INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

SecA is an essential component of the protein translocation machinery in Escherichia coli (3, 8, 21, 38). It hydrolyzes ATP and uses the energy of this hydrolysis to translocate precursor proteins across the cytoplasmic membrane (5, 6, 24, 25). SecA is composed of 901 amino acids (31) and was initially identified as a soluble and peripheral membrane protein (4, 26). It has been reported that SecA cycles on and off the membrane and that a 30-kDa SecA domain undergoes cycles of membrane insertion and deinsertion during protein translocation (11, 12). Recent studies have found, however, that a significant fraction of SecA behaves like an integral membrane protein (4, 7, 22, 38). This fraction of SecA is resistant to extraction with heparin, Na₂CO₃, alkaline, or urea, all of which are widely used to extract peripheral membrane proteins (4, 7, 22, 38). In a SecDF-overproducing strain, SecA was found almost entirely in an integral membrane form and part of SecA was exposed to the periplasm (22). Despite these apparently unusual findings, this strain still displayed normal protein translocation, as measured by rapid processing of preproteins in vivo. Membranes washed with heparin, which removes all but the integral SecA from the membrane (38), were also active in protein translocation, although Na₂CO₃ or urea treatment partially inactivated this activity (7, 38). However, supplementing the urea-washed membranes with F₁ protein restored the translocation activity (38). These findings indicate that the integral form of SecA is functional.

Electrophysiological measurements have suggested that protein translocation across membranes occurs through protein-conducting channels in both prokaryotes and eukaryotes (33, 34). Such channels have been shown to consist of a heterotrimeric Sec61p complex in yeast and mammalian endoplasmic reticulum membranes (17). SecY and SecE are the homologs of Sec61 and Sec61 (16, 18), which are components of the Sec61p complex in yeast and mammalian cells. Therefore, SecY and SecE might be part of the protein-conducting channel in E. coli. However, it is not clear whether SecA is also a part of the channels, since SecA does not have homologs in endoplasmic reticulum. More recent studies have strongly suggested that SecA might be a part of the channels. SecA and SecY were the only proteins which were cross-linked to translocating proOmpA molecules (20). Furthermore, a fraction of SecA is permanently embedded in the membrane (7), and it can be accessed from the periplasmic side (22, 29, 37). Moreover, the membrane-embedded SecA gave rise to several major fragments after proteinase K digestion, and some of these fragments were obtained in the absence of protein translocation (7). These translocation-independent SecA fragments may form the constant part of the channels. Here we present the identification, through sequencing analysis and immunodetection using region-specific antibodies, of these SecA fragments and the characterization of their interactions with membranes by a variety of chemical treatments. Our findings suggest that there are two different forms of membrane-integral SecA and that this combination of SecA has at least two membrane-interacting domains (MID), which have different interactions with the membrane. Moreover, a 48-kDa fragment represents a SecA domain induced specifically by interaction with the membrane. This fragment may form a part of the protein-conducting channels.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. The following solutions were used where indicated: TK buffer (10 mM Tris-HCl [pH 7.6], 50 mM KCl), TAKM buffer {50 mM Tris-HCl [pH 7.6], 20 mM NH₄Cl, 40 mM KCl, 10 mM magnesium acetate [Mg(OAc)₂], 2 mM dithiothreitol}, translocation buffer (TAKM buffer with 1 mM spermidine trihydrochloride, 8 mM putrescine dihydrochloride), energy source [1 mM ATP, 20 µM GTP, 5 mM phospho(enol)pyruvate, 30 µg of pyruvate kinase/ml], and cushion solution [10 mM Tris-HCl (pH 7.6), 10 mM Mg(OAc)₂, 0.5 M NaCl, and 0.5 M sucrose]. Stop solution [2 mM phenylmethylsulfonyl fluoride (PMSF) in 10 mM Tris-HCl (pH 7.6), 500 mM KCl, and 10 mM Mg(OAc)₂] was prepared immediately before use. MinA medium was prepared as described elsewhere (36). S-Sepharose FF and Sephacryl S-200 were obtained from Pharmacia Biotech. Proteinase K was obtained from Boehringer Mannheim. ATP, GTP, PMSF, spermidine trihydrochloride, putrescine dihydrochloride, trypsin (treated with N-p-tosyl-L-lysine chloromethyl ketone [TLCK]), and soybean trypsin inhibitor were obtained from Sigma. [³⁵S]Met (1,175 Ci/mmol), [³H]Gly (51 Ci/mmol), and [³H]Leu (140 Ci/mmol) were obtained from Dupont-New England Nuclear. Heparin was obtained from CalBiochem.

