Identification of the Preprotein Binding Domain of SecA*

SecA, the preprotein translocase ATPase, has a helicase DEAD motor. To catalyze protein translocation, SecA possesses two additional flexible domains absent from other helicases. Here we demonstrate that one of these “specificity domains” is a preprotein binding domain (PBD). PBD is essential for viability and protein translocation. PBD mutations do not abrogate the basal enzymatic properties of SecA (nucleotide binding and hydrolysis), nor do they prevent SecA binding to the SecYEG protein conducting channel. However, SecA PBD mutants fail to load preproteins onto SecYEG, and their translocation ATPase activity does not become stimulated by preproteins. Bulb and Stem, the two sterically proximal PBD substructures, are physically separable and have distinct roles. Stem binds signal peptides, whereas the Bulb binds mature preprotein regions as short as 25 amino acids. Binding of signal or mature region peptides or full-length preproteins causes distinct conformational changes to PBD and to the DEAD motor. We propose that (a) PBD is a preprotein receptor and a physical bridge connecting bound preproteins to the DEAD motor, and (b) preproteins control the ATPase cycle via PBD.

, light blue) that form between them a mononucleotide crevice (7,9,10,12). DEAD motors acquire specificity for different catalytic processes through add-on nonhomologous structural appendages (8,11). SecA has two such structures (12)(13)(14) (Fig. 1, A and B): (a) the C-domain (aa 611-832), which is fused C-terminally to the IRA2 domain of the DEAD motor and regulates its mobility and properties by "stapling" together NBD and IRA2 (14,15), and (b) a second appendage (aa 221-377) of unknown function, which forms an independent structural domain ( Fig.  1A, magenta) that comprises a bilobate "Bulb" bridged with NBD through a "Stem" formed of two anti-parallel ␤ strands (␤1 and ␤7; Stem out and Stem in , respectively; Fig. 1, B and C). The Stem is physically "rooted" in NBD (Fig. 1, A and C) without disturbing its structural integrity. As it protrudes out of NBD, the aa 221-377 appendase "embraces" loosely, mainly through Bulb-mediated contacts, parts of the C-domain (9,10,15). 3 At least two distinct Bulb conformational states have been identified in crystallographic studies of Bacillus subtilis SecA (10,16).
Numerous studies have demonstrated preprotein binding to SecA (12, 19, 24 -29). However, where precisely on SecA preproteins bind and how they stimulate its ATPase remains unresolved. This central question must be answered before the molecular basis of SecA-dependent preprotein movement through translocase can be understood. The Stem and Bulb substructures have been indirectly implicated in preprotein binding (12,24). Isolated NBD fragments that also include the ␤1 strand ( Fig. 1C) retain signal peptide binding of high affinity, whereas deletion of the ␤1 strand in SecA prevents signal peptide binding (12). However, this study could not conclusively determine whether signal peptide binding occurs on the Stem or whether the Stem is indirectly involved in binding. In another study, N-terminal SecA fragments that were refolded from inclusion bodies and included parts of the Bulb (aa 267-340) were cross-linked to a complete preprotein (24). However, the specificity of this reaction is unclear, since (a) cross-linking occurred only when SecA was partially "reconstituted" by mixing N-terminal fragments together with C-terminal peptides, and (b) the cross-linker used was nonspecific. Now we have investigated the functional role of the SecA domain that encompasses residues 221-377. We demonstrate that this domain is essential for cell viability, protein translocation, and preprotein-stimulated translocation ATPase. Moreover, we provide direct evidence that this region is essential for preprotein loading onto SecYEG-containing membranes and retains the ability to bind preproteins in solution even when detached from SecA. This region is therefore a preprotein-binding domain (PBD). Further, we show that PBD has two distinct domains with distinct subsites (Stem and Bulb) with discrete functional subsites. SecA binds signal peptide through the PBD Stem and mature regions through the PBD Bulb. These binding events influence PBD and DEAD motor conformation and subsequently DEAD motor ATP hydrolysis. We propose that PBD due to its strategic location acts as a physical and functional bridge between preproteins and the ATPase DEAD motor.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Recombinant DNA Experiments-E. coli strains were grown and manipulated as described (7,14,30).
