Small RNA mediated repression of subtilisin production in Bacillus licheniformis

The species Bacillus licheniformis includes important strains that are used in industrial production processes. Currently the physiological model used to adapt these processes is based on the closely related model organism B. subtilis. In this study we found that both organisms reveal significant differences in the regulation of subtilisin, their main natural protease and a product of industrial fermentation processes. We identified and characterized a novel antisense sRNA AprAs, which represents an RNA based repressor of apr, the gene encoding for the industrial relevant subtilisin protease. Reduction of the AprAs level leads to an enhanced proteolytic activity and an increase of Apr protein expression in the mutant strain. A vector based complementation of the AprAs deficient mutant confirmed this effect and demonstrated the necessity of cis transcription for full efficiency. A comparative analysis of the corresponding genome loci from B. licheniformis and B. subtilis revealed the absence of an aprAs promoter in B. subtilis and indicates that AprAs is a B. licheniformis species specific phenomenon. The discovery of AprAs is of great biotechnological interest since subtilisin Carlsberg is one of the main products of industrial fermentation by B. licheniformis.

For the closely related organism B. subtilis no comparable antisense RNA associated with the 3′-end of aprE was described, although extensive transcriptome analyses were performed by Rasmussen et al. in 2009 29 and Nicolas et al. in 2012 30 . The subtilisin gene aprE in B. subtilis was found to be regulated by the interacting global transcriptional regulators CodY and ScoC as well as by AbrB 31 . Barbieri et al. 31 could also show that the regulatory network in B. subtilis results in a repression of the aprE gene within the exponential growth phase and an exclusive expression of AprE within the stationary phase. The genome of B. licheniformis DSM13 also exhibits orthologues of aprE regulators 23,24 and the transcriptome analysis of B. licheniformis DSM13 25 showed a corresponding expression pattern for apr. However, the association of the apr gene with a highly expressed small antisense RNA indicates an additional regulatory layer in B. licheniformis.
The here presented work focuses on the impact of the newly discovered antisense RNA AprAs. By inactivation of the aprAs promoter and vector encoded transcription of AprAs we could show its regulatory effect on proteolytic activity and Apr expression.

Construction and characterization of an AprAs deficient mutant.
To investigate the impact of the 144 base antisense sRNA AprAs (Bli_r0872 in Wiegand et al. 25 ), an AprAs deficient B. licheniformis MW3 mutant was generated. A direct deletion of aprAs was not possible due to its overlap with the apr gene on the complementary DNA strand (see Fig. 1a). However, the previous RNA-Seq experiment 25 also included differential RNA-Seq (dRNA-Seq, see Sharma et al. 32 ) and thus, the transcription start site of aprAs and a corresponding promoter pattern could be identified. The SigA pattern was located on the positive DNA strand between the stop codon of the apr coding region and its transcription terminator (see Fig. 1b). In order to prevent the transcription of AprAs, the AT-rich region of the −10 promoter box was replaced with the GC-rich recognition pattern of the AscI endonuclease. For easy detection of the desired mutation, a 2.6 kbp PCR product was amplified from the respective genomic region and digested with the endonuclease AscI. The mutant strain B. licheniformis MW3 AprAs − exhibited the expected 1.3 kbp DNA band consisting of the degradation products of the 2.6 kbp PCR product (see Supplementary Figure S1). A slot blot experiment confirmed that no transcriptional activity of the aprAs gene was detectable in the mutant strain compared to the original strain B. licheniformis MW3 (see Fig. 2).
To investigate the impact of AprAs on the protease activity of B. licheniformis, the AprAs − mutant and the original strain MW3 were grown on M9 skim milk agar plates. B. licheniformis MW3 AprAs − exhibited increased halo formation, indicative for a stronger proteolytic activity compared to strain MW3 (Fig. 3a). These qualitative results were confirmed by quantitative non-specific exoprotease assays using 24 h M9 skim milk liquid cultures. Here, the AprAs deficient strain B. licheniformis MW3 AprAs − showed an approximately four fold increased exoprotease activity compared to B. licheniformis MW3 (see Fig. 3b).  25 . Transcriptome data were visualised in a logarithmic scale using TraV 56 . The red graph represents the transcription activity of aprAs and the blue of apr. The black arrow on the complementary strand represents the coding region of apr and the yellow arrow of aprAs. The dark red arrows represent the identified promoters and green arrows predicted transcription terminators. (b) Promoter region of aprAs. The −10 box (TAGAAT) of the potential SigA promoter is highlighted in red and the exchanged sequence (AscI pattern) is given. Predicted transcription terminators are framed with green arrows. Protein coding sequences are shown in black bold letters and the stop codons are underlined. The aprAs coding sequence is given in orange letters and the experimentally determined transcription start site of AprAs 25 is framed in red. (c) AprAs sequence. The 144 base large AprAs sequence was analyzed via the RNAfold WebServer, using "The Vienna RNA Websuite" 55 . The calculation of secondary structures was possible and the optimal structure with a minimum free energy of −51.9 kcal/mol is shown in dot-bracket notation.
