Signal Recognition Particle and SecA Cooperate during Export of Secretory Proteins with Highly Hydrophobic Signal Sequences

The Sec translocon of bacterial plasma membranes mediates the linear translocation of secretory proteins as well as the lateral integration of membrane proteins. Integration of many membrane proteins occurs co-translationally via the signal recognition particle (SRP)-dependent targeting of ribosome-associated nascent chains to the Sec translocon. In contrast, translocation of classical secretory proteins across the Sec translocon is a post-translational event requiring no SRP but the motor protein SecA. Secretory proteins were, however, reported to utilize SRP in addition to SecA, if the hydrophobicity of their signal sequences exceeds a certain threshold value. Here we have analyzed transport of this subgroup of secretory proteins across the Sec translocon employing an entirely defined in vitro system. We thus found SecA to be both necessary and sufficient for translocation of secretory proteins with hydrophobic signal sequences, whereas SRP and its receptor improved translocation efficiency. This SRP-mediated boost of translocation is likely due to the early capture of the hydrophobic signal sequence by SRP as revealed by site-specific photo cross-linking of ribosome nascent chain complexes.


Introduction
The heterotrimeric SecYEG translocon located in the inner membrane of Gram-negative bacteria is used by two different groups of proteins to be exported from the cytoplasm of these organisms: secretory proteins destined for the periplasmic space or the outer membrane are translocated through the pore of the Sec translocon, whereas inner membrane proteins exit laterally from the Sec translocon into the lipid bilayer. Both classes of proteins carry characteristic targeting signals. In the case of secretory proteins these are classical N-terminal signal sequences consisting of a positively charged N-region, a hydrophobic core, and a polar C-region. These signal sequences are typically cleaved off after transport. Membrane proteins instead are recognized via noncleaved hydrophobic transmembrane helices called signal anchor sequences. For a recent comprehensive review on the structure and function of the Sec translocon, see ref. [1].
The different targeting signals engage two diverse targeting routes. Signal anchor sequences of membrane proteins, which in general are more hydrophobic than classical signal sequences, recruit the signal recognition particle (SRP) consisting of the Ffh protein and the 4.5S RNA in E. coli. Binding of SRP occurs cotranslationally at the ribosome. The resulting complexes between SRP and ribosome-nascent chains (RNCs) are thought to dock at the membrane-bound SRP receptor (SR), called FtsY in bacteria, via a direct interaction between the Ffh and FtsY proteins [2], from where RNCs are then handed over to the SecYEG translocon [3]. Alternatively, an SRP-FtsY complex might initially form at the membrane in the absence of substrate and only subsequently bind nascent membrane proteins for delivery to the Sec translocon [4]. Still other models of membrane targeting propose a dominant role of the encoding mRNAs and ribosomes prebound to the membrane [5].
On the contrary, the comparably less hydrophobic signal sequences of secretory proteins are not stably bound by SRP at the ribosome and hence elicit a post-translational targeting mode. This usually involves protection by chaperones such as SecB and Trigger factor until the signal sequence binds to the motor protein SecA that has a high affinity for the Sec translocon.
SecA was, however, found to be required also for the assembly of SRP-dependent inner membrane proteins if these harbor extended hydrophilic loops that need to cross the membrane during assembly [6,7,8,9,10,11,12,13]. In this case, SecA was shown to interact with SecYEG-targeted nascent chains while they are still ribosome-associated and to push a hydrophilic loop across the translocon [12]. A simultaneous dependence on both, SRP and SecA also seems to apply to otherwise secretory proteins equipped with exceedingly hydrophobic signal sequences because export of these naturally SecA-dependent proteins was found to be impaired upon depletion of SRP [14,15,16].
In order to be able to discriminate between SRP-and SecAdependent functions during targeting and translocation of secretory proteins with highly hydrophobic signal sequences we set up an in vitro system from entirely defined components. This system was based on the PURE (Protein synthesis Using Recombinant Elements) translation system [17,18,19,20] supplemented with SecYEG-containing proteoliposomes and purified targeting factors. Our studies reveal that, whilst SecA is sufficient for translocation of secretory proteins with hydrophobic signal sequences, SRP and its receptor FtsY improve translocation efficiency by enabling co-translational membrane targeting.

