Antifolding Activity of the SecB Chaperone Is Essential for Secretion of HasA, a Quickly Folding ABC Pathway Substrate*

We have previously shown that SecB, the ATP-independent chaperone of the Sec pathway, is required for the secretion of the HasA hemophore from Serratia marcescens via its type I secretion pathway, both in the reconstituted system in Escherichia coli and in the original host. The refolding of apo-HasA after denaturation with guanidine HCl was followed by stopped-flow measurements of fluorescence of its single tryptophan, both in the absence and presence of SecB. In the absence of SecB, HasA folds very quickly with one main phase (45 s–1) accounting for 92% of the signal. SecB considerably slows down HasA folding. At stoichiometric amounts of SecB and HasA, a single phase (0.014 s–1) of refolding is observed. Two double point mutants of HasA were made, abolishing two hydrogen bonds between N-terminal and C-terminal side chain residues. In both cases, the mutants essentially maintained the same secondary and tertiary structure as wild-type HasA and were fully functional. Refolding of both mutants was much slower than that of wild-type HasA and they were secreted essentially independently of SecB. We conclude that SecB has mainly an antifolding function in the HasA ABC secretion pathway.

SecB is a chaperone of the Sec pathway in Gram-negative bacteria. It was initially identified in Escherichia coli where it interacts with a subset of preproteins to be exported across the cytoplasmic membrane in such a way as to slow down their folding and target them to SecA, the ATPase of the Sec system (1-4) (5). Currently, this chaperone is restricted to Gramnegative bacteria, from the proteobacteria class.
In the Sec pathway, the basis of the interaction of SecB with its substrates has been studied in great detail either in vivo for its natural substrates or in vitro with its natural or artificial substrates. It appears that in vitro, SecB is able to interact with a large variety of denatured substrates whereas it is highly specific in vivo (6,7).
The structure of SecB from Haemophilus influenzae is known at atomic resolution and has highlighted several important features of this chaperone (8). It is a tetramer composed of a dimer of dimers. Each monomer bears a putative peptidebinding site able to accommodate, in an extended conformation, hydrophobic peptides of 10 -15 residues; on both sides the tetramer presents a patch of negatively charged residues that can interact with a SecA dimer. In vitro studies using peptide libraries have indicated that SecB binding site is around 9 -10 residues long with a strong preference for aromatic and cationic amino acid residues (9). Furthermore, SecB is a highly negatively charged molecule; thus, negatively charged peptides bind to SecB only if the peptide contains enough aromatic residues in the peptide to overcome the electrostatic repulsion.
Maltose-binding protein (MBP), 1 the best characterized SecB substrate in both in vivo and in vitro studies, interacts with SecB in kinetic partitioning: the competition between binding to the chaperone and folding reactions (10 -13). At 25°C MBP refolds by a complex pathway characterized by three phases. SecB slows down folding but does not block it; moreover, increasing amounts of SecB progressively slow down MBP refolding. At 5°C MBP refolding is slower and SecB totally blocks refolding. This can be visualized by the existence of stable SecB⅐MBP complexes that can be isolated at 5°C but not at 25°C. MBP mutants have been isolated on the basis of their ability to translocate across the inner membrane without the presence of SecB. MBPY283D is a MBP mutant that is perfectly biologically active and folds more slowly than wild-type MBP making its translocation SecB-independent (2,14). This variant associates with SecB blocking MBP folding even at 25°C. A model has been proposed for the interaction of SecB and MBP: an unfolded form of MBP binds to SecB and is in equilibrium with unbound form. MBP does not fold on SecB and this explains the progressive blockade of MBP folding with increasing SecB amounts. Although the signal sequence of pre-LamB or pre-MBP does not interact with SecB by itself, it slows down the folding of the mature part of the protein (15,16). SecB interacts with several parts of precursor molecules, protecting a large central portion of precursors to which it is bound (17)(18)(19).
