Introduction

The Streptomyces are Gram-positive, soil-dwelling bacteria that have a complex life cycle involving vegetative mycelia, aerial hyphae and unigenomic spores [44, 57]. The vegetative mycelia grow into the substrate to obtain nutrients and give rise to aerial hyphae, which are responsible for the characteristic “fuzzy” appearance of Streptomyces colonies. Over time, the aerial hyphae undergo septation to produce pigmented spores that assist in dispersal [1, 2]. The Streptomyces are also known for their ability to produce a large variety of secondary metabolites, many of which have applications in medicine, agriculture and industry. Streptomyces clavuligerus is capable of producing a number of secondary metabolites which include the β-lactams: clavulanic acid (CA), cephamycin C, and a group of metabolites collectively referred to as the 5S clavams [16]. CA has potent β-lactamase inhibitory activity and is used in the treatment of bacterial infections that are otherwise resistant to certain β-lactam antibiotics [77]. CA and the 5S clavams partially share a common ‘early’ biosynthetic pathway and have dedicated ‘late’ steps involved in the production of the respective metabolites [41, 91]. In addition, multiple gene clusters have been shown to be involved in the production of CA and the 5S clavams, with the primary CA biosynthetic gene cluster being located adjacent to the cephamycin C gene cluster in S. clavuligerus [104]. Currently, CA is produced at the industrial level by fermenting S. clavuligerus as total chemical synthesis on a large scale is not economically or technically feasible [7].

The expression of many genes involved in secondary metabolism in the Streptomyces is regulated in a hierarchical manner [8, 66]. At the first level of control are the pathway-specific (or cluster-situated) transcriptional regulators that are encoded by genes located within or in the vicinity of the gene clusters that they regulate [101]. These regulators are in turn controlled by pleiotropic or global regulatory elements that coordinate secondary metabolism and morphological development [8, 66]. Two pathway-specific regulators are of particular importance during CA biosynthesis. The claR gene from the CA gene cluster encodes a LysR-family transcriptional regulator, which controls the expression of the ‘late’ genes involved exclusively in CA biosynthesis [67]. Recently, it was also reported that the expression of other genes involved in development and secondary metabolism is also affected in a S. clavuligerus claR mutant [56]. The ccaR gene from the cephamycin C gene cluster encodes a member of the Streptomyces antibiotic regulatory protein (SARP) family and is required for the production of cephamycin C and CA, but not for that of the 5S clavams [2]. CcaR regulates CA production both directly and indirectly by controlling the expression of the “early” genes from the CA gene cluster [94] and by regulating the expression of claR [67, 70, 82, 83], respectively.

Pleiotropic regulators controlling secondary metabolism and/or morphological development in S. clavuligerus include the products of the bldA and bldG genes, which encode a tRNA and an anti–anti-sigma factor (AASF), respectively [12, 48]. bldA and bldG are highly conserved among Streptomyces spp. and are part of the bld (bald) gene family, which get their name from the fact that mutations in these genes lead to loss of aerial hyphae formation and result in smooth colonies instead of the characteristic fuzzy appearance of wild type (wt) colonies [11, 24, 48, 61, 63, 73]. In S. coelicolor and S. griseus, the BldG AASF is known to interact with different anti-sigma factor (ASF) proteins to regulate multiple sigma factors by a partner-switching mechanism [60, 69, 86, 96]. Deletion of bldG completely abolishes the production of secondary metabolites and aerial hyphae in S. clavuligerus [11]. The loss of CA and cephamycin C production in the S. clavuligerus bldG mutant is due to blocked expression of ccaR, although at this time it is still unclear as to how BldG regulates ccaR transcription [11]. The bldA gene encodes a leucine-specific tRNA required for translating TTA codons, which is the rarest codon present in Streptomyces genes due to the high GC content of their DNA [33]. Mutations in bldA result in the loss of aerial hyphae formation and secondary metabolite production in S. coelicolor, S. lividans and S. griseus [47, 48], whereas an S. clavuligerus bldA mutant is developmentally defective but can still produce CA and cephamycin C at reduced levels as compared to the wt strain [97]. This finding was initially surprising as ccaR contains a TTA codon, but the production of the CcaR protein still occurred in a S. clavuligerus bldA mutant, suggesting that the TTA codon in ccaR is somehow mistranslated using another tRNA [97].

BldG is known to regulate CA and cephamycin C production and morphological differentiation in S. clavuligerus [11] and has been shown to interact with multiple proteins in other Streptomyces spp. [60, 69, 86]. In contrast, the bldA tRNA is required for proper morphological development but not CA and cephamycin C production in S. clavuligerus [97]. Therefore, quantitative proteome-wide changes in the S. clavuligerus bldG and bldA mutants were examined using isobaric tag for relative and absolute quantitation (iTRAQ) for comparison to the wt strain. This was done to obtain a global survey of protein expression profiles in the three strains and to attempt to differentiate between proteins involved in secondary metabolism and morphological differentiation. The study was further focused to identify proteins known or predicted to be involved in regulation to identify potential elements involved in controlling CA production in S. clavuligerus at a level above that of CcaR.

