Hfq regulates antibacterial antibiotic biosynthesis and extracellular lytic-enzyme production in Lysobacter enzymogenes OH11

Lysobacter enzymogenes is an important biocontrol agent with the ability to produce a variety of lytic enzymes and novel antibiotics. Little is known about their regulatory mechanisms. Understanding these will be helpful for improving biocontrol of crop diseases and potential medical application. In the present study, we generated an hfq (encoding a putative ribonucleic acid chaperone) deletion mutant, and then utilized a new genomic marker-free method to construct an hfq-complemented strain. We showed for the first time that Hfq played a pleiotropic role in regulating the antibacterial antibiotic biosynthesis and extracellular lytic enzyme activity in L. enzymogenes. Mutation of hfq significantly increased the yield of WAP-8294A2 (an antibacterial antibiotic) as well as the transcription of its key biosynthetic gene, waps1. However, inactivation of hfq almost abolished the extracellular chitinase activity and remarkably decreased the activity of both extracellular protease and cellulase in L. enzymogenes. We further showed that the regulation of hfq in extracellular chitinase production was in part through the impairment of the secretion of chitinase A. Collectively, our results reveal the regulatory roles of hfq in antibiotic metabolite and extracellular lytic enzymes in the underexplored genus of Lysobacter.


Introduction
Lysobacter is a genus in the family of Xanthomonadaceae and is one of the most ubiquitous environmental microorganisms (Christensen and Cook, 1978). Lysobacter enzymogenes of the genus is the best characterized species. This species is known for its ability to produce a variety of extracellular lytic enzymes, including chitinase, cellulase and protease (Kobayashi et al., 2005). These enzymes are able to destroy the cell wall of pathogenic fungi and oomycetes and are highly linked to biocontrol activity of L. enzymogenes against crop pathogens (Zhang and Yuen, 2000;Palumbo et al., 2003). Previously, efforts had been devoted to cloning the genes encoding the lytic enzymes (Epstein and Wensink, 1988;Zhang and Yuen, 2000;Palumbo et al., 2003), but little is known about the regulation of the lytic-enzyme production or secretion in L. enzymogenes, except the global regulator Clp (Kobayashi et al., 2005;Wang et al., 2014).
In addition to producing abundant lytic enzymes, L. enzymogenes produces diverse bioactive natural products and is recently emerging as a new source of antibiotics, such as the antibacterial WAP-8294A2 (Kato et al., 1998;Zhang et al., 2011;Xie et al., 2012). WAP-8294A2 is a cyclic lipodepsipeptide with a strong activity against Gram-positive bacteria, including methicillin-resistant Staphylococcus aureus (Zhang et al., 2011). However, L. enzymogenes produces a low yield of this antibiotic under the common growth condition (Zhang et al., 2011). Meanwhile, the complex chemical structure of WAP-8294A2 makes it extremely challenging for chemical synthesis. We have recently identified the gene cluster responsible for the biosynthesis of WAP-8294A2 in L. enzymogenes (Zhang et al., 2011). Among the gene, waps1, encoding a typical non-ribosomal peptide synthetase, is a key gene for WAP-8294A2 biosynthesis (Zhang et al., 2011). The molecular biosynthetic mechanism for WAP-8294A2 production has been proposed (Zhang et al., 2011). Little is known, however, about the molecular mechanisms that regulate the biosynthesis of this antibiotic. Understanding the regulatory mechanism for the production of the antibiotic WAP-8294A2 and the lytic enzymes (e.g. chitinase) is important because it could lead to new genetic approaches to improvement of biocontrol of crop disease and potential medical application.
Hfq is a protein serving as a conserved ribonucleic acid (RNA) chaperon and was first characterized as a host factor for phage Qβ RNA replication (Franze de Fernandez et al., 1968). It is widely distributed in bacterial annotated genomic databases (Caswell et al., 2012;. Hfq can bind AU-rich sequence of target messenger RNA (mRNA) and facilitate the pairing interaction between mRNA and small RNA , which suggests that this protein is a global posttranscriptional regulator in most cases. However, recent studies also showed that Hfq was able to directly bind deoxyribonucleic acid (DNA) (Updegrove et al., 2010;Geinguenaud et al., 2011), even tRNA (Lee and Feig, 2008) and proteins (Butland et al., 2005) to modulate the transcriptional expression of target genes, which indicates Hfq also can function as a transcriptional regulator in some bacterial species. Hfq was shown to play critical roles in diverse animal bacterial pathogens and plant-associated bacteria, such as Escherichia coli, Salmonella, Sinorhizobium meliloti, Staphylococcus and Pseudomonas; the hfq mutant exhibited pleiotropic phenotypes in these bacterial species, including decreased growth rate (Fantappiè et al., 2009;Chambers and Bender, 2011), increased sensitivity to various environmental stressors (Kadzhaev et al., 2009;Schiano et al., 2010) and attenuated virulence (Geng et al., 2009;Ramos et al., 2011). The role of hfq in a number of bacterial biological control agents had been investigated. For example, Hfq was found to regulate antibiotic production in Pseudomonas aeruginosa M18, a rhizobacterium bacterium that can efficiently inhibit soilborne phytopathogenic fungi . In Pseudomonas fluorescens 2P24, hfq is involved in the colonization, biofilm formation, antibiotic synthesis and quorum sensing signal production (Wu et al., 2010). However, essentially nothing is known about the role of hfq in any of Lysobacter species.
In the present study, we identified an hfq homologue from the genome of strain OH11, a Chinese isolate of L. enzymogenes. The results show that this hfq plays a pleiotropic role in regulating the antibacterial antibiotic biosynthesis and extracellular lytic enzyme activity in L. enzymogenes. To our knowledge, this study represents the first attempt to address the regulatory function of hfq in the genus of Lysobacter. Our findings also add an understanding of the conserved Hfq protein in different bacterial species.

