Báo cáo khoa học: Signal peptide hydrophobicity is critical for early stages in protein export by Bacillus subtilis

Tác giả: Geeske Zanen, Edith N. G. Houben, Rob Meima, Harold Tjalsma, Jan D. H. Jongbloed, Helga Westers, Bauke Oudega, Joen Luirink, Jan Maarten van Dijl and Wim J. Quax

Lĩnh vực: Sinh học phân tử, Di truyền học

Nội dung tài liệu: Nghiên cứu này khám phá vai trò của tính ưa nước của peptide tín hiệu trong giai đoạn đầu của quá trình xuất protein ở vi khuẩn Bacillus subtilis. Các nhà nghiên cứu đã biến đổi peptide tín hiệu của enzyme α-amylase AmyQ, tạo ra các phiên bản giàu leucine (tính ưa nước cao) hoặc giàu alanine (tính ưa nước thấp). Kết quả cho thấy peptide tín hiệu có tính ưa nước cao hơn thì hướng dẫn quá trình xuất protein hiệu quả hơn ở cả E. coli và B. subtilis. Tuy nhiên, peptide tín hiệu giàu alanine, có tính ưa nước thấp, chỉ hoạt động trong E. coli chứ không hoạt động hiệu quả ở B. subtilis. Các thí nghiệm lai chéo đã chỉ ra rằng tính ưa nước của peptide tín hiệu là yếu tố quyết định quan trọng đối với việc liên kết peptide tín hiệu với các thành phần của phức hợp nhận diện tín hiệu (SRP) hoặc yếu tố kích hoạt (TF), không chỉ ở E. coli mà còn ở B. subtilis. Điều này cho thấy hệ thống SRP của B. subtilis có khả năng phân biệt các peptide tín hiệu có mức độ ưa nước khác nhau. Đặc biệt, dường như bộ máy xuất protein của B. subtilis kém thích nghi hơn với các peptide tín hiệu giàu alanine có tính ưa nước thấp. Do đó, tính ưa nước của peptide tín hiệu dường như quan trọng hơn đối với hiệu quả của các giai đoạn đầu trong quá trình xuất protein ở B. subtilis so với E. coli.

Mục lục chi tiết:

• Keywords
• Correspondence
• Present addresses
• (Received 29 March 2005, revised 03 May 2005, accepted 18 May 2005)
• doi:10.1111/j.1742-4658.2005.04777.x
• Bacillus subtilis and Escherichia coli are used as prototype models for studies on protein translocation and secretion in Gram-positive and Gram-negative bacteria, respectively. The absence or presence of a hydrophobic 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 proteins to the translocation machinery in the membrane, the so-called Sec machinery in particular.
• Signal peptides are usually sophisticated N-terminal extensions, containing multipurpose functional information. A signal peptide can be divided into three distinct domains; the N-, H-, and C-domains. The N-domain interacts with the translocation machinery and the negatively charged phospholipids in the lipid bilayer of the membrane. The H-domain can adopt an a-helical conformation in the membrane due to a stretch of hydrophobic residues. To allow the formation 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. Analyses of the H-domain show that the hydrophobic core is the dominant structure in determining signal peptide function. 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.
• Although the overall structure of signal peptides is quite similar, small variations can result in export via different targeting pathways. 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. Reduction of the net positive charge or the hydrophobicity of certain signal peptides decreases the effectiveness of SRP recognition. However, in E. coli a high degree of H-domain hydrophobicity can compensate for the loss of basic residues in the N-domain and restore SRP binding. Signal peptides containing an (S/T)RRXFLK motif in E. coli or an RRXOO motif in B. subtilis (Φ is a hydrophobic residue, X can be any residue) are candidates to be translocated via the twin arginine translocation (Tat) pathway. In general, Tat-targeting signal peptides have H-domains which are less hydrophobic than signal peptides that target proteins to the Sec machinery.
• Upon emergence 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). 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. Altogether, this means that different specificity determinants are involved in early stages of protein export from the cytoplasm.
• Most research on the interactions between signal peptides and cytoplasmic chaperones has so far been performed in E. coli. However, as shown by Collier, signal peptides can behave differently in different hosts. Notably, B. subtilis lacks a SecB homologue, the chaperone that is involved in post-translational targeting of the secretory proteins in E. coli. Moreover, signal peptides of Gram-positive organisms are usually longer and more hydrophobic than those of Gram-negative organisms. Until now, it is not known whether this difference in hydrophobicity and length of signal peptides 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 interaction 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. amyloliquefaciens (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 precursor 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 compete 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. amyloliquefaciens by E. coli or B. subtilis, plasmids were constructed encoding AmyQ precursors with signal peptides of distinct hydrophobicity. Specifically, an Ala-rich signal peptide (MIQKRKRTVSLAAAAACAAAALQPITKTSAVN) and a Leu-rich signal peptide (MIQKRKRTVSLLLLLLCLLLLLQPITKTSAVN) 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 peptide) or higher (0.903 for the Leu signal peptide) than that of the authentic AmyQ signal peptide (MIQKRKRTVSFRLVLMCTLLFVSLPITKTSAVN; 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 samples 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 reflected 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 present (Fig. 1C).
• To examine the effects of signal peptide hydrophobicity 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 [35S]methionine for 1 min, excess nonradioactive methionine (chase) was added (t = 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 produced in E. coli. In contrast, processing of AmyQ precursors 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 signal 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 precursor 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 previously shown that signal peptides with increased hydrophobicity improved the export efficiency for PhoA in this organism. 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 peptide (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 synthesized 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 peptide 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 molecules with the authentic signal peptide were processed more efficiently in B. subtilis than in E. coli, and the same was true for AmyQ molecules with the Leu signal 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, protoplasts 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 experiments. In contrast to AmyQ with the authentic or Leu signal peptides, all AmyQ synthesized with the Ala signal peptide was present in the precursor form. As previously shown, all AmyQ molecules synthesized with the authentic signal peptide were accessible to trypsin upon protoplasting of the cells. 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 molecules 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 synthesized with the authentic, Leu or Ala signal peptides were processed in E. coli, subcellular localization experiments in this organism revealed that all corresponding 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 translocated 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 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 compared 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 translation lysate in the presence of [35S]methionine to 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 optimal cytoplasmic exposure of the signal peptides, taking into consideration that approximately 30 amino acids will be located within the ribosome (schematically 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 incubated with crude E. coli MC4100, B. subtilis 168, or B. subtilis ATF cell lysates. The latter strain lacks the TF, which is known to interact with peptides emerging from the ribosome. The ATF strain was used for these experiments, because no antibody against the B. subtilis TF is currently available. As a negative control, the purified RNCs were incubated with incubation buffer only. Interactions between RNCs and cytoplasmic components of E. coli or B. subtilis were fixed by adding the homobifunctional 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