Materials and methods

Bacterial strains, culture conditions and detection of CA in supernatants

Streptomyces clavuligerus NRRL3585 strains were maintained on International Streptomyces Project medium 4 (ISP-4) or maltose-yeast extract-malt extract (MYM) agar media [46]. Ten ml of trypticase soy broth seed cultures were grown for 48 h in 50 ml flasks supplemented with 2 % w/v starch (TSBS). Two ml of the seed cultures were used to inoculate 100 ml of SA production cultures in 500 ml flasks. All S. clavuligerus cultures were grown at 28 °C and liquid cultures were agitated using a rotary shaker at 150 rpm. Cultures used for subsequent proteomics analysis were also monitored by Gram staining and microscopy to verify growth characteristics and purity. Standard antibiotic concentrations were used when required [11]. Bioassays to detect CA production in culture supernatants were performed using previously described procedures [94]. Reversed phase high performance liquid chromatography (HPLC) analysis of imidazole-derivatized culture supernatants was performed using a 100 × 8 mm Bondclone™ 10 µm C18 148 Å LC Column (Phenomenex Inc., USA) as described previously [31]. All media and reagents were purchased from Fisher Scientific or VWR International (Canada) unless otherwise specified.

Sample preparation, iTRAQ labeling and LC–MS/MS analysis

One hundred ml of SA cultures grown for 48 h were harvested by centrifugation at room temperature and the pellets were re-suspended in 3 ml phosphate buffered saline (pH 7.4) containing 0.01 % w/v SDS. The cells were disrupted immediately by sonication on ice, followed by centrifugation and the supernatants were frozen in liquid N2 for subsequent analysis. Preliminary quantification of protein concentrations for growth determination was carried out using the OmniPur™ BRADFORD Kit (EMD/Millipore, USA). All subsequent proteomics analysis was carried out at the Genome BC Proteomics Centre, University of Victoria, Victoria, BC, Canada. Protein samples were quantified and precipitated overnight in acetone followed by resolubilization in triethylammonium bicarbonate (TEAB) and 0.2 % w/v SDS. Samples were reduced with tris(2-carboxyethyl)phosphine hydrochloride (TCEP), alkylated with methyl methanethiosulfonate (MMTS), digested in-solution with trypsin and labeled with the appropriate 8-plex iTRAQ label according to the manufacturers recommendations (Applied Biosystems). Different labels used for each extract were as follows: wt (113, 114, and 115), ΔbldG (116, 117, and 118) and ΔbldA (119 and 121).

The iTRAQ labeled peptides were combined, desalted, and concentrated on an Oasis HLB cartridge (Waters Corp.), and speed vacuum centrifuged to dryness before separation by high pH reversed phase HPLC. Fractions were collected every 30 s for 96 min and were concatenated into 24 fractions by combining every 24th fraction (e.g., fractions 1, 25, 49, 73 were combined). The length of the reverse gradient was 2 h per HPLC fraction. Each of the samples was rehydrated to 110 μl (2 % acetonitrile, 0.1 % formic acid). Following rehydration, 15 μl aliquots of the peptide solutions were separated by on-line reversed phase liquid chromatography using a Thermo Scientific EASY-nLCII system with a reversed phase Magic C-18AQ (75 μm ID, 2 cm length, 5 μm, 100 Å, Michrom BioResources, Inc.) pre-column and a Magic C-18AQ nano-analytical column (75 µm ID, 15 cm length, 5 µm, 100 Å, Michrom BioResources, Inc.) at a flow rate of 300 nl/min. The chromatography system was coupled on-line to an LTQ Orbitrap Velos mass spectrometer equipped with a Nanospray Flex source (Thermo Scientific). The solvents used are as follows: A 2 % acetonitrile, 0.1 % formic acid; B 90 % acetonitrile, 0.1 % formic acid. After a 249 bar (~5 μl) pre-column equilibration and a 249 bar (~8 μl) nano-column equilibration, samples were separated using a 120 min gradient (0 min 5 % B; 100 min 40 % B; 5 min 80 % B; 2 min 95 % B; 13 min 95 % B). The parameters for the LTQ Orbitrap Velos (Thermo Scientific) were as follows: Nano-electrospray ion source with spray voltage 2.1 kV; capillary temperature 250 °C; Survey MS1 scan m/z range 400–1800 profile mode; resolution 30,000 at m/z 400 with AGC target 1E6; and one microscan with maximum inject time 500 ms. A siloxane m/z 445.120024 lock mass was used for internal calibration. Other settings were: preview mode for FTMS master scans: on; injection waveforms: on; monoisotopic precursor selection: on; rejection of charge state: 1. The experiment was acquired as a top ten data dependent analysis of the most abundant ions with charge states of 2–4, exceeding 5000 counts, being selected for higher-energy collisional dissociation (HCD) FT MS/MS fragmentation (scans 2–11) and detection in centroid mode. Dynamic exclusion settings were: repeat count 1; repeat duration 15 s; exclusion list size 500; exclusion duration 15 s with a 10 ppm mass window. The FT HCD settings were: resolution 7500 at m/z 400 in centroid mode with AGC target 1E5, amu isolation width, and normalized collision energy 50 %, 0.1 ms activation time.

Data analysis

The raw files generated from LC–MS/MS were analyzed using Proteome Discoverer 4.1 software suite (Thermo Scientific). Spectrum Selection was used to generate peak lists of the HCD spectra (parameters: activation type HCD; s/n cut-off 1.5; total intensity threshold 0; minimum peak count 1; precursor mass 350–5000 Da). A list of S. clavuligerus proteins was created using the two available S. clavuligerus full genome sequences (NZ_ADWJ01000000 and NZ_ADGD01000000) as well as the four plasmids (NZ_CM001016.1, NZ_CM001017.1, NZ_CM001018.1, and NZ_CM001019.1) using Artemis (Wellcome Trust Sanger Institute). The sequences of all predicted S. clavuligerus proteins were uploaded to the Mascot 2.4 server and the and the peak list generated by Proteome Discoverer 4.1 was used to search the database using the following parameters: precursor tolerance 10 ppm; MS/MS tolerance 20 mmu for FT MS/MS HCD data; enzyme Trypsin; two missed cleavages; instrument type FT-ICR ESI; fixed modification: Methylthio (C), iTRAQ 8-plex (K), and iTRAQ 8-plex (N-term); variable modifications: Oxidation (M), Deamidated (NQ), iTRAQ® 8-plex (Y) [18]. The HCD Percolator settings were: max delta Cn 0.05; target false discovery rate (FDR) strict 0.01, target FDR relaxed 0.05 with validation based on q value [18].

