Properties of CsnR, the Transcriptional Repressor of the Chitosanase Gene, csnA, of Streptomyces lividans

Abstract

A palindromic sequence is present in the intergenic region preceding the chitosanase gene csnA (SSPG_06922) of Streptomyces lividans TK24. This sequence was also found in front of putative chitosanase genes in several other actinomycete genomes and upstream genes encoding putative transcriptional regulators of the ROK family, including csnR (SSPG_04872) in S. lividans. The latter was examined as a possible transcriptional regulator (CsnR) of chitosanase gene expression. In vitro, purified CsnR bound strongly to the palindromic sequences of the csnA and csnR genes (equilibrium dissociation constant [K_D] = 0.032 and 0.040 nM, respectively). Binding was impaired in the presence of chitosan oligosaccharides and d-glucosamine, and chitosan dimer was found to be the best effector, as determined by an equilibrium competition experiment and 50% inhibitory concentration (IC₅₀) determination, while glucose, N-acetyl-glucosamine, and galactosamine had no effect. In vivo, comparison of the S. lividans wild type and ΔCsnR strains using β-lactamase reporter genes showed that CsnR represses the expression of csnA and of its own gene, which was confirmed by quantitative PCR (qPCR). CsnR is localized at the beginning of a gene cluster, possibly an operon, the organization of which is conserved through many actinomycete genomes. The CsnR-mediated chitosanase regulation mechanism seems to be widespread among actinomycetes.

INTRODUCTION

The common soil environment is very complex. The organic substrate of soil is mostly composed of biopolymers produced by degradation of the cellular structures of living organisms. Chitin, a β-1,4-linked N-acetyl-glucosamine (GlcNAc) polymer, originates in soil mainly from fungal cell walls and insect cuticles and is the second most abundant biopolymer in soil after cellulose (12). Chitin deacetylase enzymes, varying by their substrate specificities and susceptibilities to environmental conditions, transform chitin into chitosan, a polymer essentially composed of β-1,4-linked d-glucosamine (GlcN) residues with a variable content of GlcNAc (53). Chitosan isolated from the cell walls of Zygomycetes (Mucor sp. and Rhizopus sp.) is characterized by a GlcNAc content of about 30% (2, 48). Chitosan with higher degrees of N-deacetylation is produced on an industrial scale by alkaline treatment of chitin mainly obtained from crustacean shells (42). Microorganisms that secrete chitosanases (EC 3.2.1.132), the chitosan endohydrolytic enzymes, are commonly found in soil. Chitosanase production is essential to degrade chitosan and assimilate it as a carbon and nitrogen source. Chitosanase production was also shown to provide some degree of resistance to the antimicrobial effect of chitosan (20).

Chitosan oligosaccharides, the end products of chitosan hydrolysis by chitosanases, are under intense investigation due to their potential biomedical applications. Tumor growth inhibition, acceleration of wound healing, and antifungal and antibacterial activities are only a few examples among recently studied properties of chitosan oligosaccharides (1, 52). This, in turn, results in an interest in chitosanases, which have been characterized from many microorganisms (1). Microbiological studies and the analysis of sequenced genomes showed that chitosanases are widespread among filamentous fungi and Gram-positive bacteria, particularly in bacilli and actinobacteria. In Streptomyces, well-studied chitosanases belong to glycoside hydrolase families GH2, GH5, and GH46 (11, 19, 23, 33, 49). Putative chitosanases from these families, as well as from GH75 (characterized only from fungal organisms) are found in many recently sequenced actinomycete genomes (CaZy database) (7).

Streptomyces lividans is an actinomycete isolated from soil, commonly used as heterologous host for production of proteins in an extracellular mode (51), including the well-studied chitosanase from Streptomyces sp. N174 (CsnN174) (19). Until the publication of the genome sequence of S. coelicolor A3(2) (4) and, more recently, of the S. lividans genomic contigs (GenBank accession no. ACEY00000000), these two closely related species were thought to be devoid of chitosanase activity because they grew very poorly on media with chitosan and no chitosanase activity was detected in their cultures (33).

However, genes encoding putative chitosanases of the GH46 family are present in both genomes: SCO0677 (csnA) and SCO2024 (csnB) in Streptomyces coelicolor A3(2) and the almost identical genes SSPG_06922 (genomic coordinate 7.62 Mb) and SSPG_05520 (genomic coordinate 6.14 Mb) in S. lividans TK24. The biochemical properties of CsnA from S. coelicolor A3(2) have been studied in detail recently (20, 23). In vivo studies performed with S. lividans TK24 have shown that CsnA is produced at a very low level (in the range of milliunits per ml), explaining the lack of chitosanase detection by earlier, less-sensitive techniques. Despite this low expression level, the deletion of csnA resulted in increased sensitivity to the antimicrobial effect of chitosan (20).

While there are numerous reports on biochemical properties of chitosanases, knowledge about the regulation of chitosanase gene expression is very scarce. In contrast, the genetic regulation of the degradation of chitin, the N-acetylated form of chitosan, has been extensively studied in Streptomyces. Members of this genus play an important part in chitin degradation in soil and produce a wide array of chitinases and chitin-binding proteins (8, 47). The regulation of chitinase (chi) gene expression in Streptomyces is rather complex, and as many as four different mechanisms have been identified, some of them linked to more general phenomena such as carbon catabolite repression, antibiotic production, and morphogenesis through the chitin-derived monomer GlcNAc (37, 40, 41). The Cpb1 regulator controls the expression of the chiA gene in S. lividans (18). The two-component system ChiS/ChiR participates to the genetic regulation of chiC gene of S. coelicolor (25, 28). Reg1, the negative regulator of α-amylase genes in S. lividans, seems also to be involved in the genetic regulation of chitinase genes (36, 37). Finally DasR, a member of the HutC/GntR subfamily, regulates the expression of some chitinase genes through interaction with the dre motif in S. coelicolor (10). DasR also has a more global effect on other genes involved in GlcNAc metabolism (10, 41).

