Kinetic Partitioning

Chaperones are a class of proteins that possess the remarkable ability to selectively bind polypeptides that are in a nonnative state. The selectivity of SecB, a molecular chaperone inEscherichia coli, for its ligands can be explained in part by a kinetic partitioning between folding of the polypeptide and association with SecB. It has clearly been established that kinetic partitioning can be poised to favor association with SecB by changing the rate constant for folding of the ligand. We now demonstrate that binding to SecB can be given a kinetic advantage over the pathway for folding by modulating the properties of the chaperone. By poising SecB to expose a hydrophobic patch, we were able to detect a complex between SecB and maltose-binding protein under conditions in which rapid folding of the polypeptide otherwise precludes formation of a kinetically stable complex. The data presented here are interpreted within the framework of a kinetic partitioning between binding to SecB and folding of the polypeptide. We propose that exposure of a hydrophobic patch on SecB increases the surface area for binding and thereby increases the rate constant for association. In this way association of SecB with the polypeptide ligand has a kinetic advantage over the pathway for folding.

The prokaryotic chaperone SecB facilitates export of a subset of proteins (1,2) that are delivered to the periplasmic space or outer membrane of Escherichia coli. SecB, the sole chaperone dedicated to export, has affinity for SecA, the peripheral membrane ATPase of the export apparatus. SecB binds precursor polypeptides in the cytoplasm before they assume their native, stably folded structures (3)(4)(5) and maintains them in an exportcompetent state that is neither folded nor aggregrated (2, 3, 6 -10). Translocation of the polypeptide across the membrane is then inititiated upon ATP hydrolysis by SecA (11).
In addition to being the only chaperone dedicated to protein localization, SecB also differs from other members of the chaperone family in that it is a tetramer and functions independently of nucleotide triphosphate hydrolysis. However, one fundamental feature common to all chaperones is the remarkable ability to selectively bind nonnative polypeptides. The ability of SecB to recognize proteins as nonnative is governed in part by a kinetic partitioning (see Fig. 1) (6,12,13). The formation of a complex between SecB and a protein that possesses nonnative structure will depend on the rate of folding of the polypeptide relative to its rate of association with SecB. Since SecB has no affinity for native, stably folded polypeptides (12, 14 -17), only those proteins that fold slowly are favored to bind SecB and enter the export pathway. All physiologic ligands of SecB are synthesized as precursor species that contain amino-terminal stretches of aminoacyl residues designated leader or signal sequences. Investigations carried out both in vivo and in vitro indicate that the physiologic function of the leader in binding to SecB is to slow folding of the precursor by interfering with the folding reaction that lies on the pathway to the native state (18 -21). Retardation of folding exerted by the leader sequence thus provides the cell with an exquisite mechanism whereby SecB can engage the polypeptide before it folds into its native form.
Studies of the interaction between purified proteins have shown that SecB can also bind the mature protein, which does not carry a leader, if folding is slowed by other means. For example, at 25°C rapid folding of wild-type mature maltosebinding protein precludes formation of a detectable complex with SecB; however, if the folding reaction is slowed by decreasing the temperature from 25 to 5°C, SecB binds the mature protein (12,19). Furthermore, three species of maltose-binding protein have been shown to contain single aminoacyl substitutions in the mature portion of the protein (MalE Y283D, MalE A276G, and MalE W10A) that slow folding sufficiently so that SecB can bind even when the temperature of the folding reaction is 20 -25°C (12,19,21). It is clear from these examples that kinetic partitioning can be poised to favor association with SecB by changing the rate constant for folding of the polypeptide (k f ) (refer to Fig. 1). However, it is also possible that binding to SecB could be given a kinetic advantage over the pathway for folding by modulating the properties of SecB and thereby affecting the rate constants for association (k on ) and/or dissociation (k off ).
