INTRODUCTION

TF,¹ the product of the tig gene, is an abundant soluble protein in the Escherichia coli cytosol with a variety of intriguing properties. TF was initially purified by Wickner and Crooke (1) because of its ability to bind to certain secreted polypeptides and to facilitate their transport into membrane vesicles (1, 2). However, subsequent studies in which the cellular content of TF was increased or reduced failed to support a role for TF in translocation of these proteins in vivo (3). TF was also found to be associated with the 50 S ribosome (4), and recent studies have shown that TF binds to many nascent polypeptides on the ribosome (5, 6). Moreover, Fischer and co-workers recently showed that TF possesses peptidyl-prolyl isomerase activity on peptides and proteins (7), and one portion of TF is homologous to the active site region of the FK-506 binding protein class of peptide prolyl isomerase (8). Thus, TF may play an important role in assisting the proper folding of the newly synthesized polypeptides.

Our recent studies have shown that TF in E. coli also plays a critical role in the rapid degradation of the fusion protein, CRAG (9). This process also requires the chaperones, GroEL and GroES (10). CRAG and proteolytic fragments of this short-lived polypeptide were found tightly associated with GroEL and TF in complexes containing one TF molecule per GroEL 14-mer and a CRAG fragment. Moreover, increased TF expression greatly accelerated CRAG degradation, whereas decreased TF content slowed this process (9). These findings suggested that a rate-limiting step in CRAG degradation is the formation of these ternary complexes containing CRAG, GroEL, and TF (9). Possibly the association of TF with GroEL and a polypeptide represents a "degradative complex" that specifically favors the proteolytic digestion of the substrates. Alternatively, TF binding to GroEL might simply enhance GroEL's capacity to associate with abnormal proteins. In the case of CRAG or CRAG fragments, this association would favor proteolysis, whereas with other polypeptides, it might promote proper folding or translocation. In support of this latter explanation, we found that increased TF expression led to a greater capacity of GroEL to bind in vitro to CRAG and also to denatured fetuin and histone (9).

These observations suggest that in vivo TF associates with GroEL and enhances its capacity to bind to unfolded proteins. The present studies were undertaken to learn more about this effect of TF and to understand how it may influence the ability of GroEL to associate with unfolded polypeptides. We have investigated 1) whether TF associates with GroEL and thus directly enhances the chaperone's binding capacity, 2) whether these two components remain together upon dissociation from protein substrates, 3) whether TF affects GroEL binding to all proteins similarly, and 4) whether TF functions by promoting GroEL phosphorylation, because prior studies have shown that upon heat shock, a fraction of GroEL is phosphorylated, and this modification enhances its capacity to bind to certain unfolded proteins (11, 12).

EXPERIMENTAL PROCEDURES

Cell proteins were labeled with [³⁵S]methionine, and extracts were prepared as described previously (10). Strains expressing TF at very high or very low levels were kindly provided by W. Wickner (Dartmouth College) and were grown as described previously (3, 9). Protein substrates immobilized on agarose were purchased from Sigma. In vitro binding of GroEL to immobilized denatured proteins was studied as described previously (9, 12).

C600 E. coli cells transformed with pDK84 plasmid kindly provided by Mark Snavely (Amgen Corp.), which carries the groELS operon under the control of the tac promotor, were grown at 30 °C until mid-log phase, and then isopropyl-1-thio--D-galactopyranoside was added (at 1 mM) to induce groELS expression. After 2 h, cells were collected by centrifugation, and extracts were prepared. The ribosomal fraction was removed by ultracentrifugation at 4 °C at 100,000 × g for 1 h, and the GroEL-containing fraction was collected by ultracentrifugation at 200,000 × g for 2 h. The pellet was washed twice with buffer A (20 mM Tris, pH 8.0, 150 mM NaCl, 1 mM dithiothreitol), resuspended in 0.5 ml of buffer A, and loaded onto a 5-ml linear sucrose gradient (10-30% sucrose in buffer A). After 2 h of centrifugation at 250,000 × g, fractions from the 20 S zone, containing GroEL, were collected, combined, and used either for binding studies or further purification on an ATP-agarose column (0.5 ml). The flow through of the ATP-agarose column contained GroEL, which was at least 95% pure as determined by SDS-PAGE (see Fig. 3A).

TF was overexpressed, and cell extracts were prepared (10). TF was isolated from ribosomes as described previously (4) and then purified further by column chromatography as described by Fischer and co-workers (7). This preparation of TF was more than 98% pure as judged by SDS-PAGE (see Fig. 3A).