Membrane Flotation Experiments-Ultracentrifugal sedimentation experiments were carried out in a bench-top ultracentrifuge (TLX120 Optima; TLA120.2 rotor; Beckman), using polypropylene tubes (0.2 ml) as described (34,35). Reactions (15 l in buffer B (50 mM Tris-Cl, pH 8, 50 mM KCl, 5 mM MgCl 2 )) containing SecA or SecA⌬Bulb (0.5 g) in the presence or absence of inner membrane vesicles (IMVs) (Յ3 g) and proOmpA-His (0.12 g) (36) were adjusted to 1.74 M final sucrose concentration and deposited at the bottom of the tube. Samples were overlaid with one layer (20 l) of 1.6 M sucrose and two consecutive layers (75 l) of 1.4 and 1.25 M sucrose in buffer B following centrifugation (4°C; 436,000 ϫ g; 180 min). Nine fractions of 25 l were removed, analyzed by SDS-PAGE, and visualized by immunostaining. SecA was visualized with ␣-SecA antibodies. proOmpA was immunostained with ␣-hexahistidinyl antibodies to selectively label freshly bound proOmpA and not the abundant indigenous OmpA population in the IMVs. Flotation of IMVs inside the gradient was monitored by immunodetection of the integral membrane protein SecY.
NMR Spectroscopy-Isotopically labeled samples for NMR studies were prepared by growing the cells in M9 media. All NMR spectra were recorded on a Varian 600-MHz spectrometer. Sequential assignment of the 1 H, 13 C, and 15 N protein chemical shifts of N219 -379 was achieved by means of through-bond heteronuclear scalar correlations along the backbone using conventional three-dimensional pulse sequences (37).

RESULTS
PBD Is Essential for SecA-dependent Translocation-To analyze the effect of PBD on SecA function, we used the following mutants of oligohistidinyl-tagged SecA (Fig. 1C): SecA⌬219 -240, which lacks the Stem out (12,13); SecA⌬351-368, in which a 17-amino acid deletion removes the highly conserved helix 4 of Bulb 2 (13); SecA⌬233-365 (hereafter SecA⌬Bulb), which is missing the entire Bulb region but retains ␤1 and ␤7 of the Stem joined directly with a linker; and SecA(W349A), in which a highly conserved aromatic residue found in all SecA proteins is substituted by alanine.
The ability of SecA PBD mutants carried on a plasmid to rescue the BL21.19 secAts strain (30) at 42°C was examined (Fig. 1D). None of the deletion mutants could complement BL21.19 in vivo, whereas secA(W349A) could, albeit poorly.
It became clear from the above results that PBD is essential for SecAmediated protein translocation. To understand at which step of the translocation pathway the role of PBD is essential, we proceeded to characterize the effect of the mutations on each step of the reaction.
SecA PBD Mutants Retain Nucleotide Binding and ATP Hydrolysis-To examine the effect of PBD on the ability of SecA to bind nucleotide, we determined the equilibrium binding constants of SecA mutants for nucleotide (TABLE ONE), using a MANT-ADP fluorescence-based assay (15,21). All mutant proteins retain the ability to bind nucleotide, with either similar (SecA⌬351-368 and SecA(W349A)) or somewhat reduced (SecA⌬219 -240 and SecA⌬Bulb) affinities to that of SecA (TABLE ONE).
Binding of ADP by SecA causes conformational changes that lead to its thermal stabilization (7,10,14,39). To examine whether ADP binding elicits a similar conformational change in SecA PBD mutants, we monitored intrinsic Trp fluorescence upon thermal melting of the mutant proteins in the presence or absence of ADP (15,21). Under these experimental conditions, the T m(app) of SecA increases upon ADP binding by 5°C (TABLE TWO). All mutants display ADP-driven thermal stabilization, albeit to a different extent than SecA (TABLE TWO).
