Functional identification of a Streptomyces lividans FKBP-like protein involved in the folding of overproduced secreted proteins.

Some bacterial peptidyl-prolyl cis/trans isomerases (PPIases) are involved in secretory protein folding after the translocation step. Streptomyces lividans has been used as a host for engineering extracellular overproduction of homologous and heterologous proteins in industrial applications. Although the mechanisms governing the major secretory pathway (Sec route) and the minor secretory pathway (Tat route) are reasonably well described, the function of proteins responsible for the extracellular secretory protein folding is not characterized as yet. We have characterized a Tat-dependent S. lividans FK506-binding protein-like lipoprotein (FKBP) that has PPIase activity. A mutant in the sli-fkbp gene induces a secretion stress response and affects secretion and activity of the Sec-dependent protein α-amylase. Additionally, propagation in high copy number of the sli-fkbp gene has a positive effect on the activity of both the overproduced α-amylase and the overproduced Tat-dependent agarase, both containing proline cis isomers. Targeted proteomic analyses showed that a relevant group of secreted proteins in S. lividans TK21 are affected by Sli-FKBP, revealing a wide substrate range. The results obtained indicate that, regardless of the secretory route used by proteins in S. lividans, adjusting the expression of sli-fkbp may facilitate folding of dependent proteins when engineering Streptomyces strains for the overproduction of homologous or heterologous secretory proteins.

Background

Streptomycetes are Gram-positive soil bacteria that secrete an array of hydrolytic enzymes [1,2], among other metabolites, to ensure their survival in this harsh environment. This natural capacity to produce hydrolytic enzymes in industrial applications has attracted great interest in using streptomycetes as hosts for the production of such extracellular proteins.

Streptomyces lividans is a bacterium with a relaxed restriction modification system, which facilitates its transformation by heterologous DNA. The genome sequence of S. lividans is known [3,4], and the bacterium has been largely used as a host for the production of engineered secretory proteins of high commercial value [5,6].

Extracellular protein secretion across the S. lividans cytoplasmic membrane mainly uses the major secretion pathway (Sec) and the minor twin arginine translocation (Tat) secretion pathway [7]. The Sec pathway targets newly synthesized pre-proteins to the membrane using the signal recognition particle (SRP), which interacts with the pre-protein signal peptides, and its membrane receptor FtsY [8]. The ribosome–protein complex may also interact with the translocase complex SecYEG for protein secretion. Sec-secreted proteins are released into the medium in an unfolded conformation. In contrast, proteins secreted by the Tat pathway appear to be exported outside the cell fully folded [9]. Thus, the Tat route is an appealing secretion system in streptomycetes where up to 27 proteins have been confirmed to be secreted by the Streptomyces coelicolor Tat system [10]. In S. lividans, the Tat system has three components (TatA, TatB and TatC), and the Tat signal peptide, containing the highly conserved twin-arginine motif SRRXFLK, at its amino end, is thought to be recognized by the TatA–TatB heterocomplex [11]. This complex theoretically interacts with TatC in the membrane, and, then, oligomerization of TatA forms the membrane pore to support secretion of the Tat proteins [11]. In Escherichia coli the signal peptide recognition is mediated by the TatB–TatC complex [12].

The main enzymes involved in extracellular folding of secretory proteins are the thiol-disulfide oxidoreductases and the peptidyl-prolyl cis/trans isomerases (PPIases). Bacterial PPIases have been linked to protein folding and secretion because of their capacity to catalyse the cis/trans isomerization of peptide bonds preceding prolyl residues [13]. The PPIases typically belong to three functional groups, cyclophilins, the FK506-binding proteins (FKBPs) and parvulins, and are ubiquitously distributed among bacteria [13].

Bacillus subtilis PrsA is a lipoprotein that belongs to the parvulin family and is considered an important factor for protein secretion. The cellular level of PrsA and the rate of secretion have been shown to be interrelated in B. subtilis. Depletion of PrsA leads to a reduction of the levels of secreted heterologous α-amylase (AmyQ) [14] as well as of the amount of several endogenous secretory proteins [15]. PrsA overproduction strongly increases the production of AmyQ or the SubC protease [16].

No equivalent PPIases have been experimentally characterized to exert this role in S. lividans yet. In the present work, we characterize a Tat-dependent S. lividans FKBP-like lipoprotein (Sli-FKBP) that positively affects the production of a significant number of Sec- and Tat-secreted proteins containing putative proline cis isomers in S. lividans TK21.

Methods

Bacterial strains, plasmids and media

The S. lividans TK21 wild-type strain [17] and its derivatives were cultured in liquid NMMP medium in the presence of mannitol as a carbon source [18]. Apramycin (50 µg ml−1), thiostrepton (50 µg ml−1), kanamycin (50 µg ml−1) and chloramphenicol (25 µg ml−1) were added to the R5 and MS solid media, when required.

Construction of gene disruption mutant

To construct the S. lividans Δsli-fkbp mutant strain oligonucleotides FKBPdisFw (5′-GGGCTGCAGAACAGCTACGACCGCAAGAC-3′) and FKBPdisRv (5′-GCCTCTAGACACACCCTTGTACTGCACGA-3′) were used to amplify a 420 nt long DNA fragment that was inserted into the non-replicating suicide plasmid pOJ260 of Streptomyces [19] through its unique XbaI and PstI sites to generate plasmid pOJFKBP. The plasmid was used to conjugate E. coli to Streptomyces as described [20], to inactivate the chromosomal copy of sli-fkbp by insertion of pOJ260 by single homologous recombination between pOJFKBP and the sli-fkbp copy in the chromosome. Escherichia coli ET12567 carrying the non-transmissible ‘driver’ plasmid pUZ8002 was used for conjugation [21]. Apramycin-resistant strains containing the disrupted gene sli-fkbp were selected upon verification of the disruption by polymerase chain reaction (PCR) amplification (not shown). Plasmids pAMI11 [22] and pAGAs5 [23] carrying the S. lividans gene amlB or the S. coelicolor agarase gene dagA, respectively, were used to transform the S. lividans TK21 and S. lividans Δsli-fkbp.