Studying Targeting and Translocation of Secretory Proteins with Hydrophobic Signal Sequences under Defined Conditions
As examples of secretory proteins harboring pronouncedly hydrophobic signal sequences we chose to study the E. coli proteins SfmC and TorT [14] ( Table 1). The genes of both proteins were subcloned under the T7 RNA polymerase promoter of vector pET22b(+) and expressed by the PURE system. The PURE system was prepared from the individually purified translation factors and aminoacyl-tRNA synthetases of E. coli [18], commercial T7 RNA polymerase, and high salt-washed and membranedeprived ribosomes. To study translocation of SfmC and TorT across the SecYEG translocon, the SecYEG complex as well as Ffh, FtsY, SecA were all purified to homogeneity from E. coli ( Figure 1). The DDM-solubilized SecYEG complex was reconstituted with E. coli phospholipids into small proteoliposomes by dialysis followed by sonication of the vesicles. The amounts of SecYEG complex reconstituted with a given amount of phospholipids and the volume of SecYEG proteoliposomes to be used per transport assay were each optimized to give maximal translocation efficiency of SfmC.
SecA is Sufficient to Transport a Secretory Protein with Hydrophobic Signal Sequence across SecYEG, but Efficient Translocation Requires SRP and SR in Addition Figure 2A compares translocation of the precursors pSfmC and pTorT into SecYEG proteoliposomes with that of the classical secretory protein OmpA. Translocation was measured by the fraction of each precursor that became resistant towards digestion by proteinase K (PK) in the presence of SecYEG proteoliposomes. In the presence of SecA, about 34% of pOmpA was found translocated into SecYEG proteoliposomes (lane 4). As expected for the SRP/SR-independent OmpA protein, Ffh plus FtsY neither significantly stimulated the SecA-mediated translocation efficiency (lane 6) nor were they able to replace SecA (lane 8). In contrast, maximal translocation efficiencies of both pSfmC and pTorT were only obtained in the presence of Ffh, FtsY and SecA (lane 6). SecA was indispensable for translocation of pSfmC and pTorT (lane 8), but different from pOmpA was not sufficient to obtain maximal translocation into the SecYEG proteoliposomes (lane 4).
The stimulating effect by the E. coli SRP/SR system on the translocation of pSfmC and pTorT into SecYEG proteoliposomes was obtained upon the simultaneous addition of Ffh and FtsY but without adding extra 4.5S RNA. Figure 2B [21]. The results depicted in Figure 2A and B therefore indicate that also for secretory proteins harbouring hydrophobic signal sequences, SecA is necessary and sufficient for translocation. In contrast to classical secretory proteins such as OmpA, the presence of more hydrophobic signal sequences obviously requires the involvement of the bacterial SRP/SR system to achieve optimal translocation.

Site-specific Photo Cross-linking Using the PURE System
We surmised that the beneficiary effect by SRP/SR on the translocation of SfmC might be mediated by the recruitment of Ffh to the hydrophobic signal sequence at the ribosome. In order to demonstrate this directly, we expanded the PURE system such as to allow the site-specific incorporation of the photo-activatable cross-linker p-benzoyl-phenylalanine (pBpa) into ribosome-associated nascent chains (RNCs) of SfmC via the suppression of amber stop codons engineered into the sfmC gene. To this end, a pBpaaccepting suppressor tRNA encoded by plasmid pEVOL-pBpF [22] was prepared by chloroform/phenol extraction and isopro- His-tagged variants of Ffh, FtsY, and SecA were over-expressed in E. coli and purified from cell extracts by metal affinity chromatography. The SecYEG complex was purified by metal affinity chromatography using a DDM-solubilized membrane pellet obtained from a SecY His EG-overproducing E. coli strain. Purified proteins were displayed by SDS-PAGE and staining with Coomassie Blue. doi:10.1371/journal.pone.0092994.g001 panol fractionation of total tRNA from E. coli cells that had been transformed with pEVOL-pBpF. The same plasmid also encodes a pBpa-specific aminoacyl-tRNA synthetase, which was modified to contain a C-terminal His-tag to allow its purification and use in the PURE system. SfmC-RNCs could quantitatively be synthesized in an oligodeoxynucleotide-dependent manner [23] if the PURE system was deprived of all release factors.
In the Absence of Membranes, the Signal Sequence of SfmC-RNCs is Found in Contact with Ffh To identify binding partners of the SfmC signal sequence before and after membrane targeting, we synthesized SfmC-RNCs of 126 amino acid length that carried pBpa in the hydrophobic core of their signal peptide ( Figure 3A, arrow). Upon supplementing the defined protein set of the PURE system with purified Ffh and FtsY, a prominent cross-linking product of the SfmC-RNCs

