The Fish Pathogen Yersinia ruckeri Produces Holomycin and Uses an RNA Methyltransferase for Self-resistance

Background: The self-resistant mechanism for holomycin has been elusive. Results: The holomycin gene cluster was identified and characterized in Y. ruckeri and the presence of hom12 shown to be crucial for holomycin resistance. Conclusion: Y. ruckeri is a holomycin producer and recruits an RNA methyltransferase for self-resistance. Significance: Our work has demonstrated the hitherto unknown resistant mechanism during holomycin production. Holomycin and its derivatives belong to a class of broad-spectrum antibacterial natural products containing a rare dithiolopyrrolone heterobicyclic scaffold. The antibacterial mechanism of dithiolopyrrolone compounds has been attributed to the inhibition of bacterial RNA polymerase activities, although the exact mode of action has not been established in vitro. Some dithiopyrrolone derivatives display potent anticancer activities. Recently the biosynthetic gene cluster of holomycin has been identified and characterized in Streptomyces clavuligerus. Here we report that the fish pathogen Yersinia ruckeri is a holomycin producer, as evidenced through genome mining, chemical isolation, and structural elucidation as well as genetic manipulation. We also identified a unique regulatory gene hom15 at one end of the gene cluster encoding a cold-shock-like protein that likely regulates the production of holomycin in low cultivation temperatures. Inactivation of hom15 resulted in a significant loss of holomycin production. Finally, gene disruption of an RNA methyltransferase gene hom12 resulted in the sensitivity of the mutant toward holomycin. A complementation experiment of hom12 restored the resistance against holomycin. Although the wild-type Escherichia coli BL21(DE3) Gold is susceptible to holomycin, the mutant harboring hom12 showed tolerance toward holomycin. High resolution liquid chromatography (LC)-ESI/MS analysis of digested RNA fragments demonstrated that the wild-type Y. ruckeri and E. coli harboring hom12 contain a methylated RNA fragment, whereas the mutated Y. ruckeri and the wild-type E. coli only contain normal non-methylated RNA fragments. Taken together, our results strongly suggest that this putative RNA methyltransferase Hom12 is the self-resistance protein that methylates the RNA of Y. ruckeri to reduce the cytotoxic effect of holomycin during holomycin production.

Holomycin 1 belongs to a class of dithiolopyrrolone natural products containing a unique heterobicyclic scaffold with an N-alkyl and an N-acyl substitution (Fig. 1). Naturally occurring dithiolopyrrolones remain a rare entity and include thiolutin 2, aureothricin 3, and more recently thiomarinol 4, a hybrid marine bacterial natural product containing a dithiolopyrrolone framework linked by an amide bridge with an 8-hydroxy-octanoyl chain linked to a pseudomonic acid (Fig. 1). Holomycin 1 was first isolated from the soil bacterium Streptomyces clavuligerus (1). Since then several other bacteria have been reported to be holomycin producers, including the marine Gram-negative bacterium Photobacterium halotolerans S2753 (2). Dithiolopyrrolone antibiotics possess broad-spectrum antibacterial activities against Gram-positive and Gram-negative bacteria (3,4). Holomycin appeared to be active against rifamycin-resistant bacteria (5) and also to inhibit the growth of the clinical pathogen methicillin-resistant Staphylococcus aureus N315 (6). Its mode of action is believed to inhibit RNA synthesis, although the exact mechanism has yet to be established in vitro. The biosynthetic gene cluster of holomycin has recently been identified in S. clavuligerus and characterized biochemically and genetically (Fig. 2). In vitro studies indicated that a thioredoxin reductase-like dithiol oxidase HlmI is responsible for the formation of the disulfide bridge from dithiol 2 to holomycin 1 (Scheme 1) (9). Interestingly, deletion of hlmI resulted in a major loss of holomycin production and significantly increased susceptibility toward holomycin, indicating that hlmI plays an important role in self-protection (7). To deal with the proposed toxic dithiol load, the hlmI mutant is likely to activate the detoxification mechanism by incapacitating the dithiol intermediates into mono-and di-S-methylated forms (10). The biosynthetic gene cluster of 4 was also identified from the marine bacterium Pseudoalteromonas sp SANK 73390 (11). By analogy, the enzymes responsible for the dithiolopyrrolone scaffold of 4 also have homologues in the biosynthesis of 1 including one multidomain nonribosomal peptide synthetase and three other oxireductases (11). The analog of the dithiol oxidase, however, cannot be found in the gene cluster of 4, suggesting that it has another mechanism of disulfide formation.