In our previous study dedicated to the regulation of chitosanase gene expression in the actinomycete Kitasatospora sp. N106 (an efficient chitosanase producer), we observed an interaction between a palindromic DNA sequence upstream from the chitosanase-encoding gene csnN106 and a protein present in crude cell extracts from cells growing on media with chitosan or GlcN (15). The protein itself remained unidentified, however. In this work, we describe CsnR (chitosanase regulator) from S. lividans as the first protein shown to be directly involved in the transcriptional regulation of chitosanase gene expression. Furthermore, the analysis of fully sequenced genomes indicates that the CsnR-mediated regulatory mechanism is widespread among actinobacteria.

MATERIALS AND METHODS

Bacterial strains, media, and culture conditions.

Escherichia coli DH5α (Invitrogen) was used as the host for cloning and DNA propagation. The methylase-negative mutant E. coli strain ET12567, containing the nontransmissible pUZ8002 plasmid (31), was used as the donor in intergeneric conjugation with the S. lividans recipient. E. coli Rosetta-gami 2 (DE3)(pLysS) (Novagen) was used for recombinant CsnR production. E. coli strains were grown on Luria-Bertani (LB) broth supplemented, when necessary, with 100 μg/ml ampicillin (Ap), 34 μg/ml chloramphenicol (Cm), 500 μg/ml hygromycin (Hm), 50 μg/ml kanamycin (Km), 100 μg/ml spectinomycin (Sm), or 12.5 μg/ml tetracycline (Tet). Standard methods were used for E. coli transformation, plasmid isolation, and DNA manipulation (44). S. lividans TK24 (27) and the isogenic S. lividans ΔcsnR strain (formerly the Δ2657h strain) (14) were used as recipients for transformation or conjugation. They were also used in quantitative PCR (qPCR) assays, while Streptomyces avermitilis MA-4680 was used in reverse transcription (RT)-PCR experiments.

DNA transformation with S. lividans protoplasts, using a rapid small-scale procedure and R5 regeneration medium, was performed as described previously (27). Hm^r colonies were selected after DNA transfer by addition of 5 mg Hm to 2.5 ml of soft agar overlay. Transformants were chosen following two subsequent cycles of purification on solid yeast-malt extract (YME) medium (27) with 250 μg/ml Hm. Intergeneric conjugation was done with S. lividans ΔcsnR spores following a published protocol with slight modifications (27). Approximately 5 × 10⁷ spores of the S. lividans ΔcsnR strain and 5 × 10⁷ cells of E. coli ET12567(pUZ8002) carrying the appropriate plasmid were combined for conjugation. The mixed bacteria were spread on SLM3 agar plates (13) supplemented with 10 mM MgCl₂. Plates were overlaid with 1 ml of sterile water, including 5 mg Sm and 0.5 mg nalidixic acid per plate. Exconjugants were purified on solid YME medium with 200 μg/ml Sm and 25 μg/ml nalidixic acid. Sporulation was obtained by heavy inoculation of plates with SLM3 agar medium. Spores were collected with glass beads and stored in 20% glycerol at −20°C.

Production and purification of CsnR.

The coding sequence of csnR was PCR amplified from S. lividans genomic DNA with the primers EcoRI-csnR and XhoI-csnR (see Table S1 in the supplemental material). The PCR product was digested with EcoRI and XhoI and ligated into pGEX-6P-1 vector (GE Healthcare) digested with the same enzymes, generating pGEX-csnR. This plasmid was used to produce the recombinant CsnR tagged with glutathione S-transferase (GST) at the N terminus. For protein expression, the plasmid was transformed into E. coli Rosetta-gami 2 (DE3)(pLysS). Transformants were selected on LB agar medium with Ap, Cm, and Tet. For production of CsnR, the transformant was grown in 1.25 liters of LB medium with Ap, Cm, and Tet inoculated 1:20 with overnight culture. Cultures were grown at 37°C until the optical density at 600 nm (OD₆₀₀) reached 0.4 to 0.6. Then 0.1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) was added, and cultures were further incubated for 4 h at room temperature with shaking. Bacteria were recovered by centrifugation, and pellets were kept frozen at −80°C. For protein extraction, pellets were thawed for 15 min on ice and suspended in a total volume of 250 ml of phosphate-buffered saline. Then 1 mg/ml lysozyme was added, and the suspension was incubated for 30 min on ice. The suspension was treated by sonication with six rounds of 10-s bursts at 45% amplitude (130 W, 20 kHz) (Vibra-Cell; Sonics and Materials, Inc.) separated by 10-s cooling periods on ice. The lysate was centrifuged for 20 min at 10,000 × g at 4°C. The supernatant (soluble crude extract) was incubated for 1 h at room temperature with 2 mM ATP and 5 mM MgCl₂. All further steps were done at 4°C with cold solutions and centrifugation steps of 1 min at 500 × g. The total volume of soluble crude extract (250 ml) was mixed with 1 ml of a 50% suspension of glutathione Sepharose 4B, divided into 50-ml aliquots, and incubated for 1 h with slight agitation. The suspensions were centrifuged, and the supernatants were transferred for a second round of binding with fresh resin. Pelleted resin was washed four times with 1.4 ml of phosphate-buffered saline (PBS) and two times with cleavage buffer (50 mM Tris-Cl, 150 mM NaCl, 1 mM EDTA, 1 mM dl-dithiothreitol [DTT] [pH 7.0]). For each wash, pelleted resin was incubated 10 min with slight agitation and centrifuged. For each resin aliquot, 16 μl of PreScission protease (GE Healthcare) and 400 μl of cleavage buffer were added to cleave the GST tag from CsnR. Suspensions were pooled and incubated for 4 h at 4°C. The suspension was centrifuged, and the supernatant was saved. The resin pellet was suspended in 600 μl of cleavage buffer, incubated for 10 min, and centrifuged. Both supernatants were pooled and divided into three fractions for size exclusion chromatography. Approximately 500 μl of partially purified CsnR was loaded onto a Superdex 200 10/300 GL column (GE Healthcare), with a mixture of 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, 5% glycerol, 0.5% Tween 20, and 1 mM DTT (pH 7.4) as the elution buffer. After SDS-PAGE analysis, purified fractions were pooled, aliquoted, and frozen at −80°C until use. Identification of the contaminant protein from E. coli was performed by the Proteomics platform of the Quebec Genomics Center, Québec, Canada.

DNase I footprinting.

