Signal peptide hydrophobicity is critical for early stages in protein export by Bacillus subtilis Geeske Zanen 1 , Edith N. G. Houben 2 , Rob Meima 2, *, Harold Tjalsma 3, †, Jan D. H. Jongbloed 3, ‡, Helga Westers 1,3 , Bauke Oudega 2 , Joen Luirink 2 , Jan Maarten van Dijl 1, § and Wim J. Quax 1 1 Department of Pharmaceutical Biology, University of Groningen, the Netherlands 2 Department of Molecular Microbiology, Vrije Universiteit, Amsterdam, the Netherlands 3 Department of Genetics, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen, the Netherlands Bacillus subtilis and Escherichia coli are used as proto- type models for studies on protein translocation and secretion in Gram-positive and Gram-negative bac- teria, respectively. The absence or presence of a hydro- phobic export signal, called signal peptide, determines whether newly synthesized proteins are retained in the cytoplasm or exported to other cellular compartments. Signal peptides and their recognition by cytoplasmic chaperones play a key role in membrane insertion of membrane proteins and in targeting of secretory Keywords SRP; signal peptide; protein targeting; protein translocation; trigger factor Correspondence W. J. Quax, Department of Pharmaceutical Biology, University of Groningen, Antonius Deusinglaan 1, 9713 AV Groningen, the Netherlands Fax: +31 50 3633000 Tel: +31 50 3632558 E-mail: W.J.Quax@farm.rug.nl Present addresses *DSM Food Specialties, Postbus 1, 2600 MA Delft, the Netherlands; †Department of Clinical Chemistry, University Medical Centre Nijmegen, PO Box 9101, 6500 HB Nijmegen, the Netherlands; ‡Department of Clinical Genetics, University Medical Center of Groningen, P.O box 30001, 9700 RB, Groningen, the Netherlands; §Laboratory of Molecular Bacteriology, Department of Medical Microbiology, University Medical Center of Groningen and University of Groningen, Hanzeplein 1, PO Box 30001, 9700 RB Groningen, the Netherlands (Received 29 March 2005, revised 03 May 2005, accepted 18 May 2005) doi:10.1111/j.1742-4658.2005.04777.x Signal peptides that direct protein export in Bacillus subtilis are overall more hydrophobic than signal peptides in Escherichia coli. To study the importance of signal peptide hydrophobicity for protein export in both organisms, the a-amylase AmyQ was provided with leucine-rich (high hydrophobicity) or alanine-rich (low hydrophobicity) signal peptides. AmyQ export was most efficiently directed by the authentic signal peptide, both in E. coli and B. subtilis. The leucine-rich signal peptide directed AmyQ export less efficiently in both organisms, as judged from pulse-chase labelling experiments. Remarkably, the alanine-rich signal peptide was functional in protein translocation only in E. coli. Cross-linking of in vitro synthesized ribosome nascent chain complexes (RNCs) to cytoplasmic pro- teins showed that signal peptide hydrophobicity is a critical determinant for signal peptide binding to the Ffh component of the signal recognition particle (SRP) or to trigger factor, not only in E. coli, but also in B. subtilis. The results show that B. subtilis SRP can discriminate between signal peptides with relatively high hydrophobicities. Interestingly, the B. subtilis protein export machinery seems to be poorly adapted to handle alanine- rich signal peptides with a low hydrophobicity. Thus, signal peptide hydro- phobicity appears to be more critical for the efficiency of early stages in protein export in B. subtilis than in E. coli. Abbreviations DSS, disuccinimidyl suberate; RNCs, ribosome nascent chain complexes; SRP, signal recognition particle; TF, trigger factor. FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS 4617 proteins to the translocation machinery in the mem- brane, the so-called Sec machinery in particular [1–4]. Signal peptides are usually sophisticated N-terminal extensions, containing multipurpose functional infor- mation. A signal peptide can be divided into three distinct domains; the N-, H-, and C-domains [5,6]. The N-domain interacts with the translocation machinery and the negatively charged phospholipids in the lipid bilayer of the membrane [7,8]. The H-domain can adopt an a-helical conformation in the membrane due to a stretch of hydrophobic residues [9]. To allow the forma- tion of a hairpin-like structure that can insert into the membrane, helix-breaking glycine or proline residues are often present in the middle of the hydrophobic stretch. Unlooping of this hairpin might result in the insertion of the complete signal peptide into the membrane [8]. Analyses of the H-domain show that the hydrophobic core is the dominant structure in determining signal peptide function [10–12]. The C-domain contains the cleavage site for specific signal peptidases that remove signal peptides from the mature part of the exported protein during or shortly after translocation [13,14]. Although the overall structure of signal peptides is quite similar, small variations can result in export via different targeting pathways [15–17]. Signal peptides directing proteins into the signal recognition particle (SRP)-dependent pathway have a significantly more hydrophobic H-domain than those mediating SRP- independent targeting, at least in E. coli [18,19]. Reduction of the net positive charge or the hydro- phobicity of certain signal peptides decreases the effectiveness of SRP recognition. However, in E. coli a high degree of H-domain hydrophobicity can compen- sate for the loss of basic residues in the N-domain and restore SRP binding [20]. Signal peptides containing an (S ⁄ T)RRXFLK motif in E. coli or an RRXFF motif in B. subtilis (F is a hydrophobic residue, X can be any residue) are candidates to be translocated via the twin arginine translocation (Tat) pathway [15,21]. In general, Tat-targeting signal peptides have H-domains which are less hydrophobic than signal peptides that target proteins to the Sec machinery [22]. Upon emer- gence from the ribosome, the signal peptide of a nascent secretory protein can be recognized by several cytoplasmic chaperones and ⁄ or targeting factors, such as Ffh or trigger factor (TF) [23]. In