Results and discussion
Sequence analysis of hfq in L. enzymogenes Hfq protein was first discovered from E. coli, and this Hfq was considered as the model in bacteria (Franze de Fernandez et al., 1968;Kajitani and Ishihama, 1991). To examine whether L. enzymogenes possesses an Hfq homologue, we used the E. coli Hfq protein (AAC43397.1) as the query sequence to perform a local BLASTP search in the draft genome of strain OH11 (Lou et al., 2011). This led to the identification of an Hfq homologue (KM186922) in L. enzymogenes. As shown in Fig. 1, Hfq protein of L. enzymogenes shares 75% similarity to that of E. coli at the amino acid level. Moreover, the locus of miaAhfq-hflX-hflK-hflC of E. coli was also conserved in L. enzymogenes. Next, we selected the reported Hfq proteins from four Lysobacter-related bacterial species, all belonging to the Xanthomonadaceae family, to do a sequence alignment. The result showed that Hfq of L. enzymogenes shared a high similarity (85-96%) to that of these taxonomically related species, including Hfq xoo of Xanthomonas oryzae pv. oryzae (WP_ 014503655.1), Hfqxcc of X. campestris pv. campestris (WP_011036893.1), Hfqsm of Stenotrophomonas maltophilia (WP_019183319.1) and Hfqxf of Xylella fastidiosa (WP_004085558.1). The results suggest that L. enzymogenes possesses a putative conserved Hfq protein.

Generation of a set of marker free of hfq-derived strains of L. enzymogenes
In order to explore the function of hfq in L. enzymogenes, a 155-bp internal fragment of hfq was in-frame deleted, which led to the generation of an hfq mutant, named as Δhfq (Fig. S1). Subsequently, a genomic marker-free strategy was utilized to construct the hfq complemented strain ( Fig. 2A). In this hfq complemented strain, no exogenous antibiotic selection markers were introduced into the hfq mutant, which can eliminate their potential effect on the tested phenotypes of the present study in L. enzymogenes. For this purpose, we selected αlp , an α-lytic-protease encoding gene as a target for hfq integration. To our knowledge, this gene is not associated with the tested phenotypes of the present study, including the antibiotic WAP-8294A2 production and the activity of extracellular chitinase, protease or cellulase. By using homologous recombination method as shown in Fig. 2A, we introduced the intact hfq gene together with its own promoter into the αlp genomic locus in the background of the hfq mutant, and generated a chromosomal marker-free hfq-complemented strain Δhfq(hfq)Δαlp (Fig. 2B). Meanwhile, we also created single mutation of αlp (Δαlp) and double mutations of hfq and αlp (ΔhfqΔαlp) as controls (Table S1). To verify the mutants, we utilized an reverse transcription polymerase chain reaction (RT-PCR) assay to examine whether the target gene (hfq or αlp) was transcribed or not in these hfqderived strains. As shown in Fig. 2C, we detected the transcription of hfq in wild-type OH11, the hfq-  complemented strain (Δhfq(hfq)Δαlp) and the αlp mutant (Δαlp) but not in the hfq mutant (Δhfq) and the double mutant (ΔhfqΔαlp). Similarly, it was also found that αlp was transcribed in wild-type OH11 and the hfq mutant (Δhfq) but not in the hfq complemented strain, the αlp deletion mutant and the double mutant. These results verified the hfq, αlp, double mutations and the marker-free hfq complemented strains.

