Microbiology (2006), 152, 1441–1450
DOI 10.1099/mic.0.28734-0
Surface plasmon resonance-based interaction studies reveal competition of Streptomyces lividans type I signal peptidases for binding preproteins Nick Geukens,1 Smitha Rao C. V.,1 Rafael P. Mellado,2 Filip Frederix,3 Gunter Reekmans,3 Sophie De Keersmaeker,1 Kristof Vrancken,1 Kristien Bonroy,3 Lieve Van Mellaert,1 Elke Lammertyn1 and Jozef Anne´1 1
Laboratory of Bacteriology, Rega Institute for Medical Research, Katholieke Universiteit Leuven, Minderbroedersstraat 10, B-3000 Leuven, Belgium
Correspondence Jozef Anne´
[email protected] 2
Department of Microbial Biotechnology, Centro Nacional de Biotecnologia, Campus de la Universidad Autonoma, Cantoblanco, 28049 Madrid, Spain
3
IMEC, MCP-ART, Kapeldreef 75, B-3001 Leuven, Belgium
Received 30 November 2005 Accepted 25 January 2006
Type I signal peptidases (SPases) are responsible for the cleavage of signal peptides from secretory proteins. Streptomyces lividans contains four different SPases, denoted SipW, SipX, SipY and SipZ, having at least some differences in their substrate specificity. In this report in vitro preprotein binding/processing and protein secretion in single SPase mutants was determined to gain more insight into the substrate specificity of the different SPases and the underlying molecular basis. Results indicated that preproteins do not preferentially bind to a particular SPase, suggesting SPase competition for binding preproteins. This observation, together with the fact that each SPase could process each preprotein tested with a similar efficiency in an in vitro assay, suggested that there is no real specificity in substrate binding and processing, and that they are all actively involved in preprotein processing in vivo. Although this seems to be the case for some proteins tested, high-level secretion of others was clearly dependent on only one particular SPase demonstrating clear differences in substrate preference at the in vivo processing level. Hence, these results strongly suggest that there are additional factors other than the cleavage requirements of the enzymes that strongly affect the substrate preference of SPases in vivo.
INTRODUCTION Proteins designed for transport across the bacterial cytoplasmic membrane are generally synthesized in the cytoplasm as a precursor with a cleavable signal peptide, which targets the precursor proteins to the translocase (Fekkes & Driessen, 1999). Preproteins can be targeted to two different translocases by the Sec/SRP (signal recognition particle)dependent pathway and the Tat (twin-arginine translocation)dependent pathway. The most important difference for both pathways is the translocation-competent state of the preprotein. Whereas for Sec-dependent secretion, preproteins must be in an unfolded state, the Tat pathway accommodates the translocation of folded proteins (Berks et al., 2003). Upon translocation across the cytoplasmic membrane, the signal peptide is enzymically removed by a type I signal peptidase (SPase), a membrane-bound endopeptidase Abbreviations: SPase, type I signal peptidase; SPR, surface plasmon resonance; Tat, twin-arginine translocation.
0002-8734 G 2006 SGM
Printed in Great Britain
(reviewed by Paetzel et al., 2000 and van Roosmalen et al., 2004), thereby releasing the mature protein at the outer side of the membrane (Dalbey & Wickner, 1985). Whereas the presence of only one chromosomal type I SPase seems to be typical for many Gram-negative bacteria, the occurrence of multiple type I SPases within a single strain is commonly observed in Gram-positive bacteria. The largest number of type I SPases has, thus far, been found in Bacillus cereus, which contains genes for seven paralogous enzymes of this type. Bacillus anthracis and Bacillus subtilis both contain five chromosomally encoded type I SPases. In other Gram-positive bacteria, like Streptomyces lividans (Schacht et al., 1998; Parro et al., 1999), Streptomyces coelicolor, Clostridium perfringens and Bacillus amyloliquefaciens (Chu et al., 2002), four different chromosomally encoded type I SPases have been identified. Furthermore, Lactobacillus plantarum, Listeria monocytogenes and Staphylococcus epidermidis have three SPase-encoding genes, whereas Bacillus halodurans, Staphylococcus aureus (spsA and spsB) (Cregg 1441
N. Geukens and others
et al., 1996) and Staphylococcus carnosus (sipA and sipB) have two SPase-encoding genes. Lactococcus lactis, Streptococcus pneumoniae (Zhang et al., 1997), Streptococcus mitis, Mycobacterium leprae and Mycobacterium tuberculosis are examples of Gram-positive bacteria having only one SPaseencoding gene. In bacteria that have one type I SPase, this enzyme is essential for the processing of all secretory preproteins. When multiple type I SPases are present, like in B. subtilis and S. lividans, none of the corresponding genes was found to be essential for cell viability by itself (Tjalsma et al., 1997, 1998; Palacin et al., 2002). This indicates that the paralogous enzymes can at least partly complement each other. However, clear differences in SPase substrate specificity have been shown in B. subtilis and S. lividans. In B. subtilis, depletion of both SipS and SipT resulted in a dramatic defect in preprotein processing and secretion. Both SPases serve to increase the preprotein processing capacity of the cell and were assigned as major SPases. SipU, SipV and SipW were found to be of minor importance for preprotein processing (Tjalsma et al., 1998). In S. lividans, inactivation of SipY resulted in a severe defect in protein secretion and is therefore suggested to be a major SPase. SipY depletion also resulted in sporulation deficiency, suggesting a specific role of SipY in the processing and export of sporulation-related proteins (Palacin et al., 2002).