Binding studies with short peptide ligands indicate that the SecB tetramer contains multiple subsites for binding of flexible unstructured polypeptides carrying a net positive charge (22). Saturation of these subsites induces a conformational change in SecB, exposing a hydrophobic patch that is thought to serve as an additional binding site for polypeptide ligands (Ref. 22; see Fig. 6 and 7). Here we test the idea that binding to SecB could be given a kinetic advantage over the pathway for folding by exposing the putative hydrophobic binding site on SecB. 1 The abbreviations used are: PMSF, phenylmethylsulfonyl fluoride; ANS, 1-anilinonapthalene-8-sulfonate; MIANS, 2-(4Ј-maleimidylanilino)napthalene-6-sulfonic acid, sodium salt; GuHCl, guanidinium chloride; Lys 10 , a lysine polymer comprising 10 residues; poly-L-lysine, a homopolymer of L-lysine of molecular weight 42,000; SecB141, a form of SecB from which 14 aminoacyl residues have been removed from the C terminus by cleavage with proteinase K at Leu-141; MIANS-SecB, a chemically modified species of SecB that was alkylated with the sulfhydryl-specific reagent MIANS.
Protein Purification-The purification of the matured form of wildtype maltose-binding protein has been described previously (23). SecB was purified as described by Topping and Randall (24). The protein concentrations were determined using extinction coefficients at 280 nm of 47,600 M Ϫ1 cm Ϫ1 for SecB tetramer and 78,972 M Ϫ1 cm Ϫ1 (25) for the mature form of wild-type maltose-binding protein.
Preparation of SecB141-Proteinase K was prepared from a stock frozen in small portions at 2.5 mg/ml in 10 mM HEPES, pH 7.6, that was thawed and diluted immediately before use. SecB was digested at 2.8 mg/ml with 0.04 mg/ml proteinase K as follows: 27 l of proteinase K at 0.05 mg/ml in 10 mM HEPES, pH 7.6 -7.7, was added to 11 l of SecB at 9.6 mg/ml in 10 mM Tris, 150 mM KC 2 H 3 O 2 , pH 7.8, and incubated on ice for 20 min. The digestion was stopped by adding 38 l of 1 mM PMSF. The PMSF, stored as a stock at 0.1 M in ethanol, was diluted into 10 mM HEPES, pH 7.6, on ice just before use. After adding the protease inhibitor, the concentration of the KC 2 H 3 O 2 was 21 mM. Since it has been observed that SecB loses structure during prolonged exposure to low ionic strength (26), we also prepared proteinase K-cleaved SecB that was restored to high ionic strength by including KC 2 H 3 O 2 in the 1 mM PMSF stock used to terminate digestion. We demonstrate here that the activity of proteolyzed SecB stored in low salt (21 mM KC 2 H 3 O 2 ) is comparable to that of proteolyzed SecB stored in high salt (150 mM KC 2 H 3 O 2 ). Typically, 5-10 digestion reactions were carried out and pooled for subsequent use. The concentration of SecB after proteinase K digestion was determined from the absorbance at 280 nm using a mock digestion containing proteinase K and PMSF as the buffer blank.
Chemical Modification of SecB with 2-(4Ј-Maleimidylanilino)napthalene-6-sulfonic acid (MIANS-SecB)-SecB (36 M monomer, each monomer contains 4 cysteine residues) was incubated with MIANS (150 M) in 10 mM HEPES, pH 7.6, for 22 h at 4°C in the dark. The alkylation of SecB with MIANS was terminated by adding 0.75 mM dithiothreitol. The labeled SecB in 16.7 ml total volume was then dialyzed three times against 1.7 liters of 10 mM HEPES, 150 mM KC 2 H 3 O 2 , pH 7.6, for longer than 8 h each time to remove excess reagent. The concentration of SecB monomer after labeling with MIANS was 28 M as determined by amino acid composition analysis. The stoichiometry of incorporation was 1 mol of MIANS:1 mol of SecB monomer as determined using an extinction coefficient at 322 nm of 20,000 M Ϫ1 cm Ϫ1 for MIANS (27).