contrast to Ffh, which is required for cotranslational protein export in E. coli, the cytoplasmic chaperone SecB has mainly been implicated in post-translational protein targeting. For E. coli it has been shown that by increasing the hydrophobicity of signal peptides, exported proteins can be re-routed from SecB into the SRP pathway [19,24,25]. Altogether, this means that different specifi- city determinants are involved in early stages of pro- tein export from the cytoplasm. Most research on the interactions between signal pep- tides and cytoplasmic chaperones has so far been per- formed in E. coli. However, as shown by Collier, signal peptides can behave differently in different hosts [26]. Notably, B. subtilis lacks a SecB homologue, the chaper- one that is involved in post-translational targeting of the secretory proteins in E. coli [2]. Moreover, signal pep- tides of Gram-positive organisms are usually longer and more hydrophobic than those of Gram-negative organ- isms [2,27,28]. Until now, it is not known whether this difference in hydrophobicity and length of signal pep- tides represents a functional difference in these species. In the present studies, we have addressed the effects of major variations in signal peptide hydrophobicity on translocation, processing, and signal peptide inter- action with cytoplasmic chaperones using a combined in vivo and in vitro approach in both E. coli and B. subtilis. The results show interesting differences for the translocation of an a-amylase of B. amyloliquefac- iens (AmyQ) with altered signal peptides in these organisms. Whereas E. coli translocates AmyQ with a less hydrophobic alanine-rich signal peptide, even in a secB mutant, B. subtilis accumulates the respective pre- cursor intracellularly. Cross-linking studies show that TF of B. subtilis interacts with the authentic signal peptide of AmyQ, whereas Ffh and TF of E. coli com- pete to interact with this signal peptide. Remarkably, a more hydrophobic leucine-rich signal peptide resulted in reduced AmyQ translocation efficiencies, both in B. subtilis and E. coli . Taken together, these findings suggest that the hydrophobicity of signal peptides is more critical for early stages in protein translocation in B. subtilis than in E. coli. Results Changing the hydrophobicity of the AmyQ signal peptide To study the effects of signal peptide hydrophobicity on the export of the a-amylase AmyQ of B. amylolique- faciens by E. coli or B. subtilis, plasmids were construc- ted encoding AmyQ precursors with signal peptides of distinct hydrophobicity. Specifically, an Ala-rich signal peptide (MIQKRKRTVSLAAAAACAAAALQPITK TSAVN) and a Leu-rich signal peptide (MIQKRKR TVSLLLLLLCLLLLLQPITKTSAVN) were designed. Hereafter, these mutant signal peptides are referred to as Ala or Leu signal peptides. These signal peptides have grand average of hydropathicity (Gravy) values that are significantly lower (0.341 for the Ala signal Secretory protein targeting in Bacillus subtilis G. Zanen et al. 4618 FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS peptide) or higher (0.903 for the Leu signal peptide) than that of the authentic AmyQ signal peptide (MIQK RKRTVSFRLVLMCTLLFVSLPITKTSAVN; Gravy value 0.591). In vivo translocation and processing of a-amylase in E. coli and B. subtilis To study the effects of the different signal peptides on in vivo translocation of AmyQ, E. coli TG90 and B. subtilis 168 were transformed with the E. coli– B. subtilis shuttle vectors pKTHM10, pKTHM101 or pKTHM102. These vectors encode the authentic pre- AmyQ, pre-AmyQ with the Ala signal peptide, and pre-AmyQ with the Leu signal peptide, respectively. Cells were grown overnight and samples were prepared for western blotting experiments and immunodetection with specific antibodies against AmyQ. As shown in Fig. 1A, mature AmyQ was detectable in cellular sam- ples of E. coli, irrespective of the signal peptide used. Pre-AmyQ was only detectable in significant amounts when the Ala signal peptide was used, and it was barely detectable when the Leu signal peptide was used. When expressed in B. subtilis, mature AmyQ was secreted into the growth medium when synthesized with the authentic or Leu signal peptide. In contrast, no mature AmyQ was secreted when this protein was synthesized with the Ala signal peptide (Fig. 1B). To verify whether the AmyQ secreted by B. subtilis was active, an activity assay was performed that is based on the degradation of starch in agar plates. As reflec- ted by the formation of halos upon staining with iodine vapour, active AmyQ was secreted when this protein was provided with the authentic or Leu signal peptide, but not when the Ala signal peptide was pre- sent (Fig. 1C). To examine the effects of signal peptide hydropho- bicity on the kinetics of pre-AmyQ processing, pulse- chase labelling experiments were performed with B. subtilis 168 or E. coli TG90 cells producing AmyQ with the authentic, Ala, or Leu signal peptides. After pulse labelling of newly synthesized proteins with [ 35 S]methionine for 1 min, excess nonradioactive methionine (chase) was added (t ¼ 0). After different periods of chase, samples were taken from which AmyQ was precipitated with specific antibodies. As shown in Fig. 2A, the authentic pre-AmyQ was almost completely processed after 5 min of chase when pro- duced in E. coli. In contrast, processing of AmyQ pre- cursors with the Leu or Ala signal peptides was significantly less efficient. After 5 min chase, 46% or 53% of the AmyQ molecules synthesized with the Leu or Ala signal peptides, respectively, were still in the precursor form (note that pre-AmyQ with the Ala sig- nal peptide has a lower mobility on SDS ⁄ PAGE than pre-AmyQ with the authentic or Leu signal peptides). In contrast, 45% of the authentic pre-AmyQ molecules was processed to the mature form within 1 min of chase. Processing of AmyQ with the Ala signal peptide was so slow, that even after a chase of 30 min precur- sor molecules were still detectable (data not shown). The observation that, in E. coli, AmyQ precursors with the Leu signal peptide were processed less efficiently