Yersinia ruckeri is the causative agent of enteric redmouth (ERM) 7 disease in salmonids, also known as yersiniosis, which can cause large losses in aquaculture (12). However, only a few pathogenic mechanisms of Y. ruckeri have been described, some of which have been proven to be involved in virulence, such as the iron uptake mechanism via the siderophore natural product ruckerbactin (13) and the YhlA hemolysin (14). A recent study indicated that a new type of two-component operon is required for full virulence of Y. ruckeri in fish (15). The operon contains an amino acid permease motif and an L-cysteine desulfidase motif, which was confirmed to be involved in the regulation of cysteine uptake. Knock-out of this operon abolishes virulence of Y. ruckeri in fish. Interestingly, a connection between holomycin production and sulfur metabolism was reported before holomycin production was found to be up-regulated by cysteine (16). In light of the importance of ERM in aquaculture, the draft genome of Y. ruckeri ATCC 29473 has been published with the length of 3.7 megabases (17).
Here we show, using bioinformatics-based genome mining, gene disruption and complementation experiments, that Y. ruckeri is a producer of holomycin under aerobic cultivation conditions. Importantly, our results also demonstrated two interesting findings. 1) The gene hom15, encoding a cold-shock like protein, is likely to play a unique protective role for the holomycin biosynthesis under low cultivation temperatures (16 and 22°C). 2) Y. ruckeri employs the RNA methyltransferase Hom12 as self-resistance during holomycin production. It is proposed that Hom12 methylates the RNA of Y. ruckeri, hence preventing holomycin binding to its own RNA and protecting itself from inhibition of RNA polymerase synthesis during the holomycin production.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and General DNA Manipulations-Y. ruckeri was maintained in Tryptone soya broth (TSB) medium. Escherichia coli DH5␣ was used as the host for general DNA propagations. GeneJET Plasmid Miniprep kits (Thermo Scientific) were used to prepare plasmids from E. coli strains. The genetic manipulations were performed by standard methods (18). Restriction endonucleases, DNA ligases, and DNA polymerase were purchased from various sources and used according to the manufacturer's recommendations. DNA fragments were purified using GeneJET Gel Extraction kits (Thermo Scientific). PCR conditions involved preheating at 95°C for 5 min followed by 35 cycles of denaturation at 95°C for 40 s, annealing at 52°C for 40 s, and extension at 72°C for 40 s to 2 min depending on the size of DNA amplifications. The PCR products were purified and ligated into the pGEM-T Easy vector (Promega) for blue/white screening. The resultant plasmids were subjected to DNA sequencing. After sequencing, the resultant plasmids were subsequently digested with restriction enzymes. The DNA inserts were purified using GeneJET Gel Extraction kits (Thermo Scientific) and ligated into pK18mobGII (19). For open reading frame (ORF) amplification, PCRs were conducted with KOD polymerase (isolated from Thermococcus kodakaraensis; 2.5 units, Merck) (50 l). DNA sequencing was performed by the sequencing service at the College of Life Sciences, University of Dundee, Scotland, using Applied Biosystems Big-Dye Version 3.1 chemistry on an Applied Biosystems model 3730 automated capillary DNA sequencer. Genomic DNAs of Y. ruckeri wild-type and mutated strains were purified using a Fermentas Genomic DNA purification kit (Thermo Scientific). Antibiotics were used at the following final concentrations as required: kanamycin (50 g/ml), tetracycline (25 g/ml), and ampicillin (100 g/ml).
HR-HPLC-ESI/MS and NMR Analysis of Holomycin-The organic extracts were subjected to high resolution LC-MS analysis (C18 Sunfire TM 150 ϫ 4.6 mm Waters; A: 0.1% formic acid in water; B: 0.1% formic acid in methanol; gradient 0 -20 min 0 -100% B). High resolution electrospray ionization mass spectra were obtained using a Thermo Scientific MS system (LTQ XL/LTQ Orbitrap Discovery) coupled to a Thermo Instruments HPLC system (Accela PDA detector, Accela PDA

HPLC Profiling of Holomycin from the Wild-type and Mutant
Strains of Y. ruckeri-For metabolite analysis, the wild-type and mutant strains of Y. ruckeri were inoculated into TSB medium in an Erlenmeyer flask (250 ml) and cultivated in a rotary incubator (180 rpm, Barnstead Lab-line MaxQ 5000) at 22°C. After 24 h of fermentation, the media were extracted by ethyl acetate. The ethyl acetate fraction was removed under reduced pressure, and the residue was dissolved in methanol (5 ml).