Preparations of membrane vesicles and protein components for in vitro translocation assays. Membrane vesicles were prepared from SecA-depleted CK1801.4 and D10.2, an Unc derivative of D10, by following procedures described previously (7, 36). proOmpA, SecB, and anti-SecA antibodies were prepared as described elsewhere (7, 36). Radioactive SecA was prepared from BL21 (DE3)/pT7-secA as previously described (3, 7) with the following modifications in order to obtain radioactive SecA with a high specific activity. Cells were grown in 50 ml of MinA medium supplemented with 0.5% glucose and an amino acid mixture (50 µg/ml) lacking either Met, Gly, or Leu. Five millicuries of either [³⁵S]Met, [³H]Gly, or [³H]Leu was used to label the proteins. Labeled SecA was purified by stepwise elution from a 1-ml column packed with SP-Sepharose FF followed by gel filtration chromatography on a Sephacryl S-200 column (1.6 by 60 cm). The final preparations contained more than 98% SecA as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

Protein translocation assay and proteolysis. SecA was reconstituted into SecA-depleted CK-1801.4 membranes according to procedures described previously (7). The reconstituted membranes were incubated at 37°C for 15 min in 100 µl of translocation mixture containing 2 µg of SecB and 1 µg of proOmpA, followed by incubation with 1 mg of proteinase K or trypsin/ml on ice for 15 min. After addition of 0.7 ml of stop solution containing 1 mM PMSF or soybean trypsin inhibitor (final concentration, 2 mg/ml) to stop the proteolysis, the membranes were recovered by centrifugation at 95,000 rpm for 20 min over a 0.2-ml sucrose cushion with a Beckman TL100 centrifuge. Two-thirds of the resulting supernatant was mixed with an equal volume of 16% cold trichloroacetic acid (TCA) and incubated on ice for 30 min. The precipitates were recovered by centrifugation at 14,000 rpm for 10 min in a Jovan A14 centrifuge, washed with 1 ml of cold acetone, and air dried. The radioactive SecA fragments recovered from both the membrane and supernatant fractions were then separated by SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membrane sheets (ProBlott; Applied Biosystems), and visualized either by Coomassie blue staining or by Western blotting with anti-SecA rabbit serum.

Sequencing of SecA fragments. SecA labeled with [³⁵S]Met, [³H]Gly, or [³H]Leu was reconstituted into SecA-depleted CK-1801.4 membranes (2 × 10⁶ to 20 × 10⁶ cpm/mg of membrane protein) by following procedures described previously (7). After translocation reaction and proteolysis, the SecA fragments were separated by SDS-PAGE and transferred to PVDF membranes. Individual bands were excised from the Coomassie blue-stained PVDF membrane sheets with a razor blade and were subjected to Edman degradation with a Beckman LF3200 protein/peptide sequencer in the Biology Department's Molecular Biology Core Facility. The samples from each cycle were collected and counted for their radioactivity with a Beckman LS 6500 liquid scintillation counter, with counting efficiencies of 80% for ¹⁴C and ³⁵S and 35% for ³H. The radioactive peaks identified the positions of labeled amino acids. Since the cleavage sites for trypsin are the carboxyl ends following either Lys or Arg, the N-terminal sequence of each fragment was identified by matching the cycle number of the radioactivity peak(s) to the occurrence between the labeled amino acid and its nearest upstream Arg or Lys of the SecA sequence deduced from its DNA sequence (31). This first identification was confirmed by repeating the radiosequencing of the fragment by using SecA labeled with a different amino acid.

Identification of SecA fragments by using region-specific antibodies. The region-specific antibodies were a generous gift from D. Oliver (Wesleyan University, Middletown, Conn.). Six SecA fragments (A1 [SecA_1-209], A2 [SecA_211-350], A3 [SecA_351-509], A4 [SecA_519-664], A5 [SecA_665-820], and A6 [SecA_822-901]) fused to the C-terminal end of maltose-binding protein were purified from E. coli strains overexpressing these chimeric proteins and were used to raise region-specific antibodies against A1 to A6, respectively (29). The antibodies were diluted 5,000- to 20,000-fold so that they gave similar band densities when the same amount of purified SecA was analyzed. Reconstituted membranes were incubated at 37°C for 15 min (7) and were digested with proteinase K or trypsin at 1 mg/ml on ice for 15 min. After addition of either stop solution or soybean trypsin inhibitor (final concentration, 2 mg/ml) to stop the proteolysis, the reaction mixture was centrifuged at 95,000 rpm for 20 min over a 0.2-ml sucrose cushion with a Beckman TL100 ultracentrifuge. SecA fragments were recovered from both membrane and supernatant fractions, separated by SDS-PAGE, and transferred to PVDF membrane sheets as described above. Individual lanes were then cut from the PVDF membrane sheets and incubated with antibodies against full-length SecA and specific SecA regions separately, followed by incubations with alkaline phosphatase-conjugated secondary antibody and chemiluminescence detection using a kit from Bio-Rad (Hercules, Calif.). The developed membrane sheets were then air dried and exposed to Kodak (Rochester, N.Y.) BioMax MR film to obtain an autoradiogram.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Comparison of proteolysis of membrane-associated SecA with proteinase K and trypsin. It has been shown (7) that the membrane-embedded SecA gives rise to several major fragments after incubation with proteinase K (at 1 mg/ml, on ice, for 15 min). To further characterize these fragments (also see below), we compared the proteolytic profiles of SecA with proteinase K and trypsin at various concentrations and for various periods (Fig. 1). The sizes of the corresponding fragments generated by these two proteases were slightly different, as revealed by both the autoradiograms (Fig. 1A, lanes 1 and 2) and the immunoblots (Fig. 1A, lanes 3 and 4). Although the 66-, 48-, and 36-kDa fragments remained unchanged, the fragments that were 28 and 25 kDa with proteinase K digestion moved up to 29 and 27 kDa, respectively, with trypsin digestion (Fig. 1A, lanes 1 and 2). From now on, these two fragments will be designated the 29- and 27-kDa fragments for consistency and simplicity.