Other substrates have been studied in great detail, like barnase of Bacillus amyloliquefaciens (20) or R-BPTI (21). Studies with R-BPTI, which is a stable unfolded form of BPTI, have shown that association of a substrate with SecB is limited by diffusion and that the bound R-BPTI is easily exchanged with free R-BPTI. Such findings allowed the researchers to estimate a K d of a few nM for the SecB⅐R-BPTI complex. In the case of barnase, which is a small secreted RNase with a standard signal peptide, SecB retards the folding of barnase but never blocks its folding in contrast to what has been observed with MBP. Barnase folds at least partially onto SecB which challenges the kinetics partition model. * 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.
We have shown that, besides its function in the Sec pathway, SecB is also the chaperone of a sec-independent pathway, namely the secretion of the HasA SM hemophore through its ABC pathway in Serratia marcescens (22,23).
HasA SM is a monomeric protein of 188 residues secreted by an ABC-type pathway under iron starvation conditions by S. marcescens (24) and other Gram-negative bacteria, Pseudomonas aeruginosa, Pseudomonas fluorescens, and Yersinia pestis (25,26). HasA SM acquires the heme from hemoproteins (e.g. hemoglobin) and delivers it to a specific receptor at the cell surface whereby the heme is internalized and used as an iron source (27). Both the secretion pathways and the heme acquisition pathways have been reconstituted in E. coli. We have shown that SecB, the chaperone of the Sec pathway, is required for HasA SM secretion both in the original host and in the reconstituted system (22,23). SecB does not affect the secretion apparatus but likely interacts with HasA SM . We have proposed that the N-terminal part of HasA SM interacts in a SecB-dependent fashion with the ABC transporter (28). Folded HasA SM is not secretion-competent and inhibits further secretion of newly synthesized HasA SM . This strongly supports a model whereby HasA SM is targeted in a cotranslational fashion to the ABC transporter before secretion via the C-terminal secretion signal (29). Altogether, these data are consistent with a model whereby both the N terminus and the C terminus of HasA SM play a role in secretion.
Here we studied the in vitro refolding of HasA SM both in the presence and absence of SecB. We show that HasA protein folds very quickly. Also, we show that SecB interacts directly and quickly with HasA SM to slow down HasA SM folding. We had previously established that the N terminus of HasA also plays a role in secretion. Here we provide evidence that hydrogen bonds which exist between N-terminal and C-terminal residues of HasA SM are involved in HasA SM folding and its SecB dependence for secretion.
pSYC134/pUC was used as a template and the quick-change kit from Stratagene used for mutagenesis. Two successive rounds of mutagenesis were carried out to obtain the double mutants. Sequencing was used to verify the mutations. The mutated fragments were then reintroduced into pAM238 and pBGS18ϩ. pSYC150 encodes HasD and HasE, which, together with chromosomally encoded TolC, formed the secretion apparatus. PsecB/trc was used to overproduce SecB. This plasmid contains the secB gene under the control of ptrc and was kindly provided by Dr. J. M. Betton.
Materials-GdnHCl ultrapure was from ICN Biomedicals. Cells used for fluorescence and circular dichroïsm were from Hellma.
Expression and Purification of the HasA SM Proteins-Wild-type HasA SM was produced in E. coli MC4100(pSYC134/pAM238 ϩ pSYC150) (24). Mutant hasA genes were constructed by in vitro sitedirected mutagenesis. Wild-type (WT), E148A,D167A, and H32A,Y75A mutant samples were produced in M9 minimal medium as reported previously (30) Y7A/H17A mutant was produced in M9 minimal medium with addition of 0.2% bactotryptone. Wild-type and mutant apo-HasA SM proteins were purified as described previously (30). Their heme content, determined by the absorbance of the Soret band wavelength, was always less than 5%. SecB was purified according to published protocol (9). Its purity was checked by SDS-PAGE and the concentration determined from absorbance at 277 nm using an extinction coefficient of 11708 M Ϫ1 cm Ϫ1 for the monomer.