Construction of Single-crossover Mutants of Y. ruckeri-The
single-crossover mutants of Y. ruckeri were constructed by homologous suicide plasmid integration using pK18MobGII as the vector (19). A 469-bp DNA fragment of hom1 was amplified from genomic DNA of Y. ruckeri by PCR using a primer pair of Hom1_F (5Ј-GGA ATT CAA CCG GAT CGG TGG ATA TTA CAC TTC TGC-3Ј) and primer Hom1_R (5Ј-AAG CTT GGA AGG ATG ACT CAC CTT CTT TCC GAG CGC-3Ј). The amplified DNA fragment was ligated into pGEM-T Easy vector and confirmed by sequencing. The correct DNA insert was cloned into the suicide plasmid pK18mobGII to create pK18_hom1. The resulting plasmid pK18_hom1 was then transferred into the competent cells of wild-type Y. ruckeri by electroporation. The transconjugants were screened on LB supplemented with kanamycin. The transconjugants were confirmed by PCR (Fig. 3U2) using the total DNA of the transcon- jugants as template and the primer pair pK18F_test (5Ј-GCC GAT TCA TTA ATG CAG CTG GCA C-3Ј) and Hom1_test_R  (5Ј-CGG AAG TTC CCA TTC GGA TGG TGC CG-3Ј, pK18F is located in pK18MobGII, and Hom1_test_R is located downstream of the 469-bp internal fragment of hom1). The PCR products were further confirmed by sequencing, and the confirmed mutant Yru⌬hom1 was used for further study.
The transconjugants for Yru⌬hom6 were confirmed by PCR (Fig. 3U, lane 1) using the primer pair of pK18F_test and Hom6_test_R (5Ј-GGC GTT CGA TCA GAA GAC TCA CCA AGG-3Ј), which is located downstream of the 1047-bp fragment of hom6. The confirmed mutant Yru⌬hom6 was used for further study. Transconjugants for Yru⌬hom12 were confirmed by PCR (Fig. 3U, lane 3) using the primer pair pK18F_test and Hom12_test_R (5Ј-TCA CCA CGT CAC GGG TTT CTC TGA TAG C-3Ј), which is located downstream of the 696-bp fragment of hom12. The confirmed mutant Yru⌬hom12 was used for further study.
Construction of a Double-crossover Mutant of Y. ruckeri-The double-crossover mutant Yru⌬hom15 of Y. ruckeri was constructed using pK18MobGII as the suicide vector. Two 500-bp DNA fragments downstream and upstream of hom15 were amplified from the total DNA of Y. ruckeri using two primer pairs; Hom15_L_F (5Ј-GAA TTC AGG ATT CAT TGA ACG ACT GGG TGG TCG C-3Ј) and Hom15_L_R (5Ј-GGA TCC GAC GGT CAG AAA GGT CCG TCT GCA G-3Ј) and Hom15_R_F (5Ј-GGA TCC CTC GCC CGG AGG CGT TAC ATA GAC AAA C-3Ј) and Hom15_R_R (5Ј-CAA GCT TAG CTT TGT GCT ATC ACT CTG GAG GTA AGG-3Ј). Both of the DNA fragments were ligated into the pGEM-T Easy vector for sequencing. After being confirmed by sequencing, these two DNA inserts were then excised from the T vectors by EcoRI-BamHI and BamHI-HindIII digestions, respectively, and cloned into the EcoRI-HindIII site of pK18mobGII to create pK18_hom15. The resulting plasmid pK18_hom15 was then transferred into Y. ruckeri competent cells by electroporation. The transconjugants were screened on LB supplemented with kanamycin. The transconjugants were confirmed by PCR (Fig.  3U, lanes 5 and 6) using the total DNA of the transconjugants as the template and the primer pair Hom15_test_F (5Ј-CTC ACC GCG CTA CCG TCA TCT CAA CC-3Ј) and Hom15_test_R (5Ј-GCA TAT TTA AGC AGT GTT TGT GAT CGA GTG GC-3Ј). The PCR products were further confirmed by sequencing, and the confirmed mutant Yru⌬hom15 was used for further study.