Here, we used a previously developed biosensor containing immobilized SPases to investigate SPase specificity at the binding level. Results showed no preferential binding of a particular preprotein to one or more SPases. In addition, no specificity in in vitro preprotein processing was observed. However, when the involvement of each SPase in the processing of the same preproteins was investigated by monitoring the accumulation of these proteins in the culture medium of S. lividans wild-type and single SPase deletion mutants, high-level secretion of several proteins was clearly dependent upon one specific SPase.
METHODS Bacterial strains, growth conditions and DNA techniques.
Escherichia coli strains TG1 (Sambrook et al., 1989) and BL21(DE3)pLysS (Studier & Moffat, 1986) were grown at 37 uC in Luria–Bertani (LB) medium supplemented with ampicillin (50 mg ml21) or kanamycin (50 mg ml21) when needed. S. lividans and its derivatives were routinely precultured in 5 ml phage medium (Korn et al., 1978) supplemented with thiostrepton (10 mg ml21) and/or kanamycin (50 mg ml21) at 27 uC for 48 h with continuous shaking at 300 r.p.m. The mycelium was subsequently homogenized and inoculated (2 %) in NM medium (Van Mellaert et al., 1994) for an additional 24 h at 27 uC. Oligonucleotides are listed in Table 1. For all DNA manipulations, standard techniques were used (Sambrook et al., 1989).
Table 1. Oligonucleotides used in this work Restriction endonuclease cleavage sites are in bold, 6-His-encoding sequences are in italic and cMyc epitope encoding sequences are underlined. Optimized Shine–Dalgarno sequences are underlined and in italic. Oligonucleotide CelAHis5 CelA3 CelBHis5 CelB3 XlnBHis5 XlnB3 XlnCHis5 XlnC3 StiHis5 Sti3 DagAHis5 DagA3 MelC1His5 MelC13 TmHis5 Tm3 XlnBrbsEco XlnBHin CelArbsEco CelAHin CelBrbsEco CelBHin
1442
Sequence (5§–3§)
Restriction site
TACATATGCACCATCATCACCATCACAAACGCCTTTTGGCCC TAGGATCCTCACAGATCCTCCTCTGAGATGAGCTTCTGCTCGCTCGTCGGGAAGTCG TACATATGCACCATCATCACCATCACCGAACGTTACGGCCCC TAGGATCCTCACAGATCCTCCTCTGAGATGAGCTTCTGCTCCACCGTGGTGCAGGCGG TACATATGCACCATCATCACCATCACAACCTGCTCGTCCAGCC TAGGATCCTCACAGATCCTCCTCTGAGATGAGCTTCTGCTCGCCCGCGCTGCAGGACACG TACATATGCACCATCATCACCATCACCAGCAGGACGGCACACAGC TAGGATCCTCACAGATCCTCCTCTGAGATGAGCTTCTGCTCACCGCTGACCGTGATGTTCG TACATATGCATCACCATCACCATCACCGCAACACCGCGCGCTGG TAGAATTCTCAGAACGTGAAGACGCTCGC TACATATGCACCATCATCACCATCACGTCAACCGACGTGATCTCATC TAGGATCCCTACACGGCCTGATACGTCC TACATATGCACCATCATCACCATCACCCGCAACTCACCCGTCGTCGC TAGGATCCTCACAGATCCTCCTCTGAGATGAGCTTCTGCTCGTTGGAGGGGAAGGGGAGGAGC AGAATTCCATATGCACCATCATCACCATCACCGCGTACGGGCACTTCG TAGAATTCGGGCTAAAGGCAGCGAG TAGAATTCAGAAAGGAGGGAACACCCCCCATGAACCTGC TAAAGCTTTCAGCCCGCGCTGCAGGACACG TAGAATTCAGAAAGGAGCGCTATGAAACGCCTTTTGGC TAAAGCTTTCAGCTCGTCGGGAAGTCGTCG TAGAATTCAGAAAGGAGCCCCCATGCGAACGTTACGGC TAAAGCTTTTACACCGTGGTGCAGGCGG
NdeI BamHI NdeI BamHI NdeI BamHI NdeI BamHI NdeI EcoRI NdeI BamHI NdeI BamHI EcoRI, NdeI EcoRI EcoRI HindIII EcoRI HindIII EcoRI HindIII
Microbiology 152
SPase substrate specificity
Vector constructions Plasmids used for secretory production of proteins in S. lividans. For the secretory production of CelA, CelB and XlnB, the
genes were amplified by PCR using chromosomal DNA as template with the primers CelArbsEco/CelAHin, CelBrbsEco/CelBHin and XlnBrbsEco/XlnBHin, respectively. The upstream primer contains an optimized Shine–Dalgarno sequence and an EcoRI restriction site, while the downstream primer contains a HindIII restriction site. Next, the obtained EcoRI–HindIII DNA fragments were ligated into pBSVsi after the vsi promoter. The resulting pBSVsiCelA, pBSVsiCelB and pBSVsiXlnB plasmids were digested with XbaI and HindIII to give DNA cassettes containing the vsi promoter and CelA/CelB/XlnB-encoding sequences that were subsequently ligated into the corresponding sites of pFD666, resulting in pFDCelA, pFDCelB and pFDXlnB, respectively. For secretory production of Tm, pAX5a (Fass & Engels, 1996) was treated with EcoRI/Klenow polymerase/HindIII to give a DNA fragment containing the ermEup promoter and the tm gene. Ligation of this fragment into BamHI/Klenow polymerase-treated HindIIIdigested pFD666 resulted in pFDTm. To construct a plasmid overexpressing XlnC, pIJ486XlnC (Schaerlaekens et al., 2004) was digested with HindIII and BamHI to give a DNA cassette containing the vsi promoter and XlnC-encoding sequence. This cassette was then ligated into the corresponding sites of pFD666 to give pFDXlnC. Plasmid pFDMelC, used for secretory production of MelC1 (transactivator protein) and MelC2 (apotyrosinase), was constructed by inserting an EcoRV–PstI pIJ702 (Kieser et al., 2000) fragment containing the MelC promoter and coding sequence into BamHI/Klenow polymerasetreated PstI-digested pFD666. Plasmids used for preprotein overproduction in E. coli. Using
S. lividans chromosomal DNA as template, DNA fragments encoding preCelA, preCelB, preXlnB, preXlnC and preSti-1 were amplified by PCR with the respective primers CelAHis5/CelA3, CelBHis5/CelB3, XlnBHis5/XlnB, XlnCHis5/XlnC3 and StiHis5/Sti3. Next, the amplified fragments were digested with NdeI and BamHI and ligated into NdeI/BamHI-treated pET3a, resulting in pCelA, pCelB, pXlnB, pXlnC and pSti, respectively. Plasmid pTm, used for preTm overproduction, was constructed by inserting an NdeI–EcoRI DNA fragment, amplified by PCR using pAX5a as template with the primers TmHis5/Tm3, into the corresponding sites of pET3a. For preDagA overproduction, an NdeI–BamHI fragment, amplified by PCR using S. coelicolor chromosomal DNA as template with the primers DagAHis5/DagA3, was inserted into the corresponding sites of pET3a to give pDagA. PreMelC1 was produced in E. coli from pMelC, a pET3a derivative in which an NdeI–BamHI fragment was cloned that was amplified by PCR with the primers MelCHis5/ MelC3 using the plasmid pIJ702 as template. Secretory production of target proteins in S. lividans. To