Targeting of SfmC-RNCs to SecY
We then moved the cross-linker from the signal sequence downstream to position V58 in the early mature part of SfmC ( Figure 3B). Much like with Bpa located in the signal sequence, the SfmC-RNCs-V58Bpa yielded a very prominent adduct of about 60 kDa (asterisk, lane 1) that was totally dependent on the addition of Ffh (cf. lane 5). Hence a stretch of SfmC extending from its signal sequence up to position V58 was exposed to Ffh after having emerged from the ribosome. The cross-linker in position V58 also yielded the prominent 100 kDa adduct if SecA and SecYEG proteoliposomes were provided (arrow head, compare lanes 2 and 4). A slight interaction of SecA with V58 in the early mature part of SfmC was obtained even in the absence of SecYEG proteoliposomes, provided that the competing Ffh was missing (compare lanes 1 and 5). This finding is in line with the idea that SecA might interact with nascent chains due to its association with ribosomes (see Discussion). The shift of the cross-linking partners from Ffh to SecA observed upon addition of SecYEG proteoliposomes (compare lanes 1 and 2) would be expected if Ffh/FtsY was to target SfmC-RNCs to the SecYEG translocon. In fact, SfmC-RNCs with pBpa at position V58 also efficiently cross-linked to a resident protein of the SecYEG proteoliposomes (dot), which by immunoprecipitation was identified as SecY (lane 8).

Discussion
Signal sequences, whose hydrophobicity exceeds a certain threshold level, have been proposed to direct secretory proteins to the SRP-dependent co-translational export pathway of E. coli [14,15,16]. This was deduced first from the diminished periplasmic export of these secretory proteins observed specifically when the Ffh protein was inactivated [14]. Secondly it was shown that in contrast to classical signal sequences, highly hydrophobic signal peptides can mediate export of a non-secretory protein, suggesting that by fusing a cytosolic protein with a hydrophobic signal sequence it can be withdrawn from folding in the cytosol through rerouting it to the SRP-dependent co-translational export pathway [15].
By use of an experimental system that allows studying the individual contributions of Ffh, FtsY and SecA to translocation across the SecYEG translocon, we were now able to establish the dominant role of SecA in the export of this kind of secretory proteins. Thus we could demonstrate that SecA is even sufficient for their translocation, while SRP and SR turned out to render transport more efficient. The latter finding would be consistent with the notion that because of inherently rapid folding kinetics, some secretory proteins require a hydrophobic signal sequence in order to co-translationally enter their export route and thereby escape premature folding [14].
Channeling secretory proteins into the SRP/SR-dependent cotranslational pathway invokes their recognition early during translation at the ribosome. We therefore analyzed RNCs carrying a protein with a hydrophobic signal sequence (SfmC) and could show a definite interaction of its signal sequence with the Ffh protein, which is in clear contrast to our previous data obtained with the less hydrophobic signal sequence of OmpA [23,24]. By virtue of its specific binding to the ribosomal protein L23 located at the orifice of the ribosomal exit tunnel, SecA was recently proposed to interact with nascent chains of classical secretory proteins, such as maltose-binding protein and b-lactamase, in order to increase the efficiency of their posttranslational targeting [25]. In fact, binding of SecA to a classical signal sequence had been shown to occur within a certain size frame of short nascent OmpA chains even in the presence of Ffh [23]. Under similar experimental conditions, the SfmC-RNCs analysed here did not interact with SecA whenever Ffh was present. This would be consistent with the predominant recognition of the more hydrophobic signal peptide of SfmC by SRP as suggested by previous in vivo analyses [14]. Only in the experimentally established absence of Ffh, a contact between SecA and SfmC-RNCs was detected and involved predominantly membranebound SecA. Thus cytosolically located SecA might play only a minor role as a targeting factor for proteins such as SfmC, whereas SfmC-RNCs can likely be targeted to the membrane through SecYEG-associated SecA, which in vivo is present at about 10fold higher concentrations than Ffh [1]. Collectively, our in vitro analysis employing highly defined experimental conditions demonstrate that the SRP dependence of secretory proteins with highly hydrophobic signal sequences is a result of a co-translational recognition of the hydrophobic signal sequence but that it is no prerequisite for the translocation of those secretory proteins, which can be executed solely by SecA.
TAG amber stop codons were introduced by mutagenizing PCR into the sfmC gene of pET-SfmC at positions encoding I12 and Y14 within the signal sequence (Phusion High-Fidelity DNA Polymerase, NEB) using the primers SfmCI12for, SfmCI12rev, SfmCY14for, SfmCY14rev (Table 2). Similarly, the pBpa-tRNAsynthetase gene of plasmid pEVOL-pBpF was modified to encode a C-terminal His-tag using the primers pBpaRSHisfor and pBpaRSHisrev ( Table 2) generating plasmid pEVOL-pBpF-His.