36,780 peptides with a single S. clavuligerus protein database match from the Mascot search were retained for further analysis. To ensure that only data that were included by Proteome Discoverer during quantitation analysis were used, peptides had to be classified as “used,” which requires the presence of an iTRAQ label and at least one valid iTRAQ reporter ion ratio. The median abundance of multiple peptides uniquely mapped to a single protein was used as the abundance measurement for that protein. All protein abundance measurements that were missing or were <1, were set to 1 and the measurements were log2 transformed. Replicates for each strain were normalized together using quantile normalization available in the normalize between arrays function in the R package, Limma (version 3.16.5). Pairwise differential protein expression between the three strains (i.e., ΔbldG–wt, ΔbldA–wt, ΔbldA–ΔbldG) was identified by means of a moderated paired t test [88] available in the Bioconductor package Limma. P values were corrected for multiple testing using the false discovery rate (FDR) approach [5]. Proteins with a FDR-corrected P value of <0.01 in any given comparison were deemed to have a statistically significant differential expression in that comparison.

Functional annotations for S. clavuligerus proteins were obtained by searching the protein sequences against Pfam database [29], NCBI Conserved Domain Database (CDD) [54] and dcGO database [26]. Based on manual examination of these functional annotations and on the genomic location of the respective encoding genes, proteins were classified in one of five functional groups; namely, metabolism, cellular processes, regulation, energetics and translation. Number of proteins per functional group differentially expressed in the ΔbldG–wt, ΔbldA–ΔbldG, ΔbldA–wt comparisons were counted and displayed using the barplot2 function in the R package gplots (version 2.14.1). Proteins found differentially expressed in more than one comparison were counted only once taking into account the comparisons in the following order: ΔbldG/wt, ΔbldA/ΔbldG and ΔbldA/wt. In addition, differentially expressed regulatory proteins (±2 fold and more) were identified for further analysis.

Z scores were calculated with the formula (x − μ)/σ where x is the normalized abundance of a protein in a sample, μ is the average abundance of that protein across all samples and σ is the standard deviation of the same protein across all samples. Heat maps of Z scores of proteins involved in regulation were obtained using the function heatmap.2 available in the R package gplots. The function was executed using hierarchical clustering with complete linkage and Euclidean distance.

Identification of TTA codons

The full genome sequence data of the S. clavuligerus chromosome (NZ_CM001015), as well as the four plasmids, pSCL1 (NZ_CM001016), pSCL2 (NZ_CM001017), pSCL3 (NZ_CM001018), and pSCL4 (NZ_CM001019), were obtained in FASTA format from NCBI as described above, and was uploaded to the TTA Lynx server which filtered out all genes containing TTA codons [109]. The proteins encoded by these genes were then manually searched for in the data set. The location and context of each TTA codon was also verified manually.

Results

Sampling conditions and summary of proteomics analysis

A survey of CA production was first carried out to verify that the strains being subjected to detailed proteomics analysis displayed previously reported phenotypes [11, 97] in terms of metabolite production when grown in starch asparagine (SA) medium. SA was selected as it is a defined medium that supports the production of CA but not of the 5S clavams [42], making it a good choice for obtaining samples for proteomic profiles specifically during CA biosynthesis in S. clavuligerus. Bioassays and HPLC analysis of culture supernatants from the three strains grown for up to 120 h were performed using samples taken at 24 h intervals (Fig. 1). As expected, CA was detected in culture supernatants from the wt strain and the bldA mutant at all of the time points examined, whereas no CA was detected in the bldG mutant samples at any of the time points (Fig. 1a). Although CA production still occurred in the bldA mutant, quantitative analysis indicated that production was reduced by 80 % in this strain as compared to wt (Fig. 1b).

Fig. 1
figure 1

Clavulanic acid production and growth of the S. clavuligerus wt, ΔbldA and ΔbldG strains grown in SA medium. a Bioassay results for the respective strains grown for up 120 h. b Quantitative HPLC results for clavulanic acid production levels using imidazole-derivatized supernatants from the respective strains after 48 h of growth. c Growth levels of the three strains at the 48 h time point, which was used for harvesting mycelia for subsequent proteomics analysis. d Morphologies of mycelia from the three strains at the 48-h time point showing growth in the form of mycelial mats

Full size image

Previous studies have reported that CA production in wt S. clavuligerus peaks after approximately 96 h of growth in SA and in other media [68]. Unlike many other secondary metabolite producers, the production of cephamycin C and CA in S. clavuligerus takes place during exponential growth and is not initiated during idiophase [81]. Therefore, the 48 h time point during growth in SA was chosen for sampling to obtain a proteomic profile at the onset of CA production based on our results and previous reports [3, 93]. The growth of all strains was comparable in SA media at the 48-h time point (Fig. 1c) and took place in the form of mycelial mats (Fig. 1d). A list of the amino acid sequences of all proteins predicted to be encoded by the S. clavuligerus chromosome and plasmids was compiled based on previously reported genome sequences (Supplementary File 1) [58, 92], and was used in the analysis. A total of 2496 S. clavuligerus proteins were detected in the subsequent iTRAQ analysis of which 2442 met the strict confidence criteria as described in the methods section, and were therefore included in the study (Supplementary File 2). This corresponds to approximately 33.5 % of the entire S. clavuligerus proteome (Supplementary File 1), as the S. clavuligerus genome has been predicated to encode 7281 proteins [58, 92].