Fluorescence Assay of Interaction between SecB and Maltose-binding Protein-The details of this assay have been reported previously (12,15,19,21). Fluorescence measurements were made using a Shimadzu RF-540 fluorescence spectrophotometer. The excitation and emission wavelengths used were 280 nm (for maximal absorption by tryptophan, bandwidth 2 nm) and 344 nm (bandwidth 5 nm), respectively. Wild-type mature maltose-binding protein was unfolded by incubation in 2 M GuHCl buffered with 10 mM HEPES, pH 7.6 -7.7, for 2 h at room temperature. Refolding of the denatured protein was initiated by rapid dilution into 2.7 ml of 10 mM HEPES (pH 7.6 -7.7) contained in a cuvette that was held in the spectrophotometer such that the final concentration of maltose-binding protein was 25 nM and the final concentration of GuHCl was 0.05 M. When SecB (unmodified, SecB141, or MIANS-SecB) was present, it was added to the solution in the cuvette before the addition of the maltose-binding protein. The molar ratio of the SecB tetramer to maltose-binding protein is indicated for each experiment. The curves are drawn to aid the eye.
Binding of ANS to SecB-The binding of ANS to SecB was monitored as described previously (22). The fluorescent compound ANS, which binds to hydrophobic clusters of aminoacyl residues, was previously used to identify a hydrophobic binding site on SecB that is exposed upon binding of peptide ligands to the SecB tetramer (22). Interaction of ANS with the hydrophobic patch on SecB is detected by an increase in the fluorescence intensity of ANS and a shift in the emission maximum from ϳ520 to 472 nm. A 1.0 mg/ml solution of a lysine polymer comprising 10 residues (Lys 10 ) that was previously shown to be long enough to bind SecB with high affinity (22), was added in 2 l increments to a cuvette containing 0.15 M intact SecB tetramer or proteolyzed SecB tetramer (digested at 1.1 mg/ml with 0.05 mg/ml proteinase K) and 15 M ANS in 10 mM HEPES, pH 7.6, held at 5°C in the spectrophotometer. The final concentration of KC 2 H 3 O 2 was 16 mM. The intensity of ANS fluorescence with excitation at 370 nm and emission at 472 nm was measured after each addition of Lys 10 . The curve is drawn to aid the eye.
Titration Calorimetry-Calorimetric titrations to obtain dissociation constants were carried out using the OMEGA titration calorimeter from MicroCal, Inc. (Northampton, MA) and the Origin software supplied with the instrument. The system has been described in detail in Wiseman et al. (28). The SecB tetramer (unmodified, SecB141, or MIANS-SecB) was held in the cell at 5 M in 1.345 ml of 10 mM HEPES, pH 7.8, 150 mM KC 2 H 3 O 2 , 0.1 M GuHCl, 0.45 mM EGTA, and was titrated with galactose-binding protein that was unfolded in 10 mM HEPES, pH 7.8, 150 mM KC 2 H 3 O 2 , and 1.0 M GuHCl and diluted into the buffer for calorimetry immediately before loading the syringe for injection. Complete binding isotherms were generated at 7°C by a sequence of 17 injections, each of 18 l spaced at 11 min intervals. The best fit of the data for a model of one binding site was obtained using a least squares deconvolution algorithm based on the Marquardt method. The absolute binding constants determined in these experiments are subject to some uncertainty because the ligand can refold during the course of the titration. Therefore we report relative affinities that are valid since the identical titration schemes were used.

Potentiation of SecB to Block Folding of Maltose-binding
Protein at 21°C-It has clearly been established that kinetic partitioning can be poised to favor association with SecB by changing the rate constant for folding of the ligand (k f ) (12,19,21). Here we demonstrate that binding to SecB can be given a kinetic advantage over the pathway for folding by modulating the properties of SecB. Ligands such as short basic peptides or polymers of L-lysine, which have lower affinity for SecB than do the physiologic polypeptide ligands, induce SecB to undergo a conformational change that exposes a hydrophobic site proposed to function in ligand binding (22). We asked whether exposure of the hydrophobic patch on SecB induced by incubation with polymers of L-lysine would allow us to detect a complex between SecB and maltose-binding protein under conditions in which rapid folding of the polypeptide would otherwise preclude formation of a kinetically stable complex. We reasoned that exposure of the hydrophobic patch might serve to increase the binding area on SecB and therefore increase the rate constant for association such that binding to SecB would have a kinetic advantage over the pathway for folding. The interaction between SecB and maltose-binding protein can be monitored in vitro by assessing the ability of SecB to block the The distribution of a polypeptide among the pathways for folding, aggregation, and export will depend on the rate constant of folding (k f ) or of aggregation (k agg ) relative to the rate of association with SecB ([SecB]⅐k on ). The model predicts that binding to SecB can be given a kinetic advantage over the pathway for folding by changing the rate constants of the competing processes. See text for discussion. Adapted from a figure (31) and reprinted with permission from Elsevier Science.