Complementation of the Mutant Yru⌬hom12-A 951-bp fragment was amplified from the genomic DNA of Y. ruckeri by PCR using the primer pair Hom12_comp_F (5Ј-GCT TAC CAA GCT TGA TTG AAG CAC AAT GAG TCG CTC CGG CTT G-3Ј) and Hom12_comp_R (5Ј-CCA TAC TGG ATC CTC ACC ACG TCA CGG GTT TCT CTG ATA GCT C-3Ј). This fragment was cloned into the pGEM-T Easy vector for sequencing. After being confirmed, the fragment was digested with HindIII/BamHI and then cloned into the broad-host vector pRK404 (20) to create pRK_hom12. The resulting plasmid pRK_hom12 was then transferred into the competent cells of wild-type Y. ruckeri by electroporation. The transconjugants were screened on LB supplemented with kanamycin and tetracycline. The mutant Yru⌬hom12-hom12 was used for further study.
Preparation of the Competent Cells of Y. ruckeri-To prepare Y. ruckeri competent cells, the bacteria were grown overnight at 22°C in TSB broth. The culture was diluted (1: 50) in fresh broth and grown at 28°C for 2-4 h until reaching A 600 ϭ 0.6 -0.8. Cells were harvested by centrifugation (4000 ϫ g, 10 min, 4°C) and washed once in cold glass-distilled water and once in transformation buffer (272 mM sucrose, 15% (v/v) glycerol). The pellet was resuspended in the transformation buffer, aliquoted, and stored at Ϫ70°C until required.
Electroporation Transfer of DNA Constructs to Y. ruckeri-Once used, the electrocompetent cells (50 l, 5 ϫ 10 10 cells/ml) were mixed with constructs (5 l, 500 ng) in a prechilled electroporation cuvette (0.2 cm, VWR International) and placed on an ice bath for 1 min. The cells were then subjected to a single electric pulse (12.5 kV/cm) according to the operating manual using Gene Pulser Apparatus (Eppendorf 2510, Germany). After discharge, TSB medium (1 ml) was immediately added, and this solution was incubated at 28°C on the water bath for 2 h. The broth was then subcultured on an LB agar plate (50 g/ml kanamycin) and incubated at 22°C for 24 h. The single colonies were selectively inoculated into 10 ml of TSB medium (50 g/ml kanamycin) at 22°C for 24 h and then stocked in sterile glycerol (60%).
For complementation experiments of the mutant Yru⌬hom12, preparation of competent cells and electroporation of pRK_hom12 were carried out using the same protocol described above, except that both kanamycin and tetracycline were used for selection.
Expression of the Methyltransferase Hom12-The ORF Hom12 was subcloned into the pET30EK/LIC (Novagen) vector using the following two primers: To test the susceptibility of the Y. ruckeri and E. coli strains toward holomycin, the strains were homogeneously added into warm LB liquid agar. After the LB liquid agar solidified, holomycin (35 g/ml in DMSO) was dropped into the center hole of the agar plates. The LB agar plates containing the Y. ruckeri and E. coli cultures were incubated at 22 and 37°C overnight, respectively.
HPLC-MS Analysis of Total RNA Fragments-A GeneJET RNA Purification kit (Fisher) was used to extract the total RNA from the wild-type Y. ruckeri, Yru⌬hom12, E. coli BL21, and E. coli BL21 harboring hom12. The purification procedures were used according to the manufacturer's instructions. The RNA fragments were then digested by RNase A (0.5 g/l as the final concentration) at room temperature for 30 min (21). The reactions were stopped by adding HCl (1 l, 0.5 M). The digested RNA fragments were then submitted for HR-LC/ ESI/MS analysis. The following HPLC conditions were used for RNA fragment analysis according to the previous report (22): HYPERCARB column, Thermo Scientific, 100 ϫ 4.6 mm, 5 m; A: 50 mM formic acid in water; B: 50 mM formic acid in acetonitrile; gradient 0 -60 min 0 -100% B; 1 ml/min. High resolution electrospray ionization mass spectra were obtained using a Thermo Scientific MS system (LTQ XL/LTQ Orbitrap Discovery) coupled to a accela HPLC system (Thermo Scientific) (negative mode, capillary voltage Ϫ50 V, capillary temperature 320°C, auxiliary gas flow rate 10 -20 arbitrary units, sheath gas flow rate 40 -50 arbitrary units, spray voltage 4.5 kV, mass range 300 -3000 atomic mass units (maximum resolution 30,000)).