To assess the secretion of wild-type and mutant HasA proteins, cultures of MC4100 or MC4100⌬secB harboring Has secretion functions from pSYC150 and wild-type or mutant hasA on a compatible plasmid were grown in LB medium at 30°C with the appropriate antibiotics to late exponential phase. The culture was centrifuged at 4000 ϫ g max for 10 min and the supernatant precipitated with 20% trichloroacetic acid was analyzed by SDS-PAGE.
Buffer-All kinetic and equilibrium experiments were carried out in 20 mM sodium phosphate, pH 7.0. The protein concentrations were determined by absorbance spectrophotometry before use. Water was deionized and purified on a Millipore system. All solutions were filtered. The equilibrium unfolding as a function of GdnHCl concentration were carried out by dilution of the proper amount of the 8 M denaturant stock solution and leaving the samples at room temperature overnight prior to performing the titration experiments.
Equilibrium Unfolding Experiments-The equilibrium unfolding of HasA SM and its mutants as a function of GdnHCl concentration was monitored by fluorescence and far-UV CD at 25°C. Fluoromax (Spex Industries) photon counting spectrophotometer was used carry out the fluorescence studies. The samples were excited at 295 nm, and the resulting emission spectra were acquired between 310 nm and 410 nm. When working at a constant emission wavelength (i.e. the protein was excited at 295 nm and emission was monitored at 360 nm), fluorescence intensity was recorded for 30 s, with a sampling period and an integration time of 1 s, and then averaged. The slit width was 5 nm for the excitation light and 10 nm for the emission light. Measurements were made with a protein concentration of 3 M and a 1.0-cm path length cuvette.
CD measurements were acquired in a Jobin-Yvon CD6 spectropolarimeter. Far-UV CD spectra were obtained by averaging five individual scans with a constant 0.5 nm step and with 5-s integration time from 180 to 200 nm and 2 s from 200 to 260 nm. A cell with a path length of 0.2 cm was used with a protein concentration of 3 M. Near-UV CD spectra were acquired between 350 nm and 450 nm using 30 M of protein in a 1.0-cm path length cell. For GdnHCl titrations, the absorbances were measured at 222 nm with a sampling period and time constant of 1 s during 60 s. The final ellipticity was calculated by averaging the 60 recorded values. Measurements were made with a protein concentration of 30 M in a 0.02-cm path length cell. All spectra were corrected by subtracting the appropriate baseline.
Thermal Denaturation-Thermal denaturation was followed by monitoring mean residue ellipticities at 222 nm (far-UV) using a Jobin-Yvon CD6 dichrograph interfaced with a Haake water bath. The signal was acquired with a 0.02 cm pathlength cell and a protein concentration of 30 M with a sampling period of 60 s and an integration time of 1 s. The results were then averaged. The heating rate was 0.8°C/min with a stabilization delay of 3 min before each recording.
Data Analysis-The measured fluorescence intensities and the mean residue ellipticity were analyzed in terms of a two-state transition to obtain the fraction of unfolded protein and free energy of folding as a function of denaturant. Non-linear least square fittings of the equilibrium transitions were achieved using the following equation, where S is the signal (CD or fluorescence) measured, n and u the slopes of the native and unfolded base lines, S u and S n the specific signals of the native and unfolded protein extrapolated to zero denaturant concentration, and K the apparent equilibrium constant of unfolding at a concentration C of denaturant. The free energy of unfolding ⌬G 0 at zero denaturant concentration is given by, where m is a constant characteristic of the protein and of the denaturant.
Stopped-flow Measurements-The measurements of HasA SM refolding kinetics were carried out at 22°C with a SFM300 mixing device from Bio-logic (Pont de Claix, France). The mixing device, equipped with a F15 (1.5 ϫ 1.5-mm cross-section) fluorescence cell, was combined with the optical bench and detection module from Bio-logic. The excitation wavelength was 295 nm, and the emitted light above 320 or 350 nm was detected through a high pass filter. The kinetics were recorded by means of the Bio-Kine software package from Bio-logic.
where A i is the amplitude of each phase, k i its rate constant, and A ϱ is the plateau value.