was unexpected, since Doud and coworkers have previ- ously shown that signal peptides with increased hydro- phobicity improved the export efficiency for PhoA in this organism [29]. Also in B. subtilis, the processing of AmyQ with the Leu signal peptide occurred at a lower rate than that of AmyQ with the authentic signal pep- tide (Fig. 2B). After 2 min of chase 53% of the AmyQ with the Leu signal peptide was processed to the mature form, whereas 71% of the AmyQ with the authentic signal peptide was processed within this time of chase. About 68% of the AmyQ molecules synthes- ized with the Leu signal peptide were mature after 5 min of chase. A completely different result was obtained for AmyQ synthesized with the Ala signal peptide. While AmyQ precursors with this signal pep- tide were processed in E. coli, no processing of these precursors could be observed in B. subtilis (Fig. 2B) and even after a chase of 60 min no mature AmyQ was detected (data not shown). Notably, AmyQ mole- cules with the authentic signal peptide were processed A B C Fig. 1. AmyQ production and secretion. (A) AmyQ production in cells of E. coli as determined by western blotting using the proteins from total cell extracts separated by SDS ⁄ PAGE. (B) AmyQ secre- tion into the growth medium of B. subtilis as determined by west- ern blotting using the proteins from culture supernatants separated by SDS ⁄ PAGE. The images in A and B relate to equal numbers of E. coli or B. subtilis cells, respectively. (C) Plate assay for AmyQ secretion by B. subtilis. The signal peptides fused to AmyQ are indicated. p, Pre-AmyQ; m, mature AmyQ. G. Zanen et al. Secretory protein targeting in Bacillus subtilis FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS 4619 more efficiently in B. subtilis than in E. coli, and the same was true for AmyQ molecules with the Leu sig- nal peptide. The fact that no processing of AmyQ with the Ala signal peptide could be detected in B. subtilis raised the question whether this precursor was translocated across the membrane. To determine the topology of (pre)AmyQ at steady state, protoplasts of B. subtilis cells were incubated with trypsin. In parallel, proto- plasts were incubated without trypsin or with trypsin plus Triton X-100. As shown in Fig. 3, cells producing the authentic AmyQ or AmyQ with the Leu signal peptide contained both precursor and mature forms of AmyQ. Notably, the accumulation of pre-AmyQ in wild-type cells of B. subtilis 168 is commonly observed, despite the fact that this precursor is shown to be processed efficiently in pulse-chase labelling experi- ments [21,30,31]. In contrast to AmyQ with the authentic or Leu signal peptides, all AmyQ synthesized with the Ala signal peptide was present in the precur- sor form. As previously shown, all AmyQ molecules synthesized with the authentic signal peptide were accessible to trypsin upon protoplasting of the cells [31]. In contrast, the situation was slightly different for AmyQ synthesized with the Leu signal peptide: while all mature molecules were accessible to trypsin upon protoplasting, a significant fraction of the pre-AmyQ molecules remained inaccessible to trypsin. The latter pre-AmyQ molecules were only degraded by trypsin in the presence of Triton X-100, indicating that they were protected against trypsin activity by the cytoplasmic membrane. Strikingly, none of the AmyQ molecules synthesized with the Ala signal peptide was accessible to trypsin upon protoplasting. These precursor mole- cules were, however, degraded by trypsin when the protoplasts were lysed with Triton X-100. As controls for these fractionation experiments, the lipoprotein PrsA, which is localized at the membrane–cell wall interface, and the cytoplasmic protein GroEL were used. Figure 3 shows that, irrespective of the cells used, the accessibility of PrsA and GroEL to trypsin was consistent with the subcellular location of these proteins. While all PrsA was accessible to trypsin upon protoplasting, GroEL was only degraded by trypsin when the protoplasts were lysed with Triton X-100. Notably, microscopic inspection of the cells suggested that none of the strains investigated contained AmyQ inclusion bodies in the cytoplasm (data not shown). Consistent with the fact that AmyQ molecules synthes- ized with the authentic, Leu or Ala signal peptides were processed in E. coli, subcellular localization experiments in this organism revealed that all corres- ponding precursor and mature AmyQ molecules were accessible to trypsin upon spheroplasting (data not shown). Taken together, these observations show that AmyQ molecules with the Leu signal peptide are trans- located across the cytoplasmic membranes of B. subtilis and E. coli, but with a slightly lower efficiency than AmyQ molecules with the authentic signal peptide. In contrast, AmyQ molecules with the Ala signal peptide are translocated across the cytoplasmic membrane in E. coli, but not in B. subtilis. Although the processing of the AmyQ precursor containing the Leu signal peptide was slower than that of wild type AmyQ in E. coli and in B. subtilis, processing of the AmyQ precursor containing the Ala signal peptide was only observed in E. coli. Since E. coli contains the cytoplasmic chaperone SecB, which is absent from B. subtilis, the influence of SecB on the processing of AmyQ containing the Ala signal peptide A B Fig. 2. Processing of pre-AmyQ. Processing of AmyQ precursors with different signal peptides in E. coli (A) or B. subtilis (B) was analysed by pulse-chase labelling at 37 °C. Cells were labelled with [ 35 S]methionine for 1 min prior to chase with excess nonradio- active methionine. Samples were withdrawn at the times indicated. The presence of pre- cursor or mature forms of AmyQ in cells plus growth medium was visualized by immunoprecipitation, SDS ⁄ PAGE and fluo- rography. The percentage of processed (mature) AmyQ relative to the total amount of AmyQ (precursor + mature) in each lane is indicated (%). The signal peptides fused to AmyQ are indicated. p, Precursor; m, mature AmyQ. Secretory protein targeting in Bacillus subtilis G. Zanen et al. 