To obtain end-labeled DNA probes, 30 pmol of the downward primers DF-csnR and DF-csnA (see Table S1 in the supplemental material) was end labeled with [γ-³²P]ATP (3,000 Ci/mmol) (PerkinElmer) and 20 U of T4 polynucleotide kinase and then purified on a G-25 column (GE Healthcare). Approximately 20 pmol of the end-labeled primers was used in 50-μl PCRs. For csnR, pMP302-Δ2657h (14) was used as the template with the XbaI-csnRC primer. For csnA, pFDES-csnA (unpublished) was used as the template with UF-csnA primer. The end-labeled probes from PCR products were purified with the High Pure PCR product purification kit (Roche). DNA binding reaction mixtures (100 μl) contained 20 mM potassium phosphate buffer (pH 6.8), 5 mM MgCl₂, 150 mM KCl, 1 mM β-mercaptoethanol, 20% glycerol, 0.5 μg poly(dI-dC), approximately 20,000 cpm of end-labeled DNA probe, and ∼0.5 nmol of purified CsnR. After 20 min of incubation at room temperature, 30 U of DNase I (Roche) was added to the reaction mixtures. After 90 s (60 s for reactions without protein), reactions were stopped by addition of 15 mM EDTA (pH 7.9). DNA fragments were extracted by phenol-chloroform and precipitated with 0.1 μg/μl yeast tRNA, 0.3 M sodium acetate (pH 5.2), and 2 volumes of isopropanol. Precipitated DNA was washed once with 70% ethanol, dried, and suspended in formamide loading buffer. Sequence reactions were done with end-labeled primers and DNA templates used in PCRs for probe labeling, and the ALFexpress AutoCycle sequencing kit (Amersham Biosciences) according to the manufacturer's recommendations. Samples and sequence reaction mixtures were heated for 5 min at 95°C just before being loaded onto a 6% polyacrylamide sequencing gel. The gel was run, dried, visualized, and analyzed by a PhosphorImager with ImageQuant version 5.2 software (Molecular Dynamics).

EMSA.

For the electrophoretic mobility shift assay (EMSA), pairs of complementary oligonucleotides were annealed, generating double-stranded oligonucleotides csnA-WT (wild type), csnA-M1, csnA-M2, csnA-MM, and csnR-WT (see Table S2 in the supplemental material). Fifty picomoles of double-stranded oligonucleotide was end labeled with [γ-³²P]ATP (3,000 Ci/mmol) and 20 U of T4 polynucleotide kinase and purified on a G-25 column. DNA binding reaction mixtures (24 μl) contained 10 mM HEPES (pH 7.9), 10% glycerol, 0.2 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 0.25 mM DTT, 1 μg poly(dI-dC), and 150 mM KCl. For equilibrium dissociation constant (K_D) determination, various concentrations of labeled csnA-WT or csnR-WT probe (0.1 nM to 1.5 nM) and ∼8.5 pmol of purified CsnR were used. For the determination of the 50% inhibitory concentration (IC₅₀) of the DNA competitors, 0.03 nM labeled csnA-WT probe and ∼8.5 pmol of CsnR were used with various concentrations of competitor double-stranded oligonucleotide (0.1 to 125 nM). For the sugar binding assay, ∼8.5 pmol of CsnR was preincubated with glucose, GlcNAc, galactosamine, GlcN, or chitosan oligosaccharides (GlcN)₂ to (GlcN)₅ at various concentrations (0.00075 to 75 mM) in the binding reaction mixture for 15 min on ice before the addition of labeled csnA-WT probe (0.03 nM). Reaction mixtures were incubated at room temperature for 15 min with the labeled probe and then subjected to electrophoresis at 4°C in a prerun gel (15 min) of 6% polyacrylamide (10 mM Tris, 80 mM glycine, 0.4 mM EDTA [pH 8.3]). Following electrophoresis, gels were dried, and band intensities were visualized with a PhosphorImager and estimated with ImageQuant software (version 5.2). All determinations were done in triplicate. K_D calculations were done with the Michaelis-Menten nonlinear fit (least squares), and the one-site log IC₅₀ nonlinear fit (least squares) was used for IC₅₀ calculations (GraphPad Prism version 5.03 for Windows; GraphPad, San Diego, CA).

Construction of blaL fusions with csnA, csnAMM, and csnR promoters.

The blaL reporter gene originated from plasmid pDML619 (21). The blaL gene was subcloned into the SphI and HindIII restriction sites of pHM8aBΔM, a modified version of the integrative vector pHM8a (14, 35), generating pHM-blaL. The intergenic region (IR) upstream of csnA (IR-csnA) was amplified by PCR with the BamHI-IR-csnA and SphI-IR-csnA primers (see Table S1 in the supplemental material) from S. lividans TK24 genomic DNA. The IR of csnR (IR-csnR) was amplified with the BamHI-IR-csnR and SphI-IR-csnR primers (see Table S1 in the supplemental material). Both IRs were cloned into the BamHI and SphI restriction sites of pHM-blaL, generating pHM-csnA and pHM-csnR. A mutated version of IR-csnA (IR-csnAMM) was obtained by PCR-directed mutagenesis (24) using Fw1-csnAMM, Rv1-csnAMM, Fw2-csnAMM, and Rv2-csnAMM as primers (see Table S1 in the supplemental material) and pHM-csnA as the DNA template. The mutated digested PCR product was cloned into the BamHI and SphI restriction sites of pHM-blaL, generating pHM-csnAMM. All of the insertions in the reporter plasmids were confirmed by sequencing. These integrative plasmids, as well as pHM-blaL (the negative control in the β-lactamase assay), were introduced into S. lividans TK24 and S. lividans ΔcsnR protoplasts by transformation followed by Hm selection.

β-Lactamase assay.