Mutation of hfq caused a significant increase of the yield of WAP-8294A2 in L. enzymogenes
To test the role of hfq in the regulation of the antibiotic WAP-8294A2 biosynthesis, we examined WAP-8294A2 production in the hfq mutant in L. enzymogenes. To eliminate the potential influence of growth alteration on WAP-8294A2 production between the wild-type strain and the hfq mutant, we subsequently determined the growth ability (expressed by OD 600 nm, the optical density at 600 nm) of wild-type OH11, the hfq mutant and its derivative strains in 20% TSB broth that was used to cultivate these Lysobacter strains for WAP-8294A2 extraction. Meanwhile, considering that inactivation of hfq in several bacteria leads to an increased cell size (Tsui et al., 1994;Boudry et al., 2014), we then tested the role of hfq in this phenotype (cell size) in L. enzymogenes, because if disruption of hfq causes bigger cells, it will affect the cell density that is expressed by OD600 nm in the present study. As shown in Fig. S2, we found that mutation of hfq almost did not alter cell size compared with the wild-type strain at the two selected time points (after growth of 24 h and 48 h). These results suggest that application of OD600 nm as an indicator to reflect the cell growth status between the wild-type strain and the hfq mutant should be reasonable in the present study. Next, the growth rate of the wild-type strain and the hfq mutant in 20% Tryptic Soy Broth (TSB) broth was determined and compared. As shown in Fig. 3, we observed that mutation of hfq resulted in a changed growth pattern that was taken place in the logarithmic phase and originated the delay in reaching the stationary phase compared with that of the wild-type strain. Under the same condition, the complemented strain exhibited the wild-type growth rate. As expected, mutation of αlp did not alter the wild-type growth rate, indicating that αlp was not involved in the growth capacity in the tested medium. This finding was further verified by the growth rate of the double mutant (ΔhfqΔαlp), as this double mutant displayed a closely similar growth rate to that of the hfq mutant. The results suggested that mutation of hfq had a slight effect on the growth of the wild-type strain in the tested medium (20% TSB).
Next, we cultivated the tested Lysobacter strains in 20% TSB broth to extract WAP-8294A2, and determined its yield in each strain by high-performance liquid chroma-tography (HPLC). To completely eliminate the influence of growth alteration on WAP-8294A2 production, we used the indicator of 'Antibiotic production (peak area/OD 600 nm)' to quantitatively evaluate the ability of WAP-8294A2 production in the wild-type strain and its derivatives. Here, peak area indicates the area of WAP-8294A2 determined by HPLC, whereas OD600 nm represents the cell density of the tested strains at the time point used for the extraction of WAP-8294A2 in L. enzymogenes. In this way, we found that mutation of hfq significantly enhanced WAP-8294A2 production (∼ 2.2 fold) compared with the wild-type strain, whereas the marker-free complemented strain of the hfq mutant displayed the wild-type level in this ability (Fig. 4A). Given that the marker-free hfq complemented strain was constructed by integration of hfq into the αlp gene in the background of hfq mutation, the parent hfq and αlp genes were both missing in this complemented strain. Therefore, it is also possible that the restoration of WAP-8294A2 production in the marker-free hfq complemented strain could be due to the effect of both missing hfq and αlp. To test this possibility, we determined the WAP-8294A2 production in the double mutant (ΔhfqΔαlp). We showed that mutation of αlp in the background of the hfq mutant did not affect the yield of WAP-8294A2 in this mutant (the hfq mutant) (Fig. 4A). This result provided supportive evidence to verify that restoration of WAP-8294A2 production in the marker-free hfq complemented strain was due to the replacement of αlp by hfq in the background of the hfq mutation but not associated with the effect of the simultaneous deletion of both hfq and αlp. Furthermore, mutation of αlp in the background of wildtype strain did not alter WAP-8294A2 production, supporting that αlp was not associated with WAP-8294A2 biosynthesis. Collectively, these results indicated that hfq played an important role in the negative regulation of WAP-8294A2 biosynthesis in L. enzymogenes. Fig. 3. The growth curves of various Lysobacter strains in 20% TSB medium. The growth level of each strain was measured by OD600 nm at regular intervals (2 h or 4 h). Three replicates for each treatment/strain were used, and the experiment was performed three times. Vertical bars represent standard errors. The strain information in Fig. 3 is shown in the legend of Fig. 2. To further verify the important role of hfq in WAP-8294A2 biosynthesis, the transcriptional level of waps1 (Zhang et al., 2011), the key gene responsible for WAP-8294A2 biosynthesis, was tested using quantitative (q)RT-PCR between the wild-type strain and the hfq mutant. Based on the result of Fig. 3, we finally collected the cells at the logarithmic phase from the wild-type strain and the hfq mutant at different time point corresponds to the same cell density (OD600 nm = 1.0), respectively, because at this cell density, the gene waps1 was previously shown to be expressed at transcriptional level in the wild-type OH11 of L. enzymogenes (Wang et al., 2014). The results of the qRT-PCR assay showed that mutation of hfq caused a significant increase (∼ 2.3 fold) of the transcription of waps1 compared with the wild-type strain (Fig. 4B). This finding was consistent with the HPLC result for the WAP-8294A2 production in the hfq mutant (Fig. 4A), and further verified that hfq was involved in the negative regulation of WAP-8294A2 biosynthesis in L. enzymogenes.

