Iron acquisition in the marine actinomycete genus Salinispora is controlled by the desferrioxamine family of siderophores

Abstract

Many bacteria produce siderophores for sequestration of growth-essential iron. Analysis of the Salinispora genomes suggests that these marine actinomycetes support multiple hydroxamate- and phenolate-type siderophore pathways. We isolated and characterized desferrioxamines (DFOs) B and E from all three recognized Salinispora species and linked their biosyntheses in S. tropica CNB-440 and S. arenicola CNS-205 to the des locus through PCR-directed mutagenesis. Gene inactivation of the predicted iron-chelator biosynthetic loci sid2-4 did not abolish siderophore chemistry. Additionally, these pathways could not restore the native growth characteristics of the des mutants in iron-limited media, although differential iron-dependent regulation was observed for the yersiniabactin-like sid2 pathway. Consequently, this study indicates that DFOs are the primary siderophores in laboratory cultures of Salinispora.

Introduction

Siderophores are small molecules secreted by bacteria to sequester growth-essential ferric iron that is poorly soluble under neutral pH and aerobic conditions (Neilands, 1995). The structures of siderophores vary considerably and are often suited to the environmental niche of the producing bacterium. For example, amphiphilic siderophores possess hydrophobic fatty acid chains that enable them to remain associated with the cell membrane (Martinez et al., 2003) – an attribute particularly advantageous in pelagic marine environments where dilution occurs rapidly. Some bacteria produce multiple siderophores as a competitive advantage by providing alternate ferric-uptake pathways not utilized by neighboring microbes (Challis and Hopwood, 2003).

Genome analysis of the obligate marine actinomycetes Salinispora tropica (Udwary et al., 2007) and Salinispora arenicola (Penn et al., 2009), suggested that they possess multiple siderophore-like biosynthetic loci. Four pathways are predicted in S. tropica CNB-440, whereas only two are retained in S. arenicola CNS-205. Both species maintain a des locus that likely codes for desferrioxamine (DFO) and a sid2 locus related to the gene cluster for yersiniabactin biosynthesis, ybt (Gehring et al., 1998). Intriguingly, ybt is usually encoded on a high pathogenicity island that mobilizes between pathogenic Gram-negative bacteria to confer virulence (Buchrieser et al., 1998; Schubert et al., 1998; Flannery et al., 2009). S. tropica CNB-440 also encodes two additional nonribosomal peptide synthetase (NRPS) pathways, sid3 and sid4, which are hypothesized to provide unique salicylate-containing iron-chelators similar to dihydroaeruginoic acid (Carmi and Carmeli, 1994) and the predicted “coelibactin” (Bentley et al., 2002).

DFOs are hydroxamate-type siderophores with a high affinity for iron (K_d ~10⁻³¹ M) (Keberle, 1964) that are produced by streptomycetes (Müller and Raymond, 1984; Barona-Gómez et al., 2004) and some Gram-negative bacteria (Martinez et al., 2001; Essén et al., 2007). Several analogs have been reported including linear DFOs B, D and G and cyclic DFO E (Fig. 3A), as well as acyl-DFO analogs with terminal branched alkyl chains or aromatic rings (D’Onofrio et al., 2010; Yang et al., 2011).

Fig. 3.

Effect of des⁻ and sid2⁻ mutagenesis on siderophore production in Salinispora. A. Structures of DFO B and DFO E. B. HPLC analysis of i) DFO standards, and extracted culture supernatants from ii) S. tropica, iii) S. arenicola and iv) S. pacifica wildtype and mutant strains grown under iron limitation. Chromatograms indicate UV absorbance of extracts monitored by HPLC at 435 nm.

DFOs are biosynthesized via an NRPS-independent mechanism (Challis, 2005), encoded by desA-D (Barona-Gómez et al., 2004; Kadi et al., 2007). Transcription from des is repressed by the divalent metal-dependent regulatory protein DmdR1, and derepressed by iron-limitation (Flores and Martín, 2004; Tunca et al., 2007). Predicted homologs to desA-D and the ferric-siderophore uptake and utilization genes (desE-F) are found in both Salinispora genomes (Fig. 1A).

Fig. 1.

Putative siderophore-like gene clusters, des and sid2, from S. tropica CNB-440 and S. arenicola CNS-205. A. Organization of the des gene cluster compared to S. coelicolor A3(2). Putative DmdR repressor binding sequences are indicated by double-headed arrows. B. Organization of the yersiniabactin gene cluster in Y. pestis and the sid2 gene clusters in S. tropica CNB-440 and S. arenicola CNS-205. Black, structural genes; grey, transport and receptor genes; white, other genes.