Scientific RepoRts | 7: 5699 | DOI:10.1038/s41598-017-05628-y Verification of AprAs phenotype. B. licheniformis MW3 AprAs − was complemented with a vector encoded aprAs gene to restore the original phenotype and confirm that the observed exoprotease phenotype of B. licheniformis MW3 AprAs − is based on the inactivation of AprAs transcription. The aprAs gene with its native promoter was cloned into the Bacillus/E. coli shuttle vector pV2 33 , creating pV2::aprAs. The vector was introduced into B. licheniformis MW3 AprAs − and the original strain B. licheniformis MW3. The vector pV2 was used as a control. The slot blot analysis (Fig. 2) confirmed the re-establishment of AprAs transcription in B. licheniformis MW3 AprAs − pV2::aprAs. B. licheniformis MW3 pV2 exhibited AprAs transcription due to the chromosomal aprAs gene. B. licheniformis MW3 pV2::aprAs showed an increased AprAs transcription due to expression from the chromosomal aprAs gene in addition to the multi-copy vector encoded aprAs gene.
The qualitative evaluation of the protease activity on M9 skim milk agar plates (Fig. 3c) showed that the complemented mutant strain B. licheniformis MW3 AprAs − pV2::aprAs could re-establish the "wild type" phenotype. Its halo formation was comparable to B. licheniformis MW3 pV2 and reduced in comparison to B. licheniformis MW3 AprAs − . These qualitative results were confirmed by the quantitative exoprotease activity evaluation (Fig. 3d), determined in the supernatant of 4 ml culture in liquid M9 skim milk media. The reduction of AprAs transcription in B. licheniformis MW3 AprAs − resulted once more in an approximately 4 times increased exoprotease activity. This effect was reversed by the ectopic AprAs transcription in B. licheniformis MW3 AprAs − pV2::aprAs. The in trans overexpression of AprAs in the original strain MW3 resulted in an additional reduction of exoprotease activity of approximately 50% (see B. licheniformis MW3 pV2::aprAs).

Impact of AprAs on Apr expression.
In order to investigate the correlation between the AprAs level and the expression of the protease Apr, the extracellular proteomes of the original strain MW3, the AprAs − mutant and the complemented strain were analysed by 2D-gelelectrophoresis and MALDI-TOF-MS/MS 34 . All strains were grown in 400 ml liquid M9 skim milk medium and supernatants were harvested after 12 h, 24 h, 36 h and 48 h. The quantitative exoprotease assay was performed using the supernatants of three experiments (Fig. 4a). The overall proteolytic activity of the strains increased constantly from 12 h to 48 h. The AprAs deficient mutant strain B. licheniformis MW3 AprAs − showed the highest activity levels with approximately 0.8 U/ml after 48 h. The initial strain B. licheniformis MW3 and the vector complemented strain MW3 AprAs − pV2::aprAs showed both lower activity levels with the highest level of approximately 0.5 U/ml. The samples with the highest activity levels (48 h) were analysed in triplicates by 2D-gelelectrophoresis. Three Apr spots were identified in the supernatants of the cultures.   30 . To find out if aprAs and especially its promoter region are conserved, we performed a comparative sequence analysis. The comparison (Fig. 5) comprised the intergenic region between yfhN and apr respectively aprE, with a special focus on the −10 promoter box (green coloration), which was experimentally verified to be responsible for the transcription of AprAs in B. licheniformis DSM13 in the present investigation (see also Fig. 1). The comparison of B. licheniformis DSM13 and B. subtilis 168 showed that the aprAs promoter box (green coloration) is missing in B. subtilis 168. (Fig. 5a). To clarify if AprAs is a strain or species specific phenomenon, we also aligned the concerning regions of 17 B. licheniformis strains (see Fig. 5b) and 35 B. subtilis strains (see Fig. 5c). The sequence comparison showed that the aprAs promoter is present in all investigated B. licheniformis strains but not in the compared B. subtilis strains.

