Post-liberation cleavage of signal peptides is catalyzed by the site-2 protease (S2P) in bacteria

Abstract

A signal peptide (SP) is cleaved off from presecretory proteins by signal peptidase during or immediately after insertion into the membrane. In metazoan cells, the cleaved SP then receives proteolysis by signal peptide peptidase, an intramembrane-cleaving protease (I-CLiP). However, bacteria lack any signal peptide peptidase member I-CLiP, and little is known about the metabolic fate of bacterial SPs. Here we show that Escherichia coli RseP, an site-2 protease (S2P) family I-CLiP, introduces a cleavage into SPs after their signal peptidase-mediated liberation from preproteins. A Bacillus subtilis S2P protease, RasP, is also shown to be involved in SP cleavage. These results uncover a physiological role of bacterial S2P proteases and update the basic knowledge about the fate of signal peptides in bacterial cells.

A signal peptide (SP) at the N terminus of secretory protein precursors (preproteins) is cleaved off by signal (leader) peptidase (1) and left behind in the membrane, typically assuming the type II (N_in-C_out) transmembrane configuration. In mammalian cells, the liberated SPs receive further cleavage by signal peptide peptidase (SPP) and are released from the membrane (2–4). Small peptides derived from SP sometimes act as a regulatory molecule (3).

SPP belongs to intramembrane-cleaving proteases (I-CLiPs), which are classified into SPP/γ-secretase (aspartyl proteases), rhomboid (serine protease), and S2P (zinc metalloprotease) (5, 6). These proteases liberate otherwise membrane-tethered domains of membrane proteins to function as a regulatory molecule. I-CLiPs have specific intramolecular routes that make water molecules accessible to the intramembrane proteolytic active sites (5, 6). For such proteolysis to occur, γ-secretase/SPP and S2P require, in most cases, removal of the ectodomain of the substrate by other protease (5, 7). SPP and S2P prefer substrate with the type II transmembrane orientation and γ-secretase and rhomboid prefer the type I orientation (7).

Bacteria contain S2Ps and rhomboids but not SPPs, and only limited knowledge is available about the fate of bacterial SPs (7). We and others have been characterizing RseP, an Escherichia coli member of the S2P involved in the σ^E pathway extracytoplasmic stress response (8, 9). In this regulation, a protease, DegS, responds to misassembled outer membrane proteins and introduces the first proteolytic cleavage into RseA, a membrane-integrated anti-σ^E protein (10). Subsequently, RseP introduces the second cleavage into RseA, activating σ^E to transcribe stress-inducible genes (8, 9). Although RseA is the only physiological substrate of RseP so far established, we have shown previously that RseP can cleave a wider range of transmembrane sequences having helix-destabilizing residues (11). Our preliminary results that RseP cleaved a β-lactamase (Bla) fusion protein at or around its SP (11), together with the fact that RseP and SPP share the substrate preference for the type II orientation, prompted us to undertake the present study. Our results suggest strongly that S2Ps (E. coli RseP and Bacillus subtilis RasP) are involved in degradation of remnant SPs left in the bacterial cytoplasmic membrane, in contrast with the currently prevailing concept that SppA (12, 13), a protease unrelated to the S2P or the SPP family, is the enzyme responsible for signal peptide cleavage in bacteria.

Results

RseP-Dependent in Vivo Cleavage of β-Lactamase SP.

The secretory precursor of Bla has a signal peptide of 23 residues. We characterized proteolysis observed with HA-MBP-SP^Bla-Bla (11), having a hemagglutinin (HA)–maltose-binding protein (MBP) domain (HA–MBP domain) attached to the N terminus of the Bla precursor (Fig. 1A). Immunoblotting using anti-HA detected two HA-MBP-SP^Bla-Bla bands in ΔrseP cells (Fig. 1B, lane 2): the full-length protein (Full) and a smaller species designated uncleaved (UC). Notably, rseP⁺ cells produced an additional and prominent, faster-migrating band designated cleaved (CL) (lane 1 in Fig. 1B). Pulse-labeling and anti-HA immunoprecipitation experiments verified the above observations (Fig. S1, Lower, lanes 2 and 4). Production of UC and CL was diminished markedly (Fig. S1, Lower, lanes 1 and 3) by treatment of cells with NaN₃, a SecA inhibitor, which prevented translocation/maturation of envelope proteins (Fig. S1, Upper, OmpA), as well as by the lep-9 mutation affecting Lep, the major E. coli signal (leader) peptidase (Fig. 1D; see below).