Despite bioinformatic predictions on the siderophores produced by Salinispora, no iron-chelators have been isolated from this genus. Therefore we explored the siderophore chemistry of these marine actinomycetes to determine which of the putative siderophore biosynthetic loci play a role in iron acquisition in Salinispora.

Methods

Bacterial strains and culturing

S. tropica strain CNB-440, S. arenicola strains CNS-205, CNT088 and CNH643 and “S. pacifica” strain CNT133 were cultured at 30 °C with continuous shaking at 200 rpm in iron-limited media (1 g L⁻¹ NH₄Cl, 2 g L⁻¹ casamino acids, 28 g L⁻¹ Instant Ocean (Aquarium Systems Inc.), 0.6% v/v glycerol), supplemented with 36 μM FeSO₄ when required.

PCR-directed mutagenesis and siderophore analysis

PCR-targeting (Gust et al., 2003; Eustáquio et al., 2008) was used to inactivate genes: stro2548/sare2737 (desD homolog), stro2655/sare2071 (sid2 NRPS1), stro2806 (sid3 NRPS) and stro2821 (sid4 NRPS). For primers see Table S1 (Supporting Information).

Wild-type and mutant S. tropica, S. arenicola and “S. pacifica” were grown to stationary phase in iron-limited media, the cells were removed by centrifugation and the supernatant acidified to pH 2 with H₂SO₄. Amberlite XAD-7 resin was added to 2% w/v, and shaken at 150 rpm for 4 h. The resin was washed with ultrapure water and compounds were eluted with acetone, vacuum-dried and dissolved in methanol. The presence of iron-chelators in the total cultures and extracted supernatants was determined by Chrome Azurol S (CAS) assay (Schwyn and Neilands, 1987).

Transcript analysis

Total RNA was extracted from duplicate, stationary phase Salinispora cultures. Harvested cells were resuspended in RNAwiz (Ribopure Bacteria Kit, Ambion) and lysed via bead beating with zirconia beads (Fast Prep, Savant) for 5 × 30 s at speed 5.5. After centrifugation, proteins were removed by chloroform extraction and nucleic acids purified via Ribopure Bacteria Kit filter cartridges. Contaminating DNA was degraded with 8 U DNase I (Ambion) for 5 h, and PCR confirmed its complete removal.

For cDNA synthesis, 1 μg RNA was pooled from duplicate samples in a 40 μL reaction with 100 ng random hexamers, RT buffer, 5 mM MgCl₂, 10 mM DTT, 80 U RNaseOUT and 400 U Superscript III reverse transcriptase (Invitrogen). The reaction was incubated for 10 min at 25 °C, 50 min at 50 °C and 5 min at 85 °C. cDNA was used in triplicate RT-PCR reactions with initial denaturation at 94 °C for 2 min, followed by 30 cycles of 94 °C for 30 s, 55 °C for 45 s and 72 °C for 30 s, and a final extension at 72 °C for 5 min. Amplicons were analyzed with ethidium bromide on a 2% agarose gel. Targeted genes were stro2551/sare2740 (desA), stro2654/sare2072 (polyketide synthase, PKS), stro2806 (NRPS) and stro2821 (NRPS). For primers see Table S1 (Supporting Information).

Siderophore isolation and identification

Supernatants from late stationary phase Salinispora cultures were extracted with XAD-7 resin, and CAS assays followed the positive siderophore fractions throughout purification. Crude extracts were dried under vacuum, resuspended in methanol, and fractionated via reversed-phase HPLC with a gradient of acetonitrile with 0.1% formic acid (0–5 min, 10%; 5–30 min, 50%; 30–50 min, 90%), using a Waters preparative C₁₈ column (25 mm × 200 mm) with a flow rate of 15 ml/min. DFO E, which eluted at 18 min, was further purified by washing the dried pellet twice in a minimal volume of methanol. DFO B eluted at 5 min.

High-resolution MS analysis of DFO B and E was performed by FT-ICR-MS and MS/MS fragmentation via collision-induced dissociation. Samples were mixed with methanol:water:formic acid (49:50:1), and injected by an Advion nanomate-electrospray ionization robot in positive ion mode with a Thermo Finnigan LTQ-FT-ICR mass spectrometer after external mass calibration. The structure of purified DFO E was confirmed by ¹H NMR in d₆-DMSO using a 500 MHz Varian Oxford AS500 spectrometer.