4620 FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS was investigated. Pulse-chase labelling experiments with E. coli MC4100 and the corresponding secB mutant strain were performed at 30 °C, because the growth of both strains at 37 °C was severely impaired when transformed with the plasmid for AmyQ-Ala expression (note that this was not the case in E. coli TG90). The results obtained with E. coli MC4100 showed a less efficient processing of AmyQ precursor containing the Ala signal peptide at 30 °C, as com- pared to the processing of this precursor in E. coli TG90 at 37 °C (compare Fig. 2A and Fig. 4A). As shown in Fig. 4A, the processing rate of AmyQ with the Ala signal peptide was mildly reduced in secB mutant cells as compared to cells of E. coli MC4100. Compared to the SecB-dependent OmpA protein, the effect of the absence of SecB on the processing of AmyQ with the Ala signal peptide was less evident (Fig. 4). In vitro cross-linking of a-amylase nascent chains in E. coli and B. subtilis The influence of signal peptide hydrophobicity on its interactions with E. coli and B. subtilis cytoplasmic proteins was investigated by chemical cross-linking of in vitro translated nascent chains. In this approach, truncated mRNAs were translated in an E. coli trans- lation lysate in the presence of [ 35 S]methionine to Fig. 3. Localization of AmyQ in B. subtilis. To analyse the subcellular localization of AmyQ molecules synthesized with different signal pep- tides, cells of B. subtilis were grown overnight at 37 °C in TY medium, diluted 50-fold in fresh TY medium and incubated at 37 °C for 3 h prior to protoplasting. Protoplasts were incubated for 30 min without further additions, in the presence of trypsin (T; 1 mgÆmL )1 ), or trypsin + Triton X-100 (1%). Samples were used for SDS ⁄ PAGE and western blotting. Specific antibodies were used to detect AmyQ, PrsA, or GroEL. The positions of (pre)AmyQ, PrsA, and GroEL (c), and degradation products of PrsA (d*) are indicated. The signal peptides fused to AmyQ are indicated. A cartoon of the protoplasting and protease protection experiment is shown to illustrate the effects of trypsin (T) and Triton X-100. A B Fig. 4. Processing of AmyQ with the Ala signal peptide in E. coli secB. Processing of pre-AmyQ containing the Ala signal peptide (A) and pro-OmpA (B) in E. coli MC4100 secB or the parental strain (wt) were ana- lysed by pulse-chase labeling at 30 °C and subsequent immunoprecipitation, SDS ⁄ PAGE, and fluorography. Cells were labelled with [ 35 S]methionine for 1 min prior to chase with excess nonradioactive methio- nine. Samples were withdrawn at the times indicated. The percentage of processed (mature) AmyQ relative to the total amount of AmyQ (precursor + mature) in each lane is indicated (%). p, Precursor; m, mature. G. Zanen et al. Secretory protein targeting in Bacillus subtilis FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS 4621 generate radioactively labelled ribosome-nascent chain complexes (RNCs). A C-terminal 4· methionine tag was introduced into the nascent chains to increase the labelling efficiency. The nascent chain corresponding to the authentic preprotein comprised 101 amino acids, while the nascent chain corresponding to the Leu and Ala preproteins comprised 105 amino acids. Thus, the lengths of these nascent chains allows opti- mal cytoplasmic exposure of the signal peptides, tak- ing into consideration that approximately 30 amino acids will be located within the ribosome (schemati- cally represented in Fig. 5A). The RNCs were purified over a high-salt sucrose cushion to remove all loosely associated E. coli components originating from the translation lyate. Subsequently, they were either incu- bated with crude E. coli MC4100, B. subtilis 168, or B. subtilis DTF cell lysates. The latter strain lacks the TF, which is known to interact with peptides emer- ging from the ribosome [23]. The DTF strain was used for these experiments, because no anti- body against the B. subtilis TF is currently available. As a negative control, the purified RNCs were incu- bated with incubation buffer only. Interactions between RNCs and cytoplasmic components of E. coli or B. subtilis were fixed by adding the homobifunc- tional lysine-lysine cross-linking reagent disuccinimidyl suberate (DSS). Incubation of AmyQ nascent chains containing the authentic signal peptide with E. coli lysate in the presence of DSS generated cross-linking adducts of  25 kDa,  60 kDa,  68 kDa, and  80 kDa (Fig. 5B, lane 3). The  25 kDa adduct could be immu- noprecipitated using antiserum raised against the E. coli ribosomal protein L23 (Fig. 5B, lane 6). In fact, cross- linking to L23 was not only shown for RNCs with the authentic signal peptide, but also for RNCs with the Leu and Ala signal peptides (Fig. 5B, lanes 2–5, lanes 10–13, lanes 18–21). As shown by immunoprecipitation A B Fig. 5. Cross-linking of AmyQ nascent chai- ns to soluble E. coli and B. subtilis compo- nents. The 101AmyQ wt, 105AmyQ Leu and 105AmyQ Ala RNCs were synthesized in an E. coli MC4100 translation lysate. (A) Schematic representation of the translation reactions. The different signal peptides used and lysine residues (K) that may participate in cross-linking reactions are indicated. (B) After translation, the RNCs were purified over a high-salt sucrose cushion, incubated with crude E. coli MC4100, B. subtilis 168, B. subtilis DTF cell lysates or incubation buffer and treated with DSS. Cross-linking was quenched by adding TCA ⁄ acetone. Immunoprecipitations were subsequently carried out as indicated in Experimental procedures. IP, Immuno-precipitation; E, crude cell lysate of E. coli MC4100; B, crude cell lysate of B. subtilis 168; BD, crude cell lysate of B. subtilis DTF; NC, nascent chain; ?, unknown cross-linking adducts; *, cross-linking adducts with E. coli L23; d, cross-linking adducts with E. coli Ffh; s, cross-linking adducts with B. subtilis Ffh; n , cross-linking adducts with E. coli TF; h, cross-linking adducts with B. subtilis TF. Secretory protein targeting in Bacillus subtilis G. Zanen et al. 4622 FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS with Ffh- and TF-specific antibodies, the  60 kDa cross-linking adduct contained E. coli Ffh, while both the  68 kDa and  80 kDa adducts contained the E. coli TF (Fig. 5B, lanes 7, 8). Adducts of TF fre- quently appear as a doublet [18,32], but it is not known why the ratio between the immunoprecipitated AmyQ- TF adducts differs from the ratio in the nonprecipitated sample (Fig. 5B, lanes 3 and 8). Unfortunately, E. coli SRP could not be removed completely from the ribo- somes by high-salt treatment (Fig. 5B, lane 2 closed cir- cle). Consequently, cross-links to E. coli Ffh were also detected upon incubation of RNCs containing the authentic signal peptide with B. subtilis 168 lysate in the presence of DSS. Cross-linking of these RNCs to B. sub- tilis Ffh could not be demonstrated by immunoprecipi- tations using antibodies specific for B. subtilis Ffh (data not shown). On the other hand, two dominant cross- linking adducts of  69 kDa were detected upon incuba- tion of authentic pre-AmyQ RNCs with B. subtilis 168 lysate (Fig. 5B, lane 4 open squares). Such adducts were not observed after incubating these nascent chains with B. subtilis DTF lysate in the presence of DSS (Fig. 5B, lane 5), which implies that the  69 kDa adducts represent cross-links to TF of B. subtilis. Interestingly, incubation with the B. subtilis DTF lysate resulted in  40-kDa cross-linking adducts that were not observed upon incubation with the B. subtilis 168 lysate (Fig. 5B, lane 5 question mark). Unfortunately, the B. subtilis protein(s) in these  40-kDa adducts could not be identified. Nascent chains of AmyQ with the Leu signal peptide generated  60-kDa and  70-kDa cross-linking adducts upon incubation with E. coli lysate in the pres- ence of DSS (Fig. 5B, lane 10 closed circles), which both represented cross-linking to E. coli Ffh (data not shown). Cross-linking of these RNCs to E. coli Ffh was even detected upon incubation with B. subtilis 168 ly- sate in the presence of DSS (Fig. 5B, lane 14). This cross-linked E. coli Ffh was derived from the transla- tion lysate, despite the high-salt purification. Never- theless, as shown by immunoprecipitation with specific antibodies, a  60-kDa cross-linking adduct containing B. subtilis Ffh was formed upon incubation with B. sub- tilis 168 lysate in the presence of DSS (Fig. 5B, lane 15). Note that the antibodies against Ffh of B. subtilis do not cross-react with Ffh of E. coli (Fig. 5B, lane 16). The  48-kDa cross-linking adduct obtained upon incu- bation of RNCs containing the Leu signal peptide with the B. subtilis 168 lysate was not identified (Fig. 5B, lane 12 question mark). Interestingly, no evidence for specific cross-links between RNCs with the Leu signal peptide and the B. subtilis or E. coli TFs was obtained (Fig. 5B, lanes 11–13 and data not shown). Finally, nascent chains of AmyQ containing the Ala signal peptide generated strong cross-links to E. coli and B. subtilis TF (Fig. 5B, lane 19–21 open and closed squares), while neither cross-links to E. coli Ffh nor B. subtilis Ffh were observed (data not shown). Incubation of these nascent chains with the B. subtilis DTF lysate again generated unidentified  40-kDa cross-linking adducts (Fig. 5B, lane 21 question mark). In conclusion, these findings show that RNCs contain- ing the authentic signal peptide can be cross-linked with L23, Ffh, and TF of E. coli and with TF of B. subtilis. RNCs containing the highly hydrophobic Leu signal peptide can be cross-linked with L23 of E. coli, Ffh of E. coli and B. subtilis, but not detecta- bly with TF of these organisms. In contrast, RNCs containing the mildly hydrophobic Ala signal peptide are efficiently cross-linked with L23 of E. coli,TFof E. coli and B. subtilis, but not detectably with Ffh of these organisms. Discussion Several studies indicate that signal peptide hydrophob- icity is an important determinant for SRP-mediated protein targeting to the E. coli inner membrane [18,33]. Cross-linking of nascent PhoA-derivatives revealed an almost linear correlation between hydrophobicity and SRP cross-linking [18]. In addition, hydrophobic alter- ations in the signal peptides of SecB-dependent pro- teins, re-routed these proteins into the SRP pathway [19,24,25]. Precursor proteins from Gram-positive bac- teria contain signal peptides that are usually longer and more hydrophobic than the signal peptides of pre- cursor proteins from Gram-negative bacteria [2,27,28]. It was therefore hypothesized that the higher hydro- phobicity of signal peptides in Gram-positive bacteria, lacking SecB, has evolved as an adaptation to the SRP-dependent translocation pathway [2]. Changes in hydrophobicity of the signal peptide of a-amylase AmyQ seem to have different effects on the translocation of pre-AmyQ in E. coli or B. subtilis. For E. coli cells, changing the alanine or leucine content and, consequently, the hydrophobicity of the signal peptide did not lead to major translocation defects. However, processing was less efficient for AmyQ precursors containing the Leu or Ala signal peptides when compared to AmyQ with the authentic signal peptide. Importantly, significant amounts of mature AmyQ were released into the periplasm irres- pective of the signal peptide used. In B. subtilis, mature AmyQ directed to and across the membrane with help of the authentic or Leu signal peptides, was efficiently secreted resulting in active AmyQ in the G. Zanen et al. Secretory protein targeting in Bacillus subtilis FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS 4623 growth medium. In contrast, AmyQ containing the Ala signal peptide was not translocated at all. This implies that B. subtilis is not able to translocate pre- cursor proteins with Ala-rich signal peptides of low overall hydrophobicity. This could be due to specific not previously reported effects of Ala residues in a B. subtilis signal peptide. For example, the relatively small size of the Ala side chain might be of relevance with respect to the recognition of the Ala signal peptide by the secretion machinery of B. subtilis. Nevertheless, certain