Purification of Proteins
Purification of SecA, Ffh, FtsY, pBpa-tRNA-synthetase and Histagged components of the PURE system [17] were performed as described previously [4] with some modifications. Proteins were affinity-purified via their His-tags using TALON Metal Affinity Resins (Clontech) and 200 mM imidazole for elution. Ffh was stored at 220uC in HT buffer (50 mM HEPES-KOH, pH 7.6, 100 mM potassium acetate, 10 mM magnesium acetate, 1 mM dithiothreitol) supplemented with 50% glycerol. Other purified proteins were stored at 270uC in HT buffer supplemented with 30% glycerol. The SecYEG complex was purified as described [4].

Isolation of Ribosomes
High salt-washed ribosomes were prepared as described [23] except that prior to sucrose gradient centrifugation, ribosomes were spun through a 1.44 M sucrose cushion.

Reconstitution of Proteoliposomes
Preparation of SecYEG proteoliposomes was performed as described [13] except that 10 mg of purified SecYEG were reconstituted with 400 mg of E. coli phospholipids (Avanti Polar Lipids, Alabaster, AL).

Synthesis of RNCs
For the synthesis of ribosome-associated nascent SfmC chains of 126 amino acid length, 4 mg of the oligodeoxynucleotide GATAAAATCGCCAGTTGCAAAAC, 0.3 mg anti-10Sa RNA oligodeoxynucleotide TTAAGCTGCTAAAGCG-TAGTTTTGGTCGTTTGCGACTA and 3 units RNaseH were additionally supplied in each 25 ml reaction according to [24].

Photo Cross-linking
For the site-specific incorporation of pBpa into amber stop codon variants of SfmC, 80 mM pBpa, 20 mg/mL pBpa-tRNA synthetase, and 40 mg/mL pBpa-tRNA sup prepared as described [30] were additionally supplied in each 25 ml reaction. After incubation at 37uC for 1 hour, the samples were exposed to UV irradiation at 365 nm for 20 min on ice.

Miscellaneous
Protein translocation was analyzed by proteinase K resistance [31]. Immunoprecipitations were performed as described [13] except that after denaturation with SDS, samples were freed of precipitated material by centrifugation in a tabletop microfuge.

Author Contributions
Conceived and designed the experiments: MM. Performed the experiments: YZ. Analyzed the data: YZ MM. Contributed reagents/materials/ analysis tools: TU. Wrote the paper: MM.