Analysis of conserved Salinispora DmdR repressor binding sequences

Blastp was used to identify putative DmdR repressors from Salinispora. EMBOSS Palindrome analysis (http://emboss.bioinformatics.nl/) was used to locate palindromic sequences upstream of desE and desF. Blastn analysis identified sequences with similarity to the DmdR binding consensus from Streptomyces coelicolor A3(2) (Flores and Martín, 2004).

Siderophore uptake assays

The S. tropica des mutant was grown in A1 media (10 g L⁻¹ potato starch, 4 g L⁻¹ yeast extract, 2 g L⁻¹ peptone, 28 g L⁻¹ Instant Ocean) with 36 μM FeSO₄ to mid log phase, and 300 μL cells were spread on iron-limited media plates. To test siderophore uptake, triplicate filter discs with 10 μg DFO E or yersiniabactin (EMC Microcollections GmbH, Germany), 130 μg FeSO₄, or ultrapure water were placed on overlays.

Results and Discussion

Growth and siderophore production in des and sid2-4 mutants

To probe the function of the putative siderophore biosynthetic loci in S. tropica CNB-440 and S. arenicola CNS-205, we inactivated each by insertional inactivation of critical structural genes. In both species the des mutants grew poorly in iron-limited media, the growth of which was visible after two months. When supplemented with FeSO₄, the des mutant growth improved with cultures growing within two weeks, compared to wild-type cultures that grew densely and reached stationary phase within one week in iron-replete media. In contrast, mutagenesis of sid2-4 did not alter the growth or phenotype of the cells in either iron-limited or iron-sufficient conditions.

CAS assays determined the presence of iron-chelators in wild-type and mutant cultures. Mutagenesis of the des cluster, but not sid2-4, abolished CAS activity in S. tropica CNB-440 in the both the total culture and extracted supernatant. Similarly, disruption of des, but not sid2, resulted in the loss of CAS activity in the S. arenicola CNS-205 total culture and supernatant. These observations suggest that the primary growth-essential iron-chelator in Salinispora laboratory cultures is a siderophore associated with the des locus.

Transcriptional analysis of putative siderophore pathways

To confirm the lack of activity from sid2-4, we used RT-PCR to determine the conditions under which the putative siderophore biosynthetic loci were transcribed (Fig. 2). The des gene cluster was transcribed in both species under iron-limitation and repressed under iron-replete conditions. Interestingly, the sid2 transcript was detected in iron-limited S. tropica CNB-440, whereas it was only identified in iron-replete S. arenicola CNS-205. Transcript was detected under iron-limitation from most genes within the S. tropica CNB-440 sid2 gene cluster (Table 1), except for a methyltransferase (stro2056), FAD-dependent monooxygenase (stro2658), hypothetical protein (stro2659) and putative transporters (stro2651-2). Despite the detection of sid2 transcript in both strains, no chemotypic differences were detected by analytical HPLC in the extracts of mutants compared with the wild-type.

Fig. 2.

RT-PCR analysis of selected genes within the siderophore gene clusters under iron-replete (+Fe) and iron-limited (−Fe) conditions. −, negative control without RNA; +^a/+^b, +Fe/−Fe cDNA controls with 16S primers; +^c, PCR control with genomic DNA.

Table 1.

Transcriptional analysis of the S. tropica CNB-440 sid2 gene cluster under iron-replete (+Fe) or iron-limited (− Fe) growth conditions

Gene locus	Homolog	Transcript detected
		+ Fe	−Fe
Stro2660	CoA ligase	−	+
Stro2659	Hypothetical protein	−	−
Stro2658	Multicopper oxidase	+	+
Stro2657	Monooxygenase, FAD-binding	−	−
Stro2656	Methyltransferase	−	−
Stro2655	NRPS (PCP/Cy/A/PCP)	−	+
Stro2654	PKS (KS/AT/KR/ACP/Cy/TE)	−	+
Stro2653	Extracellular solute binding protein	−	+
Stro2652	Inner membrane transport component	−	−
Stro2651	Inner membrane transport component	−	−
Stro2650	ATP-binding cassette transporter	−	+
Stro2649	Methyltransferase	−	+
Stro2648	Methyltransferase	−	+
Stro2647	NRPS (PCP/Cy/PCP)	−	+

The sid3 and sid4 gene clusters were not transcribed under iron-limited or -replete conditions. These gene clusters may no longer be functional or may require other factors for activation. In fact, although “coelibactin” has not been isolated from S. coelicolor A3(2), it is thought to be a zinc-regulated signaling molecule that regulates antibiotic production (Hesketh et al., 2009) and sporulation (Kallifidas et al., 2010).