RESULTS
Genome Mining of the Gene Cluster of Holomycin from the Fish Pathogen Y. ruckeri-Based upon our previous studies of holomycin biosynthesis (8), we have identified a gene segment (from yruck0001_11670 to yruck0001_11830) in the chromo-some of the fish pathogen Y. ruckeri ATCC29473 that appears to be responsible for the biosynthesis of dithiopyrrolone derivatives. Analysis of the sequence and functional annotation showed that several ORFs in this segment appear to be homologues of genes in the holomycin pathway of S. clavuligerus, including one FAD-dependent oxidoreductase, one FMN-dependent decarboxylase, one acyl-CoA dehydrogenase, one thioesterase, and one multidomain nonribosomal peptide synthetase. The nonribosomal peptide synthetase contains a unique arrangement of cyclization (Cy), adenylation (A) and thiolation (T) domains ( Fig. 2 and Table 1). The A domain of the nonribosomal peptide synthetase was predicted to activate the amino acid cysteine (23).
Differences in the gene clusters between Y. ruckeri and S. clavuligerus were also evident. First, at one end of the gene cluster of Y. ruckeri, we identified a hom12 gene encoding a SAM-dependent methyltransferase that is not present in the holomycin cluster in S. clavuligerus (Fig. 2). Second, no homologue encoding a disulfide oxidase could be identified in the genome of Y. ruckeri. The disulfide oxidase plays an important role in holomycin production in S. clavuligerus (7,8). Finally, hom15 encoding the cold-shock-like protein was identified at one end of the gene cluster (Fig. 2). Temperature is one of the major stresses that microorganisms have to face. A number of coldshock proteins are induced to counteract the harmful effects of temperature downshift. The induction of a cold-shock protein also affects microbial cellular physiology (24). Therefore, coldshock proteins could play an important role in protecting Y. ruckeri during a temperature downshift. The physical location of hom15 in the cluster implies that the encoded Hom15 may play a unique protective role in expressing the holomycin pathway at low temperatures.
Y. ruckeri Is a Holomycin Producer-To assess the outcome of the gene cluster of interest, Y. ruckeri was cultured in TSB medium under aerobic and anaerobic conditions at 22°C for 48 h. The color of the ethyl acetate extract cultured under aerobic conditions was more yellow than the one under anaerobic  (Fig. 3, S and T) led to the identification of an extra peak in the sample under the aerobic culture conditions (Fig. 3S). The corresponding m/z was also identical to the one from holomycin (calculated molecular formula ϭ C 7 H 7 O 2 N 2 S 2 , MϩH ϭ 214.9944, observed MϩH ϭ 214.9941, ␦ ϭ Ϫ0.397 ppm) (Fig. 3R). To confirm the presence of holomycin, a large scale fermentation (2 liters) was performed. The ethyl acetate extract was then concentrated under vacuum followed by normal phase chromatography with chloroform and methanol (9:1, (v/v)) as solvents. The yellowish fractions were collected for semipreparative HPLC purification, which yielded 3 mg of pure compound. Inspection of 1 H and 13  showed that the yellowish compound was holomycin 1, confirming that Y. ruckeri is a holomycin producer.

Analysis of the Genes Involved in the Formation of Holomycin-
To correlate the holomycin production with the identified gene cluster, an internal fragment of hom6 was used for the single crossover mutagenesis using a suicide pK18mobGII (19). Comparison with HPLC profiles between the wild-type (Fig. 3A) and the mutant Yru⌬hom6 strain (Fig. 3H) indicated that the production of the yellow holomycin was abolished in the mutant Yru⌬hom6 (Fig. 3H), suggesting that this gene cluster directs the biosynthesis of holomycin.
Hom1 belongs to a large protein family of flavin-dependent phosphopantothenoylcysteine decarboxylase (PPC-DC) in coenzyme A biosynthesis. Hom1 has a high homology (61% identity) with HomD in the gene cluster of S. clavuligerus (Fig.  2) that was proposed to catalyze the thioaldehyde intermediate 4 to the dithiol form 3 (Scheme 1) (7,8). To examine the function of hom1 during the biosynthesis of holomycin, a single crossover mutation was performed, generating the mutant strain Yru⌬hom1. HPLC analysis of the organic extract of Yru⌬hom1 indicated that despite significant perturbation being observed, a low level of holomycin production was still recovered in Yru⌬hom1 (Fig. 3E), suggesting that decarboxylation occurred even without the presence of Hom1 in the pathway.