Discussion
Our results clearly reveal the small RNA AprAs as a repressor of the protease Apr. The protease activity measurements also demonstrated the necessity of cis transcription of AprAs in relation to apr for highly efficient repression. The observed protease phenotypes of the initial strain and the complemented strain B. licheniformis MW3 AprAs − pV2::aprAs were on the same level, showing that the transcription of one aprAs gene in cis amounts to approximately 50 copies 33,35 in trans. The combination of the single cis encoded aprAs gene and the multi-copy vector in B. licheniformis MW3 pV2::aprAs reduced the exoprotease activity even 50% below the initial level (Fig. 3c,d).
The exoproteome analysis (Fig. 4) showed the direct correlation between the observed protease related phenotype and the expression of the protease Apr. The comparison of MW3 and mutant proteomes revealed Apr as the only differentially expressed extracellular protease (Supplementary Figure S2). Protein spots of other known extracellular proteases 36, 37 did not change in relation to the AprAs level. Thus, the observed protease phenotype was facilitated by the negative regulation of AprAs on Apr expression. Although the Apr protein expression levels in B. licheniformis MW3 and MW3 AprAs − pV2::aprAs did not exactly correspond to the levels of the proteolytic activity, both analyses clearly indicate AprAs to be a repressor of Apr expression. Antisense RNA regulation on protein expression in Bacilli has been described before 27 . However, this is, to our knowledge, the first time that an RNA repressor of a biotechnologically relevant product has been reported.
A requirement of the proteome analysis was the scale up of our experiment from 4 ml test tubes to 400 ml flask culture. Interestingly, after scaling up, we observed only a doubling of the proteolytic activity (Fig. 4a) in contrast to the four fold increased activity observed in the 4 ml volume experiment ( Fig. 3b and d). Most likely, these differences result from the adaption of the experimental design, which can have an impact on productivity 38 , but might also reflect additional regulatory layers. Investigations on the orthologous gene aprE from B. subtilis 168 revealed a complex regulatory network of the subtilisin protease including, as mentioned before, regulators such as ScoC, SinR, AbrB, DegU and further associated proteins 31,39 . The genome of B. licheniformis DSM13 23, 24 encodes orthologues of these regulators, thus a regulatory network of similar complexity can be assumed.
In case of AprAs, no orthologous sRNA was identified in B. subtilis 168 29,30 . Our sequence comparison of the promoter box of aprAs showed its presence in all investigated B. licheniformis strains and its absence in all investigated B. subtilis strains. Therefore we assume that AprAs is a species specific sRNA and thus, our results demonstrate a clear difference on subtilisin protease regulation between B. licheniformis and B. subtilis. It remains unclear if a similar regulatory layer on the subtilisin protease exists in other B. subtilis species complex members or if it is unique for B. licheniformis.
Transcriptome analysis in B. subtilis and S. aureus revealed that antisense activities can arise from inefficient termination of the sense transcription or from spurious initiation by alternative sigma factors 30,40 . However, the strong transcriptional activity and the distinct transcription start site as well as the presence of a conserved SigA promoter pattern (Figs 1 and 5) support the hypothesis that AprAs is a real non-coding RNA and therefore a relevant regulatory layer of Apr expression.
The regulation by AprAs could involve a pairing of the small RNA and its mRNA target leading to a guided degradation, as has been discussed for antisense regulators in Thomason and Storz 41 and in Desnoyers et al. 42 . The transcription of AprAs is three orders of magnitude stronger than the transcription of its target RNA 25 , which resembles viral toxin/antitoxin type I systems 43 , where the sense/antisense duplex is guided to degradation by RNaseIII 44 . Apparently, a quantitative titration of the target mRNA by the small RNA is involved in the regulatory mechanism. This is supported by the observation that in Real-time PCR analysis the apr mRNA level varies in  Figure S3). However, since a cis encoded aprAs is more effective than multiple vector encoded gene copies, an additional regulatory layer is possible. Figure 1c shows that AprAs might form a secondary structure which could delay the interaction of the sRNA and target mRNA. Hence, the transcription of AprAs close to its target mRNA could be an advantage that may explain the differences in efficiency between cis and vector encoded AprAs.
It has been shown that the production of subtilisin in B. subtilis starts at the beginning of the stationary phase 31 . In contrast, Wiegand et al. 25 observed a delay between the beginning of the transcriptional activity of the apr gene at the end of the exponential growth phase and the actual occurrence of the Apr protease spot in the proteome of B. licheniformis. It is tempting to assume that an AprAs guided degradation of the apr mRNA may be responsible for this delay between transcription and expression of the apr gene. However, life time evaluations of small RNAs in a complex industrial high density medium are heavily challenged by the background RNase activity within the fermentation process.