RESULTS

We previously showed that a fraction of the cell's GroEL can bind to affinity columns containing certain denatured proteins and can dissociate from them upon ATP addition (12). The binding capacity of GroEL to these proteins was greatly enhanced by overproduction of TF (9). To define whether TF associates with GroEL in the cell and thus influences its capacity to bind to denatured proteins, we tested whether GroEL from wild-type cells binds to unfolded proteins together with TF. Extracts of cells grown in LB at 35 °C were loaded onto an affinity column with denatured fetuin. After extensive washing with buffer, the bound chaperones were eluted with ATP, and the proteins that remained on the column were eluted with acetic acid, pH 2.5 (11, 12). More than 95% of the GroEL and TF that bound were eluted with ATP, as assayed by a Western blot with anti-TF and anti-GroEL antibodies (Fig. 1A). These data suggest that TF and GroEL from the extract bound together to the fetuin column and possibly dissociated together.

To test if GroEL and TF dissociated from the column as a complex, the ATP-eluted fraction was analyzed by gel filtration on a Superose 6 column. The major peak of TF coincided with the peak of GroEL, which eluted as the 14-mer at about 800 kDa (Fig. 1B). Thus, TF and intact GroEL remained associated with each other after elution from the fetuin column. They remained bound to one another, even when the ATP in the eluate was destroyed by treatment with apyrase or when ATP was omitted during gel filtration. In addition, there was a minor TF peak in a low molecular weight fraction, which presumably corresponds to TF monomers. Because the cells were labeled uniformly and because the number of methionine and cysteine residues in these proteins is known, we calculated that the ratio of TF to GroEL in this complex was approximately 1 TF monomer per 1 GroEL 14-mer. These data resemble our previous finding that the calculated molar ratio of proteins in GroEL-TF-CRAG complex formed in vivo was also 1:1:1 (9).

It appears very unlikely that GroEL associates with TF as an unfolded polypeptide substrate, because TF remained associated with GroEL in the presence of ATP, which causes dissociation of GroEL-substrate complexes (14). Furthermore, the addition of an excess of GroES together with ATP did not cause the dissociation of GroEL-TF complex (not shown), although it causes substrate dissociation from GroEL (13-15). In addition, as shown below, the association of GroEL with TF enhanced GroEL's binding to other polypeptide substrates.

The dissociation of the GroEL-TF complex from these proteins did not require exogenous GroES (Fig. 1A). To test whether GroES may be involved in the binding of GroEL-TF to fetuin, we prepared extracts in the presence of 1 mM ADP without freezing or cooling the cells to maintain GroEL in association with GroES (14, 15) and immediately loaded the extract onto the column. However, we were unable under any condition to detect GroES by Western blot either in the ATP wash or acid wash from the fetuin column (not shown), indicating that GroES does not participate in the binding of TF and GroEL.

Experiments were carried out to test whether the GroEL-TF complex binds to denatured fetuin better than GroEL alone or whether TF overproduction promotes GroEL binding by an indirect mechanism. GroEL was isolated from strains that overexpress or underexpress TF by sucrose gradient centrifugation, as described above. Large amounts of TF were associated with GroEL isolated from cells that overexpress TF (Fig. 2A). In contrast, almost no TF was found in the GroEL fraction from cells containing low levels of TF (Fig. 2A). Similar amounts of GroEL isolated from these two extracts were then loaded onto the fetuin column, eluted with ATP, and quantified by Western blot (Fig. 2B). Only 0.5% of the GroEL isolated from the TF-deficient extract bound to the column. By contrast, in the TF overexpressor strain, 20-fold more GroEL (8% of GroEL loaded on the column) bound to the fetuin, which correlated with much larger amounts of TF associated with GroEL in the cells (Fig. 2A). These data suggest that TF initially forms a complex with GroEL, which then binds strongly to the unfolded protein.

To test whether TF by itself can directly promote GroEL binding to proteins, experiments were performed with purified TF and GroEL. GroEL was isolated by either of two approaches described under "Experimental Procedures" and incubated with a 10-fold molar excess of pure TF at room temperatures for 1 h in buffer containing 50 mM Tris-HCl, 10 mM NaCl, 50 mM KCl, 1 mM ATP, and 5 mM MgCl₂. After the addition of 10 mM EDTA to prevent ATP-dependent dissociation of GroEL from the ligand (12), the mixtures were loaded onto fetuin columns. The bound material was eluted and analyzed by Western blot, as described above. As shown in Fig. 3, incubation of partially purified (B) or pure GroEL (C) with TF in vitro stimulated markedly the binding of GroEL to the column.