A total of 2 × 10⁸ spores of the S. lividans TK24 or ΔcsnR strains containing the blaL fusions were inoculated into 50 ml of tryptic soy broth (TSB) medium with 200 μg/ml Hm. Cultures were incubated at 30°C for approximately 64 h with shaking. Cultures were centrifuged and then washed with sterile 0.9% saline, and pellets were suspended in 2 volumes of saline. Then 2 ml of this suspension was added to 100 ml of modified M14 medium (M14M) (38), consisting of 0.1% KH₂PO₄, 0.55% K₂HPO₄, 0.14% (NH₄)₂SO₄, and 0.1% trace elements solution (2 mg/ml CoCl₂·7H₂O, 5 mg/ml FeSO₄·7H₂O, 1.6 mg/ml MnSO₄·H₂O, and 1.4 mg/ml ZnSO₄·7H₂O) (pH 6.9). Before use, 0.03% MgSO₄, 0.03% CaCl₂, and 0.5% mannitol or a mixture of 0.125% GlcN and 0.375% chitosan oligomers (a preparation containing approximately equimolar proportions of dimer and trimer) was added to the autoclaved medium as sterile filtered solutions. Cultures were incubated at 30°C with shaking. All cultures were done in duplicate. At 12, 18, 22, and 38 h, 10 ml of each culture was collected and centrifuged. Supernatants were used to assay the β-lactamase activity, and pellets were used to obtain dry weights for biomass estimation. For the assay of β-lactamase activity, 20 μl of diluted supernatant was combined with 160 μl of 100 mM sodium phosphate buffer, 1 mM EDTA (pH 7.0), and 20 μl of 1 mM nitrocefin (Oxoid Limited, England) and incubated at 37°C for 1 h with optical density read each 10 min. Activity was determined from the slope of the progression of optical density at 486 nm, assuming a molar extinction coefficient of hydrolyzed nitrocefin of ε_{486 nm} = 20,500 M⁻¹ cm⁻¹ (Oxoid Limited). One unit of β-lactamase activity was defined as the amount of enzyme that hydrolyzed 1 μmol of nitrocefin in 1 min under the specified conditions.

Genetic complementation of the S. lividans ΔcsnR strain.

The csnR gene coding sequence together with its complete upstream (211-bp) and downstream (106-bp) intergenic regions were PCR amplified from genomic DNA of S. lividans TK24 with primers XbaI-csnRC and EcoRI-csnRC (see Table S1 in the supplemental material). The PCR product was digested and introduced between the XbaI and EcoRI restriction sites of the integrative, conjugative vector pSET152m (30), generating pSETmC. Complementation plasmid was introduced into the S. lividans ΔcsnR strain by intergeneric conjugation. pSET152m vector was used as a negative control. Successful integration of vectors was confirmed by PCR analysis.

qPCR and endpoint RT-PCR.

For quantitative PCR (qPCR) analysis, 2 × 10⁸ spores of the S. lividans TK24, ΔcsnR, ΔcsnR + pSETmC, or ΔcsnR + pSET152m strain were inoculated into 50 ml TSB medium. The cultures were incubated at 30°C for approximately 64 h with shaking. The cultures were centrifuged and then washed with sterile 0.9% saline, and the pellets were suspended in 2 volumes of saline. A total of 1.5 ml of this suspension was used to inoculate 50 ml of M14M (see above) either with mannitol or with 1:3 GlcN-chitosan oligomers. The experiment was done in triplicate. Cultures were incubated at 30°C with shaking. After 14 h, 10 ml of each culture were collected and mixed immediately with stop solution (0.2 volume of 95:5 ethanol-phenol). Samples were centrifuged for 10 min at 4°C. Bacterial pellets were frozen at −80°C until lysis. For RT-PCR experiments, culture conditions were identical to those for qRT-PCR experiments, except that S. avermitilis MA-4680 was grown for 24 h in M14M supplemented with 1% mannitol or 0.2% GlcN and 0.8% chitosan oligomers.

For qPCR and RT-PCR experiments, total RNA extraction was carried out with the Qiagen RNeasy minikit (Qiagen), with the following modifications. Cell disruption was achieved by sonication with two 30-s bursts at 35% amplitude separated by a 15-s cooling period. Sonication was followed by two phenol-chloroform extractions and one chloroform extraction for elimination of cell debris. The on-column DNase treatment was done with the RNase-free DNase set (Qiagen). Additionally, another DNase digestion was done after RNA elution with the Turbo DNA-free kit (Ambion). RNA purity and concentration were assessed in a NanoDrop 1000 spectrophotometer (Thermo Scientific). RNA quality was verified by electrophoresis on agarose gel in MOPS (morpholinepropanesulfonic acid) electrophoresis buffer with 0.22 M formaldehyde (44). Reverse transcription was performed on 2 μg of total RNA with the first-strand cDNA synthesis kit (GE Healthcare) and 72% G+C-rich random hexamers.

Quantitative PCRs were performed in an Mx3000P real-time PCR system (Stratagene). PCR mixtures (20 μl) contained 2 μl of 20× diluted template cDNA, 250 nM the appropriate primer (see Table S3 in the supplemental material), and a SYBR green PCR mix. The PCR conditions were 95°C for 3 min, followed by 40 cycles at 95°C for 15 s, 60°C for 45 s, and 72°C for 15 s. An additional dissociation step (95°C for 1 min, 60°C for 30 s, and 95°C for 30 s) was added to assess nonspecific amplification. PCRs were run in triplicate. The absence of genomic DNA was verified by using samples in which the reverse transcriptase was omitted from the cDNA synthesis reaction. The gyrA and rrn genes of S. lividans (encoding gyrase A and 16S rRNA, respectively) were used as internal controls for relative quantification. Efficiencies of all primer pairs were verified. Raw data were transformed into threshold cycle (C_T) values. Relative gene expression was calculated by the comparative C_T method (39) for each strain incubated in the GlcN-chitosan oligomer medium compared to the mannitol medium.

For RT-PCR experiments with S. avermitilis MA-480, an endpoint PCR followed reverse transcription reactions. PCR mixtures (20 μl) contained 1 μl of 10× diluted template of cDNA, 2 μM each primer (see Table S4 in the supplemental material), 10% dimethyl sulfoxide (DMSO), 1× ThermoPol buffer, and 250 μM deoxynucleoside triphosphates (dNTPs). The PCR conditions were 95°C for 3 min, followed by 35 cycles at 95°C for 30 s, 58.5 to 66.4°C (depending on the set of primers used) and 72°C for 30 s, with a final elongation step at 72° for 10 min. The rpsI gene of S. avermitilis (encoding the 30S ribosomal protein S9) was used as an internal control.

RESULTS

Identification of a candidate gene regulating chitosanase expression.