refolding of denatured maltose-binding protein as established previously (12,15,19,21). Since maltose-binding protein contains 8 tryptophan residues, the refolding of maltose-binding protein can be followed by the change in intrinsic fluorescence of tryptophan. The magnitude of the blockage of folding caused by SecB is calculated from a comparison of the change in fluorescence in the presence of SecB with the same parameter determined in the absence of SecB. As shown previously (21), at 21°C SecB slows the rate of folding of mature maltose-binding protein (data not shown), but the final amplitude achieved is the same; thus, there is not a blockage of folding (Fig. 2, E). Upon addition of a homopolymer of L-lysine of molecular weight 42,000 (poly-L-lysine), SecB is poised to block folding of maltose-binding protein at 21°C (Fig. 2, q). This observation demonstrates that kinetic partitioning can also be poised to favor association with SecB by modulating the properties of the chaperone. That the poly-L-lysine has its effect directly by binding to SecB is supported by several lines of evidence. First, poly-L-lysine has been shown to interact at the physiological peptide binding site on SecB (22). Second, poly-L-lysine does not potentiate SecB by nonspecific electrostatic effects. This is demonstrated by the finding that the addition of an equal mass of a lysine oligomer comprising 4 residues that was previously shown to be too short to bind SecB with high affinity (22) does not potentiate SecB for blockage of folding under the conditions described here (data not shown). Finally, the ability of SecB to block the folding of mature maltose-binding protein at 21°C is not explained by an effect of poly-L-lysine on the rate constant for folding of the ligand, since when maltose-binding protein is refolded in the presence of poly-L-lysine, the rate of the folding reaction that renders maltose-binding protein incapable of binding SecB is unchanged (data not shown).
Characterization of Altered Species of SecB-Having established that we could use poly-L-lysine to potentiate unmodified SecB, we asked whether we could use this lysine polymer to restore activity to species of SecB that had been altered so that they had decreased ability to bind maltose-binding protein. We first evaluated the activity of two defective species of SecB under conditions in which slow folding of maltose-binding protein favors interaction with unmodified SecB in the absence of poly-L-lysine. As shown previously (12,19), at 5°C, where the refolding of mature maltose-binding protein is sufficiently slow, unmodified, intact SecB is able to exert a complete blockage of folding when present in excess and a partial block at lower molar ratios (Fig. 3, q). A truncated form of SecB (SecB141) that has lost 14 aminoacyl residues from the C terminus as a result of proteinase K cleavage at Leu-141 (29) was somewhat less active in blocking folding under these conditions (Fig. 3, OE and ). In contrast to the situation for SecB141, a species of SecB that was alkylated with the sulfhydryl-specific reagent MIANS (MIANS-SecB, 1 mol of MIANS/ mol of SecB monomer) was incapable of completely blocking the folding of mature maltose-binding protein at 5°C. As shown in Fig. 4A, the folding of mature maltose-binding protein was retarded by the presence of MIANS-SecB at a 1.25-molar excess over maltose-binding protein, whereas the presence of unmodified SecB at the same molar ratio completely eliminates the change in fluorescence that reflects the folding of maltosebinding protein (Fig. 4B). Thus the loss of 14 residues from the C terminus of SecB only slightly reduces the affinity of SecB141 for the ligand maltose-binding protein, whereas sulfhydryl modification reduces the affinity of MIANS-SecB to an extent such that it is no longer able to exert a blockage in folding. The relative affinities of the various species of SecB for a physiologic ligand, galactose-binding protein, was determined directly using titration calorimetry. The affinity of SecB141 was similar to that of intact SecB with at most a 2-fold decrease, whereas the affinity of MIANS-SecB was approximately 5-10fold lower.