Fig. 1.

RseP-dependent cleavage of HA-MBP-SP^Bla-Bla. (A) Schematic of HA-MBP-SP^Bla-Bla and its cleavage by Lep and RseP. (B) Detection of cleavage products. Cells of AD1811 (rseP⁺) (lane 1) and KK211 (ΔrseP) (lane 2), each carrying pSTD849 (HA-MBP-SP^Bla-Bla), were grown at 30 °C in L broth that contained IPTG and cAMP for induction of HA-MBP-SP^Bla-Bla. Total cellular proteins were analyzed by SDS/PAGE and anti-HA immunoblotting. Intact HA-MBP-SP^Bla-Bla (Full) and putative RseP-uncleaved (UC) and RseP-cleaved (CL) forms of Lep-processed HA-MBP-SP^Bla are indicated. (C) The UC-to-CL conversion depends on proteolytic activity of RseP. Strain AD2328 (ΔrseP)/pSTD849 was transformed further with pKK12 (RseP, lane 1), pTH18cr [vector (vec), lane 2], or pAS90 [RseP(H22F), lane 3] and analyzed by anti-RseP and anti-HA immunoblotting. (D) Lep dependence of the RseP-dependent cleavage of Bla SP. Cells of IT42 (lep⁺, lanes 1–3) and IT41(lep-9, lanes 4–6), each carrying pSTD849, were grown in M9 medium first at 32 °C and then at 42 °C for 20 min, induced with isopropyl 1-thio-β-d-galactopyranoside (IIPTG) and cAMP for 10 min, and pulse-labeled with [³⁵S]methionine for 1 min followed by chase. Labeled proteins were precipitated with anti-OmpA (Upper panels) and anti-HA (Lower panels).

The in vivo production of CL (presumably from UC) required the protease activity of RseP, as it took place in cells expressing the wild-type enzyme (RseP) but not in cells expressing an active-site mutant form of RseP [RseP(H22F)] with an alteration in the H²²ExxH motif (Fig. 1C). These results suggest that UC represents the N-terminal HA-MBP-BlaSP segment (HA-MBP-SP^Bla) generated by Lep and that CL is produced from UC by RseP-mediated proteolysis within SP^Bla. In the above experiments, a significant portion of HA-MBP-SP^Bla-Bla accumulated as the full-length form. Probably, this resulted from inefficient targeting of the model protein having the N-terminally attached HA–MBP domain, but not from jamming of the translocon with overexpressed fusion protein, as proOmpA processing was not appreciably affected (Fig. 1D and Fig. S1).

RseP Directly Cleaves Synthetic Bla SP Peptides in Vitro.

We designed a chemically synthesized substrate, Myc-SP^Bla-Flag, in which the Bla SP (SP^Bla) was sandwiched between Myc and Flag sequences (Fig. 2A). The extra sequences were anticipated to increase solubility and facilitate detection of the peptide. Incubation of the synthetic peptide with purified RseP-His₆-Myc (11) in detergent at 37 °C resulted in the generation of smaller fragments (designated F1 and F2) at the expense of the full-length substrate (Fig. 2B, lanes 1–4). This conversion depended on the proteolytic activity of RseP because it was not observed with the purified RseP(H22F)-His₆-Myc (lanes 7 and 8) and was inhibited with 1,10-phenanthroline, a Zn²⁺ chelator (lanes 9 and 10). Thus, RseP directly cleaved Myc-SP^Bla-Flag. We also examined whether RseP acted against SP^Bla-Flag having no N-terminal tag (Fig. S2A). This peptide was also cleaved by RseP, generating two fragments (F3 and F4) albeit at lower efficiency due possibly to their low solubility (Fig. S2B).