View larger version (44K): [in this window] [in a new window]   FIG. 1.   Proteolysis of membrane-integrated SecA with proteinase K and trypsin. (A) 35S-labeled SecA fragments in membranes generated by proteolysis with proteinase K or trypsin. Reconstituted membranes (r) or SecA-depleted control membranes (c) from CK1801.4 were subjected to protein translocation reaction and proteolysis as described in Materials and Methods. The resulting SecA fragments in the membrane fraction were analyzed by SDS-PAGE and autoradiography (lanes 1 and 2) or by immunoblotting using anti-SecA antibodies (lanes 3 to 6). Lanes 1 and 2 and lanes 3 and 4 were from the same gel but were developed differently. The positions of the molecular-marker proteins bovine serum albumin (68 kDa), ovalbumin (45 kDa), and carbonic anhydrase (29 kDa) are shown by bars. The sizes of the protease-resistant SecA fragments are indicated by arrows. (B) Proteolysis of membrane-associated SecA with proteinase K or trypsin was carried out as for panel A, except that the different protease concentrations indicated were used. (C) Kinetics of proteolysis of membrane-integrated 35S-labeled SecA. Proteolysis was performed as for panel A except that the incubation was carried out for the times indicated.

Identification of the major protease-resistant SecA fragments by radiosequencing. Radiosequencing was used to identify the protease-resistant SecA fragments without purification. Trypsin was used instead of proteinase K in proteolysis because of its specificity in cleaving only the peptide bonds after Arg or Lys residues. The specificity of trypsin cleavage offers the advantage of providing an additional marker or index in aligning amino acids after radiosequencing. Figure 2 shows the radiosequencing profiles of the major trypsinized SecA fragments. The 48-kDa fragment recovered from the membrane fraction gave a radioactive peak at cycle 2 with [³H]Gly-labeled SecA and one at cycle 12 with [³H]Leu-labeled SecA but showed no peaks with [³⁵S]Met-labeled SecA in the first 14 cycles (Fig. 2A). Searching the SecA sequence for (K/R)XG(9X)L determined that this sequence only matches Lys₃₆₀ to Ser₃₇₄. Thus, the 48-kDa fragment was identified as an internal fragment starting after Lys₃₆₀ at Glu₃₆₁. The 66-, 36-, and 27-kDa fragments were similarly identified as N-terminal fragments starting from Met₁ by the same method (Fig. 2A and data not shown). The 95- and 90-kDa fragments were not sequenced because of their small amounts. The 29-kDa fragment, which potentially corresponds to the translocation-dependent 30-kDa fragment identified previously (7, 11, 12), apparently contained at least three species, starting at Glu₃₆₁, Ser₂₆₉, and Ala₄₇₂. All three fragments are different from the reported translocation-dependent 30-kDa fragment, which starts at Leu₆₁₀ (27). The 29-kDa fragment starting at Glu₃₆₁ was named the s29-kDa fragment because the supernatant 29-kDa band constituted about 80% of this species (Fig. 2) (7) and contained a single fragment which starts at Glu₃₆₁ (Fig. 2B). On the other hand, the 66-, 36-, and 27-kDa fragments from the supernatant fraction were shown to start at Asn₁₄ (Fig. 2B and data not shown). These fragments clearly correlate to their counterparts on the membranes, which start at Met₁. Thus, the N-terminal 13 amino acids might play a role in anchoring these fragments on the membrane. However, these fragments are not released from the membranes by extensive digestion of their receptors on the membrane (14), since similar fragments could be generated in the absence of membrane (described below). The significance of the presence of these fragments in both fractions and of the absence of the 48-kDa fragment in the supernatant fraction is discussed below where the requirement for the formation of each fragment is examined.

View larger version (33K): [in this window] [in a new window]   FIG. 2.   Radiosequencing of the protease-resistant SecA fragments. After translocation and proteolysis as described in Materials and Methods, SecA fragments recovered from both membrane pellet (A) and supernatant (B) fractions were separated by SDS-PAGE, transferred to PVDF membrane sheets, and visualized by Coomassie blue staining or immunodetection. Individual bands were then excised from the PVDF membrane sheets and subjected to radiosequencing as described in Materials and Methods. The radiosequencing profiles for each fragment are aligned with the identified sequences. The cleavage sites of trypsin are indicated by arrows. The sequences identified were unique. There is no other match in the SecA sequence except for the 29-kDa fragments, for which all possible sequences are shown.