Mutation of hfq almost abolished extracellular chitinase activity and significantly reduced the activity of both extracellular protease and cellulase in L. enzymogenes
In addition to WAP-8294A2 biosynthesis, we are also interested in addressing the role of hfq in lytic-enzyme production in L. enzymogenes. We therefore examined the activity of three known extracellular lytic enzymes, including chitinase, protease and cellulase on the corresponding detecting media. In the present study, the ratio of the area of hydrolytic zones divided by cell density (OD 600 nm) was used as a quantitative indicator for each enzyme activity. By this way, we found that mutation of hfq almost abolished extracellular chitinase activity, and significantly reduced the activity of both extracellular protease (P < 0.01; t-test) or cellulase (P < 0.05; t-test) in L. enzymogenes (Fig. 5). Under the same conditions, the marker-free hfq complemented strain restored the wild-type level in each tested enzyme activity (Fig. 5). Furthermore, mutation of αlp in the background of the hfq mutant did not influence the activity of each tested enzymes (Fig. 5). This result suggested that the activity restoration of each tested enzymes in the marker-free hfq  Fig. 4 is shown in the legend of Fig. 2. Each column indicates the mean of three biologically independent experiments. Vertical bars represent standard errors. '*' (P < 0.05; t-test) above the bars indicate a significant difference between the wild-type strain OH11 and the hfq mutant. Vertical bars represent standard errors. '*' (P < 0.05; t-test) or '**' (P < 0.01; t-test) above the bars indicate a significant difference between the wild-type strain and its derivatives. The strain information in Fig. 5 is shown in the legend of Fig. 2. complemented strain was due to the replacement of αlp by hfq in the background of the hfq mutation but not associated with the effect of the deletion of both hfq and αlp. Moreover, mutation of αlp in the background of the wild-type strain did not alter the activity of each tested enzyme (Fig. 5), supporting that αlp was not involved in the regulation of the activity of these three tested enzymes in L. enzymogenes. Collectively, our results indicated that hfq played a key role in the regulation of the activity of extracellular chitinase, protease and cellulase in L. enzymogenes.