Overexpression of FKBP

The oligonucleotides FKBP-HindIII (5′-TTTTAAGCTTTCTCACGCCGTAGAGTTGC-3′) and FKBP-XbaI (5′-TTTTTCTAGACCCTCCCAGATGTCCTTGAT-3′) containing the HindIII and XbaI sites and FKBP-P (5′-TTTGGATCCTTCTCACGCCGTAGAGTTGC-3′) and FKBP-T (5′-TTTTTGGATCCCCCTCCCAGATGTCCTTGAT-3′) containing the BamHI sites were used to amplify the gene sli-fkbp and its possible promoter region. Chromosomal DNA of S. lividans TK21 strain was used as a template. The obtained 1222 bp long DNA fragments were subsequently cloned into the pGEM-T Easy vector (Promega) and sequenced. The plasmid pGEM, harbouring sli-fkbp, was digested with NotI to retrieve the 1244 bp long DNA fragment containing sli-fkbp; this was inserted into the multi-copy cosmid pFDT [24] through its unique NotI site to generate pFDTFKBP. Additionally, the pGEM harbouring sli-fkbp was digested with BamHI, and the 1222 bp long DNA fragment was cloned into the multicopy plasmid pIJ487 previously digested with BamHI to generate plasmid pIJFKBP.

Protoplasts from S. lividans TK21, S. lividans TK21(pAMI11) and S. lividans TK21(pAGAs5) were transformed with cosmid pFDTFKBP to obtain the S. lividans (pFDTFKBP), S. lividans TK21(pAMI11) (pFDTFKBP) and S. lividans TK21(pAGAs5) (pFDTFKBP) strains. Cosmid pFDT was propagated in S. lividans TK21, S. lividans TK21(pAMI11) or S. lividans TK21(pAGAs5) to generate the corresponding isogenic strains.

Protoplasts from S. lividans TK21 were transformed with plasmid pIJFKBP to obtain S. lividans (pIJFKBP). Plasmid pIJ487 was propagated in S. lividans TK21 to generate the corresponding isogenic strain.

Quantitative real-time polymerase chain reaction

Total RNA was isolated from bacteria growing cultures at different phases of growth (24 h, 36 h, 48 h) using the RNeasy midi Kit (Qiagen). Cell lysates were extracted twice with phenol–chloroform before being loaded onto RNeasy midi-columns for RNA purification. DNA, potentially contaminating the RNA preparations, was removed by incubation with RNase-free DNAse (Ambion) and its absence was tested by quantitative real-time PCR (qRT-PCR) amplification in the absence of reverse transcriptase. Complementary DNA was synthesized using the High Capacity Archive kit (Applied Biosystems). qRT-PCR was performed using SYBR Green technology in an ABI Prism 7300 Sequence Detection System (Applied Biosystems). Samples were initially denatured by heating at 95°C for 10 min. A 40-cycle amplification and quantification programme was then followed (95°C for 15 s and 60°C for 1 min) by a single fluorescence measurement per cycle, according to the manufacturer's recommendations. Subsequently, a final extension cycle (72°C, 1 min) was performed. Three biological samples from the different bacterial cultures were amplified in triplicate in separate PCR reactions. All PCR products were between 50 and 150 bp in length.

A melting curve analysis was conducted after amplification to distinguish the targeted PCR products from the non-targeted ones. The melting curves were obtained by heating at temperatures ranging from 60°C to 95°C at a rate of 0.2°C per second, with continuous fluorescence scanning. The hrdB transcript was carried out as an internal control to quantify the relative expression of the target genes. The hrdB transcript was used as a reference to normalize the relative expression of Streptomyces genes. The oligonucleotides used as primers to amplify the transcripts of two-component system cssRS and tatC genes were described previously [25,26]. Oligonucleotides SLI1639 FW (5′-GGACGAAGTTCGACGAGAAG-3′) and SLI1639 RV (5′-AGACCTGGCCGAGGTAGTT-3′) were used to amplify the sli-fkbp transcript.

Protein analysis and western blot experiments

Supernatants from the sli-fkbp mutant and the Sli-FKBP overproducer strains containing the multi-copy plasmids pAMI11 and pAGAs5 grown in NMMP medium [27] were processed as described [25]. For western blot analysis, extracellular proteins were fractionated by sodium dodecylsulfate–polyacrylamide gel electrophoresis (SDS-PAGE) in 10% and 12% (w/v) acrylamide gel [28]. An equivalent amount of protein loaded onto the SDS-PAGE acrylamide gel was corrected by the bacterial dry weight in each case.

Gel-fractionated proteins were transferred onto Immobilon polyvinylidene difluoride membranes (Millipore), as described [29]. The transferred material was incubated with polyclonal antibodies raised against S. lividans TK21 AmlB (a gift from C. Isiegas) and S. coelicolor agarase (DagA; [30]) followed by incubation with horseradish peroxidase-conjugated protein A (Invitrogen Laboratories) as described before [25].

Enzyme activity

To determine extracellular α-amylase activity and agarase activity, the supernatants from the aliquots of bacterial cell cultures were collected at the indicated growth phases and concentrated by precipitation with ammonium sulfate brought to 80% saturation; the precipitated protein was collected by centrifugation at 13 000g for 30 min and dissolved in 20 mM phosphate buffer (pH 7).