After purification, GroEL might still contain some unfolded peptides bound within its central cavity that might be involved in GroEL-TF interaction. To rule out this possibility, similar experiments were carried out in the presence of both GroES and ATP, under conditions that cause substrates to dissociate from GroEL. TF stimulated GroEL binding even with GroES present (not shown). Thus other polypeptide substrates are not involved in the enhancement of GroEL binding by TF. Incubation of GroEL with heat-inactivated TF or with bovine serum albumin in control experiments did not affect the binding capacity of GroEL (Fig. 3, B and C). In these experiments some TF also bound to the column as shown by Western blot analysis with an anti-TF antibody, whether or not GroEL was present (not shown). To rule out the possibility that TF first binds to the column and that GroEL then associates with the bound TF, preloading experiments were performed. When TF alone was applied onto the fetuin column prior to loading the GroEL, no increase in GroEL binding occurred (Fig. 3, B and C), in sharp contrast to the findings when GroEL and TF were preincubated together prior to loading. These results indicate that the enhancement of GroEL binding results from a direct interaction with TF prior to association with the substrate.

Recently, we have reported that heat shock leads to phosphorylation of a fraction of the cell's GroEL (11), and this modification enhances the ability of GroEL to bind to CRAG and other unfolded proteins (12). Accordingly, in extracts of wild-type cells grown at 20 °C, only trace amounts of GroEL can bind to unfolded protein columns, but following shift of cells to 42 °C, binding to fetuin increased more than 50-fold (Fig. 4A), and the bound fraction was phosphorylated (12). As a result, after incubation at 42 °C, about 5-8% of the total cellular GroEL became associated with the fetuin column (12). Interestingly, if TF was overexpressed in the cells grown at 20 °C, GroEL binding to the fetuin column also increased more than 50-fold over the control levels (Fig. 4A). However, with cells grown at 35 °C, TF overexpression increased GroEL binding only 4-5-fold (not shown), and only a 1.5-fold increase was seen with cells incubated at 42 °C (Fig. 4A). This smaller effect of TF overexpression on GroEL binding capacity at elevated temperatures is due to increased GroEL binding in the wild-type cells under these conditions. TF expression in wild-type cells did not increase at 37 or 42 °C (data not shown); in fact, at 42 °C, TF synthesis is partially repressed (3).

Thus, increased TF levels at low temperature appear to have generally similar effects on the function of GroEL as GroEL phosphorylation during heat shock, and these two stimuli did not have additive effects at 42 °C. One possible explanation of the effects of TF expression could be that TF promotes GroEL phosphorylation at 20 and 35 °C. To test this possibility, wild-type cells and the TF overexpressor strain were grown at 35 °C, and the extracts were analyzed by two-dimensional electrophoresis (16) and probed with an anti-GroEL antibody. In all extracts, minor spots were observed that are more acidic than the major GroEL spot and correspond to phosphorylated forms (12). However, their relative amounts were not affected by the level of TF expression (Fig. 4B). Thus, heat shock induces GroEL phosphorylation without causing TF induction (3), whereas at low temperatures, TF stimulates the binding capacity of GroEL without causing its phosphorylation.

Because overproduction of TF and heat shock-induced phosphorylation appear to be two independent mechanisms for regulation of the activity of GroEL, we have tested whether these two physiological stimuli affect the binding of GroEL to polypeptides in identical fashions. Extracts were prepared from E. coli overproducing TF grown at 30 °C and from wild-type cells exposed to 42 °C for 30 min. We then compared the capacities of equal amounts of GroEL from such cells and from control cells to bind to a variety of protein ligands. The binding capacity of GroEL in control cells (grown at 30 °C) varied somewhat with the different columns (Table I) (presumably because of the differences in the amounts or in conformations of the bound proteins). In any case, TF overproduction and heat shock were found to affect GroEL binding to these various ligands in qualitatively different ways. Both treatments markedly stimulated binding of GroEL to CRAG, fetuin, and histone, as noted previously (12, 9). However, only TF overproduction stimulated GroEL binding to cytochrome, globin, or thyroglobulin columns, whereas only heat shock enhanced binding to casein or albumin. These findings (Table I) suggest qualitative rather than quantitative differences between the binding properties of GroEL from control, heat shocked, or TF-overproducing cells. Thus, the association of GroEL with TF and GroEL phosphorylation during heat shock alter the affinity of the chaperone for polypeptides in distinct ways, and presumably these factors therefore have different effects on GroEL function in vivo.