express celA, celB, xlnB, xlnC, tm and melC1melC2 in S. lividans wildtype and single SPase mutants (Palacin et al., 2002), protoplasts of these strains were transformed with pFDCelA, pFDCelB, pFDXlnB, pFDXlnC, pFDTm and pFDMelC, respectively. Plasmid pAGAs5 (Palacin et al., 2002) was used to drive dagA expression. sti-1 was not cloned in multicopy, since the expression product could be detected when encoded from the chromosome. All but dagA expression cultures were grown in NM medium. dagA expression cultures were grown in NMMP (26 casein peptone) (Parro et al., 1997). After 24 h, proteins present in the supernatant were precipitated using TCA (20 % final concn). To rule out variation resulting from different growth rates of S. lividans wild-type and single http://mic.sgmjournals.org
SPase mutants, the samples taken for precipitation were corrected for mycelial dry weight. Proteins were separated by SDS-PAGE as described by Laemmli (1970) and transferred onto a nitrocellulose Porablot membrane (Macherey Nagel). Target proteins were detected colorimetrically using specific polyclonal antisera and anti-rabbit immunoglobulin G conjugated to alkaline phosphatase. Expression and purification of preproteins. For the expression
of preproteins, E. coli BL21(DE3)pLysS was transformed with the respective expression plasmids and grown at 37 uC to an OD600 of 0?6. Then IPTG was added to a final concentration of 1 mM to drive expression. At the same time 2 mM sodium azide was added to block ATPase activity of SecA and, hence, to prevent translocation and cleavage of the signal peptide from Sec-dependent preproteins. Next, cultures were grown at 37 uC for an additional 4 h. From a 600 ml IPTG-induced culture, the E. coli BL21(DE3)pLysS cells producing the desired hexahistidine (6-His)-tagged preprotein were harvested. Preproteins accumulated mainly in the cell as insoluble inclusion bodies. Hence, they were purified from the resulting cell pellet under denaturing conditions (8 M urea) on a Ni2+-NTA column (Qiagen) as recommended by the manufacturer. The obtained fractions were screened for the presence of preprotein using SDS-PAGE and fractions containing the desired preprotein were pooled. Finally, urea in the protein samples was removed by gel filtration on a PD-10 column (GE Healthcare), which was previously equilibrated with 10 mM HEPES, pH 9?0, and 0?5 % Triton X-100. Using this procedure, typically 500 mg pure preprotein could be obtained. Surface plasmon resonance (SPR) spectroscopy. All SPR
experiments were carried out at 20 uC on a Biacore2000 instrument. Covalent immobilization of S. lividans SPases on a self-made mixed SAM (self-assembled monolayer) surface (on a J1-chip of Biacore) was accomplished via coupling of their primary amines to the carboxylic groups of the SAM surface as described by Geukens et al. (2004). A control surface was prepared by running the activation and deactivation procedure only. Preproteins [10 mM in 10 mM HEPES, pH 9?0, 0?5 % Triton X-100 (running buffer)] were perfused over the immobilized SPase surface (concentration of the SPase is rate-determining) at a flow rate of 20 ml min21. After 300 s of association, the preprotein sample was replaced by running buffer for 300 s, allowing the complex to dissociate. Regeneration of the surface was performed by two pulses of 10 mM glycine, pH 2?2. Binding of preproteins to an immobilized SPase was analysed by calculating the net increase in RU (resonance units) during the association phase. A DRU value of 1000 corresponds to 1 mg m22 (Stenberg et al., 1991; Jo¨nsson et al., 1991). The net increase for each preprotein is the measured increase in RU minus the RU increase due to the bulk solution and minus the RU increase resulting from nonspecific binding of the respective preprotein to the activated and deactivated surface. The recognition signals were corrected for the difference in SPase immobilization which was in general smaller than 8 %. Competition studies of preproteins and derived peptides were performed as follows. A biosensor surface containing immobilized SipY was perfused with running buffer containing either 5, 10 or 15 mM Sti1-derived peptide (pepSti or pepContS; see Table 2). As a control, the same volume of running buffer was injected. Next, preSti-1 (5 mM in running buffer) was injected. After 600 s of association, the preprotein sample was replaced by running buffer for an additional 600 s. Regeneration of the surface was performed by two pulses of 10 mM glycine, pH 2?2. The resulting association and dissociation curves of SipY-preSti-1 with and without previously added peptide were corrected for non-specific binding and plotted as a function of time. 1443
N. Geukens and others
Table 2. Sti-1- and XlnC-derived peptides SPase cleavage sites are in bold. Peptide pepSti pepXlnC pepContS pepContX
Sequence BiotNH-TAGTAQAEA-COOH BiotNH-LPGTAHAAT-COOH BiotNH-AAQTGTEAA-COOH BiotNH-AAGPLTHTA-COOH
A similar protocol was followed for evaluating the competitive effect of an XlnC-derived peptide (pepXlnC or pepContX) on preXlnC binding to immobilized SipY. In vitro SPase assay. Analysis of in vitro SPase activity was per-
formed as described by Geukens et al. (2002) with the exception that the assay buffer was the same as the running buffer used in the in vitro interaction studies described above.