Finally, to examine the effect of PBD mutations on the ability of SecA to hydrolyze ATP, we determined the K cat for the basal ATPase activity of all SecA PBD mutants. The basal catalytic properties of SecA remained unaffected by mutations in PBD (TABLE THREE).
Mutations within PBD (including complete Bulb removal) abrogated neither basal ATP catalysis nor ADP association to SecA. Therefore, it seemed unlikely that PBD, which is located outside the ATP cleft ( Fig.  1A), is a physical determinant essential for nucleotide binding to/hydrolysis by the DEAD motor. Instead, PBD could have a regulatory role in DEAD motor catalysis.
SecA PBD Mutants Bind to SecYEG-containing Membranes-To examine the effect of PBD mutations on the ability of SecA to bind to SecYEG and to form the translocase ternary complex, we employed membrane flotation (34,35). Inverted IMVs, treated with urea to remove endogenous SecA (19), were loaded at the bottom of a sucrose gradient (corresponding to lanes 7-9 in Fig. 2) and were floated up during ultracentrifugation. Gradient fractions were subsequently analyzed by SDS-PAGE, and the migration position of IMVs in the gradient ( Fig. 2A, fractions 4 and 5) was identified on Western blots by immunostaining for SecY, the integral membrane subunit of translocase. SecA immunostaining revealed that the IMV preparation used in these experiments was not contaminated by endogenous SecA traces (Fig. 2B). SecA alone (Fig. 2C) or mixed with IMVs ( Fig. 2D) were loaded at the bottom of identical sucrose gradients. The position of SecA within the gradient was identified on Western blots by immunostaining with ␣-SecA antiserum. In the presence of IMVs, a significant portion (Ͼ50%) of SecA migrated to the less dense parts of the gradient and was present in the same fractions as the SecYEG-containing IMVs (Fig. 2, compare D with A). This indicated that SecA was bound to the membrane vesicles. This SecA-SecYEG association (Fig. 2D) is independent of nucleotide addition or temperature (data not shown) (3,18). In contrast, SecA remained at the bottom of the gradient (Fig. 2C, lanes 7 and 8). All SecA PBD mutants exhibited behavior identical to that of wild type SecA in this assay; they all associated stably with SecYEG-IMVs independently of nucleotide addition or temperature. Here only SecA⌬Bulb is shown as an example (Fig. 2, compare E with D).  (30) using secA PBD mutants. BL21.19 cultures carrying pET5 vector alone or its derivative with cloned secA, secA⌬219 -240, secA⌬351-368, secA⌬Bulb, or SecA(W349A) genes grown at 30°C were adjusted to the same density. The indicated dilutions (10 n ) were spotted on LB/ampicillin plates and incubated (20 h; 42°C). E, PBD is essential for SecA-mediated translocation. In vitro translocation of [ 35 S]proOmpA into SecYEG-proteoliposomes by the SecA PBD mutants. Assays were performed in 50 l of buffer B containing the indicated SecA proteins (40 g/ml), SecB (20 g/ml), bovine serum albumin (400 g/ml), [ 35 S]proOmpA (20,000 cpm), and SecYEG-proteoliposomes (250 g/ml). Reactions were initiated by the addition of 2 mM ATP and incubated for 20 min at 37°C. After proteinase K treatment (1 mg/ml, 15 min, 4°C), proteins were precipitated (12.5% (w/v) trichloroacetic acid) and separated on 15% SDS-PAGE. [ 35 S]proOmpA protected from proteinase K treatment was detected by phosphorimaging.
These experiments demonstrated that mutations or deletions within PBD do not prevent SecA binding to SecYEG. Thus, it seems unlikely that PBD is an essential physical determinant for assembly of the SecA-SecYEG holoenzyme.