The production of subtilisin proteases in Bacilli is of great biotechnological relevance since these production systems are extensively used to produce the main enzymatic components of household detergents 8 . Thus, a fourfold increase of the subtilisin activity, as has been achieved with the B. licheniformis wild type subtilisin gene in the AprAs − mutant, is very promising. Although apr orthologues have been identified in most genomes of members of the B. subtilis species complex, the presence of AprAs-like RNA repressors seems less ubiquitous. AprAs was not identified in B. subtilis 168 29,30 or in the 35 investigated B. subtilis genomes (Fig. 5). However, the possibility to achieve an increased subtilisin production by an AprAs − mutation in B. licheniformis has been demonstrated in our study and should be evaluated for B. licheniformis based fermentations of subtilisin-like enzymes.

Conclusion and outlook
The newly identified antisense sRNA AprAs was shown to be a negative regulator of the biotechnologically important subtilisin family protease Apr in the type strain Bacillus licheniformis DSM13. The prevention of AprAs transcription led to an up to 4-fold increase of exoprotease activity as a result of the increased Apr protein expression. This effect was reversed by ectopic overexpression of AprAs and correspondingly, the native AprAs transcription in addition to the ectopic AprAs expression led to a decrease of proteolytic activity, even below the natural level. AprAs represents a new regulatory feature of Apr expression in B. licheniformis which is not present in the model organism and its close relative B. subtilis 168. Sequence comparison revealed its presence in all investigated B. licheniformis strains and its absence in all compared B. subtilis strains, leading to the conclusion of a species-specific feature. Further analyses should focus on revealing the detailed mechanism of AprAs regulation and the search for AprAs homologues in other Bacilli and industrially used apr gene loci.

Material and Methods
Bacterial strains and culture conditions. The bacterial strains used in this study are listed in Table 1.
If not stated otherwise, the strains were grown in NB medium at 37 °C. M9 medium was prepared as described by Sambrook and Russell 45 . 1 L of M9 medium was additionally supplemented with 100 µl of SL-8 trace element solution as described by Atlas 46 and 100 µl vitamin solution. The vitamin solution contained per litre 50 mg pantothenic acid, 50 mg riboflavin, 10 mg pyridoxamine-HCl, 20 mg biotin, 20 mg folic acid, 25 mg niacin, 25 mg nicotinamide, 50 mg α-aminobenzoic acid, 50 mg thiamine-HCl and 50 mg cobalamine dissolved in H 2 O. M9 medium was supplemented with a final concentration of 0.1% (w/v) sterile skim milk to generate M9 skim milk medium. For medium solidification agar with a final concentration of 1.5% (w/v) was added prior to sterilisation by autoclaving. Table 2 and plasmids in Table 1. PCRs (50 μl) consisted of 200 μM deoxynucleotides, 100 ng of template DNA, 5 pmol of each primer and 0.5 U Phusion High-Fidelity DNA Polymerase (Thermo Scientific, Darmstadt, Germany). PCR products were purified directly using the PCR Purification Kit (Qiagen, Hilden, Germany) or after gel electrophoresis using a Qiaquick Gel extraction Kit (Qiagen, Hilden, Germany). DNA was analysed using a TAE agarose gel system as described elsewhere 42 and stained with Ethidium bromide (1 µg/mL) for 10 min. Vectors were constructed as described previously 30 using Escherichia coli TOP10 (Invitrogen, Carlsbad, USA) and isolated using the QIAprep Spin Miniprep Kit (Qiagen, Hilden, Germany). For sequence verification Sanger-sequencing was performed on an ABI3730XL capillary sequencer (BigDye 3.1 chemistry; Applied Biosystems, Darmstadt, Germany). For construction of the B. licheniformis plasmid harbouring strains, pV2::aprAs or the empty pV2 were transferred via conjugation using Escherichia coli S17-1 49 . The generated strains were controlled via plasmid re-isolation and restriction analysis. Slot Blot analysis. Cells were collected at different time points and disrupted with a Mikrodismembrator U (B. Braun Biotech International GmbH, Melsungen, Germany). Total RNA was prepared using the RNeasy Mini Kit (Quiagen, Hilden, Germany). The RNA quality was analysed using the Agilent 2100 Bioanalyser and the Agilent RNA 6000 Nano ladder (Agilent Technologies, Waldbronn). Digoxigenin labelled RNA probes were prepared by in vitro transcription with T7 RNA polymerase using the DIG Northern Starter Kit (Roche, Basel, Switzerland) and templates for in vitro transcription were generated by PCR using the primer pair HLRH308/309. Total RNA was diluted in 10x SSC to a concentration of 0.5 μg/100 μl and blotted on to a positively charged Nylon membrane (Roche, Basel, Switzerland) using the Bio-Dot SF microfiltration unit (BIO-RAD Laboratories GmbH, Munich, Germany). Subsequently, the RNA was covalently bound by exposing the membrane to UV-light (302 nm) for 120 sec on a UV-light table (Image Quant 100, GE Healthcare, Little Chalfont, UK). RNA probe hybridization was performed using the DIG Northern Starter Kit (Roche, Basel, Switzerland) following the manufacturer's instructions and detection was accomplished via the ChemoCam Imager (Intes, Göttingen, Germany).