Fig. 2.

Proteolysis of chemically synthesized Bla SP by purified RseP. (A) Amino acid sequence of Myc-SP^Bla-Flag. The Myc, Flag, SP^Bla, N22, and C21 segments are indicated. The amino acid numbers in SP^Bla start from the bla initiator methionine. (B) In vitro reaction. Substrate peptide (19.8 μM) was incubated at 37 °C with RseP-His₆-Myc (0.64 μM; lanes 1–6, 9, and 10), RseP(H22F)-His₆-Myc (0.64 μM; lanes 7 and 8), or buffer alone (lanes 11 and 12). 1, 10-Phenanthroline (PT; 5 mM) was included as indicated (lanes 9 and 10). Samples were analyzed by 12% Nu PAGE (Invitrogen) and Coomassie Brilliant Blue G-250 staining. F1 and F2 indicate cleavage products of the substrate peptide.

RseP Cleaves Bla SP Within the Hydrophobic Core.

We extracted the F1 and F2 peptides from the gel bands and determined their identities by mass spectrometric (MS) analysis (Fig. S3). The MS spectra of F1 and F2 displayed a major peak of m/z = 2,596.49 (Fig. S3 A and B) and m/z = 2,461.13 (Fig. S3C), which represented the N-terminal 22 residue fragment (N22) and the C-terminal 21 residue fragment (C21) of Myc-SP^Bla-Flag, respectively. MS/MS analysis confirmed their sequence identities (Fig. S3 D and E). Also, MS analysis of F3 and F4 (Fig. S2 C and D) showed that these bands corresponded to the N-terminal 12-residue fragment and the C-terminal 21-residue fragment of SP^Bla-Flag, respectively. These results indicate that RseP principally hydrolyzes the Pro12-Phe13 peptide bond of Bla SP in vitro.

We then characterized RseP-dependent cleavage of Bla SP in vivo by cysteine scanning-modification experiments (Fig. 3) (11). A series of derivatives having an engineered and unique cysteine at different positions within the SP were constructed and CL was examined to see whether it contained the introduced cysteine. We used modifiability with methoxypolyethylene glycol 5000 maleimide (malPEG), a thiol-alkylating reagent of about 5 kDa, to assess the presence of cysteine. HA-MBP-SP^Bla-Bla derivatives with a cysteine at position 9, 12, or 15 produced UC in ΔrseP cells and both CL and UC in rseP⁺ cells. Whereas CLs from the Cys-12 (Fig. 3, lanes 5–8) and Cys-15 (Fig. 3, lanes 9–12) variants were unmodifiable, that from the Cys-9 variant (Fig. 3, lanes 1–4) received malPEG modification (see Fig. S4 for controls and additional data). Thus, the in vivo-produced CL should have contained the N-terminal 9 residues of Bla SP but not the 12th residue and its C-terminal side. Thus, an RseP cleavage point in vivo may lie somewhere between the 9th and the 12th residues of Bla SP. Although this assignment deviates slightly from the cleavage site determined in vitro (see Discussion for possible cause of this apparent discrepancy), our in vivo and in vitro results collectively indicate that RseP cleaves Bla SP within the central hydrophobic region.

Fig. 3.

Assessment of the RseP-dependent in vivo cleavage site of HA-MBP-SP^Bla-Bla. Total cellular proteins from cells of AD1811 (rseP⁺) and KK211 (ΔrseP), expressing one of the HA-MBP-SP^Bla-Bla variants having a single cysteine at the indicated position of SP^Bla, were treated with or without 5 mM malPEG and analyzed by SDS/PAGE and anti-HA immunoblotting. All of the variants were derivatives of the SP-Cys-less (C18A) form of HA-MBP-SP^Bla-Bla that had no cysteine residue in the SP part. “Full-malPEG(3×)” indicates the full-length protein with all three cysteine residues (one in SP and two in the Bla mature part) modified with malPEG.