RESULTS To investigate if preproteins preferentially interact with one or more of the S. lividans SPases, we first selected a set of eight secretory preproteins originating from different Streptomyces species. The selected preproteins were the subtilisin inhibitor (Sti-1), cellulase A (CelA), cellulase B (CelB), xylanase B (XlnB) and xylanase C (XlnC) from S. lividans, agarase (DagA) from S. coelicolor, tendamistat (Tm) from Streptomyces tendae and tyrosinase (MelC= MelC1MelC2) from Streptomyces antibioticus. The tyrosinase is encoded by melC2, the second ORF of the melanin operon (melC). The upstream gene melC1, encoding a transactivator protein with a twin-arginine signal peptide, has been demonstrated to be essential for both tyrosinase activation and secretion (Chen et al., 1992; Leu et al., 1992). For in vitro studies, only the MelC1 subunit was used. Sti-1 has been shown to be secreted Sec-dependently (Schaerlaekens et al., 2004), whereas the export of XlnC and MelC is Tatdependent in S. lividans (Schaerlaekens et al., 2001, 2004). CelA, CelB, Tm and XlnB contain typical Sec-dependent signal peptides, whereas DagA contains a twin-arginine motif and is predicted to be exported Tat-dependently.
Target proteins have signal peptides that vary in length from 27 to 49 amino acid residues and that contain a typical SPase cleavage site following the 2321 rule in their C region (Table 3). SPase–preprotein binding experiments An SPR-based assay, which we previously developed and validated for the analysis of SPase–preprotein interaction (Geukens et al., 2004), was used to calculate the net SPase binding of the eight secretory preproteins. As such, a semiquantitative comparison of the association of precursor forms of target proteins and SPase was obtained. For this purpose, 6-His-tagged S. lividans SPases were overproduced in E. coli and purified as described previously (Geukens et al., 2002). Precursors of Sec-dependent proteins were obtained after overproduction in E. coli and subsequent block of translocation. To obtain precursors of Tat-dependent proteins, specific blocking of translocation was not required. All overproduced Tat substrates (XlnC, DagA and MelC1) were poorly secreted in E. coli and thus remained mainly in the cytoplasm as precursor (data not shown). When these purified preproteins were perfused over the SPase surface (Fig. 1), we observed that (1) all tested preproteins interact with each of the immobilized SPases, (2) different preproteins had clear differences in SPase binding affinities, i.e. certain preproteins (Sti-1, XlnC, etc.) interact more efficiently with all SPases than others (CelA, CelB, etc.) and (3) most interestingly, a particular preprotein binds with similar efficiency to each of the SPases, suggesting SPases are able to compete for binding preproteins. As seen in Fig. 1, only small differences in SPase binding affinity could be observed for a specific preprotein. Specificity assay For testing the specificity and the physiological relevance of the measured SPase–preprotein interaction, we used a small peptide consisting of the SPase cleavage site region to see if it could compete with binding of its corresponding full-length preprotein to the SPase. Whereas a pentapeptide ranging
Table 3. Signal peptides of target proteins Twin-arginine motifs are underlined. Protein Sti-1 Tm XlnC XlnB CelA CelB DagA MelC1
1444
Signal peptide MRNTARWAATLGLTATAVCGPLAGASLASPATAPA MRVRALRLAALVGAGAALALSPLAAGPASA MQQDGTQQDRIKQSPAPLNGMSRRGFLGGAGTLALATASGLLLPGTAHA MNLLVQPRRRRRGPVTLLVRSAWAVALARSPLMLPGTAQA MKRLLALLATGVSIVGLTALAGPPAQA MRTLRPQARAPRGLLAALGAVLAAFALVSSLVTAAAPAQA MVNRRDLIKWSAVALGAGAGLAGPAPAAHA MPELTRRRALGAAAVVAAGVPLVALPAARA
Reference S D A D A D A D
Strickler et al. (1992) Ve´rtesy et al. (1984) Shareck et al. (1991) Shareck et al. (1991) The´berge et al. (1992) Wittmann et al. (1994) Kendall & Cullum (1984) Bernan et al. (1985)
Microbiology 152
SPase substrate specificity
Fig. 1. Net increases in RU resulting from binding of 10 mM purified preproteins to each of the S. lividans SPases covalently immobilized on a mixed SAM chip in a 135 s period. The net increase for each preprotein is the measured increase in RU minus the RU increase due to the bulk solution and minus the RU increase resulting from non-specific binding of the respective preprotein to the activated and deactivated surface. The plotted values are the mean values of three independent measurements.