Mutation of hfq inhibited the secretion of chitinase A
The deficiency of the hfq mutant in extracellular chitinase production promotes us to focus the mechanism(s) by which hfq modulates this activity in L. enzymogenes. For this purpose, we first investigated whether the deficiency of the hfq mutant in extracellular chitinase activity might be associated with its growth reduction, as we observed that the final cell density of the hfq mutant was significantly reduced (∼ 2.3 fold) compared with that of the wildtype strain when the initial inoculated cell concentration was the same (Fig. S3). Subsequently, the experimental evidence presented here eliminated this possibility, because we clearly found that even though the hfq mutant possessed a significantly increased cell density (P < 0.05; t-test) compared with the wild-type strain (when the initial inoculated cell concentration of the hfq mutant was 10-fold higher than that of the wild-type strain) (Fig. S3), we still did not find any hydrolytic zone around the colonies of the hfq mutant (Fig. S3). These results provided supportive evidence to show that the deficiency of the hfq mutant in chitinase activity was not due to its growth reduction, which implies that other mechanism(s) may be utilized by hfq to modulate extracellular chitinase production in L. enzymogenes.
Next, we attempted to test which type(s) of chitinases was reduced in the hfq mutant, resulted in its deficiency in extracellular chitinase activity. For this purpose, we made a survey to detect how many chitinase-encoding genes are present in the genome of L. enzymogenes. This led to identification of a total of three chitinase-encoding genes, including chiA , chiB and chiC (Table S3). We subsequently generated the chiB and chiC deletion mutants (Table S1). The extracellular chitinase activity of these two mutants as well as the chiA mutant  was then tested. As shown in Fig. S4, we found that mutation of chiA completely abolished the chitinase activity, which was closely consistent with the role of the hfq mutant in this phenotype (Fig. 5). However, mutation of chiB or chiC did not alter the extracellular chitinase activity (Fig. S4). This finding raised the possibility that the regulation of hfq on extracellular chitinase activity was probably through the alteration of chiA in L. enzymogenes under the testing conditions.
Considering the fact that Hfq usually functions as a post-transcriptional regulator (Lee and Feig, 2008;Sittka et al., 2009), we then examined whether hfq plays a posttranscriptional regulation on chiA. We first constructed a plasmid (pBBR1-MCS5)-based hfq-complemented strain by introduction of pBBR1-MCS5 containing the hfq gene into the hfq mutant. Then, the extracellular chitinase activity was tested in this complemented strain. We showed that the plasmid (pBBR1-MCS5)-based hfqcomplemented strain partially restored the wild-type level in extracellular chitinase activity, whereas the hfq mutant containing the empty vector was deficient in this function (Fig. S5). This result indicates that the plasmid (pBBR1-MCS5) containing hfq was functional in restoring the deficiency of the hfq mutant in extracellular chitinase activity. Next, a flag-tagged chiA sequence was cloned into the broad-host-vector pBBR1-MCS5, and transformed into the chiA mutant. The introduction of pBBR1-MCS5 containing the flag-tagged chiA restored the chitinase activity of the chiA mutant (Fig. S6A), supporting the correction of the construction. We then individually introduced the pBBR1-MCS5 containing the flag-tagged chiA into the hfq mutant and wild-type OH11, respectively. Next, the antiflag antibody dependent western blot assay was performed to compare the protein content of the flag-tagged ChiA (Chitinase A) in wild-type OH11 and the hfq mutant. As shown in Fig. 6, we observed that the band of flagtagged ChiA was detected both in the total cells and the supernatant of wild type, suggesting ChiA was synthesized and secreted outside the cells under the testing condition. Meanwhile, we only detected the flag-tagged ChiA band in the total cells but not in the supernatant of the hfq mutant. This result indicated the ChiA protein was probably synthesized in the cells of the hfq mutant but was unable to be secreted out of the cells or degraded once secreted in the hfq mutant. To verify this, we introduced the construct (pBBR1-MCS5 containing the flagtagged chiA) into the hfq mutant and tested its chitinase activity. We found that the hfq mutant containing this construct did not restore the chitinase activity, as no hydrolytic zones around its colonies was detected on the chitinase selective medium, whereas the chiA mutant and wildtype OH11 containing this construct displayed visible chitinase activity (Fig. S6B). This result provided supportive evidence to show that the effect of hfq on chitinase activity was probably in part through the impairment of the secretion of ChiA in L. enzymogenes. However, we do not know how hfq mediates the secretion of ChiA in L. enzymogenes at this time.
A recent study shows that in Flavobacterium johnsoniae chitinase secretion is dependent on type IX secretion system (T9SS), consisting of products of the key genes of gldK, gldL, gldM, gldNO, sprA, sprE and sprT (Kharade and McBride, 2014). However, we did not find the homologue of any of these T9SS-associated key genes/ proteins in the genome of L. enzymogenes OH11, indicating that the effects of hfq on the secretion of Chitinase A in L. enzymogenes is probably not associated with T9SS and is therefore different from the finding in F. johnsoniae. Similar to our result, the hfq mutant of Listeria monocytogenes showed a less chitinolytic activity compared with that of wild-type strain. This report pointed out that uncharacterized hfq-dependent small RNAs may mediate the stimulating effect of hfq on the activity and/or secretion of the chitinase in L. monocytogenes (Nielsen et al., 2011). This information provided a clue to further explore the regulatory mechanism of hfq on the secretion of chitinase A by focusing on the hfq-dependent small RNAs in L. enzymogenes.