Analysis of Regulatory and Protective Genes Involved in the Biosynthesis of Holomycin-The gene hom8 encodes a MFS (major facilitator superfamily) efflux protein that has high homology with HomH (61% identity) in the cluster from S. clavuligerus, indicating that the gene could be a pathway-specific regulatory gene. Interestingly, we observed that the holomycin production is temperature-dependent. The temperature suitable for the holomycin production ranges from 16 to 22°C. No holomycin production was, however, observed at 28°C or above. These observations raised the question of whether a temperature sensor could protect the biosynthesis of holomycin at the low temperatures. The gene hom15 encodes a putative cold-shock-like protein. To examine the possible function of Hom15, a double-crossover mutagenesis strategy was used to generate the mutant Yru⌬hom15. Both the wild-type strain and the mutant were then cultured at two different temperatures (16 and 22°C) to assess the possible perturbation. Comparison of HPLC data of holomycin production between the wild-type strain (Fig. 3A) and the mutant Yru⌬hom15 (Fig. 3E) revealed a marked loss of holomycin production in the mutant at 22°C. No holomycin was observed at 16°C in the mutant Yru⌬hom15 (Fig. 3G), whereas the holomycin production appeared to be unaffected in the wild-type strain (Fig. 3B).
Analysis of the Resistance Gene Involved in the Biosynthesis of Holomycin-Curiously, toward one end of the 17-kb gene cluster resides the gene hom12 encoding a putative methyltransferase. Bioinformatics analysis indicated that Hom12 belongs to a large protein family of uncharacterized RNA-methyltransferase from other Gram-negative bacteria. Its physical location associated with the biosynthetic gene cluster of holomycin strongly suggests its involvement in the resistance through target modification (RNA in this case), a common strategy associated with antibiotic resistance (25). To assess the hypothesis, a single crossover experiment was conducted. HPLC analysis of the extract indicated that holomycin production was greatly reduced in the resulting mutant strain Yru⌬hom12 (Fig. 3D). To assess the possible perturbation from the mutagenesis, the full-length of hom12 including the putative native transcription site was cloned into the broad-host plasmid pRK404 (26). The resulting construct was then reintroduced into Yru⌬hom12, generating Yru⌬hom12-hom12. Inspection of HPLC data demonstrated that the holomycin production was restored in Yru⌬hom12-hom12 (Fig. 3C). Hom12 is, therefore, involved in the biosynthesis of holomycin.
To assess whether Hom12 is involved in the antibiotic resistance, we performed agar diffusion assays. Incubation of holomycin with the wild-type and mutated strains demonstrated that the wild-type strain displayed no inhibition zone (Fig. 3I), whereas deletion of hom12 rendered the mutant strain Yru⌬hom12 sensitive toward holomycin (Fig. 3K). The complementation strain Yru⌬hom12-hom12, however, completely restored the resistance toward holomycin (Fig. 3J).
To confirm the function of Hom12, the full-length of the gene was cloned and introduced into BL21 (DE3) Gold. The expression of Hom12 in E. coli was performed in the presence of IPTG (1 mM) overnight at 16°C. Incubated with holomycin (35 g/ml, w/v), the wild-type BL21 (DE3) Gold showed susceptibility toward the antibiotic in agar diffusion assays (Fig. 3l), whereas E. coli harboring hom12 displayed no inhibition zone (Fig. 3M).
To further confirm that Hom12 is an RNA methyltransferase, the total RNAs were isolated from the wild-type Y. ruckeri, the mutant Yru⌬hom12, the wild-type E. coli, and the E. coli harboring hom12. The digested RNAs were then subjected to HR-LC-MS analysis. Comparison of the HR-LC/ESI/MS chromatography led to identification of two MS peaks with identical mass in the wild-type E. coli (Fig. 3P) and the mutant Yru⌬hom12 (Fig. 3O) (Fig. 3, O and P, respectively). Interestingly, we also observed two identical ions from the wild-type Y. ruckeri (Fig.  3N) and the E. coli harboring hom12 (Fig. 3Q) (Fig. 3N) and observed M-H ϭ 1339.2103, ␦ ϭ Ϫ2.367 ppm for the E. coli harboring hom12 (Fig. 3Q)). We did not, however, observe any ion of 1325.197 that was correlated to unmethylated RNA species in the samples of the wild-type Y. ruckeri (Fig.  3N) and the E. coli harboring hom12. Taken together with our bioinformatics analysis and HR-LC/ESI/MS analysis, Hom12 is, therefore, an RNA methyltransferase.