Table I. Comparison of the effect of TF overproduction and heat shock treatment on GroEL binding to various proteins Extracts (5 mg of protein) prepared from wild-type cells growing at 30 °C, cells overproducing TF at 30 °C, or heat shocked wild-type cells (growing cells shifted from 30 to 42 °C for 30 min) were loaded onto 1-ml columns containing immobilized denatured proteins as ligands. The bound GroEL was eluted with acid and quantitated by Western blot. Minimal binding of GroEL is expressed as +, maximal binding is expressed as ++++. Ligand Relative binding of GroEL from Control cells (30 °C) TF overproducing cells (30 °C) Heat shocked cells (42 °C) CRAG ++ ++++ ++++ fetuin + ++++ ++++ histone ++ +++ ++++ casein ++ ++ ++++ albumin ++ ++ ++++ cytochrome C + ++++ + globin + ++++ + thyroglobulin ++ ++++ ++

DISCUSSION

Although appreciable progress has been made in understanding the mechanisms of GroEL function using highly purified preparations (14, 15), little is known about the regulation of GroEL function in vivo. The present approach of studying the binding of GroEL in crude extracts offers real advantages for exploring the physiological regulation of chaperone function and the possible existence of additional intracellular factors that influence GroEL activity. For example, prior studies with purified GroEL failed to uncover the marked changes in the properties of GroEL induced during heat shock by phosphorylation (11, 12) or by association with TF. However, future studies of the mechanisms by which TF influences GroEL function, the possible importance of the prolyl isomerase activity of TF, and the structure of TF-GroEL complex will clearly require mechanistic studies with purified components.

The present observations argue strongly that TF alters GroEL function through formation of a TF-GroEL complex that exhibits higher affinity for certain abnormal proteins. This conclusion is based on the finding that TF-GroEL complexes purified from TF-overproducing cells showed much greater binding to certain denatured polypeptides than GroEL isolated from TF-deficient cells. In addition, GroEL and TF dissociated together from the protein columns by an ATP-dependent process and then remained tightly associated as a complex, composed of one TF molecule per one GroEL 14-mer. Thus, a fraction of the GroEL of the cell must exist as TF-GroEL complexes, which exhibit distinct functional properties from the bulk of GroEL in the cell. Also, TF must associate with GroEL in a distinct way from polypeptide substrates, because incubation with GroES and ATP did not dissociate GroEL-TF complex. Furthermore, our experiments with pure TF and GroEL clearly show that TF directly interacts with GroEL to enhance its binding capacity. The inability of GroES and ATP to prevent TF stimulation of GroEL binding indicate that this effect is due to direct interaction of the two chaperones with each other (and does not involve any minor polypeptide substrates that might still be bound to the purified GroEL).

In related studies, we have obtained genetic data indicating that the functions of TF and GroEL/ES are linked in vivo. We found that the temperature-sensitive mutant, groES619, was not viable when TF was overexpressed on a plasmid (even at permissive temperatures). This groES mutation allows the formation of CRAG-GroEL complexes but prevents their dissociation (10). The most likely explanation for the lack of viability is that overproduction of TF stimulates the binding of GroEL to substrates, and in groES619, GroEL-substrate complexes fail to dissociate, and accumulation of GroEL in such dead-end complexes prevents normal growth.

The present findings indicate that TF and heat shock regulate the affinity of GroEL for proteins by distinct mechanisms. The relative importance of these two regulators of GroEL function depends on the growth temperature. The effects of TF overproduction on binding were much greater at 20 °C than at 42 °C, where phosphorylation occurred and where TF induction did not further stimulate binding to CRAG or fetuin (at 37 °C, the properties of GroEL isolated from wild-type cells reflect the presence of some phosphorylated and some TF-associated species). Each mode of regulating GroEL activity has distinct biochemical consequences, i.e. phosphorylated GroEL and TF-GroEL complexes show a distinct pattern of affinities for protein ligands (although there are no obvious physical or chemical properties distinguishing these classes of proteins). Therefore, in vivo, phosphorylation or association with TF probably causes a fraction of the GroEL of the cell to associate preferentially with different unfolded polypeptides.

The capacity of TF to promote the association of GroEL with abnormal proteins can account for our prior finding that TF overproduction promotes CRAG degradation and that intermediates in CRAG breakdown are associated with TF-GroEL complexes (9). However, by stimulating GroEL's association with substrates, TF may also promote the folding or translocation of certain polypeptides. A role for TF in protein folding appears likely because TF also possesses a peptidyl-prolyl isomerase activity (7) and associates with nascent polypeptides on the ribosome (5, 6). This prolyl-isomerase activity may be important in the marked stimulation of GroEL binding to certain polypeptides; for example, the TF associated with GroEL may isomerize a critical proline residue in the substrate and thus facilitate its entry into the central cavity of GroEL (17). In mammalian cells, a peptidyl-prolyl isomerase has been found in complexes with Hsp90 and Hsp70 (18). Presumably, in such complexes and in the TF-GroEL complexes in E. coli, the peptidyl-prolyl isomerase and chaperone function synergistically in catalyzing the successful folding (or degradation) of damaged or nascent polypeptides.