The activities of α-amylase and agarase were estimated by determining the amount of reducing sugars released from starch and agarose, respectively. α-amylase and agarase activities were determined as previously described [25,30]. One unit of α-amylase was defined as the amount of an enzyme necessary to produce reducing sugar equivalent to 1 µmol of glucose in 30 min under the assay conditions. The specific activity, measured as units per mg of protein, was the average of at least three independent determinations.

One unit of agarase activity is the amount of enzyme that increased absorbance at 450 nm by 0.001 per minute of incubation under the assay conditions. The specific activity was expressed as units per mg of dry weight and was the average of at least three independent determinations.

The FKBP activity was determined using protoplasts. Protoplasts were prepared as previously described [18]. Briefly, cells from 10 ml of NMMP culture were washed and re-suspended in P buffer according to the bacterial dry weight with 1 mg ml−1 lysozyme. After 20 min of incubation at 37°C protoplasts were washed with P buffer and then were centrifuged at 1500g for 5 min at 4°C. The protoplast's pellet was then re-suspended in 300 µl P buffer, and 30 µl aliquots were used to perform the assay.

The isomerase assay was determined by protease coupling assay with α-chymotrypsin (α-Ct) (Sigma Aldrich ref. C4129) using as a substrate N-succinyl-L-Ala-Ala-Pro-Phe-p-nitroanilide (sAAPF-pNA) (Sigma Aldrich ref. S7388). Chymotrypsin cleaves the synthetic peptide only when the Ala-Pro bond, which is in equilibrium between cis and trans, is in trans configuration [31]. The reaction was monitored for 6 min by the increase in absorbance at 360 nm (corresponding to the release of p-nitroanilide) with a spectrophotometer (Ultrospec 3100 pro; GE Healthcare, Amersham). FK-506 (Sigma Aldrich ref. F4679) was added in the inhibition of the PPIase activity assays [32]. PPIase catalyses the cis/trans isomerization of X-Pro peptide bonds. Therefore, the isomerization reaction of this peptide bond was measured by monitoring the release of p-nitroaniline.

Aliquots (30 µl) of the pIJ487 and pIJFKBP protoplasts were mixed with 625 µl of 110 mM Tris HCl pH 8 and water to a total volume of 710 µl and incubated for 15 min on ice. Then, 120 µl of α-Ct 0.56 mM dissolved in 1 mM HCl and 30 µl of sAAPF-pNA 1.4 mg ml−1 dissolved in dimethyl sulfoxide were added and incubated for 3 min on ice before the absorbance was measured at 390 nm for 1–6 min. For the inhibition assay 2 µl of FK506 14.2 µM dissolved in 50% (v/v) aqueous ethanol was added to the 30 µl aliquots and 625 µl of 110 mM Tris HCl pH 8 and then incubated for 15 min on ice. After that, chymotrypsin and sAAPF-pNA were added [32]. The residual activity was determined relative to a control sample treated identically in the absence of the inhibitor. All data are averaged from three independent measurements.

Protein predictions

For the prediction of the signal peptide SignalP 4.1, LipoP 1.0 and TatP 1.0 servers [33-35] were used. The presence of disulfide bonds in the proteins was predicted using the DIANNA 1.1 server [36]. The SPOT-Omega server was used to predict which proteins could contain potential cis-trans prolyl isomers [37].

Molecular modelling

Full three-dimensional protein models were generated using I-TASSER [38], Modeller [39], Raptor-X [40] and Sparks [41] by homology with known functional structures. After inspection, the best models were selected as potential representatives of the functional structure. Isomerized peptide bonds were identified using the VMD Cispeptide plugin [42,43]. Additional models with the peptide bonds reverted to the trans conformation were generated using UCSF Chimera [44] and subjected to additional refinement through a cycle comprising energy minimization, equilibration and 30 ns molecular dynamics (MD) in 150 mM saline solution using GROMACS [45] and the Amber [46] force field. All the selected models were then inspected to identify H-bonds and compute the charge distribution on their surface using UCSF Chimera and APBS [47].

Protein identification by nano-LC-MS/MS triple TOF analysis

For protein identification, the S. lividans (pAGAs5) (pFDTFKBP) extracellular protein band reacting positively with anti-agarase antibodies was sliced out of the SDS-PAGE and digested with trypsin using Proteineer DP robot (Bruker, Bremen, Germany). Digestion was performed according to a previously described protocol [48]. In summary, gel plugs were washed with 50 mM ammonium bicarbonate and samples reduced with 10 mM dithiothreitol. Alkylation was carried out with 55 mM iodoacetamide at room temperature before adding recombinant sequencing-grade trypsin (0.1 µg; Promega). Digestion took place at 37°C for 18 h. Following digestion, peptides were extracted, dried by speed-vac centrifugation and stored at −20°C until needed.

The peptide samples were analysed on a nano-liquid chromatography (LC) system (Eksigent Technologies nanoLC Ultra 1D plus; AB SCIEX, Foster City, CA) coupled to a 5600 TripleTOF mass spectrometer (AB SCIEX, Foster City, CA) with a nanoelectrospray ion source. Samples were loaded on a C18 PepMap trap column (5 µm particle size, 100 µm I.D. × 2 cm; Thermo Scientific) at 2 µl min−1, in 0.1% formic acid in water. The trap column was switched online to a C18 nanoAcquity BEH analytical column (1.7 µm, 100 Å, 75 µm I.D. ×15 cm; Waters). Chromatographic elution was achieved using a 40 min linear gradient ranging from 5% to 40% solvent B (0.1% formic acid in acetonitrile) at 250 nl min−1. The mass spectrometer operated in data-dependent acquisition mode. For TOF scans, the accumulation time was set to 250 ms, and 15 precursor ions were monitored per cycle.