from 23 to +2 relative to the SPase cleavage site was found to be the minimal substrate sequence to be recognized by a SPase, Dev et al. (1990) reported that a nonapeptide ranging from 27 to +2 was a much more efficient substrate. Therefore, the interaction of two preproteins, i.e. preSti-1 and preXlnC, to SipY was studied when, prior to the analysis of this interaction, increasing concentrations of a http://mic.sgmjournals.org
nonapeptide covering the proposed SPase–preprotein interaction region were added. More specifically, increasing peptide concentrations (pepSti or pepXlnC) were perfused over the SPase surface prior to the injection of an invariable amount of corresponding preprotein (preSti-1 or preXlnC, respectively). As a control, scrambled peptides (pepContS or pepContX) were used. Peptide sequences are listed in Table 2. 1445
N. Geukens and others
Fig. 2. Overlay plot of sensorgrams showing the association and dissociation between preSti-1 (5 mM in running buffer) and covalently immobilized SipY (a, b) and between preXlnC (5 mM in running buffer) and immobilized SipY (c, d). Prior to these interactions, the SipY surface was perfused with 15, 10, 5 or 0 mM competitive peptide [pepSti (a) or pepXlnC (c)] or similar concentrations of scrambled peptide [pepContS (b) or pepContX (d)]. Resulting sensorgrams were corrected for changes in RU due to the bulk solution and non-specific binding.
Fig. 2 shows that increasing concentrations of pepSti and pepXlnC resulted in a significant reduction in binding of the respective preprotein to SipY, suggesting that the short peptide can compete with the full-length preprotein for SPase binding. Since the effects of the scrambled peptides on preprotein binding to SipY were significantly less, we could conclude that the interaction measured in our setup mainly originates from SPase binding at the proposed cleavage site region in the preprotein. SPase enzymic activity assay To ascertain that the purified 6-His-tagged SPase enzymes used for covalent immobilization on the biosensor chip also retained their enzymic activity, their in vitro preprotein processing ability was monitored in parallel. Assessment of the in vitro activity of the purified SPase enzymes was based on the processing of purified preSti-1 (Sec-dependent substrate) and preXlnC (Tat-dependent substrate). Under similar conditions as in the binding assay, all four SPase proteins were able to process both substrates (Fig. 3), clearly demonstrating their functionality. Interestingly, as observed for the binding of preproteins to the SPases, the efficiency of in vitro preprotein processing by each SPase is not extremely different.
medium was studied. It should be noted that these S. lividans single SPase mutants have been characterized in detail before and showed a reduced extracellular protease activity in comparison to that of the wild-type (about 15, 30 and 45 % for DX, DZ and DW mutants and a protease activity below the detection limit for the DY mutant; Palacin et al., 2002). S. lividans wild-type and single SPase mutants (DW, DX, DY and DZ) were transformed with plasmids driving the secretory production of CelA, CelB, XlnB, XlnC, Tm, MelC or DagA. Sti-1 is the most abundantly secreted protein by S. lividans. Hence, there was no need to clone the gene in multicopy and the Sti-1 secretion was therefore evaluated when expressed from the native gene on the chromosome. Respective cultures were grown as described in Methods
Effect of SPase depletion on high-level secretion of target proteins The experiments described above demonstrate that a particular secretory preprotein is not preferentially recognized by any of the S. lividans SPases in vitro. In addition, a particular preprotein was efficiently processed by each SPase in vitro (only shown for preSti-1 and preXlnC) using similar conditions as for the SPR analysis. In the next step, the contribution of each SPase to in vivo processing of the same group of proteins was studied. Therefore, the effect of single SPase deletions on the release of the eight selected secretory proteins into the culture 1446
Fig. 3. Western blot analysis with specific polyclonal antibodies of in vitro preSti-1 and preXlnC cleavage by each of the SPases. Precursor (p) and mature form (m) were previously separated on a 12?5 % SDS-polyacrylamide gel. Purified SPase (2?5 mM final concn) was added to 20 mM substrate and incubated for 15 h at 37 6C in 10 mM HEPES, pH 9?0, and 0?5 % Triton X-100. Lanes: 1, precursor; 2, precursor+SipW; 3, precursor+SipX; 4, precursor+SipY; 5, precursor+SipZ. Microbiology 152
SPase substrate specificity
and, after 24 h, the presence of target proteins in the culture medium was analysed by Western blotting. As shown in Fig. 4, secretion of Sti-1 was almost completely blocked in the SipY-depleted strain. In contrast, increased Sti-1 secretion was observed in the DX and DZ single mutants, whereas inactivation of SipW did not seem to have any effect on the appearance of Sti-1 in the extracellular medium.
These observations suggest that Sti-1 is preferentially processed in vivo by SipY. As in the case of Sti-1, secretion of XlnC was dramatically reduced in the DY mutant. However, XlnC secretion remained unaffected by inactivation of one of the other SPases. A similar observation was made for high-level secretion of CelA and CelB by S. lividans wild-type and single SPase mutants. Unfortunately, CelA was observed to be extremely sensitive to degradation when present in the culture medium. Growth of CelA-producing S. lividans cultures in minimal medium or replacement of glucose as carbon source by xylose, which was previously reported to limit CelA degradation (The´berge et al., 1992), did not result in a significant improvement (data not shown). In contrast, Tm secretion was not blocked by SipY depletion, but efficient secretion was found to be SipX-dependent. The appearance of Tm in the culture medium was not affected by inactivation of SipW, but increased as a result of SipY or SipZ depletion. Whereas the five secretory proteins listed above seem to be preferred substrates for one of the four S. lividans SPases, secretion of MelC, DagA and XlnB was not blocked or dramatically reduced by single SPase inactivation. MelC secretion in S. lividans was not visibly affected in the DX, DY and DZ mutants, but increased in the DW mutant and finally, for DagA and XlnB, no significant difference of single SPase inactivation on the appearance of the mature protein in the culture medium was observed. In contrast to the in vitro situation, in vivo processing of preSti-1, preTm, preCelA, preCelB and preXlnC was clearly dependent on one SPase, in particular SipX or SipY. preDagA, preMelC1 and preXlnB were not preferentially processed by one particular SPase in vivo, suggesting that multiple SPases could be actively involved in processing of these precursors. These data support the view of an overlapping, non-identical substrate preference of the S. lividans SPases in vivo.