RseP Cleavage of SP Requires a Preceding Processing of Preproteins by Lep.

We addressed whether Lep-mediated SP processing is a prerequisite for the cleavage of SP, using the lep-9(Ts) mutation that compromises the Lep activity (14) (Fig. 1D). Pulse-chase experiments showed that the Lep-mediated processing of proOmpA was delayed markedly in the lep-9 mutant cells (Upper). Notably, the production of not only UC but also CL from HA-MBP-SP^Bla-Bla was diminished in this mutant (Lower). These results suggest that proteolysis of SP by RseP requires preceding processing by Lep.

Generality of RseP-Dependent SP Cleavage.

To address the generality of RseP-catalyzed SP cleavage, we attached HA-MBP to the precursors of selected outer membrane, periplasmic and cytoplasmic membrane proteins of E. coli, each having a cleavable SP (Fig. S5A). In ΔrseP cells, all of the fusion proteins produced fragments (designated UC) that reacted with anti-HA (Fig. 4) and had sizes similar to that of UC from HA-MBP-SP^Bla. They were found to produce smaller fragments (designated CL) in rseP⁺ cells. This was shown for OmpF, LivK, SecM, PhoA, LivJ, OmpC, and Lpp derivatives using rseP⁺ cells grown at 30 °C or 37 °C (Fig. 4A, lanes 1–14). Detection of CL from the OmpA derivative required overproduction of RseP, whereas RseP(H22F) was ineffective (Fig. 4B). Thus, RseP has the ability to cleave SPs from these proteins in vivo. The RseP dependence on CL production was less strict in the cases of the M13 coat and TolC and RbsB constructs; in these cases, low but significant amounts of CL-like fragments were produced even in the ΔrseP cells (Fig. 4A, lanes 15–20). SPs of these proteins may be subject to low-efficiency cleavage by some protease other than RseP. GlpG, the E. coli homolog of the rhomboid family I-CLiP, was not involved in the RseP-independent cleavage of SPs as the ΔglpG mutation exerted no apparent effect on the production of the CL-like fragment (Fig. S5B).

Fig. 4.

Cleavage of SPs from various presecretory proteins in vivo. (A) Cleavage of SPs from preproteins having the N-terminally attached HA–MBP domain. Cells of AD1811 (rseP⁺; odd-numbered lanes) and KK211 (ΔrseP; even-numbered lanes), expressing HA-MBP fusions of the indicated preprotein, were grown at 30 °C (lanes 1–8, 15, and 16) or 37 °C (lanes 9–14 and 17–22) in L broth containing IPTG and cAMP for anti-HA immunoblotting. (B) Cleavage of OmpA SP. (Left) The rseP⁺ (lane 1) and ΔrseP (lane 2) cells, carrying pAS80 encoding HA-MBP-proOmpA, were grown at 37 °C. (Right) Strain AD2328 (ΔrseP)/pAS80 was further transformed with pKK12 (RseP, lane 1), pTH18cr [vector (vec), lane 2], or pAS90 [RseP(H22F), lane 3] and grown at 37 °C. Proteins were visualized by immunoblotting with anti-RseP (Upper) and anti-HA (Lower). (C) Cleavage of HA-Bla-tagged OmpC SP. The rseP⁺ (lane 1) and ΔrseP (lane 2) cells, carrying a pSTD1415 (HA-Bla-preOmpC), were grown and analyzed as in A.

Earlier studies showed that a membrane-bound protease SppA degraded SP from Lpp in detergent extracts (15, 16). However, its in vivo function has not been established. We expressed HA-MBP–tagged derivatives of Lpp, LivK, M13 coat, and TolC in strains deleted for one or both of rseP and sppA (Fig. S5C). Disruption of sppA did not significantly affect the production of CL and CL-like fragments from any of the above constructs. Thus, SppA had no positive role in the RseP-dependent cleavage of SPs from the Lpp and the LivK fusion proteins, nor did it contribute to the RseP-independent cleavage observed with the M13 coat and TolC fusion proteins.