Conclusion remarks
The production of lytic enzymes and antibiotics is one of the distinctive characteristics of the important but underexplored biological control agent L. enzymogenes. The genetic determinants that regulate the biosynthesis of these factors are largely unidentified. The present study reported Hfq, a putative RNA chaperone played a pleiotropic role in the modulation of the antibacterial antibiotic WAP-8294A2, and the activity of extracellular chitinase, protease and cellulase in L. enzymogenes OH11. The regulation of hfq on extracellular chitinase production was further shown to be in part through the impairment of the secretion of chitinase A. These findings provide new insights into the role of hfq in bacteria. In future works, we aim to further address the molecular mechanisms of hfq on the regulation of WAP-8294A2 as well as the secretion of chitinase A in L. enzymogenes. These future works will help us further understand the signalling pathway of hfq in the regulation of antibiotics and lytic enzymes in L. enzymogenes.

Bacterial strains and growth conditions
Strains and plasmids used in this study are shown in Table 1. Escherichia coli, strain Top 10 was grown in LB medium at 37°C. Unless otherwise stated, Lysobacter enzymogenes strain OH11 (CGMCC No. 1978), and its derivative strains were grown in LB medium at 28°C. When required, appropriate antibiotics were added into the medium to a final concentration of kanamycin (Km) 50 μg ml −1 and gentamicin (Gm) 150 μg ml −1 .

Generation of deletion mutants of target genes in L. enzymogenes
The wild-type strain OH11 (CGMCC No. 1978) of L. enzymogenes was used as an original strain to generate the in-frame deletion mutants. Construction of gene-deletion mutants in L. enzymogenes was performed as described previously (Qian et al., 2013). As a representative example, the scheme of the hfq mutant construction and molecular confirmation in L. enzymogenes was provided in Fig. S1 in the supplementary file. All the primers and plasmids used in the present study were provided in Table S2 and Table 1 respectively.

Construction and confirmation of a marker-free hfq complemented strain
The hfq fragment with its predicted promoter (668-bp) was inserted into a pEX18Gm-αlp  recombinant suicide vector, which has an upstream fragment (300bp) and a downstream fragment (500-bp) of αlp gene of the wild-type OH11. The final construct was transformed into the hfq mutant. After twice homologous recombination, the αlp gene was replaced by hfq in the genome of the wild-type OH11, which is defined as the marker-free-complemented strain of the hfq mutant. This marker-free hfq-complemented strain was verified both by PCR and RT-PCR assays. For PCR assay, the expected size of the DNA fragment from the hfq mutant was 1994 bp by using the primers of αlp-F1/R2 (Table S2). When the αlp gene was replaced by the hfq gene in the background the hfq mutant, the expected size of DNA fragment was 1448 bp by using the same primer pairs (αlp-F1/R2). For RT-PCR assay, the expression of the target gene (hfq or αlp) in the hfq mutant and its derivative strains was tested. The expected size of the DNA fragment amplified from the complementary DNA (cDNA) of the wild-type strain corresponds to the hfq, αlp and 16s rRNA was 158 bp, 174 bp and 147 bp respectively. The corresponding primers were shown in Table S4. In the present study, 16s rRNA was used as an internal control as described previously (Qian et al., 2013;.