Siderophore isolation and identification

CAS assay-guided fractionation of S. tropica CNB-440 and S. arenicola CNS-205 wild-type cultures resulted in the isolation of two iron-chelators. These compounds were identified as DFO B (M_obs 560.35341 Da, M_calc 560.35336 Da, Δm <0.1 ppm) and DFO E (M_obs 600.3491 Da, M_calc 600.34828 Da, Δm =1.4 ppm) by high resolution FT-ICR-MS and FT-ICR-MS/MS (Fig. S1, Fig. S2). In the case of DFO E, we further confirmed its structure by ¹H NMR, via comparison with reported chemical shift data (Bergeron and McManis, 1990). CAS-activity based fractionation did not identify any other DFO analogs. DFO E was the most abundant siderophore detected from Salinispora, with 7 mg L⁻¹ purified from S. tropica CNB-440. DFO B was detected at ten-fold lower yields than DFO E. Inactivation of the desD gene in both species abolished the production of both DFO analogs (Fig. 3), verifying the gene clusters’ involvement in DFO production in Salinispora. DFO B and E were also detected in iron-limited cultures from other S. arenicola isolates (CNT-088 and CNH-643), while DFO E was produced by “S. pacifica” CNT-133, further confirming the conservation of this dominant family of siderophores in Salinispora.

While DFO production is characteristic of Salinispora and many streptomycetes (Müller and Raymond, 1984; Meiwes et al., 1990), it is not a general trait amongst all Actinomycetales (Nett et al., 2009). Notably, Saccharopolyspora (Oliynyk et al., 2007), Nocardia (Ishikawa et al., 2004) and Frankia (Udwary et al., 2011) encode various siderophore pathways, none of which include des. Although Salinispora are obligate marine organisms, they are isolated from marine sediments (Mincer et al., 2002; Maldonado et al., 2005) where the secretion of hydrophilic siderophores would not be as rapidly diluted as in the water column. In fact, DFO production has been reported from various bacteria isolated from marine sediments including Citricoccus (Kalinovskaya et al., 2011) and Micrococcus luteus (D’Onofrio et al., 2010). This specialized habitat may explain why Salinispora biosynthesize the same siderophores as soil-dwelling actinomycetes, rather than the amphiphilic siderophores produced by many pelagic microbes (Martinez et al., 2000; Xu et al., 2002; Martinez et al., 2003). Additionally, Salinispora may decompose organic materials in marine sediments (Jensen et al., 2005), akin to actinomycetes in terrestrial soils, which would support the similar requirement for DFO-type siderophores. The lack of amphiphilic siderophores produced by Salinispora may therefore be a limiting factor in its proliferation into other environmental niches, such as the water column. Interestingly, the recently reported genome sequence of aquatic Citricoccus strain CH26A predicted a novel acylation mechanism in its production of DFO (Hayano-Kanashiro et al., 2011). This biosynthetic pathway may therefore be evolutionarily distinct from other reported DFO pathways.

Analysis of Salinispora DmdR repressor binding sequences

Blastp analysis revealed a putative DmdR repressor in S. arenicola CNS-205 (Sare_1414) and S. tropica CNB-440 (Strop_1456), with 62/63% identity and 72/73% similarity to DmdR1 in S. coelicolor A3(2) (Flores and Martín, 2004). Blastn and EMBOSS Palindrome analyses identified four putative DmdR binding sites (iron boxes) in each of the Salinispora genomes. Two of the iron boxes are upstream of desE and desF in both species (Fig. 1). The des gene cluster organization is conserved in Streptomyces (Barona-Gómez et al., 2006) with all six genes in one locus whereas, in Salinispora, desF is 13–21 kb upstream of desEABCD. Despite these differences, iron-repression of des is consistent in both genera, as confirmed in Salinispora by transcript analysis (Fig. 2). The remaining two iron boxes are upstream of a periplasmic binding protein similar to ferric-enterobactin transporters, and a putative siderophore utilization protein in StBac1/SaBac2, which may encode a Class I bacteriocin (Penn et al., 2009). No iron boxes were identified near sid2-4. Alignment of the eight putative iron box sequences enabled the prediction of a DmdR binding consensus sequence for Salinispora: AGGyTAACCTA (Fig. 4).

Fig. 4.