Identification of SecA fragments by using region-specific antibodies. To confirm the radiosequencing identification of these tryptic fragments of SecA and their relationship to the proteinase K fragments, reconstituted membranes were treated with proteinase K (Fig. 3A) or trypsin (Fig. 3B), separated by SDS-PAGE, and developed by immunoblotting using the region-specific antibodies against six consecutive SecA regions (A1 to A6). SecA-depleted CK1801.4 membranes treated in the same way were used as a control to detect nonspecific bands. The data from both protease treatments clearly showed that the 66- and 36-kDa fragments were the N-terminal fragments (A1 and A2), while the 48-kDa fragment represents a center domain (regions A3 to A5). This supports the radiosequencing identification of these fragments (Fig. 2). Several bands, including a 66-kDa band detected by anti-A6 (Fig. 3A), are probably nonspecific, since these bands were also detected in the SecA-depleted control membranes. The 90- and 95-kDa fragments generated by proteinase K were recognized by antibodies against all regions except the A6 region. Therefore, these fragments represented forms of SecA with C-terminal deletions, which have been reported previously (37). The 95- and 90-kDa bands generated with trypsin were not as clear but were detected by antibodies against SecA2 and SecA5. Taking the sizes (A1 to A5, 820 amino acids; A2 to A6, 700 amino acids) into consideration, these fragments should also be the forms of SecA with C-terminal deletions. Immunodetection of the 29- and 27-kDa fragments was not clear due to nonspecific bands. There are probably many species in this area, but most of them were found within regions A1 to A4, although some faint bands were also observed in regions A5 to A6. SecA fragments corresponding to the 66-, 36-, 29-, and 27-kDa fragments in the membrane were also found in the supernatant fraction (7). The 66-, 36-, and 29-kDa fragments obtained with both protease treatments were identified as fragments corresponding to regions A1 to A4, A1 to A2, and A3 to A4 of SecA (see Fig. 7B), respectively. The 27-kDa band was not observed with proteinase K digestion in Fig. 3A but was identified as a fragment corresponding to regions A1 to A2 in a separate experiment (data not shown). A minor 48-kDa band in the supernatant fraction was recognized by anti-A5, but it was not detected by antibodies against A3 and A4 or by autoradiogram (Fig. 3B). Thus, this 48-kDa band is probably nonspecific or might represent a small amount of the 48-kDa SecA fragment released from the membranes. In conclusion, immunodetection with region-specific antibodies demonstrated that SecA fragments generated by both proteases represented the same SecA domains and confirmed the radiosequencing results. Moreover, the 66- and 48-kDa fragments overlapped in the central region (SecA_361-610), as both of them were detected by anti-SecA3 and anti-SecA4. On the other hand, each was recognized specifically by anti-SecA1 (and anti-SecA2) and anti-SecA5, respectively. Therefore, these two fragments are derived from different SecA molecules.

View larger version (70K): [in this window] [in a new window]   FIG. 3.   Identification of SecA fragments on membrane and the supernatant fractions by region-specific antibodies. CK1801.4 membranes (c) and reconstituted membranes (r) were subjected to translocation and digestion with proteinase K (A) or trypsin (B) as described in Materials and Methods. The SecA fragments in both the membrane fractions and the supernatant fractions were separated by SDS-PAGE, transferred to PVDF membrane sheets, and developed both by autoradiography and by immunoblotting with specific antibodies against different regions of SecA as indicated. Mb, membrane; A0, whole SecA; Ctrl, SecA fragments recovered from membrane fraction. For A1 to A6, see Materials and Methods.

Characterization of the formation of the SecA fragments. To elucidate the relationship between the formation of these SecA domains and protein translocation, we examined the formation of the SecA fragments in various translocation stages (Fig. 4A). Reconstituted membranes were incubated at 37°C under protein translocation conditions for the times indicated and then trypsinized on ice for 15 min. As observed previously with proteinase K digestion (7), all the major fragments were detected under no-translocation conditions (Fig. 4A, lane 1). After 3 min of incubation, the 95-, 66-, and 29-kDa fragments increased significantly while the other fragments remained unchanged. There were no obvious changes in the proteolysis profile of SecA after longer translocation periods (3 to 30 min). Taking into consideration that release of SecA from the membrane was completed within the first 3 min (7), these data suggest that SecA adapts its conformation in the membrane, which causes the formation of the individual domains, in the early stages of protein translocation.

View larger version (33K): [in this window] [in a new window]   FIG. 4.   Characterization of the protease-resistant SecA domains. (A) Proteolysis of membrane-associated SecA at different translocation stages. After translocation and proteolysis as described in Materials and Methods, SecA fragments recovered from the membrane fraction were separated by SDS-PAGE and visualized by autoradiography. Positions of molecular-marker proteins are shown by bars. (B) Effects of ATP, proOmpA, and membrane on the formation of the proteolytic SecA fragments. 35S-labeled SecA was incubated at 37°C for 15 min under the conditions indicated. After the incubation, the reaction mixture was incubated with 1 mg of trypsin/ml on ice for 15 min. The SecA fragments generated in the absence of membrane were TCA precipitated and were analyzed by SDS-PAGE and autoradiography (lanes 1 to 4). The proteolytic SecA fragments generated in the presence of membranes were separated into membrane and supernatant fractions (Fr.), followed by SDS-PAGE and autoradiography (lanes 5 to 12). Positions for molecular-weight marker proteins are indicated. (C) Limited proteolysis of SecA in the absence of membranes (Mb). After incubation in translocation mixture minus membranes, 35S-labeled SecA was incubated with 60 µg of trypsin/ml on ice for 15 min and precipitated by 8% TCA. The pelleted 35S-labeled SecA fragments were separated by SDS-PAGE and visualized by autoradiography (lanes 6 and 8) and by immunodetection with anti-SecA2 (lane 2) and anti-SecA5 (lane 4). SecA fragments from trypsinized (1 mg/ml; 15 min) membranes were analyzed for comparison (lanes 1, 3, 5, and 7).