Mass spectrometry (MS) and tandem mass spectrometry (MS/MS) spectra were processed using Analyst TF 1.5.1 software (AB SCIEX, Foster City, CA). Raw data were converted to MGFs (mascot general files) and searched against a database built from sequences obtained for S. coelicolor at Uniprot Knowledgebase (as of November 2016), using a licensed Mascot Server v. 2.4 search engine (Matrix Science, London, UK). Search parameters were set as follows: carbamidomethylation of cysteine as fixed modification and methionine oxidation as variable 1. Peptide mass tolerance was set to 25 ppm and 0.02 Da, in MS and MS/MS mode, respectively and one missed cleavage was allowed.

Targeted mass spectrometry (SRM/MRM)

Relative protein abundance of a selected (n = 36) group of proteins was determined by targeted proteomics, using selected reaction monitoring (SRM/MRM). For that purpose, 1 mg of trypsin-digested proteins per sample was analysed using an Eksigent 1D Plus nano-LC system coupled to a SCIEX 55000 QTRAP quadrupole triple mass spectrometer. Using a 45 min length gradient, 273 transitions that corresponded to 83 specific peptides of the 36 proteins of interest were monitored. Protein-specific peptides and transitions were selected, whenever possible, on the basis of previous experimental data. As the number of transitions was too high to analyse using a single SRM/MRM method, two sub-methods were designed. Nine samples (three conditions, three biological replicates per condition) were monitored using both sub-methods and the raw data files in WIFF format were analysed using the Skyline 4.2 software [49]. The analysis determined the areas corresponding to each transition and peptide. For those samples in which peptides could not be detected, quantification results were not considered. For those cases in which more than one peptide per protein was monitored, the areas corresponding to each protein-specific peptide were added to obtain a total summed area for each protein. Summed protein areas were used to calculate relative protein abundances between samples. The relative protein abundances were log2-transformed to approach normality and comparisons between different strains were performed using Student's t-test and Fisher's exact test.

Results

Identification of the Streptomyces lividans TK21 fkbp gene

Analysis of the S. lividans TK21 secretome identified a putative polypeptide similar to that of the S. coelicolor SCO1639 protein [27]. This gene encoded a putative Tat-dependent PPIase lipoprotein [50] that is orthologous to the previously characterized Streptomyces chrysomallus protein (FKBP-33) having a PPIase activity [51]. The S. lividans TK24 gene SLIV_29545 has been shown to be highly homologous to the SCO1639.

To test if the equivalent S. lividans TK21 gene homologous to the SLIV_29545 gene could also encode a PPIase with a potential role in the correct folding of secretory proteins, oligonucleotides FKBP-P and FKBP-T were used to amplify a 1222 bp long DNA fragment from the S. lividans genome containing the DNA sequence encoding the putative FKBP-like protein and the potential regulatory region of the gene.

The 1222 bp long amplified DNA fragment was sequenced and its sequence was aligned with the equivalent ones from the S. coelicolor and the S. lividans TK24 genomes and found to coincide in 99% and 100%, respectively, with that of the genes encoding their respective FKBP-like proteins. The FKBP-like protein encoded in S. lividans TK21 was predicted to be a substrate for a type II signal peptidase [34]. Moreover, the level of expression of the gene encoding the S. lividans TK21 FKBP-like protein was found to be lower in an S. lividans TK21 strain defective in the type II signal peptidase Lsp [52], when measured by qRT-PCR transcriptional analyses.

The putative S. lividans FKBP has PPIase activity

The S. lividans FKBP-like protein showed 36% identical residues and 50% equivalent amino acids to the E. coli periplasmic FKBP (FkpA), which acts as a periplasmic folding modulator with a broad substrate range [53]. Activity of FKBP-like PPIases is inhibited by the immunosuppressive compound FK506 [32].

Since S. lividans FKBP-like protein is a lipoprotein, to perform the PPIase activity assay we used protoplast extracts of the S. lividans pIJ486 and S. lividans pIJFKBP strains, containing the multi-copy plasmid pIJ486 and the multi-copy plasmid pIJ486 harbouring the fkbp-like gene, respectively (figure 1). The background values of the absorbance of the p-nitroaniline released from the substrate upon incubation in the absence of protoplasts were subtracted from those obtained in the presence of pIJ486 protoplasts and pIJFKBP protoplasts. A higher increase in the absorbance was observed for protoplasts containing pIJFKBP owing to the presence of an excess of FKBP copies versus the protoplast containing pIJ487 at 120 and 240 s (figure 1a).

Figure 1.

Assay of isomerase activity. Streptomyces lividans pIJ487 and S. lividans pIJFKBP protoplasts were tested for the PPIase activity and FK506 inhibition assays. (a) The isomerization activity was measured by monitoring the release of p-nitroaniline at 390 nm as indicated in the Methods section. (b) The inhibition assays were performed as indicated in the Methods section. The residual PPIase activity was calculated relative to the activity value of the sample treated identically in the absence of the inhibitor and expressed as a percentage.

The presence of the inhibitor FK506 decreases the absorbance to 0 at 120 s and recovers the absorbance observed with respect to the control sample (without inhibitor) at 240 s in the pIJ487 protoplasts, which represents a percentage of inhibition of 100% at 120 s. The values of residual activity after treatment with the inhibitor FK506 in the pIJFKBP protoplasts was 0%, 30% and 76.23%, which represent a percentage of inhibition of 100%, 70% and 23.77% at 120, 240 and 360 s, respectively (figure 1b). From now on we will refer to the S. lividans TK21 gene encoding the FKBP-like protein as sli-fkbp.