A. Sequence alignment of the four putative DmdR binding sequences in S. tropica CNB-440 and S. arenicola CNS-205, and comparison with the S. coelicolor A3(2) consensus sequence (Flores and Martín, 2004). Capital letters indicate invariant residues between the eight Salinispora spp. iron boxes. Lower case letters indicate less-conserved residues. B. Diagrammatical representation of the DmdR binding consensus sequence using Weblogo (Schneider and Stephens, 1990; Crooks et al., 2004).

Sid2 gene cluster analysis

Although sid2 from S. tropica CNB-440 was transcribed in iron-limited cultures, the lack of detectable siderophores in the des mutants, and their poor growth without iron, suggests that the sid2 compound was either not produced in detectable quantities or that it is unable to chelate iron. As iron-supplementation increased the growth of the des mutant, another iron-chelator may be produced at very low levels or with a lower affinity for iron than CAS, which would not be detected by our methods. Due to the differential transcriptional response of sid2 to iron in the des mutants, however, it is unlikely that this additional iron-chelator is associated with sid2.

Sid2 possesses similarity to ybt (Bearden et al., 1997; Pelludat et al., 1998); however, there are several differences between the two gene clusters (Fig. 1B). The three methyltransferases in sid2 are not integrated into the NRPS/PKS genes, and several essential ybt genes are absent in sid2, namely the reductase ybtU, salicylate synthase ybtS and regulator ybtT (Fig. 1B) (Geoffroy et al., 2000; Miller et al., 2002), which may explain the lack of yersiniabactin-like siderophore production.

Sid2 in S. arenicola CNS-205 is transcribed under iron-replete rather than iron-limited conditions, although no chemotypic difference was detected between the wild-type and sid2 mutant in iron-sufficient conditions. The altered transcriptional regulation may be due to mobilization of the sid2 cluster 846 kb downstream on a separate genomic island. The putative CoA ligase remains in the original locus with respect to the S. tropica CNB-440 sid2 gene cluster (Fig. 5). No CoA ligase (sare2861) transcript could be detected under either iron-replete or -limited conditions (data not shown), in contrast to the corresponding gene (stro2660) in S. tropica CNB-440 (Table 1). Further studies are required to fully understand how genetic rearrangements have altered the transcriptional regulation of sid2 in Salinispora.

Fig. 5.

Linear alignment of the S. arenicola CNS-205 and S. tropica CNB-440 genomes showing the mobilization of part of the sid2 gene cluster in S. arenicola CNS-205 (adapted from (Penn et al., 2009)).

Siderophore uptake assays

Although a sid2 iron-chelator was not produced in laboratory cultures of Salinispora, it was unknown whether sid2 transporters could uptake exogenous siderophores produced by other microbes. Functional transporters can import xenosiderophores in some bacteria that do not produce the iron-chelators (Yun et al., 2000; Yamanaka et al., 2005).

Therefore, we carried out siderophore uptake studies to determine whether S. tropica CNB-440 is able to utilize yersiniabactin, despite being unable to produce this siderophore.

The S. tropica des mutant was grown on iron-limited artificial sea water plates supplemented with DFO E, yersiniabactin, water or FeSO₄ on filter discs. DFO-supplementation supported confluent growth of the mutant on the entire plate (> 45 mm radius), confirming the role of this siderophore in growth-essential iron sequestration for S. tropica CNB-440. This result also confirms that the DFO-iron uptake receptors and utilization enzymes (desE (Patel et al., 2010; Tierrafría et al., 2011) and desF (Barona-Gómez et al., 2006)) are functional in this actinomycete, despite the desF gene residing 13.8 kb upstream of the remaining des genes.

Supplementation with FeSO₄ promoted growth of the S. tropica des mutant immediately around the edge of the filter disc (2 mm radius); however, the mutant strain was unable to grow on water-only (blank) control plates confirming the importance of des and DFO in iron acquisition. Exogenous yersiniabactin was unable to promote the growth of the des mutant, which suggests that the sid2 transport proteins are not functional or not specific for yersiniabactin uptake.

In conclusion, although several siderophore-like biosynthetic loci are predicted within the Salinispora genomes, DFOs are the major species involved in iron sequestration in this obligate marine genus, and are essential for the growth of the organism under iron-limitation. Many bacteria produce multiple iron-chelators as a competitive advantage; therefore, the lack of diverse siderophores identified in Salinispora may possibly be compensated by the rich secondary metabolism of this genus to enable successful colonization in marine sediments. Further work, including expression in heterologous hosts, will be required to determine the chemistry associated with the unique sid2-4 pathways. Finally, this study reinforces the importance of genetic and chemical evidence in confirming the function of gene clusters that are identified via genome sequence-based mining.