Data normalization and differential expression analysis

Quantile normalization of the replicates from each strain was performed to correct for variations in protein abundance [13]. Analysis showed that replicates from each strain clustered together, demonstrating that protein expression levels between the replicates was comparable (Supplementary File 3). Therefore, the quantile normalized data was used for further analysis. Pair-wise differential protein expression patterns between strains were determined using a moderated paired t test [78] with FDR-corrected P values, in which P values of <0.01 in any given comparison were deemed to be statistically significant (Supplementary File 4). Further analysis showed that 275 proteins were over-expressed and 641 were under-expressed in the bldG mutant when compared to the wt strain. For the bldA mutant, a total of 57 proteins were over-expressed and 1279 proteins were under-expressed when compared to the wt strain. A total of 224 proteins were over-expressed and 1240 were under-expressed in the bldA mutant when compared to the ΔbldG strain. Furthermore, 426, 222 and 227 proteins were expressed at similar levels between ΔbldG and wt, ΔbldA and wt, and ΔbldA and ΔbldG, respectively (Fig. 2a; Supplementary File 5). By sorting the differentially expressed proteins into functional groups (metabolism, cellular processes, regulation, energetics, translation) based on Pfam domains [29], it was found that the majority of proteins were differentially expressed in the bldG mutant when compared to the ΔbldA and wt strains (Fig. 2b). Of particular interest to this study was the identification of regulatory proteins that may be directly or indirectly regulated by BldG. This includes sigma factors and other transcriptional regulators that could potentially be involved in regulating secondary metabolism in S. clavuligerus. As shown in Fig. 3, 192 known and putative regulatory proteins showed statistically different expression levels among the three strains (Supplementary File 6). These 192 proteins along with others of interest based on predicted functions and the chromosomal location of their respective genes were chosen for further detailed analysis.

Fig. 2
figure 2

Pairwise analysis of differentially expressed proteins in the S. clavuligerus wt, ΔbldA and ΔbldG strains grown in SA medium. Differentially expressed proteins were analyzed per comparison (ΔbldG/wt, ΔbldA/wt and ΔbldA/ΔbldG) according to criteria described in the methods section. a Venn diagram showing number of proteins differentially expressed in the pairwise comparisons. The identities and details of proteins from each comparison and subgroup are listed in Supplementary File 5. b Number of differentially expressed proteins per comparison based on functional categories (metabolism, cellular processes, regulation, energetics and translation). Proteins deemed differentially expressed in more than one comparison were counted only once and were assigned to a comparison in the following order: ΔbldG/wt, ΔbldA/ΔbldG and ΔbldA/wt, respectively

Full size image
Fig. 3
figure 3

Heat map showing the relative abundance of differentially expressed regulatory proteins in wt, ΔbldG, and ΔbldA strains of S. clavuligerus. Columns correspond to strains and replicates, whereas each row represents a protein included in the analysis. Proteins were grouped using hierarchical clustering with Euclidean distance and complete linkage. Heat map colors indicate row-wise Z scores. The Z score of a protein is the number of standard deviations away from the mean abundance of that protein across all samples. The dendrogram on the left indicates groups of similarly expressed proteins resulting from the hierarchical clustering analysis. Proteins with known regulatory functions from S. clavuligerus and other species are marked with a dash on the left panel. Details regarding the identities of the proteins and Z scores can be found in the supplementary section (Supplementary File 6)

Full size image

Sigma factors and related proteins

A total of ten sigma factors, three ASFs, and one AASF were detected in the iTRAQ dataset (Table 1). Among these were the AASF encoded by bldG, two ASFs (Orf3/ApgA and PrsH/RshA) shown previously to interact with the BldG orthologue in S. coelicolor and S. griseus, and the sigma factor SigH, which in S. coelicolor and S. griseus is regulated by PrsH/RshA [86, 96]. As expected, BldG is under-expressed in both the S. clavuligerus bldG and bldA mutants as compared to the wt strain. However, the expression of Orf3/ApgA, PrsH/RshA and SigH was unaffected in the same pair-wise comparisons. Two additional sigma factors (BldN/AdsA and 490052605) were significantly under-expressed in the S. clavuligerus bldG mutant and were unaffected in the ΔbldA when compared to the wt strain, which agrees with previously published results on bldN expression in S. coelicolor bldA and bldG mutants [9]. A second sigma 70-family protein (490061073) was found to be significantly under-expressed in both the S. clavuligerus bldG and bldA mutants when compared to the wt strain. Additionally, four sigma factors were under-expressed in the ΔbldA when compared to the wt strain and were unaffected in the bldG mutant (Table 1). These include RpoE (known as SigR in S. coelicolor), which is a key regulator of the oxidative stress response [65]; HrdD, the function of which is yet unknown and is partially dependent on SigE for its own expression [17, 64]; a SigF homologue (490058906), which is required for proper sporulation in other Streptomyces spp. [98]; and HrdB, which is the principle sigma factor responsible for transcribing housekeeping genes in S. coelicolor [87]. In addition, two sigma 24-family proteins (497683136 and 490056752) were over-expressed in the bldG mutant and under-expressed in ΔbldA when compared to the wt strain (Table 1).