SecA PBD Mutation Prevents Recruitment of proOmpA to the Translocase-To examine whether mutations in PBD affect SecA interaction with preproteins, we used flotation gradients (Fig. 2, F-J). Untreated IMVs alone (Fig. 2F) or IMVs mixed with proOmpA complexed with its cognate chaperone SecB (Fig. 2H) or mixed with SecA, proOmpA, and SecB ( Fig. 2I) were fractionated as in Fig. 2, A-E. As a control, a mixture of SecA, proOmpA, and SecB was fractionated in the absence of IMVs (Fig. 2G). Gradient fractions were analyzed by SDS-PAGE, and the migration position of proOmpA was identified using an ␣-hexahistidinyl antiserum. This antiserum does not cross-react nonspecifically with the IMVs in the gradient (Fig. 2F). In the presence of SecA, proOmpA co-migrates in the gradient with SecYEG-containing IMVs (Fig. 2I, lanes 4 and 5; compare with Fig. 2A, lanes 4 and 5). In the absence of either SecA (Fig. 2H) or of SecYEG-containing IMVs (Fig.  2G), proOmpA remained at the bottom of the gradient (Fig. 2, lanes 7  and 8). This SecA-dependent proOmpA binding to SecYEG-IMVs was independent of nucleotide or temperature (data not shown). In contrast to SecA, SecA⌬Bulb failed to recruit proOmpA to the translocase under identical experimental conditions (Fig. 2J). Very low amounts of proOmpA, comparable with the background nonspecific binding of proOmpA to IMVs (Fig. 2H) (18), are seen to co-migrate with the IMVs.
We conclude that the PBD domain of SecA is essential for preprotein recruitment to the membrane-embedded SecYEG-SecA translocase holoenzyme.
PBD Mutations Compromise Preprotein-stimulated Translocation ATPase-To further study the role of PBD in the SecA-preprotein interaction, we monitored changes in the ATPase activity of SecA (19). Soluble SecA ATPase activity is low ("basal ATPase" ; Fig. 3A, lane 1) (19) and becomes marginally stimulated upon SecA binding to SecYEG at the membrane ("membrane ATPase"; lane 2). SecA ATPase becomes significantly stimulated (3-6-fold over membrane ATPase) only following preprotein addition ("translocation ATPase"; Fig. 3, A (lane 3) and B). It was anticipated that mutants that fail to interact with the preprotein will fail to acquire translocation ATPase activity. Indeed, only SecA(W349A) retained some translocation ATPase activity albeit to ϳ50% of the wild type levels (compare lane 15 with lane 3). No detectable preprotein-stimulated ATPase activity was observed with the SecA PBD deletion mutants (compare lanes 6, 9, and 12 with lane 3). Titration of proOmpA demonstrated that the membrane ATPase of SecA PBD mutants could not be stimulated by preprotein even at elevated concentrations (Fig. 3B).
These data indicated that SecA PBD mutants fail to acquire preprotein-simulated ATPase activity, suggesting that their interaction with preproteins is compromised.
A Model Preprotein with Stable Binding to Soluble SecA-Collectively, our data suggested that PBD is required for preprotein interaction with SecA (Figs. 2 and 3). We next sought to test whether this reflected direct binding of preproteins to PBD. However, widely used and well characterized model SecA-dependent preproteins, such as proOmpA, are not appropriate for such binding studies, since they bind very poorly to soluble SecA (19,40). 4 To overcome this limitation, we resorted to using an alternative model preprotein: proM13coatH5EE (hereafter 4 S. Karamanou and A. Economou, unpublished results.

Endothermic transitions of SecA derivatives
The intrinsic tryptophan fluorescence of the proteins was monitored as a function of temperature, as described (6 referred to as pCH5EE) (31,32). pCH5EE, like proOmpA (Fig. 2), requires SecA for its membrane binding (supplemental Fig. 2) and for translocation into the membrane (31) and outcompetes the binding of proOmpA to membranes (31). Its short length, which satisfies a minimal size required for SecA-mediated secretion (73 aa) (41), and its tight binding to soluble SecA (see below; TABLE FOUR) made pCH5EE optimally suited for mapping preprotein binding surface(s) on SecA.