Characterization of interactions between the SecA fragments and the membrane. To determine the nature of the interactions between the SecA fragments and the membrane, the stability of these fragments was examined with extraction by acid (1% acetic acid), alkali (0.1 M NaOH), a high concentration of salt (1 M potassium acetate [KOAc]), heparin (10 mg/ml), 0.1 M Na₂CO₃, and 6 M urea. Extraction with TK buffer was used as a control. All fragments were resistant to extraction by a high concentration of salt or by heparin (Fig. 5A). Since heparin specifically removes peripheral SecA (7, 38), these results indicate that all fragments in the membrane are integral and that the membrane integration is independent of ionic interactions. The 48- and 27-kDa fragments were partially extracted by Na₂CO₃ and were extensively extracted by urea (Fig. 5B), suggesting a more hydrophilic environment. In contrast, the majority of the 66- and 36-kDa fragments were not extracted with Na₂CO₃ or urea (Fig. 5B). Therefore, these fragments might interact with lipids or proteins in the membrane through hydrophobic interactions. Extraction with acetic acid or alkali gave profiles similar to those given by extraction with a high concentration of salt or with Na₂CO₃, respectively (data not shown). All these fragments were also observed when the reconstituted membranes were extracted with urea or Na₂CO₃ prior to trypsinization (Fig. 5C), indicating that all these fragments were derived from integral SecA.

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Nature of protease-resistant SecA domains in the membrane. (A) Extraction of protease-resistant SecA fragments with a high concentration of salt and with heparin. After translocation and proteolysis, the reconstituted membranes were recovered by centrifugation and either solubilized with SDS-sample buffer (no extraction) or resuspended in 100 µl of TK buffer (TK), 1 M KOAc, or 10 mg of heparin/ml in duplicates. After incubation on ice for 30 min, the extracted membranes were recovered by centrifugation at 95,000 rpm for 40 min in a Beckman TL100 centrifuge and were analyzed by SDS-PAGE and autoradiography. (B) Extraction of protease-resistant SecA fragments with Na2CO3 or urea. The procedure was the same as for panel A except that the extraction was performed with 0.1 M Na2CO3 (CO3) or 6 M urea and both the membrane fraction (Fr.) and the supernatant fraction (after TCA precipitation) were analyzed. (C) Proteolysis of membrane-associated SecA after extraction with Na2CO3 or urea. Reconstituted membranes were recovered by centrifugation after translocation, incubated with 0.1 M Na2CO3 (CO3) or 6 M urea in duplicates on ice for 30 min, recovered again by centrifugation, resuspended in translocation buffer, and incubated with 1 mg of trypsin/ml on ice for 15 min. Both the membrane fraction (Fr.) and the supernatant fraction were analyzed. The positions for the molecular-weight marker proteins are indicated.

Chemical sequencing analysis of SecA fragments. Since some of the membrane-associated SecA fragments could be extracted by urea (Fig. 5), we sequenced these fragments after extraction, as well as those recovered from the supernatant and those generated in the absence of membranes (Table 1). In order to compare our results to the reported sequencing results of SecA fragments extracted from urea-washed membranes (27), SecA was reconstituted into urea-washed D10.2 membranes and CK1081.4 membranes. Proteolysis of these two kinds of reconstituted membranes generated similar SecA fragments, as revealed by immunodetection using region-specific antibodies (data not shown). Chemical peptide sequencing of the 48-kDa fragment extracted from the D10.2 membranes by urea revealed a sequence starting at Glu₃₆₁ (Table 1), the same starting sequence as for the 48-kDa fragment from the reconstituted membranes identified by radiosequencing (Fig. 2) and immunodetection (Fig. 3). This fragment was not observed in the supernatant fraction and was not generated in the absence of membrane. The 50-kDa fragment generated in the absence of membrane, which is close to the 48-kDa fragment in size, was an N-terminal fragment starting at Val₉. The 29-kDa fragments from both the supernatant fraction and the urea extracts also start at Glu₃₆₁, matching the sequence of the s29-kDa fragment obtained by radiosequencing (Fig. 2A). The other two possible 29-kDa fragments identified by radiosequencing (Fig. 2A) were not detected in the urea extracts. The one starting at Ser₂₆₉ might have remained on the membrane because of its hydrophobicity (see Discussion), and the other one, which starts at Ala₄₇₂, might be too scarce to be detected. The 66-kDa fragment in the urea extracts is the same as the 66-kDa fragment recovered from the supernatant, since both fragments start at Asn₁₄. These results confirmed that the 66-kDa SecA fragment starting at Met₁ was not extracted by urea. Trypsin digestion of SecA in the absence of membrane generated a 66-kDa fragment starting at Val₉. This fragment was not observed in the membrane fraction. In contrast, sequencing analysis of the 36-kDa band in the urea extracts revealed three sequences starting at Met₁, Val₉, and Asn₁₄. The latter two fragments were also found in the supernatant fraction. These two 36-kDa fragments either contained no Leu or Met in their first 9 residues or were minor on the membrane, so that only the 36-kDa fragment starting from Met₁ was detected by radiosequencing (Fig. 2A). On the other hand, all three 36-kDa fragments (Table 1) are present in similar amounts in urea extracts. These results indicate that the majority of the 36-kDa fragment starting from Met₁ was not extracted by urea; indeed, most 36-kDa fragments were resistant to urea extraction (Fig. 5). This observation indicates that the urea-resistant 36-kDa fragment must be the one starting from Met₁. This major 36-kDa fragment and the 66-kDa fragment starting at Met₁ were not found in the supernatant fraction. Collectively, these results suggest that the N-terminal 8 or 13 amino acids may serve as one of the anchors for membrane-integral SecA. Therefore, removal of these peptides resulted in the release of the N-terminal SecA fragments from the membrane. Most importantly, the chemical sequencing data confirmed that the 48- and 66-kDa fragments are overlapping fragments and thus are derived from different SecA molecules.