PCR, gel electrophoresis and vector construction. Primers used in this study are listed in
Real-time PCR. Reverse transcription of 100 ng total RNA from each sample was performed using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, Darmstadt, Germany). For quantification of the apr cDNA the Real-time PCR Thermal Cycler IQ5 (Biorad, Munich, Germany) was used in combination with the QuaniTect SYBR ® Green PCR Kit (Qiagen, Hilden, Germany) and the apr specific primer pair apr fwd/apr rev. For absolute quantification of apr transcripts a standard curve was calculated using vector pV2::aprAs with concentrations of 10 1 to 10 8 copies/µl. Protease assay. For exoprotease activity determination cells were cultivated 12-48 h in M9 skim milk medium and pelleted for 2 min at 16,200× g at 4 °C. The supernatants containing the exoprotease were analysed with Sigmas' non-specific protease assay 50 . The protease activity for each sample was determined in triplicates using 2 mL Eppendorf tubes. Each sample contained one blank as reference.  2D-PAGE, gel image analysis and protein identification. Commercially available IPG strips (18 cm long, Serva, Heidelberg, Germany) in the pH-range 3-10 were used for isoelectric focusing (IEF). 100 µg protein was adjusted to 306 μL with a solution containing 2 M thiourea and 8 M urea. CHAPS solution (20 mM DTT, 1% w/v CHAPS, 0.5% v/v Pharmalyte, pH 3-10) was added (34 µL for each sample). IEF strips were rehydrated with this solution over night. IEF was performed with the following program: step 1, 150 V for 150 Vh, step 2, 300 V for 300 Vh, step 3, 600 V for 600 Vh, step 4, 1500 V for 1500 Vh, step 5, 3000 V for 37.5 kVh. Following IEF, strips were equilibrated as described by Görg et al. 52 . Flat top gels (2D HPE TM Large Gel NF-12.5%; Serva, Heidelberg, Germany) were used for protein separation in the second dimension. Gels were stained with Lava Purple (Fluorotechnics, Sydney, Australia) according to the instructions of the manufacturer. Gel images were analyzed with the Delta2D software (Decodon, Greifswald, Germany). Spot quantification was done according to Wolf et al. 34 . Briefly, gel images (three replicates) from the wild type samples were overlaid with the gels from the mutant samples. Fusion gels created with the image fusion function of the Delta2D software set to union fusion were used for spot detection. Spots were edited and transferred to the single gels. For spot quantification the % volume of each spot was calculated representing the relative portion of an individual spot of the total protein present on the gel. Proteins were excised from the gels using the Ettan Spot Picker (GE Healthcare, Little Chalfont, UK). Digestion and spotting onto MALDI targets were performed in the Ettan Spot Handling Workstation (GE Healthcare, Little Chalfont, UK). MALDI-TOF-MS/MS with the Proteome Analyzer 4800 (Applied Biosystems, Darmstadt, Germany) was performed as described by Wolf et al. 34  Sequence analysis. For identification of the putative aprAs promoter region in 17 B. licheniformis strains and B. subtilis 168, the intergenic region between yhfN and apr from B licheniformis DSM13 was used to perform a blastn search. Correspondingly, the intergenic region between yhfN and aprE from B. subtilis 168 was used as query for a blastn search on 35 B. subtilis genomes. The identified regions were extracted from the respective genomes and used for a comparative sequence alignment performed with MUSCLE 53 using default parameters.
Transcriptional terminators were predicted with the program TransTermHP 54 . Secondary structure predictions were performed using "The Vienna RNA Websuite" 55 .