type and ⌬F508 CFTR proteins into folding and degradation pathways will provide insight into the mechanisms of intracellular quality control.
In the present study we utilize an established in vitro NBD1 folding system that recapitulates the effects of several CFassociated mutants, including ⌬F508 (15,20). Notably, the ⌬F508 mutation produces a kinetic defect in CFTR-NBD1 folding. This in vitro system may be manipulated to influence the rate(s) of folding and the proportion of material destined for either off-pathway aggregation or folding to the native state.
Here we use the in vitro folding of CFTR NBD1 to demonstrate that interactions with Hsc70 promote folding.

MATERIALS AND METHODS
Expression and Purification of NBD1-Six-histidine-tagged wild type and ⌬F508 mutant CFTR-NBD1 from residues Gly 404 to Ser 589 were expressed in Escherichia coli and purified by Ni ϩ -resin affinity chromatography as described previously (15). The concentration of NBD1 solubilized in 6 M GdnHCl was calculated from the absorbance at 280 nm using a molar extinction coefficient of 13,490 M Ϫ1 cm Ϫ1 as determined in Ref. 15. Once resuspended in 6 M GdnHCl, protein was stored at 4°C and used within a week.
Hsc70 Purification-Hsc70 was purified from bovine brain as de-* This work was supported by National Institutes of Health Grant NIDDK49835 and Welch Foundation Grant I-1284. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Gel Shift Assay-A gel shift competition assay for peptide binding to Hsc70 was performed as described previously (22). The G17A peptide was chemically synthesized to mimic CFTR residues Gly 545 -Ala 561 (UT Southwestern/HHMI core facility). Reduced carboxymethylated lactalbumin (RCMLA), which is unable to reach the native state due to carboxymethylation of cysteine residues, was obtained from Sigma. Purified Hsc70 (1.4 M) was incubated for 2 h at 37°C with 0, 10, 50, 250, or 500 M G17A peptide and 300 g/ml RCMLA, which is an approximate 2-fold excess of RCMLA over Hsc70. Samples were subjected to 6% native gel electrophoresis as described in Ref. 22 and stained with Coomassie Blue.
Aggregation-Aggregation assays to monitor the formation of offpathway species were initiated by rapid dilution of NBD1 in 6 M GdnHCl to a final concentration of 2 M into a pre-warmed folding mix (Buffer R). Final concentrations in Buffer R were 100 mM Tris-HCl, pH 7.4, 0.385 M L-arginine-HCl, 2 mM EDTA, 10 mM dithiothreitol, and 25 mM GdnHCl. Aggregation was followed over time at 37°C by turbidity measured at 400 nm in a UV spectrophotometer. In experiments that contained Hsc70, a final concentration of 1 M Hsc70 in Buffer C was added either before initiating folding or 400 s after the initiation of folding. In experiments containing ATP, the following were present in Buffer R: 20 mM MgCl 2 , 200 mM KCl, and 50 M ATP (diluted from a stock adjusted to pH 7.4 with NaOH). Data are shown as dots representing single data points and a line fitted to the data from a logistic equation.
Folding Yield-The folding yield of 2 M ⌬F-NBD1 in the absence and the presence of 1 M Hsc70 was determined by diluting ⌬F-NBD1 into Buffer R pre-equilibrated to 4, 10, 20, 25, 37, or 45°C. After an overnight incubation at the designated temperature, the samples were centrifuged at 16,000 ϫ g for 15 min to pellet insoluble material, and the supernatant was subjected to Tricine SDS-polyacrylamide gel electrophoresis (16.5% gel) and stained with Coomassie Blue (23).
Kinetic Folding Experiment-Denatured ⌬F-NBD1 in 6 M GdnHCl was diluted into either 6 M GdnHCl or Buffer R at a final protein concentration of 2 M. Folding was monitored by the increase in fluorescence as the single tryptophan at position 496 in the domain enters a more hydrophobic environment during the folding process. Fluorescence measurements were made at excitation and emission wavelengths of 282 and 324 nm, respectively, on a fluorometer with band passes of 2 and 4 nm. Approximately 15 s passed between the time folding was initiated and the time data collection was begun.

RESULTS AND DISCUSSION
Binding of Hsc70 to CFTR-Although full-length CFTR has been shown to interact with Hsc70 in a tissue culture system by co-immunoprecipitation experiments (19), the location of the binding site(s) for Hsc70 to CFTR have not been identified. Using a peptide representing a portion of CFTR we localized a binding site for Hsc70 within the NBD1 by a gel shift competition assay (Fig. 1). Hsc70 was chosen as the representative Hsp70 family member for these experiments because it is located in the cytosol (where the NBD1 presumably folds and is topologically accessible), is constitutively expressed, and binds to nascent polypeptide chains (24,25).
Hsc70 purified from bovine brain was allowed to form a complex with a stable, unfolded substrate, RCMLA. Once the complex was formed, it was subjected to native gel electrophoresis (Fig. 1). Hsc70 in complex with RCMLA (lane 3) has a shifted mobility compared with the mobility of free Hsc70 (lane 2). Notably, G17A, a peptide fragment from NBD1 corresponding to residues Gly 545 -Ala 561 , was able to compete with RCMLA for binding to Hsc70. Not all peptides compete for Hsc70 binding in this gel shift assay (22). G17A is homologous to a region of p53 known to bind Hsc70 (26) and includes several CF-associated mutations that affect maturation (27,28) and folding (20). The peptide-Hsc70 complex runs at a size consistent with free Hsc70, and the free peptide does not resolve in the gel. The G17A peptide also stimulates the substratedependent ATPase activity of Hsc70 at 50 M peptide (data not shown). It is important to note that additional Hsc70-binding sites may exist within NBD1. The localization of a binding region for Hsc70 within NBD1 is significant because the domain contains the temperature-sensitive folding mutation ⌬F508 as well as other known folding mutations (27).