B. subtilis signal peptides of which the in vivo activity has been demonstrated con- tain a relatively large amount of alanine. For example the YxkA signal peptide contains 12 Ala residues [2,4], but has a grand average of hydrophaticity of 0915. This suggests that the low hydrophobicity rather than the high Ala content is responsible for the observed malfunction of the Ala signal peptide. Clearly, this malfunction cannot be explained by the absence of a SecB homologue in B. subtilis, because SecB contributes only to a minor extent to the export of AmyQ with the Ala signal peptide in E. coli. The processing of AmyQ with the authentic or Leu signal peptides was faster in B. subtilis than in E. coli. This is likely due to the overall characteristics of the AmyQ signal peptide. Precursors have normally shor- ter signal peptides in E. coli than in B. subtilis [2,4], and thus the signal peptides used in this study are probably suboptimal for E. coli. Nevertheless, when produced in E. coli, most AmyQ molecules with the authentic signal peptide are processed within 5 min of chase. Finally, in both species the processing rate for AmyQ precursors containing the Leu signal peptide was lower compared to those containing the authentic signal peptide. A possible explanation for this observa- tion could be that the Leu-rich H-domain of high hydrophobicity, perhaps in combination with the four positively charged residues already present in the N-domain, results in a tighter binding of a signalpep- tide to SRP. This might slow down the release of the precursor protein from SRP, which would result in slower translocation and processing by signal pepti- dase. Another possibility could be that the Sec trans- locon has a lower affinity for more hydrophobic AmyQ-derived signal peptides. However, it has been shown that an increased leucine content of a signal peptide increases the cross-linking to Sec of E. coli [34]. Taken together, our observations indicate that, in particular, a low signal peptide hydrophobicity com- pletely impairs precursor translocation in B. subtilis, but not in E. coli. Together with the DnaK system, the ribosome-asso- ciated chaperone TF promotes the folding of newly synthesized proteins in the cytosol of E. coli [35,36]. E. coli TF interacts with virtually all nascent polypep- tides, whereas Ffh interacts specifically with hydropho- bic signal peptides [23]. The present studies show for the first time that the hydrophobicity of a signal pep- tide has a critical impact on its binding to TF or Ffh in B. subtilis. In addition, our studies provide first support for binding of the B. subtilis TF to nascent chains. While the authentic AmyQ signal peptide binds both to TF and (E. coli) Ffh, the less hydrophobic Ala signal peptide only binds to TF, and the more hydrophobic Leu signal peptide binds mainly to Ffh. This obser- vation suggests that TF of B. subtilis plays also an important role in the early stages of signal peptide recognition. In E. coli, ribosomal protein L23 is located near the exit site of the ribosomal tunnel that runs from the peptidyl transferase centre to the surface of the large ribosomal subunit [37]. Interestingly, all AmyQ nascent chains tested were found to bind L23 present in the E. coli lysates used for in vitro translation. Remark- ably, nascent chains containing the authentic or Leu signal peptides were cross-linked to E. coli Ffh, even in the presence of a B. subtilis lysate. This implies that E. coli Ffh could not be removed completely from the RNCs by high salt treatment and that B. subtlis Ffh was unable to compete efficiently with E. coli Ffh. This is probably the reason why binding of B. subtilis Ffh to RNCs with the authentic AmyQ signal peptide could not be visualized in our cross-linking experiments even though it seems most likely that this binding does occur in vivo. As previously pointed out by Walter and Blobel [38], such inefficient binding can be exacerbated by the fact that the H-domains of signal peptides lack lysine residues, which are required for cross-linking with the lysine-specific reagent DSS. Nevertheless, binding of B. subtilis Ffh to RNCs with the Leu signal peptide could be demonstrated. This indicates that Ffh of B. subtilis has a higher affinity for hydrophobic signal peptides, such as the Leu signal peptide, and that Ffh of B. subtilis can effectively compete with Ffh of E. coli for the binding of this signal peptide. Alternatively, RNCs with exposed Leu signal peptides may not be saturated with E. coli Ffh, which would allow for more efficient binding of B. subtilis Ffh. This latter possibil- ity would imply that B. subtilis Ffh does not bind effi- ciently to the RNC with the authentic AmyQ signal peptide, which seems rather unlikely. At present it is not clear why AmyQ with the Ala signal peptide is translocated in E. coli, but not in B. subtilis. Import- antly, this precursor is still translocated in a secB mutant of E. coli, indicating that factors other than SecB are required for this process in E. coli. This suggests that the absence of a SecB homologue in Secretory protein targeting in Bacillus subtilis G. Zanen et al. 4624 FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS B. subtilis is not responsible for the lack of export of AmyQ containing the Ala signal peptide. In conclusion, our present observations imply that signal peptide hydrophobicity is critical for early stage signal peptide recognition by SRP and TF, not only in E. coli, but also in B. subtilis. This view is supported by the fact that the sequences of TF and L23 that are required for ribosome docking of TF [39] are con- served in the corresponding proteins of B. subtilis. Even though the signal peptide of AmyQ is already longer and more hydrophobic than the average E. coli signal peptide, its binding to Ffh of B. subtilis can be enhanced by further increasing its hydrophobicity. Thus, the B. subtilis SRP system is able to disciminate between signal peptides with relatively high hydropho- bicities. Conversely, the B. subtilis machinery for pro- tein export appears poorly adapted to handle signal peptides