The SPs examined above are targeted to the Sec translocation pathway. We asked whether RseP could cleave Tat pathway SPs carrying the twin-arginine motif in the N-terminal region (17). An HA-MBP-fusion of the TorA precursor, a Tat substrate, was found to produce a UC fragment having the HA epitope in ΔrseP cells and a smaller CL fragment in rseP⁺ cells (Fig. 4A, lanes 21 and 22). Thus, RseP appears to cleave a Tat SP as well.

To examine whether the presence of the MBP domain in the immediate vicinity of SP played any important role in the RseP-dependent SP cleavage that we observed in vivo, we constructed a derivative of the OmpC precursor (preOmpC) having an HA-Bla domain, instead of an HA–MBP domain, at the N terminus. This fusion protein, just like HA-MBP-preOmpC, produced CL in an RseP-dependent manner (Fig. 4C). Thus, MBP played no positive role in the in vivo cleavage of SPs.

Involvement of B. subtilis RasP in SP Cleavage.

We attached HA-MBP to the N termini of the secretory precursors of PenP and Mpr (Fig. S5) of B. subtilis. This Gram-positive bacterium contains RasP, an S2P protease involved in transmembrane stress signal transduction (18). In ΔrasP cells, the PenP fusion protein produced a fragment that reacted with anti-MBP (Fig. 5A, lane 2), most likely representing UC. By contrast, rasP⁺ cells accumulated a smaller sized fragment (CL, lane 1). UC in ΔrasP cells was converted to CL upon expression of proteolytically active RasP (RasP-His₆) (Fig. 5B, lane 1) but not of its active-site mutant, RasP(E21A)-His₆ (lane 2). The Mpr fusion protein behaved similarly (Fig. 5 A and B, lanes 3 and 4). Thus, RasP can cleave SPs from at least certain secretory proteins in B. subtilis.

Fig. 5.

Involvement of RasP in SP cleavage in B. subtilis. (A) B. subtilis strains SCB1450 (thrC::_PxylAha-mbp-penP, lane 1), SCB1628 (thrC::_PxylAha-mbp-penP, ΔrasP, lanes 2), SCB1652 (thrC::_PxylAha-mbp-mpr, lane 3), and SCB1673 (thrC::_PxylAha-mbp-mpr, ΔrasP, lane 4) were grown at 37 °C in L broth containing 0.5% xylose to A₆₀₀ of ∼0.5. Total cellular proteins were analyzed by SDS/PAGE and anti-MBP immunoblotting. (B) Cells of SCB2287 (thrC::_PxylAha-mbp-penP, ΔrasP, amyE::_PxylArasP-his₆, lane 1), SCB2288 (thrC::_PxylAha-mbp-penP, ΔrasP, amyE::_PxylArasP(E21A)-his₆, lanes 2), SCB2292 (thrC::_PxylAha-mbp-mpr, ΔrasP, amyE::_PxylArasP-his₆, lane 3), and SCB2293 (thrC::_PxylAha-mbp-mpr, ΔrasP, amyE::_PxylArasP(E21A)-his₆, lanes 4) were grown as above for anti-MBP (Upper panels) and anti-His tag (Lower panels) immunoblotting.

RseP Can Cleave an Authentic Signal Peptide in the Membrane.

We then studied proteolysis of a native, unmodified SP using in vitro translocation system. A model secretory protein, LivK-Lpp, having the LivK SP region followed by the Lpp mature sequence (Fig. 6A), was synthesized by in vitro translation in the presence of [³⁵S]methionine. It was then denatured with urea and subjected to translocation reaction into inverted membrane vesicles (IMVs) prepared from wild-type (rseP⁺) cells. This reaction yielded two lower-molecular mass species of LivK-Lpp (Fig. 6B, lane 6), one corresponding to the mature (SP-processed) form (M) and the other (about 3 kDa) comigrating with the in vitro-translated LivK SP (Fig. 6B, lanes 3 and 6). The production of these smaller fragments required active translocation of LivK-Lpp because they were not observed in the presence of NaN₃ (lane 7) or in the absence of IMV (lane 5). Indeed, both fragments resisted proteinase K (lane 9) unless the membrane barrier was disrupted with a detergent (lane 10). Thus, the 3-kDa fragment most likely represented the LivK SP.