Apo-HasA SM Unfolding and Refolding in Vitro
HasA SM Refolds Spontaneously with Fast Kinetics-HasA SM is a 188-amino acid monomeric protein with a unique domain of ␣/␤-fold (31). Fig. 1, A and B, show the chemical (Fig. 1A, left) and thermal (Fig. 1B, right) denaturations of WT HasA SM and of two mutants (see below for the description of the mutants), followed by monitoring the disappearance of the secondary structures using far-UV CD. During chemical denaturation of WT HasA, an increase in the fluorescence of the unique tryptophan was observed together with a shift in the maximum fluorescence emission wavelength (from 322 to 353 nm) (data not shown). Both fluorescence and far-UV CD (Fig. 1A) gave superimposable equilibrium denaturation curves using GdnHCl as denaturant. The data could be satisfactorily fitted to a two-state model F 7 U for unfolding. The GdnHCl concentration of half unfolding is given by C m ϭ ⌬G/m and C m ϭ 2.7 Ϯ 0.2 M. The value obtained for the free energy of denaturation ⌬G is 11.5 Ϯ 4.7 kcal⅐mol Ϫ1 . As seen in thermal denaturation curve obtained for HasA SM using far-UV CD (Fig. 1B), the midpoint of the thermal transition T m at 59.5 Ϯ 0.5°C is similar to the value reported by differential scanning calorimetry microcalorimetry. 2 Because this thermal transition of HasA SM is not reversible regardless of the heating rate chosen between 20 and 60°C/h, we could not extract the thermodynamic parameters from the thermal denaturation data.
The refolding kinetics of WT Has SM were also monitored by the fluorescence changes of its single tryptophan residue detected by a stopped-flow spectrophotometer (Fig. 2A, trace a). HasA SM is largely unfolded in 5.5 M GdnHCl and reversibly refolded upon dilution of the chaotropic agent leading to a large decrease of fluorescence. Upon dilution of apo-HasA SM from 5.5 M GdnHCl to 0.14 and 0.28 M GdnHCl, about 50% of the fluorescence intensity decreased within the dead time; nevertheless, this signal decrease is not due to a burst phase (phase faster than 4 ms) but rather to a dilution effect of GdnHCl. Indeed, extrapolation of fluorescence at 0.14 M GdnHCl from the curve obtained between 3.0 and 5.5 M GdnHCl gives the same intensity as this from the first fluorescence measurement in the denaturation kinetics (not shown). The time course of refolding as followed by Trp fluorescence, was fitted to a double exponential decay. The first phase (45 s Ϫ1 ) corresponds to 92% of the relative amplitude, and the second (1.8 s Ϫ1 ) accounts for the remaining 8% (see Fig. 3). These rate constants and amplitudes are only slightly temperature-dependent between 4 and 22°C (data not shown).