DISCUSSION Previously, we showed that none of the S. lividans SPases is essential for cell viability, suggesting that they can at least partly complement each other (Palacin et al., 2002). Nevertheless, clear differences in substrate specificity between SPases have also been observed (Geukens et al., 2001; Palacin et al., 2002).
Fig. 4. Western blotting analysis (12?5 % SDS-polyacrylamide gel) of extracellular protein extracts prepared from 24 h cultures of wild-type S. lividans and single SPase mutants overproducing the respective target proteins. Proteins were visualized using NBT/BCIP after immunodetection with specific polyclonal antibodies. http://mic.sgmjournals.org
Here, the substrate specificity of these SPases was analysed in more detail and its molecular basis was determined by evaluating preprotein binding to each of the SPases and high-level protein secretion in single SPase mutants. Most interestingly, we have observed that preproteins do not preferentially bind to a particular SPase, suggesting SPase competition for binding preproteins. In addition, no specificity was observed in in vitro preprotein processing. 1447
N. Geukens and others
However, analysis of the secretion of the same target proteins in S. lividans indicated that not all of these proteins are processed in vivo by each of the SPases with the same efficiency. Consequently, there appears to be no real correlation between preprotein binding and processing in vitro and preprotein processing in vivo. The most obvious examples are Sti-1, XlnC, CelB and Tm, which bind to all four SPases with a similar affinity and can be processed in vitro by each SPase with similar efficiency, but are preferentially processed in vivo by only one SPase. Concomitantly, these experiments suggest that factors other than cleavage requirements of the SPases strongly affect substrate preference in vivo. The same factors are most probably also responsible for the major role of SipY in the processing of exported proteins in intact cells. Indeed, our data demonstrate that the importance of SipY for efficient preprotein processing in vivo is not due to a more efficient binding of preproteins. Other factors seem to be involved, which are not known at this stage. It could be that preproteins are exported at defined membrane locations in the cell and that SipY is preferentially localized there. Maybe SipY is the best at processing preproteins cotranslationally during export across the membrane or SipY makes the most efficient interaction with the translocation pores. This is the subject of ongoing research. Preprotein binding to a SPase could be specifically reduced by a peptide covering the SPase cleavage site region. This clearly demonstrated that the signal measured in our setup originates from a physiologically relevant interaction. The competitive effect of the peptide on SPase binding of its corresponding full-length preprotein was small or negligible when the increasing peptide concentrations were directly added to the preprotein solution (data not shown). This is most probably due to the fact that peptides are far less efficient substrates than full-length preproteins (Dev et al., 1990; Paetzel et al., 2000). Therefore, the peptide was perfused over the SPase surface prior to the analysis of SPase– preprotein binding. In this case, the competitive behaviour of the peptides was obvious. In in vitro SPase processing assays, Tat-dependent preproteins were found to behave similarly to Sec-dependent preproteins (only shown here for preSti-1 and preXlnC), i.e. they are processed by each SPase with comparable efficiencies and processing takes place in a similar time course. The observation that all four SPases are capable of processing all preproteins in vitro together with the fact that they all bind each preprotein tested suggests they are all actively involved in preprotein processing in vivo. Accordingly, DagA, MelC and XlnB secretion was not severely affected by the absence of a particular SPase and multiple SPases are involved in in vivo processing of these precursors. In contrast, high-level secretion of Sti-1, Tm, CelA, CelB and XlnC was clearly dependent on one SPase, in particular SipX or SipY. As mentioned above, the factors that determine this SPase-dependency for in vivo secretion and the major role for SipY are not yet known. All together, these observations agree with an overlapping, non-identical substrate specificity 1448
of the S. lividans SPases. Finally, it should also be noted that all these findings are related to overproduced secretory proteins, except for Sti-1 which is already efficiently secreted under native conditions. Some of these effects may therefore become unnoticed when proteins are produced in low amounts. Interestingly, an increase in secretion of some secretory proteins was observed in certain single SPase mutants. Because improved protein secretion in a particular single mutant was only observed for a few proteins rather than being a general issue, it is very unlikely that reduced extracellular protease activities can account for this. In addition, improved secretion is rarely observed in the DY mutant strain, which has the lowest extracellular protease activity. As a consequence, we favour the hypothesis that the SPase competition for binding preproteins can account for improved secretion of some target proteins in specific single SPase mutants. When the S. lividans SPases compete for the binding of precursor proteins, but process these preproteins with different efficiencies depending on the nature of the precursor, this may lead to improved secretion of target proteins. In particular, deletion of an SPase which cleaves a precursor efficiently is expected to result in reduced processing. However, deletion of an SPase which binds efficiently, but cleaves the substrate poorly, allows more precursor to bind to another SPase, which may cleave it more efficiently. This may result in improved secretion. Increased secretion of proteins in single SPase mutants has already described in B. subtilis (Van Dijl et al., 1992; Bron et al., 1998). Notably, inactivation of SipS resulted in an increased secretion of a-amylase in B. subtilis. As a consequence, SPase competition for binding precursor proteins does not seem to be restricted to S. lividans, but is likely to be a general feature of bacteria that contain multiple SPases. Importantly, these findings are also of biotechnological interest. In particular, the fact that SPases compete for binding preproteins combined with the fact that they show a kind of substrate preference in vivo indicates that one or some SPase(s) may interfere with efficient secretion of a particular protein. The obtained data can be used to construct the optimal ‘signal peptide processing machinery’ as a means to increase secretory production of a target protein. The data described here suggest that co-expression of sipY (as the major SPase) is too simple and not always the best choice. In reality, the situation is more complex. It is therefore our strategy to overexpress sip genes of SPases that process the target preprotein efficiently and delete sip genes of SPases that bind but inefficiently process the preprotein of interest. This work also demonstrated that none of the S. lividans SPases is specifically dedicated to the processing of Sec- or Tat-dependent precursors and consequently these SPases act in a secretion-pathway-independent manner. This suggests that, although the preproteins are in a completely different conformational state during or after translocation across the cytoplasmic membrane, the signal peptides of Sec- and Microbiology 152
SPase substrate specificity
Tat-dependent preproteins are assumed to be presented in a similar manner to the SPase.