Mature Regions of Preproteins Bind to N68-Signal peptides bind to the N68 (a polypeptide that encompasses residues 1-610 of SecA and comprises the NBD, IRA2, and PBD domains) (12,28). In contrast, direct physical binding of mature preprotein regions to N68 has never been demonstrated. To determine whether mature preprotein regions can bind to N68, CH5EE was immobilized on an optical biosensor, and the equilibrium dissociation constants for SecA as well as for N68 were determined (TABLE FOUR and Fig. 4B). Both proteins bind to CH5EE with comparable low micromolar affinities. Similar results were obtained using a CH5EE(W49F) biosensor (data not shown).
To corroborate these findings, we used an activity-based binding assay. Binding of the model signal peptide 3K7L suppresses N68 ATPase (12). As anticipated, M13SP had a similar effect in this functional assay (Fig. 4C). The addition of pCH5EE or CH5EE also led to suppression, whereas that of proOmpA did not. This lack of suppression by proOmpA may correlate with its low affinity for soluble SecA (19,40).
These data demonstrated that the three amino-terminal domains of SecA (610 residues) are necessary and sufficient to bind signal peptide (12) and mature preprotein regions of pCH5EE.
Isolated PBD Binds Preproteins-We next sought to identify which SecA domain present in N68 is responsible for preprotein binding. To this end, we used N1-419, a fragment that contains both the NBD and T, translocation ATPase) were incubated (1 mM ATP, 30 min; 37°C). ATP hydrolysis was measured as described (19). B, translocation ATPase activity of SecA PBD mutants (as in A) as a function of proOmpA concentration (as indicated). Activity is expressed as n-fold stimulation over "membrane ATPase" measured in the absence of proOmpA.   DECEMBER 30, 2005 • VOLUME 280 • NUMBER 52

SecA Preprotein Binding Domain
PBD domains and N420 -610, a fragment that encompasses all of IRA2 (12). In addition, the sequence encoding residues 219 -379, that fully encompasses PBD (aa 221-377), was cloned as an independent gene. The derived oligohistidinyl-tagged polypeptide (N219 -379) was soluble, chromatographically stable, and monomeric (supplemental Figs. 3  and 4). To determine its folding, N219 -379 was isotopically labeled, and its fingerprint two-dimensional 1 H-15 N heteronuclear single quantum coherence spectrum was recorded by NMR (Fig. 5A). Backbone assignment revealed that the Bulb region of N219 -379 is well folded in solution, as evidenced by the dispersion of the corresponding peaks. However, the Stem element is clearly unfolded and very flexible even at temperatures as low as 10°C. Most likely, tethering of the Stem to NBD in SecA is important to stabilize the Stem.
To analyze binding of N1-419, N420 -610, and N219 -379 to preproteins, we used an optical biosensor with immobilized pCH5EE. The isolated PBD fragment retains binding to immobilized pCH5EE (TABLE FOUR), albeit with reduced affinity compared with SecA and N68. N1-419 binds to pCH5EE with an affinity indistinguishable from that of SecA and N68 (TABLE FOUR). In contrast, no detectable binding was measured with N420 -610, indicating that IRA2 is not involved in preprotein binding. We next sought to test the specific role of NBD in the binding reaction constructing and using N1-419⌬Bulb in the biosensor system. However, all ⌬Bulb derivatives tested aggregated on the sensor surface and could not be studied in this assay system (data not shown).
We concluded that isolated PBD can bind preproteins, although optimal binding activity may require either its tethering to NBD or the presence of specific NBD residues.