Observation of the cell size of L. enzymogenes by electronic microscope
The wild-type OH11 and the hfq deletion mutant were grown on 20% TSB with agar (TSA) plates, and their cells were collected after growth of 24 h and 48 h, respectively. These cells were further used for the analysis of electronic microscope as described previously in our lab (Chen et al., 2009)

Detection of growth curve
Various Lysobacter strains were cultured in LB medium at 28°C overnight. Then, 500 μl of the overnight culture for each strain was transferred into the fresh 20% TSB broth (50 ml) to grow until the cell density expressed by OD 600 nm reached to 1.0. Next, 1 ml of each culture was transferred again into the fresh 20% TSB broth (50 ml) to start the detection of growth curve. All inoculation broths were grown at 28°C with shaking at 200 r.p.m., and the OD600 nm value was determined every 2 h or 4 h until bacterial growth reached the stationary stage. Each sample involves three technical replicates and the experiment was performed three times.

RNA extraction, qRT-PCR and RT-PCR
The wild-type OH11 and the hfq deletion mutant were grown on 20% TSB. The cells of the wild-type strain and the hfq mutant were collected at the time point, 8 h and 11 h, corresponds to the same cell density (OD600 nm = 1.0). Then, the total RNA was extracted from the cells of each strain using a kit with a code of R6950-01 from OMIGA Company (China). Next, the qRT-PCR and RT-PCR assays, including cDNA synthesis and PCR amplification were performed as described previously (Qian et al., 2013;. Primer sequences used in this assay are listed in Table S4.

Extraction and detection of WAP-8294A2
Various Lysobacter strains were cultured in 20% TSB broth at 28°C until the cell density expressed by OD600 nm reached to 1.0. Next, 1 ml of each culture was transferred into the fresh 20% TSB broth (50 ml) for 3 days shaking culturing. Then, the extraction and HPLC analysis of WAP-8294A2 from L. enzymogenes were performed as described previously (Zhang et al., 2011;. Three replicates were used for each strain, and the experiment was performed three times.

Quantitative determination of the activity of three extracellular lytic enzymes
A sterile filter membrane (10 mm diameter) was put on the surface of the selective plates for chitinase, protease and cellulase. The composition of each selective plates was described in previous studies (Kobayashi et al., 2005;Qian et al., 2013). In brief, 3 μl of bacterial culture of various Lysobacter strains with the same cell density (OD600 nm = 2.0) was spotted on the filter membrane. After 3 days of growth, the filter member was taken off from the plates. The diameters of the hydrolytic zones in the plates were measured and the corresponding hydrolytic areas were calculated. Meanwhile, the cells of each strain on the filter member were washed by 800-μL sterilized ddH2O, and the cell density of each strain was measured and expressed by OD600 nm. Finally, the ratio of the area of hydrolytic zones divided by cell density (OD600 nm) was used as a quantitative indicator for each enzyme activity of the tested Lysobacter strains.

Western blot
The flag-tagged chiA was amplified by PCR with the primers of chiA-F and chiA-R (flag) (Table S2), and cloned into the broad-host-vector pBBR1-MSC5 (Table 1). Then, the pBBR1-MCS5 containing the flag-tagged chiA was transferred into wild-type OH11 and the hfq deletion mutant to generate two strains, OH11 (chiA-flag) and Δhfq (chiA-flag). These strains were cultured in 50 ml of LB broth to grow until the value of OD600 nm reached to 2.0. Then, the total cells and the culture supernatants of tested strains were collected respectively. Next, for cells, 1-ml RIPA buffer (CWBIO Company, China) with a code of 1713L was used to lyse cells, followed by a centrifugation (10 000 × g at 4°C for 10 min). These cell supernatants were collected and used for further study. For culture supernatants, they were concentrated to 1 ml using a vacuum freeze drier, and used for further studies. The following western blot assay was performed as described previously (Ansong et al., 2009). The experiments were performed three times, and each involves three replicates for each treatment.