Table 1 Comparative expression levels of sigma factors, anti-sigma factors, and anti–anti-sigma factors detected during iTRAQ analysis of protein extracts from the wt, ΔbldA and ΔbldG strains of S. clavuligerus
Full size table

Transcriptional regulators

Several proteins that are known or may be involved in indirectly regulating CA production were detected in the dataset (Fig. 4). One AraC family protein (490055879/AdpA/BldH) is under-expressed in both the bldG and bldA mutants when compared to the wt strain. In addition, BldH/AdpA is under-expressed in ΔbldA when compared to ΔbldG, which can be explained by the presence of a TTA codon in bldH/adpA. The observed decrease in the expression of BldH/AdpA in the bldA mutant was not expected as it has a role in the positive regulation of ccaR and therefore indirectly controls CA biosynthesis [52]. A homologue of the S. coelicolor RNaseIII (490059733/AbsB), which is involved in regulating secondary metabolite production [74, 106], was under-expressed to similar extent in both ΔbldG and ΔbldA when compared to the wt strain. The expression of the S. clavuligerus BldD orthologue (490054092), which is known to repress the pre-mature expression of multiple genes involved in morphological differentiation and antibiotic biosynthesis in S. coelicolor [24, 25, 38], was unaffected in all comparisons (Supplementary Files 4–6). Finally, a protein identified as an AfsR-like regulator (497683347) was found to be similarly under-expressed in both ΔbldG and ΔbldA when compared to the wt strain. On closer examination, the protein was identified to be a partial entry for CcaR [2].

Fig. 4
figure 4

Regulatory cascade leading to clavulanic acid production and development in S. clavuligerus based on results from previously published reports [66] and the current study. Proteins and gene products detected in this study are indicated in purple color and are assigned a number, which was used to report their relative expression levels in the inset legend. Red and green arrows indicate repression or activation by the designated protein, respectively. In the inset legend G or A indicates the expression level of the respective protein assigned the specific number in the bldG or the bldA mutant as compared to the wt strain, with red indicating reduced expressed and green indicating over-expressed in the respective comparison

Full size image

A putative AfsR-like protein (490057267) was under-expressed in ΔbldG when compared to the wt strain (Supplementary Files 4–6). The gene encoding this protein is located in a polyketide synthase (PKS)-associated gene cluster designated as SMC5, making it likely that this protein is a SARP [58]. A single LysR-family transcriptional regulator (497683951) was over-expressed in ΔbldG and under-expressed in ΔbldA when compared to the wt strain. Four IclR transcriptional regulators detected in the dataset were under-expressed in the bldG mutant when compared to the wt strain (Supplementary Files 4–6).

Six additional regulators were differentially expressed in the pairwise comparisons, which included proteins belonging to the AsnC-family (490055317 and 490056872), Fur-family (497681972), the PbsX-family (490050159/PcbR), NmrA-family (497683061) and DeoR-family (490052578) (Supplementary File 6). A single REX-family transcriptional regulator (490058535), which in other Streptomyces regulates genes encoding components of the respiratory chain [14], was found to be over-expressed in ΔbldG and under-expressed in ΔbldA when compared to the wt strain (Supplementary Files 4–6). Multiple GntR-family transcriptional regulators, which have various functions including gluconate repression, were also detected and were found to be differentially expressed in the comparisons (Supplementary Files 4–6). This included the auto regulatory sporulation specific transcriptional factor WhiH (490059845) [71], which was over-expressed in ΔbldA as compared to the ΔbldG strain.

Several TetR-family proteins were also identified and were found to be differentially expressed in the pairwise comparisons (Supplementary File 5). TetR-family members are involved in the transcriptional regulation of multidrug efflux, antibiotic biosynthesis pathways, osmotic stress response, catabolic pathways and differentiation processes [76]. One TetR-family protein (490060245) was under-expressed in the ΔbldG and was unaffected in the bldA mutant when compared to the wt strain. The gene encoding 490060245 (SCLAV_5246) is located downstream from the gene encoding RocG/GdhA, the deletion of which increases production of the immunosuppressant FK506 in S. tsukubaensis [37]. Two of the TetR-family proteins (497682848/AtrA and 490057306) were under-expressed in both ΔbldG and ΔbldA as compared to the wt strain. In S. coelicolor, AtrA regulates the transcription of actII-ORF4, the pathway-specific activator of actinorhodin biosynthesis [99]. The gene encoding the second TetR-family regulator (490057306: SCLAV_0504) is located upstream of genes involved in peptidoglycan synthesis (SCLAV_0506-0507). Two TetR-family regulators (490051466 and 497681723) were over-expressed in ΔbldG as compared to the wt strain and were unaffected in the bldA mutant. The protein designated as 490051466 is a homologue of AmfC, a mycelia formation protein that lacks any TetR-family regulatory domains. The gene encoding the second protein (497681723) is located downstream of putative genes encoding components of an ABC-type transport system (SCLAV1312-1313). An additional TetR-family regulator (497683062) was over-expressed in ΔbldG and under-expressed in the bldA mutant as compared to the wt strain. The gene encoding 497683062 (SCLAV_3466) is located in the opposite orientation next to genes encoding two putative serine/threonine protein kinases homologous to S. griseus AfsK (SCLAV_3467a and 3467b), as well as a gene encoding a hypothetical regulatory protein with a DUF397 domain (SCLAV_3468). Proteins containing the DUF397 domain have been previously implicated in regulating antibiotic production in S. coelicolor [28, 73].

A total of eight XRE-family transcriptional regulators, which control various and diverse metabolic functions, were also found to be differentially expressed in ΔbldG as compared to the wt strain (Table 2; Supplementary Files 4–6). The S. clavuligerus BldD orthologue (490054092), which is a pleiotropic XRE-family regulator controlling both development and antibiotic production in S. coelicolor, was not differentially expressed in ΔbldG or ΔbldA when compared to the wt strain.