A mutant in sli-fkbp induces the secretion stress response

The qRT-PCR transcriptional analyses, using as an endogenous control the hrdB gene, showed that the expression of the sli-fkbp and tatC in the S. lividans TK21 strain runs parallel to cell culture growth, although the level of the sli-fkbp expression is lower than that of the tatC gene (about 10% that of tatC) [26]. In the absence of sli-fkbp expression, the overall amount of misfolded secreted proteins would likely increase and, consequently, the level of expression of the two-component CssRS system [24,25] could be induced. Therefore, an sli-fkbp disrupted mutant (sli-fkbp::pOJ260) was constructed, and the relative level of expression of the CssRS two-component system genes was determined by qRT-PCR analyses and compared with that of the wild-type cell. A significant increase in the expression levels of the two genes (relative expression levels of the sensor cssS: 25.02 ± 1.35 and of the regulator cssR: 8.35 ± 1.11) was observed at the late exponential phase of growth (24 h), suggesting that the absence of the sli-fkbp gene may produce an accumulation of misfolded proteins at this time. This suggests that, despite the moderate level of expression of the single sli-fkbp copy present in the bacterial genome, the Sli-FKBP protein may play an active role in the correct folding of extracellular proteins.

The absence of sli-fkbp affects secretion and activity of Sec-dependent α-amylase

Prolyl isomerization to the cis conformation at residue 350 (P350) may be required to attain the mature form of the Sec-dependent protein α-amylase. Similarly, the Tat-dependent protein agarase may require prolyl isomerization at residues 127 and 183 to the cis conformation to produce mature active agarase (electronic supplementary material, figure S1).

To explore further the role of Sli-FKBP in the secretion of proteins in S. lividans, we propagated the multi-copy plasmids pAMI11 and pAGAs5 harbouring the S. lividans α-amylase gene (amlB) and the S. coelicolor agarase gene (dagA), respectively, in the sli-fkbp mutant and determined the secretion and extracellular activities of α-amylase and agarase.

Overexpression of amlB and dagA in the sli-fkbp mutant strain constrained their growth rate when compared with the wild-type (electronic supplementary material, figure S2a and S2b). The Sec-dependent protein α-amylase reaches its maximum level of secretion at 24 h of growth while agarase reaches its maximum level of secretion during the stationary phase [26] (figure 2a,c). The secretion pattern and extracellular activity of α-amylase were severely affected in the sli-fkbp mutant strain with respect to that of the isogenic strain (figure 2a,b), showing the presence of inactive α-amylase only at 60 h and probably corresponding to incorrectly folded protein as previously described [24,25]. In contrast, with the agarase, only moderate changes to the secretion pattern and extracellular activity were observed (figure 2c,d).

Figure 2.

α-amylase and agarase secretion and extracellular activities in the sli-fkbp mutant. α-amylase secretion pattern (a), α-amylase activity (b), agarase secretion pattern (c) and agarase activity (d) were measured at different times of growth in the sli-fkbp mutant containing the plasmids pAMI11 (harbouring the α-amylase-encoding gene, amlB) or pAGAs5 (harbouring the agarase-encoding gene, dagA) propagated in multi-copy and compared with those of their respective wild-type counterparts. α-amylase and agarase secretion were analysed by western blotting with antibodies raised against AmlB (59 kDa) and DagA (32 kDa). The amount of protein loaded onto the gels was corrected by the dried weight of the bacterial cultures. Activities are expressed as percentages, where 100% is the maximum level of the α-amylase and agarase activities measured in the corresponding wild-type strain. Percentage values are the mean of the three different measurements.

α-amylase and agarase activities increase when FKBP is overproduced

To study if overproduction of Sli-FKBP could increase the production of the Sec-dependent model protein α-amylase, the pFDTFKBP plasmid (a plasmid compatible with pIJ486) was propagated in the S. lividans TK21 (pAMI11) strain.

Overproduction of Sli-FKBP did not affect significantly the rate of growth in liquid medium (electronic supplementary material, figure S2c and S2d). α-amylase secretion increased, with an increase in extracellular activity of around threefold with respect to the isogenic strain overproducing α-amylase (figure 3a,b), suggesting that secreted α-amylase may benefit from the action of Sli-FKBP for correct folding. Moreover, the level of expression of the CssRS two-component system genes decreased in the strain that overproduces α-amylase and sli-fkbp (relative expression levels of the sensor: cssS −2.42 ± 0.113 and of the regulator cssR: −1.79 ± 0.89) versus the strain lacking multi-copy sli-fkbp, where the two-component system was activated [25]. This suggests that overproduction of some extracellular Sec-dependent proteins may be aided by the action of PPIases for the secretory proteins to acquire their correctly folded structure.

Figure 3.

α-amylase and agarase secretion and extracellular activities in cells overproducing Sli-FKBP. The α-amylase secretion pattern (a), α-amylase activity (b), agarase secretion pattern (c) and agarase activity (d) were measured at different times of growth by propagation of the respective multi-copy plasmids pAMI11 or pAGAs5 concomitantly with the compatible multi-copy plasmids carrying the sli-fkbp gene. α-amylase and agarase secretion were analysed by western blotting with antibodies raised against AmlB (59 kDa) and DagA (32 kDa). The amount of protein loaded onto the gels was corrected by the dried weight of the bacterial cultures. Molecular size markers are indicated on the side of (c). Activities are expressed as percentages, where 100% is the maximal level of the α-amylase or agarase activities measured in the corresponding wild-type strain harbouring the compatible plasmid pFDT. Percentage values are the mean of three different measurements.