Poly-L-lysine potentiated both SecB141 and MIANS-SecB at 21°C, where rapid folding of maltose-binding protein otherwise favors folding of the polypeptide over binding to SecB. Poly-Llysine was shown to potentiate the activity of SecB141 to almost the same extent as intact SecB (Fig. 2, compare q with OE  and ). Thus, loss of 14 amino acids at the C terminus does not interfere with the ability of poly-L-lysine to potentiate SecB141. The defect caused by the MIANS modification could also be corrected by incubation with poly-L-lysine (Fig. 5). Whereas in the absence of poly-L-lysine MIANS-SecB at a 5-molar excess over maltose-binding protein decreased the rate of folding, inclusion of poly-L-lysine potentiated the MIANS-SecB so that the increase in fluorescence due to folding was completely eliminated. Consistent with the idea that potentiation is achieved by exposure of the hydrophobic patch, we demonstrated that SecB141 was able to undergo the conformational change exposing the patch, as detected by binding of ANS (Fig.  6). These findings suggest that exposure of the hydrophobic patch on SecB compensates for the loss of activity of SecB. We were unable to directly demonstrate exposure of the hydrophobic patch on MIANS-SecB, since the assay used to detect the patch involves binding of ANS, which has an absorption spectrum overlapping with that of MIANS. DISCUSSION A model rationalizing the ability of SecB to selectively bind nonnative proteins in the absence of any consensus in sequence or structure is illustrated in Fig. 7. The binding of a natural ligand to SecB involves interaction of regions of the polypeptide ligand with two different types of binding sites on SecB (Ref. 22; Fig. 7B). Simultaneous occupancy of multiple subsites of one type, which bind flexible stretches of ligand, causes a conformational change, exposing a second type of site, a hydrophobic patch proposed to function in binding of hydrophobic regions of the nonnative protein ligand. This model is based on the observation that the SecB tetramer binds peptide ligands and as a result undergoes a conformational change to expose a hydrophobic patch that was detected by binding of the fluorescent probe ANS ( Ref. 22; Fig. 7A). The binding of a natural, long polypeptide ligand would thus involve interaction with the sites occupied by the peptide ligands as well as with the hydrophobic patch. Exposure of the patch would be triggered by simultaneous occupancy of multiple subsites by flexible stretches of a nonnative polypeptide, which would be distinguished from a native, folded protein by the availability of several such stretches. High affinity binding selective for the nonnative state would result from these multiple interactions, each having low specificity. The association between SecB and its nonnative ligands, although being of high affinity (K d ϳ 10 Ϫ8 M) has been shown to be in an equilibrium in which both binding and release are rapid (13,30). The unfolded protein continuously samples the free state and, with each cycle of dissociation, will partition between folding and rebinding. Thus, whether the polypeptide folds or forms a kinetically stable complex with SecB is based on a kinetic partitioning in which the relevant rate constants are the rate constant for folding and the rate constant for association with SecB.