Table 2 Comparative expression levels of XRE-family regulators and other proteins encoded by genes located near/next to genes encoding DUF397-containing proteins detected during iTRAQ analysis of protein extracts from the wt, ΔbldA, and ΔbldG strains of S. clavuligerus
Full size table

Two component systems and serine/threonine kinases

PhoP (490059014), a response regulator known to be involved in the phosphate-mediated regulation of antibiotic biosynthesis [55], was found to be over-expressed in ΔbldG and was unaffected in ΔbldA when compared to the wt strain (Supplementary Files 4–6). Several of the proteins known to be under the control of PhoP were also observed to be differentially expressed, including PstB (490058978) and PhoU (490051157).

Several other response regulators were over-expressed in the bldG mutant, including the regulator of a vancomycin resistance system (490054741/VanR) [39] and 497682237, which is a putative orphan response regulator. Several other putative response regulator proteins belonging to the LuxR family were also differentially expressed in both ΔbldG and ΔbldA when compared to the wt strain (Supplementary File 4–6). Genes for all of these proteins were located in close proximity to genes encoding putative sensor kinases with the exception of 490056384, which is currently annotated in the database as a protein involved in differentiation.

A total of nine serine/threonine kinases were detected in the dataset (Supplementary File 4). The expression of the S. clavuligerus orthologue of the AfsK kinase (497683065) was unaffected in all comparisons. The AfsK/AfsR/AfsS system controls both differentiation and secondary metabolism and disruption of any of the encoding genes results in decreased antibiotic production [30, 34, 36, 50, 75]. The kinase designated as 497683146 was over-expressed in ΔbldG and under-expressed in ΔbldA when compared to the wt strain and is homologous to S. coelicolor PkaD, which is thought to be a positive regulator of actinorhodin production in that species [100]. Three other kinases known to be involved in development (490060020/AmfT/RamC, 497683145, and 497682612) were under-expressed in both ΔbldG and ΔbldA when compared to the wt strain. A kinase designated as 497683929 was over-expressed in ΔbldG and was unaffected in ΔbldA as compared to the wt strain. 497683929 is encoded by a gene (SCLAV_5481) located in a predicted non-ribosomal peptide associated gene cluster and does not seem to have any homologues in other sequenced Streptomyces species [58].

Identification and expression of proteins encoded by TTA codon containing genes

Analysis of the S. clavuligerus chromosome and plasmid (pSCL1, pSCL2, pSCL3 and pSCL4) nucleotide sequences identified a total of 243 TTA codon containing genes and 24 proteins encoded by them were present in the dataset, which included eight transcriptional regulators/DNA binding proteins (Table 3; Supplementary files 4, 5). Twenty proteins, including BldH/AdpA, were under-expressed in both ΔbldA and ΔbldG when compared to wt strain, but surprisingly none of the proteins were under-expressed solely in the bldA mutant. A total of 13 of the identified TTA codon containing genes in the data set have homologous counterparts in other Streptomyces species of which only five genes were found to retain the TTA codon in across all species (Table 3).

Table 3 Comparative expression levels of proteins encoded by TTA codon containing genes detected during iTRAQ analysis of extracts from the wt, ΔbldA and ΔbldG strains of S. clavuligerus
Full size table

Discussion

The main goal of the described work was to identify proteome wide changes in the S. clavuligerus bldG and bldA mutants for comparison to each other and to the wt strain, focusing on proteins involved in regulatory pathways. Such analysis could lead to the elucidation of previously unknown regulatory pathways and mechanisms, which can provide avenues for manipulating and increasing CA production and secondary metabolism. In addition, regulatory elements found to be differentially expressed in the ΔbldG mutant when compared to the ΔbldA and wt strains could potentially be involved in the BldG-mediated regulatory cascade, which ultimately controls CA and cephamycin C production by regulating the expression of ccaR. The most relevant findings based on the described goals are discussed.

Additional sigma factor targets of BldG remain elusive

In Streptomyces, the expression of genes involved in both morphological differentiation and secondary metabolism require various sigma factors [9, 43, 85, 86, 103], but the sigma factor(s) responsible for the transcription of ccaR in S. clavuligerus is still unknown [11]. The BldG orthologue in S. coelicolor and S. griseus has been found to interact with several ASFs including Orf3/ApgA, PrsH/RshA (regulator of SigH) and RsfA (regulator of SigF) (Fig. 4) [60, 69, 86, 96]. The protein expression data obtained in the current study showed that BldG is under-expressed in both the S. clavuligerus bldG and bldA mutants when compared to the wt strain; however, the ASFs Orf3/ApgA and PrsH/RshA and the sigma factor SigH are not differentially expressed in any of the comparisons (Table 1; Fig. 4). This was surprising as bldG is necessary for sigH expression via the sigHp2 promoter in other Streptomyces species [86]. Therefore, the link between BldG and SigH in S. clavuligerus warrants further examination.

Further analysis was conducted to identify sigma-B and sigma-F family members in the dataset as these are known to be regulated by AASF/ASF protein pairs [6, 21]. The only sigma-B homologue detected in the dataset was HrdD, which was not differentially expressed in ΔbldG mutant and was under-expressed in ΔbldA when compared to the wt strain (Table 1). S. coelicolor hrdD mutants are not defective in development or secondary metabolism [17], making it unlikely that it is a target of BldG in S. clavuligerus. An S. clavuligerus SigF homologue (490058906) was also detected, which was under-expressed in both the bldG and bldA mutants when compared to the wt strain (Table 1). Previous studies have shown that SigF is required for proper sporulation S. coelicolor [98] and that BldG interacts with the SigF-specific ASF RsfA in that organism [60]. The reversible phosphorylation of BldG by RsfA is necessary for proper morphological development in S. coelicolor [10, 60], and both Orf3/ApgA and PrsH/RshA lack the necessary kinase domain required for this activity [69, 86]. It is likely that BldG has an additional sigma factor partner to interact with since sigH and sigF mutants of S. coelicolor exhibit a whi (white) phenotype in which aerial hyphae septation or maturation is affected, in contrast to the bld phenotype of the ΔbldG mutant [12, 85]. It is also likely that BldG has an additional ASF partner to interact with as in vivo phosphorylation of BldG was not affected in a previously analyzed S. coelicolor rsfA mutant [60].