Aggregation and Folding of NBD1 in the Presence of Hsc70 -Under conditions of low folding yield, NBD1 does not fold and aggregates into high molecular weight species that scatter light (15). Hsc70 inhibits ⌬F-NBD1 aggregation at a substoichiomet-  (4, 10, 20, 25, 37, or 45°C). Material in solution after centrifugation was separated on a Tricine SDS-gel stained with Coomassie Blue. The asterisk represents actin, which co-purified with Hsc70. ric ratio of 2:1 ⌬F-NBD1:Hsc70 (Fig. 2A). Not surprisingly, considering that wild type CFTR interacts with Hsp70 in vivo, under conditions where folding is inefficient, Hsc70 also inhibits the aggregation of wild type NBD1 (data not shown).
By inhibiting the aggregation of ⌬F-NBD1, Hsc70 increases the amount of soluble protein formed during an overnight folding yield experiment performed at various temperatures as shown in Fig. 2B. The fraction of NBD1 that is soluble under these conditions has previously been shown to be functional by binding nucleotide (15,20). At increased temperatures the folding yield is lowered as the amount of aggregation increases. In the presence of Hsc70 the amount of ⌬F-NBD1 in the soluble fraction increases. A large fraction of the total ⌬F-NBD1 is soluble, implying that solubility of the domain is not merely a consequence of complex formation with Hsc70.
Folding Kinetics-The intrinsic fluorescence of a single tryptophan residue in NBD1 (Trp 496 ) can be utilized to monitor folding (Fig. 3A). As NBD1 folds, Trp 496 enters a more hydrophobic environment and exhibits an increase in fluorescence (15). The change in fluorescence over time after the initiation of folding exhibited at least two phases. First, there is a burst phase that occurs during mixing and may represent the initial hydrophobic collapse of the domain. Second, there is a slower phase of fluorescence increase that occurs within 150 s. Both of these phases are complete prior to the time that aggregation can be observed by turbidity (compare with Fig. 2A). These kinetic data are consistent with the existence of at least one intermediate on a sequential pathway or multiple denatured states.
An unresolved issue in our understanding of the action of molecular chaperones is the conformation of their polypeptide substrate. Although Hsp70s are known to inhibit the aggregation of many proteins, it is unclear whether they bind to fully denatured polypeptides, partially structured intermediates, or folded domains. Here we show that a substoichiometric amount of Hsc70 inhibits aggregation when present at the initiation of folding/aggregation. If, however, Hsc70 is added either 60 (data not shown) or 400 s after initiation of folding, well after the burst phase has concluded, aggregation continues to be inhibited (Fig. 3B). Because Hsc70 is equally effective at inhibiting aggregation when added either before or after the burst phase, it most likely acts on a partially folded structure or on a conformer in equilibrium with the aggregation-prone species during folding. In addition, these results indicate that there may be a further slow phase in folding that is not reported by intrinsic fluorescence.
Aggregation in the Presence of Hsc70 and ATP-All of the previous aggregation experiments were performed in the absence of Mg 2ϩ -ATP. In the presence of K ϩ and Mg 2ϩ -ATP, conditions that promote binding and rapid release of substrate (8 -10), aggregation is no longer effectively inhibited by the chaperone (Fig. 4). ATP has been shown to have similar effects on the ability of Hsc70 to block the aggregation of citrate synthase (29). It is important to note that although NBD1 also binds ATP, this binding has no effect on the folding yield or FIG. 3. Folding kinetics of ⌬F-NBD1. A, to follow the folding process kinetically, ⌬F-NBD1 was diluted into either 6 M GdnHCl or Buffer R at 37°C to a final concentration of 2 M. The increase in intrinsic fluorescence was monitored over time with excitation and emission wavelengths of 282 and 324 nm, respectively. The burst phase in the 15 s between initiation of folding and data collection is indicated by the difference in fluorescence between dilution into 6 M GdnHCl and dilution into Buffer R. Each dot represents an individual data point. B, to investigate the effect of Hsc70 on inhibiting aggregation at various phases in the folding process, 1 M Hsc70 was added to 2 M ⌬F-NBD1 either before initiating folding (short dashed line) or 400 s after initiation (long dashed line). In both cases aggregation was inhibited as compared with the aggregation of 2 M ⌬F-NBD1 with no Hsc70 (solid line). Aggregation was performed, and the results are presented as in Fig. 2A.  (20). Therefore, the rate of release of an intermediate prone to aggregation may be critical for effective inhibition of aggregation by Hsc70. Premature release of NBD1 may result in an accumulation of the partially structured folding intermediate(s) susceptible to intermolecular interactions and, ultimately, accelerated higher order processes of aggregation or proteolysis. Thus, the kinetics of chaperone binding and protein folding must be in balance. Alteration of that balance by a kinetic folding mutation would result in a reduction in folding efficiency and, perhaps, an increased rate of quality control processes such as ubiquitination and subsequent proteolysis. Whereas other chaperones and cofactors such as Hsp40 regulate the turnover of Hsc70 substrates, they may have an important role to play in this process.
Correcting the folding defect of CFTR has promise as a treatment for the disease because ⌬F-CFTR forms a functional channel in cell culture systems under conditions that promote folding, such as lower growth temperature(s) or growth in 10% glycerol (30 -32). The results here indicate that careful alteration of the kinetics of Hsc70 interaction with CFTR may allow some of the mutant protein to reach a native conformation.