To test if higher levels of FKBP would be needed to improve the extracellular activity of secreted agarase, the multi-copy plasmid pFDTFKBP was propagated in the S. lividans (pAGAs5) strain. Extracellular agarase activity registered a threefold increase in the strain overexpressing agarase and sli-fkbp (figure 3d), suggesting that overproduction of Tat-dependent agarase needs additional activity of Sli-FKBP to fold correctly. No significant increase was detected in the amount of secreted agarase by western blot assays (figure 3c). Incidentally, the western blot analyses showed a difference between strains in the gel mobility of the protein produced (figure 3c). The faster moving band, corresponding to agarase and Sli-FKBP overproduction, was sliced out of the gel and the protein analysed by nano LC-MS/MS triple TOF MS. The proteomic analysis allowed recovery and identification of tryptic peptides of mature agarase, obtaining 65% coverage of the overall protein (electronic supplementary material, figure S3), which would correspond to 20 kDa according to ProtParam [54]. This size does not agree with the size of the protein deduced from its gel mobility. Gel mobility shifts in SDS-PAGE have been associated with changes in protein structure, as previously described [55,56], and might provide an explanation for the discordance between the sizes derived by proteomic analysis and SDS-PAGE.

Agarase modelling

The best agarase models were selected by inspecting the homology models obtained with various methods according to coherence with published structures and available biological information. MD simulations of the chosen models revealed the existence of one preferred conformation that was used as a representative for further analysis. The predicted structure contains two cis peptide bonds, at P127 and P183, equivalent to those found on other homologues. Switching of P127 to the trans conformation would disrupt the allosteric agarose-binding site, preserving the active site, while switching of P183 would disrupt both (electronic supplementary material, figure S4). The trans forms show a conformation with lower potential energy (more stability) than the cis structures. However, the mature double-cis form contains numerous H-bonds that help stabilize it. This suggests that the spontaneous change from trans to cis would be energetically unfavourable, justifying the need for a PPIase to facilitate it, while the H-bond network in the cis form would provide a significant initial barrier to spontaneous reversion to the trans conformation afterwards.

In addition, the trans forms would have larger molecular dimensions (as evidenced by a larger radius of gyration, equivalent to the Stokes radius), and the distribution of surface charges would also be different: conversion to cis would be associated with a larger reduction (greater than 400 Å2) of surface positive charge than of the surface negative charge. Both of these effects (changes in dimension and surface charge) could contribute to the differential mobility of the cis and trans forms.

Effect of Sli-FKBP on secretory proteins

To gain further insight into the effect of the overproduction of Sli-FKBP on secreted proteins we performed an SRM/MRM proteomic experiment. SRM/MRM is typically used to perform reliable, sensitive and selective protein assays for the comparison of relative protein amounts between samples [57]. The analyses were performed using supernatants from early stationary phase cultures of the Δsli-fkbp, the single-copy sli-fkbp and the sli-fkbp overexpressing strains using three biological replicates.

Cognate secreted proteins previously identified in the supernatant of S. lividans TK21 were analysed with the SPOT-Omega server [37] to identify those with a high probability of containing proline cis–trans isomers. A group of 31 secreted proteins that might be potential substrates of Sli-FKBP (electronic supplementary material, table S1) were selected for further analysis using SRM/MRM. Sli-FKBP was also included in the analysis to compare its expression in the three strains.

None of the peptides corresponding to four of the secreted proteins analysed could be detected in any of the strains (electronic supplementary material, table S1) and these proteins were not considered in the analysis, leaving 27 proteins. Twelve proteins had undetected peptides in the Δsli-fkbp strain but were detected in the other strains. The ratios and significance level (p-value < 0.05) of differences in relative protein abundances were calculated for all proteins detected. The results are summarized in table 1.

Table 1.

Secreted proteins significantly affected by Sli-FKBP. Protein function was annotated according to StoPSserver [7]. ∅︀ indicates that no peptide was detected in any of the replicates. Ratio: ratio of protein abundances. p: p-value. NC: not computable (division by zero). Analyses were performed using three biological replicates. Only statistically significant values (p < 0.05) are reported to avoid reporting potentially misleading results. Post-translocational modification proteins are shown in italics.