Palindromic sequences of similar lengths and sharing a high level of identity have been previously described in the upstream segments of several endo- and exochitosanase genes from actinomycetes (11, 32, 33). By EMSA experiments, Dubeau et al. (15) have characterized a DNA-protein interaction between a protein present in partially purified extracts from Kitasatospora sp. N106 and a double-stranded oligonucleotide probe covering the palindromic sequence. A BLAST search with this sequence as the query returned numerous hits, mostly from intergenic regions of actinomycete genomes. Their partial listing was used in an alignment (Fig. 1) and yielded the AGGAAA(G/C)TTTCCTA consensus.

Fig. 1.

Alignment of palindromic sequences found upstream of genes encoding chitosanases or ROK family regulator genes in actinomycetes and LOGO representation of consensus sequence. Complementary bases in palindromes bases are shown in blue. “Pos.” (position) indicates the distance in bp from the central nucleotide of the palindromic sequence to the start codon of the associated gene. K. sp. N106, Kitasatospora sp. N106; S. sp. N174, Streptomyces sp. N174.

The palindromic sequence was found in front of two categories of genes: those encoding studied or putative chitosanases from various families (including the csnA gene, SSPG_06922, from S. lividans) and genes encoding putative transcriptional regulators, all belonging to the ROK family established by Titgemeyer et al. (50): among these are S. lividans gene SSPG_04872, localized at map coordinate 5.42 Mb at a 2.2-Mb distance from csnA. We examined the protein encoded by this gene as a possible candidate for a transcriptional regulator of chitosanase gene expression (CsnR).

Purification of CsnR.

CsnR was overproduced with a GST tag. The majority of the recombinant protein was detected in inclusion bodies (not shown). Attempts to renaturate the insoluble protein were not successful as precipitation during dialysis occurred. Purification was then attempted with the soluble portion of the lysate. A major protein contaminant (∼60 kDa) copurified with GST-CsnR (Fig. 2). This protein was identified by partial sequencing as the chaperone GroEL, known to contaminate several recombinant proteins from E. coli during purification (9, 17). As suggested by Chen et al. (9), the soluble lysate was incubated with 2 mM ATP and 5 mM MgCl₂ before the affinity purification step. This additional step was helpful in eliminating the contaminant (Fig. 2). After an additional size exclusion chromatography step, essentially pure CsnR was obtained, as shown by SDS-PAGE followed by silver nitrate staining (Fig. 2).

Fig. 2.

Purification of CsnR. Protein samples from each stage of the CsnR purification were analyzed by 10% SDS-PAGE and visualized after silver nitrate staining. M, PageRuler prestained molecular mass protein ladder (Fermentas); S, soluble fraction of cell lysate obtained from a culture of E. coli Rosetta-gami 2 (DE3)(pLysS) (pGEX-csnR) induced with 0.1 mM IPTG; (−), purification attempt without previous treatment of the soluble fraction of cell lysate; (+), purification steps with a previous 2 mM ATP and 5 mM MgCl₂ treatment of the soluble fraction of cell lysate; E, eluate collected from the glutathione-Sepharose 4B resin following a 4-h incubation with PreScission protease; F, 20 μl of the size exclusion chromatography fraction with the highest GroEL contamination; P, 20 μl of pooled size exclusion chromatography fractions with purified CsnR.

CsnR binds in vitro to the palindromic sequences upstream of csnA and csnR.

As determined by DNase I footprinting, CsnR binds asymmetrically to the palindromic box found in the promoter region of csnA, covering 15 nucleotides upstream and 12 nucleotides downstream from the palindrome axis (Fig. 3). CsnR binds in a similar way to the palindromic box in the promoter region of its own gene, covering 17 nucleotides upstream and 12 nucleotides downstream from the axis (Fig. 3). As determined by primer extension, the protected region superimposes to the transcriptional start site in csnR gene (Fig. 3B). Despite several attempts, the transcription start site of csnA could not be determined.

Fig. 3.

DNase I footprinting analysis of the CsnR binding site to csnA and csnR promoters. (A) A 298-bp labeled probe (csnA-IR) and a 256-bp labeled probe (csnR-IR), both including the entire intergenic regions upstream from csnA and csnR, respectively, were subjected to partial DNase I digestion in the presence (+) or absence (−) of ∼0.5 nmol of purified CsnR. Vertical arrows correspond to the palindromic sequence shown in panel B. (B) Partial intergenic region sequences upstream of csnA and csnR. Boxes correspond to the protected region in panel A. Arrows correspond to the palindromic sequence. Boldface gtg represents the translation initiation codon. **, transcription initiation site as determined by primer extension. The −35 and −10 boxes of the deduced promoter sequence are shown in italic.

Oligonucleotide probes corresponding to the longest protected segment (17-1-12) were used to characterize the CsnR-DNA interaction by EMSA. The K_D values were 0.032 nM (standard error [SE] = 0.009) and 0.040 nM (SE = 0.008) for the operators of csnA and csnR, respectively (see Fig. S1 in the supplemental material). It appears that CsnR binds to the operators of the chitosanase gene as well as its own gene with similar affinity.

On the basis of our previous experiments performed in vitro with partly purified protein extracts from Kitasatospora sp. N106 and mutated oligonucleotides representing the operator of the chitosanase N106 gene, we concluded that nucleotides at the −2 and +2 positions were critical for binding, while positions −7, −6, +6, and +7 were of moderate importance (15). Accordingly, annealed double-stranded oligonucleotides corresponding to the CsnR target sequence mutated by transversion at positions +2 and/or +6 of the palindrome were used in equilibrium competition experiments against a labeled csnA-WT probe (Table 1). The effect of mutations on binding was estimated from IC₅₀s. The mutation at the +2 position was particularly deleterious for binding (Table 1). Mutation at the +6 position had a lesser effect, and the double mutation seemed to bring a cooperative effect. The doubly mutated oligonucleotide lost most of its affinity for the CsnR protein. This suggests a similarity between the DNA binding mechanism of CsnR from S. lividans and that of the putative chitosanase gene regulator from Kitasatospora sp. N106 (15).

Table 1.

Effect of mutations in the operator sequence on CsnR binding evaluated by equilibrium competition experiments

Name	Sequence (5′→3′) including equilibrium positionsa:	IC50 (nM)b	SE of log IC50b
	−6−20+2+6
csnA-WT	CCTCTTCTGGTAGGAAACTTTCCTATCAGT	1.3	0.11
csnA-M1	CCTCTTCTGGTAGGAAACTGTCCTATCAGT	52.1	0.13
csnA-M2	CCTCTTCTGGTAGGAAACTTTCCGATCAGT	2.7	0.08
csnA-MM	CCTCTTCTGGTAGGAAACTGTCCGATCAGT	105	0.33

^aMutated nucleotides are in boldface.