DISCUSSION
Holomycin is a broad-spectrum antibiotic natural product previously isolated mainly from actinomycete strains (1,27). Recently, the bioassay-guided isolation has led to the rediscovery of holomycin from a marine Gram-negative bacterium P. halotolerans S2753 isolated from the southern Pacific Ocean (2). In the present report we identified that the fish pathogen Y. ruckeri ATCC29743 is also a holomycin producer through genome mining, chemical isolation, and characterization approaches. Genetic deletion of hom6 abolished the production of holomycin, strongly supporting the involvement of the gene cluster in holomycin biosynthesis. Bioinformatics analysis showed that the gene cluster directing the biosynthesis of holomycin in Y. ruckeri is different from the one in S. clavuligerus. No obvious ORF in this gene cluster can be assigned to the disulfide oxidase responsible for the formation of a reduced dithiolopyrrolone intermediate to holomycin. A similar absence was also observed in the thiomarinol gene cluster from the marine Gram-negative bacterium Pseudoalteromonas sp SANK 73390 (6), suggesting that the underlying chemical logic of disulfide formation is different in the Gram-negative bacteria.
Hom1 is likely to function in an analogous way to PPC-DC in coenzyme A biosynthesis. In the reaction cycle of PPC-DC, the thiol moiety of pantothenoylcysteine is suggested to be first oxidized by FMN to form the thioaldehyde group-containing intermediate, and then a spontaneous decarboxylation event occurs by using the thioaldehyde as an intramolecular nucleophile, resulting in the formation of the pantothenoylaminoethenethiol intermediate. Finally, the enethiol intermediate is reduced by FMNH 2 , completing the reaction cycle to form the final product pantothenoylcysteamine (28). In the proposed mechanism of holomycin production (7,8,9), the thioaldehyde intermediate 3, however, should be ready for decarboxylation, suggesting no involvement of FMN as the redox agent. In this case, with the absence of Hom1, the thioaldehyde intermediate 3 could spontaneously progress the intramolecular nucleophilic decarboxylation to form a small amount of enethiolgroup intermediate 4 for the rest of the biosynthetic steps (Scheme 1).
The cold-shock response and adaptation is important for bacteria in response to temperature downshift. In E. coli and Bacillus subtilis, a number of cold-shock proteins (Csp) were found to be induced to counteract the harmful effects on cellu-lar physiology. For example, E. coli CspE is mainly produced constitutively at 37°C. In a strain of E. coli in which other coldshock genes were deleted, the CspE level increased upon cold shock. The deletion of all of cold-shock genes including cspE resulted in a cold-sensitive E. coli mutant, demonstrating that CspE is necessary for cellular adaptation to cold. Y. ruckeri should also need cold-shock adaptation strategies for a temperature downshift. Gene disruption of the cold-shock-like gene hom15 resulted in a significant loss of holomycin production at the low temperatures. However, there was no growth difference between the wild-type and the mutant, implying that hom15 is likely to be a pathway-specific gene for Y. ruckeri in response to low temperature.
One of the common antibiotic mechanisms is to target the RNA (29). To prevent the cytotoxic effects of antibiotics during the antibiotic production, many antibiotic-producing bacteria have evolved various coping strategies, including methylation of key RNA nucleotides at the drug target site, drug modification, and active efflux of the drug from the cell. The most common self-protection is to add methyl groups at or close to the functionally important sites of RNA (29). Bacterial-resistant methylation tightly correlates specific resistance to certain antibiotics, such as aminoglycoside, macrolides, and lincosamides (29). The presence of the hydrophobic methyl groups in ribosomal RNA changes the physiochemical properties and the overall shape of the bacterial ribosomal surface and thus leads to altered affinity to ribosome-inhibiting antibiotics. For example, the macrolide antibiotic tylosin has been used widely in veterinary medicine. The mode of action for tylosin has been long attributed to binding in the peptide exit tunnel of the bacterial 50 S ribosomal subunit, hence inhibiting protein synthesis by interfering with peptide bond formation and blocking the passage of the nascent peptide chain through the tunnel (30,31). To prevent being inhibited during the antibiotic production, the tylosin-producing bacterial strain Streptomyces fradiae recruits three methyltransferases (TlrA, TlrB, and TlrD) and an efflux pump (TlrC) (32)(33)(34). The functions of these three methyltransferases during antibiotic production have been elucidated where the methylated positions of nucleotides in 23 S rRNA of S. fradiae coincides with the glycosylation patterns of tylosin, the manner reflecting exactly how the macrolide fit into its binding site within the bacterial 50 S rRNA.