gene	annotated function	route	relative protein abundance			pFDT-FKBP versus pFDT		pFDT-FKBP versus Δfkbp
			pFDT-FKBP	pFDT	Δfkbp	ratio	p	ratio	p
SLI01600	glycoside hydrolase, family 43	Tat	29928	15411	∅︀			NC	0.0004
SLI03045	N-acetylmuramoyl-L-alanine amidase	Sec	57863	24139	∅︀	2.4	0.045	NC	0.001
SLI03130	alpha/beta hydrolase fold protein	Sec	116228	63987	∅︀			NC	7 × 10−5
SLI03390	amidase	Tat	59108	9263	713			82.9	0.023
SLI04830	metallopeptidase	Sec	202438	247559	8906	0.8	0.004	22.7	4 × 10−5
SLI04955	soluble quinoprotein glucose/sorbosone dehydrogenase	Sec	298908	145641	891	2.1	0.022	335.4	0.018
SLI06140	beta-galactosidase (lactase)	Tat	29931	19031	653			45.8	0.010
SLI07140	esterase	Sec	5483255	4583879	186258			29.4	0.006
SLI07150	uncharacterized protein	Tat	207033	132261	12380			16.7	0.018
SLI07590	peptidase S1A, chymotrypsin family	Sec	4485423	3569046	200789	1.3	0.047	22.3	0.009
SLI08115	alpha-1,2-mannosidase, putative	Sec	209808	72118	∅︀			NC	0.001
SLI09630	peptidase S1, PA clan	Tat	49908	52465	∅︀			NC	2 × 10−4
SLI10325	peptidase	Sec	56001	33131	7483	1.7	0.049	7.4	0.001
SLI12065	solute-binding protein	Sec	36343	48931	1692			21.4	0.009
SLI14975	secreted endopeptidase, NLPC/P60 domain	Tat	204846	92338	1514			135.3	0.010
SLI19350	protein of unknown function containing DUF839 domain	Tat	28591	14074	∅︀	2.0	0.020	NC	0.001
SLI19735	glycoside hydrolase/deacetylase	Sec	517430	458383	12514			41.3	0.002
SLI22430	esterase	Sec	10354280	8344363	39145			264.5	0.004
SLI23095	peptidase M6, InhA family protein	Tat	424058	345723	∅︀			NC	0.001
SLI25270	unknown function	Sec	33439	31383	∅︀	1.1	0.047	NC	1 × 10−5
SLI25665	aldose 1-epimerase	Tat	32022	12157	∅︀	2.6	0.027	NC	0.001
SLI26155	N-acetylmuramoyl-L-alanine amidase	Tat	481504	266240	24205			19.8	0.003
SLI27125	N-acetylmuramoyl-L-alanine amidase domain with Tachylectin 2	Sec	716759	634345	10642			67.3	0.010
SLI29775	glycosyl hydrolase-like 10 (UPFUPF0748-like protein)	Tat	182360	262400	∅︀			NC	4 × 10−5
SLI31645	tripeptidyl aminopeptidase	Sec	70306	47246	∅︀			NC	0.001
SLI34120	probable subtilase-type protease inhibitor	Sec	1457618	1424560	376745			3.8	0.030
SLI34825	glycosyl hydrolase, five-bladed beta-propeller domain	Tat	75741	97855	∅︀			NC	1 × 10−4
SLI29545	PPIase	Tat	636700	372348	5509	1.7	0.026	115.5	0.020
SLI18490	putative copper chaperone SCO1/SenC	Sec	2028918	892015	26378	2.275	0.005	76.9	1 × 10−4
SLI26855	HtrA1 protease	Sec	99782	88957	28550			3.4	0.005

As expected, Sli-FKBP (SLI29545) showed significant differences in the relative protein abundance among the three strains according to their corresponding genotype. Comparison of the extreme cases pFDTFKBP and Δsli-fkbp strains should highlight the effect of FKBP. All the proteins (both Sec and Tat secreted) were significantly affected in the Δsli-fkbp strain with respect to the overproducing strain; the relative expression ratio could not be calculated for proteins not detected in the Δsli-fkbp strain. On the other hand, comparison of pFDTFKBP with the isogenic pFDT strain should tag proteins that may be affected by overexpression of sli-fkbp. Besides the expected difference in Sli-FKBP, nine other secretory proteins displayed a significant difference in their relative abundance when Sli-FKBP was expressed in multi-copy (pFDTFKBP) versus single-copy (pFDT) strains.

The increments in relative protein abundances observed in the presence of Sli-FKBP suggest that it may have a role in contributing to improve the stability of a wide number of proteins.

The stabilizing effect of Sli-FKBP might be indirect, owing to isomerization of intermediate proteins with post-translocational activity over the final secretory proteins. For this reason, we also included in the MRM/SRM experiment two thiol-disulfide oxidoreductases recently described in S. lividans TK21, Sli-DsbA and Sli-DsbC [58], a protease that degrades misfolded proteins and may have a chaperone-like role, HtrA1 (SLI26855) [24], and the only extracellular chaperone detected to date in the secretome of S. lividans (SLI18490) [59]. No peptides were detected for the two thiol-disulfide oxidoreductases (electronic supplementary material, table S1). We detected a significant increase in the relative abundance of chaperone (SLI18490) between the Sli-FKBP overexpressing strain pFDTFKBP and the isogenic pFDT strain (table 1), but not for protease HtrA1 (SLI26855).

Discussion

PPIases have been found to be present in almost all sequenced genomes and to be involved in a variety of biological processes [53,60-65]. Recently, the S. lividans TK24 secretome [59] has been reported showing only one extracytoplasmic PPIase in S. lividans, SLIV_29545, which is homologous to SCO1639. The S. coelicolor FKBP-like SCO1639 has been identified as a Tat-dependent lipoprotein [50], but little is known about its function in the folding of secretory proteins. We have identified and characterized an FKBP-like homologue in S. lividans TK21 and shown that this protein (which we have named Sli-FKBP) has a PPIase activity that can be inhibited by FK506 (figure 1a,b).

The accumulation of secretory misfolded proteins induces a secretion stress response that activates the two-component system, CssRS, which regulates the synthesis of three specific HtrA-like proteases able to degrade misfolded proteins at the late exponential growth phase [25]. As expected, the disruption of the sli-fkbp gene increased the transcriptional level of the CssRS operon, suggesting accumulation of misfolded proteins. In addition, absence of sli-fkbp severely affects secretion and activity of Sec-dependent α-amylase (figure 2a,b), similarly to what has been reported with the prsA mutant [14,16,66,67]. Furthermore, absence of sli-fkbp moderately affects the secretion and activity of Tat-dependent agarase (figure 2c,d), a protein previously thought to be secreted in a fully folded conformation.