Characterization of 30-kDa SecA fragments. A C-terminal 30-kDa fragment was reported to undergo a translocation-dependent insertion-deinsertion cycle during protein translocation (11, 12, 27). However, we were unable to detect this 30-kDa fragment by radiosequencing. On the other hand, several weak bands around 30 kDa were observed when the SecA fragments were visualized by anti-A5 or anti-A6 (Fig. 3 and 4). Therefore, we examined the translocation dependence of these fragments by using anti-SecA5 (Fig. 6). Normal anti-SecA antibody and anti-SecA2 were used for comparison. Nonradioactive SecA and ³⁵S-labeled proOmpA were used to allow simultaneous detection of SecA fragments and OmpA translocation. Two SecA fragments around 30 kDa were detected by anti-SecA5, but none of them increased upon translocation of proOmpA. Since the translocation-dependent 30-kDa fragment could be extracted by urea (27), trypsinized membranes were treated with urea by following procedures described previously (27), and the extracted SecA fragments were separated by SDS-PAGE and transferred to PVDF membrane sheets. Several bands were observed around 30 kDa by Coomassie staining. Protein sequence analysis of these bands identified SecA fragments starting from Met₁, Val₉, Asn₁₄, Glu₃₆₁, and possibly Ser₅₇₅ (Table 1). Furthermore, a C-terminal 30-kDa fragment was also identified as starting at SecA₆₄₄ when SecA was digested with 0.02 mg of trypsin/ml in TAKM buffer (Table 1). However, we have not been able to detect the reported translocation-dependent C-terminal 30-kDa fragment starting at Leu₆₁₀ (27). These results suggest that this C-terminal 30-kDa fragment probably represents an unstable or minor species in the membrane after proteolysis at high protease concentrations, and thus it is undetectable in our system.

View larger version (42K): [in this window] [in a new window]   FIG. 6.   Characterization of 30-kDa fragments by region-specific antibodies. Nonradioactive SecA was reconstituted into SecA-depleted CK1801.4 membranes. The reconstituted membranes were incubated with or without 35S-labeled proOmpA in translocation buffer with the energy source at 37°C for 15 min and were digested with 1 mg of trypsin/ml on ice for 15 min. The resulting SecA fragments were separated by SDS-PAGE, transferred to PVDF membrane sheets, and visualized by immunodetection with region-specific antibodies (A). Arrows indicate the two "30-kDa" bands detected by anti-A5. The same PVDF membrane sheets were also visualized by autoradiography to show the translocated OmpA (B).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

A significant fraction of SecA has been shown to be permanently embedded in the membranes and does not cycle on and off the membrane during protein translocation (7). In contrast to findings that only a 30-kDa domain of SecA is resistant to proteolysis (11, 12, 27), several SecA domains in the membrane were found to be resistant to proteolysis, and some of them were independent of protein translocation (7). We have now identified and characterized these protease-resistant SecA fragments through immunodetection using region-specific antibodies, chemical extraction, and sequencing analysis. The observed proteolytic patterns suggested that the SecA proteins in the membrane are cleaved at two main sites, in addition to the C-terminal cleavage (37). One of these sites is at Lys₃₆₀, giving rise to the N-terminal 36-kDa fragment and the central 48-kDa fragment. The other site is located around SecA₆₁₀, giving rise to the N-terminal 66-kDa fragment and perhaps a C-terminal 30-kDa fragment which has not been identified in the present study (Fig. 7B). The 66- and 48-kDa fragments partially overlap in sequences (Fig. 7B), indicating that they originate from different SecA molecules. The most obvious explanation is that there are two different membrane-integral forms of SecA, SecA_S and SecA_M, based on the different behaviors of the overlapping 66- and 48-kDa fragments. The 66-kDa fragment increases significantly in the presence of ATP hydrolysis and protein translocation (Fig. 4) and can be chased off the membrane by excess nonradioactive SecA (7) but is resistant to Na₂CO₃ extraction (Fig. 5). In contrast, the 48-kDa fragment is not affected significantly by ATP hydrolysis and protein translocation (Fig. 4) and is relatively more sensitive to Na₂CO₃ extraction (Fig. 5) but is resistant to chasing by excess nonradioactive SecA (7). SecA_S may have the same conformation as soluble SecA in buffer, which also gives rise to an N-terminal 65-kDa fragment (27), although the protease resistance is 10- to 100-fold greater than that for free SecA. On the other hand, SecA_M is induced by interaction with the membrane, since the formation of a specific 48-kDa fragment requires the presence of membranes (Fig. 4B). Therefore, the 66- and 48-kDa fragments are specifically derived from SecA_S and SecA_M, starting at Met₁ and Glu₃₆₁, respectively. The 36-, s29-, and 27-kDa fragments could come from either SecA_S or SecA_M or both. However, they most likely come from SecA_S because, like the 66-kDa fragment, they are dependent on ATP and independent of membranes (Fig. 4 and Table 1). It has been suggested that there is an equilibrium between a soluble form and a hydrophobic form of SecA, and that the latter is stable only on the membrane (28). Here, we present evidence that the membrane-integral SecA probably possesses two different conformations as revealed by proteolysis. Our previous findings, however, showed that the proteolytic 48-kDa fragment can still be generated from membranes even after incubation with excess nonradioactive SecA under protein translocation conditions (7); therefore, at least a significant fraction of SecA_M does not participate in this equilibrium. Since SecA is known to be functional as a dimer (1, 10), it may form a conformational heterodimer of SecA_S-SecA_M in the membrane. This is supported by the observation that the molar ratio of the 66-kDa fragment and the 48-kDa fragment was close to 1:1 (Fig. 5B) when membranes were treated with proteases after urea or Na₂CO₃ extraction, which removes peripheral SecA. Alternatively, there may be two populations of SecA homodimers. The different proteolysis patterns of these two forms of SecA may also reflect their differences in interacting with the membrane. Work is under way to differentiate these possibilities and determine the different interacting components.