Results also confirm a role for BldG in the regulation of BldN/AdsA, which is consistent with previously reported transcriptional studies in a S. coelicolor bldG mutant [9]. In the current study, BldN/AdsA is the most under-expressed sigma factor in ΔbldG as compared to the ΔbldA and the wt strains (Table 1). Both BldN/AdsA and the transcriptional regulator BldH/AdpA are required for the transcription of ssgA, which is necessary for spore formation in S. coelicolor [102, 107, 108], and BldH/AdpA also has a role in the positive regulation of ccaR expression in S. clavuligerus [52]. In S. coelicolor, BldN/AdsA is part of an operon that requires BldH/AdpA for its transcription [107], and BldH/AdpA is itself dependent on bldA for its translation due to the presence of a TTA codon in the encoding gene [95]. In the current study, BldH/AdpA was under-expressed in the S. clavuligerus bldA and bldG mutants when compared to the wt strain (Table 1). Since BldN/AdsA was not under-expressed in the bldA mutant when compared to the wt strain in this study, it appears that S. clavuligerus AdpA plays a role more similar to that of BldH/AdpA in S. coelicolor (BldH/AdpA not required for BldN/AdsA), rather than that of BldH/AdpA in S. griseus (BldH/AdpA required for BldN/AdsA) [95, 107]. This suggests that BldG somehow regulates BldN/AdsA expression independently of BldH/AdpA in S. clavuligerus. It is also possible that BldN/AdsA is the sigma factor responsible for the transcription of ccaR, which would be one of the first proteins to be identified to explain the link between BldG and ccaR expression (Fig. 4). Studies are currently underway to investigate the role of BldN/AdsA in the transcription of ccaR and other genes in S. clavuligerus.

A possible system of XRE-family transcriptional regulators and DUF397-domain containing proteins in the regulation of development and secondary metabolism

Several predicted DNA binding proteins were under-expressed in ΔbldG and showed varying expression patterns in ΔbldA when compared to the wt strain (Supplementary Files 4–6). Two of these predicted proteins (490051454 and 490054446) contain the HTH_XRE conserved domain found in the XRE-family of transcriptional regulators. The genes encoding 490051454 and 490054446 are clustered together with genes predicted to encode proteins containing the DUF397 domain, which has been previously implicated in the regulation of secondary metabolism [73]. A similar arrangement was observed for almost all of the genes encoding XRE-family proteins found to be under-expressed in the bldG mutant in the current study (497683074, 497682597, 490058590, 497681584, 490058041, 490055478, 490050197 and 490054541) (Table 2). There are 45 genes encoding putative DUF397-containing proteins in S. clavuligerus and only one (497683523) was present in the dataset, which was under-expressed in ΔbldG and was unaffected in ΔbldA when compared to the wt strain. In addition, 497683523 is 74 % similar to BldB, which has a known function in regulating development and secondary metabolism in S. coelicolor [73].

Previous studies have suggested that BldB and BldB-like proteins likely have interacting partners, the combinations of which somehow regulate cellular functions [22, 23]. Studies in S. coelicolor on an XRE-family regulator (WhiJ) and a small acidic protein of unknown function with a DUF397 domain (SCO4542) may provide clues as to how these proteins might regulate gene expression. Deletion of SCO4542, which is located immediately downstream of whiJ, results in a bald phenotype, but this phenotype is completely reversed in a SCO4542/whiJ double mutant [1]. This suggests that WhiJ represses development and the SCO4542 product somehow counters its ability to do so. Therefore, it is possible that the function of the DUF397 domain containing proteins is to interact antagonistically with XRE-family proteins [57]. Based on the arrangement of the XRE-family/DUF397 containing protein encoding genes in S. clavuligerus and the proteomics data (Table 2), it is possible that these systems have some role in regulating secondary metabolism. In addition, these genes could themselves be under the indirect control of BldG, making them good candidates for future manipulation studies to elucidate their roles in CA production.

Possible role for BldD/BtdA in regulating CA production in S. clavuligerus

The most under-expressed transcriptional regulator in the S. clavuligerus bldG mutant (slightly over-expressed in ΔbldA) as compared to the wt strain is BdtA (Supplementary Files 4–6). The btdA gene encodes a DNA-binding protein that contains a MerR-family domain, and was originally thought to have a role in development in S. coelicolor [25]. Later studies showed that a S. coelicolor bdtA mutant did not have a distinctive phenotype [20, 38]. BldC is the only BdtA homologue in S. coelicolor and it is required for both differentiation and secondary metabolite biosynthesis [38]. The bdtA gene is negatively regulated by the pleiotropic transcriptional regulator BldD, which is known to repress the pre-mature expression of multiple genes involved in development, including the genes encoding the sigma factors WhiG and BldN/AdsA, as well as bldC (Fig. 4) [20]. Both BldD and BldC were detected in the dataset, with BldD not being differentially expressed in any of the comparisons and BldC being under-expressed in only the bldA mutant when compared to the wt strain (Supplementary Files 4–6).

At this point, it is not possible to determine whether BdtA is critical for either development or secondary metabolite biosynthesis in S. clavuligerus. However, the current expression data suggests an indirect role of BldG in the regulation of bdtA. In S. coelicolor, BldG and BldD are already known to indirectly or directly regulate expression of the same gene product in at least one case, namely SigH. Transcription from the SigH dependent promoter sigHp2 is abolished in a bldG mutant [86] and is activated in a bldD mutant [45]. Therefore, results from the current study are in good agreement with those from previous reports and work is currently underway to investigate a possible link between BdtA and CA production in S. clavuligerus.