Isolated PBD Binds both Signal Peptide and Mature Regions-Can N219 -379 bind to distinct preprotein moieties? To address this question, we monitored changes in the tryptic accessibility of specific Arg/ Lys residues (Fig. 6A) upon the addition of either M13SP or CH5EE or pCH5EE (Fig. 5B). In the absence of any ligand, trypsin-cleaved N219 -379 at Arg 220 (at the base of the Stem) and at Lys 360 (located at the Bulb; lane 2). The presence of each of the preprotein segments resulted in distinct and characteristic tryptic patterns. In the presence of M13SP, Arg 220 and Lys 360 became less accessible (compare amounts of N219 -  Following SDS-PAGE, polypeptides were immunostained with ␣-N68-specific antibodies (12,15,21), and their identity was determined by N-terminal sequencing. A representative of six experiments is shown. Identical results were observed when SecA was used in the same assay (data not shown). lanes 3 and 2), whereas Lys 329 became exposed (lane 3; see p221-328). The addition of CH5EE (lane 4) prevented efficient cleavage at Arg 360 (compare p221-360 in lanes 4 and 2). Finally, the addition of pCH5EE exposed Lys 329 (compare p221-328 in lanes 5 and 2). Under identical conditions, no changes to the tryptic profile of the IRA2 subdomain fragment (Fig. 1A) were observed (data not shown).

in
We concluded from the tryptic accessibility assays that the isolated PBD polypeptide can bind complete preproteins, signal peptides, and mature regions.
Bulb Deletion Affects Binding of Mature Preprotein Regions-PBD is composed of a Bulb and a Stem (Fig. 1, A and B). Are both elements necessary for preprotein binding? Deletion of the ␤1 strand of the Stem prevents signal peptide binding to N68 (12). To determine whether the Bulb substructure of PBD is required for preprotein binding, we compared binding of preprotein segments with N68 and with N68⌬Bulb. To this end, we developed a novel fluorescence-based assay.
Changes in the intrinsic fluorescence of N68 as a function of temperature were monitored upon the addition of either the nonfluorescent derivative CH5EEW49F or M13SP (Fig. 5C). In this assay, N68 displayed a characteristic T m(app) (44°C) that was reduced when M13SP (by 5°C) or CH5EEW49F (by 2.5°C) were added, indicating that binding of either preprotein moiety to N68 caused its destabilization. The addition of either ligand had a similar effect on SecA (data not shown).
As observed with N68, the addition of M13SP caused a reduction in the T m(app) of N68⌬Bulb (by 3.5°C; Fig. 5D), suggesting that Bulb deletion does not prevent the N68/M13SP interaction. However, in contrast to what was observed for N68 (Fig. 5C), the addition of CH5EEW49F failed to cause any reduction in the T m(app) of N68⌬Bulb (Fig. 5D), suggesting that Bulb deletion compromises binding of the mature region but not that of signal peptide to N68.
Our data suggested that binding of mature regions to SecA may require the Bulb (Fig. 5D), whereas binding of signal peptides to SecA may require the Stem (12).
Signal Peptide and Mature Preprotein Regions Induce Distinct PBD Conformations-We next sought to determine the effect of preprotein binding to PBD within the physiological context of the complete SecA and its fully functional N68 derivative. To this end, we examined the tryptic profile of SecA and N68 upon the addition of M13SP, CH5EE, or pCH5EE. In this experiment, the tryptic profile of SecA (data not shown) and N68 (Fig. 6B) are identical. Of all of the resulting peptides, only the five (indicated in Fig. 6B) that result from cleavage within or near PBD (Fig. 6A) were selected.
In the absence of preprotein fragments, three of the five peptides were seen (lane 1). When CH5EE was added, all five peptides were detected (lane 2). When M13SP was added none of the five peptides was detected (lane 3) due to significant protease protection of N68 (see remaining N68 amounts; top). This proteolytic protection presumably reflects a more compact conformational state. Trypsin activity is not inhibited by signal peptides (12). Finally, when pCH5EE was added three of the peptides were prominent, and the other two were barely detectable (lane 4).
It seemed that when complete preproteins, signal peptides, and mature regions bound to PBD attached to the DEAD motor, they stabilized distinct PBD conformations.