Previous work in our laboratory using forms of maltosebinding protein that are altered in their folding properties revealed different effects of SecB on the folding of these polypeptides at 21°C (21). While SecB only slightly decreased the rate of folding of wild-type maltose-binding protein, it drastically reduced the rate of folding of MalE A276G and completely blocked folding of MalE W10A. These differences in the effects of SecB can be explained in terms of kinetic partitioning and are attributed to differences in the rate constant for the folding reaction of the polypeptides (rate constants at 25°C: wild type, 1.7 s Ϫ1 ; MalE A276G, 0.24 s Ϫ1 ; MalE W10A, 0.04 s Ϫ1 ), since it is folding that is in competition with binding to SecB. Thus, the 7-fold decrease in the rate of folding of MalE A276G with respect to the rate of folding of wild-type maltosebinding protein allowed SecB to slow the appearance of the folded protein, whereas the 40-fold decrease in folding of MalE W10A poised partitioning to favor binding over folding to the extent that the observed effect was a blockage of folding. Here we have shown that poly-L-lysine can potentiate SecB to block folding of wild-type MalE at 21°C. This potentiation can also be explained in terms of a kinetic partitioning. In this case, addition of poly-L-lysine does not alter the rate constant of folding of the polypeptide ligands, but rather we propose that it has its effect directly on SecB by increasing the rate constant for association. As shown previously (22), interaction of poly-Llysine with SecB results in exposure of a hydrophobic patch, and if as proposed this patch is part of the ligand binding site, the rate constant for association would thereby increase, since the probability that productive collisions occur between SecB and maltose-binding protein is directly proportional to the reactive surface area of the two proteins.
Poly-L-lysine was effective not only in potentiating unmodified SecB to capture polypeptides that would otherwise fold too rapidly but also in restoring activity to two species of SecB that had been modified to decrease their ability to bind maltosebinding protein. Cleavage of 14 aminoacyl residues from the C terminus of SecB to yield SecB141 caused a slight defect, whereas sulfhydryl modification (MIANS-SecB) caused a more severe defect in binding of maltose-binding protein. It was shown that when incubated with poly-L-lysine, SecB141 underwent a conformational change, exposing a hydrophobic patch, thus supporting the idea that potentiation is achieved by exposure of the patch. Poly-L-lysine also potentiated MIANS-SecB at 21°C under conditions in which the ligand folds too rapidly to otherwise allow detection of a complex between unmodified SecB and maltose-binding protein.
The complex between SecB and poly-L-lysine has a dissociation constant in the range of micromolar (22), whereas the complex with maltose-binding protein has a K d in the range of 10 Ϫ8 M. Since binding to all ligands by SecB is likely to be near collision-limited, as was shown to be the case for bovine pancreatic trypsin inhibitor (30), the difference in affinity between poly-L-lysine and maltose-binding protein is likely to reflect a difference in dissociation rate that can be rationalized by proposing that poly-L-lysine occupies the subsites for flexible stretches, whereas maltose-binding protein is also bound at the hydrophobic site, and thus the probability of dissociation is lower. There are at least two ways in which one can envision the potentiation by poly-L-lysine. One possibility is that the patch becomes exposed when poly-L-lysine binds, but upon dissociation, relaxation to the closed state is slow enough that maltose-binding protein can collide with SecB before the conformation returns to the prebound state. Alternatively, maltose-binding protein might bind the SecB⅐poly-L-lysine complex, initially via the exposed patch, if it were not occupied by the poly-L-lysine and once tethered, displace the poly-L-lysine from the subsites for flexible stretches in a stepwise fashion. Whether the binding occurs when SecB is free or when poly-Llysine is bound, we propose that exposure of a hydrophobic patch on SecB increases the surface area for binding and thereby increases the probability of productive collisions and thus the rate constant for association. In this way, association of SecB with the polypeptide ligand has a kinetic advantage over the pathway for folding. It seems unlikely that in the presence of poly-L-lysine, the dissociation rate constant would be affected to the same extent as is the association rate constant, since the binding of maltose-binding protein itself exposes the hydrophobic patch. However, even if the dissociation rate constant were lower, as long as dissociation is rapid enough to allow refolding to occur over the time course of the experiment, only the association rate constant is crucial to the partitioning between folding and rebinding to SecB. In the unlikely event that the presence of poly-L-lysine has an extreme effect on k off resulting in an essentially irreversible interaction between maltose-binding protein and SecB, a blockage of folding would be observed.  7. A model of the binding of peptide ligands and natural ligands to SecB. A, binding of a peptide ligand. B, binding of a natural ligand. As described in the text, simultaneous occupancy at multiple subsites on the SecB tetramer that bind flexible stretches of ligand induces a conformational change in the chaperone that exposes a hydrophobic surface. This hydrophobic patch is proposed to serve as an additional binding site for hydrophobic portions of the nonnative polypeptide.