Data analysis
All analyses were conducted using SPSS 14.0 (SPSS Inc, Chicago, IL, USA). The hypothesis test of percentages (t-test, P < 0.05 or 0.01) was used to determine significant differences in the production of antibiotic metabolites, lytic-enzyme activity and gene expressions between the wild-type OH11 and its derivatives.  (Table S2) and 368-bp (amplified by hfq-F2/R2) ( Table S2) DNA fragment of hfq was used as 5′ and 3′ fragment for homologue recombination respectively. The internal 155-bp DNA fragment would be deleted in the hfq mutant. The primers hfq-F1/R2 (Table S2) was used for molecular confirmation of hfq mutant. B. PCR verification of the hfq mutant. An 887-bp DNA fragment was amplified from the wild-type OH11 with the primers hfq-F1/R2, while only a 732-bp fragment was obtained from the hfq mutant with the same primers due to the deletion of internal 155-bp DNA fragment. '-' in B represents the blank control. This strategy was applied into the construction of other mutants of target genes in the present study (Table S1  and Table S2). S2. Comparison of the cell size between the wild-type strain and the hfq mutant of Lysobacter enzymogenes.

Fig.
A. The representative result of cell size from three independent experiments between the wild-type strain (OH11) and the hfq mutant (Δhfq) under electronic microscope at two selected time points (24 h and 48 h after growth on solid 20% TSA medium). The scale bars represent 1 μm. B. Statistic analysis of the cell size of the wild-type strain and the hfq mutant. In each technological repeat, at least 15 cells of each strain were selected for analysis. The experiment was performed three times. Each column indicates the mean of three biologically independent experiments. Vertical bars represent standard errors. No significant difference (P < 0.05; t-test) in cell size between the wild-type strain and the hfq mutant was found. Fig. S3. Further determination of extracellular chitinase activity between the wild-type strain and the hfq mutant of Lysobacter enzymogenes on solid medium. A. The representative phenotype of extracellular chitinase activity between the wild-type OH11 and the hfq mutant from three independent experiments. B. Quantitative analysis of cell density (OD600 nm) of each strain on the surface of the filter membrane. Each column indicates the mean of three biologically independent experiments. Vertical bars represent standard errors. '**' (P < 0.01; t-test) above the bars indicate a significant difference between the wild-type strain and the hfq mutant. The initial inoculated cell concentration expressed by OD600 nm for the wild-type OH11, Δhfq-a, Δhfq-b was 2, 2 and 20 respectively. Fig. S4. Detection of the chitinase activity of three mutants of predicted chitinase synthesis genes (ΔchiA, ΔchiB, ΔchiC) on chitin plate. Only the chiA mutant cannot hydrolyse chitin under the tested conditions. The gene information of chiB and chiC was provided in Table S3. The mutant construction and confirmation was provided in Table S1. Fig. S5. Introduction of the broad-host-vector pBBR1-MCS5 containing hfq partially restored extracellular chitinase production of the hfq mutant in Lysobacter enzymogenes. OH11, the wild-type strain of L. enzymogenes; Δhfq, the hfq deletion mutant; Δhfq(pBBR), the hfq mutant containing the empty vector (pBBR1-MCS5); Δhfq(hfq-pBBR), the pBBR1-MCS5based hfq complemented strain. Figure S5 is the representative result of three independent experiments. Fig. S6. Determination of extracellular chitinase activities of Lysobacter strains.
A. The construct of flag-tagged chiA restored the chitinase activity of the chiA mutant. B. The hfq mutant containing the construct of flag-tagged chiA did not restore the chitinase activity. OH11, the wild-type strain of L. enzymogenes; ΔchiA, the chiA deletion mutant ; ΔchiB, the chiB deletion mutant; ΔchiC, the chiC deletion mutant; ΔchiA(chiA-flag), the chiA mutant containing flag-tagged chiA; Δhfq(pBBR), the hfq mutant containing an original pBBR1-MSC5 vector; Δhfq(hfq-pBBR), the hfq mutant containing hfq-pBBR complemented vector; Δhfq(chiA-flag), the hfq mutant containing flag-tagged chiA; OH11(chiA-flag), the wild-type containing flag-tagged chiA. Table S1. Mutant confirmation by PCR in this study. Table S2. Primers used for in-frame deletion and complementation in this study. Table S3. Three chitinase encoding genes in Lysobacter enzymogenes. Table S4. Primers used for qRT-PCR or RT-PCR in this study.