Accordingly, prolyl isomerization to the cis conformation may be involved in Sec-dependent α-amylase (at position P350) and Tat-dependent agarase (at positions P127 and P183) (electronic supplementary material, figure S1) to render the mature active enzymes. Further research is needed to study the implication of Sli-FKBP and these proline residues in their folding and activity.

When the sli-fkbp gene was propagated in multi-copy in the α-amylase overproducer strain, secretion and activity increased (figure 3a,b). It has been described that overproduction of PrsA enhances the secretion of α-amylase from Bacillus amyloliquefaciens (AmyQ) [14], affecting its folding. Taken together with our results, this suggests that activity of an FKBP-like PPIase may be important for the acquisition of their respective active conformation in the case of some Sec-dependent secreted proteins. In addition, it strongly suggests that bacterial cells may need extra copies of this folding enzyme when secretory proteins are overproduced.

In the case of α-amylase overproduction, which is expected to form five disulfide bonds, maturation may involve the participation of thiol-disulfide oxidoreductases, as recently published [58], in addition to increased levels of the PPIase.

Despite the Sli-FKBP lipoprotein being predicted to contain a Tat signal peptide [34] and although it has been described that the Tat route is mainly expressed at the late phase of growth [26], the presence of Sli-FKBP has been detected in the secretome of S. lividans TK21 around the end of the exponential phase of growth [27], similar to findings reported for the S. lividans TK24 secretome [59]. Tat-dependent agarase shows a similar behaviour, and it has additionally been found that, under overproduction, pre-agarase may be partially targeted to the Sec route and be secreted unfolded (in contrast to the proposed folded conformation produced through the Tat route) [8]. The presence of Sli-FKBP in the secretome to affect Sec-dependent proteins earlier than what would otherwise be expected for a typical Tat protein might be explained by a similar mechanism.

Overproduction of Sli-FKBP increased activity of agarase in a proportion greater than what would be expected only by secretion (figure 3c,d). Remarkably, under agarase overproduction conditions, we also observed a shift in its SDS-PAGE mobility that only occurred when Sli-FKBP was overproduced concomitantly (figure 3b). Analysis of the peptides produced by trypsin digestion of the protein recovered from the faster migrating band reported identification of most of the agarase protein sequence (electronic supplementary material, figure S3). The observed band migration does not correspond to the mobility that would be expected if the lack of detection of unrecovered peptides had been due to specific protein degradation. Furthermore, specific protein degradation would conflict with the higher levels of agarase activity measured. Thus, the difference in mobility can hardly be explained just by protein degradation, and seems more likely to be due to gel shifting, a phenomenon previously described where mobility changes in SDS-PAGE gels are explained by changes in the protein conformation that affect its Stokes radius or surface charge. Actually, inspection of the structural models produced for agarase agrees with this observation.

The increase in agarase activity when Sli-FKBP is concomitantly overproduced might be justified by differences in the degree of agarase folding assisted by Sli-FKBP. According to modelling, the form P183cis, P127trans might keep an active catalytic site without an allosteric agarose-binding site, lacking processive ability with reduced efficiency. This could explain why, in unsaturated Sli-FKBP conditions, agarase overproduction may display lower activity than with Sli-FKBP overproduction (electronic supplementary material, figure S4).

The dependency of agarase activity on Sli-FKBP overproduction may provide an interesting model for introducing improvements to the industrial overproduction of other Sli-FKBP-dependent secreted proteins.

In any case, a relevant proportion of the proteins analysed by SRM/MRM showed a dependence of their relative abundance on Sli-FKBP when the Δsli-fkbp strain was used or when sli-fkbp was overexpressed, indicating that Sli-FKBP may have a wide relevance for protein secretion (table 1).

This is especially striking in the case of proteins secreted through the Tat route. Tat-dependent proteins are very few in comparison with Sec-dependent ones; not all of the verified 27 Tat-dependent proteins [10] may be produced in experimental growth conditions, and they are thought to be exported fully folded having a relatively simple tertiary structure, containing few or no disulfide bonds [68]. Tat-dependent secretory proteins containing cis bonds could reach their mature form either spontaneously or with the assistance of cytoplasmic PPIases [59]. In a situation of overproduction, the isomerization step could be limited, so the overproduction of an Sli-FKBP may help extracellularly to form more active protein in the periplasm. In the case of agarase, which lacks disulfide bonds, the thiol-disulfide oxidoreductase lipoprotein Sli-DsbA may also act as a chaperone to help obtain mature agarase [58]. Unfortunately, the MRM/SRM analyses failed to detect Sli-DsbA, but could detect the extracellular protein SLI18490, the single annotated chaperone in the secretome of S. lividans [59]. SLI18490 is also a lipoprotein, and its relative protein abundance increases in the presence of Sli-FKBP. The possible relationship between Sli-FKBP and SLI18490 in S. lividans could be related to the morphological differentiation as in other streptomycetes [63,69]. Together, Sli-FKBP and other chaperones (such as Sli-DsbA) could assist the folding of overproduced secreted proteins in S. lividans in a similar way to E. coli, where two periplasmic folding factors (chaperone Skp and PPIase FkpA) are involved in protein folding in the periplasm [53]. Further experimental work is required to characterize the actual relationships between Sli-FKBP, Sli-DsbA and SLI18490.

Conclusion

We have identified an FKBP-like protein in S. lividans that may be involved in the correct folding of a wide number of Sec- and Tat-dependent secretory proteins. Our results indicate that, in practical terms and regardless of the route used, when the relative amount of overproduced secretory protein exceeds the capacity of the proteins in charge of procuring them a correctly folded structure, it may be convenient to adjust the level of expression of specific folding enzymes to ensure complete folding and optimize the overproduction of functional homologous or heterologous secretory proteins in S. lividans.