SecB Slows Down in Vitro HasA SM Refolding-We have previously shown that HasA SM secretion is SecB-dependent in the original host and in the reconstituted system in E. coli in vivo (22,23). This prompted us to study the in vitro interaction of SecB with HasA SM . We could first show that no stable complexes exist between SecB and folded apo-HasA SM detected either by coimmunoprecipitation or by gel electrophoresis under non-denaturing conditions (data not shown). Apo-HasA SM refolding was also studied in the presence of various amounts of SecB in the renaturation buffer at 22°C (Figs. 2B and 3). We used a high pass filter to collect emitted light above 350 nm to minimize the fluorescence contribution of SecB, which displays a maximum fluorescence emission around 320 nm in 0.14 M GdnHCl. Analysis of the refolding kinetics showed that SecB affects HasA SM folding (Fig. 2B, traces a-e). It is clear that SecB slows down HasA folding. This is a direct evidence for interaction, despite the very fast HasA SM folding kinetics. The folding of HasA SM is strongly retarded when more SecB is added to the refolding buffer (Fig. 2B, traces a-e). The HasA SM -SecB interaction takes place very early, before or during the fast HasA SM folding phase (45 s Ϫ1 ). Analysis of both amplitude and rate constant of the different phases observed in the presence of increasing amounts of SecB (Fig. 3) show the progressive appearance of a slow folding phase not observed in the absence of SecB. The amplitude of the major fast folding phase progressively decreases with higher concentrations of SecB, although the rate constant does not change at around 50 s Ϫ1 . A slower phase progressively appears with an apparent rate constant of 0.013 s Ϫ1 , at a 2:1 ratio of SecB tetramer to HasA; there is only a slight decrease of the rate constant with higher SecB concentrations. At the same time, the intermediate phase of 1.8 s Ϫ1 stays at the same small amplitude with a substoichiometric 2 A. Lecroisey, personal communication.  nation of HasA SM x-ray structure indicates that there are two pairs of residues making side chain/side chain hydrogen bonds between these two extremities: Tyr 7 /Asp 167 and His 17 /Glu 148 (31). Based on the fact that the N-and C-terminal parts of HasA SM are involved in the secretion process and that theses extremities are in close contact in the native conformation, we postulated that mutations of those pairs of residues, by disrupting the hydrogen bonds, might affect folding rate and hence the SecB dependence of secretion. To this end these residues were individually changed for alanine, disrupting one hydrogen bond for each mutant protein. All those single residue mutants display reduced dependence upon SecB for secretion (data not shown). Two double mutants were thus constructed, Y7A,H17A and E148A,D167A, to disrupt both hydrogen bonds by mutating two residues either in the N terminus or the C terminus (see the figure provided in the Supplementary Material).
The refolding kinetics of HasA SM 7-17 and HasA SM 148 -167 were studied using the same procedure as that for wild-type HasA SM . Refolding of both mutants showed refolding kinetics much slower than that of wild-type ( Fig. 2A, traces b and c as  compared with a). Two distinct phases were detected for each mutant with no significant burst amplitude (Fig. 2). Fitting of the data showed that the 7-17 mutant displays one rate constant of 2.1 s Ϫ1 with relative amplitude of 68% and one rate constant of 0.9 s Ϫ1 and its relative amplitude of 30%. The two phases detected for 148 -167 have rate constants of 3.9 and 1 s Ϫ1 with respective relative amplitudes of 83 and 17%. These data do not allow us to conclude that mutants and wild-type HasA SM follow the same folding pathway or different folding pathways.
In the presence of SecB, both mutants displayed essentially similar behavior and, as for WT HasA, their refolding kinetics is slowed down (not shown), indicative of an interaction with SecB. However, fitting of the data required four exponentials, instead of three for the WT. The Fig. 4 shows the analysis of the fluorescence data for the mutants in a similar way as that done for WT HasA. The amplitudes of the two refolding phases of HasA mutants, observed in the absence of SecB decreased, while two (negative) amplitudes appeared and increased with the concentration of SecB (not represented on Fig. 4). The corresponding phases could correspond to isomerizations of the HasA⅐SecB complex. Additionally, a fast phase, characterized by an increase of fluorescence emission, is observed upon SecB addition. At a 2:1 ratio of SecB tetramer to HasA, the apparent rate constants of this fast phase are 64.4 s Ϫ1 for 7-17 and 61.8 s Ϫ1 for 148 -167, respectively. This fast initial phase could correspond to the binding of HasA to SecB (see Fig. 4). Furthermore a slow phase of positive amplitude appeared with the same apparent rate constant as that of the wild-type: at a 2:1 ratio of SecB tetramer to HasA, rate constants of this slow phase were of 0.011 s Ϫ1 for 7-17 and 0.015 s Ϫ1 for 148 -167, respectively (see Fig. 4). This phase can correspond, as for WT HasA, to the slow dissociation of HasA, partially or totally folded, from the HasA⅐SecB complex. As also has been seen with the wild type HasA, folding is never blocked, only strongly retarded, whatever the concentration of chaperone. Finally, SecB is able to interact with the mutants that are indistinguishable from the wild-type with regards to this phenomenon, at our level of analysis.