A 25-aa Mature Region Peptide Binds to PBD-We next sought to determine whether the full length of the mature preprotein region is required for binding to PBD. To address this possibility, a peptide that encompasses the first 25 mature region residues (hereafter referred to as CH5EE24 -48; Fig. 4A) was chemically synthesized. SecA (Fig. 7A, lane 2) and N68 (lane 3) bind to CH5EE24 -48 immobilized on an optical biosensor, whereas a control protein did not (lane 1). Binding of CH5EE24 -48 suppresses N68 ATPase (Fig. 7B) although clearly less efficiently than CH5EE, whereas a control peptide of similar length (p13) did not.
These data indicated that the N-terminal 25 aa of CH5EE can bind to PBD albeit inefficiently. This observation identified for the first time a minimal mature domain segment capable of binding to SecA.

DISCUSSION
Using two different preprotein substrates and derivative peptides, SecA mutants, and isolated domains, we determined that the second "substrate specificity" appendage of SecA (residues 221-377) is a PBD. Further biochemical dissection led to a two-subsite model: PBD binds signal peptides with its Stem and mature preprotein regions with its Bulb (Fig. 5). PBD is required for cell viability (Fig. 1D), protein translocation (Fig. 1E), preprotein-stimulated translocation ATPase (Fig. 3), and loading of preproteins to the membrane-embedded translocase (Fig. 2). PBD is therefore an essential element of the bacterial protein translocase catalytic cycle.
Signal peptide binding to the Stem of PBD is supported by several lines of evidence: (a) it is prevented by a deletion mutation in the Stem (12), (b) it can occur minimally within the aa 220 -234 region of PBD that encompasses Stem out (12), (c) it does not require the Bulb (Fig. 5D), and (d) it prevents tryptic cleavage at Arg 220 at the base of the Stem (Figs. 5B and 6B). Several hydrophobic residues on the Stem (see hypothetical model in Fig. 8) could provide a binding surface for residues of the signal peptide hydrophobic core shown by NMR studies to face the pocket (29). One of these residues suppresses defective signal peptides when mutated (A373V) (42). An extreme carboxyl-terminal SecA peptide may "shield" these hydrophobic residues in cytoplasmic SecA (10), thus preventing premature preprotein binding.  Fig. 4B). Refractive index change was measured at equilibrium (t ϭ 750 s). Maltose-binding protein (MBP) is used as a nonbinding control. B, suppression of N68 ATPase in the presence of the indicated peptides (as in Fig. 4C). p13 is a chemically synthesized peptide (25 aa) from the yeast ␣ factor.
The mature preprotein region binds to SecA or N68 ( Fig. 4B and TABLE FOUR) or to the isolated PBD ( Fig. 5B and TABLE FOUR) but not to N68⌬Bulb (Fig. 5D). Binding of the mature preprotein region invariably affects the accessibility of Bulb residue Lys 360 (Figs. 5B and 6, A and B) in tryptic digestion assays. These data indicate that the Bulb binds mature preprotein regions. Such a specific binding function could explain why signalless preproteins still require SecA (43) and could rationalize previous cross-linking data (24). Cavities (10) and a groove formed between Stem, Bulb, NBD, and the IRA1 helical hairpin (Fig. 8) (16) could accommodate mature region peptides. Bulb mutations (44 -47) may affect this binding.
Preproteins bind distinctly to the isolated PBD substructures identified here ( Fig. 5B and TABLE FOUR). Nevertheless, the measured affinity of preproteins for PBD is reduced compared with that for the N1-419 fragment that contains both NBD and PBD (TABLE FOUR). We therefore anticipate that additional surfaces or particular residues from NBD may optimize preprotein binding. Electrostatic contacts between signal peptides and SecA were proposed to require NBD residues Asp 217 and Glu 218 at the root of the Stem (29). In other Superfamily 1 and 2 helicases, some DEAD motor residues are known to contribute to oligonucleotide substrate binding (48 -50). In addition, the possibility cannot be excluded that preprotein molecules longer than the minimal preprotein used here may make additional contacts to other parts of SecA outside of PBD.