Fig. 6.

Cleavage of an authentic SP by RseP. (A) Amino acid sequence of the N-terminal region of LivK-Lpp. The Lep cleavage site is indicated by an arrowhead. (B) Detection of LivK SP generated upon in vitro translocation. [³⁵S]methionine-labeled and urea-denatured LivK-Lpp was incubated at 30 °C for the indicated periods in the translocation reaction mixture with or without IMVs prepared from AD2466 (rseP⁺). NaN₃ (Az; 50 mM) or a mixture (Az*) of NaN₃ (50 mM), NaCl (1 M), and zinc acetate (50 μM) was included in the reactions for lanes 7 and 8. A portion of the sample for lane 6 was treated with 200 μg/mL proteinase K in the presence (lane 10) or absence (lane 9) of 1% Triton X-100. Lanes 1–3 received in vitro-synthesized peptides (open arrowheads) of the first 20 (lane 1, SP-3), 26 (lane 2, SP+3), and 23 (lane 3, SP) residues of the LivK-Lpp sequence, which were incubated with IMVs. Proteins were analyzed by 10–20% tricine gel electrophoresis. Arrow indicates a fragment of the SP+3 peptide possibly produced by degradation or premature translation termination. Asterisks indicate possible read-through products. (C) Effects of rseP disruption. Urea-denatured LivK-Lpp was incubated as above with IMVs from AD2466 (rseP⁺, lanes 1–5) or from AD2469 (ΔrseP, lanes 6–10) for 3 min at 30 °C. After addition of azide/NaCl/zinc, samples were incubated further at 30 °C for the indicated periods. (D) Effect of RseP overproduction. Urea-denatured LivK-Lpp was incubated as above with IMVs from CU141 [(WT), lanes 1–3], from CU141/pKK49 (RseP-His₆-Myc) (RseP, lanes 4–6), or from CU141/pKK50 (RseP(H22F)-His₆-Myc) (H22F, lanes 4–6) and analyzed as in C. The letters P, M, and SP indicate the precursor form, mature form, and signal peptide, respectively, of LivK-Lpp.

We examined stability of the in vitro-produced LivK SP by incubating a 6-min reaction product in the presence of NaN₃ and a high concentration of salt, which together blocked further translocation effectively (Fig. 6B, lane 8). The inclusion of high salt was also known to stimulate in vitro proteolytic activity of RseP (11). During this posttranslocation incubation, the intensity of LivK SP in wild-type IMVs decreased gradually (Fig. 6C, lanes 1–5), indicating that it was degraded or cleaved into undetectable fragments. By contrast, LivK SP produced in the ΔrseP IMVs was significantly more stable (Fig. 6C). The degradation was enhanced markedly when IMVs from RseP-overproducing cells were used for translocation (Fig. 6D; compare lanes 4–6 with 1–3) but not with IMVs from RseP(H22F)-overproducing cells (lanes 7–9) (see Fig. S6 for quantification). These results demonstrate a successful reconstitution of RseP-dependent proteolysis of an authentic SP in the membrane, although we have been unable to detect cleavage products due presumably to their small sizes.