View larger version (19K): [in this window] [in a new window]   FIG. 7.   Summary of identification and characterization of SecA fragments. (A) Schematic presentation of proposed SecA domains. The two ABDs of SecA (ABD I and ABD II) (25) are shown as shaded rectangles, and the domains interacting with precursor (23, 37) or with lipid and SecB (2) are shown as solid rectangles, on the SecA sequence. The known and proposed MID are shown by open bars below the SecA sequence. (B) Schematic presentation of the identification of SecA fragments. The identified SecA fragments and the six SecA fragments fused to maltose-binding protein to produce the region-specific antibodies are aligned with the SecA sequence illustrated in panel A. Starting amino acid numbers are given for the major fragments of the integral SecA. The open box represents the potential cleavage fragment corresponding to the reported 30-kDa fragment starting at Leu610 (27) but not identified here. (C) Proposed proteolysis pathway for SecAS and SecAM. SecAS, membrane-integral SecA with a conformation similar to that of soluble SecA; SecAM, membrane-integral SecA with a membrane-induced conformation.

The limited number of SecA fragments observed in proteolysis with two different proteases under various conditions suggests that SecA assumes definite conformations on the membrane, in which only limited sites are exposed to protease digestion. The observed SecA fragments thus represent individual SecA domains which are resistant to proteolysis in reconstituted membranes as well as in wild-type membranes (data not shown) (see also reference 9). The major protease-resistant SecA fragments can be divided into two groups: the "N-terminal" group, containing the 66-, 36-, and 27-kDa fragments, and the "internal" group, containing the 48- and s29-kDa fragments. These results suggest that SecA has at least two MID (Fig. 7), one of which is located within SecA_240-330. Removal of this region from the 36-kDa fragment, which is resistant to urea extraction, gave rise to the 27-kDa fragment, which could be extracted by urea; therefore, this region plays an important role in anchoring SecA on the membrane. In support of this notion, SecA_300-350 has been shown to be periplasmically accessible (29). The other MID is possibly located within the C-terminal 20-kDa region of the 48-kDa fragment, or SecA_610-800, since removal of this region could give rise to the soluble s29-kDa fragment. The extreme N-terminal 8 or 13 amino acids may also possess a third MID, since the N-terminal fragments found in the supernatant fraction are missing these 8 or 13 amino acids (Fig. 2B). In addition, the extreme C terminus of SecA may possess yet another MID. This region is known to interact with lipids (2) and is exposed to the periplasm (29, 37).

Proteolysis of SecA_S in the membrane could give rise to the N-terminal 66-kDa fragment and the C-terminal 30-kDa fragment, as is apparently the case for the limited proteolysis of free SecA in buffer (27). We have identified the translocation-enhanced 66-kDa fragment but have been unable to detect a major stable C-terminal 30-kDa fragment. One possibility is that this domain, if it exists, represents an unstable or minor species and therefore its presence may be masked by others in this range. Indeed, we found a 31-kDa fragment which might correspond to a C-terminal SecA fragment starting at Ser₅₇₅ (Table 1). The observation of this 30-kDa fragment as a major species (11, 12, 27) is due to the uneven artificial labeling of SecA by ¹²⁵I, as predicted (7). It has recently been confirmed that the radioactivity in ¹²⁵I-labeled SecA is essentially confined to the C-terminal 30-kDa region (13) and that proteolysis of metabolically labeled ³⁵S-SecA gave rise to an N-terminal 65-kDa fragment and several other fragments in addition to the C-terminal 30-kDa domain of SecA that were protected from proteolysis (14). This N-terminal 65-kDa fragment apparently corresponds to the N-terminal 66-kDa fragment here. Protection of this fragment appears to depend on ATP, not on membranes, because similar amounts of this fragment were observed in the presence of ATP with or without membrane, and the majority of the 66-kDa fragments generated in the presence of membranes were found in the supernatant fraction (Fig. 4). TCA precipitation could bring down the 66-kDa fragment from the supernatant fraction, thus producing a stronger 66-kDa band and an apparently greater increase with addition of proOmpA (data not shown). Therefore, the reported large increase of a 65-kDa N-terminal SecA fragment upon translocation of proOmpA was probably due to analyzing the total SecA fragments precipitated by TCA (14). This 66- or 65-kDa domain can be further cleaved at around SecA₃₃₀ into the N-terminal 36-kDa domain, which contains ABD I, and the s29-kDa domain, which contains ABD II (Fig. 7). Since proOmpA binds to SecA at SecA_267-340 (23), such binding may shield the cleavage site on the 66-kDa domain, thus increasing the amount of the 66-kDa fragment. Indeed, a modest increase of the 66-kDa fragment in the membrane fraction was observed (7). Therefore, protection of the 66- or 65-kDa N-terminal SecA domain is mainly due to ATP-induced conformational change, not merely due to membrane insertion. The apparent translocation dependence of this domain is probably due to shielding of the cleavage site by proOmpA, preventing further digestion of the 66-kDa fragment.