with a low hydrophobicity. These findings are likely to be of biological relevance since the average hydrophobicity of B. subtilis signal peptides is signifi- cantly higher than that of E. coli signal peptides. Experimental procedures Plasmids, bacterial strains and media The plasmids and bacterial strains used are listed in Table 1. TY medium contained Bacto tryptone (1%), Bacto yeast extract (0.5%), and NaCl (1%). S7-MAM medium was essentially prepared as S7 medium [40] with the differ- ence that the MAM amino acid mixture from Becton Dick- inson (Franklin Lakes, NJ, USA) was used instead of the amino acid mixture normally used to supplement S7 medium [41]. If required, media for E. coli were supple- mented with ampicillin (100 lgÆmL )1 ); or chloramphenicol (10 lgÆmL )1 ), and media for B. subtilis with chlorampheni- col (5 lgÆmL )1 ); or kanamycin (10 or 20 lgÆmL )1 ). DNA techniques Procedures for PCR, DNA purification, restriction, liga- tion, agarose gel electrophoresis, and transformation of E. coli were carried out as described by Sambrook et al. [42]. Competent B. subtilis cells were transformed as Table 1. Plasmids and bacterial strains. Relevant properties Reference Plasmids pMTL23 E. coli cloning vector [54] pMTL23Q3 pMTL23 carrying the 712 bp. EcoRV-SphI fragment of pKTH10, encompassing the 5¢-terminus of the amyQ gene This paper pQ1 pMTL23Q3 carrying silent mutations in the 5¢-terminus of the amyQ gene, creating HindIII, SpeI and KpnI sites at the nucleotides that specify the signal peptidase cleavage site This paper pQ10 pQ1 containing additional NdeIandHindIII restriction sites This paper pKTH10 B. subtilis vector; encodes the a-amylase AmyQ of B. amyloliquefaciens [55] pKTHM10 E. coli–B. subtilis shuttle vector. The EcoRV–SphI fragment of pKTH10 is replaced by the EcoRV–PvuII fragment of pQ10 This paper pCR2.1-TOPO E. coli cloning vector Invitrogen pQ101 As pCR2.1-TOPO with the Leu-rich signal peptide of AmyQ This paper pQ102 As pCR2.1-TOPO with the Ala-rich signal peptide of AmyQ This paper pKTHM101 As pKTHM10, but with the EcoRV–SphI fragment of pQ101 This paper pKTHM102 As pKTHM10, but with the EcoRV–SphI fragment of pQ102 This paper pC4Meth94Bla E. coli cloning vector used for in vitro transcription-translation [18] pC4Meth95AmyQ wt pC4Meth94Bla containing the first 95 codons of the wild type amyQ gene This paper pC4Meth95AmyQ Ala pC4Meth94Bla containing the first 95 codons of the amyQ gene from pKTHM101 This paper pC4Meth95AmyQ Leu pC4Meth94Bla containing the first 95 codons of the amyQ gene from pKTHM102 This paper Strains B. subtilis 168 trpC2 [56] 168 X like 168; amyE::X; Cm r [57] DTF Originally denoted SG1; trpC2; pheA1 tig::kan [58] E. coli TG90 pcnB80; zad::Tn10; Tc r [59] MC4100 F – ; araD139; D(argF-lac); U169; rspL150; relA1; flbB5301; fruA25; deoC1; ptsF25 [60] secB mutant Unpublished work P. Genevaux, laboratory collection G. Zanen et al. Secretory protein targeting in Bacillus subtilis FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS 4625 previously described [30]. Restriction enzymes and the Expand long template PCR system were obtained from Roche Diagnostics GmbH (Mannheim, Germany). T4 DNA ligase was obtained from Epicenter Technologies (Omaha, NE, USA). Construction of AmyQ derivatives containing altered signal peptides To study the effect of the amino acid composition of the H-domain of the signal peptide on protein export, a series of plasmids was constructed encoding AmyQ derivatives with modified H-domains. First, an SphI–EcoRV fragment of pKTH10 containing the 5¢-terminus of the amyQ gene was subcloned in the E. coli cloning vector pMTL23 result- ing in pMTL23Q3. Using pMTL23Q3 as a template, five silent mutations were introduced into the signal peptide- encoding region of amyQ by two subsequent rounds of PCR mutagenesis resulting in plasmids pQ1 and pQ10, respectively. The primers used are listed in Table 2. Thus, restriction sites were introduced for replacement of the sig- nal peptide or parts thereof (e.g. the H-domain) with exist- ing or designed amino acid sequences. To construct an E. coli–B. subtilis shuttle vector, EcoRV-digested pQ10 was fused to plasmid pKTH10, which was cut with EcoRV– PvuII. This resulted in plasmid pKTHM10, which carries a full-length amyQ gene that encodes the authentic AmyQ precursor. Next, two modified H-domains were introduced to replace the original H-domain of the AmyQ signal peptide. Firstly, complementary oligonucleotides (see Table 2) were annealed and cloned into plasmid pCR2.1-TOPO (Invitro- gen Life Technologies, Paisley, UK), using the overhanging HindIII and SpeI compatible sticky ends. Secondly, the fragments were transferred to plasmid pQ10 using the same restriction sites resulting in plasmids pQ101 and pQ102. Finally, the EcoRV–SphI fragments of pQ101 and pQ102 were ligated to EcoRV–SphI digested pKTHM10. This resulted in plasmids pKTHM101 (encoding pre-AmyQ with a Leu-rich signal sequence) and pKTHM102 (encoding pre- AmyQ with an Ala-rich signal sequence). Though not used in the present studies, the Cys residue in the centre of the H-region of the authentic AmyQ signal peptide was main- tained in the H-regions of the Ala- and Leu-rich signal pep- tides to facilitate future cross-linking experiments. The ‘grand average of hydrophathicity’ (Gravy) value for the signal peptide was calculated with the protparam tool (http://www.expasy.org/tools/protparam.html) as the sum of hydrophobicity values of all the amino acids, divided by the number of residues in the sequence [43,44]. SDS/PAGE, western blotting and immunodetection To visualize proteins of E. coli or B. subtilis by western blotting, cells were separated from the growth medium by centrifugation (3 min, 12 900 g,20°C). Cellular samples of E. coli and B. subtilis, and growth medium samples of B. subtilis were prepared for SDS ⁄ PAGE as described pre- viously [40,45]. After separation by SDS ⁄ PAGE, proteins were transferred to a ProtranÒ nitrocellulose transfer mem- brane (Schleicher and Schuell, Dassel, Germany). Western blotting was performed as described by Kyhse-Andersen [46]. AmyQ, GroEL, and PrsA were visualized with specific antibodies and horseradish peroxidase-conjugated goat anti-rabbit IgG or alkaline phosphatase-conjugated goat Table 2. Overview of primers used in the present study. Restriction sites are indicated in bold. Primer Sequence 5¢fi3¢ Restriction site ⁄ remark amyQ3 CGGCGTATACCATTCAAAATACTGCATCAG GGTACCATTTA CGGC ACTAGTTTTTGTAATCGGCAAGCTTACAAATAACAG Mutagenesis primer for introduction of KpnI, SpeI, and HindIII sites into the AmyQ signal peptide coding region amyQ1S CCATGATTACGCCAAGCTCG Reverse primer for first and second round mutagenesis of AmyQ signal peptide coding region amyQ4 ACTAGTTTTTGTAATCGGC AAGCTTACAAATAACAGCGTG CACATAAGCACAAGTCTGAAGCTTACTGTCCGCTTTCG TTTTTGAAT CATATGTC Second round mutagenesis primer (introduction of NdeIandHindIII sites upstream of H-region) h-ala-fwd AGCTTGGCGGCCGCGGCTGCGTGCGCCGCGGCT GCGCTGCAGCCGATTACAAAAA Oligo encompassing AmyQ H-region, complementary with h-ala-rev h-ala-rev CTAGTTTTTGTAATCGGCTGCAGCGCAGCCG CGGCGCACGCAGCCGCGGCCGCCA Oligo encompassing AmyQ H-region, complementary with h-ala-fwd h-leu-fwd AGCTTGCTGCTTCTCCTTTTATGCCTGCTGTTACTCCTGC AGCCGATTACAAAAA Oligo encompassing AmyQ H-region, complementary with h-leu-rev h-leu-rev CTAGTTTTTGTAATCGGCTGCAGGAGTAACAGCAGGCAT AAAAGGAGAAGCAGCA Oligo encompassing AmyQ H-region, complementary with h-leu-fwd amyQ_ATG CGCGAATTCTAATATGATTCAAAAACGAAAGCGGA Amplification primer for construction of truncated AmyQ variants for synthesis of nascent chains amyQ95 GCCGGATCCTTCTCCTAAATCATACAA Amplification primer for construction of truncated AmyQ variants for synthesis of nascent chains Secretory protein targeting in Bacillus subtilis G. Zanen et al. 4626 FEBS Journal 272 (2005) 4617–4630 ª 2005 FEBS [...]... 4617–4630 ª 2005 FEBS Secretory protein targeting in Bacillus subtilis without trypsin, or in the presence of trypsin plus 1% Triton X-100 The subcellular localization of proteins in B subtilis was determined by protoplasting and subsequent trypsin accessibility assays Protoplasts were prepared from B subtilis cells in the late exponential growth phase, essentially as described by Tjalsma et al [31] Briefly,... DG, Dluhy RA & Gierasch LM (1986) Conformations of signal peptides induced by lipids suggest initial steps in protein export Science 233, 206–208 10 Gennity JM & Inouye M (1991) Protein secretion in bacteria Curr Opin Biotechnol 2, 661–667 11 Hikita C & Mizushima S (1992a) Effects of total hydrophobicity and length of the hydrophobic domain of a signal peptide on in vitro translocation efficiency J Biol... used for SDS ⁄ PAGE and western blotting In parallel, protoplasts were incubated without trypsin, or in the presence of trypsin plus 1% Triton X-100 Cross-linking of in vitro synthesized AmyQ derivatives to cellular components of E coli or B subtilis To investigate interactions between different AmyQ nascent chains and soluble E coli or B subtilis components, crude lysates of E coli MC4100, B subtilis. .. with amyE mutant B subtilis 168 X strains Subcellular localization of proteins The subcellular localization of proteins in E coli was determined by spheroplasting and subsequent trypsin accessibility assays Spheroplasts were prepared from exponentially growing cells of E coli Cells were resuspended in spheroplast buffer (40% sucrose; 33 mm Tris pH 8.0, 1 mm EDTA), and incubated for 15 min with lysozyme... authors thank V.P Kontinen for providing antiPrsA, W Wickner for providing anti-Trigger Factor of E coli, R Brimacombe for providing anti-L23 of E coli, M Marahiel for providing B subtilis SG1, and P Genevaux for providing E coli secB Funding for the project, of which this work is a part, was provided by grant VBI.4837 from the ‘Stichting Technische Wetenschappen’ and the CEU projects QLK3-CT1999-00413,... LamB-LacZ by the signal recognition particle pathway of Escherichia coli J Bacteriol 185, 5697–5705 26 Collier DN (1994) Expression of Escherichia coli SecB in Bacillus subtilis facilitates secretion of the SecBdependent maltose-binding protein of E coli J Bacteriol 176, 4937–4940 27 van Dijl JM, Bolhuis A, Tjalsma H, Jongbloed JDH, de Jong A & Bron S (2001) Protein transport pathways in Bacillus subtilis: ... genome-based road map In Bacillus Subtilis and its Closest Relatives: from Genes to Cells (Sonenshein AL, Hoch A & Losick R, eds) pp 337–355 ASM Press, USA ´ 28 von Heijne G & Abrahmsen L (1989) Species-specific variation in signal peptide design Implications for protein secretion in foreign hosts FEBS Lett 27, 439–446 29 Doud SK, Chou MM & Kendall DA (1993) Titration of protein transport activity by incremental... (2001) The targeting pathway of Escherichia coli presecretory and integral membrane proteins is specified by the hydrophobicity of the targeting signal Proc Natl Acad Sci USA 98, 3471–3476 20 Peterson JH, Woolhead CA & Bernstein HD (2003) Basic amino acids in a distinct subset of signal peptides promote interaction with the signal recognition particle J Biol Chem 278, 46155–46162 FEBS Journal 272 (2005)... Secretory protein targeting in Bacillus subtilis 34 35 36 37 38 39 40 41 42 43 44 45 46 for inner membrane protein assembly in Escherichia coli Proc Natl Acad Sci USA 95, 14646–146451 Valent QA, Scotti PA, High S, De Gier JW, von Heijne G, Lentzen G, Wintermeyer W, Oudega B & Luirink J (1998) The Escherichia coli SRP and SecB targeting pathways converge at the translocon EMBO J 17, 2504–2512 Deuerling E,... coli signal peptidase I in Bacillus subtilis J Gen Microbiol 137, 2073–2083 Westers H, Braun PG, Westers L, Antelmann H, Hecker M, Jongbloed JDH, Yoshikawa H, Tanaka T, van Dijl JM & Quax WJ (2005) Genes Involved in SkfA Killing Factor Production Protect a Bacillus subtilis Lipase against Proteolysis Accepted for publication in Appl Environ Microbiol 71, 1899–1908 Sambrook J, Fritsch EF & Maniatis T . the hydrophobicity of signal peptides is more critical for early stages in protein translocation in B. subtilis than in E. coli. Results Changing the hydrophobicity. Signal peptide hydrophobicity is critical for early stages in protein export by Bacillus subtilis Geeske Zanen 1 , Edith N.