Discussion

We have shown that bacteria use an I-CLiP protease, S2P, to introduce a proteolytic cleavage into signal (leader) peptidase-processed SPs. In support of this conclusion, an E. coli mutant lacking RseP and a B. subtilis mutant lacking RasP accumulated uncleaved SPs, whereas wild-type bacteria produced shorter fragments of SPs. This demonstration required the use of engineered preproteins having an N-terminally attached HA-MBP or HA-Bla domain, which allowed the detection of SPs and their cleaved N-terminal fragments. The stable MBP-Bla domains might also have stabilized the short hydrophobic oligopeptides. However, these globular domains did not have any essential role in the RseP-dependent proteolysis of SP per se. This conclusion was supported by the results of our in vitro studies showing that (i) the purified RseP enzyme directly cleaved the synthetic Bla SP with or without the short N-terminal Myc tag and (ii) the unmodified LivK SP received RseP-dependent proteolysis in the membrane when it was produced in the in vitro translocation reaction.

RseP cleaved Bla SP in vitro between Pro-12 and Phe-13 within the hydrophobic core. However, position 12 that had been replaced by cysteine was not included in the N-terminal cleavage product that we detected in vivo. This apparent discrepancy could be explained in different ways. First, the cleavage point may have shifted due either to the cysteine substitution or the attachment of the HA–MBP domain. Second, some secondary proteolysis may have occurred in vivo. Third, enzyme-substrate positioning may have differed between the membrane-integrated states (in vivo) and the detergent-solubilized states (in vitro) of the substrate and the enzyme. Although further studies are required to establish the exact peptide bond that RseP hydrolyzes in vivo, our results collectively indicate that RseP cleaves the Bla SP within the hydrophobic core region. In addition to the Bla SP, SPs from 12 E. coli preproteins proved to be subject to RseP-dependent cleavage. The RseP-susceptible SPs include those of both Sec and Tat substrates, pointing to the general involvement of RseP in SP metabolism in E. coli.

One of B. subtilis S2P proteases, RasP, was found to be involved in proteolysis of SPs from at least two secretory proteins. Clearance of SPs from the membrane might be important for protein secretion in this organism, as the B. subtilis rasP disruptant exhibits a weak defect in protein secretion (19). In accordance with our conclusion, it was reported that several peptide sex pheromones in Enterococcus faecalis are produced from the lipoprotein SP through S2P (Eep)-dependent, endoproteolytic processing (20, 21). We suggest that S2P proteases in bacteria have a role equivalent to that of SPPs in higher eukaryotes. Eukaryotic S2P proteases could potentially act against SPs, but their localization at the Golgi apparatus (22) might preclude them from acting against SPs, which are likely to remain in the endoplasmic reticulum (ER) membrane. This spatial segregation seems to explain why higher eukaryotes possess SPP for degrading SPs in the ER. Both S2P (a zinc metalloprotease) and SPP (an aspartyl protease) prefer membrane proteins with type II orientation as substrates (7), in which helix-destabilizing amino acid residues promote proteolysis by these proteases (2, 11). SPs fulfill these requirements (23).

Confusingly, the name of signal peptide peptidase was first used for the E. coli enzyme SppA (or protease IV) (12, 15), a serine protease having no evolutionary relationship to SPP but with the ability to degrade Lpp SP in detergent extracts and after purification (12, 15, 16). Although SppA is generally considered to be the SP-cleaving enzyme in bacteria, disruption of the E. coli sppA gene did not affect in vivo cleavage of SPs of fusion proteins, including that from Lpp in our experiments. SppA has a large periplasmic domain, which, in archaea, forms a tetrameric assembly of an inverted bowl-like shape, the membrane-distal part of which includes the protease active sites (24). The possibility remains that SppA provides a route of SP degradation if SPs are released to the periplasm, although its in vivo role has not been studied. Our observations that SPs from some proteins received RseP/SppA-independent in vivo cleavage and that LivK SP was slowly degraded even in the RseP-deficient membranes (Fig. 6C, lanes 6–10) suggest that RseP is not the exclusive enzyme that degrades SPs. Identification and characterization of additional proteases involved in SP turnover await future studies.