Proteolysis of SecA_M gave rise to the major 48-kDa fragment. Several lines of evidence suggest that the 48-kDa domain forms part of the protein-conducting channel. First, this domain is located immediately after the precursor binding region (23) and contains ABD II (25), which is considered important for coupling the insertion-deinsertion cycle of the 30-kDa SecA domain to protein translocation (11). Second, formation of this domain is induced by interactions with membranes (Fig. 4). Third, this domain is embedded in the membrane. There are 18 lysines and arginines in the 48-kDa domain. It is unlikely that all these sites are embedded inside the SecA molecule and are thus inaccessible to proteases. In fact, this fragment disappeared when the membranes were incubated with 1 mg of proteinase K/ml in the presence of 1% Triton X-100, which disrupts the membranes (7). Fourth, this domain is permanently embedded in the membrane. It is insensitive to ATP or protein translocation (Fig. 4) and cannot be chased out from the membrane by excess nonradioactive SecA (7). Fifth, this domain is in an aqueous environment like the translocating precursor proteins (15), since it was partially extracted by urea, Na₂CO₃, or alkali but was resistant to extraction with a high concentration of salt. Sixth, this domain (SecA_610-800) might possess an important SecY-binding determinant (35), thus coupling the SecA channel and the other Sec machinery. This major 48-kDa fragment was not observed in a previous study (14). It is not clear whether this is related to the use of the urea-treated membranes there. The authors did find a major SecA fragment of about 48 kDa in the absence of membrane. This fragment probably correlates to our 50-kDa N-terminal fragment (Fig. 4C and Table 1), since both of them were generated only with low concentrations of trypsin. In contrast, the membrane-specific 48-kDa domain was resistant to protease ranging from 0.1 to 10 mg/ml (Fig. 1 and data not shown). Such protection is not likely to be due to nonspecific association of SecA with lipids. This 48-kDa domain and SecA_M play an important structural role as translocation core or channel, rather than being nonfunctional (14). The structural role for SecA_M explains why it is insensitive to ATP hydrolysis and protein translocation.

The current dogma depicts SecYEG as forming the core of the protein translocase, while SecA is the peripheral subunit that goes through the cycles of membrane insertion-deinsertion and that cycles on and off the membrane during translocation (39). However, it is not certain how the SecA dimer (1, 10) functions in the model, and the perpetual recycling of SecA is extremely inefficient physiologically. Moreover, recent studies have found that SecYG-deficient (38, 41) or SecE-deficient (40) membranes still had more than 50% translocation activity of proOmpA and other precursors with processing of the signal peptides. Thus, the SecYEG complex is not obligatorily required for all precursors but may enhance the efficiency and specificity. (The requirement of SecYEG for a subset of precursor proteins, like prePhoA, also is consistent with results of genetic studies showing that secY and secE are the essential genes [19, 30].) If SecYEG is not essential for the translocation of all proteins, and thus is not the core or channel for protein translocation, then what? The results presented here suggest that SecA might be the intrinsic core or channel. Membranes treated with urea have been shown to contain functional SecA (38). The majority is probably SecA_M, since the 48-kDa fragment became a major band when both the reconstituted and the native membranes were digested with trypsin after urea extraction (Fig. 5 and data not shown). Taking into account that some SecA goes through cycles of membrane insertion-deinsertion (39), there may be two forms of SecA in the channel, one that is dynamic and one that is static. The finding of two conformational forms of integral SecA, SecA_S and SecA_M, perhaps as a heterodimer, fulfills this requirement. It is tempting to speculate that the dynamic SecA_S may go through cycles of translocation-dependent insertion-deinsertion as proposed elsewhere (14, 39) while the static SecA_M is permanently embedded in the membrane, interacting with SecY and other membrane proteins to form the protein-conducting channel. Taking into account all the available data, the simplest model puts the integral SecA (both SecA_S and SecA_M) in the core and moves SecYEG from the core to a role as accessory proteins like SecDFYajC, important but not essential components. In this scenario, the roles of SecA and SecYEG are analogous to those of ATP hydrolysis and proton motive force in protein translocation, in which one is essential and the other contributes significantly to efficiency (5, 40). The roles of SecYEG and SecA in protein translocation are thus uncertain and controversial. Further work is needed to reveal the details and resolve differences in interpretation.