The small size of CH5EE24 -48 suggests that a limited Bulb surface may be sufficient for a minimal primary binding. This site could reject positive charges in the mature region N terminus (51-54) and could act as a "molecular ruler" dictating the 20 -30-aa step size seen in OmpA translocation (23,55). Interestingly, a 30-residue region downstream of the signal peptide was proposed to act as an "export initiation domain" (54). However, binding of CH5EE24 -48 is less efficient than that of CH5EE (Fig. 7B), suggesting that longer peptides may be better retained on the Bulb or may even engage additional parts of SecA.
Preprotein signal and mature region peptides can bind to their respective PBD subsites independently of each other in our in vitro assays (Fig. 5). Nevertheless, in bona fide secretory proteins signal peptides and mature regions are covalently adjoined and would therefore be expected to occupy both PBD subsites simultaneously. Preprotein binding to the two physically connected and spatially proximal subsites affects PBD conformation (Figs. 5B and 6B). This involves movement of residues at both the base of the Stem (Arg 220 /Lys 386 ) and within the Bulb (Arg 270 , Lys 360 , and Lys 329 ) (Fig. 6A) and could influence Bulb swiveling around the Stem (16). The binding of signal peptide, mature region, and full-length preprotein affects PBD conformation in distinct ways (Figs. 5B and 6B). At the same time, the conformation of the Bulb is altered by signal peptide binding to the Stem (Fig. 5B) (12). These data raise the possibility that Stem-Bulb allosteric communication could facilitate "latching" of mature regions onto SecA. Tethering of PBD to NBD through a "rooted" Stem could optimize formation of this ternary complex by enhancing the measured affinity of isolated PBD for the preprotein (TABLE FOUR). NMR analysis indicated that once detached from NBD, the Stem region of PBD becomes unstructured (Fig. 5A). Binding to SecA promotes the acquisition of ␣-helical structure by signal peptides (29). This change in the signal peptide may affect the possible conformations that the Stem can acquire in SecA. The availability of NMR-based atomic resolution tools (Fig. 5A) (29) will greatly facilitate the analysis of PBD conformation and dynamics and its interaction with the preproteins in atomic detail.
Our data and those of others (10,16,17,24,25,56) lead to a multistep working model where the mobile PBD domain acts as a flexible "control lever" that controls the mechanical activities of the SecY-bound SecA nanomotor. (a) Preprotein binds to PBD (Figs. 5B and 6B and TABLE FOUR) (24). (b) Preprotein binding modulates PBD conformation (Figs. 5B and 6) (12). (c) This change is transmitted through the Stem to the DEAD motor and affects the conformations of NBD (Fig. 6B), where PBD is physically rooted (Fig. 1, A and B), but also that of IRA2 as judged by tryptic accessibility of Arg 586 /Lys 609 ( Fig. 6B; peptides p361-610 and p361-586). (d) Altered PBD conformation is also transmitted to the C-domain residues (10,25), such as the helical hairpin of the IRA1 global regulator to which the Bulb is attached through defined contacts (Fig. 8, yellow) (15). (e) The net result of these conformational changes (Figs. 5C and 6B) detach the C-domain "staple" from its NBD and IRA2 binding sites (7,14,15), thereby loosening and destabilizing the DEAD motor (Figs. 5C and 6B). PBD mutations that display reduced ADP binding (TABLE ONE) and stabilization (TABLE TWO) may mimic this state. (f) "Loosening" of the DEAD motor affects ADP release (7,17,56) and nucleotide cycling (Fig. 4C). (g) ATP binding to the empty DEAD motor reverses the previous conformational events and drives the PBD-attached preprotein into the membrane. PBD could "rotate" between the two distinct conformational states identified crystallographically (10,16). PBD may be free to acquire additional as yet unidentified conformational states when the C-domain is detached. (h) ATP hydrolysis, triggered via helicase Motif III (12,21), partially releases the bound preprotein (23). (i) SecA returns to the stable ADP state (7,17) and is primed for a new catalytic cycle triggered by preprotein binding to PBD.