Our previous (11) and the present results show that the substrate specificity of RseP is broader than originally thought. It is possible that RseP homologs, prevalent in bacteria, contribute to the quality control of the cytoplasmic membrane by degrading SPs as well as some membrane proteins. Consistent with this notion, RseP appears to have a function other than activation of σ^E, as disruption of rseP causes significantly slower cell growth even in the absence of RseA, the proteolytic target that down-regulates σ^E (Fig. S7). In eukaryotes, SPPs might have roles beyond SP degradation (25, 26). However, RseP (S2P) or SPP must not promiscuously degrade membrane proteins. Cleavage of SPs by SPP (2) and RseP (this study) requires prior processing of the substrates by the signal (leader) peptidase. The Lep dependence of the RseP action should prevent premature cleavage of SPs before the commitment step of translocation. It might also prevent the formation of a potentially harmful protein species that bears the C-terminal portions of SPs. This regulation is conceptually analogous to the sequential proteolysis of RseA by DegS followed by RseP (8, 9). The periplasmic domain of RseA contains Gln-rich regions that contribute to the RseP avoidance (27). Also, binding of RseB to the RseA periplasmic domain might prevent RseP from cleaving the intact RseA (28). In these ways, removal of the periplasmic domain transforms RseA into a substrate of RseP. In the case of presecretory proteins, removal of the large mature domains by Lep might facilitate effective access of SPs to the recessed intramembrane active site of RseP (29). This regulation might also require elements in the enzyme RseP itself. A pair of PDZ domains, in tandem in the periplasmic region of RseP, plays crucial roles in suppressing DegS-independent cleavage of intact RseA (30); it could also have a positive role in the cleavage of DegS-processed RseA (31). Our results pose an interesting question about whether the RseP PDZ domains participate in regulation of SP degradation. The S2P-mediated cleavage of SPs and RseA offers excellent systems to gain molecular and cellular insights into the mechanisms of regulated intramembrane proteolysis.

Materials and Methods

Bacterial Strains, Plasmids, and Media.

Bacterial strains are listed in Table S1. See SI Materials and Methods for details of strains, plasmids, and media.

Immunoblotting, Pulse-Chase Labeling, and Immunoprecipitation.

Immunoblotting (32), pulse-chase labeling (8), and immunoprecipitation (8) analyses were carried out essentially as described. See SI Materials and Methods for details.

Cleavage of Synthetic Peptides.

Chemically synthesized Myc-SP^Bla-Flag peptide (19.8 μM) was mixed with purified RseP-His₆-Myc or RseP(H22F)-His₆-Myc (0.64 μM) (11) in buffer containing 50 mM Tris⋅HCl (pH 8.1), 0.02% n-dodecyl-β-d-maltoside, 2.5% glycerol, 5 μM zinc acetate, 10 mM 2-mercaptoethanol, and 1× protease inhibitor mixture (EDTA-free) (Nacalai Tesque, Inc.), incubated at 37 °C as described previously (11) and analyzed by SDS/PAGE and mass spectrometry. See SI Materials and Methods for details.

In Vitro Translocation and Proteolysis of LivK SP.

In vitro transcription was directed by pSTD1411, pSTD1425, pSTD1426, and pSTD1427 and [³⁵S]methionine-labeled LivK-Lpp proteins were synthesized essentially as described previously (33). After translation, proteins were precipitated with 5% trichloroacetic acid and solublized with 6 M urea in 50 mM Hepes-KOH (pH 7.5). IMVs were prepared as described previously (34). Translocation of in vitro-synthesized ³⁵S-labeled proteins into IMVs was carried out by incubation at 30 °C for the indicated time periods as described previously (34). Subsequently, NaN₃ (final concentration of 50 mM), NaCl (final concentration of 1 M), and zinc acetate (final concentration of 50 μM) were added and the incubation continued at 30 °C. Samples were withdrawn at the indicated time points and treated with 5% trichloroacetic acid. Proteins were solubilized in SDS sample buffer and analyzed by 10–20% tricine gel electrophoresis (Invitrogen).

Assessment of the in Vivo Cleavage Site of HA-MBP-Bla from malPEG Modifiability of Engineered Cysteines.

Experiments of malPEG modification were carried out essentially as described previously (11). See SI Materials and Methods for details.