Interactions between the N and C Terminus of HasA SM Affect SecB Dependence for Secretion but Not HasA SM Function-These two double mutants were tested for their secretion and found to be mostly SecB-independent (Fig. 5). Furthermore, the biological activities of the two mutants were indistinguishable from the wild-type (data not shown). Their far-UV and near-UV CD spectra were also identical to that of the wild-type (Fig. 6), indicating that mutant proteins essentially maintain their secondary and tertiary structures. Their behavior in terms of secretion strongly supported our model as it appears that slowing down their folding kinetics correlates with their reduced SecB dependence in terms of secretion. CD (Fig. 1) and fluorescence (not shown) spectra were recorded at several GdnHCl concentrations for the mutants. Like the wild-type protein, the UV CD and fluorescence of 7-17 and 148 -167 mutants showed a clear cooperative GdnHCl-induced transition (Fig. 1A). It was found that half-denaturation occurred at lower GdnHCl concentration than for wild-type, consistent with a reduced stability of the mutants. A two-state analysis of the data gave midpoints (C m ) of 1.99 Ϯ 0. the loss of two hydrogen bonds (about 4 kcal⅐mol Ϫ1 ) as a result of the mutations. Although the HasA SM mutants are destabilized with respect to the wild-type protein, their fractions of unfolded molecules at 0 M GdnHCl extracted from the free energy of denaturation remain very low: 7.8 ϫ 10 Ϫ7 for the 7-17 mutant and 1.8 ϫ 10 Ϫ8 for the 148 -167 mutant versus 3.9 ϫ 10 Ϫ9 for the wild-type protein. Thermal unfolding also indicated a lower stability of both mutant proteins (Fig. 1B).
The midpoint values of the thermal transition T m are 52.6 Ϯ 0.5°C for 7-17 and 55.0 Ϯ 0.5°C for 148 -167.
As a control we used another double mutant of HasA, H32A,Y75A, affecting heme binding (32). This mutant had a somewhat reduced stability (half-denaturation at 2.2 Ϯ 0.2 M), refolded with the same kinetics as that of the wild-type protein, and was SecB-dependent for its secretion at the same extent as wild-type protein (data not shown).

DISCUSSION
Here we have studied HasA SM folding and its interaction with SecB in vitro together with the effect of mutations disrupting hydrogen bonds between the C and N terminus of HasA SM on its folding and their secretion dependence upon SecB. HasA SM folding is simple with no burst-phase intermediate and with a fast and major folding phase, making it an attractive model to study interactions with SecB. This is the first time a natural SecB-dependent substrate that is not a Sec substrate has been studied in interaction with SecB in vitro.
The interaction between HasA SM and SecB differs in several characteristics of the interaction of SecB with Sec substrates. Since HasA SM and the mutants do not contain any arginine or lysine and are mostly negatively charged, the basis of interaction between HasA SM and SecB cannot rely on electrostatic interaction between residues but through hydrophobic interactions (33,34). Since HasA SM is a natural substrate of SecB, the absence of positive charge in HasA SM proves that positive charges are not a prerequisite for interaction with SecB. As seen with other substrates, SecB slows down HasA folding by more than three orders of magnitude. However, in contrast to the MBP or galactose-binding protein refolding, which is totally blocked at high SecB concentrations, which allows the isolation of permanent stable complexes (12,18), this does not appear to be the case for HasA SM . Instead, there appears to be a slow folding phase of 0.014 s Ϫ1 whose apparent rate constant does not significantly change with varying SecB amounts. REACTION 1 This is reminiscent of the behavior of barnase (a synthetic SecB substrate) in the presence of SecB, which has been interpreted as at least partial refolding when bound to this chaperone (20). This is in contrast to the model of kinetic partitioning like for MBP in which the unfolded protein is bound to SecB, released, and folded only when free in solution. It is likely what happens with barnase occurs with HasA SM .