^bIC₅₀ and standard error of log IC₅₀ values were determined using Graph-Pad Prism software from data compilation of three independent experiments.

DNA binding by CsnR is sensitive to the presence of chitosan oligomers.

Equilibrium competition experiments were also used to determine the ability of various carbohydrates to interfere with DNA binding of CsnR to operator sequence of csnA. At first, EMSA experiments with CsnR showed that 500 nM GlcN, chitosan dimer, and chitosan pentamer strongly affected the gel shift pattern, while glucose and GlcNAc had no effect (Fig. 4). Then detailed IC₅₀ determinations revealed that the chitosan dimer had the strongest effect on the displacement of CsnR from its target, having the lowest IC₅₀ (18.2 nM; SE of log IC₅₀ = 0.05), compared to the chitosan monomer, GlcN (977 nM; SE of log IC₅₀ = 0.05). For higher oligomers, the IC₅₀ increased progressively with their length: 30.6 nM (SE of log IC₅₀ = 0.08) for trimer, 37.3 nM (SE of log IC₅₀ = 0.08) for tetramer, and 154 nM (SE of log IC₅₀ = 0.080) for pentamer. CsnR seems to bind specifically the products of chitosan degradation by chitosanases, as the other tested sugars (glucose, galactosamine, and GlcNAc) do not interfere with CsnR binding to the csnA-WT probe even at the maximal tested concentration of 75 mM in binding reactions (not shown). When undersaturating concentrations of chitosan oligosaccharides were present in the binding reaction mixtures, a band of intermediate mobility appeared in EMSA gels (Fig. 4 and see Fig. S2 in the supplemental material), reflecting the progressive disassembly of the multimeric complex of CsnR with its DNA target.

Fig. 4.

Effect of saccharides on the interaction between CsnR and the csnA-WT operator. The indicated saccharides were added (500 nM) to binding reaction mixtures containing ∼8.5 ρmol of CsnR and 0.03 nM csnA-WT probe. Free and complexed DNA fragments were separated by 6% polyacrylamide gel electrophoresis and visualized by PhosphorImager.

In vivo regulation of csnA and csnR promoters with extracellular β-lactamase reporter gene.

To examine if CsnR is indeed involved in the in vivo transcriptional regulation of its own expression and of the expression of the chitosanase encoded by csnA, promoter regions from both genes were assessed for their activity by using the blaL reporter system (21). The coding sequence of blaL was fused to (i) the upstream intergenic region (IR) of csnA, (ii) a modified version of this IR including a mutated palindrome at the +2 and +6 positions, or (iii) the upstream IR of csnR. All fusion genes were expressed from a single integrated copy in wild-type S. lividans or in a mutant harboring an in-frame deletion of a portion of the csnR coding sequence (the ΔcsnR strain). The mutated ΔcsnR strain does not differ from the wild type in morphology, growth in various media, or actinorhodin production. All strains were grown either in a control medium with mannitol or in the induction medium M14M containing a 1:3 mixture of GlcN and chitosan oligomers (1:3 medium). No BlaL activity was detected in control strains transformed with the pHM-blaL vector. Induction of expression directed by the regions IR-csnA and IR-csnR was observed in the 1:3 medium in the wild-type host (Fig. 5 A and C), and the induction ratios (BlaL activity in 1:3 medium/BlaL activity in mannitol medium) at 16 h were 19.0 and 44.1, respectively. Lack of repression was, however, observed in mannitol medium when the same blaL fusions were introduced in the ΔcsnR host (Fig. 5A and C), and the induction ratios at 16 h dropped to 1.0 for IR-csnA and 1.3 for IR-csnR. CsnR thus regulates negatively the expression of csnA as well as its own expression. Mutations at positions +2 and +6 in the csnA operator resulted in substantial derepression of β-lactamase production in mannitol medium in the wild-type host compared to the 1:3 medium (induction ratio of only 1.6 at 16 h) (Fig. 5B), confirming the weakening effect of these mutations on CsnR binding observed in vitro.

Fig. 5.

β-Lactamase activities of S. lividans TK24 wild-type or ΔcsnR strains harboring various blaL gene fusions. Solid symbols, 0.5% mannitol medium; open symbols, medium containing 0.125% GlcN and 0.375% chitosan oligomers. Triangles, S. lividans TK24; diamonds, S. lividans ΔcsnR strain. (A to C) BlaL activities obtained from the csnA promoter region (A), the mutated csnA promoter region (B), and the csnR promoter region (C). Means were calculated from two independent experiments. Error bars represent standard errors.

CsnR regulates negatively the transcription of csnA and of a gene cluster led by csnR.

Close examination of the genomic sequence of S. lividans reveals that csnR is localized in a gene cluster composed of six genes (Table 2 and see Fig. S3 in the supplemental material). The functions putatively assigned to these genes indicate that the cluster could be dedicated to sugar transport and metabolism (Table 2). The intergenic regions between these genes are very short (the longest region of 106 bp being localized between csnR and csnE), while a much larger region containing a possible transcription terminator consisting of a 14-bp inverted repeat, separates csnK from the following gene, SSPG_04866 (see Fig. S3 in the supplemental material).

Table 2.

Components of csnR-led gene cluster in S. lividans

Gene annotation	Gene namea	Putative function
SSPG_04872	csnR	ROK family transcriptional regulator
SSPG_04871b	csnE	Secreted sugar binding protein
SSPG_04870b	csnF	Sugar transport system permease
SSPG_04869b	csnG	Sugar transport membrane protein
SSPG_04868	csnH	Glycoside hydrolase, family GH4
SSPG_04867	csnK	Sugar kinase

^aNames were adopted as described by Bertram et al. (6).

^bThese three genes determine a putative ABC transporter.