Geukens, N., Parro, V., Rivas, L. A., Mellado, R. P. & Anne´, J. (2001).
Finally, it should also be noted that the binding affinity of proteins such as Sti-1, Tm, XlnC and DagA to each SPase was significantly higher than that of CelA, CelB, XlnB and MelC1. Interestingly, mature forms of precursors that showed a high SPase binding affinity accumulated in higher amounts than mature proteins from precursors that showed a low SPase binding affinity. It is clear that the secretion efficiency increases if proteins can efficiently interact with components of the secretion pathway. Although this certainly has to be further elaborated, this assay may be included as part of a predictive assay for the efficiency of secretion for a particular protein.
Geukens, N., Lammertyn, E., Van Mellaert, L., Engelborghs, Y., Mellado, R. P. & Anne´, J. (2002). Physical requirements for in vitro
ACKNOWLEDGEMENTS This work is supported by grants from the European Commission (QLK-2000-00122). S.D.K., K.V. and K.B. are research fellows of IWT (Institute for the Promotion of Innovation through Science and Technology in Flanders). Antibodies against CelA/CelB and XlnB were kindly provided by Dr R. Morosoli. XlnC-specific antibodies were a gift from Dr J. Dusart, while MelC2 antibodies were obtained from Dr E. Vijgenboom.
Functional analysis of the Streptomyces lividans type I signal peptidases. Arch Microbiol 176, 377–380.
processing of the Streptomyces lividans signal peptidases. J Biotechnol 96, 79–91. Geukens, N., Frederix, F., Reekmans, G., Lammertyn, E., Van Mellaert, L., Dehaen, W., Maes, G. & Anne´, J. (2004). In vitro analysis
of type I signal peptidase affinity and specificity for preprotein substrates. Biochem Biophys Res Commun 314, 459–467. Jo¨nsson, U., Fa¨gerstam, L., Ivarsson, B. & 13 other authors (1991).
Real-time biospecific interaction analysis using surface plasmon resonance and a sensor chip technology. BioTechniques 11, 620–662. Kendall, K. & Cullum, J. (1984). Cloning and expression of an extra-
cellular agarase from Streptomyces coelicolor A3(2) in Streptomyces lividans 66. Gene 29, 315–321. Kieser, T., Bibb, M. J., Buttner, M. J., Chater, K. F. & Hopwood, D. A. (2000). Practical Streptomyces Genetics. Norwich, UK: John Innes
Centre. Korn, F., Weinga¨rtner, B. & Kutzner, H. J. (1978). A study of twenty actinophages: morphology, serological relationship and host range. In Genetics of the Actinomycetales. Edited by E. Freechsen, I. Tarnak & J. H. Thumin. Stuttgart: Gustave Fischer. Laemmli, U. K. (1970). Cleavage of structural proteins during the
assembly of the head of bacteriophage T4. Nature 227, 223–229. Leu, W. M., Chen, L. Y., Liaw, L. L. & Lee, Y. H. (1992). Secretion of
REFERENCES Berks, B. C., Palmer, T. & Sargent, F. (2003). The Tat protein
translocation pathway and its role in microbial physiology. Adv Microb Physiol 47, 187–254. Bernan, V., Filpula, D., Herber, W., Bibb, M. & Katz, E. (1985). The
nucleotide sequence of the tyrosinase gene from Streptomyces antibioticus and characterization of the gene product. Gene 37, 101–110. Bron, S., Bolhuis, A., Tjalsma, H., Holsappel, S., Venema, G. & van Dijl, J. M. (1998). Protein secretion and possible roles for multiple
signal peptidases for precursor processing in bacilli. J Biotechnol 64, 1–13. Chen, L. Y., Leu, W. M., Wan, K. T. & Lee, Y. H. (1992). Copper
the Streptomyces tyrosinase is mediated through its transactivator protein, MelC1. J Biol Chem 267, 20108–20113. Paetzel, M., Dalbey, R. E. & Strynadka, N. C. J. (2000). The structure
and mechanism of bacterial type I signal peptidases – a novel antibiotic target. Pharm Ther 87, 27–49. Palacin, A., Parro, V., Geukens, N., Anne´, J. & Mellado, R. P. (2002).
SipY is the Streptomyces lividans type I signal peptidase exerting a major effect on protein secretion. J Bacteriol 184, 4875–4880. Parro, V., Vives, C., Godia, F. & Mellado, R. P. (1997). Overproduction
and purification of an agarase of bacterial origin. J Biotechnol 58, 59–66. Parro, V., Schacht, S., Anne´, J. & Mellado, R. P. (1999). Four genes
encoding different type I signal peptidases are organized in a cluster in Streptomyces lividans TK21. Microbiology 145, 2255–2263.
transfer and activation of the Streptomyces apotyrosinase are mediated through a complex formation between apotyrosinase and its trans-activator MelC1. J Biol Chem 267, 20100–20107.
Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989). Molecular Cloning:
Chu, H. H., Hoang, V., Kreutzmann, P., Hofemeister, B., Melzer, M. & Hofemeister, J. (2002). Identification and properties of type
Schacht, S., Van Mellaert, L., Lammertyn, E., Tjalsma, H., van Dijl, J. M., Bron, S. & Anne´, J. (1998). The Sip(Sli) gene of Streptomyces
I-signal peptidases of Bacillus amyloliquefaciens. Eur J Biochem 269, 458–469.
lividans TK24 specifies an unusual signal peptidase with a putative C-terminal transmembrane anchor. DNA Seq 9, 79–88.
Cregg, K. M., Wilding, E. I. & Black, M. T. (1996). Molecular cloning
Schaerlaekens, K., Schierova, M., Lammertyn, E., Geukens, N., Anne´, J. & Van Mellaert, L. (2001). Twin-arginine translocation
and expression of the spsB gene encoding an essential type I signal peptidase from Staphylococcus aureus. J Bacteriol 178, 5712–5718. Dalbey, R. E. & Wickner, W. (1985). Leader peptidase catalyzes the
release of exported proteins from the outer surface of the Escherichia coli plasma membrane. J Biol Chem 230, 15925–15931. Dev, I. K., Ray, P. H. & Novak, P. (1990). Minimum substrate
sequence for signal peptidase I of Escherichia coli. J Biol Chem 265, 20069–20072. Fass, S. H. & Engels, J. W. (1996). Influence of specific signal peptide mutations on the expression and secretion of the a-amylase inhibitor
tendamistat in Streptomyces lividans. J Biol Chem 272, 13152–13158. Fekkes, P. & Driessen, A. J. (1999). Protein targeting to the bacterial
cytoplasmic membrane. Microbiol Mol Biol Rev 63, 161–173. http://mic.sgmjournals.org
a Laboratory Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory.
pathway in Streptomyces lividans. J Bacteriol 183, 6727–6732. Schaerlaekens, K., Van Mellaert, L., Lammertyn, E., Geukens, N. & Anne´, J. (2004). The importance of the Tat-dependent protein secretion
pathway in Streptomyces as revealed by phenotypic changes in tat deletion mutants and genome analysis. Microbiology 150, 21–31. Shareck, F., Roy, C., Yaguchi, M., Morosoli, R. & Kluepfel, D. (1991).
Sequences of three genes specifying xylanases in Streptomyces lividans. Gene 107, 75–82. Stenberg, E., Persson, B., Roos, H. & Urbaniczky, C. (1991).
Quantitative determination of surface concentration of protein with surface plasmon resonance using radiolabeled proteins. J Col Interf Sc 143, 513–551. 1449
N. Geukens and others
Strickler, J. E., Berka, T. R., Gorniak, J., Fornwald, J., Keys, R., Rowland, J. J., Rosenberg, M. & Taylor, D. P. (1992). Two novel
van Dijl, J. M., de Jong, A., Vehmaanpera¨, J., Venema, G. & Bron, S. (1992). Signal peptidase I of Bacillus subtilis: patterns of conserved
Streptomyces protein protease inhibitors: purification, activity, cloning, and expression. J Biol Chem 267, 3236–3241.
amino acids in prokaryotic and eukaryotic type I signal peptidases. EMBO J 11, 2819–2828.
Studier, F. W. & Moffat, B. A. (1986). Use of bacteriophage T7 RNA
Van Mellaert, L., Dillen, C., Proost, P. & 7 other authors (1994).
polymerase to direct selective high-level expression of cloned genes. J Mol Biol 189, 113–130.
Efficient secretion of biologically active mouse tumor necrosis factor alpha by Streptomyces lividans. Gene 150, 153–158.
The´berge, M., Lacaze, P., Shareck, F., Morosoli, R. & Kluepfel, D. (1992). Purification and characterization of an endoglucanase from
van Roosmalen, M. L., Geukens, N., Jongbloed, J. D. H., Tjalsma, H., Dubois, J.-Y. F., Bron, S., van Dijl, J. M. & Anne´, J. (2004). Type I
Streptomyces lividans 66 and DNA sequence of the gene. Appl Environ Microbiol 58, 815–820.
signal peptidases of Gram-positive bacteria. Biochem Biophys Acta 1694, 279–297.
Tjalsma, H., Noback, M. A., Bron, S., Venema, G., Yamane, K. & van Dijl, J. M. (1997). Bacillus subtilis contains four closely related type I
Ve´rtesy, L., Oeding, V., Bender, R., Zepf, K. & Nesemann, G. (1984). Tendamistat (HOE 467), a tight-binding a-amylase inhibitor from
signal peptidases with overlapping substrate specificities: constitutive and temporally controlled expression of different sip genes. J Biol Chem 272, 25983–25992.
Streptomyces tendae 4158. Eur J Biochem 141, 505–512.
Tjalsma, H., Bolhuis, A., van Roosmalen, M. L. & 7 other authors (1998). Functional analysis of the secretory precursor processing
machinery of Bacillus subtilis: identification of a eubacterial homolog of archaeal and eukaryotic signal peptidases. Gene Dev 12, 2318– 2331.
1450
Wittmann, S., Shareck, F., Kluepfel, D. & Morosoli, R. (1994).
Purification and characterization of the CelB endoglucanase from Streptomyces lividans 66 and DNA sequence of the encoding gene. Appl Environ Microbiol 60, 1701–1703. Zhang, Y. B., Greenberg, B. & Lacks, S. A. (1997). Analysis of a
Streptococcus pneumoniae gene encoding signal peptidase I and overproduction of the enzyme. Gene 194, 249–255.
Microbiology 152