The major folding phase of HasA SM has a rate constant inferior to 45 s Ϫ1 , which is much faster than the rate constants observed for characterized substrates of the Sec system. Fur- thermore, in the case of Sec substrates, the presence of the signal sequence slows down folding. This has been proposed as the basis of the in vivo selectivity of SecB for its substrates (10,16); this clearly does not apply to the case of HasA SM . SecB binds to HasA SM in millisecond time scale. At 1:1 HasA SM :SecB stoichiometry, we estimated the second-order binding rate constant at 3 ϫ 10 9 M Ϫ1 s Ϫ1 , suggesting that the rate of HasA SM association to SecB is diffusion-limited (21). So, a slow folding kinetics is not a prerequisite for in vivo implication of SecB in this case. It is very likely that this explains that secretion competence of HasA SM cannot be observed for a long period of time, since uncoupling of HasA SM synthesis from its secretion apparatus leads to no secretion even when SecB is overexpressed (29). Moreover this is consistent with the observed absence of blockage of folding of HasA SM by SecB.
In the Sec system, SecB fulfills two separate functions of slowing down precursor folding and of targeting to SecA, the ATPase of the Sec system (35). In the ABC system of HasA SM secretion, the precise functions of SecB are not known. Here, we have shown that SecB is able to strongly slow down the folding of wild-type HasA SM and that the secretion of the slower folding variants of HasA SM is much less SecB-dependent than the secretion of wild-type HasA SM . Given the fact that folded HasA SM is no longer secretion competent, it, thus, seems that the main role of SecB in the ABC system is slowing down HasA SM folding. The behavior of the two mutants we have constructed also supports the importance of slowing down of folding function of SecB in this system. Although it is difficult to extrapolate from the in vitro folding kinetics to in vivo, it is clear that both mutants fold more slowly than wild-type HasA SM . We conclude that interactions between the N terminus and the C terminus of the protein are at least involved in the rapid folding kinetics of HasA SM . The mutants were secreted to a significant extent in the absence of SecB, they were fully functional, and they were indistinguishable from wildtype in terms of global fold. Although both mutants display a reduced stability as observed with the thermal and chemical denaturations, the very low fraction of unfolded proteins in "native" conditions seems insufficient to explain their efficient secretion in the absence of SecB, especially, in light of the fact that HasA32-75, which has the same reduced stability, is as SecB-dependent as the wild-type. SecB interacts with wild-type and mutant HasA SM so as to slow down their folding. Although able to interact with HasA SM mutants, SecB has little effect on their secretion; this is also what was observed for the SecB independent mutant of MBP. In the Sec system, a cascade of increasing binding affinities allows the transfer of the precursor from SecB to SecA and to the translocon (35). This might not occur in the case of HasA SM if folding still occurs on SecB; efficient secretion would result from competition between folding and secretion. In the Sec system SecB has an essential targeting function to the SecA ATPase and an antifolding function. In the HasA ABC secretion system, SecB has an antifolding function. We have previously shown that the N-terminal part of HasA is involved in efficient secretion and in an early interaction with the transporter. It remains to be determined whether SecB plays a role in this interaction.
We had proposed that both the C terminus of HasA SM bearing the secretion signal and the N terminus part of HasA SM via a SecB-dependent recognition of the transporter were involved in efficient HasA SM secretion (28). The behavior of the mutants might give clues on how this could happen; one of the functions of SecB, upon binding to HasA SM , would keep both ends of HasA SM freely available to the transporter. If one assumes that the SecB binding sites defined by the peptide library technique are relevant to the in vivo situation, there appears to be only one such site in HasA SM fulfilling the criteria defined (9). This site is at the N terminus of HasA SM between residues 18 -28, YLGQWASTFG, with three aromatic residues and no negative charge. The N-terminal position of the putative binding site of SecB and the very fast folding in vitro of HasA SM suggest that the chaperone would bind the nascent chain of HasA SM . This is in agreement with our model of secretion in which HasA SM is targeted in a cotranslational fashion to the ABC transporter (29).