To get more insight into the regulatory mechanism of CsnR, the transcript abundance of various genes was evaluated by qPCR in both wild-type and ΔcsnR strains (Table 3). A mutant strain in which the deletion has been complemented by a wild-type copy of the csnR gene (including its entire 211-bp upstream IR) on an integrative vector (S. lividans ΔcsnR + pSETmC) was examined (Table 3). Data were collected from cultures growing in control medium with mannitol or in M14M with chitosan-derived carbon sources. There was no detectable csnR expression in either the ΔcsnR strain or the control complementation strain (S. lividans ΔcsnR + pSET152m strain), confirming the deletion genotype. We found that the expression of csnR itself was induced more than 100× by chitosan-derived saccharides (Table 3), but, surprisingly, the induction ratio was very low in the complementation strain when csnR was introduced in a different genomic location with the integrative plasmid pSETmC. This appeared to be due to unexpectedly high csnR transcript abundance in the mannitol medium (Table 3). In other words, CsnR failed to autorepress transcription when its own gene (including its operator) was carried by the integrated pSETmC vector. This observation warrants attention in future studies.

Table 3.

Effect of the ΔcsnR mutation and its complementation on transcript abundance patterns of chitosan-related genes in S. lividans strains

S. lividans strain	Medium	Mean ± SE relative transcript abundance and induction ratio ofa:
		csnA	Ratio	csnB	Ratio	csnR	Ratio	csnE	Ratio	csnH	Ratio	SSPG_04866	Ratio
TK24	Control	0.096 ± 0.0032	43	0.023 ± 0.0017	3.0	0.0055 ± 0.0022	156	0.0011 ± 0.000033	1,398	0.26 ± 0.0029	78	0.097 ± 0.0049	3.6
	Induction	4.2 ± 0.67		0.070 ± 0.0054		0.86 ± 0.35		1.6 ± 0.36		2.0 ± 0.23		0.35 ± 0.063
ΔcsnR strain	Control	9.4 ± 0.86	0.74	0.16 ± 0.019	1.9	ND		1.5 ± 0.17	1.5	1.5 ± 0.14	1.7	0.58 ± 0.079	1.0
	Induction	7.0 ± 0.51		0.30 ± 0.022		ND		2.3 ± 0.39		2.4 ± 0.26		0.60 ± 0.020
ΔcsnR + pSETmC strain	Control	0.13 ± 0.0073	23	0.034 ± 0.0068	1.6	0.14 ± 0.018	3.0	0.0016 ± 0.000088	734	0.025 ± 0.0022	100	0.15 ± 0.023	2.4
	Induction	3.0 ± 0.29		0.054 ± 0.0046		0.42 ± 0.14		1.2 ± 0.20	2.5 ± 0.17	0.36 ± 0.048
ΔcsnR + pSET152m strain	Control	1.6 ± 0.26	0.47	0.067 ± 0.0050	0.40	ND	0.96 ± 0.089	1.4	2.0 ± 0.27	1.4	0.11 ± 0.019	0.51
	Induction	0.77 ± 0.15		0.026 ± 0.0041		ND	1.4 ± 0.077		2.8 ± 0.15		0.056 ± 0.017

^aThe values shown are expressed as the relative transcript abundance in the control medium or in induction medium normalized to the transcript abundance of the gyrA gene. Similar values were obtained after normalization to the expression level of rrn (data not shown). The values shown are means ± SEs of three independent cultures with a culture time of 14 h. Induction ratios represent the ratio of transcript abundance in the induction medium to that in control medium. ND, not determined.

For the chitosanase gene, csnA, a 43-fold induction ratio was observed (Table 3), confirming the results obtained with the reporter blaL gene (Fig. 5). Complete derepression of csnA expression was observed in the ΔcsnR strain, and repression was restored by complementation (Table 3). A similar CsnR-dependent expression pattern was observed for genes csnE and csnH localized inside the cluster (Table 3). It is thus probable that this cluster forms a polycistronic transcription unit negatively regulated by CsnR.

A much higher induction ratio was however observed for csnE than for csnR and csnH (Table 3). We sequenced the intergenic region between csnR and csnE and observed that it includes four direct repeats (see Fig. S4 in the supplemental material), a possible site of a regulatory interaction. The presence of additional levels of regulation inside the operons of S. coelicolor seems to be frequent (29), as the gene expression patterns in S. coelicolor operons do not follow the “equal expression” rule often observed in E. coli.

Transcript abundance of SSPG_04866, the gene following csnK, did not follow a CsnR-dependent pattern (Table 3). SSPG_04866 putatively encodes a secreted protein of unknown function and does not seem to belong functionally to the csnR to -K gene cluster. An extensive inverted repeat localized in the IR following csnK could function as a transcription terminator (see Fig. S3 in the supplemental material).

While this work provides evidence that CsnR is the transcriptional repressor of the chitosanase gene csnA, the uniform expression pattern observed for its homolog, csnB (SSPG_05520), indicated that csnB is not regulated by CsnR (Table 3). This was somewhat expected, as the palindromic sequence recognized by CsnR was not found in the genomic environment of csnB.

The CsnR-mediated regulatory mechanism is widespread in actinobacteria.

After the identification of the csnR to -K gene cluster in S. lividans, a bioinformatic search was performed to establish if similar gene clusters are present in other fully or partly assembled genomes. So far, orthologs of CsnR with no less than 46% identity have been found in 23 genomes of actinobacteria, and the presence of a highly similar gene cluster of six genes has been confirmed in 12 genomes, including actinobacteria other than streptomycetes, such as Saccharopolyspora erythraea NRRL2338, Streptosporangium roseum DSM43021, and Kribbella flavida DSM17836. Table 4 shows the cluster annotation in some streptomycete species in which a palindromic box corresponding to the CsnR consensus presented in Fig. 1 is present upstream of the gene cluster. All the putative chitosanase genes belonging to well-established glycoside hydrolase (GH) families are also listed (Table 4). While the distributions of members of various GH families differ among the analyzed species, it is noteworthy that each genome includes at least one putative chitosanase gene provided with a CsnR box.

Table 4.