BldG and phosphate regulation of secondary metabolism

The PhoP response regulator was found to be over-expressed in ΔbldG and was unaffected in ΔbldA when compared to the wt strain (Supplementary Files 4–6). The PhoR–PhoP TCS is involved in the phosphate-mediated regulation of secondary metabolism in Streptomyces [55]. Deletion of the system results in varying phenotypes depending on the host, suggesting that phosphate-mediated regulation of secondary metabolism is stricter in some species [32, 89]. For example, PhoP deletion mutants over-produce pigmented secondary metabolites in S. lividans [89], but no such increase is observed in S. coelicolor [84]. Additionally, proteins involved in phosphate transport that are positively regulated by PhoP under conditions of phosphate limitation were also found to be differentially expressed [32, 80], including the pathway-specific activator PhoU (unaffected in ΔbldG and under-expressed in ΔbldA when compared to the wt strain) and PstB (under-expressed in ΔbldG and unaffected in ΔbldA when compared to the wt strain) (Supplementary Files 4–6). PhoU plays a role in the regulation of phosphate uptake and its expression has previously been found to be induced when PhoP is phosphorylated [90], and PstB is a PhoP-dependent member of the phosphate transport system [79]. The pho regulon has been characterized in S. clavuligerus (Fig. 4) and CA production is negatively regulated by phosphate [49]. It has also been suggested that PhoP somehow indirectly regulates CA biosynthesis since DNA binding sites for the protein were not identified in the CA biosynthetic gene cluster [80]. Therefore, the over-expression of PhoP in the bldG mutant could help explain the lack of CA production in this strain, although the link between BldG and PhoP is unclear at this time.

TTA dependence is re-evaluated in S. clavuligerus

In Streptomyces species, the bldA gene specifies the only known t-RNA capable of translating rare leucine TTA codons [72, 105], and S. coelicolor bldA mutants are blocked in both development and secondary metabolism [35, 48, 59]. It is also known that bldH/adpA encodes the pleiotropic transcriptional regulator responsible for the ΔbldA phenotype [77], and several other TTA-containing genes are required for the biosynthesis of the blue-pigmented secondary metabolite actinorhodin in S. coelicolor [27, 95]. In S. clavuligerus, bldA is only essential for development and not for secondary metabolism, which was surprising as ccaR contains a TTA codon [97].

It was suggested that certain TTA codons such as those present in ccaR are effectively mistranslated based on their location within the gene and the nucleotides flanking them [97]. The presence of rare codons within the first 25–50 codons has been shown to have a more severe effect on translation as compared to their presence towards 3′ end of the transcript [19]. This is unlikely to be the case for ccaR, which has a TTA codon at position 38 [97]. It was suggested that the identity of the nucleotides flanking the TTA codon in ccaR may allow it to be mistranslated, based on other systems involving frame shifting in yeasts and viruses [4, 15, 97]. According to the current hypothesis, a bldA-dependent TTA codon is immediately followed by a C or a T at the 3′ position, whereas a codon capable of being mistranslated such as the one in ccaR has a G at this position [97]. To date, ccaR was the only known bldA-independent TTA codon containing gene. Therefore, it was hypothesized that bldA-independent genes would always have a G or potentially an A at the following 3′ position of the TTA codon [97].

A total of 24 out of 243 predicted proteins (~10 %) encoded by TTA containing genes were present in the dataset (Table 3; Supplementary Files 4–5), which is a much lower percentage as compared to the total proteome detected. This is likely in part due to the fact that TTA codons only occur in genes involved in secondary metabolism and morphological differentiation, and therefore were not expressed under the growth conditions used [51]. AdpA/BldH appears to follow the conventional pattern of expression in that it is under-expressed in the bldA mutant and contains a C immediately following the TTA codon [95]; however, the model does not seem to fit well in other cases. Four proteins found to be most under-expressed in ΔbldA mutant when compared to ΔbldG and the wt strain are also encoded by genes containing a G or an A following the TTA codon (497681088, 490057169, 497685265, 490058041) (Table 3), suggesting that the presence of these nucleotides does not necessarily allow for effective mistranslation. The expression of three proteins (490060397, 490058513 and 497685280) encoded by genes with a C at the immediately 3′ end of the TTA codon was either unaffected or they were over-expressed in the bldA mutant when compared to the ΔbldG and wt strains, suggesting that all genes with Cs following the TTA codon are not bldA-dependent. Therefore, results from the current study do not support the nucleotide based mistranslation hypothesis described above [97], which is consistent with the slightly different biological programming followed in S. clavuligerus as compared to other model Streptomyces spp.

In conclusion, a total of 2442 proteins corresponding to 33.5 % of the S. clavuligerus proteome were detected from cellular lysates in the current study. There are only three previously published reports on the analysis of Streptomyces proteomes using iTRAQ in the literature, which were able to detect up to 15 % of the predicted proteome of S. coelicolor [40, 53, 62]. Many proteins involved in regulation were found to be differentially expressed in the comparisons (Figs. 3, 4). The genes for some of the differentially expressed proteins identified are good candidates for future manipulation studies for increasing CA production in S. clavuligerus. In addition, the analysis of proteins encoded by TTA codon containing genes suggests that the role of bldA and TTA dependence in S. clavuligerus needs to be reconsidered. To the best of our knowledge, the described study represents the most in depth proteomic cataloging and analysis in S. clavuligerus to date. The reported proteomics data will serve in the future as an useful tool to direct research into aspects of secondary metabolism, as well as into aspects of development and primary metabolism in S. clavuligerus and other industrially important Streptomyces species.