Holomycin has been reported to display bacteriostatic activity against Gram-negative and Gram-positive bacteria. It appeared that the antibacterial activity of holomycin against E. coli was associated with rapid inhibition of RNA synthesis in vivo (4). Some of the dithiolopyrrolones appeared to act as inhibitors of DNA-dependent RNA synthesis (5). However, direct inhibition of RNA polymerase has yet to be reconstituted in vitro, implying that holomycin could be a prodrug, requiring conversion in the cell to the active free thiol forms to inhibit RNA synthesis (4). Some synthetic dithiolopyrrolone derivatives also have strong activity against human cancer cell lines, although the cellular target(s) has not been identified (35). More recently, Li and Walsh (9) demonstrated that the mutant S. clavuligerus strains, in which the dithiol oxidase was deleted, showed significant reduction of holomycin production and increased susceptibility toward holomycin. To self-protect the accumulation of toxic dithiol intermediates, the mutant S. clavuligerus recruit a methylation strategy for the detoxification to generate a range of unique metabolites incapacitated by monoand di-S-methylation (10).
Our genetic experiments demonstrated that the gene hom12 plays an important role during the holomycin biosynthesis in Y. ruckeri. Chemical analysis, however, indicated that hom12 does not act as an N-methyltransferase to generate thiolutin 3. Gene disruption of hom12 resulted in the mutant strain Yru⌬hom12 being susceptible toward holomycin (Fig. 3K), whereas the wild-type strain and complementation strain Yru⌬hom12-hom12 displayed tolerance toward holomycin (Fig. 3, I and J). Transferring hom12 into the E. coli BL21(DE3) Gold strain caused the resultant E. coli strain to be resistant toward holomycin (Fig. 3M), whereas the wild-type E. coli displayed sensitivity against holomycin (Fig. 3L). The encoded Hom12 belongs to a large group of uncharacterized SAM-dependent RNA methyltransferases in various Gram-negative bacteria. The HR-LC/ESI/MS analysis indicated that both E. coli harboring hom12 and the wild-type Y. ruckeri strains, which are resistant to holomycin 1, contain a methylated RNA fragment, but no unmethylated RNA fragment could be identified in both RNA samples (Fig. 3, N and P, respectively). The strains sensitive toward holomycin instead contain the unmethylated RNA fragment but not the methylated RNA fragment (Fig. 3, O and P, respectively). Taken together these results demonstrate that Y. ruckeri likely recruits a common strategy to confer the resistance and prevent inhibition during holomycin production by methylating its own RNA. Notably, a significantly reduced production of holomycin was still observed in the mutant strain Yru⌬hom12, indicating that the mutant can tolerate the low concentration of holomycin produced. More recently a similar result was also observed that heterologous expression of the holomycin gene cluster from S. clavuligerus in Streptomyces coelicolor M1154 resulted in a low production of holomycin, and the inserted gene cluster in the resultant mutant S. colicolor (pVR-hol1) did not increase resistance to holomycin, indicating that S. colicolor (pVR-hol1) itself does not possess the holomycin-resistant gene (36).
Interestingly, holomycin was also active against rifamycinresistant strains (4), suggesting that the target site(s) of RNA for holomycin is different from the one for rifamycin. Interestingly holomycin was a poor inhibitor of E. coli RNA polymerase when a synthetic poly(dA-dT) template was used (5), suggesting that the inhibition occurs with a specific recognition site(s) in the RNA sequence. The methylation in the recognition site(s) of the RNA sequence could alter the binding affinity of holomycin toward RNA, hence preventing the cytotoxic effect of holomycin during holomycin production. Indeed, the binding site(s) in the RNA of Y. ruckeri could be similar to the binding pattern of holomycin in other bacteria, reflecting the manner in which holomycin fits into its binding site within other bacterial RNAs. Future studies on the exact RNA species and the position(s) of the methylation site of RNA will shed light on the mode of action of holomycin. It is not known why the fish pathogen Y. ruckeri produces holomycin or whether the production is associated with pathogenicity in this species. One possibility is that the antibacterial nature of holomycin may allow Y. ruckeri to compete with other bacteria for survival, and this will be tested as part of future studies.