CsnR gene clusters and putative chitosanase genes in sequenced Streptomyces genomes

Species and strain	CsnR cluster componentsa	Confirmed or putative endo- or exochitinase gene
		GH2		GH5		GH46		GH75
		Gene	Box	Gene	Box	Gene	Box	Gene	Box
S. lividans TK24	SSPG_04872–SSPG_04867					SSPG_06922c	+	SSPG_00778	−
						SSPG_05520	−
S. coelicolor A3(2)	SCO2657–SCO2662					SCO0677c	+	SCO7070	−
						SCO2024	−
S. avermitilis MA-4680	SAV_5384–SAV_5379	SAV_1223c	+			SAV_2015	+	SAV_1850	+
						SAV_6191	−	SAV_1288	−
S. scabies 87.22	SCAB_59491–SCAB_59441			SCAB_86311	+			SCAB_83781	−
S. griseus IFO 13350	SGR_4874–SGR_4869			SGR_1341c	+			SGR_1238	−
S. sviceus ATCC 29083	SSEG_04515–SSEG_04514					SSEG_02093	−	SSEG_10562	−
	SSEG_09506–SSEG_09503b			SSEG_10482	+
S. clavuligerus ATCC 27064	SCLAV_1826–SCLAV_1831			SCLAV_5580	+	SCLAV_4996	−	SCLAV_5034	−
S. pristinaespiralis ATCC 25486	SSDG_02817–SSDG_02822			SSDG_05015	+	SSDG_00156	−	SSDG_04141	−
						SSDG_03879	−

^aAll clusters begin with a csnR ortholog having the palindromic box in the upstream segment.

^bThese six genes form an uninterrupted cluster even if their numbers do not follow each other.

^cThese genes encode enzymes the identities of which as chitosanases have been confirmed by biochemical studies.

Among the analyzed species, S. avermitilis stands out for its highest number of putative chitosanase genes, belonging to three different families. The transcriptional behavior of all these putative chitosanases in the absence or the presence of chitosan oligomers was thus compared by endpoint RT-PCR. Induction with chitosan-derived oligosaccharides was observed only for the three putative chitosanase genes having a CsnR-type operator (Fig. 6).

Fig. 6.

RT-PCR expression profiling of putative chitosanase genes belonging to families GH2 (SAV_1223), GH46 (SAV_2015 and SAV_6161), and GH75 (SAV_1288 and SAV_1850) in S. avermitilis grown in the absence (−) or presence (+) of chitosan oligosaccharides. Expression of the SAV_4958 (rps1) gene was used as an internal control. Asterisks indicate chitosanase genes with the CsnR box.

Our data indicate that the regulatory mechanism mediated by CsnR is an evolutionary ancient mechanism of chitosanase gene regulation present in many actinobacteria and not limited to the GH46 family in which it was discovered, but extends to other chitosanase families as well.

DISCUSSION

This work describes the identification and characterization of CsnR, a novel chitosanase gene regulator in bacteria and also the first characterized transcriptional regulator of the ROK family in actinobacteria.

DNase footprinting and EMSA experiments demonstrated that CsnR binds directly to the palindromic box found upstream from the csnA and csnR gene cluster. This binding target is different from the operators characterized for other transcriptional regulators belonging to the ROK family. The CsnR box is tightly organized around the symmetry axis, the positions −2 and +2 being most critical for binding (15; this work). In contrast, the operator consensus sequences of NagC and Mlc of E. coli and XylR of Firmicutes are essentially composed of two A/T-rich inverted repeats separated by a 5- to 9-bp spacer (16, 22) with strictly conserved positions ±6 and ±5.

Equilibrium competition experiments showed that (GlcN)₂ is the preferential CsnR ligand molecule. This dimer is the major product obtained from chitosan hydrolysis by endochitosanases (33). Oligosaccharide products resulting from the hydrolysis of polymers catalyzed by endohydrolases were often described as the effectors for the transcriptional regulation of glycoside hydrolase genes in actinomycetes. Cellopentaose is the inducer of CebR, the repressor of cel1 in Streptomyces reticuli (46). Maltopentaose is the inducer of MalR, the repressor of genes coding for α-amylase in Streptomyces coelicolor A3(2) (45).

In vivo experiments performed using the blaL reporter gene (Fig. 5) and confirmed by direct estimation of transcript abundance with qPCR (Table 3) showed that CsnR is subject to autorepression. Rosenfeld et al. (43) showed that a system controlled by negative autoregulation offers the advantage of faster response to the presence of inducer molecules over a system controlled by an open loop regulator. Also, negative autoregulation ensures a more homogenous distribution of the steady-state level of the repressor between cells in a population (3).

In S. lividans, S. coelicolor A3(2), and several other actinobacteria, CsnR is localized at the beginning of a gene cluster (csnREFGHK) including an ABC transporter, a glycoside hydrolase, and a sugar kinase. Previously, Bertram et al. (6) described close orthologs of csnEFG in the S. coelicolor A3(2) genome (SCO2658 to SCO2660; localized on the SC6D10.01 cosmid) found by in silico analysis of carbohydrate uptake systems. Trehalose, maltose, and lactose were cited as possible substrates for this uptake system (6). Our study indicates that the transcription of genes localized in this cluster is induced by chitosan oligosaccharides and that they share a negative regulatory mechanism with the chitosanase gene, csnA. We suggest that this cluster represents an operon-like structure involved in the uptake, transport and intracellular metabolism of oligosaccharides resulting from the hydrolysis of chitosan (or N-deacetylated segments of chitin) by chitosanases.

Gene expression driven by (GlcN)₂ has been studied only in Vibrio cholerae, a member of a genus largely responsible for the turnover of carbon and nitrogen from chitin in the marine environment (26). By microarray profiling, seven genes for which expression was increased in the presence of (GlcN)₂ have been identified, all grouped in one cluster (VC1280 to -1286). VC1281 to -1286 form an operon, while VC1280 is orientated in the opposite direction from the other six genes (5, 34). The cluster includes a putative phosphotransferase system (PTS) transporter (VC1281 to -1283) and a LacI family regulator (VC1286) which functions as a transcriptional repressor of this operon (5). The operator sequence recognized by this repressor has not been characterized so far. Except for VC1284, encoding a putative GH4 family glycoside hydrolase which is an ortholog of csnH, the cluster is thus formed by genes which are not similar to those identified in the actinobacterial gene clusters.

As shown in Table 4, gene clusters highly similar to csnREFGHK of S. lividans were found in several other actinobacterial genomes. However, the types and numbers of chitosanase genes putatively regulated by CsnR varied among species, and the majority of putative chitosanase genes were devoid of CsnR-recognized operator. Further studies are needed to establish the biochemical and functional relationships between the chitosanases belonging to the CsnR regulon and the other putative chitosanases.